Use of the intramolecular heck reaction for forming congested quaternary carbon stereocenters. Stereocontrolled total synthesis of (±)-gelsemine

Article abstract of DOI:10.1021/ja055711h

Intramolecular Heck reactions of α,β-unsaturated 2-haloanilides derived from azatricyclo[4.4.0.0^2,8]-decanone 5 efficiently install the congested spirooxindole functionality of gelsemine. Depending upon the Heck reaction conditions and the nature of the β-substituent, either products having the natural or unnatural configuration of the spirooxindole group are formed predominantly. Efforts to elaborate the hydropyran ring of gelsemine from the endo-oriented nitrile substituent of pentacyclic Heck product 18 were unsuccessful. Important steps in the ultimately successful route to (±)-gelsemine (1) are as follows: (a) intramolecular Heck reaction of tricyclic β-methoxy α,β-unsaturated 2-iodoanilide 68 in the presence of silver phosphate to form pentacyclic product 69 having the unnatural configuration of the spirooxindole fragment, (b) formation of hexacyclic aziridine 80 from the reaction of cyanide with intermediate 79 containing an N-methoxycarbonyl-β-bromoethylamine fragment, (c) introduction of C17 by ring-opening of the aziridinium ion derived from aziridine 80, and (d) base-promoted skeletal rearrangement of pentacyclic equatorial alcohol 82 to form the oxacyclic ring and invert the spirooxindole functional group to provide hexacyclic gelsemine precursor 83.

Full text of DOI:10.1021/ja055711h

Published on Web 12/06/2005
Use of the Intramolecular Heck Reaction for Forming
Congested Quaternary Carbon Stereocenters.
Stereocontrolled Total Synthesis of (()-Gelsemine
Andrew Madin,1a Christopher J. O’Donnell,1b Taeboem Oh,1c David W. Old,1d
Larry E. Overman,* and Matthew J. Sharp1e
Contribution from the Department of Chemistry, UniVersity of California, IrVine,
516 Rowland Hall, IrVine, California 92697-2025
Received August 30, 2005; E-mail: leoverma@uci.edu
Abstract: Intramolecular Heck reactions of R,â-unsaturated 2-haloanilides derived from azatricyclo[4.4.0.02,8]-
decanone 5 efficiently install the congested spirooxindole functionality of gelsemine. Depending upon the
Heck reaction conditions and the nature of the â-substituent, either products having the natural or unnatural
configuration of the spirooxindole group are formed predominantly. Efforts to elaborate the hydropyran
ring of gelsemine from the endo-oriented nitrile substituent of pentacyclic Heck product 18 were unsuccessful.
Important steps in the ultimately successful route to (()-gelsemine (1) are as follows: (a) intramolecular
Heck reaction of tricyclic â-methoxy R,â-unsaturated 2-iodoanilide 68 in the presence of silver phosphate
to form pentacyclic product 69 having the unnatural configuration of the spirooxindole fragment, (b) formation
of hexacyclic aziridine 80 from the reaction of cyanide with intermediate 79 containing an N-methoxycarbonyl-
â-bromoethylamine fragment, (c) introduction of C17 by ring-opening of the aziridinium ion derived from
aziridine 80, and (d) base-promoted skeletal rearrangement of pentacyclic equatorial alcohol 82 to form
the oxacyclic ring and invert the spirooxindole functional group to provide hexacyclic gelsemine precursor
83.
gelsemine were disclosed in 1994 by the groups of Johnson,17
Speckamp,18 and Hart (21-oxogelsemine),19 additional total
syntheses were subsequently reported from the laboratories of
Introduction
At the time its structure was elucidated,2 the compact
hexacyclic cage structure of gelsemine (1) posed implicit
challenges to the capabilities of organic synthesis.3 In the
intervening 40 years, a number of approaches to gelsemine have
been described and much imaginative chemistry has been
developed in this context.4-16 The first total syntheses of (()-
(7) Stork, G.; Krafft, M. E.; Biller, S. A. Tetrahedron Lett. 1987, 28, 1035-
1038.
(8) (a) Vijn, R. J.; Hiemstra, H.; Kok, J. J.; Knotter, M.; Speckamp, W. N.
Tetrahedron 1987, 43, 5019-5030. (b) Hiemstra, H.; Vijn, R. J.; Speckamp,
W. N. J. Org. Chem. 1988, 53, 3882-3884. (c) Koot, W.-J.; Hiemstra, H.;
Speckamp, W. N. J. Org. Chem. 1992, 57, 1059-1061. (d) Dijkink, J.;
Cintrat, J. C.; Speckamp, W. N.; Hiemstra, H. Tetrahedron Lett. 1999, 40,
5919-5922.
(1) Current addresses: (a) Merck, Sharp & Dohme Laboratories, Terlings Park,
Harlow, Essex CM20 2QR, England. (b) Pfizer Inc., Groton/New London
Laboratories, Eastern Point Road, 8220-4044, Groton, CT 06340. (c)
California State University Northridge, Department of Chemistry, Northridge,
CA 91330. (d) Allergan Inc., 2525 DuPont Drive, RD 3D, Irvine, CA 92623.
(e) GlaxoSmithKline, 5 Moore Dr., Research Triangle Park, NC 27709.
(2) (a) Lovell, F. M.; Pepinsky, R.; Wilson, A. J. C. Tetrahedron Lett. 1959,
1 (4), 1-5. (b) Conroy, H.; Chakrabarti, J. K. Tetrahedron Lett. 1959, 1
(4), 6-13.
(3) For general reviews of Gelsemium alkaloids, see: (a) Saxton, J. E. In The
Alkaloids; Manske, R. H. F., Ed.; Academic Press: New York, 1965; Vol.
8, pp 93-117. (b) Bindra, J. S. In The Alkaloids; Manske, R. H. F., Ed.;
Academic Press: New York, 1973; Vol. 14, pp 83-121. (c) Liu, Z.-J.;
Lu, R.-R. ibid; Brossi, A., Ed.; Academic Press: San Diego, 1988; Vol.
33, pp 83-140. (d) Saxton, J. E. Nat. Prod. Rep. 1992, 9, 393-446. (e)
Takayama, H.; Sakai, S.-J. In The Alkaloids; Cordell, G. A., Ed.; Academic
Press: New York, 1997; Vol. 49, pp 1-78. For a review of synthetic work
in this area, see: (f) Lin, H.; Danishefsky, S. J. Angew. Chem., Int. Ed.
2003, 42, 36-51.
(4) (a) Autrey, R. L.; Tahk, F. C. Tetrahedron 1967, 23, 901-917. (b) Autrey,
R. L.; Tahk, F. C. Tetrahedron 1968, 24, 3337-3345.
(5) Johnson, R. S.; Lovett, T. O.; Stevens, T. S. J. Chem. Soc. C 1970, 6,
796-800.
(9) (a) Abelman, M. M.; Oh, T.; Overman, L. E. J. Org. Chem. 1987, 52,
4130-4133. (b) Earley, W. G.; Jacobsen, E. J.; Meier, G. P.; Oh, T.;
Overman, L. E. Tetrahedron Lett. 1988, 29, 3781-3784. (c) Earley, W.
G.; Oh, T.; Overman, L. E. Tetrahedron Lett. 1988, 29, 3785-3788. (d)
Madin, A.; Overman, L. E. Tetrahedron Lett. 1992, 33, 4859-4862. (e)
Overman, L. E.; Sharp, M. J. J. Org. Chem. 1992, 57, 1035-1038.
(10) (a) Choi, J.-K.; Ha, D.-C.; Hart, D. J.; Lee, C.-S.; Ramesh, S.; Wu, S. J.
Org. Chem. 1989, 54, 279-290. (b) Hart, D. J.; Wu, S. C. Tetrahedron
Lett. 1991, 32, 4099-4102. (c) Hart, D. J.; Wu, S. C. Heterocycles 1993,
35, 135-138.
(11) Takayama, H.; Seki, N.; Kitajima, M.; Aimi, N.; Sakai, S.-I. Nat. Prod.
Lett. 1993, 2, 271-276.
(12) Johnson, A. P.; Luke, R. W. A.; Steele, R. W.; Boa, A. N. J. Chem. Soc.,
Perkin Trans. 1 1996, 883-893.
(13) Ng, F.; Chiu, P.; Danishefsky, S. J. Tetrahedron Lett. 1998, 39, 767-770.
(14) Sung, M. J.; Lee, C.-W.; Cha, J. K. Synlett 1999, 561-562.
(15) Avent, A. G.; Byrne, P. W.; Penkett, C. S. Org. Lett. 1999, 1, 2073-2075.
(16) Pearson, A. J.; Wang, X. J. Am. Chem. Soc. 2003, 125, 13326-13327.
(17) (a) Sheikh, Z.; Steel, R.; Tasker, A. S.; Johnson, A. P. J. Chem. Soc., Chem.
Commun. 1994, 763-764. (b) Dutton, J. K.; Steel, R. W.; Tasker, A. S.;
Popsavin, V.; Johnson, A. P. J. Chem. Soc., Chem. Commun. 1994, 765-
766.
(18) (a) Newcombe, N. J.; Ya, F.; Vijn, R. J.; Hiemstra, H.; Speckamp, W. N.
J. Chem. Soc., Chem. Commun. 1994, 767-768. (b) Speckamp, W. N.;
Newcombe, N. J.; Hiemstra, H.; Ya, F.; Vijn, R. J.; Koot, W.-J. Pure Appl.
Chem. 1994, 66, 2163-2166.
(19) (a) Kuzmich, D.; Wu, S. C.; Ha, D.-C.; Lee, C.-S.; Ramesh, S.; Atarashi,
S.; Choi, J.-K.; Hart, D. J. J. Am. Chem. Soc. 1994, 116, 6943-6944. (b)
Atarashi, S.; Choi, J.-K.; Ha, D.-C.; Hart, D. J.; Kuzmich, D.; Lee, C.-S.;
Ramesh, S.; Wu, S. C. J. Am. Chem. Soc. 1997, 119, 6226-6241.
(6) (a) Fleming, I.; Michael, J. P. J. Chem. Soc., Perkin Trans 1 1981, 1549-
1556. (b) Fleming, I.; Loreto, M. A.; Michael, J. P.; Wallace, I. H. M.
Tetrahedron Lett. 1982, 23, 2053-2056. (c) Fleming, I.; Loreto, M. A.;
Wallace, I. H. M.; Michael, J. P. J. Chem. Soc., Perkin Trans. 1 1986,
349-359. (d) Clarke, C.; Fleming, I.; Fortunak, J. M. D.; Gallagher, P. T.;
Honan, M. C.; Mann, A.; Nubling, C. O.; Raithby, P. R.; Wolff, J. J.
Tetrahedron 1988, 44, 3931-3934. (e) Fleming, I.; Moses, R. C.; Tercel,
M.; Ziv, J. J. Chem. Soc., Perkin Trans. 1 1991, 617-626.
9
18054
J. AM. CHEM. SOC. 2005, 127, 18054-18065
10.1021/ja055711h CCC: $30.25 © 2005 American Chemical Society

