Synthesis of 19-substituted geldanamycins with altered conformations and their binding to heat shock protein Hsp90

Article abstract of DOI:10.1038/nchem.1596

The benzoquinone ansamycin geldanamycin and its derivatives are inhibitors of heat shock protein Hsp90, an emerging target for novel therapeutic agents both in cancer and in neurodegeneration. However, the toxicity of these compounds to normal cells has been ascribed to reaction with thiol nucleophiles at the quinone 19-position. We reasoned that blocking this position would ameliorate toxicity, and that it might also enforce a favourable conformational switch of the trans-amide group into the cis-form required for protein binding. Here, we report an efficient synthesis of such 19-substituted compounds and realization of our hypotheses. Protein crystallography established that the new compounds bind to Hsp90 with, as expected, a cis-amide conformation. Studies on Hsp90 inhibition in cells demonstrated the molecular signature of Hsp90 inhibitors: decreases in client proteins with compensatory increases in other heat shock proteins in both human breast cancer and dopaminergic neural cells, demonstrating their potential for use in the therapy of cancer or neurodegenerative diseases.

Full text of DOI:10.1038/nchem.1596

ARTICLES
PUBLISHED ONLINE: 10 MARCH 2013 | DOI: 10.1038/NCHEM.1596
Synthesis of 19-substituted geldanamycins with
altered conformations and their binding to heat
shock protein Hsp90
Russell R. A. Kitson1, Chuan-Hsin Chang2, Rui Xiong2, Huw E. L. Williams1, Adrienne L. Davis1,
William Lewis1, Donna L. Dehn2, David Siegel2, S. Mark Roe3, Chrisostomos Prodromou3 ,
*
David Ross2 and Christopher J. Moody1
*
*
The benzoquinone ansamycin geldanamycin and its derivatives are inhibitors of heat shock protein Hsp90, an emerging
target for novel therapeutic agents both in cancer and in neurodegeneration. However, the toxicity of these compounds to
normal cells has been ascribed to reaction with thiol nucleophiles at the quinone 19-position. We reasoned that blocking
this position would ameliorate toxicity, and that it might also enforce a favourable conformational switch of the trans-
amide group into the cis-form required for protein binding. Here, we report an efficient synthesis of such 19-substituted
compounds and realization of our hypotheses. Protein crystallography established that the new compounds bind to Hsp90
with, as expected, a cis-amide conformation. Studies on Hsp90 inhibition in cells demonstrated the molecular signature of
Hsp90 inhibitors: decreases in client proteins with compensatory increases in other heat shock proteins in both human
breast cancer and dopaminergic neural cells, demonstrating their potential for use in the therapy of cancer or
neurodegenerative diseases.
eat shock protein 90 (Hsp90) is one of the most abundant 19-substituted BQAs are effective inhibitors of Hsp90, suggesting
proteins in eukaryotic cells and, driven by the hydrolysis of that such rationally modified compounds have potential for appli-
H
adenosine tri-phosphate (ATP), is a molecular chaperone cation in the therapy of cancer and neurodegenerative diseases.
responsible for the folding of nascent proteins. It has been described Geldanamycin 1, a naturally occurring BQA polyketide, is a
as the master regulator of the stabilization, activation and degra- potent inhibitor of Hsp90, and its binding to the N-terminal
dation of a range of over-expressed or mutant oncogenic proteins1. domain has been studied by X-ray crystallography19,20 and NMR
Its inhibition can cause client proteins to adopt misfolded confor- spectroscopy21. In preclinical studies, geldanamycin was found to
mations that are ubiquinated and subject to proteasomal degra- exhibit significant hepatotoxicity15, so other less toxic and more
dation, thereby simultaneously disrupting multiple cancer-causing soluble benzoquinone ansamycins such as 17-allylamino-17-
pathways. As a consequence, it has emerged as a very attractive demethoxygeldanamycin (17-AAG, tanespimycin) 2 and 17-(2-
target for novel molecular cancer therapeutic agents2–6, and there dimethylaminoethylamino)-17-demethoxygeldanamycin (17-DMAG,
are now more than 15 Hsp90 inhibitors in clinical trials, with alvespimycin) 3 were developed. Although 17-AAG (tanespimycin)
accumulating evidence of anticancer activity3,6,7. In addition, the 2 showed some promise in Phase II trials, demonstrating activity
modulation of protein folding is also relevant to neurodegenerative in HER2/ERBB2-positive trastuzumab-refractory breast cancer22,
diseases, and a number of conditions such as Alzheimer’s and development of the compound has been halted, further emphasizing
Parkinson’s diseases are thought to be associated with protein mis- the need for less toxic derivatives.
folding and aggregation8–14
.
In neuroscience, protein misfolding can be regulated by chaper-
One compound that has played a very major role in the Hsp90 ones such as Hsp70, which paradoxically can be induced by inhi-
arena is the benzoquinone ansamycin (BQA) geldanamycin. bition of Hsp90. The chaperones Hsp90 and 70 exhibit very
However, its use is limited because of the hepatotoxicity observed different specificity with respect to client protein interactions, pro-
in preclinical studies15, possibly a result of reaction with biological viding a mechanism whereby inhibiting one chaperone (Hsp90)
nucleophiles at the reactive 19-position of the quinone ring16–18
.
while inducing another (Hsp70) can be beneficial in specific dis-
We reasoned that blocking the reactive 19-position might suppress eases. Hence, Hsp90 has now emerged as a target to prevent neuro-
this reaction with biological nucleophiles, ameliorate toxicity, and degeneration8–14, and Hsp90 inhibitors have been used as protective
also result in a conformational change arising from amide trans to therapies in mouse models of Parkinson’s disease23. The BQAs
cis isomerism. We now report the realization of this hypothesis. geldanamycin and 17-AAG have been found to protect against
Introducing a substituent at C19 does indeed cause the desired confor- 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine
mational switch, while also blocking reaction with thiol nucleophiles dopaminergic neurotoxicity in mice23, and other models of neuro-
and, importantly, markedly reducing toxicity to normal endothelial degenerative disease9,24–26
and epithelial cells. Additionally, studies in both human breast However, if such BQAs are to be considered as potential thera-
(MPTP)-induced
.
cancer cells and dopaminergic neural cells established that the novel peutic agents for neurodegenerative disease, there is an urgent
1School of Chemistry, University of Nottingham, University Park, Nottingham NG7 2RD, UK, 2Department of Pharmaceutical Sciences, School of Pharmacy,
University of Colorado Denver, 12700 East 19th Avenue, Aurora, Colorado 80045, USA, 3Genome Damage and Stability Centre, Science Park Road,
University of Sussex, Falmer, Brighton BN1 9RQ, UK. e-mail: chris.prodromou@sussex.ac.uk; c.j.moody@nottingham.ac.uk; david.ross@ucdenver.edu
*
NATURE CHEMISTRY | VOL 5 | APRIL 2013 | www.nature.com/naturechemistry
© 2013 Macmillan Publishers Limited. All rights reserved.
