2,3,4,5-Tetrahydro- and 2,3,4,5,11,11a-hexahydro-1H-[1,4]diazepino[1,7-a]indoles: New templates for 5-HT2C agonists

Article abstract of DOI:10.1016/S0960-894X(03)00403-7

The design and synthesis of the novel 2,3,4,5-tetrahydro-1H-[1,4]diazepino[1,7-a]indole 5 is described. This azepinoindole has excellent affinity for 5-HT_2C (K_i 4.8 nM) and modest selectivity over 5-HT_2A (～4-fold). Several N- and C₁₁-substituted analogues of 5 were prepared, as were a number of biaryl indoline derivatives. The anxiolytic potential for the azepinoindole template 5 is demonstrated by activity in a mouse shock-aggression assay.

Full text of DOI:10.1016/S0960-894X(03)00403-7

Bioorganic & Medicinal Chemistry Letters 13 (2003) 2369–2372
2,3,4,5-Tetrahydro- and 2,3,4,5,11,11a-Hexahydro-1H-
[1,4]diazepino[1,7-a]indoles: New Templates for 5-HT2C Agonists
Michael D. Ennis,a,* Robert L. Hoffman,a Nabil B. Ghazal,a Rebecca M. Olson,a
Christopher S. Knauer,b Chris L. Chio,b Deborah K. Hyslop,b Jeffery E. Campbell,b
Lawrence W. Fitzgerald,b Nanette F. Nichols,b Kjell A. Svensson,b Robert B. McCall,b
Christopher L. Haber,c Michelle L. Kageyc and Dac M. Dinhc
aMedicinal Chemistry III, Pharmacia Corporation, 301 Henrietta Street, Kalamazoo, MI 49009, USA
bCNS Diseases, Pharmacia Corporation, 301 Henrietta Street, Kalamazoo, MI 49009, USA
cDiscovery Technologies, Pharmacia Corporation, 301 Henrietta Street, Kalamazoo, MI 49009, USA
Received 17 March 2003; revised 11 April 2003; accepted 15 April 2003
Abstract—The design and synthesis of the novel 2,3,4,5-tetrahydro-1H-[1,4]diazepino[1,7-a]indole 5 is described. This azepino-
indole has excellent affinity for 5-HT2C (Ki 4.8 nM) and modest selectivity over 5-HT2A (ꢀ4-fold). Several N- and C11-substituted
analogues of 5 were prepared, as were a number of biaryl indoline derivatives. The anxiolytic potential for the azepinoindole tem-
plate 5 is demonstrated by activity in a mouse shock–aggression assay.
# 2003 Elsevier Science Ltd. All rights reserved.
The treatment of psychiatric disorders by modulation of
receptors for central nervous system neurotransmitters
has become standard therapy. Many the most com-
monly prescribed CNS medications worldwide are those
that affect the function of serotonin (5-HT). Among the
14 identified sub-types of the 5-HT receptor, the 5-HT2C
sub-type has received considerable attention as a
potential therapeutic target for several indications,
including anxiety, epilepsy, schizophrenia, and obesity.1
The search for agents with high selectivity for this site
has proven difficult in light of the close sequence
homology among the sub-family of 5-HT2 receptors (i.e.,
5-HT2A and 5-HT2B), and few reports of highly selective
agonists have emerged.2 Many of the reported agonists
are indole-based derivatives possessing 5-HT2C selectivity
of 5–10-fold relative to 5-HT2A. Herein we describe the
synthesis and preliminary pharmacological evaluation of
a new class of selective 5-HT2C azepinoindole-based ago-
nists with potential for the treatment of anxiety.
