Studies on the SAR and pharmacophore of milnacipran derivatives as monoamine transporter inhibitors

Article abstract of DOI:10.1016/j.bmcl.2008.01.011

Derivatives of milnacipran were synthesized and studied as monoamine transporter inhibitors. Potent analogs were discovered at NET (9k) and at both NET and SERT (9s and 9u). A pharmacophore model was established based on the conformational analysis of mil

Full text of DOI:10.1016/j.bmcl.2008.01.011

Available online at www.sciencedirect.com
Bioorganic & Medicinal Chemistry Letters 18 (2008) 1346–1349
Studies on the SAR and pharmacophore of milnacipran
derivatives as monoamine transporter inhibitors
a,
a
b
c
a
*
Chen Chen, Brian Dyck, Beth A. Fleck, Alan C. Foster, Jonathan Grey,
a
a
a
a
Florence Jovic, Michael Mesleh, Kasey Phan, Junko Tamiya,
a
a
Troy Vickers and Mingzhu Zhang
a
Department of Medicinal Chemistry, Neurocrine Bioscience, Inc., 12790 El Camino Real, San Diego, CA 92130, USA
b
Department of Pharmacology, Neurocrine Bioscience, Inc., 12790 El Camino Real, San Diego, CA 92130, USA
Department of Neuroscience, Neurocrine Bioscience, Inc., 12790 El Camino Real, San Diego, CA 92130, USA
c
Received 3 December 2007; revised 27 December 2007; accepted 4 January 2008
Available online 9 January 2008
Abstract—Derivatives of milnacipran were synthesized and studied as monoamine transporter inhibitors. Potent analogs were dis-
covered at NET (9k) and at both NET and SERT (9s and 9u). A pharmacophore model was established based on the conforma-
tional analysis of milnacipran in aqueous solution using NMR techniques and was consistent with the SAR results.
Ó 2008 Elsevier Ltd. All rights reserved.
5
The synaptic actions of the monoamine neurotransmit-
ters norepinephrine (NE), serotonin (SER), and dopa-
mine (DA) are terminated by reuptake into the nerve
endings from which they are released and by uptake into
adjacent cells. Reuptake is achieved by cell membrane
transporters specific for each monoamine (NET, SERT,
and DAT) which have been successfully targeted for the
ropathy. Duloxetine is an NSRI, and two other drugs
6
usually considered in this class are venlafaxine (4) and
7
milnacipran (5, Fig. 1). Available in many countries
as an antidepressant, including Japan and France, mil-
8
nacipran inhibits NE and SER reuptake in a 3:1 ratio.
While the SERT inhibition is likely to improve depres-
9
sion, the NET reuptake blockade is thought to improve
10
1
development of CNS drugs. Both selective serotonin
chronic pain. Milnacipran is currently in phase III
clinical trials for fibromyalgia, and recent results sug-
reuptake inhibitors (sSRI) and inhibitors of both NET
and SERT (NSRI) are effective against major depression
and a range of other psychiatric illnesses. The dopamine
transporter (DAT) is a key target for amphetamine and
methylphenidate, used in the treatment of attention
1
1
gested efficacy in this indication.
Milnacipran, marketed as a racemic mixture of two
enantiomers, is a hydrophilic molecule that differs from
the other more hydrophobic monoamine transporter
inhibitors such as duloxetine. Milnacipran is mainly
excreted in the urine as the parent and glucoronide
(>80%), and only a small fraction (<10%) is metabolized
2
deficit hyperactivity disorder.
1
2
Ò
Fluoxetine (1, Prozac ) is an sSRI (marketed as a race-
mic mixture) and has been used as an antidepressant
3
13
since 1988. Atomoxetine (2) was introduced into the
4
via N-de-ethylation by the CYP3A4 enzyme. This lack
market as a selective NET inhibitor (sNRI) for ADHD.
Recently, duloxetine (3) has been approved by the FDA
for the treatment of depression as well as diabetic neu-
of potential for drug–drug interaction via the CYP450
enzymes is quite an attractive feature for a CNS drug be-
cause many are highly lipophilic and rely heavily on liver
enzymes for elimination. The SAR of milnacipran and
its analogs based on in vivo efficacy was reported by
Bonnaud and coworkers in 1987. Recently, Roggen
et al. reported a series of milnacipran analogs as single
stereoisomers with variation in the aromatic moiety
1
4
Keywords: Structure–activity relationship; Pharmacophore; Milnacip-
ran; Norepinephrine; Serotonin; Monoamine transporter; Inhibitor;
Synthesis.
