Benzofused hydroxamic acids: Useful fragments for the preparation of histone deacetylase inhibitors. Part 1: Hit identification

Article abstract of DOI:10.1016/j.bmcl.2013.05.053

In the search for a new class of histone deacetylase inhibitors, we prepared a series of simple benzofused hydroxamic acids to find an anchoring fragment of minimal molecular weight. These initial hits, all belonging to the benzothiophene class, showed very good ligand efficiencies. Following these findings, a classical fragment growing approach was performed to increase binding affinity and cytotoxicity.

Full text of DOI:10.1016/j.bmcl.2013.05.053

Bioorganic & Medicinal Chemistry Letters xxx (2013) xxx–xxx
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Benzofused hydroxamic acids: Useful fragments for the preparation
of histone deacetylase inhibitors. Part 1: Hit identification
Elena Marastoni a, Sandra Bartoli a, Marco Berettoni a, Amalia Cipollone a, Alessandro Ettorre a,
Christopher I. Fincham a, Sandro Mauro a, Marielle Paris a, Marina Porcelloni a, Mario Bigioni a,
Monica Binaschi a, Federica Nardelli a, Massimo Parlani a, Carlo A. Maggi b, Daniela Fattori a,
⇑
a Menarini Ricerche, Via Tito Speri 10, 00040 Pomezia, Rome, Italy
b Menarini Ricerche, Via Rismondo 12A, 50131 Florence, Italy
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 8 March 2013
Revised 14 May 2013
Accepted 15 May 2013
Available online xxxx
Dedicated to Piero, a colleague, a friend
In the search for a new class of histone deacetylase inhibitors, we prepared a series of simple benzofused
hydroxamic acids to find an anchoring fragment of minimal molecular weight. These initial hits, all
belonging to the benzothiophene class, showed very good ligand efficiencies. Following these findings,
a classical fragment growing approach was performed to increase binding affinity and cytotoxicity.
Ó 2013 Elsevier Ltd. All rights reserved.
Keywords:
HDAC
Hydroxamic acids
Fragment approach
Histone deacetylases (HDACs) are zinc dependent hydrolases
that mediate chromatin remodelling and gene expression by re-
moval of acetyl groups from histone lysine residues. Histone hyper-
acetylation, inducedby HDACinhibitors, correlates with modulation
of gene expression, cell-cycle arrest, differentiation and apoptosis in
tumor cells and in recent years, HDACs have become important tar-
gets for the treatment of a number of cancers.1 Currently there are
several HDAC inhibitors in clinical trials2 and two on the market.3
Based on their homology to the yeast HDACs, mammalian HDACs
can be categorized into four classes. Mechanistically, Classes I, II
and IV are distinct from Class III in their co-factor requirement, in
that the former require zinc in the active site to mediate deacetyla-
tion catalysis, while the latter is NAD+2 dependent. Virtually all of
the known inhibitors target the Zn2+ catalytic domain of Class I, II
and IV HDACs and are only slightly selective or non-selective. They
fall into several distinct structural classes: hydroxamic acids, such
as SAHA, aminobenzamides, short chain fatty acids, cyclic peptides.4
As part of an internal project aimed at the identification of new
and proprietary classes of HDAC inhibitors, we decided to apply the
fragment approach philosophy.5 This involves first finding a small
fragment able to show binding affinity in our test and then to
evolve this fragment. We targeted molecular interactions only with
the zinc atom and the active site cavity, planning to optimize
pharmacokinetic and pharmacodynamic properties of the selected
hits via systematic functionalization.
4
3
5
OR
O
HN OH
O
a or b
R'
R'
6
S
S
7
2
1
HN OH
HN OH
O
HN OH
S
S
S
O
O
Me
Me
F
5
3
4
Cl
HN OH
O
HN OH
O
HN OH
O
S
S
S
Cl
6
CF3
7
8
OMe
HO
Me
HN OH
MeO
HN OH
HN OH
O
S
O
S
O
S
Me
11
10
9
HN OH
O
HN OH
O
S
HN OH
O
S
S
S
S
13
14
12
⇑
Corresponding author. Tel.: +39 06 91184522.
E-mail address: dfattori@menarini-ricerche.it (D. Fattori).
Scheme 1. Reagents and conditions: (a) NaOH, H2O/THF and then NH2OH, HOAt,
EDAC, DIPEA; (b) NH2OH, NaOH, MeOH.