Figure 1. Azepinoindoles 1, 2, 3, and 5 and isotryptamine 4.
ingly, the N-methyl derivative 2 was evaluated in human
clinical trials in the mid 1960s and found to cause
weight loss.4a Like 1, 2 is a potent 5-HT2C agonist but is
non-selective with regard to affinity at the 5-HT2A and
5-HT2B receptors.4b Recently, the anxiolytic activity of
a related azepino[4,5-b]indole 3 has been described.5 In
1997, the isotryptamine (S)-4 was reported as a potent
5-HT2C ligand (Ki 4 nM) with excellent selectivity over
5-HT2A (63-fold), superior to the 10-fold selectivity of the
corresponding racemic tryptamine analogue.6a However,
it has been shown that receptor affinities are highly
dependent upon assay protocols, and a re-examination of
The azepino[4,5-b]indole 13 was identified as a potent
5-HT2C partial agonist from high-throughput screening
of Pharmacia’s compound collection (Fig. 1). Interest-
*Corresponding author. Tel.:+1-269-833-1539; fax:+1-269-833-9629;
e-mail: michael.d.ennis@pharmacia.com
0960-894X/03/$ - see front matter # 2003 Elsevier Science Ltd. All rights reserved.
doi:10.1016/S0960-894X(03)00403-7

2370
M. D. Ennis et al. / Bioorg. Med. Chem. Lett. 13 (2003) 2369–2372
Table 2. 5-HT affinity and selectivity for C-11 substituted analogues
11
Compd
R
Ki (nM)
2A/2C
5-HT2C
5-HT2A
Selectivity
5
H
Cl
CHO
CH3
CN
4.8
4.7
12
6.8
85
74
54
136
18
6.5
29
12
83
46
40
316
3.8
1.4
2.4
1.8
1.0
0.6
0.7
2.3
Scheme 1. (a) NaCNBH3, HOAc (82%); (b) 1,2-dibromoethane,
K2CO3, CH3CN (81%); (c) NH3, MeOH (72%); (d) BH3–SMe2, THF
(89%); (e) DDQ, dioxane (52%).
11a
11b
11c
11d
11e
11f
11g
the binding profiles for isotryptamines such as 4 has
demonstrated greatly reduced selectivity (ꢁ10ꢂ) using
human receptors and agonist radioligands.6b Based on
these observations, we envisioned that the azepino[1,7-
a]indole 5 would be a useful template upon which an
analogue program targeting selective 5-HT2C agonists
could be based.
CONH2
COCF3
C(O)CONH2
indole 5 displayed slightly enhanced 2A/2C selectivity.
In addition, azepino[1,7-a]indole 5 possessed greater
intrinsic activity than 1. The racemic indoline 9 showed
severely attenuated affinity at both 5-HT2C and 5-HT2A
relative to 5. Substitution of the azepine nitrogen of 5
with a methyl group attenuated 5-HT2C affinity and
dramatically reduced 5-HT2C efficacy. Substitution with
larger groups also resulted in decreasing 5-HT2C affinity.
Our synthesis of azepino[1,7-a]indole 5 centered on the
elaboration of the known indol-2-yl acetate 6 (Scheme
1).7 Reduction of 6 to the corresponding indoline and
subsequent alkylation with 1,2 dibromoethane gave 7 in
66% overall yield. Treatment of 7 with ammonia-satu-
rated methanol at 50ꢃC generated the tricyclic lactam 8 in
72% yield. Reduction of 8 with borane-dimethylsulfide
and subsequent oxidation of the resulting indoline (9)
with DDQ at room temperature generated the targeted
azepino[1,7-a]indole 5 in 46% overall yield. Alkylation
of the azepine nitrogen in 5 provided N-alkyl analogues
10 (see Table 1). Protection of 5 as the trifluoro-
acetamide allowed for preparation of the C-11 analogues
11 (see Table 2) following mild acetamide hydrolysis
(MeOH, K2CO3).
An attractive feature of the azepino[1,7-a]indole template
5 is the potential for analogue generation by derivatiza-
tion of the indole C-11 position. Using well-known
transformations, compounds 11a–g were prepared and
their binding affinities are given in Table 2. Unfor-
tunately, none of the analogues prepared displayed
5-HT2C affinity or selectivity superior to that of the
parent template 5.
In addition to the N- and C11-substituted analogues
described above, we have examined a series of biaryl
indoles and indolines derived from 5. Encouraging this
effort are recent reports of selective 5-HT2C agonists
containing a related biaryl carboline motif.9 The pre-
paration of these analogues is illustrated in Scheme 2.
