Identification and optimisation of 5-amino-7-aryldihydro-1,4-diazepines as 5-HT2A ligands

Article abstract of DOI:10.1016/j.bmcl.2006.08.108

A several series of low molecular weight 5-HT_2A leads were identified from an analysis of HTS data, the exploration of SAR and optimization of one series using parallel synthesis are described, affording compound 22 (5-HT_2A IC₅₀ 1.1 nM).

Full text of DOI:10.1016/j.bmcl.2006.08.108

Bioorganic & Medicinal Chemistry Letters 16 (2006) 6058–6062
Identification and optimisation of 5-amino-7-aryldihydro-
1,4-diazepines as 5-HT2A ligands
Christopher J. Swain,a,* Ana Teran,b Marta Marotob and Angeles Cabellob
aThe Neuroscience Research Centre, Merck Sharp and Dohme, Terlings Park, Eastwick Road, Harlow, Essex, CM20 2QR, UK
bCIBE, Merck Sharp and Dohme, Josefa Valcarel 38, 28027 Madrid, Spain
Received 5 August 2006; revised 26 August 2006; accepted 28 August 2006
Available online 12 September 2006
Abstract—A several series of low molecular weight 5-HT2A leads were identified from an analysis of HTS data, the exploration of
SAR and optimization of one series using parallel synthesis are described, affording compound 22 (5-HT2A IC50 1.1 nM).
ꢀ 2006 Elsevier Ltd. All rights reserved.
Serotonin (5-hydroxytryptamine, 5-HT) receptor li-
gands have proved beneficial in a wide variety of clinical
indications,1 and in particular 5-HT2A antagonists have
potential utility in depression,2 schizophrenia3 and sleep
disorders.4 5-HT was one of the first neurotransmitters
associated with the regulation of the sleep-wake cycle.5
Furthermore, in clinical trials, several agents that antag-
onize 5-HT2A receptor activity have been shown to be
effective in the treatment of alcohol dependence,6 panic
disorders,7 and in controlling agitation and delusions
in most patients with Parkinson’s disease and psychotic
symptoms.8,9
substitution. However, we were also mindful that low
molecular weight ligands might well have only modest
to low affinity.
The strategy adopted involved selecting all compounds
from the screening collection with molecular weight in
the range 150–250, and then clustering the compounds
using maximum common substructure.12 The high-
throughput screening data were then added and the 150
clusters were then selected that contained at least one
example with >50% inhibition at 5 lM in the HTS.
SMARTS queries13 were then constructed to represent
each of the active clusters and the individual SMARTS
queries used to search a file of known 5-HT2A ligands con-
structed from both internal and literature sources. The
SMARTS queries were also used to search against a file
of known NK1 receptor antagonists using iBabel.14
As part of our ongoing efforts to identify new 5-HT2A
antagonists10 we undertook a high-throughput screening
(HTS) campaign using a functional assay (FLIPR) to
identify novel structural classes of antagonists as poten-
tial leads. This assay was also used to confirm the lack of
efficacy for selected novel ligands. Whilst a conventional
analysis of the screening results identified several inter-
esting starting points we were also interested in identify-
ing low molecular weight starting points since a number
of studies have highlighted the correlation between
molecular weight and the likelihood of success in reach-
ing the marketplace.11
Many of the SMARTS queries identified structures in
the file of known 5-HT2A ligands, whilst others were
R
O
N
O
N+
O
a
X
X
X
Our strategy was to try to identify low molecular weight
scaffolds, which could then be optimized by appropriate
R
NH
R
N
O HN
NH2
NH2
N
N
H2N
Keywords: 5HT2A; Antagonist; Library; High-throughput; Screening;
Synthesis.
H
H
X
X
*
Corresponding author at present address: Cambridge MedChem
Consulting. Tel.: +44 1223 835121; e-mail: swain@mac.com
Scheme 1. Reagents: (a) HCOOEt, NaH, NH2OH; (b) ROH, HClO4.
0960-894X/$ - see front matter ꢀ 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2006.08.108

C. J. Swain et al. / Bioorg. Med. Chem. Lett. 16 (2006) 6058–6062
Table 1. 5-HT2A binding affinity of 5-amino-7-aryldihydro-1,4-diazepine analogues 1–13
6059
HN
N
R
N
H
a
h5-HT2A IC50 (nM)
Compound
R
1
35 (5-HT2C 1050 nM)
2
3
11
Me
112
Me
4
5
534
18
MeO
F
6
7
8
9
21 (5-HT2C 637 nM)
F
F
11
76
F
Cl
2800
NC
F3C
10
11
12
514
14
Me
F
112
Me
13
71
Me
Me
a 5-HT2A affinity was determined as described in Ref. 15.

