A new and enantioselective indolizidine synthesis by meso-epoxide α-deprotonation-transannular N-C insertion

Article abstract of DOI:10.1039/a809105f

Enantioselective α-deprotonation-rearrangement of N-Boc hexahydroazonine oxide 10 using organolithiums in the presence of (-)-sparteine 3 gives the ester 12 in up to 89% ee.

Full text of DOI:10.1039/a809105f

A new and enantioselective indolizidine synthesis by meso-epoxide
a-deprotonation–transannular N–C insertion
David M. Hodgson* and Lesley A. Robinson
Dyson Perrins Laboratory, Department of Chemistry, University of Oxford, South Parks Road, Oxford,
UK OX1 3QY. E-mail: david.hodgson@chem.ox.ac.uk
Received (in Liverpool, UK) 19th November 1998, Accepted 5th January 1999
Enantioselective a-deprotonation–rearrangement of N-Boc
hexahydroazonine oxide 10 using organolithiums in the
presence of (-)-sparteine 3 gives the ester 12 in up to 89%
ee.
material (64%). An attempt to induce reaction at the epoxide
group subsequent to ortho-deprotonation using double the
quantities of reagents indicated above led to no identifiable
products; an alternative protecting group was therefore re-
quired. Removal of the tosyl group from 7 using sodium
naphthalenide and immediate Boc reprotection of the amine
hydrochloride salt gave the reduced azonine 8 (64%). Epoxida-
tion provided 10, which could potentially undergo deprotona-
tion with an organolithium either a to the epoxide oxygen, or a
to nitrogen. Beak has reported a 6-exo-tet cyclisation onto an
epoxide via deprotonation a to NBoc; the deprotonation site
was however also benzylic in this case.5 Beak has also reported
that the rate of deprotonation of Boc-protected azacycles
decreases on moving from pyrrolidine to piperidine to perhy-
droazepine.6 In the event, reaction of the epoxide 10 with BusLi
(2.4 equiv. in Et2O at 278 °C for 5 h, followed by warming to
25 °C over 15 h) led to an inseparable 1 : 1 mixture of epimers
(due to the stereogenic centre in the Bus group, vide infra) of
ketone 11a (48%, 70% based on recovered epoxide 10, Scheme
3).
We recently reported the enantioselective a-deprotonation–
rearrangement of medium-sized (8-, 9- and 10-membered)
cycloalkene-derived achiral epoxides using a secondary orga-
nolithium in combination with a chiral ligand such as (2)-spar-
teine 3, which gives bicyclic alcohols in good yields and ees
(77–84% ee, e.g. Scheme 1, X = CH2).1 However, only a single
functional group is generated in the desymmetrised bicycles.
One strategy to enhance the utility of this transformation would
be to examine heterocycloalkene-derived achiral epoxides.
Here we communicate our preliminary results concerning the
synthesis and novel rearrangement chemistry of an azacyclic
epoxide of this type (1, X = NR).
Scheme 1 Reagents and conditions: i, PriLi (2.4 equiv.), (2)-sparteine 3
(2.5 equiv.), Et2O, 298 °C (5 h) to 25 °C (15 h).
An important aspect of the study of transannular reactions of
a medium-sized heterocycle concerns the potential problem of
