Phenylglycine and phenylalanine derivatives as potent and selective HDAC1 inhibitors (SHI-1)

Article abstract of DOI:10.1016/j.bmcl.2008.02.012

An HTS screening campaign identified a series of low molecular weight phenols that showed excellent selectivity (>100-fold) for HDAC1/HDAC2 over other Class I and Class II HDACs. Evolution and optimization of this HTS hit series provided HDAC1-selective (SHI-1) compounds with excellent anti-proliferative activity and improved physical properties. Dose-dependent efficacy in a mouse HCT116 xenograft model was demonstrated with a phenylglycine SHI-1 analog.

Full text of DOI:10.1016/j.bmcl.2008.02.012

Available online at www.sciencedirect.com
Bioorganic & Medicinal Chemistry Letters 18 (2008) 1859–1863
Phenylglycine and phenylalanine derivatives as potent and selective
HDAC1 inhibitors (SHI-1)
Kevin J. Wilson,a,* David J. Witter,a Jonathan B. Grimm,a Phieng Siliphaivanh,a
Karin M. Otte,a Astrid M. Kral,b Judith C. Fleming,b Andreas Harsch,a Julie E. Hamill,a
Jonathan C. Cruz,c Melissa Chenard,c Alexander A. Szewczak,a Richard E. Middleton,a
Bethany L. Hughes,a William K. Dahlberg,a J. Paul Secristb and Thomas A. Millera
aDepartment of Drug Design and Optimization, 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 Oncology Pharmacology, Merck Research Laboratories, 33 Avenue Louis Pasteur, Boston, MA 02115, USA
Received 2 January 2008; revised 7 February 2008; accepted 7 February 2008
Available online 10 February 2008
Abstract—An HTS screening campaign identified a series of low molecular weight phenols that showed excellent selectivity (>100-
fold) for HDAC1/HDAC2 over other Class I and Class II HDACs. Evolution and optimization of this HTS hit series provided
HDAC1-selective (SHI-1) compounds with excellent anti-proliferative activity and improved physical properties. Dose-dependent
efficacy in a mouse HCT116 xenograft model was demonstrated with a phenylglycine SHI-1 analog.
ꢀ 2008 Elsevier Ltd. All rights reserved.
Ongoing clinical studies with histone deacetylase
(HDAC) inhibitors indicate that these agents show great
promise for the treatment of cancer.1 Zolinzaꢁ (SAHA,
vorinostat), belonging to the hydroxamic acid class of
HDAC inhibitors, represents the first of these to gain
approval by the FDA.2 Following this success, there re-
mains significant interest in identifying HDAC inhibi-
tors with improved efficacy and tolerability. With this
in mind, we sought to identify compounds with in-
creased HDAC isoform specificity. An HTS screening
campaign identified a series of low molecular weight
phenols, typified by compound 1 (Fig. 1), bearing a no-
vel biaryl motif in the zinc binding region of the HDAC
inhibitor pharmacophore.3 The biaryl unit was found to
impart excellent selectivity (>100-fold) for Class I
HDACs 1 and 2 over other Class I HDACs (3 and 8)
and Class II HDACs (4–7).4
version of the N-(2-aminophenyl)benzamide scaffold6,7
provided hybrid compound 2 (Fig. 1). This transforma-
tion had no effect on enzymatic or cell-based potency.
However, replacing the pendant phenyl ring with the
isosteric 2-thienyl unit8 (3) enhanced anti-proliferative
activity relative to phenol 1, although no change in
HDAC1 inhibitory potency was observed. Biaryl benz-
amide 3, thus derived from an HTS screening lead, rep-
resented an excellent starting point for further medicinal
chemistry optimization.
