BIOGENETICALLY PATTERNED SYNTHESIS OF CADAMBINE

Article abstract of DOI:10.1016/0040-4039(91)85020-6

We report the synthesis of the glucosidic indole alkaloid cadambine (5) in 48 percent yield from its biological precursor secologanin (1) in an intermolecular variant of the in vivo process where chemoselective and regioselective attack by tryptamine on a 3S-epoxide 11 yields the seven-membered azepine ring.

Full text of DOI:10.1016/0040-4039(91)85020-6

Tarahedron Letters, Vol.32, No.17, pp 1987-1990, 1991
Printed in Great Britain
0040 4039/91 83.00 + .00
Pergamon Pre~s plc
BIOGENETICALLY PATTERNED SYNTHESIS OF CADAMBINE
R i c h a r d T. Brown*= D. M a l c o l m D u c k w o r t h and Cid A. M. S a n t o s
Chemistry Department, The University, Manchester M13 9PL, U.K.
S u m m a r y : We report the synthesis of the glucosidic indole alkaloid cadambine (5) in 48% yield
from its biological precursor secologanin (I) in an intermolecular variant of the in vivo process
where chemoselective and regioselective attack by tryptamine on a 3S-epoxide 11 yields the
seven-membered azepine ring.
The glucosidic indole alkaloid cadambine 5 isolated from Anthocephalus cadamba1 is
considered to be biosynthesized from secologanin 1 via the universal precursor strictosidine 2.
Stereospecific epoxidation to give the 18,19S-epoxystrictosidine
3 and regioselective
intramolecular attack by N(b) on the epoxide yields 3(z-dihydrocadambine 4 with the formation of a
seven-membered azepine ring 1,2. Oxidation of C-3 and cyclization of the alcohol leads to
cadambine 5. We now report the synthesis of cadambine from its biological precursor secologanin
1 in an intermolecular variant of the in vivo process where chemoselective and regioselective
attack by tryptamine on a 3S-epoxide 11 yields the seven-membered azepine ring. Essentially we
have interchanged the condensation step f and oxidation step 4 of the natural sequence.
4
G-
N
...-.~
O
2
NH
Jn v,vo~.
~
H~
.,,, ..H,
"~C..~.f9 ~" ~OGIc
in vivo
,
~
H
OGLc
H~'"
~
H H
..6 ..~OGIc
~
H,~
MeO2
HW"
3
1
2
3 ~ in vivo
I
4
,
H
H~
O H
,
o
[o]
,,~l H
H
.,~l H
.,,~ OGIc
H~,~.-
5
Our first step was a simple Jones' oxidation of secologanin tetra-acetate to give
secoxyloganin tetra-acetate 6. However, the subsequent epoxidation of the 3,4-vinyl group
presented major problems both in competitive attack on the 8,9-enol ether and in controlling C-3
stereochemistry. These were eventually resolved by a highly chemoselective and stereoselective
iodolactonisation of secoxyloganin tetra-acetate to afford the key intermediate 10.
1987

1988
H
•.
, H
OGio(OAc)4
!
NaOEt Jr
f
•
2/K2C03
He""
MeO
oxJdn
Mee2
6
1 O
0 ~ 4
(88%)
(79%)
EtO~C
MeO2
3 "
H
11
~
~
1 Ttyptamine
EtOH/A/5d.
2 Ac2O/Py
N
O
Ac
~l He
., OGIc(OAc)4
'
H
,t~, H
..¢, OGIc .ml
Hl~.-
'
1. POCI~/PhMe/A
2. NaOMe
MeO2
5
1 2
(95%)
(68%)
Halolactonization of 6 can give two other products 8 and 9 in addition to the desired 3(S)
lactone 7: using bromine, the glucoside (R -- H) gave a quantitative yield of the undesired 9-
lactone, whereas the tetra-acetate (R = Ac) gave a 2:1 ratio of 9- and 3-1actones:~, the latter being a
10:1 mixture of stereoisomers 7 and 8, Le. increasing the steric bulk of the sugar moiety reduced
the rate of 8,9 vis-a-vis 3,4 attack. We then decided to replace bromine by iodine because its larger
size could also decrease 8,9-selectivity, and indeed using iodine on the glucoside afforded a 50:50
mixture of 9- and 3-1actones. Combining the two steric effects by using the tetra-acetate with iodine
as eleotrophile we obtained essentially one 3-1actone in good yield with none of the other
stereoisomer.
Product ratio (%)
X
Br
Br
I
R
H
S o l v e n t
H20
7
8
9
100
67
50
5
Ac
H
0H2012
H20/KI3
CH2CI2
30
50
95
3
I
Ac
H
H
o
o
3
H
-
H x
*¢*
OGIo(OR)4
'~
OGIc(OR)4
HI,~'I
~ 0 1''OGIc(OR)4
H**~
;
H ~ , V "
Moo c '
oeo 'o
oeo2e °
7
8
9

