Optimization of Potent ATAD2 and CECR2 Bromodomain Inhibitors with an Atypical Binding Mode

Article abstract of DOI:10.1021/acs.jmedchem.0c00021

Most bromodomain inhibitors mimic the interactions of the natural acetylated lysine (KAc) histone substrate through key interactions with conserved asparagine and tyrosine residues within the binding pocket. Herein we report the optimization of a series of phenyl sulfonamides that exhibit a novel mode of binding to non-bromodomain and extra terminal domain (non-BET) bromodomains through displacement of a normally conserved network of four water molecules. Starting from an initial hit molecule, we report its divergent optimization toward the ATPase family AAA domain containing 2 (ATAD2) and cat eye syndrome chromosome region, candidate 2 (CECR2) domains. This work concludes with the identification of (R)-55 (GSK232), a highly selective, cellularly penetrant CECR2 inhibitor with excellent physicochemical properties.

Full text of DOI:10.1021/acs.jmedchem.0c00021

Subscriber access provided by BIU Pharmacie | Faculté de Pharmacie, Université Paris V
Article
The Optimisation of Potent ATAD2 and CECR2
Bromodomain Inhibitors with an Atypical Binding Mode.
Simon Lucas, Stephen J. Atkinson, Paul Bamborough, Heather A Barnett, Chun-Wa
Chung, Laurie J. Gordon, Darren J Mitchell, Alexander Phillipou, Rab K. Prinjha, Robert
J Sheppard, Nicholas C. O. Tomkinson, Robert J Watson, and Emmanuel H Demont
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.0c00021 • Publication Date (Web): 22 Apr 2020
Downloaded from pubs.acs.org on April 23, 2020
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a service to the research community to expedite the dissemination
of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in
full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully
peer reviewed, but should not be considered the official version of record. They are citable by the
Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore,
the “Just Accepted” Web site may not include all articles that will be published in the journal. After
a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web
site and published as an ASAP article. Note that technical editing may introduce minor changes
to the manuscript text and/or graphics which could affect content, and all legal disclaimers and
ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or
consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W.,
Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 110
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
The Optimisation of Potent ATAD2 and CECR2
Bromodomain Inhibitors with an Atypical
Binding Mode.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Simon C. C. Lucas,‡,†,¶ Stephen J. Atkinson,*,‡ Paul Bamborough, Heather Barnett, Chun-wa
§
‡
§
|
‡
|
‡
Chung, Laurie Gordon, Darren J. Mitchell, Alexander Phillipou, Rab K. Prinjha, Robert J.
Sheppard,‡, ¥ Nicholas C. O. Tomkinson, Robert J. Watson, and Emmanuel H. Demont.
†
‡
‡
‡
§
Epigenetics Discovery Performance Unit, Computational & Structural Chemistry, Molecular
|
Discovery Research and Biological Sciences GlaxoSmithKline Medicines Research Centre,
Gunnels Wood Road, Stevenage, SG1 2NY, United Kingdom.
†WestCHEM, Department of Pure and Applied Chemistry, Thomas Graham Building, University
of Strathclyde, 295 Cathedral Street, Glasgow, G1 1XL, United Kingdom.
ABSTRACT
Most bromodomain inhibitors mimic the interactions of the natural acetylated lysine (KAc)
histone substrate through key interactions with conserved asparagine and tyrosine residues within
the binding pocket. Herein we report the optimisation of a series of phenyl sulfonamides which
exhibit a novel mode of binding to non-Bromodomain and Extra Terminal Domain (non-BET)
bromodomains through displacement of a normally conserved network of four water molecules.
Starting from an initial hit molecule we report its divergent optimisation towards the ATPase
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 110
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
family AAA domain containing 2 (ATAD2) and Cats Eye Syndrome Chromosome Region,
Candidate 2 (CECR2) domains. This work concludes with the identification of (R)-55 (GSK232),
a highly selective, cellularly penetrant CECR2 inhibitor with excellent physicochemical
properties.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
INTRODUCTION
The Bromodomain and Extra-Terminal Domain (BET) family of bromodomain containing
proteins is a well-established therapeutic target for small molecule inhibition, leading to efficacy
in both oncology and immuno-inflammation indications.1-3 It has been shown that BET proteins
organise on superenhancer and enhancer regions and profoundly regulate gene expression relevant
to a range of diseases.4-5 Indeed, many small molecule BET bromodomain inhibitors, which block
the interaction of BET with histones and chromatin, are currently progressing through clinical
6
trials. However, the BET family represent just 8 of the 61 known human bromodomains,
therefore, it is logical to propose that other bromodomains may control the expression of genes
crucial to other diseases.7-9 Despite this promising hypothesis, the pharmacological relevance of
inhibiting these bromodomains using small molecules is still largely unknown. Therefore, before
investment in a full drug discovery programme, these targets need to be fully validated.8-13 Work
in our group has recently been focused on developing tool molecules for the ATPase family AAA
domain containing 2 (ATAD2) and Cats Eye Syndrome Chromosome region, candidate 2
CECR2) bromodomains, with a particular focus on novel chemotypes.14-16
(
ATAD2 is a promising oncogene which is overexpressed in multiple, unrelated human
cancers.17-18 Protein knockdown experiments have linked ATAD2 to cell survival, proliferation
2
ACS Paragon Plus Environment

