Discovery of in Vivo Chemical Probes for Treating Alzheimer's Disease: Dual Phosphodiesterase 5 (PDE5) and Class i Histone Deacetylase Selective Inhibitors

Article abstract of DOI:10.1021/acschemneuro.8b00648

In order to determine the contributions of histone deacetylase (HDAC) isoforms to the beneficial effects of dual phosphodiesterase 5 (PDE5) and pan-HDAC inhibitors on in vivo models of Alzheimer's disease (AD), we have designed, synthesized, and tested novel chemical probes with the desired target compound profile of PDE5 and class I HDAC selective inhibitors. Compared to previous hydroxamate-based series, these molecules exhibit longer residence times on HDACs. In this scenario, shorter or longer preincubation times may have a significant impact on the IC₅₀ values of these compounds and therefore on their corresponding selectivity profiles on the different HDAC isoforms. On the other hand, different chemical series have been explored and, as expected, some pairwise comparisons show a clear impact of the scaffold on biological responses (e.g., 35a vs 40a). The lead identification process led to compound 29a, which shows an adequate ADME-Tox profile and in vivo target engagement (histone acetylation and cAMP/cGMP response element-binding (CREB) phosphorylation) in the central nervous system (CNS), suggesting that this compound represents an optimized chemical probe; thus, 29a has been assayed in a mouse model of AD (Tg2576).

Full text of DOI:10.1021/acschemneuro.8b00648

Subscriber access provided by YORK UNIV
Article
Discovery of in-vivo chemical probes for treating
Alzheimer´s disease: Dual phosphodiesterase 5 (PDE5)
and class I histone deacetylases-selective inhibitors.
Obdulia Rabal, Juan A Sánchez-Arias, Mar Cuadrado-Tejedor, Irene De Miguel, Marta Pérez-
González, Carolina García-Barroso, Ana Ugarte, Ander Estella, Elena Sáez, Maria Espelosin,
Susana Ursua, Haizhong Tan, Wei Wu, Musheng Xu, Ana Garcia-Osta, and Julen Oyarzabal
ACS Chem. Neurosci., Just Accepted Manuscript • DOI: 10.1021/acschemneuro.8b00648 • Publication Date (Web): 10 Dec 2018
Downloaded from http://pubs.acs.org on December 16, 2018
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a service to the research community to expedite the dissemination
of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in
full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully
peer reviewed, but should not be considered the official version of record. They are citable by the
Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore,
the “Just Accepted” Web site may not include all articles that will be published in the journal. After
a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web
site and published as an ASAP article. Note that technical editing may introduce minor changes
to the manuscript text and/or graphics which could affect content, and all legal disclaimers and
ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or
consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W.,
Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the
course of their duties.

Page 1 of 66
ACS Chemical Neuroscience
1
2
3
4
5
6
7
8
9
Discovery of in-vivo chemical probes for treating Alzheimer´s disease:
Dual phosphodiesterase 5 (PDE5) and class I histone deacetylases-
selective inhibitors.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
,5
1,5
2,3,5
Irene de Miguel,1
Obdulia Rabal, Juan A. Sánchez-Arias, Mar Cuadrado-Tejedor,
2
2
1
Marta Pérez-González, Carolina García-Barroso, Ana Ugarte, Ander Estella-Hermoso
1
1
2
2
4
4
de Mendoza, Elena Sáez, Maria Espelosin, Susana Ursua, Tan Haizhong, Wu Wei,
4
2,5
1,5,*
Xu Musheng, Ana Garcia-Osta, and Julen Oyarzabal.
1
2
Small Molecule Discovery Platform, Molecular Therapeutics Program, Neurobiology
of Alzheimer’s Disease, Neurosciences Division, Center for Applied Medical Research
(CIMA), University of Navarra, Avenida Pio XII 55, E-31008 Pamplona, Spain
3Anatomy Department, School of Medicine, University of Navarra, Irunlarrea 1, E-31008
Pamplona, Spain
4WuXi Apptec (Tianjin) Co. Ltd., TEDA, No. 111 HuangHai Road, 4th Avenue, Tianjin
3
00456, PR China
5These authors contributed equally to this work.
ABSTRACT
In order to determine the contributions of histone deacetylase (HDAC) isoforms to the
beneficial effects of dual PDE5 and pan-HDAC inhibitors on in vivo models of
Alzheimer’s disease (AD), we have designed, synthesized and tested novel chemical
probes with the desired target compound profile of phosphodiesterase 5 (PDE5)- and class
I HDAC-selective inhibitors. Compared to previous hydroxamate-based series, these
molecules exhibit longer residence times on HDACs. In this scenario, shorter or longer
pre-incubation times may have a significant impact on the IC values of these compounds
5
0
1
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 2 of 66
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
and therefore on their corresponding selectivity profiles on the different HDAC isoforms.
On the other hand, different chemical series have been explored and, as expected, some
pairwise comparisons show a clear impact of the scaffold on biological responses (e.g.,
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
5a vs 40a). The lead identification process led to compound 29a, which shows an
adequate ADME-Tox profile and in vivo target engagement (histone acetylation and
cAMP/cGMP response element-binding (CREB) phosphorylation) in the central nervous
system (CNS), suggesting that this compound represents an optimized chemical probe;
thus, 29a has been assayed in a mouse model of AD (Tg2576).
Keywords -- Chemical probes, Dual inhibitors, PDE5 inhibition, class I HDAC-selective
inhibition, Alzheimer, In-vivo test
INTRODUCTION
Inhibitors of histone deacetylases (HDACs) and phosphodiesterases (PDEs) have both
been identified as alternative treatments for Alzheimer’s disease (AD) [1–7]. As strategies
targeting multiple mechanisms are the best option to treat multifactorial diseases such as
AD [8,9], we validated the in vivo efficacy of a new systems therapeutics approach based
on the dual inhibition of HDACs and PDE5, either using a combination of single
therapeutic agents (for example PDE5 inhibitors 1-3 and HDAC inhibitors 4-12, Chart 1;
confirmed with 3 and 4) [10] or with our proprietary first-in-class dual inhibitor 13 (CM-
4
14) [11,12], Chart 2. PDE5 inhibitors increase cGMP concentrations, which might
ultimately promote memory-related gene transcription by directly and/or indirectly
activating binding to the cAMP response element (CREB) [13] and decrease levels of
phosphorylated Tau (pTau) by favouring the inactive form of GSK3 [14,15]. Treatment
of different animal models with specific PDE5 inhibitors (1 sildenafil, 2 vardenafil and 3
tadalafil, Chart 1) reverses cognitive decline and/or enhances synaptic plasticity [14–16].
2
ACS Paragon Plus Environment

