A novel synthesis of (-)-huperzine A via tandem intramolecular aza-Prins cyclization-cyclobutane fragmentation

Article abstract of DOI:10.1021/ol400012s

The acetylcholinesterase inhibitor (-)-huperzine A was synthesized from (S)-4-hydroxycyclohex-2-enone in 17 steps by a route that involved two cyclobutane fragmentations. The first of these employed a retro-aldol cleavage to generate the α-pyridone ring o

Full text of DOI:10.1021/ol400012s

ORGANIC
LETTERS
2013
Vol. 15, No. 4
882–885
A Novel Synthesis of (ꢀ)-Huperzine
A via Tandem Intramolecular Aza-Prins
CyclizationꢀCyclobutane Fragmentation
James D. White,* Yang Li, Jungchul Kim, and Miroslav Terinek
Department of Chemistry, Oregon State University, Corvallis, Oregon 97331,
United States
james.white@oregonstate.edu
Received January 2, 2013
ABSTRACT
The acetylcholinesterase inhibitor (ꢀ)-huperzine A was synthesized from (S)-4-hydroxycyclohex-2-enone in 17 steps by a route that involved two
cyclobutane fragmentations. The first of these employed a retro-aldol cleavage to generate the R-pyridone ring of huperzine A, and the second invoked a
novel intramolecular aza-Prins reaction in tandem with stereocontrolled scission of a cyclobutylcarbinyl cation to create the aminobicyclo[3.3.1]nonene
framework of the natural alkaloid.
Release of the 26ꢀ28 kcal/mol of ring strain in cyclo-
butanes through fragmentation is a well-known strategem
for constructing new cyclic and acyclic systems.1 The principle
has been applied in natural product synthesis2 where it is
often the pivotal transformation,3 but cyclobutane frag-
mentation in tandem with a second process that initiates
ring cleavage is rare.4
We were intrigued by the possibility of choreographing
fragmentation of a cyclobutylcarbinyl cation withan intra-
molecular acid-catalyzed aza-Prins reaction5 in a cascade
process that would lead to the aminobicyclo [3.3.1]nonene
skeleton of the nootropic agent huperzine A (1).6 The
concept is diagrammed in Scheme 1 where CꢀC bond for-
mation is initiated by attack of the endo isopropenyl group
of 2 on an acyliminium ion to generate transiently a
cyclobutylcarbinyl cation. Of the two cyclobutane bonds,
a and b, susceptible to cleavage in 2, the orbital alignment
of external σ-bond a with the developing carbocation
favors this mode of ring scission. The process is terminated
by the stereoelectronically favored removal of a proton
from the bicyclic ring fusiontoform the exo ethylidene unit
of 1. We now report implementation of this concept in a
synthesis of natural (ꢀ)-huperzine A (1).7
(1) The concept of ring strain release as a synthetic tool has its roots in
some of the earliest work in the synthesis literature; for example, see:
Doering, W. von E.; Knox, L. H. J. Am. Chem. Soc. 1953, 75, 297.
(2) White, J. D.; Lee, N. C. Applications of Ring Strain in Natural
Product Synthesis. In Advances in Strain in Organic Chemistry; Halton,
B., Ed.; Jai Press: Hampton Hill, U.K., 1997; Vol. 6, pp 1ꢀ34.
(3) (a) White, J. D.; Dillon, M. P.; Butlin, R. J. J. Am. Chem. Soc.
1992, 114, 9673. (b) White, J. D.; Kim, J.; Drapela, N. E. J. Am. Chem.
Soc. 2000, 122, 8665.
(4) Fragmentation/rearrangement of cyclobutane spiroepoxides is
one of the few examples. See: (a) Tobe, Y.; Yamashita, S.; Kakiuchi,
K.; Odaira, Y. J. Chem. Soc., Chem. Commun. 1984, 1259. (b) Pirrung,
M. C.; Thomson, S. A. J. Org. Chem. 1988, 53, 227.
The starting point for our route to (ꢀ)-1 was (S)-(ꢀ)-
4-hydroxycyclohex-2-enone (4), obtained from (ꢀ)-quinic
(5) Dobbs, A. P.; Guesne, S. J. J.; Parker, R. J.; Skidmore, J.;
Stephenson, R. A.; Hursthouse, M. B. Org. Biomol. Chem. 2010, 8, 1064.
(6) The Lycopodium alkaloid huperzine A was first isolated by
Wiesner (Yoshimura, H.; Valenta, Z.; Wiesner, K. Tetrahedron Lett.
1960, 12, 14) who misassigned its structure. Subsequent comparison of
huperzine A with the known alkaloid selagine showed that they were
identical and led to the correct assignment 1 (Ayer, W. A.; Browne,
L. M.; Orszanska, H.; Valenta, Z.; Liu, J. S. Can. J. Chem. 1989, 67,
1538). For an early account of the history, chemistry, and pharmacology
of huperzine A, see: Kozikowski, A. P.; Tuckmantel, W. Acc. Chem. Res.
1999, 32, 641 and references cited.
(7) Two main lines of attack on the synthesis of huperzine A, both
pioneered by Kozikowski and co-workers, have been developed and
have been extended subsequently by others. One approach builds the
bicyclo[3.3.1]nonene skeleton of 1 using a tandem Michael-aldol se-
quence; the other route employs a palladium-catalyzed annulation as the
key step. For a recent summary of previous work in the field and the
current state of play in huperzine A synthesis, see: (a) Koshiba, T.;
Yokoshima, S.; Fukuyama, T. Org. Lett. 2009, 11, 5354. (b) Ding, R.;
Sun, B.-F.; Lin, G.-Q. Org. Lett. 2012, 14, 4446.
r
10.1021/ol400012s
Published on Web 01/24/2013
2013 American Chemical Society

