Novel Benzohydroxamate-Based Potent and Selective Histone Deacetylase 6 (HDAC6) Inhibitors Bearing a Pentaheterocyclic Scaffold: Design, Synthesis, and Biological Evaluation

Article abstract of DOI:10.1021/acs.jmedchem.9b01194

Histone deacetylase 6 (HDAC6) is a peculiar HDAC isoform whose expression and functional alterations have been correlated with a variety of pathologies such as autoimmune disorders, neurodegenerative diseases, and cancer. It is primarily a cytoplasmic protein, and its deacetylase activity is focused mainly on nonhistone substrates such as tubulin, heat shock protein (HSP)90, Foxp3, and cortactin, to name a few. Selective inhibition of HDAC6 does not show cytotoxic effects in healthy cells, normally associated with the inhibition of Class I HDAC isoforms. Here, we describe the design and synthesis of a new class of potent and selective HDAC6 inhibitors that bear a pentaheterocyclic central core. These compounds show a remarkably low toxicity both in vitro and in vivo and are able to increase the function of regulatory T cells (Tregs) at well-tolerated concentrations, suggesting a potential clinical use for the treatment of degenerative, autoimmune diseases and for organ transplantation.

Full text of DOI:10.1021/acs.jmedchem.9b01194

Subscriber access provided by - Access paid by the | UCSF Library
Article
Novel Benzohydroxamate-based Potent and Selective Histone
Deacetylase 6 (HDAC6) Inhibitors Bearing a Pentaheterocyclic
Scaffold: Design, Synthesis and Biological Evaluation
Barbara Vergani, Giovanni Sandrone, Mattia Marchini, Chiara Ripamonti, Edoardo cellupica, Elisabetta
Galbiati, Gianluca Caprini, Gianfranco Pavich, Giulia Porro, Ilaria Rocchio, Maria Lattanzio, Marcello
Pezzuto, Malgorzata Skorupska, Paola Cordella, Paolo Pagani, Pietro Pozzi, Roberta Pomarico, Daniela
Modena, Flavio Leoni, Raffaella Perego, Gianluca Fossati, Christian Steinkühler, and Andrea Stevenazzi
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.9b01194 • Publication Date (Web): 11 Nov 2019
Downloaded from pubs.acs.org on November 14, 2019
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 106
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Preclinical R&D
Fossati, Gianluca; Italfarmaco SpA Head Office and Research Centre,
Preclinical R&D
Steinkühler, Christian; Italfarmaco SpA Head Office and Research
Centre, Preclinical R&D
Stevenazzi, Andrea; Italfarmaco SpA Head Office and Research Centre,
Preclinical R&D
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
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 106
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
Novel Benzohydroxamate-based Potent and
Selective Histone Deacetylase 6 (HDAC6) Inhibitors
Bearing a Pentaheterocyclic Scaffold: Design,
Synthesis and Biological Evaluation
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
†
†
†
Barbara Vergani, Giovanni Sandrone, Mattia Marchini, Chiara Ripamonti, Edoardo
Cellupica, Elisabetta Galbiati, Gianluca Caprini, Gianfranco Pavich, Giulia Porro, Ilaria
Rocchio, Maria Lattanzio, Marcello Pezzuto, Malgorzata Skorupska, Paola Cordella, Paolo
Pagani, Pietro Pozzi, Roberta Pomarico, Daniela Modena, Flavio Leoni, Raffaella Perego,
Gianluca Fossati, Christian Steinkühler and Andrea Stevenazzi*
Preclinical R&D, Italfarmaco Group, Via dei Lavoratori 54, I-20092 Cinisello Balsamo, Milan,
Italy
KEYWORDS: HDAC6 inhibitors, ITF3756, ITF3791, Treg suppression
ABSTRACT:
Histone deacetylase 6 (HDAC6) is a peculiar HDAC isoform whose expression and functional
alterations have been correlated with a variety of pathologies such as autoimmune disorders,
ACS Paragon Plus Environment
1

Page 3 of 106
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
neurodegenerative diseases, and cancer. It is primarily a cytoplasmic protein, and its deacetylase
activity is focused mainly on non-histone substrates such as tubulin, heat shock protein (HSP)90,
Foxp3 and cortactin, to name a few. Selective inhibition of HDAC6 does not show cytotoxic
effects in healthy cells, normally associated with the inhibition of Class I HDAC isoforms. Here
we describe the design and synthesis of a new class of potent and selective HDAC6 inhibitors
that bear a pentaheterocyclic central core. These compounds show a remarkably low toxicity
both in vitro and in vivo and are able to increase the function of regulatory T cells (Tregs) at well
tolerated concentrations, suggesting a potential clinical use for the treatment of degenerative,
auto-immune diseases and organ transplantation.
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
INTRODUCTION
Histones, together with DNA, are the main components of the nucleosome, the primary
chromatin unit. In eukaryotic cells, post-translational modifications of the core histones’ N-
termini are involved in regulation of gene expression by modulating chromatin accessibility and
by providing anchoring points for “reader” proteins that help guiding the transcription machinery
to gene promotors. Reversible acetylation of histones is regulated by the interplay of ‘writer’
histone acetyltransferases (HATs) and ‘eraser’ histone deacetylases (HDACs).1
There are 18 known human HDACs divided into two groups: zinc-dependent HDACs and
nicotinamide adenine dinucleotide (NAD)-dependent HDACs, also known as sirtuins (Class III).
Zinc-dependent HDACs are further grouped into four classes depending on their sequence
homology and on their catalytic activity: (1) Class I includes HDAC1, 2, 3, and 8, ubiquitary
isoenzymes primarily localized in nucleus; (2) Class IIa includes HDAC4, 5, 7, and 9,
ACS Paragon Plus Environment
2

