Discovery of a lead series of potent benzodiazepine 5-HT2C receptor agonists with high selectivity in functional and binding assays

Article abstract of DOI:10.1016/j.bmcl.2019.126929

A series of potential new 5-HT₂ receptor scaffolds based on a simplification of the clinically studied, 5-HT_2CR agonist vabicaserin, were designed. An in vivo feeding assay early in our screening process played an instrumental part in the lead identification process, leading us to focus on a 6,5,7-tricyclic scaffold. A subsequent early SAR investigation provided potent agonists of the 5-HT_2C receptor that were highly selective in both functional and binding assays, had good rat PK properties and that significantly reduced acute food intake in the rat.

Full text of DOI:10.1016/j.bmcl.2019.126929

Journal Pre-proofs
Discovery of a lead series of potent benzodiazepine 5-HT2C receptor agonists
with high selectivity in functional and binding assays
Albert Ren, Xiuwen Zhu, Konrad Feichtinger, Juerg Lehman, Michelle Kasem,
Thomas O. Schrader, Amy Wong, Huong Dang, Minh Le, John Frazer, David
J. Unett, Andrew J. Grottick, Kevin T. Whelan, Michael E. Morgan, Carleton
R. Sage, Graeme Semple
PII:
S0960-894X(19)30912-6
DOI:
https://doi.org/10.1016/j.bmcl.2019.126929
BMCL 126929
Reference:
Bioorganic & Medicinal Chemistry Letters
To appear in:
Received Date:
Revised Date:
Accepted Date:
13 November 2019
17 December 2019
20 December 2019
Please cite this article as: Ren, A., Zhu, X., Feichtinger, K., Lehman, J., Kasem, M., Schrader, T.O., Wong, A.,
Dang, H., Le, M., Frazer, J., Unett, D.J., Grottick, A.J., Whelan, K.T., Morgan, M.E., Sage, C.R., Semple, G.,
Discovery of a lead series of potent benzodiazepine 5-HT2C receptor agonists with high selectivity in functional and
binding assays, Bioorganic & Medicinal Chemistry Letters (2019), doi: https://doi.org/10.1016/j.bmcl.2019.126929
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover
page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will
undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing
this version to give early visibility of the article. Please note that, during the production process, errors may be
discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
© 2019 Published by Elsevier Ltd.

Discovery of a lead series of potent benzodiazepine 5-HT2C receptor agonists with high
selectivity in functional and binding assays
Albert Rena, Xiuwen Zhua, Konrad Feichtingera, Juerg Lehmana, Michelle Kasema, Thomas O. Schradera,
Amy Wonga, Huong Dangb, Minh Leb, John Frazerb, David J. Unettb, Andrew J. Grottickb, Kevin T. Whelana,
Michael E. Morgana, Carleton R. Sageb and Graeme Sempleb*
a) Arena Pharmaceuticals, 6154 Nancy Ridge Drive, San Diego, CA, 92121, USA.
b) Beacon Discovery, 6118 Nancy Ridge Drive, San Diego, CA, 92121, USA
Abstract
A series of potential new 5-HT2 receptor scaffolds based on a simplification of the clinically studied, 5-
HT2CR agonist vabicaserin, were designed. An in vivo feeding assay early in our screening process played
an instrumental part in the lead identification process, leading us to focus on a 6,5,7-tricyclic scaffold. A
subsequent early SAR investigation provided potent agonists of the 5-HT2C receptor that were highly
selective in both functional and binding assays, had good rat PK properties and that significantly reduced
acute food intake in the rat.
The 5-HT2C receptor (5-HT2CR) has been proven to be an important mediator of food intake. The receptor
is primarily expressed in brain, including the hypothalamus, a region known to be involved in the
regulation of appetite and feeding.1 In rodents, classical 5-HT2CR agonists including mCPP, Ro 60-0175 and
fenfluramine (via its active metabolite, norfenfluramine), reduce food intake and body weight.2 These
effects are reversed upon pre-administration of a selective 5-HT2CR antagonist. In addition, 5-HT2C
receptor null mice are hyperphagic and mildly obese, and the hypophagic effects of even nonselective 5-
HT2C agonists are significantly attenuated in such animals.3 Non-selective 5-HT2 receptor agonists including
fenfluramine and dexfenfluramine also induced weight loss in the clinical setting, and were widely
prescribed anorectic agents in the 1990s, particularly when used in combination with phentermine (the
combination known as fen-phen). Retrospective safety analyses of these patients however, showed a
strong association with heart valve disease as well as a connection to pulmonary hypertension,4 resulting
in their withdrawal from the market in the US in 1997. This valvulopathy finding was later connected to
the activation of the 5-HT2BR in cardiac valvular interstitial cells.5 It has also been proposed that activation

