Design and chemoproteomic functional characterization of a chemical probe targeted to bromodomains of BET family proteins

Article abstract of DOI:10.1039/c4md00259h

Bromodomain-containing proteins form the signal-reading element of a principal system for the control of gene expression in eukaryotes. Their potential as targets for selective drug action is increasingly being assessed and exploited. Deep characterizatio

Full text of DOI:10.1039/c4md00259h

MedChemComm
View Article Online
View Journal | View Issue
CONCISE ARTICLE
Design and chemoproteomic functional
characterization of a chemical probe targeted to
bromodomains of BET family proteins†
Cite this: Med. Chem. Commun., 2014,
5, 1871
Jiang Wu,‡§a Julia Shin,§b Cara M. M. Williams,b Kieran F. Geoghegan,a
Stephen W. Wright,c David C. Limburg,c Parag Sahasrabudhe,a Paul D. Bonin,e
d
Bruce A. Lefker and Simeon J. Ramseyb
*
Bromodomain-containing proteins form the signal-reading element of a principal system for the control of
gene expression in eukaryotes. Their potential as targets for selective drug action is increasingly being
assessed and exploited. Deep characterization of the specificity, potency and other attributes of
prototypical agents is an essential element of this process. Continuing studies of
a
dihydroquinazolinone-based series (prototype: PFI-1) with specificity for members of the BET
(bromodomain and extra terminal) family led to the discovery of quinolin-2(1H)-one inhibitors with
similar potency and selectivity, but increased chemical stability. Structure-guided design then led to the
elaboration of a desthiobiotinylated analog retaining a high fraction of the potency of its parent
compound and therefore suitable for chemoproteomic affinity capture experiments. These experiments,
conducted using nuclear extracts of THP-1 cells, extended confidence in the selectivity of the series as
first proposed. An additional and subsequent evaluation of specificity performed with a panel of
recombinant bromodomains (BROMOscan™, DiscoveRx) supported the BET family specificity of the
dihydroquinazolinone and quinolin-2(1H)-one series while adding appreciation of weaker effects shown
at other bromodomains.
Received 16th June 2014
Accepted 12th September 2014
DOI: 10.1039/c4md00259h
www.rsc.org/medchemcomm
candidate. Although tool compounds interact with their inten-
ded targets at the molecular level and produce desirable phys-
Introduction
iological responses in preclinical tests, they have usually been
excluded from clinical development. Despite this, they remain
valued agents for investigative studies of the mechanisms that
they engage.2
In drug discovery research, a chemical tool is a compound that
possesses the principal dening activity of a desired drug but
lacks additional attributes that would permit its full develop-
ment. Chemical tools are used to test the validity of discovery
concepts at an early stage, helping with assessments of the
attractiveness of particular targets and modalities in addition to
uncovering potential liabilities and safety risks. Work done with
these compounds can be inuential. For example, there have
been over 300 citations of the report1 describing A-769662, a
prototypical activator of AMP-activated protein kinase, despite
this compound not possessing all of the attributes of a drug
Another example of a chemical tool is PFI-1 (1){ (Fig. 1), an
agent that binds specically to recombinant bromodomains
(BRD) of proteins belonging to the bromo and extra terminal
(BET) domain family.3,4 BET family proteins contribute to the
important function of reading epigenetic ngerprints that take
the form of acetylated lysine residues on histone scaffolding
proteins and affect chromatin structure and DNA accessibility.
Thus, epigenetic changes effect alterations in gene expression
that do not originate in changes in the DNA sequence. Once
established, these chemical changes in DNA or histones can
persist through many cell divisions, with specic patterns of
gene expression determined by the epigenetic prole of the cell.
BRD are autonomously folded protein units that recognize
and bind to 3-N-acetyl-lysyl residues of histones, forming the
“reader” counterpart to the histone acetyltransferases that
“write” signals related to the control of gene expression by
aStructural Biology and Biophysics, Worldwide Medicinal Chemistry, Pzer, Groton,
CT 06340, USA
bInammation and Remodeling Research Unit, Pzer, Cambridge, MA 02139, USA
cWorldwide Medicinal Chemistry, Pzer, Groton, CT 06340, USA
dWorldwide Medicinal Chemistry, 610 Main Street, Pzer, Cambridge, MA 02139, USA.
Fax: +1 860-715-8059; Tel: +1 617-674-7450. E-mail: bruce.a.leer@pzer.com
ePrimary Pharmacology Group, Pzer, Groton, CT 06340, USA
† Electronic supplementary information (ESI) available: Details of the chemical
synthesis, biological methods, supplementary tables and supplementary
references. See DOI: 10.1039/c4md00259h
‡ Current address: Shire, 300 Shire Way, Lexington, MA 02421, USA.
§ These authors contributed equally to this work.
{ PFI-1, also known as PF-06405761, is commercially available from Sigma Aldrich
(catalog # SML0352).
