Discovery of a potent, selective, and less flexible selective norepinephrine reuptake inhibitor (sNRI)

Article abstract of DOI:10.1016/j.bmcl.2008.05.057

The design, synthesis, and SAR of a series of ring-constrained norepinephrine reuptake inhibitors are described. A substantially rigid inhibitor with potent functional activity at the transporter (IC₅₀ = 8 nM) was used to develop a model for th

Full text of DOI:10.1016/j.bmcl.2008.05.057

Bioorganic & Medicinal Chemistry Letters 18 (2008) 4224–4227
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Discovery of a potent, selective, and less flexible selective norepinephrine
reuptake inhibitor (sNRI)
b
Dongpei Wu a, Joseph Pontillo a, Brett Ching a, Sarah Hudson a, , Yinghong Gao ,
*
Beth A. Fleck c, Kathleen Gogas d, Warren S. Wade a,
*
a Department of Medicinal Chemistry, Neurocrine Biosciences Inc. 12790 El Camino Real, San Diego, CA 92130, USA
b Department of Computational Chemistry, Neurocrine Biosciences Inc. 12790 El Camino Real, San Diego, CA 92130, USA
c Department of Pharmacology, Neurocrine Biosciences Inc. 12790 El Camino Real, San Diego, CA 92130, USA
d Department of Neuroscience, Neurocrine Biosciences Inc. 12790 El Camino Real, San Diego, CA 92130, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
The design, synthesis, and SAR of a series of ring-constrained norepinephrine reuptake inhibitors are
described. A substantially rigid inhibitor with potent functional activity at the transporter (IC50 = 8 nM)
was used to develop a model for the distance and orientation of key features necessary for interaction
with the norepinephrine transporter (NET).
Received 2 April 2008
Revised 12 May 2008
Accepted 15 May 2008
Available online 20 May 2008
Ó 2008 Elsevier Ltd. All rights reserved.
Keywords:
Norepinephrine
Serotonin
Monoamine reuptake inhibition has been an effective therapeu-
tic intervention in a variety of CNS diseases starting with depres-
sion and recently expanding to include chronic pain, ADHD, and
stress urinary incontinence.1–4 Compounds with varied profiles
have been described as active in animals and humans. The range
of active compounds contains selective norepinephrine reuptake
inhibitors (sNRIs) like atomoxetine 1 and nisoxetine 2, selective
serotonin (5-HT) reuptake inhibitors like fluoxetine 3, and mixed
inhibitors like duloxetine 4.5,6
Interest in sNRIs as marketable drugs is relatively recent, with
atomoxetine approved for ADHD in 2003 and (S,S)-reboxetine 5 re-
ported to be effective in a phase II trial for neuropathic pain.7 We
were interested in multiple indications in the therapeutic spec-
trum of sNRI compounds and initiated work directed towards find-
ing potent and selective compounds with good pharmaceutical
properties and minimal risk of drug–drug interactions.
tended to produce compounds with a range of selectivities and
study their properties in vivo.
Structural requirements were derived from the literature,8 illus-
trated for compounds 1–4, the Lilly series, in Figure 1. Three struc-
tural features were readily apparent: a basic site A, typically a
secondary amine for potent NET inhibition, an aromatic ring where
substitution leads to substantial changes in NET, SERT, and DAT
selectivities, and an additional hydrophobe C that was aromatic
for these compounds.10 The Lilly series is maximally efficient,11
with only four atoms separating the base features, but suffers from
metabolism and clearance issues,12,13 probably related to the over-
all flexibility of the molecules. Our intention was to maintain the
R
O
B
Initial criteria for target compounds were based on the proper-
ties of known reuptake inhibitors.8 Compounds needed to be po-
tent functional inhibitors of NET (<100 nM) and much less active
O
O
O
N
S
NH
N
A
C
H
H
O
as DAT (dopamine transporter) inhibitors (>1 lM) to best avoid po-
tential dependence issues.9 Since the relative contribution of SERT
(serotonin transporter) inhibition to efficacy was not clear, we in-
1 2-Me, atomoxetine
4 duloxetine 5 S,S-reboxetine
2 2-OMe, nisoxetine (rac)
3 4-CF3, fluoxetine (rac)
* Corresponding authors. Tel.: +1 858 617 7600; fax: +1 858 617 7601 (S.H.).
