Synthesis and SAR of 2,3,3a,4-tetrahydro-1H-pyrrolo[3,4-c]isoquinolin- 5(9bH)-ones as 5-HT2C receptor agonists

Article abstract of DOI:10.1016/j.bmcl.2012.10.091

A series of 2,3,3a,4-tetrahydro-1H-pyrrolo[3,4-c]isoquinolin-5(9bH)-ones is described, several examples of which exhibit potent 5-HT_2C agonism with excellent selectivity over the closely related 5-HT_2A and 5-HT_2B receptors. Compounds such as 38 and 44 were shown to be effective in reducing food intake in an acute rat feeding model.

Full text of DOI:10.1016/j.bmcl.2012.10.091

Bioorganic & Medicinal Chemistry Letters 23 (2013) 330–335
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Synthesis and SAR of 2,3,3a,4-tetrahydro-1H-pyrrolo[3,4-c]isoquinolin-5(9bH)-
ones as 5-HT2C receptor agonists
⇑
John M. Fevig , Jianxin Feng, Karen A. Rossi, Keith J. Miller, Ginger Wu, Chen-Pin Hung, Thao Ung,
Sarah E. Malmstrom, Ge Zhang, William J. Keim, Mary Jane Cullen, Kenneth W. Rohrbach, Qinling Qu,
Jinping Gan, Mary Ann Pelleymounter, Jeffrey A. Robl
Bristol-Myers Squibb Research and Development, P.O. Box 5400, Princeton, NJ 08543-5400, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 1 May 2012
Revised 16 October 2012
Accepted 19 October 2012
Available online 2 November 2012
A series of 2,3,3a,4-tetrahydro-1H-pyrrolo[3,4-c]isoquinolin-5(9bH)-ones is described, several examples
of which exhibit potent 5-HT2C agonism with excellent selectivity over the closely related 5-HT2A and
5-HT2B receptors. Compounds such as 38 and 44 were shown to be effective in reducing food intake in
an acute rat feeding model.
Ó 2012 Elsevier Ltd. All rights reserved.
Keywords:
Serotonin
5-HT2c receptor agonist
Obesity
Obesity is widely recognized as a major and growing public
health issue, with about two-thirds of the adult US population
being considered overweight or obese.1 In addition, obesity can
exacerbate many other chronic health conditions, such as heart
disease, hypertension, diabetes and arthritis.2 Until recently, mar-
keted pharmacotherapies for obesity included orlistat3 and phen-
termine.4 However, these drugs have had limited commercial
success because of modest efficacy and several undesirable side ef-
fects. Given the severity of the obesity epidemic, significant atten-
tion has been devoted to finding new pharmacological treatments
for obesity, with many of these approaches involving modulation
of centrally-acting neurotransmitter receptors.5
The biological effects of the neurotransmitter serotonin (5-HT)
are mediated through at least 14 distinct receptor subtypes. Of
these, the 5-HT2C receptor appears to have the greatest role in
the regulation of appetite.6 In preclinical models, 5-HT2C agonists
reduce food intake and body weight in rats.7 This effect is
attenuated by co-administration of the selective 5-HT2C antagonist
SB-242086.8 5-HT2C knockout mice are obese, hyperphagic, hype-
rinsulinemic and are insensitive to the action of 5-HT2C agonists.9
Recently, the selective 5-HT2C receptor agonist lorcaserin (1)10
(Fig. 1) has gained FDA approval on the basis of showing
statistically significant weight loss in several phase 3 trials.11 These
results demonstrate that 5-HT2C agonism is a well-validated target
for anti-obesity therapy.
In targeting the 5-HT2C receptor, selectivity over other serotonin
receptor subtypes, especially the closely related 5-HT2A and 5-HT2B
receptors, is of paramount importance in order to minimize un-
wanted side effects. Specifically, agonism of 5-HT2A can lead to
undesirable CNS side effects such as hallucinations, while 5-HT2B
agonism has been associated with heart valvulopathy. The latter
finding was responsible for the withdrawal of the non-selective
5-HT2C agonists fenfluramine and the (S)-enantiomer dexfenflur-
amine from the market.12
Our efforts were focused on identifying novel structural motifs
with potent 5-HT2C agonism and high selectivity over both the 5-
HT2A and 5-HT2B receptors (preferably with no functional activity
at 5-HT2B). A series of cis-fused tetracyclic indoline 5-HT2C agonists
has been previously reported.13 The lead for this series, compound
2, was potent toward 5-HT2C but only modestly selective. Extensive
SAR led ultimately to 3, which was highly potent, efficacious and
13
displayed >100-fold binding selectivity over 5-HT2A and 5-HT2B
.