Scheme 1. Proposed Route to (ꢀ)-Huperzine A (1) via
Intramolecular Aza-Prins Cyclization and Cyclobutane
Fragmentation
Scheme 3. Attempted Synthesis of the Fused R-Pyridone of
Huperzine A from Tricyclic Ketone 6
acid (3) by the method of Danishefsky.8 Etherification of 4
with trans-crotyl bromide in the presence of silver(I) oxide9
gave diene 5 which was irradiated in dichloromethane at
0 °C through Pyrex glass (Scheme 2). The resultant [2 þ 2]
photoadduct 6 was converted to enone 7 with the intention
of using this dienophile in a DielsꢀAlder cycloaddition
with azadiene 810 to fabricate the R-pyridone ring of 1, but
this reaction failed to produce more than trace amounts
of fused tetrahydropyridine 9. Consequently, a revised
plan was conceived for building a fused R-pyridone from
6 which first introduced a primary amine into the six-
membered ring and then assembled the heterocycle in a
stepwise sequence patterned after work by Magnus.11
Scheme 2. Synthesis of Tricyclic Ketone 6 from D-(ꢀ)-Quinic
Acid (3)
The revised route began with exposure of ketone 6 to
triisopropylsilyl chloride in the presence of a strong base to
furnish silyl enol ether 10, and this was reacted with iodoso-
benzene and trimethylsilyl azide to give azide 11 as a single
epimer (Scheme 3).12 The latter was reduced to amine 12
which was acylated with R-bromoacryloyl chloride to give
amide 13. Treatment of 13 with trimethylaluminum in
Figure 1. ORTEP representation of the X-ray crystal structure
of 14.
dichloroethane at 70 °C cleanly produced [2 þ 2] cycload-
duct 14 whose structure was secured by X-ray crystallo-
graphic analysis (Figure 1). Exposure of 14 to aqueous HF
in nitromethane caused silyl ether cleavage13 and subse-
quent retro-aldol fragmentation to give R-bromo lactam 15
as a 3:1 mixture of stereoisomers. Our expectation was
that the pyridine ring of 2 could be fashioned from 15 by
(8) Audia, J. E.; Boisvert, L.; Patten, A. D.; Villalobos, A.;
Danishefsky, S. J. J. Org. Chem. 1989, 54, 3738.
(9) Martin, S. F.; Rueger, H.; Williamson, S. A.; Grzejszczak, S.
J. Am. Chem. Soc. 1987, 109, 6124.
(10) (a) Boger, D. L. Tetrahedron 1983, 39, 2869. (b) Serckx-Poncin,
B.; Hesbain-Frisque, A.-M.; Ghosez, L. Tetrahedron Lett. 1982, 23,
3261.
(11) Magnus, P.; Rigollier, P.; Lacour, J.; Tobler, H. J. Am. Chem.
Soc. 1993, 115, 12629.
(12) Magnus, P.; Lacour, J. J. Am. Chem. Soc. 1992, 114, 767.
(13) The use of nitromethane as solvent in this reaction is critical to its
success. See: White. J. D.; Carter, R. G. Silyl Ethers. Science of Synthesis;
Thieme: Stuttgart, Germany, 2002; Vol. 4, pp 371ꢀ412.
Org. Lett., Vol. 15, No. 4, 2013
883