preparing the substrate.2 However, application of methodology3
used in the synthesis of the azacycloundecene system found in
manzamine C led to a highly satisfactory route to the azacyclic
epoxide 9 (Scheme 2). Thus, cyclisation under dilute conditions
of the ditosylate 6 of the known diol 5 (readily available from
cycloocta-1,5-diene)4 gave the reduced azonine 7 in 62% yield;
to the best of our knowledge this is the most efficient cyclisation
reported which gives a simple reduced azonine.2
Scheme 3
In contrast, reaction of the epoxide 10 with BusLi, under the
same conditions but in the presence of TMEDA (2.5 equiv.), led
to the formation of ester 12 as the major product (12 : 11a, 8 : 1
by 1H NMR analysis; 74% isolated yield of 12). Using
(2)-sparteine 3 as the ligand in an otherwise identical
experiment gave an equal mixture of 11a and 12 (66% ee for
12).† Experiments were then carried out to examine the
possibility of increasing both the proportion and ee of ester 12
formed from epoxide 10 (Table 1).
Maintaining the reaction at 278 °C for 18 h and then
quenching at this temperature gave ester 12 in improved ee
(74%, Table 1, entry 1), but the ketone 11a predominated.
However, repeating the same procedure at 298 °C significantly
improved the proportion of ester 12 (12 : 11a, 5 : 1) and
increased the ee of 12 to 79% (entry 2). Using PriLi at 298 °C
gave mainly the ester 12 (12 : 11a, 10 : 1) and with the highest
level of asymmetric induction (89% ee, entry 3),‡ as also
observed with our earlier work on cycloalkene-derived ep-
oxides.1 Using (2)-a-isosparteine 4 as ligand with either BusLi
or PriLi slowed the reaction considerably (entries 4 and 5),
particularly in conjunction with BusLi; the ees were also
reduced compared with the corresponding (2)-sparteine 3
reactions. In an attempt to allow PriLi/(2)-a-isosparteine 4 to
completely consume the epoxide 10, the reaction was left for 40
h at 298 °C (entry 6), but it still remained only 50% complete
after this time and no change in the ee of ester 12 was observed.
The use of catalytic amounts of ligand was also investigated
Scheme 2 Reagents and conditions: i, TsCl (4.9 equiv.), Py, 0 °C (5 h) to 25
°C (15 h), 74%; ii, TsNH2 (1.7 equiv), NaOH (200 equiv.), Bu4NI (1.4
equiv.), toluene–H2O, reflux, 5 h, 62%; iii, Na naphthalenide (2.5 equiv.),
THF, 278 °C, then HCl(g), then Et3N (1.5 equiv.), Boc2O (1.5 equiv.),
DMAP (0.1 equiv.), CH2Cl2, 25 °C, 64% from 7; iv, MeCO3H (1.2 equiv.),
Na2CO3 (3 equiv.), NaOAc (0.02 equiv.), CH2Cl2, 0 °C (10 min) to 25 °C
(15 h), 82% (R = Ts), 87% (R = Boc).
Subjection of the epoxide 9, derived from reduced azonine 7,
to typical asymmetric rearrangement conditions1 [BusLi (2.4
equiv.) and (2)-sparteine 3 (2.5 equiv.) in Et2O at 278 °C for
5 h, followed by warming to 25 °C over 15 h, cf. Scheme 1] led
only to the recovery of starting epoxide 9, whereas quenching
the reaction with D2O led to essentially complete o-deuterium
incorporation into the tosyl group of the recovered starting
Chem. Commun., 1999, 309–310
309