E-mail addresses: shudson@neurocrine.com (S. Hudson), warrenswade@gmail.-
com (W.S. Wade).
Figure 1. Structure of the Lilly series with the key features labeled. (A) Amine
features typically a secondary amine for NET activity; (B) aromatic ring, controls
NET/SERT selectivity; (C) hydrophobic site, often aromatic.
URL: http://www.neurocrine.com (S. Hudson).
0960-894X/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2008.05.057

D. Wu et al. / Bioorg. Med. Chem. Lett. 18 (2008) 4224–4227
4225
overall connectivity of the Lilly series while introducing ring
O
constraints, aiming to improve the overall properties while main-
taining potent transporter inhibition. The ring connection from C
to the methylene next to A was successful in producing potent
NET inhibitors.
a
b
NHCOCF3
20
NH2
NHCOCF3
Synthesis of the 1-amino-3-oxyarylindane core was straightfor-
ward from readily available 3-amino-3-phenylacetic acid and was
initially done with racemic starting material. Following the litera-
ture procedure for the amino alcohols,14 protection of the amino
group, best as the trifluoroacetamide, followed by Friedel–Crafts
cyclization gave the aminoindanone. Reduction with boraneÁTHF
produced predominantly the cis amino alcohol. Alternatively,
reduction with sodium borohydride was less selective but pro-
ceeded in higher overall yields, and allowed for isolation of the
minor trans amino alcohol 9t.15 The route used to generate the pro-
tected aryloxy intermediate was determined by the electronic
properties of the aryl ring. SNAr chemistry was employed for aryl
groups with electron-withdrawing substituents like 4-chloro. The
process was efficient and proceeded predominantly with retention
of stereochemistry starting from either the pure cis or trans amino
alcohols. Yields were lower for ortho-substituted aryl fluorides,
especially more electron-rich aromatic rings, such as compounds
13d and 13e. Alternatively, the amino alcohols could be arylated
with a phenol under Mitsunobu conditions.16 This route proceeded
predominantly with inversion of stereochemistry. Diastereomer
assignment was verified by NOESY and coupling constant correla-
tions for the ring hydrogens.
Substitution at the amine was achieved by basic hydrolysis to
give the primary amine followed by reductive amination to give
the tertiary amine. The secondary amine was generated by alkyl-
ation of the TFA-protected intermediate followed by base depro-
tection under forcing conditions. Final compounds were purified
by preparative HPLC with mass selective collection.
Similar chemistry starting from 3-amino-4-phenyl butyric acid
generated the 2,4-aminotetralin isomer. Synthesis of compounds
22–26 was performed with chiral starting materials and the com-
pounds were tested individually. The 1,4-aminooxyaryltetralins
were prepared from the protected intermediate 21 which in turn
had been synthesized from 1-aminotetralin (Scheme 2). Following
the procedure outlined in Scheme 1, we then produced compounds
27–29.
21
Scheme 2. Reagents and conditions: (a) ethyl trifluoroacetate, TEA, MeOH, RT, 91%;
(b) CrO3, Ac2O, 10 °C, 40 min, 78%.
pounds were initially tested in two independent dose–response
experiments, and those with reasonable potency at NET were re-
tested multiple times. Atomoxetine was included on all assay
plates as a standard control, and assay variability was reasonable
for a functional assay with SEM values typically below 0.2 log
units. A table of SEM values is included as Supplementary Material.
