Towards the discovery of drug-like epigallocatechin gallate analogs as Hsp90 inhibitors

Article abstract of DOI:10.1016/j.bmcl.2014.03.088

(-)-Epigallocatechin gallate (EGCG) is the major flavonoid of green tea and has been widely explored for a range of biological activities including anti-infective, anti-inflammatory, anti-cancer, and neuroprotection. Existing structure-activity data for EGCG has been largely limited to exploration of simple ethers and hydroxyl deletion. EGCG has poor drug-like properties because of multiple phenolic hydroxyl moieties and a metabolically labile ester. This work reports a substantial expansion of structure-activity understanding by exploring a range of semi-synthetic and synthetic derivatives with ester replacements and variously substituted aromatic and alicyclic groups containing more drug-like substituents. Structure-activity relationships for these molecules were obtained for Hsp90 inhibition. The results indicate that amide and sulfonamide linkers are suitable ester replacements. Hydroxylated aromatic rings and the cis-stereochemistry in EGCG are not essential for Hsp90 inhibition. Selected analogs in this series are more potent than EGCG in a luciferase refolding assay for Hsp90 activity.

Full text of DOI:10.1016/j.bmcl.2014.03.088

Bioorganic & Medicinal Chemistry Letters 24 (2014) 2263–2266
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Towards the discovery of drug-like epigallocatechin gallate analogs
as Hsp90 inhibitors
Rohit Bhat a, Amna T. Adam a, Jungeun Jasmine Lee a, Thomas A. Gasiewicz b, Ellen C. Henry b,
David P. Rotella a,
⇑
a Department of Chemistry and Biochemistry, Montclair State University, Montclair, NJ 07043, United States
b Department of Environmental Medicine, University of Rochester Medical Center, Rochester, NY 14642, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
(À)-Epigallocatechin gallate (EGCG) is the major flavonoid of green tea and has been widely explored for a
range of biological activities including anti-infective, anti-inflammatory, anti-cancer, and neuroprotec-
tion. Existing structure–activity data for EGCG has been largely limited to exploration of simple ethers
and hydroxyl deletion. EGCG has poor drug-like properties because of multiple phenolic hydroxyl moie-
ties and a metabolically labile ester. This work reports a substantial expansion of structure–activity
understanding by exploring a range of semi-synthetic and synthetic derivatives with ester replacements
and variously substituted aromatic and alicyclic groups containing more drug-like substituents. Struc-
ture–activity relationships for these molecules were obtained for Hsp90 inhibition. The results indicate
that amide and sulfonamide linkers are suitable ester replacements. Hydroxylated aromatic rings and
the cis-stereochemistry in EGCG are not essential for Hsp90 inhibition. Selected analogs in this series
are more potent than EGCG in a luciferase refolding assay for Hsp90 activity.
Received 26 January 2014
Accepted 26 March 2014
Available online 4 April 2014
Keywords:
Hsp90 inhibitor
Natural product
Epigallocatechin gallate
Ó 2014 Elsevier Ltd. All rights reserved.
OH
OH
(À)-Epigallocatechin gallate (EGCG, 1, Fig. 1) is the primary
flavonoid component of green tea and has been widely studied
due to its diversity of biological properties including anti-
inflammatory, anti-oxidative, anti-cancer, anti-infective and
OH
OH
OH
OH
B
HO
O
HO
O
C
A
neuroprotective activity.1–3 Structure–activity studies on
1
OH
O
OH
reported to date have focused primarily on a limited spectrum of
analogs involving methylation and/or deletion of phenolic hydro-
xyl moieties in 1–3.4–6 These compounds have been studied for
biological activity in systems related to the therapeutic applica-
tions mentioned above. The resulting derivatives retain unfavor-
able characteristics of EGCG including chemical and metabolic
instability. Our approach to modification of EGCG is focused on dis-
covery of more drug-like derivatives in the context of an expanded
structure–activity study that includes exploration of the influence
of stereochemistry in the natural product on biological/target spe-
cific properties.
