A mechanistic and structural analysis of the inhibition of the 90-kDa heat shock protein by the benzoquinone and hydroquinone ansamycins

Article abstract of DOI:10.1124/mol.110.070086

The benzoquinone ansamycins inhibit the ATPase activity of the 90-kDa heat shock protein (Hsp90), disrupting the function of numerous client proteins involved in oncogenesis. In this study, we examine the role of NAD(P)H:quinone oxidoreductase 1 (NQO1) in the metabolism of trans-and cis-amide isomers of the benzoquinone ansamycins and their mechanism of Hsp90 inhibition. Inhibition of purified human Hsp90 by a series of benzoquinone ansamycins was examined in the presence and absence of NQO1, and their relative rate of NQO1-mediated reduction was determined. Computational-based molecular docking simulations indicated that the trans-but not the cis-amide isomers of the benzoquinone ansamycins could be accommodated by the NQO1 active site, and the ranking order of binding energies correlated with the relative reduction rate using purified human NQO1. The trans-cis isomerization of the benzoquinone ansamycins in Hsp90 inhibition has been disputed in recent reports. Previous computational studies have used the closed or cocrystallized Hsp90 structures in an attempt to explore this isomerization step; however, we have successfully docked both the trans-and cis-amide isomers of the benzoquinone ansamycins into the open Hsp90 structure. The results of these studies indicate that both trans-and cisamide isomers of the hydroquinone ansamycins exhibited increased binding affinity for Hsp90 relative to their parent quinones. Our data support a mechanism in which trans-rather than cis-amide forms of benzoquinone ansamycins are metabolized by NQO1 to hydroquinone ansamycins and that Hsp90-mediated trans-cis isomerization via tautomerization plays an important role in subsequent Hsp90 inhibition. Copyright

Full text of DOI:10.1124/mol.110.070086

Supplemental material to this article can be found at:
http://molpharm.aspetjournals.org/content/suppl/2011/02/01/mol.110.070086.DC1.html
0
026-895X/11/7905-823–832$25.00
MOLECULAR PHARMACOLOGY
Copyright © 2011 The American Society for Pharmacology and Experimental Therapeutics
Vol. 79, No. 5
70086/3680556
Mol Pharmacol 79:823–832, 2011
Printed in U.S.A.
A Mechanistic and Structural Analysis of the Inhibition of the
9
0-kDa Heat Shock Protein by the Benzoquinone and
□
S
Hydroquinone Ansamycins
Philip Reigan, David Siegel, Wenchang Guo, and David Ross
Department of Pharmaceutical Sciences, School of Pharmacy, University of Colorado Denver, Denver, Colorado
Received November 18, 2010; accepted February 1, 2011
ABSTRACT
The benzoquinone ansamycins inhibit the ATPase activity of benzoquinone ansamycins in Hsp90 inhibition has been dis-
the 90-kDa heat shock protein (Hsp90), disrupting the function puted in recent reports. Previous computational studies have
of numerous client proteins involved in oncogenesis. In this used the closed or cocrystallized Hsp90 structures in an at-
study, we examine the role of NAD(P)H:quinone oxidoreduc- tempt to explore this isomerization step; however, we have
tase 1 (NQO1) in the metabolism of trans- and cis-amide iso- successfully docked both the trans- and cis-amide isomers of
mers of the benzoquinone ansamycins and their mechanism of the benzoquinone ansamycins into the open Hsp90 structure.
Hsp90 inhibition. Inhibition of purified human Hsp90 by a series The results of these studies indicate that both trans- and cis-
of benzoquinone ansamycins was examined in the presence amide isomers of the hydroquinone ansamycins exhibited in-
and absence of NQO1, and their relative rate of NQO1-medi- creased binding affinity for Hsp90 relative to their parent qui-
ated reduction was determined. Computational-based molec- nones. Our data support a mechanism in which trans- rather
ular docking simulations indicated that the trans- but not the than cis-amide forms of benzoquinone ansamycins are metab-
cis-amide isomers of the benzoquinone ansamycins could be olized by NQO1 to hydroquinone ansamycins and that Hsp90-
accommodated by the NQO1 active site, and the ranking order mediated trans-cis isomerization via tautomerization plays an
of binding energies correlated with the relative reduction rate important role in subsequent Hsp90 inhibition.
using purified human NQO1. The trans-cis isomerization of the
tein, operating in a dynamic “chaperone cycle” dependent on
Introduction
the ATPase activity of Hsp90 (Pearl and Prodromou, 2006).
The 90-kDa heat shock protein (Hsp90) is overexpressed in
Hsp90 is an important anticancer target, and the conserved
cancer cells and is implicated in their survival by maintain-
ATP-binding domain of Hsp90 is also the binding site of the
ing the active conformation of key oncogenic proteins, includ-
natural products geldanamycin (GM) and radicicol and a
ing ErbB2, Raf-1, Akt, hypoxia-inducible factor-1␣, and mu-
range of semisynthetic and synthetic compounds (Roe et al.,
tant p53 (Maloney and Workman, 2002). The function of
1999; Maloney and Workman, 2002). These compounds pre-
Hsp90 is complex and involves homodimerization, recruit-
ment of cochaperones, accessory proteins, and the client pro-
vent Hsp90 from cycling between ADP- and ATP-bound con-
formations, resulting in the degradation of multiple onco-
genic client proteins via the ubiquitin-proteasome pathway
and ultimately growth arrest or apoptosis (Maloney and
Workman, 2002).
This work was supported by the National Institutes of Health National
Cancer Institute [Grant R01-CA51210].
The authors disclose a pending patent interest in hydroquinone ansamycins
and their prodrugs (international application number PCT/US2005/031524).
Article, publication date, and citation information can be found at
http://molpharm.aspetjournals.org.
The benzoquinone ansamycin class of Hsp90 inhibitors
include GM and its semisynthetic derivative 17-allylamino-
doi:10.1124/mol.110.070086.
17-demethoxy-geldanamycin (17AAG), which have been
□S The online version of this article (available at http://molpharm.
aspetjournals.org) contains supplemental material.
shown to bind to Hsp90 with micromolar affinity in vitro and
ABBREVIATIONS: Hsp90, 90-kDa heat shock protein; 17AAG, 17-(allylamino)-17-demethoxygeldanamycin; 17AAGH , 17-(allylamino)-17-
2
demethoxygeldanamycin hydroquinone; 17AG, 17-(amino)-17-demethoxygeldanamycin; 17AGH , 17-(amino)-17-demethoxygeldanamycin
2
hydroquinone; 17DMAG, 17-demethoxy-17-[[2-(dimethylamino)ethyl]amino]-geldanamycin; 17AEP-GA, 17-demethoxy-17-[[2-(pyrrolidin-1-
yl)ethyl]amino]-geldanamycin; 17AEP-GAH , 17-demethoxy-17-[[2-(pyrrolidin-1-yl)ethyl]amino]-geldanamycin hydroquinone; ES936, 5-me-
2
thoxy-1,2-dimethyl-3-[(4-nitrophenoxy)methyl]indole-4,7-dione; GM, geldanamycin; GMH , geldanamycin hydroquinone; NQO1, NAD(P)H:
2
quinone oxidoreductase 1; PDB, Protein Data Bank; PPIases, peptidyl-prolyl isomerases; LC/MS, liquid chromatography/mass
spectrometry; HPLC, high-performance liquid chromatography; rh, recombinant human.
