Design, synthesis, and biological evaluation of dual targeting inhibitors of histone deacetylase 6/8 and bromodomain BRPF1

Article abstract of DOI:10.1016/j.ejmech.2020.112338

Histone modifying proteins, specifically histone deacetylases (HDACs) and bromodomains, have emerged as novel promising targets for anticancer therapy. In the current work, based on available crystal structures and docking studies, we designed dual inhibitors of both HDAC6/8 and the bromodomain and PHD finger containing protein 1 (BRPF1). Biochemical and biophysical tests showed that compounds 23a,b and 37 are nanomolar inhibitors of both target proteins. Detailed structure-activity relationships were deduced for the synthesized inhibitors which were supported by extensive docking and molecular dynamics studies. Cellular testing in acute myeloid leukemia (AML) cells showed only a weak effect, most probably because of the poor permeability of the inhibitors. We also aimed to analyse the target engagement and the cellular activity of the novel inhibitors by determining the protein acetylation levels in cells by western blotting (tubulin vs histone acetylation), and by assessing their effects on various cancer cell lines.

Full text of DOI:10.1016/j.ejmech.2020.112338

Journal Pre-proof
Design, synthesis, and biological evaluation of dual targeting inhibitors of histone
deacetylase 6/8 and bromodomain BRPF1
Ehab Ghazy, Patrik Zeyen, Daniel Herp, Martin Hügle, Karin Schmidtkunz, Frank
Erdmann, Dina Robaa, Matthias Schmidt, Elizabeth R. Morales, Christophe Romier,
Stefan Günther, Manfred Jung, Wolfgang Sippl
PII:
S0223-5234(20)30307-X
DOI:
https://doi.org/10.1016/j.ejmech.2020.112338
EJMECH 112338
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 17 February 2020
Revised Date: 9 April 2020
Accepted Date: 13 April 2020
Please cite this article as: E. Ghazy, P. Zeyen, D. Herp, M. Hügle, K. Schmidtkunz, F. Erdmann, D.
Robaa, M. Schmidt, E.R. Morales, C. Romier, S. Günther, M. Jung, W. Sippl, Design, synthesis, and
biological evaluation of dual targeting inhibitors of histone deacetylase 6/8 and bromodomain BRPF1,
European Journal of Medicinal Chemistry (2020), doi: https://doi.org/10.1016/j.ejmech.2020.112338.
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition
of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of
record. This version will undergo additional copyediting, typesetting and review before it is published
in its final form, but we are providing this version to give early visibility of the article. Please note that,
during the production process, errors may be discovered which could affect the content, and all legal
disclaimers that apply to the journal pertain.
©
2020 Published by Elsevier Masson SAS.

Design, synthesis, and biological evaluation of dual targeting inhibitors of histone deacetylase
/8 and bromodomain BRPF1
6
a
a
b
b
b
a
Ehab Ghazy ; Patrik Zeyen Daniel Herp ; Martin Hügle ; Karin Schmidtkunz ; Frank Erdmann ;
a
a
c
c
b
Dina Robaa ; Matthias Schmidt ; Elizabeth R. Morales ; Christophe Romier ; Stefan Günther ;
b
a
Manfred Jung ; Wolfgang Sippl
a
Institute of Pharmacy, Martin-Luther University of Halle-Wittenberg, 06120 Halle/Saale,
Germany
b
Institute of Pharmaceutical Sciences, University of Freiburg, 79104 Freiburg, Germany
c
Département de Biologie Structurale Intégrative, Institut de Génétique et Biologie Moléculaire et
Cellulaire (IGBMC), Université de Strasbourg, CNRS, INSERM, 67404 Illkirch Cedex, France
Corresponding author:
Wolfgang Sippl
Wolfgang.sippl@pharmazie.uni-halle.de
Wolfgang-Langenbeck-Str. 4, 06120 Halle (Saale)
Abstract
Histone modifying proteins, specifically histone deacetylases (HDACs) and bromodomains, have
emerged as novel promising targets for anticancer therapy. In the current work, based on available
crystal structures and docking studies, we designed dual inhibitors of both HDAC6/8 and the
bromodomain and PHD finger containing protein 1 (BRPF1). Biochemical and biophysical tests
showed that compounds 23a,b and 37 are nanomolar inhibitors of both target proteins. Detailed
structure-activity relationships were deduced for the synthesized inhibitors which were supported
by extensive docking and molecular dynamics studies. Cellular testing in acute myeloid leukemia
(AML) cells showed only a weak effect, most probably because of the poor permeability of the
inhibitors. We also aimed to analyse the target engagement and the cellular activity of the novel
inhibitors by determining the protein acetylation levels in cells by western blotting (tubulin vs
histone acetylation), and by assessing their effects on various cancer cell lines.
Keywords
Epigenetics, dual targeting inhibitors, HDAC6, HDAC8, bromodomain, BRPF1, hydroxamic acids,
acute myeloid leukemia.
1

1. Introduction
Post-translational modifications (PTMs) of histones, along with DNA methylation, are the most
extensively studied pathways of epigenetic control of gene expression [1]. Histones are subject to
various PTMs that include acylation, acetylation, methylation, phosphorylation, and ubiquitination.
Acetylation of lysine residues of histones is a dynamic process which results in opening up the
chromatin structure. This leads to an increase in the accessibility of the DNA by the transcription
machinery and is usually associated with increased gene expression [2-4]. The state of histone
acetylation is controlled by different regulators, namely writer proteins; (histone acetyltransferases
(
(
HATs)), reader proteins; (bromodomains (BRDs)), and eraser proteins; (histone deacetylases
HDACs)). Deregulation of these key players, and consequently abnormal acetylation levels, is
linked to several pathologies such as inflammatory, metabolic, and cardiovascular diseases, and
more clearly cancer [5-10].
Bromodomains usually occur as an integral part of larger protein complexes. Due to their ability to
specifically recognize ε-N-acetylated lysine residues, they are generally responsible - together with
other epigenetic readers - for the recruitment of transcription factors to chromatin. The human
proteome contains 61 bromodomains, which are present within 46 different proteins and are
classified into eight distinct families [11, 12]. Due to their relation to different malignancies and
their druggability, they have emerged as promising targets for anticancer therapy. Many inhibitors
showed notable potency and selectivity in biochemical and biophysical assays against different
bromodomain classes, together with significant in vitro activity against various tumor cell lines
(reviewed in [13, 14]). The bromodomain and PHD finger containing protein (BRPF) family has
recently received increasing interest to elucidate its physiological role and pathological functions.
BRPFs have multiple reader domains, including a bromodomain, and act as a scaffold for the
recruitment and assembly of the histone acetyltransferases of the MYST family. Normal activity of
these HATs is essential for different physiological processes, whereas their deregulation is
associated with overexpression of oncogenes and development of different tumors, more notably
leukemia [15-17]. Several BRPF inhibitors containing a 1,3-dimethyl benzimidazolone scaffold
were reported (Figure 1; compound I), which not only showed high potency and selectivity against
the BRPF family, but in some cases also subtype selectivity for BRPF1 [18, 19]. This scaffold was
also utilized to design dual targeting inhibitors of BRPF1 and the transcription factor TRIM24 (e.g.
compound II, Figure 1) as well as dual inhibitors of BRPF and other bromodomains [20-23].
Another group reported a pan BRPF inhibitor containing a 1,3-dimethylquinolin-2-one scaffold
(e.g. compound III; Figure 1) that showed low nanomolar potency against BRPFs and excellent
selectivity [24, 25]. Recently, a hit-to-lead campaign identified a 1,4-dimethyl-2,3-dioxo-
quinoxaline (e.g. compound IV; Figure 1) and 2,4-dimethyloxazole derivatives as low micromolar
probes for BRPF1 with good selectivity [26, 27].
2

O
S
O
O
O
N
N
N
O
O
O
H
N
HN
N
N
O
O
O
S
HN
S
O
HN
N
N
N
O
O
O
O
O
N
O
NC
HN
N
O
O
II (IACS-9571)
Dual TRIM24-BRPF1inhibitor
I (GSK6853)
BRPF1; PIC50 8.1 (IC50 8 nM), TR-FRET
III (NI-57)
BRPF1; K 40 nM, ITC
IV
BRPF1; K
TRIM24; K
d
2.1 nM, bromoELECT
1.3 nM, bromoELECT
d
d
BRPF1; K 1.8 µM, Bromoscan
d
BRPF1 inhibitors
Figure 1: Examples of previously reported BRPF bromodomain inhibitors
HDACs are a group of enzymes responsible for the removal of acyl groups from acylated lysine
residues in histones and non-histone proteins. HDACs comprise 18 isoforms categorized into 4
classes differing in size, cellular distribution, substrate, and mechanism of catalytic activity. The
deacetylase activity of classical HDACs (class I, II and IV) is mediated through zinc ion, while
+
sirtuins (class III) depend on nicotinamide adenine dinucleotide (NAD ) for their action [28, 29].
Given the large variation of their substrates, HDACs regulate diverse physiological processes, in
addition to their epigenetic role. Their aberrant activity is linked to different pathologies such as
cardiac hypertrophy [30, 31], neurodegenerative diseases [32, 33], viral infections [34, 35], and
cancer [36, 37]. As a result, a lot of effort has been devoted to the development of HDAC inhibitors
(HDACi) to fully reveal their physiological role. This work has already resulted in a large number
of promising probes and some candidates in clinical trials (either as single agents or in combination
therapy). Four drugs for the treatment of specific hematological malignancies have been approved
by the FDA [38]. However, currently FDA approved drugs are pan HDAC inhibitors, with several
side effects such as cardiotoxicity. This necessitates the design of selective HDAC inhibitors for
disease-related isoforms [39-41].
The quest for selective HDAC6 and 8 inhibitors is of high importance given their specific role in
different disorders and limited side effects that oberved from their inhibition or knockdown studies.
HDAC6 is a class IIb isoform that localizes in the cytoplasm, and deacetylates mainly non-histone
proteins including α-tubulin, cortactin and heat-shock protein 90 (Hsp90) [42, 43]. Over the last
decade, a variety of HDAC6 selective inhibitors (in vitro) were reported ([44-50] and reviewed
[
51]) showing nanomolar inhibitory activity, good in vitro selectivity over other isoforms in
biochemical assays (Figure 2), and some also exhibited promising in vitro activity against various
cancer cell lines. Additionally HDAC6 deregulation is associated with different neurodegenerative
disorders, inflammatory and rare diseases [52-55], and its selective inhibitors could represent a
valuable tool to study the mechanisms underlying these diseases and/or a potential therapeutic tool
to treat them. Recently published studies have however shown that selective HDAC6 inhibition in
cells is not sufficient for an anti-cancer effect and that the observed anti-cancer effect of reported
HDAC6 inhibitors might be the result of inhibiting other HDACs or other off-targets [56, 57]. Class
I member HDAC8 localizes to either the nucleus or the cytoplasm, and therefore can interact with
non-histone proteins such as cortactin, SMC3, ERRα and p53. HDAC8 is associated with a wide
variety of tumors, and recent evidence suggests a potential therapeutic benefit from its inhibition
[
58, 59]. Recent research also revealed some unique structural features in this isoform, which
distinguish it from other HDACs and which can be utilized to design selective inhibitors [60]. As a
result, a variety of probes emerged (Figure 3) that showed preferential in vitro inhibition of
HDAC8 compared to other isoforms ([61-66], reviewed in [67, 68]).
3

