Discovery of a new chemical series of BRD4(1) inhibitors using protein-ligand docking and structure-guided design

Article abstract of DOI:10.1016/j.bmcl.2015.04.107

Bromodomains are key transcriptional regulators that are thought to be druggable epigenetic targets for cancer, inflammation, diabetes and cardiovascular therapeutics. Of particular importance is the first of two bromodomains in bromodomain containing 4 protein (BRD4(1)). Protein-ligand docking in BRD4(1) was used to purchase a small, focused screening set of compounds possessing a large variety of core structures. Within this set, a small number of weak hits each contained a dihydroquinoxalinone ring system. We purchased other analogs with this ring system and further validated the new hit series and obtained improvement in binding inhibition. Limited exploration by new analog synthesis showed that the binding inhibition in a FRET assay could be improved to the low μM level making this new core a potential hit-to-lead series. Additionally, the predicted geometries of the initial hit and an improved analog were confirmed by X-ray co-crystallography with BRD4(1).

Full text of DOI:10.1016/j.bmcl.2015.04.107

Accepted Manuscript
Discovery of a new chemical series of BRD4(1) inhibitors using protein-ligand
docking and structure-guided design
Bryan C. Duffy, Shuang Liu, Gregory S. Martin, Ruifang Wang, Ming Min Hsia,
He Zhao, Cheng Guo, Michael Ellis, John F. Quinn, Olesya A. Kharenko, Karen
Norek, Emily M. Gesner, Peter R. Young, Kevin G. McLure, Gregory S.
Wagner, Damodharan Lakshminarasimhan, Andre White, Robert K. Suto,
Henrik C. Hansen, Douglas B. Kitchen
PII:
S0960-894X(15)00452-7
DOI:
http://dx.doi.org/10.1016/j.bmcl.2015.04.107
BMCL 22692
Reference:
Bioorganic & Medicinal Chemistry Letters
To appear in:
Received Date:
Revised Date:
Accepted Date:
12 March 2015
28 April 2015
30 April 2015
Please cite this article as: Duffy, B.C., Liu, S., Martin, G.S., Wang, R., Hsia, M.M., Zhao, H., Guo, C., Ellis, M.,
Quinn, J.F., Kharenko, O.A., Norek, K., Gesner, E.M., Young, P.R., McLure, K.G., Wagner, G.S.,
Lakshminarasimhan, D., White, A., Suto, R.K., Hansen, H.C., Kitchen, D.B., Discovery of a new chemical series
of BRD4(1) inhibitors using protein-ligand docking and structure-guided design, Bioorganic & Medicinal Chemistry
Letters (2015), doi: http://dx.doi.org/10.1016/j.bmcl.2015.04.107
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Graphical Abstract
To create your abstract, type over the instructions in the template box below.
Fonts or abstract dimensions should not be changed or altered.
Discovery of a new chemical series of
BRD4(1) inhibitors using protein-ligand
docking and structure-guided design
Leave this area blank for abstract info.
Bryan C. Duffy, Shuang Liu, Gregory S. Martin, Ruifang Wang, Ming Min Hsia, He Zhao, Cheng Guo,
Michael Ellis, John F. Quinn, Olesya A. Kharenko, Karen Norek, Emily M. Gesner, Peter R. Young, Kevin G.
McLure, Gregory S. Wagner, Damodharan Lakshminarasimhan, Andre White, Robert K. Suto, Henrik C.
Hansen, Douglas B. Kitchen*

Bioorganic & Medicinal Chemistry Letters
journal homepage: www.els evier.com
Discovery of a new chemical series of BRD4(1) inhibitors using protein-ligand
docking and structure-guided design
Bryan C. Duffya, Shuang Liua, Gregory S. Martina, Ruifang Wanga, Ming Min Hsiaa, He Zhaoa, Cheng
Guoa, Michael Ellisa, John F. Quinnb, Olesya A. Kharenkoc, Karen Norekc, Emily M. Gesnerc, Peter R.
