Amino acid derivatives as histone deacetylase inhibitors

Article abstract of DOI:10.1016/j.bmcl.2007.11.017

Ongoing clinical studies indicate that inhibitors of Class I and Class II histone deacetylase (HDAC) enzymes show great promise for the treatment of cancer. Zolinza (SAHA, Zolinza) was recently approved by the FDA for the treatment of the cutaneous manifestations of cutaneous T-cell lymphoma. As a part of an ongoing effort to identify novel small molecules to target these important enzymes, we have prepared several classes of amino acid-derived HDAC1 inhibitors. The design rationale and in vitro activity against the HDAC1 enzyme and HCT116 cell line are described in this letter.

Full text of DOI:10.1016/j.bmcl.2007.11.017

Available online at www.sciencedirect.com
Bioorganic & Medicinal Chemistry Letters 18 (2008) 34–38
Amino acid derivatives as histone deacetylase inhibitors
Jed L. Hubbs,a,* Hua Zhou,a Astrid M. Kral,b Judith C. Fleming,b
William K. Dahlberg,c Bethany L. Hughes,c Richard E. Middleton,c
Alexander A. Szewczak,c J. Paul Secristb and Thomas A. Millera
aDepartment of Drug Design and Optimization-Medicinal Chemistry, Merck Research Laboratories,
33 Avenue Louis Pasteur, Boston, MA 02115, USA
bDepartment of Cancer Biology and Therapeutics, Merck Research Laboratories, 33 Avenue Louis Pasteur, Boston, MA 02115, USA
cDepartment of Drug Design and Optimization-Automated Lead Optimization, Merck Research Laboratories,
33 Avenue Louis Pasteur, Boston, MA 02115, USA
Received 25 October 2007; revised 6 November 2007; accepted 7 November 2007
Available online 13 November 2007
Abstract—Ongoing clinical studies indicate that inhibitors of Class I and Class II histone deacetylase (HDAC) enzymes show great
promise for the treatment of cancer. ZolinzaÒ (SAHA, ZolinzaÒ) was recently approved by the FDA for the treatment of the cuta-
neous manifestations of cutaneous T-cell lymphoma. As a part of an ongoing effort to identify novel small molecules to target these
important enzymes, we have prepared several classes of amino acid-derived HDAC1 inhibitors. The design rationale and in vitro
activity against the HDAC1 enzyme and HCT116 cell line are described in this letter.
Ó 2007 Elsevier Ltd. All rights reserved.
Histone deacetylase (HDAC) enzymes are becoming an
O
H
N
increasingly important therapeutic target for the treat-
ment of cancer and other diseases. These enzymes cata-
lyze the removal of acetyl groups from the lysine
residues of proteins including histones. In some cancer
cells, there is an overexpression of HDACs causing hyp-
oacetylation of histones.1 Hypoacetylation of histones is
associated with a condensed chromatin structure and
repression of gene transcription.1 Inhibition of HDAC
activity allows for the accumulation of acetyl groups
on the histone lysine residues resulting in an open chro-
matin structure and transcriptional activation.1 In vitro,
the HDAC inhibitor ZolinzaÒ causes the accumulation
of acetylated histones and induces cell cycle arrest and/
or apoptosis of some transformed cells.2 In vivo xeno-
graft studies have demonstrated many of these agents
are effective in inhibiting tumor cell growth.3 In the
clinic, ZolinzaÒ was effective in decreasing skin lesions
resulting from cutaneous T-cell lymphoma and was re-
cently approved for treatment of the disease.4 In addi-
tion to ZolinzaÒ, a wide range of structures inhibit the
activity of Class I/II HDAC enzymes. These inhibitors
OH
N
H
Zolinza®
Linker
O
Surface
Metal
Recognition
Binding
Figure 1. Pharmacophore for HDAC inhibitors.