Compound 3 was found to possess sub-optimal physical
properties, with very poor aqueous solubility (0 nM at
pH 7) and a moderate logD7 of 2.7.9 We reasoned that
the addition of an amino acid-derived substituent to the
core scaffold would increase polarity and improve aque-
ous solubility. We targeted the 4-position of the benz-
amide core following our recent success with both
amino acid-6 and malonyl-derived7 benzamide HDAC
inhibitors (Fig. 1). Our goal was to establish the mini-
mum structural elements required for potent inhibition
of HDAC1 and cell proliferation while obtaining suit-
able physical properties.
HTS screening hit 1 showed moderately potent inhibi-
tion of both the HDAC1 isoform and proliferation of
the human colon carcinoma cell line HCT116.5 Combin-
ing our newly discovered biaryl unit with a truncated
We began our investigation by synthesizing parent
phenylglycine analog 4 bearing a free amino acid group,
as outlined in Scheme 1. Palladium coupling of tert-bu-
Keywords: HDAC1 inhibitor; Selective; Biaryl; Benzamide.
*
Corresponding author. Tel.: +1 617 992 2066; fax: +1 617 992
2403; e-mail: kevin_wilson2@merck.com
0960-894X/$ - see front matter ꢀ 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2008.02.012

1860
K. J. Wilson et al. / Bioorg. Med. Chem. Lett. 18 (2008) 1859–1863
R1
N
O
IP (nM)
58
O
R2
N
H
HDAC1
HDAC2
O
NH2
200
R
n
N
H
2100
HCT116-72hr
R = NHCOR3 or CONR4R5
OH
N-(2-aminophenyl)benzamides
1
IP (nM)
60
HDAC1
HDAC2
O
780
N
1700
HCT116-72hr
H
NH2
2
IP (nM)
48
S
HDAC1
HDAC2
O
362
N
720
HCT116-72hr
H
NH2
3
Figure 1. HDAC1 and proliferation inhibitory potency of a biaryl phenol HTS screening hit (1), related aniline derivatives (2 and 3) and
representative N-(2-aminophenyl)benzamides. IP, inflection point.
tyl 4-bromobenzoate 5 with ethyl [(diphenylmethylid-
ene)amino]acetate10 gave protected amino di-ester 6.
Acid hydrolysis of both the tert-butyl ester and imine
provided amino ester substituted benzoic acid 7. The
free amine was first protected (8) before adding the biar-
yl head group3 via a Bop coupling to provide benzamide
9. Hydrolysis of the ethyl ester yielded advanced carbox-
ylic acid 10. Finally, concomitant removal of both Boc
O
O
O
OtBu
OH
OtBu
a
b
H2N
O
Ph
N
O
Br
Ph
OEt
OEt
5
7
6
S
O
O
OH
H
N
d
c
N
H
tBuO
H
N
tBuO
HN
O
O
O
OEt
O
OtBu
O
OEt
8
9
S
S
O
O
f or g
e
N
H
N
H
H
tBuO
N
HN
O
H2N
NH2
O
OtBu
O
R
O
OH
4: R = OH
10
11-13: R = NR1R2
Scheme 1. Reagents and conditions: (a) ethyl [(diphenylmethylidene)amino]acetate, K3PO4, Pd(PtBu3)2, 100 ꢂC; (b) 2 N HCl, Et2O; (c) Boc2O,
DIPEA, THF; (d) 1,1-dimethylethyl [2-amino-4-(2-thienyl)phenyl]carbamate, BOP, DIPEA, DMF; (e) 1 N LiOH, THF; (f) TFA, DCM, (g) i—
AmineÆHCl, EDC, HOBt, DIPEA, DMF; ii—TFA, DCM.

K. J. Wilson et al. / Bioorg. Med. Chem. Lett. 18 (2008) 1859–1863
1861
groups according to standard conditions yielded desired
amino acid 4.
properties. Representative analogs 11 and 12 possessed
logD7 values of 1.49 and 1.57, respectively, with the
aqueous solubility (pH 7) of both compounds around
185 lM.
