Discovery of a potent and selective Bcl-2 inhibitor using SAR by NMR

Article abstract of DOI:10.1016/j.bmcl.2010.09.033

The Bcl-2 family of proteins plays a major role in the regulation of apoptosis, or programmed cell death. Overexpression of the anti-apoptotic members of this family (Bcl-2, Bcl-x_L, and Mcl-1) can render cancer cells resistant to chemotherapeutic agents and therefore these proteins are important targets for the development of new anti-cancer agents. Here we describe the discovery of a potent, highly selective, Bcl-2 inhibitor using SAR by NMR and structure-based drug design which could serve as a starting point for the development of a Bcl-2 selective anti-cancer agent. Such an agent would potentially overcome the Bcl-x_L mediated thrombocytopenia observed with ABT-263.

Full text of DOI:10.1016/j.bmcl.2010.09.033

Bioorganic & Medicinal Chemistry Letters 20 (2010) 6587–6591
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Discovery of a potent and selective Bcl-2 inhibitor using SAR by NMR
Andrew M. Petros, Jeffrey R. Huth, Thorsten Oost, Cheol-Min Park, Hong Ding, Xilu Wang, Haichao Zhang,
Paul Nimmer, Renaldo Mendoza, Chaohong Sun, Jamey Mack, Karl Walter, Sarah Dorwin, Emily Gramling,
⇑
Uri Ladror, Saul H. Rosenberg, Steven W. Elmore, Stephen W. Fesik, Philip J. Hajduk
Global Pharmaceutical Research and Development, Abbott Laboratories, Abbott Park, IL 60064, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
The Bcl-2 family of proteins plays a major role in the regulation of apoptosis, or programmed cell death.
Overexpression of the anti-apoptotic members of this family (Bcl-2, Bcl-xL, and Mcl-1) can render cancer
cells resistant to chemotherapeutic agents and therefore these proteins are important targets for the
development of new anti-cancer agents. Here we describe the discovery of a potent, highly selective,
Bcl-2 inhibitor using SAR by NMR and structure-based drug design which could serve as a starting point
for the development of a Bcl-2 selective anti-cancer agent. Such an agent would potentially overcome the
Bcl-xL mediated thrombocytopenia observed with ABT-263.
Received 12 August 2010
Revised 1 September 2010
Accepted 7 September 2010
Available online 15 September 2010
Keywords:
NMR
Structure-based
Linking
Ó 2010 Elsevier Ltd. All rights reserved.
Fragment
Apoptosis
Bcl
Cancer is a complex disease that arises from a decades-long pro-
cess of accumulated mutations. One hallmark of malignant cells is
their ability to evade apoptosis, or programmed cell death.1 There
are many ways that cancer cells achieve this, one of which is
through overexpression of one or more members of the Bcl-2 fam-
ily of anti-apoptotic proteins. This family includes Bcl-2, Bcl-xL,
Bcl-w, Mcl-1, and A1. Recently, we have developed a potent, orally
bioavailable, dual Bcl-xL/Bcl-2 inhibitor (ABT-263) that shows ro-
bust in vivo activity against a number of different tumors.2 While
a very potent anti-tumor agent, this molecule suffers the drawback
of inducing thrombocytopenia upon dosing.3 Furthermore, we
have shown that this thrombocytopenia is mediated by inhibition
of Bcl-xL and not of Bcl-2.3 Resistance of cancer cells to apoptosis,
on the other hand, is mediated by Bcl-xL, Bcl-2, or both, depending
on the tumor type. Therefore, a Bcl-2 selective inhibitor could have
utility as a platelet-sparing anti-tumor agent.
In order to identify ligands that could serve as starting cores in
the development of Bcl-2 selective inhibitors, an NMR-based
screen of human Bcl-2 was conducted using a library of about
17,000 compounds with an average molecular weight of
225 Da.4 Compound binding was monitored by following chemi-
cal shift changes of Leu, Val, and Ile methyl groups in a 13C-HSQC
spectrum upon compound addition. A diphenylmethane (com-
pound 1, Table 1) was discovered that binds to Bcl-2 with a KD
of 20 lM, as measured in an NMR titration experiment. A similar
compound has been discovered by Jahnke using an NMR spin-
labeling method.5 Complementary titrations using Bcl-xL found
this compound to be 20-fold selective for Bcl-2 (Table 1) and it
was, in fact, the only compound to show this degree of selectivity.