Journal of Medicinal Chemistry
Page 4 of 106
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
isoenzymes localized both in nucleus and in cytoplasm; (3) Class IIb, includes HDAC6 and 10,
mainly localized in cytoplasm; and (4) Class IV, which includes only HDAC11.2
Proteomic studies have shown that in addition to histones, post-translational acetylation
regulates the function and stability of a growing number of proteins with diverse biological
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
activities, from transcription factors to metabolic and contractile proteins. Indeed, protein lysine
4
acetylation has long been recognized to have a comparable role to protein phosphorylation and
the widespread activity of these enzymes has led to the proposal to rename the family as “lysine
deacetylases” or KDACs.
Numerous HDAC inhibitors have been synthesized. Most of them are characterized by the
5
presence of a hydroxamate group, able to bind the zinc in the active site of the enzyme. There
are also several examples of non-hydroxamate HDAC inhibitors, among them the most
important are the 2-aminophenylbenzamide-derivatives, like Entinostat and Mocetinostat,6
currently in clinical phase III and II, respectively. Some HDAC inhibitors have been approved by
Food and Drug Administration (FDA) for the treatment of cancer: Vorinostat and Romidepsin
for cutaneous T-cell lymphoma (CTCL), Belinostat for peripheral T-cell lymphoma and
Panobinostat for multiple myeloma (MM). These compounds are relatively non-selective
inhibitors of all or most of the zinc-dependent HDACs and have a modest therapeutic window. In
clinical trials, side effects such as decreased platelet levels, fatigue, or gastrointestinal toxicity
7
are dose-limiting. The scientific community is, therefore, actively trying to develop novel
inhibitors that are highly selective for individual HDAC subtypes with the goal of identifying
more active and better tolerated compounds. In addition, selective inhibitors may also allow
specific biological functions of individual HDAC subtypes to be addressed.
ACS Paragon Plus Environment
3

Page 5 of 106
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
In this context, HDAC6 is an interesting, emerging target, since its expression and functional
alterations have been correlated to a variety of pathologies such as autoimmune disorders,
neurodegenerative diseases and cancer.8
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
HDAC6 is the only isoform with two catalytic domains and a C-terminal zinc finger domain,
which can bind ubiquitinated proteins. Even though it can be localized in the nucleus under
9
certain conditions, HDAC6 is primarily a cytoplasmic protein and its main deacetylase activity
is toward a number of non-histone substrates including -tubulin, heat shock protein 90 (Hsp90),

-catenin, forkhead box 3 (Foxp3), p53 and cortactin. In contrast to many other HDACs, whose
ablation is lethal in mice or leads to severe functional impairment, HDAC6 KO mice are viable,
suggesting that the selective inhibition of this isoenzyme is likely to be well-tolerated.9a, 10
Among different functions, HDAC6 has an important role in immune response regulation.
HDAC6, as well as other isoforms, deacetylates the Foxp3 transcription factor,9a, 10 a master
regulator of Treg cell development and activity. Selective inhibition of HDAC6 leads to the
1
1
acetylation of Foxp3 which in turn results in an increased Treg suppressor function. Tregs play
1
2
an important role in maintaining immune system homeostasis, and their alteration is involved
in the development of auto-immune pathologies such as rheumatoid arthritis, psoriasis, lupus,
myasthenia gravis, as well as graft versus host disease (GVHD) and organ rejection. In these
pathologic conditions, pharmacologic increase of Treg suppression function is an active field of
1
3
investigation. Increased Treg activity exerted by selective HDAC6 inhibitors appears a
promising approach for the treatment of these disorders and, in particular, in organ
transplantation.14
Many selective HDAC6 inhibitors are described in the literature,5, 8a, 15 and a new generation of
16
candidates have recently appeared. Tubastatin A (see Table 1 and Figure 1), a compound
ACS Paragon Plus Environment
4

Journal of Medicinal Chemistry
Page 6 of 106
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
obtained through rational design combined with homology modeling was a milestone in the
1
7
discovery of this class of compounds. The tricyclic molecule is highly selective for HDAC6
and is largely used as a research tool. In order to improve the druglike properties of this kind of
structure, Kozikowski et al. later generated a range of analogues with a bicyclic-capped
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
8
scaffold and identified HDAC6 inhibitors with improved pharmacokinetics and in vitro
selectivity. The same group developed Nexturastat A (see Table 1 and Figure 1), an urea
derivative compound, that shows in vitro antiproliferative activity on B16 murine melanoma
1
9
cells. Another worth mentioning compound is HPOB (see Table 1 and Figure 1), which shows
good selectivity for HDAC6 and seems to enhance the anticancer activity of some known drugs
2
0
such as doxorubicin and etoposide. ACY-738 (see Table 1 and Figure 1) and ACY-775 are two
pyrimidine-derivatives that inhibit HDAC6 with high selectivity and potency. These compounds
2
1
rapidly distribute to the brain and show antidepressant properties in animal models. To the best
of our knowledge, however, none of these selective inhibitors are under active development,
probably due to stability issues or non-optimal druglike properties (as an example, compare
stability data of the above mentioned compounds with the ones of Ricolinostat, Table 1).
Two selective HDAC6 inhibitors, namely Ricolinostat (see Table 1 and Figure 1) and
Citarinostat, are currently under clinical phase I/II investigation, in combination with various
2
2
agents such as dexamethasone and bortezomib for the treatment of multiple myeloma and
immune checkpoint inhibitor nivolumab in non-small cell lung cancer, respectively. These two
molecules, however, exhibit very low selectivity over HDAC Class I isoforms (see Table 1) and
may have a suboptimal therapeutic index.
ACS Paragon Plus Environment
5