O
O
H
N
O
N
H
Br
H
N
N
OH
N
N
N
O
S
OH
OH
N
O
N
O
O
O
VII
V
VI
HDAC6; IC50 59 nM,
HDAC6; IC50 17 nM,
HDAC6; IC50 5.2 nM,
O
O
O
H
N
N
H
N
S
N
N
H
N
H
H
H
O
N
N
H
N
OH
OH
N
OH
OH
O
O
O
O
VIII
IX
HDAC6; IC50 6 nM,
X (Tubastatin A)
HDAC6; IC50 15 nM,
XI (Nexturastat A)
HDAC6; IC50 5 nM,
HDAC6; IC50 14 nM,
Figure 2: Examples of previously reported HDAC6 inhibitors.
O
O
H
H
H
N
N
N
OH
N
OH
OH
N
O
O
S
N
N
O
O
XIV
HDAC8; IC50 27 nM
XII (PCI-34051)
HDAC8; IC50 10 nM
XIII (NCC149)
HDAC8; IC50 70 nM
O
OH
NH
O
OH
O
N
H
N
N
HN OH
O
H O
H
N
N
N
S
OH
N
O
O
H
O
XV (OJI-1)
HDAC8; IC50 0.8 nM
XVI
HDAC8; IC50 27 nM
XVII
HDAC8; IC50 50 nM
XVIII
HDAC8; IC50 70 nM
Figure 3: Examples of previously reported HDAC8 inhibitors
Several studies suggest that combination therapies of epigenetic modulators could achieve better
clinical results than a monotherapy, especially against solid and resistant tumors (reviewed in [69]).
As a result, the polypharmacology concept was extended to the epigenetics field, assuming that a
multi-target inhibitor could be more effective than single agents. Moreover, such an inhibitor is
postulated to show higher therapeutic efficacy, better predictable pharmacokinetic profile, and
improved patient compliance as compared to a combination therapy. In this context, several dual
acting HDACi were designed to interact with a second target such as kinases, metalloproteinases,
topoisomerases, and others (reviewed in details in [70]). Bromodomains received also some
interest in this regard, as some dual bromodomain/kinase inhibitors were investigated [71-74]. Of
particular relevance to the current work are the attempts to design dual HDAC/BRD epigenetic
inhibitors [75-79]. In all of the reported studies, the rationale was to change the cap group of
HDACi, mostly SAHA, to a reported BRD4 inhibiting pharmacophore. Indeed, the authors were not
only able to achieve dual inhibitory activity for some compounds (Figure 4), but also promising in
vitro activity against some cancer cell lines. However, the results failed to show superior activity
4

over the original HDAC or BRD inhibitors. Additionally, little information was given regarding the
selectivity of these dual inhibitors on different HDAC isoforms.
5
5
XIX
XX (DUAL946)
IC 50 HDAC1: 250 nM
IC 50 HDAC: 260 nM
IC 50 BRD4: 2700 nM
IC 50 HDAC6: 420 nM
IC 50 BRD4: 50 nM
3
4
4
XXIII
IC 50 HDAC1: 180 nM
XXI
XXII
IC 50 HDAC2: 300 nM
IC 50 HDAC1: 150 nM
IC 50 HDAC1: 32nM
IC 50 HDAC3: 5 nM
IC 50 BRD4: 670 nM
IC 50 BRD4: 225 nM
BRD4 inhibition @ 10 µM: 88%
Figure 4: Previously reported dual HDAC/BRD inhibitors
Therefore we wanted to focus on the design of dual targeting inhibitors based on isoform selective
HDACi. We have previously reported the structural guided design, optimization and synthesis of
benzhydroxamic acids as potent and selective HDAC8 inhibitors [64, 80]. These inhibitors were
used as starting point, as we set to modify their structure by including a bromodomain inhibiting
scaffold. We decided to target the BRPF1 bromodomain rather than the extensively studied BRD4
from the bromodomain and extra-terminal motif (BET) family. BRPF1 was chosen in the current
study due to the reported activity of BRPF1 inhibitors against several cancer cell lines as well as
due to the availability of crystal structures in complex with inhibitors. In addition, some BRPF1
inhibitors were reported to selectively inhibit it over other bromodomains and BRPF isoforms.
Guided by molecular modelling studies, the essential structural features for binding to both targets
were merged to come up with dual targeting inhibitors for HDAC8 and BRPF1. Moreover, we
extended the scope of our dual targeting inhibitors to selectively target HDAC6 and BRPF1. We
report on the synthesis and in vitro testing of dual targeting inhibitors against specific HDAC
isoforms and non-BET bromodomains.
2. Results and discussion
2.1. HDAC8/BRPF1 inhibitors
Our previously reported HDAC8 inhibitors were designed, so that the benzhydroxamic acid moiety,
which occupies the lysine tunnel and chelates the zinc ion, is linked to an aromatic cap group
through amine, amide, inverse amide and ether moieties. These two-atom linkers provided
additional interactions with the enzyme, and together with the meta-substitution pattern contributed
to the HDAC8 selectivity [64]. According to our experience with HDAC8 inhibitors, meta-
substitution of the cap group with respect to the benzhydroxamic acid moiety is important for
selective inhibition. These findings were also supported by other groups [61, 81]. In the current
work, we first maintained this substitution pattern and chose the 1,3-dimethylquinolin-2-one to
5

serve as a cap group and BRPF targeting scaffold. This moiety was selected as it contains the
essential features to bind BRPF1, namely the N-methyl and carbonyl groups to fit in the
acetyllysine binding pocket, and the 3-methyl substituent to increase the hydrophobic interactions
with the bromodomain [24]. This ring is also chemically advantageous since it can be easily
synthesized and functionalized with an amino group at position 6 to attach the benzhydroxamic acid
moiety. Finally, BRDi containing this scaffold showed nanomolar potency against BRPF1, and
excellent selectivity over other bromodomains [24, 25]. Compounds 17a,b and 20a,b were
generated (Figure 5) by retaining the amine and inverse amide two-atom linkers as found in our
original HDAC8i. Meanwhile in compounds 23a,b, we introduced a sulfamoyl linker [61] to retain
the bent conformation of the original BRPF1 inhibitor [24]. The previous compounds are designed
to block either of the targets. We then designed a structurally different dual HDAC8/BRPF1
inhibitor 32. The idea here was not to modify the cap group of our previously reported selective
HDAC8i XVIII but rather to attach it as an intact unit through a relatively longer linker to the 1,3-
dimethylquinolin-2-one scaffold (Figure 5). Compound 32 might be able to bind to both targets
simultaneously.
NC
O H
N
S
O
N
O
O
O
O
reported BRPF inhibitor
N
N
N
HO
O
HO
O
HO
NH
HN
R
NH
HN
R
NH
HN
S
R'
N
O
O
Pharmacophore fusion
H
N
O
O
OH
Modification of the cap group
R
R
H
R
O
R'
17a; R = OCH3
17b; R = Cl
20a; R = OCH3
20b; R = Cl
23a; R = OCH3
23b; R = Cl
H
N
H
N
OH
O
O
reported HDAC8 inhibitors
OH
OH
NH
O
NH
O
O
N
Attachment of BRPF inhibiting scaffold
N
N
HN
H
H
O
O
O
O
XVII
HDAC8; IC50 70 nM
32
Figure 5: Design of dual targeting HDAC8/BRPF1 inhibitors
We determined the activity of compounds 17a,b, 20a,b and 23a,b on human recombinant HDAC8
Table 1). As expected, all the compounds showed a good inhibitory activity, with the amide
derivatives 17a and 17b showing the lowest IC values (113 and 65 nM). This was in accordance
(
5
0
with our previous observation that this orientation of the amide bond is favorable for the interaction
with HDAC8 [64]. Compounds 20a,b, bearing an amine linker, showed lower activity, while the
sulfonamides 23a,b displayed the lowest inhibitory activity, albeit IC values still remained in the
5
0
submicromolar range. As expected, docking studies of these meta-substituted derivatives in
HDAC8 show the capping group, i.e. the 1,3-dimethylquinolin-2-one scaffold, accommodated in
the HDAC8-specific side pocket where it undergoes π-π stacking interactions with Tyr306 (Figure
6).
6