Given below in Table 1 are the 5-HT2C and 5-HT2A
binding affinities and 5-HT2C efficacies for 1, 2, 3, 5, 9,
and the N-alkylated analogues 10a–c.8 Although both 1
and 5 had comparable 5-HT2C affinities, azepino[1,7-a]-
Table 1. 5-HT2C and 5-HT2A binding affinities and 5-HT2C efficacies
for 1, 2, 3, 5, 9, and N-alkyl analogues 10a–c
Bromination of the protected indoline 12 occurred
regioselectively to give the C-9 bromo derivative 13
exclusively. Palladium-mediated coupling of 13 with 2-
ethoxy phenylboronic acid and subsequent deprotection
provided the 9-aryl substituted azepinoindoline 14.
Access to the 8-aryl substituted azepinoindoline 17 was
achieved by carrying the bromo indole 1510 through the
sequence of reactions illustrated in Scheme 1 to the
bromo-lactam 16. Suzuki coupling of 16 provided the
targeted 8-aryl analogue after lactam reduction with
borane-dimethylsulfide. The 7-aryl analogues 20 could
be prepared from the chloride 18. In contrast to the
selective bromination described above, chlorination of
12 afforded a separable mixture of 18 and 19 in 18 and
64% isolated yields, respectively. We obtained yields
between 73 and 83% for Suzuki couplings of 18 with a
variety of phenylboronic acids when employing tris(di-
benzylideneacetone) dipalladium (0) and tri-cyclohexyl
Compd
R
Ki (nM)
5-HT2C
5-HT2C
5-HT2A
Intrinsic efficacy
1
2
3
5
9a
10a
10b
10c
6.1
19
4.3
4.8
10
19
4.1
18
805
28
30
56
66%
83%
105%
80%
ND
49%
66%
ND
721
Me
Et
Pr
13
38
64
aTested as a racemic mixture. ND=not determined.

M. D. Ennis et al. / Bioorg. Med. Chem. Lett. 13 (2003) 2369–2372
2371
Scheme 2. (a) N-bromosuccinimide, DMF; (b) 2-ethoxyphenylboronic acid, (PPh3)2PdCl2, 2M Na2CO3, benzene (yield range); (c) TFA, CH2Cl2;
(d) BH3-SMe2; (e) N-chlorosuccinimide, DMF; (f) ArB(OH)2, Pd2dba3, P(cy)3, Cs2CO3, dioxane.
phenyl)azepino[1,7-a]indole 23 displayed reduced
5-HT2C affinity (112 nM) and significant 5-HT2A affinity
Table 3. Effect of regiochemistry on binding affinity for biaryl indo-
lines
(32 nM). Based upon these results, we embarked on a
more thorough exploration of 7-arylazepinoindolines,
including an examination of the effect of absolute ste-
reochemistry on receptor affinity. The analogues pre-
pared for this study and their corresponding binding
profiles are given in Table 4.12
Compda
R9
R8
R7
Ki (nM)
The data in Table 4 show that incorporation of a
chlorine substituent at the indoline C-9 position does
not generally lead to analogues with either greater affi-
nity or selectivity for the 5-HT2C receptor relative to
analogues without the chlorine. Although such an
enhancement was seen in the 7-(2-ethoxyphenyl) series
(e.g., 20a and 22a, Table 3), the corresponding 7-(2-
chlorophenyl) and 7-(2-methylphenyl) series showed the
opposite trend (see 20b and 22e). It is also observed that
monosubstitution at the 20-position on the 7-aryl sub-
stituent is greatly preferred (compare the 20-ethoxy
analogue 22a with the corresponding 30- or 40-ethoxy
compounds 22b and 22c). Disubstituted analogues 22h–j
also displayed inferior receptor binding profiles. Not
unexpectedly, 7-aryl substituted indolines 20 and 22
exhibit an enantiopreference wherein the favored isomer
possesses a 5-HT2C Ki about 2–5 times lower than its
enantiomer. In general, this series favors the (+)-isomer,
although the (À)-isomer in the 20-trifluoromethyl series
(20d) is an exception. Typically, that enantiomer showing
greater affinity for the 5-HT2 receptor also displays
improved 2A/2C selectivity, often by 2-fold or higher (see
(+)-20c, for example). The 7-(20-methylphenyl) analogue
(+)-20c is the most selective compound to emerge from
this series. With an affinity at the 5-HT2C site of 30 nM
5-HT2C 5-HT2A
14
17
20a
22a
23a
2-EtOPh
H
H
2-EtOPh
H
H
H
4815
932
110
54
>10,000
1581
440
218
32
H
Cl
2-EtOPh
2-EtOPh
H
(indole of 22a)
112
aAll compounds except 23 were tested as racemic mixtures.
phosphine as the ligand.11 Finally, deprotection of the
coupled products gave the targeted 7-aryl analogues 20.