Unfortunately, the amphiphilic nature of tetracyclic indolines such
as 3 led to significant liabilities, particularly hERG inhibition and
parietal cell necrosis. More recently, a tricyclic indolone series of
5-HT2C agonists represented by 4 was reported, wherein the added
polarity of the central lactam moiety was sufficient to prevent pari-
etal cell necrosis.14
Herein we report on a series of tricyclic 5-HT2C agonists, repre-
sented generically by the cis- and trans-fused 2,3,3a,4-tetrahydro-
1H-pyrrolo[3,4-c]isoquinolin-5(9bH)-ones 5 and 6. These cores
can be seen as arising from the tricyclic indolone 4 by a 1,2-sigma
bond shift as shown by the arrow. Figure 2 shows the modeling of
⇑
Corresponding author.
E-mail address: john.fevig@bms.com (J.M. Fevig).
0960-894X/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.bmcl.2012.10.091

J. M. Fevig et al. / Bioorg. Med. Chem. Lett. 23 (2013) 330–335
331
MeO
CF3
NH
OH
O
NH
N H
H
CHO
H
a,b
c
d,e
Cl
CO2H
CONEt2
NH
H
N
O
CF3
CF3
CF3
S
8
9
7
2
3
Lorcaserin (1)
BOC
H
N
5-HT
K
i
i
i
= 1.3 nM 5-HT
= 15 nM 5-HT
= 14 nM 5-HT
K
i
= 1.3 nM
= 162 nM
= 1280 nM
5-HT EC = 9 nM
2C
2B
2A
2C
2B
2A
N
2C
50
5-HT
5-HT
K
K
i
H
H
CO2H
H
H
NH2
5-HT EC = 943 nM
2B
50
CO2Me
f,g,h
i
K
K
i
5-HT EC = 168 nM
2A
50
CO2iPr
CO2iPr
CO2iPr
NH
H
NH
NH
CF3
CF3
CF3
H
σ-bond
H
R8
R8
R
( )-12
( )-11
shift
10
N
H
H
N
N
R4
R4
H
N
O
R6
O
R6
O
R'
NH
NH
H
H
4
k
j
6
5
NH2
H
+
H
N
N
H
H
CO2H
Figure 1.
CF3
O
CF3
CF3
O
( )-15
( )-13
( )-14
Scheme 1. Reagents and conditions: (a) (COCl)2, CH2Cl2, cat. DMF, RT; then
diethylamine, CH2Cl2, RT; (b) sec-BuLi, TMEDA, THF,À78 °C; then DMF, À78 °C to
RT; (c) 6 N HCl, reflux (80% for 3 steps); (d) iPrI, K2CO3, DMF, RT (87%); (e)
(F3CCH2O)2POCH2CO2Me, KHMDS, 18-crown-6, THF, À78 °C (94%); (f) N-(methoxy-
methyl)-N-(trimethylsilylmethyl)benzylamine, TFA (0.2 equiv), EtOAc, 65 °C (93%);
(g) H2 (1 atm), 10% Pd/C, BOC2O, EtOH, RT (98%); (h) LiOH, THF/H2O, RT (92%); (i) (i)
ClCO2Et, Et3N, THF, 0 °C; (ii) NaN3, THF, 0 °C; (iii) tol, 100 °C; (iv)1 N HCl, THF, RT
(45%); (j) KOH, EtOH, 80 °C (90%); (k) (i) BOC2O, 1 N NaOH, dioxane/THF, RT; (ii)
iodomethane, NaHCO3, DMF, RT; (iii) 4 N HCl in dioxane, Et2O, RT; (iv) NaOMe,
MeOH, 65 °C.
NH
H
NH
H
Blue: (2)
Blue: (2)
H
H
Green: (6)
Magenta: (5)
2,3,3a,4-tetrahydro-1H-pyrrolo[3,4-c]isoquinolin-5(9bH)-one ( )-
13, along with variable amounts of the unexpected racemic
trans-fused isomer ( )-14. In this initial synthesis, ( )-14 was a
minor (5%) byproduct which was not isolated or fully character-
ized. On subsequent scaleup, however, the Curtius rearrangement
step ( )-11 to ( )-12 was found to proceed in poor isolated yield.
The major product, isolated from the aqueous layer, was deter-
mined to be the diamino acid ( )-15, wherein the lactam ring
had failed to form but the isopropyl ester had been hydrolyzed.
Surprisingly, upon conversion of ( )-15 to the tricyclic lactam only
the trans-fused pyrroloisoquinolinone ( )-14 was obtained, indi-
cating that epimerization had occurred at some point prior to or
during the Curtius rearrangement sequence. The ring fusion ste-
reochemistry of ( )-13 and ( )-14 were verified by NOE
experiments.