Potency values are reported as IC50, though with the neurotrans-
mitter concentrations in each assay well below their respective
Km values, little difference would be expected between the mea-
sured IC50 and Ki of the compounds. Active compounds were fur-
ther tested in transporter binding assays by competition with the
appropriate radioligand.17
SAR was generated to determine whether the general trends
previously observed in the Lilly series were present in these new
compounds (Table 1). Initial experiments with the cis indanes indi-
cated that the activity of amine substitution does not match that
observed for the Lilly series. In particular, the most active amines
were the dimethylamines 13a, 13b, and 13d, followed by the
monomethyl derivatives 12a, 12b, and 12d. The primary amine
compounds were barely active at any of the transporters. Substitu-
tions of the amine with larger groups like ethyl (compounds 14d,
15d) or isopropyl (16d) also led to a loss of activity.
In contrast, the expected trends for aryloxy substitution were
observed. Substitution at the 4 position with chloro, as in com-
pound 13b, led to a 40-fold increase in SERT inhibition without
much activity at NET. The 4-methoxy compound 13c was also
SERT-selective but less potent. Conversely, substitution with 2-
Me resulted in about a 5-fold improvement of NET inhibition, with
the best compound 13d measuring 150 nM. Interestingly, com-
pound 13e with 2-methoxy substitution was expected to be as po-
tent by comparison to nisoxetine, but was much less active at
All compounds were tested for their ability to inhibit norepi-
nephrine, serotonin and dopamine uptake in HEK cell lines which
had been stably transfected with the human transporters.17 Com-
3 lM. Both enantiomers of 13d were synthesized and the (R,S)
OH
OH
O
c or d
b
OH
+
CF3
CF3
O
HN
CF3
HN
9t
R1
HN
9c
HN
O
O
6 R1 = H
O
a
8
7 R1 = CF3CO
e
f
O
O
O
R
R
g or i, g
R
CF3
CF3
N
HN
HN
R2
R1
10t
O
10c
O
11 R1, R2 = H
12 R1 =H, R2 = Me
13 R1, R2 = Me
h
Scheme 1. Reagents and conditions: (a) ethyl trifluoroacetate, RT, 18 h, quantitative; (b) SOCl2ÁDMF, DCM, dioxane, then AlCl3, RT, 1 h, 92%; (c) BH3ÁTHF, À78 °C, 2 h, 9c 60%;
(d) NaBH4, MeOH, 0 °C, 9c 50%, 9t 25%; (e) aryl fluoride, NaH, DMSO, 90 °C, 24 h, 30–70%; (f) Hydroxyaryl, Bu3P, TMAD, DCM, RT, 18 h, 60–70%; (g) NaOH, EtOH, H2O, 50 °C,
1 h, 70%; (h) HCOH, NaBH(OAc)3, EtOH, RT, 50–80%; (i) CH3I, NaH, DMF, 90–100%.

4226
D. Wu et al. / Bioorg. Med. Chem. Lett. 18 (2008) 4224–4227
Table 1
SAR of the aryloxyindanesa
R
O
R2
N
R1
Compound
R1
R2
R
NETb
SERTb
DATb
Atomoxetine, 1
Nisoxetine, 2
Fluoxetine, 3
cis-11a
cis-12a
cis-13a
cis-11b
cis-12b
cis-13b
cis-13c
Me
Me
Me
H
Me
Me
H
Me
Me
Me
Me
Me
H
H
H
H
H
Me
H
2-Me
5
15
180
1400
47
>10,000
>10,000
2000
9000
320
50
130
1400
300
110
3000
2400
6000
2000
5000
>10,000
2400
4000
7000
>10,000
4600
10,000
8000
2-OMe
4-CF3
H
H
H
4-Cl
4-Cl
4-Cl
4-OMe
2-Me
2-Me
2000
5000
2000
800
4000
4000
1600
6000
660
150
60
Figure 2. Overlay of 1 (light gray), 5 (dark gray), and 22 (teal) emphasizing the
pseudoaxial placement of the aryloxy ring, the common overlap of the aryloxy-
propylamine scaffold and the different conformations of ring C.
H
Me
Me
H
cis-12d
cis-13d
13d-I
Me
reach the optimal position. To study this further, both tetralin
regioisomers were prepared. The (2,4) isomer contained the ato-
moxetine functional group spacing and was expected to be more
active. As shown in Table 2, all four diastereomers (22, 24–26)
were active at NET, with the best activity, almost equivalent to 1,
residing in the (2S,4R) isomer 22 (NBI 80532).18 In the NET binding
assay, 22 was able to compete nisoxetine binding to baseline with
a Ki of 4 nM (similar to its IC50 of 8 nM in the NET functional assay),
suggesting that its observed functional NET inhibition was due to
interaction with a binding site that overlapped with nisoxetine.