can be characterized by a common pharmacophore
comprised of metal binding, linker, and surface recogni-
tion domains (Fig. 1).5
Ongoing efforts to find HDAC1 inhibitors more potent
than ZolinzaÒ (1, IC50 ꢀ 48 nM) led to the development
of aminosuberoyl hydroxamic acids such as 2.6 These
inhibitors were prepared based on the notion that the
amino acid moiety in the surface recognition domain
more closely mimics the natural protein substrates for
the HDAC1 enzyme. Concurrent efforts, based on the
premise that the linker region could itself impart favor-
able interactions with the enzyme, led to the develop-
ment of benzothiophene hydroxamic acids exemplified
by 3.7 We began our work with the idea that favorable
pharmacological properties conveyed by these two
improvements could be combined. We found that this
was the case when we prepared benzothiophene deriva-
Keywords: Histone deacetylase; Cancer; Ugi reaction; Lysine; Amino
acid; Benzothiophene.
*
Corresponding author. E-mail: jed_hubbs@merck.com
0960-894X/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2007.11.017

J. L. Hubbs et al. / Bioorg. Med. Chem. Lett. 18 (2008) 34–38
35
Bz
HN
O
H
N
OH
OH
Ph
N
H
O
O
O
H
N
H
N
OH
2 IC50 = 12 nM
PMPHN
S
OH
Ph
N
H
HN
O
O
Bn
H
N
H
N
Zolinza® (1)
4 IP = 4 nM
Bz
S
O
3 IC50 = 10 nM
O
O
NH2
H
N
R2
R3
R1
N
N
H
N
H
S
H
N
NH2
N
O
R1
R2
R3
n
O
27 Examples
13 Examples
Figure 2. The evolution of amino acid-derived HDAC1 inhibitors.
tive 4, which showed potent HDAC1 enzyme inhibition.
Having established that we could use this concept to
prepare highly potent compounds, we wished to exam-
ine other potential binding groups, particularly the 2-
aminoanilide. The preparation and structure–activity
relationship of benzothiophene derivatives with this
binding group and subsequent development of phenyl-
glycine- and phenylalanine-derived HDAC1 inhibitors
are described herein (Fig. 2).
Additionally, we focused on piperazine-based amino
acid amides because we expected them to have improved
physical chemical properties, and because they could be
synthesized more efficiently. These compounds could be
prepared by first carrying out the Ugi reaction of alde-
hyde 5 in the presence of a piperazine and TFA to give
the tertiary amino acid amide (Scheme 2). The anilide
could then be installed by LiOH hydrolysis followed
by coupling with 1,2-diaminobenzene to give the corre-
sponding HDAC1 inhibitor 7.
The desired benzothiophene amino acids were prepared
using an Ugi reaction-based strategy.8 In designing these
inhibitors, experience suggested that a basic amine in the
amino acid product would enhance cell permeability and
growth inhibition. For this reason, trifluoroacetic acid
(TFA) was used as the acid coupling partner in this mul-
ticomponent reaction in order to afford readily cleavable
trifluoroacetamide products. With this strategy in mind,
we prepared the amino acid derivatives as shown in
Scheme 1. Aldehyde 5 underwent Ugi reaction with an
isonitrile and a primary amine to give the amino acid tri-
fluoroacetamide. The TFA group was removed with
NaBH4 in MeOH then the methyl ester was hydrolyzed
with LiOH to give the acid. The acid, in turn, was coupled
with 1,2-diaminobenzene to give anilide 6.
Using this chemistry, a number of piperazine substitu-
ents (R2) were incorporated into the benzothiophene
scaffold to prepare analogs 7a–g, where R1 was held
constant as a benzyl group. The alkyl- and aryl-substi-
tuted compounds (7a–e) were more potent in the
HDAC1 assay than those with a carbamate substituent
(7f and 7g). The more potent methyl (7a) and ethyl (7b)
substituted compounds were screened against the hu-
man colon cancer cell line HCT116 and showed cell
growth inhibition in the 400 nM range at 72 h. We fur-
ther examined the effect of the amide substituent (R1) on
potency and found that aryl groups (7h–m) afforded
more potent compounds than the benzyl group, with
the 2-naphthyl-(7i) and 4-MePh-(7k) substituted com-
pounds giving the best enzyme inhibition and cell-
growth inhibition. Compound 7k was resolved to give
enantiomers 7l and 7m and isomer 7l showed excellent
HDAC1 inhibition and cell potency (Table 2).
