Divergent enantioselective synthesis of (-)-galanthamine and (-)-morphine

Article abstract of DOI:10.1021/ja054449+

An efficient divergent synthetic strategy for the synthesis of the opiate and amaryllidaceae alkaloids emerges by employing a Pd-catalyzed asymmetric allylic alkylation (AAA) to set the stereochemistry. Three generations of syntheses of galanthamine are discussed in detail with particular focus on the scope of the palladium-catalyzed AAA reactions and intramolecular Heck reactions. The pivotal tricyclic intermediate is available in six steps from 2-bromovanillin and the monoester of methyl 6-hydroxycyclohexene-1-carboxylate. This intermediate requires only two steps to convert to (-)-galanthamine. Using a Heck vinylation, we found that the fourth ring of codeine/morphine could be formed. The final ring formation involves a novel visible light-promoted hydroamination. Thus, six steps are required to convert the pivotal tricyclic intermediate into codeine, which has been demethylated in high yield to morphine.

Full text of DOI:10.1021/ja054449+

Published on Web 09/24/2005
Divergent Enantioselective Synthesis of (-)-Galanthamine
and (-)-Morphine
Barry M. Trost,* Weiping Tang, and F. Dean Toste
Contribution from the Department of Chemistry, Stanford UniVersity, Stanford, California 94305
Received July 5, 2005; E-mail: bmtrost@stanford.edu
Abstract: An efficient divergent synthetic strategy for the synthesis of the opiate and amaryllidaceae alkaloids
emerges by employing a Pd-catalyzed asymmetric allylic alkylation (AAA) to set the stereochemistry. Three
generations of syntheses of galanthamine are discussed in detail with particular focus on the scope of the
palladium-catalyzed AAA reactions and intramolecular Heck reactions. The pivotal tricyclic intermediate is
available in six steps from 2-bromovanillin and the monoester of methyl 6-hydroxycyclohexene-1-carboxylate.
This intermediate requires only two steps to convert to (-)-galanthamine. Using a Heck vinylation, we
found that the fourth ring of codeine/morphine could be formed. The final ring formation involves a novel
visible light-promoted hydroamination. Thus, six steps are required to convert the pivotal tricyclic intermediate
into codeine, which has been demethylated in high yield to morphine.
reported using biomimetic oxidative phenol coupling17-31 to
Introduction
create the critical spiro quaternary carbon of narwedine (2),
Galanthamine (1; Figure 1),1,2 the parent member of the
galanthamine-type amaryllidaceae alkaloids (e.g. 1-3), is a
centrally acting reversible inhibitor of acetylcholinesterase
(Ache), which significantly enhances cognitive functions of
Alzheimer’s patients.3-8 It was first approved in Austria and
most recently in the rest of Europe and United States for the
treatment of Alzheimer’s disease. In the endeavor of searching
for more potent inhibitors of Ache, there is considerable interest
in derivatives which are based on (-)-galanthamine as a lead
structure9,10 since (-)-galanthamine is less toxic than other Ache
inhibitors, such as physostigmine and tacrine.9,11-13 Among
them, galanthamine derivatives 4 12-14 and its iminium salt,
synthesized by selective N-demethylation followed by N-
alkylation of galanthamine,15 were found to be more potent (up
to 70-fold) than galanthamine in inhibiting Ache.
which is converted into 1 by diastereoselective reduction.16 We
disclosed the first enantioselective synthesis of (-)-1 using a
sequential palladium-catalyzed asymmetric allylic alkylation
(AAA) and intramolecular Heck cyclization.32,33 At the same
time, several other groups utilized similar Heck cyclizations to
construct the quaternary carbon center of (()-3-deoxygalan-
thamine34,35 and (()-galanthamine 1.36 Up to now, however,
syntheses employing the oxidative phenol coupling step have
(16) Shieh, W.-C.; Carlson, J. A. J. Org. Chem. 1994, 59, 5463.
(17) Kueenburg, B.; Czollner, L.; Froehlich, J.; Jordis, U. Org. Process Res.
DeV. 1999, 3, 425.
(18) Barton, D. H. R.; Kirby, G. W. J. Chem. Soc. 1962, 806.
(19) Kametani, T.; Yamaki, K.; Yagi, H.; Fukumoto, K. J. Chem. Soc. D 1969,
425.
(20) Kametani, T.; Yamaki, K.; Yagi, H.; Fukumoto, K. J. Chem. Soc. C 1969,
2602.
(21) Kametani, T.; Yamaki, K.; Terui, T. J. Heterocycl. Chem. 1973, 10, 35.
(22) Shimizu, K.; Tomioka, K.; Yamada, S.; Koga, K. Heterocycles 1977, 8,
277.
Because of the limited supplies and the high cost of its
isolation from natural sources,16,17 several syntheses have been
(23) Shimizu, K.; Tomioka, K.; Yamada, S.; Koga, K. Chem. Pharm. Bull. 1978,
26, 3765.
(1) Hoshino, O. The Alkaloids; Academic Press: New York, 1998; Vol. 51.
(2) Martin, S. F. The Alkaloids; Academic Press: New York, 1987; Vol. 30.
(3) Weinstock, M. CNS Drugs 1999, 12, 307.
(24) Krikorian, D.; Vlahov, R.; Parushev, S.; Chinova, M.; Vlahov, I.; Schaefer,
H. J.; Duddeck, H.; Snatzke, G. Tetrahedron Lett. 1984, 25, 2969.
(25) Vlahov, R.; Krikorian, D.; Spassov, G.; Chinova, M.; Vlahov, I.; Parushev,
S.; Snatzke, G.; Ernst, L.; Kieslich, K.; Abraham, W. R.; Sheldrick, W. S.
Tetrahedron 1989, 45, 3329.
(4) Thatte, U. Curr. Opin. Cent. Peripher. NerV. Syst. InVest. Drugs 1999, 1,
357.
(5) Unni, K. CNS Drugs 1998, 10, 447.
(6) Nordberg, A.; Svensson, A.-L. Drug Saf. 1998, 19, 465.
(7) Rainer, M. Drugs Today 1997, 33, 273.
(26) Szewczyk, J.; Wilson, J. W.; Lewin, A. H.; Carroll, F. I. J. Heterocycl.
Chem. 1995, 32, 195.
(8) Mucke, H. A. M. Drugs Today 1997, 33, 251.
(9) Han, S. Y.; Sweeney, J. E.; Bachman, E. S.; Schweiger, E. J.; Forloni, G.;
Coyle, J. T.; Davis, B. M.; Joullie, M. M. Eur. J. Med. Chem. 1992, 27,
673.
(10) Bores, G. M.; Kosley, R. W., Jr. Drugs Future 1996, 21, 621.
(11) Han, S. Y.; Mayer, S. C.; Schweiger, E. J.; Davis, B. M.; Joullie, M. M.
Bioorg. Med. Chem. Lett. 1991, 1, 579.
(12) Mary, A.; Renko, D. Z.; Guillou, C.; Thal, C. Bioorg. Med. Chem. 1998,
6, 1835.
(13) Guillou, C.; Mary, A.; Renko, D. Z.; Gras, E.; Thal, C. Bioorg. Med. Chem.
Lett. 2000, 10, 637.
(14) Jordis, U.; Froehlich, J.; Treu, M.; Hirnschall, M.; Czollner, L.; Kaelz, B.;
Welzig, S. (Sanochemia Pharmazeutika Aktiengesellschaft, Austria) PCT
Int. Appl. Wo, 2001, p 285.
(15) Mary, A.; Renko, D. Z.; Guillou, C.; Thal, C. Tetrahedron Lett. 1997, 38,
5151.
(27) Chaplin, D. A.; Fraser, N.; Tiffin, P. D. Tetrahedron Lett. 1997, 38, 7931.
(28) Czollner, L.; Frantsits, W.; Kenburg, B.; Hedenig, U.; Frohlich, J.; Jordis,
U. Tetrahedron Lett. 1998, 39, 2087.
(29) Kita, Y.; Arisawa, M.; Gyoten, M.; Nakajima, M.; Hamada, R.; Tohma,
H.; Takada, T. J. Org. Chem. 1998, 63, 6625.
(30) Krikorian, D.; Tarpanov, V.; Parushev, S.; Mechkarova, P. Synth. Commun.
2000, 30, 2833.
(31) Node, M.; Kodama, S.; Hamashima, Y.; Baba, T.; Hamamichi, N.; Nishide,
K. Angew. Chem., Int. Ed. 2001, 40, 3060.
(32) Trost, B. M.; Toste, F. D. J. Am. Chem. Soc. 2000, 122, 11262.
(33) Trost, B. M.; Tang, W. Angew. Chem., Int. Ed. 2002, 41, 2795.
(34) Pilger, C.; Westermann, B.; Floerke, U.; Fels, G. Synlett 2000, 1163.
(35) Parsons, P. J.; Charles, M. D.; Harvey, D. M.; Sumoreeah, L. R.; Shell,
A.; Spoors, G.; Gill, A. L.; Smith, S. Tetrahedron Lett. 2001, 42, 2209.
(36) Guillou, C.; Beunard, J.-L.; Gras, E.; Thal, C. Angew. Chem., Int. Ed. 2001,
40, 4745.
9
10.1021/ja054449+ CCC: $30.25 © 2005 American Chemical Society
J. AM. CHEM. SOC. 2005, 127, 14785-14803
14785