Figure 6: A. Predicted binding mode of 23a (green sticks) in HDAC8 (PDB ID 2V5X). B.
Predicted binding mode of 17a (cyan sticks) in HDAC8 (PDB ID 2V5X). The ligands are shown as
cyan sticks, side chains of binding site residues as white sticks, water molecules as red spheres and
the catalytic zinc ion as orange sphere. Yellow-dashed lines indicate hydrogen bond interactions or
metal-coordination, while cyan-dashed lines depict π-π stacking interactions.
Table 1: Inhibitory activity of the HDAC8/BRPF1 inhibitors against different human HDACs and
BRPF1. Recombinant HDAC1,6,8 and fluorogenic peptide substrates (HDAC1,6: ZMAL (Z
(Ac)Lys-AMC), HDAC8: Fluor-de-Lys) were used for enzymatic testing. BRPF1 binding was
measured by isothermal titration calorimetry.
IC (nM) / % IC (nM) / %
5
0
50
IC (nM)
HDAC8
K BRPF1
ꢀ99ꢁꢂ
(nM)
5
0
d
Compound
inhibition
HDAC1
6900 ± 500
n.d.
inhibition
(nM)
HDAC6
709 ± 174
n.d.
17a
65 ± 7
113 ± 9
857
n.d.
690 - 1065
n.d.
17b
0
@ 1 µM
40 @ 1 µM
2
0a
0b
3a
3b
555 ± 169
n.d.
n.d.
n.d.
n.d.
3
1 @10 µM* 68 @ 10 µM*
@ 1 µM 28 @ 1 µM
0 @10 µM* 84 @ 10 µM*
0
2
3193 ± 660
2
6
@ 1 µM
21 @ 1 µM
64 @ 10
µM**
13 @ 1 µM
59 @10 µM
2
38 @ 10
µM**
443 ± 23
560 ± 42
956 ± 74
67
55 – 82
198 – 275
2544 - 6022
13 @ 1 µM
2
234
3
1 @ 10 µM
5
0 @ 1 µM
68 @ 10
µM**
3
2
5200 ± 1100
4080
n.d.: not determined. For K BRPF1 the P99 interval is given.
d
*
*
Self-fluorescence of the compounds hindered measurements at concentrations above 10 µM.
* Solubility problems at higher concentration hindered IC value determination.
5
0
The most active compound on hHDAC8 17a and the sulfonamide derivatives 23a,b were then
selected for the BRPF1 binding assay, where the amide derivative 17a showed only a modest
affinity. Interestingly, the sulfonamides 23a,b had a much better affinity with K values in the
d
7

nanomolar range. In order to rationalize the difference in the activities between the sulfonamide
and amide derivatives (23a and 17a, respectively), docking studies were performed into the crystal
structure of BRPF1 in complex with NI-57 (PDB ID 5MYG) using Glide (Schrödinger LLC, NEW
York, USA). Our docking studies revealed that compounds bearing a sulfonamide linker, as
exemplified by 23a in (Figure 7A), could adopt a highly similar binding conformation as the co-
crystallized ligand. The quinolinone ring of 23a is embedded in the acetyllysine site, showing the
conserved hydrogen bond interaction with Asn708 and a water molecule mediated hydrogen bond
with Tyr665. In addition, two π-π stacking interactions between the side chain of Phe714 and the
quinolone ring on the one hand and the phenyl ring on the other hand, were observed. Meanwhile,
the amide linker in 17a prevents the compound from adopting the bent conformation observed in
compounds bearing a sulfonamide linker. The predicted binding mode shows that the
benzhydroxamic acid moiety is solvent-exposed (Figure 7B) which explains its significantly
decreased activity.
Figure 7: A. Predicted binding mode of 23a (green sticks) in BRPF1. B. Predicted binding mode of
17a (cyan sticks) in BRPF1. The ligands are shown as green sticks, side chains of binding site
residues as white sticks, and water molecules as red spheres. Yellow-dashed lines indicate hydrogen
bond interactions and cyan-dashed lines π-π stacking interactions.
We then tested the activity of the most promising compounds on the class I member HDAC1 and
class IIb HDAC6, where they generally showed weak inhibition (Table 1). From this series, we
were able to identify compounds 23a,b as nanomolar inhibitors of both HDAC8 and BRPF1, which
showed low in vitro activity against HDAC1 and 6.
Compound 32 showed only a modest activity against hHDAC8 and only weak micromolar activity
against BRPF1.
2.2. HDAC6/BRPF1 inhibitors
In order to target HDAC6, we took advantage of the common structure characteristics of selective
HDAC6 inhibitors [81]. They are usually aromatic hydroxamic acids with a cap group located in the
para position. The linker consists of one to three atoms often including a methylene group. As we
already have the bulky 1,3-dimethylquinolin-2-one as a cap group, we hypothesized that shifting it
to the para-position would shift the activity to HDAC6, and hence dual HDAC6/BRPF1 inhibitors
could be obtained (Figure 8). For this series, we generally retained the sulfamoyl linker since the
8

compounds containing this linker 23a,b from the HDAC8/BRPF1 series showed the highest affinity
for BRPF1 bromodomain. While in compound 37 we retained the two-atom sulfamoyl bridge, we
incorporated an additional methylene group in compounds 44a,b. To further prove the effect of the
substitution pattern on HDAC isoform selectivity, compounds with meta substitution pattern 51a,b
were synthesized as negative controls for HDAC6 inhibition (isomers of 44a,b respectively).
37
BRPF inhibiting scaffold
cap group) to position-4
(
Dual HDAC8/BRPF inhibitors
Dual HDAC6/BRPF inhibitors
negative controls for HDAC6
2
3a; R = OCH 3
44a; R = OCH 3
44b; R = H
51a; R = OCH 3
51b; R = H
23b; R = Cl
Figure 8: Design of dual HDAC6/BRPF1 inhibitors
These five compounds were first tested in vitro for HDAC inhibition, where the para substituted
compounds 37 and 44b showed, as expected, IC50 values in the nanomolar range (Table 2).
However, both compounds did not show pronounced selectivity and inhibited HDAC1, 6 and 8 in
submicromolar concentrations. The observed HDAC6 inhibitory activity could be strongly
attributed to the 4-substitution pattern of the cap group, as the meta substituted compounds from the
first series 23a,b and the control compound 51b showed only a very weak HDAC6 inhibitory
activity. Interestingly, compound 44a did not show the expected HDAC6 activity, despite the 4-
substitution pattern of the cap group, suggesting that the introduction of an ortho methoxy group
(with respect to the cap group) has a negative effect on the activity against HDAC6. Docking of 37
and 44b into HDAC6 showed that both compounds adopt a highly similar binding mode (Figure 9),
where the hydroxamic acid group is able to chelate the catalytic zinc ion in a monodentate fashion
and undergo a hydrogen bond interaction with the side chain of Tyr 782 in addition to a water
mediated interaction. Meanwhile the quinolone capping group is embedded in a hydrophobic region
lined with Phe620 and His500 and undergoes and additional hydrogen bond interaction with
Asn494. An additional ortho methoxy substitution, as found in 44a, would lead in this case to a
steric clash with the side chain of Leu749, which might explain the loss of HDAC6 inhibitory
activity.
9

Figure 9: A. Predicted binding mode of 37 (cyan sticks) in HDAC6 (PDB ID 5EDU). B. Predicted
binding mode of 44b (green sticks) in HDAC6. The ligands are shown as green sticks, side chains
of binding site residues as white sticks, and water molecules as red spheres. Yellow-dashed lines
indicate hydrogen bond interactions or metal coordination and cyan-dashed lines π-π stacking
interactions.
Table 2: Inhibitory activity of the HDAC6/BRPF1 inhibitors against different human HDACs and
BRPF1. Recombinant HDAC1,6,8 and fluorogenic peptide substrates (HDAC1,6: ZMAL (Z
(Ac)Lys-AMC), HDAC8: Fluor-de-Lys) were used for enzymatic testing. BRPF1 binding was
measured by isothermal titration calorimetry.
IC (nM) /
% inhibition
HDAC1
IC (nM) /
% inhibition
HDAC6
344 ± 41
15 @ 1 µM
5
0
50
IC (nM)
K BRPF1
ꢀ99ꢁꢂ
(nM)
5
0
d
Compound
HDAC8
908 ± 274
231 ± 24
360 ± 44
158 ± 21
(nM)
175.2
1582
1497
n.d.
37
797 ± 282 nM
160 – 193
1402 - 1609
1358 - 1652
n.d.
11 @ 1 µM
4
4a
4b
1a
2
9 @ 10 µM* 46 @ 10 µM*
4
545 ± 59 nM
@ 1 µM
9 @ 10 µM
@ 1 µM
6 @ 10 µM
152 ± 13
14 @ 1 µM
63 @ 10 µM
3 @ 1 µM
4
5
2
2
51b
465 ± 72
n.d.
n.d.
2
33 @ 10 µM
n.d.: not determined. For K BRPF1 the P99 interval is given.
d
*
Solubility problems at higher concentration hindered IC value determination.
50
The para-substituted compounds 37 and 44a,b were also tested in the BRPF1 binding assay (Table
), where only 37 showed a nanomolar K value. The introduction of the additional methylene
2
d
group in the linker of 44a,b remarkably decreased the BRPF1 affinity (almost 9fold decrease for
4b compared to 37). In order to comprehend the observed negative effect of an additional
4
methylene group in the linker on the BRPF1 affinity, we first performed docking into the crystal
structure of BRPF1 as previously described. Compound 37 could maintain the same binding
conformation as the co-crystallized ligand, where the quinolinone ring shows the conserved
hydrogen bond interaction with Asn708 and a water mediated hydrogen bond with Tyr665, in
addition to π-π stacking interactions between the side chain of Phe714 (Figure 10A). Moreover, the
10

benzhydroxamic moiety is placed in a perpendicular T-shaped orientation to the side chain of
Phe714 where it can undergo π-π stacking interactions. Meanwhile, the docking pose of compound
44b shows that it similarly adopts a bent conformation, where the benzhydroxamic moiety is still
placed in the vicinity of Phe714 (Figure 10B). Notably, the predicted binding mode compound 37
perfectly overlaps with the cocrystallized analogue NI-57, while in 44b the benzhydroxamic group
shows a strong deviation (Figure 10C). We postulated that, owing to the flexibility of the
methylene-sulfamoyl linker in 44b, the predicted binding mode is not stable. In order to investigate
this hypothesis, molecular dynamics (MD) simulations of the predicted complexes of BRPF1 with
37 and 44b were performed using Amber16. These revealed that, as expected, the predicted binding
mode of 37 in BRPF1 is highly stable with the ligand having a root mean square deviation (RMSD)
below 2.5 Å throughout the 100 ns MD simulation (Figure 10D and Video 1 suppl.). Meanwhile,
the binding mode of 44b is highly unstable (Figure 10E) and the benzhydroxamic moiety becomes
majorly solvent-exposed (Video 2 suppl.). The applied MD simulation setup was also validated
using the cocrystal structure of BRPF1 with NI-57 (Figure S7, Supporting Information).
Figure 10: A. Predicted binding mode of 37 (cyan stick) in BRPF1. B. Predicted binding mode of
44b (green sticks) in BRPF1. C. Overlay of the predicted binding mode of 37 (cyan stick) and 44b
(green sticks) with the cocrystallized ligand NI-57 (yellow sticks). Side chains of binding site
residues are depicted as white sticks, and water molecules as red spheres. Yellow-dashed lines
indicate hydrogen bond interactions and cyan-dashed lines π-π stacking interactions. D. RMSD plot
of the MD simulation of 37 in BRPF1. E. RMSD plot of the MD simulation of 44b in BRPF1.
11

