Total synthesis as a resource in the discovery of potentially valuable antitumor agents: Cycloproparadicicol

Article abstract of DOI:10.1002/anie.200390329

A highly convergent synthesis was used to prepare a series of epimers and analogues of radicicol (1). Biological assessment revealed a previously unrecognized correlation between stereochemistry and antitumor potential in these compounds. Significantly, cycloproparadicicol (2) showed promising therapeutic properties based on inhibition of chaperones.

Full text of DOI:10.1002/anie.200390329

Communications
Synthesis of Antitumor Agents
served to heighten interest in Hsp90 as a target for oncol-
ogy.[3,4]
Radicicol (1),[5] originally isolated from M. bonorden,has
a high affinity for Hsp90 (Kd = 20 nm) and is particularly
potent in inhibiting its functions.[6] However,translation of
the very promising in vitro profiles of radicicol and radicicol
oxime to the in vivo realm has not been accomplished,[7]
apparently as a result of unfavorable pharmakokinetics and
nonspecific toxicities.
These early findings notwithstanding,we felt it worth-
while to persist in seeking a useful radicicol-based drug that is
not subject to the potential liabilities arising from quinone-
containing substructures such as those found in the geldana-
mycin series. We previously developed a highly convergent,
three-stage total synthesis of radicicol,[8] which was crucial to
the work described herein.
Our investigation of the radicicol series focused on two
aspects. The first was directed to the effect,if any,of the
stereochemistry of the system on its biological properties. We
hoped to examine binding to Hsp90 as well as cytotoxicity. We
saw such a study as a way to probe the areas of the drug that
make important contacts with the chaperone. Assessing the
consequences of altering the stereochemistry at particular
points in a complex drug may be a more subtle probe of the
stringency of presumably required molecular contacts,than
gross deletion of functional groups.
Total Synthesis as a Resource in the Discovery of
Potentially Valuable Antitumor Agents:
Cycloproparadicicol**
Kana Yamamoto, Robert M. Garbaccio,
Shawn J. Stachel, David B. Solit, Gabriela Chiosis,
Neal Rosen, and Samuel J. Danishefsky*
Hsp90 is a molecular chaperone required for the refolding of
proteins in cells exposed to environmental stress. It governs
the conformational maturation of a subset of proteins critical
in regulating mitotic signal transduction and cellular sur-
vival.[1] These client proteins include steroid receptors,tran-
scription factors,mutant p53,Hif1 a,soluble kinases such as
Akt,Raf-1,transmembrane kinases such as HER2,as well as
cdk4 protein kinases. Hsp90 contains an ATP-binding pocket
in its amino terminus. Several natural products,including
geldanamycin,herbimycin A,and radicicol bind to this
pocket and inhibit its chaperone function. This inhibition is
mirrored in enhanced intracellular proteasomal degradation
of Hsp90 client proteins. An encouraging consequence of this
effect is an attenuation in the proliferation of cancer cells. In
animal models,geldanamycin congeners exhibit antitumor
activity at nontoxic doses.[2] The novel mechanism of action of
geldanamycin-related drugs and the demonstration that they
have,in principle,exploitable in vivo therapeutic indices have
We also became interested in a cyclopropane derivative of
radicicol,that is,one in which the oxido linkage that connects
C7’ and C8’ is replaced by a methylene group (i.e.,a
cyclopropane analogue,see Scheme 4,
21–24).[9,10] The
thought was that the lack of in vivo efficacy of radicicol
might arise from its degradation (k1/2 < 2 hrs) or undirected
cytotoxicity emanating from the w-epoxydienone. Given the
structural information available[11] on the binding of radicicol
to the N-terminal ATP/ADP-binding domain of Hsp90,which
seems to identify a hydrogen bond between the epoxide and
Lys44,there was concern that the cyclopropane derivative
might suffer an unacceptable loss of affinity to its intended
target. In essence,both aims of this inquiry focused on a
careful assessment of Lys44 contact in the action of the drug.