A R T I C L E S
Trost et al.
Figure 1. Naturally occurring galanthamine-type amaryllidaceae alkaloids and their derivatives.
studies suggesting that morphine may play a general role in
immune, vascular, and nervous systems of mammals.49-54
The structure-activity relationships of morphine analogues
have been studied extensively.37,55-59 Minor modification of
morphine can yield opiates with altered receptor affinities and
fundamentally different pharmacologies. Oxymorphone (8a),
etorphine (9), and buprenorphine (10) are more potent µ-agonists
than morphine.60,61 Structurally related morphine antagonists (i.e.
naloxone (8b), naltrexone (8c), and nalmefene (11)) are used
for the treatment of alcohol abuse and eating disorders.37,55
However, very few compounds made to date from modifications
around the phenanthrene structure of morphine have exceeded
the pain control properties of morphine without a concomitant
increase in addiction potential.62 Most of these synthetic opiates
(Figure 3) are derived from semisyntheses, whereby the natural
opium alkaloids are isolated and modified. Clearly, an efficient
enantioselective synthesis of morphine will open the way to
the preparation of a variety of new morphine analogues.
Figure 2. Naturally occurring opium alkaloids.
not been reported to be able to allow direct access to galan-
thamine derivative 4.
(-)-Morphine (5; Figure 2) is one of the first medicines used
by humankind. Since the first documented use of opium in
1500BC, its impact on society has been extremely significant.37-41
This significance rests on the potency of the active principles,
notably, the major one, morphine, which gives it its analgesic,
euphoric, and potentially addictive properties. Morphine 5 is
the major component of opium, representing 10-15% of its
dry weight. Other morphine derivatives, such as codeine (6;
3-4%) and thebaine (7; 0.1-2%), are also present in opium.
The majority of isolated morphine is used to manufacture
codeine 6 by simple phenolic methylation.37 To date, no
synthetic drugs or naturally occurring compound has been found
which can match the broad spectrum analgesic properties of
opium alkaloids. (-)-Morphine 5, (-)-codeine 6, (-)-thebaine
7, and simpler morphinan and benzomorphan structural ana-
logues42 have broad spectrum pharmacological properties (an-
titusive, analgesic, and sedative, etc) and are employed in diverse
therapies.
Divergent Synthetic Plan for Opium Alkaloids and
Galanthamine-Type Amaryllidaceae Alkaloids
Morphine 5 and galanthamine 1 share the same tricyclic core,
and biosynthetically both derive from oxidative phenol couplings
(Scheme 1 and Scheme 2).37,39,44,63-65 These processes were
initially proposed on theoretical grounds by Barton.18,66 It has
been shown that the 4′-O-methylnorbelladine (12) was the
precursor for the production of N-demethylnarwedine (15),
presumably through spontaneous cyclization of a hypothetical
dienone 14 generated from the oxidative phenol coupling
Recent investigations have shown that morphine and its
cogeners are produced not only in plants but also in mammalian
organisms along essentially analogous pathways.43-47 It has been
demonstrated that human cells can synthesize morphine, and it
is present in human cells at a nanomolar level.48 The precise
role of endogeneous morphine is not clear. However, there are
(49) Suzuki, S.; Chuang, L. F.; Doi, R. H.; Chuang, R. Y. Exp. Cell Res. 2003,
291, 340.
(50) Chuang, L. F.; Killam, K. F.; Chuang, R. Y. J. Biol. Chem. 1997, 272,
26815.
(51) Miyagi, T.; Chuang, L. F.; Doi, R. H.; Carlos, M. P.; Torres, J. V.; Chuang,
R. Y. J. Biol. Chem. 2000, 275, 31305.
(52) Miyagi, T.; Chuang, L. F.; Lam, K. M.; Kung, H. F.; Wang, J. M.; Osburn,
B. I.; Chuang, R. Y. Immunopharmacology 2000, 47, 53.
(53) Suzuki, S.; Chuang, L. F.; Yau, P.; Doi, R. H.; Chuang, R. Y. Exp. Cell
Res. 2002, 280, 192.
(54) Yin, D. L.; Mufson, R. A.; Wang, R.; Shi, Y. F. Nature 1999, 397, 218.
(55) Harris, L. S. Narc. Drugs: Biochem. Pharmacol. 1971, 89.
(56) Fries, D. S. In Principlies of Medicinal Chemistry; 4th ed.; Foye, W. O.,
Lemke, T. L., Williams, D. A., Eds.; Williams and Wilkins: Baltimore,
MD, 1995; p 247.
(57) Gentilucci, L. Curr. Top. Med. Chem. 2004, 4, 19.
(58) Valverde, O.; Maldonado, R.; Kieffer, B. L. CNS Drugs 1998, 10, 1.
(59) Newman, A. H.; Rice, K. C.; Jacobson, A. E. NIDA Res. Monogr. 1992,
119, 54.
(37) Hudlicky, T.; Butora, G.; Fearnley, S. P.; Gum, A. G.; Stabile, M. R. Stud.
Nat. Prod. Chem. 1996, 18, 43.
(38) Blakemore, P. R.; White, J. D. Chem. Commun. 2002, 1159.
(39) Novak, B. H.; Hudlicky, T.; Reed, J. W.; Mulzer, J.; Trauner, D. Curr.
Org. Chem. 2000, 4, 343.
(40) Rice, K. C. In The Chemistry and Biology of Isoquinoline Alkaloids;
Phillipson, J. D., Roberts, M. R., Zenk, M. H., Eds.; Springer-Verlag:
Berlin, 1985; p 191.
(41) Taber, D. F.; Neubert, T. D.; Rheingold, A. L. J. Am. Chem. Soc. 2002,
124, 12416.
(42) Trost, B. M.; Tang, W. P. J. Am. Chem. Soc. 2003, 125, 8744.
(43) Oka, K.; Kantrowitz, J. D.; Spector, S. Proc. Natl. Acad. Sci. U.S.A. 1985,
82, 1852.
(60) Ananthan, S.; Khare, N. K.; Saini, S. K.; Seitz, L. E.; Bartlett, J. L.; Davis,
P.; Dersch, C. M.; Porreca, F.; Rothman, R. B.; Bilsky, E. J. J. Med. Chem.
2004, 47, 1400.
(44) Kutchan, T. M. In The Alkaloids: Chemistry and Biology; Cordell, G. A.,
Ed.; San Diego, 1997; Vol. 50, p 257.
(61) Ferris, C. D.; Hirsch, D. J.; Brooks, B. P.; Snyder, S. H. J. Neurochem.
1991, 57, 729.
(45) Hazum, E.; Sabatka, J. J.; Chang, K. J.; Brent, D. A.; Findlay, J. W. A.;
Cuatrecasas, P. Science 1981, 213, 1010.
(46) Donnerer, J.; Oka, K.; Brossi, A.; Rice, K. C.; Spector, S. Proc. Natl. Acad.
Sci. U.S.A. 1986, 83, 4566.
(47) Kodaira, H.; Spector, S. Proc. Natl. Acad. Sci. U.S.A. 1988, 85, 1267.
(48) Poeaknapo, C.; Schmidt, J.; Brandsch, M.; Drager, B.; Zenk Meinhart, H.
Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 14091.
(62) Jaffe, J. H.; O’keeffe, C. Drug Alcohol Depend. 2003, 70, S3.
(63) Eichhorn, J.; Takada, T.; Kita, Y.; Zenk, M. H. Phytochemistry 1998, 49,
1037.
(64) Frenzel, T.; Zenk, M. H. Phytochemistry 1990, 29, 3505.
(65) Stadler, R.; Zenk, M. H. Liebigs Ann. Chem. 1990, 555.
(66) Barton, D. H. R.; Cohen, T. Festschr. Prof. Dr. Arthur Stoll Siebzigsten
Geburtstag 1957, 117.
9
14786 J. AM. CHEM. SOC. VOL. 127, NO. 42, 2005

(−)-Galanthamine and (−)-Morphine Synthesis
A R T I C L E S
Figure 3. Semisynthetic opiates.
Scheme 1. Biosynthesis of Galanthamine via Oxidative Bisphenol Coupling
Scheme 2. Biosynthesis of Morphine via Oxidative Bisphenol Coupling
(Scheme 1).63 In plants, the intramolecular ortho-para bis-
phenolic coupling of (R)-reticuline (16) is catalyzed by
salutaridine synthase, a microsomal NADPH-dependent cyto-
chrome P-450 enzyme. Cyclization of salutaridine (18) to
thebaine 7 requires two more enzymes (Scheme 2).67-71
The biosyntheses of galanthamine and morphine have
desired para-ortho coupling is complicated by three other
possible couplings (ortho-ortho, ortho-para, para-para). How-
ever, some bioanalogous syntheses involving Grewe-type cy-
clization catalyzed by acid83-95 or palladium96,97 have led to
several successful syntheses.
The distinct difference between the oxidative phenol coupling
precursors (bisphenols 12 and 16) seems to prevent any
possibility of accessing these two families of alkaloids from
the same advanced intermediate through a biomimetic approach.
Indeed, no divergent approach to both families of alkaloids has
been realized despite their biogenetic similarity and significant
biological activities. Furthermore, despite the extensive synthetic
work devoted to these two families of alkaloids and recent
inspired
many
biomimetic
syntheses
of
both
amaryllidaceae16-18,20,21,24-29,72-76 and opium37,39,77-82 alkaloids.
The biomimetic synthesis of galanthamine involving a dynamic
resolution has been performed on a pilot scale.17 The biomimetic
synthesis of morphine generally suffered from low yields due
to low selectivity of the bisphenol oxidative coupling. The
(67) Zenk, M. H.; Gerardy, R.; Stadler, R. J. Chem. Soc., Chem. Commun. 1989,
1725.
(68) Stadler, R.; Zenk, M. H. J. Biol. Chem. 1993, 268, 823.
(69) Gerardy, R.; Zenk, M. H. Phytochemistry 1993, 32, 79.
(70) Amann, T.; Zenk, M. H. Tetrahedron Lett. 1991, 32, 3675.
(71) Hirata, K.; Poeaknapo, C.; Schmidt, J.; Zenk, M. H. Phytochemistry 2004,
65, 1039.
(83) Grewe, R.; Fischer, H. Chem. Ber. 1963, 96, 1520.
(84) Grewe, R.; Fischer, H.; Friedric, W. Chem. Ber. 1967, 100, 1.
(85) Grewe, R.; Friedric, W. Chem. Ber. 1967, 100, 1550.
(86) Morrison, G. C.; Waite, R. O.; Shavel, J. Tetrahedron Lett. 1967, 4055.
(87) Kametani, T.; Ihara, M.; Fukumoto, K.; Yagi, H. J. Chem. Soc. C 1969,
2030.
(72) Kametani, T.; Seino, C.; Yamaki, K.; Shibuya, S.; Fukumoto, K.; Kigasawa,
K.; Satoh, F.; Hiiragi, M.; Hayasaka, T. J. Chem. Soc. C 1971, 1043.
(73) Kametani, T.; Shishido, K.; Hayashi, E.; Seino, C.; Kohno, T.; Shibuya,
S.; Fukumoto, K. J. Org. Chem. 1971, 36, 1295.
(88) Beyerman, H. C.; Buurman, E.; Lie, T. S.; Maat, L. Recl. TraV. Chim.
Pays-Bas-J. R. Neth. Chem. Soc. 1976, 95, 43.
(89) Lie, T. S.; Maat, L.; Beyerman, H. C. Recl. TraV. Chim. Pays-Bas 1979,
98, 419.
(74) Kametani, T.; Yamaki, K.; Terui, T.; Shibuya, S.; Fukumoto, K. J. Chem.
Soc., Perkin Trans. 1 1972, 1513.
(90) Beyerman, H. C.; Vanberkel, J.; Lie, T. S.; Maat, L.; Wessels, J. C. M.;
Bosman, H. H.; Buurman, E.; Bijsterveld, E. J. M.; Sinnige, H. J. M. Recl.
TraV. Chim. Pays-Bas 1978, 97, 127.
(75) Szewczyk, J.; Lewin, A. H.; Carroll, F. I. J. Heterocycl. Chem. 1988, 25,
1809.
(91) Beyerman, H. C.; Lie, T. S.; Maat, L.; Bosman, H. H.; Buurman, E.;
Bijsterveld, E. J. M.; Sinnige, H. J. M. Recl. TraV. Chim. Pays-Bas 1976,
95, 24.
(76) Holton, R. A.; Sibi, M. P.; Murphy, W. S. J. Am. Chem. Soc. 1988, 110,
314.
(77) Barton, D. H. R.; Kirby, G. W.; Steglich, W.; Thomas, G. M. Proc. Chem.
Soc., London 1963, 203.
(92) Beyerman, H. C.; Vanleeuwen, F. F.; Lie, T. S.; Maat, L.; Olieman, C.
Recl. TraV. Chim. Pays-Bas 1976, 95, 238.
(78) Barton, D. H. R.; Bhakuni, D. S.; James, R.; Kirby, G. W. J. Chem. Soc.
C 1967, 128.
(93) Hoogwate, D. A.; Reinhoud, D. N.; Lie, T. S.; Gunneweg, J. J.; Beyerman,
H. C. Recl. TraV. Chim. Pays-Bas 1973, 92, 819.
(79) Schwartz, M. A.; Mami, I. S. J. Am. Chem. Soc. 1975, 97, 1239.
(80) Schwartz, M. A.; Zoda, M. F. J. Org. Chem. 1981, 46, 4623.
(81) Schwartz, M. A.; Pham Phuong Thi, K. J. Org. Chem. 1988, 53, 2318.
(82) White, J. D.; Caravatti, G.; Kline, T. B.; Edstrom, E.; Rice, K. C.; Brossi,
A. Tetrahedron 1983, 39, 2393.
(94) Rice, K. C. J. Org. Chem. 1980, 45, 3135.
(95) Ludwig, W.; Schaefer, H. J. Angew. Chem. 1986, 98, 1032.
(96) Hong, C. Y.; Kado, N.; Overman, L. E. J. Am. Chem. Soc. 1993, 115,
11028.
(97) Hong, C. Y.; Overman, L. E. Tetrahedron Lett. 1994, 35, 3453.
9
J. AM. CHEM. SOC. VOL. 127, NO. 42, 2005 14787