In contrast to our initial hypothesis, the in vitro data for com-
pounds ( )-13 and ( )-14 in Table 1 indicate that the trans-fused
core was about 15-fold more potent than the cis-fused core in
terms of 5-HT2C binding and functional activity. Subsequently,
the major research focus was redirected toward the SAR around
the trans-fused core. A selective synthesis of the trans-fused core
was developed as described in Scheme 2. Beginning with com-
pound 9, alkylation with methyl iodide afforded the corresponding
aldehyde methyl ester which was treated with benzyl, 2-(dimeth-
oxyphosphoryl)acetate 16 and sodium hydride to give the trans
olefin 17. Azomethine ylide [3+2] cyclization (as described in
Scheme 1) afforded ( )-18, where the olefin geometry was pre-
served in producing the trans pyrrolidine. Catalytic hydrogenation
in the presence of BOC anhydride effected double debenzylation
and in situ N-protection of the pyrrolidine, giving ( )-19 in excel-
lent yield. The Curtius rearrangement sequence was performed un-
der slightly modified conditions relative to those described in
Scheme 1. At the isocyanate stage, it was realized that hydrolysis
with very dilute acid (0.1 N HCl) allowed for the hydrolysis of the
isocyanate without concomitant N-BOC removal. Ring-closure of
the aminoester intermediate was then accomplished with sodium
NH
NH
CF3
O
CF3
O
Figure 2.
both isomers 5 and 6 (where R6 = CF3 and R4 = R8 = H) and their
overlay onto compound 2. The cis-fused core 5 maintains the ex-
pected concave shape and appears to have a more favorable over-
lap with the cis-fused tetracyclic indoline 2. The trans-fused core 6
was expected to be less potent because it loses the concave shape
and does not appear to allow for optimal overlap of the basic
nitrogen.
The initial synthesis of the cis-fused core 5 is shown in
Scheme 1. A CF3 substitutent ortho to the central ring carbonyl
was expected to provide considerably more potency relative to
the unsubstituted analog, as previously observed in a related ser-
ies.14 Directed metallation of the diethylamide derived from car-
boxylic acid 7, followed by reaction with DMF, afforded aldehyde
8, which was subsequently hydrolyzed under strongly acidic con-
ditions to give 9. Treatment of 9 with isopropyl iodide and potas-
sium carbonate afforded an isopropyl ester aldehyde, which was
converted to the cis olefin 10 by olefination under Still’s condi-
tions.15 Installation of the N-benyzl pyrrolidine moiety was readily
accomplished by [3+2] cycloaddition of 10 with the azomethine
ylide
generated
from
N-methoxymethyl-N-(trimethylsilyl-
methyl)benzyl amine and catalytic TFA in refluxing EtOAc. Cata-
lytic hydrogenation in the presence of BOC anhydride afforded
the N-BOC derivative, which was subjected to selective hydrolysis
of the methyl ester to afford ( )-11. Curtius rearrangement was ini-
tially accomplished in a multistep sequence involving formation of
the acyl azide via the mixed anhydride of acid ( )-11, thermal rear-
rangement of the acyl azide in refluxing toluene, and acidic hydro-
lysis of the resulting isocyanate, which generally afforded the
diamine ( )-12. Ring-closure of the primary amine onto the isopro-
pyl ester under basic conditions afforded the racemic cis-fused

332
J. M. Fevig et al. / Bioorg. Med. Chem. Lett. 23 (2013) 330–335
Table 1
Serotonin (5-HT) binding and functional activity of 2,3,3a,4-tetrahydro-1H-pyrrolo[3,4-c]isoquinolin-5(9bH)-ones
NH
H
NH
NH
NH
H
H
H
H
9
9
8
7
8
7
H
H
H
H
N
N
N
N
4
H
H
R
6
CF
O
CF
O
CF
O
R
O
( 3)-13
( 3)-14
3
34
35-39
b
Compounda
#
5-HT2C (nM)b
Binding Ki
192 92
5-HT2B (nM)b
5-HT2A (nM)
R4
R6
Function EC50 (IA)
Binding Ki
Function EC50 (IA)
Binding Ki
Function EC50 (IA)
( )-13
( )-14
34
35
36
37
38
39
H
H
H
H
CF3
CF3
CF3
CF3
CF3
Cl
30 2.4 (1.2)
2.0 1.3 (1.1)
62 21 (0.9)
1.0 0.2 (0.9)
3.3 1.6 (0.8)
2.6 1.2 (0.9)
1.1 0.4 (0.9)
43 18 (0.9)
2373 304
73 27
6244 463
ND
ND
ND
1882 71
91 15
ND
12
6
63 (n = 1) (1.3)
>10000
1052 415
9.0 5.8
46 22
1796
41
52
33
1
9.9 4.4 (0.6)
32 6 (0.8)
31 (n = 1) (0.6)
94 68 (0.5)
>10000
8
10 (n = 1) (1.8)
89 76 (0.9)
124 25 (1.0)
127 82 (1.0)
>10000
Mec
H
24 11
79 45
258 80
56 20
73
5
40
23
9
7
H
H
Me
H
3.9 2.2
1062 1407
5812 3869
7957 2065
a
All compounds were tested as the hydrochloride salt and exhibited satisfactory analytical data.