This is in contrast to the binding profile of biaryl acid compounds
that were previously discovered in a screen of Bcl-xL.4 For exam-
ple, biaryl acids (6, 7, and 8) bind with nearly equal affinity to
both Bcl-2 family members (Table 2).
To understand the structural basis for this selectivity and to
guide the design of more potent Bcl-2 selective inhibitors, NMR
structural studies of compound 1 bound to Bcl-2 were conducted.
NMR was chosen for these studies since we were not able, at the
time, to obtain crystals of Bcl-2 in complex with compounds of
micromolar affinity. Twelve protein–ligand NOEs were observed
in a three-dimensional, 13C-edited, 12C-filtered NOESY spectrum
(Supplementary data), which were then used to guide docking of
1 into the Bcl-2 groove. As shown in Figure 1, the chlorinated phe-
nyl rings of 1 pack into the hydrophobic groove created by the side
Here, we report the use of SAR by NMR and structure-based drug
design in the discovery of selective inhibitors of Bcl-2 based on a
diphenylmethane core. These compounds have nanomolar potency
against Bcl-2, but exhibit >1000-fold and >28-fold selectivity over
Bcl-xL and Mcl-1, respectively, and could serve as a useful starting
point for development of a Bcl-2 selective, anti-tumor agent.
⇑
Corresponding author. Address: Abbott Laboratories, Dept. R460, Bldg. AP10,
100 Abbott Park Rd., Abbott Park, IL 60064, United States. Tel.: +1 (847)937 0368;
fax: +1 (847)938 2478.
E-mail address: philip.hajduk@abbott.com (P.J. Hajduk).
0960-894X/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2010.09.033

6588
A. M. Petros et al. / Bioorg. Med. Chem. Lett. 20 (2010) 6587–6591
Table 1
NMR KD values for diphenylmethanes
a
a
No.
Structure
KD
(lM) Bcl-2
KD
(l
M) Bcl-xL
N
OH
Cl
1
20
450
Cl
N
OH
2
3
80
500
Cl
N
200
5000
OH
N
N
4
5
60
700
NDb
Cl
Cl
NH2
OH
F
250
a
From NMR titration.
Not determined.
b
Figure 1. NMR-derived structure of diphenylmethane compound 1 bound to Bcl-2.
(A) Ribbon representation. Amino acid side chains that form the binding pocket are
indicated with carbon in green, nitrogen in blue, sulfur in orange, and oxygen in red.
Atoms of the ligand are colored with carbon in orange, oxygen in red, nitrogen in
blue, and chlorine in green. (B) Bcl-2 surface is shown in green with oxygen atoms
in red and nitrogen atoms in blue.
Table 2
NMR KD values for biaryl acids
a
a
No.
Structure
Bcl-XL KD
(
lM)
Bcl-2 KD
(
l
M)
À1b
+1c
HO
O
F
6
7
8
300
290
360
400
100
300
430
Based on the NMR-derived structure, a strategy was pursued to
extend the diphenylmethane core into the unoccupied region of
the groove in order to improve potency. A structural comparison
of the Bcl-2/diphenylmethane complex with the Bcl-xL/biaryl acid
complex4 suggested that the diphenylmethane and biaryl acid li-
gands would occupy neighboring sites in the Bcl-2 groove and
could thus bind simultaneously. To test this hypothesis, 70 biaryl
acids were screened for binding to a complex of Bcl-2 and com-
pound 1. As the data in Table 2 show, the para-fluoro substituted
biaryl (6) binds equally as well to Bcl-2 in the presence and ab-
sence of the diphenylmethane. No evidence for competition for
binding to the same site was observed. The larger ethyl group at
the meta position binds fourfold more tightly (compound 7). Fur-
thermore, the ethyl substitution may favorably interact with the
diphenylmethane as evidenced by the fivefold increase in potency
when compound 1 is present versus when it is absent. In contrast,
an ethyl substitution at the para position (8) seems to inhibit bind-
ing in the presence of 1 but not in its absence. This suggests steric
clash between the two ligands when bound to Bcl-2. Overall, the
SAR for binding of biaryl acids in the presence of 1 is consistent
with the hypothesis that both ligands bind proximal to one another
in the Bcl-2 ligand–binding groove.
