Discovery of a potent, cell penetrant, and selective p300/CBP-associated factor (PCAF)/general control nonderepressible 5 (GCN5) bromodomain chemical probe

Article abstract of DOI:10.1021/acs.jmedchem.6b01566

P300/CREB binding protein associated factor (PCAF/KAT2B) and general control nonderepressible 5 (GCN5/KAT2A) are multidomain proteins that have been implicated in retroviral infection, inflammation pathways, and cancer development. However, outside of viral replication, little is known about the dependence of these effects on the C-terminal bromodomain. Herein, we report GSK4027 as a chemical probe for the PCAF/GCN5 bromodomain, together with GSK4028 as an enantiomeric negative control. The probe was optimized from a weakly potent, nonselective pyridazinone hit to deliver high potency for the PCAF/GCN5 bromodomain, high solubility, cellular target engagement, and ≥18000-fold selectivity over the BET family, together with ≥70-fold selectivity over the wider bromodomain families.

Full text of DOI:10.1021/acs.jmedchem.6b01566

Subscriber access provided by UNIV OF CALIFORNIA SAN DIEGO LIBRARIES
Article
Discovery of a Potent, Cell Penetrant and Selective p300/
CBP-Associated Factor (PCAF)/General Control Non-
Derepressible 5 (GCN5) Bromodomain Chemical Probe
Philip G Humphreys, Paul Bamborough, Chun-Wa Chung, Peter D Craggs, Laurie J. Gordon,
Paola Grandi, Thomas G Hayhow, Jameed Hussain, Katherine L. Jones, Matthew Lindon,
Anne-Marie Michon, Jessica F Renaux, Colin J. Suckling, David F Tough, and Rab K. Prinjha
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.6b01566 • Publication Date (Web): 21 Dec 2016
Downloaded from http://pubs.acs.org on December 22, 2016
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 free 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 accessible to all
readers and 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.
Journal of Medicinal Chemistry 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 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Discovery of a Potent, Cell Penetrant and Selective
p300/CBPꢀAssociated Factor (PCAF)/General
Control NonꢀDerepressible 5 (GCN5) Bromodomain
Chemical Probe
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Philip G. Humphreys,*† Paul Bamborough,‡ Chunꢀwa Chung, ‡ Peter D. Craggs, ‡ Laurie
Gordon,‡ Paola Grandi, § Thomas G. Hayhow,† ┴Jameed Hussain, ‡┴ Katherine L. Jones,†
Matthew Lindon,† AnneꢀMarie Michon,§ Jessica F. Renaux,† Colin J. Suckling, ₸ David F.
Tough,† Rab K. Prinjha†
† Epigenetics Discovery Performance Unit and ‡Platform Technology and Science,
GlaxoSmithKline R&D, Stevenage, Hertfordshire SG1 2NY, United Kingdom
§Cellzome GmbH, Molecular Discovery Research, GlaxoSmithKline, Meyerhofstrasse 1, 69117
Heidelberg, Germany
₸WestCHEM, Department of Pure and Applied Chemistry, Thomas Graham Building, University
of Strathclyde, 295 Cathedral Street, Glasgow, G1 1XL, United Kingdom
KEYWORDS Bromodomain, PCAF, GCN5, chemical probe, epigenetics, GSK4027, GSK4028
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 49
1
2
3
4
5
6
7
8
9
ABSTRACT: p300/CREB binding protein associated factor (PCAF/KAT2B) and general control
nonꢀderepressible 5 (GCN5/KAT2A) are multidomain proteins that have been implicated in
retroviral infection, inflammation pathways and cancer development. However, outside of viral
replication, little is known about the dependence of these effects on the Cꢀterminal
bromodomain. Herein, we report GSK4027 as a chemical probe for the PCAF/GCN5
bromodomain, together with GSK4028 as an enantiomeric negative control. The probe was
optimized from a weakly potent, nonꢀselective pyridazinone hit to deliver high potency for the
PCAF/GCN5 bromodomain, high solubility, cellular target engagement and ≥18000 fold
selectivity over the BET family, together with ≥70 fold selectivity over the wider bromodomain
families.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
INTRODUCTION
Bromodomains are structurally conserved protein modules which selectively bind to acetylꢀ
lysine (KAc) marks on proteins and, in doing so, form part of the epigenetic regulation of gene
transcription.1 Bromodomains have recently emerged as tractable2 and importantly,
therapeutically attractive targets for a number of disease areas including inflammation and
oncology.3–6 The main body of research in the bromodomain field has focused on the BET
(bromodomain and extra terminal) family and multiple small molecule inhibitors have been
disclosed, as recently reviewed elsewhere.7–10 It is of note that the publication of a variety of
diverse chemical probes has enabled the wide ranging evaluation of the BET family as
therapeutic targets. Historically, bromodomains outside of the BET family have received less
attention in the literature, in part, due to the lack of chemical probes.11 High quality chemical
probes from diverse chemotypes have been, and will continue to be a critical component in
ACS Paragon Plus Environment

