Discovery of small-molecule inhibitors of Bcl-2 through structure-based computer screening

Article abstract of DOI:10.1021/jm010016f

Bcl-2 belongs to a growing family of proteins which regulates programmed cell death (apoptosis). Overexpression of Bcl-2 has been observed in 70% of breast cancer, 30-60% of prostate cancer, 80% of B-cell lymphomas, 90% of colorectal adenocarcinomas, and

Full text of DOI:10.1021/jm010016f

J . Med. Chem. 2001, 44, 4313-4324
4313
Articles
Discover y of Sm a ll-Molecu le In h ibitor s of Bcl-2 th r ou gh Str u ctu r e-Ba sed
Com p u ter Scr een in g
Istvan J . Enyedy,†,‡,§ Yan Ling,†,‡ Kassoum Nacro,† York Tomita,† Xihan Wu,†, Yeyu Cao,†, Ribo Guo,†
Bihua Li,† Xiaofeng Zhu,† Ying Huang,† Ya-Qiu Long,| Peter P. Roller,| Dajun Yang,*,† and Shaomeng Wang*,†,
Structural Biology and Cancer Drug Discovery Program, Lombardi Cancer Center and Department of Oncology,
Georgetown University Medical Center, 3970 Reservoir Road, Washington, D.C. 20007, and Laboratory of Medicinal
Chemistry, National Cancer Institute, FCRDC, Building 376, Room 207, Fredrick, Maryland 21702
Received J anuary 10, 2001
Bcl-2 belongs to a growing family of proteins which regulates programmed cell death (apoptosis).
Overexpression of Bcl-2 has been observed in 70% of breast cancer, 30-60% of prostate cancer,
80% of B-cell lymphomas, 90% of colorectal adenocarcinomas, and many other forms of cancer.
Thereby, Bcl-2 is an attractive new anti-cancer target. Herein, we describe the discovery of
novel classes of small-molecule inhibitors targeted at the BH3 binding pocket in Bcl-2. The
three-dimensional (3D) structure of Bcl-2 has been modeled on the basis of a high-resolution
NMR solution structure of Bcl-XL, which shares a high sequence homology with Bcl-2. A
structure-based computer screening approach has been employed to search the National Cancer
Institute 3D database of 206 876 organic compounds to identify potential Bcl-2 small-molecule
inhibitors that bind to the BH3 binding site of Bcl-2. These potential Bcl-2 small-molecule
inhibitors were first tested in an in vitro binding assay for their potency in inhibition of the
binding of a Bak BH3 peptide to Bcl-2. Thirty-five potential inhibitors were tested in this binding
assay, and seven of them were found to have a binding affinity (IC50 value) from 1.6 to 14.0
µM. The anti-proliferative activity of these seven active compounds has been tested using a
human myeloid leukemia cell line, HL-60, which expresses the highest level of Bcl-2 protein
among all the cancer cell lines examined. Compound 6 was the most potent compound and
had an IC50 value of 4 µM in inhibition of cell growth using the 3-(4,5-dimethylthiazol-2-yl)-
2,5-diphenyltetrazolium bromide assay. Five other compounds had moderate activity in
inhibition of cell growth. Compound 6 was further evaluated for its ability to induce apoptosis
in cancer cells. It was found that 6 induces apoptosis in cancer cells with high Bcl-2 expression
and its potency correlates with the Bcl-2 expression level in cancer cells. Furthermore, using
NMR methods, we conclusively demonstrated that 6 binds to the BH3 binding site in Bcl-XL.
Our results showed that small-molecule inhibitors of Bcl-2 such as 6 modulate the biological
function of Bcl-2, and induce apoptosis in cancer cells with high Bcl-2 expression, while they
have little effect on cancer cells with low or undetectable levels of Bcl-2 expression. Therefore,
compound 6 can be used as a valuable pharmacological tool to elucidate the function of Bcl-2
and also serves as a novel lead compound for further design and optimization. Our results
suggest that the structure-based computer screening strategy employed in the study is effective
for identifying novel, structurally diverse, nonpeptide small-molecule inhibitors that target
the BH3 binding site of Bcl-2.
In tr od u ction
apoptotic molecules such as Bax, Bak, Bid, and Bad.1-3
These molecules play crucial roles in regulating apop-
tosis. A simplified Bcl-2-mediated apoptosis pathway is
provided in Figure 1. The ability of Bcl-2 and Bcl-XL to
form heterodimmers with a number of pro-apoptotic
proteins such as Bak, Bad, and Bax is believed to play
a crucial role for the anti-apoptotic function of Bcl-2 and
Bcl-XL, although the precise molecular mechanism of
action of Bcl-2 remains far from completely understood.
Apoptosis, or programmed cell death, is important for
normal development, host defense, and suppression of
oncogenesis, and faulty regulation of apoptosis has been
implicated in cancer and many other human diseases.1-3
The Bcl-2 family of proteins now includes both anti-
apoptotic molecules such as Bcl-2 and Bcl-XL and pro-
* Address correspondence and requests for reprints to Shaomeng
Wang [phone (734) 615-0362; fax (734) 647-9647; e-mail shaomeng@
umich.edu] or Dajun Yang [phone (202) 687-0265; fax (202) 687-7505;
e-mail yangd@georgetown.edu].
As an anti-apoptotic member of this family, Bcl-2
contributes to cancer cell progression by preventing
normal cell turnover caused by physiological cell death
mechanisms. Overexpression of Bcl-2 has been observed
in 70% of breast cancer, 30-60% of prostate cancer, 80%
of B-cell lymphomas, 90% of colorectal adenocarcinomas,
and many other forms of cancer.4 The expression levels
† Georgetown University Medical Center.
‡ These authors contributed equally to this work.
§ Present address: Bayer Pharmaceutical Division, 400 Morgan
Lane, West Haven, CT 06516.
| National Cancer Institute.
Present address: 3316 CCGC, University of Michigan Cancer
Center, 1500 E. Medical Center, Ann Arbor, MI 48109.
10.1021/jm010016f CCC: $20.00 © 2001 American Chemical Society
Published on Web 11/01/2001

4314 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 25
Enyedy et al.
Ch a r t 1. Small-Molecule Inhibitors of Bcl-2 (or Bcl-XL)
Reported to Date
Huang’s laboratory.15 Using a computerized structure-
based database screening strategy, Dr. Huang and his
colleagues screened the Available Chemical Directory
of more than 200 000 small organic compounds and
discovered one class of small organic molecules (called
HA14-1) that bind to the BH3 binding site in Bcl-2 (1,
Chart 1). HA14-1 effectively induced apoptosis in hu-
man acute myeloid leukemia (HL-60) cells overexpress-
ing Bcl-2 protein. Although the potency of this small
organic inhibitor is only moderate (IC50 ) 9 µM in
binding assay), this study clearly showed that it is
possible to design a nonpeptide, small-molecule inhibitor
of Bcl-2.
F igu r e 1. A simplified Bcl-2-mediated signaling pathway of
programmed cell death.
Very recently, two independent papers reported the
discovery of three new classes of small-molecule inhibi-
tors of Bcl-2 or Bcl-XL.16,17 Using a high-throughput
screening assay based upon fluorescene polarization, Dr.
Yuan and colleagues screened a library of 16 320
chemicals and identified two different classes of small-
molecule inhibitors of Bcl-XL. The most potent repre-
sentative compounds are shown in Chart 1 (compounds
2 and 3).16 Dr. Hockenbery and colleagues identified
antimycin A (compound 4 in Chart 1), an antibiotic, as
a small-molecule inhibitor of Bcl-2/Bcl-XL with an IC50
value of 2 µM.17 Taken together, these three studies
showed that a small organic molecule inhibitor that
binds to the BH3 domain in Bcl-2/Bcl-XL can inhibit
their anti-apoptotic function, which in turn induces
apoptosis in cells with Bcl-2/Bcl-XL overexpression.
These exciting discoveries strongly indicated that it is
possible to design small-molecule inhibitors that block
the interactions between Bcl-2/Bcl-XL and pro-apoptotic
proteins (peptides) such as Bak, Bad, and Bax, and
inhibit the biological function of Bcl-2/Bcl-XL.
In recent years, the structure-based approach has
become very powerful for the discovery of novel lead
compounds and for lead optimization. Herein, we report
the discovery of several classes of novel small-molecule
inhibitors of Bcl-2 through structure-based computer
screening of the National Cancer Institute three-
dimensional (3D) database and the characterization in
biochemical and cellular assays. Furthermore, we have
investigated the interactions of a promising small-
molecule inhibitor (6) with Bcl-XL through NMR meth-
ods and conclusively confirmed that 6 binds to the BH3
binding site in Bcl-XL.
of Bcl-2 proteins also correlate with resistance to a wide
spectrum of chemotherapeutic drugs and γ-radiation
therapy.5-7 Biological approaches targeted at Bcl-2
using anti-sense oligonucleotides or single-chain anti-
bodies have been shown to enhance tumor cell chemosen-
sitivity.8-13 Recently, cell-permeable Bcl-2 binding pep-
tides containing a fatty acid have been shown to induce
apoptosis in vitro and have in vivo activity in suppress-
ing human myeloid leukemia growth in severe combined
immunodeficient mice.14 Therefore, Bcl-2 represents an
attractive target for the development of a novel therapy
for the treatment of many forms of cancer.
Certain possible limitations are associated with large-
molecule and peptide approaches, including poor oral
availability, poor in vivo stability, and/or high cost. We
are therefore interested in the discovery and design of
nonpeptide, small-molecule inhibitors that target Bcl-2
and modulate its anti-apoptotic function. A potent, cell-
permeable Bcl-2 small-molecule inhibitor can serve as
a valuable pharmacological tool to further elucidate the
biological function of Bcl-2 in vitro and in vivo. Fur-
thermore, such an inhibitor may have great therapeutic
potential as a novel therapy for the treatment of many
forms of cancer, either alone or in combination with
conventional chemotherapy or radiation therapy.
It has been traditionally difficult to discover small-
molecule inhibitors to block protein-protein interac-
tions. Thus, despite the successful development of
several screening assays, discoveries of small-molecule
inhibitors of Bcl-2 remained elusive for several years.
However, more recently, three independent groups
reported their discovery of small-molecule inhibitors of
Bcl-2/Bcl-XL. The very first report on the discovery of a
small-molecule inhibitor of Bcl-2 was contributed by
Resu lts a n d Discu ssion
Hom ology Mod elin g of th e Bcl-2 3D Str u ctu r e.
Structure-based database searching requires an ac-