These amino acid derivatives exhibited good inhibitory
potency against the HDAC1 enzyme, with similar re-
sults observed for the p-methoxyphenyl (PMP) and ben-
zyl amides.9 For example, PMP amide 6c and benzyl
amide 6g had an inflection point (IP) of 71 and
70 nM, respectively. Morpholinoethyl amide 6f was
slightly less potent. A focus on the benzyl amides iden-
tified compound 6m, bearing a 4-pyridylethyl substitu-
ent on the amino acid nitrogen, as the most potent
derivative with an IP of 39 nM (Table 1).
Concurrent efforts indicated that compounds with a
truncated linker could afford highly effective HDAC1
inhibitors. For example, malonyl substituted benzamide
8 showed excellent HDAC1 inhibition, with an IC50 of
41 nM, and good cell proliferation inhibition, with an
HCT116 IC50 of 170 nM.10 This prompted us to exam-
O
NH2
H
N
R1
H
OMe
a, b, c, d
N
H
S
S
HN
O
O
O
R2
5
6
Scheme 1. Reagents and conditions: (a) R1NC, R2NH2, TFA, MeOH; (b) NaBH4, MeOH; (c) LiOH, H2O, MeOH, THF; (d) 1,2-diaminobenzene,
EDC, HOBT, DMF.

36
J. L. Hubbs et al. / Bioorg. Med. Chem. Lett. 18 (2008) 34–38
Table 1. Benzothiophene amino acid derivatives-secondary amines
ine phenylglycine analogs of 8 as potential HDAC1
inhibitors.
O
NH2
H
N
R1
O
N
H
S
HN
O
O
O
N
H
R2
Ph
NH2
N
H
Compound
R1
R2
HDAC1
IPa (nM)
NH
Ph
8
6a
6b
6c
6d
6e
6f
PMP
PMP
PMP
PMP
PMP
PMB
c-C6H11
90
74
HDAC1 IC50 = 41 nM
HCT116-72hr IC50 = 170 nM
CH2CH2Ph
Bn
71
79
CH2CH2-4-pyridyl
CH2CH2Ph
32
230
The desired phenylglycine anilides could be prepared
using a similar Ugi strategy (Scheme 3). In this series,
we first pursued compounds where the amine of the amino
CH2CH2-1-
morpholino
6g
6h
6i
Bn
Bn
Bn
Bn
Bn
Bn
Bn
Bn
CH2CH2Ph
CH2CH2OMe
CH2CH2OH
70
170
150
150
520
180
39
O
O
6j
CH2CH2NMe2
CH2CH2-2-imidazolyl
CH2-3-pyridyl
OMe
O
N
H
a, b, c, d
R1
NH2
6k
6l
H
N
H
O
HN
6m
6n
CH2CH2-4-pyridyl
CH2CH2-2-pyridyl
R2
9
10
110
a IP values are from a single determination or the average of two
determinations.
Scheme 3. Reagents and conditions: (a) LiOH, H2O, MeOH, THF; (b)
N-Boc-1,2-diaminobenzene, EDC, HOBT, DMF; (c) R1NC, 2,4-
MeOPhCH2NH2, acid, MeOH; (d) TFA, CH2Cl2.
O
NH2
H
N
R1
a, b, c
H
OMe
N
H
S
S
N
O
O
O
5
7
N
R2
Scheme 2. Reagents and conditions: (a) R1NC, amine, TFA, trifluoroethanol; (b) LiOH, H2O, MeOH, THF; (c) 1,2-diaminobenzene, EDC, HOBT,
DMF.