Page 3 of 110
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
19
and migration pathways. CECR2 is a bromodomain containing transcription factor which is
2
0
involved in neurulation through the chromatin remodeling complex SNF2L. It has also been
implicated in oncology using a functional genomics approach and is involved in the formation of
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
γ-H2AX foci in response to DNA double strand breaks.21-22
The ATAD2 and CECR2 bromodomains are exciting targets for drug discovery and,
consequently, inhibitors for both bromodomains are starting to appear in the literature.
Computational modelling had previously predicted the ATAD2 bromodomain to be amongst the
least tractable of the bromodomain family, however, work by our group amongst others has shown
that the optimisation of sub-micromolar inhibitors is possible.23-25 The first small molecule
2
6
inhibitor (Fig. 1) published by our group was naphthyridone 1a. Despite showing relatively
modest activity compared to the leading BET inhibitors (ATAD2 pIC = 5.9) this was a significant
5
0
step forward against what was predicted to be one of the least tractable bromodomains.23
Naphthyridone 1a was optimized to afford 1b (GSK8814), a cell-permeable and selective tool
molecule.27-28 Recently, an isoform selective ATAD2 inhibitor, 2 (BAY-850) was published.29
This showed excellent ATAD2 potency and no measurable inhibition of other bromodomains. The
compound displayed a unique binding mode whereby it bound to a dimer of ATAD2 as determined
by mass spectrometry.
There are currently three reported, selective, small molecule inhibitors of CECR2 (Fig. 1).
Thepan-BRD inhibitor bromosporine 3 is a potent inhibitor of CECR2 with a pK of 8.0 but lacks
d
3
0
selectivity over other BRDs. The first selective CECR2 inhibitor was 4 (NVS-CECR2-1) which
31
is a highly selective tool molecule. It has nanomolar activity in an isothermal calorimetry (ITC)
assay (pK = 7.1) and displays robust activity in cellular fluorescence recovery after
d
photobleaching (FRAP) and nano bioluminescence resonance energy tranfer (nanoBRET) assays.
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 110
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
However, the authors describe the compound as being poorly soluble. The poor solubility of this
chemotype was addressed through a collaborative effort between the SGC and Takeda which
delivered 5 (TP-238) as a dual CECR2/bromodomain PHD finger transcription factor (BTPF) tool
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
32
compound. Crawford et al. then showed that modest selectivity for different BRDs could be
achieved by changing the N-substituent of a series pyrrolopyridines.33 In CECR2 the allyl
substituent 6a induces a small lipophilic pocket adjacent to the water nework, by moving the
gatekeeper Tyr. This was further optimised into 6b (GNE-886), which has a CECR2 TR-FRET
3
4
pIC of 7.8 equating to a 100-fold selectivity window over the closest bromodomains. The
5
0
authors demonstrated cellular permeability and target engagement, identifying it as a suitable tool
molecule.
4
ACS Paragon Plus Environment

Page 5 of 110
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
NH2
HN
NH
NH
O
N
N
N
N
HN
O
NH
NH
O
N
Cl
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
O
NH
O
NH
O
O
F
F
NH
O
1
a
BAY-850, 2
ATAD2 pIC50 = 7.7 (HTRF)
GSK8814, 1b
ATAD2 pK = 8.1 (ITC)
ATAD2 pIC50 = 5.9 (TR-FRET)
d
H
O
N
N
N N
H
O O
S
N
EtO
N
N
N
N
N
O
O O
S
N
HN
O O
S
N
N
H
HN
N
N
Bromosporine, 3
NVS-CECR2-1, 4
CECR2 pK = 7.1 (ITC)
TP-238, 5
CECR2 pKd = 8.0 (ITC)
CECR2 pK = 8.0 (ITC)
d
d
O
H
N
N
O
H
N
N
N
N
N
O
O
N
N
O
6a
GNE-886, 6b
CECR2 pIC50 = 7.8 (TR-FRET)
CECR2 pIC50 = 6.7 (TR-FRET)
Fig. 1. Small molecule inhibitors of ATAD2 and CECR2 with KAc mimetics shown in blue.
Whilst tool molecules have been identified for these bromodomains, fully validating any results
observed with these tools will rely on the availability of additional molecules which have
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 110
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
differentiated chemotypes.13-14 We recently disclosed the output of a high throughput screen
3
5
(
HTS) which identified a series of phenyl sulfonamides as ATAD2 inhibitors. Phenyl
sulfonamide (rac)-7 was identified as a suitable starting point for optimisation into a chemical
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
tool for both ATAD2 and CECR2 (Table 1). It showed activity against ATAD2 (pIC = 6.3) and
5
0
CECR2 (pIC = 5.6) using TR-FRET assays. As in our earlier work on the naphthyridone series,
5
0
we moved from a peptide, to a ligand displacement TR-FRET assay due to improved assay
2
7
robustness. Since the data was concordant (e.g. (rac)-7: 6.5 vs. 6.3), the ligand displacement
assay will be used herein. Importantly, (rac)-7 showed little affinity for either bromodomain of
BRD4 (pIC ≤4.7), used in this work as a representitive example of the BET family due to the
50
high sequence homology between family members (for full BET data on (rac)-7, (S,S)-32, and
(R)-55 see Supporting Information). High BET selectivity (>100-fold) is desirable due to the
strong phenotype observed with BET inhibitors. Phenyl sulfonamide (rac)-7 was in a good
physicochemical space as defined by ChromLogD (a chromatographic measure of lipophilicity
measured at 4.8), the Property Forecast Index (PFI (ChromLogD + #Ar), 5.8) and clogP (2.8).36-
3
8
The series had high kinetic solubility as measured by chemiluminescence nitrogen detection
-1
(
CLND) solubility (93 μg mL ) and good permeability through an artificial lipid membrane (200
-
1 39-41
nm s ).
Charged aerosol detection (CAD) solubility was also used in this work and an
4
2
internal analysis has shown that this is equivalent to CLND (see Supporting Information). The
aim of this work was to generate cell active chemical probes and optimisation of the
physicochemical properties was considered important to maximise the probability of success of
-1
good cell permeation. We stipulated targets for solubility of >100 μg mL and permeability
-1
>
100 nm s with the knowledge that one would likely compensate for the other if both of these
were not achievable.Table 1. Hit phenyl sulfonamide identified from a HTS against ATAD2.
6
ACS Paragon Plus Environment