Our previous total synthesis of 1 provided a method for
the preparation of compounds 2–4. As before,the synthesis
[*] Prof. Dr. S. J. Danishefsky, Dr. K. Yamamoto, Dr. R. M. Garbaccio,+
Dr. S. J. Stachel
Laboratory for Bioorganic Chemistry
Sloan-Kettering Institute for Cancer Research
1275 York Avenue, New York, NY 10021 (USA)
Fax: (+1)212-772-8691
E-mail: s-danishefsky@ski.mskcc.org
Prof. Dr. S. J. Danishefsky
Department of Chemistry, Columbia University
Havemeyer Hall, New York, NY 10027 (USA)
Dr. D. B. Solit, Dr. G. Chiosis, Prof. N. Rosen
Program in Cell Biology and Department of Medicine
Sloan-Kettering Institute for Cancer Research
1275 York Avenue, New York, NY 10021 (USA)
H
H
7'
O
7'
O
[þ] Current address:
O
O
10'
Cl
O
O
10'
8'
Merck Research Laboratories
West Point, PA 19486 (USA)
8'
H
H
HO
HO
[**] Support for this research was provided by the National Institutes of
Health (CA28824). Postdoctoral Fellowship support is gratefully
acknowledged by K.Y. (The Helen Hay Whitney Foundation), R.M.G.
(NIH, 1 F32 CA85894-01), and S.J.S. (NIH, 1 F32 CA81704). Dr.
George Sukenick (NMR Core Facility, Sloan-Kettering Institute) is
acknowledged for NMR and mass spectrometric analyses. We thank
Dr. Louis J. Todaro (The CUNY X-ray Facility, Hunter College) and
Dr. David G. Churchill (Department of Chemistry, Columbia Uni-
versity) for X-ray analyses.
O
O
Cl
OH
OH
1
2
H
H
7'
7'
O
O
O
O
10'
Cl
O
O
10'
8'
8'
H
H
HO
HO
O
O
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author. Experimental
details include the description of experimental procedures and
spectral data for all new compounds.
Cl
OH
OH
3
4
1280
ꢀ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1433-7851/03/4211-1280 $ 20.00+.50/0
Angew. Chem. Int. Ed. 2003, 42, No. 11

Angewandte
Chemie
featured a novel ring-closing olefin metathesis of a vinyl
epoxide.[12,13] Thus,alcohol 5 was converted into 6 or 7 by the
sequence described earlier (Scheme 1).[8] Coupling of 6 with
acid 8 under either Mitsunobu (C10’ inversion) or acylation
(C10’ retention) experimental protocols afforded esters 9 and
11,respectively. Similarly,when the two senses of coupling
were applied to alcohol 7,esters 10 and 12 were obtained. The
strategy used in the total syntheses of radicicol allowed access
to stereocongeners 2 , 3(ent-2),and 4 (ent-1),as shown in
Scheme 2.
The syntheses of the cyclopropane analogues commenced
with the same allylic alcohol precursor 5 used in our earlier
synthesis of radicicol and its stereocongeners. We now took
advantage of the asymmetric cyclopropanation chemistry
reported by Charette et al.[14,15] for the conversion of 5 into 13.
Alcohol 13 was converted into vinyl compound 14
(Scheme 3). Again acylation of 14 with 8 by using the two
stereodifferentiating protocols led to 17 and 19 in high
enantiomeric and diastereomeric purity. The conversion of 17
into the carba version of radicicol (cf. 21) was carried out as
shown. Conversion of 19 into 23 followed an identical route
(Scheme 4).
TBDPSO
OH
TBDPSO
OH
5
5
a
b
a
b
H
H
H
H
O
H
O
H
TBDPSO
TBDPSO
OH
OH
TBDPSO
TBDPSO
OH
OH
H
H
13
15
H
H
O
O
HO
HO
H
H
H
H
HO
HO
6
7
H
H
16
14
Scheme 1. Synthesis of the enantiopure epoxide side chains. a) (R,R)-
diisopropyltartrate, TiCl4, CH2Cl2, molecular sieves (4 Š), 71%;
b) (S,S)-diisopropyltartrate, TiCl4, molecular sieves (4 Š ), 76%.