Youngc, Kevin G. McLurec, Gregory S. Wagnerc, Damodharan Lakshminarasimhand, Andre Whited, Robert
K. Sutod, Henrik C. Hansenc, Douglas B. Kitchena*
aAlbany Molecular Research (AMRI), 26 Corporate Circle, PO Box 15098, Albany, NY 12212-5098, USA
bJFQuinn Consulting, 113 Jay St. Albany, NY 12210
cZenith Epigenetics Corp., Suite 300, 4820 Richard Road SW, Calgary, Alberta, T3E 6L1, Canada
dXtal BioStructures, Inc., 12 Michigan Dr., Natick, MA 01760, USA
* douglas.kitchen@amriglobal.com
ARTICLE INFO
ABSTRACT
Article history:
Received
Revised
Accepted
Available online
Keywords:
Virtual screening
Bromodomain inhibitors
BRD4(1)
Protein-ligand docking
Hit triage
Bromodomains are key transcriptional regulators that are thought to be druggable epigenetic
targets for cancer, inflammation, diabetes and cardiovascular therapeutics. Of particular
importance is the first of two bromodomains in bromodomain containing 4 protein (BRD4(1)).
Protein-ligand docking in BRD4(1) was used to purchase a small, focused screening set of
compounds possessing a large variety of core structures. Within this set, a small number of weak
hits each contained a dihydroquinoxalinone ring system. We purchased other analogs with this
ring system and further validated the new hit series and obtained improvement in binding
inhibition. Limited exploration by new analog synthesis showed that the binding inhibition in a
FRET assay could be improved to the low µM level making this new core a potential hit-to-lead
series. Additionally, the predicted geometries of the initial hit and an improved analog were
confirmed by X-ray co-crystallography with BRD4(1).
2009 Elsevier Ltd. All rights reserved
.

Bromodomains (BRDs) have recently been of increasing interest
as possible new drug targets due to the profound effects these
nuclear structures have on cellular transcriptional control. 1 BRD-
containing proteins are key players in epigenetics by binding
acetylated lysines (KAc) on chromatin structures and are thus
termed “readers” of these epigenetic markers. Aberrant
acetylation of lysine residues has been implicated in
posttranslational increases of gene expression leading to several
conditions, such as cancer, diabetes, inflammation and
cardiovascular diseases. The BET (bromodomain and extra-
terminal) protein family (BRD2, BRD3, BRD4 and BRDT) have
two consecutive bromodomains that bind to proximal KAc sites
on histone tails. BRDs contain well-defined pockets for KAc that
provide a “druggable” site suitable for small molecule inhibitor
development. Many examples of potent BRD binders have been
disclosed and are illustrated as scaffolds 1-5 and specific
optimized examples, 6-9 (adapted from Filippakopoulos and
Knapp). 1
scaffold 11 was identified and rapidly explored by additional compound
purchases.
First, we designed our workflow to use 3D protein structural
information in order to test only a few samples while
simultaneously accounting for docking inaccuracies, unavailable
compounds and experimental uncertainties. Docking scores from
any available algorithm have large errors in predicting binding
potency. However, we believed that if many examples from a
scaffold score well in docking calculations that it is less likely
that all the high scores are incorrect. Therefore, we attempted to
identify clusters of chemically similar compounds containing
common core scaffolds with significantly higher docking scores
and simultaneously to improve the chances that examples of each
scaffold were available for purchase and testing. Experimental
uncertainties arise from multiple sources including weighing,
structural assignment and biochemical measurements. In order to
decrease the chances of missing an active compound as a false
negative, clusters of screening compounds provide statistical
reinforcement of single concentration screening data, and
possibly provide a small amount of early SAR. Grouping
experimental measurements by scaffold can identify active
substructures where no single member achieved high inhibition
in the single point binding assay. As hits were identified,
chemical structures were grouped by common scaffolds. The
structural similarity within each scaffold provided confidence
that the weak activity measurements were valid and that the
structural representations were likely accurate. Additionally,
when similar binding activity exists within a series, fewer
analogs need to be re-synthesized since any example becomes
In an effort to identify new starting points for a BRD inhibitor
program, we undertook virtual screening using protein-ligand
docking and tested selected compounds using two competitive
binding inhibition assays and identified a new core ring system,
10, as a possible starting point for drug discovery. Herein, we
report the validation of the binding activities of this series,
crystallographic confirmation of the binding mode in BRD4(1)
and a limited SAR exploration around this new N-substituted
dihydroquinoxalinone hit series.
