A novel chiral base mediated glutarimide desymmetrisation: Application to the asymmetric synthesis of (-)-paroxetine

Article abstract of DOI:10.1055/s-2002-35588

The asymmetric desymmetrisation of certain 4-aryl substituted glutarimides has been accomplished with high levels of selectivity (up to 97% ee) by enolisation with a chiral bis-lithium amide base. The selectivity of the reaction is shown to be the result of asymmetric enolisation, followed by a kinetic resolution. One of the chiral imides synthesised was converted into the selective seratonin reuptake inhibitor (-)-paroxetine.

Full text of DOI:10.1055/s-2002-35588

2074
LETTER
A Novel Chiral Base Mediated Glutarimide Desymmetrisation: Application to
the Asymmetric Synthesis of (–)-Paroxetine
C
hira
l
Base Media
a
ted
G
lutari
n
D
esym
i
metrisa
e
tion l A. Greenhalgh, Nigel S. Simpkins*
School of Chemistry, The University of Nottingham, University Park, Nottingham NG7 2RD, UK
Fax +44(115)9513564; E-mail: nigel.simpkins@nottingham.ac.uk
Received 3 October 2002
Ar
Ar
Abstract: The asymmetric desymmetrisation of certain 4-aryl sub-
E
stituted glutarimides has been accomplished with high levels of se-
lectivity (up to 97% ee) by enolisation with a chiral bis-lithium
amide base. The selectivity of the reaction is shown to be the result
of asymmetric enolisation, followed by a kinetic resolution. One of
the chiral imides synthesised was converted into the selective sera-
tonin reuptake inhibitor (–)-paroxetine.
chiral base
electrophile
O
O
O
N
O
N
R
R
4
5
O
O
Key words: chiral lithium amide, kinetic resolution, asymmetric
synthesis, glutarimide, paroxetine
Ar
O
6 Ar = 4-fluorophenyl
N
H
Previously, we have achieved high levels of enantioselec-
tivity in the chiral lithium amide base mediated substitu-
tion reactions of certain types of imide, e.g. the conversion
of imide 1 into silylated derivative 3, by base 2
(Scheme 1).1
Scheme 2
Here we describe the successful realisation of the asym-
metric process outlined in Scheme 2, and its use in a new
asymmetric synthesis of 6.
2
H
H
Ph
N
Ph
H
SiMe3
O
Li
The literature concerning the reactions of metal enolates
derived from glutarimides is sparse.3 In our hands enolate
alkylation reactions of glutarimides 4 were low yielding
when using LDA as the base, and some doubly substituted
products 7 were also obtained. Chiral base 2 also gave
rather disappointing yields, and we were prompted to use
the bis-lithium amide 8,4 which gave the results shown in
the Table 1.
O
O
O
Me3SiCl, LiCl, THF
N
N
Ph
Ph
ca -100 °C
3 80%, 91-95%ee
1
Scheme 1
In all such reactions a ring-fused imide was chosen so as
to avoid issues of diastereoselectivity in either the depro-
tonation or electrophilic quenching steps. However, this
places limitations on the use of this desymmetrisation ap-
proach, and we considered the more challenging chiral
base metallations of other types of imide an attractive
goal. Foremost amongst these was the transformation of
meso-glutarimides, such as 4 into chiral products 5,
(Scheme 2).
The use of 8 in its bis-lithiated form enabled the formation
of the desired products 5 in good yields, with good to ex-
cellent levels of ee, and as single diastereoisomers.5 The
trans arrangement of the newly installed substituent, rela-
tive to the fluorophenyl group was evident from the J val-
ues of associated ring protons (JH3–JH4 typically 11–13
Hz). The absolute configurations shown follow from the
conversion of one adduct 5i into paroxetine, as shown
below.
The success of this process would depend upon the selec-
tivity in both the metallation and quenching steps, and nei-
ther aspect appeared straightforward (see later). However,
this chemistry appeared enticing, given that an asymmet-
ric route to the drug substance (–)-paroxetine 6 could be
anticipated should access to appropriately substituted im-
ides 5 be possible.2
The somewhat variable nature of the levels of ee attracted
our attention. Overall the NBn series of compounds ap-
peared to give better levels of induction than the NMe se-
ries, but in both cases the results were found to be
variable. We noticed a broad correlation between reac-
tions that gave significant amounts of bis-substituted by-
product 7 and those that gave the highest levels of asym-
metric induction. For example, in the reaction leading to
5i, high levels of ee were found if the desired product was
accompanied by 10–20% of 7, whereas in reactions that
produced very little of 7 the ee could drop as low as 80%.6
Synlett 2002, No. 12, Print: 02 12 2002.