The addition of the amino acid unit maintained equiva-
lent HDAC1 inhibitory potency relative to 3 (Table 1).
However, amino acid 4 lacked cell proliferation inhibi-
tion activity, which is most likely a reflection of the poor
membrane permeability of the zwitterion. Reasoning
that cell-based potency could be improved by converting
the acid to more cell penetrant functionalities, we under-
took the SAR evaluation of amide derivatives. In order
to minimize molecular weight, we chose to synthesize a
small series of minimally substituted amides, that is,
NH2, NHMe, and NMe2. To this end, 10 proved to be
a useful intermediate allowing access to these amides
via a routine EDC coupling followed by removal of
the Boc groups under acidic conditions (Scheme 1).11
Further optimization of this series of novel, low molec-
ular weight HDAC1 inhibitors included the evaluation
of individual enantiomers. For this purpose, we devel-
oped a synthesis of chiral benzoic acid derivatives 15
from the commercially available enantiomers of 4-
hydroxyphenylglycine.15 The key step in this sequence
was the carbonylation of an aryl triflate, which required
the use of DIPEA rather than TEA to suppress racemi-
zation of the stereochemistry. Both enantiomers of the
primary amide (16 and 17) and methylamide (18 and
19) were readily accessed from these chiral benzoates
via a Bop coupling to install the biaryl head-group fol-
lowed by TFA-mediated deprotection (Scheme 2,
n = 0). No stereochemical preference was observed with
respect to inhibition of either HDAC1 or cell prolifera-
tion (Table 1).
An increase in HDAC1 inhibition was observed for all
amide analogs relative to the acid, ranging from sixfold
for methyl amide 12 and dimethyl amide 13 to 10-fold
for primary amide 11. Gratifyingly, the amides pos-
sessed excellent anti-proliferative activity showing a
twofold improvement over the unsubstituted compound
3 (Table 1). Interesting SAR was observed with respect
to hERG binding.12 The more lipophilic dimethylamide
13 possessed a moderate degree of affinity for this ion
channel (IC50 = 6.8 lM). hERG affinity was signifi-
cantly reduced with the more polar primary and methyl
amides (<20% inhibition at 10 lM). The ethyl ester 1413
further illustrates this trend, being both the most lipo-
philic and the most potent hERG ligand
(IC50 = 5.7 lM) within this series.14 In addition to their
excellent HDAC1 and proliferation inhibition proper-
ties, the amino amides exhibited favorable physical
S
O
OH
O
a, b
R1
N
N
H
n
n = 0, 1
NH2
R2
NH
R1
N
*
n
O
R2
NH2
O
OtBu
*
O
16-19: n = 0
24, 25: n = 1
15
Scheme 2. Reagents: (a) 1,1-dimethylethyl [2-amino-4-(2-thie-
nyl)phenyl]carbamate, BOP, DIPEA, DMF; (b) TFA, DCM.
Table 1.
S
O
N
H
NH2
n
R
N
Y
H
O
Compound
R
Y
n
HDAC1 IPa (nM)
HCT116-72 h IPa (nM)
4
OH
NH2
H
0
0
0
0
0
0
0
0
0
0
0
0
0
1
1
66
5.7
10
8.8
10
12
11
14
22
14
10
18
29
37
29
>20,000
350
450
11
12
13
14
16
17
18
19
20
21
22
23
24
25
H
NHMe
NMe2
OEt
H
H
420
H
1100
350
NH2 (R)
NH2 (S)
NHMe (R)
NHMe (S)
NHMe
NHMe
NHMe
NHMe
NH2 (R)
NH2 (S)
H
H
510
420
H
H
480
620
COMe
CONHMe
CO2Me
SO2Me
H
960
410
780
390
H
260
a Values are means of n P 2.

