A practical asymmetric synthesis of (R)-fluoxetine and its major metabolite (R)-norfluoxetine

Article abstract of DOI:10.1016/S0040-4039(01)01877-9

A convenient and chromatography-free synthesis for the enantiomers of fluoxetine and norfluoxetine is described. The synthesis relied on the use of the CBS reduction, and Hofman rearrangement to establish the key common intermediate 5, and enrichment of optical purity of the final product by crystallization as the tartrate salt.

Full text of DOI:10.1016/S0040-4039(01)01877-9

TETRAHEDRON
LETTERS
Pergamon
Tetrahedron Letters 42 (2001) 8919–8921
A practical asymmetric synthesis of (R)-fluoxetine and its major
metabolite (R)-norfluoxetine†
James W. Hilborn,b Zhi-Hui Lu,a,* Alex R. Jurgens,b Q. Kevin Fang,a Paul Byers,a
Stephen A. Walda and Chris H. Senanayakea,*
aSepracor, Inc., Chemical Process R&D, 111 Locke Dr., Marlborough, MA 01752, USA
bSepracor Canada Ltd., 24 Ivey Lane, Windsor, N.S. Canada B0N 2T0
Received 6 August 2001; accepted 24 September 2001
Abstract—A convenient and chromatography-free synthesis for the enantiomers of fluoxetine and norfluoxetine is described. The
synthesis relied on the use of the CBS reduction, and Hofman rearrangement to establish the key common intermediate 5, and
enrichment of optical purity of the final product by crystallization as the tartrate salt. © 2001 Elsevier Science Ltd. All rights
reserved.
Fluoxetine (Prozac®), marketed as a racemate, is a
potent and selective inhibitor of neural serotonin-reup-
take, and is used for the treatment of depression and
anxiety.1 Although both enantiomers are reportedly
equal in pharmacological activity, marked differences
have been observed between the two major metabolites
(R)- and (S)-norfluoxetine in regard to their inhibition
of 5-HT re-uptake.2 (R)-Norfluoxetine is 10- to 20-fold
less active than the (S)-enantiomer. Due to the lengthy
half-life of (S)-norfluoxetine (16–19 days), pure (R)-
fluoxetine is expected to reduce the time for elimination
of the functional drug from the body.3 In view of their
pharmaceutical potential, it is not surprising that much
effort has been expanded in devising asymmetric syn-
theses of both enantiomers of fluoxetine and its major
active metabolites.4–19 Most of these approaches start
with a three-carbon-chain segment and establish the
chirality by enzymatic resolution,4–12 asymmetric reduc-
tion,13–16 asymmetric epoxidation,17,18 or resolution.19
Among the most effective route is Corey’s procedure
for the preparation of the enantiomers of fluoxetine,
which involves the asymmetric reduction of 3-chloro-
propiophenone,13 and this method might be adaptable
to the synthesis of (R)-norfluoxetine. We felt that it
might offer certain advantages over previous method-
ologies, if the chiral aminoalcohol I could be derived
from readily accessible four-carbon-chain ketoamide II
by incorporating asymmetric reduction and Hofman
rearrangement (Scheme 1). Such a strategy could cir-
cumvent the enolization problem frequently associated
with previous 3-carbon strategies in the reduction of
b-ketoesters, and the amide functionality of 4-carbon
fragment II could be envisaged as a protected amine
equivalent, which can be directly manipulated to either
(R)-fluoxetine or (R)-norfluoxetine. Herein, we describe
the first entry utilizing inexpensive methyl benzylpropi-
CF3
Asymm.
Reduction;
O
O
OH
NH2
NHR
[N]
O
Hofman
Rearrangement
II
I
Fluoxetine
R = Me
R = H
Norfluoxetine
Scheme 1.
Keywords: asymmetric synthesis; fluoxetine; norfluoxetine.
* Corresponding authors. Tel.: +1-508-357-7575; fax: +1-508-357-7808; e-mail: zlu@sepracor.com; csenanay@sepracor.com
† This paper is dedicated to Professor Carl R. Johnson on the occasion of his retirement.
0040-4039/01/$ - see front matter © 2001 Elsevier Science Ltd. All rights reserved.
PII: S0040-4039(01)01877-9

8920
J. W. Hilborn et al. / Tetrahedron Letters 42 (2001) 8919–8921
onate to synthesize (R)-fluoxetine and norfluoxetine via
optically pure strategic cyclic carbamate 5.
As an alternative method, asymmetric reduction of 1
with (+)-Ipc2BCl23 directly afforded the lactone 2 in
excellent enantiomeric excess (99% e.e.). Although (+)-
Ipc2BCl is expensive, and operationally inconvenient on
large-scale, the reagent could be generated in situ by
treatment of a-pinene (87% e.e.) with NaBH4, boron
trichloride in DME.24 The reagent generated by such a
process could effectively reduce the ketoester 1 and
allow formation of lactone 3 in 97% e.e. Aminolysis of
either hydroxy ester 2 or lactone 3 by treatment with
ammonium hydroxide in MeOH at 40°C or room tem-
perature afforded (R)-4-hydroxy-4-phenylbutyramide
(4) in 84 or 80% yield, respectively (Scheme 2).25
Our approach began with the conversion of methyl
3-benzoylpropionate
(1)
into
(R)-4-hydroxy-4-
phenylbutyramide (4) in two steps (Scheme 2). The
first key-step involved the asymmetric reduction of
ketoester 1 to afford methyl 4-(R)-hydroxy-4-phenyl-
butyrate (2), or (R)-x-phenyl-x-butyrolactone (3). Two
approaches have been reported previously for this
purpose, by using oxazaborolidine–borane reduction,20
and BINAP–Ru(II)-catalyzed hydrogenation,21 res-
pectively. The latter procedure involved a rigorous
reaction condition (1500 psi, 35°C) and was inefficient
(258 h, 30% conversion). Therefore, the former method
using Me–CBS reagent was examined. It was found
that excellent enantioselectivity (96% e.e.) and high
yield (95%) were achieved when the reaction was per-
formed by simultaneous slow addition of borane and
ketoester 1 to the solution of (S)-Me–CBS catalyst in
THF.22
To complete the synthesis of (R)-fluoxetine and (R)-
norfluoxetine, amide 4 was converted into optically
pure cyclic carbamate 5 in 83% yield by Hofman rear-
rangement using 1 equiv. of iodobenzene diacetate in
acetonitrile at 40°C, followed by recrystallization in
AcOEt (Scheme 3). Hydrolysis of cyclic carbamate 5 by
refluxing with KOH in aqueous isopropyl alcohol for
(+)-Ipc2BCl, THF, 22 °C, 48 h
or
(-)-α-Pinene, NaBH4, BCl3
DME, 0-22 °C
O
O
(70%)
3
(97-99% ee)
NH3·H2O MeOH
22 °C, 4 h
(84%)
O
OH
OH
O
N B
NH3·H2O
NH2
OMe
OMe
MeOH
40 °C, 3 h
O
BH3·THF
O
O
THF, 0 °C, 1 h
2
1
4
(80%)
(95%)
(96% ee)
Scheme 2.
CF3
CF3
O
O
O
NH
OH
Cl
O
HO
PhI(OAc)2
OH
OH
KOH
NaH, DMSO
·
NH2
4
NH2
HO
ACN, 40 °C
iPrOH-H2O
Reflux, 3 h
(95%)
90 °C, 5 h;
5 h
5
O
6
Recrystallization
in AcOEt
(>83%)
TA, MeOH
(59%)
(>99% ee)
(R)-Norfluoxetine·(D)-TA
LiAlH4
THF, reflux
OH
CF3
O
Ref.
NHMe
NHMe
8
(R)-Fluoxetine
(90% )
Scheme 3.

J. W. Hilborn et al. / Tetrahedron Letters 42 (2001) 8919–8921
8921
3 h afforded aminoalcohol 6 in 84% yield.26 It is worth
noting that isopropyl alcohol was determined to be the
proper solvent for efficient extraction of this highly
polar product from the aqueous phase. Contrary to
intuition, isopropanol is not miscible with aqueous
phase when the pH is adjusted to above 10, and has a
clear phase-cut for easy separation. Arylation of the
aminoalcohol 6 was carried out according to literature
procedure by treatment with sodium hydride, and 4-
chlorobenzotrifluoride in DMSO at 90°C.19 Isolation of
the desired product in the form of a pharmacologically
useful salt from the crude mixture without using chro-
matography is crucial to the practicability of this pro-
cess. Although formation of the HCl salt of
norfluoxetine has been reported,27 the stability of the
HCl salt was poor. We found that (R)-norfluoxetine
7. Chenevert, R.; Fortier, G.; Rhlid, R. B. Tetrahedron
1992, 48, 6769.
8. Schneider, M. P.; Goergens, U. Tetrahedron: Asymmetry
1992, 3, 525.
9. Kumar, A.; Ner, D. H.; Dike, S. Y. Indian J. Chem. 1992,
31B, 803.
10. Kumar, A.; Ner, D. H.; Dike, S. Y. Tetrahedron Lett.
1991, 32, 1901.
11. Chenevert, R.; Fortier, G. Chem. Lett. 1991, 1603.
12. Fronza, G.; Fuganti, C.; Grasselli, P.; Mele, A. J. Org.
Chem. 1991, 56, 6019.
13. Corey, E. J.; Reichard, G. A. Tetrahedron Lett. 1989, 30,
5207.
14. Sakuraba, S.; Achiwa, K. Synlett 1991, 689.
15. Devocelle, M.; Agbossou, F.; Mortreux, A. Synlett 1997,
1306.
was readily isolated as the
of the crude product with
Furthermore, formation of the
D
-tartrate salt, by treatment
-tartaric acid in methanol.
16. Srebnik, M.; Ramachandran, P. V.; Brown, H. C. J. Org.
D
Chem. 1988, 53, 2916.
D
-tartrate salt allowed
17. Gao, Y.; Sharpless, K. B. J. Org. Chem. 1988, 53, 4081.
18. Mitchell, D.; Koenig, T. Synth. Commun. 1995, 25, 1231.
19. Koenig, T.; Mitchell, D. Tetrahedron Lett. 1994, 35, 1339.
20. 94% e.e. was reported by Corey, see: (a) Corey, E. J.;
Bakshi, R. K.; Shibata, S.; Chen, C.-P.; Singh, V. K. J.
Am. Chem. Soc. 1987, 109, 7925; (b) Corey, E. J.; Link, J.
O. Tetrahedron Lett. 1989, 30, 6275.
for easy enrichment of the optical purity of the final
product by recrystallization in methanol, in case the
precursor is not optically pure. By this procedure,
(R)-norfluoxetine·D-tartrate was obtained as a white
solid in 59% overall yield for two steps, and in 99.6%
e.e., as well as 99.2% chemical purity.
21. Ohkuma, T.; Kitamura, M.; Noyori, R. Tetrahedron Lett.
On the other hand, reduction of the key optically pure
carbamate 5 with lithium aluminum hydride in THF,
afforded N-methyl aminoalcohol 8 in 90% yield, which
could be converted into the corresponding (R)-fluox-
etine by the same method mentioned above.19
1990, 31, 5509.
22. No reduction of the carboxyl function by borane was
observed at 0°C under these conditions.
Experimental procedure: To a dry 2.5 L round bottom
flask were added (S)-Me–CBS (50 mL, 1.0 M in toluene,
50 mmol) and 500 mL of THF under argon. The solution
was cooled to −5°C. To this solution were added simulta-
neously BH3·THF (500 mL, 1.0 M in THF, 500 mmol)
and ketoester 1 in neat (96 g, 500 mmol) via syringe
pump. The reaction mixture was stirred at the same
temperature for another 30 min and worked up by addi-
tion of 2N K2CO3 aqueous solution. The organic layer
was separated, and the aqueous layer was extracted with
AcOEt. The combined organic layers were washed with
brine and dried (Na2SO4). Removal of the solvents gave
a pale yellow oil containing (R)-2 and the catalyst. The
crude product was used for next reaction without further
purification. Solution yield of (R)-2 was determined by
HPLC (>95%). E.e. was measured by HPLC on lactone
(R)-3 derived from (R)-2 [Column: Chiralcel AS; Mobil
phase: hexane–isopropanol (90:10); wavelength: 254 nm;
rate: 1.0 mL/min].
In conclusion, an economical and operationally simple
process for the asymmetric synthesis of (R)-fluoxetine
and (R)-norfluoxetine·tartrate has been developed via
novel optically pure cyclic carbamate 5. The process
uses low-cost raw materials and conventional reagents,
thereby providing (R)-fluoxetine and (R)-norfluox-
etine·tartrate with >99% chemical purity and >99% e.e.
without resorting to chromatography or distillation.
This novel strategy is applicable to other medicinally
important chiral 1,3-aminoalcohol building blocks.
References
1. Murphy, D. L.; Mueller, E. A.; Garrick, N. A.; Aulakh,
C. S. J. Clin. Psychiatry 1986, 47, (4, Suppl.), 9; DE 2 500
110; US 4 018 895.
23. Chandrasekharan, J.; Ramachandran, P. V.; Brown, H.
C. J. Am. Chem. Soc. 1988, 110, 1539.
2. (a) Robertson, D. W.; Jones, N. D.; Swrtzendruber, J.
K.; Yang, K. S.; Wang, D. T. J. Med. Chem. 1988, 31,
185; (b) Robertson, D. W.; Krushinski, J. H.; Fuller, R.
W.; Leander, J. D. J. Med. Chem. 1988, 31, 1412.
3. Sepracor’s internal documents.
24. Zhao, M.; King, A. O.; Larsen, R. D.; Verhoeven, T. R.;
Reider, P. J. Tetrahedron Lett. 1997, 38, 2641.
25. Reaction took over 4 days to complete with 7 M NH3 in
MeOH, see: Barton, D. H. R. J. Chem. Soc., Chem.
Commun. 1965, 181.
4. Quiros, M.; Rebolledo, F.; Liz, R.; Gotor, V. Tetra-
26. Aqueous KOH solution is much more effective than
sodium methoxide in hydrolysis of 5. No epimerization
occurred under these conditions.
27. Fuller, R. W.; Robertson, D. W.; Wong, D. T., EU, 0369
685 B1.
hedron: Asymmetry 1997, 8, 3035.
5. Master, H. E.; Newadkar, R. V.; Rane, R. A.; Kumar, A.
Tetrahedron Lett. 1996, 37, 9253.
6. Bracher, F.; Litz, T. Bioorg. Med. Chem. 1996, 4, 877.