0960-894X/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.bmcl.2013.05.053
Please cite this article in press as: Marastoni, E.; et al. Bioorg. Med. Chem. Lett. (2013), http://dx.doi.org/10.1016/j.bmcl.2013.05.053

2
E. Marastoni et al. / Bioorg. Med. Chem. Lett. xxx (2013) xxx–xxx
Table 1
From the data reported in literature and a visual inspection of
the various X-ray structures of HDACs complexed with various
inhibitors,6,7 a structural feature caught our attention: two phenyl
alanine side chains (Phe 152 and Phe 208 in HDAC-8) form part of
the wall of the binding cavity and are almost parallel. We reasoned
that these two aromatic rings could be used to gain binding energy
IC50 values for compounds 3–14 and ligand efficiencies (LEs) calculated according to
Ref. 6
Compd
IC50
(
l
M)
LE
3
4
5
6
7
8
9
10
11
12
13
14
2.17 0.65
1.76 0.52
0.458 0.169
2.91 0.70
9.19 3.73
11.63 1.40
3.79 1.31
1.35 0.42
0.989 0.151
1.08 0.28
0.557 0.185
76.13 26.53
0.59
0.56
0.58
0.54
0.40
0.29
0.49
0.40
0.54
0.68
0.71
0.43
from p–p interactions via the insertion of an aromatic fragment.
Given these assumptions we hypothesised that benzofused
derivatives containing a hydroxamic acid were the initial frag-
ments of choice. A docking of benzothiophene-2-hydroxamic acid
into the active site of HDAC-8 showed that the hydroxamic acid
moiety could complex the Zn atom without any steric clash with
the heterocycle’s sulfur atom, and that the benzene ring of the ben-
xothiophene was able to sandwich between the two phenyl
alanine aromatic side chains. These positive docking results
RHN
RHN
a or b
HN OH
O
OMe
H2N
OMe
c or d
S
S
O
S
O
16a R = acyl
16b R = alkyl
17a R = acyl
15
17b R = alkyl
OMe
O
OMe
O
HN OH
c or d
RHN
a or b
RHN
S
S
X
S
O
20a R = acyl
18
X = NO2
d
21a R = acyl
21b R = alkyl
20b R = alkyl
19 X = NH2
Scheme 2. Reagents and conditions: (a) EDAC, HOAt, RCO2H; (b) RCHO, NaBH3CN; (c) KOH, NH2OH–HCl, MeOH/THF; (d) NaOH, H2O/THF and then NH2OH–HCl, EDAC, HOAt,
DMF.
HN
O
O
OH
O
OH
O
FmocHN
ONH2
b
H2N
FmocHN
a
S
S
S
24a
5-amino
22a
23a
5-amino
5-amino
24b 6-amino
22b 6-amino
23b 6-amino
O
R
HN
O
O
HN
O
O
d
H2N
c
HN
e
17a 21a
or
S
S
25a
5-amino
26a
5-amino
25b 6-amino
26b 6-amino
Scheme 3. (a) Fmoc-N-hydroxysuccinimide; (b) DIPC, HOAt, pyridine, hyroxylamine chlorotrityl resin; (c) 20% piperidine in DMF; (d) DIPC, HOAt, pyridine; (e) TES, TFA, DCM.
OMe
H2N
S
e
O
f
32
X = H
OMe
O
ROCHN
OMe
O
S
OMe
O
a
Br
33
X = H
S
S
Me
b
27
X = H
d
28
X = H
c
OMe
O
OMe
O
NR'R
S
S
S
OMe
O
29
ArO
F
S
31
X = H
30
Scheme 4. (a) NBS, CCl4; (b) p-F-thiophenol; (c) ArOH, Cs2CO3; (d) NHRR’; (e) NaN3, DMF and then H2, Pd/C, THF; (f) RCO2H and coupling agent.