Radioligand binding studies were conducted to determine the binding affinities (Ki values) of compounds for the human recombinant 5-HT2A, 5-HT2B, and 5-HT2C
b
receptors. Functional screenings were carried out in HEK293E cells expressing the human 5-HT2A, 5-HT2B, or 5-HT2C receptor. The reported intrinsic activity (IA) is relative to
receptor activation by serotonin at 3 M (defined as 1). ND = value not determined. See Ref. 14 for complete description of all in vitro and in vivo assays.
l
c
N-methyl lactam was prepared using sodium hydride and methyl iodide in THF prior to N-BOC deprotection.
Bn
N
Schemes 1 and 2. Where the carboxylic acids were not commer-
OH
O
cially available, as was the case with many of the 8-alkyl analogs
in Table 2, the synthesis was carried out in two ways as shown
in Scheme 3. In the first route, the diethylamide 25 (readily derived
from the appropriately substituted 4-bromobenzoic acid 24) was
subjected to palladium-catalyzed coupling of 25 with either an al-
kyl zinc reagent or with a vinyl boronic acid followed by catalytic
hydrogenation, affording amide 26. This material was converted to
the final products following the procedures described in Schemes 1
and 2. In the second route, selective demethylation of the dime-
thoxy ester 27 with BCl3 followed by standard triflation afforded
28. Palladium-catalyzed Heck coupling of 28 with benzyl acrylate
gave 29. This material was carried through as described in
Scheme 3 to provide, after the Curtius rearrangement, the racemic
trans-fused tricyclic compound ( )-31. O-Demethylation was best
accomplished with thiophenol and potassium carbonate at ele-
vated temperature, which was followed by standard triflation of
the resulting phenol to produce ( )-32. Palladium-catalyzed cou-
pling with dialkyl zinc reagents as described above afforded the
8-alkyl-substituted analogs. Chiral separation on an OD column
and N-BOC deprotection gave the final products 33.
The in vitro binding, functional and selectivity data for analogs
that explore the ring-fusion stereochemistry, the effects of the C6
aryl substituent, and the effects of lactam N-substitution are
shown in Table 1. As previously mentioned, the greater potency
of racemic trans-fused ( )-14 relative to racemic cis-fused ( )-13
was an unexpected result based on the initial modeling shown in
Figure 1. Comparison of the enantiomers 34 and 35 clearly indi-
cates that the potency resides in the 3aR, 9bS-enantiomer (the iso-
mer which is modeled in Fig. 1). While compound 35 is a very
potent full agonist of 5-HT2C, it is only ꢀfour fold selective versus
5-HT2A and 5-HT2B in binding affinity, and only ꢀten fold selective
in functional activity. The N-methyl lactam 36 maintains 5-HT2C
potency, but this substitution had little effect on selectivity. Chloro
(37) and methyl (38) are effective C6 substituents, both of which
are full agonists with comparable 5-HT2C functional potency rela-
tive to the trifluoromethyl analog 35. The methyl analog 38 also
has good functional selectivity, being 85-fold selective over 5-
CO2Bn
CO2Me
a,b
c
CO2Bn
CO2Me
O
CF3
CF3
CF3
17
( )-18
9
BOC
N
BOC
N
H
d
e
f
H
H
CO2H
CO2Me
N
CF3
O
CF3
( )-19
( )-20
O
O
BOC
BOC
S
O
N
N
H
H
N
H
g, h
H
+
H
H
N
N
N
H
H
H
CF3
O
CF3
O
CF3
O
23
21
22
Structure confirmed by
X-ray crystallography
Scheme 2. Reagents and conditions: (a) MeI, K2CO3, DMF, RT (95%); (b)
(MeO)2POCH2CO2Bn (16), NaH, THF, 0 À25 °C (75%); (c) N-(methoxymethyl)-N-
(trimethylsilylmethyl)benzylamine, TFA (0.2 equiv), CH2Cl2, 40 °C (95%); (d) H2
(1 atm), 10% Pd/C, BOC2O, EtOH, RT (97%); (e) (i) ClCO2Et, Et3N, THF, 0 °C; (ii) NaN3,
THF, 0 °C; (iii) tol, 100 °C; (iv) 0.1 N HCl, THF, RT; v) NaOMe, MeOH, 65 °C (70%); (f)
chiral separation, OD column, 30% MeOH/EtOH/70% heptane; (g) 12 N HCl, dioxane,
RT (91%); (h) (1S)-(+)-10-camphorsulfonyl chloride, Et3N, THF, RT (55%).
methoxide in refluxing methanol to deliver the desired trans-fused
pyrroloisoquinolinone ( )-20. Chiral separation of enantiomers
was conveniently performed on an OD column, affording the enan-
tiomers 21 and 22. Subsequent BOC removal gave the homochiral
amines 35 and 34, respectively (Table 1). Compound 34 was con-
verted to the sulfonamide 23, the absolute structure of which
was established by X-ray crystallography.16
Additional analogs in Tables 1 and 2 were prepared from the
appropriate carboxylic acid following the procedures outlined in
HT2B (as a partial agonist) and >100-fold selective over 5-HT2A
The diminished potency of the C6 unsubstituted analog 39 is con-
.