6060
C. J. Swain et al. / Bioorg. Med. Chem. Lett. 16 (2006) 6058–6062
Table 2. 5-HT2A binding affinity of 5-amino-7-aryldihydro-1,4-diazepine analogues 5–29
X
X
N
N
N
F
N
H
F
N
H
A
B
X
X
N
F
N
N
H
H
F
D
C
a
h5-HT2A IC50 (nM)
Compound
R
A
X
5
18
NH2
NH2
14
A
4.5
15
16
A
A
3.8
3.6
H2N
Ph
NH2
N
17
18
A
A
>1000
1515
H
H2N
NH2
19
20
B
B
1.6
1.5 (5-HT2C 16 nM)
NH2
21
22
B
C
0.8
NH2
1.1 (5-HT2C 27 nM)
NH2
23
C
1.4
H2N

C. J. Swain et al. / Bioorg. Med. Chem. Lett. 16 (2006) 6058–6062
6061
Table 2 (continued)
a
h5-HT2A IC50 (nM)
Compound
R
C
X
24
3.7 (5-HT2C 86 nM)
NH2
25
26
D
D
2.4
H2N
Ph
14.9 (5-HT2C 75 nM)
NH2
O
27
28
29
D
D
D
>1000
>1000
>1000
N
H
N
H
N
N
H
a 5-HT2A affinity was determined as described in Ref. 15.
found in both 5-HT2A and NK1 ligand files and pos-
sibly represent promiscuous motifs. For the purpose
of this exercise the most interesting SMARTS queries
are those representing the 17 clusters containing ac-
tives in the HTS that are not present in either known
5-HT2A or NK1 ligands. Based on this analysis seven-
teen compounds (50–70% inhibition at 5 lM in the
HTS) were selected for titration15 (one per cluster)
and ten of the seventeen were subsequently shown to
have a measured IC50 < 200 nM. The exploration
and optimization of one of these leads (1) is described
herein. The synthetic route used to prepare these ana-
logues is shown in Scheme 1.17 The aryl isoxazolines
were prepared from the corresponding substituted ace-
tophenones by reaction with ethyl formate and
hydroxylamine in the presence of sodium hydride.16
Subsequent reaction with an alkyl alcohol in the pres-
ence of perchloric acid, followed by reaction with
diaminoethane and subsequent cyclisation, afforded
the 1,4-diazepine ring. The nitrile of compound 9
was introduced via cuprous cyanide displacement of
the corresponding bromide.
Small lipophilic substituents at the para-positions of the
phenyl ring were beneficial (2, IC50 11 nM, 6 IC50
21 nM) affording a modest 2- to 3-fold increase in affin-
ity, however polar substituents were not tolerated (9.
IC50 2800 nM). ortho-Substitution was more limited
with only fluoro being tolerated (5, IC50 18 nM). Com-
bination of ortho- and para-substitution afforded a fur-
ther modest increase in affinity (7, IC50 11 nM).
Introduction of a trifluoromethyl at the meta position
(10) was not well tolerated.
Using the 4-methyl (2) and three fluoro-substituted
analogues (5, 6 and 7) as templates the influence of
the alkyl amine was investigated (Table 2). Replace-
ment of the tert-butyl amine by a secondary amines
resulted in a dramatic loss in affinity (e.g., 17, 27,
28 and 29, IC50 >1000 nM), perhaps indicating the
requirement for a hydrogen bond donor. In contrast,
reaction with a variety of larger primary amines gave
a significant increase in affinity. The introduction of
an additional two carbons gave a 5- to 10-fold in-
crease in affinity; compare 5 with 14 (IC50 4.5 nM),
or 7 with 20 (IC50 1.5 nM). Replacement of one of
the methyl groups of the t-butyl by benzyl also gave
a similar increase in affinity (15, IC50 3.8 nM, 25,
IC50 2.4 nM) but was less attractive due to the in-
crease in molecular weight. Interestingly, introduction
of the phenyl ring also gave the only examples of
compounds with significant HERG activity (15,
270 nM, 25, 300 nM) all other compounds being
>1000 nM.
The chemistry proved eminently suitable for parallel
synthesis and a scanning library was prepared using 13
amines and 20 substituted acetophenones. In this com-
munication, we describe initial results from a limited
subset of this library.
Initial SAR was developed to aid understanding of the
influence of aromatic ring substitution (Table 1).