Intramolecular Heck Reaction Forming Quaternary C Centers
A R T I C L E S
Fukuyama,20 Overman,21 and Danishefsky,22 and a total syn-
thesis of (+)-gelsemine was disclosed by Fukuyama and co-
workers in 2000.23
Scheme 1
The preceding paper in this series presented full details of
the early stages of our gelsemine total synthesis, which resulted
in a direct route for preparing azatricyclo[4.4.0.02,8]decanone 5
from 3-methylanisole (6) (Figure 1).24 The bromide functional
group of azatricyclodecanone 5 provides a potential handle for
fabricating the hydropyran ring, whereas the ketone carbonyl
was seen as a locus for generating the spirooxindole unit. At
the outset of these studies, we envisaged that this latter
elaboration would be particularly demanding as installation of
the spirooxindole on the azatricyclodecane framework introduces
a 1,3-diaxial interaction between the oxindole carbonyl group
and the angular vinyl group. As events developed, elaboration
of the spirooxindole proved to be relatively easy as a result of
the remarkable ability, first revealed in these studies,9a of
intramolecular Heck reactions to form congested quaternary
carbon centers.25 Herein we provide full details of our develop-
ment of Heck cyclization routes to the spirooxindole unit of
gelsemine, our unsuccessful efforts to elaborate the hydropyran
ring from pentacyclic intermediates 2, and the final endgame
strategy that allowed (()-gelsemine (1) to be fashioned from
3.
bonds.26 To this end, the lithium enolate of ketone 5 was trapped
with N-phenyl-bis(trifluoromethanesulfonimide) (7) to generate
enol triflate 8 in 78% yield (Scheme 1). Palladium-catalyzed
carbonylative cross-coupling of triflate 8 with 2-bromoaniline
gave 2-bromoanilide 9 in 74% yield.27 However, this coupling
proceeded in poor yield with 2-iodoaniline or N-benzyl-2-
bromoaniline.28 Protection of the amide nitrogen of 9 with a
methyl or 2-(trimethylsilyl)ethoxymethyl (SEM) group provided
R,â-unsaturated amides 10a and 10b in good yields.
We now were ready to explore generating the critical spiro-
fused oxindole fragment of gelsemine by intramolecular Heck
reaction. At the time these studies began, we had shown that
intramolecular Heck reactions could be employed to elaborate
spirooxindole units onto simple, unfunctionalized carbocyclic
rings.9a However, the projected cyclization of 9 or 10 was far
more demanding: these substrates are more highly functional-
ized, their C-C double bonds are more hindered, and two
epimeric Heck products could be formed in the Heck cycliza-
tions (eq 1). Migratory insertion from the R-face would lead to
Figure 1. Synthetic strategy.
spirooxindole 11, whereas insertion from the â-face would
provide the undesired epimer 12. Molecular mechanics (MM2)
Results and Discussion
Elaboration of the Spirooxindole Using an Intramolecular
Heck Reaction. With a workable synthesis of tricyclic ketone
5 in hand,24 the next task was to introduce the R,â-unsaturated
anilide functionality that would serve as a precursor of the
spirooxindole unit. Intermediates 4 wherein Z ) H were chosen
for our initial attempts to form the spirooxindole functionality,
because the projected cyclization was deemed speculative
enough without requiring insertion of tetrasubstituted double
(20) (a) Fukuyama, T.; Liu, G. J. Am. Chem. Soc. 1996, 118, 7426-7427. (b)
Fukuyama, T.; Liu, G. Pure Appl. Chem. 1997, 69, 501-505.
(21) Madin, A.; O’Donnell, C. J.; Oh, T.; Old, D. W.; Overman, L. E.; Sharp,
M. J. Angew. Chem., Int. Ed. 1999, 38, 2934-2936.
(22) (a) Ng, F. W.; Lin, H.; Tan, Q.; Danishefsky, S. J. Tetrahedron Lett. 2002,
43, 545-548. (b) Hong, L.; Ng, F. W.; Danishefsky, S. J. Tetrahedron
Lett. 2002, 43, 549-551. (c) Ng, F. W.; Lin, H.; Danishefsky, S. J. J. Am.
Chem. Soc. 2002, 124, 9812-9824.
(23) Yokoshima, S.; Tokuyama, H.; Fukuyama, T. Angew. Chem., Int. Ed. 2000,
39, 4073-4075.
(24) See the preceding paper in this journal.
9
J. AM. CHEM. SOC. VOL. 127, NO. 51, 2005 18055

