An unusual oxidative rearrangement of azabicyclo[2.2.1]heptenes, providing a stereoselective route to 2'- and 3'-hydroxycyclopentylglycines

Article abstract of DOI:10.1039/b006683o

Treatment of the azabicyclo[2.2.1]heptene derivative 4 with mCPBA for 3-5 seconds generates the oxazabicyclo[3.2.1]octene derivative 7 (X-ray structure) via rapid Meisenheimer rearrangement of the N-oxide 5, whilst heating 7 in MeCN for 4-6 h leads to fur

Full text of DOI:10.1039/b006683o

An unusual oxidative rearrangement of azabicyclo[2.2.1]heptenes, providing a
stereoselective route to 2A- and 3A-hydroxycyclopentylglycines
Patrick D. Bailey,*a I. M. McDonald,*b Georgina M. Rosaira and David Taylora
a Department of Chemistry, Heriot-Watt University, Riccarton, Edinburgh, UK EH14 4AS.
E-mail: P.D.Bailey@hw.ac.uk
b James Black Foundation, 68 Half Moon Lane, Dulwich, London, UK SE24 9JE
Received (in Liverpool, UK) 11th August 2000, Accepted 30th October 2000
First published as an Advance Article on the web
Treatment of the azabicyclo[2.2.1]heptene derivative 4 with
mCPBA for 3–5 seconds generates the oxazabicyclo-
[3.2.1]octene derivative 7 (X-ray structure) via rapid Mei-
senheimer rearrangement of the N-oxide 5, whilst heating 7
in MeCN for 4–6 h leads to further rearrangement to the
more thermodynamically stable oxazabicyclo[3.3.0]octene
isomer 6; 6 and 7 can be readily reduced to enantio-pure
hydroxylated cyclopentylglycine derivatives.
Pharmacologically active peptoids can be generated by attach-
ing the side-chains of a parent peptide to carrier moieties such
as sugars, steroids, or porphyrins.1 We have been exploring the
use of azabicyclo[2.2.1]heptanes as rigid templates for such
systems which, if they can be derivatised appropriately, offer
potential as 3D structures on which to construct peptoid
libraries. We were also attracted to such systems because we
had developed an extremely short, efficient, and highly
stereoselective route to these systems some years ago;2 the aza-
Diels–Alder reaction of the imine 3 with cyclopentadiene gives
the azabicyclo[2.2.1]heptene 4 as a single stereoisomer, and we
wished to explore the derivatisation of this adduct.3 This
communication, however, reports some unexpected results that
have instead provided a short, efficient, stereo- and regio-
specific route to hydroxylated cyclopentylglycines.
As outlined in Scheme 1, the adduct 4 could be efficiently
prepared using ethyl glyoxylate generated using the procedure
developed by Roberts’ group.4 After formation of the imine, the
Scheme 1 Formation of the oxazabicyclo[3.3.0]octane system 6. The aza-
cycloaddition reaction was found to be most reliable when
conducted in trifluoroethanol in the presence of TFA (1 equiv.).
As reported by ourselves4 and others,5 this reaction proceeds
both with excellent exo diastereo-control (in contrast to the endo
selectivity observed for acyclic dienes2), and with extremely
high asymmetric induction resulting from the presence of the a-
methylbenzyl auxiliary (72% isolated yield, 95+5 exo+endo,
90% asymmetric induction). After chromatography, 4 was
obtained as a single stereoisomer, from which we hoped an
epoxidation/nucleophilic ring-opening sequence would allow
derivatisation at site C, whilst further pharmacophores could be
introduced at sites A and B. However, treatment with mCPBA
failed to give the desired epoxide (even under acidic conditions
whereby the nitrogen would be expected to be protonated), but
instead yielded an isomeric oxygenated product which we
initially assigned as the N-oxide 5. Small amounts of a by-
product were also generated, which was identified as the
oxazabicyclo[3.3.0]octene 6 from NMR data.†
Diels–Alder reaction (3 ? 4) was in TFE, TFA (1 eq.), 0 °C, 3.5 h (72%
yield, 95% exo, 90% asymmetric induction). See text and Scheme 2 for
yields and identification of the products 5 and 6 from mCPBA oxidation of
4. [R* = (S)-1-phenylethyl].
We had expected that an X-ray crystal structure of the ‘N-
oxide’ would confirm its identity and allow us to assign its
stereochemistry at the chiral nitrogen, but this surprisingly
revealed that the isomeric oxazabicyclo[3.2.1]octene 7 had been
formed instead (Fig. 1‡). Optimisation of the reaction condi-
tions established that the highest yield of 7 was obtained when
the cyclo-adduct 4 was reacted at room temperature with
mCPBA for only 3–5 seconds and, that clean conversion to 6
could be achieved by refluxing 7 in acetonitrile for 24 h.
Presumably 7 is formed via a Meisenheimer rearrangement,6–9
but the rapidity and selectivity of the initial reaction, in which
Fig. 1 X-Ray crystal structure of 7 with crystallographic numbering.‡
DOI: 10.1039/b006683o
Chem. Commun., 2000, 2451–2452
This journal is © The Royal Society of Chemistry 2000
2451