Please cite this article in press as: Marastoni, E.; et al. Bioorg. Med. Chem. Lett. (2013), http://dx.doi.org/10.1016/j.bmcl.2013.05.053

E. Marastoni et al. / Bioorg. Med. Chem. Lett. xxx (2013) xxx–xxx
3
Table 2
HDAC inhibition of amines 34–39 and amides 40–63, and cytotoxicity on HCT-116 cells
HN OH
O
R
S
Compd
R
Inh at (%)
IC50
Compd
R
Inh at (%)
IC50
0.1
48
71
28
58
28
32
28
lm
1.0
82
85
68
79
61
73
69
l
m
HCT-116
0.1
24
73
55
l
m
1.0
78
87
78
lm
HCT-116
H O
N
H O
5
34
35
36
37
38
39
40
N C CH2
1.3
0.10
nt
6
N C CH2
50
51
52
nt
N
H
H O
6
5
N C CH2
H
H O
S
6
6
N
CH2
N C CH2
0.11
nt
H
N
H O
6
5
CH2
1.3
nt
N C CH2
O
H
N
O
CH2CH2
CH2CH2
H
N
H O
6
6
nt
53
54
N C CH2
NMe2
OMe
57
75
82
84
0.50
0.49
6
5
6
H O
H O
N C CH2CH2
nt
N C CH2
H O
OMe
N C CH2CH2
41
42
76
2.6
55
H O
65
83
0.6
6
N C CH2
Me
O
MeO
5
N C CH2CH2
42
43
44
27
21
39
67
63
74
nt
nt
nt
56
57
58
H O
54
50
70
81
79
85
0.55
2.4
6
6
N C CH2
Me
O
H O
N C CH2CH2CH2NMe2
6
5
N C CH2CH2
N
H O
N C
H
C
H O
N C CH2
0.5
6
6
6
6
6
6
NH2
N
H O
N C
H
C
H O
6
5
N C CH2
45
46
47
48
49
59
49
62
49
66
78
7
0.48
17
59
60
61
62
63
69
41
54
44
71
85
77
81
78
85
0.77
0.34
5.0
NH2
N
N
H O
H O
N C CHCH2
NH2
N C CH2
H O
H O
N C CHCH2
6
5
N C CH2
83
78
84
3.0
8.6
1.8
NH2
H O
O
H
N C CH2
N
N C CHCH2CH2
NH2
0.32
3.4
H O
O
H
6
N C CH2
N
N C CHCH2CH2
NH2
prompted us to form a collection of aromatic bicyclic compounds
containing a thiophene and a carboxylic acid, available commer-
cially, and in our chemical archives, and to synthesize the corre-
sponding hydroxamic acids (Scheme 1, compounds 3–14).8
The benzothiophene hydroxamic acids 3–14 were prepared
from the corresponding carboxylic acids through coupling with
hydroxylamine in the presence of N-(3-dimethylaminopropyl)-N0-
ethylcarbodiimide (EDAC) and 7-aza-1-hydroxybenzotriazole
(HOAt), and then profiled using partially purified HDAC enzyme
obtained from HeLa cells.9 The IC50 values and ligand efficiencies
(calculated using the method of Rees co-workers.10) for these ini-
tial fragments are reported in Table 1.
position appearing the more sensitive of the two. Functionalisation
of the 5 and 6 positions (singularly or together) was tolerated and
this prompted us to use these positions for the introduction of new
functional groups aimed at finding new interactions within the ac-
tive site. This was carried out either via the introduction of amino
groups at the 5 and 6 positions of scaffold 3, or via the modification
of the methyl group in compounds 4 and 5. The subject of commu-
nication is the study done on 3 and 4, which were selected as start-
ing point for the easiness of modification. The information
obtained was then used for the synthesis of the 7-fluorobenzothi-
ophene derivatives, which will be the subject of a subsequent
report.
These data clearly showed that the 2 position was better than
the 3 position from the point of view of the hydroxamic acid moi-
ety, and that while some substituents were tolerated in the 4 and 7
positions, it seemed that there was a limit to their size, with the 7
The 5- and 6-aminobenzothiophene-2-carboxylate methyl es-
ters 15 and 19 (the latter prepared by catalytic hydrogenation of
the commercially available 6-nitro derivative 18) were submitted
to reductive amination using cyanoborohydride11 to obtain amines
Please cite this article in press as: Marastoni, E.; et al. Bioorg. Med. Chem. Lett. (2013), http://dx.doi.org/10.1016/j.bmcl.2013.05.053

4
E. Marastoni et al. / Bioorg. Med. Chem. Lett. xxx (2013) xxx–xxx
Table 3
16b and 20b, or coupled with an appropriate carboxylic acid
(EDAC, HOAt) to obtain the corresponding amides 16a and 20a.
Treatment of 16a/b and 20a/b with a methanol solution of NaOH
and hydroxylamine hydrochloride, or submission to basic hydroly-
sis followed by coupling with hydroxylamine, afforded the
hydroxamate derivatives 17a/b and 21a/b (Scheme 2).
Subsequently, we developed a solid phase method for the syn-
thesis of benzothiophene amides 17a and 21a and this allowed us
to speed up the SAR process (Scheme 3). The free amine of 5- or 6-
amino benzothiophene-2-carboxylic acids was protected with an
Fmoc group through treatment with Fmoc-N-hydroxysuccinimide
to produce acids 23a/b, which were loaded onto hydroxylamine
chlorotrityl resin using standard conditions.