1862
K. J. Wilson et al. / Bioorg. Med. Chem. Lett. 18 (2008) 1859–1863
(Table 2).16 The unique selectivity profile of this new
S
S
class of inhibitors arises from the addition of the second
aryl group on to the zinc binding diamine. One possible
explanation is that the pendant aryl ring is able to occu-
py the internal cavity17 within HDAC1 but not the inter-
nal cavity of other isoforms, thus imparting potency and
selectivity for this isoform.3
O
O
a or b or c or d
N
H
N
H
H
N
H2N
O
NH2
NH2
Y
NH
O
NH
12
20: Y = COMe
Based on the excellent in vitro profile of compound 12,
pharmacokinetic parameters were determined in multi-
ple species (Table 3). This analog exhibited moderate
(rat) to low (dog and rhesus) plasma clearance with an
acceptable half-life in rat and dog. The half-life in rhesus
was found to be shorter as a result of a reduced volume
of distribution in this species. Oral bioavailability was
satisfactory in rat but lower in dog and very poor in
rhesus.
21: Y = CONHMe
22: Y = CO2Me
23: Y = SO2Me
Scheme 3. Reagents: (a) Y = COMe: Ac2O, DIPEA, THF; (b)
Y = CONHMe: MeNCO, THF; (c) Y = CO2Me: ClCO2Me, DIPEA,
THF; (d) Y = SO2Me: MeSO2Cl, DIPEA, THF.
Having established that low molecular weight phenyl-
glycine derivatives 11–19 were potent inhibitors of
HDAC1 and cell proliferation, we next wished to assess
the effect of additional modifications. To this end, the
primary amine of 12 was selectively capped by reacting
with various electrophiles (Scheme 3). Acetamide 20
and methyl urea 21 exhibited equipotent inhibition
of HDAC1 relative to 12, whereas the methyl carbamate
22 and methyl sulfonamide 23 showed a two- and
threefold loss of potency, respectively. Only 22
retained equivalent inhibition of cell proliferation,
with all other analogs showing attenuated cell-based
potencies (Table 1).
Despite the sub-optimal pharmacokinetic profile, we
chose to further characterize methyl amide 12 in a 21-
day HCT116 mouse xenograft study. Gratifyingly, 12
exhibited dose-dependent tumor growth inhibition
(TGI) following ip administration (Fig. 2). A dose of
65 mg/kg qd was sufficient to produce TGI equivalent
to vorinostat (150 mg/kg ip qd). These data provide evi-
dence of the utility of our selective HDAC1 inhibitors as
potential anti-cancer therapeutics.
In conclusion, the evolution and optimization of a novel
scaffold identified by HTS yielded compounds with
excellent inhibitory potency of HDAC1 and cell prolif-
eration. The presence of a biaryl unit in the zinc binding
domain was found to impart a high degree of selectivity
for HDAC1 over other Class I and Class II isoforms.
The strategy of appending amino acid-derived substitu-
ents was successful in improving aqueous solubility (e.g.
185 lM at pH 7 for compounds 11, 12, and 24) while
minimizing off-target activity. Efficacy was demon-
strated in a mouse xenograft model despite sub-optimal
pharmacokinetic properties. Further optimization and
characterization of biaryl SHI-1 will be the subject of
subsequent publications.
We also extended our investigation to include phenylal-
anine homologs, exemplified by primary amides 24 and
25. These analogs were readily accessed from chiral ben-
zoates 15 (n = 1) according to Scheme 2. These enantio-
merically pure building blocks were again synthesized
via an hydroxycarbonylation approach, this time
employing triflates derived from L- or D-tyrosine.15 Both
enantiomers showed a three- to fourfold decrease in
HDAC1 inhibition relative to the lower homologs
(n = 0) but maintained equivalent anti-proliferative po-
tency (Table 1). Adding additional molecular weight,
either through derivatizing the amino unit or adding lar-
ger amide substituents, provided analogs with compara-
ble HDAC1 inhibitory potency in both the
phenylglycine and phenylalanine homologs. However,
these larger derivatives suffered from increased hERG
binding and offered no advantage with respect to phar-
macokinetics or other properties (data not shown).