of the central 5-HT2AR could be the cause of some of the adverse CNS effects observed including changes
in perception and even hallucination.6 The challenge in bringing new 5-HT2CR compounds to the market
then, can be largely defined by the need to rigorously address receptor selectivity to deal with these
potential side effect issues. This approach eventually resulted in the discovery and FDA approval of the
selective 5-HT2CR agonist lorcaserin (1, BELVIQ®, Figure 1).7 In addition to obesity, the 5-HT2CR has been
demonstrated to be involved in other diseases.
N
Cl
N
NH
NH
NH
1 Lorcaserin
,
2 WAY-163909
,
3 Vabicaserin
,
5-HT2CR; Ki = 15 nM
5-HT2CR; Ki = 10.5 nM
5-HT2CR; Ki = 3 nM
selectivity >7x vs 5-HT2AR
selectivity: >20x vs 5-HT2AR;
selectivity >50x vs 5-HT2AR
7b
12
13
and >11x vs 5-HT
R
and >40x vs 5-HT
R
and >4x vs 5-HT R
2B
2B
2B
Figure 1. Structures of historical selective 5-HT2CR agonists with their literature binding affinity data
The 5-HT2CR agonist vabicaserin (3) shows good 5-HT2R selectivity in functional assays and has been shown
to be clinically effective in the treatment of the positive symptoms of schizophrenia8 and the non-selective
compound fenfluramine showed clinical efficacy as a treatment for the rare childhood epilepsy disorder
Dravet’s Syndrome.9 It remains to be seen whether fenfluramine can be used safely for long-term
treatment in this population in spite of its clear lack of selectivity against the 5-HT2BR. Interestingly, a small
trial with the much more selective 5-HT2CR agonist lorcaserin in Dravet patients demonstrated similar and
promising reductions in seizure frequency and severity.10 Both lorcaserin and other selective 5HT2CR
agonists have also demonstrated significant potential for the treatment of stress incontinence in
preclinical models.11 With these additional potential indications, some of which remain to be fully
validated in the clinic, a high safety bar will still be required so there remains a need to identify new 5-
HT2CR agonist compounds with improved receptor selectivity compared to lorcaserin.
During the lorcaserin approval process, we initiated a follow-on program to identify a second-generation
compound to be able to investigate additional potential indications such as those outlined above. As
discussed, our primary focus in this effort was on receptor selectivity as measured in both functional and
binding assays. Our secondary objective was to expand structural diversity amongst the ligands we
investigated compared to our original program.

In one such iteration, we were interested in investigating a series of analogues of the two related
molecules WAY-163909 (2)12 and vabicaserin (3, SCA-136).13 In our hands, vabicaserin was somewhat less
potent than lorcaserin but was highly selective in functional assays, with only a very weak partial agonist
effect at the 5-HT2AR (Table 1). It was, however, essentially devoid of selectivity between 5-HT2CR and 5-
HT2BR in binding assays. We envisaged instead, an alternative simplified scaffold in which the fused
cyclopentane ring would be absent. This reduction in complexity from 4 rings to 3 should provide two
significant advantages. The first would be a simplification of the synthesis, partly by virtue of the removal
of 2 chiral centres, with the second related advantage being easier access to substitution around the
aromatic ring and potentially in other positions.
Scheme 1 : General synthesis of the 6,5,7-ring series
R
R
R
R
(i)
(ii)
(iii)
N
N
N
N
H
H
N
H
NH2
4
5
6
7a-g
(iv)
R
N
N
H
7h-7m
Reagents and Conditions i) Et3SiH, TFA, r.t. ii) a) 2-hydroxy acetonitrile, 105°C b) CoCl2.6H2O, NaBH4, r.t.
c) HCl or iii) formaldehyde, MeOH, TFA, 80 °C iv) For R = 8-Br either RB(OH)2, bis(di-t-butyl-p-
dimethylaminophenylphosphino)palladium chloride, Na2CO3, dioxane, H2O or RZnCl, dihydrogen di-µ-
chlorodichlorobis(di-tert-butylphosphinito-κΡ)dipalladate (2-) (POPd2), THF.
Our first iteration of this approach was to maintain the diazepine ring, based on the assumption that the
secondary amine in that ring formed the key binding interaction with the conserved aspartate residue in
TM3 in the 5-HT2 receptor sub-family.14 We thereafter investigated a range of ring sizes and hetero-atom
insertions in the second partially saturated ring, as well as checking the effect of halo-substitution on the

aromatic ring (Table 1). In general, the synthesis of compounds of this type involved the construction of
the diazepine ring as the final step of the route via the addition of an amino ethyl moiety to the desired
bicyclic system (indoline, tetrahydroquinoline, tetrahydro-1H-benzazepine etc.) followed by ring closure
with a Pictet-Spengler type cyclization with formaldehyde. The general synthetic routes to such
compounds are shown in Schemes 1-3.
Scheme 2: General synthesis of the 6,6,7-ring series
Cl
Cl
Cl
R
O
O
O
N
(v)
(vi)
(i)
OH
N
H
N
H
O
NH2
18
16
17
15
NH2
(iv)
R
R
R
R
R
X
N
12a
12b
13
14
(ii)
(iii)
, X = CH2, R = H
, X = CH2, R = Cl
, X = O, R = Cl
, X = S, R = Cl
(i)
(iv)
N
N
N
N
10
9
11
8
N
H
O
HN
NH2.HBr
X=S; (ix)
NH2
Cl
O
O
Cl
O
S
S
(iv)
N
+
N
Cl
N
Cl
Cl
N
Cl
Cl
H
H
S
N
S
S
S
CO2Et
(i)
F
(viii)
(vi)
(vii)
25
24
N
H
N
H
O
NO2
NO2
20
21
23
19
22
NH2
Reagents and Conditions i) 2-Bromoethanamine hydrobromide, neat, ∆; ii) HCOOH, HCONH2, 120 °C; iii)
1.25M HCl in MeOH, 75 °C; iv) formaldehyde, MeOH, TFA, 80 °C; v) ClCH2COCl, NaHCO3, MeCN, 0 °C; vi)
BH3.THF, 100 °C; vii) Ethyl 2-mercaptoacetate, Et3N, r.t.; viii) Fe (powder), AcOH, 90 °C; ix) mCPBA, DCM,
r.t.
Scheme 3: Synthesis of the 6,7,7-ring analogues