Page 3 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
enabling our biological understanding of the function and target validation of the 53 nonꢀBET
members of the bromodomain phylogenetic tree.12,13 To this end, a number of chemical probes
and inhibitors for several nonꢀBET bromodomains, for example CREBBP/EP300,14–19
BRD7/9,20–22 BRD9,23 BRPF1,24 ATAD225 and SMARCA2/426 have been disclosed, as has been
recently reviewed elsewhere.27–30 Importantly, many of these probes are readily available31 and
are already being used to elucidate the biological function of these protein reader domains in
health and disease.32–34
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
KAT2B (lysine acetyl transferase 2B), also known as PCAF (p300/CBPꢀassociated factor), is a
multidomain protein containing an acetyltransferase domain, E3 ligase ubiquitination domain
and a Cꢀterminal bromodomain.35,36 PCAF functions as a transcriptional coꢀactivator and has
been implicated in many cellular processes, including several inflammationꢀrelated
pathways,37,38 oncology,39–41 neuroꢀdegeneration42 and viral replication.43,44 KAT2A, also known
as GCN5 (general control nonꢀderepressible 5), possesses a similar domain structure and exhibits
high homology with PCAF, particularly in the Cꢀterminal bromodomain. GCN5 is reported to
play important roles in cell proliferation and differentiation, nucleosome assembly as well as DNA
–
damage repair,45 49 and has been implicated in certain cancers, including nonꢀsmall cell lung
cancer and breast cancer.47,50–52 Acetyltransferase activity is recognized to play a key role in
PCAF/GCN5 function, linked to acetylation of lysine residues on both histone proteins and
transcription factors.46,53 In contrast, the function of the PCAF/GCN5 bromodomain is less well
understood. Thus, while studies in yeast (which possess only a single PCAF/GCN5 family
member called GCN554,55) have shown that pointꢀmutations disrupting the ability of the GCN5
bromodomain to interact with acetylꢀlysine can lead to decreases in GCN5ꢀdependent
transcription and histone acetylation,46,56 very little is known about the role of PCAF and GCN5
bromodomains in mammalian cell function.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 49
1
2
3
4
5
6
7
8
9
To this end, there have been several reports concerning the development of PCAF
bromodomain inhibitors. In 2005, Zhou and coꢀworkers used an NMR screen to identify the first
small molecule PCAF bromodomain inhibitor, diamine 1 (Figure 1).57 This compound was
shown to disrupt a specific interaction between PCAF and the K50Ac HIV TAT protein, which
is essential for viral replication. Subsequent work to optimize this series succeeded in improving
the biochemical potency, resulting in compound 2.58 A fragment screen against the PCAF
bromodomain using surface plasmon resonance delivered fragment binders, such as amide 3
which shows 250 µM affinity for PCAF.59 Knapp and coꢀworkers also adopted a fragment based
approach to identify PCAF binders, such as compound 4, as potential starting points for more
potent and selective PCAF inhibitors.60
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment

Page 5 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 1. PCAF inhibitors showing their reported affinity to the PCAF bromodomain
Demonstrating significant improvements in potency from what had been previously reported,
Genentech and Constellation Pharmaceuticals recently disclosed, in three patent applications, a
number of highly potent dual PCAF/GCN5 inhibitors from three different chemotypes, as
exemplified by the most active analogues from each series, compounds 5ꢀ7 (Figure 1).61–63
During the preparation of this manuscript, the Genentech/Constellation pyridazinone patent
application was published and the similarity of the pyridazinone structures, such as 7,63 to the
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 49
1
2
3
4
5
6
7
8
9
work presented in this manuscript is acknowledged. However, to the best of our knowledge, the
discovery and optimization of the pyridazinone template as PCAF/GCN5 inhibitors by
Genentech and Constellation scientists has not yet been reported in the peerꢀreviewed literature.
As detailed above, high quality chemical probes for the PCAF/GCN5 bromodomains would
facilitate further elucidation of the bromodomain function. To this end, in 2011 GSK set out to
generate potent and selective inhibitors of the PCAF bromodomain. Due to the profound
phenotype associated with BET inhibition, it was considered that the probe should show PCAF
pIC50 >7 with a minimum of 100ꢀfold selectivity (2 log units) over the BET family. Cellular
target engagement was critical, together with >30ꢀfold selectivity across other bromodomain
families and an understanding of broader nonꢀbromodomain activity.64 Another consideration
was the development of a negative control, a compound with highly similar properties to the
chemical probe, but with reduced activity against PCAF to provide further confidence that any
phenotype could be associated with inhibition of bromodomain function.13 Selectivity against
GCN5 for the chemical probe was not expected, or indeed targeted, due to the high homology
with the PCAF bromodomain mentioned previously (Figure 2c).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Here, we report the optimization of a weakly active, nonꢀselective pyridazinone hit to deliver a
potent, selective, soluble and cell penetrant chemical probe for the PCAF/GCN5 bromodomains
and the enantiomeric negative control.
RESULTS AND DISCUSSION
PCAF, like the BET family, is classified as a typical bromodomain with conserved asparagine
and tyrosine residues responsible for KAc recognition (Figure 2).60,65 Comparison of the Xꢀray
crystal structures of apo PCAF (Figure 2a) and BRD4 BD1 (Figure 2b) reveals a profound
ACS Paragon Plus Environment

Page 7 of 49
Journal of Medicinal Chemistry
1
2
difference in the ZA channel regions due to differences in the amino acids.66 PCAF, like the BET
family, contains a lipophilic WPF shelf (Figure 2c). However, in the PCAF bromodomain the
bulky Tyr809 gatekeeper residue blocks access and creates a far narrower binding pocket when
compared with BRD4 BD1. In contrast, BRD4 BD1 contains a smaller Ile146 which permits
access to the WPF shelf, with binding to this lipophilic region typically observed with BET
bromodomain inhibitors such as IꢀBET151 (Figure 2b). Another important difference is an
acidic residue in the ZA loop region, Glu756 in PCAF, whose 3D equivalent in the BET family
is a lipophilic leucine (BRD4 BD1 Leu92), which provides the potential for a selective
interaction (vide infra).
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
c.
Figure 2. (a) Xꢀray crystal structure of apo PCAF bromodomain (pdb: 3gg3) (grey). Hydrogenꢀ
bonds are shown as yellow dashed lines and selected water molecules as red spheres; (b) Xꢀray
crystal structure of BRD4 BD1 bromodomain (cyan) in complex with IꢀBET151 (pdb: 3zyu)
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 49
1
2
3
4
5
6
7
8
9
(yellow); (c) Sequence alignment of PCAF, GCN5 and BRD4 BD1. The WPF moiety is colored
orange, the conserved acetylꢀlysineꢀbinding asparagines and tyrosine are colored red and green
respectively and the gatekeeper residue is colored blue.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Discovery of the Pyridazinone Series. As previously disclosed,67 as part of our bromodomain
discovery strategy we assembled a focused set of compounds containing known and potential
KAc mimetics. This set of ~30,000 compounds was screened in single shot format at 10 µM
against PCAF using a fluorescence polarization assay based on a labeled small molecule
promiscuous bromodomain binder. A number of single shot hits, including pyridazinone 8 were
progressed to full curve analysis and an orthogonal 15Nꢀ1H HSQC NMR screen to rule out
potential assay interference. Full curve screening of 8 in a timeꢀresolved fluorescence resonance
energy transfer (TRꢀFRET) assay demonstrated moderate potency with pIC50 4.8 (IC50 16 µM)
and an encouraging ligand efficiency of 0.37 (Figure 3). Profiling compound 8 more broadly
revealed equipotent activity against the BET family, with BRD4 BD1 (Nꢀterminal, bromodomain
1) used as a representative member of the BET family for optimization throughout.
Pyridazinone 8 also demonstrated high levels of aqueous solubility as determined by
chemiluminescent nitrogen detection (CLND) and a ChromLogDpH7.4 3.3. In the NMR
experiment, a subset of peaks were shifted upon binding to pyridazinone 8, consistent with
interactions at the KAc binding site. Although we thought it likely that the carbonyl group
would mimic the KAc hydrogenꢀbonding interactions to Asn803 and via water to Tyr809, it was
unclear whether the chloro or methyl group of 8 was occupying the PCAF methyl pocket.68
ACS Paragon Plus Environment

