Asymmetric synthesis of pochonin e and F, revision of their proposed structure, and their conversion to potent Hsp90 inhibitors

Article abstract of DOI:10.1002/chem.201200546

A concise and modular synthesis of pochonin E and F, and their epimers at C-6 established the correct stereochemistry of these two natural products. Several members of the pochonin family have been shown to bind the heat shock protein 90 (Hsp90), which has been the focus of intense drug discovery efforts. Pochonin E and F as well as their epimers were derivatized into the corresponding pochoximes and further modified at the C-6 position. Molecular dynamics simulations, docking studies, and Hsp90 affinity measurements were performed to evaluate the impact of these modifications.

Full text of DOI:10.1002/chem.201200546

FULL PAPER
DOI: 10.1002/chem.201200546
Asymmetric Synthesis of Pochonin E and F, Revision of Their Proposed
Structure, and Their Conversion to Potent Hsp90 Inhibitors
[
a]
[a]
[a]
[b]
Ganesan Karthikeyan, Claudio Zambaldo, Sofia Barluenga, Vincent Zoete,
[
c]
[a]
Martin Karplus,* and Nicolas Winssinger*
Abstract: A concise and modular syn-
thesis of pochonin E and F, and their
epimers at C-6 established the correct
stereochemistry of these two natural
products. Several members of the po-
chonin family have been shown to bind
the heat shock protein 90 (Hsp90),
which has been the focus of intense
drug discovery efforts. Pochonin E and
F as well as their epimers were derivat-
ized into the corresponding pochox-
imes and further modified at the C-6
position. Molecular dynamics simula-
tions, docking studies, and Hsp90 affin-
ity measurements were performed to
evaluate the impact of these modifica-
tions.
Keywords: conformation analysis ·
Hsp90
· inhibitors · lactones ·
pochonins
[
7]
Introduction
mation by total synthesis and the tentative assignment of
[8]
pochonin J needs to be revised based on the recent synthe-
sis of the proposed structure. The stereochemistry of the
C-6 hydroxyl group of pochonin E and F were left unas-
[9]
The resorcyclic acid lactones (RAL) have been known for
over half a century with the first isolation of radicicol re-
ported in 1953 from researchers in Belgium (Scheme 1).
[1]
[6]
signed in the original isolation report. Both compounds
Until the 1990s, the most prominent member of this family
was zearalenone, which is known to have estrogen agonistic
properties. The more recent discovery that several members
of this family are potent inhibitors of ATPases and kinases
has revived interest in the resorcylides. RALs are secon-
dary metabolites produced by a modular polyketide syn-
thase (Scheme 1) and a number of variations in the oxida-
tion state of the macrocycle have already been isolated from
various environments ranging from the Sonoran desert in
were re-isolated by Shinonaga and coworkers and the ste-
reochemistry of the C-6 hydroxyl group was tentatively as-
signed as b (S stereochemistry) based on NOESY experi-
[10]
ments. A major interest in the pochonins stems from the
demonstration that pochonin A and D bind to the heat
shock protein 90 (Hsp90) in an analogous fashion to radici-
[2]
[11]
col. Hsp90 has been the focus of intense drug discovery
efforts over the past decade with several small molecules
[12]
progressing through clinical development.
Hsp90 is an
[
3]
[4]
Arizona to mangroves in Thailand. New members of this
ATP-dependent chaperone, which is involved in the post-
translational maturation and stabilization of over one hun-
dred proteins (“its clients”). In the absence of Hsp90s chap-
eroning, its clients are misfolded and degraded through an
[5]
family continue to be isolated. However, relative stereo-
chemical assignments can be challenging in flexible macro-
cycles. For example, the stereochemistry of the C-5 chlorine
[6]
[13]
in pochonin C had not been firmly established until confir-
ubiquitin-proteasome pathway. Its activity has been impli-
[
14]
cated in diverse pathologies ranging from oncology
to
[15]
[16]
neurodegenerative
and infectious diseases.
Radicicol
provided an important lead in the drug discovery efforts to-
wards a clinical candidate against Hsp90 with two pharma-
cophores in clinical developments bearing the resorcinol
[
a] Dr. G. Karthikeyan, C. Zambaldo, Dr. S. Barluenga,
Prof. N. Winssinger
Institut de Science et Ingꢀnierie Supramolꢀculaires (ISIS-UMR 7006)
Universitꢀ de Strasbourg CNRS
[17]
[18]
moiety of radicicol (NVP-AUY922, STA9090 ). Parallel
efforts have focused on analogs retaining the macrocycle
but deprived of problematic epoxide and/or dienone moiet-
8
Allꢀe Gaspard Monge, 67000 Strasbourg (France)
Fax : (+33)368855112
E-mail: winssinger@unistra.fr
[11b,19]
ies of radicicol.
Most notably, an oxime derivative of
[
b] Dr. V. Zoete
Swiss Institute of Bioinformatics
Quartier Sorge—1015 Lausanne (Switzerland)
a pochonin analog (pochoxime) was shown to be efficacious
in several in vivo models: mice bearing a breast tumor
[19g]
[20]
[
c] Prof. M. Karplus
Harvard University
(BT474) xenograft,
a model of glioblastoma, and a mei-
[21]
otic progression of spermatocytes beyond pachytene. Fur-
ther, this pochoxime and a closely related analog were
shown to bind to a different conformation of Hsp90 than
radicicol wherein the so-called ATP lid is rearranged into
1
2 Oxford Street, Cambridge
Massachusetts 02138 (USA)
E-mail: marci@tammy.harvard.edu
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201200546.
[
19h,22]
a contiguous a helix.
The co-crystal structure (pdb ID:
Chem. Eur. J. 2012, 00, 0 – 0
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
1
&
ÞÞ
These are not the final page numbers!