Art Id.1437-2096,E;2002,0,12,2074,2076,ftx,en;D20802ST.pdf.
© Georg Thieme Verlag Stuttgart · New York
ISSN 0936-5214

LETTER
Chiral Base Mediated Glutarimide Desymmetrisation
2075
Table 1 Enantioselective Substitution of Glutarimides 4
Although this level of enrichment is rather low, represent-
ing a selectivity factor S below 2, our findings broadly
parallel the observations of Gotov and Schmalz.8 We also
found that carboxymethyl derivative 5i could be enriched
from ca. 44% ee to 81% ee by further metallation with
base 8 and reaction with MeO2CCN.
F
F
F
Ph
Ph
Me
Ph
Me
Ph
N
Li
N
Li
8
E
E
E
+
electrophile
O
O
These results support the picture of ‘constructive kinetic
resolution’ superimposed upon the initial asymmetric
enolisation, illustrated in Scheme 3. We assume that this
type of process may be operative in all of the asymmetric
substitutions described here, although the extent of kinetic
resolution may be dependent upon the nature of the sub-
stituent introduced as well as the extent to which ‘over-
alkylation’ is allowed to proceed.
O
O
O
N
R
O
N
R
N
R
4 R = Me, Bn
5
7
R
Electrophile
Product 5
ee of 5
5:7a
(%yield)
Me
Me
Me
Me
Bn
Bn
Bn
Bn
Bn
MeI
5a (73)
5b (58)
5c (63)
5d (87)
5e (64)
5f (52)
5g (61)
5h (40)
5i (71)
86
74
77
75
97
90
97
97
97
2:1
BnBr
2.5:1
3:1
The availability of highly enantioenriched imides in syn-
thetically useful quantities via this method should be use-
ful for the preparation of a range of targets, including
biologically potent piperidines. To illustrate this point,
and to establish the sense of asymmetric induction in the
chiral base reactions, we carried out the conversion of im-
ide 5i into the aforementioned drug substance (–)-parox-
etine, Scheme 4.
ArCH2Brb
MeO2CCN
MeI
20:1
3:1 (14)
nd (7)
2:1 (22)
nd
allylBr
BnBr
PhCHOc
MeO2CCN
Ar
Ar
LiAlH4 (5 equiv.)
THF
6.5:1
CO2Me
O
X
X = OMs
a Ratio estimated from 1H NMR of crude reaction mixture. Figures in
brackets are isolated yields of 7.
reflux overnight
90%
O
N
N
Bn
Bn
b Ar = 4-bromophenyl.
c Isolated as a single diastereomer.
5i Ar = 4-fluorophenyl
9
X = OH
MsCl, py
72%
10 X = OMs
These findings suggested that the observed ee for the
products 5 was the result of an initial asymmetric enolisa-
tion of the starting imide 4, followed by an ee enhancing
kinetic resolution of 5, Scheme 3.
O
O
O
11
Ar
O
HO
(5 equiv.)
O
NaOMe (5 equiv.)
MeOH
reflux overnight 55%
N
Bn
Ar
12
E
O
O
N
8
Cl
O
R
8
Ar
O
(i)
slow
(+)-5 major
Ar
O
N
Cl
oC to RT
major
E
E
0
+
4
O
(ii) reflux, 3h
NaOH
(-) 6 54%
two steps
Ar
O
O
N
R
8
8
(iii) MeOH
reflux, 2h
E
HCl
H
minor
fast
7
13
O
O
N
R
Scheme 4
(-)-5 minor
Scheme 3
Global reduction of imide 5i (97% ee) gave piperidine al-
cohol 9, to which the appropriate sesamol side-chain 11
was introduced by conventional means.9 Deprotection of
the piperidine nitrogen then gave the desired drug sub-
stance as the free amine after base treatment.
The synthetic paroxetine prepared this way had [ ]20D –84
(c = 0.77, MeOH), which is comparable with reported val-
ues,10 allowing us to assign the absolute stereochemistry
of intermediates as shown in Scheme 4.
Precisely this type of effect has been reported previously,
for example in catalytic asymmetric desymmetrisation
processes involving coupling of prochiral bis-triflates or
dihalides.7,8
In order to test this idea we exposed racemic 5a to an ex-
cess of bis-lithiated base 8 and then alkylated with MeI to
generate 7a. When a 46% conversion into 7a was
achieved the remaining (–)-5a showed an ee of 13%.