Small-Molecule Inhibitors of Bcl-2
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 25 4315
F igu r e 2. Sequence alignment between Bcl-2 and Bcl-XL.
curate 3D structure of Bcl-2. The experimental 3D
structures of Bcl-2 are now available.18 However, when
we started the project, the experimental Bcl-2 structure
was not determined. Fortunately, high-resolution ex-
19-21
perimental 3D structures of Bcl-XL
alone and in
complex with a Bak BH3 (Bcl-2 homology domain 3)
peptide21 have been determined. Bcl-2 and Bcl-XL share
a high degree of homology in their amino acid sequences
(45% identity and 56% similarity). It has been demon-
strated that when there is a sequence identity of more
than 30% between the target protein (Bcl-2) and the
template protein (Bcl-XL), current computational homol-
ogy modeling methods, such as MODELLER,24 can
provide an accurate 3D structure for the target pro-
tein.25 Therefore, computational homology modeling was
employed to model the 3D structure of Bcl-2 (the target
protein) on the basis of the experimental 3D structural
coordinates of Bcl-XL (the template protein) in this
study.
The sequence alignment obtained using the BLAST
program22,23 between Bcl-2 and Bcl-XL is provided in
Figure 2, which was used in our homology modeling.
Since the Bak BH3 peptide binds to both Bcl-2 and Bcl-
XL with good affinities,14,15 the 3D structure of Bcl-2 in
complex with the Bak BH3 peptide was modeled on the
basis of the experimental NMR structure of Bcl-XL in
complex with the Bak BH3 peptide.21 Using the MOD-
ELLER program,24-26 10 different models were gener-
ated. It was found that these 10 models were very
similar, with a root-mean-square deviation (RMSD)
within 1 Å for all the non-hydrogen atoms of residues
that form the BH3 binding site. To further refine the
side chain conformations, the modeled 3D complex
structure was extensively simulated through molecular
dynamics (MD) simulation in explicit water for 3 ns.
Comparison of our modeled Bcl-2 structure with the
recently published experimental high-resolution Bcl-2
NMR structure18 showed that they are essentially the
same with respect to both the overall fold and binding
site conformation. The RMSD is 1.0 Å for all the non-
hydrogen atoms of residues that form the BH3 binding
site between the NMR structure18 and our modeled
structure. Thus, computational homology modeling
provided us with an accurate 3D structure of Bcl-2 for
our structure-based 3D database search. The refined
structure of Bcl-2 in complex with the Bak BH3 peptide
is depicted in Figure 3.
F igu r e 3. Modeled 3D structure of Bcl-2 in complex with the
Bak BH3 domain based upon the experimental structure of
Bcl-XL in complex with the Bak BH3 peptide (PDB code 1BXL).
(A) Ribbon representation of the overall Bcl-2 structure in
complex with the Bak BH3 peptide. (B). Detailed representa-
tion of the binding site. The carbon atoms in the Bak BH3
peptide are shown in magenta, the carbon atoms in the Bcl-2
protein in green, the oxygen atoms in red, and the nitrogen
atoms in blue.
Defin in g th e “Bin d in g” Site for Str u ctu r e-Ba sed
Com p u ter Scr een in g. The experimental structure of
Bcl-XL in complex with the Bak BH3 peptide21 as well
as our modeled structure of Bcl-2 in complex with the

4316 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 25
Enyedy et al.
Ta ble 1. Chemical Structures and Binding Affinities of Active Compounds
Bak BH3 peptide provided us with a structural basis
for their interactions between these proteins.
repository has physical samples of about 60% of the
registered compounds. Recently, analysis of several
large chemical databases showed that the NCI database
has by far the highest number of compounds that are
unique to it.32 Approximately 200 000 of the NCI
structures are not found in any of the other six analyzed
databases.32 Therefore, the NCI open database provides
a large number of unique synthetic compounds and
natural products and is an excellent resource for drug
lead discovery.
The Bak BH3 peptide binds to the largely hydrophobic
binding pocket formed by the BH1, BH2, and BH3
domains in Bcl-2 as well as in Bcl-XL. This binding
pocket in Bcl-2 appears to be essential for its anti-
apoptotic function since mutations at this site abolish
its function.28,29 Furthermore, synthetic, cell-permeable
peptides binding to this pocket in Bcl-2 induce apoptosis
in vitro and have in vivo activity in suppressing human
myeloid leukemia growth.14,30 Nonpeptide small-mol-
ecule Bcl-2 inhibitors 1-4 that bind to this pocket have
been shown to compete with the Bak BH3 peptide in
binding to Bcl-2 in vitro and induce apoptosis.15 Taken
together, these data provide strong evidence that the
hydrophobic binding pocket formed by the BH1, BH2,
and BH3 domains of Bcl-2 is an attractive site for the
design of nonpeptide, small-molecule inhibitors of Bcl-2
for modulating its function. Accordingly, all residues
that are within 8 Å from the Bak peptide in the Bak-
Bcl-2 complex were selected and used to define the bind-
ing site for identification of potential Bcl-2 inhibitors
in our subsequent structure-based computer screening.
Str u ctu r e-Ba sed Com p u ter Scr een in g of th e
Na tion a l Ca n cer In stitu te 3D Da ta ba se To Id en tify
P oten tia l Bcl-2 Sm a ll Molecu les. Since 1955, the
National Cancer Institute (NCI) at the National Insti-
tutes of Health has conducted extensive testing of
materials for possible activity against different forms
of cancer. Most of the substances tested have been pure
organic compounds. The program examined an extraor-
dinarily eclectic assembly of organic structures. Cur-
rently, more than 465 000 compounds have been tested.31
Of these compounds, about half (206 876 compounds)
are classified as “open” compounds, whose structures
and biological data can be accessed by the public.31
Because samples of compounds were needed for testing,
it was necessary to develop a large acquisition effort and
a repository, both of which are still functioning today.
Continual scanning of the chemistry literature allows
identification of compounds that are novel and of
interest to the program, and the authors are approached
for a sample, typically under 100 mg. Currently, the NCI
Using the modeled 3D structure of Bcl-2, we searched
the NCI 3D database of 206 876 small molecules using
the program DOCK.33,34 In the database search, con-
formational flexibility of the small molecules was taken
into account. The small molecules were ranked accord-
ing to their scores as calculated using the energy scoring
function in the DOCK program. The top 500 candidate
small molecules with the best scores were considered
as potential Bcl-2 inhibitors. Only nonpeptide molecules
were selected for testing. In the first batch of biological
testing, samples of 80 compounds were requested from
the NCI and 35 compounds were available for testing.
Resu lts of th e Bin d in g Assa y. We have used a
sensitive and quantitative in vitro fluorescence polariza-
tion (FP) based binding assay.15 The basic principle
behind this assay is competition: a fluorescent peptide
tracer (Flu-Bak-BH3) and a nonfluorescent small-
molecule inhibitor compete for binding to the target
protein (Bcl-2). In a reaction mixture containing no
small-molecule inhibitor, the fluorescent tracer Flu-Bak-
BH3 is bound to the target protein Bcl-2 and the
emission signal is polarized; however, in a reaction
mixture containing a small-molecule inhibitor, the
tracer is displaced by the small-molecule inhibitor from
the target protein and the emission signal becomes
depolarized. The resulting change in the FP signal is
directly related to the inhibitory activity of the small-
molecule inhibitor. Using this method, a binding affinity
of 0.3 µM (IC50) was obtained for Bak-BH3 peptide
binding to the Bcl-2, which is consistent with the value
reported in the literature.15
The binding affinity of the 35 candidate small mol-
ecules was tested initially at a dose of 100 µM in this