A R T I C L E S
Madin et al.
Table 1. Heck Cyclization of 10 Using Different Palladium
Catalysts
oxindole 11, alkylpalladium(II) halide complex 14 (eq 2), we
investigated the use of chelating diphosphine ligands. To our
reaction conditions
additiveb
solvent T (
oxindole
entry substrate
catalysta
°
C) t (h) yield, %c
11:12d
1
2
3
4
5
6
7
8
10a
10b
10b
10b
10b
10b
10b
10b
Pd(Ph3P)4 Et3N
Pd(Ph3P)4 Et3N
Pd(Ph3P)4 Et3N
Pd(dppe)e Et3N
Pd(dppf)e Et3N
Pd2(dba)3 Et3N
Pd2(dba)3 Et3N
MeCN
MeCN
THF
THF
THF
82 18 81
82 15 66
66 24 20f
66 24 46f
66 36 (18)
66 24 75f
66:34
60:40
60:40
55:45
50:50
73:27
THF
PhMe 110
1
80-95g 89:11
66 36 77 3:97
Pd2(dba)3 Ag3PO4 THF
a 10-20% was employed. b 10-15 equiv of Et3N or 2 equiv of Ag3PO4
(per equiv of 10) was employed. c Isolated yields of the isomer mixture,
yields in parentheses are % conversion by capillary GC analysis (peak area
1
% without calibration). d Capillary GC or 500 MHz H NMR analysis; in
surprise, replacing Ph3P with bis(diphenylphosphino)ethane
(dppe) in cyclizations of R,â-unsaturated anilide 10b carried
out in refluxing THF had little effect on stereoselection (entry
4). Stereoselectivity was also poor when the larger 1,1′-bis-
(diphenylphosphino)ferrocene (dppf) ligand was employed
(entry 5). However, cyclizations of R,â-unsaturated anilide 10b
catalyzed by tris(dibenzylideneacetone)dipalladium [Pd2(dba)3]
without added phosphine ligands were more selective (entries
6 and 7).31-33 Spirooxindoles 11b and 12b were generated in a
9:1 ratio, respectively, and in good yield using 10 mol % of
[Pd2(dba)3] as the catalyst in refluxing toluene.
In contrast, Heck cyclization of precursor 10b conducted in
the presence of Ag3PO4 without phosphine ligands occurred with
virtually complete selectivity to give the epimeric oxindole 12b
(entry 8). We attribute the high selectivity (97:3) in this case to
coordination of the angular vinyl group during the insertion step
(eq 3). Apparently, this coordination is enhanced by the
contrast to the methyl analogue, isomers 11b and 12b could not be separated
on silica gel. e Prepared in situ from 1 equiv of Pd2(dba)3 and 2 equiv of
the diphosphine. f The majority of the remaining mass was recovered 10b.
g Multiple experiments conducted with 100-300 mg of 10b.
calculations indicated that oxindole 11, wherein the aromatic
ring is equatorially disposed, is more stable (∼3 kcal/mol);
however, it was not clear to what extent at all this preference
would be felt in the stereodetermining step.
Salient results of our extensive studies of Heck cyclizations
in the gelsemine series are summarized in Table 1. Initial
attempts to cyclize secondary amide 9 failed, giving only
reduction product. This result was not unexpected, because
cyclization of a secondary amide requires population of a high-
energy E amide conformation. In contrast, tertiary anilides 10a,b
cyclized under a variety of Heck reaction conditions. Cycliza-
tions were examined initially using (Ph3P)4Pd as the catalyst in
refluxing acetonitrile. Changing the amide protecting group from
Me to SEM had little effect on the low stereoselectivity observed
in the reaction (entries 1 and 2).29 Cyclization was more sluggish
in refluxing tetrahydrofuran; however, stereoselectivity remained
unaltered (entry 3). The configuration of oxindole products 11
and 12 was determined readily by1H NMR NOE experiments.
For diastereomer 11, a positive NOE is observed between the
hydrogens at C9, C5, and C16, whereas, for epimer 12, a
positive NOE is seen between the C9 and C19 hydrogens.30
With the aim of reducing the severe steric interactions that
exist between the phosphine ligands and the two-carbon bridge
of the azatricyclic ring in the presumed precursor of spiro-
dissociation of the bromide ligand to give a more electrophilic
cationic palladium(II) intermediate.9d Consistent with this
proposal, Heck cyclization under identical conditions of the
analogue of R,â-unsaturated anilide 10b in which the angular
vinyl group is replaced by an ethyl substituent provided a 1:1
mixture of oxindole diastereomers.9d
(25) For a comprehensive review of the intramolecular Heck reaction, see: (a)
Link, J. T. Org. React. 2002, 60, 157-534. For reviews of the utility of
intramolecular Heck reactions in complex molecule construction, see: (b)
Link, J. T.; Overman, L. E. Chemtech 1988, 28, 19-26. (c) Overman, L.
E.; Link, J. T. In Cross Coupling Reactions; Diederich, F.; Stang, P. J.,
Eds.; VCH: Weinheim; 1998; pp 231-269. (d) Dounay, A. B.; Overman,
L. E. Chem. ReV. 2003, 103, 2945-2964.
(26) At the time there was only a single example of successful intramolecular
Heck cyclization of a tetrasubstituted double bond.9a Even today, despite
the recent intense investigations in this area, there are only a few examples
of intramolecular Heck cyclizations of tetrasubstituted olefins, most from
various investigations in our laboratories.25a
(27) Cacchi, S.; Ciattini, P. G.; Morera, E.; Ortar, G. Tetrahedron Lett. 1986,
27, 3931-3934.
(28) The analogous iodo amide can be prepared by a two step sequence,
involving initial palladium-catalyzed carbonylation of 8 in methanol27
followed by aminolysis of the resulting methyl ester with the dimethyl-
aluminum amide of 2-iodoaniline.
(29) Similar success in this Heck cyclization was observed when the amide
protecting group was a benzyl or methoxymethyl group. The iodide
analogue of 10a cyclized with similar low stereoselectivity.
(30) The C19 proton of the terminal vinyl group is observed at lower field in
the 1H NMR spectrum of the oxindole isomer that places the carbonyl group
proximal to C19 than in its epimer. For example, for epimers 11a/11b this
signal is seen at 6.7 ppm, whereas it is seen at 6.2 ppm in isomers 12a/
12b.
Attempts to Construct the Tetrahydropyran Ring of
Gelsemine from Pentacyclic Intermediate 11b. With five of
the six rings of gelsemine assembled, what remained was
introduction of one carbon and elaboration of the tetrahydro-
pyran ring. Our anticipation that this task would be straight-
forward turned out to be a remarkably poor assessment. Our
initial plan was to displace the bromide substituent of 11b with
cyanide anion, convert the resulting endo-oriented nitrile to a
primary alcohol, and engage this group and the alkene to form
the final ring of gelsemine. With this strategy in mind, a 9:1
mixture of pentacyclic oxindoles 11b/12b (formed as described
in entry 9 of Table 1) was allowed to react with NaCN in hot
dimethyl sulfoxide (Scheme 2). To our surprise, this reaction
produced aziridines 15/16 in a 9:1 ratio and 85% yield; none
of the expected nitrile product was formed. These aziridine
epimers could be separated on silica gel, providing the major
9
18056 J. AM. CHEM. SOC. VOL. 127, NO. 51, 2005

Intramolecular Heck Reaction Forming Quaternary C Centers
A R T I C L E S
Scheme 2
Scheme 3
stereoisomer 15 in 76% yield from the 11b/12b Heck product.
Whether this unusual reaction occurs by an SN1 process, a
double displacement sequence, or another mechanism has not
been determined.
congener of pentacyclic intermediate 18 could be selectively
converted to the corresponding primary amine without reduction
of the oxindole. To examine this issue, the SEM protecting
group of 18 was removed by reaction with 6 M HCl, providing
pentacyclic oxindole 22 (Scheme 3). This crystalline product
yielded single crystals, allowing its structure and the configu-
ration of the major Heck product to be rigorously established.
Reaction of unprotected oxindole 22 with triisopropylsilyl
(TIPS) trifluoromethanesulfonate in CH2Cl2 in the presence of
diisopropylethylamine cleanly introduced the TIPS group onto
the oxindole nitrogen to give intermediate 23 in 80% yield.
Subsequent reduction of 23 with AlH3‚Me2EtN38 in THF at
room temperature generated primary amine 24. As we hoped,
the bulky TIPS group had shielded the oxindole carbonyl from
reduction.39 This polar intermediate was not purified but directly
acylated with either N-(trifluoroacetoxy)succinimide40 or pen-
tafluorophenyl acetate41 to give trifluoroacetamide 25 or acet-
amide 26 in 40-50% overall yields from pentacyclic intermedi-
ate 23. As the TIPS group had been cleaved during these
acylation steps, the oxindole nitrogen was reprotected by
reaction of amides 25 and 26 with benzoic anhydride to yield
imides 27 and 28.
The formation of aziridine 15 did not appear to sabotage our
plans because the aromatic ring of the oxindole should block
backside attack at C5 of this and related intermediates. Thus,
we anticipated that nucleophiles would attack an electrophilic
aziridinium ion derived from hexacyclic aziridine 15 at the C16
carbon. In practice it was found that reaction of aziridine 15
with methyl trifluoromethanesulfonate at room temperature in
CH2Cl2 formed the stable N-methylaziridinium salt 17, which,
when exposed to an excess of NaCN in dimethyl sulfoxide at
90 °C, produced a single pentacyclic nitrile 18 in 84% yield.
We turned to examine elaboration of the endo-oriented nitrile
substituent of pentacyclic intermediate 18 to a hydroxymethyl
group. Attempted reduction of the nitrile group of 18 with
diisobutylaluminum hydride (DIBALH), sodium bis(2-methoxy-
ethoxy)aluminum hydride (RedAl), LiAl(OEt)3H, NaBH4, or
NaBH4‚CoCl2 under a wide variety of reaction conditions
resulted either in recovery of starting material or production of
unidentifiable products. One exception was the reaction of
pentacyclic nitrile 18 with LiAlH4 in Et2O at -78 °C, which
delivered indolenine 19 in 80% yield upon aqueous workup
(Scheme 2). Further reduction of this product with LiAIH4 in
Et2O at -20 °C provided indoline 20. It is remarkable that the
endo nitrile group was untouched by LiAIH4 under both
conditions. The only reagent that reduced the nitrile group of
intermediate 18 cleanly was AlH3,34 which delivered pentacyclic
triamine 21 in 93% yield (eq 4).
(31) In a preliminary description of these studies,9d these conditions were
inappropriately described as “ligandless”; however, dba is likely bound to
palladium during the Heck insertion sequence.32
(32) Amatore, C.; Broeker, G.; Jutand, A.; Khalil, F. J. Am. Chem. Soc. 1997,
119, 5176-5185.
(33) These conditions were employed by Speckamp and Hiemstra for construc-
tion of the spirooxindole in their early synthesis of (()-gelsemine.18
(34) Yoon, N. M.; Brown, H. C. J. Am. Chem. Soc. 1968, 90, 2927-2938.
(35) Nikolaides, N.; Ganem, B. J. Org. Chem. 1989, 54, 5996-5998.
(36) (a) White, E. H. J. Am. Chem. Soc. 1955, 77, 6008-6010. (b) White, E.
H. J. Am. Chem. Soc. 1955, 77, 6011-6014. (c) White, E. H. J. Am. Chem.
Soc. 1955, 77, 6014-6022. (d) White, E. H. Organic Syntheses; Wiley:
New York, 1973; Collect. Vol. V, pp 336-339.
(37) Romea, P.; Urpi, F.; Vilarrasa, J. J. Org. Chem. 1989, 54, 3209-3211.
(38) This complex was found to be slightly more selective than AlH3; see:
Marlett, E. M.; Park, W. S. J. Org. Chem. 1990, 55, 2968-2969.
(39) Attempts to reduce the nitrile group of 22 or 23 with DIBALH or LiAlH4
resulted in unidentifiable products.
(40) Bergeron, R. J.; McManis, J. S. J. Org. Chem. 1988, 53, 3108-3111.
(41) Kisfaludy, L.; Mohacsi, T.; Low, M.; Drexler, F. J. Org. Chem. 1979, 44,
654-656.
As a primary amine could plausibly serve as a precursor of
a primary alcohol,35-37 we explored the use of other oxindole
protecting groups in the hopes that the nitrile substituent of a
9
J. AM. CHEM. SOC. VOL. 127, NO. 51, 2005 18057