307

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1596
ARTICLES
O
O
O
geldanamycin derivatives have been prepared previously by
modification of the natural product, either by exploiting the
nucleophilic character of the amino-quinone moiety (for example,
halogenation36 and Mannich37,38 reactions) or the ability of the
19
R
X
17
X
R
O
H
Me
N
N
H
O
3
O
quinone to act as
a conjugate acceptor by reacting with
H
Me
MeO
Me
OMe
nucleophiles such as thiols39, amines26,37 and alcohols26,40. Given the
limited applicability of the aforementioned methods, particularly
for the formation of a C–C bond at the 19-position, we were
attracted to the possibility of utilizing a palladium-catalysed cross-
coupling. Such protocols, including the Heck, Suzuki, Stille, Negishi
and Sonagashira reactions, constitute a versatile and powerful
strategy in modern chemistry, and have the advantage that a wide
array of coupling partners are readily available, allowing access to a
diverse set of 19-substituted analogues.
Me
OH
MeO
HO
Me
Me
Me
Me
OCONH2
OMe
OCONH2
trans-amide
1 R = H; X = OMe
2 R = H; X = NHCH2CH=CH2
3 R = H; X = NHCH2CH2NMe2
4 R = H; X = azetidinyl
cis-amide
(protein bound
conformation)
To assess the viability of the cross-coupling approach, we com-
menced with the stable and readily accessible 19-iodogeldanamycin
5, prepared in excellent yield via the remarkably selective reaction of
commercially available geldanamycin 1 with iodine (Fig. 2a)36.
Unfortunately, difficulties were immediately encountered using
standard conditions for cross-couplings with a range of partners
(boronic acids or boronate esters, stannanes, Grignards, alkynes
Figure 1 | Trans–cis amide isomerization in geldanamycin BQAs. Does the
steric strain caused by introduction of a substituent R at the 19-position
enforce a favourable conformational switch of the trans-amide group into the
cis-form required for protein binding, while also ameliorating toxicity by
blocking attack of biological nucleophiles?
need to reduce their toxicity while retaining or improving their and alkenes) and different metal catalysts (predominantly
potency. The toxicity has been proposed to arise from the reaction palladium and iron), with the sensitivity of the different functional-
of biological nucleophiles, particularly glutathione, at the unsubsti- ities within the BQA substrate proving incompatible with many
tuted 19-position of the quinone ansamycin16–18. We reasoned that conditions (high temperature and strong base). In addition,
the conjugation of glutathione (or other nucleophiles) might be sup- couplings under milder conditions (those at lower temperature or
pressed by blocking the 19-position of the BQAs, and, as a result, the with mild or no base) also proved to be problematic, with only
toxicity issues might be ameliorated. In addition, we also hypoth- the formation of geldanamycin itself observed, presumably due to
esized that the introduction of a substituent at C19 would result competing reductive catalytic processes. We hypothesized that
in conformational changes in the macrocyclic ring, arising from these findings may be due to the transmetallation step in the cataly-
the 19-substituent causing amide trans to cis isomerism as a result tic cycle being slower than that for a competing pathway. Thus, we
of steric strain. We now report the first studies on the design and subjected our substrate to modified conditions that have been
synthesis of a range of such 19-substituted BQAs, a study of their reported to address such problems, focusing on the Stille reaction,
conformation in solution by NMR spectroscopy, their binding to as this is generally considered the mildest of palladium-catalysed
yeast Hsp90 by protein X-ray crystallography, their reaction with cross-coupling processes.
thiol nucleophiles, and studies on toxicity and Hsp90 inhibition
in cellular systems.
The replacement of phosphine ligands with triphenylarsine has
been well documented to assist slow transmetallations41, the softer
and more labile arsine ligand facilitating the formation of the
palladium–tin double-bond p-complex, which has been reported
to be the rate-determining step in Stille cross-couplings42.
We were delighted to observe that performing the Stille coupling
with tetramethyltin in the presence of triphenylarsine gave
19-methylgeldanamycin 6, albeit in moderate yield and with
significant quantities of recovered geldanamycin (29%). Following
this promising result, a range of optimization reactions were
carried out (Supplementary Table S2), resulting in the establish-
ment of optimal conditions of 1.2 equiv. stannane, 20 mol%
Results and discussion
Compound design and rationale. The BQA macrocycles are
known to adopt an extended trans-amide conformation in the
solid state, as shown by X-ray crystal structures of geldanamycin 1
itself27 and the 17-azetidinyl compound 4 (ref. 28). Early NMR
spectroscopic studies on the solution conformation of BQA 4 also
suggested a trans-amide conformation on the basis of the strong
nuclear Overhauser effect (nOe) enhancement between the NH
and the alkene H-3 (Fig. 1)28. In contrast, protein crystallography
studies using either yeast or human Hsp90 have shown that, on
binding, geldanamycin 1 and 17-DMAG 3 adopt a more closed
‘C-clamp’ conformation with a cis-amide bond19,20,29, so the
isomerization equilibrium (Fig. 1) between trans and cis forms is
a matter of some debate. Although some computational studies
have put the barrier to trans–cis isomerization as .80 kJ mol21
(ref. 30), other calculations suggest that it is much lower than
this31. A requirement for trans–cis isomerization of the BQA for
binding and inhibition of Hsp90 has been suggested29,30, but a
separate study has disputed this conclusion32,33. We therefore set
out to synthesize a wide range of stable geldanamycin analogues,
containing diverse substituents at the 19-position, to investigate
both the toxicological implications and also whether any
conformational ‘switch’ was observed.
.
arsine, 5 mol% Pd2(dba)3 CHCl3 and 5 mol% CuI (ref. 43) in
N,N-dimethylformamide (DMF) at 35 8C. This gave the desired
19-substituted BQA 6 in an excellent yield of 86%.
