4-Amino derivatives of the Hsp90 inhibitor CCT018159

Article abstract of DOI:10.1016/j.bmcl.2006.01.099

Novel piperazinyl, morpholino and piperidyl derivatives of the pyrazole-based Hsp90 inhibitor CCT018159 are described. Structure-activity relationships have been elucidated by X-ray co-crystal analysis of the new compounds bound to the N-terminal domain o

Full text of DOI:10.1016/j.bmcl.2006.01.099

Bioorganic & Medicinal Chemistry Letters 16 (2006) 2543–2548
4-Amino derivatives of the Hsp90 inhibitor CCT018159
Xavier Barril,a Mandy C. Beswick,a Adam Collier,a Martin J. Drysdale,a,*
Brian W. Dymock,a Alexandra Fink,a Kate Grant,a Robert Howes,a Allan M. Jordan,a
Andrew Massey,a Allan Surgenor,a Joanne Wayne,a Paul Workmanb and Lisa Wrighta
aVernalis Ltd, Granta Park, Cambridge CB1 6GB, UK
bCancer Research UK Centre for Cancer Therapeutics, The Institute of Cancer Research,
15 Cotswold Road, Sutton, Surrey SM2 5NG, UK
Received 5 January 2006; revised 18 January 2006; accepted 18 January 2006
Available online 9 February 2006
Abstract—Novel piperazinyl, morpholino and piperidyl derivatives of the pyrazole-based Hsp90 inhibitor CCT018159 are
described. Structure–activity relationships have been elucidated by X-ray co-crystal analysis of the new compounds bound to the
N-terminal domain of human Hsp90. Key features of the binding mode are essentially identical to the recently reported potent
analogue VER-49009. The most potent of the new compounds has a methylsulfonylbenzyl substituent appended to the piperazine
nitrogen, possesses an IC50 of less than 600 nM binding against the enzyme and demonstrates low micromolar inhibition of tumour
cell proliferation.
Ó 2006 Elsevier Ltd. All rights reserved.
O
O
Exposure of cells to stress, such as heat shock or oxida-
tive stress, results in the accumulation of molecular
chaperones, commonly known as heat shock proteins
(Hsps). Hsp90 has emerged over the last few years as
being of particular interest because of its role in the evo-
lution, development and disease pathology of cancer.1,2
Hsp90 is an ATP-dependent chaperone essential for the
maturation and activity of a varied group of proteins in-
volved in signal transduction, cell cycle regulation and
apoptosis.3,4 Although weak, the ATPase activity can
be measured and is selectively inhibited by the ansamy-
cin natural product geldanamycin and its synthetic
derivatives 17-AAG and 17-DMAG (Fig. 1).5,6 In vivo
and clinical data with 17-AAG support the hypothesis
that the Hsp90 family may be an appropriate target
for anti-cancer drug development.7–9 The chaperone
has therefore been the focus of several recent investiga-
tions10 and amongst these, aryl pyrazoles have been
reported in the literature as potent small molecule
inhibitors.11
R
O
Me
N
H
Me
4
5
MeO
Me
OH
MeO
Me
OCONH2
R =
17-DMAG
17-AAG
Geldanamycin -OCH3
-NHCH2CH2NMe2
-NHCH2CH=CH2
Figure 1. Ansamycin natural product Hsp90 inhibitors.
Hsp90 ATPase activity.12 Subsequently, discoveries
were made in this series which led to VER-49009.13 This
potent compound was found to display useful levels of
inhibition of tumour cell proliferation.
O
Cl
Our own efforts in this area revealed the 3,4-diaryl pyr-
azole CCT018159 (1), a small molecule inhibitor of
O
HO
HO
5'
O
HO
O
HO
N
Keywords: Hsp90; Cancer; Structure-based drug design; Pyrazole;
X-ray crystallography.
N
N
H
5
N
H
2
HN
*
Corresponding author. Tel.: +44 1223 895555; fax: +44 1223
895556; e-mail: m.drysdale@vernalis.com
CCT018159 (1)
VER-49009
0960-894X/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2006.01.099

2544
X. Barril et al. / Bioorg. Med. Chem. Lett. 16 (2006) 2543–2548
Herein, we describe further developments in this area,
revealing a novel series of 3-aryl, 4-aminopyrazoles
and characterisation of their binding mode by X-ray
co-crystallography.