A R T I C L E S
Madin et al.
Scheme 4
One of the better methods for converting aliphatic primary
amines to alcohols involves thermal rearrangement of N-
nitrosoamide intermediates. However, all our attempts to convert
trifluoroacetamide 27 or acetamide 28 into the corresponding
N-nitrosoamide failed. For example, reaction of amides 27 or
28 with NaNO2 in Ac2O-HOAc35 or with N2O4 in the presence
of NaOAc36 or pyridine37 only returned unchanged starting
material. Subjecting amide 28 to the more electrophilic reagent
37
NOBF4 under a variety of reaction conditions resulted only
in decomposition of this substrate.
In light of our inability to generate pentacyclic alcohol 29
from an amine precursor, we returned to examine other
potentially useful transformations of the nitrile group of
pentacyclic intermediate 18. Initially we explored basic hy-
drolysis. Reaction of 18 with LiOH in refluxing methanol, or
NaOH in refluxing ethanol, resulted only in epimerization of
the nitrile group of precursor 18 to give mixtures of 18 and its
epimer 30 (eq 5).42 Hydrolysis of pentacyclic nitrile 18 under
of ester 32. Screening the same set of bases using tert-
butyldimethylsilyl chloride as an electrophilic quenching agent
also returned only 32, suggesting that the hindered endo proton
at C16 was not being removed.
We also examined acidic conditions for elaborating the
hindered endo nitrile group of pentacyclic intermediate 22 in
the hopes that epimerization at C16 would be less problematic
(Scheme 4). Dissolving this precursor in concentrated H2SO4
led, in low yield, to the formation of hexacyclic carboxamide
33, the structure of which was confirmed by X-ray crystal-
lographic analysis. Although the nitrile group had been con-
verted to a carboxamide without epimerization, these strongly
acidic conditions had also promoted cyclization of the oxindole
onto the C19 terminal vinyl group.46 However, all attempts at
promoting the cyclization of the γ,δ-unsaturated carboxamide
unit of precursor 33 (or the tetracyclic oxindole derived from
33 upon treatment with aqueous HCl) by reaction with I2,47
more forcing conditions (5 M KOH in ethylene glycol at 150
°C), followed by treating the crude reaction mixture with
diazomethane, provided a single methyl ester product in 63%
yield. However, it was readily ascertained that this product was
the exo ester 32, most likely resulting from epimerization of
the nitrile group prior to hydrolysis. Additionally, treatment of
nitrile 18 with basic hydrogen peroxide under phase transfer
conditions again yielded an epimerized product, in this case the
corresponding primary amide.
As a last-ditch effort to salvage these intermediates, we
explored whether pentacyclic ester 32 could be epimerized by
kinetic protonation of its derived enolate.43 Although such a
sequence could be realized with a simpler azatricyclo[4.4.0.02,8]-
decyl ester,44 we never succeeded in accomplishing this epimer-
ization in the pentacyclic series. For example, exposure of ester
32 to excess lithium diisopropylamide (LDA), lithium pyrroli-
dide, lithium diethylamide, potassium diisopropylamide, potas-
sium bis(trimethylsilyl)amide (KHMDS), lithium hydride, or
potassium hydride (with or without N,N′-dimethylpropyleneurea,
DMPU), followed by quenching with malononitrile failed to
produce C16-epi 32.45 In all cases, only the starting ester was
obtained. As it was not clear whether deprotonation was taking
place, we attempted to prepare the silyl ketene acetal derivative
51
PhSeBr,48 Br2,49 Hg(OAc)2,50 or Hg(OCOCF3)2 were unsuc-
cessful.
With these disheartening results in hand, we abandoned
attempts to generate the hydropyran ring of gelsemine from a
pentacyclic precursor having an endo-oriented C16 nitrile
substituent. We returned to aziridinium ion 17 to examine its
opening with other one-carbon nucleophiles. Unfortunately,
reactions of hexacyclic aziridinium ion 17 with nucleophiles
such as (allyldimethylsilyl)methylmagnesium bromide,52 (iso-
propoxydimethylsilyl)methylmagnesium bromide,53 lithium acetyl-
ide, propynyllithium, 1,4-dioxen-2-yllithium,54 and the Creger-
Silbert dianion, LiCH2CO2Li,55 failed to give any identifiable
products resulting from ring-opening at C16.
(45) For reviews of diastereoselective protonations, see: (a) Zimmerman, H.
E. Acc. Chem. Res. 1987, 20, 263-268. (b) Fehr, C. Chimia 1991, 45,
253-261. (c) Krause, N.; Ebert, S.; Haubrich, A. Liebigs Ann./Recueil 1997,
2409-2418. (d) Eames, J.; Weerasooriya, N. J. Chem. Res., Synop. 2001,
2-8.
(46) Several examples of electrophile-promoted cyclization of the oxindole and
vinyl groups of gelsemine were encountered during early degradation
studies.3a
(42) The configuration of the nitrile substituent of 18 and 30 was determined
by correlating the coupling constants predicted by the Karplus-Conroy
equation with the observed coupling constants. In 18 the dihedral angle
between the hydrogens at C16 and C5 is ∼90° and that between the
hydrogens at C16 and C15 is ∼15°, so a doublet with a 7 Hz coupling
constant is predicted; the observed coupling is 7.2 Hz. Likewise for 30,
the dihedral angle between the hydrogens at C16 and C5 is ∼30° and that
between the hydrogens on C16 and C15 is approximately 90°, so a doublet
with a 4 Hz coupling constant is predicted; the observed coupling is 4.2
Hz.
(47) Knapp, S.; Levorse, A. T. J. Org. Chem. 1988, 53, 4006-4014.
(48) Clive, D. L. J.; Wong, C. K.; Kiel, W. A.; Menchen, S. M. J. Chem. Soc.,
Chem. Commun. 1978, 379-380.
(49) Craig, P. N. J. Am. Chem. Soc. 1952, 74, 129-131.
(50) (a) Danishefsky, S.; Taniyama, E.; Webb, R. R. Tetrahedron Lett. 1983,
24, 11-14. (b) Danishefsky, S.; Taniyama, E. Tetrahedron Lett. 1983, 24,
15-18.
(51) (a) Takacs, J. M.; Helle, M. A.; Takusagawa, F. Tetrahedron Lett. 1989,
30, 7321-7324. (b) Amoroso, R.; Cardillo, G.; Tomasini, C.; Tortoreto,
P. J. Org. Chem. 1992, 57, 1082-1087.
(52) Tamao, K.; Ishida, N. Tetrahedron Lett. 1984, 25, 4249-4252.
(53) Tamao, K.; Ishida, N.; Kumada, M. J. Org. Chem. 1983, 48, 2120-2122.
(54) (a) Fetizon, M.; Hanna, I.; Rens, J. Tetrahedron Lett. 1985, 26, 3453-
3456. (b) Fetizon, M.; Goulaouic, P.; Hanna, I. Synthesis 1987, 503-505.
(43) A related epimerization was employed by Fukuyama and Liu in their
synthesis of (()-gelsemine.20
(44) For a detailed account of these studies, see: Old, D. W. Ph.D. Dissertation,
University of California, Irvine, 1997.
9
18058 J. AM. CHEM. SOC. VOL. 127, NO. 51, 2005

