Table 1. Preparation of tetracyclic 12-oxa-4-azatetracyclo[5.4.1.02,6.08,11]dodecan-3,5-
diones by a sequence of Diels–Alder reaction, ring opening metathesis, and intramo-
lecular [2+2] photocycloaddition.
The desymmetrization of compounds 2 by an
enantiotopos-selective reduction of the imide car-
bonyl groups[15,16] turned out to be the most reliable
way to access enantioenriched compounds with a
bicycloACHTNUTRGNE[UNG 4.2.0]octane skeleton. After initial reduction
to a hydroxylactam the reduction was completed
with triethylsilane in trifluoroacetic acid (TFA) and
the enantiomeric excess of the resulting products 6
was determined. Optimization experiments were
performed with the N-benzyl-substituted substrate
2e. The best compromise between high chemoselec-
tivity and high enantioselectivity was to run the re-
action with stoichiometric amounts of borane and
an oxazaborolidine catalyst[17] derived from (1S,2R)-
cis-1-amino-2-indanol in THF as the solvent at
408C. After 12 h the reaction was stopped and
product 6e was separated from remaining starting
material (15%). Higher enantioselectivities as com-
pared to the 86% ee recorded for exo-6e in Table 2
were achieved with a higher catalyst loading (up to
91% ee). Oxazaborolidines derived from other
amino alcohols or with other boron substituents
than methyl did not show an improved perform-
ance.
From the data presented in Table 2, it is apparent
that the general procedure optimized for substrate
2e was ideally suited for the enantioselective reduc-
tion of N-aryl (2c, 2d) and N-benzyl (2e–2g)
imides but it was less suited to achieve an enantio-
selective reduction of N-alkyl imides (2a, 2b). In
most cases, the chiral reagent showed no significant
differentiation in the reduction of either exo- or
endo-substrate and the d.r. of products 6 was similar
to the d.r. of substrates 2. Also the enantiomeric
excess for exo- or endo-product did not vary exten-
Substrate
R1
Yield[a]
4 [%]
Yield
Yield[b]
d.r.[c]
ACHTUNGTRENUN(NG exo/endo)
5 [%][d]
2 [%][d]
3a
3b
90
90
87
75
89
88
82/18
89/11
3c
3d
3e
81
59
77
76
69
79
74[e]
97
82/18
91/9
91
66/34
3 f
3g
76
83
72
62
95
94
80/20
83/17
[a] The minor diastereoisomer of the Diels–Alder reaction could be separated by
column chromatography. Yields refer to the isolated diastereoisomer 4 depicted in the
reaction equation. [b] All [2+2] photocycloaddition reactions were conducted using a
RPR-100 reactor with 16 Rayonet RPR-2540 ꢃ lamps (quartz apparatus) as the irradi-
ation source in deaerated toluene (c=15 mm, t=20 h). [c] Diastereomeric products
exo-2 (Figure 2) and endo-2 were not separable by flash chromatography. The dia-
ACHTUNGTRENNUNG
stereomeric ratio (d.r.) was determined by 1H NMR integration of the product mix-
ture. [d] Yields refer to isolated products after chromatographic purification. [e] 24%
of starting material was re-isolated.
The reaction sequence 3 ! 2 proceeded smoothly for all
substrates and only a few issues require special comment. In
the Diels–Alder reaction, major product 4 was formed to-
gether with a diastereomeric by-product. The separation of
the latter compound was facile and products 4 were ob-
tained as single diastereoisomers. In the photocycloaddition
of substrate 5c, the reaction did not go to complete conver-
sion after 20 h, which may be due to competing absorption
by the N-phenyl chromophore. No side reactions were ob-
served, though, and the yield based on conversion was 97%.
Attempts to achieve the conversion of products 2 to chiral
ring-opening products by elimination were performed with
different bases.[13] Unfortunately, the reactions did not lead
to homogenous products but rather to product mixtures.
Due to the fact that the a-carbonyl protons are more readily
accessible in the endo-diastereoisomer endo-2, elimination
products turned out to be enriched in the respective endo-
isomer. Attempted ring-opening reactions with Lewis
acids,[14] for example with BBr3 in CH2Cl2, led to a cleavage
of the central ether bond by nucleophilic displacement. The
resulting halohydrins were unstable, however, and could not
be isolated in enantiomerically pure form.
sively. Only with R=ortho-methoxybenzyl (substrate 2g)
was the endo-product apparently more readily reduced than
the exo-product and the d.r. changed from 83/17 to 70/30.
Separation of the diastereoisomers exo-6 and endo-6 was
not possible by flash chromatography.
Gratifyingly, base treatment of compounds 6 not only in-
duced the desired ring-opening of the central ether bond
but it also led to products 7, the two diastereoisomers of
which were readily separable. Lithium bis(trimethylsilyl)-
AHCTUNGTRENNUNG
amide (LHMDS) was the base of choice[18] for this transfor-
mation. Upon elimination, major exo-isomers exo-7 were ac-
cessed as shown for substrate 6c in Scheme 1. In these com-
pounds five contiguous stereogenic centers around the cen-
tral bicycloACTHNUGRTNEUNG[4.2.0]octane skeleton exhibit a defined absolute
and relative configuration. Apart from the depicted exo-
product exo-7c the endo-isomer endo-7c was isolated in
18% yield. The total yield of the elimination was conse-
quently 94%. As expected there was no significant change
in the enantiomeric excess and product exo-7c was obtained
in 97% ee.
To substantiate the presumed relative configuration of
product exo-7c and its congeners, the compound was con-
&
2
&
ꢂ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 0000, 00, 0 – 0
ÝÝ
These are not the final page numbers!