Inhibition of compounds 64–76 and cytotoxicity on HCT-116 cells
HN OH
R
S
O
Compd
R
% Inh at
1.0
IC50
0.1
l
m
l
m
HCT-116
OH
OH
64
52
53
80
76
0.83
1.7
N
N
N
65
Treatment with a 20% piperidine solution in dimethylformam-
ide (DMF) released the amino group, which was then coupled to
a series of carboxylic acids. The use of pyridine as a base in the sub-
sequent coupling reaction was necessary, since with diisopropyl-
66
67
53
93
83
87
1.3
HN
0.25
ethyl amine (DIPEA) consistent amounts of
a side product,
O
deriving from acylation of the hydroxamate nitrogen, were ob-
served. Finally, cleavage from the resin with a trifluoroacetic acid
(TFA)/triethylslane (TES) solution in dichloromethane (DCM),
afforded the corresponding hydroxamic acids, in many cases pure
enough to be tested.
An entry to 6-benzyl derivatives of thiophene-2-carboxylate
was obtained by radical bromination of methyl derivative 27 with
N-bromosuccinimide (NBS), followed by nucleophilic substitution
with an aromatic thiol (29), a phenol (30), a secondary amine
(31) or the azide ion which, after catalytic reduction gave primary
benzylamine 32 (Scheme 4).
H
N
68
60
88
1.2
O
HN
69
70
89
72
92
84
0.07
0.73
O
O
HN
N
O
OH
H
N
N
71
77
91
1.4
Coupling with an appropriate carboxylic acid under standard
conditions gave amide derivatives 33.
O
72
73
<20
23
31
68
nt
nt
F
F
O
In a first set of compounds (34–39), a non-functionalized phe-
nyl ring was linked to the benzothiophene in both the 5 and 6 posi-
tions. The distance of the ring from the heterocyclic nucleus was
modulated with carbon chains of different lengths, while the use
of amines, amides and methylamides modulated slightly the
geometry of the side chain versus the heteroaromatic moiety.
The newly prepared compounds were evaluated for their ability
O
O
74
75
22
58
24
nt
nt
F
to inhibit HDAC in a two point experiment (1.0 and 0.1
lM). Those
having a inhibition >40% at a concentration of 0.1 M were also
l
<20
F
S
tested for their cytotoxicity on HCT-116 cells.12 From this data
three things were clear: first that the functionalization in the 6 po-
sition of the benzothiophene scaffolds generally afforded higher
potency than the same substituent in the 5 position; second that
the amides, if not methylated, were clearly superior to amines,
and third that among the amides, phenylacetamide was best.
Taking advantage of this information, a second set of com-
pounds was prepared: all the isomers of pyridyl acetic acid were
linked at the amino group in position 5 (44, 46 and 48) and 6
(45, 47 and 49) of the benzothiophene nucleus, with the aim of
increasing the aqueous solubility, which for 34–43 was not satis-
factory. The binding affinities were generally good for all six mol-
ecules, although there was a drop in cellular activity, very likely
due to decreased cell membrane permeability. Solubility remained
low, ranging from 0.004 to 0.05 mg/mL, and this we hypothesised
O
CN
H2N
N
H
S
76
<20
<20
nt
amino and an aromatic group: both enantiomers of phenyl glycine
(58, 59), phenylalanine (60, 61) and homophenylalanine (62, 63)
were prepared. While the two enantiomers of phenylglycine were
equipotent in the enzyme and in the cell test, with both phenylal-
anine and homophenylalanine the
the isolated enzymes and the -isomers more active in the cell-test.
The reason for this difference is as yet unclear (see Table 2)
Among the aminomethyl derivatives (64–76, Table 3), the
amides showed better activity than the amines, ethers, thioethers
and ureas, with compound 69 showing activity in the nanomolar
range in the HCT-116 cell test.
Finally a xenograft efficacy study using HCT-116 tumor cells in
CDI mice was carried out with compound 35. It was dosed to fe-
male CDI nude (nu/nu) athymic mice at doses of 50 and 100 mg/
kg os (n = 4 per group) for 21 days. The control vehicle was 0.5%
CMC (10 mL/kg). This study resulted in a TVI of 26% at a dose of
50 mg/kg.
D-isomers were more active on
L
was due to p-stacking.