Screening strategy: Our goal in this hit-to-lead program was to
rapidly identify new starting compound scaffolds using small
focused sets of readily available samples with diverse structural
cores which inhibit the binding of histones to the BRD4(1)
domain. Virtual screening is a common method for the
generation of hit compounds in medicinal chemistry 2, 3 and
recently has been used to identify compounds that bind to
bromodomains. 4, 5, 6 To rapidly identify new hits, we chose to
virtually screen a collection of chemical structures and then
purchased and tested samples for binding activity in BRD4(1)
and BRD2(2). We used the concept of the “latent-hit series” 7 to
overcome the inherent errors in docking, scoring and single
concentration screening.
Figure 1. Screening Process. 153 compounds were selected by docking and
purchased. After testing in BRD2(2) and BRD4(1) single point assays

validation of the series and synthetic efforts can be focused on
new analogs. Compounds obtained in virtual screening have been
previously disclosed and therefore, do not offer novel chemical
structures. However, each scaffold can be used as rapid entry
points to identify novel BRD4(1) hit series. Our subsequent goal
of the hit-to-lead program was to rapidly develop structure
activity relationships (SAR) on each biologically active scaffold
yielding scaffolds with multiple members for further lead
optimization.
0.60. The diverse set was not strictly filtered to prevent the
elimination of compounds that do not have optimal R-groups, but
could lead to groups of more ideal compounds by using chemical
similarity calculations.
The protein-ligand docking proceeded as illustrated in Figure
1 starting from a diverse set of available compounds with
successive rounds of higher precision docking.27 Compounds
within each of the top-scoring scaffolds were manually viewed
and evaluated as to the likelihood of binding with valid poses and
distinguishing features from known BRD binding compounds.
The minimum requirement for consideration as a valid pose was
a predicted potential hydrogen bond to the Asn140 sidechain
amide, which mimics the native histone acetylated lysine binding
location. Compounds with groups residing in the WPF shelf were
given special attention. A total of 153 available compounds met
the hydrogen bond requirements and either a very favorable
Glide score or were members of a scaffold having a favorable
average Glide score. These compounds represented 82 scaffolds
and were purchased from commercial vendors and evaluated in in
vitro screening.
Figure 2. Initial co-crystal structures of BRD4(1) used for docking. Top
panel: co-crystal of 7 8 ; Lower panel: pdb code: 3u5l, 6. Water molecule,
HOH4 (blue) was the only water molecule retained in the docking models.
Chart 1. Initial hits obtained in single concentration BRD4(1) and BRD2(2)
per cent inhibition measurements.
Virtual Screening Method: Ligands are known to induce
differences in shape among crystal structures and this in turn
changes the structure and binding energy predictions of any
docking protocol. In order to increase the chances that we would
identify new diverse scaffolds for BRD4(1), we used two X-ray
crystal structure models with different co-crystallized ligand
structural classes: an analog of RVX-208 (7) with binding site
coordinates similar to RVX-208 and 6, BzT-7 (PDB:3u5l) 9
which occupies an additional pocket. The docking grids were
prepared using Schrödinger’s protein preparation tools and Glide
grid preparation workflows for each crystal structure. 10 In both
docking models, only one water molecule (HOH4) was retained
from each crystal structure (Figure 2). 9, 8, 11 This water
molecule is in the chain of five commonly conserved water
molecules and is the closest water to Asn140. It bridges Tyr97
and the ligand through hydrogen bonds in both X-ray crystal
structures. Elimination of all other water molecules provided
room for new interactions.
Hit Screening and treatment: The purchased compounds were
initially tested by HTRF binding inhibition to BRD4(1) and
BRD2(2) at 50 µM concentrations.28 Compounds with greater
than 50% binding to BRD4(1) were submitted for dose-response
testing and IC50 calculations, while hits in the 30–49% binding
range at 50 µM were grouped by scaffold.
We grouped compounds into top-level scaffolds (most
complex) and the next lower sub-scaffolds (removal of one ring).
13 The enrichment of the occurrence of hits (binding >30%)
within a scaffold was used to indicate likely hit series for further
SAR exploration. These few compounds are only indicative of
potential active scaffolds but the fact that there is significant
enrichment of hits overcomes the otherwise high experimental
uncertainty.