TBDPS=tert-butyldiphenylsilyl.
Scheme 3. Synthesis of the enantiopure cyclopropyl side chains.
a) (R,R)-dioxaborolane, Et2Zn, CH2I2, 69%; b) (S,S)-dioxaborolane,
Et2Zn, CH2I2, 72%.
O
H
O
H
O
O
O
O
H
Cl
H
Cl
HO
HO
2
1
a
a
O
OH
OTBDPS
OTBDPS
HO
Cl
9
10
7 +
+ 6
O
H
O
H
O
O
O
O
OTBDPS
H
Cl
H
Cl
HO
HO
8
4
3
b
b
OTBDPS
11
OTBDPS
12
H
O
O
S
H
O
O
O
O
S
Li
S
H
steps
H
HO
RCM
1
HO
9
S
S
S
OTBDPS
OTBDPS
Scheme 2. Divergent synthesis of radicicols. a) DIAD, P(fur)3, benzene, 75% (9), 76% (10); b) EDCI, DMAP, CH2Cl2, 41% (11), 44% (12).
DIAD=diisopropylazodicarboxylate, fur=furyl, EDCI=1-(3-dimethylaminopropyl)-3-ethylcarbodiimide hydrochloride, DMAP=4-dimethylamino-
pyridine, RCM=ring-closing olefin metathesis.
Angew. Chem. Int. Ed. 2003, 42, No. 11
ꢀ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1433-7851/03/4211-1281 $ 20.00+.50/0
1281

Communications
H
H
7'
H
H
7'
O
O
O
O
O
O
10'
O
O
10'
Cl
8'
8'
H
Cl
H
Cl
H
H
HO
HO
HO
HO
a
a
O
O
Cl
OTBDPS
OTBDPS
OH
OH
22
18
17
21
16 +
8
+ 14
H
7'
H
7'
H
H
O
O
O
O
10'
10'
O
O
O
O
8'
8'
H
H
H
Cl
H
Cl
HO
HO
HO
HO
O
O
b
b
Cl
Cl
OH
OH
OTBDPS
OTBDPS
19
24
20
23
H
H
O
O
O
O
S
Li
S
H
H
HO
RCM
21
HO
17
S
OTBDPS
S
S
S
OTBDPS
Scheme 4. Divergent synthesis of cycloproparadicicols. a) DIAD, P(fur)3, benzene, 48% (17), 39% (28); b) EDCI, DMAP, CH2Cl2, 54% (19),
61% (20).
In a similar way 16 was obtained from 5 by using the
enantiotopic Charette catalyst (Scheme 3). The conversion of
16 into 18 and 20 followed the same pathway. These,in turn,
gave rise to 22 and 24,respectively,again following closely the
strategy and protocols used to reach 21. Thus the four
radicicol stereocongeners 1–4 as well as their respective carba
variants 21–24 became available.
require Hsp90 for maturation or stability. The expression
levels of Raf1 and HER2,two of the most sensitive protein
targets,was analyzed with the above radicicol variants
(Figure 1). The radicicol-like action of 21 is also reflected in
its potent degradation of oncogenic proteins. Interestingly,
the stereochemical trend found in the Hsp90-binding assay,
was also observed in the degradation level of the oncogenic
proteins in both series. It seems that the actions of 1 and 21 on
Hsp90 appear to occur in the same recognition setting.