A R T I C L E S
Trost et al.
Scheme 3. Divergent Synthetic Plan
presence of a catalytic amount of iron afforded bromophenol
26.102 Alternatively, reaction of isovanillin with iodine mono-
chloride produced o-iodophenol 27.102
In accordance with our retrosynthetic analysis, the 2-allyl-
cyclohexenol (30; Scheme 6) was prepared since the allyl group
could be readily converted to the necessary aldehyde by
oxidative cleavage. Cyclohexenol 30 prepared by Birch reduc-
tion103 followed by Luche reduction104 was transformed into
the required allyl carbonate (31 and 32) and more reactive allyl
chloride 33 using standard procedures. Chemoselective osmium
tetroxide catalyzed dihydroxylation of carbonate 32 afforded a
1:1 mixture of diastereomeric diols, which was readily protected
as the acetonide 34 by treatment with dimethoxymethane in the
presence of catalytic p-toluenesulfonic acid.
progress in the development of enantioselective reactions, only
one asymmetric synthesis of galanthamine 1 98 and two of
morphine 5 96,99 without using resolution were reported prior
to our investigation.32,33,100 The similarity of galanthamine 1
and morphine 5, as highlighted in Scheme 1 and Scheme 2,
arising from oxidative phenol coupling followed by formation
of an ether bridge, prompted us to examine the possibility of
accessing these products from a common intermediate through
chemical synthesis. We envisioned that both alkaloid families
could be derived from a common intermediate 19 (Scheme 3),
bearing the common tricyclic core of both families of alkaloids
and appropriately functionalized side chains (R1 and R2) that
can be transformed to the additional rings of galanthamine 1
and morphine 5. Stereoselective formations of two distinct allylic
alcohols in galanthamine 1 and morphine 5 from a simple olefin
in common intermediate 19 requires careful synthetic planning
and conformational analysis.
Alternatively, the cyclohexene R-group of intermediate 22
could be an ester. A practical synthesis of esters 38 and 39
utilized the Horner-Wadsworth-Emmons reaction of dialde-
hyde.32,105 Reaction of glutaraldehyde 35 with trimethyl phos-
phonoacetate (36) in aqueous potassium carbonate produced
alcohol 37 in 46% yield (Scheme 6). Conversion of alcohol 37
to its methyl ester 38 or trichloroethyl carbonate (Troc) 39 was
accomplished by treatment with corresponding chloroformate.
Direct treatment of crude alcohol 37 with trichloroethyl
chloroformate provided the desired carbonate 39 with a slightly
higher yield (54%).33
Since there are very few examples of palladium-catalyzed
allylic alkylation utilizing ortho-disubstituted phenols,106,107 the
AAA of iodiophenol 27 with cyclohexene methyl carbonate (40)
was investigated as a test system (Table 1). In the presence of
the standard chiral ligand 42, phenol 27 showed good reactivity
producing aryl ether 41 in 75% yield, however, with lower
enantioselectivity (66%ee) than that observed in the palladium-
catalyzed AAA of simpler phenols and carbonate 40.106,107
Confident that phenol 27 would participate as a nucleophile,
more complicated carbonates were investigated and the enan-
tioselectivity of this model reaction was not optimized further.
First Generation Synthesis of Galanthamine
The retrosynthetic analysis of galanthamine begins with
removal of the C3 hydroxyl group of galanthamine furnishing
3-deoxygalanthamine (20; Scheme 4), which may be directly
converted to the natural product by allylic oxidation.32,101
Tertiary amine 20 could in turn be prepared from dialdehyde
21 by a reductive amination with methylamine. An intramo-
lecular Heck reaction was envisioned to forge the quaternary
center in the dialdehydes 21. This strategy has the potential
pitfall that palladium-catalyzed ionization of the phenols may
be competitive with the Heck cyclization. To avoid problems
of isomerization of the olefin into conjugation with the aldehyde,
aryl ether 22 would have an R-group which could readily be
converted into the required aldehyde of 21 after the Heck
reaction. Finally, the enantioselective synthesis of aryl ether 22,
from which the remaining stereocenters of galanthamine would
derive, would be accomplished by a palladium-catalyzed AAA
of an ortho-substituted halophenol 24 with cyclohexenyl carbon-
ate 23.
The 2-allylcyclohexenyl carbonates were the first set of
substrates examined in the palladium-catalyzed AAA of iodo-
phenol 27 (Table 1). The palladium-catalyzed reaction of phenol
27 with methyl carbonate 31 did not produce any of the desired
aryl ether even when the reaction mixture was heated at 40 °C.
Similarly, the more reactive carbonate 32 did not react with
phenol 27 under the typical conditions of the palladium-
catalyzed AAA. Finally, switching to allylic chloride 33 did
produce some of the desired ether 43 in poor yield and with no
enantioselectivity. Carrying out the reaction in the absence of
the palladium catalyst indicated that aryl ether 43 was formed
The o-iodo and -bromophenols are both readily prepared from
isovanillin (25; Scheme 5). Bromination of isovanillin in the
Scheme 4. Retrosynthetic Analysis of (-)-Galanthamine
9
14788 J. AM. CHEM. SOC. VOL. 127, NO. 42, 2005

(−)-Galanthamine and (−)-Morphine Synthesis
A R T I C L E S
Scheme 5. Preparation of the o-Halophenolsa
ester 38 were examined as substrates (Table 1). Palladium-
catalyzed reaction of iodophenol 27 and carbonate 34 failed to
produce any of the desired cyclohexenyl ether with both tri-
(dibenzylidideneacetone)palladium(0) and bis[π-allylpalladium-
(II) chloride] as catalyst precursors. Employing ester 38 as a
substrate also resulted in unsuccessful reaction with phenol
27.
a Conditions: (a) Br2, NaOAc, HOAc, Fe, rt; (b) ICl, pyr.
The failure of 2-substituted cyclohexenyl carbonates to
participate in palladium-catalyzed AAA with iodophenol 27
prompted us to reexamine their reactivity with simple 4-meth-
oxyphenol (44; Scheme 7). 2-Allyl-substituted carbonate 31 did
as a result of a slow uncatalyzed reaction. We postulated
that the poor reactivity of 2-allylcyclohexenyl carbonates
may result from the 1,4-diene acting as a ligand for
palladium. To prevent this coordination, acetonide 34 and
Scheme 6. Synthesis of 2-Substituted Cyclohexenyl Carbonatesa
a Conditions: (a) (i) Na, NH3/THF; (ii) allyl bromide; (iii) H2SO4, H2O; (b) NaBH4, CeCl3, MeOH/CH2Cl2; (c) CH3O2CCl, DMAP Pyr, CH2Cl2; (d)
Troc-Cl, DMAP, Pyr, CH2Cl2; (e) CH3SO2Cl, Pyr, CH2Cl2; (f) 4mol % OsO4, NMO, CH2Cl2, rt, 4 h; (g) (CH3O)2CH2, TsOH, acetone, rt, 1 h; (h) K2CO3,
H2O, 2 days.
Table 1. Palladium-Catalyzed AAA of Iodophenol 27 with Cyclohexenyl Carbonate or Chloridea
a Conditions: (a) 1 mol % Pd2dba3, 3 mol % (R,R)-42, CH2Cl2; (b) 1 mol % [(η3-C3H5)PdCl]2, 3 mol % (R,R)-42, TEA, CH2Cl2, 40 °C.
9
J. AM. CHEM. SOC. VOL. 127, NO. 42, 2005 14789

A R T I C L E S
Trost et al.
Scheme 7. Pd-Catalyzed AAA of 4-Methoxyphenol with 2-Substituted Cyclohexenyl Carbonatesa
a Conditions: (a) 1 mol % Pd2dba3, 3 mol % (R,R)-42, Cs2CO3, CH2Cl2, 40 °C; (b) 1 mol % Pd2dba3, 3 mol % (R,R)-42, CH2Cl2, rt; (c) 1 mol % Pd2dba3,
3 mol % (R,R)-42, CH2Cl2, 0 °C; (d) 1 mol % [(η3-C3H5)PdCl]2, 3 mol % (R,R)-42, TEA, CH2Cl2, 0 °C; (e) (NH4)2Ce(NO3)6, CH3CN, H2O.
not react with phenol 44 under the typical palladium-catalyzed
AAA conditions. Carrying out the reaction in the presence of
base, cesium carbonate, and increasing the temperature to 40
°C still failed to produce any of the desired coupling product.
On the other hand, the palladium-catalyzed AAA of phenol 44
with ester 38 proceeded smoothly under the typical conditions,
affording aryl ether 45 in 81% yield and 77% ee. Lowering the
temperature to 0 °C increased the enantiomeric excess to 86%
without affecting the yield. The absolute stereochemistry of aryl
ether 45 was tentatively assigned by ceric ammonium nitrate
(CAN) deprotection of the 4-methoxyphenol to produce alcohol
37 (Scheme 7). Comparison of the rotation of methyl ester 37
to the known ethyl ester 47, prepared by enzymatic resolu-
tion,108,109 suggested that the stereochemistry of ether 45 is (R).
Surprisingly, this stereochemistry is opposite to the absolute
stereochemistry obtained when unsubstituted cyclic allyl carbon-
ates (e.g. 40) are used as electrophiles in the palladium-catalyzed
AAA of phenols with ligand 42 (Scheme 7).106,107
The absolute stereochemistry of 45 is also opposite to that
which would be predicted by the mnemonic established for
ligand 42.107,110,111 A possible explanation for this unexpected
reversal is shown in Scheme 8. In the unsubstituted cyclohexenyl
π-allyl palladium complex, the palladium cants away from the
C2-allyl carbon. Addition of phenol to the π-allyl palladium
complex occurs anti-periplanar to the palladium through the
open region of the chiral space (the right front quadrant depicted
in cartoon 48). In the case of ester-substituted cyclohexenyl
π-allyl palladium complex, coordination of the ester to the
palladium would result in canting of the C2 carbon toward the
palladium.112,113 If this is the case, for nucleophilic addition to
remain anti-periplanar to palladium and to occur through an
open region of the chiral ligand requires that the nucleophile
approach from the left rear quadrant of 48 rather than the front
right quadrant. In other words, for addition to occur from the
front quadrant, a 180° rotation with respect to the chiral ligand
of the π-allyl (see 49) is required. Nucleophilic addition to ester-
substituted cyclohexenyl π-allyl palladium complex, as depicted
in 49, results in the formation of opposite stereochemistry to
what is predicted from mnemonic. Interestingly, this reversal
in stereochemistry was not observed when acyclic 2-ester-
substituted allylic carbonate, acyclic 2-cyano-substituted allylic
carbonate, or cyclic 2-cyano-substituted allylic carbonate was
utilized as substrate for the AAA reaction of phenol with the
same ligand.114,115 In the case of cyclic 2-cyano-substituted
allylic carbonate, which is electronically similar to the ester,
the nitrile is geometrically constrained from coordinating to the
palladium metal. Similarly, acyclic 2-cyano-substituted allylic
carbonate underwent normal allylic alkylation through a syn-
π-allyl complex. Generally, the syn-π-allyl complex is more
stable than the anti-π-allyl complex due to unfavorable A 1,3
strain in the anti complex.116,117 However, A 1,2 steric interaction
destabilizes the 2-substituted syn-π-allyl complex, and the anti-
π-allyl become favored as the size of the 2-substituents increases
(Scheme 8). The absolute stereochemistry of the products from
the anti-π-allyl complex is opposite to that from the syn-π-
allyl complex. In the case of acyclic 2-ester-substituted allylic
carbonate, the combination effect of ester coordination to
(98) Tomioka, K.; Shimizu, K.; Yamada, S.; Koga, K. Heterocycles 1977, 6,
1752.
(99) White, J. D.; Hrnciar, P.; Stappenbeck, F. J. Org. Chem. 1997, 62, 5250.
(100) Trost, B. M.; Tang, W. J. Am. Chem. Soc. 2002, 124, 14542.
(101) Muxfeldt, H.; Schneider, R. S.; Mooberry, J. B. J. Am. Chem. Soc. 1966,
88, 3670.
(102) Toth, J. E.; Hamann, P. R.; Fuchs, P. L. J. Org. Chem. 1988, 53, 4694.
(103) Taber, D. F.; Gunn, B. P.; Chiu, I. C. Org. Synth. 1983, 61, 59.
(104) Amann, A.; Ourisson, G.; Luu, B. Synthesis 1987, 1002.
(105) Amri, H.; Rambaud, M.; Villieras, J. Tetrahedron 1990, 46, 3535.
(106) Trost, B. M.; Toste, F. D. J. Am. Chem. Soc. 1998, 120, 815.
(107) Trost, B. M.; Toste, F. D. J. Am. Chem. Soc. 1999, 121, 4545.
(108) Yamane, T.; Ogasawara, K. Synlett 1996, 925.
(109) Takano, S.; Yamane, T.; Takahashi, M.; Ogasawara, K. Tetrahedron:
Asymmetry 1992, 3.
(112) Brock, C. P.; Attig, T. G. J. Am. Chem. Soc. 1980, 102, 1319.
(113) Rulke, R. E.; Kliphuis, D.; Elsevier: C. J.; Fraanje, J.; Goubitz, K.;
Vanleeuwen, P. W. N. M.; Vrieze, K. J. Chem. Soc., Chem. Commun.
1994, 1817.
(114) Trost, B. M.; Tsui, H.-C.; Toste, F. D. J. Am. Chem. Soc. 2000, 122,
3534.
(115) Trost, B. M.; Machacek, M. R.; Tsui, H. C. J. Am. Chem. Soc. 2005,
127, 7014.
(110) Trost, B. M.; Bunt, R. C. J. Am. Chem. Soc. 1994, 116, 4089.
(111) Trost, B. M.; Van Vranken, D. L.; Bingel, C. J. Am. Chem. Soc. 1992,
114, 9327.
(116) Faller, J. W.; Tully, M. T. J. Am. Chem. Soc. 1972, 94, 2676.
(117) Faller, J. W.; Thomsen, M. E.; Mattina, M. J. J. Am. Chem. Soc. 1971,
93, 2642.
9
14790 J. AM. CHEM. SOC. VOL. 127, NO. 42, 2005