Virtual Screening to identify potential hit series: The virtual
screening and hit triage process that we used is illustrated in
Figure 1. A source library of ~25 million compounds from
various publically available sources was filtered on common
drug-like property criteria. A diverse-set of ~230,000 compounds
was selected through multiple rounds of random selections
followed by confirmation of diversity through 2-D atom-pair
similarity calculations. 12 Compounds within each selection
subset had 2D atom-pair similarity Dice coefficients of less than
Compounds containing the N-acylated dihydroquinoxalone
ring, 11, showed promising activity and we will only discuss
further SAR work on this scaffold, though several other series
were also obtained from the virtual screening.
Additional
purchases were made within the series following the same
docking protocol. Of the 22 tested dihydroquinoxalones, 14
exhibited inhibition in either BRD4(1) or BRD2(2) assays. Chart
1 divides the structures into three groups: strong inhibitors in at

least one assay (>50% inhibition), weak hits (25-50% inhibition
in one assay) and very weak hits that lack meaningful inhibition.
The % inhibition of binding to BRD2(2) was generally higher
than to BRD4(1) with the dihydroquinoxalone system. BRD2(2)
was inhibited in the 50-75% range for four compounds whereas
the inhibition was ~0-25% in BRD4(1). Even though our
ultimate goal was BRD4(1) inhibition, the addition of testing in
BRD2(2) provided validation for hits near the activity cutoff
which would have been missed using only one assay.
16b
16c
4-MeO-Ph
4-MeO-Ph
thiophen-2-yl
E
E
13 (2)
13
23 (2)
24
4-acetamido-
Ph
16d
16e
16f
16g
16h
16i
4-MeO-Ph
4-MeO-Ph
4-MeO-Ph
4-MeO-Ph
4-pyridyl
4-MeO-Ph
3,4-di-Me-Ph
4-MeO-Ph
3,4-di-MeO-Ph
4-MeO-Ph
Ph
7-MeO-E
13
13
E
14 (2)
16 (2)
18
>50 (2)
13 (2)
26
E
E
Structure-Activity exploration: The compound 11c was the most
active compound from the hit series after single concentration
testing. To begin to explore the series, we first resynthesized 11c
and retested to obtain more accurate IC50’s. In parallel with
E
25
5
4-MeO2C-Ph
4-MeO-Ph
3-NHCOMe
4-MeO-Ph
4-NH2COPh
4-MeOPh
E
25 (2)
34 (2)
40
25 (2)
>50 (2)
10
synthetic expansion of the series, a resynthesized sample of
16j
16k
16l
Ph
7-MeO-E
29
11c(16a) was co-crystallized in BRD4(1).
(Figure 3,
Ph
E
F
E
E
PDB:4hy3) The X-ray crystal structure was important to the
screening process, because it established that the weak hit was
binding in the desired pocket in a manner that agreed with the
Glide docking SP pose with an rmsd of 0.96 Å from the predicted
geometry. The structure also provided the initial information to
suggest improvements in the binding interactions of 16a (22 µM,
BRD2(2) see Table 1). The –NH group of the dihydroquinoxaline
forms a hydrogen bond to the BRD4(1) Asn140 sidechain in a
manner similar to the native acetylated lysine and other inhibitor
structures (see figures ref 1 and pdb codes 3uvw and 3u5l.). The
pendant aryl system points towards the WPF shelf while the
carbonyl of the amide group points to the narrow ZA water
channel. The SAR exploration for this scaffold attempted to
enhance these three interaction points and to identify any
freedom to eliminate liabilities in the series.
4-MeO-Ph
Ph
44
12
16m
46
49
16n
2,4-di-OMe-Ph
>50
>50
a
n=1 unless otherwise indicated in parenthesis
A small set of compounds similar to amide 11c(16a) were
synthesized in order to complement the purchased compounds
and verify that the observed activity was derived from the
chemical structure and was representative of a larger series of
compounds. A three-step reaction sequence was used to prepare
and test a small library of compounds for testing primarily
intended to optimize binding through changes in R1 and R2
(Scheme 1).
