Efficient synthesis of the κ-opioid receptor agonist CJ-15,161: Four stereospecific inversions at a single aziridinium stereogenic center

Article abstract of DOI:10.1016/S0957-4166(03)00581-0

An efficient four-step sequence has been developed for the synthesis of the κ-opioid receptor agonist CJ-15,161. The process features four consecutive regioselective and stereospecific inversions at a single aziridinium stereogenic center, which leads to

Full text of DOI:10.1016/S0957-4166(03)00581-0

TETRAHEDRON:
ASYMMETRY
Pergamon
Tetrahedron: Asymmetry 14 (2003) 3517–3523
Efficient synthesis of the k-opioid receptor agonist CJ-15,161:
four stereospecific inversions at a single aziridinium stereogenic
center
Michel Couturier,a,* John L. Tucker,a Brian M. Andresen,a Keith M. DeVries,a Brian C. Vanderplasa
and Fumitaka Itob
aChemical Research & Development, Pfizer Inc., Eastern Point Road, Groton, CT 06340, USA
bPfizer Global Research and Development, Pfizer Inc., Nagoya, Japan
Received 21 March 2003; accepted 22 May 2003
Abstract—An efficient four-step sequence has been developed for the synthesis of the k-opioid receptor agonist CJ-15,161. The
process features four consecutive regioselective and stereospecific inversions at a single aziridinium stereogenic center, which leads
to overall retention of stereochemistry, in a single operation. The chemistry is straightforward, practical and amenable to
large-scale synthesis.
© 2003 Elsevier Ltd. All rights reserved.
1. Introduction
point of view, but was not amenable to large-scale
preparation and long-term manufacturing (Scheme 1).
More specifically, there were no crystalline intermedi-
ates throughout the synthesis, the styrene oxide 2 ring
opening by pyrrolidine 3 lacked regioselectivity, the
reactions did not lend themselves to easy manipula-
tions, and the actual drug substance was an amorphous
hydrochloride salt. While we sought alternate routes to
CJ-15,161 1,5 we improved upon the current synthesis
to enable the production of drug substance.
It is well established that opioid analgesics such as
morphine are therapeutically useful, but their usage is
limited due to adverse side effects, such as physical
addictiveness, withdrawal properties, and respiratory
depression, amongst others. Since opiates exert their
effects by interacting with opioid receptors, consider-
able research has been directed towards determining the
distinct pharmacological profiles of the m (OP3), d (OP1)
and k (OP2) sub-types of receptors.1 While m opioid
receptors mediate opiate phenomena associated with
morphine, including analgesia, euphoria, respiratory
functions and physical dependence,2 stimulation of the
k receptors has been shown to produce analgesia.3
Separating the action based on the former receptor
from the latter has been investigated to develop potent,
non-addictive analgesics with reduced tendency to
cause dependence.
The initial goal was to identify a final crystalline form
for the drug substance: Salt screens showed that the
benzoic acid salt mono-hydrate was a stable polymorph
suitable for formulation (Fig. 1).
As mentioned above, the penultimate MOM-protected
intermediate 6 was non-crystalline, and purification of
this oil required silica gel chromatography. We there-
fore sought an alternative protecting group that would
impart crystallinity and would allow purification by
simple crystallization. Using the bottoms-up approach,6
we derivatized the alcohol group of 1 in various pro-
tected forms compatible with the previous chemistry,
and identified the benzoate ester 7 as a stable crystalline
intermediate. Based on this, we protected pyrrodinol 3
as the benzoate ester 8 (Scheme 2). By performing the
In a recent development program of the k-opioid recep-
tor agonist drug candidate CJ-15,161 1,4 we required
large scale production of the drug substance to support
toxicological screening and clinical trials. The original
discovery route was straightforward from a synthetic
* Corresponding
groton.pfizer.com
author.