(−)-Galanthamine and (−)-Morphine Synthesis
A R T I C L E S
Scheme 8. Rationalization of Reversal in Enantioselectivity
Scheme 9. Pd-Catalyzed Allylic Alkylation of Phenol 25 and
Bromophenol 26a
o-brominated phenol 26 was examined as a nucleophile (Scheme
9). Palladium-catalyzed AAA reaction of bromophenol 26 with
unsubstituted cyclohexenyl carbonate 40 proceeded smoothly
to afford aryl ether 51 with excellent enantioselectivity.
Unfortunately, as was the case with iodophenol 27 (Table 1)
and 4-methoxyphenol 44 (Scheme 7), the palladium- catalyzed
reaction of bromophenol 26 with allyl-substituted carbonate 32
failed to produce any aryl ether. Similarly, bromophenol 26 did
not react with cyclohexenyl carbonate 34 even at 40 °C.
Unlike the reaction of iodophenol 27 (Table 1), palladium-
catalyzed AAA reaction of bromophenol 26 and 2-ester-
substituted carbonate 39 in the presence of the standard ligand
(R,R)-42 proceeded smoothly, affording aryl ether 52 in 60%
yield and 71% ee (entry 1, Table 2). The absolute stereochem-
istry of aryl ether 52 is assumed as shown on the basis of an
analogy to aryl ether 45, and this was confirmed by the optical
rotation of the synthetic natural (-)-galanthamine. Variation of
the ligand showed that the stilbene diamine ligand 55 provided
slightly higher enantioselectivity (entry 4). Switching the
palladium source to bis[π-allylpalladium(II) chloride] increased
the enantioselectivity to 85% (entry 5). Lowering the catalyst
loading to 1% of bis[π-allylpalladium(II) chloride] further
increased the enantiomeric excess of aryl ether 52 to 88% (entry
6).
a Conditions: (a) 1 mol % Pd2dba3, 3 mol % rac-42, CH2Cl2, rt; (b) 1
mol % [(η3-C3H5)PdCl]2, 3 mol % (R,R)-42, TEA, CH2Cl2, rt; (c) 2.5 mol
% Pd2dba3, 7.5 mol % (R,R)-42, TEA, CH2Cl2, 40 °C.
palladium metal and formation of unusual anti-π-allyl was
proposed to explain its difference from cyclic 2-ester-substituted
allylic carbonate.115
Having prepared the required chiral o-haloaryl ethers, we next
examined the intramolecular Heck reaction. Larock reported that
the Heck reaction of aryl allyl ethers often suffers from
competing palladium-catalyzed ionization of the aryloxy group.118
However, the Heck reaction of o-iodoaldehyde 41 (Table 1)
proceeded smoothly to afford an inseparable 3:1:1 mixture
benzofuran 56a:56b:56c products in 82% yield (Scheme 10).
The relative stereochemistry of the major product 56a was
assigned by NOE experiments. Since benzofuran 56a was not
useful for the galanthamine synthesis, this Heck reaction was
not optimized. Since we can prepare aryl ether 52 with good
enantioselectivity in good yield, we thus turned our attention
The successful reactions of 4-methoxyphenol 44 with ester-
substituted carbonate 38 (Scheme 7) and of iodophenol 27 with
unsubstituted carbonate 40 (Table 1) led us to conjecture that
the failure of iodophenol 27 to react with 38 (Table 1) is due to
steric interaction between the o-iodo substituent and the center
carbon of the π-allyl. Indeed, palladium-catalyzed reaction of
uniodinated vanillin 25 with carbonate 38 produced aryl ether
50 in 82% yield (Scheme 9). The successful reaction of
isovanillin 25 with carbonate 38 is consistent with the hypothesis
that the failure of iodophenol 27 to react with 2-ester-substituted
carbonate 38 is due to the presence of the sterically demanding
o-iodo group. Since bromide is significantly smaller than iodine,
(118) Larock, R. C.; Stinn, D. E. Tetrahedron Lett. 1988, 29, 4687.
9
J. AM. CHEM. SOC. VOL. 127, NO. 42, 2005 14791

A R T I C L E S
Trost et al.
Table 2. Palladium-Catalyzed AAA Reaction of Bromophenol 26 and Carbonate 39
entry
Pd source
liganda
yield (%)
ee (%)
1
2
3
4
5
6
2.5 mol % Pd2dba3
2.5 mol % Pd2dba3
2.5 mol % Pd2dba3
7.5 mol % (R,R)-42
7.5 mol % (R,R)-53
7.5 mol % (S,S)-54
7.5 mol % (S,S)-55
7.5 mol % (S,S)-55
3 mol % (S,S)-55
60
61
27
64
73
72
(+)-71
(+)-47
0
(-)-77
(-)-85
(-)-88
2.5 mol % Pd2dba3
2.5 mol % [(η3-C3H5)PdCl]2
1 mol % [(η3-C3H5)PdCl]2
Scheme 10. Intramolecular Heck Reaction of Cyclohexyl Aryl Ether 41 and 52
to the Heck reaction of chiral aryl ether 52 (Scheme 10). The
reaction conditions utilized for the Heck reaction of unsubsti-
tuted cyclohexenyl ether 41 resulted exclusively in ionization
of the phenoxy moiety to produce phenol 26. Larock had
reported that the use of Jeffery-type conditions favored the Heck
cyclization over ionization.118 However, even under Jeffrey
conditions (10% Pd(OAc)2, KOAc, (Bu)4NCl, dimethylforma-
mide (DMF)) no Heck product was observed. Clearly, the
presence of the ester on the olefin is detrimental to the Heck
reaction. It should retard the rate of the reaction for both steric
(formation of a quaternary versus a tertiary center) and electronic
(a contraelectronic carbopalladation is required) reasons. Since
the synthesis of galanthamine requires the formation of a
quaternary center, the former cannot be avoided. However, the
electronic nature of the olefin can be readily changed by
reduction of the ester to the corresponding alcohol. The aromatic
aldehyde can impose competing effects on the course of the
Heck reaction. The presence of the electron-withdrawing
aldehyde should increase the rate of oxidative addition of
palladium(0) into the aryl halide bond. However, the presence
of an electron-withdrawing group should also aid in the
competing ionization reaction by stabilization of the phenoxide
leaving group. We therefore wanted to examine substrates with
the aldehyde present and the aldehyde reduced to the benzyl
alcohol.
To this end, aldehyde 52 was protected as its dimethylacetal
57 by treatment with trimethyl orthoformate and catalytic
p-toluenesulfonic acid (Scheme 11). Reduction of ester 57 with
di-isobutylaluminum hydride (DIBAL-H) at -78 °C, followed
by treatment of the crude alcohol with a catalytic amount of
p-toluenesulfonic acid in aqueous methanol afforded hydroxyl
aldehyde 58. Alternatively, both the ester and the aldehyde were
reduced simultaneously with DIBAl-H to furnish diol 59.
Having reduced the ester, we examined the Heck reaction of
the hydroxymethyl-substituted cyclohexenes 58 and 59. Reaction
of aldehyde 58 under standard Jeffrey conditions resulted in
exclusive ionization to produce phenol 26. Attempted in situ
protection of the hydroxyl group, with N,O-bis(trimethyl-
silylacetamide) (BSA), also resulted in rapid conversion of 58
to phenol 26. At this point, it appeared that the presence of the
electron-withdrawing aldehyde on the phenol resulted in greater
acceleration of the ionization than the Heck reaction under the
Jeffrey condition. Therefore, our attention turned to the Heck
reaction of benzyl alcohol 59. Unfortunately, reaction of 59
9
14792 J. AM. CHEM. SOC. VOL. 127, NO. 42, 2005

(−)-Galanthamine and (−)-Morphine Synthesis
A R T I C L E S
Scheme 11. Preparation of Cyclohexenyl Alcohols for the Heck Reactiona
a Conditions: (a) HC(OCH3)3, TsOH, MeOH, 50 °C, 4 h; (b) DIBAL-H, PhCH3, -78 °C, 1 h; (c) TsOH, MeOH, H2O.
Scheme 12. Attempted Intramolecular Heck Reaction of 60
either with use of a phosphine ligand or under Jeffrey conditions
produced a complex mixture. The crude 1H NMR of the reaction
mixture contained aldehyde proton signals. Palladium(II) has
been shown to oxidize primary alcohol producing hydrido-
palladium species, which in some cases have been utilized to
reduce aryl halides.119-121 The p-methoxy-substituted benzylic
alcohol of 59 would be especially prone to oxidation by
palladium.
(22%); however, some ionization of the phenol was still
observed.
The chemoselectivity observed for the Heck reaction, over
the ionization reaction, using tri-o-tolylphosphine as ligand
prompted us to further investigate this reaction. It is presumably
the steric hindrance imposed by the o-methyl group which
inhibits coordination of the palladium(0) complex to the
trisubstituted olefin of 60, thereby preventing ionization. On
the other hand, the steric requirements for the oxidative additions
of the same complex to the arylbormide appear to be less
stringent. Changing the solvent from dimethylacetamide (DMA)
to toluene increases the yield of the benzofuran to 42% at 90
°C and 51% at 110 °C without a detectable amount of ionization.
Unfortunately, significant amounts of olefin isomerization
occurred and the benzofuran was isolated as a 3.6-4 to 1
mixture of the desired product 61 to the isomerization product
62. Carrying out the Heck reaction in the presence of 5 equiv
of silver nitrate suppressed the olefin isomerization; however,
a drastic drop in yield was also observed. Although employing
tri-o-tolylphosphine in the Heck reaction of 60 completely
prevented ionization of the phenol, its failure to suppress both
the olefin isomerization and obtain good yields of 61 led us to
abandon its use.
To avoid the complications caused by free alcohols, diol 59
was converted into the corresponding bis(tert-butyldimethyl-
silyl)ether 60 (Scheme 12). Reaction of arylbromide 60 under
Jeffrey conditions, employing potassium acetate as base,
produced only a trace of benzofuran 61, returning mainly starting
material. Changing the base to silver carbonate resulted in no
reaction. Utilizing the sterically hindered amine base, penta-
methylpiperidine (PMP), afforded 61 in approximately 10%
yield. Under these conditions, the major problem was catalyst
lifetime (conversion) rather than the competing ionization
reaction. Encouraged by these results, several ligands were
investigated in order to increase turnover. Addition of triphen-
ylphosphine or triphenylarsine to the reaction did not result in
a substantial increase in the yield of benzofuran 61. On the other
hand, using tri-o-tolylphosphine as the ligand resulted in an
increase in the yield of 61 to 21% and only a trace amount of
ionization of the phenol. The bidentate ligand bis(1,4-diphen-
ylphosphino)butane (dppb) produced a similar increase in yield
The failure of tri-o-tolylphosphine as a ligand led us to more
closely examine employing bidentate ligands in the Heck
reaction (Table 3). Use of dppb in acetonitrile resulted mainly
in phenol ionization, producing mainly diene 64 (entry 1, Table
3). Changing the solvent from acetonitrile to DMA improved
the yield of 61 to 19%, while significantly decreasing the amount
of ionization (entry 2). Increasing the equivalents of penta-
(119) Zask, A.; Helquist, P. J. Org. Chem. 1978, 43, 1619.
(120) Tamaru, Y.; Yamamoto, Y.; Yamada, Y.; Yoshida, Z. Tetrahedron Lett.
1979, 1401.
(121) Ben-David, Y.; Gozin, M.; Portnoy, M.; Milstein, D. J. Mol. Catal. 1992,
73, 173.
9
J. AM. CHEM. SOC. VOL. 127, NO. 42, 2005 14793