Amine substitution also reproduced the Lilly series, with a 50-fold
drop in potency for the tertiary amine 23. Activity of the 1,4 regioi-
somers 27–29 were ꢀ500 nM, less than that of the indanes.
Gratifyingly, the selectivity of 22 for NET versus SERT and DAT
was as good as 1, but was achieved with only 3 rotatable bonds.
The rigidity of 22 provided an opportunity to generate a model
for the overlay of 1, 5, and 22 in the NET binding site, as shown
in Figure 2.19 Because the tetralin core is almost flat, there is little
difference in the distance between the amine and the aryloxy fea-
tures among the four diastereomers, but the relative orientation
changes for each. Importantly, the most active isomer is one where
the phenoxy group is pseudoaxial. The position of the C ring in 1 is
similar to that of 1 and 3 in their crystal structures,20 but 22 cannot
reach the same conformation because of the ring constraint, con-
sistent with the presence of an adjustable hydrophobic pocket on
the transporter. This model was used to develop multiple new ser-
ies because it addresses the active conformation of the aryloxy
substituent and the relative positioning of the amine.
13d-II
cis-13e
cis-14d
cis-15d
350
3000
3000
2000
1100
840
72
600
3500
>10,000
2500
2500
>10,000
600
Me
Et
Et
Me
H
Me
Me
Me
H
Et
iPr
H
H
2-OMe
2-Me
2-Me
2-Me
2-Me
2-Me
2-Me
3500
8000
3000
>10,000
10,000
370
cis-16d
trans-17d
trans-18d
trans-19d
180
3400
Me
150
150
a
Data are the average of two or more independent measurements.
IC50 (nM) for monoamine uptake.
b
enantiomer 13d-I is the more potent at both NET and SERT. Com-
pared to the Lilly series, the selectivity of this series generally fa-
vored SERT. The trans diastereomer was slightly more potent at
NET, while the most potent compound was now the secondary
amine 18d. However, these compounds showed substantial inhibi-
tion of DAT in addition to minimal selectivity versus SERT (see Ta-
ble 2).
It appeared that the indane compounds could interact with the
aryl pocket on the transporter, but that the amine was not able to
Table 2
SAR of the aryloxytetralinsa
O
O
Compound 22 was evaluated further in vitro to determine its
potential as a drug candidate. Selectivity against an in-house panel
of receptors for amine neurotransmitters revealed no submicrom-
olar activity except at the 5HT2b receptor (250 nM). However, 22
was more than 5-fold more potent than 1 at inhibiting CYP2D6
activity (IC50 = 250 nM), and it was expected that this level of inhi-
bition could potentially lead to drug–drug interactions with other
CNS agents such as desipramine and atomoxetine itself. Accord-
ingly, it was necessary to look for alternative ring connections with
more favorable CYP interaction profiles, and the results of these
studies will be reported in future publications.
R2
N
R1
N
R2
R1
22-26
27-29
Compound
Isomer
R1
R2
NETb
SERTb
DATb
trans-22
trans-23
trans-24
cis-25
cis-26
cis-27
2S,4R
2S,4R
2R,4S
2R,4R
2S,4S
Me
Me
Me
Me
Me
H
H
Me
H
H
H
H
H
H
8
430
200
190
160
630
520
950
590
700
8000
>10,000
4500
>10,000
6000
1000
3300
3000
1100
2300
5000
>10,000
Acknowledgments
cis-28
trans-29
Me
H
1000
1000
The authors are indebted to Mr. Rajesh Huntley for technical
support in the transporter assays, Dr. Michael Mesleh for the 2D
NMR experiments, and Ms. Hua Wang for the CYP2D6 results.
a
Data are the average of two or more independent measurements.