In addition, bromination of 19 provided the 7-bromo-9-
chloro derivative 21. Suzuki coupling of 21 occurred
exclusively at the C-7 bromo position to ultimately
afford the 9-chloro-7-biaryl derivatives 22.
The binding profiles for the initial set of regioisomeric
2-ethoxyphenyl biaryl indolines is given in Table 3. It is
clear that the analogue bearing the aromatic ring at the
9-position (20a) is superior to the 7- and 8-position iso-
mers in terms of both affinity and selectivity for 5-HT2C
.
Interestingly, introduction of a 9-chloro substituent
enhances affinity at both 5-HT2C and 5-HT2A (22a). For
comparison, the corresponding 9-chloro-7-(2-ethoxy-

2372
M. D. Ennis et al. / Bioorg. Med. Chem. Lett. 13 (2003) 2369–2372
Table 4. 5-HT2C affinity and selectivity for 7-aryl azepinoindolines 20
and 22
an in vivo behavioral model at doses which do not
sedate or impair motor function. In addition, we have
also examined a series of aryl-substituted azepinoindo-
lines to identify 5-HT2C-selective analogues. From this
effort has emerged the (+)-7-(20-methylphenyl)-
azepinoindoline (+)-20c, which has good affinity and
improved 2A/2C selectivity.
Compd
R
Ar
Ki (nM)
2A/2C
Selectivity
5-HT2C
5-HT2A
Acknowledgements
(+)-20a
(À)-20a
20b
(+)-20c
(À)-20c
(+)-20d
(À)-20d
(+)-22a
(À)-22a
22b
22c
22d
22e
22f
22g
22h
22i
22j
H
H
H
H
H
2-EtO
2-EtO
2-Cl
2-Me
2-Me
2-CF3
2-CF3
2-EtO
2-EtO
3-EtO
4-EtO
2-Me
103
196
23
30
39
59
14
32
178
472
265
43
458
980
82
191
136
107
45
116
354
233
182
88
102
141
146
426
70
4.45
5.00
3.57
6.37
3.45
1.81
3.21
3.63
1.99
0.49
0.69
2.05
1.79
0.63
0.47
1.10
0.44
0.95
We thank L. Dolak, E. Seest, and M. Hester for chrom-
atographic resolution of the enantiomers in Table 4.
H
H
References and Notes
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Cl
1. (a) Isaac, M. Drugs Future 2001, 26, 383. (b) Roth, B. L.;
Shapiro, D. A. Exp. Opin. Ther. Targets 2001, 5, 685.
2. Fitzgerald, L. W.; Ennis, M. D. Annu. Rep. Med. Chem.
2002, 37, 21.
3. Hester, J. B.; Tang, A. H.; Keasling, H. H.; Veldkamp, W.
J. Med. Chem. 1968, 11, 101.
4. (a) Gallant, D. M.; Bishop, M. P.; Bishop, G.; O’Meallie,
L. Curr. Therapeutic Res. 1967, 9, 579. (b) McCall, R. B.;
Franklin; S. R.; Hyslop; D. K.; Knauer; C. S.; Chio; C. L.;
Haber; C. L.; Fitzgerald; L. W. Abstracts of Papers, 31 Annual
Meeting of the Society for Neuroscience, San Diego, CA,
2001; 309.2.
5. Nichols, N. F.; Sobie, J. L.; Cimini, M. G.; Kagey, M. L.;
Haber, C. L.; Svensson, K. A. Abstracts of Papers, 32 Annual
Meeting of the Society for Neuroscience, Orlando, Fl, 2002;
290.13.