A R T I C L E S
Trost et al.
Table 3. Intramolecular Heck Reactions with Bidentate Phosphine Ligands
yield (%)
entry
ligand
conditions
60
61
63
64
1
2
3
4
5
6
7
8
9
dppb
dppb
dppb
dppb
dppb
dppe
dppf
BINAP
DIP-dpppa
dcpe
dcpe
dcpp
0.1 M CH3CN, 1.2 equiv of PMP, 2 h
0.1 M DMA, 1.2 equiv of PMP, 3 h
0.1 M DMA, 2.0 equiv of PMP, 2 h
0.1 M PhCH3, 2.0 equiv of PMP, 2 h
0.1 M dioxane, 2.0 equiv of PMP, 10 h
0.1 M DMA, 2.0 equiv of PMP, 8 h
0.1 M DMA, 2.0 equiv of PMP, 2 h
0.1 M DMA, 2.0 equiv of PMP, 2 h
0.1 M DMA, 2.0 equiv of PMP, 2 h
0.1 M DMA, 2.0 equiv of PMP, 12 h
0.1 M DMA, 3.0 equiv of proton sponge, 12 h
0.2 M DMA, 3.0 equiv of proton sponge, 12 h
0
16
32
32
32
38
27
27
24
5
8
19
22
7
14
20
7
17
17
42
50
33
70
44
33
51
33
29
27
35
58
25
0
10
11
12
7
23
19
23
0
methylpiperidine (PMP) from 1.2 to 2.0 further improved the
ratio of benzofuran 61 to ionization (entry 3). Employing toluene
(entry 4) or dioxane (entry 5) as solvent had a dentrimetnal effect
on the course of the Heck reaction. Milstein has observed that
the rate of oxidative addition of palladium(0) to aryl chlorides
is increased with decreasing ligand bite angles.122,123 Changing
the ligand from dppb (bite angle ) 97°) to bidentate ligands
with smaller bite angles, such as 1,2-bis(dipehnylphosphino)-
ethane (dppe, 85°) or 2,2′-bis(diphenylphosphino)-1,1′-binaph-
thyl (BINAP, 92°) did not significantly alter the ratio of
ionization to Heck reaction (entries 6 and 8). Furthermore, 1,1′-
bis(diphenylphosphino)ferrocene (dppf, 95°) was less effective
as a ligand (entry 7). Although decreasing the bite angle of the
ligand may increase the rate of oxidative addition, this change
appears not to have much effect of the relative rates of oxidative
addition and the competing ionization.
It is important to note that, in all cases in which bidentate
ligands were employed, less than 5% of the isomerized olefin
62 was observed. The bidentate ligands did not suppress the
ionization reaction, but the sterically demanding o-tolylphos-
phine ligand did. These results led us to hypothesize that
sterically demanding bidentate phosphine ligands would sup-
press both the olefin isomerization reaction and the phenol
ionization reaction. Furthermore, sterically hindered phosphine
ligands also appear to accelerate the rate of oxidative additions
into aryl halides.124-126 The di-ortho-substituted phosphine
ligand, 1,3-bis[di(2,6-diisopropoxyphenyl)phosphine]propane
(DIP-dppp), however, did not have the expected beneficial effect
on the reaction (entry 9). We thus turned our attention to
bidentate alkylphosphine ligands. 1,2-(Dicyclohexylphosphino)-
ethane (dcpe) proved to be an excellent ligand for the intra-
molecular Heck reaction, affording benzofuran 61 in 42% yield
accompanied by 25% of the ionization product 64 (entry 10).
Changing the base from PMP to proton sponge further improved
the reaction, furnishing 61 in 50% yield along with 19% of the
monodeprotected product 63 (entry 11). More importantly, none
of diene 64 derived from ionization was observed. Using a
ligand with a slightly increased bite angle relative to dcpe, 1,3-
(dicyclohexylphosphino)propane (dcpp), slightly decreased the
catalyst turnover, but the reaction remained chemoselective for
the Heck reaction over ionization (entry 12). The relative
stereochemistry of benzofuran 61 was assigned on the basis of
an NOE experiment.
The selectivity observed for the Heck reaction over ionization
in the presence of cyclohexylphosphine ligands may derive from
a combination of factors. First, alkylphosphines are more
electron-rich than their aryl counterparts. This has been shown
to increase the rate of oxidative addition of palladium(0) to aryl
halides. Conversely, electron-rich ligands should decrease the
rate of the first step in the catalytic cycle of the ionization
reaction, the coordination of palladium(0) to the trisubstituted
olefin of 61. Second, the steric hindrance of the dicyclohexyl-
phosphine ligands is significantly greater than the corresponding
diaryl phosphine ligands. This factor should also decrease the
rate of coordination of a trisubstituted olefin. This is consistent
with the selectivity observed when tri-o-tolylphosphine was used
as ligand. This combination of electronic and steric factors
present in the dicyclohexylphosphine ligand is presumably
responsible for the sharp increase in relative rate of the Heck
reaction to the ionization reaction.
(122) Portnoy, M.; Milstein, D. Organometallics 1993, 12, 1665.
(123) Portnoy, M.; Milstein, D. Organometallics 1993, 12, 1655.
(124) Wolfe, J. P.; Singer, R. A.; Yang, B. H.; Buchwald, S. L. J. Am. Chem.
Soc. 1999, 121, 9550.
(125) Hamann, B. C.; Hartwig, J. F. J. Am. Chem. Soc. 1998, 120, 7369.
(126) Littke, A. F.; Fu, G. C. J. Org. Chem. 1999, 64, 10.
9
14794 J. AM. CHEM. SOC. VOL. 127, NO. 42, 2005

(−)-Galanthamine and (−)-Morphine Synthesis
A R T I C L E S
Scheme 13. Construction of the Galanthamine’s Benzoazepine Ring
Scheme 14. Preparation of Amino-Aldehyde 69 by Reductive Aminationa
a Conditions: (a) TBAF; (b) MnO2,; (c) (i) MeNH2 HCl, MeOH; (ii) NaCNBH3, (iii) Boc2O, Et3N; (d) Dess-Martin periodane, CH2Cl2.
Scheme 15. Formation of the Hydrobenzoazepine Ringa
a Conditions: (a) MeOCH2PPh3 Br, NaHMDS, THF, 0 °C, Z:E ) 10:1; (b) TFA, CH2Cl2, rt, (c) (i) 4 Å MS, MeOH, 60 °C; (ii) NaCNBH3, MeOH, 0
°C.
Scheme 16. Synthesis of (-)-Deoxygalanthamine from (-)-Galanthamine Hydrobromidea
a Conditions: (a) MsCl, pyr, CH2Cl2; (b) LiEt3BH, THF, 40 °C, 20:74 ) 3:1.
Having constructed galanthamine’s benzofuran ring, we
turned our attention to the preparation of the hydrobenzoazepine.
We envisioned that the benzylic alcohol could be selectively
oxidized and converted into amino alcohol 65 by a reductive
amination (Scheme 13). We anticipated that the required one-
carbon homologation could be achieved by oxidation of the
alcohol to the aldehyde followed by Wittig reaction. The
hydrobenzoazeping would then be prepared by hydrolysis of
enol ether 66 by reductive amination.
68 underwent smooth oxidation, with Dess-Martin periodiane,
providing aldehyde 69 in 93% yield.
The necessary one-carbon homologation of aldehyde 69 was
achieved by a Wittig olefination of the aldehyde with the ylide
derived from methoxymethyltriphenylphosphonium bromide and
NaHMDS, producing 70 in 64% yield as a 10:1 mixture of olefin
isomers (Scheme 15). Concomitant deblocking of the secondary
amine and hydrolysis of the methoxyvinyl group gave a seven-
membered ring hemiaminal 71 which was not isolated but
immediately treated with sodium cyanoborohydride to afford
(-)-3-deoxygalanthamine 20 in 53% overall yield.
The absolute stereochemistry of our synthetic 3-deoxy-
galanthamine 20 was confirmed by the deoxygenation of natural
galanthamine hydrobromide (-)-72 (Scheme 16). This was
accomplished by first converting galanthamine hydrobromide
72 into the known chloride 739 by treatment with methane-
sulfonyl chloride. Dehalogenation of 73 was effected by reaction
with lithium triethylborohydride (Super-Hydride), which af-
forded a 3:1 mixture of regioisomeric olefin (-)-20 and (-)-
74. Comparison of the rotation of our synthetic (-)-20 (Scheme
15) with that of (-)-20 prepared from galanthamine (Scheme
16) confirms the absolute stereochemistry. Furthermore the
To prepare the required alcohol 65, the mixture of bis-61
and monosilyl ether 63 obtained from the Heck reaction was
deprotected with tetrabutylammonium fluoride (TBAF) (Scheme
14). Chemoselective manganese dioxide oxidation of the result-
ing benzyl alcohol afforded aldehyde 67. Reductive amination
of aldehyde 67 was accomplished by formation of the imine
with methylamine followed by reduction with sodium cy-
anoborohydride. The resulting secondary amine was protected
without isolation to supply tert-butyl carbamate 68. It was
necessary to protect the secondary amine because all attempts
to chemoselectively oxidize the alcohol in the presence of
unprotected amine resulted mainly in oxidation of the amine
and recovery of aldehyde 67. On the other hand, protected amine
9
J. AM. CHEM. SOC. VOL. 127, NO. 42, 2005 14795