Page 7 of 110
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
O
N
H
N
S O
O
N
O
HN
Br
O
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(rac)-7
ATAD2 (peptide) pIC (n)
6.5(8)
5
0
ATAD2 (ligand) pIC (n)
6.3(5)
5
0
LE / LLEat
CECR2 pIC (n) / Sel.
0.25 / 0.26
5.6(2) / 5x
5
0
a
BRD4 BD1 pIC (n) / Sel.
4.3(1) / 100x
5
0
BRD4 BD2 pIC (n) / Sel.
4.7(3) / 40x
4.8 / 5.8 / 2.8
93
5
0
ChromLogD / PFI / clogP
-
1
CLND (μg mL )
-
1
AMP (nm s )
200
aalso tested <4.3 (n=2) test occasions.
As discussed previously, the phenyl sulfonamides, such as (rac)-8 bind to the
bromodomain of ATAD2 in a non-canonical fashion, completely displacing the normally
3
5
conserved water network from the base of the binding pocket (Fig 2. A). The total removal of
the water network has been observed for family VIII ligands,43-45 which Grand Canonical Monte
4
6
Carlo simulations predicted to have the least stable of the BRD water networks. ATAD2 was
also predicted to have a high energy water network, consistent with our observations that these are
displaced by this series. Conversely, the water network in CECR2 was shown to be amongst the
most stable and so the CECR2 pIC of 5.6 for (rac)-8 is harder to reconcile. For ATAD2 binding
5
0
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 110
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
of (rac)-8, one of the sulfonamide oxygens replaces the interactions of water W1 and makes a
hydrogen-bond to Y1021. Additionally, the characteristic 110.5° angle of the sulfonamide,
positions the piperidine into a position where it is able to make lipophilic contacts with V1018,
Y1063 and I1074. Initial SAR studies revealed that a halide ortho to the sulfonamide was optimal
for potency, bromide typically being 5-fold more potent than chloride. In the ATAD2 crystal
structure, this bromide overlaps with the position of the W2 water in the conserved water network
and forms a halogen bond to I1056. Previous quantum mechanical analysis of the binding
interactions of this series found a relatively small contribution to the binding energy from the
halogen bond, with instead, the increase in binding affinity likely due to contibutions from multiple
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
5
interactions with other nearby residues. Truncation of the hydantoin back to an ethylamide
resulted in a 50-fold reduced potency, suggesting a significant binding contribution from this
3
5
moiety. This can be rationalised by the two binding interactions of the hydantoin carbonyl groups
with R1007 and the backbone NH of D1014 (Fig 2. B). However, it should be noted that multiple
R1007 conformations were observed in different structures, indicating a degree of flexibility for
this residue. Furthermore, the hydantoin NH did not appear to pick up any specific interactions
with the protein.
8
ACS Paragon Plus Environment

Page 9 of 110
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
O
N
H
N
S O
O
N
O
HN
Br
O
(
rac)-8
ATAD2 pIC50 (n) = 6.2(4)
3
5
Fig. 2. A) X-ray crystal structure of (rac)-8 (PDB 6S55 ) overlaid with KAc binding in ATAD2
2
5
(PDB 4QSP ). The density fitted the S-enantiomer better and so this was modelled. The
conserved network of waters (W1–W4) at the base of the binding pocket are displaced by the
phenyl sulfonamide ligand. B) Key contacts between the ligand ((rac)-8) and ATAD2 are shown
as yellow dashes. A halogen bond between the aryl bromide and the backbone carbonyl group of
I1056 is present at the base of the KAc recognition site. One of the sulfonamide oxygens makes a
H-bond to Y1021 and the two carbonyl groups of the hydantoin make H-bonds to R1007 and
D1014. C) Alternative view of B.
9
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 110
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
The potency of (rac)-7 against ATAD2 and its selectivity over the BET family (>40-fold),
combined with the good physicochemical properties of this molecule and the novel binding mode
of the phenyl sulfonamides, made it an attractive starting point for our investigation. Additionally,
potency for CECR2 presented the opportunity for divergent optimisation. Within this manuscript
we report the optimisation of (rac)-7 into potent inhibitors of ATAD2 and CECR2 respectively.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
RESULTS AND DISCUSSION
Optimisation towards an ATAD2 inhibitor:
As previously disclosed during the discovery of this template, preliminary SAR generation
efforts had focused on the aryl substituent and the hydantoin vector. The binding contribution of
35
the sulfonamide substituent was not evaluated at this stage. As a starting point for SAR expansion
towards ATAD2 potency and selectivity, piperidine 10 was either truncated or expanded to the 5-
and 7-membered analogues (9 and 11 respectively, Table 2). For synthetic feasibility, these were
prepared with a chloro-aryl substituent and an unfunctionalised hydantoin. Pyrrolidine 9 was the
most potent (pIC = 6.0) and ligand efficient of the three heterocycles against ATAD2, which was
5
0
attributed to a good shape complementarity of the substituent with the lipophilic pocket (formed
by the BC loop) of the protein. Truncation of the heterocycle to dimethyl sulfonamide 12 led to a
significant decrease in potency (pIC = 4.9) against ATAD2, a trend mirrored in CECR2 (pIC =
5
0
50
4
.2). As the pyrrolidine was the optimal size, functionalisation of the ring was then investigated.
A methyl substituent at the 3- position of the pyrrolidine (13) was tolerated but no increase in
ATAD2 potency or selectivity over CECR2 was observed for either enantiomer. Whilst both
enantiomers of the 3-methyl substituent were well tolerated, gem-dimethyl 14 showed a 0.5 log
unit decrease in potency, suggesting the steric demands of this analogue are not accommodated by
10
ACS Paragon Plus Environment