Synlett 2002, No. 12, 2074–2076 ISSN 0936-5214 © Thieme Stuttgart · New York

2076
D. A. Greenhalgh, N. S. Simpkins
LETTER
1681 (C=O), 1606 (Ar), 1512 (Ar); H (400 MHz, CDCl3):
1.11 (3 H, d, J = 7 Hz, CH3), 2.72 (1 H, dq, J = 11 Hz, 7 Hz,
CHCH3), 2.78 [1 H, dd, J = 18 Hz, 13 Hz, C(O)CH2ax], 2.94
(1 H, m, CHAr), 2.94 [1 H, dd, J = 18 Hz, 4 Hz, C(O)CH2eq],
4.97 (1 H, d, J = 14 Hz, NCHHPh), 5.02 (1 H, d, J = 14 Hz,
NCHHPh), 7.02 (2 H, m), 7.11 (2 H, m), 7.29 (3 H, m) and
7.38 (2 H, m); C (125 MHz, CDCl3): 14.4 (CH3), 40.8
[C(O)CH2], 41.9 (CHAr), 43.4 (NCH2Ph), 43.5 (CHCH3),
116.1 (JC-F = 21.5 Hz, CH), 127.6 (CH), 128.5 (CH), 128.6
(CH), 128.9 (CH), 136.3 (C), 137.3 (C), 162.1 (d, JC-F = 246
Hz, C), 171.0 (C=O) and 174.5 (C=O); HRMS(EI) Found:
[M]+ 311.1334, C19H18NO2F requires 311.1322.
The results described above further extend the utility of
the chiral base method in the area of imide desymmetrisa-
tion. Further applications of this chemistry to the synthe-
sis of naturally occurring alkaloids are underway and will
be reported shortly.
Acknowledgement
We thank The University of Nottingham for support of D.A.G.
References
The enantiomeric excess was determined by HPLC analysis
using a Chiralcel OD column with 5% i-PrOH in hexane as
eluant, at a flow rate of 0.8 mL/min, using UV detection at
215 nm. Retention times were 49.8 min(minor) and 59.5
min(major).
(1) (a) Adams, D. J.; Simpkins, N. S.; Smith, T. J. N. Chem.
Commun. 1998, 1605. (b) For our full report of this work,
see: Adams, D. J.; Blake, A. J.; Cooke, P. A.; Gill, C. D.;
Simpkins, N. S. Tetrahedron 2002, 58, 4603; this issue of the
journal is dedicated entirely to chiral base chemistry and is
an excellent source of leading references in the area.
(2) For recent asymmetric syntheses of paroxetine, see: (a) de
Gonzalo, G.; Brieva, R.; Sanchez, V. M.; Bayod, M.; Gotor,
V. J. Org. Chem. 2001, 66, 8947. (b) Cossy, J.; Mirguet, O.;
Gomez Pardo, D.; Desmurs, J.-R. Tetrahedron Lett. 2001,
42, 7805. (c) Johnson, T. A.; Curtis, M. D.; Beak, P. J. Am.
Chem. Soc. 2001, 123, 1004. (d) Liu, L. T.; Hong, P.-C.;
Huang, H.-L.; Chen, S.-F.; Wang, C.-L. W.; Wen, Y.-S.
Tetrahedron: Asymmetry 2001, 12, 419. (e) Amat, M.;
Bosch, J.; Hidalgo, J.; Canto, M.; Perez, M.; Llor, N.;
Molins, E.; Miravitlles, C.; Orozco, M.; Luque, J. J. Org.
Chem. 2000, 65, 3074.