This 40N is clearly important since morpholine 3 looses
nearly an order of magnitude of potency. The presence
of an acid centre in the vicinity of this position
(Asp54) may explain the preference for basic groups.
Optimisation of the screening hit 1 initially focused on
improving potency and identifying areas for the intro-
duction of groups to enhance solubility, through the
use of the co-crystal structure of 1 and its analogues
bound to Hsp90.12–15 The pyrazole 4-position bearing
the 1,4-benzodioxane does not appear to make impor-
tant interactions with protein.
Alkylation with small lipophiles (4 and 5) does not alter
potency but further exploration, particularly with benzyl
systems, such as 7 and methylsulfonylbenzyl 6, resulted
in significant potency gains of nearly 10-fold.
An inspection of the X-ray co-crystal of 6 bound to
Hsp90 reveals that the sulfone moiety makes a direct
hydrogen bond with Ser50 (Fig. 3). Furthermore, the
steric bulk of the benzylic group forces Asp54 to change
its conformation, bringing it closer to the 40N atom of 6,
thus inducing the formation of a salt bridge. Gratifying-
ly, cellular potency had also improved.
We have previously shown that the para-position of
this phenyl ring may be substituted with a large fluo-
rescent group to provide a useful assay probe.15 There-
fore, we surmised that a solubilising group could be
placed at the 4-position of the pyrazole ring with min-
imal impact on potency. Considering requirements for
potency, novelty and synthetic access, we determined
through modelling studies that cyclic amines were
desirable targets.
Increasing the spacer length between the piperazine and
phenyl rings results in loss of potency (8), as does
exchanging the phenyl for an amide (9). Alkylation of
the pyrazole nitrogen (either isomer) abolished activity
due to clash with the residues forming the binding site
(Gly97, Thr184 and an interstitial water molecule).17,18
Indeed when the novel piperazine 2 was prepared, we
were delighted to find that ATPase enzyme activity
was maintained with only a small loss in growth inhibi-
tion against HCT116 colon tumour cells. Binding activ-
ity in the FP assay15 was reduced by nearly 10-fold but
this did not affect the functional (ATPase) activity.
In the quest for further potency, we focused on the 50-
position of the resorcinol ring. Ethyl and chloro groups
have been reported in the 3,4-diaryl pyrazole series to
enhance potency over hydrogen.12,13 Development of
this position by extension into a lipophilic pocket was
achieved from bromo 10, which itself had a similar
profile to 2. This lipophilic pocket does not exist in the
native conformation of Hsp90, but we found that it
appeared as a result of a conformational change induced
by the binding of PU3.17 Phenethyl 11 induced the same
conformational change and bound slightly better to the
enzyme but had similar ATPase inhibition.
From the co-crystal structure of 2 (Fig. 2), key interac-
tions between the resorcinol and the Asp93-water net-
work at the base of the pocket are identical to those
observed for 1.12–14 Pleasingly, the new piperazine group
40N overlays closely with the 40C of the aryl of 1 with no
effect on the conformation of the resorcinol–pyrazole
system.
Figure 2. Binding mode of 2 to Hsp90.21 The dotted lines indicate
polar interactions between the ligand, the protein and interstitial water
molecules.
Figure 3. Binding mode of 6 to Hsp90.21 The conformation of Asp54
in the Hsp90–2 complex is provided for comparison (orange residue).

X. Barril et al. / Bioorg. Med. Chem. Lett. 16 (2006) 2543–2548
2545
Deletion of the ethyl linker to give phenyl 12 had no
effect on ATPase or cell activity. These results led to
combinations of R1 and R2 (where X = N, see Table
1). Alkylating the bromo compound 10 with the
sulfonylbenzyl group from 6 gave 13, which was 10-fold
more active than 10 as expected. However, when 11 was
similarly treated, the resulting 14 was less active than
either 11 or 6. Clearly these groups are influencing each
other. We believe that these aromatic rings stack against
each other, forming a conformation incompatible with
binding to Hsp90. The loss of activity could then be
explained on the basis of an internal energy penalty.
Attention then turned to the 5-position of the pyrazole.
As previously reported, an ethyl amide improves poten-
cy by at least an order of magnitude in the 3,4-diaryl
series.13 In the piperazine series, however, such SAR is
not seen. Methyl 15 was similar to 2 and hydroxyl meth-
yl 16 greater than 10-fold less potent. Unfortunately,
ethyl amide 17 was no better or worse than 2. Overlay
of the co-crystal structure of 17 with the structure of 2
bound to Hsp90 shows the binding modes to be virtually
identical (Fig. 4). The new amide in 17 even makes the
desired hydrogen bond with Gly97.