Page 11 of 110
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
the binding pocket. Interestingly, cyclopropyl 15 was comparable with the des- and mono- methyl
analogues (9 and (S)-13), however, it had a slightly lower ligand efficiency. Bis-substitution at
the 3- and 4- position was then investigated. The trans-configuration ((rac)-16) gave a 0.5 log unit
increase in potency compared to (S)-13. This included a slight increase in LE but a small decrease
in LLE . Conversely, the cis-configuration 17 was poorly tolerated and less potent (pIC = 5.7)
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
at
50
than the monomethyl examples. Extending the 3-position substituent to ethyl (rac)-18 or ether
rac)-19, was tolerated, but neither offered any improvements in terms of potency or ligand
(
efficiency compared to 9 and (S)-13. Cyclopropyl (rac)-20 was notable for its increased CECR2
potency vs. (S)-13, and was now equipotent against the two proteins. Substitution at the 2-position
to afford pyrrolidine (R)-21 was less well tolerated, with a drop in pIC50 to 5.6. The more
substantive change of a sulfonamide to a cyclohexyl sulfone (22, entry 7, pIC <4.0) was not
5
0
tolerated in accordance with the predicted importance of the directionality confered by the
sulfonamide (vide supra). Likewise, sulfonyl-imidazole 23 (entry 8) was inactive against ATAD2
and CECR2 at the concentrations tested. All of the examples tested showed no appreciable affinity
for BRD4 BD1, maintained a moderate bias for ATAD2 over CECR2, and showed desirable levels
-1
of solubility (≥116 μg mL ). Whilst no increases in ATAD2 selectivity were observed,
pyrrolidines 9, (S)-13 and (rac)-16 were identified as the optimal sulfonamides due to their potency
and ligand efficiency.
Table 2. Exploration of the sulfonamide vector.
O
R
H
N
S O
O
N
O
HN
Cl
O
11
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 110
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
ATAD2 CECR2
BRD4
BD1
CLND
sol.
ATAD2
Chrom
LogD
R
LE /
pIC (n)
50
pIC (n)
50
-
1
LLEat
/ Sel.
5.4(4) /
4x
pIC (n)
(μg mL )
50
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
0
.32 /
.38
0.28 /
.32
.29 /
0.30
.28 /
9
6.0(4)
5.6(6)
5.9(6)
4.9(6)
6.1(8)
<4.3(4)
<4.3(2)
<4.3(4)
4.4(2)
2.6
3.5
3.7
2.2
3.2
≥259
≥176
≥174
≥129
≥153
N
N
0
4.5(4) /
13x
1
0
1
2
0
0
4.9(4) /
10x
1
1
N
N
0
4.1(4)a /
6x
0
.37
0
.31 /
5.5(8) /
4x
(
S)-13
<4.3(2)
N
N
N
N
N
0
.35
0.30/
.34
.27 /
0.28
.29 /
5.3(2) /
5x
(R)-13
6.0(6)
5.5(6)
6.0(8)
<4.3(4)
<4.3(2)
<4.3(2)
4.7(1)b
3.2
3.8
3.3
3.8
175
≥153c
≥158
167
0
0
4.6(3) /
8x
14
0
5.4(4) /
4x
1
5
0
.34
(
rac)-
0.32 /
0.34
5.7(2) /
8x
6
.6(6)
1
6
12
ACS Paragon Plus Environment

Page 13 of 110
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
0
.28 /
.29
4.7(2) /
10x
1
7
5.7(6)
<4.3(2)
<4.3(4)
3.7
3.8
156
N
0
(
rac)-
0.30 /
0.32
5.9(4) /
2x
6
6
.2(8)
.0(4)
≥175
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
N
1
8
O
(
rac)-
0.28 /
0.38
5.4(2) /
4x
<4.3(2)
2.6
4.0
≥352
N
1
9
(rac)-
0.30 /
0.32
6.3(4)
6.3(3) / - <4.3(2)
≥169c
N
2
0
0.28 /
5.1(4) /
4.5(1)b
3x
(
R)-21
5.6(6)
<4.0(2)
<4.0(4)
3.1
3.6
2.1
≥201
116
N
0
.32
0
.20 /
<4.0(6) /
<4.3(2)
-
2
2
2
3
0.21
N
0.21 /
<4.0(6) /
<4.3(2)
-
≥155
N
0
.29
a
b
c
also tested <4.0 (n=1) test occasions; also tested <4.3 (n=2) test occasions; CAD solubility.
Attention then turned to the amide vector, which directed the hydantoin moiety towards
R1007 and D1014 (Fig 2). The HTS had identified the hydantoin as an effective substituent but
it was important to fully explore the SAR of this region to further elicidate the differences in
35
ATAD2 and CECR2 binding (Table 3). As before, the aryl chloride was used for synthetic
feasibility and the more optimal pyrrolidine sulfonamide was employed. As expected, the
unsubstituted hydantoin 9 showed good potency (pIC = 6.0), though, it was characterised by
5
0
13
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 110
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
-1
a low artificial membrane permeability (AMP <3 nm s ). This is most likely due to the low
lipophilicity of this molecule. The methyl (rac-24) and dimethyl (25) analogues increased
permeability further and maintained a similar ATAD2 potency (pIC50 = 5.9 and 5.8
respectively). Here, alpha substitution of the H-bond donor NH increases lipophilicity by 0.4–
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
0
.8 log units allowing it to more easily pass though the hydrophobic portion of membranes.
The geminally substituted (rac)-26 further raised the predicted permeability achieving a
-1
desirable value of 100 nm s and also maintained potency at ATAD2. Interestingly, 2,4-
oxazolidinedione 27 (pIC = 5.9) was also well tolerated, supporting the crystallographic
5
0
observation that the hydantoin NH wasn’t making any specific interactions. This was further
supported by pyrrolidinedione 28, which also maintained a similar potency of 5.8 (c.f.
hydantoins 9, 24–26). However, further simplification of the hydantoin substituent to 2-
pyrrolidinone 29 saw a large decrease in potency (pIC = 4.3). This shows, in accord with the
5
0
crystallography (Fig. 2), that both carbonyl groups are important for binding to ATAD2. In
fact, 2-pyrrolidinone 29 offered little difference in potency compared to truncated acetamide
-1
3
0 (pIC = 4.5). Pleasingly, 24–30 maintained good solubility (>94 µg mL ), and, none of the
50
compounds tested showed any appreciable affinity for the BET family. Unfortunately, potency
for CECR2, largely tracked with ATAD2 and selectivity between the two bromodomains was
not forthcoming. In summary, hydantoin (rac)-26 offered the highest potency for ATAD2
combined with the best balance of physicochemical properties.
Table 3. Investigation of the Hydantoin substituent.
N
H
N
S O
O
R
O
Cl
14
ACS Paragon Plus Environment