Based on initial modeling of the series and the crystal
structures, R1 was expected to rest on the WPF shelf. R2 was
expected to face a narrow water channel (the ZA channel). R3
was varied in order to explore the hydrogen bonding
opportunities close to Asn140. The compounds were tested in
the BRD4(1) and BRD2(2) HTRF binding assays (Table 1). The
SAR was flat and narrow, however, significant improvement in
binding was observed with the best examples achieving 13 µM
binding inhibition for BRD4(1). Compounds with the electron
rich 4-methoxyphenyl as R1 and substituted phenyl groups as R2
performed the best in the binding assay.
The initial hit, 11c contains some possible liabilities to
development into a true lead series. The potential Michael
acceptor, the enone moeity was the first concern. We made an
analog where the double bond was hydrogenated to provide 11x.
Additionally, the necessity of the carbonyl was examined with
10a. In both cases, the inhibition was < 30% at 50 µM
suggesting that a planar geometry was necessary at both the
amide and vinyl group. Additional, analogs that are not reported
here support these trends.
Scheme 2. Synthesis of 19a–i.
A small series of compounds related to scaffold 19 and
analogs of compound 16a were prepared and tested (Table 2).
These aryl amides were prepared using the route in Scheme 2, in
an effort to replace the α,β-unsaturation with alternate, less-
reactive biaryl groups and to introduce a scaffold that had not
been disclosed previously. The 5-Ar position for R1 was
predicted to point towards the WPF shelf with the expectation of
improving binding. Docked pose predictions of the pyridyl-
phenyl group of 19d were ambiguous in multiple protein sites
because of very similar scores. Some of the docked poses filled
the ZA channel with the pyridyl-phenyl ring while other models
accommodated the same ring system in the WPF shelf.
Therefore, 19d was prepared and submitted for co-crystallization
with BRD4(1) and the X-ray crystal structure determined
(pdb:4hy4). The X-ray crystal structure reveals that one of the
Scheme 1. Synthesis of 16a-n.30
Table 1. FRET binding data for α,β-unsaturated amides
Compd. Substitution
IC50 (µM)a
R1
R2
Ph
R3
E
BRD4(1) BRD2(2)
11c, 16a
4-MeOPh
26
22

ring bonds of the pyridine ring replaces the double bond of the
α,β-unsaturated amide (see Figure 3c) as predicted by some of
the docked models of 19d. This observation confirms the
necessity of two sp2 centers. The X-ray crystal structure (Figure
3) indicates a stabilizing hydrogen bond between the 3-pyridyl
nitrogen atom and a water molecule in the ZA channel that was
not present in any of the crystal structures used for docking
(Scheme 2). In fact, the chlorine atom of 6 in its crystal structure
(3u5l) is located in approximately the same location as this new
water molecule. Additionally, the water molecule initiates a
hydrogen-bond chain with other water molecules that form
hydrogen bonds to the protein. The most potent compound of this
series, 19a, replaced this 3-pyridyl group with a 3-hydroxy group
that is predicted to form a hydrogen bond to the same water
network and a modest improvement in binding was observed.
Figure 3. X-ray crystal structures of 11c (16a) (top) and 19d (middle) and
both molecules (bottom) in BRD4(1). Water molecules are colored green in
the bottom image for 11c structure. (PDB:4hy3 and 4hy4)
Table 2. FRET binding data for biaryl amides.
Compd. Substitution
IC50 (µM)a
R1
Ar
BRD4(1) BRD2(2)
19a
19b
19c
19d
19e
19f
19g
19h
19i
Ph
3-hydroxyPh
3-MeO-Ph
3-pyridyl
3-pyridyl
3-pyridyl
3-pyridyl
3-pyridyl
3-pyridyl
3-MeO-Ph
9
nd
nd
nd
20.
nd
12
nd
nd
nd
Ph
13
25
26
29
44
47
49
>50
3-MeO-Ph
Ph
3-hydroxyPh
3-cyanoPh
4-F-Ph
4-Cl-Ph
PhCH2NH-
a
nd= no data; n=1 in all cases
A new hit series was identified for inhibition of BRD4(1), an
important new biological target. We used docking to identify a
series of hits that were confirmed by a limited SAR exploration
and two protein co-crystal structures. The hit identification step
used compound clustering extensively to improve the quality of
virtual screening and the probability of a successful selection of
active compounds for purchase. Scaffold relationships and latent-
hit analysis increased the probability of selecting true-positive
clusters of active compounds, while decreasing the probability of
selecting false-positive clusters. The confirmation of ligand
binding pose through X-ray crystallography validated the
docking model predictions and identified regions where
improved binding may be possible. We improved the best initial
screening activity from 26 µM (11c) to 9 µM (19a) in BRD4(1)
inhibition. The application of virtual screening, cheminformatic
techniques, X-ray crystallography, and early focused library
synthesis rapidly identified several scaffold clusters of BRD4(1)
binding hits derived from commercially available compounds
that may act as starting points for medicinal chemistry hit-to-lead
optimization. Other clusters of compounds are being pursued in
parallel.