Having established a successful coupling protocol, we investi-
gated the scope of the reaction with a range of stannane coupling
partners, selected to introduce representative alkyl, alkenyl, aryl
and hetaryl groups (Fig. 2a). Although a methyl group could be suc-
cessfully incorporated at C19, attempted coupling of other alkyl
groups was unsuccessful. Similarly, the reaction of 19-iodogeldamy-
cin 5 with allyl tri-n-butylstannane only gave a very poor yield
(ꢀ5%) of the 19-allyl compound. Both electron-rich and electron-
deficient aromatic groups could also be coupled successfully in
good to excellent yield. Heteroaromatic stannanes proved to be
more variable under our conditions. Coupling of the 2-pyridyl
group was problematic, with the product 12 isolated in a moderate
Synthesis. Geldanamycin and related BQAs such as the herbimycins yield of 30%. However, furan and thiophene groups were success-
and macbecins have been the subject of a number of studies, ranging fully transferred, affording substrates 13 and 14 in excellent yields
from total synthesis of the natural products themselves to their of 90% and 94%, respectively. The Stille products, following an
synthetic modification34,35. A small number of 19-substituted aqueous work-up and purification (K2CO3/SiO2 chromatography)44,
308
NATURE CHEMISTRY | VOL 5 | APRIL 2013 | www.nature.com/naturechemistry
© 2013 Macmillan Publishers Limited. All rights reserved.

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1596
ARTICLES
a
O
O
O
O
O
19
I
19
MeO
R
19
MeO
MeO
O
O
O
Stannane (1.2 equiv.)
Pd2(dba)3•CHCl3 (5 mol%)
Ph3As (20 mol%)
Me
Me
Me
N
H
N
N
H
I2, pyridine
98%
H
O
Me
MeO
Me
Me
MeO
CuI (5 mol%)
DMF, 35 °C
OH
OH
MeO
OH
MeO
MeO
Me
MeO
Me
Me
Me
Me
Me
OCONH2
OCONH2
OCONH2
1
5
6
7
8
9
R = Me (86%)
R = vinyl (76%)
R = Ph (85%)
11 R = 4-morpholino-
C6H4 (67%)
12 R = 2-pyridyl (30%)
R = 4-MeOC6H4 (56%) 13 R = 2-furyl (90%)
10 R = 4-FC6H4 (80%)
14 R = 2-thienyl (94%)
O
b
c
19
R
NHAc
MeO
O
17
CO2Me
O
O
O
O
Me
N
H
S
X
X
17
19
O
O
O
17
Me
MeO
Me
Me
OH
Me
MeO
N
N
H
Ac-Cys-OMe
DBU, THF, rt
H
Me
MeO
Me
MeO
Me
OH
Me
OH
Me
MeO
MeO
OCONH2
Me2N
6–14
Me
Me
OCONH2
OCONH2
NH2
NH2
(5 equiv.)
THF, 60 °C, 2 h
(5 equiv.)
THF, reflux
16 h
1 X = OMe
2 X = NHCH2CH=CH2
3 X = NHCH2CH2NMe2
29 X = OMe (35%)
30 X = NHCH2CH=CH2 (62%)
31 X = NHCH2CH2NMe2 (21%)
O
O
O
H
N
17
H
19
R
19
R
N
Me2N
R
X
O
O
17
O
17
Me
Me
Me
N
N
H
H
O
N
H
O
Ac-Cys-OMe
DBU, THF, rt
Me
MeO
Me
MeO
O
no reaction
Me
MeO
OH
Me
OH
Me
MeO
MeO
OH
Me
MeO
Me
Me
OCONH2
OCONH2
Me
OCONH2
15 R = Me (100%)
16 R = Ph (100%)
22 R = Me (100%)
23 R = Ph (100%)
6 R = Me; X = OMe
17 R = 4-MeOC6H4 (100%)
18 R = 4-FC6H4 (77%)
19 R = 4-morpholino-
C6H4 (100%)
20 R = 2-furyl (96%)*
21 R = 2-thienyl (48%)*
* reaction at room temp
15 R = Me; X = NHCH2CH=CH2
22 R = Me; X = NHCH2CH2NMe2
8 R = Ph; X = OMe
16 R = Ph; X = NHCH2CH=CH2
23 R = Ph; X = NHCH2CH2NMe2
24 R = 4-MeOC6H4 (100%)
25 R = 4-FC6H4 (75%)
26 R = 4-morpholino-
C6H4 (100%)
27 R = 2-furyl (92%)*
28 R = 2-thienyl (92%)*
* reaction at room temp
Figure 2 | Synthesis and reactivity of 19-substituted geldanamycin derivatives. a, Synthesis of 19-substituted geldanamycins by selective iodination and
optimized palladium-catalysed Stille coupling. b, Synthesis of 17-allylamino- and 17-(2-dimethylaminoethylamino)-19-substituted geldanamycins (15–21 and
22–28, respectively) by displacement of the 17-methoxy group with amines. c, Addition of N-acetylcysteine methyl ester to BQAs 1–3, with formation of the
corresponding 19-substituted adducts 29–31. The 19-substituent in the methyl (6, 15, 22) and phenyl (8, 16, 23) compounds prevents similar nucleophilic
attack of the N-acetylcysteine methyl ester. dba, dibenzylideneacetone; DBU, 1,8-diazabicyclo[5.4.0]undec-7-ene.
contained 10.5 ppm palladium, 7.9 ppm tin and arsenic, and derivatives in modest to excellent yield. In the main, the substi-
undetectable levels of copper, as detected by inductively coupled tutions proceeded smoothly, although the 19-vinyl derivative 7
plasma mass spectrometry (ICPMS) trace element analyses gave no product, presumably due to competing conjugate
(for details, see Supplementary Section, ‘ICPMS trace element addition reactions.
analysis’, page S52).
In the geldanamycin series of BQAs, it is the 17-allylamino (17- Conformational studies. Interestingly, the 19-substituted
AAG) and 17-(2-dimethylaminoethylamino) (17-DMAG) deriva- derivatives 6–14 were significantly more polar than geldanamycin
tives 2 and 3 that have shown the most clinical promise, so we itself (silica gel TLC Rf ¼ 0.29 (where Rf is retardation factor)
synthesized the corresponding AAG and DMAG analogues of our (19-methylgeldanamycin) versus 0.56 (geldanamycin) in 2:1 ethyl
19-substituted geldanamycin derivatives (Fig. 2b). This was readily acetate/light petroleum). To investigate whether this was as a result
achieved by heating the 17-methoxy compounds 6–14 with a of the proposed conformational ‘switch’, we undertook a series of
fivefold excess of allylamine or N,N-dimethylethylenediamine in detailed NMR spectroscopic experiments. Conformational NMR studies
tetrahydrofuran (THF), and gave the corresponding 17-amino of geldanamycin 1 and 17-azetidinyl-17-demethoxygeldanamycin 4
NATURE CHEMISTRY | VOL 5 | APRIL 2013 | www.nature.com/naturechemistry
© 2013 Macmillan Publishers Limited. All rights reserved.