6, which can be converted into the [3.3.0] system 7 by simple
heating. These two isomers can be separately reduced, to
provide access to a wide range of cyclopentylglycine deriva-
tives, with complete regio-, diastereo- and enantio-control.
We thank Dr A. S. F. Boyd and Dr R. Fergusson for NMR and
mass spectra, the EPSRC for financial support, and we
acknowledge the use of the EPSRC’s Chemical Database
Service at Daresbury.
Notes and references
† Selected data for 6: C17H21NO3; dH(CDCl3) 1.1 (3H, t, J 7.2), 1.5 (3H, d,
J 6.6), 2.4 (2H, m), 3.05 (1H, q, J 6.6), 3.55 (1H, m), 3.7 (1H, d, J 9.1), 3.95
(2H, q, J 7.2), 5.3 (1H, dd, J 7.4, 0.7), 5.8 (1H, dd, J 3.5, 1.7), 5.95 (1H, dd,
J 1.7, 0.7), 7.2–7.4 (5H, m); dC(CDCl3) 14.15 (CH3), 20.47 (CH3), 34.16
(CH2), 47.20 (CH), 60.52 (CH2), 65.68 (CH), 69.85 (CH), 87.35 (CH),
127.66 (CH), 128.25 (2 3 CH), 128.30 (2 3 CH), 129.49 (CH), 135.17
(CH), 141.91 (C), 169.90 (C).
‡ Crystal data for 7. A single crystal of 7 was grown by slow evaporation
of 40–60 pet. ether at rt and coated in Nujol and mounted on a glass fibre
covered with vacuum grease for data collection with a Bruker P4
diffractometer14 at 160 K. Colourless plate, 0.12 3 0.72 3 0.34 mm,
C17H21NO3, M = 287.35, monoclinic, space group P21, a = 6.0734 (4), b
= 13.0445 (9), c = 10.1045(7) Å , b = 98.280(5)°, U = 792.18(13) Å3, Z
= 2, m(Mo–Ka) = 0.082 mm21. Data measured = 3180, unique data =
1593, R = 0.0778 on all data (wR2 = 0.0306). Absolute structure parameter
= 3.5(19), so the absolute structure could not be confirmed from the X-ray
data. Crystallographic computing was performed using the SHELXTL15
system version 5.1. CCDC 182/1836. See http://www.rsc.org/suppdata/b0/
b006683o/ for crystallographic data in .cif format.
Scheme 2 Possible mechanism for the formation of 6 and 7; heterolytic
bond cleavage of 7 would also be consistent with its thermolysis to 6. [R*
= (S)-1-phenylethyl].
1 R. Hirschmann, Angew. Chem., Int. Ed. Engl., 1991, 30, 1278.
2 P. D. Bailey, R. D. Wilson and G. R. Brown, J. Chem. Soc., Perkin
Trans. 1, 1991, 1337; P. D. Bailey, G. R. Brown, F. Korber, A. Reid and
R. D. Wilson, Tetrahedron: Asymmetry, 1991, 2, 1263.
3 Andersson’s group has made extensive use of the azabicyclo-
[2.2.1]heptene 4, as exemplified by P. Pinho and P. G. Andersson,
Chem. Commun., 1999, 597, and reference 5 therein; the Malpass group
has also developed the chemistry of these azabicyclic systems, as
exemplified by J. R. Malpass and C. D. Cox, Tetrahedron Lett., 1999,
40, 1419.
4 M. B. Hursthouse, K. M. A. Malik, D. E. Hibbs, S. M. Roberts, A. J. H.
Seago, V. Sik and R. Storer, J. Chem. Soc., Perkin Trans 1, 1995,
2419.
5 L. Stella, H. Abraham, J. Feneu-Dupont, B. Tinant and J. P. Declercq,
Tetrahedron Lett., 1990, 31, 2603; H. Abraham and L. Stella,
Tetrahedron, 1992, 48, 9707.
6 For a useful discussion of the Meisenheimer rearrangement, see V.
Rautenstrauch, Helv. Chim. Acta, 1973, 56, 2492, and references 8–18
therein.
7 For a study of the 2,3-Meisenheimer rearrangement of allyl N-oxides,
see Y. Yamamoto, J. Oda and Y. Inouye, J. Org. Chem., 1976, 41, 303,