Table 3. PK parameters for compound 12
Species IV AUC
Cl
Vdss
T1/2 (h) F (%)
Having identified potent inhibitors of the HDAC1 iso-
form and cell proliferation, we profiled this series
against additional HDAC isoforms. As with the unsub-
stituted benzamide 3, representative analog 12 exhibited
excellent HDAC1 selectivity, ranging from 13-fold rela-
tive to HDAC2, 350-fold relative to HDAC3 and
HDAC11, and P3500-fold over the remaining isoforms
(lM h kg/mg) (mL/min/kg) (L/kg)
Rata
1.2
4.0
38
11
16
7.6
3.2
1.6
5.9
5.4
1.6
27
17
5
Dogb
Rhesusb 2.7
a IV dose 2 mg/kg, oral dose 4 mg/kg.
b IV dose 0.5 mg/kg, oral dose 1 mg/kg.
Table 2. Compound 12 HDAC isoform selectivity profile
HDAC1
HDAC2
HDAC3
HDAC4
HDAC5
HDAC6
HDAC7
HDAC8
HDAC11
IP (nM)
Fold sel.
10
—
130
13·
3400
340·
>50,000
>5000·
>50,000
>5000·
>50,000
>5000·
>50,000
>5000·
35,000
3500·
3600
360·

K. J. Wilson et al. / Bioorg. Med. Chem. Lett. 18 (2008) 1859–1863
1863
8. During the preparation of this manuscript a report
describing the 2-thienyl head-group was published: Mor-
adei, O. M.; Mallais, T. C.; Frechette, S.; Paquin, I.;
Tessier, P. E.; Leit, S. M.; Fournal, M.; Bonfils, C.;
Trachy-Bourget, M.-C.; Liu, J.; Yan, T. P.; Lu, A.-H.;
Rahil, J.; Wang, J.; Lefebvre, S.; Li, Z.; Vaisburg, A. F.;
Besterman, J. M. J. Med. Chem. 2007, 50, 5543.
Effect of 12 on tumor growth
400
350
300
250
200
150
100
50
Vehicle
Vorinostat, 150 mg/kg
12, 50 mg/kg
12, 65 mg/kg
12, 80 mg/kg
9. Aqueous solubility and logD7 values were determined
using a high-throughput HPLC assay. For this assay two
samples were prepared: one 100 lM solution in organic
solvent (8:1:1, MeOH:MeCN:DMSO) and one 200 lM
solution in PBS buffered water at pH 7. The sample in PBS
buffered water was agitated for 24 h at room temperature
and filtered to remove any compound that had precipi-
tated. The sample in organic solvent was fully dissolved.
Both the sample in organic solvent and the filtered
supernatant of the sample in PBS buffered water were
run using the same HPLC method and the areas of the
compound peaks for each sample were compared in order
to determine the concentration of compound dissolved in
PBS buffered water. This concentration indicated the
solubility of the compound in water up to a maximum
concentration of 200 lM. To determine logD7 for the
compound, several standards with known logD7 values
were run using the same HPLC method and a standard
curve was generated that related logD7 to retention time.
The retention time of the compound peak from the sample
dissolved in organic solvent was then used with the
equation for the best fit line of the logD7 standard curve to
determine the logD7 value for the compound.
*
*
2-way ANOVA
* : P < 0.05
N = 12 , 11 , 12 , 10 and 7
0
0
2
4
6
8
10
12
14
16
18
20
22
Days after treatment
Plasma concentrations of 12:
50 mg/kg
65 mg/kg
28.6 μM
10.0 μM
0.02 μM
80 mg/kg
2 hr
4 hr
nd
nd
nd
17.9 μM
4.1 μM
0.01 μM
24 hr
Figure 2. 21-day mouse xenograft data and corresponding plasma
exposures. nd, not determined.