E-mail:
michel a couturier@
– –
0957-4166/$ - see front matter © 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0957-4166(03)00581-0

3518
M. Couturier et al. / Tetrahedron: Asymmetry 14 (2003) 3517–3523
Scheme 1. Original discovery route. Reagents and conditions: (a) MOMCl, NaH, DMF, 70°C, 3 h, SiO2 chromatography, 97%;
(b) H2, Pd(OH)2/C, EtOH, rt, 40 h, quant.; (c) (S)-styrene oxide, EtOH, reflux, 2 h, SiO2 chromatography, 59%, 4:5=0.65:0.35;
(d) MsCl, TEA, CH2Cl2, rt, 12 h then 15, EtOH, reflux, 2 h, SiO2 chromatography, 88%; (e) HCl, MeOH, rt, 6 h, 89%.
reaction in the absence of base, the hydrogen chloride
generated in the process provided the hydrochloride
salt, which was conveniently isolated as a crystalline
solid by addition of MTBE upon reaction completion.
In addition to the hydrochloride’s crystallinity, the
subsequent N-benzyl hydrogenolysis with 10% Pd/C in
isopropanol was accelerated when compared to the
corresponding oily free-base, which actually stalled at
some point.7 Upon reaction completion and filtration
of the catalyst, the solvent was concentrated with con-
comitant azeotropic removal of water originating from
the wet catalyst. The resulting solution was then diluted
with diisopropyl ether, from which pyrrolidine 9 crys-
tallized out of solution. When performed on a multi-
kilo scale, the reactions yielded 97 and 87%,
respectively, of easily filterable crystalline solids.
mediate 13. Thus, the isolation of the benzyl chloride
12 further substantiates that the reaction leading to
diamine 7 proceeds with anchimeric assistance on two
successive occasions in order to account for the overall
net retention of configuration at the benzylic site.9 The
foregoing argument that a total of four stereospecific
inversions occurred at the benzylic position adds fur-
ther insight to earlier reports that consider the net
retention a result of two successive inversions.10
With the requisite pyrrolidine 9 in hand, the key cou-
pling with (S)-styrene oxide was investigated. To that
effect, the preparation of b-substituted amines from
mixtures of styrene oxide opening products via a com-
mon aziridinium ion intermediate was recently commu-
nicated.8 In essence, the production of the two
regioisomers 10 and 11 is not problematic since they
both independently lead to the desired diamine 7. As
evidenced by the X-ray of CJ-15,161 1 (Fig. 1), the
overall transformation proceeds with net retention of
configuration at the benzylic site. More specifically, we
have found that the two isomers 10 and 11 initially
produce a single common intermediate isolated and
characterized as the benzylic chloride 12 (Scheme 3).
From a mechanistic viewpoint, formation of chloride
12 from both regioisomers 10 and 11 can be explained
by the common intermediacy of the aziridinium inter-
Figure 1. ORTEP representation of 1·BzOH·H2O.

M. Couturier et al. / Tetrahedron: Asymmetry 14 (2003) 3517–3523
3519
Scheme 2. Reagents and conditions: (a) BzCl, CH2Cl2 then MTBE, 97%; (b) H2, 10% Pd/C, i-PrOH then IPE, 87%.
Scheme 3.
From a process chemistry perspective, the regioisomeric
mixture of 10 and 11 was an oil, whereas the isolated
benzylic alcohol 10 is a stable, crystalline intermediate.
Therefore, it would be desirable to identify a regioselec-
tive synthesis of 10, which could be purified by simple
crystallization, and thus provide an additional point of
purity control in the synthesis. We initially attempted to
perform the coupling reaction directly with the pyrroli-
dinium salt 9 by free-basing in situ. Unfortunately, even
after an exhaustive screen of bases, we were not able to
effect the reaction starting from the hydrochloride salt
with an acceptable rate and complete conversion.
Isolations from polar, high boiling solvents can be
difficult. In the course of our experiments, however, we
found that the combination of NMP with water as an
anti-solvent selectively crystallized the predominant
regioisomer 10. Fortunately, the resulting slurry filters
relatively quickly despite the aqueous solvent mixture.
The optimized procedure for coupling was to free-base
pyrrolidinium 9 with MeTHF, displace into NMP, add
Table 1. Regioselective styrene oxide opening
Entry
Solvent
10:11 ratio
Conversion (%)
The preferred solvent utilized for the free-basing proce-
dure was methyl-THF, which provided good solubility
for the free base, and excellent separation from water, as
compared to THF. It was later observed that while
residual amounts of up to 10% MeTHF left behind after
displacement in the reaction solvent slowed the reaction
slightly, it did not adversely affect the isomer ratio or
impurity profile.11 Solvent screening was performed to
optimize the 65:35 regioselectivity (10:11) observed in the
original coupling reaction when performed in ethanol
(Table 1). In general, higher boiling, more polar solvents
both increased the reaction rate and the ratio of 10:11.