Figure 4. Overlay of X-ray structures of 17 (grey) and 2 (green) bound
to Hsp90.21
As can be seen in Table 2, compounds such as 17 have
to pay a significant internal energy penalty to be able
to bind to Hsp90, either in their neutral (2.4 kcal/mol)
or protonated form (0.4 kcal/mol). The preference for
the syn conformation is due to the formation of an inter-
This disappointing result is explained by the fact that,
unlike the analogous 4-phenyl compounds, the confor-
mation observed in the co-crystal structure does not
correspond to a global minimum in aqueous solution.
Table 1. Hsp90 inhibitors and their activity16
R2
R1
O
O
X
S
HO
A =
N
R3
N
OH
N
H
R1
XR2
R3
FP IC50 (lM)
ATPase IC50 (lM)
HCT116 GI50 (lM)
a
b
Compound
17-AAG
1.54 (0.5)
0.148 (0.016)
2.0 (0.76)
21.7 (3.3)
8.8 (4.2)
17.7 (7.7)
6.3 (2.9)
8.2 (2.6)
68.7 (3.0)
n/d
0.17 (0.07)
5.8 (1.2)
23.4 (4.8)
>80
1
2
Cl
Cl
NH
O
H
H
H
H
H
H
H
H
H
H
H
H
H
Me
3
4
Cl
Cl
NMe
NEt
N-A
>80
5
4.3 (1.8)
0.74 (0.2)
0.6 (0.2)
11.7 (5.0)
1.3 (0.4)
2.5 (2.3)
15.5 (2.1)
101 (92)
n/d
13.8 (4.6)
3.1 (0.63)
6.5 (1.7)
14.1 (1.6)
>80
6
Cl
Cl
7
NBn
8
Cl
Cl
N(CH2)2Ph
NCH2CONH2
NH
4.4 (0.42)
32.6 (14.4)
9.7 (0.0)
9
10
11
12
13
14
15
16
17
18
19
20
Br
PhCH2CH2
69.9 (5.2)
31.6 (4.5)
>80
NH
NH
2.4 (1.6)
8.5 (3.4)
9.6 (0.3)
9.6 (2.5)
3.0 (1.7)
31.8 (7.6)
28.3 (2.8)
121.7 (20)
6.3 (1.2)
96.9 (19.6)
40.3 (11.6)
26.1 (5.3)
Ph
Br
N-A
N-A
0.6 (0.2)
5.2 (2.7)
8.1 (1.9)
7.9 (0.9)
29.9 (0.3)
>80
PhCH2CH2
Cl
Cl
Cl
Br
Br
Br
NH
NH
3.0 (0.25)
25.1 (4.1)
9.5 (0.1)
CH2OH
CONHEt
NH
34.4 (3.5)
>80
CHOH
CHNH-A
CHNHEt
H
H
H
21.2 (8.9)
13.9 (2.6)
5.6 (0.1)
73.2 (6.0)
54.4 (5.2)
FP, fluorescence polarization; all results shown with their standard deviation from at least 2 determinations. n/d, not determined.
a IC50’s are obtained from 22 different inhibitor concentrations, Z0 typically >0.9.
b IC50 determined from 10 different inhibitor concentrations at 2-fold dilutions, Z0 typically >0.75.

2546
X. Barril et al. / Bioorg. Med. Chem. Lett. 16 (2006) 2543–2548
Table 2. Relative QM energies (in kcal/mol) of 5-amide pyrazoles in aqueous solution (italics) and in vacuo (parentheses)
Compound
X
X
X
X
H
H
Res
Res
Res
H
Res
H
O
N
N
O
N
O
O
N
N
N
N
N
N
N
N
N
H
H
H
H
0.0 (0.0)
2.4 (2.6)
5.5 (8.9)
6.4 (8.6)
2.1 (4.4)
0.1 (5.1)
2.7 (1.4)
0.0 (0.0)
X =
H
N
X =
N
H
H
+
N
0.4 (0.0)
4.7 (11.4)
0.1 (0.0)
0.0 (15.1)
X =
N
The first two columns correspond to the anti conformation observed in the X-ray structures. The two rightmost columns correspond to the syn
conformation, which is incompatible with formation of a hydrogen bond with Gly97. Res, resorcinol.
nal hydrogen bond between NH of the 5-amide and N1
of the piperazine. The fact that, for this type of mole-
cule, both tautomers are isoenergetic may also have an
impact on the binding affinity.
ethanol in one pot, giving the skeleton of the target mol-
ecule. Total deprotection may be accomplished with
boron trichloride in DCM to give 2 in 52% yield. Substi-
tution of the piperazine nitrogen to give 4–9 is accom-
plished via either reductive amination or alkylation.