6062
C. J. Swain et al. / Bioorg. Med. Chem. Lett. 16 (2006) 6058–6062
15. Human 5-HT2A receptor binding assay. Membranes (1 mg)
obtained from confluent cultures of CHO K1 cells tansfect-
ed with the human 5-HT2A receptor (licenced from Euro-
screen, Belgium) were incubated with 1 mM [3H]ketanserin
(76 Ci/mmol, Perkin-Elmer, Cat. No. NET-791) in buffer
containing 50 mM Tris[hydroxylmethylaminomethane],
pH 7.7, 0.1% ascorbate, 10 mM pargyline, 10 mM
MgSO4 and 0.5 mM EDTA, in a final volume of
1 mL, for 15 min at 37 ꢁC and at 180 rpm. The reaction
was terminated by filtration onto GF/B filter plates.
The filters were washed with 50 mM Tris[hydroxylme-
thylaminomethane], pH 7.7, using a Packard Filtermate
Harvester. Then, the filter plates were dried and
counted for radioactivity in a Topcount NXT (Pack-
ard) after addition of 50 mL Microscint 20 (Perkin-
Elmer). Non-specific binding was assessed in the
presence of 10 mM mianserin. Five-point log titrations
of the compounds were tested for affinity determina-
tion. Percent inhibition of specific radioligand binding
was calculated and data analyzed using a four-param-
eter curve fitting to determine IC50 and Ki.
Constraining the alkyl groups into either a 5- or a 6-mem-
bered ring gave a further small increase in affinity and
afforded compounds with single figure nanomolar affinity
(16, 19, 21 and 22). Compound 22 IC50 1.1 nM, MW 319
was subsequently evaluated in variety of ion channels,
GPCR and enzyme assays and shown to be >1000-fold
selective except for modest activity at the 5-HT2C receptor
(IC50 27 nM). Importantly, withrespect topossibleinvivo
evaluation compound 22 also displayed excellent affinity
for the rat 5-HT2A receptor (IC50 2.3 nM).
In conclusion, we have described the discovery and opti-
mization of a novel class of 5-HT2A antagonists devel-
oped from a low molecular weight hit (1) identified
from high-throughput screening, by an analysis intended
to identify selective low-molecular weight ligands.
Acknowledgment
16. Simpson, W. R. U.S. Patent 4,096,140, 1978.
17. In a typical procedure, to a stirred mixture of sodium
hydride (60%, 14.6 g, 385.4 mmol), ethyl formate (43.8 g,
730.8 mmol) and tetrahydrofuran (300 mL) was added
substituted acetophenone (182.7 mmol) in tetrahydrofu-
ran (100 mL) at 0 ꢁC. The reaction mixture was stirred for
2.5 h at room temperature, and then diluted with water
(200 mL) and washed with ethyl acetate. The aqueous
layer was separated and hydroxylamine hydrochloride
(12.7 g, 182.7 mmol) was added. The mixture was stirred
for 17 h at room temperature, and then diluted with 1 N
HCl and extracted with ethyl acetate. The organic layer
was dried over magnesium sulfate and concentrated to
give a yellow solid.
We thank Ellen Crawford of the outsourcing group for
arranging the synthetic work to be completed by WuXi
PharmaTech Co., Ltd.
References and notes
1. (a) Leysen, J. E. Curr. Drug Targets CNS Neurol. Disord.
2004, 3, 11; (b) Lanfumey, L.; Hamon, M. Curr. Drug
Targets CNS Neurol. Disord. 2004, 3, 1.
2. Smith, S. In Medicinal Chemistry; Roe, P., Small, J. K.,
Eds.; Pergamon: Oxford, 1990; Vol. 1, pp 301–383.
3. Akhondzadeh, S. IDrugs 2001, 4, 295.