Intramolecular Heck Reaction Forming Quaternary C Centers
A R T I C L E S
Scheme 5
Scheme 6
As a final effort to elaborate aziridinium ion 17, we examined
its reaction with the potent one-carbon nucleophile, disodium
tetracarbonylferrate (35, Collman’s reagent).56 As this reagent
reacts with five- and six-carbon unsaturated tosylates and halides
to yield cycloalkanones,57,58 the expected product of its union
with aziridinium ion 17 was hexacyclic ketone 37 (Scheme 5).
We conjectured that Baeyer-Villiger oxidation of this product
could lead to lactone 38, which surely would serve as a precursor
of gelsemine (1).
spirooxindole was in place (2 f 44, Figure 2). This timing of
events had been chosen because we anticipated that the presence
of an endo substituent at C16 would compromise the stereo-
selectivity of the intramolecular Heck reaction by shielding the
lower face of the cycloalkene. Much to our surprise, in 1994
Speckamp and co-workers18 reported just such an intramolecular
Heck reaction that proceeded with modest stereoselectivity to
form the desired spirooxindole using the Pd2(dba)3 conditions
we had introduced.9d This disclosure prompted us to investigate
installation of the C17 hydroxymethyl group prior to Heck
cyclization. We anticipated that the absence of the endo-oriented
aryl ring of the oxindole might simplify elaboration of the nitrile
to a hydroxymethyl group. Therefore, intermediate 45 became
our next subgoal.
The preparation of tricyclic ketone 45 is summarized in
Scheme 7. Ketone 5 was readily elaborated to nitrile 48 by way
of tetracyclic aziridine 47.60 Initial survey experiments showed
that reduction of nitrile 48 with diisobutylaluminum hydride
(DIBALH) was facile, even at -78 °C. However, hydrolysis
of the resulting dialkylaluminum imine was not straightforward,
with many standard conditions promoting extensive epimeriza-
tion of the initially generated endo formyl group. Best results
were obtained using aqueous acetic acid. After careful basifi-
cation, amino aldehyde 49 was obtained as a 4:1 mixture of
In the event, reaction of hexacyclic aziridinium ion 17 with
Collman’s reagent 35 in N-methylpyrrolidinone (NMP) at 50
°C under a CO atmosphere did not yield hexacyclic ketone 37
but instead produced a rearranged product, the isomeric hexa-
cyclic ketone 39, in 59% yield (Scheme 6).9e A possible
mechanism for this unexpected and unprecedented reorganiza-
tion has been suggested.9e Although hexacyclic product 39 was
not a useful intermediate en route to gelsemine, it did provide
one important insight. Exposure of this material to 1 M NaOH
solution in acetone at room temperature caused rapid and
complete epimerization at the spiro quaternary center to deliver
oxindole epimer 42 in 90% yield. Removal of the SEM
protecting group from this product gave hexacyclic oxindole
43, which provided single crystals suitable for X-ray analysis.
Epimerization of the 1,5-dicarbonyl compound 39 likely occurs
by retro-Michael fragmentation to generate oxindole enolate 40,
rotation about the C6-C7, and Michael ring closure of rotamer
41 to provide epimer 42. The driving force for this epimerization
is undoubtedly relief of steric interactions between the aromatic
ring of the oxindole and the underside of the cup-shaped
azatetracyclic moiety of 39. Base-promoted epimerization of
an oxindole intermediate turned out to play an important role
in our ultimately successful route to gelsemine (vide infra).59
Attempts to Elaborate the Tetrahydropyran Ring Prior
to Forming the Spirooxindole. Up to this point, our strategy
had been to elaborate a hydroxymethyl group at C16 after the
(55) (a) Creger, P. L. J. Org. Chem. 1972, 37, 1907-1918. (b) Danishefsky,
S.; Schuda, P. F.; Kitahara, T.; Etheredge, S. J. J. Am. Chem. Soc. 1977,
99, 6066-6075.
(56) (a) Cooke, M. P., Jr. J. Am. Chem. Soc. 1970, 92, 6080-6082. (b) Collman,
J. P.; Finke, R. G.; Cawse, J. N.; Brauman, J. I. J. Am. Chem. Soc. 1977,
99, 2515-2526. (c) Collman, J. P. Acc. Chem. Res. 1975, 8, 342-347.
(57) Me´rour, J. Y.; Roustan, J. L.; Charrier, C.; Collin, J.; Bena¨ım, J. J.
Organomet. Chem. 1973, 51, C24-C26.
(58) (a) McMurry, J. E.; Andrus, A.; Ksander, G. M.; Muser, J. H.; Johnson,
M. A. J. Am. Chem. Soc. 1979, 101, 1330-1332. (b) McMurry, J. E.;
Andrus, A.; Ksander, G. M.; Muser, J. H.; Johnson, M. A. Tetrahedron
1981, 37, Supplement No. 1, 319-327.
Figure 2. Potential alternate strategy in which spirooxindole formation is
accomplished after elaboration at C16.
(59) The Hart total synthesis of (()-21-oxogelsemine also features a base-
promoted epimerization of a spirocyclic oxindole intermediate.
9
J. AM. CHEM. SOC. VOL. 127, NO. 51, 2005 18059

A R T I C L E S
Scheme 7
Madin et al.
Figure 3. Revised plan for elaborating the hydropyran ring.
formyl epimers favoring the desired endo isomer. Reduction of
this crude product with NaBH4 and cleavage of the dioxolane
group of product 50 gave tricyclic ketone 45 in 56% overall
yield from nitrile 48. However, this sequence, particularly the
nitrile reduction step, could not be accomplished successfully
on anything larger than exploratory scales. This unsatisfactory
outcome forced us to abandon this approach.
Scheme 8
The Final Successful Route to (()-Gelsemine. Up to this
point, most of the strategies we had investigated for forming
the oxacyclic ring of gelsemine envisaged elaboration of
pentacyclic nitrile 2 to intermediates that would allow the
hydropyran ring to be constructed by forming the C3-O bond
(Figure 1). Two factors, both arising from the sterically
congested environment of C17 of the gelsemine ring system,
had thwarted these efforts: facile epimerization at C16 when
C17 is an electron-withdrawing substituent and the difficulty
in accomplishing common functional-group interconversions at
C17 in pentacyclic intermediates. We turned to an alternate
strategy that would side step these problems by constructing
the tetrahydropyran ring of gelsemine by forming the O-C17
bond. Such an approach could allow the critical functionalization
of the hindered endo nitrile to be accomplished in intramolecular
fashion as suggested in Figure 3 (51 f 52). An attraction of
this strategy was the expected ease of reduction of the C3
carbonyl group of intermediates such as pentacyclic keto nitrile
51 as the reductant could approach from the less-encumbered
convex â-face. We entertained the possibility of accessing 51
from pentacyclic enol ethers such as 53, which might be
available from Heck cyclization of R,â-unsaturated anilide 54
having a heteroatom substituent at C3. A pivotal issue would
be whether the tetrasubstituted double bond of 54 would
participate in an intramolecular Heck reaction. Heck insertions
of tetrasubstituted double bonds are rare, and none had been
described for a substrate approaching the complexity of 54.26
Moreover, the double bond of 54 might be a particularly poor
ligand because it is part of a delocalized vinylogous carbamate
or thiocarbamate.
triflate intermediate 56. Tricyclic ketone 5 was initially con-
verted to R,R-dithio derivative 55 by reaction with an excess
of potassium hexamethyldisilazide (KHMDS) and methyl meth-
anethiosulfonate. This intermediate was then desulfenylated by
reaction with sodium methylmercaptide in THF, and the
resulting sodium enolate was trapped with N-phenyltriflimide
to give enol triflate 56. Palladium-catalyzed carbonylation of
56 in methanol27 then provided vinylogous thiocarbonate 57.
As in other Heck reactions investigated in our laboratories, the
presence of the â-sulfur substituent did not have a deleterious
effect on this palladium-catalyzed reaction.61 Weinreb aminoly-
sis of ester 5762 and protection of the resulting secondary amide
with an SEM group generated the highly functionalized Heck
cyclization substrate 58 in 30% overall yield from azatricyclic
ketone 5.
As preliminary scouting experiments had shown that installing
an alkoxy group into the R-position of ketone 5 would be
difficult, we examined initially this revised plan in the sulfur
series (Scheme 8). After failing to cleanly monosulfenylate
enolate derivatives of 5, we developed a novel route to enol
We next examined the propensity of â-methylthio R,â-
unsaturated anilide 58 to undergo intramolecular Heck reaction.
Cyclization conditions such as those listed in Table 1 that
(60) We examined also, without success, opening the methyl aziridinium ion
(61) See, for example: Hynes, J., Jr.; Overman, L. E.; Nasser, T.; Rucker, P.
derived from 46 with a variety of one-carbon nucleophiles.
V. Tetrahedron Lett. 1998, 39, 4647-4650.
9
18060 J. AM. CHEM. SOC. VOL. 127, NO. 51, 2005