and references therein; it might be noted that this process is believed to
be pericyclic rather than radical, and that 6 is (formally) the product of
the 2,3-Meisenheimer rearrangement of 5.
8 For the synthesis of 7-, 8- and 9-membered 1,2-oxaza ring systems via
the Meisenheimer rearrangement, see J. B. Bremner, E. J. Browne, P. E.
Davies and L. van Thuc, Aust. J. Chem., 1980, 33, 833; J. B. Bremner
E. J. Browne, P. E. Davies, C. L. Raston and A. H. White, Aust. J.
Chem., 1980, 33, 1323; J. B. Bremner E. J. Browne and P. E. Davies,
Aust. J. Chem., 1980, 33, 1335.
9 Spontaneous Meisenheimer rearrangements are rare, but has been
observed in the ring expansion of a highly strained azetidine N-oxide: T.
Kurihara, Y. Sakamoto, K. Tsukamoto, H. Ohishi, S. Harusawa and R.
Yoneda, J. Chem. Soc., Perkin Trans 1, 1993, 81.
10 O. Mitsunobu, Synthesis, 1981, 1.
11 N. Katagiri, M. Okada, Y. Morishita and C. Kaneko, Chem. Commun.,
1996, 2137.
Scheme 3 Conversion of the oxazabicyclooctanes 6 and 7 into hydroxylated
cyclopentylglycines such as 9 and 10. Conditions: (i) Zn–AcOH; (ii) H2 (50
psi)– Pd(OH)2/C; (iii) MeCN, reflux, 24 h. [R* = (S)-1-phenylethyl].
the di-radical must be retained within a solvent cage, is
extraordinary. Unprecedented, as far as we are aware, is the
establishment of an equilibrium with the Meisenheimer inter-
mediate (or possibly via heterolytic bond cleavage of 7),
providing a route by which the thermodynamically more stable
adduct 6 can be generated (see Scheme 2).
With an efficient route to two isomeric oxazabicyclooctanes
achieved, we wished to explore whether they could be
converted into hydroxy-substituted cyclopentylglycines. The
key transformations are shown in Scheme 3. In particular, Zn–
AcOH treatment of the [3.2.1] isomer 7 allowed selective
cleavage of the N–O bond, yielding the 4A-hydroxycyclopent-2A-
enyl glycine derivative 8. Hydrogenation of the double bond
(Pearlman’s catalyst, 50 psi) resulted in concomitant removal of
the a-methylbenzyl auxiliary to afford the 3A-hydroxycyclo-
pentyl glycine 9 as a single (2R,1AR,3AS) stereoisomer; the same
hydrogenation conditions also allowed direct conversion of 7 to
9 (90% yield). On the other hand, these hydrogenation
conditions enabled us to reduce the [3.3.0] isomer 6 directly to
the 2A-hydroxycyclopentyl glycine 10, also as a single
(2R,1AR,2AR) stereoisomer. Closely related analogues of 8 have
been used in the synthesis of cyclopentenylglycine,11 and of
carbocyclic analogues of nikkomycin12 and polyoxin C.13
In summary, the asymmetric aza-Diels–Alder adduct 4 reacts
with mCPBA to generate the Meisenheimer rearranged product
12 S. E. Ward, A. B. Holmes and R. McCague, Chem. Commun., 1997,
2085.
13 D. Zhang and M. J. Miller, J. Org. Chem., 1998, 63, 755; V. K.
Aggarwal and N. Monteiro, J. Chem. Soc., Perkin Trans. 1, 1997,
2531.
14 XSCANS X-ray Single Crystal Analysis System. Version 2.2. Bruker
AXS Inc., Madison, Wisconsin, USA, 1994.
15 G. M. Sheldrick, 1999. SHELXTL Version 5.1, Bruker AXS Inc.,
Madison, Wisconsin, USA.
2452
Chem. Commun., 2000, 2451–2452