Page 7 of 106
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
OH
O
O
N
N
N
N
H
N
H
H
H
N
N
N
OH
OH
OH
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
Tubastatin A
Nexturastat
HPOB
S
F
O
N
O
H
H
H
N
N
N
N
N
O
N
H
N
H
H
N
H
N
N
N
O
OH
OH
OH
O
O
O
ACY-738
ACY-775
ITF3107
N
N
N
N
H
H
H
H
N
N
N
N
N
N
OH
Cl
OH
O
O
O
O
Ricolinistat
Citarinostat
N
H
O
N
H
O
N
OH
Givinostat
O
Figure 1. Representative examples of selective histone deacetylase 6 (HDAC6) inhibitors along
with the pan-inhibitor Givinostat and an internally developed HDAC6 inhibitor, i.e. ITF3107.
Table 1. In-house biological evaluation of the most representative selective HDAC6 inhibitors
along with the pan-inhibitor Givinostat and an internally developed HDAC6 inhibitor, i.e.
ITF3107.
ACS Paragon Plus Environment
6

Journal of Medicinal Chemistry
Page 8 of 106
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
HDAC6 Sel. vs Sel. vs Sel. vs Rat
Human
697
toxicity
Compd
a
b
b
b
c
c
d
IC50
HDAC1 HDAC2 HDAC3 plasma plasma
Tubastatin A
Nexturastat
HPOB
216
251
1212
21
218
84
942
172
146
1176
101
547
9
246
64
53
200
16
73
1
383
401
451
121
911
181
721
982
2006327
40134
1001
7922
1001
1001
1001
972
38
8330464
3390246
20129
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
ACY-738
Ricolinostat
ITF3107
156
28
121
51
120
2
>10000
Givinostat
445
10015
a – Enzymatic data (IC ) in nM unit were obtained from curve-fitting of a 5-point enzymatic
5
0
assay starting from 100, 30 or 10 μM with 10-fold serial dilution. Experiments were done in
triplicate. Single experiment, except for ITF3107 and Givinostat (n=4). In-house data are in
1
9-22
accordance with the literature.
퐻퐷퐴퐶푛 퐼퐶50
b - Selectivity indicator, expressed in terms of ratio between IC values (퐻퐷퐴퐶6 퐼퐶50
)
. See the SI
5
0
spreadsheet for the original data and the statistical evaluation.
c – Residual percentage of compounds after 4 h. Experiments were done in duplicate. SDs were
calculated on technical replicates
d - IC values in nM unit are the mean of at least two experiments obtained from the curve
fitting of a 4-points viability assay starting from 10 µM with 10-fold serial dilution.
5
0
2
3
ITF3107, a benzohydroxamate-based molecule bearing an amino acid scaffold (see Figure 1),
shares the same selectivity, but also the same stability issues of the other selective HDAC6
inhibitors (see Table 1), probably due to the presence of an amide bond. The compound is,
indeed, a substrate of chymotrypsin (data not shown).
Here we report how our rational drug design approach led to: (1) the modification of the
structure of ITF3107 in order to generate a new class of stable and selective hydroxamate-based
HDAC6 inhibitors, and (2) the identification of potential preclinical candidates.
RESULTS AND DISCUSION
ACS Paragon Plus Environment
7

Page 9 of 106
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Structure–Activity Relationship (SAR): The design strategy adopted in the discovery of the
new class of compounds was essentially ligand-based, although several docking calculations
were performed in order to cast light on protein motif involved in the binding with the inhibitors.
All new molecular entities were conceived to satisfy simultaneously two conditions: to be highly
potent versus HDAC6 (IC in the low nM range) and extremely selective (two or three orders of
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
5
0
magnitude) over HDAC3, the latter being selected as representative of Class I HDAC isoforms.
HDAC3 was preferred to HDAC1 as during the early stage of our screening campaign the
selectivity over HDAC3 was the most demanding to achieve. Activity on both isoforms (HDAC6
and HDAC3) was evaluated in terms of IC (nM), whereas the simple potency ratio
50
IC (HDAC3)/IC (HDAC6) was used to monitor the selectivity.
5
0
50
The extensive literature on HDAC inhibitors helped us to define a paradigm for an efficient
2
4
inhibitor, partitioning the structure (Figure 2) in four distinct domains: (1) the zinc binding
group (ZBG); (2) a spacer (usually hydrophobic) filling the dip between the catalytic metal ion
and the outermost surface; (3) a junction capable of addressing (4) the last inhibitor component
(namely the cap term) toward the flexible rim of enzyme.
N
H
O
H
O
N
N
H
N
OH
O
N
H
OH
O
O
Figure 2. Qualitative partition of Vorinostat (Zolinza) and Givinostat, two HDAC pan-inhibitors.
It is possible to distinguish in both molecules the zinc binding group (ZBG) in blue, the spacer
region in red, the “junction” in green, and the cap term, in pink.
ACS Paragon Plus Environment
8

Journal of Medicinal Chemistry
Page 10 of 106
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
25
The hydroxamic acid moiety is a well-known and highly efficient ZBG that has been widely
adopted in the development of reversible inhibitors of zinc-based hydrolases such as matrix
2
6
27
metalloproteinases (MMPs) or tumor necrosis factor- converting enzymes (TACE),
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
enzymes in which the closed shell transition metal ion acts as a Lewis acid and triggers the
substrate hydrolysis. The relatively high zinc affinity of this functional group favors both flexible
2
8
29
29-30
(Vorinostat, Ricolinostat ) and rigid aromatic
(Givinostat) spacers in the dip between the
zinc cation and the enzyme rim. The N-hydroxybenzamide itself (Table 2, compound 1) exhibits
an inhibitory activity close to 2 M and a moderate selectivity versus HDAC6 (in terms of
2
4
IC ). Substitution in the para position with an amino group (compound 2) improves the
5
0
potency with a mild reduction in the selectivity, whereas acetylation of this amino group leads to
sub-micromolar activity versus HDAC6 (compound 3). The resulting molecular entity still
exhibits some selectivity and can be considered as a promising starting point for a new class of
HDAC6 selective inhibitors, provided that stability of the amide bond is being monitored.
Table 2. Potency and selectivity trend in small similar moieties.
Selectivity
Compd
HDAC6a
HDAC3a
6 vs 3b
H
N
1
2370255 7157210438
30
OH
O
H N
2
H
N
2
139559
184021092
13
OH
O
ACS Paragon Plus Environment
9

