A radical approach to araliopsine and related quinoline alkaloids using manganese(III) acetate

Article abstract of DOI:10.1016/S0040-4039(00)01324-1

Reaction of 4-hydroxy-1-methyl-2(1H)-quinolone 3 with electron-rich alkenes and manganese(III) acetate produces tricyclic quinoline alkaloids, including araliopsine 1, in one-pot reactions. This combined intermolecular addition-cyclisation reaction produces angular and/or linear tricycles and the regioselectivity of the cyclisation is shown to depend on whether alkyl or aryl substituents are attached to the alkene. (C) 2000 Elsevier Science Ltd.

Full text of DOI:10.1016/S0040-4039(00)01324-1

TETRAHEDRON
LETTERS
Pergamon
Tetrahedron Letters 41 (2000) 7751–7755
A radical approach to araliopsine and related quinoline
alkaloids using manganese(III) acetate
Gregory Bar, Andrew F. Parsons* and C. Barry Thomas
Department of Chemistry, University of York, Heslington, York YO10 5DD, UK
Received 25 July 2000; accepted 1 August 2000
Abstract
Reaction of 4-hydroxy-1-methyl-2(1H)-quinolone 3 with electron-rich alkenes and manganese(III)
acetate produces tricyclic quinoline alkaloids, including araliopsine 1, in one-pot reactions. This combined
intermolecular addition–cyclisation reaction produces angular and/or linear tricycles and the regioselectiv-
ity of the cyclisation is shown to depend on whether alkyl or aryl substituents are attached to the alkene.
©
2000 Elsevier Science Ltd. All rights reserved.
Keywords: alkaloids; manganese and compounds; quinolines; radicals and radical reactions.
More than 200 naturally occurring quinoline alkaloids (including araliopsine 1 and N-
methylschinifoline 2) have been isolated from the Rutaceae family of plants, and these
1
compounds have been shown to exhibit a wide range of important medicinal properties. These
include antiparasitic, anthelmintic and cytotoxic activities. As a consequence, the synthesis of
these types of compound is of considerable interest and one potentially straightforward and
flexible approach involves the introduction of a C-5 unit on to (commercially available)
quinoline 3 by deprotonation and alkylation with prenyl bromide. The resultant alkene 4 could
then be converted to a range of natural products using standard transformations (e.g. epoxida-
tion, dihydroxylation etc.). Unfortunately, the desired alkene 4 is isolated in a very low yield (ca.
2
4
%) from this reaction because of competitive di-alkylation and O-alkylation reactions. With a
view to developing a new, more efficient approach to these types of alkaloid, we now report the
novel alkylation of quinoline 3 using a manganese(III)-promoted radical addition reaction.
*
Corresponding author. Tel: +44-1904-432608; fax: +44-1904-432516; e-mail: afp2@york.ac.uk
0
040-4039/00/$ - see front matter © 2000 Elsevier Science Ltd. All rights reserved.
PII: S0040-4039(00)01324-1