Piperidine analogues 18–20 were all less potent than the
corresponding piperazines.
Synthesis of a variety of lipophilic substituents appended
to the 50-position of the resorcinol started from 2,4-dihy-
droxy acetophenone (Scheme 2). Di-O-benzyl protection
was followed by selective bromination of the acyl group
with phenyltrimethylammonium tribromide. Further
bromination with N-bromosuccinimide yielded the de-
sired dibromo intermediate which could be elaborated
as described in Scheme 1 to furnish 10. Alkylation of 10
with methylsulfonylbenzyl chloride gave 14. Pd-catalysed
Suzuki coupling to 10 with aryl boronates yielded 12
and Heck coupling with styrene followed by hydrogena-
tion gave 11. Alkyation of these coupled products
proceeded smoothly to give the more complex derivative
14 (Scheme 3).
Piperazine 2 was prepared from commercially available
chloro resorcinol. Friedel–Crafts acylation and O-ben-
zyl protection of the hydroxyls, followed by bromina-
tion of the acyl group, gave key bromomethyl ketone
intermediate in moderate overall yield (Scheme 1). Dis-
placement of the bromide was carried out with a range
of amine nucleophiles, such as morpholine, which ulti-
mately gives 3, or protected piperazine (Step d). Homol-
ogation with DMF dimethylacetal gives an enamine
intermediate which can be cyclised with hydrazine in
Cl
Cl
Cl
HO
HO
a
BnO
b
c
Exploring the pyrazole 5-position was achieved with a
directed lithiation procedure, quenching with a suitable
electrophile such as ethyl isocyanate to give ethyl amide
17. Primary alcohol 16 was prepared by reduction of an
intermediary tert-butyl ester.
90%
50%
58%
OH
Cl
O
OH
OBn O
Cl
BnO
e
BnO
d
N
30%
81%
Br
Piperidines 18–20 were prepared as for Scheme 1, substi-
tuting 4-piperidone for N-Boc piperazine in step d. The
resulting ketone was reduced to give alcohol 18 and
reductively aminated with methylsulfonyl benzylamine
and ethylamine to give 19 and 20, respectively.
OBn O
NBoc
OBn O
NBoc
R
N
H
N
Cl
Cl
Cl
BnO
HO
g or h
f
HO
N
N
N
52%
OBn
N
N
H
N
OH
N
H
N
OH
N-terminal human Hsp90a his-tagged protein was co-
crystallised in complex with compounds 2, 6 and 17 as
previously described.17 Data were collected on all three
co-crystals and the structures subsequently solved by
molecular replacement using the previously solved na-
tive Hsp90a structure17 (PDB code: 1UY1). All three
co-crystals diffracted in space group I222 to resolutions
N
H
2
4-9
Scheme 1. Synthesis of 3-(50-chloro)aryl-4-piperazinyl pyrazoles.
Reagents and conditions: (a) AcOH, BF3ÆOEt2, 90 °C, 3.5 h; (b) BnBr,
K2CO3, MeCN, reflux, 6 h then rt o/n; (c) PhMe3N+Br3ꢀ, THF, rt, 2 h;
(d) Boc-piperazine, Cs2CO3, DMF, rt, 2 h; (e) i—DMF DMA, reflux,
7 h; ii—hydrazine, EtOH, microwave, 120 °C, 5 min; (f) BCl3, CH2Cl2,
0 °C to rt, 1 h; (g) RCHO, NaBH(OAc)3, AcOH, CH2Cl2, rt, 3 h; (h)
RBr, Cs2CO3, DMF, rt, 3 d.