Acknowledgments
We wish to thank Xiaoliu Geng and Adam Beard for the
solubility and log D7 determinations.
10. Lee, S.; Beare, N. A.; Hartwig, J. F. J. Am. Chem. Soc.
2001, 123, 8410.
11. Spectral data for the TFA salt of representative amino
1
amide 12: H NMR (DMSO-d6, 600 MHz) d 9.83 (s, 1H),
8.72 (br s, 3H), 8.50 (q, J = 4.2 Hz, 1H), 8.04 (d,
J = 8.4 Hz, 2H), 7.61 (d, J = 8.4 Hz, 2H), 7.47 (d,
J = 1.2 Hz, 1H), 7.35 (d, J = 4.8 Hz, 1H), 7.32 (dd,
J = 8.1 and 2.1 Hz, 1H), 7.24 (d, J = 3.6 Hz, 1H), 7.03
(dd, J = 4.8 and 3.6 Hz, 1H), 6.84 (d, J = 8.4 Hz, 1H), 4.97
(br s, 1H), 2.61 (d, J = 4.2 Hz, 3H). MS: calc’d 381
(MH+), exp 381 (MH+). For full experimental details, see
WO2007100657.
References and notes
1. Glaser, K. B. Biochem. Pharm. 2007, 74, 659.
2. Grant, S.; Easley, C.; Kirkpatrick, P. Nat. Rev. Drug Disc.
2007, 6, 21.
3. Witter, D. J.; Harrington, P.; Wilson, K. J.; Chenard, M.;
Cruz, J. C.; Fleming, J. C.; Harsch, A.; Kral, A. M.;
Secrist, J. P.; Miller, T. A. Bioorg. Med. Chem. Lett. 2008,
18, 726.
4. For assay details, see reference within Jones, P.; Altamura,
S.; Chakravarty, P. K.; Cecchetti, O.; De Francesco, R.;
Gallinari, P.; Ingenito, R.; Meinke, P. T.; Petrocchi, A.;
Rowley, M.; Scarpelli, R.; Serafini, S.; Steinkuhler, C.
¨
Bioorg. Med. Chem. Lett. 2006, 16, 5948–5952.
5. For assay details, see Ref. 3.
6. Hubbs, J. L.; Zhou, H.; Kral, A. M.; Fleming, J. C.;
Dahlberg, W. K.; Hughes, B. L.; Middleton, R. E.;
Szewczak, A. A.; Secrist, P.; Miller, T. A. Bioorg. Med.
Chem. Lett. 2008, 18, 34.
7. Siliphaivanh, P.; Harrington, P.; Witter, D. J.; Otte, K.;
Tempest, P.; Kattar, S.; Kral, A. M.; Fleming, J. C.;
Deshmukh, S. V.; Harsch, A.; Secrist, P.; Miller, T. A.
Bioorg. Med. Chem. Lett. 2007, 17, 4619.
12. For assay details, see Wang, J.; Della Penna, K.; Wang,
H.; Karczewski, J.; Connolly, T. M.; Koblan, K. S.;
Bennett, P. B.; Salata, J. J. Am. J. Physiol. 2003, 284,
H256.
13. Synthesized by treating compound 9 with TFA in DCM.
14. For a review of hERG optimizations, including a discus-
sion of the effect of lipophilicity, see Jamieson, C.; Moir,
E. M.; Rankovic, Z.; Wishart, G. J. Med. Chem. 2006, 49,
5029.
15. Grimm, J. B.; Wilson, K. J.; Witter, D. J. Tetrahedron
Lett. 2007, 48, 4509.
16. Relative HDAC1 selectivity values were consistent across
all analogs discussed herein.
17. Wang, D.; Wiest, O.; Helquist, P.; Lan-Hargest, H.;
Wiech, N. L. J. Med. Chem. 2004, 47, 3409.