7752
Our initial experiments involved the reaction of quinoline 3 with 1-octene and manganese(III)
acetate under a variety of conditions (Scheme 1). For example, when 3 (0.57 mmol, 1 equiv.) was
3
added to 1-octene (1 equiv.) and manganese(III) acetate (2 equiv.) in acetic acid (50 cm ) and the
solution stirred in an ultrasonic bath (300 W, 30–40 kHz) at 60°C (under an atmosphere of
3
nitrogen) the tricycle 5 was formed in 28% yield. In the absence of ultrasound no reaction took
4
place presumably because of the low solubility of 3 in acetic acid. The yield of 5 could be
improved to 50% when the (ultrasound) reaction was repeated using 10 (rather than 1)
equivalents of 1-octene. This reaction could also be mediated using less than 2 equivalents of
manganese(III) acetate when the oxidant, potassium permanganate, was added slowly to the
reaction. Hence, reaction of 3 with 1-octene (10 equiv.), manganese(III) acetate (0.3 equiv.) and
potassium permanganate (0.6 equiv., added portionwise over 3 h) produced 5 in 58% yield.
Potassium permanganate can therefore be used to oxidise manganese(II), which is produced as
a reaction by-product, back to manganese(III).
Scheme 1.
The formation of 5 is expected to involve initial oxidation of 3 to form radical 6. There are
several potential routes by which 6 might be formed but the involvement of an intermediate
5
manganese(III) enolate in this type of reaction is generally favoured. Radical 6, which is
electrophilic, is then expected to add regioselectively (on electronic and steric grounds) to the
6
electron-rich double bond of 1-octene to form a secondary radical. Oxidation of this radical, by
reaction with a second equivalent of manganese(III), will generate the corresponding secondary
cation 7, and nucleophilic attack by the enol hydroxyl group is expected to form the five-mem-
7
bered ring in 5. This mechanism can also explain the formation of tricycle 8 (in 64% yield) from
reaction of 3 with manganese(III) acetate and 2-methyl-1-pentene. In this case, the cyclisation is
expected to involve attack of the enol hydroxyl group on an intermediate tertiary carbocation.
Interestingly, when styrene was used in place of 1-octene, reaction with 3 gave an approxi-
mately equal (1.1:1) mixture of the two regioisomers, 9a and 10a, in a combined 53% yield. The
intermediate carbocation (of type 7) therefore reacts with either of the two possible enol
tautomers (of the ketone or amide) to give the two tricyclic isomers. This was also observed with
other alkenes bearing an aromatic substituent(s). Reaction of 3 with a-methylstyrene gave the

7753
angular isomer 9b in 34% yield and the linear isomer 10b in 52% yield, while reaction with
,1-diphenylethylene gave a 1:1 mixture of 9c and 10c in a combined yield of 76%. The reason
1
for the unexpected formation of linear isomers 10a–c using phenyl substituted alkenes is unclear.
Molecular orbital calculations (AM1 or PM3) suggest that the linear isomers are thermodynam-
ically less stable than the angular isomers and this was supported by the slow isomerisation of
8
1
0c to 9c on heating in acetic acid. One possible explanation could stem from the different rates
of cyclisation. As the enol is expected to react more slowly with a more stable benzylic (rather
than an alkyl) cation this could allow time for the formation of a second manganese(III) enolate
of type 11. Steric interactions between the benzene ring(s) and the bulky manganese(III) substituent
could result in cyclisation to give the linear, rather than angular, isomer.
A similar result was observed on reaction of 3 with phenylacetylene to give the angular and
linear furoquinolines, 12 and 13, in 20 and 43% yields, respectively. In this case, addition of radical
6
to the less substituted end of the alkyne triple bond is expected to generate an intermediate
’
a-styryl radical of the type PhC =CR . Although manganese(III) does not usually oxidise vinyl
radicals to vinyl cations, the a-styryl radical has been shown to be a p-radical (rather than a
2
9
10
s-radical), with extensive delocalisation over the benzene ring. As a consequence, it is likely that
the manganese(III) can oxidise this (benzylic rather than vinylic-type) radical to a cation, which
then reacts with either of the two possible enol tautomers to form 12 and 13. Reaction of 3 with
alternative terminal alkynes, bearing an alkyl rather than an aryl substituent, gave further support
to the importance of the benzene ring. In these cases, the reactions gave a number of products
in low yield, presumably because the manganese(III) cannot efficiently oxidise the intermediate
vinyl radicals.
This method of radical addition/cyclisation was then applied to the synthesis of (±)-araliopsine
by reacting quinoline 3 with manganese(III) acetate (2 equiv.) and 2-methyl-3-buten-2-ol (10
1
1
1
equiv.) as shown in Scheme 2. After purification by column chromatography (±)-araliopsine 1 was
isolated in 40% yield. Like the related 1-octene reaction (Scheme 1), none of the corresponding
linear quinoline isomer was isolated. When the same reaction was carried out using 0.3 equiv.
of manganese(III) acetate in the presence of potassium permanganate (0.7 equiv.), 1 was isolated
in a similar yield (35%). Reaction of 3 with the related 2-methyl-1,3-butadiene (isoprene) was also
investigated, but this gave rise to a number of products in low yield including 14 in 18% yield.
Scheme 2.