309

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1596
ARTICLES
have been reported previously. For amino derivative 4, qualitative
nOe analysis of the compound was consistent with the trans-
amide conformation28. Quantitative rotating-frame nuclear
Overhauser effect correlation spectroscopy (ROESY) experiments
performed on geldanamycin 1 itself gave calculated inter-atomic
distances that were again consistent with an extended
conformation and a trans-configured amide31.
a
The NMR chemical shift data obtained in our work immediately
suggested conformational disparities between geldanamycin 1 and
our synthesized derivatives, with all of the 19-substituted com-
pounds exhibiting 1H chemical shift patterns that were significantly
different from those of the parent molecule. These differences in the
NMR spectra (Supplementary Figs S3 and S4) indicate a major
change in the environment of many of the positions, both those
influenced by a potential change in the amide configuration, and
also those in the opposite section of the macrocycle, reinforcing
the evidence for a change in the conformation of the ansa ring.
The 13C shifts, being relatively unaffected by the magnetic aniso-
b
1
tropy of other atoms or groups (unlike H, for example, aromatic
ring currents), are particularly compelling in this regard. We also
investigated the through-space correlations detected in nuclear
Overhauser effect correlation spectroscopy (NOESY) and ROESY
spectra, as well as undertaking a quantitative nOe study of
19-phenyl-AAG 16, with subsequent molecular modelling investi-
gations. These studies (for details, see Supplementary pages
S57–S75) strongly suggest that the dominant form in solution is a
cis-amide adopting a C-clamp type conformation. From the
model, it is clear that the compact C-shaped conformation allows
the hydrophobic surface area to be minimized by being ‘buried’ in
the heart of the structure, while the hydrophilic moieties are
exposed to the solvent. This observation would be consistent with
the large difference in polarity observed by thin layer chromato-
graphy (TLC) analysis.
To confirm that 19-substitution caused a trans to cis amide
change in conformation in the solid state, we sought evidence
from X-ray crystallography. 19-(2-Furyl)geldanamycin 13 gave
crystals (of its THF complex) that were suitable for X-ray diffraction.
The molecule exhibited both the cis-configured amide and also the
‘C-clamp’ conformation (Supplementary Fig. S12; CCDC reference
Figure 3 | Overlaid crystal structures reveal the effect of 19-substitution on
geldanamycin conformation and binding. a, The different conformations of
19-(2-furyl)geldanamycin 13 (green) with a cis-amide and 19-unsubstituted-
17-azetidinylgeldanamycin 4 (orange) with a trans-amide28. b, Similarity
between the unbound cis-amide conformation of 19-(2-furyl)geldanamycin
13 (green) and protein-bound geldanamycin 1 (blue) from the yeast
Hsp90–geldanamycin complex20
.
864025). Unsurprisingly, therefore, the conformation of macrocycle Substitution at C19 reduces toxicity in cells. A key tenet of our
13 is different to that of 19-unsubstituted geldanamycins that adopt original hypothesis was that introduction of a substituent at C19
a trans-amide conformation in the crystal; this is clearly seen by would not only cause the desired conformational switch, but
overlaying the structure of 13 and the published structure of com- would also prevent nucleophilic attack by thiols, thereby
pound 4 (ref. 28), as depicted in Fig. 3a. On the other hand, and ameliorating the toxicity of the parent BQAs. With the first two
in line with our original hypothesis, the isomerization from trans- features now established, we sought evidence for reduced toxicity,
to cis-amide induced by the 19-substituent does impart an overall and therefore evaluated the toxicity of the 19-substituted
‘C-clamp’ conformation to the BQA macrocycle that is very compounds in comparison to their parent BQAs in reducing
similar to that of the protein-bound geldanamycin, as shown in toxicity in normal endothelial and epithelial cellular systems. We
the overlaid structures in Fig. 3b.
Table 1 | Toxicity of benzoquinone ansamycins.
Substitution at C19 prevents nucleophilic attack. Part of the
rationale for the incorporation of a substituent at the 19-position
was that it would block the potential for metabolism by
Compound
HUVECs* IC50
(mM)
ARPE-19 IC50 (mM)
Geldanamycin 1
0.041+0.003
16.9+3.3
2.1+0.4
1.2+0.1
.20
.20
0.88+0.04
.20
0.10+0.04
.20
8.3+0.7
1.0+0.4
.20
.20
0.15+0.01
.20
nucleophilic attack of glutathione16–18
,
and hence might
19-Me-geldanamycin 6
19-Ph-geldanamycin 8
17-AAG 2
19-Me-17-AAG 15
19-Ph-17-AAG 16
17-DMAG 3
ameliorate the quinone-related hepatotoxicity seen with BQAs.
We therefore tested whether our 19-substituted analogues would
react with thiol nucleophiles. N-Acetylcysteine methyl ester was
utilized as a model for glutathione and was initially introduced to
geldanamycin 1, 17-AAG 2 and 17-DMAG 3 under neutral
conditions. Although this resulted in no reaction, basic conditions
(1,8-diazabicyclo[5.4.0]undec-7-ene (DBU), THF, room temperature)
afforded the corresponding adducts 29–31 in 35%, 62% and 21%
unoptimized yields, respectively (Fig. 2c). On subjecting our
19-methyl and 19-phenylgeldanamycin derivatives to identical
conditions, no reaction was observed with the thiol at C19, even
upon heating of the reaction mixture.
19-Me-17-DMAG 22
19-Ph-17-DMAG 23
17.7+1.6
8.2+0.4
*HUVEC are primary cells and ARPE-19 cells are a non-transformed human retinal pigmented
epithelial cell line. Toxicity values were generated using the 3-(4,5-dimethylthiazol-2-yl)-2,5-
diphenyltetrazolium bromide (MTT) assay. The values are represented as a mean+standard
deviation (n ¼ 3). IC50 (half-maximal inhibitory concentration) of 19-substituted BQAs and their
parent quinones. Culture conditions and the MTT assay are described in Supplementary section,
‘Toxicity Studies: Methods’, page S82.
310
NATURE CHEMISTRY | VOL 5 | APRIL 2013 | www.nature.com/naturechemistry
© 2013 Macmillan Publishers Limited. All rights reserved.