Page 3 of 66
ACS Chemical Neuroscience
1
2
3
4
5
6
7
8
9
HDACs are “erasers” that remove acetyl groups from lysine residues in histone and other
non-histone substrates and trigger the transcription of transcriptionally silent chromatin.
Eighteen HDACs have been identified and are divided into four groups according to
phylogenetic sequence and function: class I (HDACs 1, 2, 3, and 8), class IIa (HDACs 4,
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
5
, 7, and 9), class IIb (HDACs 6 and 10) and class IV (HDAC11). Pan-HDAC inhibitors,
such as compound 4 (SAHA) [17] and ortho-amino anilides predominantly targeting class
I HDAC enzymes (6 entinostat [18], 7 CI-994 [19], 8 RGFP136 [20], 10 BRD4884 [21]),
and other HDAC2-selective benzamides [22] have been used to improve cognitive
dysfunction. Class I HDACs are ubiquitously expressed, display a predominantly nuclear
localization, and function as transcriptional co-repressors of CREB-regulated genes that
are important for learning and memory [17,23]. However, a persistent missing link is the
identification of the specific HDAC(s) responsible for mediating the pro-cognitive effects
[24]. Here, we pursued a strategy to develop first-in-class dual PDE5- and class I HDAC-
selective inhibitors that are able to cross the blood-brain barrier (BBB) as chemical probes
to be tested in a mouse model of AD. Using a similar design strategy as reported for our
previous hydroxamic derivatives, such as compound 13 [12,25,26], in the present study,
ortho-aminoanilides (also referred as benzamides) were incorporated as the zinc-binding
group (ZBG) into the appropriate growing vector (guided by structural information) of
scaffolds borne by known PDE5 inhibitors to confer them with the additional ability to
bind HDACs. This ZBG confers great selectivity for class I HDACs [27] and tends to
exhibit longer residence times than hydroxamate-based HDAC inhibitors, a characteristic
associated with the time-dependent responses of histone acetylation, cell viability and
gene expression [28].
3
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 4 of 66
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
O
N
O
O
H
N
O
S
HN
N
O
S
HN
N
N
N
N
N
N
O
O
O
O
N
O
O
N
N
N
O
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
Sildenafil
Vardenafil
Tadalafil
IC50 = 8.5 nM (vs PDE5A)
IC50 = 0.89 nM (vs PDE5A)
IC50 = 9.4 nM (vs PDE5A)
O H
HN
O
N
H
O H
N
N
N
N
O
O
O
NH
O
N
H
NH
NH2
H
N
N
O
4
5
6
SAHA
Quisinostat
Entinostat
IC50 = 30 nM (vs HDAC1)
IC50 = 170 nM (vs HDAC2)
IC50 = 100 nM (vs HDAC3)
IC50 = 38 nM (vs HDAC6)
IC50 = 0.11 nM (vs HDAC1)
IC50 = 0.33 nM (vs HDAC2)
IC50 = 4.86 nM (vs HDAC3)
IC50 = 76.8 nM (vs HDAC6)
IC50 = 243 nM (vs HDAC1)
IC50 = 453 nM (vs HDAC2)
IC50 = 248 nM (vs HDAC3)
IC50 > 10000 nM (vs HDAC6)
O
NH
F
O
H
N
O
H
N
O
N
H
NH
NH2
NH2
O
NH2
7
8
9
CI-994 / Tacedinaline
RGFP136
IC50 = 900 nM (vs HDAC1)
IC50 = 900 nM (vs HDAC2)
IC50 = 1300 nM (vs HDAC3)
IC50 = 1140 nM (vs HDAC1)
IC50 = 560 nM (vs HDAC3)
IC50 = 900 nM (vs HDAC2)
O
O
O
H
N
NH
O
N
H
NH2
H
N
HN
O
NH2
O
S
N
O H
O
O
1
0
11
1
2
F
BRD4884
IC50 = 29 nM (vs HDAC1)
IC50 = 62 nM (vs HDAC2)
IC50 = 1009 nM (vs HDAC3)
IC50 = 13 nM (vs HDAC1)
IC50 = 115 nM (vs HDAC2)
IC50 = 22450 nM (vs HDAC3)
IC50 > 50000 nM (vs HDAC6)
HDAC2 K < 100 nM
d
4
ACS Paragon Plus Environment