Page 9 of 49
Journal of Medicinal Chemistry
1
2
3
4
Figure 3. Profile of pyridazinone hit 8
5
6
7
8
9
With confirmation of PCAF binding by orthogonal assay formats, we selected pyridazinone 8
as a synthetically tractable, soluble, small and ligand efficient start point for further optimization.
Initial SAR was rapidly developed around this template through either external purchase
(compounds 10ꢀ12) or straightforward SNAr chemistry with commercially available 4,5ꢀ
dichloropyridazinone 9 (Scheme 1).69 Although the methodology typically gave a ~3:1 mixture
of 4ꢀ and 5ꢀpositional isomers, substitution at the desired 5ꢀposition is favored and these products
were easily separable by column chromatography.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Scheme 1a
aReagents and conditions: (a) RNH2, DIPEA, DMSO, 120ꢀ130 °C, 1ꢀ5 h
SAR for the initial pyridazinone compounds are shown in Table 1. The PCAF and BRD4 BD1
binding data was generated using TRꢀFRET assays based on labeled small molecule
bromodomain binders. Selectivity between PCAF and BRD4 BD1 is represented by a ꢁ, the
difference between PCAF pIC50 and BRD4 BD1 pIC50
Table 1. PCAF and BRD4 BD1 SAR for the pyridazinone seriesa
PCAF
pIC50
PCAF
LE
BRD4 BD1
pIC50
Chrom
LogDpH7.4
Compound
R
ꢁ
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 49
1
2
3
4
5
6
7
4.8
4.9
0.37
0.40
4.8
4.7
0
3.3
3.2
8
0.2
10
8
9
5.2
0.40
4.6
0.6
3.8
11
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
5.2
5.1
0.42
0.41
4.7
0.5
3.4
1.2
12
13
<4.3
>0.8
Ph
Me
iꢀPr
5.0
4.7
4.5
0.43
0.59
0.47
4.5
3.9
0.5
0.8
3.1
0.4
2.2
14
15
16
<4.3
>0.2
aLE = (1.37 × pIC50) / heavy atom count. ꢁ = PCAF pIC50 – BRD4 BD1 pIC50. For statistics,
see Table S1, Supporting information.
Removal of the fluoro group from 8 to give compound 10 demonstrated that the halogen had
little effect on potency at either PCAF or BRD4 BD1. Substitution of the phenyl ring with a
variety of groups revealed that the 2ꢀposition, in particular, afforded an increase in PCAF
potency and selectivity over BRD4 BD1, for example with a 2ꢀmethyl group (compound 11).
The phenyl ring could be replaced with the bioisosteric thiophene to give 12, which maintained
PCAF potency. Introduction of the more polar thiazole ring to give 13 reduced BRD4 BD1
potency while maintaining activity at PCAF. This gave us encouragement to believe that
selectivity over the BET family would potentially be achievable with the pyridazinone series.
Removal of the methylene from 10 to give phenyl 14 maintained PCAF potency with a
concomitant increase of ligand efficiency. Further truncation to methyl substituted fragment 15
retained the potency of compound 8 at PCAF, together with a substantial boost in ligand
efficiency (0.37 to 0.59). The added benefit of this change was the encouraging improvement in
ACS Paragon Plus Environment

Page 11 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
selectivity over BRD4 BD1. Growing to iꢀPr 16 was tolerated at the expense of ligand
efficiency, with activity against BRD4 BD1 not measurable at the concentrations tested.
Initial attempts to obtain crystal structures of pyridazinones such as thiophene 12 in complex
with PCAF by soaking preformed apo crystals of human PCAF bromodomain were
unsuccessful. Analysis of the crystal packing around the acetylꢀlysine binding site of the
bromodomain suggested it may be due to the proximity of adjacent protein molecules within the
crystal lattice. However, structures of PCAF from soaked crystals in complex with smaller
compounds such as fragment 15 were successful, presumably as the absence of larger groups
prevented disruption of protein contacts required to maintain the crystal integrity. A structure
was also obtained of fragment 15 in complex with BRD4 BD1 for comparison (Figure 4).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 49
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 4. Xꢀray crystal structures of (a) PCAF bromodomain (cyan) in complex with 15 (cyan);
(b) BRD4 BD1 bromodomain (grey) in complex with 15 (grey); (c) BRD4 BD1 bromodomain
(grey) in complex with 12 (grey); (d) overlay of BRD4 BD1 bromodomain (grey) in complex
with 12 (grey) and PCAF bromodomain (cyan). Hydrogenꢀbonds are shown as red dashed lines
and selected water molecules as red spheres.
The overall structure of 15 bound to PCAF showed a similar protein conformation to the
published apo structure,66 with a slight rotation of the gatekeeper Tyr809 that allows a faceꢀtoꢀ
face aromatic interaction with the pyridazinone core. The ZA loop does show slight movements
between bound and unbound structures: the sidechains are observed in the same conformation,
except Val752 which is rotated ~120°, although this rotated conformation has been seen in other
structures with fragments bound.60 A number of hydrogenꢀbonds are observed between fragment
15 and the PCAF bromodomain including the expected hydrogenꢀbond via the carbonyl group to
the sideꢀchain NH2 group of Asn803 (Figure 4a). Additionally, a shorter hydrogenꢀbond was
also made to a conserved water molecule bound by Tyr760. A further hydrogenꢀbond between
the NH group of 15 and the backbone carbonyl group of Pro747 was also observed. Finally, the
ACS Paragon Plus Environment