J. M. Fevig et al. / Bioorg. Med. Chem. Lett. 23 (2013) 330–335
333
Table 2
Serotonin (5-HT) binding and functional activity of disubstituted compounds
NH
H
R
9
8
H
N
7
4
R
6
R
O
40-52
Compounda
5-HT2C (nM)
5-HT2B (nM)
5-HT2A (nM)
b
b
b
#
R6
R
R4
Binding Ki
Function EC50 (IA)
Binding Ki
Function EC50 (IA)
Binding Ki
Function EC50 (IA)
40
41
42
43
44
45
46
47
48
49
50
51
52
Cl
Cl
Cl
Cl
Cl
Cl
Cl
Me
Me
Me
Me
CF3
CF3
8-Me
8-Et
8-nPr
9-Et
7-Cl
7-Me
8-Et
8-Me
8-Et
8-cycPr
8-iPr
8-Et
H
H
H
H
H
H
253
1
17 1 (1.0)
26 16 (0.9)
218 60 (1.0)
11 3.5 (0.9)
3.3 1.8 (0.9)
2.2 0.8 (0.9)
82 24 (0.9)
20 7 (1.1)
12 5 (1.0)
27 6 (0.8)
59 28 (1.0)
4.3 1 (0.8)
41 18 (1.0)
478 68
905 1399
776 261
79 59
523 (n = 1) (0.3)
>10000
>10000
2330 (n = 1) (1.0)
414 468 (1.3)
ND
>10000
207 74 (0.6)
>10000
217 71
1488 1287
4441 509
359 53
605 (n = 1) (0.5)
1872 442 (1.1)
>10000
1674 888 (0.8)
128 44 (1.0)
21 4 (1.0)
44 28
556 181
55 19
13
9
48 29
72 11
4.9 1.4
393 195
86 27
46 11
87
83 18
34 11
151 35
4.0 1.8
1717 1033
191 91
188 124
177 62
535 107
87 29
13
1
Mec
H
H
H
H
6528 4006
540 66
703 362
3180 508
3334 114
157 34
>10000
303 63 (0.8)
856 221 (0.8)
>3000
>3000
605 349 (0.8)
2089 1025 (0.8)
9
>10000
706 515 (0.3)
101 31 (0.6)
268 118 (0.5)
H
8-Et
Mec
140 36
328 34
a,b,c
See Table 1 legend.
pounds 35–38, none of these compounds possessed adequate
selectivity over 5-HT2A and 5-HT2B
Br
Br
R8
b or c
a
.
CO2H
R6
CONEt2
OTf
CONEt2
R6
Introduction of additional substituents at the C7, C8 or C9 posi-
tion of the aromatic ring in certain instances provided greater
selectivity over 5-HT2A and 5-HT2B. This trend has been observed
in multiple 5-HT2C chemotypes, especially with substituents at
the 8-position.13,14 In the C6 chloro series (compounds 40–46),
increasing the size of the C8 substituent from methyl to n-propyl
had a significant effect on potency and selectivity. Within this
set, the optimal compound was the 8-ethyl analog 41. 5-HT2C po-
tency was maintained (EC50 = 26 nM, full agonist) but 5-HT2A func-
tional potency was diminished (72-fold selective) and the
compound was devoid of 5-HT2B functional potency, despite main-
taining measurable 5-HT2B binding potency. Increasing the substi-
tuent size to n-propyl (42) led to an unacceptable drop off in 5-
HT2C functional activity. The highly potent 9-ethyl analog 43 was
quite selective over 5-HT2A (>150-fold) and 5-HT2B (>200-fold)
but still exhibited full agonism at both receptors. Examples with
a C7 substituent (44 and 45) maintain excellent functional potency
at 5-HT2C but do not have sufficient selectivity. A similar series of
C8 analogs in the C6 methyl series (compounds 47–50) was also
prepared. The optimal substituents for increasing selectivity were
the 8-ethyl (48) and 8-cyclopropyl (49) groups. Both compounds
were potent full agonists at 5-HT2C but were devoid of 5-HT2B func-
tional potency, despite having modest 5-HT2B binding potency,
demonstrating that within this series certain aryl substitutions
can cause some compounds to show a disconnect between binding
and functional potency, especially with the 5-HT2B receptor. The
isopropyl analog 50 unexpectedly regained some modest 5-HT2B
functional potency. The optimal C8 ethyl substitution was also ap-
plied to the C6 trifluoromethyl series, where compound 51 was
found to have very potent 5-HT2C potency but poor selectivity. In
an effort to modulate physicochemical properties, the 8-ethyl sub-
stituent was combined with N-alkylation of the lactam ring, afford-
ing 46 and 52. Both compounds exhibited diminished 5-HT2C
binding and functional potency relative to their corresponding
NH analogs 41 and 51, respectively.