Acknowledgements
We thank Hélène Decornez and Keith Barnes for careful
reading of the manuscript.

References and Notes
calculations resulted in 17,774 compounds and the compounds were
filtered using the same property values and this filtering resulted in a
total of 3394 molecules. The HTVS selection set of compounds was
re-docked at the Glide Standard Precision(SP) precision level for
more accurate scoring. The top scoring 1,229 compounds from both
grids contained 82 top-level HierS scaffolds. 13
1.
2.
Filippakopoulos, P.; Knapp, S. Nat. Rev. Drug Disc. 2014, 13 337.
Kitchen, D. B.; Decornez, H.; Furr, J. R.; Bajorath, J. Nat.Rev.Drug
Disc. 2004, 3 935.
3.
4.
Lavecchia, A.; Di Giovanni, C. Curr.Med.Chem. 2013, 20 2839.
Vidler, L. R.; Filippakopoulos, P.; Fedorov, O.; Picaud, S.; Martin, S.;
Tomsett, M.; Woodward, H.; Brown, N.; Knapp, S.; Hoelder, S. J.
Med. Chem. 2013, 56 8073.
28 Bromodomain expression: BRD4(1) and BRD2(2) bromodomain
constructs were based on (Filippakopoulos et al., 2010, Nicodeme et
al., 2010) with an added N-terminal His-tag. These constructs were
cloned, expressed, and purified by nickel affinity and size exclusion
chromatography by either Genscript or Xtal BioStructures, frozen at
-80 °C for competition experiments. Time Resolved Fluorescence
Resonance Energy Transfer (TR-FRET) assay: 200 nM N-
terminally His-tagged bromodomains, BRD4(1) or BRD2(2) and 25-
50 nM biotinylated tetra-acetylated histone H4 peptide (Millipore)
were incubated in the presence of Europium Cryptate-labeled
streptavidin (Cisbio Cat. #610SAKLB) and XL665-labeled
monoclonal anti-His antibody (Cisbio Cat.#61HISXLB) in a white
96 well microtiter plate (Greiner). For inhibition assays, serially
diluted compound was added to these reactions in a 0.2% final
concentration of DMSO. Final buffer concentrations were 30 mM
HEPES pH 7.4, 30 mM NaCl, 0.3 mM CHAPS, 20 mM PO4 pH 7.0,
320 mM KF, 0.08% BSA). After 2h incubation at room temperature,
the fluorescence by FRET was measured at 665 and 620 nm by a
SynergyH4 plate reader (Biotek). Duplicate measurements were
obtained at each concentration. IC50 values were determined from a
dose response curve.
29 X-ray Crystallography. The recombinant human BRD4(1) 8
protein at 1 mg/mL was mixed with 0.6 mM 16a from 100 mM
DMSO stock. The complex was incubated at room temperature for
30 min and then concentrated to 4.0 mg/mL using 3000MWCO
concentrators. Protein co-crystals were obtained by vapor diffusion
with sitting drops against a reservoir solution containing 200 mM
Potassium thiocyanate and 20% (W/V) PEG 3350. Selected
monocrystals were briefly treated with a cryo-protectant solution
containing the well solution supplemented with 20% 2-methyl-2,4-
pentanediol and flash-frozen in liquid nitrogen.
5.
6.
7.
8.
Muwa, C.; Singam, E. R. A.; Raman, S. S.; Subramanian, V. Mol.
BioSy.s 2014, 10 2384.
Pachaiyappan, B.; Woster, P. M. Bioorg. Med. Chem. Lett. 2014, 24
21.
Varin, T.; Didiot, M. C.; Parker, C. N.; Schuffenhauer, A. J. Med.
Chem. 2012, 55 1161.