1-Benzyl-4-(4-fluorophenyl)-2,6-dioxo-piperidine-3-
carboxylic acid methyl ester 5i:
The reaction was carried out as described above, starting
with imide 4 (200 mg, 0.67 mmol) and quenching of the
intermediate enolate with excess methyl cyanoformate (0.11
mL). After 30 min at –78 °C the mixture was worked up as
described above to give a crude product which was purified
by flash column chromatography on silica gel (30% EtOAc
in petroleum ether followed by DCM) to give the product 5i
as a white solid (168 mg, 71%), mp 137–139 °C; [ ]D28 –31
(c = 0.74 in CHCl3); IR (CHCl3)/cm–1: 1749 (C=O), 1729
(C=O), 1680 (C=O), 1608 (Ar), 1512 (Ar); H (400 MHz,
CDCl3): 2.82 [1 H, dd, J = 17.5 Hz, 11, C(O)CH2ax], 3.02 [1
H, dd, J = 17.5 Hz, 4.5 Hz, C(O)CH2eq], 3.65 (3 H, s, OCH3),
3.68 (1 H, ddd, J = 11 Hz, 11 Hz, 4.5 Hz, CHAr), 3.81 (1 H,
d, J = 11 Hz, CHCO2Me), 4.96 (1 H, d, J = 14 Hz,
NCHHPh), 5.03 (1 H, d, J = 14, NCHHPh), 7.00 (2 H, m),
7.13 (3 H, m), 7.29 (2 H, m), 7.37 (2 H, m); C (125 MHz,
CDCl3): 37.5 (CHAr), 38.8 [C(O)CH2], 43.5 (NCH2Ph),
52.9 (OCH3), 56.4 (CHCO2Me), 116.2 (d, JC-F = 21.5 Hz,
CH), 127.8 (CH), 128.4 (CH), 128.6 (CH), 129.1 (CH),
134.4 (C), 136.5 (C), 163.5 (d, JC-F = 247 Hz, C), 168.1
(C=O), 168.3 (C=O), 170.1 (C=); HRMS(EI): Found: [M]+
355.1220, C20H18NO4F requires 355.1220.
(3) (a) Rama Rao, R.; Singh, A. K.; Varaprasad, C. V. N. S.
Tetrahedron Lett. 1991, 32, 4393. (b) Goehring, R. R.;
Greenwood, T. D.; Nwokogu, G. C.; Pisipati, J. S.; Rogers,
T. G.; Wolfe, J. F. J. Med. Chem. 1990, 33, 926.
(4) Bambridge, K.; Begley, M. J.; Simpkins, N. S. Tetrahedron
Lett. 1994, 35, 3391.
(5) 1-Benzyl-4-(4-fluorophenyl)-3-methyl-piperidine-2,6-dione
5e: A solution of the chiral bis-lithium amide base 8 was
prepared by dropwise addition of a solution of n-BuLi (1.0
mL, of a 1.6 M in hexanes, 1.60 mmol) to the chiral diamine
(342 mg, 0.81 mmol) in THF (4 mL) at –78 °C. The resulting
red coloured solution was warmed to room temperature for
20 min before cooling to –78 °C and addition via cannula, to
a stirred solution of the starting imide 4 (200 mg, 0.67 mmol)
in THF (10 mL) at ca. –78 °C (internal temperature). The
mixture was stirred for 45 min at this temperature before
being diluted with THF (14 mL). Excess methyl iodide (3
mL) was then added, the mixture warmed to –40 °C (internal
temperature) and then stirred at this temperature for 4 h. The
reaction mixture was quenched with saturated aqueous
NH4Cl (10 mL) and extracted into Et2O (3 20 mL). The
extracts were washed sequentially with 2 M HCl (3 60
mL), saturated aqueous NaHCO3 (60 mL) and brine (60
mL). The combined extracts were then dried (MgSO4) and
concentrated under reduced pressure to yield a crude product
which was purified by flash column chromatography on
silica gel (40% Et2O in petroleum ether) to give the product
The enantiomeric excess was determined by HPLC analysis
using a Chiralcel OD column with 3% EtOH in hexane as
eluant, at a flow rate of 0.8 mL/min, using UV detection at
215 nm. Retention times were 75.8 min(minor) and 88.5
min(major).
(6) The use of bis-lithiated base 8 may seem to invite the
formation of unwanted products 7 but we achieved much
poorer yields of desired compounds 5 by using the base in
mono-lithiated form.
(7) Kamikawa, T.; Hayashi, T. Tetrahedron 1999, 55, 3455.
(8) Gotov, B.; Schmalz, H.-G. Org. Lett. 2001, 3, 1753.
(9) Many syntheses of paroxetine use this approach, see for
example: Yu, M. S.; Lantos, I.; Peng, Z.-Q.; Yu, J.; Cacchio,
T. Tetrahedron Lett. 2000, 41, 5647; see also ref.2 and
references therein.
(10) The reported values for paroxetine include [ ]D20 –75.5 (c =
1.2, MeOH) for >97:3 er,2c and [ ]D22 –80.8 (c = 1.25,
MeOH) via enantiopure intermediates.2e
22
as an off white solid (128 mg, 64%), mp 115–118 °C; [ ]D
–21 (c = 1.02 in CHCl3); IR (CHCl3)/cm–1: 1726 (C=O),
Synlett 2002, No. 12, 2074–2076 ISSN 0936-5214 © Thieme Stuttgart · New York