Small-Molecule Inhibitors of Bcl-2
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 25 4317
Ta ble 2. Chemical Structures of Inactive Compounds
binding assay, of which 7 compounds showed inhibitory
activity of more than 50% at the initial 100 µM dose
level and were classified as active. The other 28 com-
pounds had less than 50% inhibition at the 100 µM dose
level and were classified as inactive. Further dose-
dependent binding experiments were carried out for
these seven active compounds to determine their IC50
values. All seven active compounds displayed a dose-
dependent inhibition of Bak peptide binding to Bcl-2.
The chemical structures and IC50 values of these seven
active compounds are provided in Table 1. The chemical
structures of the 28 inactive compounds are provided
in Table 2.
are also different from those of the previously reported
Bcl-2 inhibitors 1-4.
A previous study showed that screening of 16 320
compounds using a high-throughput screening ap-
proach has led to the discovery of two chemical classes
of small-molecule inhibitors of Bcl-2/Bcl-XL.16 In our
study, testing of 35 compounds selected from structure-
based 3D database screening allowed the discovery of
7 classes of small-molecule inhibitors of Bcl-2. Our
results thus suggested that the structure-based com-
puter screening strategy employed in this study is
effective in identifying novel small-molecule inhibitors
of Bcl-2.
As can be seen from Table 1, all of these seven active
compounds have an IC50 value better than 15 µM.
Compound 7 is the most potent compound in the binding
assay, with an IC50 value of 1.6 µM. The other six
compounds have an IC50 value from 5.8 to 14.0 µM. It
is of note that all of these seven active compounds
belong to different chemical classes and their structures
In h ibition of Cell Via bility a n d Gr ow th in Ca n -
cer Cells w ith High Bcl-2 Exp r ession . Binding
experiments showed that these seven active compounds
are capable of competing with the Bak BH3 peptide in
binding to Bcl-2 in vitro. Since we are interested in
identifying cell-permeable Bcl-2 small-molecule inhibi-
tors, we have tested the inhibitory activity of these

4318 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 25
Enyedy et al.
among all the cancer cell lines examined in our labo-
ratories (Figure 5).
Using the trypan blue exclusion assay, these seven
compounds were screened for their activity in inhibition
of cell viability. All seven compounds except for com-
pound 10 had an IC50 value better than 50 µM.
Compound 6 is the most potent compound in the cellular
assay, with an IC50 value of 10 µM, as shown in Figure
4. We have further tested compound 6 for its ability to
inhibit cell growth. In the MTT assay where cells were
treated for 4 days, 6 showed a potent inhibition in cell
growth with an IC50 value of 4 µM. Because of its potent
cellular activity, 6 was used in subsequent biological
experiments.
Ch em ica l Str u ctu r e Ver ifica tion of th e Active
Com p ou n d s. One problem we and other investigators
have encountered when using the NCI database is that
samples of some compounds from the NCI repository
are not pure, or may have decomposed over the years,
or in rare cases the structures are simply wrong. Care
was taken to verify the structures of the small-molecule
inhibitors of Bcl-2 shown in Table 1.
F igu r e 4. Effects of compound 6 on cell viability in cancer
cell lines with different levels of Bcl-2 protein expression.
First, we have performed mass spectral analyses on
the samples of these seven compounds. The results are
shown in Table 3. As can be seen, five compounds were
found to have an observed mass in agreement with the
calculated mass based upon the chemical structures.
However, two compounds, 9 and 10, were found to have
an observed mass not in agreement with the calculated
mass, suggesting that the chemical structures for these
two compounds as recorded in the NCI database are
incorrect.
Next, we obtained 1H NMR spectra for these five
active compounds whose structures are consistent with
the mass spectral data. We were unable to obtain the
1H NMR spectrum for compound 5 because it is in-
seven active compounds on cell viability and prolifera-
tion using two different assays. First, the trypan blue
exclusion method was used to determine the effect of
an inhibitor on cell viability in which cells were treated
with the inhibitor for 24 h. Second, the 3-(4,5-dimeth-
ylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT)
assay was used to determine the activity of an inhibitor
on cell proliferation where cells were treated for 4 days.
It is important to keep in mind that an active compound
in the binding assay could fail to show any cellular
activity simply because of its poor cellular permeability.
We first tested these seven active compounds using the
HL-60 cell line. HL-60 is a human myeloid leukemia
cell line and expresses the highest level of Bcl-2 protein
Ta ble 3. Results of Mass Spectral Analysis of the Active Compounds in Table 1