4. Dursun, S. M.; Patel, J. K.; Burke, J. G.; Reveley, M. A.
J. Psychiatry Neurosci. 1999, 24, 333.
5. Jones, B. E. In Principles and Practice of Sleep Medicine;
Kryger, M. H., Roth, T., Dement, W. C., Eds.; W. B.
Saunders: Philadelphia, 2000; pp 134–154.
A mixture of the isoxazoline (57.5 mmol) and tert-butyl
alcohol (4.5 g, 60 mmol) was stirred in an ice bath, and
0.30 mole of 71% perchloric acid was added dropwise. A
white precipitate was formed during addition. After
addition, the precipitate gradually thickened as stirring
was continued. About 100 mL of water was then added.
The suspension was stirred until homogeneous, and the
solid was filtered. Washing with water, air drying, and
washing with dichloromethane left 12 g (90%) of the crude
perchlorate salt.
6. Martinotti, G.; Andreoli, S.; Di Nicola, M.; Buonanno,
A.; Janiri, L.; Bria, P. 18th Congr. Eur. Coll. Neuropsy-
chopharmacol. (ECNP) (Oct 22–Oct 26, Amsterdam)
2005, Abstr. P.6.0357.
7. Galynker, I.; Khan, A.; Grebchenko, Y.; Ten, A.; Malaya,
L.; Yanowitch, P.; Cohen, L. J. J. Clin. Psychiatry 2005, 66.
8. Merims, D.; Shabtai, H.; Balas, M.; Peretz, C.; Giladi, N.
Mov. Disord. 2005, 20(Suppl. 10), Abstr. P4293.
9. Juri, C.; Chana, P.; Tapia, J.; Kunstmann, C.; Parrao, T.
Clin. Neuropharmacol. 2005, 28, 18.
10. Fish, L. R.; Gilligan, M. T.; Humphries, A. C.; Ivarsson,
M.; Ladduwahetty, T.; Merchant, K. J.; O’Connor, D.;
Patel, S.; Philipps, E.; Vargas, H. M.; Hutson, P. H.;
MacLeod, A. M. Bioorg. Med. Chem. Lett. 2005, 15, 3665.
11. Lipinski, C. A.; Lombardo, F.; Dominy, B. W.; Feeney, P.
J. Adv. Drug. Deliv. Rev. 1997, 23, 3.
12. Sheriden, R. A.; Miller, M. J. Chem. Inf. Comput. Sci
1998, 38, 915.
13. SMARTS is a line notation language for representing
molecular search queries, Weinger, D. J. Chem. Inf.
Comput. Sci. 1988, 28, 31, http://www.daylight.com/day-
html/doc/theory/theory.smarts.html.
Ethylenediamine 16 mL (240 mmol) was dissolved in meth-
ylene chloride (200 mL). The perchlorate salt (48 mmol)
was added portionwise with stirring over 20 min. The
temperature of the exothermic reaction was maintained at
20–30 ꢁC by cooling. Stirring was continued for 1 h. The
reaction mixture was diluted by addition of sufficient
methylene chloride to bring the volume to 500 mL. The
reaction mixture was then extracted (2· 300 mL) with water
and the organic layer dried over anhydrous magnesium
sulfate. The methylene chloride was evaporated at reduced
pressure to obtain an oil. The oil was dissolved in 300 mL of
anhydrous ether and filtered free of white solids. The white
solids were washed with a small amount of anhydrous ether
and the washings were combined with the filtrate. Evapo-
ration at a reduced pressure gave the product as a colourless
oil. This was then dissolved in absolute ethanol (5 mL). The
solution was acidified to pH 1 with methanesulfonic acid,
refluxed under nitrogen for 1 h and cooled to room
temperature. Volatiles were evaporated and the residue
was purified by preparative thin-layer chromatography to
give target compound.
14. iBabel is an Applescript Studio GUI for OpenBabel, CDK
News, Vol. 3, issue 1, pp 19–20. http://sourceforge.net/
project/showfiles.php?group_id=40728.