Scheme 1. Proposed biosynthesis of radicicol (top) through the polyketide synthases (PKS) and selected examples of resorcylides related to monocillins
and pochonins (bottom).
3
INW and 3INX, see the Supporting Information, Figure 1)
S or R stereochemistry. Fragment 2 could be expediently ac-
cessed by using the methodology developed by the group of
Denmark for the activation of mild Lewis acids with chiral
Lewis bases. Salient features of this reaction are the ex-
tremely low catalyst loading (1 mol%) and an excellent re-
gioselectivity coupled to high enantiomeric excess (ee) ob-
showed that this rearrangement creates a large lyophilic
pocket at the interface of the side chains of Met98, Leu103,
Phe138, and Trp162 with the oxime substituent sandwiched
between the aryl moiety of Trp162 and the side chain of
Met98. The rearrangement thus creates the opportunity for
favorable interactions with lipophilic substituents on the
oxime. Previous studies had suggested that cyclic aliphatic
[23]
tained with vinylogous nucleophiles.
From a practical
standpoint, both enantiomers of the catalyst are commer-
cially available. However, the reaction by using the required
2-butenal substrate had not been previously reported with
vinylogous nucleophiles and aliphatic acrylic aldehydes tend
to be poor substrates. The reaction of the vinylogous slilyl-
dienol ether 3 with 2-butenal and SiCl4 catalyzed by the
Denmarkꢂs catalyst afforded the desired compound 4 in
60% yield and >94% ee (Scheme 3). The lower yield of the
reaction for 2-butenal had been previously noted with silyl
[19h]
rings such as piperidine amide provided optimal activity.
These added interactions provided a rationale for the en-
hancements in activity of the pochoximes relative to pocho-
nin A or D as well as radicicol.
Results and Discussion
[24]
The asymmetric syntheses of pochonin E and F were envi-
sioned to proceed as shown in Scheme 2 and were based on
the logic of previous pochonin synthesis thus requiring the
enol ether. The alcohol group was then protected as TBS
ether and the ester converted to the Weinreb amide under
the action of the magnesium chloride salt of the Weinreb
amine in excellent yield. The stereoisomer R-2 was prepared
starting with the catalyst (R,R)-5 and its enantiomer S-2 was
synthesized by using the catalyst (S,S)-5. The absolute ste-
reochemistry obtained was ultimately verified by X-ray crys-
tallography of a final compound (see below). As shown in
[11b]
known fragment 1
and the chiral fragment 2 with either
[11b,19g]
Scheme 4, intermediates 1a and 1b
were individually
treated with LDA to deprotonate the benzylic position and
engaged in a reaction with either (R)-2 or (S)-2 to obtain
the product 6a and 6b, respectively, in acceptable yield and
[7,11b,19e,f]
consistent with previous use of this transformation.
Notably, d elimination of the hydroxyl ether from the a,b-
Scheme 2. Disconnections for the synthesis of pochonin E.
conjugated system was not observed under the basic condi-
&
2
&
www.chemeurj.org
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 0000, 00, 0 – 0
ÝÝ
These are not the final page numbers!

Asymmetric Synthesis of Pochonin E and F
FULL PAPER
[
19g,h]
me),
(
particularly with 2-
piperidyl acet-
aminooxy)
amide, we sought to apply this
modification to this series of
compounds. It had been estab-
lished that the E oximes were
more potent than the Z oxi-
[19g]
mes.
For the preparation of
the pochoxime analogs of po-
chonin E and F as well as their
epimers, intermediates 6a and
4
Scheme 3. Synthesis of key fragment 2. a) 3 (2.0 equiv), 2-butenal (1.0 equiv), 5 (0.01 mol%), SiCl (1.1 equiv),
6
b
were condensed with 2-
(aminooxy) piperidyl acetamide
8) in pyridine to afford both
epimers 9a and 9b, respectively,
CH
2
Cl
2
, ꢀ788C, 1 h, 60 %; b) tert-butyldimethylsilyl chloride (TBSCl) (1.5 equiv), imidazole (1.5 equiv), DMF,
2
38C, 16 h, 89%; c) Me ACHTUNGTRENNUNG( MeO)NH, iPrMgCl, THF, ꢀ208C, 10 min, 93%.
(
tions of the reaction despite an excess of LDA. Cyclization
as a mixture of E/Z-isomeric oximes. Contrarily to the ke-
tones 6a and 6b, oximes 9 afforded excellent yield in the
metathesis (75–95%). Global deprotection under the action
of an sulfonic acid resin afforded a separable mixture of E
and Z oxime isomers (about 1:1) of pochoxime E and F as
well as their epimers (10b and 10a, respectively).
Having access to pochonins E and F with both R and S
stereochemistry for the C-6 hydroxyl group, the NMR spec-
tra of the synthetic compounds were compared to the ones
of the natural products (Table 1). In deuterated methanol,
the most salient difference in the NMR spectra of pocho-
[25]
under the action of second generation Grubbs catalyst
Grubbs II) for each permutation of isomers and aryl sub-
(
stituent of 6 followed by a sulfonic acid catalyzed deprotec-
tion of the ethoxymethyl (EOM) with concomitant depro-
tection of the silyl ether afforded the two different diaste-
reoisomers of pochonin E (7b) and pochonin F (7a), respec-
tively. Surprisingly, the yield of the metathesis (48–65%)
was lower than in previous cases lacking the allylic substitu-
[19f]
tion.
Based on the fact that significant gain in activity can
be achieved by converting the ketone to an oxime (pochoxi-
Scheme 4. Synthesis of pochonin E and F, [7
b (1.5 equiv), (R)-2 or (S)-2 (1.0 equiv), lithium diisopropylamide (LDA) (2.2 equiv), ꢀ788C, 15 min, 52–75%; b) Grubbs catalyst II (0.5 mol%), tolu-
ene (808C) or CH Cl (heating to reflux), 5-8 h, 48–100%; c) PS-SO H (10 equiv), 238C, 16 h, 47–93%; d) 8 (6.0 equiv), pyridine (py), 458C, 48 h, 53–
5%.