Performing the coupling in N-methylpyrrolidinone at
100°C offered the best regioselectivity (12:1), with reac-
tion completion after 12 h.
1
2
3
4
5
6
7
8
THF
2-Me-THF
NMP
MeCN
DMF
DMAC
DMSO
MTBE
Diisopropyl ether
n-BuOAc
MeOPiv
EtOH
82:18
80:20
92:8
97
94
98
90
95
77
89
95
95
75
26
70
50
79
74:26
84:16
86:14
88:12
73:27
64:36
80:20
79:21
65:35
78:22
9
10
11
12
13
14
i-PrOH
1,2-Dichloroethane 58:42

3520
M. Couturier et al. / Tetrahedron: Asymmetry 14 (2003) 3517–3523
Scheme 4. Reagents and conditions: (a) MsCl, TEA, CH2Cl2 then 14, 80% (b) aq. NaOH, i-PrOH then BzOH, i-PrOH, 81%.
This work demonstrates an efficient and practical large-
scale synthesis of CJ-15,161, where all the intermediates
are crystalline solids and only one biphasic extraction is
utilized throughout the synthesis. The end-game fea-
tures a saponification quench with benzoic acid, which
concomitantly produces the desired final benzoate salt
form directly in the correct polymorphic state. The
chemistry was exemplified on a multi-kilogram scale
and the operations lend themselves to simple
manipulations.
styrene oxide, heat to 100°C for 12 h, cool, add water,
and filter. This procedure was exemplified in the kilo-
lab facility12 on a 7.2 kg scale, resulting in the isolation
of 5.9 kg (60%) of the crystalline aminoalcohol 10.13
With alcohol 10 in hand, we proceeded with the key
aziridinium formation and coupling (Scheme 4). We
found that methylene chloride was an ideal solvent for
the mesylation/chloride displacement reaction, and that
trimethylamine, required for both the activation and
coupling steps, could be added in one initial charge. We
also found that dichloromethane could advantageously
be used in the coupling step instead of ethanol. Exclu-
sion of the latter offered a single solvent process, elimi-
2. Experimental
2.1. General methods
nating
a
reaction quench and solvent swap.
Additionally, the coupling occurred at a lower tempera-
ture, and there was no longer the need to remove the
ethanol prior to the aqueous quench. Furthermore, any
residual pyrrolidine 9 was purged with two 1 M hydro-
chloric acid washes.14 Methylene chloride, with its low
boiling point, made for easy displacement with ethyl
acetate, and subsequent crystallization of the product
via hexane addition and seeding. The penultimate inter-
mediate was isolated in 87% yield (>99% purity) as a
crystalline free-base. The process was exemplified on a
5.9 kg scale providing 7.2 kg of pure (99%), crystalline
material.
Unless otherwise noted, all operations were performed
in Clean-By-Test nitrogen purged vessels. Charges and
transfers were performed using an isolated vacuum
1
whenever possible. The H and 13C NMR spectra were
recorded on a Varian Innova 400 spectrometer. Ele-
mental analyses were performed by Schwarzkopf
Microanalytical Laboratory, Inc. (Woodside, NY).
Melting points were obtained from a Thomas Hoover
Uni-Melt capillary apparatus and are uncorrected.