This Letter details our initial efforts toward the discovery of
more drug-like analogs of EGCG using inhibition of Hsp90 as a
representative biological target. Hsp90 is an intracellular chaper-
one protein essential for cell growth and division as well as protein
trafficking.7 It is a target of interest for treatment of cancer and
OH
OH
D
O
OH
2 EGC
1 EGCG
OH
OH
OH
OH
OH
OH
HO
O
HO
O
O
O
OH
OH
OH
OH
OH
OH
O
O
OH
OH
4 GCG
3 ECG
Figure 1.
potentially neurodegenerative disorders.8 We prepared a series of
semisynthetic and synthetic EGCG analogs in which the metaboli-
cally labile ester is replaced by more stable linkers, aromatic rings
⇑
Corresponding author. Tel.: +1 973 655 7204; fax: +1 973 655 7772.
E-mail address: rotellad@mail.montclair.edu (D.P. Rotella).
http://dx.doi.org/10.1016/j.bmcl.2014.03.088
0960-894X/Ó 2014 Elsevier Ltd. All rights reserved.

2264
R. Bhat et al. / Bioorg. Med. Chem. Lett. 24 (2014) 2263–2266
(R2)m
heat-induced denaturation is dependent on Hsp90 chaperone func-
R1= H, OH; n=0, 2
R2= H, OH; m=0, 2,3
X= O, NH
Y= (C=O), SO2
Z= aromatic, heteroaromatic, alicyclic
tion, this assay is commonly used to test compounds for their abil-
(R1)n
ity to inhibit Hsp90.9,10
O
The semi-synthesis of EGCG analogs is outlined in Scheme 1
below.11 Silyl protection of the phenolic hydroxyl groups in 1 fol-
lowed by lithium aluminum hydride reduction furnished protected
EGC derivative 6. Esterification by DCC-mediated coupling pro-
vided silyl-protected intermediates in which silyl deprotection
was best accomplished with aqueous HF in tetrahydrofuran, lead-
ing efficiently to targets 7a–e. Oxidation of 6 with Dess–Martin re-
agent followed by reductive amination with benzylamine and
hydrogenolytic debenzylation provided primary amine 8 that
was treated with benzoyl chloride or phenylsulfonyl chloride to
give the respective amide and sulfonamide which were de-silylat-
ed as above leading to amide 9a and sulfonamide 9b in good yield.
Beginning with (+)-catechin (10), selective phenolic hydroxyl
benzylation, esterification and hydrogenolytic debenzylation led
to trans esters 11a–c. Benzyl protection of 10 followed by
X
Y
5
Z
Figure 2.
are replaced by alicyclic systems and phenolic hydroxyl groups are
removed and/or replaced with other substituents as outlined in
Figure 2. Stereochemical effects on Hsp90 activity are evaluated
using a commercially available derivative, gallocatechin gallate
(GCG, 4). EGCG was characterized previously as a C-terminal
binding Hsp90 inhibitor.9 EGCG analogs were evaluated as Hsp90