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1596
ARTICLES
a
b
c
Asp 40
Lys 44
Lys 44
Ser 36
Asp 40
Ser 36
Asn 37
Asn 37
90°
d
e
f
Asp 40
Ser 36
Asp 40
Lys 44
Glu 33
Lys 44
Lys 44
Ser 36
Asn 37
Asp 40
Gly 118
Gly 123
Asn 37
Ser 36
Asn 37
Figure 4 | Structures of 19-substituted geldanamycins bound in the ATP site of yeast Hsp90, as determined by protein X-ray crystallography. a, Two
orthogonal views of the superimposition of geldanamycin 1 (green), 19-methyl geldanamycin 6 (cyan) and 19-phenyl geldanamycin 8 (yellow), from the
co-crystal structures of N-terminal yeast Hsp90, showing the similarity of the bound conformations. b–f, Comparison of the binding of geldanamycin with
19-substituted geldanamycin analogues to the N-terminal domain of yeast Hsp90. PyMol diagrams showing the interactions of geldanamycin 1 (green) and
19-methyl geldanamycin 6 (cyan) with Hsp90 (green and cyan residues, respectively (b); geldanamycin 1 (green) and 19-methyl 17-AAG 15 (yellow) with
Hsp90 (green and yellow residues, respectively) (c); geldanamycin 1 (green) and 19-(2-furyl) geldanamycin 13 (grey) with Hsp90 (green and grey residues,
respectively) (d); geldanamycin 1 (green) and 19-methyl 17-DMAG 22 (gold) with Hsp90 (green and gold residues, respectively) (e); geldanamycin 1 (green)
and 19-phenyl geldanamycin 8 (salmon) with Hsp90 (green and salmon residues, respectively) (f). In general, in these geldanamycin analogues, the
19-substituents tend to alter binding to the protein through a positional change of the quinone group of the benzoquinone ansamycin. Dotted blue lines,
hydrogen bonds involving geldanamycin; dotted red lines, hydrogen bonds involving 19-substituted analogues; cyan-coloured spheres, water molecules. The
structures for geldanamycin and the 19-substituted derivatives were obtained at 2.0 Å (geldanamycin), 2.7 Å (19-methyl geldanamycin), 2.25 Å (19-methyl
17-AAG), 3.1 Å (19-methyl 17-DMAG), 2.2 Å (19-phenyl geldanamycin) and 2.6 Å (19-(2-furyl) geldanamycin) resolutions. For atomic coordinates and
structure factors, see PDB codes 1A4H, 4AS9, 4ASA, 4ASB, 4ASG and 4ASF, respectively.
used human umbilical vein endothelial cells (HUVECs) and retinal for comparison. Geldanamycin and its 19-substituted derivatives
pigmented epithelial cells (ARPE-19 cells) in these studies. Our data essentially bind in the same way to the N-terminal domain of
(Table 1) clearly demonstrate that 19-substitution markedly reduces Hsp90 (Fig. 4a), with, as expected, a cis-amide conformation.
BQA toxicity in all series of compounds tested. Both 19-methyl and However, all the 19-substituted analogues show a shift in the pos-
19-phenyl BQAs in the geldanamycin, 17-AAG and 17-DMAG ition of the quinone ring (Fig. 4b–f) relative to the geldanamycin
series were markedly less toxic than their parent quinones.
complex to avoid a steric clash with Asn 37, consequently altering
some of the interactions in these complexes relative to the geldana-
Binding to Hsp90. We analysed the thermodynamics of the mycin–Hsp90 complex. For example, in the geldanamycin–Hsp90
binding of 19-methylgeldanamycin to the N-terminal domain of complex, Lys 44 is normally hydrogen-bonded to the 13-hydroxy
yeast Hsp90 using isothermal titration calorimetry (ITC). group of geldanamycin. With 19-methyl geldanamycin 6, the repo-
Geldanamycin binds to the protein with ꢀ1:1 stoichiometry with sitioning of the quinone group results in Lys 44 forming a hydrogen
a Kd of 2.9 mM. The binding is accompanied by a favourable bond, via a water molecule, with one of the quinone oxygens of
enthalpy change, although it does incur an entropic penalty. For 19-methyl geldanamycin (Fig. 4b). For geldanamycin, the same
the 19-methyl derivative 6, there is an approximately fivefold loss quinone oxygen normally forms a hydrogen bond with one of the
of binding affinity (Kd ¼ 16.3 mM) relative to geldanamycin itself, oxygens of Asp 40, while in the 19-methyl geldanamycin–Hsp90
with a loss of enthalpic contribution, somewhat compensated for complex, Asp 40 adopts an alternative conformation that disrupts
by an increase in entropic contribution favouring binding. To
a pre-existing network of water-mediated hydrogen bonds
investigate the reasons for this slight loss of binding upon between the same quinone group in question and the hydroxyl
introduction of the 19-methyl group, we turned to protein oxygen and main-chain oxygen of Ser 36 (Fig. 4b). Loss of these
X-ray crystallography.
waters might account for the increase in the entropic contribution
The crystal structure of geldanamycin with the N-terminal favouring binding. A similar effect is also seen with the 19-methyl
domain of yeast Hsp90 has been determined previously20, so we derivative of 17-DMAG 22 (Fig. 4e). With 19-methyl 17-AAG 15
studied the Hsp90 complexes of the 19-substituted geldanamycins and 19-methyl 17-DMAG 22 we see essentially the same changes,
NATURE CHEMISTRY | VOL 5 | APRIL 2013 | www.nature.com/naturechemistry
© 2013 Macmillan Publishers Limited. All rights reserved.
311

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1596
ARTICLES
except that the Asp 40 residue appears to flip between the two
alternative conformations seen. Consequently, 19-methyl 17-AAG
15 makes hydrogen bonds (via water molecules) to Lys 44, but
the quinone oxygen interactions (via two water molecules) with
Ser 36 appear disrupted (Fig. 4c). Similarly, for 19-methyl 17-
DMAG 22, Lys 44 forms direct hydrogen bonds with Asp 40,
while another hydrogen bond (via a water molecule) is seen
between one of the Asp 40 oxygens and the amide nitrogen of the
17-DMAG group (Fig. 4e). For 19-phenyl 8 and 19-(2-furyl)-gelda-
namycin 13, the Asp 40 adopts its geldanamycin bound confor-
mation. However, for these compounds, Lys 44 forms two direct
hydrogen bonds with the side-chain oxygens of Asp 40
(Fig. 4d,f). For the 19-(2-furyl) analogue 13, we also see some
additional hydrogen-bond interaction with the 2-furyl group
oxygen. This oxygen atom forms a network of hydrogen bonds
(via water molecules) to the main-chain amide of Gly 123, the
main-chain carbonyl of Gly 118 and to one of the oxygens of the
catalytic Glu 33 residue (Fig. 4d). However, collectively, the results
indicate that in these geldanamycin analogues, the 19-substituents
tend to alter binding to the protein through forced positional
change of the quinone group of the benzoquinone ansamycin.