Page 15 of 110
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
CAD
Sol.
ATAD2 CECR2
BRD4
Chrom
LogD /
PFI
AMP
(nm
ATAD2
R
LE /
pIC (n)
BD1
50
pIC (n)
(μg
5
0
/ Sel.a
pIC (n)
s )
-1
LLEat
50
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
mL-1)
O
0
.32 /
0.38
.30 /
5.4(2) /
4x
N
9
6.0(4)
<4.3(4)
<4.3(2)
<4.3(2)
<3.3(2)c
<4.3(2)
2.6
3.0
3.4
4.2
3.6
3.3
3.4
≥343
<3
12
HN
O
O
0
(
rac)
5.2(4) /
5x
N
≥259e
5
.9(6)
HN
O
0
.34
-24
O
0
0
0
.28 /
5
.3(4) /
N
2
5
5.8(4)
≥158
148e
94
18
HN
0
.30
3
x
O
O
.27 /
(
rac)
26
5.8(2) /
2x
N
6
.0(4)
100
54
HN
N
0
.28
-
O
O
O
O
.31 /
5
.8(4) /
2
7
5.9(2)
0
.37
0.31 /
.35
0.24 /
.25
.32 /
1
x
O
5
.4(4) /
N
2
8
9
0
5.9(6)
4.3(6)
4.5(4)
<4.3(2)
<4.3(2)
3.7(2)d
≥106
181e
≥122
42
0
3
x
O
O
4.7(2) /
3x(C)
4.1(4)b /
3x
2
3
N
120
270
0
0
H
3.4
0.29
a(C) denotes selectivity for CECR2; also tested <4.0 (n=2) test occasions; also tested <4.3 (n=1)
b
c
d
e
test occasions; also tested <4.3 (n=4) test occasions; CLND solubility;
15
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 110
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
If the SAR of the sulfonamide and amide groups were additive, the combination of the
three most potent sulfonamide groups (Table 2), the 3-methyl-3-cyclopropyl hydantoin (Table 3)
and bromide should afford optimal ATAD2 inhibitors (Table 4). Unfortunately, (S)-31 did not
show the expected increase in potency when the pyrrolidine was combined with the other
substituents. As the optimisation was carried out using Cl analogues, the non-additivity of the SAR
could arise from a subtle shift in binding position when moving to the larger Br atom. However,
addition of the 3-methyl substituent increased potency by 0.3–0.5 log units leading to a pIC of
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
5
0
7
.2 for the most potent diastereomer ((S,S)-32). Phenyl sulfonamide (S,S)-32 maintained the
excellent solubility and physico-chemical profile of the starting hit and had a 5-fold bias for
ATAD2 over CECR2. This profile was a marginal improvement over the hydantoin epimer (R,S)-
3
2, and consequently, (S,S)-32 was nominated for further profiling. In agreement with previous
SAR, trans 3,4-dimethyl pyrrolidine 33 increased affinity for ATAD2 (pIC = 7.6). However, 33
5
0
-1
was less soluble (49 µg mL ), potentially driven by the increased lipophilicity of this analogue.
Table 4. Combination of hydantoin substituent and sulfonamide vector.
O
R
H
N
S O
O
N
O
HN
Br
O
ATAD2 ATAD2 CECR2
BRD4
BD1
CLND
sol.
Chrom
LogD
R
pIC50
n)
LE /
LLEat
0.30 /
pIC (n)
50
-
1
(
/ Sel.
6.4(6) /
2x
pIC (n)
(μg mL )
50
(S)-31
6.6(8)
<4.3(2)
4.1
>195b
N
0
.31
16
ACS Paragon Plus Environment

Page 17 of 110
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
(
S,S)-
0.32 /
0.30
6.5(2) /
5x
7
7
.2(6)
.0(8)
<4.3(2)
<4.3(4)
4.4(2)
4.7
4.8
5.2
129
146
47
N
N
3
2
(R,S)-
0.31 /
0.30
6.9(4) /
1x
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
2
0.30 /
6.9(2) /
5x
33a
7.6(2)
N
0
.29
aTested as a diastereomeric mixture of hydantoin epimers with unknown trans-stereochemistry of
b
the pyrrolidine substituents; CAD solubility.
A crystal structure of (R,S)-32, was resolved in ATAD2 (Fig. 3. A–B). The binding mode
of (R,S)-32 was consistent with that of the initial hit (Fig 2). The aromatic bromide sits deep within
the binding pocket and displaces the conserved water molecules in ATAD2. The (S)-methyl
pyrrolidine group occupies a lipophilic pocket surrounded by the conserved Y1021, N1064 and
the gatekeeper I1074 residues. The hydantoin carbonyl groups can then be seen making the
previously observed H-bonding interactions with the backbone NH of D1014 and R1007. The
methyl and cyclopropyl groups, which were added to improve permeability, do not appear to pick
up any specific contacts, in agreement with the biochemical data generated.
17
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 110
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Fig. 3. X-ray crystal structure (PDB 6YB4) of (R,S)-32 (teal) in ATAD2 (silver) and a docking
4
7
model of (S,S)-32 (lime) in CECR2 (lilac, PDB 3NXB ): A) Pose showing (R,S)-32 sitting deep
within the binding pocket, where the ligand displaces the normally conserved water network in
the KAc recognition site; B) Rotated pose showing the hydantoin carbonyl groups making
hydrogen bonds to both R1007 and D1014 and one of the sulfonamide oxygens hydrogen
bonding to Y1021; C) pose showing there is space for (S,S)-32 to displace the waters in CECR2
and make key contacts with M506 and Y471. D) Rotated pose showing that (S,S)-32 makes a H-
bond to D464 NH but does not make any specific interactions with the WPF shelf in CECR2.
18
ACS Paragon Plus Environment