inhibitors using a luciferase refolding assay because a direct
Hsp90 binding assay for C-terminal inhibitors is not yet avail-
able. Since the recovery of luciferase enzymatic activity after
OH
OH
OTBS
OTBS
OH
OH
OH
HO
O
TBSO
O
HO
O
OH
OH
OH
OTBS
c,d
a,b
O
OH
HO
O
OH
OTBS
OH
Z
O
O
7
a-e
1
6
(EGCG)
OH
OH
OTBS
OTBS
OH
OH
e,f,g
h/i,d
6
O
TBSO
O
OTBS
NH
Y
NH2
OH
OTBS
9
a) Y= C=O
b) Y= SO2
8
OH
OBn
OH
OH
OBn
j,e,k
OH
HO
O
BnO
O
HO
O
j,c,g
O
OH
OH
OH
11
OBn
OH
10
Z
12
O
a-c
COOH
COOH
CO2Et
CO2H
a,m
l,a,m
OH
13
OTBS
14
OH
15
OTBS
16
OBn
OH
d,g
OBn
OH
c
BnO
O
12
14 16
/
+
HO
O
c,g
18 c/f
O
O
OBn
OH
O
O
Z
d,n,g
18 a/b/d/e
OTBS
17a
cis-OTBS
b trans-OTBS
Scheme 1. Reagents and conditions: (a) TBDMS-Cl, Et3N, DMF, rt 95%; (b) LiAlH4/THF, 0 °C 75%; (c) Z-CO2H, DCC, DMAP, CH2Cl2, rt 70%; (d) 48% aq HF, THF, rt 80%; (e) Dess–
Martin, CH2Cl2, rt 85%; (f) benzylamine, cat. HOAc, NaBH3CN, THF, rt 75%; (g) H2 (30 psi), 10% Pd-C, MeOH, rt 95%; (h) benzoyl chloride, Et3N, CH2Cl2, rt 76%; (i) phenyl sulfonyl
chloride, Et3N, CH2Cl2, rt 68%; (j) BnBr, K2CO3, DMF, rt 30%; (k)
0 °C to rt, 75%.
L
-Selectride, THF, À78 °C to rt, 76%; (l) EtOH/H+ reflux, 85%; (m) NaOH/EtOH rt, 89%; (n) methyl DAST, CH2Cl2

R. Bhat et al. / Bioorg. Med. Chem. Lett. 24 (2014) 2263–2266
2265
R1
OH
CH3
b, c
R1
O
R1
O
R1
O
R2
O
g, h
d
19 a) R1, R2 = H
OH
OH
b) R1, R2 = OH
b, c
R2
R2
R2
a
22
21
a) R1, R2 = H
b) R1, R2 = OBn
23
a) R1, R2 = H
a) R1, R2 = H
b) R1, R2 = OBn
BnO
OH
CH3
b) R1, R2 = OBn
e, f
e, f
O
OBn O
20
R1
HO
O
O
O
O
R3
R4
OH
OH
OH
R2
O
R5
OH
24 R1,R2 = OH
25 R1,R2 = H
26 R3,R4,R5 = H
27
R3,R4,R5 = OH
Scheme 2. Reagents and conditions: (a) BnBr, K2CO3, DMF, rt, 45%; (b) benzaldehyde, piperidine, ethanol, reflux, 1N HCl, 43%; (c) NaBH4, EtOH/THF, reflux, rt 16 h, 15% HOAc,
40%; (d) BH3-THF, THF, 0 °C, H2O2, NaOH, THF, 59%; (e) Z-CO2H, DCC, DMAP, DCM, rt, 74%; (f) 30 psi H2, 10% Pd/C, THF, 90%; (g) Dess–Martin, 0 °C to rt, DCM, 80%; (h)
L
-Selectride, THF, À78 °C to rt, 63%.
Table 2
Dess–Martin oxidation and L-Selectride reduction efficiently gave
Semisynthetic catechol D-ring analogs
an epimeric alcohol 12 that was esterified and debenzylated as
before to give protected cis esters 18c and 18f. cis- and trans-
hydroxy-substituted cyclohexyl intermediates (14 and 16) neces-
sary for esters 18d and 18e were prepared by standard methods
and coupled with 12 then converted as indicated in Scheme 1.
Fluoro derivatives 18a and 18b were prepared from their respec-
tive hydroxyl intermediates by methyl DAST.