R6
24
25
26
MeO
MeO
CO2Bn
CO2Et
MeO
OMe
d,e
f
CO2Et
Me
28
CO2Et
Me
29
BOC
Me
27
BOC
N
N
MeO
MeO
j,k
g,h
i
CO2H
CO2Me
H
H
N
Me
O
CF3
( )-30
( )-31
BOC
H
N
NH
H
R8
TfO
l,m,n
H
H
H
H
N
N
Me
O
Me
O
( )-32
33
Scheme 3. Reagents and conditions: (a) (COCl)2, CH2Cl2, cat. DMF, RT; then
diethylamine, CH2Cl2, RT; (b) R2Zn, PdCl2(dppf), THF, 65 °C (50–90%); (c) (i) trans
CH3CH@CHB(OH)2, Pd(PPh3)4, K2O3, DME/H2O, 90 °C, (ii) H2 (1 atm), 10% Pd/C,
EtOH, RT (50%); (d) BCl3, CH2Cl2, À78 °C, (78%); (e) triflic anhydride, pyridine, 0 °C,
(98%); (f) CH2@CHCO2Bn, PdCl2(dppf), Et3N, DMF, 90 °C (85%); (g) N-(methoxy-
methyl)-N-(trimethylsilylmethyl)benzylamine, TFA (0.2 equiv), CH2Cl2, 40 °C; (h)
H2 (1 atm), 10% Pd/C, BOC2O, EtOH, RT (66% for 2 steps); (i) (i) ClCO2Et, Et3N, THF,
0 °C; (ii) NaN3, THF, 0 °C; (iii) tol, 100 °C; (iv) 0.25 N HCl, THF, RT; (v) K2CO3, MeOH,
65 °C (78%); (j) PhSH, K2CO3, NMP, 155 °C (75%); (k) triflic anhydride, pyridine, 0 °C,
(81%); (l) R2Zn, PdCl2(dppf), THF, 65 °C (50–90%); (m) chiral separation, OD column,
30% MeOH/EtOH/70% heptane; (n) 12 N HCl, dioxane, RT (90%)
sistent with previous observations in related series14 and confirms
our view that a C6 substituent is critical for potent 5-HT2C activity.
Despite the excellent 5-HT2C potency and full agonism of com-
Several analogs were selected for abbreviated rat pharmacoki-
netic studies, as shown in Table 3. These studies were designed
to assess general exposure (Cmax, AUC (0–4 h)) and also the rela-

334
J. M. Fevig et al. / Bioorg. Med. Chem. Lett. 23 (2013) 330–335
Table 3
Rat pharmacokinetic data
Table 4
Acute feeding studies in ratsa
Compound
Dose
(mpk)
Cmax
(nM)
AUC 0–4 h
(nM⁄h)
Brain/Plasma
ratio
Cmpd Dose
Pellets consumed (%
+10 mpk 5HT2C Antagonista
(% change from vehicle, 20 h)
(mpk) change from vehicle)
38
41
44
46
48
51
52
10
10
10
10
8
1261
7302 (0–8 h)
6111
7591
42838
5149
22425
15244
1.9
0.1
0.2
0.2
0.6
0.1
0.3
2 h
8 h
20 h
2370
2395
15783
1694
6688
5117
41
43
48
38
10
10
10
3
10
30
3
10
30
10
0
+2
+2
ND
ND
ND
ND
ND
ND
ND
ND
ND
+7
À18
À16
À6
+20
0
À4
À80⁄
À98⁄
À100⁄
À9
À31⁄
À51⁄
À88⁄
À12
À33⁄
À53⁄
À46⁄
À26⁄
À34⁄
À54⁄
À10
À19⁄
À34⁄
À30⁄
10
10
44
44
PK studies were performed in SpragueDawley rats using a PO dosing solution of 15%
propylene glycol, 1% tween, 84% water. Plasma sampling of compounds was
determined over a 4 h period except where noted. B/P ratio determined at end of
study.
À71⁄
À94⁄
À94⁄
⁄p <0.05.
a
Effects on feeding in an acute (20 h) rat operant model upon treatment with
selected compounds. Male Sprague-Dawley rats (n = 6) were dosed orally with test
compound or vehicle (14% propylene glycol, 1% tween, 85% water) 60 min prior to
the onset of the dark cycle. Pellets consumed for compound treated animals were
compared to that of vehicle treated animals in order to determine percent reduction
of food intake over a 20 h period. Data is shown for the 2 h, 8 h and 20 h timepoints.