Table 1 Effect of experimental conditions on the yields and enantioselectiv-
ities of formation of indolizidine 12 from epoxide 10 using ligand/RLi in
Et2O
process seems most likely, since reducing the equivalents of
organolithium from 2.5 improves (at the expense of conversion
of starting epoxide 10) the ratio of ester 12 : ketone 11, and in a
separate experiment ester 12 could be quantitatively converted
to ketone 11b using PriLi (1.1 equiv., 278 °C for 1 h, followed
by warming to 0 °C over 2 h).
Insertion of a lithiated epoxide into a C–N bond has not
previously been reported and the present study illustrates an
example of this process leading to a new and enantioselective
entry to the important indolizidine framework. Further studies
on the scope of this process are in progress and will be reported
in due course.
We thank the EPSRC for a Research Grant (GR/L39940:
postdoctoral support to L. A. R.), the EPSRC National Mass
Spectrometry Service Centre for mass spectra and Dr D. J.
Watkin and Professor K. Prout (Chemical Crystallography
Laboratory, University of Oxford) for assistance with the X-ray
structure analysis.
Entrya Ligand RLi
10 : 11 : 12b
Yield of 12c (%) Ee of 12 (%)
1d
2
3
4
5
6e
7f
8f
9e,f
3
3
3
4
4
4
3
4
4
BusLi 0 : 1.6 : 1.0
32 (20)
74
79
89
64
79
78
82
77
89
BusLi 0.1 : 0.2 : 1.0 58 (50)
PriLi 0.3 : 0.1 : 1.0 57 (49)
BusLi 5.0 : 0 : 1.0
14
PriLi 2.0 : 0.1 : 1.0 29
PriLi 1.3 : 0.1 : 1.0 40
PriLi 0.7 : 0.6 : 1.0 36
PriLi 1.2 : 0.1 : 1.0 33
PriLi 0.4 : 0.2 : 1.0 54
a Ratio of ligand : RLi : epoxide 10, 2.45 : 2.4 : 1 and carried out at 298 °C
with a reaction time of 18 h unless otherwise indicated. b Ratios determined
c
by 1H NMR analysis of the crude reaction mixture. Yield of 12 as
measured by 1H NMR analysis using methyl diphenylacetate as an internal
d
standard. Isolated yields given in parentheses. Carried out at 278 °C.
e Reaction time 40 h. f Ratio of ligand : RLi : epoxide 10, 0.24 : 2.4 : 1.
Notes and references
† Ees were determined by GC (Chrompack chirasil-dex 25 m 3 0.32 mm ID
column; 6 psi, 120 °C). The absolute configurations of the predominant
indolizidinol enantiomers are not known but can be tentatively assigned as
shown in Scheme 3 by analogy with the selectivity for deprotonation at the
R configured epoxide stereocentre with (2)-sparteine 3 observed in our
earlier medium-ring studies (ref. 1).
‡ Freshly distilled (2)-sparteine (70 mm3, 0.30 mmol) in Et2O (1 cm3) was
added dropwise over 0.5 h to a stirred solution of PriLi [1.09 mol dm23 in
light petroleum (boiling range 40–60 °C); 270 mm3, 0.29 mmol] in Et2O (1
cm3) at 298 °C. The reaction mixture was allowed to stir for 1 h at 298 °C
before the epoxide 10 (30 mg, 0.12 mmol) in Et2O (1 cm3) was added
dropwise over 0.5 h. The reaction mixture was stirred for 18 h at this
temperature and then H3PO4 (0.5 mol dm23 in water; 1 cm3) added slowly
dropwise. After warming to room temperature the organic layer was
removed and the aqueous layer extracted with Et2O (3 3 5 cm3). The
combined organic extracts were dried (MgSO4) and then evaporated under
reduced pressure. Purification of the residue by column chromatography
[(SiO2, 50% Et2O–light petroleum (boiling range 40–60 °C) ? 100% Et2O]
gave the ester 12 (14.8 mg, 49%); [a]2D1+48.6 (c 0.3 in CHCl3).
§ Crystal data for 11b: C12H20NO2, M = 210.29, orthorhombic, space
group P212121 (No. 19), a = 5.807(7), b = 13.393(2), c = 15.477(3) Å, V
= 1203.7(3) Å3, Z = 4. 1038 independent reflections measured at 173 K on
an Enraf-Nonius DIP2000 diffractometer. Mo-Ka radiation. 718 reflections
with I > 8s(I) and 137 variables yield R = 0.064, Rw = 0.063. CCDC
182/1138. Crystal data are available in CIF format from the RSC web site,
see: http://www.rsc.org/suppdata/cc/1999/309/
(entries 7–9) with interesting results. Using 24 mol% (2)-spar-
teine 3 (10 mol% with respect to PriLi), high levels of ee (82%)
were still achieved, but the reaction was found to be much
slower. In contrast, (2)-a-isosparteine 4 was more effective
when used in a catalytic fashion (entry 8), with no apparent
change in the ee (compare entry 5). Repeating this last reaction
but leaving it for 40 h at 298 °C allowed the reaction to proceed
further to completion and also gave a much higher level of ee
(entry 9).
The structures of indolizidinols 11 and 12 were assigned by
extensive spectroscopic investigations and were later further
supported by X-ray crystallographic analysis of ketone 11b.§ A
mechanistic explanation for the formation of the indolizidinols
is that they arise via lithiation a to the epoxide oxygen to give
13, followed by transannular reaction using the N lone pair to
give an ammonium ylide 14 which undergoes [1,2] migration of
the exocyclic N substituent (Scheme 4); direct insertion of the
lithiated epoxide into the exocyclic C–N bond is also possible.
Incorporation of the organolithium to give the ketones 11 could
occur before or after the transannular reaction. The latter
1 D. M. Hodgson, G. P. Lee, R. E. Marriott, A. J. Thompson, R. Wisedale
and J. Witherington, J. Chem. Soc., Perkin Trans. 1, 1998, 2151.
2 P. A. Evans and A. B. Holmes, Tetrahedron, 1991, 47, 9131.
3 Y. Torisawa, A. Hashimoto, M. Nakagawa, H. Seki, R. Hara and T. Hino,
Tetrahedron, 1991, 47, 8067.
4 D. Raederstorff, A. Y. L. Shu, J. E. Thompson and C. Djerassi, J. Org.
Chem., 1987, 52, 2337.
5 P. Beak, S. Wu, E. K. Yum and Y. M. Jun, J. Org. Chem., 1994, 59,
276.
6 P. Beak and W. K. Lee, J. Org. Chem., 1993, 58, 1109.
Scheme 4
Communication 8/09105F
310
Chem. Commun., 1999, 309–310