2.2. 4-N%-Methylamino-N-propylbenzamide 14
To a suspension of 4-N-methylaminobenzoic acid (34.5
kg, 229 mol) in dichloromethane (455 L) was added
propylamine (22 L, 267 mol) followed by EDC (52.6
kg, 274 mol). The resulting suspension was stirred at rt
for 17 h, then water (150 L) added and the biphasic
mixture stirred for 15 min. The layers were allowed to
settle, and then the aqueous phase extracted with
dichloromethane (50 L). The combined organic layers
were washed twice with aqueous citric acid (10%, 2×115
L) and then concentrated down to a crude solid (37.76
kg, 85%). The crude material (30.11 kg) was purified by
dissolution in ethyl acetate (45 L) and toluene (45 L) at
70°C, followed by cooling to rt whereupon additional
toluene (90 L) was added to the resulting slurry. The
mixture was further cooled to 0°C and stirred for an
additional 2 h. The crystalline material was collected by
filtration and rinsed with ice-cold toluene (10 L) to
provide the title compound (26.3 kg, 87%) as a colorless
The final transformation leading to the drug substance
entails a saponification, followed by salt formation
using benzoic acid in the presence of water to produce
the mono-hydrated salt. Since the conjugate acid of the
sodium benzoate produced in the saponification is the
actual acid used to produce the final salt form, there
lies an opportunity to neutralize the saponification
reaction mixture with benzoic acid, in which case there
cannot be scrambling between different counter-ions.
Accordingly, instead of going through the usual series
of pH adjustments and biphasic extractions, we devised
a simple one-pot, one-isolation process. Following
treatment of benzoate 7 with 1.2 equiv. of 1 M sodium
hydroxide for 4 h at 50°C in 5 volumes (L of solvent
per kg of substrate) of isopropanol, 2.3 equiv. of ben-
zoic acid were added15 to neutralize the excess sodium
hydroxide and form the desired benzoate salt. The
procedure was tailored such that when the clear solu-
tion was cooled to room temperature, the product
crystallized out while the sodium benzoate by-product
remained soluble. Following this simple process, the
crystalline material was isolated by filtration to provide
an 83% yield of drug substance, in the correct polymor-
phic state.
1
solid, mp 94–95°C; H NMR (400 MHz, CDCl3) l 7.63
(d, J=3.5 Hz, 2H), 6.58 (d, J=3.5 Hz, 2H), 6.09 (brs,
1H), 3.36 (q, J=6.5 Hz, 2H), 2.85 (s, 3H), 1.60 (m, 2H),
0.95 (t, J=7.5 Hz); 13C NMR (100 MHz, CDCl3) l
167.7, 151.5, 128.7, 123.5, 112.0, 41.8, 30.8, 23.3, 11.7.
Anal. calcd for C11H15N2O: C, 68.72; H, 8.39; N, 14.57.
Found: C, 68.70; H, 8.59; N, 14.50.

M. Couturier et al. / Tetrahedron: Asymmetry 14 (2003) 3517–3523
3521
2.3. (S)-3-Benzoyloxy-N-benzylpyrrolidine hydrochlo-
2-methyltetrahydrofuran (14 L).11 The mixture was
allowed to settle and the layers separated. The
aqueous phase was extracted with an additional 2-
methyltetrahydrofuran (14 L). The combined organic
solutions were washed with brine (7 L), dried over
magnesium sulfate (3.7 kg), filtered over Celite, and
the latter rinsed with additional 2-methyltetra-
hydrofuran (18 L). The combined filtrate and rinse
solutions were concentrated by vacuum distillation (to
a total volume of 22 L). At this stage, 1-methyl-2-
pyrrolidinone (15 L) was added and the resulting
solution was further vacuum distilled to remove resid-
ual 2-methyltetrahydrofuran. A solution of (S)-styrene
oxide (4.0 kg, 33 mol) in 1-methyl-2-pyrrolidinone (36
L) was then added and the resulting mixture was
heated to 100°C for 12 h. Once the reaction was
deemed complete, the solution was cooled to 25°C
and water (22 L) added portion-wise over 30 min.