This is in spite of the fact that these 19-substituted compounds
already exist in the cis-amide conformation required for binding
to Hsp90, as can be seen for the 19-(2-furyl) analogue 13, where
the protein bound conformation is very similar, although not com-
pletely identical, to that seen in the crystal structure of this analogue
alone (Fig. 3; Supplementary Fig. S12).
a
DMAG
19Ph-DMAG
DMSO
1
5
DMSO
1
5
(μM)
Raf-1
Hsp70
Cdc-2
β-actin
DMAG
1
19Ph-DMAG
b
DMSO 0.5
5 (μM) DMSO 0.5
1
5 (μM)
Her-2
Raf-1
Hsp70
Cdc-2
β-actin
17AAG
GA
19Ph-GA
c
DMSO 0.25
0.5
0.25
0.5
0.25
0.5
(μM)
Hsp70
β-actin
Hsp27
Inhibition of Hsp90 in cellular systems. X-ray crystallography
clearly showed that 19-substituted derivatives do bind to the ATP
site in Hsp90. We therefore examined the ability of 19-substituted
compounds to inhibit Hsp90 in cellular systems. A decrease in
Hsp90 client proteins (including Raf1, Her2 and Cdc-2) and a
compensatory increase in Hsp70 are the two most common
biomarkers used for inhibition of Hsp90 in cells45–47. Initially, we
used MDA-468-NQ16 cells that over-express the quinone
reductase NQO1, because we have shown that NQO1 potentiates
Hsp90 inhibition due to the formation of the more active
hydroquinone ansamycin45,46, and worked with the 17-DMAG
(alvespimycin) series. In Fig. 5a we demonstrate that both 17-
DMAG and 19-phenyl 17-DMAG are effective Hsp90 inhibitors,
as shown by decreases in Hsp90 client proteins and compensatory
induction of Hsp70. We extended these data to the Her2-positive
BT474 breast cancer cell line, and both DMAG and 19-phenyl
DMAG resulted in degradation of Hsp90 clients, including Her2,
as well as inducing a compensatory increase in Hsp70 (Fig. 5b).
In addition to their use as anticancer agents, Hsp90 inhibitors
have attracted attention as potential protective agents against neuro-
β-actin
Figure 5 | Evaluation of 19-substituted compounds as inhibitors of Hsp90
in cellular systems. a, Immunoblot analysis of biomarkers of Hsp90
inhibition in MDA468-NQ16 human breast cancer cells treated with either
17-DMAG or 19-phenyl 17-DMAG for 48 h and immunoblotted for the
Hsp90 clients Raf-1, Cdc-2 and for Hsp70. b, Immunoblot analysis of
biomarkers of Hsp90 inhibition in BT474 human breast cancer cells treated
with either 17-DMAG or 19-phenyl 17-DMAG for 48 h and immunoblotted
for the Hsp90 clients Her-2 Raf-1, Cdc-2 and for Hsp70. c, Immunoblot
analysis of Hsp70 and Hsp27 induction in SH-SY5Y neuroblastoma cells
following treatment with either 17-AAG, geldanamycin or 19-phenyl
geldanamycin for 16 h. In general, we observe a decrease in Hsp90 client
proteins (Raf-1, Her-2 and Cdc-2) and a compensatory increase in other
Hsps (27 and 70), the two most common biomarkers used for inhibition of
Hsp90 in cells. Results are representative of three separate experiments.
b-actin was used as a loading control.
degeneration due to their ability to induce a compensatory increase Hsp27 levels in dopaminergic cells, particularly when coupled
in other heat shock proteins that protect against aberrant protein with their lower toxicity to normal cells, suggests that these novel
folding and aggregation8–14. SH-SY5Y neuroblastoma cells are dopa- compounds may have useful therapeutic applications.
minergic and have been used as a cellular model system of relevance
In this study, the power of palladium-catalysed coupling reac-
to Parkinson’s disease48–50. In Fig. 5c, we demonstrate that both gel- tions has been demonstrated with the synthesis of a range of
danamycin and 19-phenyl geldanamycin induce Hsp70 and Hsp27 novel 19-substituted benzoquinone ansamycins, specifically
in SH-SY5Y human neuroblastoma cells. Both compounds were designed to both block nucleophilic attack at C19, a potential
extremely potent at inducing Hsp70 and Hsp27 and were found source of metabolic instability and toxicity, and to cause a confor-
to be effective inducers in the nanomolar range (Fig. 5c). mational switch to facilitate protein binding. NMR spectroscopy
Expansion of the dose–response curves to low nanomolar doses established that the resulting macrocycles did indeed adopt the
demonstrated that geldanamycin was more potent than 19-phenyl desired cis-amide conformation, while protein X-ray crystallography
geldanamycin at Hsp induction (Supplementary Fig. S16), but, showed that the compounds bound to the N-terminal ATP site of
importantly, both geldanamycin and 19-phenyl geldanamycin Hsp90. Importantly, the 19-substituted benzoquinone ansamycins
were more potent at the induction of Hsp70 and Hsp27 than were found to be markedly less toxic to normal endothelial and epi-
17-AAG, the benzoquinone ansamycin that advanced to Phase II thelial cell systems than their parent quinones. Studies in both
clinical trial (Fig. 5c). The effect of 19-substituted benzoquinone human cancer cells and dopaminergic neural cells established that
ansamycins on both oncogenic client proteins in tumour cells and the 19-substituted benzoquinone ansamycins are effective inhibitors
their ability to induce a compensatory increase in Hsp70 and of Hsp90, exhibiting the appropriate molecular signature: decreases
312
NATURE CHEMISTRY | VOL 5 | APRIL 2013 | www.nature.com/naturechemistry
© 2013 Macmillan Publishers Limited. All rights reserved.

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1596
ARTICLES
23. Shen, H-Y. et al. Geldanamycin induces heat shock protein 70 and protects
against MPTP-induced dopaminergic neurotoxicity in mice. J. Biol. Chem.
280, 39962–39969 (2005).