A R T I C L E S
Trost et al.
Scheme 17. Epoxidation of (-)-Deoxygalanthamine with Dimethyldioxirane
Scheme 18. Conversion of Epoxide 76 into (-)-Isogalanthamine
79
enantiomeric excess of our (-)-20 can be calculated to be
approximately 89%, indicating no deterioration of the enantio-
meric excess (88%) initially achieved in the palladium-catalyzed
AAA.
At this point, all that remained was the installation of the C3
hydroxyl group. The most direct way to accomplish this is using
an allylic oxidation. We were confident this oxidation could be
achieved by a method similar to that reported by Muxfeldt and
co-workers in their synthesis of related amarylidaceae alkaloid,
crinine.101 Unfortunately, direct oxidation of 3-deoxygalan-
thamine 20 (SeO2 or tert-butylperbenzoate, CuBr, etc.; Scheme
17) gave a complex mixture. Similar difficulties with direct
allylic oxidations of alkaloids have been previously reported.
Torssell and co-workers reported that all attempts to effect the
allylic oxidation of deoxylycorine, using reagents such as
N-bromosuccinimde, lead(IV) acetate, chromium(VI) oxide, and
selenium dioxide, failed.127 Instead a three-step protocol,
epoxidation, phenylselenide opening-selenoxide elimination,
and allylic rearrangement, was used to install the desired allylic
oxygen.127 Treatment of the ammonium salt derived from 20
and 5 equiv of p-toluenesulfonic aicd with dimethyldioxirane
afforded a 1:1 mixture of tosylate 75 and epoxide 76. Formation
of the tosylate could be avoided by lowering the amount of
p-toluenesulfonic acid used; however, this resulted in lower
yields of 76 presumably as a result of competing formation of
the N-oxide. Furthermore, treatment of the crude mixture of
tosylate 75 and 76 with DBU cleanly converted the hydroxyl
tosylate into the desired epoxide 76. The relative stereochemistry
of 76 was assigned by an NOE experiment. Irradiation of the
epoxide proton (Ha, 3.20 ppm) showed a 5.4% enhancement of
one of the benzylic protons (4.10 ppm). Based on the minimized
structure, the epoxidation was assigned as having occurred on
the olefin face anti to the hydrobenzazepine ring. It would appear
that steric factors, rather than a hydrogen bond to the am-
monium, are responsible for the facial selectivity of the
epoxidation.128,129
Scheme 19. Rhenium-Mediated Allylic Rearrangement To
Produce (-)-Galanthamine
the mixture of selenoxide to 80 °C rapidly effected the
1
elimination of one diastereomer (as observed by H NMR),
while the other diastereomer required 2 h for complete elimina-
tion. Under these conditions, (-)-isogalanthamine 79 could be
prepared in 64% yield.
We investigated the use of Osborne’s trioxorhenium(IV)
catalyst130 for the isomerization of isogalanthamine 79 to
galanthamine (-)-1 (Scheme 19). Under catalytic conditions
only trace of amounts of galanthamine were observed in both
refluxing chloroform and benzene. However, treatment of 79
first with p-toluenesulfonic acid followed by stoichiometric
amounts of the rhenium(VII) complex afforded a 3:1 mixture
of 1 and 79. Alcohols 79 and 1 were readily separated, affording
All attempts to utilize the lithium pyrolidide to convert
epoxide 76 into allylic alcohol 79 failed, returning epoxide even
in refluxing benzene (Scheme 18). On the other hand, the
phenylselenide opening of epoxide 76 proceeded smoothly,
furnishing hydroxyselenide 77 in 98% yield. As expected,
selenide attack occurred with complete selectivity for attack at
the stereoelectonic favored carbon rather than the neo-pentyl
carbon. The regioselectivity of attack was established by an 1H-
COSY experiment. Sodium periodate oxidation of selendie 77
produced a 1:1 mixture of diastereomeric selenoxides 78, which
surprisingly did not eliminate at room temperature. Heating of
1
galanthamine 1 in 50% yield from 79. The H NMR and 13C
NMR of synthetic 1 are identical to those obtained from a
sample of natural galanthamine. The optical rotation ([R]D
)
-112.8 (c ) 0.5, EOH); lit.9 [R]D ) -131.4 (c ) 0.6, EOH) is
in accordance with those reported in the literature.
This is the first synthesis of any member of the galanthamine
alkaloids in which the enantioselectivity is generated by a
catalytic asymmetric reaction. All the rest of the stereochemistry
was derived from the stereogenic center of chiral ether 52. This
synthesis also constitutes a formal enantioselective synthesis
(127) Moller, O.; Steinberg, E.-M.; Torssell, K. Acta Chem. Scand., Ser. B 1976,
32, 98.
(128) Asensio, G.; Mello, R.; Boix-Bernardini, C.; Gonzalez-Nunez, M. E.;
Castellano, G. J. Org. Chem. 1995, 60, 3692.
(129) Asensio, G. B.-B., C.; Andreu, C.; Gonzalez-Nunez, M. E.; Mello, R.;
Edwards, J. O.; Carpenter, G. B. J. Org. Chem. 1999, 64, 4705.
(130) Bellemin-Laponnaz, S.; Le Ny, J. P.; Osborn, J. A. Tetrahedron Lett. 2000,
41, 1549.
9
14796 J. AM. CHEM. SOC. VOL. 127, NO. 42, 2005

(−)-Galanthamine and (−)-Morphine Synthesis
A R T I C L E S
Scheme 20. Parsons’ Attempts for the Synthesis of
(()-Galanthamine35
Scheme 22. Preparation of Carbonate 89a
a Conditions: (a) CHOCO2Et, Dabco/toluene; (b) MsCl/Et3N/DMAP;
(c) DBU/cyclohexane, reflux; (d) NaBH4, CeCl3; (e) Cl3CCH2OC(O)Cl,
pyr.
Scheme 21. Synthesis of Galanthamine via Iodolactonization
Strategy
Scheme 23. Palladium-Catalyzed AAA of 26 and 89
of related amaryllidaceae alkaloids, narwedine 2 and lycoramine
3, via oxidation or reduction.131
ring of 83. Compound 85, which is readily available from Pd-
catalyzed AAA followed by Heck cyclization, can be converted
to lactone 84 by iodolactonization promoted by either iodine
or NIS followed by base-mediated elimination.
Second Generation Synthesis of (-)-Galanthamine
We can access the core structure of galanthamine, compounds
61 and 63 (Table 3), very efficiently via a sequential palladium-
catalyzed AAA reaction and intramolecular Heck reaction.
However, rather lengthy manipulations are required for the
synthesis of the azepine ring from compounds 61 and 63 to
3-deoxygalanthamine 20. Multiple steps are involved for the
introduction of the 3-hydroxyl group (20 to 1). We decided to
pursue a second-generation synthesis, which had three goals:
(1) realization of an effective general strategy for this family
of alkaloids by total synthesis of galanthamine; (2) development
of a flexible strategy to provide ready access to galanthamine
analogues (i.e. compound 4) by simple modification of the
synthetic scheme; and (3) development of a divergent strategy
to access both amaryllidaceae and opium alkaloids.
During the course our investigation, two groups34,35 inde-
pendently reported the synthesis of (()-3-deoxygalanthamine.
Both groups used the same allylic alkylation (involving
Mitsunobu reaction) followed by Heck cyclization strategy as
our previous galanthamine synthesis. They all failed to complete
the synthesis because of the difficulty of introducing the allylic
hydroxyl group. Parson and co-workers tried various conditions
for the allylic oxidation of lactam 80. They did not observe
any corresponding allylic oxidation product. Instead, a mixture
of recovered starting material, ring cleaved product 81, and
imide 82 were obtained (Scheme 20).35 It clearly showed that
the benzylic hydrogens are much more labile toward oxidation
than the allylic hydrogens.
The synthesis begins with the preparation of carbonate 89
(Scheme 22). Compound 88 had been prepared from a tandem
Birch reduction of methyl 2-methoxybenzoate followed by
alkylation with ethyl 2-bromoacetate and hydrolysis.103 We
found that a three-step protocol in Scheme 22 is more suitable
for large-scale preparation of known enone 88.135 Luche104
reduction followed by acylation provided carbonate 89.
Various conditions failed to catalyze the allylic alkylation of
phenol 26 with carbonate 89 using ligands 42 and 53-55 (Table
2) and other ligands in our hands. BINAPO ligand 91, the first
chiral ligand developed for the palladium-catalyzed AAA
reaction in this group,136 is the only ligand that worked for the
reaction among all ligands we examined (Scheme 23).137 Aryl
ether 90 can be obtained in high yield (75-80%) but only
moderate enantioselectivity (57%ee) in the presence of BINAPO
ligand under various conditions. The absolute stereochemistry
of compound 90 is not determined due to the low enantiomeric
excess.
To test the validity of the proposed synthetic plan (Scheme
21), aryl ether 90 was cyclized to give 85 in the presence of
Pd(OAc)2/dppp and excess silver carbonate (Scheme 24).34,35
Significant amounts of olefin isomers were found when potas-
sium carbonate was used as base or less silver carbonate was
employed. Early results118 and results from our previous
synthesis of galanthamine suggested that the presence of
electron-withdrawing substituents on the phenol ring favored
the palladium-catalyzed ionization over the intramolecular Heck
reaction. The success of current reaction suggests that the
electron-withdrawing group on the olefin favored the ionization,
while the electron-withdrawing group on the phenol ring
facilitated oxidative addition of the palladium to both the C-O
and C-Br bonds. The relative stereochemistry of benzofuran
Due to the difficulty of the direct allylic oxidation, we
envisioned an iodolactonization strategy to introduce the allylic
OH (Scheme 21). An SN2′-type inversion was proposed to
convert 1-OH(R) to 3-OH(â).132-134 Reductive amination of 84
followed by cyclization was proposed to synthesize the lactam
9
J. AM. CHEM. SOC. VOL. 127, NO. 42, 2005 14797

A R T I C L E S
Trost et al.
Scheme 24. Intramolecular Heck Reaction and Cyclization of the
Scheme 27. Alternative Iodolactonization Strategy
Benzoazepine Ringa
a Conditions: (a) 15 mol % Pd(OAc), 15 mol % dppp, PhCH3, 3 equiv
of Ag2CO3; (b) MeNH2/MeOH, NaBH4, 0 °C.
Scheme 25. Iodolactonization and Cyclization of the
Hydrobenzoazazepine Ringa
48%) was identified as natural galanthamine 1, and the minor
isomer (isolated in 35%) was assigned as 3-epi-galanthamine
95.
Third Generation Synthesis of (-)-Galanthamine and
Its Analogue
The low overall efficiency and selectivity of the above
strategy prompted us to pursue a new strategy to introduce the
allylic hydroxyl group. We decided to take the advantage of
the known 1,3-rearrangement of penultimate precursor 79 to
galanthamine 1 (Scheme 19) in our first generation synthesis
of galanthamine. We envisioned a new oxidation-iodolacton-
ization-elimination sequence to introduce the allylic oxygen
of compound 96 directly from compound 98 (Scheme 27).
Conformation analysis showed that the carboxylic acid derived
from oxidation of aldehyde 98 could only access the more
hindered â-face of the olefin intramolecularly to give lactone
96 (as depicted in conformation 97).
a Conditions: (a) (i) 1 N NaOH, MeOH then 3 N HCl; (ii) NIS, DMAP,
MeCN; (b) DBU/THF; (c) MeNH2/MeOH, then NaBH4.
Scheme 26. Synthesis of Galanthamine and 3-epi-Galanthamine
Compound 98 can be prepared either directly from aryl ether
90 (Scheme 23) via Heck cyclization or from aryl ether 52
(Table 2) with some functional group manipulations followed
by Heck cyclization. Due to the difficulty in achieving good
enantioselectivity of ether 90, we decided to prepare compound
98 from aryl ether 52, which can be obtained with high optical
purity (88%ee) using palladium-catalyzed AAA reaction (Table
2). On the basis of the success of the Heck reaction for substrate
90 (Scheme 24), we decided to keep the aldehyde functionality
of ether 52 while removing the electron-withdrawing substituent
on the olefin by homologation of the R,â-unsaturated ester to a
R,â-unsaturated nitrile (Scheme 28). To this end, aldehyde 52
was protected as its dimethylacetal by treatment with trimethyl
orthofomate and catalytic p-toluenesulfonic acid. It would be
more convergent if the acetal can be formed before the AAA
reaction. Neither methanol nor ethylene glycol can produce pure
acetal. The acetal obtained was quickly hydrolyzed when
exposed to air. Selective reduction of the R,â-unsaturated ester
with diisobutylaluminum hydride (Dibal-H) at -78 °C afforded
allylic alcohol 100. The R,â-unsaturated nitrile 101 was prepared
85 was assigned on the basis of analogy to Heck cyclization of
compound 60 (Table 3). Reductive amination of 85 provided
lactam 80 in high yield. Attempts for the allylic oxidation of
lactam 80 were not successful, which has been observed by
Parsons35 as well (Scheme 20).
Direct iodolactonization of 85 was not successful. A one-pot
hydrolysis, iodolactonization followed elimination procedure
provided lactone 84 in 64% overall yield (Scheme 25). Reduc-
tive amination and amidation were achieved in a single
operation, giving lactam 83 in 50% yield.
The allylic alcohol 83 was activated and subjected to basic
hydrolysis (Scheme 26).132-134 Two isomers (93 and 94) in a
ratio of 1.5:1 were obtained without any recovery of the starting
(131) Lee, T. B. K.; Goehring, K. E.; Ma, Z. J. Org. Chem. 1998, 63, 4535.
(132) Shull, B. K.; Sakai, T.; Nichols, J. B.; Koreeda, M. J. Org. Chem. 1997,
62, 8294.
(133) Tsunoda, T.; Yamamiya, Y.; Kawamura, Y.; Ito, S. Tetrahedron Lett. 1995,
36, 2529.
1
material on the basis of crude H NMR. Both isomers were
(134) Hughes, D. L.; Reamer, R. A. J. Org. Chem. 1996, 61, 2967.
(135) Ziegler, F. E.; Piwinski, J. J. J. Am. Chem. Soc. 1982, 104, 7181.
(136) Trost, B. M.; Murphy, D. J. Organometallics 1985, 4, 1143.
(137) Mori, M.; Kuroda, S.; Zhang, C.-S.; Sato, Y. J. Org. Chem. 1997, 62,
3263.
collected and reduced by LiAlH4 in refluxing tetrahydrofuran
(THF). By comparing the 1HNMR of known galanthamine 132
and 3-epi-galanthamine 95,25,75 the major isomer (isolated in
9
14798 J. AM. CHEM. SOC. VOL. 127, NO. 42, 2005