Page 11 of 106
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
H
N
H
N
3
4
27021
6957
1671110
5000469
6
7
O
OH
OH
O
O
H N
2
H
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
a – Enzymatic data (IC ) in nM unit were obtained from curve-fitting of a 5-point enzymatic
assay starting from 100, 30 or 10 μM with 10-fold serial dilution. Experiments were done in
triplicate (single experiment). SDs were calculated on technical replicates.
50
퐻퐷퐴퐶3 퐼퐶50
b - Selectivity indicator, expressed in terms of ratio between IC values (퐻퐷퐴퐶6 퐼퐶50
)
.
5
0
On the other hand, the amide bond can be replaced with a substructure having similar behavior
and compatible geometry but higher resistance versus the activity of the metabolic enzymes.31
Substitution can be particularly convenient when the replacement not only mimics all or part of
interactions of the amide group with the binding site residues, but it also adds new ones. The
five-membered heterocycles reported in Figure 3 can potentially mimic the lone pairs localized
3
2
on the amide oxygen thanks to the two adjacent nitrogen atoms, but only the 1,2,4-triazole (A)
and the 1,2,3-triazole (F) have a hydrogen bond donor (HBD) similar to the amide, whereas the
other pentaheterocycles are apparently closer to the ester functional group (B-E).
(A)
(B)
(C)
O
N N
N N
O
N N
S
*
*
*
*
*
*
*
*
N
N
H
H
(
D)
(E)
(F)
N O
N
N N
N N
N
N
*
*
*
*
*
*
N
Figure 3. Five-membered rings acting as non-classical bioisosteres of the amide/ester bond.
ACS Paragon Plus Environment
1
0

Journal of Medicinal Chemistry
Page 12 of 106
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
Direct replacement of the amide of compound 3 with a 1,2,4-triazole results in better
selectivity (compound 5, Table 3) that is retained also when a methylene group is introduced
between the phenyl and the triazole rings (compound 6, Table 3). In the literature there are
examples of azole-containing benzohydroxamate-based HDAC6 inhibitors.18
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Table 3. Effects on enzymatic inhibitory activity of para substitution on the benzoic
hydroxamate with 1,2,4 triazole, directly bound to phenyl substructure (compound 5) or bearing
a methylene spacer (compound 6).
Sel 6 vs
Compd HDAC6a HDAC3a
b
3
NH
N
N
5
6
35336
1989
3563203
2896135
10
15
H
N
OH
O
H
N
H
N
N
N
OH
O
a – Enzymatic data (IC ) in nM unit were obtained from curve-fitting of a 5-point enzymatic
5
0
assay starting from 100, 30 or 10 μM with 10-fold serial dilution. Experiments were done in
triplicate (single experiment). SDs were calculated on technical replicates.
퐻퐷퐴퐶3 퐼퐶50
b - Selectivity indicator, expressed in terms of ratio between IC values (퐻퐷퐴퐶6 퐼퐶50
)
.
50
In the early stage of the investigation a crude homology model was prepared using HDACs or
homologue HDACs structures available at the time. Despite its low quality (especially if
3
3
compared with the recently published complex), the model was able to capture several features
ACS Paragon Plus Environment
1
1

Page 13 of 106
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
of the enzyme and address the design of cap term region of the inhibitors. It is of note that the
hypothetic binding conformer was derived from docking calculations where the large amount of
high ranking poses of the training compounds (selective inhibitors) were similar to the
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
30
Trichostatin A binding conformer available in h-HDAC8 (pdb code 1T64): the cap term of the
ligand is pointing toward loop L1, although able to engage an intramolecular bond with Asp101,
a conserved residue in Class I isoforms and part of loop L2.
The structures of hydroxamates co-crystallized with HDAC6 show relevant details in the inner
catalytic core, not previously detected in complexes involving other HDAC isoforms.30, 34 The
denticity (i.e. the number of donor groups in a single ligand that bind to a central atom in a
3
5
coordination compound) of the hydroxamate in Class I enzymes is usually bidentate, with
oxygen atoms both involved in the Zn-O bond, allowing a distorted trigonal bipyramid
coordination for the cation. A common feature of several hydroxamate-based inhibitors co-
crystallized with HDAC62
9, 36
indicates that only the hydroxyl oxygen atom is directly bound to
the metal, whereas the carbonyl oxygen is engaged in a hydrogen bond with the catalytic tyrosine
Y745 for z-CD2, PDB code:5G0I/ Y782 for h-CD2, PDB Code 5EDU). This particular binding
(
mode likely allows the presence of one catalytic water molecule in the inner zinc coordination
sphere, a feature that can play a possible role for the selectivity. It must be said, though, that the
bidentate mode is also detected in several HDAC6 - complexes29, 37 as the energy barrier between
these two binding conformers is quite low. As a consequence, the delicate balance between the
two different denticities, combined with the interaction between the cap-term of the inhibitor and
the rim residues of the enzyme could be the key for the selectivity.
Moving from the catalytic core to the outermost region of the enzyme both zebra fish and
human CD2 domains exhibit a dyadic motif that involves F620 and F680 in h-CD2 (F583 and
ACS Paragon Plus Environment
1
2