After the addition of seed crystals, additional water
(29 L) was added portion-wise over 1 h, and the
resulting slurry stirred for a further 4 h. The solids
were isolated by filtration, washed with water (22 L)
and dried under vacuum at 45°C to provide the title
pyrrolidine (5.9 kg, 58%) as a colorless crystalline
material, mp 103–104°C; [h]2D2=+44.8 (c 1.1, MeOH);
1H NMR (400 MHz, CDCl3) l 8.06–8.03 (m, 2H),
7.59–7.55 (m, 1H), 7.47–7.43 (m, 2H), 7.40–7.32 (m,
4H), 7.29–7.25 (m, 1H), 5.47–5.43 (m, 1H), 4.80 (dd,
J=3.2, 10.4 Hz, 1H), 3.29–3.24 (m, 1H), 3.17–3.11
(m, 1H), 2.93–2.83 (m, 2H), 2.72–2.62 (m, 2H), 2.43–
2.36 (m, 1H), 2.10–2.07 (m, 1H); 13C NMR (100
MHz, CDCl3) l 166.6, 142.0, 133.4, 130.3, 129.8,
128.6, 127.9, 126.1, 74.6, 70.8, 64.1, 60.3, 52.9, 32.2;
MS [(m+1)/z] 312.2. Anal. calcd for C19H21NO3: C,
73.29; H, 6.80; N, 4.50. Found: C, 73.49; H, 6.74; N,
4.59.
ride 8
To an ice cold, stirred solution of (S)-N-benzyl-3-
pyrrolidinol (6.90 kg, 38.9 mol) in dichloromethane
(21 L), was slowly added a solution of benzoyl chlo-
ride (5.75 kg, 40.8 mol) in dichloromethane (5 L).
The reaction mixture was warmed to rt and stirred
for an additional 2 h. Upon confirmation of reaction
completion, methyl t-butyl ether (26 L) was added,
and the resulting slurry stirred for 2 h at −10°C. The
solids were isolated by filtration, washed with methyl
t-butyl ether (14 L) and dried under vacuum at 40°C
to provide the title pyrrolidine (11.8 kg, 96.6%) as a
colorless crystalline material, mp 203–204°C; 1H
NMR (400 MHz, CDCl3) (major isomer) l 8.14–7.90
(m, 2H), 7.70–7.39 (m, 8H), 5.59–5.58 (m, 1H), 4.30
(d, J=5.5 Hz, 2H), 4.14–4.08 (m, 1H), 3.80–3.76 (m,
1H), 3.24–3.17 (m, 2H), 2.66–2.61 (m, 1H), 2.37–2.32
(m, 1H); 13C NMR (100 MHz, CDCl3) l 166.2,
165.6, 134.0, 133.7, 131.1, 130.9, 130.3, 130.2, 130.1,
130.0, 129.8, 129.6, 129.5, 129.4, 129.1, 129.0, 128.8,
128.6, 73.0, 71.9, 59.1, 58.5, 57.5, 57.2, 52.1, 52.0,
31.3, 30.6; MS [(m+1)/z] 282.2. Anal. calcd for
C18H20ClNO2: C, 68.03; H, 6.34; N, 4.41. Found: C,
67.81; H, 6.28; N, 4.44.
2.4. (S)-3-Benzoyloxypyrrolidine hydrochloride 9
A warm suspension of (S)-3-benzoyloxy-N-benzyl-
pyrrolidine hydrochloride (11.8 kg, 37.0 mol) in 2-
propanol (118 L) at 40°C was hydrogenated over 10%
palladium on carbon (2.35 kg, 50% water wet) under
50 psig for 12 h. Upon confirmation of reaction com-
pletion, the mixture was filtered over Celite and the
latter rinsed with additional 2-propanol (38 L). The
combined filtrate and rinse solutions were concen-
trated by atmospheric distillation (to a total volume
of 35 L). At this stage, diisopropyl ether (105 L) was
slowly added over 15 min, and the resulting slurry
was further stirred at 55°C for a further 15 min. The
reaction mixture was gradually cooled to −10°C, and
stirred for an additional 2 h. The solids were isolated
by filtration, washed with diisopropyl ether (30 L)
and dried under vacuum at 40°C to provide the title
pyrrolidine (7.3 kg, 86.6%) as a colorless crystalline
material, mp 140–141°C; 1H NMR (400 MHz, d6
DMSO) l 9.65 (bs, s, 2H), 8.04–8.01 (m, 2H), 7.68–
7.64 (m, 1H), 7.53–7.49 (m, 2H), 5.53–5.51 (m, 1H),
3.42–3.41 (m, 2H), 3.33–3.28 (m, 2H), 2.25–2.15 (m,
2H); 13C NMR (100 MHz, d6 DMSO) l 165.8, 134.3,
130.2, 130.0, 129.3, 74.2, 50.4, 43.7, 31.3; MS [(m+1)/
z] 192.2. Anal. calcd for C11H14ClNO2: C, 58.03; H,
6.20; N, 6.15. Found: C, 58.00; H, 6.07; N, 6.30.