Page 19 of 110
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
(
S,S)-32 represented the most optimised ATAD2 inhibitor to date, boasting excellent potency and
a novel mode of binding. To assess the inherent selectivity of this series of inhibitors, the wider
bromodomain selectivity of (S,S)-32 was investigated, firstly, using in house TR-FRET assays and
then DiscoverX’s full BROMOscan panel (Fig. 4). Pleasingly, (S,S)-32 was >200-fold selectivity
over the BET family of bromodomains, using BRD4 as a representitive example. The high potency
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
of (S,S)-32 was confirmed in the BROMOscan ATAD2A (pK = 8.0) and ATAD2B (pK = 7.6)
d
d
assays. Significant potency was observed for seven other bromodomains: Bromodomain adjacent
to zinc finger domain protein 2A (BAZ2A, pK = 7.8), BRD1 (pK = 6.9), Bromodomain and PHD
d
d
finger-containing protein 3 (BRPF3, pK = 6.9), CECR2 (pK = 7.6), general control
d
d
nonderepressible 5 (GCN5L2, pK = 6.6), TATA-binding protein-associated factor 1 (TAF1(2),
d
pK = 7.8) and TAF1-like (TAF1L(2), pK = 7.6) bromodomains. Greater than 30-fold selectivity
d
d
was observed for all other bromodomains tested against. Interestingly, the stability of the water
network in these bromodomains has previously been predicted to vary greatly, therefore, this level
46
of off-target activity was suprising. It may suggest that displacement of the canonical waters is a
viable approach for multiple targets in the future. Cellular target engagement was demonstrated
using a CECR2 nanoBRET assay, in which a fluorescent bromosporine ligand was competitively
4
8
displaced from CECR2 in HEK293 cells. (S,S)-32 showed activity (pIC50 = 6.5) in concordance
with it’s CECR2 potency, demonstrating its ability to engage a cellular target. Overall (S,S)-32
was identified as a potent ATAD2 inhibitor with good physicochemical profile and evidence of
cellular target engagement. Despite the residual activity against other bromodomains, given the
historical difficulties with achieving highly potent ATAD2 inhibitors, the unique binding mode
and novel chemotype will make (S,S)-32 (GSK388) a useful addition to the ATAD2 chemical
toolbox when used in conjuction with other differentially selective probes.
19
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 110
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Table 5. Profile of (S,S)-32
N
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
O
H
N
S O
O
N
O
HN
Br
O
GSK388
(
S,S)-32
ATAD2 pIC (n)
7.2(6)
50
LE / LLEat
0.31 / 0.30
4.7
CHROMlogD
-1
CLND sol. (µg mL )
129
-1
AMP (nm s )
99
BRD4 BD1 pIC (n) / Sel.
<4.3(2) / >630x
4.9(2) / 200x
8.0
5
0
BRD4 BD2 pIC (n) / Sel.
50
ATAD2 pKd
ATAD2B pK / Sel.
7.7 / 2x
7.8 / 2x
6.9 / 13x
6.9 / 13x
7.6 / 3x
6.6 / 25x
7.8 / 2x
7.6 / 3x
d
BAZ2A pK / Sel.
d
BRD1 pK / Sel.
d
DiscoverX
BRPF3 pK / Sel.
d
BROMOscan
CECR2 pK / Sel.
d
GCN5L2 pK / Sel.
d
TAF1(2) pK / Sel.
d
TAF1L(2) pK / Sel.
d
20
ACS Paragon Plus Environment

Page 21 of 110
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
CECR2 nanoBRET pIC (n)
6.6(4)
5
0
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Fig. 4. DiscoverX’s BROMOscan results of (S)-32.
21
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 110
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Optimisation towards a CECR2 selective inhibitor:
Having optimised our phenyl sulfonamide series for ATAD2 potency, it was intriguing that
potency at CECR2 (and indeed 7 other bromodomains) had not been reduced more significantly.
Since we did not have a CECR2 crystal system, docking of (S,S)-32 in the apo structure of CECR2
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
4
7
(
PDB 3NXB ) was solved to aid understanding of the differences between the two proteins (Fig.
3
. C–D). Simulation suggested that the water network in CECR2 is more stable than in ATAD2,46
however the binding pocket of CECR2 is able to accommodate (S,S)-32 when all its water is
removed. Considering the trends in SAR, we believe that the phenyl sulfonamide ligands bind to
both bromodomains with displacement of the canonical waters in a similar manner. In the CECR2
model, one of the sulfonamide oxygens hydrogen bonds to Y471 and the aryl bromide forms a
halogen bond to the carbonyl group of M506 (the CECR2 residue analogous to I1056 in ATAD2).
The (S)-methyl pyrrolidine then occupies a lipophilic pocket formed of Y471, N514 and Y520.
When compared with the crystal structure of ATAD2, several notable residue changes were
apparent (Fig. 5). Firstly, the I1074 gatekeeper residue in ATAD2 is replaced by Y520 in CECR2.
In ATAD2 the (RVF) shelf region is formed from R1007, V1008 and F1009, compared to W457,
P458 and F459 in CECR2 which form a WPF shelf. It was hypothesised that these residue changes
could be exploited to obtain enhanced CECR2 potency and selectivity.
22
ACS Paragon Plus Environment