R
(i)
(ii)
(iii)
N
N
N
N
H
CN
N
29a
29b
R=H
R=Br
BocHN
H
26
27
28
(iii)
Br
Br
Br
Br
(vi)
(v)
(iv)
N
N
H
N
O
OH
33
30
31
32
H2N
Reagents and Conditions i) 2-hydroxy acetonitrile, water, ∆; ii) a) Raney Ni, 7M NH3 in MeOH, 70psi o/n
b) Boc2O, Et3N, DCM, r.t.; iii) formaldehyde, MeOH, TFA, 80 °C; iv) NH2OH.HCl, pyridine, r.t.; v) Dibal, -10
°C to r.t.; vi) 2-Bromoethanamine hydrobromide, neat, ∆
Our testing approach was to initially use the functional assay as a guide to potency and selectivity.
However, there have historically been significant differences in the selectivity as measured by binding or
functional assays, so as we were attempting to improve on the overall profiles of both lorcaserin and
vabicaserin (data shown in the Tables is from this study, but is comparable to the published data for both
lorcaserin7b and vabicaserin13),we were equally interested in the compounds’ receptor profile in binding
assays. Thus, for compounds with reasonable activity and selectivity in the functional assays, we then
measured displacement of 125I-labelled 1-(4-(125I)Iodanyl-2,5-dimethoxyphenyl)propan-2-amine (125I-
DOI) from all three 5-HT2 receptors.
Table 1
5-HT2CR
5-HT2BR
5-HT2AR
EC50, nM
(Emax, %)
125I-DOI
binding, Ki,
nM
EC50, nM
(Emax, %)
125I-DOI
binding, Ki,
nM
EC50, nM
(Emax, %)
125I-DOI
binding, Ki,
nM
7.6 (95)
n=12
66 (101)
n=7
15.2 (n=5)
970 (100)
n=12
>50000
n=6
220 (n=3)
1920 (28)
n=4
1240 (7)
n=6
98.5 (n=3)
1
3
Lorcaserin
10.6 (n=5)
>50000
13.8 (n=5)
>50000
149 (n=5)
>50000
Vabicaserin
12600
n=3
>50000
>50000
7a
N
HN

Cl
52 (88)
n=5
30 (n=7)
5930 (58)
637 (n=6)
>50000
1140 (n=6)
7b
N
HN
490 (87)
n=3
250 (n=2)
11 (n=4)
>50000
>50000
>50000
>50000
>50000
12a
12b
N
N
H
Cl
12 (85)
n=8
2110 (35)
226 (n=5)
389 (n=5)
N
N
H
Cl
360 (60)
n=2
192 (n=2)
34 (n=2)
n.d.
>50000
3730 (n=1)
442 (n=2)
n.d.
>50000
>50000
>50000
2430 (n=1)
180 (n=2)
n.d.
13
14
24
O
N
N
H
Cl
37 (76)
n=5
391 (6)
n = 4
S
N
N
H
Cl
O
S
2650 (77)
n=2
>50000
>50000
N
N
H
Cl
O
S
O
1320 (78)
n=2
905 (n=1)
>50000
>50000
>50000
25
N
N
H
1400 (79)
n=2
n.d.
500 (7)
>50000
n.d.
>50000
>50000
n.d.
29a
29b
N
HN
Br
260 (48)
n=5
51 (n=2)
360 (n=2)
59 (n=2)
N
HN
The data from the first, unsubstituted scaffolds - the direct analogues of vabicaserin and WAY-163909
with the fused cyclopentyl 4th ring removed - were not initially promising with 7a having no measurable
5-HT2 receptor activity at all, although 12a did appear to have a selective, albeit weak, interaction with
the 5-HT2CR (Table 1). However, a chloro-substitution in the appropriate position on the aromatic ring
resulted in an improvement in activity for both analogues, immediately providing compounds with
potency and selectivity comparable to that seen for lorcaserin in both functional and binding assays (7b,
12b). As a follow-up, we also prepared a number of other 6,6,7-fused heterocycles with a halo- substituent
in the position equivalent to the 9-position in compound 12, however neither the morpholino (13) nor
thiomorpholino (14) analogues provided any further improvement in activity over 12b, with 14 showing