(−)-Galanthamine and (−)-Morphine Synthesis
A R T I C L E S
Scheme 28. Synthesis of Compound 102a
a Conditions: (a) 1.5 mol % TsOH, CH(OMe)3, MeOH; (b) 2. DIBAL-H/Tol, -78 °C; (c) 1. Ph3P, acetonecyanohydrin, DIAD, Et2O; (d) 2.20 mol %
TsOH, THF, H2O; (e) 15 mol % Pd(OAc)2, 15 mol % dppp, 3 equiv of Ag2CO3, PhCH3, 107 °C.
Scheme 29. Attempted Iodolactonization of Acid 103
in 84% yield using a modified Mitsunobu protocol138,139
followed by acid hydrolysis. The enantiomeric excess of 101
was improved to 96% by recrystallization from diethyl ether
and petroleum ether with 90% mass recovery. We then examined
the Heck reaction of the acetonitrile-substituted cyclohexene
101. High yield (91%) of the Heck product 102 was obtained
using a catalytic amount of diphenylphosphinopropane (dppp)
and Pd(OAc)2 in refluxing toluene in the presence of excess of
silver carbonate, conditions similar to the cyclization of 90
(Scheme 24).34,35 Other solvent (DMF) or ligands (dppe, dppf)
gave lower yield. Lower conversion and olefin isomerization
were observed with lower catalyst loading or less amount of
silver carbonate.
first generation synthesis, the competitive oxidation of the
tertiary amine and the benzylic protons (Scheme 20).34,35 We
decided to reexamine the allylic oxidation strategy on substrate
102, where the benzylic position is protected as the aldehyde
and the tertiary amine is masked as nitrile. This can be realized
by oxidation of the olefin to an enone followed by known
diastereoselective 1,2-reduction or, more efficiently, direct
oxidation to the allylic alcohol, which raises the question
regarding the diastereoselectivity. Usually, the electrophile
would approach the olefin from the less hindered convex face.
We envisioned that SeO2 would react with the olefin from the
more hindered concave face through an ene mechanism140
(structure 108, Scheme 30), because the axial proton Hax is
perfectly aligned with the π system. Gratifyingly, treatment of
olefin 102 with SeO2 in dioxane at 150 °C in a sealed tube
provided alcohol 107 in 57% yield (dr ) 10:1) with 7%
recovered starting material in the presence of NaH2PO4 and
quartz sand.141 Other additives, such as HCO2H,142 InCl3, YbCl3,
and Na2HPO4, gave less of the desired product. Attempts to
fully convert all the starting material by extending the reaction
time or increasing the temperature resulted in more decomposi-
tion. Only a trace amount of product was found in refluxing
dioxane (110 °C) after 8 h. This represents the first successful
allylic oxidation32,35 of the galanthamine skeleton. The stereo-
chemistry of 107 was tentatively assigned on the basis of
coupling constants of Ha/Heq and Hb/Heq (J < 5.0 Hz). The high
diastereoselectivity reflects the stereoelectronic requirement for
the ene reaction.
Aldehyde 102 was oxidized to acid 103 uneventfully (Scheme
29). All attempts for the iodolactonization of acid 103 under
the same condition for acid derived from ester 85 (Scheme 25)
or more forcing conditions (high temperature) failed, and most
of the starting materials were recovered. In the case of acid
derived from ester 85 the electrophile (NIS) approached the
olefin from the more hindered â-face (syn to the bulky aryl
group). Formation of iodonium species 105 should be a facile
process since the electrophile (NIS) now approaches the olefin
from the less hindered R-face (anti to the bulky aryl group).
However, to bring the carboxylic acid in close proximity to the
olefin, the cyclohexene ring has to adopt the less stable
conformation 106, where the bulky aromatic ring is on the
pseudoaxial position. Furthermore, the diaxial trajectory required
for the attack of the carboxylate to iodonium ion cannot be
fulfilled in conformation 106.
Compound 107 together with its epimer (dr ) 10:1) was then
converted to the natural product in a one-pot process (Scheme
31). Aldehyde 107 was treated with methylamine in methanol
There are primarily two reasons for the failure of the direct
allylic oxidation of deoxygalanthamine 20 (Scheme 16) in our
(140) Nicolaou, K. C.; Petasis, N. A. Selenium in Natural Products Synthesis;
CIS: Philadelphia, PA, 1984.
(141) Trost, B. M.; Van Vranken, D. L. J. Am. Chem. Soc. 1993, 115, 444.
(142) Shibuya, K. Synth. Commun. 1994, 24, 2923.
(138) Wilk, B. K. Synth. Commun. 1993, 23, 2481.
(139) Aesa, M. C.; Baan, G.; Novak, L.; Szantay, C. Synth. Commun. 1995,
25, 1545.
9
J. AM. CHEM. SOC. VOL. 127, NO. 42, 2005 14799

A R T I C L E S
Trost et al.
Scheme 30. Direct Diastereoselective Allylic Oxidation of 102
Scheme 31. One-Pot Completion of the Synthesis of (-)-Galanthamine and Its Derivativea
a Conditions: (a) RNH2, MeOH; (b) 4 equiv of DIBAL-H, then aqueous NaH2PO4; (c) NaCNBH3.
solution. Excess methylamine and methanol was removed under
vacuum. Concomitant reduction of the imine and nitrile by
DIBAL-H followed by acid quenching presumably gave a seven-
membered ring hemi-aminal 109. The resulting solution was
directly treated with sodium cyanoborohydride to afford (-)-
galanthamine 1 and 3-epi-galanthamine 95 (Scheme 26) in 62
and 6% isolated yields, respectively. All spectral data (1H NMR,
reductive cyclization represents a simple and efficient strategy
to form the latter and to access many galanthamine analogues.
Enantioselective Synthesis of (-)-Codeine and
(-)-Morphine
In our synthesis of galanthamine 1 and its derivative SPH1339
110, we established a practical synthesis of cyanoaldehyde 102
in only five steps from Pd-AAA product 52 (Scheme 28). On
the basis of the diversified strategy proposed in Scheme 3, we
then pursued the synthesis of opium alkaloids starting from
intermediate 102, which has the common tricyclic core of opium
alkaloids and appropriate functionalized side chains (R1 and R2
of compound 19, Scheme 3) that can be transformed to the
additional rings of the opium alkaloids.
Scheme 32 outlines two strategies based upon formation of
the piperidine ring as the final ring-forming event.143,144 In path
A, a straightforward displacement of amino alcohol 111
envisions formation of the latter via a carbonyl ene reaction of
aldehyde 112, which, in turn, may derive by homologation of
cyanoaldehyde 102. Path B outlines a more intriguing, but more
speculative, strategy in which the piperidine ring would arise
by a hydroamination of amine 113, which, in turn, may derive
from Heck vinylation of Z-vinyl bromide 114. A simple
olefination protocol should allow the latter to derive from the
same aldehyde 102.
13C NMR, IR) [R]D ) -123.1 (c ) 0.4, EtOH), lit.9 [R]D
)
-131.4 (c ) 0.6, EtOH)) are identical to those of a sample of
the natural product and data from previous synthesis.
Simply switching methylamine to propylamine, SPH-1339
11014, a slightly more potent inhibitor of Ache compared to
(-)-galanthamine 1, was prepared in 56% yield by this one-
pot process. Previous syntheses of this type of galanthamine
derivatives require demethylation of galanthamine via
Polonovski reaction15 followed by alkylation of the resulting
norgalanthamine.12,14
The third generation synthesis33 of galanthamine (10 steps,
96% ee, 8% overall yield from commercially available glut-
araldehyde 35 and isovanillin 25) is a significant improvement
over and successfully addresses many of the shortcomings of
the first generation synthesis32 (16 steps, 88% ee, 0.8% overall
yield). Furthermore, switching methylamine to other alkylamines
in the last step can provide easy access to various galanthamine
derivatives, demonstrated by the synthesis of SPH-1339. The
sequential palladium-catalyzed AAA and intramolecular Heck
reaction followed by a diastereoselective allylic oxidation
provided the key intermediate 107 with all the functionality
installed except the hydrobenzoazepine ring. The one-pot
Scheme 33 outlines the initial efforts based upon path A.
Homologation of the aldehyde proceeded uneventfully via
(143) Parker, K. A.; Fokas, D. J. Am. Chem. Soc. 1992, 114, 9688.
(144) Toth, J. E.; Fuchs, P. L. J. Org. Chem. 1987, 52, 473.
9
14800 J. AM. CHEM. SOC. VOL. 127, NO. 42, 2005