Filippakopoiulos, P.; Picaud, S.; Federov, O.; Keller, M.; Wrobel, M.;
Morgenstern, O.; Bracher, F.; Knapp, S. Bioorg. Med. Chem.
2012, 20 1878.
9.
Suite 2012: Glide, version 5.8; Schrodinger, LLC: New York, NY,
2012.
10. McLure, K. G.; Gesner, E. M.; Tsujikawa, L.; Kharenko, O. A.;
Attwell, S.; Campeau, E.; Wasiak, S.; Stein, A.; White, A.;
Fontano, E.; Suto, R. K.; Wong, N. C. W.; Wagner, G. S.; Hansen,
H. C.; Young, P. R. PLOS One 2013, 8 e83190.
11. Vidler, L. R.; Brown, N.; Knapp, S.; Hoelder, S. J. Med. Chem. 2012,
55 7346.
12. Carhart, R. E.; Smith, D. H.; Venkataraghavan, R. J. Chem. Inf.
Comput. Sci. 1985, 25 64.
13. Wilkens, S. J.; Janes, J.; Su, A. I. J. Med. Chem. 2005, 48 3182.
14. Zhao, H.; Gartenmann, L.; Dong, J.; Spiliotopoulos, D.; Caflisch, A.
Bioorg. Med. Chem. Lett. 2014, 24 2493.
15. Pachaiyappan, B.; Woster, P. M. Bioorg. Med. Chem. Lett. 2014, 24
21.
16. Herold, J. M.; Wigle, T. J.; Norris, J. L.; Lam, R.; Korboukh, V. K.;
Gao, C.; Ingerman, L. A.; Kireev, D. B.; Senisterra, G.; Vadadi,
M.; Tripathy, A.; Brown, P. J.; Arrowsmith, C. H.; Jin, J.; Janzen,
W. P.; Frye, S. V. J. Med. Chem. 2011, 54 2504.
Apo-BRD4(1) crystals were obtained by vapor diffusion using sitting
drop against a reservoir solution containing 10% (W/V) PEG 3350,
100 mM HEPES pH 7.5 and 200 mM L-proline. Well-formed
crystals were soaked for 3 hours at room temperature with the same
reservoir solution supplemented with 20 mM 19d. Soaked crystals
were briefly treated with the soaked solution where the PEG3350
concentration was increased to 35% (W/V) and flash-frozen in liquid
nitrogen.
17. Filippakopoulos, P.; Knapp, S. Nat. Rev. Drug Disc. 2014, 13 337.
18. Chung, C.-w.; Dean, A. W.; Woolven, J. M.; Bamborough, P. J. Med.
Chem. 2012, 55 576.
19. Duffy, B. C.; Zhu, L.; Decornez, H.; Kitchen, D. B. Bioorg. Med.
Chem. 2012, 20 5324.
20. Mujtaba, S.; Zeng, L.; Zhou, M.-M. Oncogene 2007, 26 5521.
21. Friesner, R. A.; Murphy, R. B.; Repasky, M. P.; Frye, L. L.;
Greenwood, J. R.; Halgren, T. A.; Sanschagrin, P. C.; Mainz, D.
T. J. Med. Chem. 2006, 49 6177.
Single-crystal X-ray diffraction was obtained at Beam line X29 of
the National Synchrotron Light Source, Brookhaven National
Laboratory, using an automated sample mount system. The X-ray
diffraction data were reduced using HKL2000 26 The BRD4(1)-16a
and BRD4(1)-19d crystals both belonged to the space group
P212121 and they diffracted to 1.33 Å and 1.60 Å resolution,
respectively (Table S1). The protein structures were solved by
molecular replacement and refined using REFMAC5 22 as
previously done. 8 Model rebuilding was pursued using COOT 23. In
each structure, a single compound conformation was observed and
refined. The BRD4(1)-19d co-structure was refined to an R/Rfree of
16.0%/19.9% with good stereochemistry. Given the high resolution
of the BRD4(1)-16a co-structure, individual anisotropic temperature
factors were also refined to an R/Rfree of 11.0%/14.8% with good
stereochemistry. The final crystallographic data reduction statistics
are summarized in Table S2. The structures have been deposited in
the PDB with the following codes: 4yh3 and 4yh4.