We then tested the activity of the most promising compounds on HDAC1 where they generally
showed no or low inhibition (Table 2). Testing on HDAC8 showed that compound 44a has a good
inhibitory activity against HDAC8 with an IC value in the low nanomolar range.
5
0
Among this group of inhibitors, compound 37 appeared to be the best HDAC6/BRPF1 dual
inhibitor showing a nanomolar in vitro inhibitory activity against both targets, and in vitro
selectivity over HDAC1 and 8.
In addition we synthesized several control compounds including a dimeric BRPF1i compound 38
(Scheme 3) and carboxylic esters 21, 35 and 42b (analogs of 23a, 37 and 44b) lacking the
hydroxamic acid functionality.
2.3. Cellular assay
The first objective of our cellular assays was to determine the potential toxicity of our compounds
on normal cells. Therefore, the compounds were tested in a human epithelial kidney cell line
(
HEK293). We also incorporated some of the intermediate carboxylic acids and esters in the assay
as negative controls. The majority of the compounds showed no effect on HEK 293 cells, except for
2 which displayed only 15% viability of the cells. It is also worth mentioning that some esters
3
showed more toxic effects compared to their corresponding hydroxamic acids, such as 21 and 23a.
Viability of HEK293 cells upon treatment with our compounds is provided in supporting
information, S5.
Our second objective was to assess the potential antiproliferative effects on cancer cell lines. As
previously mentioned, HDAC and BRD are interesting targets for the anticancer therapy, and their
inhibitors showed cytotoxic activity against different tumors. Our main focus was acute myeloid
leukemia (AML) cell lines (THP-1 and HL60) based on some previous reports showing evidence of
significance of HDAC6 [82] and HDAC8 [83, 84] as well as BRPF [15-17] inhibition in these cells.
We tested our most promising compounds 23a,b (HDAC8/BRPF1), 37 and 44b (HDAC6/BRPF1) -
together with their esters as negative controls- on THP-1 and HL60 cell lines to investigate if the
compounds in vitro inhibitory activity could be translated into cellular anticancer activity (Table 3).
Although all the hydroxamic acids showed no or weak growth inhibition on the cells, a much
stronger inhibition was generally observed for the esters, and the GI value of 21 was determined
5
0
to be around 5.5 µM. The activity of the esters could be attributed to the BRPF1 inhibiting scaffold.
In fact, both 21 and the potent BRPF1 inhibitor 13-d [24] have almost the same GI value on THP-
5
0
1
cells. The corresponding hydroxamic acids, however, did not show an inhibition, despite their
bromodomain inhibitory scaffold. We assumed that since the esters are more lipophilic than the
hydroxamic acids (log P value for 21 is 1.83 compared to 1.09 for 23a, calculated with ChemDraw
Ultra 8), they should have better cell permeability, and consequently better cellular activity than the
hydroxamic acids.
To get insight into the cellular activity of our hydroxamic acids, western blotting experiments were
performed with our best HDAC6 inhibitor 44b (IC value: 152 nM, Table 2) in the two cell lines
50
THP-1 and HL60. The cells were incubated with 44b or SAHA (reference compound), and then cell
lysates were blotted against acetylated-α-tubulin (HDAC6 substrate), acetylated-histone H3
(HDAC1 substrate), and glyceraldehyde 3-phosphate dehydrogenase (GAPDH) as a loading control
(Figure 11). At lower concentrations, 44b showed negligible or no induction of tubulin acetylation
in both cell lines, whereas some degree of acetylation was observed at 10 µM. The lack of
hyperacetylation at lower concentrations is in line with the antiproliferative assays, where the
12

compound showed inhibition only at a high concentration of 100 µM. The western blots were
quantified by determining the ratio of acetylated tubulin or acetylated histone 3 versus GAPDH
(Table 4). These results further supported our postulation that our hydroxamic acids have poor
permeability and therefore did not show remarkable cellular activity. It is worth mentioning that the
dimeric compound 38 (Scheme 3 and later discussed in 2.5.2.) showed a promising growth
inhibition on THP-1 cells at 1 µM. However, its relatively poor solubility hindered further testing at
higher concentrations and consequently GI determination.
5
0
Based on these findings, we then aimed to modify some of our hydroxamic acids in order to acquire
better cellular permeability. To achieve that, we masked the hydroxamic acid functionality in 23a
and 44b by synthesizing the corresponding benzyl and para-acetoxy-benzyl ester prodrugs of
hydroxamic acids 59a,b and 60b (Figure 12) as prodrugs. Masking hydroxamic acids was recently
reported as an approach to overcome problems like fast elimination, decreased cellular uptake and
poor tissue penetration caused by the highly polar hydroxamic acid group [85-90]. In addition, para-
acetoxybenzyl-based prodrugs were shown to be completely converted to the parent hydroxamic
acid in plasma [91]. Unfortunately, even these masked hydroxamic acids 59a,b and 60b (log P
values ranging from 2.51-3.08, calculated with ChemDraw Ultra 8) did not show an improvement
in the activity against either THP-1 or HL60 cell lines. Compounds 60a,b showed only promising
inhibition at 100 µM, but not at lower concentrations.
13

Table 3: Antiproliferative activity on acute myeloid leukemia cells THP-1 and HL60. Growth
inhibition was determined using the CellTiter 96AQueous Non-Radioactive Cell Proliferation
Assay.
THP-1
HL60
Primary testing
Conc. Inhibition GI50 Conc. Inhibition GI50
Primary testing
Structure
Compound
[
1
µM]
00
[%]
43
32
0
21
8
16
78
0
[µM]
100
20
4
100
20
4
100
10
1
[%]
39
10
25
33
32
29
34
6
O
O
H
HN
S
N
23a
23b
37
20
4
00
20
4
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
N
HO
O
O
O
1
Cl
O
H
HN
S
N
N
HO
HO
O
O
O
O
100
10
1
O
N
H
HN
S
N
O
O
0
3
O
1
1
00
0
1
89
8
0
100
10
1
27
0
5
HO
N
H
O
S
NH
44b
n.d.
n.d.
O
N
O
H
N
100
10
1
100
10
1
0
20
0
27
3
100
10
1
100
10
1
13
13
2
95
0
O
O
NH
S
O
n.d
n.d
n.d
n.d
N
O
59a
59b
O
O
H
N
O
O
O
O
NH
S
N
O
O
O
O
0
0
O
1
10
1
00
91
1
0
S
H
N
NH
O
O
O
HO
60a
60b
n.d
n.d
n.d
N
O
O
S
100
1
91
1
4
100
10
1
97
0
3
H
N
NH
O
O
0
1
n.d
O
O
O
N
O
1
20
4
100
10
1
00
102
101
40
95
85
0
100
20
4
100
10
1
66
44
39
68
22
5
5
µM
.5
O
O
HN
S
O
N
21
35
n.d.
n.d
O
O
O
O
O
O
n.d
HN
N
S
O
O
O
1
1
00
0
1
93
9
0
100
10
1
78
14
15
O
O
S
NH
n.d
n.d
42b
O
N
O
14

N
O
HN
n.d
n.d
n.d
n.d
3
8
50
10
1
73
84
0
50
10
1
48
55
20
O
N
O
O
S
N
O
H
100
98
97
84
100
10
1
95
98
55
SAHA
10
1
Figure 11: Western blots of acetyl-α-tubulin, acetyl-H3, and GAPDH after treatment of HL60 and
THP-1 cells with SAHA and 44b. GAPDH and DMSO were used as loading and negative controls.
Table 4: Quantification of Western Blots in THP-1 and HL60 Cells
THP-1
HL60
Ac-
tubulin/GAPD
H
Ac-
tubulin/GAPD
H
Compound
DMSO
Ac-
H3/GAPDH
Ac-
H3/GAPDH
1.00
7.67
2.51
2.64
78.80
68.15
6.21
1.00
2.63
1.98
1.53
8.78
10.09
2.78
1.00
7.36
4.96
1.36
309.78
146.10
8.36
1.00
1.38
1.66
0.90
5.73
2.57
1.12
4
4
4
4b 10 µM
4b 1 µM
4b 0.1 µM
SAHA 10 µM
SAHA 1 µM
SAHA 0.1 µM
15

59a
59b
60b
60a*
*
Phenol intermediate 60a (deacetylated)
was obtained as a by-product during the
synthesis of 60b (acetylated) and was
included in the biological testing
Figure 12: Benzyl and para-acetoxy-benzyl ester prodrugs of hydroxamic acids synthesized in the
current work.
2.4. Stability Analysis
Stability stress tests were carried out to investigate the stability of the synthesized hydroxamic
acids, and to see if the masked derivatives are hydrolyzed to release the free hydroxamic acid under
assay conditions. We used the hydroxamic acid 44b and the derivative 60b as representatives. For
this purpose, the compounds were subjected to standard cellular assay conditions (Dulbecco’s
Modified Eagle’s Medium, DMEM) and also to buffer (in vitro assay) conditions (Phosphate-
Buffered Saline, PBS pH 7.4). After incubation at 37 °C for different time intervals, the original
compounds and potential hydrolysis products were detected and quantified via HPLC.
We observed that the free hydroxamic acid 44b showed acceptable stability over 4 days for both in
cellular assay and in vitro conditions. The prodrg 60b was relatively stable under buffer conditions
(70%) and showed only a 15% release of 44b as hydrolysis product after 12 hours. After 96 hours
several non-identified degradation products could be detected. However, under cellular assay
conditions 60b was rapidly hydrolyzed (50% after 4 hours) showing 44b as the major degradation
product. After 24 h, we could only detect 44b without any traces of 60b. The data clearly showed
that the masked prodrug was indeed hydrolyzed to release the active hydroxamic acid under cellular
assay conditions. However, this also suggested that the rapid hydrolysis might be responsible that
the required concentration of the masked hydroxamic acid couldn’t be delivered to cells, where the
hydrolysis is supposed to occur. Experimental details and graphs are provided in the supplementary
information (Figure S6, Supporting Information).
2.5. Chemistry
2.5.1. Synthesis of HDAC8/BRPF1 inhibitors
Synthesis of the 6-amino-1,3-dimethylquinolin-2-(1H)-one (5), the common starting material for the
compounds, was achieved using the procedure reported by Igoe et al [24] with some modifications
(Scheme S.1.1, supporting information). Briefly, 3-methylquinoline (1) was converted to 3-
methylquinolin-2(1H)-one (2) in two steps using 3-chloroperbenzoic acid then benzoyl chloride.
Methylation with iodomethane afforded the N-methyl derivative (3), which was then nitrated to
yield 1,3-dimethyl-6-nitroquinolin-2-(1H)-one (4). Our main modification to the procedure was the
conversion of the latter compound to the targeted amine 5 using a microwave assisted reduction
protocol [92]. This enabled a much faster, more efficient, and easier reduction process than the
16