See Ref. 14 for a detailed description of all in vivo studies. For the reversal study of
compound 44, test compound was dosed alone or with 10 mpk of the selective 5-
HT2C antagonist SB-243213.17
tive exposure of the compounds in the brain versus the plasma. In
general, the compounds had reasonable exposures, with
Cmax > 1200 nM when administered orally at a dose of 10 mpk.
Compounds 46, 51 and 52 all had especially high exposures. Brain
to plasma ratios ranged from 0.1 to 1.9 (but trending toward lower
values), with the lower ratios generally arousing concern over the
ability of the compounds to cross the blood–brain barrier. For rea-
sons which are not well understood, those compounds with an
electron-withdrawing C6 substituent (41, 44, 46, 51 and 52) had
lower brain/plasma ratios than those with an electron-donating
C6 substituent (38 and 48). N-methylation of the lactam only mod-
estly increased the brain/plasma ratio (compare 46 vs. 41, and 52
vs. 51).
sired profile of potent 5-HT2C functional activity, no 5-HT2B func-
tional agonism and potent 5-HT2C-mediated reduction of food
intake in a rat acute feeding model.
Acknowledgments
Several of the most promising compounds were selected to as-
sess their ability to reduce feeding in a 20 h rat operant feeding
model, as shown in Table 4. Compounds were dosed 60 min prior
to the onset of the dark cycle (the most active time of feeding)
and the number of food pellets consumed relative to vehicle-trea-
ted animals was assessed at several timepoints to determine per-
cent reduction in feeding. Data for the 2 h, 8 h and 20 h
timepoints is shown. Compounds 41, 43 and 48 failed to elicit a
significant reduction in food intake at any timepoint. Compound
38, which had the highest brain/plasma ratio among this set, sig-
nificantly reduced food intake at all doses tested, with the bulk
of the reduction occurring at the earlier timepoints. Compound
44 was effective at reducing food intake at 10 and 30 mpk. Finally,
a reversal study was performed on compound 44, shown at the
bottom of Table 4, which involved dosing 44 alone or in conjunc-
tion with SB-243213, a selective 5-HT2C antagonist.17 The data
show that co-administration of SB-243213 completely reversed
the food reduction effect, providing strong evidence that the ob-
served feeding effects of 44 were mediated by agonism of the 5-
HT2C receptor.
In summary, it was found that the trans-fused 2,3,3a,4-tetrahy-
dro-1H-pyrrolo[3,4-c]isoquinolin-5(9bH)-one core provided a po-
tent framework to effect 5-HT2C receptor agonism. Substitution
at C6 was required for good potency, while additional substituents
on the aryl ring were found to modulate both 5-HT2C potency and
selectivity versus the 5-HT2B and 5-HT2A receptors. Compounds
such as 41, 48 and 49, were potent 5-HT2C full agonists but were
devoid of 5-HT2B functional potency. This is a critical attribute in
minimizing the risk of heart valvulopathy in this class of com-
pounds. Several compounds were assessed in vivo for their ability
to reduce food intake in a rat acute feeding model. Two analogs, 38
and 44, were found to significantly reduce food intake. The satiety
effect of 44 was completely reversed when co-administered with a
selective 5-HT2C antagonist, demonstrating that this compound
was acting via the 5-HT2C receptor. Despite these attributes, we
have been unable to identify a compound in this series with the de-
The authors wish to thank Mike Galella and Jack Gougoutas for
the determination of the crystal structure of compound 23.
References and notes
1. http://www.cdc.gov/nchs/fastats/overwt.htm.
2. Ogden, C. L.; Carroll, M. D.; Flegal, K. M. Endocrinol. Metab. Clin. North Am. 2003,
32, 741.
3. Foxcroft, D. R.; Milne, R. Obes. Rev. 2000, 1, 121.
4. Ioannides-Demos, L. L.; Proietto, J.; McNeil, J. J. Drugs 2005, 65, 1391.
5. (a) Cooke, D.; Bloom, S. Nat. Rev. Drug Disc. 2006, 5, 919; (b) Neff, L. M.; Aronne,
L. J. Curr. Atheroscler. Rep. 2007, 9, 454.
6. (a) Wacker, D. A.; Miller, K. J. Curr. Opin. Drug Discovery Dev. 2008, 11, 438; (b)
Miller, K. J. Mol. Interventions 2005, 5, 282; (c) Monck, N. J. T.; Kennett, G. A.
Prog. Med. Chem. 2008, 46, 281.
7. Walsh, A. E. S.; Smith, K. A.; Oldman, A. D.; Williams, C.; Goodall, E. M.; Cowen,
P. J. Psychopharmacology 1994, 116, 120.
8. Kennett, G. A.; Wood, M. D.; Bright, F.; Trail, B.; Riley, G.; Holland, V.; Avenell, K.
Y.; Stean, T.; Upton, N.; Bromidge, S.; Forbes, I. T.; Brown, A. M.; Middlemiss, D.