1
5
*
Corresponding author. Tel.: +1 858 617 7600; fax: +1 858 617
602; e-mail: cchen@neurocrine.com
and their activity as NET and SERT inhibitors. How-
ever, the SAR and pharmacophore of milnacipran
7
0
960-894X/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2008.01.011

C. Chen et al. / Bioorg. Med. Chem. Lett. 18 (2008) 1346–1349
1347
F3C
O
N
O
O
O
O
NH2
OH
NH
NH
NH
N
S
Fluoxetine (1)
Atomoxetine (2)
Duloxetine (3)
Venlafaxine (4)
Milnacipran (5)
Figure 1. Chemical structures of some monoamine transporter inhibitors.
a
derivatives at the transporter level are still largely
unclear. Here we report the synthesis of milnacipran
analogs and their structure–activity relationships at
NET and SERT.
Table 1. SAR of the N-alkyl milnacipran derivatives 6 and 7
1
R
Compound
NET
SERT
DAT
5
H
77
420
6,100
>10,000
>10,000
6
6
6
a
b
c
Me
Et
6500
860
6100
5400
1300
n
N-Alkyl derivatives of milnacipran 6 were synthesized
by reductive alkylation of milnacipran with various
aldehydes in the presence of sodium triacetoxyborohy-
dride as shown in Scheme 1. N,N-Dimethyl milnacipran
Bu
Bu
>10,000 >10,000
i
6d
6e
6f
>10,000 >10,000 >10,000
CyclopentaneCH
2
8400
960
>10,000 >10,000
(MeO)
2
CHCH
2
1100
1600
4800
>10,000
>10,000
>10,000
1
6
6
6
6
6
6
6
6
7
g
h
i
2-OxazoleCH
5-OxazoleCH
2
2
450
2900
7
was obtained from the corresponding aldehyde by a
reductive amination with dimethylamine in the presence
of sodium triacetoxyborohydride.
2-FuranCH
2-PyrroleCH
2
>10,000 >10,000 >10,000
j
2
2600
8100
4400
3500
>10,000
6200
k
l
2-PyridineCH
3-PyridineCH
4-PyridineCH
2
2
2
Amide analogs of milnacipran were prepared from the
>10,000 >10,000 >10,000
>10,000 >10,000 >10,000
>10,000 >10,000 4500
1
4
known acid intermediate 8 as depicted in Scheme 2.
Selected primary amines 9 were also converted to the
aminoacetamides 10 using a coupling reaction with N-
Boc-glycine, followed by a deprotection with trifluoro-
acetic acid.
m
a
Data are average of two or more independent measurements.
IC50 = 6500 nM) and SERT activity by about 10-fold.
All N-alkyl derivatives (6b–e) were weakly active at
NET and SERT and devoid of activity at DAT (6k
was very weakly active). Among them, N-(2-oxazolem-
ethyl) milnacipran (6g, NET IC50 = 450 nM) displayed
the best potency at NET but was still 6-fold less potent
than the parent. Therefore, it could be concluded that
N-alkylation of milnacipran, including the N,N-di-
methyl analog 7, detrimentally decreased its interaction
with all three transporters (Table 1).
The target compounds 6–7 and 9–10 as racemic mixtures
were tested for inhibition of human NET, SERT, and
DAT using a procedure similar to that described by
1
7
Owens et al. These results are summarized in Tables
–3.
1
Milnacipran (5) as a pair of enantiomers exhibited mod-
erate potencies at NET (IC50 = 77 nM) and SERT
(
IC50 = 420 nM) and was weakly active at DAT
(IC50 = 6100 nM). N-Methylation of milnacipran re-
duced its NET activity by almost 100-fold (6a, NET
Among the secondary amides 9a–d, the N-phenyl com-
pound 9c showed some activity at NET, and the N-ben-
zyl 9d showed activity at SERT (Table 2), suggesting a
probable role of p-electrons for the interaction between
the transporters and ligands. N-Methyl-N-ethyl amide
9e was much less potent at NET than milnacipran 5.