ACTHUNGTRENNU(GN 6-S)] epi-pochonin E and F [7 ACHTNUGTRENNUNG( 6-R)], and their conversion in to the corresponding pochoximes 10. a) 1a or
1
2
2
3
7
Chem. Eur. J. 2012, 00, 0 – 0
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
&
3
&
ÞÞ
These are not the final page numbers!

N. Wissinger, M. Karplus et al.
1
Table 1. Comparison of H NMR data for the different C-6 configurations corresponding to pochonin E and F.
7
b
A
H
U
G
R
N
U
G
7b AHCTUNGRETNNUN(G 6-S)
[
11]
Carbon
Natural pochonin E
]MeOD
d(H) [ppm]
Synthetic pochonin E
[D ]MeOD
Multiplicity (J [Hz])
Synthetic epi-pochonin E
[D
4
4
[D
4
]MeOD
Multiplicity (J [Hz])
d(H) [ppm]
d(H) [ppm]
Multiplicity (J [Hz])
1
2
3
1.19
5.20
2.47
d (6.4)
m
1.19
5.20
2.47
2.16
5.38
5.24
4.23
2.36
2.25
6.69
5.80
4.17
4.09
6.37
d (6.4)
m
1.19
5.25
2.47
2.21
5.38
5.09
4.09
2.44
2.00
6.61
5.74
4.08
3.91
6.39
d (6.4)
m
ddd (14.3, 7.6, 3.7)
ddd (14.3, 7.6, 5.5)
dt (15.3, 7.6)
dd (15.3, 6.1)
m
ddd (12.8, 7.3, 3.4)
dt (12.8, 7.3)
dt (15.9, 7.3)
d (15.9)
ddd (14.3, 7.5, 3.6)
ddd (14.3, 7.5, 5.6)
dt (15.2, 7.5)
dd (15.2, 6.0)
m
ddd (12.8, 7.3, 3.6)
dt (12.8, 7.3)
dt (15.9, 7.3)
d (15.9)
ddd (12.2, 9.8, 3.8)
ddd (12.2, 5.2, 1.4)
ddd (15.1, 9.8, 5.2)
dd (15.1, 7.9)
m
2
.16
4
5
6
7
5.38
5.23
4.24
2.36
m
2
.25
dt (11.7, 10.1)
ddd (15.8, 10.1, 6.04)
d (15.8)
d (17.8)
d (17.8)
8
9
1
6.69
5.80
4.14
1
5
d (17.7)
d (17.7)
s
d (17.6)
d (17.6)
s
4
.10
1
6.35
s
7
a
A
H
U
G
R
N
U
G
7a AHCTUNGRETNNUN(G 6-S)
[
11]
Carbon
Natural pochonin F
]acetone
d(H) [ppm]
Synthetic pochonin F
[D ]acetone
Multiplicity (J [Hz])
Synthetic epi-pochonin F
[D ]acetone
Multiplicity (J [Hz])
[D
6
6
6
Multiplicity (J [Hz])
d(H) [ppm]
d(H) [ppm]
1
2
3
1.28
5.38
2.65
d (6.7)
m
ddd (14.0, 7.9, 4.3)
ddd (14.0, 7.9, 4.3)
dt (15.3, 7.9)
m
1.28
5.38
2.65
2.24
5.52
5.37
4.40
2.48
2.34
6.73
5.88
4.04
3.98
6.30
d (6.5)
m
ddd (14.1, 7.8, 4.3)
ddd (14.1, 7.8, 4.3)
dt (15.3, 8.1)
m
1.30
5.31
2.68
2.29
5.49
5.27
4.25
2.50
2.13
6.66
5.85
4.13
3.70
6.31
d (6.7)
m
ddd (14.1, 9.6, 3.9)
ddd (14.1, 5.2, 3.7)
ddd (15.2, 9.6, 5.2)
m
2
.24
4
5
6
7
5.52
5.37
4.41
2.48
m
m
m
m
ddd (14.0, 7.3, 4.9)
dt (14.0, 8.5)
ddd (15.9, 8.5, 7.3)
d (15.9)
d (16.5)
d (16.5)
ddd (12.8, 7.3, 4.0)
dt (12.8, 8.4)
ddd (15.9, 8.5, 7.3)
d (15.9)
d (16.8)
d (16.8)
2
.34
dt (12.2, 9.2)
ddd (15.7, 9.2, 6.6)
d (15.7)
d (16.9)
d (16.9)
8
9
1
6.73
5.89
4.04
1
5
3
.98
1
6.31
d (2.4)
d (2.4)
d (2.4)
nin E was the chemical shift of the C-2 and C-5 protons. For
the S isomer, the C-2 proton was between the two alkenic
protons of C-4 and C-5, whereas for the R isomer, it was
slightly upfield of the C-5 alkenic protons, which shifted
downfield by 0.15 ppm (Table 1). Also notable are differen-
ces in the chemical shifts between both protons at C-11. In
the case of the S isomer, they are sufficiently resolved to
give the expected pair of doublets. In the case of the R
isomer, they are close and only give a hint of their splitting
showing an apparent singlet. In the case of pochonin F,
NMR spectrum of which was reported in deuterated ace-
tone, the distinction between the C-2 proton is less pro-
nounced but a similar difference between the C-11 protons
is visible with the S isomer having a larger difference in the
chemical shifts between both protons. The NMR comparison
between both synthetic epimers at C-6 and the natural prod-
uct leads to the conclusion that the natural compound has
the R stereochemistry at C-6. Crystallization of the pochox-
ime isomer 10a (E-oxime isomer) emanating from the inter-
mediate, which yielded the pochonin F epimers, afforded
&
4
&
www.chemeurj.org
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 0000, 00, 0 – 0
ÝÝ
These are not the final page numbers!