Page 13 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
pyridazinone ring N made a long throughꢀwater hydrogenꢀbond to gatekeeper Tyr809 and the
backbone carbonyl of Glu756. Interestingly, the chloro group acted as the methyl mimetic,
sitting in the small pocket surround by four crystallographically conserved waters. The
pyridazinone NMe group sits adjacent to Tyr802 with excellent shape complementarity to the
binding pocket.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
In comparison, the structure of 15 in complex with BRD4 BD1 shows the same hydrogenꢀ
bonding interactions as observed in PCAF (Figure 4b). The canonical hydrogenꢀbond between
the carbonyl group to Asn140 is observed, together with a throughꢀwater hydrogenꢀbond to
Tyr97. A hydrogenꢀbond to the backbone carbonyl group of Pro82 by the pendant NH group is
also seen, together with the chloro group sitting in the methyl pocket. Presumably the difference
in potency of fragment 15 between the two proteins is related to the positive aromatic
interactions with Tyr809 in PCAF that are not possible in BRD4 BD1 due to the Ile146
gatekeeper residue.
To further understand the binding mode of these compounds, thiophene 12 was selected for
crystallography with BRD4 BD1. As discussed above, crystallography in PCAF was
unsuccessful with this compound. Intriguingly, a flipped binding mode in BRD4 BD1 was
observed with thiophene 12 when compared to fragment 15 (compare Figure 4b and 4c).
Although the carbonyl group made the expected interactions with Asn140 and Tyr97 via a water
molecule, the chloro group was no longer sitting in the methyl pocket. The pyridazine core had
flipped 180o and the pendant methyl group was now sitting in the methyl pocket. As a result, the
NH group no longer made a hydrogenꢀbonding interaction with Pro82 and appeared to be bound
to bulk solvent. The thiophene ring was now sitting on the lipophilic WPF shelf, presumably
driving the change in binding mode. This binding mode observed in BRD4 BD1 would not be
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 49
1
2
3
4
5
6
7
8
9
possible in PCAF due to the difference in gatekeeper residue; the bulky Tyr809 in PCAF would
block the thiophene ring from occupying the same vector (Figure 4d). This structure also
explained the increase in selectivity over BRD4 BD1 when moving from a thiophene ring to a
thiazole ring (compare compound 12 and compound 13, Table 1). Presumably, the more polar
thiazole (ChromLogDpH7.4 1.2) is less well tolerated on the lipophilic WPF shelf than the
thiophene (ChromLogDpH7.4 3.4), which drives a reduction in BET activity.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Optimization of the aniline group. With an understanding of how to drive selectivity over
the BET family with the use of polar aryls rings, a number of ‘benzyl’ analogues were accessed
to try to improve PCAF activity. Unfortunately, no increases in PCAF potency were observed,
despite repeated efforts, which led us to move away from the ‘benzyl’ substituted pyridazinone
subꢀseries. Optimization was far more successfully conducted using aniline 14 as the lead.
Although lacking a crystal structure of 14 bound to PCAF, we hypothesized that the phenyl ring
may provide suitable vectors to interact with Glu756 via a basic functionality. As discussed
above, the analogous residue in 3D space in BRD4 BD1 (and the rest of the BET bromodomain
family) is a leucine.
Although the SNAr synthetic route described above was suitable for chemistry with simple
anilines (Scheme 1), this approach was less successful with more elaborated nucleophiles. As
such, an alternative route was developed using a chemoselective BuchwaldꢀHartwig coupling
with known 4ꢀchloroꢀ5ꢀiodo pyridazinone 17 to facilitate access to the desired targets (Scheme
2).70
Scheme 2a
ACS Paragon Plus Environment

Page 15 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
aReagents and conditions: (a) RNH2, (±)ꢀBINAP, Pd(OAc)2, Cs2CO3, PhMe, 80 °C; (b) HCl,
IPA, rt, 2 h, 73%.
Table 2. SAR for aniline substituted pyridazinonesa
PCAF
pIC50
5.0
PCAF
LE
0.43
BRD4 BD1 BRD9
Chrom
LogDpH7.4
3.1
Compound
R
H
ꢁ
pIC50
pIC50
4.5
5.6
0.5
14
18
19
20
3ꢀMe
5.2
6.0
6.5
0.42
0.46
0.42
4.8
5.9
n.d.
5.7
0.4
3.8
0.3
1.7
2ꢀCH2NH2
<4.3
>1.7
see above
<4.3
>2.2
aLE = (1.37 × pIC50) / heavy atom count. ꢁ = PCAF pIC50 – BRD4 BD1 pIC50. n.d.=not
determined. For statistics, see Table S1, Supporting information.
Although substitution of the phenyl ring with an alkyl group had little effect on either PCAF
potency or selectivity over BRD4 BD1 (Compound 18, Table 2), we were hugely encouraged
with compound 19 containing a pendant amine group. A substantial jump in PCAF potency was
observed, together with an increase in ligand efficiency upon addition of this basic moiety
(compare 14 and 19). Confirming our design hypothesis that a basic group may drive selectivity
over the BET family, a dramatic jump in the ꢁ value to >50 fold selectivity was seen. Due to the
low ChromLogDpH7.4 of amine 19, more lipophilic analogues were investigated to ensure cell
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 49
1
2
3
4
permeability.
The potency and selectivity was driven further with THIQ (tetraꢀ
5
6
7
8
hydroisoquinoline) 20 which displayed ~300 nM potency toward PCAF with over 100 fold
selectivity over BRD4 BD1. With a ChromLogDpH7.4 1.7, ≥660 µM CLND aqueous solubility
and 377 nm/s artificial membrane permeability (AMP), we felt that THIQ 20 may be suitable, or
very close to being suitable as a chemical probe for the PCAF bromodomain. However, as we
profiled this compound more broadly, substantial activity against BRD9 (pIC50 5.7, IC50 2 µM)
was uncovered. Although a crystal structure of an elaborated pyridazinone in complex with
PCAF still eluded us at this stage, the more robust BRD9 crystallographic system was used to
understand the interactions driving potency against this undesired bromodomain offꢀtarget
(Figure 5).
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 5. Xꢀray crystal structures of BRD9 bromodomain (yellow) in complex with 20 (yellow).
Hydrogenꢀbonds are shown as red dashed lines and selected water molecules as red spheres.
BRD9, in contrast to the BET family, but in a similar manner to PCAF contains a Tyr
gatekeeper residue. The shape created by this residue drives a similar deep, narrow binding
ACS Paragon Plus Environment