In comparison, the N-ethyl-N-propylamide 9f possessed
slightly higher NET activity but lower SERT potency
than 5, while N-ethyl-N-butyl analog 9g exhibited high
IC50 values at both NET and SERT, indicating milna-
cipran was the most balanced dual inhibitor among
these dialkyl amides. Both 9h and 9i displayed poor
H
N
a
O
NH2
O
O
N
R1
N
N
N
5
(+/-)
6 (+/-)
7 (+/-)
Scheme 1. Reagents and conditions: (a) aldehyde/THF/Et
NaBH(OAc) /EtOAc/rt.
3
N/rt; then
3
NH2
O
O
N
a
b
O
O
NH2
O
NH
R2N R3
OH
N
R3
O
R2
8
(+/-)
9 (+/-)
10 (+/-)
2
/reflux; ii—R R NH/CH
3
Scheme 2. Reagents and conditions: (a) i—SOCl
CH Cl /rt; ii—TFA/CH Cl /rt, 1 h.
2
2
Cl
2
; iii—NH NH
2
2
2 3
/MeOH; (b) i—BocNHCH COOH/EDC/HOBt/Et N/
2
2
2
2

1348
C. Chen et al. / Bioorg. Med. Chem. Lett. 18 (2008) 1346–1349
a
Table 2. SAR of the amides 9
matched that of milnacipran 5, but its SERT potency
was 10-fold lower. In contrast, the N-ethyl-N-benzyl
m exhibited 20-fold better SERT potency than the sec-
2
Compound R NR
3
NET
SERT
DAT
9
9
9
9
9
9
5
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
9
2
3
a
b
c
d
e
NHEt
c
>10,000 >10,000 >10,000
>10,000 >10,000 >10,000
ondary 9d, and 9m was the best SERT inhibitor among
the acyclic amides 9a–n. The diisopropyl 9n was only
weakly active at NET, indicative of a significant differ-
ence in conformations from 5 due to the bulkiness of
its amide side chain. Compared to the pyrrolidine and
morpholine analogs (9o–p), the piperidine 9q and hom-
opiperidine 9r were much more potent at NET, suggest-
ing a requirement for a relative lipophilic group with a
limited size at this site.
NH( Pr)
NHPh
570
>10,000 3200
>10,000 >10,000
>10,000
2400
NHBn
EtNMe
1400
77
830
420
NEt
2
6100
f
EtNPr
n
EtN( Bu)
41
1700
2000
>10,000
>10,000
g
h
i
360
8000
4900
180
14
t
EtN( Bu)
>10,000 >10,000
c
EtN( Hx)
7500
1300
280
>10,000
>10,000
>10,000
>10,000
>10,000
j
EtNCH
EtNCH
EtNPh
2
CH
2
OMe
k
l
2
CH@CH
2
The indoline 9s displayed high potencies at both NET
and SERT, much better than the acyclic 9l. The tetrahy-
droquinoline (THQ) 9t exhibited much reduced potency
from 9s, indicative of certain geometrical requirements.
The tetrahydroisoquinoline (THIQ) 9u exhibited equal
potency at both NET and SERT and was much better
than its acyclic 9m. Compound 9u had a similar profile
to duloxetine 3 at NET and SERT but was less potent
at DAT.
63
4400
160
m
n
o
p
q
r
EtNBn
i
N( Pr)
200
2,800
3700
2300
170
2
>10,000 >10,000
Pyrrolidin-1-yl
Morpholin-1-yl
Piperidin-1-yl
3100
1700
1200
640
13
>10,000
7400
4600
Homopiperidin-1-yl 130
>10,000
3900
s
Indolin-1-yl
b
4.4
21
t
THQ
THIQ
100
8.3
>10,000
>10,000
3100
c
u
8.4
5.1
8.9
190
6.6
Since milnacipran is a relatively rigid molecule, NMR
is a good tool to study its conformation in solution.
Kazuta et al. have investigated the conformation of an
ethyl milnacipran analog in aqueous solution based on
660
a
b
c
Data are average of two or more independent measurements.
THQ, tetrahydroquinoline.
THIQ, tetrahydroisoquinoline.