Small-Molecule Inhibitors of Bcl-2
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 25 4319
Sch em e 1. Synthetic Scheme of Compound 6
F igu r e 5. Level of Bcl-2 protein expression in cancer cell lines
as detected by Western blotting. Human myeloid leukemia cell
line HL-60 expresses the highest level of Bcl-2 protein among
all the cell lines examined. MDA-231 (subclone 2LMP derived
from our laboratory) expresses a high level of Bcl-2; T47D
expresses a very low but detectable level of Bcl-2, whereas
MDA-453 does not express detectable Bcl-2.
myeloid leukemia cell line HL-60 expresses the highest
level of Bcl-2 protein among all the cell lines examined.
MDA-231 (subclone 2LMP derived from our laboratory)
expresses a high level of Bcl-2; T47D expresses a very
low but detectable level of Bcl-2, whereas MDA-453 does
not express detectable Bcl-2. Accordingly, we chose
MDA-231 and HL-60 cell lines with high Bcl-2 expres-
sion as the positive cells and MDA-453 and T47D as
the negative control cells to test the specificity of 6 in
inhibition of cell viability using the trypan blue exclu-
sion assay, as well as in our apoptosis experiments. The
results are shown in Figure 4.
As can be seen, in HL-60 and MDA-231 cells, two cell
lines with high Bcl-2 expression, 6 displays a dose-
dependent inhibition of cell viability with IC50 values
of 10 and 15 µM, respectively. In T47D cells, which have
low but some level of Bcl-2 expression, 6 has a minimal
activity at 25 µM. Furthermore, in MDA-453 cells, which
have no detectable level of Bcl-2, 6 has no effect on cell
viability at 25 µM. Therefore, the ability of 6 in inhibi-
tion of cell viability correlates with the Bcl-2 protein
expression level in these cancer cells.
In d u ction of Ap op tosis. One central hypothesis we
wanted to test is that the binding pocket formed by the
BH1, BH2, and BH3 domains in Bcl-2 is essential for
its anti-apoptotic function, and binding of a small-
molecule inhibitor such as 6 to this pocket may block
the anti-apoptotic function of Bcl-2 and induce apoptosis
in cells with Bcl-2 protein overexpression. To test this
hypothesis, we evaluated the ability and very impor-
tantly the specificity of 6 in inducing apoptosis in cancer
cells with a high or low level of Bcl-2 expression.
We used the Annexin-V flow cytometry assay to
obtain a quantitative assessment on the ability of 6 in
induction of apoptosis in HL-60 and MDA-231 cells.
MDA-231 cells treated with 0 (untreated), 5, and 10 µM
6 for 24 h exhibited 0%, 13%, and 20.0% apoptotic cells,
respectively, while HL-60 cells treated with 0, 5, 10, and
20 µM 6 for 24 h had 0%, 24%, 31%, and 67% apoptotic
cells, respectively. Therefore, 6 induced apoptosis in a
dose-dependent manner in MDA-231 and HL-60 cell
lines with Bcl-2 protein overexpression.
Taken together, our results showed that 6 specifically
induces apoptosis in a dose-dependent manner in cancer
cells with high Blc-2 expression while it has little effect
on cancer cells with low or little Bcl-2 expression. Our
data suggested that overexpression of Bcl-2 may be
necessary to maintain the transformed state in these
cancer cells and that blocking the Bcl-2 anti-apoptotic
function with a small-molecule inhibitor such as 6 will
induce apoptosis in these cancer cells.
soluble in either DMSO or chloroform. The 1H NMR
spectra for compounds 6-8 and 11 are consistent with
their chemical structures.
Compound 6 potently inhibits the cell viability (Figure
4) and growth, thus representing a novel class of cell-
permeable small-molecule inhibitors of Bcl-2 and a
promising lead compound for further design and chemi-
cal modifications. Accordingly, we have synthesized this
compound to confirm conclusively its chemical structure.
Ch em ica l Syn th esis of Com p ou n d 6. The general
synthesis procedure of compound 6 is illustrated in
Scheme 1, according to a method developed by Bown
and Greene.35 Briefly, commercially available 3-methyl-
4-nitroanisole (40) was dimerized to give compound 41
in 64% yield by oxidation with tBuOK and ether in
DMSO. The diamino derivative 42 was obtained in
excellent yield by reducing 41 in the presence of hydra-
zine hydrate and a catalytic amount of palladium.
Oxidation of 42 with hydrogen peroxide and sodium
tungstate dihydrate in ethanol gave compound 6 in 61%
yield. It was previously shown that compound 6 was in
tautomeric equilibrium (1:3.68 at 295.70 K in chloro-
form).35 On the basis of our own NMR data, we
estimated that the ratio between 6a and 6b is 1:3.69-
3.79 in chloroform or 1:2.52 in DMSO at room temper-
1
ature, respectively. Mass spectral and H NMR analyses
of the synthesized sample of compound 6 showed that
it is identical to that acquired from the NCI repository.
We further confirmed that the two samples of compound
6 have the same affinity for binding to Bcl-2 in the FP-
based binding assay and have the same cellular activity
in the HL-60 cell line. Accordingly, our subsequent
studies on this compound were performed using the
sample synthesized in our laboratory.
Sp ecificity of 6 in In h ibition of Cell Via bility. We
showed that Bcl-2 inhibitor 6 blocks the binding of the
Bak BH3 peptide to Bcl-2 in vitro and potently inhibits
cell viability and growth in the HL-60 cell line with high
Bcl-2 expression. Next we further tested the specificity
of Bcl-2 inhibitors such as 6 in cancer cell lines with
different levels of Bcl-2 expression to investigate whether
6 can achieve specific inhibition of cell viability and
whether its cellular activity is dependent upon the level
of Bcl-2 protein in cancer cells.
For this purpose, we have first characterized Bcl-2
protein expression in human breast and other cancer
cell lines, and the results are shown in Figure 5. Human
Con fir m a tion of th e Bin d in g of 6 to Bcl-XL by
NMR. Using the FP-based assay, we showed that 6

4320 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 25
Enyedy et al.
F igu r e 6. Superposition of 15N HSQC spectra of 15N-labeled
Bcl-XL protein in free form (black) and in complex with 6 (red).
binds to Bcl-2. However, the in vitro FP-based method
simply shows that addition of a small-molecule inhibitor
reduces the intensity of FP. The most straightforward
interpretation is that the small-molecule inhibitor binds
to Bcl-2 and displaces the binding of the fluorescence-
labeled Bak BH3 peptide. However, it is also possible
that addition of the small-molecule inhibitor simply
causes the unfolding of the protein and thus reduces
the binding of the fluorescence-labeled Bak BH3 peptide
to the protein. To rule out the latter possibility, we have
employed NMR methods to conclusively show that 6
indeed binds to the protein and does not unfold the
protein.
Wild-type Bcl-2 behaves poorly in solution even if the
putative hydrophobic transmembrane region is deleted
and is thus not amenable to structure determination by
either NMR spectroscopy or X-ray crystallography. Very
recently, the problem of poor solubility of wild-type Bcl-2
was elegantly circumvented by using Bcl-2/Bcl-XL chi-
meras in which part of the putative unstructured loop
of Bcl-2 was replaced with a shortened loop from Bcl-
XL.18 However, wild-type Bcl-XL is soluble if the putative
hydrophobic transmembrane region is deleted.19,21 Bcl-2
and Bcl-XL are closely related homologous proteins and
have very similar 3D structures. We have recently
determined the affinity of binding of 6 to Bcl-XL using
a FP binding assay for Bcl-XL and showed that 6 binds
to Bcl-XL with an IC50 value of 7 µM, similar to that of
binding to Bcl-2. Furthermore, Bcl-XL and Bcl-2 are
structurally very similar. Thus, it is highly likely 6 binds
to Bcl-2 and Bcl-XL in very similar binding modes. Since
Bcl-XL behaves much better in solution,19,21 we chose
to use Bcl-XL for our NMR experiments.
We measured the hetereonuclear single-quantum
correlation (HSQC) spectrum of 15N-labeled Bcl-XL
without and with 6. The HSQC spectrum is also called
a fingerprint spectrum because of its sensitivity to
structural changes. As can be seen from Figure 6, the
binding of 6 caused the peak shifts of only several
residues, strongly suggesting that 6 causes only local
perturbation in the structure of Bcl-XL but not overall
folding of Bcl-XL. Furthermore, most of the residues
whose chemical shifts are affected by the binding of 6
were around the BH3 binding pocket of Bcl-XL (Figure
7). Therefore, our NMR HSQC experiments conclusively
showed that 6 binds to the BH3 binding site in Bcl-XL
and does not unfold the protein. We are currently
undertaking efforts to determine the 3D structure of 6
F igu r e 7. The residues in Bcl-XL whose chemical shifts are
affected by the binding of 6 are shown in red. The 3D structure
of Bcl-XL is shown in ribbon representation in blue, and the
BH3 peptide is shown in licorice representation in yellow. The
NMR experimental structure of Bcl-XL in complex with the
Bak BH3 peptide is used to depict the locations of these
residues. This figure was prepared using the MOLMOL
program.38
in complex with Bcl-XL, and the results will be reported
in due course.
Su m m a r y
Through structure-based computer screening of the
NCI 3D database and biological testing of a limited
number of potential Bcl-2 inhibitors, we have identified
seven classes of novel Bcl-2 inhibitors with IC50 values
ranging from 1.6 to 14.0 µM in the FP-based binding
experiments. To verify the chemical structures of these
1
small-molecule inhibitors, we performed mass and H
NMR analyses on the chemical samples. The mass and
1H NMR data for five small-molecule inhibitors are
consistent with their chemical structures recorded in
the NCI database. However, the mass data for two
active compounds are inconsistent with their chemical
structures recorded in the NCI database. Therefore, it
is recommended that when the NCI database is used
for drug discovery, the chemical structures of the active
compounds should be carefully verified.
Of these inhibitors, compound 6 displays a potent
activity in inhibition of cell viability and cell growth in
the HL-60 cell line. We further demonstrated that 6
induces apoptosis in cancer cells with high Bcl-2 expres-
sion but has no or very little effect in cancer cells with
low Bcl-2 expression. Furthermore, using NMR meth-
ods, we conclusively demonstrated that 6 binds to the
BH3 binding site in Bcl-XL. Taken together, 6 represents
a novel class of cell-permeable small-molecule inhibitors
of Bcl-2 and a lead compound for further design and
chemical modifications.