N.; Blackburn, T. P. Neuropharmacology 1997, 36, 609.
9. (a) Tecott, L. H.; Sun, L. M.; Akana, S. F.; Strack, A. M.; Lowenstein, D. H.;
Dallman, M. F.; Julius, D. Nature 1995, 374, 542; (b) Vickers, S. P.; Clifton, P. G.;
Dourish, C. T.; Tecott, L. H. Psychopharmacology 1999, 143, 309.
10. (a) Thomsen, W. J.; Grottick, A. J.; Menzaghi, F.; Reyes-Saldana, H.; Espitia, S.;
Yuskin, D.; Whelan, K.; Martin, M.; Morgan, M.; Chen, W.; Al-Shamma, H.;
Smith, B.; Chalmers, D.; Behan, D. J. Pharmacol. Exp. Ther. 2008, 325, 577; (b)
Smith, S. R.; Prosser, W. A.; Donahue, D. J.; Morgan, M. E.; Anderson, C. M.;
Shanahan, W. R. Obesity 2009, 17, 494; (c) Smith, B. M.; Smith, J. M.; Tsai, J. H.;
Schultz, J. A.; Gilson, C. A.; Estrada, S. A.; Chen, R. R.; Park, D. M.; Prieto, E. B.;
Gallardo, C. S.; Sengupta, D.; Dosa, P. I.; Covel, J. A.; Ren, A.; Webb, R. R.; Beeley,
N. R. A.; Martin, M.; Morgan, M.; Espitia, S.; Saldana, H. R.; Bjenning, C.;
Whelan, K. T.; Grottick, A. J.; Menzaghi, F.; Thomsen, W. J. J. Med. Chem. 2008,
51, 305.
11. http://arna.client.shareholder.com/releasedetail.cfm?ReleaseID=419009.
12. (a) Fitzgerald, L. W.; Burn, T. C.; Brown, B. S.; Patterson, J. P.; Corjay, M. H.;
Valentine, P. A.; Sun, J.-H.; Link, J. R.; Abbaszade, I.; Hollis, J. M.; Largent, B. L.;
Hartig, P. R.; Hollis, G. F.; Meunier, P. C.; Robichaud, A. J.; Robertson, D. W. Mol.
Pharmacol. 2000, 57, 75; (b) Rothman, R. B.; Baumann, M. H.; Savage, J. E.;
Rauser, L.; McBride, A.; Hufeisen, S. J.; Roth, B. L. Circulation 2000, 102, 2836.
13. Lee, T.; Robichaud, A. J.; Chen, W.; Lu, Y.; Dowdell, S.; Boyle, K. E.; Mitchell, I. S.;
Fevig, J. M.; Wexler, R. R.; Miller, K. J.; Largent, B. L.; Rohrbach, K. W.; Devenny,
J. J.; McElroy, J. F. Abstracts of Papers, 229th ACS National Meeting, San Diego,
CA; American Chemical Society: Washington, DC, 2005; MEDI-283.
14. Wacker, D. A.; Varnes, J. G.; Malmstrom, S. E.; Cao, X.; Hung, C.-P.; Ung, T.; Wu,
G.; Zhang, G.; Zuvich, E.; Thomas, M. A.; Keim, W. J.; Cullen, M. J.; Rohrbach, K.

J. M. Fevig et al. / Bioorg. Med. Chem. Lett. 23 (2013) 330–335
335
W.; Qu, Q.; Narayanan, R.; Rossi, K.; Janovitz, E.; Lehman-McKeeman, L.;
Malley, M. F.; Devenny, J.; Pelleymounter, M. A.; Miller, K. J.; Robl, J. A. J. Med.
Chem. 2007, 50, 1365.
17. Bromidge, S. M.; Dabbs, S.; Davies, D. T.; Davies, S.; Duckworth, D. M.; Forbes, I.
T.; Gaster, L. M.; Ham, P.; Jones, G. E.; King, F. D.; Mulholland, K. R.; Saunders, D.
V.; Wyman, P. A.; Blaney, F. E.; Clarke, S. E.; Blackburn, T. P.; Holland, V.;
Kennett, G. A.; Lightowler, S.; Middlemiss, D. N.; Trail, B.; Riley, G. J.; Wood, M.
D. J. Med. Chem. 2000, 43, 1123.
15. Still, W. C.; Gennari, C. Tetrahedron Lett. 1983, 24, 4405.
16. CCDC 897809 contains the supplementary crystallographic data for compound
23. These data can be obtained free of charge from The Cambridge
Crystallographic Data Centre via http://www.ccdc.cam.ac.uk/data_request/cif.