dehydrobromination to an R,β-unsaturated lactam which
could then undergo further oxidation to a pyridone. How-
ever, treatment of 15 with a variety of bases led exclusively
to fused cyclopropane 16 resulting from intramolecular
alkylation rather than the desired 1,2-elimination.
Scheme 4. Synthesis of Fused Methoxypyridine 22 from
Azide 11
Recourse to R-phenylselenylacrylic acid (17)14 circum-
vented the problem that produced 16. By converting azide
11 via amine 12 to amide 18 with 17 in the presence of
3,5-dinitrobenzoyl chloride and then reacting 18 with
trimethylaluminum, [2 þ 2] cycloadduct 19 was obtained
in a consistent yield of 65ꢀ70% (Scheme 4). As expected,
19 underwent silyl ether cleavage and retro-aldol fragmen-
tation with aqueous fluoride to furnish a separable mixture
of stereoisomeric R-selenyl δ-lactams 20.15 Oxidation of
this mixture with sodium periodate then led directly to R-
pyridone 21 in good yield.16 Methylation of 21 with methyl
iodide in the presence of silver carbonate furnished meth-
oxypyridine 22.
Our next task, reductive cleavage of the pyridylic ether
22, proved more troublesome than expected. Although the
CꢀO bond in 22 is activated by the adjacent methoxypyr-
idine, attempted hydrogenolysis of this ether led mainly to
reduction of the ketone with no evidence of CꢀO scission.
Samarium diiodide and other electron transfer reductants
gave similar results. However, when 22 was exposed to
activated zinc dust17 in methanol containing 0.2 M sodium
hydroxide, alcohol 23 was formed in excellent yield
(Scheme 5). Hydroxy ketone 23 was oxidized with Dess-
Martin periodinane18 to keto aldehyde 24 which underwent
a selective Grignard reaction with methylmagnesium iodide
at the aldehyde carbonyl to yield hydroxy ketone 25 as a
1:1 mixture of diastereomers. Oxidation of this mixture of
alcohols gave a single diketone 26.
At this point, our blueprint specified methylenation of
the methyl ketone of diketone 26. This process was ex-
pected to be selective since the cyclic ketone in 26 was
presumed to be less reactive due its conjugation through
the pyridine to the remote methoxy substituent and there-
fore formally equivalent to the carbonyl in a vinylogous
ester. However, we had failed to anticipate the severe steric
hindrance to attack at the carbonyl of the methyl ketone
of 26 by virtue of its endo placement on the cis fused
cyclobutane in this congested structure. In the event,
olefination of 26 under Wittig conditions19 gave mixtures
in which 28 was formed in only 27% yield; under forcing
conditions, 28 was accompanied by the diene from
double ketone methylenation as well as the alkene from
sole methylenation at the cyclohexanone carbonyl. On
the other hand, it was recognized that enolization of
diketone 26 was more likely to occur at the methyl ketone
than at the cyclohexanone carbonyl, where an endo double
bond at the ring fusion in this bicyclo[4.2.0]octane would
impose significant strain. In accord with this proposition,
treatment of 26 with Comins reagent20 and a sterically
demanding base gave enol triflate 27 in good yield. A Stille
reaction of the triflate with tetrakistriphenylphosphinepal-
ladium and hexamethyldistannane21 produced 28 for which
X-ray crystallographic analysis confirmed that no epimer-
ization had taken place at the cyclobutane during its deri-
vation from diketone 26 (Figure 2).
Ketone 28 readily formed an oxime as well as imine
derivatives with substituted benzylamines, but these sub-
strates were uncooperative in the skeletal rearrangement
programmed in Scheme 1. Generally, imine derivatives of
28 underwent hydrolysis back to the parent ketone or were
decomposed in the presence of Lewis acids such as tita-
nium tetrachloride.
Finally, when 28 was treated with methyl carbamate
and anhydrous p-toluenesulfonic acid in hot benzene an
immediate reaction took place that passed transiently
through imine 29 and ended after rearrangement at 30.
The latter is the penultimate substance in Kozikowski’s
synthesis of huperzine A,22 and a sample of 30 was
prepared from natural huperzine A to confirm the identity
(14) Reich, H. J.; Willis, W. W.; Clark, P. D. J. Org. Chem. 1981, 46,
2775.
(15) The initial mixture of selenyl lactams was found to equilibrate in
the presence of excess HF to a 10:1 mixture favoring the β (3S)
stereoisomer.
(16) It is assumed that selenoxide elimination to an R,β-unsaturated
lactam is the first step and that this is followed by allylic oxidation with
excess periodate at the activated ring fusion to give 21.
(17) Newman, M. S.; Arens, F. J. J. Am. Chem. Soc. 1955, 77, 946.
(18) Boeckman, R. K., Jr.; Shao, P.; Mullins, J. J. Org. Synth. 2000,
77, 141.
(20) Comins, D. L.; Dehgani, A. Tetrahedron Lett. 1992, 32, 6299.
(21) Eschavarran, A. M.; Stille, J. K. J. Am. Chem. Soc. 1987, 109,
5478.
(22) Xia, Y.; Kozikowski, A. P. J. Am, Chem. Soc. 1989, 111, 4116. It
was shown in this work that the E and Z exocyclic ethylidene substi-
tuents attached to the bicyclo[3.3.1]nonene frame equilibrate under
relatively mild conditions and that the E isomer is thermodynamically
more stable.
(19) Maryanoff, B. E.; Reitz, A. B. Chem. Rev. 1989, 89, 863.
884
Org. Lett., Vol. 15, No. 4, 2013