Page 5 of 66
ACS Chemical Neuroscience
1
2
3
4
5
6
7
8
9
Chart 1. Chemical structures of PDE5 inhibitors (compounds 1-3) and HDAC inhibitors
(
compounds 4-12). The IC values against PDE5 for compounds 1, 2 and 3 have been
50
reported [29]. HDAC IC values for compounds 4 [30], 5 [31], 6 [28], 7 [32], 8 [33], 9
5
0
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
[34], 10 [21], 11 [30] and 12 [35] have been also described in previous studies.
O
N
O
HN
N
IC50 = 310 nM (vs HDAC1)
IC50 = 490 nM (vs HDAC2)
IC50 = 322 nM (vs HDAC3)
IC50 = 91 nM (vs HDAC6)
IC50 = 60 nM (vs PDE5A)
N
1
3
CM-414
O
HN
O H
Chart 2. Previously reported data for lead compound 13 (CM-414) [11,12].
RESULTS
Rationale and Design
The well-established pharmacophore for HDACi comprises a hydrophobic capping group
that tightly binds to the rim of the catalytic tunnel (also known as the recognition group),
a hydrophobic linker spanning the 11 Å catalytic channel and a ZBG. In our dual
inhibitors, the PDE5 inhibitor core serves as the recognition group. Here, we envisioned
the attachment of ortho-aminoanilides to the piperazinylsulfonamide group of the
sildenafil (compound 18a) or vardenafil (compound 23a) cores through a pyrimidyl
linker, similar to compound 5, which is well known for conferring class I selectivity
[12,25]. Designed compounds 18a and 23a were docked into the HDAC2 crystal structure
complexed with N-(4-aminobiphenyl-3-yl)benzamide inhibitor 9 (PDB entry 3MAX
5
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 6 of 66
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
[
34], Figure 1A). As shown in Figure 1A, the ortho-amidoanilide group of compounds
1
8a and 23a is predicted to chelate the Zn metal in a bidentate manner, with both the
amide oxygen and the amino nitrogen interacting with the Zn atom. Furthermore, both
structures superimpose well with the corresponding co-crystallized ligand 9 (pink),
particularly regarding the overlap of the pyrimidyl moiety of compounds 18a and 23a
with the amide-bonded phenyl ring of compound 9, at the gate of the 11 Å catalytic
channel. The other para-phenyl ring of compound 9 occupies the 14 Å long internal cavity
adjacent to the Zn binding site that is characteristic of HDAC1 and HDAC2 [28,34,36]
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(
also known as the foot pocket) and is probably involved in the release of the acetate
product and ligands [37]. Here, the sildenafil and vardenafil cores of compounds 18a and
3a, respectively, project beyond the 11 Å channel and are oriented towards the loop
2
comprising Tyr206-Tyr209 (HDAC2) and Tyr201-Tyr204 (HDAC1). Potent inhibitors
of class I HDACs tend to interact through their capping groups with the L1 loop (Tyr28-
Pro37 in HDAC2, Tyr23-Pro32 in HDAC1) [38], as we have previously reported for our
therapeutic tool 13 [11]. Intrigued by the different predicted positions of the capping
groups in these two compounds, and in an attempt to assess the quality of our predictions,
we reviewed all the HDAC crystals complexed with a benzamide inhibitor. However, to
date, no inhibitor of this chemical family of benzamides contains a bulky capping group
(e.g., compound 9 lacks this group), and the largest capping group was observed for 4-
acetamido-N-[2-amino-5-(2-thienyl)phenyl]benzamide crystallized with HDAC2 (PDB
entry 4LY1) [28]. As shown in Figure S1, the orientation of the “capping” group overlaps
well with our designed inhibitors (and in opposition to compound 4 crystallized into
HDAC2, which projects towards the L1 loop). Taken together, the conformation of the
benzamide moiety, which is imposed by the chelation geometry, seems to be crucial for
orienting the capping group of this benzamide-based family of inhibitors opposite to the
6
ACS Paragon Plus Environment

Page 7 of 66
ACS Chemical Neuroscience
1
2
3
4
5
6
7
8
9
L1 loop. Concerning HDAC6 selectivity, an analysis of crystal structures of HDAC6 in
complex with several inhibitors [38,39] confirms the hypothesis that the foot-pocket is
unusable or closed in class II HDACs [40], establishing the basis for the selectivity of
benzamides for class I HDACs over HDAC6. Finally, for PDE5, para derivatization of
the ethoxyphenyl group of 1 and 2 projects towards the solvent, thereby predictably
accommodating proposed substitutions, as shown in Figure 1B for the structure of
compound 18a docked into the PDE5 cavity.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 1. (A) Docking 18a (green) and 23a (orange) into the three-dimensional structure
of HDAC2 (PDB entry 3MAX). The co-crystallized benzamide 9 of 3MAX is shown in
pink sticks, with the para-phenyl ring occupying the 14 Å cavity characteristic of HDAC1
and HDAC2 [36]. (B) Docking 18a (green) into the PDE5 binding site (PDB entry 1TBF)
[41].
SAR and Biochemical profiling
The synthesis of both compounds 18a and 23a is reported in the CHEMISTRY section
(Schemes 1 and 2, respectively). Both compounds were potent, low nanomolar PDE5
7
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 8 of 66
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
inhibitors, with the vardenafil-based derivative 23a being a log unit more potent than its
corresponding matched pair 18a (IC values of 4 and 41 nM, respectively, Table 1), and
5
0
this finding was consistent with the expected trend for both chemical series [25].
Concerning HDACs, compounds were routinely screened versus HDAC1, HDAC2 and
HDAC6 isoforms for various pre-incubation times, 30 min (standard period used for
hydroxamic derivatives) [12,25,26] and 4 hours (HDAC1 and HDAC2), as an accurate
determination of IC values for extremely slow HDAC binders remains challenging [28].
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
5
0
Compound 23a exhibited time-dependent HDAC2 inhibition, with increasing inhibition
observed when the pre-incubation time was prolonged from 30 minutes to 4 hours (the
IC decreased from 2620 nM to 467 nM, Table 1). Then, compound 23a achieved mid-
5
0
nanomolar class I HDAC inhibition, with excellent selectivity over HDAC6 (IC > 20
5
0
µM). Strikingly, its sildenafil-based matched pair 18a was a weaker HDAC1 inhibitor
and had no effect on HDAC2 activity at the assayed conditions. A set of close analogues
was prepared to examine SAR (Table 1). Other sulfamoyl-linked analogues listed in
Table 1 (compounds 18b, 23b, 18c and 23c, synthesis shown in Schemes 1 to 2) did not
show improved HDAC binding affinities, although we partially confirmed the observed
impact of the pre-incubation time on inhibitor potency and the variability of this response
to different HDAC isoforms (e.g., for compounds 18c and 23c). Replacement of this
linker by a methylene group afforded anilides 29a (phenyl), 29b (thiophene) and 29c-34c
(cyclobutane) (synthesis shown in Scheme 3). The phenyl linker of compound 29a
conferred the highest activity against HDAC1 and HDAC2 in the mid-nanomolar range
(
IC = 180 and 668 nM at 4 h), although PDE5 activity was attenuated (IC = 114 nM).
5
0
5
0
Bioisosteric substitution of the phenyl group of the anilide by a thiophene (derivatives
5a, 40a and 35b, synthesis shown in Scheme 4) retained the HDAC1 and HDAC2
3
inhibitory activity, particularly for the vardenafil analogue 40a. Among all compounds
8
ACS Paragon Plus Environment