Small-Molecule Inhibitors of Bcl-2
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 25 4321
ing of Bcl-2 structure were automatically derived from the
experimental 3D structure of Bcl-XL by the MODELLER
program. The output of MODELLER is the 3D structure
models of Bcl-2 that satisfy these restraints as well as possible.
The optimization is carried out by the variable target function
procedure employing methods of conjugate gradients and
molecular dynamics with simulated annealing.26
A number of factors may contribute to the successful
discovery of these Bcl-2 small-molecule inhibitors. First,
the high degree of homology between Bcl-2 and Bcl-XL
allows for an accurate modeling of the 3D structure of
Bcl-2, in particular its binding site, which provides a
solid structural basis for the identification of potential
small-molecule inhibitors binding to this site. Second,
the current version of the DOCK program employed in
our structure-based computer screening takes the con-
formational flexibility of the small molecules into ac-
count. Inclusion of the conformational flexibility of the
small molecules in structure-based computer screening
is important because it is rare that a single conforma-
tion of a small molecule stored in the 3D database is
the desired conformation for binding to the target
molecule (Bcl-2). Third, the NCI open database provides
a large number of unique synthetic compounds and
natural products, which is an excellent database for
drug lead discovery. Certain improvements can be made
to our database searching strategy. For example, we
used a single conformation of Bcl-2 in our database
search. However, our lengthy MD simulation showed
that Bcl-2 exhibits certain conformational flexibility in
its binding site (data not shown) and some of the
conformations are significantly different from the con-
formation used in our database search. Exploring the
conformational flexibility of the Bcl-2 binding site could
lead to the discovery of additional novel inhibitors of
Bcl-2. Furthermore, we used a simple interaction energy
function as implemented in the DOCK program to rank
the potential Bcl-2 inhibitors identified in the computer
screening. Thus, using a scoring function that has a
better correlation with the binding affinity (binding free
energy) may improve the yield of the active compounds
and could lead to the discovery of additional and/or more
potent novel Bcl-2 inhibitors.
The discovery of these novel, nonpeptide, cell-perme-
able, small-molecule Bcl-2 inhibitors represents the first
but very exciting step toward the development of a novel
cancer therapy that targets Bcl-2 for the treatment of
breast, prostate, and many other forms of cancers with
Bcl-2 protein overexpression. A small-molecule drug
that inhibits the anti-apoptotic function of Bcl-2 is
predicted to be particularly effective when used in
combination with other conventional chemotherapeutic
drugs or radiation therapy. Moreover, a potent cell-
permeable small-molecule inhibitor also provides an
invaluable research tool to further elucidate the function
of Bcl-2 in vitro and in vivo. Extensive efforts to further
optimize the small-molecule inhibitors of Bcl-2 discov-
ered in this study through structure-based design are
currently under way in our laboratories, and the results
will be reported in due course.
Further refinement was done using the molecular dynamics
program CHARMM (version 27b2).27 Hydrogen atoms were
assigned to the modeled structure using the program
QUANTA.36 The Bak BH3 peptide was placed in the Bcl-2 BH3
domain binding site in the same orientation as in the NMR
structure of Bcl-XL in complex with the Bak BH3 peptide
(1BXL in the Protein Databank).21 The complex structure was
solvated by inserting it into a 60 Å diameter TIP3P water
sphere and deleting solvent molecules that have heavy atoms
at less than 2.5 Å from any protein heavy atom. The MD
simulation was done using the all-atom parameter set from
the CHARMm force field as implemented in QUANTA,36
a
constant dielectric, ꢀ ) 1, and constant temperature, T ) 300
K. The leapfrog method with a 1 fs time step was applied for
numerical integration. Long-range electrostatic forces were
treated with the force switch method with a switching range
of 8-12 Å. van der Waals forces were calculated with the shift
method and a cutoff of 12 Å. The nonbond list was kept to 14
Å and updated heuristically. Solvent waters were kept from
evaporating by using a spherical miscellaneous mean field
potential as implemented in CHARMM.27 The solvated protein
was energy minimized using 100 cycles using the steepest
descent method and an additional 1000 cycles using the
adopted-basis Newton Raphson method. This was followed by
a 3.0 ns MD simulation. The simulation was performed on an
Origin2000 computer at the Advanced Biomedical Computing
Center at the National Institutes of Health.
St r u ct u r e-Ba sed 3D Da t a b a se Sea r ch . The refined
structure of Bcl-2 from the Bak/Bcl-2 complex, obtained after
the 3 ns MD simulation, was used for structure-based database
searching of the NCI 3D database.31 The program DOCK
(version 4.0.1) was employed.34 All residues within 8 Å from
the Bak peptide were included in the binding site used for
screening. United atom KOLLMAN charges were assigned for
the protein using the BIOPOLYMER menu in the Sybyl
program.37 Because of its general applicability, the Geisterger
method as implemented in Sybyl was used to assign charges
to the NCI compounds. We searched the National Cancer
Institute’s 3D database of 206 876 small molecules and natural
products that can be accessed by the public.31
The interactions between the Bak BH3 peptide and Bcl-2
in the modeled complex structures define the crucial binding
elements between them. Thus, the spheres used in the DOCK
program were defined in part by the coordinates of the Bak
BH3 peptide in the modeled complex structure with Bcl-2. The
conformational flexibility of the compounds from the database
was considered, and their positions and conformations were
optimized using single anchor search and torsion minimiza-
tion. Fifty configurations per ligand building cycle and 100
maximum anchor orientations were used in the anchor-first
docking algorithm. All docked configurations were energy
minimized using 10 iterations and 2 minimization cycles. This
combination of parameters resulted in an average of 26 CPU
seconds per compound on a Silicon Graphics Indigo2 Impact
with a 195 MHz R10,000 CPU. Several filters were used during
database screening: compounds with more than 10 flexible
bonds, with less than 10 heavy atoms, or with more than 50
heavy atoms were not considered. These essentially excluded
highly flexible, very small or very large molecules. A scaling
factor of 0.5 was used for the electrostatic interaction calcula-
tions. The sum of the electrostatic and van der Waals interac-
tions as calculated in the DOCK program was used as the
ranking score. The top-scoring 500 compounds were analyzed
for structural diversity. All organometallic compounds were
discarded. Chemical samples of 80 compounds were requested
from the National Cancer Institute, and 35 had samples
available for biological testing.
Exp er im en ta l Section
Hom ology Mod elin g. The sequence of human Bcl-2 was
obtained from Gene Bank (entry gi4557355). The NMR
structure of Bcl-XL (PDB code 1BXL from the Protein Data-
bank), which has 45% amino acid sequence identity, 56%
sequence similarity, and 3% gaps with Bcl-2, was used as the
template.21 The structure of Bcl-2 was built using the homol-
ogy-modeling program MODELLER (version 4.0).26 MOD-
ELLER is most frequently used for comparative modeling of
protein three-dimensional structure. More generally, MOD-
ELLER models protein 3D structure by satisfaction of spatial
restraints.24-26 The restraints used in our comparative model-