Asymmetric Synthesis of Pochonin E and F
FULL PAPER
treated with tetrabutylammonium fluoride (TBAF) to selec-
tively remove the silyl group, thus affording compound 11.
The hydroxyl group was converted to an azide in two steps
(
Ms-Cl (Ms=methane sulfonyl), NaN ) to obtain compound
3
1
2, which was reduced to an amine with trimethylphosphine.
Global deprotection of the EOM groups with a sulfonic acid
resin thus afforded the C-6 amino pochoxime 13. Alterna-
tive strategies to convert the hydroxyl group by using palla-
dium-catalyzed p-ally chemistry through its acetylated or
carbonated form were not productive. The key azide 12, fol-
lowing a reduction to the amine, could also be derivatized
as a chloroacetamide and conjugated to 1-b-thioglucose to
afford compound 14. The same sequence by using a simple
acetylation afforded 15. Although the chemistry shown in
Scheme 5 is starting with the S isomer 9a, the same proce-
dures were carried out with the R isomer, thus affording
products 13–15 as the S isomers at C-6. All the products
were obtained as a mixture of oxime geometries, which
were separated by chromatography.
Figure 1. Key differentiating proton NMR signals between pochonin E
(
bottom) and its C-6 epimer (epi-pochonin E, top)—spectra recorded in
CD OD.
3
We next measured the affinity of the new pochonins and
their pochoximes derivatives for human Hsp90a by using
[26]
a previously reported competition assay (Table 2). Consis-
[11b]
tent with previous observations for pochonin D,
the aryl
chloride is important in the interaction of the pochonins
with HSP90 (pochonin E vs. F). Amongst the pochonins, po-
chonin E was found to be the best ligand for Hsp90 with an
affinity of 196 nm as compared to 140 nm for radicicol. Con-
version of the pochonin to the corresponding pochoxime led
to consistent and significant gain in affinity with the best
ligand (epi-pochoxime F) being 80 times more potent than
epi-pochonin F. A significant difference between the pocho-
nins and their pochoxime derivatives is the fact that the aryl
chloride no longer seems important for Hsp90 binding in
the pochoxime. Indeed, the most potent pochoxime does
not have an aryl chloride. There are also significant differ-
ence in binding affinity (16-fold) depending on the stereo-
Figure 2. X-ray crystallographic structure of pochoxime F [10a AHCUTNGRNEUNG( 6-R)] in
gray overlaid with the predicted structure (RMSD: 0.8 ꢃ).
diffracting crystals (Figure 2) and clearly showed the R ste-
reochemistry at C-6.
In efforts to further explore modifications at the C-6 posi-
tion of the pochoxime scaffold, the fully protected pochox-
ime obtained from the metathesis cyclization of 9a was
Scheme 5. Synthesis of aminopochoxime F and its derivatives. a) 9a
S) (1.0 equiv), TBAF (5.0 equiv), THF, 238C, 3 h, 72 %; c) 11(6-S) (1.0 equiv), Ms-Cl (4.0 equiv), Et
10 equiv), DMF, 238C, 24 h, 74% for two steps; e) Me P (4.0 equiv), THF·H O (5:1), 238C, 5 h; f) PS-SO
2% for 14 from 12, and 64% for 15 from 12; g) (Ac) O (5.0 equiv), iPr EtN (5.0 equiv), CH Cl , 08C, 10 min; h) 1) (ClCH
5.0 equiv), CH Cl , 08C, 10 min; 2) tetra-O-acetyl-1-S-acetyl-1-thio-b-d-glucopyranose (3.0 equiv), Na CO (15 equiv), MeOH, 238C, 6 h.
A
H
U
G
R
N
U
2
Cl
2
, heating to reflux, 8 h, 87%; b) 9a
ACHTUNGTRENNUNG( 6-
A
H
U
G
R
N
U
G
3
2
Cl , 0 to 238C, 7 h; d) NaN
2
3
(
5
(
3
2
3
2
2
2
2
2
2
2
2
2
2
3
Chem. Eur. J. 2012, 00, 0 – 0
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
&
5
&
ÞÞ
These are not the final page numbers!

N. Wissinger, M. Karplus et al.
D
Table 2. Measured K for human HSP90a for radicicol, pochonin D, po-
chonin E and F, their epimers and E-oximes. The values represent the
2
average of three experiments (root-mean-square deviations (R ) are
given in parenthesis).
2
K
D
[nm] (R )
Compound
140 (0.991)
360 (0.996)
318 (0.998)
196 (0.993)
1119 (0.996)
1715 (0.994)
38 (0.949)
34 (0.945)
14 (0.929)
22 (0.948)
34 (0.945)
229 (0.989)
90 (0.988)
32 (0.982)
343 (0.992)
radicicol
pochonin D
epi-pochonin E [7b AHCUTNGRNENUG( 6-R)]
pochonin E [7b
epi-pochonin F [7a
pochonin F [7a(6-S)]
epi-pochoxime E [10b
pochoxime E [10b(6-R))
epi-pochoxime F [10a(6-S)]
pochoxime F [10a(6-R)]
epi-aminopochoxime E [13
aminopochoxime E [13(6-R)]
ACHTUNGTRENNUNG
AHCTUNGTRENNUNG
AHCTUNGTRENNUNG
AHCTUNGTRENNUNG( 6-S)]
AHCTUNGTRENNUNG
AHCTUNGTRENNUNG
AHCTUNGTRENNUNG
AHCTUNGTRENNUNG( 6-S)]
AHCTUNGTRENNUNG
14
14
15
ACHTUNGTRENNUNG
AHCTUNGTRENNUNG
AHCTUNGTRENNUNG
chemistry and the nature of the substituent at C-6. In order
to gain insight into the origin of these differences, the
impact on the conformational profile of modifications at C-6
was evaluated for compounds 10 (6-S and 6-R diastereoiso-
mers with and without chlorine) and 13 (6-S and 6-R diaste-
reoisomers). Each molecule was simulated by molecular dy-
namics with the Merck molecular force field in the
CHARMM program to analyze its conformation profile. As
[11b]
previously described
this analysis led to the identification
of three main conformations: an L-shape conformation,
which is the bioactive conformation of radicicol and pochox-
ime derivatives (see Figure 3), an essentially planar (P-
shape) conformation, and an L’-shape conformation that
mainly differs from the L-shape by the fact that the macro-
cycle is positioned on the opposite side of the aromatic
Figure 3. Docking of pochoxime E 10b
ACHTUNGTNERNUNG( 6-R) (top) and epi-pochonin E
1
0b ACHTUNTGR(EUNNNG 6-S) (bottom) to Hsp90. The bioactive conformation can be describe
[11b]
as an “L-shape” based on the fact that the aliphatic portion of the macro-
cycle is effectively perpendicular to the aromatic portion.