Page 9 of 66
ACS Chemical Neuroscience
1
2
3
4
5
6
7
8
9
listed in Table 1, analogue 29a showed a significant time-dependent effect on class I
HDAC inhibition, particularly towards HDAC2. For HDAC1, its inhibitory potency also
increased significantly (~1 log unit) with the pre-incubation time: 673 nM (30 min), 180
nM (4 hours) and 69 nM (6 hours). This finding is consistent with its residence time (347
min, as determined by the jump dilution approach [26]). Given its interesting biochemical
profile, the HDAC3 inhibitory activity of compound 29a was further determined, with
similar findings obtained as for HDAC2; HDAC3 IC > 20 µM and 777 nM after 30 min
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
5
0
and 4 hours of compound pre-incubation, respectively. As expected, all compounds listed
in Table 1 were inactive against HDAC6 (IC > 20 µM), even if the pre-incubation time
5
0
was extended to 4 hours, as assayed for compound 29a.
Table 1. Dual PDE5 and HDAC inhibitors with slow binding kinetics and structures
based on sildenafil and vardenafil scaffolds.
O
N
O
N
N
O
HN
O
HN
N
N
N
R1
R1
A
B
9
ACS Paragon Plus Environment

Cpd
CORE
R1
PDE5A HDAC1 HDAC1 HDAC2 HDAC2
HDAC6
ACS Chemical Neuroscience
Page 10 of 66
IC nM
5
0
a
IC nM IC nM t
=
4
t
=
4 IC nM
50
50
50
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
hours
hours
0
1
2
3
4
5
6
7
8
H2N
1
8a
A
B
A
B
41
4
3300
393
1290
331
>20000
2620
>20000
467
>20000
>20000
>20000
>20000
H
N
O
N
N
O
S
N
N
O
H N
2
9 23a
H
N
O
N
N
0
1
2
3
O
S
N
N
O
H2N
4 18b
5
6
7
O
24
6
>20000
2650
>20000
3280
>20000
>10000
>20000
>10000
H
N
S
O
N
N
H
N
N
O
8
H2N
9 23b
O
H
N
S
O
N
N
0
1
2
3
4
N
H
N
O
H N
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
2
8c
3c
9a
A
B
A
2
49
8
1760
1880
673
7350
3520
180
>20000
>20000
>20000
6420
3580
668
>20000
>20000
>20000
H
N
H
S
N
O
O
O
H N
2
H
N
H
N
S
O
O
O
H N
2
114
H
N
O
1
0
ACS Paragon Plus Environment

Page 11 of 66
ACS Chemical Neuroscience
1
2
3
2
9b
A
H2N
160
3560
1040
>20000
1670
>20000
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
HN
O
S
0
1
2
3
4
5
6
7
8
9
0
1
2
2
3
9c
4c
A
B
A
H2N
74
>20000
7780
>20000
>20000
831
>20000
14200
>20000
4860
>20000
>20000
>14000
H
N
O
H2N
5
H
N
O
3 35a
H2N
1080
4170
>14000
11500
H
N
4
5
6
7
8
9
S
S
O
4
3
0a
5b
B
A
H2N
26
661
97
>20000
12400
406
>20000
>20000
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
H
N
O
H N
498
1850
2390
3100
2
H
N
S
O
Biochemical data in Tables 1-3 is the average of at least two assays performed on different days.
In cases in which pIC difference was greater than 0.3 log units, in absolute terms, additional
assays were carried out until a satisfactory error was obtained (individual results greater than 2
5
0
a
MADs of the average are discarded). Compound 29a was also pre-incubated with HDAC6 for 4
hours, and its IC was constant at a value greater than 20 M.
5
0
In parallel with the sildenafil- and vardenafil-based analogues shown in Table 1, we
examined whether the incorporation of an ortho-amino anilide moiety into the tadalafil
core would result in dual PDE5 and HDAC inhibitors, as previously observed for
hydroxamic-based derivatives [25]. Thus, compounds 50a-50b, with linker moieties
1
1
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 12 of 66
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
connecting the scaffold to the chelating group similar to those in compounds in Table 1,
were designed and synthetized as described in Scheme 5. Despite moderate class I HDAC
inhibitory activity (Table 2), particularly for compound 50b, the poor activity against
PDE5 (e.g., 50b IC > 10 µM) prevented us from further developing this chemical series.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
5
0
Table 2. Dual PDE5 and HDAC inhibitors with slow binding kinetics and a structure
based on the tadalafil scaffold.
O
H
R1
N
N
H
N
H
O
O
O
Cpd R1
PDE5A HDAC HDAC1 HDAC2 HDAC2 HDAC6
1
IC50
nM
IC50
nM
t = 4
hours
t = 4 IC nM
50
IC50
hours
nM
H N
2
5
5
0a
0b
5950
>20000 5810
>20000 12900
>20000
>20000
H
N
N
N
N
O
>10000 2300
516
12600
1930
H N
2
HN
O
N
1
2
ACS Paragon Plus Environment

Page 13 of 66
ACS Chemical Neuroscience
1
2
3
4
5
6
7
8
9
By revisiting the sildenafil analogues and after having identified the benzyl linker of
compound 29a as an optimal linker for class I HDAC activity, we expanded our
exploration to include 2-aminoanilides substitutions in position 5 to produce compounds
that are able to occupy the foot pocket of the active site of class I HDACs (compounds
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
5
2a-52b, Table 3 and synthesis shown in Scheme 6); therefore, we aimed to develop
HDAC1- and HDAC2-selective inhibitors that did not target HDAC3 [21,30] – as
additional chemical probes to determine the contribution of just two class I HDAC
isoforms. Compared to compound 29a, derivatization with a thiophene ring (52a) or a
fluorine atom (52b) was detrimental for PDE5 and HDAC activity (low micromolar
range). Next, we turned our attention to other alternative Zn-chelating groups, such as the
sulphur-bonded 4-hydroxy-1H-pyrimidin-6-one of compound 54 (synthesis in Scheme
7
). Disappointingly, although compound 54 was a potent nanomolar PDE5 inhibitor, only
slight activity was detected towards HDAC6 (IC50 = 7.7 µM) and no activity was
observed towards class I HDACs (IC > 20 M).
5
0
Table 3. SAR exploration of sildenafil-based analogues occupying the class I HDAC
foot pocket or containing alternative zinc-binding groups.
O
N
O
HN
N
N
R1
1
3
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 14 of 66
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Cpd R1
PDE5A HDAC HDAC1 HDAC2 HDAC2 HDAC6
1
IC50
nM
IC50
nM
t = 4
hours
t = 4 IC nM
50
IC50
hours
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
nM
NH2
5
5
2a
2b
3640
>20000 833
>20000 2470
>20000 14500
>20000 9110
>14000
>20000
H
N
O
S
NH2
1860
H
N
O
F
54
O
N
5
40% at 40% at 25% at 25% at 7670
0 µM 20 µM 20 µM 20 µM
HN
2
OH
S
O
Cellular Response: From functional impact on H3K9 acetylation and CREB
phosphorylation To cytotoxicity.
Selected dual inhibitors with promising biochemical profiles were progressed into cellular
assays, including the functional responses in the induction of histone acetylation
(AcH3K9) and CREB phosphorylation (pCREB), determined in SH-SY5Y cells, and
cytotoxicity towards the healthy hepatic cell line THLE-2. As shown in Table 4, our
anilides generally exerted minor effects on THLE-2 cell proliferation (LC > 10 µM, 72
5
0
h incubation) compared with our hydroxamic-based lead compound 13 (LC = 7 M) or
5
0
1
4
ACS Paragon Plus Environment