Table 2. Benzothiophene amino acid derivatives: piperazines
O
NH2
H
N
R1
N
H
S
N
O
N
R2
Compound
R1
R2
HDAC1 IP (nM)
HCT116—72 h IPa (nM)
7a
7b
7c
7d
7e
7f
Bn
Bn
Me
Et
110
80
430
490
NDb
ND
ND
ND
ND
340
580
410
240
130
560
Bn
Bn
CH2CHCHPh
Ph
230
150
180
260
470
62
Bn
Bn
CH2CH2Ph
Boc
Cbz
7g
7h
7i
Bn
4-ClPh
2-Naphthyl
4-MeOPh
4-MePh ( )
4-MePh
4-MePh
Me
Me
25
43
7j
Me
7k
7l
Me
25
21
Mec
Med
7m
18
a IP values are from a single or the average of two determinations.
b Not determined.
c Enantiomer 1.
d Enantiomer 2.

J. L. Hubbs et al. / Bioorg. Med. Chem. Lett. 18 (2008) 34–38
37
Table 3. Phenylglycine derivatives
O
O
N
H
R1
NH2
N
H
HN
R2
Compound
R1
R2
HDAC1 IP (nM)
HCT116—72 h IPa (nM)
10a
10b
10c
10d
10e
10f
4-MePh
COCH2Ph
COCH2Ph
COCH2Ph
690
870
170
340
260
250
8100
6400
2700
3000
2800
2700
4-MeOPh
2-Naphthyl
4-MePh
COCH2CH2Ph
COCH2CH2Ph
COCH2CH2Ph
4-MeOPh
2-Naphthyl
a IP values are from a single determination or the average of two determinations.
acid was acylated—these compounds differ from the mal-
onates only by transposition of the carbonyl and the NH
moiety of one amide unit. These inhibitors were prepared
by first hydrolyzing ester 9 and then coupling to N-Boc-
1,2-diaminobenzene. This amide could then undergo
Ugi reaction with the appropriate acid and 2,4-dimeth-
oxybenzylamine to give the protected HDAC1 inhibitor.
The 2,4-dimethoxybenzyl and Boc protecting groups
could then be removed by treating with TFA in CH2Cl2
to give amide 10. These amino acid derivatives showed
good to moderate inhibition of the HDAC1 enzyme
(Table 3). However, they were relatively ineffective at
inhibiting cell proliferation in the HCT116 assay.
The cell-based activity imparted by the piperazine ring
in the benzothiophene series suggested that we should
examine its effect in the phenylglycine series. Such com-
pounds were prepared as illustrated in Scheme 4 in a
manner similar to the benzothiophenes.
We examined these piperazine-based compounds in the
HDAC1 and HCT116 assays (Table 4). Although these
inhibitors did not appear to be as potent as the amide-
substituted compounds in the HDAC1 enzymatic assay
(12c vs. 10c), they showed much better cell potency. At
this point, it seemed that the potency decrease observed
in these compounds could result from a truncated linker
relative to the benzothiophenes. With that in mind, the
one-carbon homologs, phenylalanines, were prepared
and a dramatic increase in potency was observed. For
example, compound 12g is approximately 10 times more
potent in the HDAC1 assay than compound 12c and is
about five times more potent in the cell proliferation assay.
R2
O
N
O
O
OMe
a, b, c, d
N
N
H
H
N
H
n
NH2
In conclusion, potent HDAC1 inhibitors were devel-
oped based on the premise that amino acid derivatives
with optimal linker domains mimic the natural sub-
strates of the HDAC1enzyme. We found that the benzo-
thiophene and phenylalanine series, in particularly,
afforded highly potent inhibitors, while the phenylgly-
R1
n
9, n = 0
O
11, n = 1
12
Scheme 4. Reagents and conditions: (a) R1NC, amine, TFA, trifluo-
roethanol; (b) LiOH, H2O, MeOH, THF; (c) 1,2-diaminobenzene,
EDC, HOBT, DMF.