An additional group of compounds was prepared with hetero-
aryl amides in the 6 position in place of the substituted phenylace-
tic amides (50–56): they showed no significant improvements,
either in the ex-cell and cell tests, or in aqueous solublity. We were
pleasantly surprised to find that the introduction of a simple ter-
tiary aliphatic amine produced a compound (57) with reasonable
potency and solubility (1.31 mg/mL in aqueous solution at pH
7.4). As a result of this we introduced onto the 6-amino group of
the benzothiophene scaffold substituents containing both a free
Please cite this article in press as: Marastoni, E.; et al. Bioorg. Med. Chem. Lett. (2013), http://dx.doi.org/10.1016/j.bmcl.2013.05.053

E. Marastoni et al. / Bioorg. Med. Chem. Lett. xxx (2013) xxx–xxx
5
been published. We could have easily deleted them from the present paper, but
we think that it is important to report their activity to complete the SAR.
9. The compounds were evaluated for their ability to inhibit isolated HDACs in a
commercially available assay (Fluor de Lys Assay System, Biomol International
LP, Plymouth Meeting PA), employing SAHA and LAQ-xx as positive controls.
In summary, starting from a selection of small, rigid fragments
derived from benzothiophene, we have optimized a series of 5- and
6-substituted benzothiophene-2-hydroxamates that display signif-
icant potency in vitro. One of the first derivatives prepared, 35, was
also shown to be efficacious after oral administration in a mouse
HCT-116 xenograft study.
The reaction mixture, 50 lL run in a 96 well plates, contains a HeLa cell nuclear
extract and a commercial substrate containing acetylated lysine side chains.
The substrate and extracts are incubated in the presence of the appropriate
concentration of the inhibitor. Deacetylation of the substrate followed by
mixing with the provided developer generates a fluorophore, The release of the
fluorophore was monitored with a Victor 1420 fluorescent plate reader set at
excitation/emission wavelength of 335/460 nm. The activity of compounds was
expressed as IC50 (drug concentration causing a 50% inhibition of enzymatic
activity) and calculated with Easy-fit software application. All the experiment
were carried out in triplicate and reported values are the average of those
determinations.
This series was further elaborated and the results will be re-
ported in a subsequent paper.
References and notes
1. Marks, P. A.; Breslow, R. Nat. Biotechnol. 2007, 84.
2. Wagner, J. M.; Hackanson, B.; Lübbert, M.; Jung, M. Clin. Epigenetics 2010, 1,
117.
3. Bolden, J. E.; Peart, M. J.; Johnstone, R. W. Nat. Rev. Drug Disc. 2006, 5, 769.
4. Pontiki, E.; Hadjipavlou-Litina, D. Med. Res. Rev. 2012, 1.
5. Fattori, D. Drug Discovery Today 2004, 9, 229.
6. Vannini, A.; Volpari, C.; Filocamo, G.; Casavola, E. C.; Brunetti, M.; Renzoni, D.;
Chakravarty, P.; Paolini, C.; De Francesco, R.; Gallinari, P.; Steinkhuler, C.; Di
Marco, S. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 15064.
7. Finnin, M. S.; Donigian, J. R.; Cohen, A.; Richon, V. M.; Rifkind, R. A.; Marks, P. A.;
Breslow, R.; Pavletich, N. P. Nature 1999, 188.
8. While we were well advanced in our work on this series, analogous compounds
were published by Merck [Bioorg. Med. Chem. Lett. 2007, 17, 4562].
Nevertheless, we decided to write the current paper since the SAR data
(including in vivo activity data) from the current work is much larger than that
in the Merck publication and this may be of interest to readers. Some of the
compounds presented in this paper, namely 34–39, 54, 67 and 68, have already
10. Carr, R. A. E.; Congreve, M.; Murray, C. W.; Rees, D. Drug Discovery Today 2005,
10, 987.
11. Borch, R. F.; Bernstein, M. D.; Dupont Durst, H. . J. Am. Chem. Soc. 1971, 93, 2897.
12. Colon cancer HCT-116 cells were plated in a 96-well tissue culture plates
containing 200
added at different concentrations, ranging from 0.1 to 100
quadruplicate. After 5 days, 20 L of Alamar blue were added to each well
l
L of complete medium. After 24 h HDAC inhibitors were
l
M
in
l
and the plates were further incubated for 4 h. The chemical reduction of
Alamar Blue in the growth medium is a fluorometric/colorimetric indicator of
cellular growth based on the detection of metabolic activity. Fluorescence was
monitored in
a multilabel counter Victor 1420 at 530 nm excitation and
590 nm emission wavelength. All results were expressed as IC50 (drug
concentration causing a 50% inhibition of cellular proliferation) calculated
using the Easy-fit software.
Please cite this article in press as: Marastoni, E.; et al. Bioorg. Med. Chem. Lett. (2013), http://dx.doi.org/10.1016/j.bmcl.2013.05.053