Intramolecular Heck Reaction Forming Quaternary C Centers
A R T I C L E S
Scheme 9
trimethyl- or tert-butyldimethylsilyl analogues) with m-chloro-
perbenzoic acid (m-CPBA),71 peracetic acid buffered with
sodium carbonate, dimethyldioxirane, or osmium tetroxide also
did not generate the desired R-siloxyketone. We finally dis-
covered that addition of a CH2Cl2 solution of crude enoxytri-
ethylsilane 62 to a solution of iodosobenzene and boron
trifluoride etherate in methanol at -78 °C and then allowing
the temperature to rise to 0 °C72 produced R-methoxy ketone
63 in useful yield. Although attempted purification of this
product on silica gel led to its decomposition, a singlet at δ 3.4
1
ppm in the H NMR spectrum of this crude product left little
doubt that a methoxy group had been incorporated. When crude
63 was treated with KHMDS at -78 °C in THF, and the
resulting enolate was quenched with 2-[N,N-bis(trifluoro-
methylsulfonyl)amino-5-chloropyridine (Comins’ reagent),73
â-methoxy triflate 64 was formed in 61% overall yield from
ketone 5.74
To our initial surprise, reaction of 64 with 2-bromoaniline
under an atmosphere of carbon monoxide and a catalytic amount
of tetrakis(triphenylphosphine)palladium(0) in DMF at 80 °C
gave rise to none of the desired anilide but instead provided
R-methoxy enone 65 in 58% (eq 6). This unusual oxidation,
succeed with analogous substituents lacking a â-substituent
failed to promote the cyclization of 58. However, the desired
intramolecular Heck reaction was accomplished successfully
under forcing conditions (150 °C) using (Ph3P)4Pd as the catalyst
to give a 1:2 mixture of stereoisomeric tetracyclic products 59
and 60 in a combined yield of 69%. Configurational assignments
for these adducts followed from diagnostic signals for the C19
vinyl hydrogens in 1H NMR spectra.30 Attempts to improve the
diastereoselectivity of this reaction by using Pd2(dba)3 as the
catalyst in the absence of phosphine ligands resulted in no
reaction even at 150 °C. At this high temperature, the fine
suspension of palladium that is originally present coagulates, a
factor that is likely responsible for the unreactivity of substrate
58 under these conditions.
Extensive studies to optimize the Heck cyclization of unsatu-
rated anilide 58 were not carried out because we soon discovered
that cleavage of the vinyl sulfide functionality of these adducts
would not be straightforward. Preliminary calibration experi-
ments in which 59 was subjected to standard conditions for
cleaving a vinylsulfide, for example Hg(OCOCF3)2-aqueous
trifluoroacetic acid (TFA) or TFA alone in several different
solvents, resulted in either decomposition or cleavage of the
SEM group.63 Because of these difficulties, we turned to
examine the related sequence in the oxygen series.
As alluded to earlier, introduction of an alkoxy group into
the R position of ketone 5 proved to be extremely challenging
(Scheme 9). Enolization of 5 with KHMDS or LDA, followed
by quenching the derived enolate with Davis’ oxaziridine,64
returned starting material. Likewise, enolization with LDA
followed by attempted oxidation with dimethyldioxirane,65
MoO5‚pyridine‚HMPA,66 or LiOOt-Bu67 returned starting mate-
rial, as did the reaction of ketone 5 with potassium tert-butoxide,
molecular oxygen, and triethyl phosphite.68 Treatment of ketone
5 with potassium hydroxide and 2-iodosobenzoic acid or
iodosobenzene diacetate69 led to unidentifiable products, as did
attempted oxidation of 5 with N-chlorosuccinimide.70 Our
attempts to oxidize the enoxytriethylsilane derivative 62 (or
which we have shown to be general for enol triflates having
alkoxy or thioalkoxy â-substituents, likely involves solvolysis
of the â-methoxy vinyl triflate with loss of trifluoromethane-
sulfinic acid.75
Fortunately, this side reaction could be prevented by carrying
out the carbonylation at 80 °C using bis(diphenylphospine)-
ferrocenylpalladium(II) chloride as the catalyst in DMF-
methanol (1:2) under 1 atm of carbon monoxide (Scheme 10).
Under these conditions, the crystalline vinylogous carbonate 66
was produced in 94% yield; the structure of this intermediate
was confirmed by single-crystal X-ray analysis. Subsequent
reaction of vinylogous carbonate 66 with the dimethylaluminum
amide of 2-iodoaniline62 in CH2Cl2 at 0 °C and slowly letting
the reaction warm to room temperature over a period of 2 h
delivered iodoanilide 67 in 81% yield. It was essential that this
reaction be initiated below room temperature; otherwise a
significant amount of the vinylogous urea resulting from 1,4
addition of the aluminum amide to 67 was observed as a
byproduct. Finally protection of the secondary amide with a
methoxymethyl (MOM) group provided â-methoxy R,â-
unsaturated anilide 68 in 86% yield.
The intramolecular Heck reaction of â-methoxy unsaturated
anilide 68 was investigated in detail. We quickly found that
Heck cyclization of this intermediate could be accomplished
under milder conditions than those required to cyclize the
(62) Lipton, M. F.; Basha, A.; Weinreb, S. M. Organic Syntheses; Wiley: New
York, 1988; Collect. Vol. VI, pp 492-495.
(63) (a) Corey, E. J.; Shulman, J. I. J. Org. Chem. 1970, 35, 777-780. (b)
Trost, B. M.; Stanton, J. L. J. Am. Chem. Soc. 1975, 97, 4018-4025.
(64) (a) Davis, F. A.; Sheppard, A. C. Tetrahedron 1989, 45, 5703-5742. (b)
Davis, F. A.; Chen, B.-C. Chem. ReV. 1992, 92, 919-934.
(65) Guertin, K. R.; Chan, T.-H. Tetrahedron Lett. 1991, 32, 715-718.
(66) Vedejs, E.; Engler, D. A.; Telschow, J. E. J. Org. Chem. 1978, 43, 188-
196.
(71) Rubottom, G. M.; Vazquez, M. A.; Pelegrina, D. R. Tetrahedron Lett. 1974,
15, 4319-4322.
(72) Moriarty, R. M.; Prakash, O.; Duncan, M. P.; Vaid, R. K.; Musallam, H.
A. J. Org. Chem. 1987, 52, 150-153.
(67) Julia, M.; Jalmes, V. P.-S.; Ple´, K.; Verpeaux, J.-N.; Hollingworth, G. Bull.
Soc. Chim. Fr. 1996, 133, 15-24.
(73) Comins, D. L.; Dehghani, A. Tetrahedron Lett. 1992, 33, 6299-6302.
(74) Trapping the resulting enolate with N-phenyl-bis(trifluoromethanesulfon-
imide) was successful but gave inconsistent yields.
(68) Bailey, E. J.; Barton, D. H. R.; Elks, J.; Templeton, J. F. J. Chem. Soc.
1962, 1578-1591.
(69) Moriarty, R. M.; Hou, K.-C. Tetrahedron Lett. 1984, 25, 691-694.
(70) Vaz, A. D. N.; Schoellmann, G. J. Org. Chem. 1984, 49, 1286-1288.
(75) Hynes, J., Jr.; Nasser, T.; Overman, L. E.; Watson, D. A. Org. Lett. 2002,
4, 929-931.
9
J. AM. CHEM. SOC. VOL. 127, NO. 51, 2005 18061

