Structure-based design of 7-carbamate analogs of geldanamycin

Article abstract of DOI:10.1016/j.bmcl.2005.08.013

The 7-carbamate groups of geldanamycin and its 17-(2-dimethylaminoethyl) amino-17-demethoxy derivative (17-DMAG) bind the N-terminal domain of Hsp90 by establishing a network of hydrogen bonds which involve four buried water molecules. In this study, a structure-based approach was used to investigate the effects of displacing some of these waters by modification of the 7-carbamate. A general loss of binding to human Hsp90 was observed, except for replacement of the carbamate with a hydroxamate group which gave an analog with weak activity. Modeling of Hsp90-ligand interactions suggested that the hydroxamate was not able to displace the buried water molecules, while bulkier substituents able to do so proved inactive.

Full text of DOI:10.1016/j.bmcl.2005.08.013

Bioorganic & Medicinal Chemistry Letters 15 (2005) 5016–5021
Structure-based design of 7-carbamate analogs of geldanamycin
Giulio Rastelli,b Zong-Qiang Tian,a Zhan Wang,a David Mylesa and Yaoquan Liua,*
aKosan Biosciences, Inc., 3832 Bay Center Place, Hayward, CA 94545, USA
`
Dipartimento di Scienze Farmaceutiche, Universita di Modena e Reggio Emilia, Via Campi 183, 41100 Modena, Italy
b
Received 29 June 2005; accepted 1 August 2005
Available online 13 September 2005
Abstract—The 7-carbamate groups of geldanamycin and its 17-(2-dimethylaminoethyl)amino-17-demethoxy derivative (17-DMAG)
bind the N-terminal domain of Hsp90 by establishing a network of hydrogen bonds which involve four buried water molecules. In
this study, a structure-based approach was used to investigate the effects of displacing some of these waters by modification of the
7-carbamate. A general loss of binding to human Hsp90 was observed, except for replacement of the carbamate with a hydroxamate
group which gave an analog with weak activity. Modeling of Hsp90–ligand interactions suggested that the hydroxamate was not
able to displace the buried water molecules, while bulkier substituents able to do so proved inactive.
Ó 2005 Elsevier Ltd. All rights reserved.
Geldanamycin, a benzoquinone ansamycin, binds to the
N-terminal domain ATP binding site of heat shock pro-
tein 90 (Hsp90), inhibiting the chaperone activity of the
protein.1,2 By regulating the function and stability of
many key signaling proteins, Hsp90 is of particular
importance to the survival of cancer cells. To stabilize
its client proteins, the chaperone assembles with other
cochaperones and associated proteins to form a Ôsuper-
chaperoneÕ complex. The disruption of the Hsp90 client
protein complexes leads to the ubiquitination and subse-
quent proteasomal degradation of the client proteins.
Because many of the client proteins are oncogenic,
Hsp90 is an attractive target for anticancer chemothera-
py. Indeed, the geldanamycin analogs 17-allylamino-17-
demethoxygeldanamycin (17-AAG) and 17-[2-(dimeth-
ylamino)ethyl]amino-17-demethoxygeldanamycin (17-
DMAG) have both entered human clinical trials.3 The
crystal structures of human Hsp90–geldanamycin and
Hsp90–17-DMAG complexes provided several features
of the interaction,4–6 one of which was that the 7-carba-
mate of geldanamycin is located deep in the binding
pocket and coordinates four buried water molecules
through a hydrogen-bonding network with residues
D93, G97, T184, S52, I91, V186, L48, and N51
(Fig. 1A). The few structure–activity relationships avail-
able suggested that the 7-carbamate substituent of
geldanamycin plays an important role in cytotoxic activ-
ity. Removal of the carbamate or replacement with urea,
thiourea or oxalate reduced cytotoxicity dramatically.7
Disruption of the carbamate hydrogen bond network
by D79N mutation in yeast Hsp90, which corresponds
to D93N mutation in the human enzyme, resulted in loss
of binding.8 Notwithstanding, whether or not the carba-
mate is absolutely required for activity has not been
investigated in sufficient detail, as SARs are still limited
to a very small number of compounds.