Scheme 5. Completion of the Synthesis of (ꢀ)-Huperzine A (1)
from Methoxypyridine 22
Figure 2. ORTEP representation of the X-ray crystal structure
of 28.
(a) the bicyclo[3.3.1]nonene portion of the natural product
is created using an intramolecular aza-Prins reaction of an
isopropenyl substituted cyclobutane in conjunction with
stereoelectronically directed fragmentation of an interven-
ing cyclobutylcarbinyl cation and (b) the R-pyridone is
orchestrated via pentacycle 19 in a novel extrapolation of
Magnus’ chemistry.11 Although shorter and more efficient
routes to 1 exist, a useful paradigm suggested by this work
is that incorporation of ring strain in a synthesis design in
the form of a cyclobutane, along with appropriately placed
substituents, can provide access to structures of substantial
complexity.
Acknowledgment. We are grateful to Professor Bai
Donglu of Shanghai Institute of Materia Medica for a
sample of natural huperzine A and to Dr. Lev Zacharov,
University of Oregon, for X-ray crystal structures of 14
and 28. M.T. is indebted to the Swiss National Science
Foundation for a postdoctoral fellowship (Grant No.
PBEZ2-105115).
of our synthetic material. Thus, exposure of (ꢀ)-1 to
methyl chloroformate and potassium carbonate followed
by methyl iodide in the presence of silver(I) carbonate
produced material identical with the compound obtained
from 28. Treatment of 30 with trimethylsilyl iodide as
described by Kozikowski22 completed our route to (ꢀ)-1.
In summary, a synthesis of the acetylcholinesterase
inhibitor huperzine A (1) is demonstrated in which
Supporting Information Available. X-ray diffraction
data for 14 and 28; full characterization data and copies
of 1H and 13C NMR spectra for all compounds. This mate-
rial is available free of charge via the Internet at http://
pubs.acs.org.
The authors declare no competing financial interest.
Org. Lett., Vol. 15, No. 4, 2013
885