Journal of Medicinal Chemistry
Page 14 of 106
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
F643 in z-CD2), two highly conserved residues among HDAC isoforms (see Figure 4). The
facing arylic side chains define a crevice where the phenyl moiety of benzylhydroxamate can fit
in, stabilized by  interactions, as shown, as illustrative examples, for compounds 42 in Figure
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
4
and 13 in Figure 5. Comparison of available HDAC6 complexes with both pan and selective
hydroxamic-based inhibitors29, 33, 38 along with recent docking studies indicate the relevance of
residues in loops L1 (H500/H463, P501/P464, h-CD2/z-CD2) and L2 (S568/S531), where
binding can be related to a direct contact with enzyme residues or mediated by water molecules,
respectively. Although the loop L1 is shared between class I isoforms and the CD2 of HDAC6,
39
3
9
recent docking results and related structural analysis suggest that the selectivity for HDAC6
can be explained in terms of narrow (HDAC1) or different shape of the binding pocket
(HDAC8, where loop L1 is shorter). Rim details do not justify, however, the selectivity of
compounds shown in tables 2 and 3, due to the lack of the cap-term, but recent biophysical
investigations indicate that the larger affinity toward HDAC6 exhibited by the benzohydroxamic
acid itself (compound 1) is probably entropy driven.40
ACS Paragon Plus Environment
1
3

Page 15 of 106
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Figure 4. The docking pose of compound 42 having the highest Dreiding score (the best ranking
function detected in the preliminary docking of the training set compounds). The aryl cap term is
pointing toward loop L1, interacting with a hydrophobic pocket defined by Pro501, His500,
His499 and Leu749 (h-CD2, PDB code 5EDU). H-bond, electrostatic and π-π interactions are
represented as green, orange and purple dashed lines, respectively; Zn ion as a magenta sphere.
The surface of the binding site was generated and rendered in transparent solid yellow.
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 5. The docking pose of compound 13 having the highest Dreiding score (the best ranking
function detected in the preliminary docking of the training set compounds). The aryl cap term is
pointing toward loop L1, interacting with a hydrophobic pocket defined by Pro501, His500 and
His499 (h-CD2, PDB code 5EDU). H-bond, electrostatic and π-π interactions are represented as
green, orange and purple dashed lines, respectively; Zn ion as a magenta sphere. The surface of
the binding site was generated and rendered in transparent solid yellow.
ACS Paragon Plus Environment
1
4

Journal of Medicinal Chemistry
Page 16 of 106
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
Docking calculations were performed using the libdoc protocol of the Discovery Studio (DS)
Suite (see Supporting Information for details) and the conformational search of candidates was
focused in a sphere centered in the center of mass of aligned inhibitors available in the crystal
structures (radius equal to 8.0 A). The sphere volume includes both L1 and L2 loops, the
catalytic core, the zinc cation and the residues directly bound to the metal ion.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
4
1
The best score function for the ranking of the poses was the “Dreiding scores”, identified by
analyzing the Receiving Operator Curves (ROCs) evaluated in a preliminary calculation where a
training set of HDAC6 selective and pan inhibitors were mixed with compounds exhibiting low
HDAC6 inhibition potency (at least 10 fold less).
The results emerging from the qualitative docking calculations led us to synthesize a collection
of compounds having several five-membered ring heteroaromatic cycles, bearing the p-
methylbenzohydroxamic acid and an aromatic group. Among them, 1,2,4-triazole, tetrazole (both
1
,5- and 2,5-disubstituted), 1,3,4-oxadiazole, 1,2,4-oxadiazole, and 1,3,4-thiadiazole (Table 4,
compounds 7a–c; 7e-g) generated molecules exhibiting good activity on isoform 6 (4 nM) and
selectivity (105 fold) over HDAC3, a robust proof that i) five-membered rings can be
interchanged without relevant loss of potency and selectivity and that ii) the small aromatic cap-
term is a relevant pharmacophoric feature for both potency and selectivity.
The key role played by the methylene group between the benzohydroxamate moiety and the
five-membered ring is clearly evident when comparing compounds 7b and 8 (Table 4), where the
3
lack of an sp carbon atom of the latter induces a significant decrease in both potency and
selectivity. A comparable effect has already been described for other HDAC6 selective inhibitors
3
9
3
that bear a phenothiazine-based benzhydroxamic acid. An sp atom acting as a bridge between
ACS Paragon Plus Environment
1
5

Page 17 of 106
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
the two aromatic rings seems beneficial for the HDAC6 inhibitory activity,17, 19-20 as it allows a
tilted conformation of the molecule which, in turn, maximizes the interaction with the enzyme’s
binding site.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Table 4. Confirmatory analysis of aromatic heterocycle equivalence and relevance of methylene
spacer between the five-membered ring and the benzohydroxamate moiety.
R
n
Het
H
N
OH
O
Sel
Compd
Het
n
R
HDAC6a 6 vs 3b
H
N
7
a
1
51
21
N
N
N
N
O
N
N
O
N
7
b
1
1
1
1
1
41
32
72
26
50
105
N
N
N
7
c
31
N
N
N
7
d
521
61
N
N
S
N
7e
7f
161
N
N
7
g
1
91
83
N
N
ACS Paragon Plus Environment
1
6