Subsequently,these radicicol variants were evaluated for
their growth-inhibition activity towards MCF-7 breast-cancer
cells (Table 1). As expected,the same configurational trend
The radicicol congeners were tested for binding with
Hsp90 by means of a binding-competition assay.[4] The parent
natural product 1 binds with high affinity to Hsp90 (ED50 =
45 nm) as reported previously. When 21,the carba version of
the radicicol series,was tested for Hsp90 binding,it was found
to still retain high affinity (ED50 = 160 nm),that is,a fourfold
reduction from 1. Therefore the hydrogen bond found
between the epoxide and the Lys44 of the chaperone in the
X-ray structure,while possibly useful,is not critical. Com-
pound 22,in which the C7 ’ and C8’ stereocenters are inverted,
exhibits binding to a much lower extent (ED50 = 2 mm).
Compound 23 with inversion at the C10’ stereocenter suffered
a further decrease in its binding (ED50 = 5 mm). Inversion of
all these stereocenters (see 24 = ent-21),undermines the
affinity to Hsp90 still further (ED50 = > 10 mm).[16] It would be
tempting to argue that in the cycloproparadicicols,and
probably in the radicicol series,the protein Hsp90 target
recognizes the gross substructure of the 14-membered lactone
fused to the resorcinol. Supporting recognition elements are
found in the methyl group at C10’ and at the cyclopropyl
(epoxide) group,in the “natural” series when these elements
are in an optimal stereochemical setting. Inversion of either of
these loci,diminishes binding and biological activity. Inver-
sion at both of the supporting sectors leads to severe loss of
recognition.
Figure 1. Immunoblot of Raf1 and HER2 expression of MCF-7 cells fol-
lowing 24 h of exposure to the indicated drugs.
Table 1: IC50 (nm) values of radicicol and cycloproparadicicol analogues
in MCF7 cells.
Entry
Epoxide
IC50 [nm][a]
Cyclopropyl
IC50 [nm][a]
1
2
3
4
1
2
3
4
23
902
1720
2172
21
22
23
24
43
836
2142
3364
As previously reported,the inhibition of Hsp90 induces
the proteasomal degradation of a subset of proteins that
[a] Determined after 96 h of exposure to the indicated drugs.
1282
ꢀ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1433-7851/03/4211-1282 $ 20.00+.50/0
Angew. Chem. Int. Ed. 2003, 42, No. 11

Angewandte
Chemie
was also observed in this assay. Thus, 2 exhibits intermediate
cell-growth inhibition (20-fold less),whereas 3 showed further
decrease in their potency (50-fold less),and 4 was found to be
even less active (~ 100-fold less). Most significantly,cyclo-
propane analogue 21,which has the same stereochemical
motif as radicicol (1), retains the potent cytotoxicity of 1
within a factor of two.
[1] For general reviews of Hsp90,see: a) S. Walter,J. Buchner,
Angew. Chem. 2002, 114,1142 – 1158; Angew. Chem. Int. Ed.
2002, 41,1098 – 1113; b) L. H. Pearl,C. Prodromou, Curr. Opin.
Struct. Biol. 2000, 10,1,46 – 51; c) J. Buchner, Trends Biochem.
Sci. 1999, 24,136 – 141.
[2] For our efforts involving geldanamycin,another natural product
which also binds to Hsp90,see: a) S. D. Kuduk,F. F. Zheng,L.
Sepp-Lorenzino,N. Rosen,S. J. Danishefsky,
Bioorg. Med.
As previously reported for ansamycins,growth inhibition
by radicicol (1) is a result of an Rb-dependent (Rb = retino-
blastoma) arrest in the G1 phase of the cell cycle (Figure 2).
Thus,MDA-468 breast cancer,which lacks Rb function,does
Chem. Lett. 1999, 9,1233 – 1238; b) S. D. Kuduk,C. R. Harris,
F. F. Zheng,L. Sepp-Lorenzino,Q. Ouerfelli,N. Rosen,S. J.
Danishefsky, Bioorg. Med. Chem. Lett. 2000, 10,1303 – 1306.
[3] a) L. Neckers, Trends Mol. Med. 2002, 8,4,S55-S61; b) M. V.
Blagosklonny, Leukemia 2002, 16(4),455 – 462; c) T. Scheibel,J.