OH
OH
HO
O
O
OH
O
Z
Synthetic EGCG analogs lacking phenolic hydroxyl groups were
prepared as outlined in Scheme 2. Chalcone synthesis using 19a
and benzyl protected acetophenone 20 furnished 21a and 21b,
respectively, which underwent hydroboration/oxidation to give a
mixture of alcohols in which the trans isomers 22a and 22b pre-
dominate.12,13 Esterification of 22b and deprotection as above pro-
vided trans esters 26 and 27. Dess–Martin oxidation of 22a/b
b
Compd
cis/transa
Z
IC50 (lM)
11a
11b
trans
trans
Cyclohexyl
Phenyl
98
110
11c
18a
trans
cis
OH
280
65
followed by stereoselective reduction by L-Selectride, esterification
F
F
and debenzylation furnished cis-analogs 24 and 25.
EGCG (1), GCG (4) and the new analogs were evaluated for func-
tional inhibition of Hsp90 activity using a luciferase refolding assay
as described elsewhere.9,10 The results are shown in Tables 1–3
18b
18c
cis
cis
69
81
Cyclohexyl
Cyclopropyl
18d
cis
OH
OH
310
Table 1
18e
18f
cis
cis
440
410
Semisynthetic D-ring and linker analogs
OH
OH
a
Catechol B ring/ester (absolute stereochemistry derived from GCG).
Avg based on at least two independent determinations.
b
HO
O
OH
Y
below. The following structure–activity observations can be made:
(1) The metabolically labile ester can be replaced by an amide (9a)
or sulfonamide (9b) leading to more potent Hsp90 inhibitors. This
further suggests that a poly-hydroxy D ring is not necessary for
X
Z
OH
Compd
cis/transa
X
Y
Z
IC50 (lM)
b
Hsp90 inhibition and the relative geometry of
a pyramidal
1
4
cis
trans
cis
cis
cis
cis
cis
cis
cis
O
O
O
O
O
O
O
NH
NH
C@O
C@O
C@O
C@O
C@O
C@O
C@O
C@O
SO2
Gallate
Gallate
Phenyl
4-F-Ph
3,4-Di-F-Ph
4-Pyridyl
Cyclohexyl
Phenyl
155
34
24
41
58
64
30
75
44
sulfonamide compared to the more planar amide or ester does
not disrupt potential binding interactions in this region. The
sulfonamide 9b is approximately 3-fold more potent compared
to EGCG. (2) Hsp90 inhibition associated with these and other
des-hydroxy derivatives shows that there are a variety of poten-
tially more drug-like options for replacement of the gallate ester
including an electron deficient heterocycle (7d) and substituted
alicyclics (18a, 18b). Fluoro-substituted aromatic groups (7b, 7c)
also result in improved Hsp90 inhibition. Selected D-ring analogs
7a
7b
7c
7d
7e
9a
9b
Phenyl
a
Gallate B ring/X (absolute stereochemistry as in EGCG).
Average value based on at least two independent determinations.
b

2266
R. Bhat et al. / Bioorg. Med. Chem. Lett. 24 (2014) 2263–2266
Table 3
Racemic, synthetic analogs
This is especially important given the known tendency of the
4-hydroxy group in the B-ring to play a role in epimerization of
EGCG.14 The metabolically labile ester can be replaced by amide
and sulfonamide linkages and the aromatic D ring can be replaced
by an alicyclic system. Ongoing work is focused on using this data
to identify a combination of sites and analogs where significant
improvements in potency of Hsp90 inhibition can be achieved.
Other research in this laboratory is exploring connections between
increased Hsp90 potency, structural variations and biological
outcomes such as neuroprotection and cytotoxicity.
R1
O
O
R2
R1
O
Z
cis/transa
R2
Z
IC50
(lM)
b
Compd
24
25
26
27
cis
cis
trans
trans
OH
H
OH
OH
OH
H
OH
OH
Phenyl
Gallate
Phenyl
Gallate
63
35
120
150
Acknowledgments
a
Relative phenyl/ester stereochemistry.
Avg based on at least two independent determinations.