3HC
HO
20
O
HO
CH3
>1000
O
a
b
c
From NMR titration.
No compound 1.
Compound 1 at 500 lM.
chains of L116, V130, M112, and L134. The methane core positions
the hydrophilic substituents away from the hydrophobic pocket
towards the polar side chains of E111, D108, and E133 on the sur-
face of the protein. Although we show the hydroxyl of 1 pointing
towards E111, a binding orientation with this group interacting
with E133, obtained by rotating compound 1 by approximately
180° in the pocket, is also consistent with the NMR NOE data.
The diphenylmethane SAR is consistent with this NMR-derived
structure (Table 1). Removal of one (2) or both (3) chlorines weak-
ens binding by 4- and 10-fold, respectively, as would be expected
from reduced hydrophobic interactions with the groove. On the
other hand, the hydroxyl and dimethylpropylamine seem to be less
critical for binding. For example, replacing them with a methylpi-
perazine leads to a compound with equal affinity (compare 4 with
2). From the structure, it is clear that a variety of polar substitu-
tions from the diphenylmethane core would be able to interact
with the flexible polar side chains on the protein surface.
NMR structural studies of a ternary complex between a biaryl
acid and a diphenylmethane were undertaken to confirm the bind-
ing orientation. To identify a ternary complex for which strong
intermolecular NOEs could be measured, 14 ternary complexes
were evaluated using 2D 13C-filtered NOESY experiments. First,
eight different biaryl acids were prepared in complex with Bcl-2

A. M. Petros et al. / Bioorg. Med. Chem. Lett. 20 (2010) 6587–6591
6589
as it does in the absence of the biaryl acid (Fig. 1). Furthermore, the
biaryl binds further up the groove with its fluorophenyl pointing
towards the diphenylmethane. This orientation is consistent with
the biaryl SAR where meta and para substitutions seem to interact
favorably or clash with the diphenylmethane, respectively. Both
the structural studies (Fig. 2) and biaryl SAR (Table 2) indicated
that linking the two ligands could be achieved from the para or
meta position of the biaryl. Similarly, both the meta- and para-posi-
tions of the diphenylmethane could direct a linker towards the
biaryl.
To investigate this linking strategy, two-, three-, and four-atom
linkers were evaluated at the meta- and para-positions of each
fragment using the general synthetic strategy outlined in Scheme
1. Variations in the diphenyl methane core were explored through
routine reaction of appropriate beta-amino ketones (9) with
various aryl bromides (10) (Scheme 1) to yield intermediate diph-
enylmethane derivatives (11) with either a meta- or para-hydroxyl
group. The biaryl ester portion of the molecules (12) was then
assembled through simple Suzuki couplings of 4-(methoxycar-
bonyl)phenylboronic acid with different aryl bromides. A subse-
quent Mitsunobu coupling of the core moieties yielded target
compounds 13–29.
These compounds were then evaluated in a fluorescence polar-
ization competition assay, using a fluorescently-labeled peptide
from the anti-apoptotic protein Bad as a probe.7 These data are
summarized in Tables 3 and 4. Several linking strategies produced
compounds with inhibition constants in the low micromolar range
indicating a favorable linking geometry. Because meta/para and
para/meta two- and three-atom linkers yielded compounds with
similar potencies, we can conclude that some flexibility in binding
exists. This variability is likely enhanced by the absence of the phe-
nyl-substitutions that contribute to binding of the diphenylme-
thane core (Tables 1 and 2).
Figure 2. NMR-derived structure of a ternary complex of Bcl-2 with compounds 5
and 6. (A) Ribbon representation with color coding as described in Figure 1. (B)
Bcl-2 surface with coloring as described in Figure 1. Ligand atoms are colored with
carbon in orange, oxygen in red, nitrogen in blue, chlorine in green, and fluorine in
cyan.
and compound 1. Of these, the para-fluoro and para-methyl substi-
tuted biaryl acids yielded the highest quality 2D filtered NOE spec-
tra, based on NOE intensity and chemical dispersion of compound
aromatic protons. Next, these acids were evaluated in ternary com-
plexes with three other analogs of 1. A ternary complex between 5
and 6 yielded the highest quality spectra. From a three-dimen-
sional, 13C-edited, 12C-filtered NOESY spectrum, a total of 32 inter-
molecular NOEs between the ligands and Bcl-2 were observed. The
ligands were docked into the Bcl-2 groove using these intermolec-
ular NOEs and minimized using Xplor.6 As shown in Figure 2, the
diphenylmethane in the ternary complex binds to the same pocket
When hydrophobic substitutions were added back to the diph-
enylmethane core in the linked compounds, potency improved as
expected. Dichloro-substitutions improved the potency of the
linked compound by 10-fold (24 and 25, Table 4). The majority of
this improvement came from the chloro groups and not modifica-
tion of the dimethylamine in 24. This conclusion is consistent with
the observation that compounds with and without the polar mor-
pholino group differed by at most twofold in potency (25 and 26).
The same SAR was observed for the unlinked diphenylmethanes.
Addition of two chlorines improved potency by 10-fold (1 and 3,
Table 1), while the substitution of the dimethyl amine with a
R2
HO
R2
a
+
Br
R3
O
R1
R1
R3
9
10
11
CO2Me
COOH
R3
CO2Me
R2
c
HO
b
R1
B
X
HO
OH
R4
R4
12
13-29
Scheme 1. General synthesis of linked biaryl acid inhibitors. Reagents and conditions: (a) substituted 2-amino ketone (9), substituted bromobenzene (10), nBuLi, THF,
À78 °C. (b) Suzuki coupling: substituted bromobenzene, Pd(dppf), aq Cs2CO3, dioxane, 4-(methoxycarbonyl)phenylboronic acid. (c) Mitsunobu: DIAD, PPH3, THF, 11.

6590
A. M. Petros et al. / Bioorg. Med. Chem. Lett. 20 (2010) 6587–6591
Table 3
Table 4
Influence of linker length and geometry on Bcl-2 inhibition
Activity of linked compounds
No.
13
Structure
Kia
(
lM)
No.
Structure
Kia
(lM)
O
OH
OH
N
O
OH
OH
OH
5.0
N
N
O
O
24
34
O
OH
O
N
14
15
20
O
OH
O
25
26
27
0.22
0.46
0.20
N
Cl
Cl
OH
5.2
4.6
3.4
O
O
O
HO
OH
OH
OH
OH
N
N
O
O
16
17
Cl
O
Cl
O
OH
O
HO
O
N
OH
O
OH
O
O
Br
Cl
OH
18
19
>1
OH
OH
O
N
N
O
O
O
O
HO
O
N
OH
O
O
5.8
8.0
OH
28
29
0.11
0.04
Cl
OH
O
HO
20
21
OH
OH
N
OH
N
N
O
O
O
Cl
O
83
HO
a
Fluoresence polarization assay.
O
OH
O
OH
OH
OH
methyl group. The biaryl also moves closer to V130 in the linked
compound. These structural results explain the linker SAR that
was observed (Table 3). The ability of long, narrow hydrophobic
compounds to slide within the Bcl-2 hydrophobic groove is consis-
tent with multiple linker lengths and linker geometries being com-
patible with this pocket. In addition to explaining the SAR of
compound linking, further improvement in potency could be ob-
tained by increasing the size of the para-substituent on the distal
ring of the diphenylmethane (Fig. 3). Consistent with this predic-
22
23
12
>1
N
N
O
O
O
OH
O
a
Fluoresence polarization assay.
tion, a bulky t-butyl improved potency further to 0.11 lM (28).
The structure of 27 also indicated that additional interactions
with the Bcl-2 groove could be accessed from the biaryl. A limited
number of compounds were made to explore this possibility, but
none showed improved affinity over the unsubstituted analogs.
In addition, acid replacements that preserved a negative charge,
such as a tetrazole or acylsulfonamide, were equipotent. However,
uncharged compounds where the carboxyl was replaced with an
alcohol, nitrile, or methyl ester were inactive. In fact, this SAR is
consistent with the SAR of biaryl-containing inhibitors of Bcl-xL
in that an unsubstituted biaryl acid seems to optimally fill this re-
gion of the BH3-binding groove.4 Finally, the binding affinity of the
piperizine (2 and 4) had little effect on affinity. In order to further
improve potency using structure-based design, binding of 27 was
studied by NMR. A total of 46 protein–ligand NOEs were used to
dock 27 into the groove of Bcl-2 ( Fig. 3). The biaryl and diph-
enylmethane components were found to bind to the same pockets
as in the ternary complex. Unexpectedly, however, the binding
positions in these pockets changed dramatically. In the presence
of the linked compound, the side chain of M112 moves away from
V130 creating a deeper pocket (compare Figs. 2A and 3A). This is
accompanied by reorientation of the diphenylmethane (Fig. 3B)
such that one chlorophenyl fills the pocket vacated by the M112
linked compound was improved to 0.04 lM by replacing the

A. M. Petros et al. / Bioorg. Med. Chem. Lett. 20 (2010) 6587–6591
6591
rationale for this difference may be the presence in Bcl-xL of a leu-
cine residue at position 112 as opposed to the methionine residue
in Bcl-2. The sidechain of this methionine flips out of the groove in
order to accommodate the chlorophenyl ring of 29. Such sidechain
movement is not possible for the more rigid leucine of Bcl-xL. Sim-
ilarly, greater than 1000-fold specificity was observed for Bcl-2
versus Bcl-w, greater than 100-fold versus Bcl-B, about 58-fold ver-
sus A1, and about 28-fold versus Mcl-1. This is in contrast to the
specificity profile of ABT-737 that binds with sub-nanomolar affin-
ity to Bcl-2, Bcl-xL, and Bcl-w, but with micromolar affinity to Mcl-
1, Bcl-B, and A1.8 To date, this is the first potent small molecule
inhibitor to be described which is highly selective for Bcl-2.9
The potency of 29 in killing Bcl-2 dependent tumor cells was
evaluated using a Bcl-2 dependent follicular lymphoma cell line
(FL.12).8 An EC50 of 16
based on our experience with ABT-737 and ABT-263 analogs, is
consistent with a Ki of 0.04 M. Given this cellular potency it is un-
lM was determined for this cell line, which
l
likely that these compounds would be active in vivo. However,
they may serve as a useful starting point for developing a potent,
Bcl-2 selective, anti-tumor agent. Further gains in potency could
potentially be achieved by replacing the carboxylate moiety with
an acylsulfonamide and then building off of this acylsulfonamide
to fill unoccupied portions of the groove in a manner similar to
ABT-737.4
Acknowledgment
Figure 3. NMR-derived structure of Bcl-2 in complex with compound 27. (A)
Ribbon representation with color coding for Bcl-2 as described in Figure 1. Atoms of
the ligand are colored with carbon in orange, oxygen in red, nitrogen in blue,
chlorine in magenta, and bromine in brown. (B) Surface representation including 1
superimposed on 27. Bcl-2 is colored as described in Figure 2B. For 27, carbon atoms
are in orange, oxygens in red, nitrogen in blue, chlorine in magenta, and bromine in
brown. For 1, carbon and atoms are in magenta, oxygen in red, and nitrogen in blue.
Thanks to Steve Swann for assistance in preparing this
manuscript.
Supplementary data
Supplementary data associated with this article (protein
expression and procedures for NMR structural studies) can be
found, in the online version, at doi:10.1016/j.bmcl.2010.09.033.
morpholino group of 28 with a pyrrolidine (29). A structural expla-
nation for this gain in potency is not apparent from the structure of
27 (Fig. 3). Polar interactions with surface amino acids E111, Q115,
or E133, may be important. However, NMR structural studies of a
pure enantiomer would need to be pursued to understand the de-
tails of the pyrrolidine interaction.
The theoretical gain in potency for the linking of two com-
pounds is simply the product of the dissociation constants for
the individual compounds plus a contribution from the linker it-
self, the latter of which is difficult to quantify. Thus optimal linking
References and notes
1. Hanahan, D.; Weinberg, R. A. Cell 2000, 100, 57.
2. Tse, C.; Shoemaker, A. R.; Adickes, J.; Anderson, M. G.; Chen, J.; Jin, S.; Johnson, E.
F.; Marsh, K. C.; Mitten, M. J.; Nimmer, P.; Roberts, L.; Tahir, S. K.; Xiao, Y.; Yang,
X.; Zhang, H.; Fesik, S.; Rosenberg, S. H.; Elmore, S. W. Cancer Res. 2008, 68, 3421.
3. Zhang, H.; Nimmer, P. M.; Tahir, S. K.; Chen, J.; Fryer, R. M.; Hahn, K. R.; Iciek, L.
A.; Morgan, S. J.; Nasarre, M. C.; Nelson, R.; Preusser, L. C.; Reinhart, G. A.; Smith,
M. L.; Rosenberg, S. H.; Elmore, S. W.; Tse, C. Cell Death Differ. 2007, 14, 943.
4. Petros, A. M.; Dinges, J.; Augeri, D. J.; Baumeister, S. A.; Betebenner, D. A.; Bures,
M. G.; Elmore, S. W.; Hajduk, P. J.; Joseph, M. K.; Landis, S. K.; Nettesheim, D. G.;
Rosenberg, S. H.; Shen, W.; Thomas, S.; Wang, X.; Zanze, I.; Zhang, H.; Fesik, S. W.
J. Med. Chem. 2006, 49, 656.
5. Jahnke, W. ChemBioChem 2002, 3, 167.
6. Brunger, A. T. X-PLOR—a system for X-ray crystallography and NMR; Yale
University Press: New Haven, 1987.
7. Zhang, H.; Nimmer, P.; Rosenberg, S. H.; Ng, S. C.; Joseph, M. Anal. Biochem. 2002,
307, 70.
of 1 (KD = 20 lM) with 6 (KD = 400 lM) would give a KD of at least
8 nM. The linked compound 25 most closely represents the original
ligands and with an inhibition constant of 220 nM is about 27-fold
weaker than the theoretical value. This difference most likely
arises from the inability to maintain the optimal position of the
individual compounds given the constraint of finite bond lengths
and finite angles in the linker.
8. Oltersdorf, T.; Elmore, S. W.; Shoemaker, A. R.; Armstrong, R. C.; Augeri, D. J.;
Belli, B. A.; Bruncko, M.; Deckwerth, T. L.; Dinges, J.; Hajduk, P. J.; Joseph, M. K.;
Kitada, S.; Korsmeyer, S. J.; Kunzer, A. R.; Letai, A.; Li, C.; Mitten, M. J.;
Nettesheim, D. G.; Ng, S.; Nimmer, P. M.; O’Connor, J. M.; Oleksijew, A.; Petros, A.
M.; Reed, J. C.; Shen, W.; Tahir, S. K.; Thompson, C. B.; Tomaselli, K. J.; Wang, B.;
Wendt, M. D.; Zhang, H.; Fesik, S. W.; Rosenberg, S. H. Nature 2005, 435, 677.
9. Porter, J.; Payne, A.; de Candole, B.; Ford, D.; Hutchinson, B.; Trevitt, G.; Turner,
J.; Edwards, C.; Watkins, C.; Whitcombe, I.; Davis, J. Bioorg. Med. Chem. Lett. 2009,
19, 230.
NMR-based fragment binding studies of the dichlorodiphenyl
methane indicated that this core is 20-fold specific for Bcl-2 over
Bcl-xL (Table 1). To determine whether this specificity was pre-
served when linked to the biaryl acid, compound 29 was tested
against a panel of anti-apoptotic Bcl-2 family proteins, including
Bcl-xL, in a fluorescence polarization assay.7 Consistent with the re-
sults for binding of the diphenylmethane fragment, 29 is more than
1000-fold more potent against Bcl-2 versus Bcl-xL. A structural