˚
of 1.8, 2.5 and 2.3 A for 2, 6 and 17, respectively. The
previously reported flexible loop region of Hsp90

X. Barril et al. / Bioorg. Med. Chem. Lett. 16 (2006) 2543–2548
2547
Br
HO
BnO
Scheme 1
BnO
a
b, c
77%
78%
Br
OH
O
OBn
R
O
OBn O
HO
A
H
H
N
N
N
Br
R
HO
HO
e
g
N
N
N
N
OH
10
13
N
OH
12 R=Ph
N
N
H
N
OH
N
H
H
14
d
R=PhCH=CH2
f
11
Scheme 2. Synthesis of 50-analogues of 3-aryl,4-piperazinyl pyrazoles. Reagents and conditions: (a) BnBr, K2CO3, MeCN, rt, o/n; (b) PhMe3N+Br3
,
ꢀ
THF, rt, 2 h; (c) NBS, DMF, rt, o/n; (d) A-Cl, Cs2CO3, DMF, rt, o/n; (e) ArB(OH)2 or ArCH@CH2, PdCl2[((o-tolyl)3P)2 (2 mol%), iPr2NEt, nBuOH,
reflux, 15 h; (f) H2, 10% Pd/C, EtOAc, rt, 1 h; (g) A-Cl, Et3N, DMF, rt, o/n.
NBoc
NBoc
NBoc
Cl
Cl
Cl
BnO
BnO
BnO
a
b
N
N
N
77%
R
OBn
N
N
H
OBn
N
OBn N
NTs
NTs
H
N
Cl
HO
15 R = Me
16 R = CH2OH
c, d
N
17 R = CONHEt
R
OH
N
N
H
Scheme 3. Synthesis of 5-analogues of 3-aryl,4-piperazinyl pyrazoles. Reagents and conditions: (a) TsCl, pyr, CH2Cl2, rt, o/n; (b) i—n-BuLi, THF,
ꢀ78 °C, 10 min; ii—MeI (R = Me) or tBuOCOCl (R = CO2 Bu) or EtNCO (R = CONHEt), warm to rt; (c) BCl3, CH2Cl2, 0 °C to rt; (d)
t
(R = CH2OH only) LAH, Et2O, rt, 2 h.
(residues 108–114) is in a ‘closed’ conformation5 in all of
these structures. The binding mode of the resorcinol and
pyrazole rings is as previously described.14 All quantum-
mechanical calculations were carried out using GA-
MESS.19 All molecules in Table 2 were fully minimized,
in vacuo and using the PCM model20 to simulate aque-
ous solution. In both cases, the minimization started
either from the crystallographic structure (anti) or a
rotation of 180° of the amide (syn). The level of calcula-
tion was HF/6-31G(d).
These results are in agreement with those seen for the
published Hsp90 inhibitors 17-AAG and VER-49009.11
In summary, we have discovered novel and soluble 4-
amino analogues of the 3-aryl pyrazole series of Hsp90
inhibitors. Crystal structures of the new compounds
bound to the enzyme help to explain the observed
SAR and suggest further compounds which may be of
interest for future studies.
To support the fact that the cellular growth inhibition of
the more active compounds 2, 6, 7 and 13 was through
Hsp90 inhibition, the ability of these compounds to
modulate the Hsp90 specific cellular markers Raf-1
and Hsp70 was determined. All four compounds down-
regulate Raf-1 at doses between 1 and 2 times the cellu-
lar GI50 level and upregulate Hsp70 (as measured by an
Hsp70 specific ELISA) at 1-times the GI50 (Fig. 5).
References and notes
1. Jolly, C.; Morimoto, R. I. J. Natl. Cancer Inst. 2000, 92,
1564.
2. Maloney, A.; Workman, P. Expert Opin. Biol. Ther. 2002,
2, 3.
3. Workman, P. Curr. Cancer Drug Targets 2003, 3, 297.
4. Isaacs, J. S.; Xu, W.; Neckers, L. Cancer Cell 2003, 3, 213.
5. Stebbins, C. E.; Russo, A. A.; Schneider, C.; Rosen, N.;
Hartl, F. U.; Pavletich, N. P. Cell (Cambridge, MA) 1997,
89, 239.
6. Jez, J. M.; Chen, J. H.; Rastelli, G.; Stroud, R. M.; Santi,
D. V. Chem. Biol. 2003, 10, 361.
2
6
1
7
1
13
1
1
2
0.5
2
7. Banjeri, U.; Judson, I.; Workman, P. Curr. Cancer Drug
Targets 2003, 3, 385.
8. Erlichman, C.; Toft, D.; Reid, J.; Sloan, J.; Atherton, P.;
Adjei, A.; Ames, M.; Croghan, G. Proced. Am. Assoc.