cycle. In the previous study,
we found that for radicicol
the bioactive conformation (L) was most stable and also
predicted good binding for pochonin D, L conformation of
ꢀ
1
which is only mildly (1.2 kcalmol ) disfavored.
A more quantitative analysis
of the binding data for a series
D
Table 3. Calculated energetic penalty to adopt the bioactive conformation, measured K , experimental and
of compounds reported in this
paper is made here. For all the
studied molecules, the L’-shape
conformer, rather than the L
conformation, was found to be
energetically most favorable.
This is in accord with the fact
that the L’ shape has been ob-
served in the crystal structure
calculated DGbind values.
Energetic penalty
for adopting
[
a]
[b]
[c]
K
[
D
DGbind
DGbind
the bioactive
conformation
ꢀ
1
ꢀ1
nm]
[kcalmol
]
[kcalmol ]
ꢀ
1
[kcalmol
]
radicicol
0.00
0.36
0.98
0.78
2.07
1.45
1.96
2.68
140
22
38
34
14
22
34
229
ꢀ9.5
ꢀ10.6
ꢀ10.3
ꢀ10.3
ꢀ10.9
ꢀ10.6
ꢀ10.3
ꢀ9.2
ꢀ9.4
ꢀ10.7
ꢀ10.4
ꢀ10.5
ꢀ10.1
ꢀ10.3
ꢀ10.3
ꢀ9.7
of
compound
10a AHCTUNTGERNNUN(G 6-R) pochoxime D: X=Cl; R=H
(
Figure 2 shows the experimen- epi-pochoxime E (10b): X=Cl; R=S-OH
pochoxime E (10b): X=Cl; R=R-OH
epi-pochoxime F (10a): X=H; R=S-OH
pochoxime F (10a): X=H; R=R-OH
epi-aminopochoxime F (13): X=H; R=S-NH
tal and predicted structure).
Depending on the molecule,
the L-shape conformation was
2
energetically disfavored by 0.8– aminopochoxime F (13): X=H; R=R-NH2
ꢀ
1
2.7 kcalmol
(Table 3). Each
D
[a] Measured against human Hsp90. [b] Experimentally obtained from the measured K . [c] Calculated by
compound was docked into using Equation (2).
&
6
&
www.chemeurj.org
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 0000, 00, 0 – 0
ÝÝ
These are not the final page numbers!

Asymmetric Synthesis of Pochonin E and F
FULL PAPER
Hsp90, based on the Hsp90 conformation observed in co-
crystals of Hsp90 with pochoxime A and B, which is similar
to the one of radicicol (pdb ID: 2INW, 3INX, and 3BGQ re-
spectively, see the Supporting Information, Figure 1). Three
son of 10b
R) (X=Cl) with 10aAHCTUNGTRENNUNG
A
H
U
G
E
N
N
ACTHUNGTRENUNNG( 6-S) (X=H) and 10b ACHTUNGTRENNUNG( 6-
ingly, that the conformational energy penalty is significantly
larger for H than for Cl. An analysis of the conformational
energy terms shows that this results from a destabilization
of the L’-shape conformer, making it easier for the molecule
to adopt the bioactive L-shape conformer from the L’-shape
conformer, which dominates in solution. Interestingly, the
larger energy penalty has only a small effect on the binding
free energy, suggesting that it is cancelled by the interactions
with the protein. For both pairs of molecules it can be seen
that the van der Waals energy Evdw is significantly more neg-
ative for X=H than for Cl; the differential values are 5.49
and 1.95 kcalmol , respectively. The contribution of the
nonpolar desolvation term is negligible. The difference in
the van der Waals interaction with the protein is due to
more favorable van der Waals interactions between the
phenyl cycle of the ligands and the Phe138 residue in the ab-
[27]
[28]
programs (Autodock 4,
Autodock Vina,
and the “at-
[29]
tracting cavities” algorithm of EADock ) were used for
the docking study. Starting with the L conformer, all three
programs reproduced the experimental binding mode for
these three compounds with a root-mean-square deviation
lower than 1.2 ꢃ. The “attracting cavities” approach of
EADock predicted an L binding mode similar to that of the
3INW ligand for the new pochoxime derivatives (see
Table 3), independent of the starting conformation used to
seed the docking process. The structures obtained with
EADock were confirmed by Autodock 4 and AutodockVina
runs that were seeded with the L conformer. However,
Autodock 4 and AutodockVina seeded with the P conformer
led to significantly different bound structures with more un-
favorable scores (data not shown).
ꢀ
1
ꢀ
1
sence of the Cl atom (+1.5 kcalmol
for X=Cl and
ꢀ
1
The binding affinities of the new pochoxime derivatives in
their calculated binding modes were first estimated by using
ꢀ0.4 kcalmol for X=H, on average) as well as more fa-
vorable van der Waals interactions between the oxime-
based group of the ligand and the hydrophobic pocket it oc-
cupies, that is Met98, Leu103, Leu107, Phe138, Tyr139,
ꢀ
1
Equation (1) (in kcalmol ).