Page 17 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
pocket to PCAF and, as such, the pyridazone core of THIQ 20 makes faceꢀtoꢀface aromatic
interactions with Tyr106 in BRD9. The canonical hydrogenꢀbonding interactions of KAc and its
mimetics are made by the pyridazinone carbonyl directly to Asn100 and via water to Tyr57. As
in the structures of fragment 15 in complex with PCAF and BRD4 BD1 (see Figure 4), the NH
group makes a hydrogenꢀbonding interaction with a backbone carbonyl, in this case, Phe44. The
basic amine makes a throughꢀwater hydrogenꢀbond to the Gly43 backbone carbonyl.
Additionally, the pyridazinone ring N interacts with Ile53 via a long throughꢀwater hydrogenꢀ
bond. As discussed above, the THIQ was designed to target a specific interaction with PCAF
Glu756 for potency and selectivity over the BET family, with the data suggesting that this was
possible (Table 2). In BRD9, the comparable residue in 3D space is the neutral and hydrophobic
Ile53 and as expected, no specific interaction is made.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Optimization of the basic group. Attempts to overcome activity at BRD9 and engender
selectivity via optimization of the aniline or THIQ groups were unfortunately unsuccessful. As
such, we turned our attention back to the initial SAR compounds (Table 1). Screening several of
these compounds against BRD9 revealed an interesting trend (Table 3).
Table 3. SAR at BRD9 for select pyridazinonesa
PCAF
pIC50
BRD9
pIC50
Compound
13
R
ꢁ
5.1
5.9
–0.8
Ph
5.0
4.5
5.6
–0.6
>0.2
14
16
iꢀPr
<4.3
aꢁ = PCAF pIC50 – BRD9 pIC50. For statistics, see Table S1, Supporting information.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 49
1
2
3
4
5
6
7
8
9
Thiazole 13, despite promising selectivity against BRD4 BD1 (see Table 1), was actually
substantially more potent against BRD9. Aniline 14 also showed increased potency against
BRD9 when compared to PCAF, however, iꢀPr substituted compound 16 showed a bias over
BRD9. In comparison to the WPF motif (Trp746, Pro747, Phe748) in PCAF, BRD9 contains a
HGFF motif (His42, Gly43, Phe44, Phe45). We hypothesize that the bulky branched sp3 moiety
linked to the NH group is not tolerated in BRD9 due to Phe44, which sits almost perpendicular to
Trp746 in PCAF and creates a more restricted binding site. In comparison, thiazole 13 which
contains a more flexible methylene substituent and aniline 14, with a planer aryl ring are better
accommodated within the narrow BRD9 binding site.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Encouraged by the hint of selectivity over BRD9 observed with 16, we choose this motif to
continue our optimization and accessed a selection of cyclic amines targeting the PCAF Glu756
residue (Table 4). As before, straightforward SNAr chemistry with the appropriate amines was
suitable to access the desired pyridazinone targets (Scheme 1).
Table 4. Amine substituted pyridazinonesa
PCAF
pIC50
4.5
PCAF
LE
0.47
BRD4 BD1
pIC50
BRD9
pIC50
<4.3
Chrom
LogDpH7.4
Compound
16
R
ꢁ
iꢀPr
<4.3
>0.2
2.2
4.8
4.8
0.41
0.39
<4.3
<4.3
n.d.
>0.5
>0.5
–0.1
–0.3
21
22
<4.3
5.9
0.48
<4.3
<4.3
>1.6
0.6
23
ACS Paragon Plus Environment