markedly poorer selectivity, although the 5-HT2BR agonist efficacy was very low. Oxidation of the sulfur in
the ring to either the (racemic) sulfoxide (24) or the sulfone (25) further weakened 5-HT2CR activity.
Increasing the ring size to provide a 6,7,7 system did provide a compound with excellent agonist functional
selectivity, albeit with a bromo rather than chloro substituent, but in our binding assay 29b was not
selective for 5-HT2CR over 5-HT2AR. This lack of binding selectivity was also confirmed in triplicate using
tritiated 5-HT as the radioligand. Thus, even though the clear binding to the 5-HT2AR and 5-HT2BR did not
appear to translate into an agonist effect on the Gq pathway as measured with an IP3 assay in vitro, we
did not pursue this scaffold further as we were concerned that alternative second messenger pathways
that we were not capturing with the functional assay may still be activated by this compound - as could
be the case with vabicaserin. For example, several 5-HT2A agonists and partial agonists are known to have
hallucinogenic properties in humans and there is still no general agreement on how hallucinogenic and
non-hallucinogenic 5-HT2A agonists may differ in their ability to activate receptor second messenger
pathways to exert their effects.15 Computational approaches have proposed that distinct IC2 loop
conformations can be induced by different ligands suggesting that multiple second messenger pathways
may be activated, but this has not been confirmed experimentally.16 We therefore focused our attention
on the 6,5,7 and 6,6,7 series that each appeared to offer more promise with respect to selectivity in both
functional and, equally importantly, radioligand binding assays.
With significant experience in screening for compounds of interest in the 5-HT2CR agonist class, we chose
to position an in vivo pharmacodynamic measure (an acute re-feeding assay in the fasted rat) very early
on in the testing scheme. Thus, compounds of appropriate selectivity in both binding and function for the
5-HT2CR were assessed in this assay at a screening dose of 10mg/kg PO. Active compounds produced both
a significant decrease in food intake over 1-2 hours and induced other well-characterized 5-HT2CR
mediated behaviors in the rat, such as penile grooming and decreased motor activity, providing some
confidence that the effect was on target. A positive effect in this assay represented an early indicator that
the compound was orally bioavailable, had reasonable metabolic stability and was CNS penetrant,
allowing us to focus our efforts on the analogues of highest interest. Active compounds were then
progressed into both pharmacokinetic studies and a dose-response determination in the food intake
assay. Testing of 7b and 12b in this manner showed 7b to be the more active of the pair in vivo at the
10mg/kg PO screening dose. Following up with short time frame brain/plasma PK studies for each of these
compounds offered a clear explanation for this difference. We observed that the whole brain exposure
for 7b was around 3-fold higher at all 3 time points measured (0.5, 1 and 2 h) than for 12b and perhaps as
significant, the CSF concentration (a potential surrogate for the free brain concentration) of 7b was also

5-10 fold higher than 12b at each timepoint. As a result of this clear difference in CNS partitioning, we
chose to focus the bulk of our further exploratory efforts on the 6,5,7 series. This also had the advantage
that more substituted indoles were commercially available than substituted quinolines and the reduction
of the indoles to the indolines to provide the starting material for the cyclisation was somewhat more
consistent in yield than the quinoline to tetrahydroquinoline reduction. Interestingly, a 6,6,7 tricyclic
series similar to the one we decided not to pursue, has been extensively studied by the group from
Abbvie,17 but no detailed SAR has been published.
Table 2
8
R
5-HT2CR
5-HT2BR
5-HT2AR
9
10
N
HN
EC50, nM
Ki,* nM (n)
EC50, nM
Ki,* nM (n)
EC50, nM
Ki,* nM (n)
R
(Emax, %, n)
(Emax, %, n)
(Emax, %, n)
52 (88, 5)
30 (7)
n.d.
5930 (58)
637 (6)
n.d.
>50000
1140 (6)
n.d.
7b
7c
7d
7e
7f
7g
7h
7i
8-Cl
9-Cl
8-F
8-Br
8-CF3
>50000
>50000
>50000
>50000
>50000
>50000
>50000
>50000
>50000
>50000
>50000
>50000
>50000
>50000
2500 (95, 4)
84 (77, 6)
n.d.
n.d.
n.d.
7.1 (3)
6 (7)
140 (3)
n.d.
2150 (40, 9)
1680 (57, 11)
19000 (72, 5)
>50000
127 (2)
105 (7)
3500 (3)
n.d.
300 (3)
168 (7)
>10000 (4)
n.d.
35 (88, 11)
6300 (90, 2)
250 (85, 4)
190 (90, 4)
450 (105, 2)
780 (88, 3)
72 (87, 2)
8-OMe
8-Me
8-Et
n.d.
1750 (70, 4)
1220 (80, 2)
320 (93, 3)
260 (105, 2)
>50000
n.d.
n.d.
7j
7k
7l
8-nPr
n.d.
n.d.
n.d.
8-iBu
n.d.
n.d.
n.d.
8-cPr
n.d.
n.d.
n.d.
295 (86, 2)
16 (3)
925 (3)
68 (3)
7m
8-CH2Ph
* = 125I-DOI ligand
Exploring first the effect of substitution on the aromatic ring (Table 2), we showed that moving the chloro-
substituent from the 8-position to the 9-position (7c) abolished all 5-HT2 receptor activity. Surveying the
effect of alternative substitutions at the 8-position provided several other active substituents, such as
halo-, trifluoromethyl- and small alkyl- groups that maintained 5-HT2CR functional activity comparable to
that for lorcaserin or 7b. All the compounds had excellent functional selectivity for the 5-HT2CR over 5-
HT2AR, however, as was seen previously several compounds had significantly lower selectivity in the 125I-
DOI binding assay. Selectivity over the 5-HT2BR was more variable, but the 8-Me analogue (7h) while only
a moderately potent agonist of 5-HT2CR showed some promise in this regard whereas increasing the size
of this alkyl group (7i-l) reduced selectivity over 5-HT2BR in functional assays. Testing further compounds