ꢀ
1
DGbind ¼ kþaðEvdwþDGdesol,npÞþbðEelecþGdesol,elecÞþgEconf ð1Þ
Val150, and Trp162 (ꢀ12.5 kcalmol
for X=Cl and
ꢀ
1
ꢀ
13.3 kcalmol for X=H, on average). The latter is due to
in which Evdw and Eelec are the van der Waals and the elec-
trostatic interaction energies between the ligand and the
protein, DGdesol,np andDGdesol,elec are the non-polar and the
electrostatic desolvation energies upon complexation and
Econf is the energetic penalty for adopting the bioactive con-
a 1.0 ꢃ motion of the oxime-based group made possible by
the absence of the Cl atom, which places it closer to
Leu103, Val150, and Trp162 and essentially counterbalances
the configurational energy penalty.
ꢀ
1
The most pronounced energetic penalty (2.68 kcalmol )
is observed for compound 13 with the amino group at C-6 in
the R stereochemistry, though that for the S stereochemistry
is also quite large (1.96 kcalmol ; see Table 3). Both of
these compounds have X=H, which is likely to contribute
to the energy penalty (see above). The amino group is pro-
tonated at physiological pH making this substituent more
sterically demanding than the corresponding hydroxyl group
formation. The values for Evdw, Eelec, DG
, and DGdesol,elec
desol,np
27]
[
were calculated as previously described, and a, b, and g
were fitted by multiple linear regression to the experimental
binding free energies. Because the contribution of the elec-
trostatic term was statistically not significant, it was removed
from Equation (1), leading to the final Equation (2) (in kcal
ꢀ
1
ꢀ
1
mol ):
and contributes to the large conformational energy penalty.
À
Á
+
ꢀ
OH or ꢀNH3 func-
DGbind ¼ ꢀ5:7 þ 0:076 E þ DGdesolv;np þ 0:61Ecorf
ð2Þ
In the L- and P-shape conformers, the
vdw
tions in the 6-position of the macrocycle are in an unfavora-
ble axial conformation for the R configuration of the C-6
atom, and in a more favorable equatorial position when C-6
is in the S configuration. In the L’-shape conformer, in con-
with a correlation coefficient and a leave-one-out correla-
tion coefficient of 0.76 and 0.64, respectively. Table 3 lists
the experimental and calculated DGbind values as well as
Econf. The results are in reasonable agreement with the ex-
perimental determined affinities, that is, it is found that both
radicicol and aminopochoxime F (13) (X=H, R=RNH2)
have low affinities relative to the other compounds, which
have similar affinities.
+
trast, the ꢀOH or ꢀNH functions are in an equatorial po-
3
sition when C-6 is in the R configuration, and in an axial po-
sition otherwise. This effect, more pronounced in the case of
+
the bulkier and charged ꢀNH function, appears to explain
3
the large propensity of the R-aminopochoxime F (13) mole-
cule to be in the L’-shape conformation.
There are a number of interesting points about the series
of compounds listed in Table 3. To interpret them we show
the individual contributions to Equation (2) in Table S1 in
the Supporting Information. Although radicicol is the only
compound with no configurational energy penalty, its bind-
ing affinity is low. This is due to the fact that Evdw and
DGdesol,np are small relative to the other compounds, as
a result of the absence of the favorable non-polar contacts
made by the oxime-based fragment of the latter. Compari-
Compound 13 with R stereochemistry has the worst affini-
ty for Hsp90 in the series studied computationally; it is
slightly weaker than radicicol. The difference between it and
the S stereochemistry compound arises from the configura-
tional penalty because the interaction with Hsp90 is essen-
tially identical (see the Supporting Information, Table 1).
Substitution of the C-6 amino group with the glycan (14) is
tolerated, preferably with the S stereochemistry (32 vs.
Chem. Eur. J. 2012, 00, 0 – 0
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
&
7
&
ÞÞ
These are not the final page numbers!

N. Wissinger, M. Karplus et al.
9
0 nm for the R isomer), which minimized the energetic pen-
alty of adopting the bioactive conformation by analogy to
(6-S) versus 13(6-R). The ability to conjugate glycans or
[5] J. Dong, Y. Zhu, H. Song, R. Li, H. He, H. Liu, R. Huang, Y. Zhou,
L. Wang, Y. Cao, K. Zhang, J. Chem. Ecol. 2007, 33, 1115–1126.
[
6] V. Hellwig, A. Mayer-Bartschmid, H. Muller, G. Greif, G. Kley-
13
A
C
H
T
U
N
G
T
R
E
N
N
U
N
G
ACHTUNGTRENNUNG
mann, W. Zitzmann, H. V. Tichy, M. Stadler, J. Nat. Prod. 2003, 66,
other targeting groups at this position without significantly
perturbing their binding affinity may offer interesting phar-
macological opportunities. Docking of pochoxime F and its
C-6 epimer in Hsp90 shows that, in the R stereochemistry
8
29–837.
[7] S. Barluenga, P. Lopez, E. Moulin, N. Winssinger, Angew. Chem.
004, 116, 3549–3552; Angew. Chem. Int. Ed. 2004, 43, 3467–3470.
2
[
8] H. Shinonaga, Y. Kawamura, A. Ikeda, M. Aoki, N. Sakai, N. Fuji-
moto, A. Kawashima, Tetrahedron Lett. 2009, 50, 108–110.
9] D. Martinez-Solorio, K. A. Belmore, M. P. Jennings, J. Org. Chem.