Page 19 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
6.3
5.2
0.51
0.42
<4.3b
<4.3
<4.3
<4.3
>2
0.6
0.6
(R)ꢀ23
(S)ꢀ23
>0.9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
aLE (1.37 × pIC50) / heavy atom count. ꢁ = PCAF pIC50 – BRD4 BD1 pIC50. n.d.=not
determined. For statistics, see Table S1, Supporting information. Data is 7/8 test occasions,
pIC50 4.8 on 1/8 test occasions.
b
The position of the basic amine was critical to PCAF activity with pyrrolidine 21 and 4ꢀ
substituted piperidine 22 both showing little difference in profile when compared to iꢀPr 16
(Table 3). In contrast, 3ꢀsubstituted racemic piperidine 23 demonstrated a substantial increase in
PCAF potency while maintaining the high ligand efficiency seen with iꢀPr fragment 16.
Critically, the increase in PCAF potency also improved selectivity over both BRD4 BD1 and
BRD9. Independent synthesis of the two enantiomers with the requisite chiral amines (see
Supporting Information for details) demonstrated that the majority of the PCAF activity resided
within (R)ꢀ23, with (S)ꢀ23 showing 10 fold less potency. Critically, (R)ꢀ23 also maintained
selectivity over BRD9 and BRD4 BD1, with activity against these undesired targets below the
levels of quantification at the concentrations tested.
In our attempts to establish a more robust crystallographic system for PCAF, we also explored
the mouse orthologue, which differs to human in seven amino acids outside of the preserved
KAc binding. For some compounds the mouse PCAF bromodomain was more successful in
delivering coꢀcrystal structures. For example the crystal structure of an elaborated pyridazinone,
(R)ꢀ23, bound to mouse PCAF was obtained (Figure 6a).
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 49
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 6. Xꢀray crystal structures of (a) mouse PCAF bromodomain (yellow) in complex with
(R)ꢀ23 (yellow) (b) Overlay between mouse PCAF bromodomain (yellow) in complex with (R)ꢀ
23 (yellow) and human PCAF bromodomain (cyan) in complex with 15 (cyan). (c) PCAF
bromodomain (yellow) in complex with (R)ꢀ23 (yellow) highlighting Trp745. Hydrogenꢀbonds
are shown as red dashed lines and selected water molecules as red spheres.
As observed for fragment 15 in complex with human PCAF (see Figure 4a), the pyridazinone
core of (R)ꢀ23 sits in the narrow KAc PCAF pocket, forming faceꢀtoꢀface aromatic interactions
with Tyr808.71 As expected, multiple hydrogenꢀbonds are also observed with interactions to
Asn802, Pro746, via water to Tyr759 and via water to the backbone carbonyl of Glu755.
Validating our design hypothesis, the basic piperidine makes a salt bridge to the acidic side chain
of Glu755, which presumably contributes to the enhanced potency toward PCAF. Notably there
is a subtle movement of Glu755 towards the bound ligand to optimize this favorable salt bridge.
(Figure 6b).
Looking to further enhance the PCAF potency, while maintaining the selectivity profile seen
with (R)ꢀ23, we targeted substitutions about the piperidine ring. Due to the polarity of (R)ꢀ23
(ChromLogDpH7.4 0.6), growing the molecule with more lipophilic groups was considered
advantageous for cell permeability. Analysis of the PCAF crystal structure suggested that the 5ꢀ
ACS Paragon Plus Environment

Page 21 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
position of the piperidine ring was a suitable vector to access an edgeꢀtoꢀface aromatic
interaction with Trp745, which forms part of the PCAF WPF motif (Figure 6c).
Both the transꢀ and the cisꢀdiastereomeric 3,5ꢀdisubstitued piperidines were accessed via
alternative routes (Schemes 3 and 4 respectively). The synthesis of the transꢀdiastereomer began
with SuzukiꢀMiyaura cross coupling with commercially available pyridine 24, followed by
trifluoroacetate protection of the pendant aniline to give 25. Pyridine hydrogenation under
forcing conditions was followed by chemoselective EschweilerꢀClarke methylation to give 26,
isolated as a single diastereomer. Amide deprotection, followed by thermal SNAr reaction with 9
gave target pyridazinone 27 which was determined to be the transꢀdiastereomer by NMR
analysis.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Scheme 3a
aReagents and conditions: (a) PhB(OH)2, Pd(PPh3)4, K2CO3, PhMe, EtOH, 100 °C, 1 h, 99%;
(b) (CF3CO)2O, CH2Cl2, rt, 15 min, 57%; (c) 10% Pd/C. H2, AcOH, 90 °C, 70 bar, 28%; (d)
formaldehyde, HCO2H, 2ꢀMeTHF, H2O, 80 °C, 45 min, 86%; (e) K2CO3, MeOH, H2O, rt, 72 h,
then 60 °C, 2 h, 99%; (f) 4,5ꢀdichloroꢀ2ꢀmethylpyridazinone (9), DIPEA, DMSO, 120 °C, 1 h,
27%.
Synthesis of the cisꢀdiastereomer began with Curtius rearrangement of readily available
nicotinic acid 28 and was followed by a standard SuzukiꢀMiyaura reaction to give phenyl
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 49
1
2
3
4
5
6
7
8
9
substituted pyridine 29 (Scheme 4). Methylation with MeI, was followed by a two stage
reduction under forcing conditions to give 3,5ꢀdisubtituted piperidine 30 which was isolated as a
single diastereomer. Subsequent Bocꢀdeprotection was followed by SNAr with the appropriate
dihaloꢀsubstituted pyridazinone. Investigation of 31 by NMR analysis and comparison to the
spectra and LCMS retention time of transꢀdiastereomer 27, revealed only the presence of the cisꢀ
diastereomer. Subsequent separation of the racemates by preparative chiral HPLC gave the
desired single enantiomers (R,R)ꢀ31, (S,S)ꢀ31, (R,R)ꢀ32 and (S,S)ꢀ32.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Scheme 4a
aReagents and conditions: (a) DPPA, tꢀBuOH, NEt3, PhMe, 100 °C, 2 h, 88%; (b) PhB(OH)2,
Pd(PPh3)4, K2CO3, PhMe, EtOH, 100 °C, 1 h, 82%; (c) MeI, DMF, rt, 16 h, 93%; (d) NaBH4,
EtOH, 0 °C to rt, 72 h; (e) 10% Pd/C, H2, MeOH, 60 °C, 50 bar, 24 h, 25% over two steps; (f)
o
TFA, CH2Cl2, rt, 16 h; (g) dihalopyridazinone, DIPEA, DMSO, 120ꢀ130 C, 1ꢀ10 h; (h) Chiral
HPLC purification.
Table 5. 3,5ꢀDisubstituted piperidinesa
ACS Paragon Plus Environment