Cancer Res. 2001, 42, 833.
9. Munster, P. N.; Tong, L.; Schwartz, L.; Larson, S.;
Kenneson, K.; De La Cruz, A.; Rosen, N.; Scher, H.
Proced. Am. Soc. Clin. Oncol. 2001, 20, 83a.
Raf-1
Hsp70
Induction
1.0
12
15
20
8
Figure 5. Western blot showing the depletion of Raf-1 levels and
induction of Hsp70 (as measured by a Hsp70 specific ELISA) in
HCT116 cells following exposure to either 0.5, 1 or 2 times GI50 of 2, 6,
7 or 13 for 48 h.

2548
X. Barril et al. / Bioorg. Med. Chem. Lett. 16 (2006) 2543–2548
10. Yanin, Y. L. J. Med. Chem. 2005, 48, 7504.
11. Tomura, A.; Odanaka, J.; Takashio, K.; Kuramochi, H.
Preparation of pyrazoles as HSp90 inhibitors and their use
as antitumor agents. Jpn Kokai Tokkyo Koho JP
2005225787, 2005.
12. Cheung, K.-M.; Matthews, T. P.; James, K.; Rowlands,
M. G.; Boxall, K. J.; Sharp, S. Y.; Maloney, A.; Roe, S.
M.; Prodromou, C.; Pearl, L. H.; Aherne, G. W.;
McDonald, E.; Workman, P. Bioorg. Med. Chem. Lett.
2005, 15, 3338.
13. Dymock, B. W.; Barril, X.; Brough, P. A.; Cansfield, J. E.;
Massey, A.; McDonald, E.; Hubbard, R. E.; Surgenor, A.;
Roughley, S. D.; Webb, P.; Workman, P.; Wright, L.;
Drysdale, M. J. J. Med. Chem. 2005, 48, 4212.
14. Kreusch, A.; Han, S.; Brinker, A.; Zhou, V.; Choi, H.-a.;
Yun, H.; Lesley, S. A.; Caldwell, J.; Gu, X. J. Bioorg.
Med. Chem. Lett. 2005, 15, 1475.
15. Howes, R.; Barril, X.; Dymock, B. W.; Grant, K.;
Northfield, C. J.; Robertson, A. G.; Surgenor, A.; Wayne,
J.; Wright, L.; James, K.; Matthews, T.; Cheung, J.;
McDonald, E.; Workman, P.; Drysdale, M. J. Anal
Biochem 2006, doi:10.1016/j.ab.2005.12.023.
16. The values obtained for the ATPase and the FP assay
differ between compounds as they measure different
aspects of compound activities. The ATPase assay mea-
sures a compound’s ability to inhibit ATPase functionality
of yeast Hsp90. In comparison, the FP assay measures the
ability of a compound to compete with a small molecule
for binding to human Hsp90. Therefore, some compounds
may compete for binding to Hsp90 with this small
molecule poorly due to their kinetic properties (e.g., fast-
on, fast-off) but may still retain the ability to inhibit the
ATPase activity.
17. Wright, L.; Barril, X.; Dymock, B.; Sheridan, L.; Surg-
enor, A.; Beswick, M.; Drysdale, M.; Collier, A.; Massey,
A.; Davies, N.; Fink, A.; Fromont, C.; Aherne, W.;
Boxall, K.; Sharp, S.; Workman, P.; Hubbard, R. E.
Chem. Biol. 2004, 11, 775.
18. Data not shown. Compounds were prepared by alkylation
with Cs2CO3 and alkyl halide. Regioisomers were sepa-
rated by preparative HPLC-MS and regiochemistry was
determined by NOE spectroscopy.
19. Schmidt, M. W.; Baldridge, K. K.; Boatz, J. A.; Elbert, S.
T.; Gordon, M. S.; Jensen, J. J.; Koseki, S.; Matsunaga,
N.; Nguyen, K. A.; Su, S.; Windus, T. L.; Dupuis, M.;
Montgomery, J. A. J. Comput. Chem. 1999, 14, 1347.
20. Tomasi, J.; Persico, M. Chem. Rev. 1994, 94, 2027.
21. Atomic coordinates have been deposited with the Protein
Data Bank at Rutgers University (http://rscb.rutgers.edu/
pdb/index.html). PDB id codes for 2, 6 and 17 co-
crystallised with Hsp90 are 2CCS, 2CCU and 2CCT,
respectively.