7754
It is of interest to contrast the manganese(III)-mediated reactions of 3 with those of the
related a-chloro-quinoline 15. This was formed from reaction of 3 with N-chlorosuccinimide (in
2% yield) and subsequent reaction with manganese(III) acetate and 2-methyl-3-buten-2-ol
under the same conditions as for the reaction with 3), gave only a low yield of quinoline dimers
510% yield). The fact that no products derived from addition to the alkene double bond were
8
(
(
isolated could reflect the stability of the first-formed tertiary radical 16. The introduction of a
chlorine atom is expected to reduce the reactivity of the radical due to electronic and steric
factors. Addition of 16 to the alkene double bond could also be reversible and this is supported
t
by the reaction of 15 with allyl tert-butyl sulfide ( BuSCH CHꢀCH ). This reaction produced the
2
2
allyl derivative 17 in 40% yield, which was derived from intermolecular radical addition to the
alkene double bond followed by b-fragmentation to expel the BuS radical.
t
’
Biologically important quinoline alkaloids, including araliopsine 1, can therefore be prepared
from reaction of quinoline 3 with electron-rich alkenes or phenylacetylene in the presence of
manganese(III) acetate. This one-pot reaction gives rise to linear and/or angular tricyclic
alkaloids; the regioselectivity of the cyclisation is determined by the substituents on the alkene
double bond. These combined intermolecular addition–cyclisation reactions offer a quick,
economical and flexible approach to quinoline alkaloids. It should also be noted that the use of
manganese(III) acetate compares favourably with the use of cerium(IV) ammonium nitrate
(
(
CAN) in these alkylation reactions. Thus, reaction of 3 (1 equiv.) with 2-methyl-3-buten-2-ol
2–20 equiv.) in the presence of CAN only gave rise to dimeric products derived from
1
2
homo-coupling of quinoline radical 6. The reasons for this difference are currently under
investigation.
Acknowledgements
We would like to thank BP Amoco Chemicals, the EU (under the ERASMUS scheme), and
The University of York for funding, and Dr. Peter Hanson for the molecular orbital
calculations.
References
1
2
3
4
5
6
. Michael, J. P. Nat. Prod. Rep. 1999, 16, 697 and earlier reviews in the same series.
. Shobana, N.; Yeshoda, P.; Shanmugam, P. Tetrahedron 1989, 45, 757.
. All new compounds exhibited satisfactory spectral and analytical (high resolution mass) data.
. Reactions in alternative solvents (including methanol or acetonitrile) were lower-yielding.
. Snider, B. B. Chem. Rev. 1996, 96, 339.
. Reaction of quinoline 3 with electron-poor alkenes (such as methyl acrylate) was unsuccessful.

7755
7
. For related alkylations of quinoline 3 mediated by silver(I) carbonate/Celite (which gave exclusively angular
products) see: (a) Lee, Y. R.; Kim, B. S. Tetrahedron Lett. 1997, 38, 2095. (b) Lee, Y. R.; Kim, B. S.; Kweon,
H. I. Tetrahedron 2000 56, 3867.
8. Ultrasonic irradiation of 10c in acetic acid for 0.75 h at 60°C produced a mixture of 10c and 9c in the ratio of
4
:1.
9. Snider, B. B.; Merritt, J. E.; Domboski, M. A.; Buckman, B. O. J. Org. Chem. 1991, 56, 5544.
10. (a) Bennett, J. E.; Howard, J. A. Chem. Phys. Lett. 1972, 76, 1957. (b) Kasai, P. H.; McBay, H. C. J. Phys. Chem.
1
984, 88, 5932.
1
1
1. For an alternative approach to 1 see: Barr, S. A.; Boyd, D. R. J. Chem. Soc., Chem. Commun. 1994, 153.
2. Dimerisation of quinolines (including 3) using CAN, in the absence of alkenes, has been reported previously. Ye,
J.-H.; Xue, J.; Ling, K.-Q.; Xu, J.-H. Tetrahedron Lett. 1999, 40, 1365.
.