This research was supported by the Margaret and Herman Sokol
Endowment, NIH Grant AT006366 and Center Grant ES01247,
Montclair State University and the Sokol Institute for Pharmaceuti-
cal Life Sciences.
b
are measurably more potent than EGCG, for example cyclohexyl
ester 7e and benzoate ester 7a. (3) Deletion of both A- and B-ring
hydroxyl groups as in 25 (a racemate) notably furnishes a more po-
References and notes
tent Hsp90 inhibitor (IC50 35 lM) compared to EGCG. (4) B-ring
catechol esters are less active compared to analogous gallate ana-
logs, cf. 11a versus 7e and 11b versus 7a. This observation suggests
the possibility of more complex SAR in view of the improved
potency of racemic des-hydroxy 25. In this catechol series, fluoro
substitution of the cyclohexane ring is preferable to hydroxyl (cf.
18b vs 18e) and the relative stereochemistry of fluorine has little
effect on Hsp90 inhibition. It is also interesting to observe that
the difference in potency between cis- and trans-isomers in the
racemic, synthetic des-hydroxy analogs in Table 3 differs from that
observed with EGCG and GCG. These results collectively suggest a
number of avenues for future SAR studies.
This initial SAR survey of synthetic and semi-synthetic analogs
was intended to highlight stereochemical and structural variations
in rings A, B and D of EGCG that can serve as starting points for
preparation and study of more drug like derivatives of this biolog-
ically interesting natural product. trans-Geometry on the C ring
furnishes a more potent Hsp90 inhibitor compared to 1. Selected
compounds were more potent than EGCG in an Hsp90 functional
assay. Furthermore our data show that multiple hydroxyl groups
can be eliminated and/or replaced with other functional groups.
1. Steinmann, J.; Buer, J.; Pietschmann, T.; Steinmann, E. Br. J. Pharmacol. 2013,
168, 1059.
2. Huo, C.; Wan, S. B.; Lam, W. H.; Li, L.; Wang, Z.; Landis-Piwowar, K. R.; Chen, D.;
Dou, Q. P.; Chen, T. H. Inflammopharmacology 2008, 16, 248.
3. Hugel, H. M.; Jackson, N. Mini-Rev. Med. Chem. 2012, 12, 380.
4. Khandelwal, A.; Hall, J. A.; Blagg, B. S. J. J. Org. Chem. 2013, 78, 7859.
5. Kazi, A.; Wang, Z.; Kumar, N.; Falsetti, S. C.; Chang, T.-H.; Dou, Q. P. Anticancer
Res. 2004, 24, 943.
6. Osanai, K.; Landis-Piwowar, K. R.; Dou, Q. P.; Chan, T. K. Bioorg. Med. Chem.
2007, 15, 5076.
7. Chiosis, G.; Rosen, N.; Mimnaugh, E.; Whitesell, L.; Neckers, L. Mol. Cancer Ther.
2003, 2, 123.
8. Ernst, J. T.; Liu, M.; Zuccola, H.; Neubert, T.; Beaumont, K.; Turnbull, A.; Kallel,
A.; Vought, B.; Stamos, D. Bioorg. Med. Chem. Lett. 2014, 24, 204.
9. Yin, Z.; Henry, E. C.; Gasiewicz, T. A. Biochemistry 2009, 48, 336.
10. Turbyville, T. J.; Kithsiri Wijeratne, E. M.; Liu, M. X.; Burns, A. M.; Seliga, C. J.;
Luevano, L. A.; David, C. L.; Faeth, S. H.; Whitesell, L.; Gunatilaka, A. A. L. J. Nat.
Prod. 2006, 69, 178.
11. All final compounds were judged to be at least 95% pure and were
characterized by NMR, HPLC and MS.
12. Zaveri, N.; Chao, W.-R. WO2003089423.
13. Devakaram, R.; Black, D. S.; Andrews, K. T.; Fisher, G. M.; Davis, R. A.; Kumar, N.
Bioorg. Med. Chem. 2011, 19, 5199.
14. Suzuki, M.; Sano, M.; Yoshida, R.; Degawa, M.; Miyase, T. Agric. Food Chem.
2003, 51, 510.