24. Waza, M. et al. 17-AAG, an Hsp90 inhibitor, ameliorates polyglutamine-
mediated motor neuron degeneration. Nature Med. 11, 1088–1095 (2005).
25. Waza, M. et al. Alleviating neurodegeneration by an anticancer agent—an Hsp90
inhibitor (17-AAG). Ann. NY Acad. Sci. 1086, 21–34 (2006).
26. Tadtong, S. et al. Geldanamycin derivatives and neuroprotective effect
on cultured P19-derived neurons. Bioorg. Med. Chem. Lett. 17,
2939–2943 (2007).
in Hsp90 client proteins and a compensatory increase in other heat
shock proteins including Hsp70 and Hsp27. When considered
together with their markedly decreased toxicity to normal cells,
this suggests that 19-substituted benzoquinone ansamycins may
have a greater therapeutic window than their parent quinones and
hence considerable potential for application in the therapy of
cancer and neurodegenerative diseases.
Methods
27. Rinehart, K. L. & Shield, L. S. Chemistry of the ansamycin antibiotics.
Prog. Chem. Org. Nat. Prod. 33, 231–307 (1976).
28. Schnur, R. C. & Corman, M. L. Tandem [3,3]-sigmatropic rearrangements in an
ansamycin—stereospecific conversion of an (S)-allylic alcohol to an (S)-allylic
amine derivative. J. Org. Chem. 59, 2581–2584 (1994).
All experimental procedures, characterization data for all compounds and copies of
the 1H and 13C NMR spectra, details of NMR studies, biological evaluations and
protein crystallography are given in the Supplementary Information.
29. Jez, J. M. et al. Crystal structure and molecular modeling of 17-DMAG in
complex with human Hsp90. Chem. Biol. 10, 361–368 (2003).
30. Lee, Y-S., Marcu, M. G. & Neckers, L. Quantum chemical calculations and
mutational analysis suggest heat shock protein 90 catalyzes trans–cis
isomerization of geldanamycin. Chem. Biol. 11, 991–998 (2004).
31. Thepchatri, P. et al. Relationship among ligand conformations in solution, in the
solid state, and at the Hsp90 binding site: geldanamycin and radicicol. J. Am.
Chem. Soc. 129, 3127–3134 (2007).
Received 8 November 2012; accepted 4 February 2013;
published online 10 March 2013
References
1. McDonald, E., Workman, P. & Jones, K. Inhibitors of the Hsp90 molecular
chaperone: attacking the master regulator in cancer. Curr. Top. Med. Chem. 6,
1091–1107 (2006).
32. Onuoha, S. C. et al. Mechanistic studies on Hsp90 inhibition by ansamycin
derivatives. J. Mol. Biol. 372, 287–297 (2007).
2. Janin, Y. L. ATPase inhibitors of heat-shock protein 90, second season. Drug
Discov. Today 15, 342–353 (2010).
3. Porter, J. R., Fritz, C. C. & Depew, K. M. Discovery and development of Hsp90
inhibitors: a promising pathway for cancer therapy. Curr. Opin. Chem. Biol. 14,
412–420 (2010).
4. Fadden, P. et al. Application of chemoproteomics to drug discovery:
identification of a clinical candidate targeting Hsp90. Chem. Biol. 17,
686–694 (2010).
5. Massey, A. J. ATPases as drug targets: insights from heat shock proteins 70 and
90. J. Med. Chem. 53, 7280–7286 (2010).
33. Reigan, P., Siegel, D., Guo, W. & Ross, D. A mechanistic and structural analysis
of the inhibition of the 90-kDa heat shock protein by the benzoquinone and
hydroquinone ansamycins. Mol. Pharmacol. 79, 823–832 (2011).
34. Janin, Y. L. Heat shock protein 90 inhibitors. A text book example of medicinal
chemistry. J. Med. Chem. 48, 7503–7512 (2005).
35. Wrona, I. E., Agouridas, V. & Panek, J. S. Design and synthesis of ansamycin
antibiotics. C. R. Chimie 11, 1483–1522 (2008).
36. Sasaki, K. & Inoue, Y. Geldanamycinderivate und Antitumormittel mit einem
Gehalt derselben. German patent 3,006,097 (1980).
6. Trepel, J., Mollapour, M., Giaccone, G. & Neckers, L. Targeting the dynamic
Hsp90 complex in cancer. Nature Rev. Cancer 10, 537–549 (2010).
7. Biamonte, M. A. et al. Heat shock protein 90: inhibitors in clinical trials. J. Med.
Chem. 53, 3–17 (2010).
37. Schnur, R. C. et al. Inhibition of the oncogene product P185(Erbb-2) in vitro and
in vivo by geldanamycin and dihydrogeldanamycin derivatives. J. Med. Chem.
38, 3806–3812 (1995).
38. Rinehart, K. L. et al. Synthesis of hydrazones and oximes of geldanaldehyde
as potential polymerase inhibitors. Bioorg. Chem. 6, 341–351 (1977).
39. Sasaki, K. & Inoue, Y. Novel geldanamycin derivatives as pharmaceutically active
ingredients and their preparation. Japan patent JP 57-163369 (1981).
40. Wu, L. et al. Geldanamycin biosynthetic analog 19-O-glycylgeldanamycin.
China patent CN 101792418A (2010).
41. Farina, V. & Krishnan, B. Large rate accelerations in the Stille reaction with
tri-2-furylphosphine and triphenylarsine as palladium ligands—mechanistic
and synthetic implications. J. Am. Chem. Soc. 113, 9585–9595 (1991).
42. Negishi, E-I. et al. Nickel-catalyzed or palladium-catalyzed cross coupling. 31.
Palladium-catalyzed or nickel-catalyzed reactions of alkenylmetals with
unsaturated organic halides as a selective route to arylated alkenes and
conjugated dienes—scope, limitations, and mechanism. J. Am. Chem. Soc.
109, 2393–2401 (1987).
43. Liebeskind, L. S. & Fengl, R. W. 3-Stannylcyclobutenediones as nucleophilic
cyclobutenedione equivalents—synthesis of substituted cyclobutenediones and
cyclobutenedione monoacetals and the beneficial effect of catalytic copper iodide
on the Stille reaction. J. Org. Chem. 55, 5359–5364 (1990).
44. Harrowven, D. C. et al. Potassium carbonate–silica: a highly effective stationary
phase for the chromatographic removal of organotin impurities. Chem.
Commun. 46, 6335–6337 (2010).