Page 23 of 110
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Fig. 5. A crystal structure of (R,S)-32 (lime) in ATAD2 (PDB 6YB4) with point changes
4
7
between the CECR2 (lilac, PDB 3NXB ) and ATAD2 (grey) domains annotated.
In ATAD2, the hydantoin makes H-bonds using both carbonyl groups; one to the backbone
NH of D1014 and the other to the flexible sidechain of R1007 on the RVF shelf (Fig. 3. A–B). In
CECR2, according to the model, one of the hydantoin carbonyl groups can still make a H-bond to
the backbone NH of D464. However, there is no suitable donor for the other carbonyl group on
the WPF region due to the residue change from R1007 to W457. SAR work had shown that the
two carbonyl groups were vital for ATAD2 potency and that moderate selectivity for CECR2 could
be gained through removal of one of them, albeit with a significant reduction in potency (Table 3).
Therefore, a strategy was adopted which aimed to build out towards the WPF residues with a
substituent which could interact favourably with W457 in CECR2.
Aromatic rings have been shown to form π-stacking interactions with the WPF shelf in the
49
BET family. Whilst the shelf in CECR2 is blocked by the gatekeeper Y520, it was hypothesised
that substituents such as benzimidazolinone (S)-34 (Table 6) might still interact with W457 whilst
23
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 110
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
maintaining a H-bond with the backbone NH of D464. This would also remove one of the carbonyl
groups which had appeared crucial for ATAD2 activity, hopefully leading to an increase in
selectivity. Pleasingly, benzimidazolinone (S)-34 maintained potency at CECR2 (pIC = 6.4)
5
0
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
relative to hydantoin (S,S)-32 whilst activity at ATAD2 dropped 20-fold. The corresponding
benzoxazolidine (S)-35 showed a further 10-fold increase in potency towards CECR2 (pIC =
5
0
7
.4) and 5-fold selectivity over ATAD2. There was only a marginal increase in LLE for (S)-35
at
relative to (S,S)-32, due to the raised lipophilicity of this analogue. Compound (S)-35 was also
more potent at BRD4 BD1 (pIC = 5.0) although it still maintained a 250-fold selectivity window.
5
0
Based on previously discussed ATAD2 crystallography and CECR2 modelling (vide supra), it was
hypothesised that the carbamate in (S,R)-35 retained a H-bonding interaction with both proteins.
Benzotriazole (S)-36 and benzisoxazole (S)-37 were designed to explore whether this interaction
could be maintained when the H-bond acceptor was located within an aromatic ring. Both
compounds showed good potency for CECR2 (pIC = 6.7 and 6.6 respectively) and maintained a
5
0
bias over ATAD2 and high selectivity over BRD4 BD1. Removal of one of the heteroatoms to
give benzofuran (S)-38 showed a 32-fold drop-off in CECR2 potency, suggesting that a H-bond
acceptor at the 2-position was crucial. Unfortunately, compounds 34-38 showed reduced solubility
-1
(
≤25 μg mL ), most likely due to their greater lipophilicity and aromaticity. As a result, none of
the compounds tested were ideal tool molecules.
24
ACS Paragon Plus Environment

Page 25 of 110
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Table 6. Targeting the WPF shelf of CECR2.
N
H
N
S O
O
R
O
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Br
BRD4
BD1
CAD
Sol.
CECR2 ATAD2
CECR2
Chrom
LogD
R
LE /
pIC (n)
50
pIC (n)
pIC50
(μg
5
0
LLEat
/ Sel.a
mL-1)
(n)
O
(
S,S)
- 32
S)-
0.30 /
0.28
7.2(6) /
5x(A)
N
129c
6.5(2)
<4.3(2)
4.7
HN
O
(
0.29 /
0.23
5.9(6) /
3x
N
6
7
.3(4)
.2(4)
<4.3(3)
5.0(2)
4.6
5.7
5.3
6.2
6.7
25
21
18
3c
HN
3
4
O
(
S)-
0.34 /
0.29
5.7(4) /
32x
N
O
3
5
O
(
S)-
0.32 /
0.26
5.7(4) /
10x
4.8(1)b
4.6(1)b
<4.3(2)
N
N
6.7(4)
3
6
N
O
(
S)-
0.31 /
0.26
6.0(4) /
4x
6
5
.6(6)
.1(5)
N
3
7
(S)-
0.24 /
0.14
5.6(2) /
3x(A)
0c.d
O
3
8
a(A) denotes selectivity for ATAD2 over CECR2; also tested <4.3 (n=1) test occasions; CLND
b
c
d
solubility; also tested <0 (n=1) test occasions.
25
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 110
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Whilst the use of an aromatic amide substituent increased potency and selectivity at
CECR2 it was unclear whether this was indeed the optimal substituent. To investigate the
requirement for the phenyl ring of the benzotriazole (S)-36 was truncated. Unsubstituted triazole
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(S)-39 (Table 7) showed weak CECR2 potency (c.f. benzotriazole (S)-36) but maintained ligand
efficiency compared to the leading biaryls (34–37). It was therefore logical to investigate alternate
substituents to look for improved physicochemical properties whilst maintaining the potency of
benzotriazole (S)-36. Dimethyltriazole (S)-40 showed increased CECR2 inhibition (pIC = 6.0)
5
0
whilst maintaining the ligand efficiency and ATAD2 selectivity of (S)-39. This suggested that the
3
aromatic ring could be replaced by sp groups with similar overall efficiency. A diversity
orientated approach which initially looked at range of substitution at the 4-position was undertaken
and the key examples are discussed herein. Neither a propyl or ether group increased affinity for
CECR2 ((S)-41 and (S)-42), but, the installation of a methylene linked dimethylamine afforded
(S)-43 (pIC = 6.6), a potent CECR2 inhibitor which was 5-fold selective over ATAD2. (S)-43
5
0
was equipotent with benzotriazole (S)-36 but significantly less lipophilic and highly soluble (≥234
-1
µg mL ). The same substituents at the 5-position were then prepared and interestingly, all three
analogues (44–46) showed increased potency for CECR2 and improved selectivity over ATAD2.
In particular, dimethylamine (S)-46 displayed 50-fold selectivity for CECR2 (pIC50 = 7.1).
-1
Importantly, it was also soluble (≥219 µg mL ) and therefore demonstrated, for the first time, that
desirable CECR2 potency and selectivity were possible with compounds with a suitable
physicochemical profile.
26
ACS Paragon Plus Environment