in our in vivo screening assay confirmed that the in vitro functional effect remained in the appropriate
range to see a significant decrease in food intake at 10 mg/kg, with 7f being equipotent with 7b. In
addition, the abbreviated CNS assay showed that this compound could achieve high brain concentrations
(Table 4), albeit around half of those seen for 7b. With these data in hand we turned our attention to
substitutions on the 5-membered ring.
Table 3
8
7
5-HT2CR
5-HT2BR
5-HT2AR
9
6
10
N
4
1
HN
3
EC50, nM
Ki,* nM (n)
EC50, nM
Ki,* nM (n)
EC50, nM
Ki,* nM (n)
Substitution
(% Emax,, n)
(Emax, %, n)
(Emax, %, n)
390 (88, 3)
19000 (80, 2)
370 (89, 2)
700 (91,2)
94 (3)
n.d.
>50000
>50000
4350 (2)
n.d.
>50000
>50000
>50000
>50000
3700 (2)
n.d.
7n
7o
7p
7q
7-Me
6-Me
cis-6-Me, 7-Me
trans-6-Me, 7-Me
n.d.
6000 (21, 2)
>50000
n.d.
n.d.
n.d.
n.d.
n.d.
44 (95, 3)
175 (100, 53)
780 (92, 2)
17 (100, 21)
7850 (100, 3)
0.8 (90, 3)
2.5 (91, 3)
1.4 (95, 6)
2 (98, 7)
19 (3)
52 (6)
n.d.
3400 (63, 4)
>50000
395 (3)
1800 (6)
n.d.
4000 (22, 4)
>50000
290 (3)
2000 (8)
n.d.
7r
7s
7t
7u
7v
7w
7x
7y
7-CH2CF3
7,7-Me2
7,7-Et2
>50000
>50000
7,7-(spiro)cBu
7,7-(spiro)-4-THP
7-Me, 8-Me
7-Me, 8-Cl
14 (14)
n.d.
6420 (73, 16)
>50000
550 (12)
n.d.
4500 (40, 17)
>50000
265 (12)
n.d.
2.4 (3)
0.8 (3)
0.6 (6)
0.8 (3)
n.d.
175 (80, 3)
325 (76, 4)
160 (75, 5)
125 (96, 7)
>50000
22 (3)
46 (3)
21 (6)
18 (2)
n.d.
693 (21, 3)
1220 (10, 4)
750 (19, 4)
250 (47, 6)
>50000
44 (3)
44 (3)
28 (6)
13 (2)
n.d.
7-Me, 8-Br
7-CH2CF3, 8-Br
7-Me, 9-Cl
7z
12000 (80, 2)
30 (88, 3)
30 (93, 3)
>50000
7aa
7ab
7ac
7ad
7ae
7af
7ag
7ah
7ai
7aj
2.8 (3)
8 (3)
800 (37, 3)
300 (22, 4)
>50000
168 (3)
150 (3)
n.d.
1120 (38, 3)
1380 (38, 4)
>50000
58 (3)
96 (3)
n.d.
7-Me, 8,9-Cl2
7-Me, 8-Cl, 9-Me
7,7-Me2, 1-Me
7,7-Me2, 3-Me
7,7-Me2, 4-Me
7,7-Me2, 8-Br
7,7-Me2, 8-Cl
7,7-Me2, 8-F
7,7-Me2, 8-Me
n.d.
>50000
n.d.
>50000
n.d.
>50000
n.d.
400 (100, 2)
5 (88, 6)
n.d.
>50000
n.d.
>50000
n.d.
0.9 (3)
1.4 (4)
11 (6)
4.2 (4)
200 (103, 3)
320 (86, 6)
10600 (70, 7)
215 (93, 3)
28 (3)
18 (3)
320 (6)
24 (4)
>50000
36 (3)
60 (4)
350 (6)
70 (4)
6.5 (88, 6)
51 (90, 7)
8 (100, 2)
1340 (13, 7)
>50000
649 (17, 3)
* = 125I-DOI ligand