This journal is © The Royal Society of Chemistry 2014
Med. Chem. Commun., 2014, 5, 1871–1878 | 1871

View Article Online
MedChemComm
Concise Article
For reasons discussed below, PFI-1 has been replaced by 2,k
a very close analog of PFI-1 with increased chemical stability
and equivalent or superior potency. A derivative of 2 has also
been prepared which is suitable for affinity capture through a
linked desthiobiotin moiety (3). Using nuclear extracts from
cultured human cells that were shown proteomically to contain
a signicant fraction of known BRD-containing proteins, the
novel probe was used to conduct protein affinity capture
studies. Specicity was ensured by using 2 as a competitive
agent, an approach complemented by the use of SILAC (stable
isotope labeling with amino acids in cell culture) methodology
in the proteomic analyses.12
Fig. 1 Dihydroquinazolinone PFI-1 and the alternative quinolin-
2(1H)-one probe 2, which lacks sensitivity to oxidative insertion of
nucleophilic solvents. The desthiobiotinylated compound 3 was
captured on streptavidin beads in the course of the affinity capture
experiments.
Results and discussion
Subsequent to the original validation of PFI-1 as a chemical
probe with high selectivity for BET family proteins,3,4 its dihy-
droquinazolinone ring system was found to be sensitive to
oxidative insertion of nucleophilic solvents (see footnote 24 of
Fish et al. (2012)4). In light of these data, we selected for the
studies described below its aromatic quinolin-2(1H)-one
analogue13 2 (Scheme 1). 2 is a chemically stable alternative to
PFI-1 which exhibits similar potency and selectivity (Table 1)
against a panel of BRD. (In addition to the potencies measured
by FP for PFI-1, 2 and 3 against BRD4-BD-1 and shown in Table
1, values of 115 nM, 105 nM and 83 nM, respectively, were
measured by isothermal calorimetry.)
It was also important to conrm that 2 was active in a whole
cell assay. In an LPS challenge assay in human peripheral blood
mononuclear cells (PBMCs), PFI-1 inhibited the release of
interleukin-6 with an IC50 of 2830 nM. The quinolin-2(1H)-one
analog 2 was a more potent inhibitor in the same assay, with an
IC50 of 610 nM (Fig. 2). Its combination of superior stability and
modestly improved bioactivity compared to PFI-1 demonstrated
its suitability for the present study. This investigation
comprised a series of experiments in which chemoproteomic
methodology was applied to characterizing the specicity of our
quinolin-2(1H)-one probe against a wide sample of BRD-con-
taining proteins from nuclear extracts of THP-1 cells.
N-acetylating histone tails at lysine side chains. For example,
the BET family member BRD4 can recognize and bind to acet-
ylated lysine residues on histones 3 and 4, allowing positive
transcription elongation factor b (P-TEFb) to dock, phosphory-
late RNA polymerase II, and facilitate efficient transcription of
mRNA.5 This has made BET family members, and in particular
BRD4, attractive targets for therapeutic intervention aimed at
directly modulating gene expression, a potentially powerful but
presently underexploited mode of pharmaceutical action. Their
potential signicance as antitumor targets is at the leading edge
of this effort.4,6–9 Several medicinal chemistry efforts have now
identied small molecule BET inhibitors that belong to
different chemical series that represent alternative scaffolds for
future drug discovery.10
Efforts to develop selective inhibitors of gene expression via
targeted inhibition of BET family proteins (as opposed to broad
antagonism of all BRD-containing proteins) are an endeavor
that requires multiple well-characterized chemical tools. For
example, access to a panel of chemical tools with diverse
selectivity proles within the BET subfamily of bromodomains
would permit investigation of whether or not a specic BET
family member (e.g. BRD4) can drive the transcription of a
unique set of genes or, alternatively, that BET proteins exhibit
overlapping and/or redundant roles. Likewise, in the context of
a specic disease, the expression of BET proteins may be
skewed to particular family members that signicantly inu-
ence gene transcription. The ability to interrogate the respective
roles of specic BET proteins in diseased cells with a set of well
characterized chemical tools would yield important insights
into pathological gene transcription. In the case of the proto-
type compound PFI-1, its potential has to some extent been
examined through cell biological and structural biological
approaches.3,4 Here we evaluate the specicity of a close analog
using the central chemoproteomic strategy of protein affinity
capture.11 We also provide additional biochemical character-
ization of the specicity of this compound using a proprietary
panel of recombinant bromodomain targets (DiscoveRx
BROMOscan™).
The planned chemoproteomic studies required a bio-
tinylated (or desthiobiotinylated) affinity capture probe that
retained all or most of the affinity of 2 for BRD proteins,
Scheme 1
k Compound 2, also known as PF-06482483, will be made commercially available
from Sigma Aldrich.
1872 | Med. Chem. Commun., 2014, 5, 1871–1878
This journal is © The Royal Society of Chemistry 2014

View Article Online
Concise Article
MedChemComm
Table 1 Relative affinities (by fluorescence polarization assay) of PFI-1,
2, 3, and reference compounds IBET-762 and IBET-151 for BET family
bromodomains and CREBBP
Ki (nM)
Bromodomain
PFI-1
2
3
IBET-151
IBET-762
BRD4-BD1
BRD4-BD2
BRD2-BD2
BRD3-BD1
BRD3-BD2
BRDT-BD1
CREBBP
88
894
122
175
294
565
3800
27
213
102
71
207
423
43
717
321
113
146
521
17
181
217
37
118
209
692
58
35
83
112
48
528
2060
11 500
>14 500
Fig. 3 Crystal structure of compound PFI-1 bound to bromodomain 1
of BRD4 (PDB code 4e96). See also P. V. Fish et al., J. Med. Chem.,
2012, 55, 9831–9837. The structure was interpreted as suggesting that
a linker group to facilitate immobilization through desthiobiotin could
be placed at the ortho position of the aryl sulfonamide ring, from
which the methoxy group of PFI-1 points into solvent (arrow).
In similar fashion to the preparation of 3, we terminated the
poly-alkoxy linker with Cy5 uorescent dye (GE Healthcare, cat.
PA15100: see preparation of PF-411FP in ESI†). The availability
of PF-411FP then made it possible to assess the extent to which
3 retained the activity of 2. In a FP-based assay of binding to
BRD4-BD1 (Table 1), 2 and the pull-down probe 3 both exhibited
slightly higher affinity for the protein target than PFI-1, similar
potency to previously reported IBET-762 (ref. 14 and 16)
(currently in Phase 1 clinical trials), and slightly lower affinity
when compared to IBET-151.17
Fig. 2 The effects of compound 2 and PFI-1 on LPS induced inter-
leukin-6 (IL-6) release assayed in human PBMCs. IC50 values of 610 nm
(n ¼ 4) and 2830 nm (n ¼ 5) were obtained for compound 2 and PFI-1
respectively.
analogous to a previously described derivative of IBET-762.9
In vitro BET family binding selectivity proles of the
Thus, it was essential to identify a region of molecule 2 which compounds were then assessed by FP assay using six recombi-
would tolerate homologation of an appropriate linker without nant-expressed BET family bromodomains (Table 1, note:
interfering with binding to BRD proteins. Examination of the BRD2-BD1 and BRDT-BD2 proteins were not available at the
previously described4 structure of PFI-1 bound to the rst bro- time of this study). Our soluble competitor compound 2 and the
modomain of BRD4 (Fig. 3) suggested that the ortho-position of affinity probe 3 demonstrated very similar potency and selec-
the aryl sulfonamide pointed to solvent and could serve as an tivity proles against these six BET family proteins (Fig. 4), with
exit vector from the binding site that would not alter affinity to both 2 and 3 showing 8–15ꢀ selectivity for the rst bromodo-
BRD targets. As this observation should equally apply to 2, we main of BRD4 (BRD4-BD1) over BRD4-BD2 and BRDT-BD1.
devised the route shown in Scheme 2 to provide compound 3.
The majority of BET ligands disclosed in the literature are
Compound 3 was assembled in a modular fashion, in which the either non-selective for the rst bromodomain of BRD4 (BRD4-
active probe was coupled to desthiobiotin through a (3-(2-(2-(2- BD1) versus the second bromodomain of BRD4 (BRD4-BD2), or
hydroxyethoxy)ethoxy)ethoxy)propyl)amine linker. The linker, 9, exhibit modest selectivity for BRD4-BD1 over BRD4-BD2. A
which was generated from triethylene glycol and acrylonitrile, notable exception is RVX-208 (currently in Phase 2 clinical trials
followed by hydrogenation and Boc protection, afforded the for cardiovascular indications) which exhibits selectivity for
amine handle to which desthiobiotin would be coupled. BRD4-BD2.18 Although the sequence homology between the rst
Alcohol, 9, was combined with 2-(benzylthio)phenol under and second bromodomains of BRD4 is much lower than that
Mitsunobu conditions to yield the desired intermediate 10. between BRD4-BD1 and BRD2-BD1,14,19 it is difficult to ratio-
Conversion to the sulfonyl chloride (11) and coupling to the nalize the observed selectivity for these compounds in the
active probe (5) resulted in compound 12. Removal of the BOC absence of crystallographic data of their binding modes with
group, followed by standard amide coupling conditions with each of these proteins.
desthiobiotin resulted in the desired probe 3.
Also of note is the very similar BET family selectivity prole
Fluorescence polarization (FP) is a favored mode of assay for for both 2 and 3 when compared to our original probe molecule.
rating the respective affinities and selectivities of compounds Fig. 4 also highlights some additional differences in BET family
that bind to bromodomains,14,15 but it requires provision of a selectivity for the set of molecules evaluated. IBET-151 demon-
uorescent probe against which test compounds can compete. strated a very similar BET family prole when compared to 2
This journal is © The Royal Society of Chemistry 2014
Med. Chem. Commun., 2014, 5, 1871–1878 | 1873

View Article Online
MedChemComm
Concise Article
Scheme 2
BRD, and that 3 retains a substantial fraction of the affinity
shown by 2 for each protein target with a very similar selectivity
prole. In summary, the in vitro binding results appeared to
qualify 3 as suitable for use in affinity capture studies in extracts
from whole cells. Although the above results provide some
insights into selectivity proles for BET family proteins, the
main objective of this work was to understand the extent to
which our probe molecules exhibited selectivity against the
broader BRD family in an authentic biological extract.
These data, which reinforce earlier information about the
properties of 1 and 2, opened the way for preliminary tests of the
ability of 3 to capture BRD proteins from a biological milieu.
Using SDS-polyacrylamide gel electrophoresis as a readout, it
was shown that 3 immobilized on streptavidin-coated agarose
beads could capture recombinant His-tagged BD1 of BRD4 from
solution (Fig. 5). Soluble compound 2 prevented capture
(Fig. 5A), demonstrating that the capture was a specic event
based on affinity between protein and immobilized compound.
Building on this result with recombinant protein, it was also
shown by immunodetection on western blots that agarose
beads bearing 3 could capture native BRD4 protein from lysates
of either HeLa (cervical cancer immortalized cell line) or THP-1
(immortalized human monocytic cell line derived from a
patient with acute monocytic leukemia) cells (Fig. 5B).
Fig. 4 BET family bromodomain selectivity profile of PFI-1, 2, 3, IBET-
762 and IBET-151 as measured by the fluorescence polarization assay.
Values on the vertical axis represent the ratio of each compound's Kd
from a particular bromodomain to its Kd from BRD4-BD1: for example,
The above results le us ready for experiments in which 3
was used to “sh” by affinity capture for proteins in a biological
extract, using SILAC labeling to allow specic and nonspecic
the affinity of 3 for BRD4-BD2 is 17-fold weaker than its affinity for protein binding to be distinguished.12 Nuclear extracts of THP-1
BRD4-BD1.
cells made a proteomically suitable starting material for this
work (see below), and THP-1 cells also appeared biologically
pertinent in view of ndings that (i) bromodomain inhibitor
and 3, but it has signicantly less selectivity versus CREBBP. In JQ1 inhibits growth of monocytic cells from acute myelogenous
contrast, IBET-762 demonstrated similar potency against all of leukemia patients,20 and (ii) BET inhibitors may have utility in
the BET domain proteins with the exception of BRDT for which the treatment of chronic inammatory conditions in which
it has comparable selectivity (ꢁ10ꢀ) when compared to 2 and 3. monocytes and macrophages are prominent.21 Thus, THP-1
Overall, the functional consequences of these different selec- cells were grown using SILAC conditions, with the “light”
tivity proles have not yet been explored and require future culture matched by “heavy” cells grown on medium containing
investigation. These data support the view that PFI-1 and [13C6,15N2]-L-lysine and [13C6,15N4]-L-arginine. Proteomic anal-
compound 2 generally have comparable affinities for these ysis of a lysate of heavy cells showed >97% incorporation of the
1874 | Med. Chem. Commun., 2014, 5, 1871–1878
This journal is © The Royal Society of Chemistry 2014

View Article Online
Concise Article
MedChemComm
genome; some of the proteins not detected are specic to cell
types of other lineage (for example, BRDT which is testes-
specic).22 Also, because high relative abundance favors detec-
tion in proteomics, we reasoned that the list may not include
every bromodomain-containing protein in the nuclear fraction
and theoretically capable of being affinity captured. This
possibility later emerged as a fact (see below).
Details of the method for affinity capture studies are given in
ESI.† Briey, probe 3 (3 mM) was incubated with THP-1 cell
nuclear extract in the absence (SILAC light) or presence (SILAC
heavy) of 10 mM 2, which was added as a competitor of specic
binding. Aer 3 h incubation, streptavidin-coated magnetic
beads were added to capture 3 and proteins bound specically
to it. The two fractions of beads were combined aer washing,
and protein was eluted using SDS-polyacrylamide gel buffer.
Following SDS-polyacrylamide gel electrophoresis to frac-
tionate the sample, captured proteins from the +/ꢂ compound-
treated nuclear extracts were identied by LC-MS/MS. A total of
60 proteins were detected with a false-positive rate of 1%,
among which 28 were identied with two or more unique
peptides (Table 2) when the minimum Mascot23 peptide score
was set to the moderately stringent value of 30. The BET-family
members BRD3, BRD4, and BRD2 were each present among the
captured proteins, with 12, 8 and 4 unique peptides identied,
respectively. As shown by the SILAC scores of zero for the BET-
family members, their capture by 3 was completely inhibited by
the soluble compound 2, as demonstrated by the absence of the
heavy partners of the detected peptide peaks. This was not the
case for other proteins, for which SILAC ratios close to 0.5
indicated that the heavy and light nuclear extracts had not been
mixed in perfectly equal proportion, but that nonspecic
capture of proteins other than the BET family members was at a
consistent level. However, we cannot exclude that failure to
capture additional BRD-containing proteins is to some extent
due to these proteins being complexed with protein binding
partners and therefore inaccessible to the probe.
Fig. 5 Specific affinity capture of recombinant BRD4-BD1 or full-
length BRD4 by compound 3. (A) Western blot using anti-His-tag
antibody. A 17 kDa His-tagged recombinant form of BD1 of BRD4 was
captured by 3 linked to agarose beads (lane ꢂ cpd). Binding was
prevented by competition with soluble compound 2 (lane + cpd). (B)
Western blot showing affinity capture of BRD4 from HeLa or THP-1
cell lysates. Capture of BRD4 was specific, as indicated by the absence
of capture in the presence of soluble compound 2.
Subsequent to the chemoproteomic work, a broader survey
of the bromodomain specicity of PFI-1 and 3 was conducted
isotopically-substituted amino acids (data not shown), which using the DiscoveRx BROMOscan™ platform (Table S2† and
was suitable for ratiometric proteomics. Fig. 6). Consistent with earlier results, both PFI-1 and 3 showed
As bromodomain-containing proteins are believed to be strong specicity for bromodomains of proteins from the BET
largely nuclear proteins,7 nuclear extracts were prepared from family. Both exhibited >25 fold selectivity in favor of BET-family
the light and heavy THP-1 cells, and proteomic analysis was bromodomains as compared with all other bromodomains
used to check the extent to which these extracts represented the evaluated (Fig. 6), further supporting our chemoproteomics
nuclear fraction. The ve highest-scoring hits in the sample work demonstrating that the biotinylated probe had a similar
(data not shown) were heterogeneous nuclear ribonucleopro- selectivity prole to compound PFI-1. It was a point of interest
teins A2/B1, DNA-dependent protein kinase catalytic subunit, that 3 exhibited higher potency than PFI-1 against BRD4-BD1 in
heterogeneous nuclear ribonucleoprotein K, poly [ADP-ribose] the BROMOscan™ study (7 nM vs. 62 nM), in contrast to the
polymerase 1, and nucleophosmin. Each of these proteins is similar potency detected in both FP (43 nM vs. 88 nM) and ITC
located entirely or partly in the nucleus, according to the Uni- (83 nM vs. 115 nM). This difference is most likely attributable to
ProtKB database, and none was among the top y hits in the differences in assay platforms (D. Treiber, DiscoveRx, personal
cytoplasmic fraction prepared from the same lysate. This indi- communication). Overall, the data strongly supported the
cated that valid nuclear extracts had been prepared. The list of conclusion that among the set of bromodomains tested, 3
detected proteins (550 in all) was then scanned for BRD-con- exhibits a minimum of about 25-fold specicity for those
taining proteins, and 15 such proteins were detected (Table belonging to proteins of the BET family, and oen a much
S1†). We considered this an acceptable fraction of the 46 greater selectivity.
known7 BRD-containing proteins encoded by the human
This journal is © The Royal Society of Chemistry 2014
Med. Chem. Commun., 2014, 5, 1871–1878 | 1875

View Article Online
MedChemComm
Concise Article
Table 2 SILAC-based affinity capture of proteins from nuclear extract of THP-1 cellsa
Accession
Description
# Unique peptides
Heavy/light
Q15059
P51659
O60885
P23246
D6RAN4
P46783
Q08211
P68104
Bromodomain-containing protein 3
Peroxisomal multifunctional enzyme type 2
Bromodomain-containing protein 4
Splicing factor, proline- and glutamine-rich
60S ribosomal protein L9 (fragment)
40S ribosomal protein S10
ATP-dependent RNA helicase A
Elongation factor 1-alpha 1
40S ribosomal protein S7
Probable ATP-dependent RNA helicase DDX5
Bromodomain-containing protein 2
Non-POU domain-containing octamer-binding protein
Heterogeneous nuclear ribonucleoprotein U
Heterogeneous nuclear ribonucleoprotein H
40S ribosomal protein S5 (fragment)
Isoform 4 of methyl-CpG-binding domain protein
Heterogeneous nuclear ribonucleoprotein A1
Heterogeneous nuclear ribonucleoproteins A2/B1
Actin, cytoplasmic 1
40S ribosomal protein S4, X isoform
Isoform short of RNA-binding protein FUS
Signal recognition particle 14 kDa protein
60S ribosomal protein L11
Probable ATP-dependent RNA helicase DDX17
RNA-binding protein 14
12
12
8
6
6
5
4
4
4
4
4
3
3
3
3
3
2
2
2
2
2
2
2
2
2
2
2
2
0.00
0.52
0.00
0.41
0.82
0.40
2.21
0.56
0.69
0.47
0.00
0.43
0.51
0.31
0.44
0.73
0.38
0.38
0.40
0.65
0.52
1.05
0.64
0.48
0.21
0.76
0.68
0.47
P62081
B4DLW8
P25440
Q15233-2
Q00839-2
H0YB39
M0R0F0
O95243-4
F8VTQ5
P22626-2
P60709
P62701
P35637-2
P37108
P62913-2
Q92841-1
Q96PK6
Q5T6W1
K7EQ02
Q09472
Heterogeneous nuclear ribonucleoprotein K
DAZ-associated protein 1
Histone acetyltransferase p300
a
Limited to proteins with minimum of 2 peptides detected; hits on keratins and trypsin not shown.
According to the BROMOscan™ data (Table S2†), BRDT house binding data (Table 1 and S2†). Although it is possible
(bromodomain testis-specic protein, 10–20ꢀ selectivity) and that the linker group in 3 could interfere with its binding to
BRPF1 (>100ꢀ selectivity) displayed moderate, but measurable some bromodomain family members, the lack of differentiation
interactions with compound 3, but these proteins were not between the selectivities of PFI-1 versus 3 in the BROMOscan™
detected within the soluble nuclear fraction used in these sets study (Fig. 6 and Table S2†), and the conserved acetyl lysine
of experiments. CREBBP (>300ꢀ selectivity) was found to be binding pocket of the BRD family suggests this is unlikely.
present in the soluble nuclear fraction, albeit in low abundance.
This work highlights the combined use of SILAC technology
Failure to detect it in the affinity capture experiment may with a biotinylated small molecule probe to assess the selec-
therefore conrm the weak binding affinity between this tivity prole of an unlabeled compound for a particular class of
protein and compound 3 as identied by BROMOscan™ and in- intracellular target proteins. This technique provides data
Fig. 6 TREEspot™ plots of BROMOscan™ data for PFI-1 and compound 3. images generated using TREEspot™ software tool and reprinted with
permission from KINOMEscan®, a division of DiscoveRx corporation, © DiscoveRx corporation 2010.
1876 | Med. Chem. Commun., 2014, 5, 1871–1878
This journal is © The Royal Society of Chemistry 2014

View Article Online
Concise Article
MedChemComm
based on the capture of authentic cellular proteins that HL60 nuclear extracts, they demonstrated that both IBET-151
complement results from biochemical screening platforms that and IBET-762 inhibited binding of BRD9. They speculated that
use recombinant proteins. In this instance our study was not BRD9 was not a direct target of the IBET molecules and that it
quantitative, and dose response studies with the probe and was subject to competition because it can form complexes with
soluble competitor would be required to gauge selectivity BRD4. At least in HL60 nuclear extracts this appears to be the
between the different BET family members.
case, as they demonstrated pull down with BRD4 in immuno-
Our efforts in probe design have demonstrated that adoption precipitation studies. However, it is intriguing that we were
of a polyglycol spacer-bearing desthiobiotinylated derivative, 3, unable to detect BRD9 in THP1 cells, despite this cell line's
does not sacrice affinity for any of the bromodomains which ability to produce IL-6 upon LPS stimulation at similar levels to
have been assessed to date. In addition, we have demonstrated that observed in human whole blood. This leads us to question
a somewhat different binding/selectivity prole of compounds 2 whether or not BRD9 is required for the transcription of IL-6 in
and 3 in THP1 nuclear extracts compared to that described for certain cell types.
IBET-762 and IBET-151 in HL60 nuclear extracts.9 Our uores-
cent polarization assay revealed enhanced selectivity for 2 over
CREBBP compared to that observed for IBET-151. This selec-
Conclusion
tivity prole may have signicant impact on gene transcription In summary, this work describes the design and utility of an
because (i) CREBBP is known to have intrinsic histone acetyl- affinity based probe coupled with a protein capture method to
transferase activity and (ii) it acts as a scaffold to stabilize quantify the selectivity of soluble competitor compounds for the
additional protein interactions within the transcription bromodomain containing family of proteins identied in THP-1
complex. For example, CREBBP interacts24,25 with a diverse cells. These results support the hypothesis that 3 is highly
group of transcription factors including CREB, p53, NF-kB and selective for BET family bromodomains in nuclear extracts of
AP-1, which can lead to augmentation of gene transcription. THP-1 cells and these results were conrmed when assessed in
Thus, in the seminal paper by Dawson et al.,9 wherein they the proprietary BROMOscan™ panel of bromodomains.
report on IBET-151 in vitro potency and in vivo efficacy, Further analysis using extracts from different cellular
concentrations above 1 mM would be expected to also interact compartments and different cell types would provide additional
with CREBBP and affect gene transcription. Consequently, in information on the selectivity prole of soluble inhibitors of
their studies that addressed in vivo efficacy in MLL leukemia BRD proteins. Additionally, we have demonstrated a different
models, IBET-151 was dosed at 30 mg kgꢂ1 intraperitoneally bromodomain selectivity prole for PFI-1, 2, and 3 compared to
once daily. From the PK analysis of this compound reported in other BET inhibitors which could be exploited in future studies
their paper,9 systemic exposures would be expected to exceed 1 to explore the role of specic family members in gene
mM for a signicant portion of the day, and this leads us to transcription.
question whether a proportion of the observed in vivo phar-
macology is due to inhibition of CREBBP in addition to the
expected BET family pharmacology. Of note, Delvecchio et al.26
Acknowledgements
recently published the crystal structure of the p300 catalytic The authors thank D. Owen for establishing a collaboration
core, a closely related HAT enzyme with structural and func- with DiscoveRx and providing input on the writing of this
tional similarities to CREBBP. This compact module consists of manuscript; W. Lin for preparation of the FP ligand PF-411FP;
the bromodomain together with PHD, RING and HAT domains, L. Jones, E. Hett, and B. Jacobson for their insightful input on
and tight integration of the chromatin substrate-binding the design of these experiments; G. Pallares, J. P. Hunt and D. K.
domains into the enzymatic core of the acetyltransferase Treiber of DiscoveRx Corporation for the BROMOscan™ data;
explains why substrate recognition is coupled to HAT activity and D. Hepworth for reviewing the manuscript.
and why mutation of these domains can lead to aberrant HAT
function and pathogenesis. These observations provide further
validation that binding to the bromodomain of CREBBP will
Notes and references
impact gene transcription even in the absence of BET activity
and highlight the risk that a portion of IBET-151 efficacy is
driven by off-target effects on CREBBP.
1 B. Cool, B. Zinker, W. Chiou, L. Kie, N. Cao, M. Perham,
R. Dickinson, A. Adler, G. Gagne, R. Iyengar, G. Zhao,
K. Marsh, P. Kym, P. Jung, H. S. Camp and E. Frevert, Cell
Metab., 2006, 3, 403–416.
2 M. E. Bunnage, E. L. P. Chekler and L. H. Jones, Nat. Chem.
Biol., 2013, 9, 195–199.
3 S. Picaud, C. D. Da, A. Thanasopoulou, P. Filippakopoulos,
P. V. Fish, M. Philpott, O. Fedorov, P. Brennan,
M. E. Bunnage, D. R. Owen, J. E. Bradner, P. Taniere,
B. O'Sullivan, S. Muller, J. Schwaller, T. Stankovic and
S. Knapp, Cancer Res., 2013, 73, 3336–3346.
Interestingly, we were unable to detect BRD9 in the nuclear
extracts of THP-1 cells. Although this may simply reect limi-
tations in the sensitivity of our proteomic analysis, we were also
unable to show binding of 2 to BRD9 in a DiscoveRx
BROMOscan™. In contrast, Dawson et al.9 reported that both
IBET-762 and IBET-151 could block the capture of BRD9 by
acetylated histone tail peptides in a nuclear extract derived from
HL60 cells. Similarly, in proteomic proling studies in which
they used a set of biotinylated histone peptides immobilized to
streptavidin-coated beads to capture binding proteins from
4 P. V. Fish, P. Filippakopoulos, G. Bish, P. E. Brennan,
M. E. Bunnage, A. S. Cook, O. Federov, B. S. Gerstenberger,
This journal is © The Royal Society of Chemistry 2014
Med. Chem. Commun., 2014, 5, 1871–1878 | 1877

View Article Online
MedChemComm
Concise Article
H. Jones, S. Knapp, B. Marsden, K. Nocka, D. R. Owen, 15 M. Thompson, Methods Enzymol., 2009, 466, 383–407.
M. Philpott, S. Picaud, M. J. Primiano, M. J. Ralph, 16 E. Nicodeme, K. L. Jeffrey, U. Schaefer, S. Beinke, S. Dewell,
N. Sciammetta and J. D. Trzupek, J. Med. Chem., 2012, 55,
9831–9837.
5 B. N. Devaiah and D. S. Singer, Transcription, 2013, 4, 13–17.
C.-w. Chung, R. Chandwani, I. Marazzi, P. Wilson, H. Coste,
J. White, J. Kirilovsky, C. M. Rice, J. M. Lora, R. K. Prinjha,
K. Lee and A. Tarakhovsky, Nature, 2010, 468, 1119–1123.
6 I. A. Asangani, V. L. Dommeti, X. Wang, R. Malik, M. Cieslik, 17 O. Mirguet, Y. Lamotte, F. Donche, J. Toum, F. Gellibert,
´
R. Yang, J. Escara-Wilke, K. Wilder-Romans, S. Dhanireddy,
C. Engelke, M. K. Iyer, X. Jing, Y.-M. Wu, X. Cao, Z. S. Qin,
S. Wang, F. Y. Feng and A. M. Chinnaiyan, Nature, 2014,
510, 278–282.
A. Bouillot, R. Gosmini, V.-L. Nguyen, D. Delannee, J. Seal,
F. Blandel, A.-B. Boullay, E. Boursier, S. Martin,
J.-M. Brusq, G. Krysa, A. Riou, R. Tellier, A. Costaz,
P. Huet, Y. Dudit, L. Trottet, J. Kirilovsky and E. Nicodeme,
Bioorg. Med. Chem. Lett., 2012, 22, 2963–2967.