45. Guo, W. et al. Formation of 17-allylamino-demethoxygeldanamycin (17-AAG)
hydroquinone by NAD(P)H:quinone oxidoreductase 1 (NQO1): role of 17-AAG
hydroquinone in heat shock protein 90 inhibition. Cancer Res. 65,
10006–10015 (2005).
46. Guo, W. et al. The bioreduction of a series of benzoquinone ansamycins by
NAD(P)H: quinone oxidoreductase 1 to more potent heat shock protein
90 inhibitors, the hydroquinone ansamycins. Mol. Pharmacol. 70,
1194–1203 (2006).
47. Siegel, D. et al. Role for NAD(P)H:quinone oxidoreductase 1 and
manganese-dependent superoxide dismutase in 17-(allylamino)-17-
demethoxygeldanamycin-induced heat shock protein 90 inhibition in
pancreatic cancer cells. J. Pharmacol. Exp. Ther. 336, 874–880 (2011).
48. Kitamura, Y. et al. Protective effects of the anti-Parkinsonian drugs talipexole
and pramipexole against 1-methyl-4-phenylpyridinium-induced apoptotic death
in human neuroblastoma SH-SY5Y cells. Mol. Pharmacol. 54, 1046–1054 (1998).
49. Smith, W. W. et al. alpha-Synuclein phosphorylation enhances
eosinophilic cytoplasmic inclusion formation in SH-SY5Y cells. J. Neurosci.
25, 5544–5552 (2005).
8. Gallo, K. A. Targeting Hsp90 to halt neurodegeneration. Chem. Biol. 13,
115–116 (2006).
9. Waza, M. et al. Modulation of Hsp90 function in neurodegenerative disorders:
a molecular-targeted therapy against disease-causing protein. J. Mol. Med.
84, 635–646 (2006).
10. Luo, G-R., Chen, S. & Le, W-D. Are heat shock proteins therapeutic target for
Parkinson’s disease? Int. J. Biol. Sci. 3, 20–26 (2007).
11. Adachi, H. et al. Heat shock proteins in neurodegenerative diseases: pathogenic
roles and therapeutic implications. Int. J. Hyperther. 25, 647–654 (2009).
12. Sajjad, M. U., Samson, B. & Wyttenbach, A. Heat shock proteins: therapeutic
drug targets for chronic neurodegeneration? Curr. Pharm. Biotechnol. 11,
198–215 (2010).
13. Kalia, S. K., Kalia, L. V. & McLean, P. J. Molecular chaperones as rational drug
targets for Parkinson’s disease therapeutics. CNS Neurol. Disord.: Drug Targets
9, 741–753 (2010).
14. Aridon, P. et al. Protective role of heat shock proteins in Parkinson’s disease.
Neurodegen. Dis. 8, 155–168 (2011).
15. Supko, J. G., Hickman, R. L., Grever, M. R. & Malspeis, L. Preclinical
pharmacological evaluation of geldanamycin as an antitumor agent. Cancer
Chemother. Pharmacol. 36, 305–315 (1995).
16. Cysyk, R. L. et al. Reaction of geldanamycin and C17-substituted analogues
with glutathione: product identifications and pharmacological implications.
Chem. Res. Toxicol. 19, 376–381 (2006).
17. Lang, W. et al. Biotransformation of geldanamycin and 17-allylamino-17-
demethoxygeldanamycin by human liver microsomes: reductive versus oxidative
metabolism and implications. Drug Metab. Dispos. 35, 21–29 (2007).
18. Guo, W., Reigan, P., Siegel, D. & Ross, D. Enzymatic reduction and glutathione
conjugation of benzoquinone ansamycin heat shock protein 90 inhibitors:
relevance for toxicity and mechanism of action. Drug Metab. Dispos. 36,
2050–2057 (2008).
19. Stebbins, C. E. et al. Crystal structure of an Hsp90–geldanamycin complex:
targeting of a protein chaperone by an antitumor agent. Cell 89, 239–250 (1997).
20. Roe, S. M. et al. Structural basis for inhibition of the Hsp90 molecular chaperone
by the antitumor antibiotics radicicol and geldanamycin. J. Med. Chem. 42,
260–266 (1999).
21. Dehner, A. et al. NMR chemical shift perturbation study of the N-terminal
domain of Hsp90 upon binding of ADR AMP-PNP, geldanamycin, and
radicicol. ChemBioChem 4, 870–877 (2003).
22. Modi, S. et al. HSP90 inhibition is effective in breast cancer: a phase 2 trial of
tanespimycin (17AAG) plus trastuzumab in patients with HER2-positive
metastatic breast cancer progressing on trastuzumab. Clin. Cancer Res. 17,
5132–5139 (2011).
50. Jeong, H. J. et al. Transduced Tat-DJ-1 protein protects against oxidative
stress-induced SH-SY5Y cell death and Parkinson disease in a mouse model.
Mol. Cell 33, 471–478 (2012).
NATURE CHEMISTRY | VOL 5 | APRIL 2013 | www.nature.com/naturechemistry
313
© 2013 Macmillan Publishers Limited. All rights reserved.

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1596
ARTICLES
Acknowledgements
co-directed by D.S., and carried out by C-H.C., D.L.D. and R.X. Protein crystallography
was directed by C.P. and carried out by C.P. and S.M.R. All authors contributed to the
preparation of the manuscript.
The authors acknowledge support from the Parkinson’s Disease Society UK (R.R.A.K. and
C.J.M.). The work was also supported by the National Institutes of Health (NIH grants
CA51210 and ES018943; D.R., D.S., C-H.C. and R.X.). The authors thank S. Aslam and
K. Butler for help with NMR studies, and S. Young (University of Nottingham, ICPMS),
M. Cooper and G. Coxhill (University of Nottingham, mass spectrometry) for
technical assistance.
Additional information
Supplementary information and chemical compound information are available in the
online version of the paper. Reprints and permissions information is available online at www.
nature.com/reprints. Correspondence and requests for materials should be addressed to C.P.,
D.R. and C.J.M.
Author contributions
D.R. and C.J.M. jointly conceived the project, and directed the biology and chemistry,
respectively. R.R.A.K. was responsible for the design and execution of the chemistry. NMR
studies and molecular modelling were carried out by R.R.A.K., H.E.L.W. and A.L.D., and Competing financial interests
W.L. carried out small-molecule X-ray crystallography. The biological studies were
The authors declare no competing financial interests.
314
NATURE CHEMISTRY | VOL 5 | APRIL 2013 | www.nature.com/naturechemistry
© 2013 Macmillan Publishers Limited. All rights reserved.