Page 15 of 66
ACS Chemical Neuroscience
1
2
3
4
5
6
7
8
9
the reference compound 6 (LC50 = 159 nM, transgenic APP/PS1 mice tolerate the
administration of this compound for 10 days[18]). Compound 29a, which was further
progressed to in vivo testing, showed similar low cytotoxicity in THLE-2 cells after 24
and 48 h of incubation (LC = 15.1 and 14.5 µM, respectively) and in PBMCs (LC =
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
5
0
50
1
4 µM at 72 hours and > 100 µM at 24 h), although a greater anti-proliferative effect was
observed on neurons and glial cells from wild type (WT) mice (LC = 4.8 µM after 72
50
hours). Changes in the AcH3K9 mark induced by these molecules were routinely
determined after treatment for 24 hours (optimal read-out), as a delayed induction of
H3K9 acetylation was observed for reference compound 6 (described in Table S1,
supporting information), aligned with the results obtained for other marks (H2BK5 and
H4K12) by Lauffer et al.[28] Values presented in Table 4 are reported as the mean fold
change versus control vehicle-treated cultures, with values greater than 1 illustrating the
induction of acetylation (or phosphorylation, in the case of the pCREB mark). Consistent
with the biochemical profile, compounds 23a and 29a strongly increased the level of this
mark at a dose of 1 µM (5.1 and 2.9 fold change compared with the basal level,
respectively), similar to reference compound 6 (5.1 fold change) under the same
conditions. A dose-response evaluation of the effects of 29a on the acetylation of H3K9
after a 24 h incubation was performed (Figure 2). At concentrations exceeding 26 nM,
compound 29a elicited a significant (1.5-fold change, p value < 0.05) increase in the
AcH3K9 mark, with a maximum response of 7.9-fold at 1 µM. Thus, based on its IC50
values for class I HDACs (leading to low toxicity) and previously reported
data,[10,11,12,25,26] this level of AcH3K9 induced by the compound at concentrations
exceeding 26 nM corroborates the functional synergistic effect induced by the
simultaneous inhibition of PDE5 and HDAC class I enzymes.
1
5
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 16 of 66
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
The level of CREB phosphorylation has little room for improvement and is substantially
biased by the incubation time, as shown in Table 4 for well-known PDE5 inhibitors 1-3.
Here, the incubation time was also required to be extended to 4 hours to achieve a
significant response (greater than 1.4 fold change compared to vehicle-treated cells),[26]
and only compound 29a elicited a significant response: 2.8-fold change at 500 nM
compared with the basal level. A detailed dose-response assay showed that compound
29a increased the level of the pCREB mark by 47% (1.47-fold) at a concentration of 100
nM, with the maximum response of 5.11-fold observed for a 1 M treatment (described
in Table S2, supporting information). Compared to the other more potent PDE5 inhibitors
reported in Table 4 (e.g., 18a, 23a, and 40a), a possible explanation for the gain in CREB
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
phosphorylation by 29a is its high PAMPA permeability (P = 82.9 nm/s).
e
Table 4. Cellular response of prioritized dual PDE5 and class I HDAC inhibitors.
Cpd
THLE-2
LC nM
AcH3K9 levels (fold pCREB levels
PAMPA
change compared with the
(fold
change P (nm/s)
5
0
e
basal level (1))
compared with the
after treatment with 1 basal level (1))
µM for
after treatment with
500 nM for
2
4 hours
4
hours
1
>100000
N.D.
1.9 (30 min)
27.5
0
.6 (2 hours)
1
6
ACS Paragon Plus Environment

Page 17 of 66
ACS Chemical Neuroscience
1
2
3
4
5
6
7
8
9
1
.1 (4 hours)
2
3
>100000
60300
159
N.D.
N.D.
0.9 (30 min)
21.4
26.8
1
2
.2 (2 hours)
.2 (4 hours)
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1.3 (30 min)
1.4 (2 hours)
1
.0 (4 hours)
6
1
1
5.1
N.D.
2.3
5.1
2.9
1.4
1.6
N.D.
2.0
0.5
0.9
2.8
0.7
0.5
0.59
15.7
29.5
10.3
82.9[a]
18.0
21.4
3 [12] 7200
8a
23000
21300
11700
42700
7990
23a
29a
3
4
5a
0a
Quantification of the AcH3K9, AcTub and pCREB levels on Western blots is reported as the
mean value of at least three experiments. Each reported permeability rate corresponds to the mean
value of three assays. Cytotoxicity results correspond to the mean value of at least two tests
performed on different days. If the difference in pLC was greater than 1 log unit, in absolute
terms, new experiments were carried out until a satisfactory error was obtained (results exceeding
MADs of the mean were discarded). N.D. = not determined; [a]the concentration of each
5
0
3
compound was measured by LC-MS (standard approach involves UV spectroscopy, see
methods).
1
7
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 18 of 66
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 2. Levels of H3K9 acetylation were measured in SH-SY5Y cells, by AlphaLisa
approach, after their treatment with 29a for 24 hours (*p ≤0.05, **p≤0.005, ***p≤ 0.001).
Functional response of Tg2576 neurons to compound 29a and ADME properties
Next, we concentrated on compound 29a (Mw = 537 Da, TPSA = 124 Å) as other potent
class I HDAC inhibitor that substantially increased the acetylation of H3K9; compound
2
3a exhibited poorer permeability and a clearly undesirable Mw (673 Da) and TPSA (198
Å). The impact of compound 29a on AD-related markers was determined in Tg2576
neurons using western blot assays. After exposure to compound 29a (50 nM) for 48 hours,
the expression of the Aβ42 precursor 99-amino-acid-long APP-carboxy-terminal
fragments (C99), a marker of hAPP processing, was reduced by 20%. Under the same
assay conditions, the levels of phosphorylated Tau (pTau) were decreased by 32%
compared to untreated cultures.
1
8
ACS Paragon Plus Environment