2-Cl
2-F
4-F
57
223
310
386
160
793
2,4-di Cl
2,5-di Cl
5-Cl-2-MeO
756
Table 5. Anxiolytic activity in the shock-induced aggression assay
and motor impairment (ataxia) for 5 and alprazolam in mice
Dosea
Latency to
fight (mean s)
Time on rotating
rod (mean s)
6. (a) Bos, M.; Jenck, F.; Martin, J. R.; Moreau, J.-L.;
Sleight, A. J.; Wichmann, J.; Widmer, U. J. Med. Chem. 1997,
40, 2762. (b) Chang-Fong, J.; Addo, J.; Dukat, M.; Smith, C.;
Mitchell, N. A.; Herrick-Davis, K.; Teitler, M.; Glennon,
R. A. Bioorg. Med. Chem. Lett. 2002, 12, 155.
Vehicle
Alprazolam
Alprazolam
5
5
5.8Æ1.0
11.9Æ2.1*
35.2Æ3.7y
12.8Æ2.3yy
27.0Æ4.5y
95.5Æ6.6
74.2Æ9.2#
14.5Æ2.6y
102.0Æ0.0
87.6Æ9.6
0.3
3.0
3.0
30
7. Moody, C. J.; Rahimtoola, K. F. J. Chem. Soc., Perkin
Trans. I 1990, 673.
*p<0.05; #p<0.01; p<0.001; yyp=0.09 compared to vehicle.
aN=10–12/group.
y
8. Radioligand binding competition assays used [125I] DOI
and human 5-HT2C (ISV isoform in CHO-K1 cells and human
5-HT2A in HEK-293 cells); values reported are means (n=2–3;
individual variations of Æ10%). See: Fitzgerald, L. W.; Con-
klin, D. S.; Krause, C. M.; Marshall, A. P.; Patterson, J. P.;
Tran, D. P.; Iyer, G.; Kostich, W. A.; Largent, B. L.; Hartig,
P. R. J. Neurochem. 1999, 72, 2127. Efficacies were assessed by
fluorescent imaging plate reader (FLIPR Ca2+) with fluores-
cence readings taken 10 s before and 4 min after the agonist
addition.
and a 2A/2C selectivity of 6.4, this analogue represents a
dramatic improvement in receptor binding profile relative
to the unsubstituted parent azepinoindoline 9.
The anxiolytic properties of the tetracyclic azepino-
indole 3 have already been mentioned, as have the
anorexic effects of 2. We hypothesize that both of these
activities could be due to the 5-HT2C agonist properties
of these compounds. While we were investigating the
SAR around the azepino[1,7-a]indole template 5, we
examined its anxiolytic properties in a mouse shock-
9. Robichaud; A. J.; Chen; W.; McClung; C.; Stranz; M.;
Clark; D.; Deng; W.; Calvello; E. B.; Mitchell; I. S.; Lee; T.;
Rajagopalan; P.; Robertson; D. W.; Molino; B. F.; Haydar;
S. N.; Conlon; M. P.; Duffy; B. C.; Liao; J.; Pawlush; N. G.;
Fitzgerald; L. W.; Conklin; D. S.; Krause; C. M.; McElroy;
J. F.; Rohrbach; K. W.; Miller K. J.; Largent; B. L. Abstracts
of Papers; 221 National Meeting of the American Chemical
Society; San Diego; CA; 2001; MEDI-105.
10. Ennis, M. D.; Hoffman, R. L.; Ghazal, N. B.; Olson,
R. M. WO0172752, 2001; Chem. Abstr. 2001, 135, 288799.
11. Littke, A. F.; Fu, G. C. Angew. Chem. Int. Ed. Eng. 1998,
37, 3387.
induced aggression assay.13 Azepino[1,7-a]indole
5
caused a dose-dependent increase in the latency-to-fight
times comparable to that seen with alprazolam, albeit at
10-fold higher doses (Table 5). Unlike alprazolam,
however, 5 failed to induce ataxia at anxiolytic doses as
determined by a standard Rotarod assay.
12. Optical isomers were prepared to >95% ee by chromato-
graphic resolution of racemic mixtures.
13. Abe, M.; Nakai, H.; Tabata, R.; Saito, K.-I.; Egawa, M.
Jpn. J. Pharm. 1998, 76, 297.
We have described a new azepino[1,7-a]indole template
5 with high affinity and intrinsic activity at the 5-HT2C
receptor. This template possesses anxiolytic activity in