Table 4. Phenylglycine and phenylalanine piperazine derivatives
R2
N
O
N
N
H
H
N
NH2
R1
n
O
Compound
R1
R2
n
HDAC1 IP (nM)
HCT116—72 h IPa (nM)
12a
12b
12c
12d
12e
12f
12g
4-ClPh
Bn
Me
Me
0
0
0
0
0
1
1
6200
860
360
620
680
310
37
ND
ND
640
ND
ND
370
220
2-Naphthyl
2-Naphthyl
2-Naphthyl
4-ClPh
Me
Ph
CH2CH2Ph
Me
Me
2-Naphthyl
a IP values are from a single determination or the average of two determinations.

38
J. L. Hubbs et al. / Bioorg. Med. Chem. Lett. 18 (2008) 34–38
Mosley, R. T.; Secrist, J. P.; Miller, T. A. Bioorg. Med.
Chem. Lett. 2007, 17, 5300; (b) Arts, J.; de Schepper, S.;
Emelen, K. V. Curr. Med. Chem. 2003, 10, 2343.
4. Grant, S.; Easley, C.; Kirkpatrick, P. Nat. Rev. Drug Disc.
2007, 6, 21.
5. Miller, T. A.; Witter, D. J.; Belvedere, S. J. J. Med. Chem.
2003, 46, 5097.
6. Belvedere, S.; Witter, D. J.; Yan, J.; Secrist, P.; Richon,
V.; Miller, T. A. Bioorg. Med. Chem. Lett. 2007, 17,
3969.
cine series was less potent. These results suggest that an
optimal linker length is necessary to maximize inhibitor
potency. We also found that piperazine-based amino
acids imparted the best inhibition of cell proliferation
across the different scaffolds. These principles are being
used to develop additional HDAC1 inhibitors. Those
analogs will be reported in due course.
References and notes
7. Witter, D. J.; Belvedere, S.; Chen, L.; Secrist, P.; Mosley, R.
T.; Miller, T. A. Bioorg. Med. Chem. Lett. 2007, 17, 4562.
8. Ugi, I.; Meyr, R.; Fetzer, U.; Steinbruckner, C. Angew.
Chem. 1959, 71, 386.
9. All diaminobenzene derivatives are less potent than
hydroxamic acid 4, this is consistent with previous results
which suggest that hydroxamic acids are more potent
HDAC1 inhibitors.
10. Siliphaivanh, P.; Harrington, P.; Witter, D. J.; Otte, K.;
Tempest, P.; Kattar, S.; 0, A. M.; Fleming, J. J.;
Deshmukh, S. V.; Harsch, A.; Secrist, P. J.; Miller, T. A.
Bioorg. Med. Chem. Lett. 2007, 17, 4619.
1. (a) Rodriquez, M.; Aquino, M.; Bruno, I.; De Martino,
G.; Taddei, M.; Gomez-Paloma, L. Curr. Med. Chem.
2006, 13, 1119; (b) Dokmanovic, M.; Marks, P. A. J. Cell.
Biochem. 2005, 96, 293.
2. Sakajiri, S.; Kumagai, T.; Kawamata, N.; Saitoh, T.; Said,
J. W.; Koeffler, H. P. Exp. Hematol. 2005, 33, 53.
3. (a) Hamblett, C. L.; Methot, J. L.; Mampreian, D. M.;
Sloman, D. L.; Stanton, M. G.; Kral, A. M.; Fleming, J. C.;
Cruz, J. C.; Chenard, M.; Ozerova, N.; Hitz, A. M.; Wang,
H.; Deshmukh, S. V.; Nazef, N.; Harsch, A.; Hughes, B.;
Dahlberg, W. K.; Szewczak, A. A.; Middleton, R. E.;