Buchner, Biochem. Pharmacol. 1998, 56,675 – 682; D. J.
Newman,G. M. Cragg,S. Holbeck,E. A. Sausville,
Curr.
Cancer Drug Targets 2002, 2,279 – 308.
[4] A novel class of synthetic nucleotide analogues was shown to
bind to Hsp90 and inhibit the proliferation of cancer cells; see:
a) G. Chiosis,M. N. Timaul,B. Lucas,P. N. Munster,F. F. Zheng,
L. Sepp-Lorenzino,N. Rosen, Chem. Biol. 2001, 8,289 – 299;
b) G. Chiosis,B. Lucas,A. Shtil,H. Huezo,N. Rosen,
Bioorg.
Med. Chem. Lett. 2002, 10,3555 – 3564.
[5] P. Delmontte,J. Delmontte-PlaquØe, Nature 1953, 171,344 – 347.
[6] a) S. V. Sharma,T. Agatsuma,H. Nakano, Oncogene 1998, 16,
20,2639 – 2645; b) T. W. Schulte,S. Akinaga,T. Murakata,T.
Agarsuma,S. Sugimoto,H. Nakano,Y. S. Lee,B. B. Simen,Y.
Argon,S. Felts,D. O. Toft,L. M. Neckers,S. V. Sharma,
Mol.
Endocrinol. 1999, 13(9),1435 – 1448.
[7] Some derivatives of microbial radicicol were reported to exhibit
efficacy in animal models: a) Y. Shiotsu,L. M. Neckers,I.
Wortman,W. G. An,T. W. Schulte,S. Soga,C. Murakata,T.
Tamaoki,S. Akinaga, Neoplasia 2000, 96,6,2284 – 2291; b) S.
Soga,S. V. Sharma,Y. Shiotsu,M. Shimizu,H. Tahara,K.
Yamaguchi,Y. Ikuina,C. Murakata,T. Tamaoki,J. Kurebayashi,
T. W. Schulte,L. M. Neckers,S. Akinaga, Cancer Chemother.
Pharmacol. 2001, 48,435 – 445; c) S. Soga,L. M. Neckers,T. W.
Schulte,Y. Shiotsu,K. Akasaka,H. Narumi,T. Agatsuma,Y.
Ikuina,C. Murakata,T. Tamaoki,S. Akinaga, Cancer Res. 1999,
59,2931 – 2938; d) T. Agatsuma,H. Ogawa,K. Akasaka,A.
Asai,Y. Yamashita,T. Mizukami,S. Akinaga,Y. Saitoh, Bioorg.
Med. Chem. 2002, 10,3445 – 3454.
[8] a) R. M. Garbaccio,S. J. Danishefsky, Org. Lett. 2000, 2,20,
3127 – 3129; b) R. M. Garbaccio,S. J. Stachel,D. K. Baeschlin,
S. J. Danishefsky, J. Am. Chem. Soc. 2001, 123,10903 – 10908.
[9] The electophilicity of this system would allow Michael addition
of a bionucleophile to the dienone or attack of the nucleophile at
the epoxide linkage.[7b,d] Esterase action on the macrolactone is
still another potential site of vulnerability. In the epothilone
series,we found that agents lacking the epoxide manifest a much
improved biological profile.[10]
[10] a) T. C. Chou,X. G. Zhang,C. R. Harris,S. D. Kuduk,A. Balog,
K. Savin,J. R. Bertino,S. J. Danishefsky, Proc. Natl. Acad. Sci.
USA 1998, 95, 15798 – 15802; b) T. C. Chou,X. G. Zhang,A.
Balog,D. Su,D. F. Meng,K. Savin,J. R. Bertino,S. J. Dani-
shefsky, Proc. Natl. Acad. Sci. USA 1998, 95, 9642 – 9647.