Page 23 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
O
R1
N
N
R2
N
H
7
8
9
PCAF
pIC50
PCAF
LE
BRD4 BD1 BRD9
Chrom
LogDpH7.4
Compound
R1
R2
ꢁ
pIC50
pIC50
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Cl
6.3
0.51
<4.3b
<4.3
>2.0
0.6
(R)ꢀ23
Cl
Cl
Cl
Br
Br
4.4
0.26
<4.3
<4.3
<4.3
<4.3c
<4.3
n.d.
4.5
>0.1
>2.8
>0.3
>3.1
>0.6
4.8
27
7.1
4.6
7.4
4.9
0.43
0.27
0.44
0.29
3.6
3.6
3.8
3.8
(R,R)ꢀ31
(S,S)ꢀ31
(R,R)ꢀ32
(S,S)ꢀ32
<4.3
5.1
4.5d
aLE (1.37 × pIC50) / heavy atom count. ꢁ = PCAF pIC50 – BRD4 BD1 pIC50. n.d.=not
determined. For statistics, see Table S1, Supporting information. Data is 7/8 test occasions,
pIC50 4.8 1/8 test occasions. Data is 6/8 test occasions, pIC50 4.4 on 2/8 test occasions. Data is
3/4 test occasions, pIC50 <4.3 on 1/4 test occasions.
b
c
d
Addition of a phenyl ring to the 5ꢀposition of the piperidine ring of (R)ꢀ23 to give racemic 27,
with a 3,5ꢀtrans conformation across the piperidine ring, led to a substantial drop in PCAF
potency (Table 5). However, cisꢀdiastereomer (R,R)ꢀ31 had increased PCAF potency compared
to (R)ꢀ23, albeit at the expense of some ligand efficiency. As anticipated, selectivity over BRD4
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 49
1
2
3
4
5
6
7
8
9
BD1 and BRD9 was observed, with (R,R)ꢀ31 representing the first compound accessed that
achieved the PCAF chemical probe potency criteria set out at the start of our efforts. The
majority of the activity was contained in the presumed (R,R)ꢀenantiomer, with the absolute
stereochemistry assigned by analogy to compound (R)ꢀ23.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Properties of Optimized Compounds. Throughout the optimization process, the chloroꢀ
pyridazinone had been used, with crystallography of compounds bound to PCAF showing that
the halide acted as the methyl mimetic in the KAc binding pocket (see Figures 4 and 6).
Intrigued as to whether other halogens would also be tolerated, bromoꢀanalogues (R,R)ꢀ32 and
(S,S)ꢀ32 were accessed. Comparison of the single point halide change revealed that the bromoꢀ
analogues were consistently more potent at PCAF than the chloroꢀanalogues (compare (R,R)ꢀ31
with (R,R)ꢀ32 and (S,S)ꢀ31 with (S,S)ꢀ32, Table 4). Pleasingly, (R,R)ꢀ32 demonstrated 40 nM
potency toward PCAF with excellent ligand efficiency and >1000 fold selectivity over BRD4
BD1, albeit with increased activity at BRD9 compared to (R,R)ꢀ31. The absolute (R,R)
stereochemistry for (R,R)ꢀ32 was determined by vibrational circular dichroism and then
confirmed by crystallography (vide infra). As for chloroꢀanalogue (R,R)ꢀ31, the majority of the
PCAF potency was retained in a single enantiomer, with the (S,S)ꢀenantiomer (S,S)ꢀ32 showing
much reduced PCAF activity and potential for use as an enantiomeric negative control.
As highlighted throughout, crystallography of compounds bound to human PCAF had been
challenging to achieve. Due to the high bromodomain homology with PCAF (Figure 2c), a
GCN5 crystallographic system was also generated. Crystallography of (R,R)ꢀ32 bound to human
GCN5 confirmed not only the binding mode, but also the (R,R)ꢀabsolute stereochemistry across
the piperidine ring (Figure 7).
ACS Paragon Plus Environment

Page 25 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 7. Xꢀray crystal structure of GCN5 bromodomain (green) in complex with (R,R)ꢀ32
(green). Hydrogenꢀbonds are shown as red dashed lines and selected water molecules as red
spheres.
As expected based on the previous Xꢀray crystal structures discussed above, (R,R)ꢀ32 adopts a
similar conformation to piperidine (R)ꢀ23 in PCAF, making analogous interactions to Glu761,
Asn808, Pro752 and via water to Tyr765. When compared to the apo GCN5 structure,66 Glu761
moves ~2.2Å in order to form a salt bridge with the basic amine. The lipophilic bromoꢀgroup is
well tolerated as the methyl mimetic and the pendent phenyl ring of (R,R)ꢀ32 appears to form an
edgeꢀtoꢀface interaction with Trp751 which may explain the increased PCAF potency of both
(R,R)ꢀ31 and (R,R)ꢀ32 observed relative to (R)ꢀ23.
The selectivity of (R,R)ꢀ32 against the wider bromodomain family was assessed in the
BROMOscan72 panel (Figure 8a).
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 49
1
2
3
4
5
a.
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
b.
Figure 8. (a) BROMOscan data for (R,R)ꢀ32, together with fold selectivity relative to
PCAF/GCN5 (generated at DiscoveRx Corp.). (b) Full length PCAF cellular NanoBRET doseꢀ
response curve of (R,R)ꢀ32 (squares, pIC50 7.2) and (S,S)ꢀ32 (triangles, pIC50 5.0). Error bars=SD.
As expected due to high bromodomain homology, compound (R,R)ꢀ32 showed equipotent
activity against PCAF and GCN5 with Ki 1.4 nM for both bromodomains (for full data, see
Supporting Information). Confirming the BRD4 BD1 TRꢀFRET data, exceptional levels of
selectivity over the BET family was observed with ≥18000 fold selectivity. ≥70 fold selectivity
was observed over every other bromodomain screened, with the closest off target activity at
BRPF3 (Ki 100 nM). The difference in observed PCAF potency between the TRꢀFRET and
BROMOscan assays may be due to the use of a different detection system, alternate protein
constructs and/or the sample preparation methods.
ACS Paragon Plus Environment