2.6. (2%S,3S)-3-Benzoyloxy-N-(2-chloro-2-phenyl)-
ethylpyrrolidine 12
A solution of (2%S,3S)-3-benzoyloxy-N-(2-hydroxy-2-
phenyl)ethyl-pyrrolidine (350 mg, 1.10 mmol) in
dichloromethane (3.5 mL) was treated with triethyl-
amine (157 mL, 1.20 mmol) followed by a dropwise
addition of methanesulfonyl chloride (96 mL, 1.10
mmol). The reaction mixture was stirred for 30 min,
and then diluted with additional dichloromethane (10
mL) and quenched with water (5 mL). The organic
layer was separated, dried over sodium sulfate, con-
centrated and the residual material purified by silica
gel chromatography (20% ethyl acetate in hexane) to
provide the title compound (236 mg, 64%) as a color-
1
less oil; H NMR (400 MHz, CDCl3) l 8.06–8.04 (m,
2H), 7.57–7.55 (m, 1H), 7.45–7.30 (m, 7H), 5.42–5.38
(m, 1H), 4.97 (dd, J=6.4, 7.9 Hz, 1H), 3.21 (dd,
J=7.9, 13.1 Hz, 1H), 3.11–3.01 (m, 2H), 2.85–2.79
(m, 2H), 2.70–2.65 (m, H), 2.32–2.27 (m, 1H), 2.00–
1.96 (m, 1H); 13C NMR (100 MHz, CDCl3) l 166.6,
140.7, 133.3, 130.5, 129.9, 128.9, 128.7, 128.6, 127.5,
74.9, 64.1, 61.7, 60.3, 53.3, 32.1; HRMS calcd for
C19H21ClNO2 (M++1) 330.1261, found 330.1267.
2.5. (2%S,3S)-3-Benzoyloxy-N-(2-hydroxy-2-phenyl)-
ethylpyrrolidine 10
To a stirred solution of sodium hydroxide (1.52 kg,
38.0 mol) in water (35 L), was added (S)-3-benzoyl-
oxypyrrolidine hydrochloride (7.20 kg, 33.0 mol) and

3522
M. Couturier et al. / Tetrahedron: Asymmetry 14 (2003) 3517–3523
2.7.
(2%S,3S)-3-Benzoyloxy-N-{2-[N-methyl-N-4-(N-
with 2-propanol (7.2 L) and dried under vacuum at
45°C to provide the title pyrrolidine (6.30 kg, 81%) as a
propylaminocarbonyl)phenyl]amino-2-phenyl}ethyl-
pyrrolidine 7
colorless crystalline material. Mp 110–111°C; [h]2D2=
1
+168.4 (c 1.1, MeOH); H NMR (400 MHz, d6 DMSO)
l 8.043 (t, J=5.6, 1H), 7.93–7.91 (m, 2H), 7.65 (d,
J=9.2 Hz, 2H), 7.60–7.56 (m, 1H), 7.48–7.44 (m, 2H),
7.29–7.28 (m, 3H), 7.26–7.19 (m, 1H), 6.79 (d, J=9.2),
5.16–5.13 (m, 1H), 4.12–4.07 (m, 1H), 3.16–3.11 (m,
2H), 3.05–3.00 (m, 1H), 2.93–2.89 (m, 1H), 2.76–2.74
(m, 1H), 2.60–2.56 (m, 2H), 2.48–2.46 (m, 2H), 2.73
(dd, J=4, 9.6 Hz, 1H), 1.93–1.84 (m, 1H), 1.51–1.42
(m, 3H), 0.83 (t, J=7.6, 3H); 13C NMR (100 MHz, d6
DMSO) l 168.5, 166.7, 152.3, 140.8, 133.0, 132.6,
129.9, 129.3, 129.2, 129.0, 127.8, 122.5, 111.8, 70.0,
63.1, 59.5, 58.1, 53.3, 41.5, 34.9, 32.7, 23.3, 12.2; MS
[(m+1)/z] 382.2. Anal. calcd for C30H39N3O4: C, 69.07;
H, 7.54; N, 8.06. Found: C, 69.07; H, 7.43; N, 8.15.