Page 19 of 66
ACS Chemical Neuroscience
1
2
3
4
5
6
7
8
9
A first in vitro evaluation of the ADME properties of compound 29a was performed. This
compound displayed minimal inhibition, less than 20% at 10 µM, towards five of the
cytochrome P450 isoforms (1A2, 2C19, 2C9, 2D6, and 3A4). Compound 29a exhibited
moderate metabolic stability in human and mouse liver microsomes, with >55% of the
compound remaining after its incubation (at 1 M) with the microsomes for 20 min, and
estimated half-lives of 39.8 min (human) and 55.9 min (mouse). The good permeability
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
obtained using our PAMPA assay (P = 82.9 nm/s) was replicated using CACO-2 cells
e
-6
(
P = 1.4 × 10 cm/s), without efflux (efflux ratio of 0.86). Finally, a patch-clamp
e
electrophysiology assay (using CHO cells with stable expression of hERG potassium
channels) yielded an IC value > 30 μM, suggesting that the compound did not present
5
0
cardiovascular safety issues. Based on these data, compound 29a exhibited a promising
in vitro ADME profile that was only slightly limited by its low solubility (2.14 µg/mL)
and high plasma protein binding activity in human and mouse (0.2% of the unbound
fraction in both species). In terms of phosphodiesterase selectivity, IC50 values of
compound 29a towards PDE9, PDE3A and PDE6 are > 10 µM, > 10 µM and 7.6 nM,
respectively. PDE9 and PDE6 are cGMP-specific enzymes, while PDE3A also
hydrolyses cAMP. PDE3A is involved in cardiac contractility and is associated with
undesired cardiac side-effects [42]; therefore, the lack of activity of compound 29a
towards PDE3A was regarded as a positive feature.
In vivo efficacy tests with 29a: From target engagement To impact on memory
A pharmacokinetic (PK) study of the effects of compound 29a on C57BL/6J mice was
performed after an intraperitoneal (i.p.) injection of a 20 mg/kg dose, and the results
showed that despite its moderate in vitro microsomal stability, compound 29a has a short
half-life in vivo (t = 7.5 h), reaching a C of 934 nM at 15 min (plasma concentrations
1
/2
max
1
9
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 20 of 66
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
are shown in Figure 3A and additional PK parameters are shown in the supporting
information). Nevertheless, compound 29a has a low, but acceptable brain penetration
(LogBB = -0.98; monitored at Tmax) that exceeds the dose required to achieve the
aforementioned functional cellular response (26 nM for AcH3K9 in SH-SY5Y cells and
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
5
0 nM for AD-related biomarkers in primary cultures). The functional in vivo efficacy
and brain accessibility of compound 29a were confirmed by monitoring the levels of the
AcH3K9 and pCREB marks over time after administering this dose to WT mice (n=3).
The level of AcH3K9 increased from 1.4 (30 min) to 1.9 (60 min). For pCREB, a stronger
induction was detected from 1.4 (15 min) to 2.3 (30 min) and 2.4 (60 min). Thus, target
engagement was confirmed in the hippocampus after the administration of compound 29a
(20 mg/Kg, i.p.), as shown in Figure 3B.
The compound (at 20 mg/kg, i.p.) was administered to 16-month-old Tg2576 mice for
two weeks and contextual fear conditioning was assessed to determine if compound 29a
reverses cognitive impairments. As depicted in Figure 3C, 16-month-old Tg2576 mice
showed a significant decrease in freezing behaviour (37.9 ± 9.5 % of mice displayed
freezing behaviours) compared to wild type (WT) mice (71.31±6.2 % of mice displayed
freezing behaviours). Although Tg2576 mice treated with compound 29a showed a partial
restoration of memory function (59.65 ±9.2 % of mice displayed freezing behaviours),
the effect was not significant (p value = 0.17).
2
0
ACS Paragon Plus Environment

Page 21 of 66
ACS Chemical Neuroscience
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 3. (A) Plasma concentrations of 29a after i.p. administration (20 mg/kg) (n=5).
(B) Western blots showing histone 3 acetylation at lys 9 (AcH3K9) and pCREB levels in
the hippocampus of wild type mice 15, 30 min and 1 hour after the intraperitoneal (i.p.)
2
1
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 22 of 66
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
injection of compound 29a (20 mg/kg) or vehicle (n = 3). Actin was used as a loading
control. (C) Freezing behaviours of 16-month-old WT (n=8) or Tg2576 mice treated (i.p.)
with vehicle (n=10) or 20 mg/kg compound 29a (n=11) for 2 weeks. Data are presented
as the percent of time spent freezing during a 2 min test, (* p ≤ 0.05).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
DISCUSSION
Based on our investigations of a novel therapeutic approach for treating AD based on dual
inhibition of PDEs and HDACs [11,12,25,26], we describe a novel chemical probe, 29a
(
CM-675), with class I HDAC selectivity over HDAC6 (>1 log unit) in the present study.
Compared to our hydroxamic acid-based lead compound 13 [11], compound 29a has a
0 times longer residence time on class I HDACs; e.g., 35 versus 347 min on HDAC1,
1
respectively. The slow binding kinetics of compound 29a emphasize the importance of
screening HDAC inhibitors under adequate experimental enzymatic assay conditions
(i.e., pre-incubation time) to prevent an underestimation of their potency and avoid
discarding them early in the drug discovery process. Other authors have also reported
similar findings for this family of ZBG: compounds 6 [28] and 9 [34] (whose HDAC2
IC decreased from 900 nM to 27 nM when delaying the compound incubation time from
5
0
1 h to 24 h). Thus, the use of longer pre-incubation times with compounds exhibiting long
residence times (or are expected to exhibit long residence times, according to their
structural characteristics), as reported here, is key to obtain precise IC values and
5
0
therefore selectivity profiles for HDAC isoforms.
Together with setting adequate experimental assay conditions, kinetic binding profiling
of HDAC inhibitors (i.e., structural-kinetic relationships, SKR) is becoming recognized
2
2
ACS Paragon Plus Environment