Journal of Medicinal Chemistry
Page 18 of 106
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
8
9
0
1
6609
171
3
N
N
N
N
129
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
a – Enzymatic data (IC ) in nM unit were obtained from curve-fitting of a 5-point enzymatic
assay starting from 100, 30 or 10 μM with 10-fold serial dilution. Experiments were done in
triplicate (single experiment). SDs were calculated on technical replicates.
50
퐻퐷퐴퐶3 퐼퐶50
b - Selectivity indicator, expressed in terms of ratio between IC values (퐻퐷퐴퐶6 퐼퐶50
)
. See the
5
0
SI spreadsheet for the original data and the statistical evaluation.
The shape and the extension of the hydrophobic pocket in the outer portion of the binding site
was then evaluated by adding a methylene spacer between the apical phenyl group and the 2,5-
disubstituted tetrazole (compound 7d, table 4); although the selectivity is still acceptable, the
potency decreases (compare compounds 7c and d), suggesting a commensurable misfit on both
HDAC isoforms tested.
The next step was the exploration of a possible (bioisosteric) replacement of the -CH - group.
2
The sulfur atom, despite the larger volume of the atomic orbitals in the valence shell, leads to
potent analogues (Table 5, compounds 10a and b) although reduced selectivity is detected when
combined with 1,2,4-oxadiazole. On the contrary, the -NH- moiety fails in both exploratory
compounds synthesized (compounds 11a and b). The different behavior of these analogues can
3
apparently be explained in terms of the different orbital hybridization in which the sp valence
orbitals on the sulfur atom allow for a higher degree of freedom with a small energetic barrier
between inactive and active (bent) conformers. The setup is not perfectly compatible with the
2
nitrogen atom in which the sp component is more prominent and the related lone pair in the
residual p atomic orbital can be delocalized between the phenyl group and the five-membered
ACS Paragon Plus Environment
1
7

Page 19 of 106
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
ring. The resulting structures (11a and b) are then flat, linear and rigid and unable to efficiently
fit into the external portion of the binding site.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Table 5. Possible replacement of methylene group (-S- and –NH-) between the aromatic
substructures.
X
R
Het
H
N
OH
O
Compd
10a
Het
O
N
X
R
HDAC6a Sel 6 vs 3b
S
61
61
17
62
19
10
N
N
N
N
S
10b
S
N
S
1
1a
NH
NH
6610
1361
N
O
N
1
1b
a – Enzymatic data (IC ) in nM unit were obtained from curve-fitting of a 5-point enzymatic
5
0
assay starting from 100, 30 or 10 μM with 10-fold serial dilution. Experiments were done in
triplicate (single experiment). SDs were calculated on technical replicates.
퐻퐷퐴퐶3 퐼퐶50
b - Selectivity indicator, expressed in terms of ratio between IC values (퐻퐷퐴퐶6 퐼퐶50
)
. See the
5
0
SI spreadsheet for the original data and the statistical evaluation.
Methylation in position 4 of the 1,2,4-triazole ring (compound 9, Table 4) confirms the
hypothesis of the low relevance of a hydrogen bond donor in that position and the experimentally
detected efficiency of oxadiazoles, tetrazoles, and thiadiazoles as isosteres of the amide group.
Although the lack of a polar hydrogen slightly decreases potency on HDAC6 of compound 9,
ACS Paragon Plus Environment
1
8

Journal of Medicinal Chemistry
Page 20 of 106
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
selectivity is substantially improved, suggesting a different local environment in the two
isoforms. These considerations are true if the substituent group is small, as a methyl or an amine
(compound 13, 14 and 15, Table 6), while both selectivity and especially potency on HDAC6 are
negatively affected by a bulkier substituent, as e.g. a phenyl group (compound 12, Table 6). On
the other hand, the importance of the lone pairs on adjacent nitrogen atoms of the pentacycles
has been verified by the introduction of substituents in positions 1 and 2 of the 1,2,4-triazole
core. The analysis of the data collected in Table 6 shows that all of these types of substitutions
have a small influence on the selectivity profile, but they deeply affect the potency of the
HDAC6 inhibitory activity. Substitutions in these positions, especially in position 1, in fact,
involve a drastic drop in potency (see compounds 16–18, Table 6). Alkylation of the nitrogen in
position 1 means somehow silencing it by preventing hydrogen bond formation or other types of
interactions with the protein residues. These results are in accordance with the ones exhibited by
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
8
compounds 19, 20 and 21 (Table 6): these molecules, where the nitrogen in position 1 has been
replaced by a carbon atom, show indeed lower inhibitory activities.
Table 6. Effects of triazole nitrogen atoms substitution on the enzymatic activity.
R4
N
X
R6
H
N
N
N
R1
R2
OH
O
Sel 6
vs 3
Compd R4
R1
R2
R6
X
HDAC6a
b
1
2
abs
abs
S
561
25
ACS Paragon Plus Environment
1
9

Page 21 of 106
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
S
1
1
3c
4
CH3
CH3
abs
abs
abs
S
S
81
51
61
71
abs
N
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
O
N
1
5
NH2
abs
abs
abs
S
91
55
16
S
O
O
S
16
abs
abs
CH2 33324
CH2 36820
Cl
1
1
7
8
abs
abs
24
12
abs
S
16216
13712
H
S
N
H
N
1
9
4
N
OH
O
S
S
H
2
2
0
1
N
N
6212
1026
27
49
OH
O
N
N
H
N
OH
O
a – Enzymatic data (IC ) in nM unit were obtained from curve-fitting of a 5-point enzymatic
assay starting from 100, 30 or 10 μM with 10-fold serial dilution. Experiments were done in
triplicate (single experiment). SDs were calculated on technical replicates.
5
0
퐻퐷퐴퐶3 퐼퐶50
b - Selectivity indicator, expressed in terms of ratio between IC values (퐻퐷퐴퐶6 퐼퐶50
)
. See the
5
0
SI spreadsheet for the original data and the statistical evaluation.
c - HDAC6 IC value is the mean of three experiments and SD was calculated on these
5
0
independent experiments.
abs = not present.
ACS Paragon Plus Environment
2
0