(−)-Galanthamine and (−)-Morphine Synthesis
A R T I C L E S
conformation of morphine,147 both Ha and Hb of alkene 117
are stereoelectronically and sterically favored for allylic oxida-
tion. Despite Ha being doubly allylic, its removal with selenium
dioxide is clearly strained due to creation of the bridgehead
double bond. Indeed, subjecting 117 to this reagent only involves
abstraction of Hb to give the corresponding alcohol 118
accompanied by the overoxidation product ketone 119. Directly
adding the Dess-Martin periodinane to the reaction mixture
prior to workup allowed ketone 119 to be isolated in 58% yield.
Its reduction proceeded stereoselectively with DIBAL-H in
THF-ether to give the required alcohol 120 almost quantita-
tively. No reduction of the nitrile was observed in this solvent
system. Furthermore, none of alcohol 120 was detected in the
initial allylic oxidation.
Scheme 32. Retrosynthetic Analysis of Opium Alkaloids
Adapting a known protocol,148 the nitrile 120 was converted
to the secondary amine 113 in a one-pot operation (Scheme
36). Switching from THF to methylene chloride allowed nitrile
reduction to the imine aluminum complex. Addition of am-
monium bromide in dry methanol destroyed excess DIBAL-H
and freed the imine. Subsequent addition of excess methylamine
converted the primary imine to the more stable secondary one
(Scheme 36). The final stage involved addition of sodium
borohydride wherein amine 113 was obtained quantitatively
from alcohol 120. Performing this same protocol on ketone 119
also converted it to amine 113, thereby saving one step.
rearrangement of an epoxide intermediate. All attempts to effect
the intramolecular carbonyl ene reaction to form alcohol 115
failed.
Thwarted in attempts to execute path A, we turned to path
B. Olefination145 followed by chemoselective reduction of the
E-vinyl bromide146 provided the cyclization substrate 114
(Scheme 34). Intramolecular Heck vinylation to the sterically
congested neopentyl carbon created the tetracycle 117 in good
yield under the same conditions used for the Heck reaction of
compounds 90 and 102 (Scheme 24 and Scheme 28).33
With the skeleton of morphine in hand, our attention focused
on allylic functionalization (Scheme 35). Based upon the known
The stage was set for the closure of the final piperidine ring.
Parker,143 Mulzer,149-151 and Ogasawara152,153 synthesized 7,8-
Scheme 33. Attempted Carbonyl Ene Reaction
Scheme 34. Heck Cyclization of Vinyl Bromidea
a Conditons: (a) CBr4, Ph3P, CH2Cl2; (b) 5 mol % Pd(PPh3)4, nBu3SnH, PhCH3; (c) 15% Pd(OAc)2, 15% dppp, Ag2CO3, toluene.
Scheme 35. Allylic Oxidation of Compound 117 and Diastereoselective Reduction
9
J. AM. CHEM. SOC. VOL. 127, NO. 42, 2005 14801

A R T I C L E S
Trost et al.
Scheme 36. Sequential Reduction-Transamination-Reduction
Scheme 37. Light Promoted Hydroamination
dihydrocodeinone by cyclizing the final piperidine ring via a
radical anion process (Li/NH3/t-BuOH) from a tosyl amide
derivative. We decided to investigate the direct ring closure of
secondary amine 113 to codeine via hydroamination. Carefully
examining the functional group compatibility of all known olefin
hydroamination reactions,154-164 the base-catalyzed hydro-
amination appeared to be the most promising method for the
hydroamination of amine 113.
The first example of base-catalyzed intermolecular hydroami-
nation of vinylarenes appeared in 1948.165 The anti-Markovnikov
addition product was obtained as the major product albeit in
low yield. The hydroamination of styrene derivatives using
alkyllithium as precatalyst was first reported in 1972.166 The
alkyllithium-catalyzed hydroamination of styrene has been
NHCH3) do not cyclize in the presence of either stoichiometric
or catalytic amounts of nBuLi.172 However, under anodic
oxidative conditions,173,174 the lithium amide of electron-rich
vinylamines can cyclize to corresponding pyrrolidines in moder-
ate yield.172
extended to more complicated amines and more substituted vinyl
155,167,168
aromatics by Beller
and other groups.169,170
Simply treating 113 with various amounts of lithium diiso-
propylamide (LDA) or n-butyllithium in refluxing THF led to
recovered starting material up to 2 h and extensive decomposi-
tion after 8 h. With the notion that the addition might be
facilitated by single electron transfer,172-177 promotion of the
latter by irradiation of the basic solution with an ordinary
tungsten light bulb178,179 was envisioned. Indeed, subjecting the
solution of amine 113 and 6 equiv of LDA in THF to such
irradiation with a 150-W tungsten light bulb led to cycloisomer-
ization to form (-)-codeine whose spectral data are identical
to those previously reported (Scheme 37).180,181 Less amount
of base led to lower conversion. Demethylation as reported by
Rice182 with boron tribromide converts this route into a synthesis
of (-)-morphine as well.
This very short asymmetric total synthesis of (-)-morphine
(15 longest linear steps with 3.2% overall yield and 16 total
steps from commercially available materials) arises because of
the minimal use of protecting groups. Palladium-catalyzed
reactions, AAA and two Heck-type, create the entire carbon
framework and four rings. The one-pot reduction-transamina-
tion-reduction of nitriles significantly shortens the route. Most
Base-catalyzed intramolecular hydroamination of vinyl arenes
was only studied for the formation of five-membered rings.171
It has been shown that more than stoichiometric amounts of
base (i.e. BuLi) give lower yields. Although monosubstituted
alkenes cyclize smoothly, more substituted alkenylamines and
electron-rich vinylamines (e.g. p-CH3O-C6H4-CHdCH(CH2)3-
(145) Corey, E. J.; Fuchs, P. L. Tetrahedron Lett. 1972, 3769.
(146) Uenishi, J. I.; Kawahama, R.; Shiga, Y.; Yonemitsu, O.; Tsuji, J.
Tetrahedron Lett. 1996, 37, 6759.
(147) Mackay, M.; Hodgkin, D. C. J. Chem. Soc. 1955, 3261.
(148) Zandbergen, P.; Van Den Nieuwendijk, A. M. C. H.; Brussee, J.; Van
Der Gen, A.; Kruse, C. G. Tetrahedron 1992, 48, 3977.
(149) Trauner, D.; Bats, J. W.; Werner, A.; Mulzer, J. J. Org. Chem. 1998, 63,
5908.
(150) Mulzer, J.; Bats, J. W.; List, B.; Opatz, T.; Trauner, D. Synlett 1997, 441.
(151) Mulzer, J.; Duerner, G.; Trauner, D. Angew. Chem., Int. Ed. Engl. 1997,
35, 2830.
(152) Nagata, H.; Miyazawa, N.; Ogasawara, K. Chem. Commun. 2001, 1094.
(153) Nagata, H.; Miyazawa, N.; Ogasawara, K. Org. Lett. 2001, 3, 1737.
(154) Borman, S. Chem. Eng. News 2004, 82, 23.
(155) Mueller, T. E.; Beller, M. Chem. ReV. 1998, 98, 675.
(156) Molander, G. A.; Romero, J. A. C. Chem. ReV. 2002, 102, 2161.
(157) Beller, M.; Breindl, C.; Eichberger, M.; Hartung, C. G.; Seayad, J.; Thiel,
O. R.; Tillack, A.; Trauthwein, H. Synlett 2002, 1579.
(158) Seayad, J.; Tillack, A.; Hartung, C. G.; Beller, M. AdV. Synth. Catal. 2002,
344, 795.
(159) Hartwig, J. F. Pure Appl. Chem. 2004, 76, 507.
(160) Hong, S.; Marks, T. J. Acc. Chem. Res. 2004, 37, 673.
(161) Brunet, J. J.; Neibecker, D. In Catalytic Heterofunctionalization from
Hydroamination to Hydrozirconation; Togni, A., Gruetzmacher, H., Eds.;
Wiley-VCH: Weinheim, Germany, 2001; p 98.
(162) Beller, M.; Seayad, J.; Tillack, A.; Jiao, H. Angew. Chem., Int. Ed. 2004,
43, 3368.
(163) Pohlki, F.; Doye, S. Chem. Soc. ReV. 2003, 32, 104.
(164) Alonso, F.; Beletskaya, I. P.; Yus, M. Chem. ReV. 2004, 104, 3079.
(165) Weston, A. W. (Abbott Laboratories) U.S., 1948.
(166) Schlott, R. J.; Falk, J. C.; Narducy, K. W. J. Org. Chem. 1972, 37, 4243.
(167) Beller, M.; Breindl, C. Tetrahedron 1998, 54, 6359.
(168) Beller, M.; Thiel, O. R.; Trauthwein, H.; Hartung, C. G. Chem. Eur. J.
2000, 6, 2513.
(169) Seijas, J. A.; Vazquez-Tato, M. P.; Entenza, C.; Martinez, M. M.; Onega,
M. G.; Veiga, S. Tetrahedron Lett. 1998, 39, 5073.
(170) Seijas, J. A.; Vazquez-Tato, M. P.; Martinez, M. M. Synlett 2001, 875.
(171) Fujita, H.; Tokuda, M.; Nitta, M.; Suginome, H. Tetrahedron Lett. 1992,
33, 6359.
(172) Tokuda, M.; Fujita, H.; Miyamoto, T.; Suginome, H. Tetrahedron 1993,
49, 2413.
(173) Tokuda, M.; Yamada, Y.; Takagi, T.; Suginome, H.; Furusaki, A.
Tetrahedron Lett. 1985, 26, 6085.
(174) Tokuda, M.; Yamada, Y.; Takagi, T.; Suginome, H. Tetrahedron 1987,
43, 281.
(175) Ashby, E. C.; Goel, A. B.; Depriest, R. N. J. Org. Chem. 1981, 46, 2429.
(176) Ashby, E. C.; Goel, A. B.; Depriest, R. N. Tetrahedron Lett. 1981, 22,
4355.
(177) Newcomb, M.; Burchill, M. T.; Deeb, T. M. J. Am. Chem. Soc. 1988,
110, 6528.
(178) Scamehorn, R. G.; Bunnett, J. F. J. Org. Chem. 1977, 42, 1457.
(179) Rossi, R. A.; Bunnett, J. F. J. Org. Chem. 1973, 38, 1407.
(180) White, J. D.; Hrnciar, P.; Stappenbeck, F. J. Org. Chem. 1999, 64, 7871.
(181) Carroll, F. I.; Moreland, C. G.; Brine, G. A.; Kepler, J. A. J. Org. Chem.
1976, 41, 996.
(182) Rice, K. C. J. Med. Chem. 1977, 20, 164.
9
14802 J. AM. CHEM. SOC. VOL. 127, NO. 42, 2005

(−)-Galanthamine and (−)-Morphine Synthesis
A R T I C L E S
noteworthy is the effectiveness of the intramolecular hydroami-
nation promoted by visible lightsa reaction that should prove
more generally useful in alkaloid synthesis.
of the amaryllidaceae and opium alkaloids from a chemical
synthesis perspective has now been established. All the stereo-
chemistry emanates from the palladium-catalyzed AAA. Since
both enantiomers of ligand 55 are available,141 both galan-
thamine and morphine are available, as either enantiomer, from
commercial available materials. This strategy should allow for
entry into a variety of amaryllidaceae alkaloids and opium
alkaloids with similar skeletons.
Conclusion
All three generations of syntheses of galanthamine and the
synthesis of morphine feature the sequential palladium-catalyzed
asymmetric allylic alkylation and Heck cyclization. We devel-
oped three different ways to introduce the allylic alcohol
diastereoselectively. Among them, the direct allylic oxidation
is the most efficient and selective one. The third generation
synthesis of galanthamine is different from previous ones by
closing the benzoazepine ring in the last step. This provided
easy access to various galanthamine analogues demonstrated
by the synthesis of SPH-1339. This is the shortest and most
efficient nonbiomimetic total synthesis of (-)-galanthamine to
date. (-)-Galanthamine and (-)-morphine are now available
in two steps (35% yield) and seven steps (14% yield),
respectively, from the common intermediate 102, which can be
prepared enantioselectively from commercial available glut-
araldehyde 35 and isovanillin 25 in eight longest linear (23%
overall yield) and nine total steps. Strategically, the commonality
Acknowledgment. We thank the National Science Foundation
and the National Institute of Health, General Medical Sciences
(Grant GM-13598), for their generous support of our programs.
Mass spectra were provided by the Mass Spectrometry Facility
of the University of CaliforniasSan Francisco, supported by
the NIH Division of Research Resources. W.T. thanks Amgen
for a predoctoral fellowship.
Supporting Information Available: Compound spectra and
experimental details. This material is available free of charge
via the Internet at http://pubs.acs.org.
JA054449+
9
J. AM. CHEM. SOC. VOL. 127, NO. 42, 2005 14803