A solution of (2%S,3S)-3-benzoyloxy-N-(2-hydroxy-2-
phenyl)ethyl-pyrrolidine (5.82 kg, 18.7 mol) in
dichloromethane (70 L) was distilled to a total volume
of 64 L. The resulting solution was cooled to 0°C and
treated with triethylamine (6.25 L, 44.9 mol) followed
by the slow addition of methanesulfonyl chloride (1.74
L, 22.4 mol) over 30 min. The solution was then
warmed to 20°C and stirred for a further 30 min. Once
the reaction was deemed complete, 4-(N-methylamino)-
N-propylbenzamide (3.59 kg, 18.7 mol) was added and
the resulting solution heated to reflux for 12 h. Upon
confirmation of reaction completion, the reaction mix-
ture was cooled to 20°C, and successively washed with
water (20 L), hydrochloric acid (1 M, 20 L), hydrochlo-
ric acid (1 M, 10 L), brine (10 L), aqueous sodium
hydroxide (29 L) and water (20 L). The organic phase
was concentrated by atmospheric distillation to a total
volume of 12 L. The solution was cooled to 25°C after
which ethyl acetate (29 L) was added and the mixture
concentrated further by atmospheric distillation to a
total volume of 12 L. The solution was cooled to 25°C
and hexane (2 L) was added portion-wise over 30 min.
After the addition of seed crystals, additional hexane
(12.5 L) was slowly added over an hour, and the
resulting slurry was further stirred for 12 h. The solids
were isolated by filtration, washed with hexane (17.5 L)
and dried under vacuum at 45°C to provide the title
pyrrolidine (7.24 kg, 80%) as a colorless crystalline
material, mp 105–107°C; [h]2D2=+161.9 (c 1.1, MeOH);
1H NMR (400 MHz, d6 DMSO) l 8.06–8.03 (m, 1H),
7.86 (d, J=7.6, 2H), 7.66 (d, J=9.2 Hz, 2H), 7.64–7.60
(m, 1H), 7.51–7.47 (m, 2H), 7.32–7.19 (m, 4H), 6.82 (d,
J=9.2, 2H), 5.23–5.19 (m, 2H), 3.17–3.06 (m, 3H),
3.00–2.95 (m, 1H), 2.86–2.76 (m, 6H), 2.52 (dd, J=8.0
Hz, 14.4, 1H), 2.21–2.13 (m, 1H), 1.79–1.75 (m, 1H),
1.51–1.42 (m, 2H), 0.083 (t, J=7.2 Hz, 3H); 13C NMR
(100 MHz, CDCl3) l 167.6, 166.6, 152.7, 133.2, 129.8,
128.8, 128.7, 128.6, 127.6, 127.2, 112.1, 74.8, 60.5, 60.4,
57.5, 53.1, 41.8, 32.7, 32.1, 23.3, 11.7; MS [(m+1)/z]
486.2. Anal. calcd for C30H35N3O3: C, 74.20; H, 7.26;
N, 8.65. Found: C, 73.94; H, 7.12; N, 8.68.
Acknowledgements
We thank Mr. Richard Patterson, Tom Staigers,
Thomas St. Louis and co-workers for their assistance in
successfully transferring the process to the pilot plant,
and Mr. Kyle Leeman and Ms. Ioulia Loubkina for
their analytical support. The authors also wish to thank
Drs. Ste´phane Caron, John Ragan and Joanna Negri
for insightful comments and suggestions.
References
1. Dhawan, B. N.; Cesselin, F.; Raghubin, R.; Reisine, T.;
Bradley, P. B.; Portoghese, P. S.; Mamon, M. Pharmacol.
Rev. 1996, 48, 567.
2. Sporer, K. A. Ann. Int. Med. 1999, 130, 584.
3. Spanagel, R.; Herz, A.; Shippenberg, T. S. J. Neurochem.
1990, 55, 1743.
4. Ito, F.; Kondo, H. US 6,201,007 B1, 2001.
5. Ghosh, A.; Sieser, J. E.; Caron, S.; Watson, T. J. N.
Chem. Commun. 2002, 1644.