Page 27 of 110
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Table 7. Investigating triazole substitution.
N
H
N
S O
O
R
O
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Br
CECR2 ATAD2
BRD4
BD1
CAD
Sol.
CECR2
Chrom
LogD
R
LE /
LLEat
0.30 /
0.30
pIC (n)
50
pIC (n)
50
-
1
/ Sel.
5.4(6) /
-
pIC (n)
(μg mL )
50
(
S)-
5
N
4
5.4(6)
<4.3(2)
<4.3(2)
4.4(2)
3.9
4.4
5.1
4.1
3.1
N
≥166
N
3
9
(S)-
0.30 /
0.30
5.2(6) /
6x
N
6
5
5
.0(6)
.8(6)
.8(6)
≥
181
N
N
4
0
(S)-
0.28 /
0.24
5.8(6) /
-
N
N
48
N
4
1
(S)-
0.28 /
0.31
5.6(8) /
2x
N
N
<4.3(2)
<4.3(2)
O
N
≥235
≥234
N
N
4
2
(S)-
0.31 /
0.34
5.9(6) /
5x
N
N
6.6(4)
7.1(5)
43
(S)-
0.35 /
0.30
5.5(6) /
40x
5.0(1)
5.1
4.3
N
N
1
9
44
N
N
(S)-
O
0.32 /
0.35
6.0(6) /
4x
4.3(1)a
6
.6(6)
N
N
≥216
4
5
27
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 110
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
(
S)-
N
0.34 /
0.37
5.4(6) /
50x
7.1(5)
<4.3(2)
4.5
N
N
≥
219
46
N
aalso tested <4.3 (n=1) test occasions.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Building on the success of the tertiary amine derivatives, and the tolerance of substitution at both
the 4- and 5- positions, we wanted to explore the effect of conformationally restricting the
dimethylamine group. To do this, a range of triazolopiperidines (47–53, Table 8) were prepared.
(
S)-46 had afforded higher potency relative to its regioisomer, (S)-43 (Table 7). Therefore,
conformational restriction of this amine was investigated first. Interestingly, triazolopiperidine (S)-
7 was 0.4 log units less potent at CECR2 (pIC50 = 6.7) relative to dimethylamine (S)-46
4
suggesting that conformational restriction did not maintain the dimethylamine in its favoured
binding conformation. Regioisomeric triazolopiperidine (S)-48 was prepared and pleasingly, was
both highly potent (CECR2 pIC = 7.3) and 40-fold selective for CECR2. Additionally, (S)-48
50
-
1
had a low ChromLogD (3.7) and was highly soluble (≥211 µg mL ). The methyl group could be
extended to give ethylamine (S)-49 which had a comparable profile to (S)-48 albeit with a slight
increase in ChromLogD. Branching of the alkyl group gave isopropyl amine (S)-50 which was
also highly potent at CECR2 (pIC = 7.6), equating to 79-fold selectivity over ATAD2. Despite
5
0
the increased lipophilicity compared to (S)-48, (S)-49 and (S)-50 were still highly soluble (>238
-1
µg mL ). In an effort to further understand the tolerated substitution of the amine,
trifluoroethylamine (S)-51 was tested. A 0.5 log unit reduction in CECR2 potency was observed
relative to ethylamine (S)-49. The increased lipophilicity of (S)-51 was also detrimental to the
-1
solubility (49 µg mL ). Acylation of the amine gave acetamide (S)-52 which was potent at CECR2
(
pIC = 7.0) but was only 16-fold selective over ATAD2. The unsubstituted triazolopiperidine
50
28
ACS Paragon Plus Environment

Page 29 of 110
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
(
S)-53 was poorly tolerated in CECR2 (pIC = 6.6) but still maintained a 10-fold selectivity
5
0
window. Overall, whilst there seemed scope to extend from the -NMe to larger substituents, (S)-
48 already exhibited a highly promising profile worthy of further investigation.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Table 8. Investigation the SAR of triazolopiperidines.
N
H
N
S O
O
R
O
Br
BRD4
BD1
CAD
Sol.
CECR2 ATAD2
CECR2
Chrom
LogD
R
LE /
pIC (n)
50
pIC (n)
pIC50
(μg
5
0
LLEat
/ Sel.
mL-1)
(n)
(S)-
N
0.31 /
0.34
5.7(4) /
8x
6
.6(4)
<4.3(2)
3.9
≥186
N
N
4
7
N
(S)-
0.33 /
0.37
5.7(10) /
40x
N
N
7.3(8)
7.3(8)
<4.3(4)
4.5(3)a
4.3(1)a
4.5(2)
3.7
4.1
4.5
5.3
≥211
569
238
49
N
N
4
8
(
S)-
49
S)-
0.33 /
0.35
5.6(10) /
50x
N
N
N
N
(
0.33 /
0.32
5.7(2) /
63x
N
7
.5(3)
N
5
0
N N
(S)-
0.28 /
0.29
5.6(4) /
16x
N
F3C
N
N
6.8(3)
5
1
N
29
ACS Paragon Plus Environment