[11] S. M. Roe,C. Prodromou,R. O'Brien,J. E. Ladbury,P. W. Piper,
L. H. Pearl, J. Med. Chem. 1999, 42,260 – 266.
[12] a) M. S. Stanford,J. A. Love,R. H. Grubbs, J. Am. Chem. Soc.
2001, 123,6543 – 6554,and references therein; b) for reviews,
see: A. Fꢀrstner,L. Ackermann,B. Gabor,R. Goddard,C. W.
Lehmann,R. Mynott,F. Stelzer,O. R. Thiel, Chem. Eur. J. 2001,
7,3236 – 3253; c) A. Fꢀrstner, Angew. Chem. 2000, 112,3140 –
3172; Angew. Chem. Int. Ed. 2000, 39,3012 – 3043; d) for an
important early example of the use of olefin metathesis for
macrocyclization in a functionally complex setting,see: Z. Xu,
Figure 2. Cell cycle distribution of MCF-7 or MDA-468 cell lines assessed
by fluorescence-activated cell sorter (FACS) analysis.
not arrest in G1,but instead is blocked in prometaphase prior
to undergoing apoptosis.[17] Inhibition of Rb-dependent
growth occurred at concentrations that correlate with those
required for protein degradation. Cells with intact Rb (MCF-
7) arrested in G1 and then underwent differentiation.
Importantly,Rb-negative cells (MDA-468) also underwent
apoptosis,in this case upon arrest in prometaphase. The same
propensity was observed with cycloproparadicicol 21.
In summary,these stereo and “carba” radicicol variants
were designed and first prepared in our laboratory following
the outlines of our concise total synthesis of radicicol.[8] The in
vitro biological testing of these compounds revealed previ-
ously unrecognized structure–activity relationships. Notably,
the oxygen atom of the epoxide of radicicol (1),in spite of its
apparent Lys44–epoxide hydrogen bonding in the governing
crystal structure in the radicicol field,is demonstrated to be
non-essential for its activity.[18] A qualitative though instruc-
tive recognition model can now be proposed to focus future
discovery-oriented investigations in this rather promising
series of anticancer agents. The work described above serves
to chart a direction by which organic synthesis in conjunction
with cell biology can be mutually enhancing in the quest for
novel drugs.
Received: December 13,2002 [Z50769]
Keywords: antitumor agents · chaperone proteins · natural
.
products · structure–activity relationships · synthesis design
Angew. Chem. Int. Ed. 2003, 42, No. 11
ꢀ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1433-7851/03/4211-1283 $ 20.00+.50/0
1283

Communications
C. W. Johannes,S. S. Salman,A. H. Hoveyda, J. Am. Chem. Soc.
1996, 118,10926 – 10927.
[13] K. Yamamoto,K. Biswas,C. Gaul,S. J. Danishefsky,unpub-
lished results.
[14] a) A. B. Charette,S. Prescott,C. Brochu, J. Org. Chem. 1995, 60,
1081 – 1083; b) A. B. Charette,H. Juteau, J. Am. Chem. Soc.
1994, 116,2651 – 2652; c) A. B. Charette,H. Juteau,H. Lebel,C.
Molinaro, J. Am. Chem. Soc. 1998, 120,11943 – 11952; d) A. B.
Charette,H. Lebel, Org. Synth. 1999, 76,86 – 100.
[15] The configuration of the cyclopropane was initially assigned
based on Charette's empirical rule and later confirmed by X-ray
analysis of the derivative of 14 (see Supporting Information).
[16] This trend is consistent with molecular modeling (MOE 2001.01)
results (Chiosis,unpublished data).
[17] Importantly,the small-cell lung cancer,which is not effected by
geldanamycin variant are found to be Rb-negative.
[18] On the other hand,other hydrogen bonds in X-ray data appear
to be critical as the dimethyl ethers of 1 and its derivatives lost
their activities (data not shown).
1284
ꢀ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1433-7851/03/4211-1284 $ 20.00+.50/0
Angew. Chem. Int. Ed. 2003, 42, No. 11