1
8
NOE data. Their results indicate that the amino nitro-
gen is away from the carbonyl oxygen due to the steric
clash caused by the 4-ethyl group, and this compound
Table 3. Summary of NMR studies of milnacipran
1
9
is devoid of SERT activity. In our NMR experiments,
8
it was observed that the proton 1 and the proton 4 of
milnacipran 5 (see numbering in Table 3) in water had
NOE correlation, suggesting their close proximity. The
two methyl groups of 5 had different chemical shifts,
indicating their different chemical environments. One
of the two methyl groups was located close to the phenyl
ring since NOE was observed between the methyl and
the ortho-proton (Table 3). Its associated ethylene was
nearby to the cyclopropane because NOE between this
ethylene and one of the protons at position-1 was
observed. These results provide evidence for a possible
3D structure of milnacipran in aqueous solution, and
a model was constructed based on these data using
computational simulation (Fig. 2). In this conformer,
9
H1
7
1
H '
1
0
1
6
H
11
1
7
3
2
H4
18
N14
12
4
H '
4
1
5
O13
N5
16
H
H
Proton
1
Chemical shift
NOE
0
0
0
1.48 (m, 1H)
1.79 (m, 1H)
1.79 (m, 1H)
3.12 (m, 2H)
7.33 (m, 2H)
7.44 (m, 2H)
7.34 (m, 1H)
18,1 ,4,4 ,17,17
1,2,7, or 11
0
1
0
0
2
1 ,4, or 4
1,2
0
4
7
8
9
1
1
1
1
,4
0
0
,11
,10
18,1 ,17,17
0
0
5,15
6
3.35 and 3.42 (m, 1H)
1.13 (t, 3H)
16
15,15
0
7,17
8
3.39 and 3.55 (m, 1H)
0.83 (t, 3H)
18,1,7,11
0
1,17,17 ,7,11
potencies at all three transporters. While the methoxy-
ethyl 9j was slightly better than the butyl 9g at both
NET and SERT, the N-allyl 9k showed significant
improvement over the N-propyl 9f for the two trans-
porters, indicating an important role of the p-electrons
of the allylic double bond.
The N-ethyl-N-phenyl amide 9l was about 10-fold better
than the secondary amide 9c. Its NET activity also
Figure 2. Computational simulation of milnacipran conformation
required for NET inhibition.

C. Chen et al. / Bioorg. Med. Chem. Lett. 18 (2008) 1346–1349
1349
a
Table 4. SAR of aminoacetamides 10a–d
cological profiles of compounds 9s (clogP = 2.7) and 9u
2
R NR
3
(
(
clogP = 2.6) matched with that of duloxetine
clogP = 3.7) at the transporters. These compounds
Compound
NET
SERT
DAT
1
1
1
1
0a
0b
0c
0d
NEt
2
700
>10,000
12
2200
>10,000
77
>10,000
4600
i
N( Pr)
possessed lower lipophilicity than many of the marketed
drugs in this class. A pharmacophore model for the mil-
nacipran analogs was also established based on the con-
formational analysis in aqueous solution, which could
be useful for designing novel potent monoamine trans-
porter inhibitors with ideal pharmacokinetic properties.
2
Indolin-1-yl
THIQ
>10,000
>10,000
150
150
a
Data are average of two or more independent measurements.
the distance between the amine nitrogen and the benzene
2
centroid is about 5.6 A . One of the ethyl groups is clo-
sely located to the phenyl and cyclopropane, and the
aminomethyl protons are in a close range with one of
the methylene protons of the cyclopropane. This confor-
˚
References and notes
1
2
. Iversen, L. Br. J. Pharmacol. 2006, 147, S82.
. Runyon, S. P.; Carroll, F. I. Curr. Top. Med. Chem. 2006,
6, 1825.
mation puts the NH proton near the carbonyl oxygen
2
atom, which may form hydrogen bonding.
3. Stokes, P. E.; Holtz, A. Clin. Therapeut. 1997, 19, 1135.
4
. Gibson, A. P.; Bettinger, T. L.; Patel, N. C.; Crismon, M.