Journal of Medicinal Chemistry
Page 22 of 106
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
A careful analysis of the poses generated by the highly potent compounds of the training set
revealed two possible (average) “active” conformations, shown in Figure 6 for compound 42.
The first geometry, light blue in the figure, is in agreement with the results observed in the
crystal structures of HDAC6-inhibitor complexes,29, 33, 36-39 where the aryl moiety of the cap term
lies close to L1 and L2 loops and interacts with the hydrophobic pocket generated by H500 and
P501 (hCD2). The second conformation (in yellow) can be obtained from the former geometry
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
°
42
by a 180 rotation of a C-C bond (see Figure 6). These results are consistent with the literature.
Figure 6. Binding conformers of compound 42 detected in docking experiment (hHDAC6 CD2,
PDB code 5EDU): conformer A in light blue and conformer B in yellow. The green surface shows
residues which interact with both conformers; light blue and yellow surfaces show specific
interactions with conformer A and B respectively. H-bond, electrostatic and π-π interactions are
represented as green, orange and purple dashed lines, respectively; the Zn ion as a magenta sphere.
ACS Paragon Plus Environment
2
1

Page 23 of 106
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
The aryl moiety of the cap term is pointing toward the loop (L5) where T678 (A641 in zCD2)
was recently indicated as a key residue involved in a hydrogen bond network triggered by the
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
7
peptidoid moiety of several HDAC6 selective inhibitors and the water molecules. It should be
noticed that the aryl moiety of the cap-term lies too far from threonine and the conserved F679
(F642 in zCD2), whereas the tetrazole moiety is still close to S658.
6
To further study the variability on the apical portion (R substituent, Figure 5) we investigated
6
the SAR of some tetrazole-core compounds with a couple of R aryl substructures with high
similarity (such as the phenyl and the pyridyl groups).
Both (average) identified conformations show an interesting feature: a residue with an acidic
functionality can be detected in the direction of an ideal vector exiting from para position of the
phenyl terminal substructure: D497 (D460 in zCD2) and D675 (E638 in zCD2) for the first and
second pose, respectively.
ACS Paragon Plus Environment
2
2

Journal of Medicinal Chemistry
Page 24 of 106
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
Figure 7. Binding conformers of compound 35 (in green) detected in docking experiment
(hHDAC6 CD2 PDB code 5EDU). The interaction between the aniline cap term and Asp497
contributes to stabilization of the inhibitor into the enzyme binding pocket (yellow surface).
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
The presence of a local positive charge attractors (loops L1 and L2) suggests the introduction
of an ionizable group in the para position of aryl moiety, as shown, by way of example, in Figure
7 for compound 35 (first pose). This hypothesis was confirmed by data related to all compounds
bearing a positively ionizable group in that position, as evident in table 7.
Substituents in para- or meta- position with high electron density (-CF , -SF , -COOH) tend to
3
5
slightly reduce potency and selectivity (compare compounds 22–25, Table 7 with compound 7c,
Table 4). On the contrary, amino groups seem to slightly increase activity on HDAC6 down to
one-digit nanomolar (compounds 26 and 27) although selectivity decreases when the nitrogen
atom is not directly bound to the aryl substructure.
Table 7. Tetrazole-based inhibitors bearing a range of phenyl and pyridinyl substituents.
N
R6
N
H
N
N
N
OH
O
Sel
6 vs 3
HDAC6a
HDAC6a
Sel
vs 3
Compd
R6
Compd
28
R6
b
b
6
F
F
F
2
2
112
51
91
189
N
ACS Paragon Plus Environment
2
3

Page 25 of 106
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
F
S
F
F
F
F
F
N
2
3
252
50
29c
72
126
N
F
S
F
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
F
F
2
2
4
5
263
121
35
43
30
31
201
201
7
N
HO
N
132
O
H N
2
2
2
6
7
11
21
75
30
7d
32
521
254
26
H N
2
157
N
a – Enzymatic data (IC ) in nM unit were obtained from curve-fitting of a 5-point enzymatic
assay starting from 100, 30 or 10 μM with 10-fold serial dilution. Experiments were done in
triplicate (single experiment). SDs were calculated on technical replicates.
5
0
퐻퐷퐴퐶3 퐼퐶50
b - Selectivity indicator, expressed in terms of ratio between IC values (퐻퐷퐴퐶6 퐼퐶50
)
. See the
5
0
SI spreadsheet for the original data and the statistical evaluation.
c - HDAC6 IC value is the mean of three independent experiments. SD was calculated on the
5
0
independent experiments.
If the phenyl ring is then replaced by a pyridyl moiety (compound 28 vs compound 7c),
selectivity increases up to a factor of 189 with a small loss in the potency. The pyrimidinyl
substituent (compound 29) shows a similar behavior, whereas the inclusion of a methylene group
between the tetrazole and pyridinyl substructure causes a decrease in potency and an increase in
selectivity (compare compound 32 and compound 7d). The remarkable selectivity exhibited by
pyridyl and pyrimidinyl analogues could be related to new interactions in the HDAC6 isoform
ACS Paragon Plus Environment
2
4