Substitution with a methyl group at the 7-position of the scaffold was tolerated with 7n showing promising
5-HT2CR activity and selectivity, again demonstrating that substitutions on the bare scaffold could alter
potency compared to the parent 7a (Table 3), even on the non-aromatic ring. Substitution in the 6-
position, however, gave 7o which was less potent. Combining these two substitutions to provide either
cis- or trans-6,7-dimethyl analogues 7p and 7q provided compounds with potency comparable to 7n, but
all these analogues were less potent than the cyclized analogue vabicaserin (2). Adding a second
substituent to the 7-position (7s) or elongating that substituent to a trifluoroethyl group (7r) gave
compounds with a modest improvement in activity at the 5-HT2CR. In addition, 7s showed promise in
terms of selectivity. There was no measurable functional effect at either the 5-HT2AR or 5-HT2BR and it had
a 5-HT2CR Ki of 52 nM and a >30-fold selectivity in binding over both 5-HT2AR and 5-HT2BR, making it
therefore a compound of interest for further investigation. This compound showed good efficacy in the
food intake assay at 10 mg/kg which triggered a dose-response. Reductions in food intake were observed
as low as 2.5 mg/kg, with an ED50 of 6.3 mg/kg at 60 min post-dose (Table 4). As a result of these favorable
properties, we prepared a handful of other 7,7-disubstituted compounds. The diethyl analogue 7t was
less potent than 7s but reducing the volume of the substituents by the use of a spirocyclic group (7u)
resulted in a compound that was an approximately 10-fold more potent agonist of the 5-HT2CR than 7s
with comparable selectivity in both functional and binding assays. Expanding the size of this spiro-ring did
not further improve potency (e.g. 7v). When tested in vivo, 7u showed excellent efficacy and as a result
of its improved receptor potency provided full inhibition of food intake at 5 mg/kg and maintained an
effect even at 2 mg/kg PO.
Table 4: In vivo pharmacodynamic and pharmacokinetic data
Inhibition of Food Intake
at 1h (%)
Abbreviated CNS PK study (10mg/kg PO)
Tissue concentrations at 1h
Compound
10 mg/kg
PO screen
Additional doses
(mg/kg PO)
Plasma
(ng/mL)
Brain
(ng/g)
CSF
(ng/mL)
97
48
94
87
99
89
98
35
-
332 ± 105
-
3940 ± 55
-
59.7 ± 1.2
-
9.31†
7b
7e
7f
7s
7u
7ag
7ai
12b
-
28 (3)
78.2 ± 43.6
679 ± 178
634 ± 122
-
1710 ± 717
5330 ± 597
7380 ± 1850
-
45 (5); 20 (2.5)
423 ± 103
229 ± 45
-
97 (5); 48 (2); 30 (1)
-
-
-
608 ± 55
12.7 ± 3.3
8860 ± 2020
957 ± 449
170 ± 59
9.06†
Mean data (± SEM) for n=6 (food intake) or n=3 (PK study) animals per timepoint except † : n=2

To complete this portion of the SAR study we prepared a series of compounds that incorporated
substituents on both the indoline ring as well as either the aromatic ring or the 7-membered ring. In our
hands, methyl substitution on any of the 3 available carbon atoms in the 7-membered ring in combination
with a 7,7-dimethyl substitution on the indoline ring resulted in a loss of activity compared to 7s. In
contrast, analogues with either 1 or 2 methyl groups on the indoline and a methyl or halogen substituent
in the 8-position of the aromatic ring provided some of the most potent 5-HT2CR agonists we had seen in
this series with some examples having either sub-nanomolar EC50s in the functional assay or Ki values of
less than 1 nM in binding (Table 3). Interestingly, when we separated the enantiomers of 7w by chiral
HPLC, we observed that all the 5-HT2CR activity resided in one isomer. However, while these compounds
showed good potency, unlike 7s and 7u their selectivity was not superior to Lorcaserin when taking both
binding and function into account. 7ai, which had the best selectivity profile amongst the group was, as
expected, a potent inhibitor of food intake in vivo with good CNS exposure (Table 4).
Having focused heavily on receptor selectivity from the start of this project, we further profiled
compounds 7s and 7u based on their excellent selectivity profiles even though they were not the most
potent agonists at the 5-HT2CR. A broader selectivity panel of 20 monoamine related receptors,
transporters and ion channels showed that in addition to the 5-HT2 receptor subfamily, both compounds
showed some moderate binding to the human 5-HT7R (~85% at 10 M) and 7u had some additional weak
interactions with the 5-HT1AR and 5-HT5AR as well as dopamine D3 (all 55-66% inhibition of binding at 10
M) but they were otherwise highly selective. Neither compound showed any significant inhibition of
either the hERG channel (patch clamp; 7s < 12% inhibition at 10 M; 7u < 30% inhibition at 3 M) or the
major CYP enzymes (IC50 >50 M for CYP3A4, 2C8, 2C9, 2C19, 1A2, 2D6 for both compounds).
High metabolic stability for the two compounds was demonstrated in vitro (T > 45 mins for both
½
compounds in human, rat, dog, cynomolgus monkey and mouse microsomes). As expected, based on the
potent inhibition of food intake observed in the rat, both 7s and 7u had excellent exposure in the brain
after oral administration of a 10 mg/kg dose (Table 4) in an abbreviated CNS/plasma PK study. Each
compound also had high CSF concentrations, suggestive of a significant free fraction in the brain. We
measured plasma protein binding for 7u and as might be predicted for small, non-lipophilic basic
compounds, these measurements confirmed the likelihood of high unbound tissue fractions (PPB: human
79.1%, rat 63.6%, mouse 51%). A full PK evaluation for each compound showed similar parameters for
half-life (7s = 1.88h, 7u =1.4h) and oral bioavailability (~100%) whereas clearance of 7u was twice that of