Page 27 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Additional cross screening against an internal GSK panel of 53 biochemical and phenotypic
assays considered as potential liabilities did not reveal any off target binding < 3 µM.
Critically for a bromodomain chemical probe, cellular target engagement of PCAF and
disruption of chromatin binding was demonstrated through a NanoBRET73 assay measuring
displacement of NanoLucꢀtagged full length PCAF from Haloꢀtagged histone H3.3 (Promega)
(Figure 8b). As expected due to the encouraging measured artificial membrane permeability
(500 nm/s), treatment of HEK293 cells with (R,R)ꢀ32 displaced full length PCAF from histone
H3.3 with little dropꢀoff from the biochemical assay and a pIC50 7.2 (IC50 60 nM). In
comparison and consistent with the biochemical data, enantiomeric negative control (S,S)ꢀ32 was
far less effective at displacing PCAF with a pIC50 5 (IC50 10 µM).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Based on the high PCAF/GCN5 affinity, excellent selectivity against both the wider
bromodomain family and a broader selection of targets, high solubility, high permeability and
proof of cellular target engagement, (R,R)ꢀ32 (GSK4027) was chosen as a PCAF/GCN5 bromodomain
chemical probe, together with (S,S)ꢀ32 (GSK4028) as the enantiomeric negative control.74 A summary of
both compounds is given in Table 6.
Table 6. Summary of properties for (R,R)ꢀ32 and (S,S)ꢀ32a
(R,R)ꢀ32
(S,S)ꢀ32
PCAF TRꢀFRET pIC50
PCAF BROMOscan pKi
GCN5 BROMOscan pKi
PCAF NanoBRET pIC50
BRD4 BD1 TRꢀFRET pIC50
BROMOscan selectivity
7.4
4.9
8.9
8.9
n.d.
n.d.
5.0
7.2
<4.3b
<4.3
n.d.
BETs ≥×18000
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 49
1
2
3
4
5
6
7
8
All others ≥×70
ChromLogDpH7.4/LogP
CLND solubility (µM)
AMP (nm/s, pH 7.4)
3.8/4.1
395
3.8/4.1
363
500
473
b
aFor statistics, see Table S1, Supporting information; n.d.= not determined. Data is 6/8 test
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
occasions, pIC50 4.4 on 2/8 test occasions.
CONCLUSION
In conclusion, we report the development of (R,R)ꢀ32 as a chemical probe for the PCAF/GCN5
bromodomain, together with the enantiomeric negative control (S,S)ꢀ32
.
The probe was
optimized from a nonꢀselective, weakly active pyridazinone hit through iterative medicinal
chemistry design hypotheses and guided by crystallography where available. Critical to the
success of this endeavor was targeting an interaction with Glu756 in PCAF which required
specific placement of a basic centre. With high potency against the PCAF/GCN5 bromodomain,
exceptional selectivity over the BET bromodomain family and proof of cellular target
engagement, we believe (R,R)ꢀ32 and (S,S)ꢀ32 represent an important novel addition to the
chemical probe toolset and hope these compounds will aid the scientific community in
elucidating the biological phenotype of PCAF/GCN5 bromodomain inhibition.
EXPERIMENTAL SECTION
Physicochemical Properties. Artificial membrane permeability, chromatographicLogD at pH
7.4, and CLND aqueous solubility were measured using published protocols.75
Chemistry. See Supporting Information for general experimental details.
The purity of all compounds tested was determined by LCMS and 1H NMR to be > 95%.
SELECTED EXPERIMENTAL PROCEDURES
ACS Paragon Plus Environment

Page 29 of 49
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
tertꢀButyl (5ꢀphenylpyridinꢀ3ꢀyl)carbamate (29). Step 1: Diphenyl phosphorylazide (40.0
mL, 186 mmol) was added dropwise over 15 min to a stirred solution triethylamine (51.5 mL,
371 mmol) and 5ꢀbromonicotinic acid (25 g, 124 mmol) in toluene (250 mL) and tertꢀbutanol (50
mL) at rt under N2. The resultant solution was stirred at rt for 20 min and then heated to 100 °C
for 2 h. Upon cooling to rt, sat. aq. NaHCO3 (200 mL) and EtOAc (200 mL) were added. The
separated aqueous phase was extracted with EtOAc (2 × 100 mL). The combined organic phase
was washed with brine (100 mL), passed through a hydrophobic frit and evaporated under
reduced pressure to give an orange solid. The solid was loaded in CH2Cl2 and purified by silica
gel chromatography using a gradient of 0ꢀ15% EtOAc / cyclohexane. The appropriate fractions
were combined and evaporated under vacuum to give tertꢀbutyl (5ꢀbromopyridinꢀ3ꢀyl)carbamate
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
as a white solid (29.81 g, 88%). H NMR (400 MHz, CDCl3) δ 8.34 (s, 1H), 8.30 (app. s, 2H),
6.68 (br.s, 1H), 1.53 (s, 9H); LCMS (formic) (M+H)+ = 273.0, 275.0, Rt = 1.04 min (100%).
Step 2: Pd(PPh3)4 (2.52 g, 2.183 mmol) was added in a single portion to a stirred suspension of
tertꢀbutyl (5ꢀbromopyridinꢀ3ꢀyl)carbamate (29.81 g, 109 mmol), phenylboronic acid (15.97 g,
131 mmol) and potassium carbonate (45.3 g, 327 mmol) in toluene (250 mL) and EtOH (250
mL) at rt under N2. The resultant suspension was evacuated under vacuum and then backꢀfilled
with N2 three times. The suspension was then heated to 100 °C for 1 h. Upon cooling to rt,
EtOAc (300 mL) and H2O (200 mL) were added. The separated aqueous phase was extracted
with EtOAc (2 × 100 mL), the combined organic phase was passed through a hydrophobic frit
and evaporated under reduced pressure to give an orange solid. The solid was mostly dissolved
in CH2Cl2 (100 mL) and loaded onto a silica plug (15 cm x 30 cm). The plug was eluted with
EtOAc (1000 mL). The eluent was combined and evaporated under reduced pressure to give a
yellow solid. The solid was suspended in Et2O (100 mL) and filtered. The collected solid was
ACS Paragon Plus Environment