Page 23 of 66
ACS Chemical Neuroscience
1
2
3
4
5
6
7
8
9
as an increasingly important method to discern kinetic selectivity between highly
homologous HDAC isoforms belonging to the same family (e.g., HDAC1 and HDAC2),
thus complementing thermodynamic profiling (i.e., classical structure-activity
relationships, SAR) [22]. Here, the kinetic selectivity profile of compound 29a merits
further investigation, specifically its residence time on HDAC2 and HDAC3 isoforms, as
a pre-requisite to associate its biochemical profile with the observed functional response
and mode of action (MoA).
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Functionally, compound 29a is able to induce significant cellular responses (AcH3K9
and pCREB) at low concentrations (26-100 nM), while its worst cytotoxicity (towards
neurons and glia cells) is in the micromolar range. This finding endows compound 29a
with a therapeutic window of around one log unit, suggesting (together with the results
from the patch-clamp electrophysiology assay) that compound 29a is safe for in vivo
administration. Moreover, compound 29a reaches the brain at a dose sufficient to exert in
vivo effects on functional hallmarks in the mouse hippocampus, as confirmed following
acute i.p. administration of a 20 mg/kg dose. Disappointingly, we did not observe a
significant improvement in memory restoration after two weeks of treatment with
compound 29a, although a certain trend is acknowledged. From the compound
optimization perspective, compound 29a contains the sildenafil scaffold. Based on the
SAR results presented in Table 1, vardenafil-based analogues seem to be more potent at
inhibiting class I HDACs than compounds based on the sildenafil core, suggesting that
the vardenafil-matched pair of 29a might be a good alternative. However, during the
course of our investigations with hydroxamic acid-based derivatives, the vardenafil
matched pair of our sildenafil-based lead compound 13 exhibited a narrower therapeutic
range and poorer in vivo pCREB efficacy that prevented its progression to investigations
2
3
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 24 of 66
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
using in vivo models of AD. Therefore, sildenafil-based compounds were prioritized over
compounds with the vardenafil chemotype in the present study.
Additional studies are underway to identify the relationships among chemical structures,
residence times, binding kinetics, binding sites, biochemical inhibition and functional
responses. These studies may provide researchers new opportunities to develop HDAC
isoform-selective molecules, based on their binding kinetics, and then to obtain dual
phosphodiesterase and HDAC inhibitors with the required target compound profile to
treat AD. In addition, the kinetic profiling of these molecules is also very important to
define the optimal dosing scheme for in vivo testing in terms of determining the efficacy
and safety.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
4
ACS Paragon Plus Environment

Page 25 of 66
ACS Chemical Neuroscience
1
2
3
4
5
6
7
8
9
CHEMISTRY
Scheme 1.
O
N
O
O
N
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
N
N
O
HN
O
O
HN
i)
O
HN
N
ii)
iii)
N
N
N
N
14
15
O
S
A
SO Cl
2
16a-c, R = CH or CH CH
3 2 3
OR
O
O
O
N
N
N
O
HN
O
O
HN
A = a)
c)
HN
N
N
b)
iv)
NH
N
N
N
N
O
S
A
O
S
A
O
H
N
NH2
N
N
O H
18a-c
N
N
1
7a-c
O
O
Conditions: i) ClSO
LiOH·H O, THF/MeOH/H
NMM, DMF, rt, overnight.
3
H, 0 ºC, then rt, 2 h; ii) corresponding amine, Et
3
N, EtOH, 100 ºC, MW, 1 h; iii)
2
2
O (10:1:5 or 15:1:5), rt, overnight; iv) EDC·HCl, HOBt, benzene-1,2-diamine,
Sildenafil derivatives 18a-c were prepared as described in Scheme 1 from commercially
available 5-(2-ethoxyphenyl)-1-methyl-3-propyl-6H-pyrazolo[4,3-d]pyrimidin-7-one
14). This compound was first transformed into key intermediate 15 by sulphonylation at
(
5
’-position of the phenyl ring and then esters 16a-c were synthesized after reaction with
the convenient amine under microwave irradiation. Then, desired amido anilines 18a-c
were obtained through hydrolysis and reaction with benzene-1,2-diamine using
EDC/HOBt/NMM as coupling system.
2
5
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 26 of 66
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Scheme 2.
O
N
O
O
O
HN
O
O
HN
i)
O
HN
N
ii)
iii)
N
N
N
N
N
N
N
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
19
20
O
S
A
SO Cl
2
21a-c, R = CH or CH CH
3
3
2
OR
O
O
O
N
O
HN
O
HN
N
HN
N
A = a)
b)
c)
N
iv)
N
NH
N
N
N
N
O
S
A
O
O
S
A
O
H
N
NH2
22a-c
23a-c
N
N
O H
N
N
O
O
Conditions: i) ClSO
LiOH·H O, THF/MeOH/H
diamine, NMM, DMF, rt, overnight.
3
H, 0 ºC, then rt, 2 h; ii) corresponding amine, Et
3
N, EtOH, 100 ºC, MW, 1 h; iii)
2
2
O (10:1:5 or 30:1:10), rt, overnight; iv) EDC·HCl, HOBt, benzene-1,2-
Vardenafil analogues 23a-c were synthesized following the same method described for
sildenafil derivatives 18a-c (Scheme 2).
Scheme 3.
2
6
ACS Paragon Plus Environment