7s and greater than hepatic blood flow in the rat (7s 2.28 L/hr/kg; 7u 5.09 L/hr/kg). Notably the renal
contribution to clearance was close to 10% of the total clearance for both compounds.
Based on its lower clearance in the rat and lower volume of distribution we selected 7s to be tested in PK
studies in higher species. However, the oral bioavailability was greatly reduced in dog (12%) and
cynomolgus monkey (10%) and the half-lives were not longer than in the rat (T dog = 2.3h, monkey
½
1.85h). We hypothesized that the high renal clearance observed in these species, likely accompanied by
poorer than expected adsorption, were detrimental to the overall PK profile. As a result, we decided not
to pursue this compound further but rather to prepare other analogues to further optimize the series.
In summary, we have shown that from a range of newly designed tricyclic diazepine compounds, the 6,5,7-
series described herein showed excellent overall profiles in functional and binding assays for the 5-HT2
receptors. Compounds from this series consistently showed good PK properties in the rat, including high
plasma exposure after oral administration, good CNS partitioning and low protein binding. This core
therefore became our main series in our subsequent lead optimization, with the focus on further
improving selectivity and identifying a compound with an adequate PK profile in higher species -
particularly with respect to renal clearance. The results of these studies, including the selection of a clinical
candidate molecule, will be described in due course.
References
1. Halford, J. C.G.; Harrold, J. A.; Lawton C. L.; Blundell J. E. Serotonin (5-HT) Drugs: Effects on appetite
expression and use for the treatment of obesity. Curr. Drug Targets, 2005, 6, 201-213.
2. Miller, K.J. Serotonin 5-HT2C receptor agonists: Potential for the treatment of obesity. Mol. Int., 2005,
5, 282-291.
3. Vickers, S.P.; Clifton, P.G.; Dourish, C.T.; Tecott, L.H. Reduced satiating effect of d-fenfluramine in
serotonin 5-HT(2C) receptor mutant mice. Psychopharmacology, 1999, 143, 309–314.
4. Connolly, H. M.; Crary, J. L.; McGoon, M. D.; Hensrud, D. D.; Edwards, B. S.; Edwards, W. D.; Schaff, H.
V. Valvular heart disease associated with fenfluramine–phentermine. New Engl. J. Med. 1997, 337,
581-588.

5. a) Rothman, R. B.; Baumann, M. H.; Savage, J. E.; Rauser, L.; McBride, A.; Hufeisen, S. J.; Roth, B. L.
Evidence for possible involvement of 5-HT2B receptors in the cardiac valvulopathy associated with
fenfluramine and other serotonergic medications. Circulation, 2000, 102, 2836-2841. b) Huang, X.;
Setola, V.; Yadav, P. N.; Allen, J. A.; Rogan, S. C.; Hanson, B. J.; Revankar, C.; Robers, M.; Doucette, C.;
Roth, B. L. Parallel functional activity profiling reveals valvulopathogens are potent 5-
Hydroxytryptamine2B receptor agonists: Implications for drug safety assessment. Mol. Pharm. 2009,
76, 710-722.
6. a) Roth, B. L.; Willins, D. L.; Kristiansen, K.; Kroeze, W. K. Activation is hallucinogenic and antagonism is
therapeutic: Role of 5-HT2A receptors in atypical antipsychotic drug actions. Neuroscientist 1999, 5,
254-262. b) Nichols, D.E. Psychedelics. Pharmacol. Rev. 2016, 68, 264–355.
7. a) 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. Discovery and structure -activity relationship of (1R)-8-Chloro-2,3,4,5-
tetrahydro-1-methyl-1H-3-benzazepine (Lorcaserin), a selective serotonin 5-HT2C receptor agonist for
the treatment of obesity. J. Med. Chem., 2008, 51, 305-313. b) 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. P. Lorcaserin, a novel selective human 5-
hydroxytryptamine2C agonist: in vitro and in vivo pharmacological characterization. J. Pharmacol.
Exp. Ther. 2008, 325, 577-587. c) Smith, S. R.; Weissman, N. J.; Anderson, C. M.; Sanchez, M.; Chuang,
E.; Stubbe, S.; Bays, H.; Shanahan, W.R. Multicenter, placebo-controlled trial of lorcaserin for weight
management. N. Engl. J. Med. 2010, 363, 245-256.
8) Shen, J.H.Q.; Zhao,Y.; Rosenzweig-Lipson, S.; Popp, D.; Williams, J.B.W.; Giller, E.; Detke, M.J.; Kane,
J.M. A 6-week randomized, double-blind, placebo-controlled, comparator referenced trial of
vabicaserin in acute schizophrenia. Journal of Psychiatric Research, 2014, 53, 14-22.
9) Schoonjans A.S.; Lagae, L.; Ceulemans, B. Low-dose fenfluramine in the treatment of neurologic
disorders: experience in Dravet syndrome. Ther Adv Neurol Disord. 2015, 8, 328–338.
10) Griffin, A.; Hamling, K.R.; Knupp, K.; Hong, S.; Lee, L.P.; Baraban, S.C. Clemizole and modulators of
serotonin signalling suppress seizures in Dravet syndrome. Brain, 2017, 140, 669-683.