1989
On the basis of stereochemical arguments given below we had anticipated that the major
product would be the desired 3S-lactone 7, but to confirm this n.O.e, experiments were required.
These showed key through-space correlations for H-3 with H-1 and H-6ax but not with H-7 or H-6eq
(Fig, I) only consistent with an equatorial 3S orientation. If H-3 had been axial, some interaction
with H-7 and H-6eq would have been expected and none with H-I.
Formation of the 3S-iodolactone 7 rather than the 3R 8 is rationalised by the following
argument. Attack by iodine and carboxylate on the 3,4-alkene must be in an antiperiplanar manner
with a chair-like transition state for the 6-1actone ring. Both II and III satisfy these requirements, but
by rotation about the C-1/O bond the acetylated sugar moiety sweeps through the region that would
be occupied by C-4 and the iodine electrophile in III, but only by H-3 in It. Hence the latter with
lesser steric crowding is preferred exclusively with iodine, although with the smaller bromine
reaction does occur to some extent via the alternative pathway. Again, rotation about the C-1'/O
linkage will sweep directly under the 8,9-alkene across the trajectory of the electrophile with a
resulting steric retardation. This is particularly effective with the bulky iodine, diverting attack almost
completely to the 3,4-alkene, but much less so with the smaller bromine electrophile.
.eo1.9c H y
#
°
'
~
0
~.0
/
o,o
'
'X
/ \ o A c
/ \o,c
't X ,
H
i
II
III
Subsequent treatment with alkoxide afforded the desired 3S-epoxide with concomitant de-
acetylation of the sugar. The next step required a chemoselective and regioselective reaction of
tryptamine with C-4 of the epoxide in the presence of other sites that were susceptible to
nucleophilic attack: the other carbon C-3 of the epoxide, the C-5 ester, the C-1O ester and the
conjugated alkene at C-9. Of these possibilities the l~-alkoxyacrylate ester system was known to
react sluggishly, and under basic conditions C-3 was unlikely to compete with the less hindered
C-4. The greatest danger lay with the C-5 ester, since, although we wished it to react subsequently
with the amine to form the azepine ring, if it reacted first, an amide would be formed. This would not
be nucleophilic enough to open the epoxide under these reaction conditions, and in any case
would almost certainly lead to a six rather than seven-membered ring. initially we used sodium
methoxide to give the epoxide with a methyl C-5 ester, but the reaction with tryptamine in refluxing
methanol yielded a mixture of products including amides. We considered that increasing the size of

1990
the alkyl group at the C-5 ester might decrease its rate of reaction relative to the epoxide. An
isopropyl group seemed suitable but potassium isopropoxide reacted too slowly with the lactone
and Zemplen de-acetylation was incomplete. However the ethyl ester epoxide 11 could be
obtained readily, using 1.7 equivalents of sodium ethoxide in ethanol, interestingly without any
appreciable exchange of the C-10 methyl ester. In the event the ethyl group provided sufficient
steric retardation to be effective. Thus heating 11 with tryptamine in ethanol under reflux for five
days, and subsequent acetylation gave in the reasonable yield of 68% from 10 the desired
azepine lactam 12 ([~]D~j -85° (CHCI3), Vmax 1757 (OAc), 1642 (NCO) cm 1,6 7.28 (d, J= 1.5, H-17),
5.18 (d, J= 7, H-21), 4.47 (m, H-19), 4.01 (m, H-5a), 3.41 (m, H-5b), 3.36 (m, 2H-18a,b), 3.09 (m, H-
6a), 3.00 (m, H-15), 2.99 (m, 2H-14a,b), 2.91 (m, H-6b), 1.90 (m, H-20), M+ 772.2680 Calc. for
C37H44N2016 772.2690). Direct cyclisation to a seven-membered ring is much more difficult than a
six or five, but we rationalised that it could in this case be obtained by initial formation of a 6-1actone
13 from the C-3 alcohol, and subsequent opening of the lactone by N(b) via a five-membered ring
intermediate.
m
i',
t
It,
im
¢
"
H
1
2
H
".~'
Me02C f
~
0
13
11
Finally, Bischler-Napieralski cyclization of 12 followed by Zemplen de-acetylation gave
cadambine 5 (tetra-acetate: m.p. 149-151 °C, [¢z]~ -124° (CHCI3)4) identical with natural material in
an overall yield of 48% from secologanin 1. This synthesis corroborates the structure and absolute
stereochemistry unambiguously determined for cadambine from n.m.r, and c.d. spectra and other
data ], Cadambine and its immediate precursor can be selectively reduced] to either 3~- or 31~-
dihydrooadambine, which have been obtained as a mixture in a previous synthesis5.
Acknowledgements CAMS wishes to thank CAPES, Brazilian Government, for financial support
and UFPR for leave of absence; and DMD acknowledges a SRC Quota Studentship.
References
1. R.T. Brown and S,B. Fraser, Tet. Letters, 1957 (1974).
2. G.I. Dimitrienko, D.G. Murray and S. McLean, Tet. Letters, 1961 (1974).
3. D. M. Duckworth, Ph.D. Thesis, Manchester University, 1978.
4. The rotation given in ref. 1 is incorrect.
5. R.G, Hamilton, G.N. Saunders and S. McLean, Can. J. Chem., 61,284 (1983).
(Received in UK 24 January 1991)