A R T I C L E S
Scheme 10
Madin et al.
assignment for 71 followed from its mass spectrum, where the
parent peak at 394 amu corresponded to loss of HBr from the
expected pentacyclic ketone. The 13C NMR spectrum of
cyclopropyl ketone 71 could be fully assigned by an HMQC
experiment, showing a carbonyl carbon at 201.6 ppm and signals
for C14, C15, and C16 at 24.1, 24.5, and 28.1 ppm, respectively.
Of most significance, these cyclopropane carbons had diagnostic
C-H coupling constants in the range of 166-180 Hz.77
Fortunately, cyclopropane formation could be avoided by
carrying out the hydrolysis of enol ether 69 at room temperature
using a 1:2 mixture of concentrated HCl and methanol (Scheme
11). In this way, tetracyclic ketone 72 was obtained in 98%
yield, with its structure being confirmed by X-ray crystal-
lographic analysis. In light of our earlier results, it was not
surprising to find that product 72 lost HBr when heated at 50
°C in 1 M HCl in methanol, presumably via its enol, to form
cyclopropyl ketone 71.
We were now poised to investigate epimerization of the
spirooxindole. Hart had shown that a pentacyclic ketone
somewhat analogous to intermediate 72 having a carbonyl group
at C3 and the unnatural oxindole stereochemistry undewent
epimerization at C7 upon treatment with potassium cyanide.19b
The mechanism proposed for this oxindole inversion involves
initial formation of the C3 cyanohydrin, retro-aldolization to
break the C3-C7 bond, rotation about the C6-C7 σ bond, and
acylation to regenerate the C3-C7 bond. However in our case,
treating ketone 72 with a catalytic amount of KCN in DMF at
50 °C resulted only in forming hexacyclic cyclopropyl ketone
71. Attempts to form the cyanohydrin derivative of 72 under
nonbasic conditions by reaction with trimethylsilyl cyanide in
the presence of zinc iodide78 also led to hexacyclic product 71,
whereas attempted reaction of ketone 72 with diethylaluminum
cyanide79 (-30 °C f room temperature in THF) returned
starting material.
â-methylthio analogue 58. For example, reaction of â-methoxy
precursor 68 with 20 mol % Pd2(dba)3 at 110 °C in the absence
of phosphine ligands (conditions of Table 1, entry 9) provided
pentacyclic products 69 and 70 in a 5:1 ratio and 60% yield.30
As was observed in the sulfur series, these conditions favored
the formation of the product having the unnatural configuration
of the spirooxindole; these results stand in marked contrast to
what was observed in the cyclization of the simpler congener
having hydrogen as the â substituent (see Table 1).76 Despite
considerable effort, we were never able to find Heck cyclization
conditions that provided useful amounts of the natural isomer
70. However, the unnatural isomer 69 could be produced in
high yields by carrying out the Heck cyclization of 68 in the
presence of silver salts. Under optimal conditions, â-methoxy
R,â-unsaturated anilide 68 was converted to Heck product 69
in yields ranging from 61 to 78% using 35 mol % Pd2(dba)3
and 2 equiv of Ag3PO4 in refluxing THF. The high prefer-
ence for forming pentacyclic product 69 having the aryl
fragment of the oxindole cis to the angular vinyl group
presumably arises from coordination of palladium with this latter
group (eq 3).9d
To prepare gelsemine from pentacyclic Heck product 69, we
needed to invert the configuration of the oxindole as well as
fashion the tetrahydropyran ring. Although prospects for the
former would have appeared poor at the outset of our investiga-
tions, our experience in inverting the spirocyclic oxindole of
pentacyclic ketone 39 (Scheme 6) and the significant develop-
ment of an epimerization strategy by Hart and co-workers19b
suggested that epimerization of the spirooxindole should be
possible if the enol ether functionality of pentacyclic Heck
product 69 could be converted to a C3 ketone or alcohol.
Another example of the unique reactivity engendered by the
compact cage structure of gelsemine confronted us when we
attempted to hydrolyze pentacyclic enol ether 69 with oxalic
acid in MeOH-H2O at 50 °C or 1 M HCl in acetone at 55 °C.
These conditions provided the structurally remarkable hexacyclic
cyclopropyl ketone 71 in high yield (eq 7). The structural
Influenced by Hart’s demonstration that the spirooxindole of
a pentacyclic gelsemine precursor having a hydroxyl group at
C3 could be inverted by a retro-aldol/aldol process,19b we
examined the reduction of pentacyclic ketone 72 (Scheme 12).
Reaction of this ketone with sodium borohydride and cerium(III)
chloride heptahydrate (Luche conditions)80 in methanol at -10
°C gave a 3:1 ratio of readily separable alcohol epimers 73 and
74 in 85% combined yield. The selectivity of this transformation
could be reversed using triisobutylaluminum as the reductant
(-50 °C f room temperature in toluene).81 These latter
conditions provided a 6.5:1 mixture of alcohol epimers with
(77) Silverstein, R. M.; Bassler, G. C.; Morrill, T. C. Spectrometric Identification
of Organic Compounds, 5th ed.; John Wiley & Sons: New York, 1991;
Chapter 5.
(78) Evans, D. A.; Carroll, G. L.; Truesdale, L. K. J. Org. Chem. 1974, 39,
914-917.
(79) Nagata, W.; Yoshioka, M.; Murakami, M. Org. Synth. 1972, 52, 96-99.
(80) Luche, J.-L.; Gemal, A. L. J. Am. Chem. Soc. 1979, 101, 5848-5849.
(81) (a) Katzenellenbogen, J. A.; Bowlus, S. B. J. Org. Chem. 1973, 38, 627-
632. (b) Heinshon, G. E.; Ashby, E. C. J. Org. Chem. 1973, 38, 4232-
4236. (c) Overman, L. E.; McCready, R. J. Tetrahedron Lett. 1982, 23,
2355-2358.
(76) At this point, it is unclear why stereoselection in the Heck cyclization varies
with the nature of the â-substituent.
9
18062 J. AM. CHEM. SOC. VOL. 127, NO. 51, 2005