4322 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 25
Enyedy et al.
Ch em ica l Sa m p les of th e NCI Op en Com p ou n d s. The
chemical samples of these 35 compounds were provided by the
Drug Synthesis & Chemistry Branch, Developmental Thera-
peutics Program, Division of Cancer Treatment and Diagnosis,
National Cancer Institute, National Institutes of Health. All
compounds were dissolved at 10 mM in dimethyl sulfoxide
(DMSO) prior to biological experiments.
Anal. Calcd for C16H20N2O2: C, 70.56; H, 7.40; N, 10.29.
Found: C, 70.19; H, 7.45; N, 10.21.
2 ,9 -D i m e t h o x y -1 1 ,1 2 -d i h y d r o d i b e n z o [c ,g ][1 ,2 ]-
d ia zocin e 5,6-Dioxid e (6a ) a n d 5,5′-Dim eth oxy-2,2′-d in i-
tr osobiben zyl (6b). A solution of 42 (2.46 g, 9.04 mmol) and
sodium tungstate dihydrate (0.42 g, 1.26 mmol) in 95% ethanol
(27.13 mL) and water (9.04 mL) was cooled to 5 °C and treated
dropwise with 30% H2O2 (4.61 mL). The reaction mixture was
stirred for 5 h at 5 °C, then 10 mL of a saturated aqueous
solution of NH4Cl was added, and the mixture was extracted
with CH2Cl2 (3 × 30 mL). The organic phase was dried over
Na2SO4, filtered, and concentrated under reduced pressure to
afford the crude product. Recrystallization in CH2Cl2/hexanes
afforded 1.22 g (4.06 mmol, 45%) of green crystals. The mother
solution was concentrated under reduced pressure, and the
residue purified by flash chromatography on silica gel with
hexanes/ethyl acetate (3:1) as eluant to give an additional
0.431 g (1.47 mmol, 16%) of compounds 6a and 6b as green
crystals, mp 136-138 °C.
Ch em istr y. Gen er a l P r oced u r e. All chemical reagents
were commercially available. Melting points were determined
on a MelTemp II apparatus, Laboratory Devices, and are
uncorrected. Silica gel chromatography was performed on silica
gel 60, 230-400 mesh (E. Merck). 1H spectra were recorded
on a Varian Mercury instrument at 300 MHz or on a Bruker
AC-250 instrument at 250, and 13C NMR spectra were recorded
on a Bruker AC-250 instrument at 62.9 MHz. Spectra were
referenced to the solvent in which they were run (7.24 ppm
for 1H (CDCl3)). Elemental analyses were performed by
Atlantic Microlab, Inc., Atlanta, GA. FAB mass spectra were
performed on a VG-7070-EHF mass spectrometer, at unit
resolution (isotopic mass), in the positive and/or negative ion
mode. The sample matrix was 3-nitrobenzyl alcohol (NBA).
Str u ctu r a l An a lysis of th e NCI Com p ou n d s.
Data for compound 6a (ring cyclized isomer): 1H NMR
(DMSO-d6, 250 MHz) δ 7.42 (d, 2H, J ) 8.55 Hz), 6.89-6.83
1
(m, 4H), 3.76 (s, 6H), 3.07 (distorted d, 4H); H NMR (CDCl3,
Compound 5: FABMS 613.2 (M + H)+; insoluble in either
DMSO or chloroform.
250 MHz) δ 7.37 (d, 2 H, J ) 8.79 Hz), 6.80 (dd, 2H, J ) 2.69,
8.79 Hz), 6.66 (d, 2H, J ) 2.69 Hz), 3.84 (s, 6H), 3.38 and 3.01
(A2B2, 4H).
Data for compound 6b (open isomer): 1H NMR (DMSO-d6,
250 MHz) δ 7.19 (d, 2H, J ) 2.69 Hz), 6.84-6.79 (m, 4H), 6.34
(d, 2H, J ) 9.03 Hz), 4.36 (s, 4H), 3.91 (s, 6H); 1H NMR (CDCl3,
250 MHz) δ 6.97 (d, 2H, J ) 2.44 Hz), 6.73 (dd, 2H, J ) 2.44,
9.03 Hz), 6.54 (d, 2H, J ) 9.03 Hz), 4.42 (s, 4H), 3.94 (s, 6H).
Anal. Calcd for C16H16N2O4: C, 63.99; H 5.37, N 9.33. Found:
C, 63.72; H 5.48, N 9.28. FABMS 300.9 (M + H)+.
Compound 6: Identical mass and NMR spectra were
obtained for this compound for the samples obtained from the
NCI and synthesized in our laboratory. The data for this
compound are provided in the Synthesis section below.
Compound 7: 1H NMR (DMSO-d6, 300 MHz) δ 8.57 (d, 1H,
J ) 3.3 Hz), 8.55 (d, 1H, J ) 3.3 Hz), 8.39-8.43 (m, 2H), 7.69
(t, 1H, J ) 3.3 Hz), 7.67 (t, 1H, J ) 3.3 Hz), 7.42-7.50 (m, 3
H); FABMS 305.4 (M)-.
Compound 8: 1H NMR (CDCl3, 300 MHz) δ 9.56 (d, 1H, J
) 2.7 Hz), 8.57 (m, 1H), 8.22 (m, 1H), 7.91 (m, 1H), 7.71 (m,
2H), 7.52 (m, 1H), 7.32 (m, 1H); FABMS 222.1 (M)+.
Compound 9: FABMS 1241.8 (M)+ or 352.3 (M)-.
Bcl-2 F P Bin d in g Assa y. The fluorescence-labeled 16-mer
peptide tracer Flu-Bak-BH3 (sequence GQVGRQLAIIGDDINR
derived from the Bak BH3 domain) was synthesized and
labeled at the amino terminus. The 46 kDa recombinant
soluble GST-fused Bcl-2 protein was purchased from Santa
Cruz Biotechnology (Santa Cruz, CA). The reaction was carried
out in a total volume of 20 µL per well containing 10 µL of 1X
phosphate-buffered saline, 5 µL of the GST-Bcl-2 protein, and
5 µL of peptide tracer. The reaction was incubated at room
temperature for 20 min. The reading was taken at λex ) 485
nm and λem ) 535 nm using the Ultra Reader (Tecan U.S. Inc.,
Research Triangle Park, NC). A series of validation experi-
ments were performed by analyzing the maximal and minimal
signals obtained by the background, buffer, Bcl-2 protein,
tracer, and mixture of Bcl-2 protein and tracer. The Kd of
binding between Bcl-2 protein and the 16-mer fluorescence-
labeled peptide was determined by titration of Bcl-2 protein
at a concentration range of 5.4-540 nM and fluorescent tracer
concentration range of 0.145-1450 nM. The optimal binding
was obtained at a final concentration of 290 nM fluorescent
tracer and 270 nM Bcl-2 protein. To verify the specificity,
binding of the labeled peptide was compared with that of the
nonlabeled 16-mer peptide. The data indicate that nonlabeled
16-mer peptide was able to abrogate binding of the labeled
tracer, with an IC50 of approximately 0.3 µM, a value similar
to the value reported in the literature.
Compound 10: FABMS 538 (M)-.
Compound 11: 1H NMR (DMSO-d6, 300 MHz) δ 10.92 (s,
1H), 9.48 (br s, 1H), 9.38 (s, 1H), 9.18 (br s, 1H), 8.0-8.2 (m,
4H), 7.90 (d, 1H, J ) 7.2 Hz), 7.83 (d, 1H, J ) 8.4 Hz), 7.35
(dd, 1H, J ) 2.4, 7.2 Hz), 7.22 (t, 1H, J ) 8.4 Hz), 7.05 (t, 1H,
J ) 7.2 Hz), 6.80 (d, 1H, J ) 2.4 Hz), 6.61 (dd, 1H, J ) 2.4,
8.4 Hz); FABMS 330.1 (M + H)+.
Syn th esis. 5,5′-Dim eth oxy-2,2′-d in itr obiben zyl (41). To
a solution of potassium tert-butoxide (16.11 g, 136.40 mmol)
in ether (126.8 mL) and DMSO (6.34 mL) at -10 °C was added
slowly 5-methyl-2-nitrotoluene (20 g, 119.65 mmol). The reac-
tion mixture was stirred at -10 °C for 45 min, then allowed
to warm to room temperature, and stirred for an additional 3
h. The reaction was quenched by adding water (50 mL)
dropwise, and the mixture was extracted with CH2Cl2 (3 ×
150 mL). The organic layers were gathered, dried with MgSO4,
filtered, and concentrated. The crude product was recrystal-
lized in CHCl3 to afford compound 41 (12.83 g, 38.57 mmol,
64%) as brown crystals: mp 193-195 °C; 1H NMR (DMSO-
d6) δ 8.08 (d, 2H, J ) 9.03 Hz), 7.07 (dd, 2H, J ) 2.69, 9.03
Hz), 6.96 (d, 2H, J ) 2.69 Hz), 3.88 (s, 4H), 3.40 (s, 6H); 13C
NMR (DMSO-d6, 62.9 MHz) δ 162.65, 141.62, 138.52, 127.46,
116.80, 112.79, 55.97, 33.47. Anal. Calcd for C16H16N2O6: C,
57.83; H, 4.85; N 8.43. Found: C, 57,70; H 4.90; N 8.37.
5,5′-Dim eth oxy-2,2′-d ia m in obiben zyl (42). A solution of
5,5′-dimethoxy-2,2′-dinitrobibenzyl (41) (12.0 g, 36.11 mmol)
in ethanol (271 mL) was treated with 10% Pd/C (1.2 g), and
hydrazine hydrate (7.37 mL, 236.52 mmol) was added dropwise
over 10 min. The reaction mixture was stirred at rt for 30 min
and then refluxed for 2 h. The hot solution was filtered through
a short pad of Celite, and the catalyst was washed with hot
benzene (500 mL) and hot 95% ethanol (250 mL). The filtrates
were concentrated under reduced vacuum to give the crude
product, which was recrystallized in benzene to afford 42 (9.64
g, 35.39 mmol, 98%) as dark green needles: mp 69-71 °C; 1H
NMR (DMSO-d6, 250 MHz) δ 6.87-6.55 (m, 6H), 4.47 (br s, 4
H), 3.66 (s, 6H), 2.65 (s, 4H); 13C NMR (DMSO-d6, 62.9 MHz)
δ 150.95, 139.62, 126.87, 115.58, 114.96, 111.73, 55.19, 30.13.
Initial screening of all compounds was carried out at 100
µM. A 5 µL sample of the test compound was added in reaction
buffer to each of the wells containing tracer and Bcl-2 protein
at the same concentration determined before. The final
concentration of DMSO in all compounds was less than 1%.
The final reading was taken after a 10 min incubation at room
temperature. For IC50 determination of active compounds, six
to seven point serial dilutions were made in triplicate starting
at 100 µM.
Cells a n d Rea gen ts. Human breast cell lines (T47D, MDA-
231, MDA-453) and human Leukemia cells HL-60 were
obtained from the American Type Culture Collection (ATCC).
All tumor cell lines were grown and maintained in RPMI 1640
medium containing 10% FBS, except MDA-MB-231, which
used Dulbecco’s modified Eagle’s medium as the basal medium.

Small-Molecule Inhibitors of Bcl-2
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 25 4323
Cultures were maintained in a humidified incubator at 37 °C
and 5% CO2.
for providing us the plasmid (pET30, Novagen) contain-
ing human Bcl-XL gene and His tag. We thank Professor
Frederick D. Greene II for sending us part of the thesis
by David H. Bown, which described the synthetic
methods for compound 6. We are grateful to the Ad-
vanced Biomedical Computing Center at the NIH for
providing us access to their Origin2000 supercomputer.
Financial support from the CaPCure Foundation, the
NCI SPORE Project (Grant P50CA058185), and the
Lombardi Cancer Center is greatly appreciated.
Cell Via bility a n d Gr ow th Assa ys. Cell viability was
determined by the trypan blue assay. Trypan blue is a polar
dye that cannot cross intact cell membranes but crosses the
membranes of necrotic cells and apoptotic cells undergoing
secondary necrosis. Thus, it is a useful, rapid, and simple
screening assay for viable cells. In this assay, cells (5000 cells/
well) were plated in triplicate in 24-well plates with culture
medium and various amounts of FBS. Various concentra-
tions of drugs were added to the cells. Trypan blue dye was
added 24 h later, and the percentage of viable cells was
determined.
Refer en ces
Cell growth was determined by the MTT assay. The MTT
assay is a colorimetric assay that measures the reduction of
MTT by mitochondrial succinate dehydrogenase. The MTT
enters the cells and passes into the mitochondria, where it is
reduced to an insoluble, colored, formazen product. The cells
are then solubilized with an organic solvent (2-propanol), and
the released, solubilized formazen reagent is measured spec-
trophotometrically. Since reduction of MTT can only occur in
metabolically active cells, the level of activity is a measure of
the viability of the cells. Cells (2000-4000 cells/well) were
grown in medium with FBS, and various concentrations of
drugs were added to the cells at the beginning. Four days later,
MTT was added to each well and incubated for 4 h at 37 °C.
Absorbency was measured with a Dynatech model MR700.
Assa y of Ap op tosis. For flow cytometry apoptosis assay,
cell pellets were resuspended in 1× binding buffer (10 mM
HEPES (pH 7.4), 0.15 M NaCl, 5 mM KCl, 1 mM MgCl2, 1.8
mM CaCl2) containing 1:100 dilution of Annexin V-FITC
(Trevigen) and 50 µg/mL of propidium iodide and incubated
at 4 °C for 15 min. The fluorescence of Annexin V-FITC and
propidium iodide of individual cells was analyzed by FACscan.
NMR Exp er im en ts. a . Exp r ession a n d P u r ifica tion of
th e Bcl-XL P r otein . The recombinant Bcl-XL protein was
overexpressed from Escherichia coli BL21(DE3) and pET15b
(Novagen, Darmstadt, Germany) expression vector with an
N-terminal His tag. In this construct the putative C-terminal
membrane-anchoring region (residues 214-237) and a loop
between helix 1 and helix 2 (residues 49-88) were removed
to facilitate the purification of the protein. This loop is
previously shown to be dispensable for the anti-apoptotic
activity of the protein.19 The cells were grown in the minimum
medium using 15NH4Cl as the sole nitrogen source to produce
uniformly 15N labeled protein.39,40 The protein was purified
with Ni-NTA resin (Novagen), and His tag was cleaved by
thrombin digestion. The protein was further purified with NI-
NTA resin and benzamidine Sepharose resin (Pharmacia,
Piscataway, NJ ).
(1) Gross, A.; McDonnell, J . M.; Korsmeyer, S. J . Bcl-2 family
members and the mitochondria in apoptosis. Genes Dev. 1999,
13, 1899-1911.
(2) Adams, J . M.; Cory, S. The Bcl-2 protein family: arbiters of cell
survival. Science 1998, 281, 1322-1326.
(3) Hawkins, C. J .; Vaux, D. L. The role of the Bcl-2 family of
apoptosis regulatory proteins in the immune system. Semin.
Immunol. 1997, 9, 25-33.
(4) Buolamwini, J . K. Novel anticancer drug discovery. Curr. Opin.
Chem. Biol. 1999, 3, 500-509.
(5) Reed, J . C. Bcl-2 family proteins: strategies for overcoming
chemoresistance in cancer. Adv. Pharmocol. 1997, 41, 501-553.
(6) Strasser, A.; Huang, D. C. S.; Vaux, D. L. The role of the bcl-2/
ced-9 gene family in cancer and general implications of defects
in cell death control for tumourigenesis and resistance to
chemotherapy. Biochim. Biophys. Acta 1997, 1333, F151-F178.
(7) DiPaola, R. S.; Aisner, J . Overcoming bcl-2- and p53-mediated
resistance in prostate cancer. Semin. Oncol. 1999, 26, 112-116.
(8) Ziegler, A.; Luedke, G. H.; Fabbro, D.; Altmann, K. H.; Stahel,
R. A.; Zangemeister-Wittke, U. Induction of apoptosis in small-
cell lung cancer cells by an antisense oligodexynucleotide target-
ing the BCl-2 coding sequence. J . Natl. Cancer Inst. 1997, 89,
1027-1036.
(9) Webb, A.; Cunningham, D.; Cotter, F.; Clarke, P. A.; Di Stefano,
F.; Ross, P.; Corbo, M.; Dziewanowska, Z. BCl-2 antisense
therapy in patients with nonhodgkin lymphoma. Lancet 1997,
349, 1137-1141.
(10) J ansen, B.; Schlagbauer-Wadl, H.; Brown, B. D.; Bryan, R. N.;
Van Elsas, A.; Muller, M.; Wolff, K.; Eichler, H. G.; Pehamberger,
H. BCl-2 antisense therapy chemosensitizes human melanoma
in SCID mice. Nat. Med. 1998, 4, 232-234.
(11) Waters, J . S.; Webb, A.; Cunningham, D.; Clarke, P. A.; Raynaud,
F.; Di Stefano, F.; Cotter, F. E. Phase I clinical and pharmaco-
kinetic study of bcl-2 antisense oligonucleotide therapy in
patients with non-hodgkin’s lymphoma. J . Clin. Onclol. 2000,
18, 1812-1823.
(12) Piche, A.; Grim, J .; Rancourt, C.; Gomez-Navarro, J .; Reed, J .
C.; Curiel, D. T. Modulation of Bcl-2 protein levels by an
intracellular anti-Bcl-2 single-chain antibody increases drug-
induced cytotoxicity in the breast cancer cell line MCF-7. Cancer
Res. 1998, 58, 2134-2140.
(13) Chen, H. X.; Marshall, J . L.; Trocky, N.; Baidas, S.; Rizvi, N.;
Ling, Y.; Bhagava, P.; Lippman, M. E.; Yang, D.; Hayes, D. F. A
phase I study of bcl-2 antisense G3139 (Genta) and weekly
docetaxel in patients with advanced breast cancer and other solid
tumors. Proc. Am. Soc. Clin. Oncol., in press.
(14) Wang, J .-L.; Zhang, Z.-J .; Choksi, S.; Shan, S.; Lu, Z.; Croce, C.
M.; Alnemri, E. S.; Korngold, R.; Huang, Z. Cell permeable bcl-2
binding peptides: a chemical approach to apoptosis induction
in tumor cells. Cancer Res. 2000, 60, 1498-1502.
b. NMR Da ta Acqu isition a n d An a lysis. The NMR
experiments were performed on a Varian Inova 500 with pulse
field gradient (PFG) HSQC with water flip back to maximize
the signal intensity and to minimize the destruction from the
water signal41,42 (300 µM Bcl-XL, 50 mM phosphate buffer (pH
7.3), 2 mM DTT at 25 °C). HSQC spectra of Bcl-XL were
recorded prior to (free Bcl-XL) and after the addition of the
concentrated 6 solution (the final concentration of 6 in the
sample was 300 µM). Then two spectra were compared to
identify the chemical shifts induced by the additions of the
inhibitor. The NMR data were processed with the programs
pipp and nmrDraw.43,44
(15) Wang, J .-L.; Liu, D.; Zhang, Z.-J .; Shan, S.; Han, X.; Srinivasula,
S. M.; Croce, C. M.; Alnemri, E. S.; Huang, Z. Structure-based
discovery of a novel organic compound that binds to Bcl-2 protein
and induces apoptosis of tumor cells. Proc. Natl. Acad. Sci.
U.S.A. 2000, 97, 7124-7129.
(16) Degterev, A.; Lugovskoy, A.; Cardone, M.; Mulley, B.; Wagner,
G.; Mitchison, T.; Yuan, J . Identification of small-molecule
inhibitors of interaction between the BH3 domain and Bcl-xL.
Nat. Cell Biol. 2001, 3, 173-182.
Ack n ow led gm en t. The chemical samples used in
our biological evaluations were provided by the Drug
Synthesis & Chemistry Branch, Developmental Thera-
peutics Program, Division of Cancer Treatment and
Diagnosis, National Cancer Institute, National Insti-
tutes of Health, and their help on this project is highly
appreciated. We thank Professor Ziwei Huang from
Thomas J efferson University for providing us the Flu-
Bak-BH3 peptide. We are grateful to Professor Gabriel
Nunez from the University of Michigan Cancer Center
(17) Tzung, S.-P.; Kim, K. M.; Basanez, G.; Giedt, C. D.; Simon, J .;
Zimmerberg, J .; Zhang, K. Y. J .; Hockenbery, D. M. Antimycin
A mimics a cell-death-inducing Bcl-2 homology domain 3. Nat.
Cell Biol. 2001, 3, 183-191.
(18) Petros, A. M.; Medek, A.; Nettesheim, D. G.; Kim, D. H.; Yoon,
H. S.; Swift, K.; Matayoshi, E. D.; Oltersdorf, T.; Fesik, S. W.
Solution structure of the antiapoptotic protein bcl-2. Proc. Natl.
Acad. Sci. U.S.A. 2001, 98, 3012-3017.
(19) Muchmore, S. W.; Sattler, M.; Liang, H.; Meadows, R. P.; Harlan,
J . E.; Yoon, H. S.; Nettesheim, D.; Chang, B. S.; Thompson, C.
B.; Wong, S.-L.; Ng, S.-C.; Fesik, S. W. X-ray and NMR structure
of human Bcl-XL, an inhibitor of programmed cell death. Nature
1996, 381, 335-341.

4324 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 25
Enyedy et al.
(20) Aritomi, M.; Kunishima, N.; Inohara, N.; Ishibashi, Y.; Ohta,
S.; Morikawa, K. Crystal structure of rat Bcl-XL. Implications
for the function of the Bcl-2 protein family. J . Biol. Chem. 1997,
272, 27886-27892.
(21) Michael, S.; Heng, L.; David, N.; Robert, P. M.; J ohn, E. H.;
Matthias, E.; Ho Sup, Y.; Suzanne, B. S.; Brian, S. C.; Andy, J .
M.; Craig, B. T.; Stephen, W. F. Structure of Bcl-XL-Bak peptide
complex: recognition between regulators of apoptosis. Science
1997, 275, 983-986.
(22) Altschul, S. F.; Gish, W.; Miller, W.; Myers, E. W.; Lipman, D.
J . Basic local alignment search tool. J . Mol. Biol. 1990, 215, 403-
410.
(23) Altschul, S. F.; Madden, T. L.; Scha¨ffer, A. A.; Zhang, J .; Zhang,
Z.; Miller, W.; Lipman, D. J . Gapped BLAST and PSI-BLAST:
a new generation of protein database search programs. Nucleic
Acids Res. 1997, 25, 3389-3402.
(24) Sali, A.; Potterton, L.; Yuan, F.; van Vlijmen, H.; Karplus, M.
Evaluation of comparative protein modeling by MODELLER.
Proteins: Struct., Funct., Genet. 1995, 23, 318-326.
(25) Sali, A. Modeling mutations and homologous proteins. Curr.
Opin. Biotechnol. 1995, 6, 437-451.
(26) See the MODELLER web site http://guitar.rockefeller.edu/
modeller/modeller.html.
(27) Brooks, B. R.; Bruccoleri, R. E.; Olafson, B. D.; States, D. J .;
Swaminathan, S.; Karplus, M. CHARMM: a program for mac-
romolecular energy, minimization, and dynamics calculations.
J . Comput. Chem. 1983, 4, 187-217.
(28) Cosulich, S. C.; Worrall, V.; Hedge, P. J .; Green, S.; Clarke, P.
R. Regulation of apoptosis by BH3 domains in a cell-free system.
Curr. Biol. 1997, 7, 913-920.
(29) Yin, X.-M.; Oltval, Z. N.; Korsmeyer, S. J . BH1 and BH2 domains
of Bcl-2 are required for inhibition of apoptosis and heterodimer-
ization with Bax. Nature 1994, 369, 321-323.
(30) Holinger, E. P.; Chittenden, T.; Lutz, R. J . Bak BH3 peptides
antagonize Bcl-XL function and induce apoptosis through cyto-
chrome c-independent activation of caspases. J . Biol. Chem.
1999, 274, 13298-13304.
(31) Milne, G. W. A.; Nicklaus, M. C.; Wang, S.; Driscoll, J .;
Zaharevitz, D. The NCI drug information system 3D database,
J . Chem Inf. Comput. Sci. 1994, 34, 1219-1224.
(34) Ewing, T. J . A.; Kuntz, I. D. Critical evaluation of search
algorithms for automated molecular docking and database
screening. J . Comput. Chem. 1997, 18, 1175-1189.
(35) Bown, D. H. Synthesis and investigation of aryl azo dioxide-
bisnitroso systems. Ph.D. Thesis (thesis supervisor Professor
Frederick D. Greene II), Department of Chemistry, Massachu-
setts Institute of Technology, Cambridge, MA, 1983.
(36) QUANTA, a molecular modeling system, is supplied by Molec-
ular Simulations Inc., San Diego, CA.
(37) Sybyl, a molecular modeling system, is supplied by Tripos, Inc.,
St. Louis, MO 63144.
(38) Koradi, R.; Billeter, M.; Wuthrich, K. MOLMOL: a program for
display and analysis of macromolecular structures. J . Mol.
Graphics 1996, 14, 51-55.
(39) Cai, M. L.; Huang, Y.; Sakaguchi, K.; Clore, G. M.; Gronenborn,
A. M.; Craigie, R. An efficient and cost-effective isotope labeling
protocol for proteins expressed in Escherichia coli. J . Biomol.
NMR 1998, 11, 97-102.
(40) J ansson, M.; Li, Y. C.; J endeberg, L.; Anderson, S.; Montelione,
G. T.; Nilsson, B. High-level production of uniformly N-15- and
C-13-enriched fusion proteins in Escherichia coli. J . Biomol.
NMR 1996, 7, 131-141.
(41) Grzesiek, S.; Bax, A. The Importance of Not Saturating H2O in
Protein NMRsApplication to Sensitivity Enhancement and Noe
Measurements. J . Am. Chem. Soc. 1993, 115, 12593-12594.
(42) Sheppard, G. S.; Davidsen, S. K.; Fesik, S. W.; Steinman, D. H.;
Carrera, G. M.; Florjancic, A. S.; Hajduk, P. J .; Shuker, S. B.;
Olejniczak, E. T.; Nettesheim, D.; Meadows, R. P.; Nagy, I.;
Marcotte, P.; Magoc, T. J .; Morgan, D. W.; Summers, J . B. SAR
by NMR as a novel tool for lead generation: Discovery of a potent
series of non-peptide inhibitors of stromelysin (MMP-3). Abstr.
Pap.sAm. Chem. Soc. 1997, 213, 81.
(43) Delaglio, F.; Grzesiek, S.; Vuister, G. W.; Zhu, G.; Pfeifer, J .;
Bax, A. NMRPipe:
a multidimensional spectral processing
system based on UNIX pipes. J . Biomol. NMR 1995, 6, 277-
293.
(44) Garrett, D. S.; Powers, R.; Gronenborn, A. M.; Clore, G. M. A
common sense approach to peak picking in two-, three-, and four-
dimensional double and triple-resonance heteronuclear magnetic
resonance spectroscopy. J . Magn. Reson., Ser. B 1991, 95,
214-220.
(32) Voigt, J . H.; Bienfait, B.; Wang, S.; Nicklaus, M. C. Comparison
of the NCI Open Database with Seven Large Chemical Struc-
tural Databases, J . Chem. Inf. Comput. Sci. 2001, 41, 702-712
(33) Kuntz, I. D.; Blaney, J . M.; Oatley, S. J .; Langridge, R.; Ferrin,
T. E. A geometric approach to macromolecule-ligand interactions.
J . Mol. Biol. 1982, 161, 269-288.
J M010016F