6. Chen, C.-K.; Singh, A. K. Org. Proc. Res. Dev. 2001, 5,
508.
7. As compared to methanol and ethanol, isopropanol was
deemed the solvent of choice as it leads to fewer baseline
impurities originating from transesterification.
8. Anderson, S. R.; Ayers, J. T.; DeVries, K. M.; Ito, F.;
Mendenhall, D.; Vanderplas, B. C. Tetrahedron Asymme-
try 1999, 10, 2655.
9. For isolation of b-chloro amines, see: (a) Weber, K.;
Kuklinski, S.; Gmeiner, P. Org. Lett. 2000, 2, 647. (b)
Dieter, R. K.; Deo, N.; Lagu, B.; Dieter, J. W. J. Org.
Chem. 1992, 57, 1663.
10. (a) Dieter, K. R.; Lagu, B.; Dieter, J. W.; Deo, N.;
Pennington, W. T. Synlett 1990, 109; (b) Rossiter, B. E.;
Eguchi, M.; Miao, G.; Swingle, N. M.; Herna´ndez, A. E.;
Vickers, D.; Fluckinger, E.; Patterson, G.; Reddy, K. V.
Tetrahedron 1993, 49, 965; (c) Miao, G.; Rossiter, B. E. J.
Org. Chem. 1995, 60, 8424; (d) O’Brian, P.; Poumellec, P.
Tetrahedron Lett. 1996, 37, 5619; (e) Saravanan, P.;
Singh, V. K. Tetrahedron Lett. 1998, 39, 167; (f) de
Sousa, S. E.; O’Brian, P.; Poumellec, P. J. Chem. Soc.,
Perkin Trans. 1 1998, 1484; (g) O’Brian, P.; Towers, T. D.
J. Org. Chem. 2002, 67, 304.
2.8. (2%S,3S)-3-Hydroxy-N-{2-[N-methyl-N-4-(N-propyl-
aminocarbonyl)phenyl]amino-2-phenyl}-ethylpyrrolidine
1
To a stirred solution of (2%S,3S)-3-benzoyloxy-N-{2-[N-
methyl-N-4-(N-propylaminocarbonyl)phenyl]amino-2-
phenyl}ethylpyrrolidine (7.24 kg, 14.9 mol) in
2-propanol (22 L) was added an aqueous solution of
sodium hydroxide (1 M, 18.2 L). The resulting slurry
was warmed to 55°C and stirred for 4 h. Upon confir-
mation of reaction completion, the solution was cooled
to 40°C and spec-free filtered. A warm solution of
benzoic acid (4.37 kg, 35.8 mol) in 2-propanol (14.5 L)
at 55°C was then added to the reaction mixture. The
resulting slurry was gradually cooled to 20°C over 2 h
and stirred for an additional 12 h at the same tempera-
ture. The solids were then isolated by filtration, washed

M. Couturier et al. / Tetrahedron: Asymmetry 14 (2003) 3517–3523
3523
11. Trace amounts of methylene chloride, as low as 0.1%
(v/v), were absolutely detrimental to the reaction. Reac-
tion completion was decreased, the ratio of product iso-
mers was negatively affected, and unknown impurities in
the range of 10% surfaced. Interestingly, it was noted
that the addition of methyl-THF to the reaction mixture
where the pyrrolidine had previously been extracted in
dichloromethane led to a cleaner reaction profile. This
could be attributed to the presence of BHT originating
from the methyl-THF.
13. The yield on scale was lower than the 75% typically
observed in laboratory pilots, owing to the fact that less
water than expected was added as anti-solvent.
14. In a control experiment, 98% of free-base was recovered
after dissolution in dichloromethane and vigorous stir-
ring with 1 M aqueous HCl.
15. At 2 volumes of isopropanol, the benzoic acid would
not come out of solution if kept above 30°C. To be
certain that no crystallization occurred in the lines dur-
ing speck-free filtration, the solution was heated to 50–
60°C and transferred through an insulated stainless steel
line.
12. Kobelski, E.; Staigers, T.; Sullivan, C. Pharm. Eng.
2002, 22, 1.