Journal of Medicinal Chemistry
Page 26 of 106
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
triggered by the nitrogen atom of the ring (extra hydrogen bond) or by an enhancement of a 
or cation- non-local bond already available for phenyl-based inhibitors. Unfortunately, docking
results did not confirm this hypothesis. If the pyridinyl group is replaced by a larger moiety, as a
quinoline or an iso-quinoline (compounds 30 and 31, respectively), a moderate decrease in
potency is observed.
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
Data in Table 2 clearly show that the benzohydroxamic substructure has a role in the
2
4
selectivity for HDAC6. This observation suggests that modifications of the aromatic ring could
improve the selectivity for this isoform. The excellent potency and selectivity of ACY-73843
(Figure 1), where a pyrimidine is adopted instead of a benzene ring, confirm this hypothesis.
Fluorine atoms, due to their small volume and high electronegativity, are able to generate, when
introduced in 1,3-position, an electrostatic field equivalent to that generated by the pyrimidine
nitrogens.44
The hydrogen to fluorine substitution is extensively adopted in drug discovery and almost 20%
4
5
of drugs actually available on the market exhibit at least one fluorine atom, despite the lack of a
clear rationale. Several investigations pointed out that fluorine can hardly act as hydrogen bond
4
6
acceptor or trigger a halogen bond, whereas computational studies at high level of theory drew
opposite conclusions.47
The superimposition (see the Supporting Information) of the complex obtained by the best
pose of compound 35 in h-CD2 (Figure 7) and the experimental active conformer detected for
Bavarostat and ACY-1083 (PDB codes 6DVO and 5WGM, respectively – see the Supporting
Information) indicate that fluorinated phenyl and pyrimidine moieties adopt a very similar
ACS Paragon Plus Environment
2
5

Page 27 of 106
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
position, with fluorinated substrate exhibiting the F atom pointing toward Ser568, a convenient
orientation at a favorable distance for both hydrogen and halogen bonds.
Moreover, compound 35 shows the other fluorine atom close to Leu749 (2.54 A), that might
be stabilized by a possible weak interaction between the halogen and the non-polar hydrogens of
leucine, a feature detected for the Carbonic Anhydrase II and 2,3,4,5,6 Pentafluoro-SBB
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
4
8
complex using QM/MM methods. If h-CD2 HDAC6 and hHDAC1 (5ICN) are superimposed,
the conserved leucine in isoform 1 (Leu271) appears too close to the cleft where the
difluorophenyl substructure lies, suggesting a shape of HDAC1 binding site unfavorable for the
weak bond C-F•••H.
The first group of difluorine-substituted analogues took into consideration only the 2,5
disubstituted tetrazole scaffold. The results summarized in Table 8 show a systematic relevant
increase in selectivity for all reported fluorinated compounds compared to the non-fluorinated
analogues, whereas the potency on HDAC6 is substantially unaltered (compare, for example, the
quinoline analogues 30 and 38, Table 8).
Table 8. Effects of 3,5-fluoro substitution of benzohydroxamate moiety, when a tetrazole ring is
adopted as a scaffold.
X
N
R6
N
H
N
N
N
X
OH
O
Sel
vs 3
Sel
6 vs 3
R6
Compd X HDAC6a
Compd X HDAC6a
b
b
6
ACS Paragon Plus Environment
2
6

Journal of Medicinal Chemistry
Page 28 of 106
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
33 c
2
8
H
H
91
72
189
126
F
F
71
51
921
967
N
N
N
2
9c
34 c
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
H N
2
35 c
2
6
7
H
H
11
21
75
30
F
F
32
42
210
96
H N
2
2
36
37
N
31
H
H
132
7
F
F
676
58
2
01
221
101
3
0
201
38
N
a – Enzymatic data (IC ) in nM unit were obtained from curve-fitting of a 5-point enzymatic
assay starting from 100, 30 or 10 μM with 10-fold serial dilution. Experiments were done in
triplicate (single experiment). SDs were calculated on technical replicates.
5
0
퐻퐷퐴퐶3 퐼퐶50
b - Selectivity indicator, expressed in terms of ratio between IC values (퐻퐷퐴퐶6 퐼퐶50
)
. See the
5
0
SI spreadsheet for the original data and the statistical evaluation.
c - HDAC6 IC value is the mean of three independent experiments. SD was calculated on the
5
0
independent experiments.
The same difluorine motif was then assembled on the other heteropentacyclic scaffolds (such
as triazoles, oxadiazoles, and thiadiazoles) and a similar behavior was observed regardless of the
scaffold used (data not shown).
The effects of fluorine atoms on the other positions of the phenyl ring were also investigated
(
Table 9). Both the tetrafluoro-substituted analogue (compound 40, Table 9) and the monofluoro
ACS Paragon Plus Environment
2
7

Page 29 of 106
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
in the ortho position to the hydroxamate moiety (compound 41) exhibited a heavy drop in
potency and in selectivity, as already observed in other examples in the literature.18
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Table 9. Effects of fluorine substitution on benzohydroxamate.
R5
S
R7
O
N
N
N
H
N
N
N
R4
OH
R8
Sel
vs 3
71
Compd R4 R5 R7 R8 HDAC6a
a
6
1
4
H
F
H
F
H
H
F
H
H
F
51
62
3
9c
413
58
7
4
0
F
F
252721
223819
41
H
H
H
F
a – Enzymatic data (IC ) in nM unit were obtained from curve-fitting of a 5-point enzymatic
5
0
assay starting from 100, 30 or 10 μM with 10-fold serial dilution. Experiments were done in
triplicate (single experiment). SDs were calculated on technical replicates.
퐻퐷퐴퐶3 퐼퐶50
b - Selectivity indicator, expressed in terms of ratio between IC values (퐻퐷퐴퐶6 퐼퐶50
)
. See the
5
0
SI spreadsheet for the original data and the statistical evaluation.
c - HDAC6 IC value is the mean of three independent experiments and SD was calculated on
5
0
the independent experiments.
On the basis of the results and the considerations discussed above, a large number of
4
9
compounds were finally synthesized and tested (data not shown). Compounds with the best
profiles, in terms of potency, selectivity and stability, are shown in Table 10.
Table 10. The most promising HDAC6 inhibitors.
ACS Paragon Plus Environment
2
8