Intramolecular Heck Reaction Forming Quaternary C Centers
A R T I C L E S
Scheme 11
Scheme 13
Scheme 12
between the aromatic ring of the oxindole and the neighboring
ethenyl group.82 If so, the C3 hydroxyl group would no longer
be positioned appropriately to displace the bromide.
The results summarized in Scheme 12 demonstrated that the
nitrile needed to be introduced at C16 prior to attempting
epimerization of the spirooxindole. We initially explored this
possibility with intermediate 73 after first masking the axial
alcohol as an ethoxyethyl ether (Scheme 13). Reaction of this
product, 76, with NaCN in dimethyl sulfoxide at 150 °C gave
rise to hexacyclic aziridine 77 in 80% yield. Conversion of this
product to the corresponding methyl aziridinium ion, and
reaction of this salt with NaCN at 90 °C produced hexacyclic
furan 78 in high yield. Evidently the methyl aziridinium ion
reacts more rapidly with the proximal oxygen of the axial acetal
substituent than with external cyanide. This result directed us
to the ultimately successful solution: employ equatorial alcohol
epimer 74 to introduce C19.
This final sequence began with protection of the equatorial
hydroxyl group of pentacyclic intermediate 74 by reaction with
pyridinium p-toluenesulfonate and ethyl vinyl ether in CH2Cl2
to give 79 (Scheme 14). Reaction of this product with NaCN
in the usual way provided aziridine 80 in nearly quantitative
yield. Activation of 80 by reaction with methyl trifluoro-
methanesulfonate in the presence of 2,6-di-tert-butyl-4-methyl-
pyridine (DBMP) and opening the resulting methyl aziridinium
ion with NaCN delivered nitrile 81 in 85% overall yield from
hexacyclic aziridine 80. When the aziridine was activated in
the absence of DBMP, the 1-ethoxyethyl group was cleaved
prematurely giving hexacyclic tetrahydrofuran 78 as the major
product. After removing the 1-ethoxyethyl group of acetal 81
under standard acidic conditions, pentacyclic nitrile 82 was
isolated in 71% overall yield from equatorial alcohol 74.
Three critical conversions remained in order to elaborate
hydroxy nitrile 82 to gelsemine: epimerization of the spiro-
oxindole stereocenter C7, epimerization of the C3 alcohol, and
intramolecular condensation of the C3 axial alcohol with the
endo cyanide. To our delight, we quickly found that all three
conversions could be realized by simply heating hydroxy nitrile
82 with DBU in toluene at 110 °C (Scheme 14). After silica
gel chromatography, hexacyclic lactone 83 was isolated in 80%
yield. A likely mechanism for this multifaceted conversion is
equatorial epimer 74 predominating; after chromatography on
silica gel, pentacyclic equatorial alcohol 74 was isolated in 71%
yield.
We were once again at a stage to examine the critical
epimerization of the spirooxindole fragment. When pentacyclic
alcohol 73 or 74 was exposed to 1,8-diazabicyclo[5.4.0]undec-
7-ene (DBU) in CH2Cl2 at room temperature,19b epimerization
of the oxindole was not realized, but instead the two alcohol
epimers were equilibrated (Scheme 12). However, when this
mixture of alcohol epimers was heated with DBU at 110 °C in
toluene for 4 h, hexacyclic furan 75 was isolated in 57% yield.
No trace of the corresponding product having an epimeric
oxindole fragment was seen. The stereostructure of 75 followed
unambiguously from diagnostic NOEs between the methine
hydrogens at C5 and C9. Epimerization of the oxindole,
presumably by a retro-aldol fragmentation/aldol condensation
sequence, obviously had taken place prior to cyclization of the
axial alcohol epimer to form 75. Why 73 does not undergo
similar intramolecular etherification to form the spirooxindole
epimer of 75 is less clear. Perhaps the cyclohexyl ring of 73
adopts a boat conformation to minimize steric interaction
(82) In product 73, an NOE is observed between the cyclohexyl proton at C3,
the vinyl proton at C19, and the aromatic proton at C9. These NOEs are
consistent with the cyclohexyl ring of 73 existing in a twist boat
conformation.
9
J. AM. CHEM. SOC. VOL. 127, NO. 51, 2005 18063

A R T I C L E S
Scheme 14
Madin et al.
Scheme 16
The conversion of hexacyclic lactone 83 to gelsemine (1)
was straightforward (Scheme 16). The methoxymethyl group
of 83 was first discharged by reaction with concentrated HCl
in ethylene glycol dimethyl ether (DME) at 55 °C to give the
corresponding N-hydroxymethyl oxindole; exposure of this
intermediate to N,N-diisopropylethylamine in methanol at 55
°C furnished oxindole lactone 88 in 90% yield. Reduction of
this intermediate with excess diisobutylaluminum hydride
generated a mixture of lactols 89, which upon reaction with
triethylsilane in trifluoroacetic acid83 produced (()-gelsemine
Scheme 15
1
(1) in 65% yield from hexacyclic lactone 88. The H and 13C
NMR and IR spectra and mass spectral fragmentation patterns
of 1 were identical to those of an authentic sample of natural
gelsemine.
Conclusion
Structurally unique natural products have long served to
benchmark the existing tools of organic synthesis and stimulate
the development of new reagents, reactions, and synthetic
strategies. As gelsemine has no commercial or medicinal value,
it was the opportunity to discover and develop new synthetic
chemistry that led us to tackle its total synthesis. Discoveries
of new chemistry and new modes of reactivity made during
our studies, and those of other laboratories working in the area,
have broad implications for organic chemistry that transcend
the accomplishment of a gelsemine total synthesis.3f
The extraordinary ability of intramolecular Heck reactions
to construct highly congested quaternary carbon centers is a
discovery of fundamental importance made during our efforts
in the gelsemine area. At the time we first reported in 1988
that palladium-catalyzed insertions could succeed even in the
face of numerous developing syn-pentane (1,3-diaxial) inter-
actions,9c Heck reactions were rarely used for ring formation
and never as the key strategic step in the synthesis of complex
organic molecules. Today, intramolecular Heck reactions are
established as one of the synthetic chemist’s most powerful tools
for forming C-C bonds in complex, polyfunctional molecules.25
This research program also uncovered unique, unexpected
reactivity associated with the enforced functional-group proxim-
ity engendered by the cage-like gelsemine ring system. Notable
examples are: (1) the formation of aziridines from the reaction
of N-methoxycarbonyl-â-bromoethylamines with cyanide by
formal front-side displacement, (2) the unprecedented re-
suggested in Scheme 15. Retro-aldol cleavage of 82 would
generate oxindole enolate 84. Rotation about the C6-C7 σ bond
and re-aldolization would form spirooxindole epimer 85,
whereas alcohol epimer 86 would be produced by addition of
the oxindole enolate to the other prochiral face of the aldehyde.
Finally, addition of the axial hydroxyl group of intermediate
86 to the proximal nitrile would generate cyclic imidate 87.
Hydrolysis of this product upon purification on silica gel would
give the observed product 83. Although intermediate 86 was
not detected, its equatorial epimer 85 could be isolated when
the reaction was stopped prior to completion.
(83) Kursanov, D. N.; Parnes, Z. N.; Loim, N. M. Synthesis 1974, 633-651.
9
18064 J. AM. CHEM. SOC. VOL. 127, NO. 51, 2005

Intramolecular Heck Reaction Forming Quaternary C Centers
A R T I C L E S
arrangement of an alkyl iron intermediate,9e (3) the acid-
promoted cyclizations of γ-bromoketones to cyclopropyl ke-
tones, and (4) the base-promoted reaction cascade summarized
in Scheme 15. Our studies in the gelsemine area revealed also
the thermal instability of â-alkoxy enol triflates, providing a
caveat to their use in transition metal catalyzed chemistry and
foreshadowing our development of a new synthesis of R-thio-
alkoxy enones.75 As summarized in some detail in the ac-
companying paper,24 our gelsemine studies also provided a
greater understanding of the scope and limitations of aza-Cope-
Mannich transformations.24
Despite extensive synthetic work in the gelsemine area over
nearly two decades, a short synthesis of gelsemine has yet to
emerge.
Acknowledgment. This work was supported by the National
Institute of Health (HL-25854 and GM-30859). C.J.O. gratefully
acknowledges the American Cancer Society for postdoctoral
fellowship support (PF-98-002-01). We also thank Dr. John
Greaves and Mr. John Mudd of the UCI mass spectroscopy
facility, Dr. Joeseph Ziller of the UCI X-ray crystallography
laboratory, and Dr. Jiejun Wu of the UCI NMR spectroscopy
facility for their technical support.
As for our total synthesis of (()-gelsemine (1), it was
accomplished in 1.1% overall yield from 3-methylanisole by
the way of 26 isolated intermediates.84 It is among the most
efficient total syntheses of racemic gelsemine reported to
date,85,86 although it is longer than the Speckamp-Hiemstra
synthesis, which proceeded in 23 steps from sorbic acid.18
Supporting Information Available: Experimental procedures,
spectroscopic and analytical data for new compounds, 1H NMR
1
spectra of selected compounds, and H and 13C NMR spectra
of synthetic (()-gelsemine and natural gelsemine (79 pages,
PDF). This material is available free of charge via the Internet
at http://pubs.acs.org.
(84) The yield of the largest scale reaction we conducted of each step was used
in this calculation. For the intramolecular Heck reaction of 68 an average
yield of 69% was used as this reaction was only carried out on one scale.
(85) The first step of our gelsemine synthesis is a Diels-Alder reaction of a
1,3-cyclohexadiene with methyl acrylate.24 As chiral auxiliaries for acrylate
dienophiles have been developed and much progress in asymmetric catalysis
of Diels-Alder reactions has been recorded, it would appear likely that an
enantioselective version of the opening moves of our gelsemine synthesis
could be developed.85
JA055711H
(86) For selected reviews of asymmetric Diels-Alder reactions, see: (a) Oh,
T.; Reilly, M. Org. Prep. Proced. Int. 1994, 26, 129-158. (b) Kagan, H.
B.; Riant, O. Chem. ReV. 1992, 92, 1007-1019. (c) Corey, E. J. Angew.
Chem., Int. Ed. 2002, 41, 1650-1667. For a review of the use of the Diels-
Alder reaction in natural products total synthesis, see: (d) Nicolaou, K.
C.; Snyder, S. A.; Montagnon, T.; Vassilikogiannakis, G. Angew. Chem.,
Int. Ed. 2002, 41, 1668-1698.
9
J. AM. CHEM. SOC. VOL. 127, NO. 51, 2005 18065