L. Ann. Pharmacother. 2006, 40, 1134.
This conformation might be the essential active pharma-
cophore required at least for NET, and the SAR data
support this model. Thus, N-methylation of 5 decreases
its chance of forming this conformation due to a steric
effect, resulting in a much less potent inhibitor at both
NET and SERT. Any N-alkyl group basically reduces
the possibility of this conformation. It is also possible
that the allyl group of 9k contributes an additional p–
p interaction with the phenyl group compared to the
ethyl group of 5, resulting in further stabilization of this
favored conformation. For high SERT potency, an
additional phenyl group is required at the amide side
chain to interact with the transporter. The fact that
the cyclic THIQ 9u exhibited much higher potency at
NET and SERT than its acyclic N-ethyl-N-benzyl ana-
log 9m also indicates that there is a limited space avail-
able in the transporters and so key binding features of
an inhibitor are optimally contained in a more compact
molecule.
5
6
7
8
. Frampton, J. E.; Plosker, G. L. CNS Drugs 2007, 21, 581.
. Dierick, M. Eur. Psychiatry 1997, 12, 307s.
. Bisserbe, J. C. Int. Clin. Psychopharmacol. 2002, 17, S43.
. Puozzo, C.; Panconi, E.; Deprez, D. Int. Clin. Psycho-
pharmacol. 2002, 17, S25.
9
. Stahl, S. M.; Grady, M. M.; Moret, C.; Briley, M. CNS
Spectr. 2005, 10, 732.
10. Mochizucki, D. Hum. Psychopharmacol. 2004, 19, S15-9.
11. (a) Rooks, D. S. Curr. Opin. Rheumatol. 2007, 19, 111; (b)
Leo, R. J.; Brooks, V. L. Curr. Opin. Investig. Drugs 2006,
7, 637.
1
1
2. The calculated logD values are 1.2 for milnacipran and 3.3
for duloxetine using ACD software.
3. Puozzo, C.; Lens, S.; Reh, C.; Michaelis, K.; Rosillon, D.;
Deroubaix, X.; Deprez, D. Clin. Pharmacokinet. 2005, 44,
9
77.
1
4. Bonnaud, B.; Cousse, H.; Mouzin, G.; Briley, M.; Stenger,
A.; Fauran, F.; Couzinier, J. P. J. Med. Chem. 1987, 30,
318.
15. Roggen, H.; Kehler, J.; Stensbol, T. B.; Hansen, T. Bioorg.
Med. Chem. Lett. 2007, 17, 2834.
1
1
1
6. Shuto, S.; Ono, S.; Imoto, H.; Yoshii, K.; Matsuda, A. J.
Med. Chem. 1998, 41, 3507.
7. Owens, M. J.; Morgan, W. N.; Plott, S. J.; Nemeroff, C. B.
J. Pharmacol. Exp. Ther. 1997, 283, 1305.
8. Kazuta, Y.; Tsujita, R.; Yamashita, K.; Uchino, S.;
Kohsaka, S.; Matsuda, A.; Shuto, S. Bioorg. Med. Chem.
To test this model, we synthesized several aminoaceta-
mides of milnacipran analogs and the results are sum-
marized in Table 4. The aminoacetamide of
milnacipran 10a displayed 9- and 5-fold reduction in po-
tency at NET and SERT, respectively, compared to 5,
and these values were similar to those of the N-ethyl
analog 6b. In comparison, the indoline derivative 10c
only reduced NET activity less than 3-fold from its par-
ent 9s. These results might suggest the terminal amine of
2
002, 10, 3829.
1
9. The NMR experiments were performed on a Varian
500 MHz spectrometer at 27 °C. Milnacipran (5 mg) was
dissolved in 25 mM TRIS-d11 buffer (pH = 7.0), and water
suppression was accomplished using Excitation Sculpting
with Gradients. Protons 16 and 18 were assigned from
chemical shifts and multiplicities. Protons 15 and 17 were
assigned based on correlations in COSY. Protons 7–11
were assigned from absolute value COSY. Protons 1, 2,
and 4 were assigned from chemical shift and COSY
correlations. The overlap of one proton from 1 and the
proton 2 is confirmed via the HSQC experiment, which
clearly shows two different carbons with the similar proton
chemical shift.
1
9
0 is able to replace the function of the basic nitrogen in
through a different conformation.
In summary, a series of milnacipran derivatives were
synthesized, their SAR of NET and SERT inhibition
was studied, and potent compounds were discovered.
Thus, compound 9k exhibited a similar pharmacological
profile to atomoxetine but with lower lipophilicity
(
clogD = 1.5 vs 3.3 for atomoxetine), while the pharma-