22. Halgren, T. A.; Murphy, R. B.; Friesner, R. A.; Beard, H. S.; Frye, L.
L.; Pollard, W. T.; Banks, J. L. J. Med. Chem. 2004, 47 1750.
23. Friesner, R. A.; Banks, J. L.; Murphy, R. B.; Halgren, T. A.; Klicic, J.
J.; Mainz, D. T.; Repasky, M. P.; Knoll, E. H.; Shaw, D. E.;
Shelley, M.; Perry, J. K.; Francis, P.; Shenkin, P. S. J. Med. Chem.
2004, 47 1739.
24. Otwinowski, Z.; Minor, W. Methods Enzymol. 1997, 276 307.
25. Vagin, A. A.; Steiner, R. S.; Lebedev, A. A.; Potterton, L.;
McNicholas, S.; Long, F.; Murshudov, G. N. Acta Crystallogr.,
Sect. D 2004, 60 2284.
26. Emsley, P.; Cowtan, K. Acta Crystallogr., Sect. D 1991, 60 2126.
27 The diverse set of 230,000 compounds was docked into each
30 The synthesis of these analogues is exemplified by the preparation
of 16e as described below:
independent BRD4(1) docking models using Schrödinger’s Glide
,
program in the high throughput virtual screen (HTVS) mode. 25 24
Poses of compounds with a Glide score less than -7.5 in either model
were retained. This diverse set of 1,475 compounds was used to
select compounds from the virtual library with 2D atom pair
similarity scores ≥0.80 Dice similarity level. These similarity
Step 1: A mixture of 4-methoxybenzaldehyde (1.36 g, 10.0 mmol),
2-(4-methoxyphenyl)acetic acid (1.66 g, 10.0 mmol), acetic
anhydride (2.04 g, 20.0 mmol), and triethylamine (1.0 mL, 7.0
mmol) was heated at 140 °C under nitrogen for 16 h. The reaction

mixture was cooled to room temperature and diluted with ethyl
acetate and water. The organic layer was washed with brine, dried
over sodium sulfate, filtered and concentrated. The residue was
recrystallized in ethyl acetate and hexanes to afford (E)-2,3-bis(4-
methoxyphenyl)acrylic acid (0.65 g, 23%) as a yellow solid: 1H
NMR (500 MHz, CDCl3) δ 7.87 (s, 1H), 7.17 (dd, J = 6.7, 2.1 Hz,
2H), 7.05 (dd, J = 7.0, 1.9 Hz, 2H), 6.93 (dd, J = 6.7, 2.1 Hz, 2H),
6.71 (dd, J = 7.0, 2.0 Hz, 2H), 3.85 (s, 3H), 3.77 (s, 3H).
Step 2&3: To a solution of (E)-2,3-bis(4-methoxyphenyl)acrylic acid
(142 mg, 0.5 mmol) in dichloromethane (3 mL) was added oxalyl
chloride (0.17 mL, 2.0 mmol). The reaction mixture was stirred at
room temperature for 1 h and then concentrated. The residue was
dissolved in tetrahydrofuran (2 mL) and was added into an ice cold
solution of 3,4-dihydroquinoxalin-2(1H)-one (67 mg, 0.45 mmol)
and triethylamine (50 mg, 0.5 mmol) in tetrahydrofuran (7 mL). The
mixture was warm to room temperature, stirred for 16 h. The
mixture was diluted with ethyl acetate and saturated sodium
carbonate (200 mL/50 mL). The organic layer was separated,
washed with brine, dried over sodium sulfate, filtered and
concentrated. The residue was purified by chromatography (silica
gel, 0–40% ethyl acetate/hexanes) to give Example 16e (170 mg,
82%) as an off-white solid: 1H NMR (500 MHz, DMSO–d6) δ 10.57
(s, 1H), 7.43 (d, J = 8.0 Hz, 1H), 7.06 (td, J = 7.7, 1.3 Hz, 1H), 7.01–
6.99 (m, 4H), 6.94 (td, J = 7.9, 1.3 Hz, 1H), 6.86 (dd, J = 7.9, 1.2 Hz,
1H), 6.81–6.76 (m, 5H), 4.34 (s, 2H), 3.71 (s, 3H), 3.70 (s, 3H); ESI
m/z 415 [M + H]+.