11) Mbaki Y.; Gardiner, J.; McMurray, G.; Ramage, A.G.; 5-HT2A receptor activation of the external
urethral sphincter and 5-HT2C receptor inhibition of micturition: A study based on pharmacokinetics
in the anaesthetized female rat Eur. J. Pharmacol. 2012, 682, 142–152 b) Ouchi, M.; Kitta, T.; Kanno,
Y.; Higuchi, M.; Togo, M.; Moriya, K.; Shinohara, N. Effect of a 5-HT2c receptor agonist on urethral
closure mechanisms in female rats Neurourol Urodyn. 2018, 37, 2382-2388. c) Ishigami, T.; Ueshima
K.; Ukai, M.; Asai, N.; Takamatsu, H.; Yokono, M.; Takeda, M.; Masuda, N. Effect of ASP2205
fumarate, a novel 5-HT2C receptor agonist, on urethral closure function in rats. J Pharmacol Sci. 2019
139, 333-339.
12) Dunlop, J.; Sabb, A.L.; Mazandarani, H.; Zhang, J.; Kalgaonker, S.; Shukhina, E.; Sukoff, S.; Vogel, R.L.;
Stack, G.; Schechter, L.; Harrison, B.L.; Rosenzweig-Lipson, S. WAY-163909 [(7bR, 10aR)-
1,2,3,4,8,9,10,10a-Octahydro-7bH-cyclopenta-[b][1,4]diazepino[6,7,1hi]indole], a Novel 5-
Hydroxytryptamine 2C Receptor-Selective Agonist with Anorectic Activity J. Pharmacol. Exp. Ther.
2008, 313, 862-869
13) Dunlop, J.; Watts, S.W.; Barrett, J.E.; Coupet, J.; Harrison, B.; Mazandarani, H.; Nawoschik, S.;
Pangalos, M.N.; Ramamoorthy, S.; Schechter, L.; Smith, D.; Stack, G.; Zhang, J.; Zhang, G.;
Rosenzweig-Lipson, S. Characterization of vabicaserin (SCA-136), a selective 5-hydroxytryptamine 2C
receptor agonist. J Pharmacol Exp Ther. 2011, 337,673-80.
14) Liu, W., Wacker, D., Gati, C., Han, G.W., James, D., Wang, D., Nelson, G., Weierstall, U., Katritch, V.,
Barty, A., Zatsepin, N.A., Li, D., Messerschmidt, M., Boutet, S., Williams, G.J., Koglin, J.E., Seibert,
M.M., Wang, C., Shah, S.T.A., Basu, S., Fromme, R., Kupitz, C., Rendek, K.N., Grotjohann, I., Fromme,
P., Kirian, R.A., Beyerlein, K.R., White, T.A., Chapman, H.N., Caffrey, M., Spence, J.C.H., Stevens, R.C.,
Cherezov, V. Crystal structure of the 5-HT2B receptor solved using serial femtosecond crystallography
in lipidic cubic phase. Science 2013, 342 1521-1524
15) Halberstadt, A.L. Recent advances in the neuropsychopharmacology of serotonergic hallucinogens.
Behav. Brain Res., 2015, 277, 99–120
16) Perez-Aguilar, J.M.; Shan, J.; LeVine, M.V.; Khelashvili, G.; Weinstein, H. A functional selectivity
mechanism at the serotonin-2A GPCR involves ligand-dependent conformations of intracellular loop
2. J. Am. Chem. Soc., 2014, 136, 16044−16054.
17) a) Koolman, H.; Braeje, W.; Mack, H.; Haupt, A.; Relo, A. L.; Drescher, K.; Bakker, M. H. M.; Lakics, V.;
Hoft, C.; Xu, R.; Zhao, X. PCT Patent Application WO2014041131. b) Backfisch, G.; Bakker, M. H. M.;

Blaich, G.; Braeje, W.; Drescher, K.; Erhard, T.; Haupt, A.; Lakics, V.; Reingruber, A.; Mack, H.; Peter,
R.; Relo, A. L.; Smyth, L.; PCT Patent Application WO2015136091. c) Backfisch, G.; Bakker, M. H. M.;
Blaich, G.; Braeje, W.; Drescher, K.; Erhard, T.; Haupt, A.; Koolman, H.; Lakics, V.; Reingruber, A.;
Mack, H.; Peter, R.; Relo, A. L.; Wang, J.; Zhao, X. PCT Patent Application WO2015136090
18) Dragan, V.; McWilliams, J.C.; Miller, R.; Sutherland, K.; Dillon, J.L; O’Brien, M.K. Asymmetric Synthesis
of Vabicaserin via Oxidative Multicomponent Annulation and Asymmetric Hydrogenation of a 3,4-
Substituted Quinolinium Salt. Org. Letts., 2013, 15, 2942–2945.
R
R'
N
N
N
N
N
N
H
H
H
Vabicaserin
7a-7aj
7s
5-HT2CR; Ki = 52 nM
selectivity >30x vs 5-HT2AR
and 5-HT2BR in both binding and
functional assays
Declaration of interests
☒ The authors declare that they have no known competing financial interests or personal relationships
that could have appeared to influence the work reported in this paper.
☐The authors declare the following financial interests/personal relationships which may be considered
as potential competing interests: