Enantioselective synthesis of β-hydroxy amines and aziridines using asymmetric transfer hydrogenation of α-amido ketones

Article abstract of DOI:10.1016/S0957-4166(00)00310-4

A rapid, expedient and enantioselective method for the synthesis of β-hydroxy amines and monosubstituted aziridines in up to 99% e.e., via asymmetric transfer hydrogenation of α-amido ketones, is described. Copyright (C) 2000 Elsevier Science Ltd.

Full text of DOI:10.1016/S0957-4166(00)00310-4

TETRAHEDRON:
ASYMMETRY
Pergamon
Tetrahedron: Asymmetry 11 (2000) 3257–3261
Enantioselective synthesis of b-hydroxy amines and
aziridines using asymmetric transfer hydrogenation of
a-amido ketones
Aparecida Kawamoto† and Martin Wills*
Department of Chemistry, University of Warwick, Coventry, CV4 7AL, UK
Received 12 July 2000; accepted 8 August 2000
Abstract
A rapid, expedient and enantioselective method for the synthesis of b-hydroxy amines and monosubsti-
tuted aziridines in up to 99% e.e., via asymmetric transfer hydrogenation of a-amido ketones, is described.
© 2000 Elsevier Science Ltd. All rights reserved.
Aziridines are valuable synthetic reagents and intermediates. In particular they benefit from a
high reactivity due to the small strained nitrogen-containing ring, thus permitting their rapid
conversion into a range of derivatives. The synthesis of enantiomerically pure derivatives is a
desirable objective since the physiological properties of both the aziridines themselves and the
products formed from them are likely to be dependant on the absolute configuration.1
Enantiomerically enriched aziridines may be formed by the asymmetric catalysis of the addition
of a nitrene onto one face of a prochiral alkene.2 However, the most conceptually simple approach
to these targets is probably through the cyclisation of an appropriate enantiomerically pure
b-amino alcohol precursor.1 Whilst a number of homochiral b-amino alcohols are available from
natural sources such as amino acids, the range of materials is limited. In this paper we describe
a simple and expedient route for the asymmetric synthesis of aziridines through a sequence of
a-amido ketone reduction followed by cyclisation.
We have previously described the use of asymmetric transfer hydrogenation for the reduction
of ketones containing a-amido groups (Scheme 1).3 In the example shown we first employed a
catalyst formed by the combination of (1R,2S)-cis-aminoindanol 1 with [Ru(cymene)Cl2]2 and
isopropanol/KOH as the hydrogen source, which gave a product of 79% e.e. In subsequent studies
we found that the combination of (R,R)-TsDPEN 2 with the same ruthenium complex, a system
first reported by Noyori, gave marginally superior results (Scheme 1) when used with the formic
acid/triethylamine hydrogen source.3e Since the enantioselectivities of the reductions were high,
we considered that this might be a suitable method for the asymmetric synthesis of aziridines.
* Corresponding author.
†
On leave from the Brazilian Space Aeronautical Institute-Praca Mal. Eduardo Gomes 50, Vila das Acacias, SJ
Campos, SP Brazil.
0957-4166/00/$ - see front matter © 2000 Elsevier Science Ltd. All rights reserved.
PII: S0957-4166(00)00310-4

3258
A. Kawamoto, M. Wills / Tetrahedron: Asymmetry 11 (2000) 3257–3261
Scheme 1.
Indeed this proved to be the case. A sample of the (N-benzyl)(N-tBoc)-a-amino ketone 3 was
prepared by the route shown in Scheme 2. A one-pot procedure was favoured for this synthesis
as the intermediate N-benzyl-a-amino ketone appeared to be unstable under the conditions used
for working up the reaction. Enantioselective reduction of 3 to the alcohol 4 was successfully
achieved using both the (1R,2S)-1/isopropanol and (R,R)-2/formic acid systems. In both cases
the enantiomeric excesses were measured by chiral HPLC, whilst the absolute configuration in
each case was shown to be (S)-(+) by conversion to the O-tert-butyldimethylsilyl ether 5.4 The
independent synthesis of (S)-(+)-5 is described in a later discussion.
Scheme 2.
Deprotection of the nitrogen atom in (S)-(+)-4 gave the amino alcohol (S)-(+)-6, which was
the subject of cyclisation studies. Attempts to achieve this by O-tosylation followed by base
treatment failed;5 however, the use of Mitsunobu conditions resulted in cyclisation in low yield
but high enantiomeric excess to the aziridine (R)-(−)-7.6 Cyclisation of the racemic analogue of

A. Kawamoto, M. Wills / Tetrahedron: Asymmetry 11 (2000) 3257–3261
3259
6 was achieved in similar yield (27%), whilst N-benzyl-2-aminoethanol could be converted to the
corresponding aziridine in only 17% yield.
In an attempt to identify an improved route, we studied the asymmetric reduction of the
N-tBoc–a-aminoketone 8. Our studies revealed that the monotosylated diamine system was
highly efficient at this reduction, furnishing alcohol (S)-(+)-9 in 99% e.e. as measured by chiral
HPLC. In contrast the cis-1/Ru(II)/isopropanol system totally failed in this application (Scheme
3). The reasons for this dramatic difference are not fully clear, however we have previously
speculated that chelating reduction products (as might be obtained from the reduction of 8) may
inhibit, and lead to subsequent decomposition of, the amino alcohol/Ru(II) complex.3 In
contrast the monotosylated diamine may well form much stronger complexes with the same
metal, and be resistant to such chelation-initiated decomposition.
Scheme 3.
The configuration of (S)-(+)-9 was confirmed by deprotection of the nitrogen atom with TFA
to give the known amino alcohol (S)-(+)-10.7 Conversion of the same sample of (S)-(+)-9 to
(S)-(+)-5 served to confirm the absolute sense of the product of reduction of 3 as described
previously (Scheme 2).
The alcohol (S)-(+)-9 was cyclised to the N-tBoc aziridine (R)-(−)-11 in 83% yield and 99%
e.e. through treatment with tosyl chloride and base,5 thus delivering an efficient synthesis of
aziridines in high yield and enantioselectivity.8
In view of the importance of 2-hydroxy-3-phenoxy-propylamine derivatives, such as bisprolol
12,9 as b-selective adrenoceptor blocking agents (b-blockers), we chose to examine the synthetic
approach to such targets through the asymmetric transfer hydrogenation of the corresponding
ketone precursor 13.10 In the event, the use of the (R,R)-1/formic acid system gave alcohol 14
in 82% e.e. and 50% yield, although the configuration remains to be determined (we have
assumed that it follows the pattern of other reductions). We believe that this reduction
represents a competitive, and highly practical, new approach for the reduction of this class of
ketones, which are normally regarded as ‘difficult’ substrates due to the lack of steric differenti-
ation between the groups flanking the CꢀO group. The results of our ongoing studies in this area
will be reported in due course (Scheme 4).

3260
A. Kawamoto, M. Wills / Tetrahedron: Asymmetry 11 (2000) 3257–3261
Scheme 4.
In conclusion, we have demonstrated that the use of a monotosylated diamine/formic
acid/triethylamine system is highly effective at the enantioselective reduction of a-amido ketones
and that this process provides an efficient method for the asymmetric synthesis of b-amino
alcohol and azidiridines.
Acknowledgements
We thank the Brazilian CNPq for a Scholarship to A.K. and Professor D. Games, Dr. J.
Ballantine and Dr. B. Stein of the EPSRC Mass Spectrometry Service at Swansea for carrying
out analyses of certain compounds.
References
1. (a) For a comprehensive survey of methods of aziridine synthesis from double bonds, including a discussion of
cyclisations of b-amino alcohols, see; Kemp, J. E. G. In Comprehensive Organic Synthesis; Academic Press: New
York, 1991; Vol. 7, Chapter 3.5, pp. 470–483. (b) Atkinson, R. S.; Kelly, B. J. J. Chem. Soc., Chem. Commun.
1987, 1362.
2. (a) Evans, D. A.; Faul, M. M.; Bilodeau, M. T. J. Org. Chem. 1991, 113, 6744. (b) Li, Z.; Conser, K. R.;
Jacobsen, E. N. J. Am. Chem. Soc. 1993, 115, 5326. (c) Evans, D. A.; Faul, M. M.; Bilodeau, M. T.; Anderson,
B. A.; Barnes, D. J. Am. Chem. Soc. 1993, 115, 5328.
3. (a) Kenny, J. A.; Palmer, M. J.; Smith, A. R. C.; Walsgrove, T.; Wills, M. Synlett 1999, 1615. (b) Palmer, M.
J.; Walsgrove, T.; Wills, M. J. Org. Chem. 1997, 62, 5226. (c) Wills, M.; Palmer, M. J.; Smith, A. R. C.; Kenny,
J. A.; Walsgrove, T. Molecules 2000, 5, 1–15. (d) Kenny, J. A.; Versluis, K.; Heck, A. J. R.; Walsgrove, T.; Wills,
M. Chem. Commun. 2000, 99–100. (e) Kenny, J. A.; Walsgrove, T.; Wills, M., unpublished result.
4. All compounds gave appropriate physical and spectroscopic data. The optical rotations and chiral HPLC
retention data (OD column 250×4 mm, 0.5 mL/min, retention times in minutes unless otherwise indicated) are as
follows; (S)-(+)-4 (reduction using 2); [h]2D0=+3.3 (c=2, EtOH), HPLC (hexane:ethanol:diethylamine 95:4.9:0.1)
13.93 (R), 15.27 (S). (S)-(+)-5; (product from 9) [h]D20=+52.3 (c=2, EtOH), HPLC (hexane:ethanol:diethylamine
97:2.9:0.1) 9.45 (R), 8.63 (S). (S)-(+)-6; [h]2D0=+33.8 (c=2, EtOH), HPLC (hexane:ethanol:diethylamine
95:4.9:0.1) 33.19 (R), 31.97 (S). (R)-(−)-7; [h]2D0=−49.4 (c=2, EtOH), e.e. determined using NMR shift reagent
Eu(hfc)3. (S)-(+)-9; [h]2D0=+3.5 (c=1, EtOH), HPLC (hexane:ethanol:diethylamine 95:4.9:0.1) 19.49 (R), 17.86
(S). (R)-(−)-11; [h]2D0=−137.3 (c=2, EtOH), e.e. determined using NMR shift reagent Eu(hfc)3. (R)-(+)-14;
[h]2D0=+3.5 (c=1, EtOH), HPLC (hexane:ethanol:diethylamine 95:4.9:0.1) 41.62 (R), 22.48 (S).

A. Kawamoto, M. Wills / Tetrahedron: Asymmetry 11 (2000) 3257–3261
3261
5. (a) Wessig, P.; Schwarz, J. Synlett 1997, 893. (b) Effenberger, F.; Stelzer, U. Tetrahedron: Asymmetry 1995, 6,
283.
6. (a) General reference: Mitsunobu, O. In Comprehensive Organic Synthesis; Academic Press: New York, 1991;
Vol. 6, Chapter 1.3, pp. 93–98. (b) Specific to epoxides: Pfister, J. R. Synthesis 1984, 969.
7. There appears to be a discrepancy in the signs of optical rotation for compound 11. Previous work, and the
conversion of 9 to (S)-(+)-10, suggests that 11 should be of (R) configuration (i.e. through inversion). However,
our specific rotation for 11 ([h]2D0=−137.3 (c=1, EtOH)) suggests that it is in fact (S) when compared to literature
results ([h]2D0=−163.5 (c=1, CH2Cl2) for the (S) compound; Ref. 6a). A sample of (R)-(−)-11 was also prepared
from a commercial sample of (S)-(+)-10 via t-Boc protection of the N atom followed by cyclisation with
TsCl/KOH. This material exhibited [h]2D0=−132.8 (c=1, EtOH), [h]2D0=−142.6 (c=1, CH2Cl2).
8. The reduction appears to be quite versatile in terms of substrate scope. In a preliminary investigation the
asymmetric reduction of the challenging substrate 3-tert-butyloxycarbonylaminoacetone worked smoothly with
the (R,R)-2 system to give an alcohol in 87% yield ([h]2D0=+27.5 (c=1, DCM)). Although the e.e. of the product
remains to be determined, cyclisation to the N-Boc aziridine using the conditions given in Scheme 3 gave a
product of [h]2D0=−54.5 (c=1, EtOH) and [h]2D0=−42.2 (c=1, DCM), which compares favourably to the literature
value for this material, i.e. [h]2D0=+39.2 (c=1, DCM) for the (S) enantiomer (Ref. 6a). In this case, the expected
aziridine configuration, (R), matches that found by comparison with the literature.
9. Kitaori, K.; Furukawa, Y.; Yoshimoto, H.; Otera, J. Tetrahedron Lett. 1998, 39, 3173.
10. Ketone 13 was conveniently prepared by the epoxidation of N-tBoc-allylamine, followed by ring-opening with
sodium phenoxide and oxidation of the alcohol (80% overall yield). An alternative method, avoiding an alcohol
oxidation step but resulting in a slightly lower yield, involves sequential reaction of 2-bromomethylallylbromide
with phenol and sodium azide, reduction of azide and amine protection, followed by ozonolysis of the alkene.
.