reported stannous chloride based procedure. To synthesize the first dual HDAC8/BRBF inhibitors
7a,b, 20a,b and 23a,b, the required starting materials were prepared as reported in the literature
Scheme S.1.1, supporting information) [93-95]. The synthesis of the target compounds is
1
(
illustrated in Scheme 1. Briefly, compounds 17a,b were prepared starting from the 5-
methoxycarbonyl-2-substitutedbenzoic acids (9a,b) which were first activated using oxalyl
chloride, and then coupled to 6-amino-1,3-dimethylquinolin-2-(1H)-one (5) to give methyl 3-[(1,3-
dimethyl-2-oxo-1,2-dihydroquinolin-6-yl)carbamoyl]-4-substitutedbenzoates (15a,b). These were
finally hydrolyzed to yield the corresponding carboxylic acids 16a,b. For inhibitors 20a,b, sodium
triacetoxyborohydride was utilized for the reductive amination of aldehydes 8a,b and amine 5 to
synthesize the esters 18a,b, again followed by hydrolysis to afford the 3-[(1,3-dimethyl-2-oxo-1,2-
dihydroquinolin-6-ylamino)methyl]-4-substitutedbenzoic acids (19a,b). Carboxylic acid 22b was
prepared directly from the aromatic sulfonyl chloride 13b and amine 5 using pyridine as a base. The
same protocol for the synthesis of intermediate 22a afforded a major byproduct (which was
extremely insoluble in common solvents, and unfortunately could not be characterized). This
necessitated an alternative route, where compound 5 was reacted with the ester derivative 14 to
first give methyl 3-[(1,3-dimethyl-2-oxo-1,2-dihydroquinolin-6-yl)sulfamoyl]-4-methoxybenzoate
(21), which was further hydrolysed to yield the carboxylic acid 22a. The desired hydroxamic acids
17a,b, 20a,b and 23a,b were finally obtained from the corresponding carboxylic acids 16a,b, 19a,b
and 22a,b using PyBOP as an activating agent and O-(tetrahydro-2H-pyran-2-yl)hydroxylamine
followed by cleavage of the protecting group [65]. We adopted this method rather than the classical
one (reaction with hydroxylamine hydrochloride and a strong base like KOH), as this method
facilitated the purification of the final hydroxamic acids.
R
O
R
R
R
O
O
O
H2N
OH
O
e
O
a
O
d
+
HN
N
HN
N
HN
N
N
O
O
O
O
O
OH
HN
OH
R
O
O
a-b
9
5
15a-b
16a-b
17a-b
R
R
R
CHO H2N
+
O
O
b
O
d
e
HN
N
HN
N
HN
N
N
O
O
O
O
HN
OH
O
OH
O
O
a-b
8
5
18a-b
19a-b
20a-b
O
O
O
O
O
O
O
S
O
S
H2N
+
O
S
Cl
c
O
d
S
O
e
O
HN
HN
HN
N
O
O
O
O
O
O
O
OH
HN OH
N
O
N
O
N
O
14
5
21
22a
23a
Cl
Cl
Cl
O
O
O
S
O
S
S
H2N
O
O
Cl
c
e
+
HN
HN
N
O
O
O
OH
HN OH
O
OH
3b
N
O
N
O
1
5
22b
23b
a; R = OCH3 b; R = Cl
Scheme 1: Synthesis of dual HDAC8/BRPF1 inhibitors 17a,b, 20a,b and 23a,b. Reagents and
conditions: (a) i) oxalyl chloride, cat. DMF, CH Cl , ii) DIPEA, CH Cl ; (b) Na(AcO) BH,
2
2
2
2
3
17

CH COOH, THF; (c) pyridine; (d) i) aqu. NaOH, MeOH, reflux, ii) c. HCl; (e) i) PyBOP, DIPEA,
3
H NOTHP, THF, ii) cat. HCl, THF.
2
In the case of synthesis of compound 32, it was quiet challenging to find the optimal synthetic
sequence. As shown in Scheme 2, the 4-hydroxybenzaldehyde (26) was first alkylated with 5-
bromovalerate (25) to give the ester 27, which was then hydrolyzed to the corresponding acid 28.
Again oxalyl chloride served as an activating agent and the resulting acid chloride was reacted with
5
to give N-(1,3-dimethyl-2-oxo-1,2-dihydroquinolin-6-yl)-5-(4-formylphenoxy)pentanamide (29).
The previously mentioned conditions of reductive amination were not successful for the reaction of
9 with 3-amino-4-methoxybenzoic acid (30), and required a modification by adding trifluoroacetic
2
acid (TFA) [96] to obtain the carboxylic acid 31, which was finally converted to the target
compound 32 using the above mentioned protocol.
O
O
O
O
O
O
a
b
c
HO
Br
O
Br
HO
O
OH
CHO
CHO
24
25
CHO
27
28
26
O
O
O
O
O
H
NH
NH
N
H N
2
CHO
d
e
+
N
O
N
O
N
H N
O
2
O
HO
O
5
29
31
HO
O
0
O
O
3
O
H
N
NH
f
N
O
HN
OH
O
3
2
Scheme 2: Synthesis of dual HDAC8/BRPF1 inhibitor 32. Reagents and conditions: (a) p-
toluenesulfonic acid, CH OH; (b) K CO , acetone, reflux; (c) LiOH.H O, THF, H2O; (d) i) oxalyl
3
2
3
2
chloride, cat. DMF, CH Cl , ii) DIPEA, CH Cl ; (e) Na(AcO) BH, CF COOH, THF; (f) i) PyBOP,
2
2
2
2
3
3
DIPEA, H NOTHP, THF, ii) cat. HCl, THF.
2
2.5.2. Synthesis of HDAC6/BRPF1 inhibitors
The starting materials for the dual HDAC6/BRPF1 inhibitors were the appropriate sulfonyl
chlorides. Trials to directly synthesize the carboxylic acid 36 from the commercially available 4-
(
chlorosulfonyl)benzoic acid (33) again afforded a major by-product. This product had a reasonable
solubility and was identified as 4-[(1,3-dimethyl-2-oxo-1,2-dihydroquinolin-6-yl)sulfamoyl]-N-
1,3-dimethyl-2-oxo-1,2-dihydroquinolin-6-yl)benzamide 38 (Scheme 3). In case of 34, 41a,b and
8a,b we decided to start with the sulfonyl chloride esters. Reaction of the corresponding alkyl
(
4
bromides 40a,b and 47a,b with sodium sulphite and phosphorus pentachloride yielded the aliphatic
sulfonyl chlorides 41a,b and 48a,b (Scheme S.1.2., supporting information). Again reaction of
these sulfonyl chlorides with 5 in pyridine afforded the esters 35, 42a,b and 49a,b, which were
18

hydrolysed to the carboxylic acids and finally converted to the desired hydroxamic acids 37, 44a,b
and 51a,b as previously discussed (Scheme 4).
O
N
HO
O
O
S
O
O
NH
H N
2
S
O
Pyridine
Cl
O
S
+
+
HN
O
HO
N
O
NH
O
N
O
O
33
5
36
38
N
O
Scheme 3: Synthesis of dimeric BRPF1i 38
a
b
c
34
5
35
36
37
a
b
c
41a-b
5
42a-b
43a-b
44a-b
a
b
c
48a-b
5
49a-b
50a-b
51a-b
a; R = OCH 3
b; R = H
Scheme 4: Synthesis of dual HDAC6/BRPF1 inhibitors 37, 44a,b and 51a,b. Reagents and
conditions: (a) pyridne; (b) i) aqu. NaOH, CH OH, reflux, ii) c. HCl; (c) i) PyBOP, DIPEA,
3
H NOTHP, THF, ii) cat. HCl, THF.
2
2.5.3. Synthesis of masked hydroxamic acids
Compounds 59a,b and 60a,b were synthesized as shown in Scheme 5 from the corresponding
carboxylic acids 22a and 43b, respectively. Whereas intermediate 58a was commercially available,
intermediate 58b was synthesized following the procedure reported by Rais et al [91] with some
modifications (Scheme S.1.3., supporting information). Unlike the protocol for hydroxamic acid
synthesis (Schemes 1, 2 and 4), activation of the carboxylic acids using PyBOP to synthesize the
masked hydroxamic acid was not optimal and complicated the purification process. Instead, we
19

used HATU as an activating agent, since its traces and side products could be efficiently removed
by aqueous workup which facilitated purification. It is also worth mentioning that compound 60a
was obtained as a deacetylated side product of compound 60b (probably during the hydrazinolysis
process due to slightly increased reaction time). It was then purified, characterized, identified and
incorporated in our cellular assays.
O
O
R
O
S
O
O
S
HN
O
a
O
+
HN
HN
R
O
N
O
O
a; R = H
b; R = OCOCH3
O
OH
N
O
NH
2
2
2a
58a-b
59a-b
O
H
O
O
O
N
O
H
HN
O
S
H
N
N
O
a
HN
S
O
HO
S
O
N
O
+
O
O
+
O
O
N
O
N
O
O
O
O
NH2
OH
43b
58b
60a
60b
Scheme 5: Synthesis of masked hydroxamic acids 59a,b and 60a,b. Reagents and conditions: (a)
HATU, DIPEA, THF, DMF.
20

3. Conclusion
HDACs and bromodomains represent interesting epigenetic targets whose deregulation is linked to
different disorders. The current work was initiated based on promising results of the combination
therapy of HDAC and bromodomain inhibitors, assuming that a dual target inhibitor could achieve
superior activity to the single agents. We used previously developed selective HDAC8 inhibitors as
starting points and modified the cap group to achieve simultaneous HDAC8 and BRPF1 inhibition.
Two dual HDAC8/BRPF1 nanomolar inhibitors (23a,b) were identified. In vitro assays and
modelling studies showed that small modifications of the linker are not tolerated, especially for the
BRPF1 inhibition. Since HDAC6 is in the scope of this work, we took advantage of the structural
differences between different HDAC isoforms to achieve the dual HDAC6/BRPF1 inhibitor 37.
Further attempts to enhance HDAC6 activity abolished the BRPF1 activity emphasizing the limited
available space to optimize these dual inhibitors. The promising in vitro enzymatic results of the
developed dual target inhibitors were not translated into cellular activity against AML cell lines,
while the corresponding esters, which served as HDAC negative controls with only BRPF1 activity,
showed a modest effect. Therefore, we assumed that the poor permeability of the hydroxamic acid
is the reason for the lack of cellular activity. Indeed, western blot experiments showed a very weak
acetylation of α-tubulin, substantiating the lack of cellular HDAC6 activity of the tested
compounds. In an attempt to enhance the cellular effect of the dual target inhibitors, some masked
hydroxamic acids were synthesized as potential prodrugs. However, no improvement in the
antiproliferative effect could be observed. Reasons for that could be the known limited cell
permeability and short half-life of hydroxamic acids due to fast metabolization. One prospective
modification of the developed dual inhibitors could be a replacement of the hydroxamic acid moiety
by more stable and permeable zinc binding groups. It would be interesting to investigate if this
modification could achieve a balance between activities on both targets and confer cellular activity.
21

4
4
4
. Experimental protocols
.1. Chemistry
.1.1. General
All materials and reagents were purchased from Sigma-Aldrich Co. Ltd. and abcr GmbH. All
solvents were analytically pure and dried before use. Thin layer chromatography was carried out on
aluminum sheets coated with silica gel 60 F254 (Merck, Darmstadt, Germany). For column
chromatography under normal pressure silica gel 60 (0.036−0.200 mm) was used.
Final compounds were confirmed to be of >95% purity based on HPLC. Purity was measured by
UV absorbance at 254 nm. The HPLC consists of an XTerra RP18 column (3.5 ꢀm, 3.9 mm × 100
mm) from the manufacturer Waters (Milford, MA, USA) and two LC-10AD pumps, a SPD-M10A
VP PDA detector, and a SIL-HT autosampler, all from the manufacturer Shimadzu (Kyoto, Japan).
For preparative tasks a XTerra RP18 column (7 ꢀm, 19 mm × 150 mm) from the manufacturer
Waters (Milford, MA, USA) and two LC-20AD pumps were used. The mobile phase was in all
cases a gradient of methanol/ water (starting at 95% water going to 5% water).
Mass spectrometry analyses were performed with a Finnigan MAT710C (Thermo Separation
Products, San Jose, CA, USA) for the ESIMS spectra and with a LTQ (linear ion trap) Orbitrap XL
hybrid mass spectrometer (Thermo Fisher Scientific, Bremen, Germany) for the HRMS-ESI (high
resolution mass spectrometry) spectra. For the HRMS analyses the signal for the isotopes with the
3
5
79
highest prevalence was given and calculated ( Cl, Br).
1
13
H NMR and C NMR spectra were taken on a Varian Inova 500 using deuterated chloroform and
deuterated DMSO as solvent. Chemical shifts are referenced to the residual solvent signals. The
following abbreviations and formulas for solvents and reagents were used: ethyl acetate (EtOAc),
dimethylformamide (DMF), dimethylsulfoxide (DMSO), methanol (MeOH), tetrahydrofuran
(
THF), chloroform (CHCl ), water (H O), dichloromethane (CH Cl ), N,N-diisopropylethylamine
3 2 2 2
(
DIPEA), trimethylamine (TEA), hydrochloric acid (HCl) and trifluoroacetic acid (TFA).
.1.2. General method for the synthesis of methyl 3-[(1,3-dimethyl-2-oxo-1,2-dihydroquinolin-6-
4
yl)carbamoyl]-4-substitutedbenzoates (15a-b)
The appropriate 5-methoxycarbonyl-2-substitutedbenzoic acid 9a-b (3 mmol), DMF (one drop) and
oxalyl chloride (4 mmol) were stirred in CH Cl at room temperature for 3 h. The mixture was then
2
2
added dropwise to a solution of 6-amino-1,3-dimethylquinolin-2-(1H)-one (5; 3 mmol) and DIPEA
8 mmol) in CH Cl , and stirring was continued for another 2 h at room temperature. The reaction
(
2
2
mixture was washed with a saturated aqueous solution of ammonium chloride and brine. The
organic layer was dried over anhydrous sodium sulfate, filtered, and concentrated in vacuo. The
residue was purified using column chromatography (CHCl : MeOH, 100:0-99:1).
3
Detailed characterization data of the compounds are provided in the supporting information.
4.1.3. General method for the synthesis of methyl 3-[(1,3-dimethyl-2-oxo-1,2-dihydroquinolin-6-
ylamino)methyl]-4-substitutedbenzoates (18a-b)
A mixture of the appropriate 4-substituted-3-formylbenzoate 8a-b (1.2 mmol) and 6-amino-1,3-
dimethylquinolin-2-(1H)-one (5; 1 mmol) was dissolved in toluene, and the reaction mixture was
refluxed for 2 h using a dean stark apparatus. The solvent was then evaporated under reduced
pressure, and the crude product was dissolved in CH Cl and cooled to 0 °C. Then sodium
2
2
triacetoxyborohydride (4 mmol) was added and the reaction mixture was stirred overnight at room
22

temperature. The reaction was quenched by adding H O, and the pH was adjusted to 5 with aqueous
2
potassium bicarbonate solution. The mixture was extracted with EtOAc, and the organic layer was
evaporated under vacuum. The products were purified by column chromatography (CHCl : MeOH,
3
100:0-99:1).
Detailed characterization data of the compounds are provided in the supporting information.
4.1.4. General method for the of synthesis of methyl 3(or 4)-[(1,3-dimethyl-2-oxo-1,2-
dihydroquinolin-6-yl)sulfamoyl]-4-substitutedbenzoates (21 and 35) and methyl 3(or 4)-{[(1,3-
dimethyl-2-oxo-1,2-dihydroquinolin-6-yl)sulfamoyl]methyl}-4-substitutedbenzoates (42a-b) and
(49a-b)
To a solution of 6-amino-1,3-dimethylquinolin-2-(1H)-one (5; 1 mmol) in pyridine was added the
appropriate methyl chlorosulfonylbenzoate or methyl (chlorosulfonylmethyl)benzoate 14, 34, 14a-b
or 48a-b (1 mmol), and the reaction was stirred at room temperature overnight. The solvent was
then removed under vacuum, and the residue obtained was dissolved in EtOAc and washed
successively with aqueous 1M HCl, 1M copper sulfate and brine solutions. The organic layer was
dried over anhydrous sodium sulfate, filtered, and concentrated in vacuo. The residue was purified
using column chromatography (CHCl : MeOH, 100:0-99:1).
3
Detailed characterization data of the compounds are provided in the supporting information.
4.1.5. General method for the synthesis of carboxylic acids 16a-b, 19a-b, 22a-b, 36, 43a-b and
50a-b
The appropriate methyl esters 15a-b, 18a-b, 21, 35, 42a-b or 49a-b (1 mmol) was dissolved in
MeOH followed by the addition of aqueous solution of sodium carbonate (10 mmol) and the
mixture was stirred under reflux for 2-4 h until complete hydrolysis of the ester. MeOH was then
evaporated and the reaction was neutralized with aqueous solution of 1M HCl until pH 6. The
liberated free acid was extracted with a mixture of EtOAc and THF, and the extract was dried over
sodium sulfate and evaporated under reduced pressure. The product required no further purification.
Compound 22b was prepared directly from 6-amino-1,3-dimethylquinolin-2-(1H)-one and methyl
4-chloro-3-chlorosulfonylbenzoate using the procedure described in 4.1.4.
Detailed characterization data of the compounds are provided in the supporting information.
4.1.6. General method for the of synthesis of hydroxamic acids 17a-b, 20a-b, 23a-b, 37, 44a-b and
51a-b
The appropriate carboxylic acid 16a-b, 19a-b, 22a-b, 36, 43a-b and 50a-b (1 mmol) was dissolved
in dry THF, followed by the addition of PyBOP (1.5 mmol) and DIPEA (3 mmol). The mixture was
stirred for 15 min, then NH OTHP (1.2 mmol) was added and the reaction mixture was stirred at
2
room temperature for 2-4 h. The solvent was evaporated under vacuum and the mixture was
dissolved in EtOAc (50 ml) and washed with 1M sodium carbonate solution and brine. The organic
layer was evaporated under vacuum and the product was purified by column chromatography
(
CHCl / MeOH / TEA, 99:0.95:0.05).
3
The obtained product was dissolved in THF and a catalytic amount of diluted HCl was added. The
mixture was stirred at room temperature overnight. The reaction was controlled by TLC. After that
the solvent was evaporated under vacuum and the hydroxamic acid product was purified by column
chromatography (CHCl : MeOH, TEA 95:4.95:0.05).
3
Detailed characterization data of the compounds are provided in the supporting information.
4.1.7. Synthesis of 3-{4-[5-(1,3-dimethyl-2-oxo-1,2-dihydroquinolin-6-ylamino)-5-
oxopentyloxy]benzylamino}-N-hydroxy-4-methoxybenzamide (32)
23

Detailed synthesis of intermediates 25-28 are provided in the supporting information.
.1.7.1. N-(1,3-dimethyl-2-oxo-1,2-dihydroquinolin-6-yl)-5-(4-formylphenoxy)pentanamide (29)
4
O
O
NH
CHO
N
O
5-(4-Formylphenoxy)pentanoic acid (28; 3 mmol.), DMF (one drop) and oxalyl chloride (4 mmol)
were stirred in CH Cl at room temperature for 3 h. The mixture was then added dropwise to a
2
2
solution of 6-amino-1,3-dimethylquinolin-2-(1H)-one (5; 3 mmol) and DIPEA (8 mmol) in CH Cl ,
2
2
and stirring was continued for another 2 h at room temperature. The reaction mixture was washed
with saturated aqueous solution of ammonium chloride and brine. The organic layer was dried over
anhydrous sodium sulfate, filtered, and concentrated in vacuo. The residue was purified using
column chromatography (CHCl : MeOH, 100:0-98:2).
3
+
MS m/z: 393.20 [M+H]
Yield: 74%
1
H NMR (400 MHz, DMSO) δ 10.00 (s, 1H, -CO-NH-Ar), 9.84 (s, 1H, -CHO), 7.91 (d, J = 2.4 Hz,
1H, Ar-H), 7.86 – 7.80 (m, 2H, Ar-H), 7.70 (s, 1H, Ar-H), 7.65 (dd, J = 9.1, 2.4 Hz, 1H, Ar-H),
7
.42 (d, J = 9.1 Hz, 1H, Ar-H), 7.14 – 7.07 (m, 2H, Ar-H), 4.11 (t, J = 5.9 Hz, 2H, -CH -O-), 3.60
2
(
s, 3H, -NCH ), 2.39 (t, J = 6.9 Hz, 2H, -CH CO-), 2.10 (d, J = 1.0 Hz, 3H, Ar-CH ), 1.84 – 1.71
3
2
3
(
m, 4H, -CH -C H -CH -).
2
2
4
2
4.1.7.2. 3-{4-[5-(1,3-Dimethyl-2-oxo-1,2-dihydroquinolin-6-ylamino)-5-oxopentyloxy]-
benzylamino}-4-methoxybenzoic acid (31)
HO
O
N
O
N
O
H
O
O
N
H
A mixture of N-(1,3-dimethyl-2-oxo-1,2-dihydroquinolin-6-yl)-5-(4-formylphenoxy)pentanamide
29; 1.1 mmol), 3-amino-4-methoxybenzoic acid (30; 1 mmol) and TFA (2 mmol) in a (1:1)
(
mixture of THF and EtOAc was cooled to 0 °C, then sodium triacetoxyborohydride (4 mmol) was
added and the reaction mixture was stirred overnight at room temperature. The reaction was
quenched by adding H O, and the pH was adjusted to 5 with aqueous potassium bicarbonate
2
solution. The organic layer was separated and the aqueous layer was extracted with EtOAc. The
organic layers were collected, dried over anhydrous sodium sulfate, filtered, and concentrated in
vacuo. The residue was purified using column chromatography (CHCl : MeOH, 100:0-97:3).
3
-
MS m/z: 542.82 [M-H]
Yield: 81%
1
H NMR (400 MHz, DMSO) δ 12.26 (s, 1H, -COOH), 9.98 (s, 1H, -CO-NH-Ar), 7.91 (d, J = 2.2
Hz, 1H, Ar-H), 7.71 (s, 1H, Ar-H), 7.65 (dd, J = 9.1, 2.3 Hz, 1H, Ar-H), 7.42 (d, J = 9.1 Hz, 1H,
Ar-H), 7.25-7.16 (m, 3H, Ar-H), 6.95 (d, J = 2.0 Hz, 1H, Ar-H), 6.89-6.82 (m, 3H, Ar-H), 5.59 (t, J
=
6.0 Hz, 1H, -CH -NH-Ar), 4.24 (d, J = 5.6 Hz, 2H, -CH -NH-Ar), 3.98-3.91 (m, 2H, -CH -O-),
2 2 2
3
.85 (s, 3H, -OCH ), 3.60 (s, 3H, -NCH ), 2.40-2.34 (m, 2H, -CH CO-), 2.10 (3H, Ar-CH ), 1.74
3 3 2 3
(
m, 4H, -CH -C H -CH -).
2 2 4 2
4.1.7.3. 3-{4-[5-(1,3-Dimethyl-2-oxo-1,2-dihydroquinolin-6-ylamino)-5-oxopentyloxy]-
benzylamino}-N-hydroxy-4-methoxybenzamide (32)
24

This compound was prepared from 3-{4-[5-(1,3-dimethyl-2-oxo-1,2-dihydroquinolin-6-ylamino)-5-
oxopentyloxy]-benzylamino}-4-methoxybenzoic acid (31) according to the method described in
4.1.6.
Detailed characterization data of the compound are provided in the supporting information.
4.1.8. General method for the synthesis of masked hydroxamic acids 59a-b and 60a-b
The appropriate carboxylic acids 23a or 44b (1 mmol.) were dissolved in dry THF, followed by the
addition of HATU (1.2 mmol) and DIPEA (3 mmol). The mixture was stirred for 15 min, then the
appropriate O-substitutedhydroxylamine 58a-b (1.2 mmol) was added and the reaction mixture was
stirred at room temperature for 2 h. The solvent was evaporated under vacuum and the mixture was
dissolved in EtOAc and washed with aqueous sodium bicarbonate and ammonium chloride
solutions then brine. The organic layer was separated, dried over anhydrous sodium sulfate, filtered
and concentrated in vacuo. The residue was purified by column chromatography (CHCl : MeOH,
3
100:0-99:1).
Detailed characterization data of the compounds are provided in the supporting information.
4.2. In vitro HDAC inhibitory activity
Recombinant human HDAC1 and -6 were purchased from BPS Biosciences. Enzyme inhibition was
determined by using a published homogenous fluorescence assay [97]. The enzymes were incubated
for 90 min at 37 °C, with the fluorogenic substrate ZMAL (Z-(Ac)Lys-AMC) in a concentration of
1
0.5 ꢀM and increasing concentrations of inhibitors. Fluorescence intensity was measured at an
excitation wavelength of 390 nm and an emission wavelength of 460 nm in a microtiter plate reader
BMG Polarstar).
(
Recombinant hHDAC8 was produced by Romier et al. in Strasbourg [60]. The HDAC8 activity
assay was performed according to the commercial HDAC8 Fluorometric Drug Discovery Kit [Fluor
de Lys®-HDAC8, BML-KI178] corresponding to the manufacturer´s instructions as described
earlier [98]. The substrate was a synthesized tetrapeptide bound to aminomethylecoumarine (AMC)
H N-Arg-His-Lys(Ac)-Lys(Ac)-AMC. The enzyme was incubated for 90 min at 37 °C, with a
2
substrate concentration of 50 ꢀM and increasing concentrations of inhibitors. The Stop-solution
containing Trichostatin A (TSA), to stop the hHDAC8 activity, and Trypsin, to release the AMC,
was added. The solution was incubated for 20 min at 37 °C to develop the assay. Fluorescence
intensity was measured at an excitation wavelength of 355 nm and an emission wavelength of 460
nm in a microtiter plate reader (BMG Polarstar). Detailed synthesis and characterization of the
substrate are provided in the supporting information.
4.3. Isothermal titration calorimetry (ITC)
ITC experiments were carried out and processed as described previously [99-101] using Microcal
Origin for experiment setups, NITPIC [102, 103] and SEDPHAT [104] for data evaluation, and
GUSSI [105] for plotting. Ligand concentrations were between 20 and 40 ꢀM in the sample cell and
BRPF1 concentrations between 200 and 400 ꢀM in the injection syringe. Images and calculations
are provided in the supporting information. For the final K BRPF1 value the P99 interval was
d
determined.
4
4
.4. Cellular assays
.4.1. Viability of human HEK293 cells
HEK293 cells (DSMZ Braunschweig, ACC305) were incubated at 37 °C in a humidified incubator
with 5% CO in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% FCS and
2
3
5
mM glutamine. Cells were seeded at 1.5 x 10 cells per well in a 96-well cell culture plate (TPP,
25

Switzerland). The compounds were added immediately to the medium at 50 µM to determine the
®
percentage viability. After 24 h, AlamarBlue reagent (Invitrogen, CA, USA) was added according
to the manufacturer’s instructions and incubated again for 21 h before samples were analyzed.
Detection of viable cells that convert the resazurin reagent into the highly fluorescent resorufin was
performed by using a FLUOstarOPTIMA microplate reader (BMG Labtec) with the following filter
set: Ex 560 nm/Em 590 nm. All measurements were performed in triplicates over three independent
experiments.
4.4.2. Antiproliferative activity on acute myeloid leukemia cells THP-1 and HL60
HL60 or THP-1 cells were diluted in 96-well tissue culture plates at a density of 5000 per well and
immediately incubated with inhibitors or DMSO vehicle to a total volume of 100 µL. Compounds
were diluted from 200x stock solutions in DMSO and were compared to DMSO vehicle only. Three
replicates per concentration were used. Growth inhibition was determined using the CellTiter
96AQueous Non-Radioactive Cell Proliferation Assay according to the manufacturer’s instructions.
Data was plotted as absorbance units against compound concentration for pretest results or against
logarithm of compound concentration using OriginPro 9. 50% Growth inhibition (GI ) was
5
0
determined as compound concentration required to reduce the number of metabolically active cells
by 50% compared to DMSO control. The assay was already described for MCF7 [106].
4.4.3. Western blot assay in THP-1 and HL60
5
HL60 or THP-1 cells were seeded in a 12-well plate at a density of 2.5 × 10 cells per well, and
immediately incubated with different concentrations of test compounds for 4 h. After incubation,
cells were collected in Eppendorf tubes and centrifuged with 500g for 5 min at room temperature.
Cells were washed with phosphate-buffered saline and lysed in 90 ꢀL of sodium dodecyl sulfate
(SDS) sample buffer (cell signaling, 62.5 mM Tris hydrochloride (pH 6.8 at 25 °C), 2% w/v SDS,
10% v/v glycerol, 50 mM dithiothreitol, 0.01% bromophenol blue). After sonicating for 5 min, to
shear DNA and reduce sample viscosity, the samples were heated to 95 °C for 1 min. Cell extracts
were used directly for SDS polyacrylamide gel electrophoresis (PAGE) or kept frozen at −20 °C
until usage. For the SDS-PAGE, an amount of 7 ꢀL of cell extracts was loaded onto a 12.5% SDS
gel and run at 160 V followed by the transfer to a nitrocellulose membrane via western blotting for
antibody-based detection. After transfer, the nonspecific binding was blocked by incubating the
membrane in 25 mL of blocking buffer [5% nonfat dry milk in Tris buffered saline with 0.1%
Tween 20 (TBS-T)] for 1 h at room temperature or at 4 °C overnight. After washing the membrane
three times for 5 min with TBS-T, the primary antibody [antiacetylated α-tubulin (Sigma-Aldrich
T7451-200UL, 1:1000)] was added in 3% milk in TBS-T for 3 h at room temperature or overnight
at 4 °C. Before exposing the membrane with the secondary antibody, it was washed again three
times for 5 min with TBS-T to remove an unbound primary antibody. The secondary antibody
antimouse-IgG-HRP (Sigma-Aldrich, 1:2000) was added in 3% milk in TBS-T at room temperature
for 1 h. Afterward, the membrane was washed again. The detection was performed via enhanced
chemiluminescence (ECL Prime) after incubation for 5 min in the dark with a FUSION-SL
(PEQLAB) and the FUSION-CAPT software. After detection of acetylated tubulin, the whole
procedure was repeated with the primary antibody antiacetyl-histone H3 (Millipore 06.599, 1:2000)
and the secondary antibody anti-rabbit IgG-HRP (Sigma-Aldrich, 1:5000) to detect the acetylation
of histone H3 and again with the primary antibody antiGAPDH (Sigma-Aldrich 69545-200UL,
1:5000) and the secondary antibody antirabbit IgGHRP (Sigma-Aldrich, 1:10 000) to control the
loading amount.
4.5. Docking and molecular dynamics simulations
26

4.5.1. Docking into HDAC8
The crystal structure of an HDAC8/inhibitor complex (PDB ID 2V5X) was downloaded from
Protein Data Bank [107] and only chain A was retained. Protein preparation was done using
Schrödinger's Protein Preparation Wizard (Schrödinger Suite 2017-1: Protein Preparation Wizard;
Epik, Schrödinger, LLC, New York, NY, 2016; Impact, Schrödinger, LLC, New York, NY, 2016;
Prime, Schrödinger, LLC, New York, NY, 2017) by adding hydrogen atoms, assigning protonation
states and minimizing the protein. Solvent molecules were removed except for two conserved water
molecules: HOH2061 and HOH2152. Hydrogen bond networks, amino acid residues protonation
states and tautomers were optimized, and the complex was finally subjected to energy minimization
using the OPLS-2005 force field using the default settings. Receptor grids were generated using
default settings by assigning the cocrystallized ligand as the center of the grid.
Ligands were first prepared using LigPrep (Schrödinger Release 2017-1: LigPrep, Schrödinger,
LLC, New York, NY, 2017) were the hydroxamic acids were kept in the neutral forms and energy
minimized using the OPLS2005 force field. Twenty conformers for each ligand were subsequently
generated using ConfGen (Schrödinger Release 2017-1: ConfGen, Schrödinger, LLC, New York, NY,
2
017).
Molecular docking was performed using Glide (Schrödinger Release 2017-1: Glide, Schrödinger,
LLC, New York, NY, 2017) in the Standard Precision mode. Default settings without any
constraints were used, the number of docking poses for post-docking minimization per ligand was
increased to 20 and the maximal number of output poses per ligand was set to 2.
4.5.2. Docking into HDAC6
The crystal structure HDAC6/trichostatin A complex (PDB ID 5EDU) was downloaded from the
Protein Data Bank and only chain A was kept. Since the crystal structure was resolved without
water molecules in the binding site, we chose to insert the crucial water molecules by superposing
with the crystal structures of the homologous DrHDAC6 in complex with HPOB (PDB ID 5EF7).
Altogether, four water molecules (HOH 921, 990, 999, and 1011) together with the cocrystallized
ligand HPOP were taken from chain A of the crystal structure of DrHDAC6 (PDB ID 5EF7).
Further protein preparation was performed as described for HDAC8; the ligand was however kept
in the deprotonated hydroxamate form. Ligands were generated in the hydroxamate form and
further prepared using Ligprep and ConfGen as previously described.
Molecular docking was performed using Glide in the Standard Precision mode using the same
settings as with HDAC8.
4.5.3. Docking into BRPF1
The crystal structure BRPF1/NI-57 complex (PDB ID 5MYG) was downloaded from the Protein
Data Bank, and only chain C together with the ligand and four water molecules (HOH 906, 907,
910, and 911) were kept. The complex was prepared using Schrödinger's Protein Preparation
Wizard as described for HDAC8. Ligands were prepared using LigPrep and 20 conformers were
generated using ConfGen as previously described. Docking was finally performed using Glide in
the Standard Precision mode using the same setting as described for HDAC8. The applied setting
was able to reproduce the binding mode of NI-57 in BRPF1 (RMSD < 1 Å).
4.5.4. Molecular dynamics simulation
Initial coordinates of the BRPF1/ligand complexes were taken from the obtained docking poses
described above. MD simulations were performed using Amber16 package (AMBER 2016, UCSF,
27

San Francisco, California, USA, 2016). Force field parameters for the ligands were assigned using
the Antechamber package and AM1-BCC atomic charges [108, 109]. The complex structure was
combined using the TLeap module, where the ff14SB force field [110, 111] and the General Amber
Force Field (GAFF2) [112] were used for the parameterization of protein residues and ligand,
respectively. The complex structures were solvated in an octahedral periodic box of SPC/E water
+
molecules [113] at a margin of 10 Å and the system was neutralized using Na counter ions.
The system was first subjected to two steps of minimization: The first step involved the
minimization of only the solvent atoms in 3000 iterations (first 1000 steepest descent and then 2000
conjugate gradient), while restraining the protein, ligand and conserved water molecules to their
-1
-2
initial coordinates with a force constant of 10 kcal*mol *Å . In the second step, the whole system
was minimized with no restraints using 4000 iterations (first 2000 steepest descent and then 2000
conjugate gradient). The system was subsequently heated to the production temperature (300 K)
through 100 ps of MD simulation, while keeping the complex atoms (protein, ligand and conserved
-1
-2
water molecules) restrained with a force constant of 10 kcal*mol *Å . Constant volume periodic
boundary was set to equilibrate the temperature of the system by Langevin thermostat using a
-1
collision frequency of 2 ps . The system was subsequently subjected to a pressure equilibration
routine for 100 ps at 300 K, where a constant pressure of 1 bar was applied.
Finally, a production run of 100 ns with a time step of 2 fs was simulated at a constant temperature
-1
of 300 K using Langevin thermostat with a collision frequency of 2 ps . Constant pressure periodic
boundary was used to maintain the pressure of the system at 1 bar using isotropic position scaling
with a relaxation time of 2 ps. A non-bonded cut-off distance of 10.0 Å for long-range electrostatic
interactions was used by applying the Particle Mesh Ewald (PME) [114] method during the
temperature equilibration and MD routines. The SHAKE algorithm [115] was applied to constrain
all bonds involving hydrogen. All simulations were run using PMEMD.cuda implementation in
Amber16 on CUDA-enabled NVIDIA graphics processing units (GPUs).
RMSD analysis of the trajectories was performed using the CPPTRAJ module of Amber16. Plots
were generated using the R package, and the videos using PYMOL.
4.5.5. PAINS filter.
All the herein described compounds were filtered for pan-assay interference compounds
PAINS)[116]. For this purpose, PAINS1, PAINS2 and PAINS3 filters, as implemented in
(
Schrödinger’s Canvas program, were employed. None of the compounds was flagged as PAINS.
Author contribution
Ehab Ghazy: Synthesis of all inhibitors, preparation of manuscript.
Patrik Zeyen: Carried out HDAC8 in vitro assays.
Daniel Herp: Carried out HDAC1 and 6 in vitro assays.
Karin Schmidtkunz: AML cellular assays and western blot.
Martin Hügle: Produced the recombinant BRPF1 protein and ran the ITC assay
Frank Erdmann: Carried out cytotoxicity tests on HEK293 cells.
Dina Robaa: Did all modelling studies, preparation of manuscript.
Matthias Schmidt: Did compound analytics, supervision of organic synthesis.
Elizabeth R. Morales: Produced the recombinant HDAC8 protein for in vitro testing.
Christophe Romier: Coordination of protein expression and preparation of manuscript.
28

Stefan Günther: Coordination of ITC assays, preparation of manuscript.
Manfred Jung: Coordination of HDAC1&6, AML cellular assays and western blot, preparation of
manuscript.
Wolfgang Sippl: Study design, research planning, preparation of manuscript.
Acknowledgment
E.G. received his scholarship through the GERLS program from the German Academic Exchange
Service (DAAD) and Ministry of higher education, Egypt. W.S. and M.J. acknowledge the funding
from Deutsche Forschungsgemeinschaft (DFG Grants Ju 295/13-1, Si 846/13-1)). This work was
supported by the European Regional Development Fund of the European Commission.
Abbreviations
HDAC
BRD
Histone Deacetylase
Bromodomain
PTMs
DNA
HAT
Post-Translational Modification
Deoxyribonucleic Acid
Histone Acetyltransferase
BRPF
FDA
HSP 90
BET
Bromodomain and PHD Finger-Containing Protein
U.S. Food and Drug Administration
Heat Shock Protein 90
Bromodomain and Extra-Terminal Motif
Protein Data Bank
PDB
RMSD
MD
Root-Mean-Square Deviation
Molecular Dynamics
AML
GAPDH
Ac
Acute Myeloid Leukemia
Glyceraldehyde 3-Phosphate Dehydrogenase
Acetylated
DMSO
TFA
Dimethyl Sulfoxide
Trifluoroacetic Acid
PyBOP
HATU
HPLC
UV
Benzotriazol-1-Yl-Oxytripyrrolidinophosphonium Hexafluorophosphate
Hexafluorophosphate Azabenzotriazole Tetramethyl Uronium
High-Performance Liquid Chromatography
Ultraviolet
ESI-MS
Electrospray Ionization Mass Spectrometry
HRMS-ESI High Resolution Mass Spectroscopy- Electrospray Ionization
EtOAc
DMF
MeOH
THF
Ethyl Acetate
Dimethylformamide
Methanol
Tetrahydrofuran
Chloroform
CHCl3
H O
2
Water
CH Cl2
Dichloromethane
N,N-Diisopropylethylamine
Hydrochloric Acid
Triethylamine
Hour
2
DIPEA
HCl
TEA
h
29