(
pochoxime F), the hydroxyl group is pointing towards the
[
solvent. However, in the S stereochemistry (epi-pochoxi-
2
011, 76, 3898–3908.
me F), it is within the hydrogen-bond distance of Asp54
[10] H. Shinonaga, Y. Kawamura, A. Ikeda, M. Aoki, N. Sakai, N. Fuji-
moto, A. Kawashima, Tetrahedron 2009, 65, 3446–3453.
(
Figure 3). This interaction may contribute to the fact that
[
11] a) E. Moulin, S. Barluenga, N. Winssinger, Org. Lett. 2005, 7, 5637–
639; b) E. Moulin, V. Zoete, S. Barluenga, M. Karplus, N. Wins-
this isomer of pochoxime F is the most potent in the series.
5
singer, J. Am. Chem. Soc. 2005, 127, 6999–7004; c) N. Winssinger,
J. G. Fontaine, S. Barluenga, Curr. Top. Med. Chem. 2009, 9, 1419–
1435; d) S. M. Roe, C. Prodromou, R. OꢂBrien, J. E. Ladbury, P. W.
Piper, L. H. Pearl, J. Med. Chem. 1999, 42, 260–266.
Conclusion
[
12] a) M. A. Biamonte, R. Van de Water, J. W. Arndt, R. H. Scannevin,
D. Perret, W. C. Lee, J. Med. Chem. 2010, 53, 3–17; b) Y. L. Janin, J.
Med. Chem. 2005, 48, 7503–7512; c) T. Taldone, A. Gozman, R. Ma-
haraj, G. Chiosis, Curr. Opin. Pharmacol. 2008, 8, 370–374.
The concise synthesis presented herein afforded rapid access
to pochonin E and F as well as their epimers at the allylic
hydroxyl group (C-6). Comparison between the reported
proton NMR data of the natural products and the synthetic
compounds clearly established the stereochemistry of the
hydroxyl group to be the R isomer. Furthermore, the de-
scribed chemistry enabled further exploration into the struc-
ture–activity relationship of this pharmacophore for Hsp90
inhibition, an important therapeutic target. Despite an un-
favorable conformational bias imposed by modifications at
C-6, epi-pochoxime F (10a) stands as the most potent Hsp90
ligand in the pochoxime series (14 nm), which may be ra-
tionalized by the formation of a productive hydrogen bond
to an aspartic acid residue in Hsp90. Finally, the pochox-
[13] a) W. B. Pratt, D. O. Toft, Exp. Biol. Med. 2003, 228, 111–133;
b) F. U. Hartl, A. Bracher, M. Hayer-Hartl, Nature 2011, 475, 324–
3
32.
[
14] a) P. Workman, F. Burrows, L. Neckers, N. Rosen, Ann. N. Y. Acad.
Sci. 2007, 1113, 202–216; b) W. Xu, L. Neckers, Clin. Cancer. Res.
2007, 13, 1625–1629; c) L. Whitesell, S. L. Lindquist, Nat. Rev.
Cancer 2005, 5, 761–772.
[
15] a) A. Sittler, R. Lurz, G. Lueder, J. Priller, H. Lehrach, M. K.
Hayer-Hartl, F. U. Hartl, E. E. Wanker, Hum. Mol. Genet. 2001, 10,
1
307–1315; b) M. Waza, H. Adachi, M. Katsuno, M. Minamiyama,
C. Sang, F. Tanaka, A. Inukai, M. Doyu, G. Sobue, Nat. Med. 2005,
11, 1088–1095; c) W. Luo, F. Dou, A. Rodina, S. Chip, J. Kim, Q.
Zhao, K. Moulick, J. Aguirre, N. Wu, P. Greengard, G. Chiosis, Proc.
Natl. Acad. Sci. USA 2007, 104, 9511–9516; d) W. Luo, W. Sun, T.
Taldone, A. Rodina, G. Chiosis, Mol. Neurodegener. 2010, 5, 24.
16] a) R. Geller, M. Vignuzzi, R. Andino, J. Frydman, Genes Dev. 2007,
21, 195–205; b) X. Sun, E. A. Barlow, S. Ma, S. R. Hagemeier, S. J.
Duellman, R. R. Burgess, J. Tellam, R. Khanna, S. C. Kenney, Proc.
Natl. Acad. Sci. USA 2010, 107, 3146–3151.
17] P. A. Brough, W. Aherne, X. Barril, J. Borgognoni, K. Boxall, J. E.
Cansfield, K. M. Cheung, I. Collins, N. G. Davies, M. J. Drysdale, B.
Dymock, S. A. Eccles, H. Finch, A. Fink, A. Hayes, R. Howes, R. E.
Hubbard, K. James, A. M. Jordan, A. Lockie, V. Martins, A. Massey,
T. P. Matthews, E. McDonald, C. J. Northfield, L. H. Pearl, C. Pro-
dromou, S. Ray, F. I. Raynaud, S. D. Roughley, S. Y. Sharp, A. Sur-
genor, D. L. Walmsley, P. Webb, M. Wood, P. Workman, L. Wright,
J. Med. Chem. 2008, 51, 196–218.
ime–glucose conjugate 14 ACHUTNRGNENUG( 6-S) was shown to be a potent
Hsp90 ligand (32 nm) and may be useful to direct the inhibi-
tor to metabolically demanding malignant cells by an active
uptake mechanism.
[
[
Experimental Section
Full experimental procedures and spectroscopic characterization date are
provided in the Supporting Information.
[
18] Y. Wang, J. B. Trepel, L. M. Neckers, G. Giaccone, Curr. Opin. In-
vestig. Drugs 2010, 11, 1466–1476.
Acknowledgements
[19] a) K. Yamamoto, R. M. Garbaccio, S. J. Stachel, D. B. Solit, G. Chio-
sis, N. Rosen, S. J. Danishefsky, Angew. Chem. 2003, 115, 1318–
1322; Angew. Chem. Int. Ed. 2003, 42, 1280–1284; b) X. Lei, S. J.