7 P. Filippakopoulos, S. Picaud, M. Mangos, T. Keates,
J.-P. Lambert, D. Barsyte-Lovejoy, I. Felletar, R. Volkmer, 18 S. Picaud, C. Wells, I. Felletar, D. Brotherton, S. Martin,
¨
S. Muller, T. Pawson, A.-C. Gingras, C. H. Arrowsmith and
P. Savitsky, B. Diez-Dacal, M. Philpott, C. Bountra,
H. Lingard, O. Fedorov, S. Muller, P. E. Brennan, S. Knapp
and P. Filippakopoulos, Proc. Natl. Acad. Sci. U. S. A., 2013,
110, 19754–19759.
S. Knapp, Cell, 2012, 149, 214–231.
8 M. W. Moyer, Nat. Med., 2011, 17, 1325.
9 M. A. Dawson, R. K. Prinjha, A. Dittmann, G. Giotopoulos,
M. Bantscheff, W.-I. Chan, S. C. Robson, C.-w. Chung, 19 P. Filippakopoulos, J. Qi, S. Picaud, Y. Shen, W. B. Smith,
C. Hopf, M. M. Savitski, C. Huthmacher, E. Gudgin,
D. Lugo, S. Beinke, T. D. Chapman, E. J. Roberts,
P. E. Soden, K. R. Auger, O. Mirguet, K. Doehner,
R. Delwel, A. K. Burnett, P. Jeffrey, G. Drewes, K. Lee,
B. J. P. Huntly and T. Kouzarides, Nature, 2011, 478, 529–533.
O. Fedorov, E. M. Morse, T. Keates, T. T. Hickman,
I. Felletar, M. Philpott, S. Munro, M. R. McKeown,
Y. Wang, A. L. Christie, N. West, M. J. Cameron,
B. Schwartz, T. D. Heightman, N. La Thangue,
C. A. French, O. Wiest, A. L. Kung, S. Knapp and
J. E. Bradner, Nature, 2010, 468, 1067–1073.
¨
10 S. Muller and S. Knapp, MedChemComm, 2014, 5, 288–296.
11 C. W. Sutton, Br. J. Pharmacol., 2012, 166, 457–475.
12 S.-E. Ong, M. Schenone, A. A. Margolin, X. Li, K. Do,
M. K. Doud, D. R. Mani, L. Kuai, X. Wang, J. L. Wood,
N. J. Tolliday, A. N. Koehler, L. A. Marcaurelle, T. R. Golub,
20 H. Herrmann, K. Blatt, J. Shi, K. V. Gleixner, S. Cerny-
¨
Reiterer, L. Mullauer, C. R. Vakoc, W. R. Sperr,
H.-P. Horny, J. E. Bradner, J. Zuber and P. Valent,
Oncotarget, 2012, 3, 1588–1599.
R. J. Gould, S. L. Schreiber and S. A. Carr, Proc. Natl. Acad. 21 B. Huang, X.-D. Yang, M.-M. Zhou, K. Ozato and L.-F. Chen,
Sci. U. S. A., 2009, 106, 4617–4622. Mol. Cell. Biol., 2009, 29, 1375–1387.
13 P. V. Fish, A. S. Cook, C. Phillips, A. F. Bent, J. E. J. Mills and 22 M. H. Jones, M. Numata and M. Shimane, Genomics, 1997,
N. Sciammetta, WO2013027168A1, 2013. 45, 529–534.
14 C.-w. Chung, H. Coste, J. H. White, O. Mirguet, J. Wilde, 23 D. N. Perkins, D. J. C. Pappin, D. M. Creasy and J. S. Cottrell,
R. L. Gosmini, C. Delves, S. M. Magny, R. Woodward, Electrophoresis, 1999, 20, 3551–3567.
S. A. Hughes, E. V. Boursier, H. Flynn, A. M. Bouillot, 24 F. Wang, C. B. Marshall and M. Ikura, Cell. Mol. Life Sci.,
P. Bamborough, J.-M. G. Brusq, F. J. Gellibert, E. J. Jones, 2013, 70, 3989–4008.
A. M. Riou, P. Homes, S. L. Martin, I. J. Uings, J. Toum, 25 N. Vo and R. H. Goodman, J. Biol. Chem., 2001, 276, 13505–
´
C. A. Clement, A.-B. Boullay, R. L. Grimley, F. M. Blandel,
13508.
R. K. Prinjha, K. Lee, J. Kirilovsky and E. Nicodeme, J. Med. 26 M. Delvecchio, J. Gaucher, C. Aguilar-Gurrieri, E. Ortega and
Chem., 2011, 54, 3827–3838.
D. Panne, Nat. Struct. Mol. Biol., 2013, 20, 1040–1046.
1878 | Med. Chem. Commun., 2014, 5, 1871–1878
This journal is © The Royal Society of Chemistry 2014