823

8
24
Reigan et al.
with nanomolar activity in vivo (Roe et al., 1999; Chiosis et quinone and hydroquinone ansamycins in the nucleotide-
al., 2003) and have demonstrated selectivity toward tumor binding site of human Hsp90 was also investigated using
cells (Chiosis and Neckers, 2006). The benzoquinone ansa- molecular docking to explore the potential mechanism of
mycins exist in two isomeric forms: in solution they adopt an Hsp90-mediated trans-cis isomerization.
almost planar trans-amide conformation (Schnur and Cor-
man, 1994), whereas cocrystallization studies indicate the
formation of the C-shaped cis-amide isomer in the nucleotide-
Materials and Methods
binding site of Hsp90 (Stebbins et al., 1997; Jez et al., 2003).
Isomerization of the benzoquinone ansamycins has been pro-
posed to be a key requirement for Hsp90 inhibition (Lee et
al., 2004). An early cocrystallization study suggested that
Hsp90 may bind to the benzoquinone ansamycins because
they resemble misfolded peptides (Stebbins et al., 1997).
Although the nucleotide-binding site of Hsp90 is not a recog-
nized peptide-binding domain, the Hsp90-mediated isomer-
Materials. GM, 17-demthoxy-17-[[2-(dimethylamino)ethyl]amino]-
geldanamycin (17DMAG), and 17-demethoxy-17-[[2-(pyrrolidin-1-
yl)ethyl]amino]-geldanamycin (17AEP-GA) were obtained from
Invitrogen (Carlsbad, CA), and 17AAG and 17-(amino)-17-
demethoxygeldanamycin (17AG) were obtained from the National
Cancer Institute and Kosan Biosciences (Hayward, CA). 2,6-
Dichlorophenol-indophenol, NADH, NADPH, bovine serum albu-
min, and D(Ϫ)penicillamine were obtained from the Sigma-
Aldrich (St. Louis, MO). Malachite green phosphate assay kit was
ization of the benzoquinone ansamycins has been supported obtained from BioAssay Systems (Hayward, CA). 5-Methoxy-1,2-
by quantum and molecular mechanic studies (Jez et al., 2003; dimethyl-3-[(4-nitrophenoxy)methyl]indole-4,7-dione (ES936)
Lee et al., 2004) and the cis-trans isomerization of nonproline (Winski et al., 2001) was supplied by Professor Christopher J.
Moody (School of Chemistry, University of Nottingham, Notting-
peptide bonds (Schiene-Fischer et al., 2002). Other studies
ham, United Kingdom). Human Hsp90 was a kind gift from Pro-
have proposed that a cochaperone may have isomerase activ-
fessor David Toft (Mayo Clinic College of Medicine, Rochester,
ity (Pearl and Prodromou, 2000; Kamal et al., 2003), and a
MN). Recombinant human NQO1 was purified from Escherichia
recent study has disputed the isomerization of the benzoqui-
coli as described previously (Beall et al., 1994). The activity of
rhNQO1 was 4.5 ␮mol 2,6-dichlorophenol-indophenol per minute
per milligram.
none ansamycins as a requirement for Hsp90 inhibition
(
Onuoha et al., 2007).
In addition to trans-cis isomerization, the C17 substituents
HPLC and LC/MS Analysis. The NQO1-mediated reduction of
of the benzoquinone ansamycins are extensively metabolized the benzoquinone ansamycins was monitored by HPLC on a Luna
in vivo (Egorin et al., 1998), and the 19-position is prone to C18 5 ␮m, 4.6 ϫ 250 mm reversed-phase column (Phenomenex,
glutathionylation (Cysyk et al., 2006; Guo et al., 2008). The Torrance, CA) at room temperature. HPLC conditions were as fol-
lows: buffer A, 50 mM ammonium acetate, pH 4, containing 10 ␮M
D(Ϫ)-penicillamine; buffer B, acetonitrile (100%). Both buffers were
redox active quinone moiety is susceptible to one- and two-
electron reduction by flavin-containing reductases (Guo et
continuously bubbled with argon, gradient, 30% buffer B to 90%
al., 2005; Lang et al., 2007). The direct two-electron reduction
buffer B over 10 min and then 90% buffer B for 5 min (flow rate of 1
of benzoquinone ansamycins catalyzed by NAD(P)H:quinone
ml/min). The sample injection volume was 50 ␮l. LC/MS was per-
oxidoreductase 1 (NQO1; DT-diaphorase, EC: 1.6.99.2) gen-
formed using positive-ion electrospray ionization, and mass spectra
erating their hydroquinone derivatives circumvents the for-
were obtained using a PE Sciex API-3000 triple quadrupole MS
mation of semiquinone radicals and reactive oxygen species
(
Applied Biosystems, Foster City, CA) with a turbo ion spray source
(
Ross, 2004). Furthermore, NQO1 is expressed at high levels interfaced to a PE Sciex 200 HPLC system. Samples were separated
in many solid tumors (Siegel and Ross, 2000), and the ex- on a Luna C18 5 ␮m, 50 ϫ 2 mm reverse-phase column (Phenome-
pression of NQO1 has been found to correlate with 17AAG nex) using a gradient elution consisting of a 2 min initial hold at 20%
sensitivity (Kelland et al., 1999), offering the potential for
buffer B followed by an increase to 80% buffer B over 20 min at a flow
drug activation with tumor selectivity (Rooseboom et al.,
2
004).
In our previous studies, we reported the metabolism of a
series of benzoquinone ansamycins by recombinant human
rh)NQO1 generating the corresponding hydroquinone ansa-
(
mycins (Scheme 1), and these were more potent inhibitors of
yeast Hsp90 ATPase activity than their parent quinones
(
Guo et al., 2005; Guo et al., 2006). This potentiated inhibi-
tion was rationalized by molecular modeling simulations that
displayed increased direct hydrogen bond interactions be-
tween the hydroquinone ansamycins and the amino acid
residues in the nucleotide-binding site of Hsp90 (Guo et al.,
2
005, 2006). In this study, we extend our previous investiga-
tions using yeast Hsp90 to human Hsp90 (yeast Hsp90 has
ϳ60% homology with human Hsp90). We have examined the
relative rate of rhNQO1-mediated reduction of a series of
benzoquinone ansamycins and the inhibition of purified hu-
man Hsp90 by both benzoquinone and hydroquinone ansa-
mycins. Computational-based molecular docking was used to
investigate the conformation of the benzoquinone ansamy-
cins in the NQO1 active site and the structural properties
that influence the rate of NQO1-mediated reduction. The
Scheme 1. The NQO1-mediated reduction of the benzoquinone ansamy-
interaction of the trans- and cis-amide isomers of the benzo- cins.