Danishefsky, Adv. Synth. Catal. 2008, 350, 1677–1681; c) N. Proisy,
S. Y. Sharp, K. Boxall, S. Connelly, S. M. Roe, C. Prodromou, A. M.
Slawin, L. H. Pearl, P. Workman, C. J. Moody, Chem. Biol. 2006, 13,
This work was supported by a grant from Institut Universitaire de France
IUF) and is part of the COST network D28. An MENRT fellowship
C.Z.) is also gratefully acknowledged.
(
(
1
203–1215; d) J. E. H. Day, A. J. Blake, C. J. Moody, Synlett 2009,
[
1] a) P. Delmotte, J. Delmotte-Plaquee, Nature 1953, 171, 344; b) N.
Winssinger, S. Barluenga, Chem. Commun. 2007, 22–36.
2] a) S. Barluenga, P. Dakas, M. Boulifa, E. Moulin, N. Winssinger, C.
R. Chem. 2008, 11, 1306–1317; b) T. Hofmann, K.-H. Altmann, C.
R. Chim. 2008, 11, 1318–1335.
3] T. J. Turbyville, E. M. Wijeratne, M. X. Liu, A. M. Burns, C. J.
Seliga, L. A. Luevano, C. L. David, S. H. Faeth, L. Whitesell, A. A.
Gunatilaka, J. Nat. Prod. 2006, 69, 178–184.
1567–1570; e) J. E. H. Day, S. Y. Sharp, M. G. Rowlands, W.
Aherne, W. Lewis, S. M. Roe, C. Prodromou, L. H. Pearl, P. Work-
man, C. J. Moody, Chem. Eur. J. 2010, 16, 10366–10372; f) E.
Moulin, S. Barluenga, F. Totzke, N. Winssinger, Chem. Eur. J. 2006,
12, 8819–8834; g) S. Barluenga, C. Wang, J. G. Fontaine, K. Aouadi,
K. Beebe, S. Tsutsumi, L. Neckers, N. Winssinger, Angew. Chem.
2008, 120, 4504–4507; Angew. Chem. Int. Ed. 2008, 47, 4432–4435;
h) S. Barluenga, J. G. Fontaine, C. Wang, K. Aouadi, R. Chen, K.
Beebe, L. Neckers, N. Winssinger, ChemBioChem 2009, 10, 2753–
2759.
[
[
[
4] M. Isaka, C. Suyarnsestakorn, M. Tanticharoen, P. Kongsaeree, Y.
Thebtaranonth, J. Org. Chem. 2002, 67, 1561–1566.
&
8
&
www.chemeurj.org
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 0000, 00, 0 – 0
ÝÝ
These are not the final page numbers!

Asymmetric Synthesis of Pochonin E and F
FULL PAPER
[
20] H. H. Zhu, S. Woolfenden, R. T. Bronson, Z. M. Jaffer, S. Barluenga,
N. Winssinger, A. E. Rubenstein, R. H. Chen, A. Charest, Mol.
Cancer Therap. 2010, 9, 2618–2626.
[25] a) M. Scholl, S. Ding, C. W. Lee, R. H. Grubbs, Org. Lett. 1999, 1,
953–956; b) T. M. Trnka, R. H. Grubbs, Acc. Chem. Res. 2001, 34,
18–29.
[
[
[
21] I. Grad, C. R. Cederroth, J. Walicki, C. Grey, S. Barluenga, N. Wins-
singer, B. De Massy, S. Nef, D. Picard, Plos One 2010, 5, e15770.
22] M. Spichty, A. Taly, F. Hagn, H. Kessler, S. Barluenga, N. Winssing-
er, M. Karplus, Biophys. Chem. 2009, 143, 111–123.
[26] J. Kim, S. Felts, L. Llauger, H. He, H. Huezo, N. Rosen, G. Chiosis,
J. Biomol. Screening 2004, 9, 375–381.
[27] G. M. Morris, R. Huey, W. Lindstrom, M. F. Sanner, R. K. Belew,
D. S. Goodsell, A. J. Olson, J. Comput. Chem. 2009, 30, 2785–2791.
[28] O. Trott, A. J. Olson, J. Comput. Chem. 2010, 31, 455–461.
[29] A. Grosdidier, V. Zoete, O. Michielin, J. Comput. Chem. 2009, 30,
2021–2030.
23] a) S. E. Denmark, G. L. Beutner, J. Am. Chem. Soc. 2003, 125,
7
800–7801; b) S. E. Denmark, J. R. Heemstra, Jr., J. Am. Chem.
Soc. 2006, 128, 1038–1039; c) S. E. Denmark, J. R. Heemstra, Jr., J.
Org. Chem. 2007, 72, 5668–5688.
[
24] S. E. Denmark, G. L. Beutner, T. Wynn, M. D. Eastgate, J. Am.
Chem. Soc. 2005, 127, 3774–3789.
Received: February 19, 2012
Published online: && &&, 0000
Chem. Eur. J. 2012, 00, 0 – 0
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
&
9
&
ÞÞ
These are not the final page numbers!

N. Wissinger, M. Karplus et al.
Natural Products
G. Karthikeyan, C. Zambaldo,
S. Barluenga, V. Zoete, M. Karplus,*
N. Winssinger* ................ &&&& — &&&&
Asymmetric Synthesis of Pochonin E
and F, Revision of Their Proposed
Structure, and Their Conversion to
Potent Hsp90 Inhibitors
Cooling down the heat shock: Rapid
access to pochonin E and F as well as
their epimers at the allylic hydroxyl
group established the correct stereo-
chemistry of the natural product (see
figure). The unnatural epimer afforded
a more potent heat shock protein 90
(Hsp90) ligand, which stands as the
most potent reported.
&
10
&
www.chemeurj.org
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 0000, 00, 0 – 0
ÝÝ
These are not the final page numbers!