IC50 (nM) for monoamine uptake.
b

D. Wu et al. / Bioorg. Med. Chem. Lett. 18 (2008) 4224–4227
4227
12. Sauer, J.-M.; Ring Barbara, J.; Witcher Jennifer, W. Clin. Pharmacokinet. 2005, 44,
571.
Supplementary data
13. Bymaster, F. P.; Lee, T. C.; Knadler, M. P.; Detke, M. J.; Iyengar, S. Curr. Pharm.
Des. 2005, 11, 1475.
14. Kinbara, K.; Katsumata, Y.; Saigo, K. Chem. Lett. 2002, 31, 266.
15. Compound 9c was crystalline and easily separated from the trans oil 9t by
crystallization.
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.bmcl.2008.05.057.
16. Tsunoda, T.; Otsuka, J.; Yamamiya, Y.; Ito, S. Chem. Lett. 1994, 23, 539.
17. Kiankarimi, M.; Hudson, S.; Dwight, W. J.; Wade, W. S. Preparation of benzofuranyl-
methylamines, chromanylmethylamines, and thiochromanylmethylamines as
monoamine reuptake inhibitors. US 2006252818.
References and notes
1. Jann, M. W.; Slade, J. H. Pharmacotherapy 2007, 27, 1571.
2. Babu, R. P. K.; Maiti, S. N. Heterocycles 2006, 69, 539.
3. Zhou, J. Drugs of the Future 2004, 29, 1235.
4. Wernicke, J. F.; Iyengar, S.; Ferrer-Garcia, M. D. Curr. Drug Ther. 2007, 2,
161.
18. 1H NMR (500 MHz, CHCl3-d) d ppm 9.84 (s, 1H), 9.67 (s, 1H), 7.30 (t, J = 7 Hz,
1H), 7.20 (m, 3H), 7.16 (d, J = 7 Hz, 1H), 7.15 (t, J = 7 Hz, 1H), 7.04 (d, J = 8 Hz,
1H), 6.94 (t, J = 7 Hz, 1H), 5.45 (s, 1H), 3.77 (s, 1H), 3.40 (dd, J = 15, 4 Hz, 1H),
3.05 (dd, J = 15, 12 Hz, 1H), 2.72 (s, 3H) 2.14 (t, J = 12 Hz, 1H), 2.04 (s, 3H); APCI
MS m/z 268.0.
5. Walter, M. W. Drug Dev. Res. 2005, 65, 97.
6. Iversen, L. Mol. Psychiatry 2000, 5, 357.
19. Molecule structures were imported into the Molecular Operating Environment
(MOE) software version 2007.09 and subjected to Stochastic Conformation
Search using the MMFF94x force field with GBSA implicit solvation. Low-
energy conformations within 1.5 kcal/mol from the minimum that were most
consistent with the known solid-state X-ray crystal structures were taken
forward and aligned using the Flexible Alignment module using default field
similarity terms except hydrogen-bonding terms were scaled down to 0.5. The
best scoring alignment most consistent with known structure–activity
patterns for NET:SERT selectivity was retained.
7. Mackay, M. Pfizer R&D Analyst Meeting 2006, Neurosciences slide 9, http://
www.pfizer.com/investors/presentations/presentations.jsp, web site accessed
5/2008.
8. Maxwell, R. A.; White, H. L. Handbook Psychopharmacol. 1978, 14, 83.
9. Wise, R. A. Curr. Opin. Neurobiol. 1996, 6, 243.
10. Boot, J.; Cases, M.; Clark, B. P.; Findlay, J.; Gallagher, P. T.; Hayhurst, L.; Man, T.;
Montalbetti, C.; Rathmell, R. E.; Rudyk, H.; Walter, M. W.; Whatton, M.; Wood,
V. Bioorg. Med. Chem. Lett. 2005, 15, 699.
11. Reynolds, C. H.; Tounge, B. A.; Bembenek, S. D. J. Med. Chem. 2008, 51,
2432.
20. Fuller, R. W.; Wong, D. T.; Robertson, D. W. Med. Res. Rev. 1991, 11, 17.