Hsp90 Inhibition by Benzoquinone and Hydroquinone Ansamycins
825
rate of 200 ␮l/min and a sample injection volume of 20 ␮l. Solvent A Gly150, Ser151, Met154, Tyr155, His161, Pro68, Gly122, Tyr126,
was made up of 10 mM ammonium acetate containing 0.1% (v/v) Thr127, Tyr128, Met131, Phe178, Phe232, and the FAD cofactor. An
acetic acid, pH 4.4, and solvent B was composed of 10 mM ammo-
in situ minimization using the conjugate gradient method (1000
nium acetate in acetonitrile containing 0.1% (v/v) acetic acid. Mass iterations with a root mean square cutoff of 0.001 kcal/mol) coupled
spectra were continuously recorded from 150 to 1000 atomic mass with the Poisson-Boltzmann surface area implicit solvent model, was
units every 3 s during the chromatographic analysis, with a turbo ion used to calculate binding energy for each docked pose (Feig et al.,
spray temperature of 250°C, spray needle voltage at 4500 V, declus-
tering potential at 35 V, and focus plate at 125 V.
2004).
Hsp90 Docking. The cis- and trans-isomers of the benzoquinone
Reduction of Benzoquinone Ansamycins by rhNQO1. Ben- and hydroquinone ansamycins were docked separately into the nu-
zoquinone ansamycin 50 ␮M, NADH 200 ␮M, and rhNQO1 (1.65 ␮g cleotide-binding site of the open human Hsp90 structure, defined as
of for GM, 17AAG, and 17AG; 3.3 ␮g for 17DMAG; 6.6 ␮g for 17AEP- whole residues within a 4-Å radius subset. The flexible docking
GA) in 50 mM potassium phosphate buffer, pH 7.4 (1 ml), containing program was used to dock the benzoquinone and hydroquinone ansa-
1
mg/ml bovine serum albumin were incubated at 37°C for 10 min.
mycins into the nucleotide-binding site of Hsp90, which was followed
Reactions were monitored by HPLC, and the reduction rate was by a minimization using the conjugate gradient method (1000 iter-
determined by measuring NADH oxidation at 340 nm during the ations with a root mean square cutoff of 0.001 kcal/mol). Flexibility
reaction. The reaction was stopped with an equal volume of metha-
was allowed in the following active site R-groups: Ile26, Glu47,
nol. The apparent Km values for 17AAG and 17DMAG (0.004–0.8 Leu48, Ser50, Asn51, Ser52, Asp54, Ala55, Lys58, Ile96, Gly97,
mM) were determined using NADH 0.5 mM and rhNQO1 10 ␮g, Met98, Asp102, Leu103, Asn105, Asn106, Leu107, Gly108, Ile110,
incubated in 50 mM potassium phosphate buffer, pH 7.4, containing Lys112, Gly132, Gly135, Val136, Gly137, Phe138, Tyr139, Val150,
FAD (1 ␮M), 0.4% (v/v) Tween 80, and 0.4% (v/v) Triton X-100 at and Thr184.The interaction of the cis-amide isomers of the benzo-
3
0°C and by monitoring quinone-dependant oxidation of NADH. quinone and hydroquinone ansamycins with Hsp90 were further
Reactions were terminated by diluting the reaction 10 times with examined using the human Hsp90 structure cocrystallized with GM.
ice-cold methanol, and the resulting NADH concentrations were The cocrystallized GM structure was removed, and the cis-amide
determined by fluorescence spectroscopy (excitation, 340 nm; emis- isomers of the benzoquinone and hydroquinone ansamycins were in
sion, 460 nm). Km values were determined using Prism software turn docked into the nucleotide-binding domain of Hsp90 using the
(
GraphPad Software Inc., San Diego, CA).
Hsp90 ATPase Activity Assay. Inhibition of human Hsp90
same docking method for the open Hsp90 structure. The Poisson-
Boltzmann implicit solvent model used in NQO1 docking was used to
ATPase activity was measured as described previously (Rowlands et calculate the binding energy for each docked pose.
al., 2004; Guo et al., 2005). Briefly, 2.5 ␮g of purified human Hsp90
was incubated in 100 mM Tris-HCl, pH 7.4, containing 20 mM KCl,
6
mM MgCl , 200 ␮M NADH, benzoqinone ansamycin (2 and 4 ␮M)
2
Results
with or without rhNQO1 0.33 ␮g, and with or without 2 ␮M ES936.
Reactions (25 ␮l) were started by the addition of 1 mM ATP and
NQO1-Mediated Reduction of Benzoquinone Ansa-
allowed to proceed at 37°C for 12 h. Reactions were then diluted with mycins. The benzoquinone ansamycins GM, 17AAG,
2
25 ␮l of 100 mM Tris-HCl, pH 7.4, containing KCl 20 mM and 6 mM 17AG, 17DMAG, and 17AEP-GA were all reduced by puri-
MgCl2 mixed thoroughly, and 80 ␮l was transferred to a 96-well fied rhNQO1, using either NADH or NADPH as cofactors,
plate followed by 20 ␮l of malachite green reagent. After 10 min,
to the corresponding hydroquinone ansamycins without
further metabolism. The reduction of the benzoquinone
ansamycins by rhNQO1 was monitored spectrophotometri-
trisodium citrate (83 mM) was added to stabilize the color, and plates
were read at 650 nm.
Molecular Modeling. All simulations were performed using
cally, and the generation of the polar hydroquinone ansa-
Discovery Studio software (version 2.5; Accelrys, San Diego, CA).
The crystallographic coordinates of the 1.8-Å human NQO1 struc-
mycin was proportional to the loss of benzoquinone ansa-
mycin (see Supplemental Data). The hydroquinone
ansamycins were identified by LC/MS, and mass ions were
ture cocrystallized with ES936 (PDB 1KBQ), the 2.2Å open human
Hsp90 structure (PDB 1YES), and the 1.9-Å human Hsp90 struc-
ture cocrystallized with GM (PDB 1YET) (Stebbins et al., 1997; observed for 17-(allylamino)-17-demethoxygeldanamycin
ϩ
Winski et al., 2001) were obtained from the Research Collabora- hydroquinone (17AAGH ) at m/z 603.7 [MϩNH ] ,
2
4
tory for Structural Bioinformatics Protein Data Bank (http:// geldanamycin hydroquinone (GMH ) at m/z 580.5
2
ϩ
www.rcsb.org). The cis-amide isomers of the benzoquinone ansa-
[
MϩNH ] , 17-(amino)-17-demethoxygeldanamycin hy-
4
ϩ
droquinone (17AGH ) at m/z 548.2 [MϩH] , 17-deme-
thoxy-17-[[2-(dimethylamino)ethyl]amino]-geldanamycin
hydroquinone at m/z 619.3 [MϩH] , and 17-demethoxy-
7-[[2-(pyrrolidin-1-yl)ethyl]amino]-geldanamycin hydro-
quinone (17AEP-GAH ) at m/z 645.3 [MϩH] (see Supple-
mental Data). The relative rhNQO1-mediated reduction
mycins were constructed from GM isolated from the human
Hsp90-GM cocrystallized complex, the trans-amide isomers were
constructed from trans-17-azetidinyl-17-demethoxygeldanamycin
2
ϩ
(
Schnur and Corman, 1994) obtained from the Cambridge Struc-
1
tural Database (http://www.ccdc.cam.ac.uk), and the correspond-
ing hydroquinone ansamycins were generated by modifying the
quinone ring structure. The ionizable residues were corrected for
ϩ
2
physiological pH, and the potentials and charges of the complexes rate for this series of benzoquinone ansamycins was de-
were corrected using CHARMm in all simulations (Brooks et al., pendent on the C17 substituent and fastest for 17AAG
2
009).
followed by 17AG, GM, 17DMAG, and then 17AEP-GA
Fig. 1). The NQO1-dependent formation of the hydroqui-
NQO1 Docking. The ES936 molecules were removed from the
(
NQO1 structure, and the anionic form of reduced FAD was con-
structed from each FAD cofactor (Reigan et al., 2007). The binding
site was defined as whole residues within a 5-Å radius subset en-
compassing the FAD cofactor. The flexible docking program was used
to dock the trans-amide isomers of the benzoquinone ansamycins
into the NQO1 active site, followed by simulated annealing step to
optimize interactions (Koska et al., 2008). In addition to the benzo-
none ansamycin from the parent quinone was prevented by
ES936, a mechanism-based inhibitor of NQO1 (see Supple-
mental Data). Apparent Km values were determined for
the NQO1-mediated reduction of 17AAG (189 ␮M) and
1
7DMAG (262 ␮M) (see Supplemental Data).
Molecular Docking of the trans-Amide Isomers of
quinone ansamycin, flexibility was allowed in the following active the Benzoquinone Ansamycins into Human NQO1. Mo-
site R-groups: Leu103, Gln104, Trp105, Phe106, Thr148, Gly149, lecular docking simulations demonstrated that the trans-

8
26
Reigan et al.
amide isomers of this benzoquinone ansamycin series could benzoquinone and hydroquinone ansamycins was evaluated
be accommodated within the active site of NQO1. The re- using the malachite green phosphate assay (Rowlands et al.,
duced anionic form of the FAD cofactor was used in these 2004). The inhibition of human Hsp90 ATPase activity was
docking studies because this represents the NQO1 enzyme at assessed with only 17AAG and 17AAGH (Guo et al., 2005).
2
the beginning of the catalytic cycle, when the enzyme can The benzoquinone ansamycins did inhibit the ATPase activ-
accept the quinone substrate before reduction to the hydro- ity of human Hsp90 at the concentrations investigated; how-
quinone (Li et al., 1995; Cavelier and Amzel, 2001). The ever, inhibition significantly increased when the benzoqui-
flexible docking program generated 50 structural complexes none ansamycins were incubated with rhNQO1, and this
for 17AAG, 41 for 17AG, 21 for GM, 15 for 17DMAG, and 11 increased inhibition was abrogated by pretreatment with
for 17AEP-GA. The resultant complexes were evaluated ES936 (Fig. 3). These results clearly demonstrate that the
based on the calculated binding energy, the number of hy- hydroquinone ansamycins generated by NQO1 were more
drogen bond interactions between NQO1 and the benzoqui- potent inhibitors of Hsp90 ATPase activity than the corre-
none ansamycin, and the position of the quinone ring in the sponding benzoquinone ansamycins, in agreement with our
active site of NQO1. A recurring high-ranking binding con- previous data using yeast Hsp90 (Guo et al., 2005, 2006).
formation was identified for each benzoquinone ansamycin in
Molecular Docking of the trans- and cis-Amide Iso-
this series, in which the quinone moiety was parallel to the mers of the Benzoquinone and Hydroquinone Ansamy-
isoalloxazine ring of the flavin cofactor, optimizing ␲–␲ in- cins into Human Hsp90. Previous studies had reported
teractions and increasing the van der Waals contribution to that it was not possible to dock the trans-amide isomers of
the binding energy. In this conformation, the C21 carbonyl the benzoquinone ansamycins into the Hsp90 structure; how-
of the benzoquinone ansamycin displayed a hydrogen bond ever, these studies used the human Hsp90 structure cocrys-
with Tyr128, the C17 substituent occupied the cleft formed tallized with GM or 17DMAG (Jez et al., 2003; Lee et al.,
by Tyr155, His161, His194, Phe232, and Phe236, and the 2004). In this study, the trans- and cis-amide isomers of the
majority of the ansa ring was outside of the active site (Fig. 2). benzoquinone and hydroquinone ansamycins were success-
In this series of benzoquinone ansamycins, 17AAG had the fully docked into the nucleotide-binding domain of the open
most favorable binding energy followed by 17AG, GM, Hsp90 structure (Stebbins et al., 1997). The calculated bind-
17DMAG, and then 17AEP-GA, for this conformation (Table 1). ing energies values indicate greater stability between the
In addition, the binding energies obtained for the highest hydroquinone ansamycin and Hsp90 than the parent qui-
ranked pose for each benzoquinone ansamycin correlates with none. Similar high-scoring conformations were observed
the distance between the C21 carbonyl of the quinione and across the benzoquinone and hydroquinone ansamycin series
Tyr128, which would suggest that the benzoquinone ansamy- for the trans- and cis-amide isomers, all with favorable dock
cins with small C17 substituents are accommodated by NQO1 scores and binding energies for binding to the Hsp90 protein
more readily that benzoquinone ansamycins with large C17 (Table 2 data shown for 17AAG; for other compounds, see
substituents. The docking of the cis-amide isomers of the ben- Supplemental Data). The overall conformation of the qui-
zoquinone ansamycins was attempted, but this was unsuccess- none or hydroquinone and the ansa ring were similar for
ful because of steric conflicts with active site residues.
the trans-amide isomers of the benzoquinone and hydro-
Inhibition of Human Hsp90 ATPase Activity. The in- quinone ansamycins in the nucleotide-binding domain of
hibition of human Hsp90 ATPase activity by this series of the open human Hsp90 structure (Fig. 4, A and B, repre-
sentations shown for 17AAG; for other compounds, see
Supplemental Data). However, differences were observed
in the orientation of the C17 substituent outside of the
binding pocket, with the exception of 17AG and 17AGH2,
the C17 substituents of the benzoquinone ansamycins
were almost perpendicular to the plane of the quinone, and
in the case of the hydroquinone ansamycins, they extended
in the plane of the hydroquinone. There were few hydrogen
bond interactions between the trans-amide isomers and
the Hsp90 protein, the most notable being the interactions
between the C18 carbonyl of 17AAG and 17AG with Ser113
and the C18 hydroxyl of 17AGH2 and 17AEP-GAH2 with
Gly135. Interestingly, although no hydrogen bond interac-
tions were observed between Lys112 and the hydroquinonϩe
ansamycins, with the exception of 17AEP-GAH , the NH
2
3
group of Lys112 was directed toward the hydroquinone
ring. This was in contrast to the benzoquinone ansamy-
ϩ
cins, in which the NH3 group of Lys112 was directed
away from the quinone ring.
The conformations of the cis-amide isomers of the benzo-
quinone and hydroquinone ansamycins were similar in the
Fig. 1. Relative NQO1-mediated reduction rate of the benzoquinone nucleotide-binding site of the open Hsp90 structure, and with
ansamycins. Columns, mean (n ϭ 3); bars, S.D. GM, 17AG, 17DMAG, and
the exception of GM and GMH , the C17 substituents of the
benzoquinone and hydroquinone ansamycins adopted a sim-
ilar arrangement outside of the binding pocket. A greater
2
1
0
7AEP-GA were significantly different from 17AAG: ء, p Ͻ 0.01, ءء, p Ͻ
.001. Statistical analysis was performed using one-way analysis of vari-
ance with Tukey multiple comparison test.

Hsp90 Inhibition by Benzoquinone and Hydroquinone Ansamycins
827
number of hydrogen bond interactions was observed between Hsp90 structure. The docking of the trans- and cis-amide
the hydroquinone ansamycins with a C17 amino substituent hydroquinone ansamycins in this series into the open Hsp90
and the open Hsp90 structure compared with the correspond- structure resulted in more favorable dock scores and calcu-
ing benzoquinone ansamycins (Fig. 4, C and D, representa- lated binding energy values than the corresponding benzo-
tions shown for 17AAG; for other compounds, see Supple- quinone ansamycins, and this was largely due to the in-
mental Data). The hydroquinone ansamycins displayed creased contribution of electrostatic energy (Table 2 data
additional hydrogen bond interactions between the C18 hy- shown for 17AAG; for other compounds, see Supplemental
droxyl and Asp54, and this hydrogen bond was observed with Data).
GMH ; however, no hydrogen bond interactions were ob-
The docking of the benzoquinone and hydroquinone ansa-
2
served between the C17 substituent of GM or GMH and the mycins into the GM bound form of human Hsp90 (Stebbins et
2
open Hsp90 structure. GM was the only benzoquinone ansa- al., 1997) resulted in conformations and interactions similar
mycin to display an additional hydrogen bond between the to those reported in the cocrystallized structures (Stebbins et
C1 amino and Gly135, resulting in GM and GMH having an al., 1997; Jez et al., 2003). In these studies, only the cis-amide
2
equal number of hydrogen bond interactions with the open isomers of the benzoquinone and hydroquinone ansamycins
Fig. 2. Flat-ribbon representation of hu-
man NQO1 in complex with the trans-
amide isomer of 17AAG (A), GM (B),
1
7AG (C), 17DMAG (D), 17AEP-GA (E)
(
stick display style; colored by atom type,
carbon atoms colored yellow) in the re-
duced anionic FAD (stick display style;
colored by atom type, carbon atoms col-
ored orange) site, key amino acid residues
(
stick display style; colored by atom type)
and hydrogen bond interactions (black
dashed lines) are displayed.
TABLE 1
Interaction and associated energies of the trans-amide isomers of the benzoquinone ansamycins in complex with NQO1
H-Bond Interaction
Compound
⌬Vvdw
⌬Velect
⌬Gbind
H-Bond Distance
Amino Acid
Ligand
kcal/mol
Å
1
7AAG
Ϫ11.5
Ϫ8.1
Ϫ13.6
Ϫ12.0
Ϫ5.7
Ϫ56.2
Ϫ68.0
Ϫ56.1
Ϫ52.7
Ϫ39.4
Ϫ18.3
Ϫ15.5
Ϫ16.1
Ϫ12.7
Ϫ8.7
Tyr128
Tyr128
Tyr128
Tyr128
Tyr128
C21 ϭ O
C21 ϭ O
C21 ϭ O
C21 ϭ O
C21 ϭ O
2.48
2.50
2.52
2.54
2.57
GM
1
1
1
7AG
7DMAG
7AEP-GA
⌬Vvdw, van der Waals energy; ⌬Velect, electrostatic energy; ⌬Gbind, binding energy.

8
28
Reigan et al.
were used, because the trans-amide isomers could not be a more compact C-clamp conformation around helix 2, and
positioned into the nucleotide-binding domain without steric these additional interactions translate into more favorable
conflicts with active site residues. Consistent with previous calculated binding energies than those of the parent qui-
studies, using the yeast Hsp90 crystal structure (Guo et al., nones (Table 3 data shown for 17AAG; for other compounds, see
2
005, 2006), the hydroquinone ansamycins displayed addi- Supplemental Data), providing a structural explanation as to why
tional hydrogen bond interactions with the human Hsp90 the hydroquinone ansamycins are more potent Hsp90 inhibitors.
protein compared with the parent quinone (Fig. 4, E and F,
representations shown for 17AAG; for other compounds, see
Supplemental Data). The C18 hydroxyl interacts with the
Discussion
carboxylate side chain of Asp40 and allows for further inter-
The amide bond in the ansa ring of the benzoquinone and
action between the C17 substituent and the Asp54 and hydroquinone ansamycins restricts rotation and, conse-
Lys58. These additional hydrogen bond interactions between quently, only trans- and cis-amide conformations are possi-
the hydroquinone ansamycin and the Hsp90 protein result in ble. The benzoquinone ansamycins adopt the trans-amide
Fig. 3. Inhibition of human Hsp90 ATPase activity by 17AAG and 17AAGH . Human Hsp90 ATPase activity was measured in reactions with either
2
vehicle [dimethyl sulfoxide (DMSO)] or 17AAG (A), GM (B), 17AG (C), 17DMAG (D), or 17AEP-GA (E) (2 and 4 ␮M) in the presence and absence of
rhNQO1. The reactions were analyzed after 12 h, and phosphate concentrations were measured using the malachite green assay. ꢀ, benzoquinone
ansamycin and NADH; f, benzoquinone ansamycin, NADH, and rhNQO1; p, benzoquinone ansamycin, NADH, NQO1, and ES936. Columns, mean
(
n ϭ 3); bars, S.D. Hsp90 ATPase activity in incubates containing benzoquinone ansamycin, NADH, and NQO1 were statistically different from
incubates containing benzoquinone ansamycin and NADH or benzoquinone ansamycin, NADH, NQO1, and ES936: ء, p Ͻ 0.05, one-way analysis of
variance with Tukey for pairwise comparison. Data for 17AAG has been presented previously (Guo et al., 2005); the experiment was repeated, and the
data are shown here to complete the comparison of all compounds in this structural series.
TABLE 2
Interactions and associated energies for the trans-/cis-amide isomers of 17AAG and 17AAGH2 in complex with the open human Hsp90 structure
H-Bond Interaction
Compound
Dock Score
⌬Vvdw
⌬Velect
⌬Gbind
H-Bond Distance
Amino Acid
Ligand
kcal/mol
Å
trans-17AAG
trans-17AAGH2
cis-17AAG
60.4
86.2
56.7
Ϫ10.5
Ϫ13.8
Ϫ15.3
Ϫ58.1
Ϫ75.6
Ϫ57.5
Ϫ16.2
Ϫ24.6
Ϫ26.9
Ser113
C18ϭO
1.76
Phe138
Asp54
Lys112
Asp93
Phe138
Asp54
Asp54
Lys112
Asp93
C1ϭO
C17-NHR
C21ϭO
C24-NH2
C1ϭO
C17-NHR
C18-OH
C21-OH
C24-NH2
2.26
2.23
1.80
1.91
2.14
2.08
1.60
1.96
2.06
1
1
cis-17AAGH2
89.7
Ϫ12.9
Ϫ65.8
Ϫ34.9
⌬Vvdw, van der Waals energy; ⌬Velect, electrostatic energy; ⌬Gbind, binding energy.

Hsp90 Inhibition by Benzoquinone and Hydroquinone Ansamycins
829
TABLE 3
Interactions and associated energies for the cis-amide isomers of 17AAG and 17AAGH2 in complex with the bound human Hsp90 structure
H-Bond Interaction
Compound
Dock Score
⌬Vvdw
⌬Velect
⌬Gbind
H-Bond Distance
Amino Acid
Ligand
kcal/mol
Å
cis-17AAG
61.4
Ϫ12.2
Ϫ50.5
Ϫ13.4
Phe138
Lys58
Asp54
Lys112
Asp93
Phe138
Lys58
Asp54
Asp54
Lys112
Asp93
C1ϭO
2.04
1.88
2.37
1.71
1.87
1.95
1.92
2.18
1.65
2.01
1.85
C11-OH
C17-NHR
C21ϭO
1
1
C24-NH2
C1ϭO
C11-OH
C17-NHR
C18-OH
C21-OH
C24-NH2
cis-17AAGH2
88.3
Ϫ13.0
Ϫ64.9
Ϫ16.9
⌬Vvdw, van der Waals energy; ⌬Velect, electrostatic energy; ⌬Gbind, binding energy.
conformation in solution, because this is energetically fa- al., 2009) and our molecular docking studies in which the
vored (Schnur and Corman, 1994; Jez et al., 2003), and co- cis-amide isomer would not dock into NQO1. However, the
crystallization studies have clearly shown the cis-amide iso- results of a binding affinity study using BODIPY-geldana-
mer is bound to Hsp90 (Stebbins et al., 1997; Jez et al., 2003), mycin in the absence and presence of reducing agents re-
indicating that trans-cis isomerization of the benzoquinone ported by Onuoha et al. may, in fact, support Hsp90-medi-
ansamycins must occur during the binding process. However, ated isomerization. The binding affinity was time-dependent
a recent study concluded that no isomerization step occurs in in the absence of reducing agents, indicating that the benzo-
the binding of the benzoquinone ansamycins to Hsp90 (On- quinone ansamycins weakly bind Hsp90 initially and then
uoha et al., 2007) and proposed that small quantities of the undergo a structural change, in agreement with a prior study
cis-amide isomer in solution, identified in an NMR study (Gooljarsingh et al., 2006). The non–time-dependent in-
(
Thepchatri et al., 2007), may bind directly to Hsp90, and in creased binding affinity in the presence of reducing agents
vivo potency could be due to the direct binding of the hydro- may be due to the formation of the hydroquinone and the
quinone of the cis-amide isomer. This counters the results of cis-amide isomer, because reducing agents were used to cat-
a study examining the effects of conformationally con- alyze tautomerization (Clarke et al., 1984; Schmid, 1993).
strained cis-amide chimeric inhibitors of Hsp90 (Duerfeldt et The Hsp90-mediated isomerization of the benzoquinone
Fig. 4. The human Hsp90 –17AAG/
1
7AAGH2 complex. The nucleotide binding
domain of open human Hsp90 in complex
with the trans-amide isomer of 17AAG (A)
and 17AAGH (B) and the cis-amide isomer
2
of 17AAG (C) and 17AAGH (D). The nucle-
2
otide binding domain of bound human Hsp90
in complex with the cis-amide isomer of
1
7AAG (E) and 17AAGH2 (F). 17AAG and
1
7AAGH2 (stick display style; carbon atoms
are colored yellow) in the nucleotide-binding
domain of Hsp90, key amino acid residues
(
stick display style; colored by atom type), and
hydrogen bond interactions (black dashed
lines) are displayed.

8
30
Reigan et al.
Scheme 2. The proposed mechanism of
trans-cis isomerization of the benzoqui-
none ansamycins (A) and the hydroqui-
none ansamycins (B). Arrows with broken
lines indicate bond rotation.
ansamycins has also been supported by quantum and molec- teraction with Ser113. The deprotonation of the amide nitro-
ular mechanic studies (Jez et al., 2003; Lee et al., 2004) and gen of the benzoquinone ansamycins may allow the C1 amide
the identification of nonprolyl peptide cis-trans isomerases or the C18 carbonyl to abstract a proton from Ser113, pro-
(
Bouckaert et al., 2000; Schiene-Fischer et al., 2002). Lee et moting tautomerization of the amide via a proton shuttle
al. (2004) proposed that Lys112 and Ser113 of Hsp90 interact between the quinone and the amide (Scheme 2A). In response
with the C1 amide of the benzoquinone ansamycin promoting to the change in the electronic properties of the quinone ring,
isomerization via tautomerization, and based on the loss of Lys112 may then move to protonate the nitrogen of the
GM binding to a Ser113 Hsp90 point mutant, they deter- imine. The destabilization of the hydroquinone ring is not re-
mined that Ser113 was essential for Hsp90-mediated trans- quired for the isomerization of the hydroquinone ansamycins
cis isomerization. However, the mutation of Ser113 may in (Scheme 2B), because Lys112 is in close proximity to the nitro-
turn affect the conformation of Lys112 in the nucleotide- gen of the C1 amide to donate a proton if required (Fig. 5). The
binding site. Although the initial benzoquinone ansamycin interaction between the Lys112 residue and the lone pair of
binding pose described by Lee et al. was not identified in our electrons of the amide nitrogen acts to deconjugate the reso-
docking studies, and no interactions were observed between nance of the amide bond and promote trans-cis isomerization.
the C1 amide of the benzoquinone or hydroquinone ansamy-
This proposed acid catalyzed trans-cis isomerization mech-
cins and Lys112 or Ser113 residues of Hsp90, the results of our anism is similar to that described for the peptidyl-prolyl
docking studies provide support for the model of Hsp90-mediated isomerases (PPIases). Although the precise mechanism of
trans-cis isomerization through promoting tautomerization and peptidyl-prolyl cis-trans isomerization catalyzed by PPIases
suggest that Lys112 is an important mediator in this process.
remains to be fully elucidated, it is widely accepted that the
The benzoquinone ansamycin series examined in our study protonation of the imine nitrogen is the fundamental step in
were metabolized by rhNQO1 to generate the corresponding enzyme-catalyzed isomerization (Schmid, 1993; Hur and
hydroquinone ansamycins, and these were more potent in- Bruice, 2002). Cis-trans isomerization of prolyl peptides by
hibitors of purified human Hsp90 than their parent qui- the cyclophilin PPIases involves the stabilization of the
nones. The potentiated inhibition of Hsp90 by the hydroqui- amide oxygen by an asparagine residue, the proline ring is
none ansamycins was rationalized by the increased hydrogen held in position by the phenyl group of a phenylalanine
bond interactions between the hydroquinone ansamycins and residue, and the guanidino group of a conserved arginine
the Hsp90 protein using computational-based modeling. The residue weakens the double bond character of the C-N bond
interaction of the trans- and cis-amide isomers of the benzo- (Hur and Bruice, 2002). Similar interactions were observed
quinone and hydroquinone ansamycins in the open human between the trans-amide isomers of the benzoquinone and
Hsp90 structure provide support for Hsp90-mediated isomer-
ization. In all cases, the hydroquinone ansamycins had more
favorable binding energies than the parent quinones, indi-
cating that both the trans- and cis-amide isomers of the
hydroquinone ansamycins would exhibit increased binding
affinity for Hsp90. Modeling studies support that the trans-
amide isomer of the benzoquinone and hydroquinone ansa-
mycins bind to Hsp90, and this would be a critical initial step
for Hsp90-mediated isomerization. This could also rational-
ize the results of mutational studies (Lee et al., 2004), be-
cause mutating the Lys112 and/or the Ser113 residues could
impede Hsp90-mediated isomerization, but it would not ob-
struct the binding of the trans-amide isomer. The high-scor-
ing complexes of the trans-amide isomers in the open human
Hsp90 structure, identified in our study, may represent the
initial binding conformation of the benzoquinone and hydro-
quinone ansamycins. The C18 carbonyl of the trans-amide
isomer of 17AAG and 17AG displayed a hydrogen bond in- Lys112 in the nucleotide-binding domain of human Hsp90.
Fig. 5. The interaction of trans-17AAG (A) and trans-17AAGH2 (B) with

Hsp90 Inhibition by Benzoquinone and Hydroquinone Ansamycins
831
hydroquinone ansamycins and the open Hsp90 structure, References
with Ser113 and Phe138 stabilizing benzoquinone and hy-
droquinone ansamycin binding, and the NH3 group of
Beall HD, Mulcahy RT, Siegel D, Traver RD, Gibson NW, and Ross D (1994)
Metabolism of bioreductive antitumor compounds by purified rat and human
DT-diaphorases. Cancer Res 54:3196–3201.
ϩ
Lys112 is available to protonate the imine nitrogen promot-
ing the keto tautomer. The intrinsic structural properties of
the ansamycin ring, the anchoring of the carbamate through
interaction with Asp90 at the base of the nucleotide-binding
domain, and the interaction of the ansa ring with amino acid
residues via through-bond interactions with the specific sol-
vent network in the nucleotide-binding domain may also aid
to drive trans-cis isomerization (Pappenberger et al., 2001).
The confirmation of Hsp90-mediated trans-cis isomerization
of the benzoquinone and hydroquinone ansamycins is impor-
tant because it could be exploited in the development of new
Hsp90 inhibitors, and the effect of conformationally restrained
cis-amide isomers of benzoquinone ansamycin analogs on
Hsp90 inhibition has been reported (Duerfeldt et al., 2009).
The rate of NQO1-mediated reduction of this benzoquinone
ansamycin series was dependent on the C17 substituent and
was fastest for 17AAG because of the small size of the C17
substituent and the additional interactions between the allyl
group and active site residues. The allyl substituent of
Bouckaert J, Dewallef Y, Poortmans F, Wyns L, and Loris R (2000) The structural
features of concanavalin A governing non-proline peptide isomerization. J Biol
Chem 275:19778–19787.
Brooks BR, Brooks CL, 3rd, Mackerell AD, Jr., Nilsson L, Petrella RJ, Roux B, Won
Y, Archontis G, Bartels C, Boresch S, Caflisch A, Caves L, Cui Q, Dinner AR, Feig
M, Fischer S, Gao J, Hodoscek M, Im W, Kuczera K, Lazaridis T, Ma J, Ovchin-
nikov V, Paci E, Pastor RW, Post CB, Pu JZ, Schaefer M, Tidor B, Venable RM,
Woodcock HL, Wu X, Yang W, York DM, and Karplus M (2009) CHARMM: the
biomolecular simulation program. J Comput Chem 30:1545–1614.
Cavelier G and Amzel LM (2001) Mechanism of NAD(P)H:quinone reductase: Ab
initio studies of reduced flavin. Proteins 43:420–432.
Chiosis G, Huezo H, Rosen N, Mimnaugh E, Whitesell L, and Neckers L (2003)
17AAG: low target binding affinity and potent cell activity–finding an explanation.
Mol Cancer Ther 2:123–129.
Chiosis G and Neckers L (2006) Tumor selectivity of Hsp90 inhibitors: the explana-
tion remains elusive. ACS Chem Biol 1:279–284.
Clarke ED, Wardman P, and Wilson I (1984) Cis-trans isomerization of the (5-nitro-
2
-furyl)acrylamide, AF-2, initiated by ascorbate, glutathione, Fe(II) and OH-.
Biochem Pharmacol 33:83–87.
Cysyk RL, Parker RJ, Barchi JJ, Jr., Steeg PS, Hartman NR, and Strong JM (2006)
Reaction of geldanamycin and C17-substituted analogues with glutathione: prod-
uct identifications and pharmacological implications. Chem Res Toxicol 19:376–
381.
Duerfeldt AS, Brandt GE, and Blagg BS (2009) Design, synthesis, and biological
evaluation of conformationally constrained cis-amide Hsp90 inhibitors. Org Lett
11:2353–2356.
Egorin MJ, Rosen DM, Wolff JH, Callery PS, Musser SM, and Eiseman JL (1998)
Metabolism of 17-(allylamino)-17-demethoxygeldanamycin (NSC 330507) by mu-
rine and human hepatic preparations. Cancer Res 58:2385–2396.
Feig M, Onufriev A, Lee MS, Im W, Case DA, and Brooks CL 3rd (2004) Performance
comparison of generalized born and Poisson methods in the calculation of electro-
static solvation energies for protein structures. J Comput Chem 25:265–284.
Gooljarsingh LT, Fernandes C, Yan K, Zhang H, Grooms M, Johanson K, Sinnamon
RH, Kirkpatrick RB, Kerrigan J, Lewis T, et al. (2006) A biochemical rationale for
the anticancer effects of Hsp90 inhibitors: slow, tight binding inhibition by
geldanamycin and its analogues. Proc Natl Acad Sci USA 103:7625–7630.
Guo W, Reigan P, Siegel D, and Ross D (2008) Enzymatic reduction and glutathione
conjugation of benzoquinone ansamycin heat shock protein 90 inhibitors: rele-
vance for toxicity and mechanism of action. Drug Metab Dispos 36:2050–2057.
Guo W, Reigan P, Siegel D, Zirrolli J, Gustafson D, and Ross D (2005) Formation of
1
7AAG is almost perpendicular to the plane of the phenoxy of
Tyr155, further increasing the van der Waals contribution to
the calculated binding energy through ␲–␲ interactions. The
C17 amino and methoxy substituents of 17AG and GM are
small enough to be accommodated by the NQO1 active site;
however, the large 2-(dimethylamino)ethyl and 2-(pyrrolidi-
nyl-1-yl)ethyl substituents of 17DMAG and 17AEP-GA are
directed away from His161 and Tyr155, and their steric bulk
would restrict the quinone from entering the active site, and
this is reflected in the relative reduction rate. The difference
between the relative rates of NQO1-mediated reduction of
1
7-allylamino-demethoxygeldanamycin (17-AAG) hydroquinone by NAD(P)H:
quinone oxidoreductase 1: role of 17-AAG hydroquinone in heat shock protein 90
inhibition. Cancer Res 65:10006–10015.
Guo W, Reigan P, Siegel D, Zirrolli J, Gustafson D, and Ross D (2006) The biore-
duction of a series of benzoquinone ansamycins by NAD(P)H:quinone oxidoreduc-
tase 1 to more potent heat shock protein 90 inhibitors, the hydroquinone ansamy-
cins. Mol Pharmacol 70:1194–1203.
Hur S and Bruice TC (2002) The mechanism of cis-trans isomerization of prolyl
peptides by cyclophilin. J Am Chem Soc 124:7303–7313.
1
7AAG and 17DMAG was approximately 3-fold. Interest-
ingly, the ranking order of the calculated binding energy
associated with the benzoquinone ansamycin binding confor-
mation was also in agreement with the order of the relative
reduction rate using purified rhNQO1.
Jez JM, Chen JC, Rastelli G, Stroud RM, and Santi DV (2003) Crystal structure and
molecular modeling of 17-DMAG in complex with human Hsp90. Chem Biol
1
0:361–368.
Kamal A, Thao L, Sensintaffar J, Zhang L, Boehm MF, Fritz LC, and Burrows FJ
2003) A high-affinity conformation of Hsp90 confers tumour selectivity on Hsp90
The selective generation of the hydroquinone ansamycins
within the cancer cell is desirable, because this will result in
potentiated Hsp90 inhibition, and increased cell retention and
accumulation compared with the parent quinone. The proper-
ties of the C17 substituent and the metabolism of the benzoqui-
none ansamycin, by NQO1 and other oxidoreductases, should
be considered in the development of further benzoquinone ansa-
mycins. The rational progression of this approach would be to
develop a stable, water-soluble prodrug that protects the hydro-
quinone group and is activated via intracellular enzymatic hy-
drolysis, generating a relatively stable hydroquinone ansamy-
cin and potent Hsp90 inhibitor, which in its quinone form would
be an efficient substrate for NQO1.
(
inhibitors. Nature 425:407–410.
Kelland LR, Sharp SY, Rogers PM, Myers TG, and Workman P (1999) DT-
Diaphorase expression and tumor cell sensitivity to 17-allylamino, 17-
demethoxygeldanamycin, an inhibitor of heat shock protein 90. J Natl Cancer Inst
91:1940–1949.
Koska J, Spassov VZ, Maynard AJ, Yan L, Austin N, Flook PK, and Venkatachalam
CM (2008) Fully automated molecular mechanics based induced fit protein-ligand
docking method. J Chem Inf Model 48:1965–1973.
Lang W, Caldwell GW, Li J, Leo GC, Jones WJ, and Masucci JA (2007) Biotrans-
formation of geldanamycin and 17-allylamino-17-demethoxygeldanamycin by hu-
man liver microsomes: reductive versus oxidative metabolism and implications.
Drug Metab Dispos 35:21–29.
Lee YS, Marcu MG, and Neckers L (2004) Quantum chemical calculations and
mutational analysis suggest heat shock protein 90 catalyzes trans-cis isomeriza-
tion of geldanamycin. Chem Biol 11:991–998.
Li R, Bianchet MA, Talalay P, and Amzel LM (1995) The three-dimensional struc-
ture of NAD(P)H:quinone reductase, a flavoprotein involved in cancer chemopro-
tection and chemotherapy: mechanism of the two-electron reduction. Proc Natl
Acad Sci USA 92:8846–8850.
Acknowledgments
Maloney A and Workman P (2002) HSP90 as a new therapeutic target for cancer
therapy: the story unfolds. Expert Opin Biol Ther 2:3–24.
Onuoha SC, Mukund SR, Coulstock ET, Sengerov a` B, Shaw J, McLaughlin SH, and
Jackson SE (2007) Mechanistic studies on Hsp90 inhibition by ansamycin deriv-
atives. J Mol Biol 372:287–297.
Pappenberger G, Ayg u¨ n H, Engels JW, Reimer U, Fischer G, and Kiefhaber T (2001)
Nonprolyl cis peptide bonds in unfolded proteins cause complex folding kinetics.
Nat Struct Biol 8:452–458.
Pearl LH and Prodromou C (2000) Structure and in vivo function of Hsp90. Curr
Opin Struct Biol 10:46–51.
We acknowledge Dr. Daniel L. Gustafson and Joeseph Zirolli for
their assistance with mass spectral determinations.
Authorship Contributions
Participated in research design: Reigan and Ross.
Conducted experiments: Reigan, Siegel, and Guo.
Performed data analysis: Reigan and Siegel.
Wrote or contributed to the writing of the manuscript: Reigan and
Ross.
Pearl LH and Prodromou C (2006) Structure and mechanism of the Hsp90 molecular
chaperone machinery. Annu Rev Biochem 75:271–294.
Reigan P, Colucci MA, Siegel D, Chilloux A, Moody CJ, and Ross D (2007) Develop-

8
32
Reigan et al.
ment of indolequinone mechanism-based inhibitors of NAD(P)H:quinone oxi-
doreductase 1 (NQO1): NQO1 inhibition and growth inhibitory activity in human
pancreatic MIA PaCa-2 cancer cells. Biochemistry 46:5941–5950.
Roe SM, Prodromou C, O’Brien R, Ladbury JE, Piper PW, and Pearl LH (1999)
Structural basis for inhibition of the Hsp90 molecular chaperone by the antitumor
antibiotics radicicol and geldanamycin. J Med Chem 42:260–266.
Rooseboom M, Commandeur JN, and Vermeulen NP (2004) Enzyme-catalyzed acti-
vation of anticancer prodrugs. Pharmacol Rev 56:53–102.
ansamycin-stereospecific conversion of an (S)-allylic alcohol to an (S)-allylic amine
derivative. J Org Chem 59:2581–2584.
Siegel D and Ross D (2000) Immunodetection of NAD(P)H:quinone oxidoreductase 1
(NQO1) in human tissues. Free Radic Biol Med 29:246–253.
Stebbins CE, Russo AA, Schneider C, Rosen N, Hartl FU, and Pavletich NP (1997)
Crystal structure of an Hsp90-geldanamycin complex: targeting of a protein chap-
erone by an antitumor agent. Cell 89:239–250.
Thepchatri P, Eliseo T, Cicero DO, Myles D, and Snyder JP (2007) 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.
Winski SL, Faig M, Bianchet MA, Siegel D, Swann E, Fung K, Duncan MW, Moody
CJ, Amzel LM, and Ross D (2001) Characterization of a mechanism-based inhib-
itor of NAD(P)H:quinone oxidoreductase 1 by biochemical, X-ray crystallographic,
and mass spectrometric approaches. Biochemistry 40:15135–15142.
Ross D (2004) Quinone reductases multitasking in the metabolic world. Drug Metab
Rev 36:639–654.
Rowlands MG, Newbatt YM, Prodromou C, Pearl LH, Workman P, and Aherne W
(
2004) High-throughput screening assay for inhibitors of heat-shock protein 90
ATPase activity. Anal Biochem 327:176–183.
Schiene-Fischer C, Habazettl J, Schmid FX, and Fischer G (2002) The hsp70 chap-
erone DnaK is a secondary amide peptide bond cis-trans isomerase. Nat Struct
Biol 9:419–424.
Schmid FX (1993) Prolyl isomerase: enzymatic catalysis of slow protein-folding
reactions. Annu Rev Biophys Biomol Struct 22:123–142.
Schnur R and Corman ML (1994) Tandem [3,3]-sigmatropic rearrangements in an
Address correspondence to: Dr. David Ross, Department of Pharmaceutical
Sciences, C238-P15, Research 2, 12700 East 19th Avenue, Room 3100, Aurora,
CO 80045. E-mail: david.ross@ucdenver.edu