In light of the hydrogen bonding network shown in Fig-
ure 1A, we reasoned that binding affinity could be in-
creased by introducing additional hydrogen bonding
groups onto the ligand to displace one or more of the
water molecules that coordinate the carbamate of gel-
danamycin, so that they interact directly with the pro-
tein. The rationale for this is that displacing water
molecules at the protein–ligand interface is, in general,
entropically favored. The novobiocin and clorobiocin
inhibitors of DNA gyrase B provide a good example
of this. Indeed, the crystal structures of human and yeast
Hsp90 N-domain revealed structural homology and se-
quence conservation in the ATP sites of Hsp90s and
DNA gyrase B.9,10 Novobiocin and clorobiocin both
are potent DNA gyrase B inhibitors.11–13 Interestingly,
novobiocin has a carbamate functionality which estab-
lishes, with gyrase B, a hydrogen-bonding network sim-
ilar to that of the carbamate of geldanamycin with
Hsp90 (Fig. 1B), including the four buried water mole-
cules. In clorobiocin, the 5-methyl-2-pyrrolcarboxylate
Keywords: Geldanamycin; Hsp90; 17-AAG; 17-DMAG; Structure-
based design.
*
Corresponding author. Tel.: +1 510 732 8400; fax: +1 510 732
8401; e-mail: liu@kosan.com
0960-894X/$ - see front matter Ó 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2005.08.013

G. Rastelli et al. / Bioorg. Med. Chem. Lett. 15 (2005) 5016–5021
5017
O
group, which replaces the carbamate group of novobio-
cin, binds DNA gyrase B with nearly identical orienta-
tion, but three of the four water molecules are
displaced (Fig. 1C). Interestingly, isothermal titration
calorimetry (ITC) experiments suggested that clorobio-
cin binds gyrase B with 30-fold better affinity than novo-
biocin, and that while novobiocin binding is entropically
disfavored, that of clorobiocin is both enthalpically and
entropically favored.10
A
N51
NH
O
NH2
W4
T184
O
OH
O
O
L48
NH
W3
NH2
W1
O
W2
NH
O
O
V186
G97
NH
Because of the structural homology between Hsp90 and
gyrase B, we reasoned that carbamate replacements in
geldanamycin should also be attempted. By modeling
Hsp90–ligand interactions on a set of 17-DMAG deriv-
atives modified at the carbamate using the crystal struc-
ture of the human Hsp90–17-DMAG complex5 as
starting point, we found that polar extensions from the
carbamate nitrogen could replace one or more bound
water molecules without disrupting the rest of the
hydrogen bond network. The complexes were modeled
with a molecular mechanics and molecular dynamics ap-
proach in water using the sander classic module of AM-
BER614 and the Cornell et al.15 force-field. A carbazide
derivative was first proposed. Referring to the com-
pounds reported in Table 1, modeling predicted that
the 7-carbazide derivative 8 might displace water mole-
cule w3 from the complex. Compared with the 17-
DMAG hydrogen bonding network reported in Figure
2A, the predicted binding mode of this analog has the
7-carbazide carbonyl hydrogen bonded with w1 which,
in turn, hydrogen bonds with D93, G97, and T184
(Fig. 2B). Likewise, the adjacent amino function hydro-
gen bonds with D93 and the terminal amino group of
the carbazide is predicted to interact directly with the
L48 backbone carbonyl and with w2 and w4. On the
contrary, models of the 7-hydroxamate derivative 5 (Ta-
ble 1) in complex with Hsp90 predicted that the
hydroxamate would not be able to displace water mole-
cules (Fig. 2C). A molecular dynamics simulation in
which w3 was removed showed that w2 readily moved
to occupy the void created by the removal of w3; evi-
dently, the 7-hydroxamate substituent would not be able
to reach and effectively replace w3. The hydrogen bonds
formed by the hydroxamate are predicted to be similar
to those of the 17-DMAG complex, with two additional
(long) hydrogen bonds formed by the hydroxyl group of
the hydroxamate with w3 and w4. The most significant
difference is that the bulkier hydroxamate substituent
OH
S52
D93
O
O
HN
I91
B
Y94
O
NH2
OH
NH
O
HO
O
N45
W4
T166
O
O
OH
O
I42
W3
NH2
W1
NH
O
NH
O
O
V168
NH
W2
G76
D72
O
O
HN
V70
C
Y94
O
O
NH2
OH
NH
O
HO
O
N45
O
T166
OH
O
I42
NH
˚
is predicted to force w3 1.2 A far from the position occu-
W1
NH
O
pied in the Hsp90–geldanamycin and Hsp90–17-DMAG
complexes; however, this displacement could be tolerat-
ed by Hsp90, with a minor shift of the nearby V186 side
chain.
NH
O
O
V168
NH
G76
D72
O
O
Significant differences were noted for the pyrrole deriva-
tive 17 (Table 1), which is predicted to bind Hsp90 by
displacing water molecules w3 and w4 (Fig. 2D). Except
for the 5-methyl group, this compound mimics the
novobiocin/clorobiocin substitution reported for gyrase
B. Hydrogen bonds with D93, either direct via the ami-
no of the pyrrole or w1-mediated via the carbonyl, were
still predicted. The pyrrole ring was still compatible with
the presence of w2, as molecular dynamics simulations
HN
V70
Figure 1. (A) Hydrogen-bonding network established between the 7-
carbamate of geldanamycin in complex with human Hsp90; (B)
hydrogen bonding network established between the carbamate of
novobiocin and DNA gyrase B; (C) hydrogen bonding network
established between the 5-methyl-pyrrole ring of clorobiocin and DNA
gyrase B.

5018
G. Rastelli et al. / Bioorg. Med. Chem. Lett. 15 (2005) 5016–5021
Table 1. Structure and biological activities of the 7-carbamate analogs
Compound
R2 or R3
R1
Kd (lM) FL.Hsp90
IC50 (nM) SKBr3
1
2
–NH2
–NH2
–NHCH2CH2N(CH3)2
–NHCH2CH2F
–Azetidinyl
0.5
0.8
24
17
3
–NH2
–NH2
1.4
0.1
24
33
4
–NH2
5
–NHOH
–NHOH
–NHCH2CH2N(CH3)2
–NHCH2CH2F
–Azetidinyl
18
220
1500
640
140
2100
530
420
2500
1400
390
490
460
410
1900
1200
4500
6
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
>100
7
–NHOH
–NHNH2
8
–NHCH2CH2N(CH3)2
–NHCH2CH2F
–Azetidinyl
9
–NHNH2
–NHNH2
–NHCH3
10
11
12
13
14
15
16
17
18
19
20
–NHCH2CH2N(CH3)2
–NHCH2CH2N(CH3)2
–NHCH2CH2N(CH3)2
–NHCH2CH2N(CH3)2
–NHCH2CH2N(CH3)2
–NHCH2CH2N(CH3)2
–NHCH2CH2N(CH3)2
–NH2
–N(CH3)NH2
–NHCH2CONH2
–NHCH2CH2OH
–NHOCH3
–NHCH2CH2N(CH3)2
2-Pyrrole
2-Pyrrole
4-Imidazole
Cyclic carbamate
–NHCH2CH2N(CH3)2
–NHCH2 CH2N(CH3)2
in which w2 was removed showed that the pyrrole did
not reach this position.
The 11-hydroxyl group is more sterically hindered than
the 7-hydroxyl group, resulting in the observed selectiv-
ity for the 7-hydroxyl. Given the diminished activity ob-
served for the pyrrole-2-carboxylate analog and other
carbamate analogs, the 5-methylpyrrole-2-carboxylate
ester17 derivative of geldanamycin was not pursued
further.
To examine these possibilities, these and other deriva-
tives, collected in Table 1, were synthesized and tested
for binding affinity to full length human Hsp90 as well
as cytotoxicity in SKBr3 mammalian cells. Since substi-
tution at position 17 of geldanamycin largely affects the
cytotoxicity of the analogs with little effect on the in vitro
binding to purified Hsp90,16 the 7-carbamate modifica-
tions were coupled with the four most potent 17-substit-
The biological activities (Table 1) of the analogs synthe-
sized were evaluated using the Hsp90-binding assay and
tumor cell growth inhibition assay reported in a previ-
ous study.16 The binding affinity of these analogs for
purified Hsp90 was measured using a scintillation prox-
imity assay previously described.16 Analogs of 17-
DMAG and other 17-aminogeldanamycins with modifi-
cations at the 7-positions ranging from simple substitu-
tion at the carbamate nitrogen to replacing the
carbamoyl group with a pyrrole-2-carboxylate group
showed diminished binding affinity to purified human
Hsp90. The only exception was the 7-hydroxamate
derivative 5, which retains binding to Hsp90
(Kd ꢀ 18 lM), though weaker than 17-DMAG
(Kd ꢀ 0.5 lM), and has an IC50 of 220 nM. However,
the 7-hydroxamate substituent is predicted to be unable
to displace water molecules from the binding site. All
other analogs were practically inactive (Kd > 100 lM)
in the binding assay. The cyclic carbamate analog 20
was also inactive. Several compounds (5, 8, 11, and
14–17), however, retained cytotoxicity to a certain ex-
tent, with an IC50 less than 500 nM in SKBr3 cells.
uents:
17-[2-(dimethylamino)ethyl]amino-,
17-(2-
fluoroethyl)amino-, 17-azetidin-1-yl-, and 17-amino-
17-demethoxygeldanamycins. As shown in Scheme 1,
analogs 1–4 in Table 1 were readily synthesized in good
yields by treating geldanamycin (GDM) with the corre-
sponding amines in 1,2-dichloroethane.16 The 7-carba-
mate group was removed by treating the 17-
alkylamino-17-demethoxygeldanamycins with 2.5 equiv
of potassium t-butoxide in DMSO at room tempera-
ture.7 Decarbamoylation of 17-DMAG gave the expect-
ed 7-hydroxy product along with a minor product 20,
presumably formed via Michael addition of the carba-
mate nitrogen to C-5. However, decarbamoylation
of 17-(2-fluoroethyl)amino-17-demethoxylgeldanamycin
under the same condition resulted in decomposition of
the material. The desired 7-hydroxyl intermediate was
synthesized from 7-O-decarbamoylgeldanamycin and
2-fluoroethylamine. 7-Carbazides and 7-hydroxamates
5–10 were prepared by reacting the 7-O-decarbamoyl
intermediate with 1,10-carbonyldiimidazole (CDI) and
subsequently trapping the CDI-adduct with hydrazine
or hydroxylamine, respectively. Treatment of the CDI
adduct of 7-O-decarbamoyl-17-DMAG with alkyl-
amines yielded a series of N-substituted carbamate ana-
logs 11–16. The pyrrole-2-carboxylate and imidazole-4-
carboxylate esters 17–19 were prepared by dicyclohexyl-
carbodiimide (DCC) coupling of the 7-hydroxy interme-
diates with the appropriate carboxylic acids (Scheme 1).
In light of the poor biological results obtained by mole-
cules designed to displace water molecules in the 17-
DMAG series, the hydrogen-bonding network between
the carbamate and Hsp90 appears crucial, if not essen-
tial. Despite the positive result obtained with clorobio-
cin binding to gyrase B, in Hsp90 the only example of
water displacement remains radicicol, another Hsp90
inhibitor, which binds Hsp90 by displacing one of the

G. Rastelli et al. / Bioorg. Med. Chem. Lett. 15 (2005) 5016–5021
5019
Figure 2. Structures of Hsp90–ligand complexes at the bottom of the active site. The four buried water molecules are noted as w1–w4; for clarity,
only polar hydrogens are shown. (A) Hsp90–17-DMAG complex; (B) Hsp90–8 complex; (C) Hsp90–5 complex; (D) Hsp90–17 complex.
four water molecules (w4).6 Of course, we cannot ex-
clude that substituents different from the ones presented
here would be able to do so, and further work will be
needed to investigate this possibility.
predicted to displace one water molecule, does not
make stronger hydrogen bonds than the interfacial
water w3 because it is more constrained by attachment
to the rest of the ligand. Since the macrocycle of gel-
danamycin binds Hsp90 by almost entirely filling the
binding pocket and establishing a number of hydro-
phobic contacts and hydrogen bonds, there is little
room for conformational adjustment when additional
hydrogen bonding groups are introduced. Novobiocin
and clorobiocin, in this respect, are significantly more
flexible than geldanamycins and might be better able
to do so. Second, it is possible that the large confor-
mational changes required to bring the macrocycle of
geldanamycin derivatives in a conformation suited
Many issues are still open: first of all, whether it is
best to treat water molecules at the interface between
protein and ligand as part of the structure of the pro-
tein with a ligand designed to hydrogen bond to the
water, or whether it is best to gain entropy from dis-
placement of water by introducing a hydrogen bond-
ing group onto the ligand. This is a choice that is
influenced by many factors.18 For example, it is possi-
ble that the 7-carbazide of compound 8, which was

5020
G. Rastelli et al. / Bioorg. Med. Chem. Lett. 15 (2005) 5016–5021
O
O
O
NH2
OMe
NH2
O
O
O
OMe
NH
R1H
MeO
MeO
MeO
H CO
3
OH
OH
OH
O
O
O
O
O
O
NH
NH
NH
N
MeO
R1
N
H
O
O
O
1-4
20
KOt-Bu, DMSO
KOt-Bu, DMSO
FCH2CH2NH2
O
O
R3
OMe
R2
OMe
O
HO
O
OMe
1) CDI
2) HR2
DCC, R3CO2H
MeO
MeO
MeO
OH
OH
OH
O
O
O
O
O
O
NH
NH
NH
R1
R1
R1
O
O
O
5-16
17-19
Scheme 1. Synthesis of 7-carbamate derivatives.
for binding5 might be perturbed by modifying the
carbamate. Finally, another issue is the cross-talk
between the carbamate derivatives and the 17-substitu-
ents. Unexpectedly, while the hydroxamate 5 retained
some binding for Hsp90, hydroxamates with different
substituents at position 17 (6, 7) are inactive. There
is no clear explanation for this, because the 7- and
17-positions are distant and the 17-substituents are
solvent exposed. However, manual docking of gel-
danamycin 17-derivatives into the newly released crys-
tal structure of E. coli Hsp90 containing both the
N-terminal and the middle domains19 suggests that
the 17-substituents would not be as solvent exposed
as originally inferred from the structure of the N-ter-
minal domain alone, rather they significantly interact
with residues of the middle domain. Therefore, the
binding process might be more complicated than
expected.
References and notes
1. Pearl, L. H.; Prodromou, C. Curr. Opin. Struct. Biol. 2000,
10, 46.
2. Prodromou, C.; Pearl, L. H. Curr. Cancer Drug Targets
2003, 3, 301.
3. Sausville, E. A.; Tomaszewski, J. E.; Ivy, P. Curr. Cancer
Drug Targets 2003, 3, 377.
4. Stebbins, C. E.; Russo, A. A.; Schneider, C.; Rosen, N.;
Hartl, F. U.; Pavletich, N. P. Cell 1997, 89, 239.
5. Jez, J. M.; Chen, J. C.; Rastelli, G.; Stroud, R. M.; Santi,
D. V. Chem. Biol. 2003, 10, 361.
6. Roe, S. M.; Prodromou, C.; OÕBrien, R.; Ladbury, J.
E.; Piper, P. W.; Pearl, L. H. J. Med. Chem. 1999, 42,
260.