Page 27 of 66
ACS Chemical Neuroscience
1
2
3
4
5
6
7
8
9
iv) and v)
O
N
O
N
A)
O
N
N
O
A
HN
N
O
A
HN
O
N
O
B
HN
N
N
N
i)
N
ii)
iii)
viii)
O
I
HN
14
N
N
NH2
H
N
OR
24
25
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
O
O
O
O
26a-c, R = CH or CH CH
3
3
2
2
9a-c
vi)
2
7a-c, R = H
8a-c, R = PFP
vii)
2
O
N
O
N
B)
O
O
HN
O
HN
N
O
I
HN
N
N
N
i)
N
iv) and v)
viii)
N
19
N
NH2
H
N
3
0
A
OR
31c, R = CH CH3
A
2
vi)
O
O
3
2c, R = H
vii)
34c
3
3c, R = PFP
A =
a)
b)
c)
S
Conditions: i) NIS, TFA, 0 ºC, then rt, overnight; ii) 4,4,5,5-tetramethyl-2-(4,4,5,5-tetramethyl-1,3,2-
dioxaborolan-2-yl)-1,3,2-dioxaborolane, KOAc, Pd(dppf)Cl , 1,4-dioxane, 80-100 ºC, 48 h; iii)
corresponding bromide, Pd(PPh , K CO , 1,4-dioxane/H O (5:2 or 16:1), 85 ºC, MW, 1 h; iv) ethyl 3-
methylenecyclobutanecarboxylate, 9-BBN, THF, reflux, 4 h; v) xantphos or X-Phos, Pd (dba) , Na CO
,4-dioxane/H O (35:6), reflux, overnight; vi) LiOH·H O, THF/MeOH/H O (3:3:2 or 1:3:1), rt, overnight;
vii) 2,3,4,5,6-pentafluorophenol, DIC, CH Cl , 0 ºC, then rt, 6 h or overnight; viii) benzene-1,2-diamine,
2
3
)
4
2
3
2
2
3
2
3
,
1
2
2
2
2
2
DIEA, DMAP (optional), DMF, 60-80 ºC, overnight.
Synthesis of compounds 29a-c is shown in Scheme 3A. In this case intermediate 14 was
transformed into iodide 24 and then esters 26a-c were prepared through Suzuki coupling.
Subsequent hydrolysis with LiOH led us to corresponding carboxylic acids and desired
amido anilines 29a-c were obtained after preparation of pentafluorophenyl active esters
28a-c and reaction with benzene-1,2-diamine. On the other hand, vardenafile analogue
3
1
4c was synthesized by a similar strategy as used for ester 29c starting from compound
9 (Scheme 3B).
2
7
ACS Paragon Plus Environment

ACS Chemical Neuroscience
Page 28 of 66
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Scheme 4.
A)
O
O
O
N
N
N
O
A
HN
N
O
HN
N
i)
O
HN
i)
N
N
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
N
N
NH2
H
N
O
O
S
O
PFP
PFP
3
5a-b
O
28a
O
28c
B)
O
O
N
O
ii)
O
HN
O
B
HN
iii)
O
HN
i)
30
N
N
N
N
N
N
N
N
NH2
H
N
36
A
OR
A
O
O
O
37a, R = CH3
38a, R = H
O
S
iv)
v)
4
0a
3
9a, R = PFP
A = a)
b)
Conditions: i) thiophene-1,2-diamine, DIEA, DMAP, DMF, 60 ºC, overnight; ii) 4,4,5,5-tetramethyl-2-
4,4,5,5-tetramethyl-1,3,2-dioxaborolan-2-yl)-1,3,2-dioxaborolane, KOAc, Pd(dppf)Cl , 1,4-dioxane, 90
ºC, overnight; iii) methyl 4-(bromomethyl)benzoate, Na CO , Pd(PPh , 1,4-dioxane/H O (10:1), 85 ºC,
MW, 1 h; iv) LiOH·H O, THF/MeOH/H O (3:3:2), rt, overnight; v) 2,3,4,5,6-pentafluorophenol, DIC,
CH Cl , 0 ºC, then rt, overnight.
(
2
2
3
3
)
4
2
2
2
2
2
Synthesis of thiophene derivatives 35a-b and 40a is shown in Scheme 4. Compounds
5a-b were simple obtained from described intermediates 28a and 28c by reaction with
3
thiophene-1,2-diamine. A similar synthetic strategy was followed to afford vardenafil
analogue 40a from iodide 30.
Scheme 5.
2
8
ACS Paragon Plus Environment

Page 29 of 66
ACS Chemical Neuroscience
1
2
3
4
5
6
7
8
9
O
O
O
O
H
H
Cl
HO
ii)
NH
N
i)
O
iii)
O
O
O
N
H
N
H
H
H
4
1
O
42
O
43
44
O
O
O
O
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
O
O
O
N
N
A
n
OR
vi) or vii)
viii)
iv)
N
N
N
N
N
H
N
H
B
R
O
O
4
2
7a, n = 0, A = B = N, R = CH CH3
4
5, R = BOC
O
O
47b, n = 1, A = B = CH, R = CH3
v)
O
O
4
6, R = H
O
O
O
O
O
O
N
A
OR
N
A
N
H
x)
N
N
N
N
NH2
N
H
N
H
B
n
B
n
4
4
8a, n = 0, A = B = N, R = OH
8b, n = 1, A = B = CH, R = OH
50a, n = 0, A = B = N
50b, n = 1, A = B = CH
O
O
O
ix)
O
4
9a, n = 0, A = B = N, R = PFP
49b, n = 1, A = B = CH, R = PFP
Conditions: i) MnO
propoan-2-ol, 80-100 ºC, overnight; iii) 2-chloroacetyl chloride, Et
butyl 4-(aminomethyl)piperidine-1-carboxylate, MeOH, 80-100 ºC, overnight; v) HCl/EtOAc (1.0 M), 0
ºC, then rt, 3 h; vi) ethyl 2-chloropyrimidine-5-carboxylate, K CO , DMF, 80 ºC, overnight; vii) methyl 4-
formylbenzoate, CH COOH, CH Cl , rt, 1 h, then NaBH CN, rt, overnight; viii) LiOH·H O,
THF/MeOH/H O (10:3:2 or 3:2:2), rt, overnight; ix) 2,3,4,5,6-pentafluorophenol, DIC, CH Cl , 0 ºC, then
rt, overnight; x) benzene-1,2-diamine, DIEA, DMAP, DMF, 60 ºC, overnight.
2
, CH
2
Cl
2
, rt, overnight; ii) methyl (2R)-2-amino-3-(1H-indol-3-yl)propanoate,
3
N, CH Cl , 0 ºC, then rt, 3 h; iv) tert-
2
2
2
3
3
2
2
3
2
2
2
2
Tadalafil-based amides 50a-b were prepared following the synthetic route outlined in
Scheme 5 from commercially available 1,3-benzodioxol-5-ylmethanol (41). Oxidation of
this alcohol with MnO led us to piperonal and then compound 44 was synthesized
2
following a known synthetic procedure [43]. The stereochemistry of this intermediate has
been previously studied and it was confirmed to be the 6R/12aR cis isomer. Nevertheless
the subsequent transformations have not been described before and as it is known that
both stereogenic centers of tadalafil core can epimerize and their stereochemistry have
2
9
ACS Paragon Plus Environment