7. Schnur, R. C.; Corman, M. L.; Gallaschun, R. J.; Cooper,
B. A.; Dee, M. F.; Doty, J. L.; Muzzi, M. L.; DiOrio, C. I.;
Barbacci, E. G.; Miller, P. E.; Pollack, V. A.; Savage, D.
M.; Sloan, D. E.; Pustilnik, L. R.; Moyer, J. D.; Moyer,
M. P. J. Med. Chem. 1995, 38, 3813.
8. Obermann, W. M. J.; Sondermann, H.; Russo, A. A.;
Pavletich, N. P.; Hartl, F. U. J. Cell Biol. 1998, 43, 901.
9. Lamour, V.; Hoermann, L.; Jeltsch, J.-M.; Oudet, P.;
Moras, D. J. Biol. Chem. 2002, 277, 8947.
Acknowledgments
10. Lafitte, D.; Lamour, V.; Tsvetkov, P. O.; Makarov, A. A.;
Klich, M.; Deprez, P.; Moras, D.; Briand, C.; Gilli, R.
Biochemistry 2002, 41, 7217.
11. Gormley, N. A.; Orphanides, G.; Meyer, A.; Cullis, P. M.;
Maxwell, A. Biochemistry 1996, 35, 5083.
We thank John R. Carney and Nina Viswanathan for
the analytical chemistry support, Wei Ma for the
assistance in cytotoxicity assays and Kwok Yu for
the assistance in the Hsp90-binding assays.
12. Kampranis, S. C.; Gormley, N. A.; Tranter, R.; Orphan-
ides, G.; Maxwell, A. Biochemistry 1999, 38, 1967.
13. Schio, L.; Chatreaux, F.; Loyau, V.; Murer, M.; Ferreira,
A.; Mauvais, P.; Bonnefoy, A.; Klich, M. Bioorg. Med.
Chem. Lett. 2001, 11, 1461.
Supplementary data
Supplementary data associated with this article can be
found in the online version at doi:10.1016/j.bmcl.
2005.08.013
14. Case, D. A.; Pearlman, D. A.; Caldwell, J. W.; Cheatman,
I.; Ross, W. S.; Simmerling, C. L.; Darden, T. A.; Merz,

G. Rastelli et al. / Bioorg. Med. Chem. Lett. 15 (2005) 5016–5021
5021
K. M.; Stanton, R. V.; Cheng A. L.; Vincent, J. J.;
Crowley, M.; Tsui, V.; Radmer, R. J.; Duan, Y.; Pitera, J.;
Massova, I.; Seibel, G. L.; Singh, U. C.; Weiner, P. K.;
Kollman, P. A. AMBER 6.0, University of California, San
Francisco, USA, 1999.
16. Tian, Z.-Q.; Liu, Y.; Zhang, D.; Wang, Z.; Dong, S. D.;
Carreras, C. W.; Zhou, Y.; Rastelli, G.; Santi, D. V.;
Myles, D. C. Bioorg. Med. Chem. 2004, 12, 5317.
17. Kaiser, H.-P.; Muchowski, J. M. J. Org. Chem. 1984, 49, 4203.
18. Ward, W. H. J.; Holdgate, G. A. In Progress in Medicinal
Chemistry; King, F. D., Oxford, A. W., Eds.; Elsevier
Science: Amsterdam, 2001, pp 309–376.
15. Cornell, W. D.; Cieplak, P.; Bayly, C. I.; Gould, I. R.;
Merz, K. M., Jr.; Ferguson, D. M.; Spellmeyer, D. C.;
Fox, T.; Caldwell, J. W.; Kollman, P. A. J. Am. Chem.
Soc. 1995, 117, 5179.
19. Huai, Q.; Wang, H.; Liu, Y.; Kim, H.-Y.; Toft, D.; Ke, H.
Structure 2005, 13, 579.