Free radical cyclizations in alkaloid total synthesis: (±)-21- oxogelsemine and (±)-gelsemine

Article abstract of DOI:10.1021/ja970089h

Total syntheses of (±)-21-oxogelsemine (3) and (±)-gelsemine (1) are described. Free radical cyclizations play important roles in constructing the tricyclic core (40 → 41) and oxindole (93 → 96) substructures of the target alkaloids, and an isomerization-cyclization strategy (101 → 102) was used to complete construction of the gelsemine cage. Observations made along the way include intramolecular 1,4- and 1,6-hydrogen atom transfers (64 → 65 and 40 → 47) and a free radical cyclization-fragmentation sequence (85 → 86 + 87). A retroaldol-aldol strategy for adjusting oxindole stereochemistry in gelsemine-like structures is also described (72 → 74 and 86 → 87).

Full text of DOI:10.1021/ja970089h

6226
J. Am. Chem. Soc. 1997, 119, 6226-6241
Free Radical Cyclizations in Alkaloid Total Synthesis:
(()-21-Oxogelsemine and (()-Gelsemine
Shogo Atarashi, Joong-Kwon Choi, Deok-Chan Ha, David J. Hart,*
Daniel Kuzmich, Chih-Shone Lee, Subban Ramesh, and Shung C. Wu
Contribution from the Department of Chemistry, The Ohio State UniVersity, 100 West 18th
AVenue, Columbus, Ohio 43210
ReceiVed January 10, 1997X
Abstract: Total syntheses of (()-21-oxogelsemine (3) and (()-gelsemine (1) are described. Free radical cyclizations
play important roles in constructing the tricyclic core (40 f 41) and oxindole (93 f 96) substructures of the target
alkaloids, and an isomerization-cyclization strategy (101 f 102) was used to complete construction of the gelsemine
cage. Observations made along the way include intramolecular 1,4- and 1,6-hydrogen atom transfers (64 f 65 and
40 f 47) and a free radical cyclization-fragmentation sequence (85 f 86 + 87). A retroaldol-aldol strategy for
adjusting oxindole stereochemistry in gelsemine-like structures is also described (72 f 74 and 86 f 87).
Introduction
Gelsemine (1) was first isolated as an amorphous base from
the roots and rhizomes of Gelsemium semperVirens (Carolina
jasmine) in 1870 by Wormley.1,2 It was later obtained in
crystalline form by Gerrard in 1883,3 and its correct molecular
formula was reported by Moore in 1910.4 The exact nature of
gelsemine, however, remained elusive until 1959 when inde-
pendent reports of its structure appeared from the laboratories
of Conroy and Wilson.5,6 Structurally related alkaloids have
been subsequently isolated from related plant sources (Figure
1). These include gelsevirene (2) and 21-oxogelsemine (3) from
Gelsemium semperVirens,7 gelsemine N-oxide, 19-(S)-hydroxy-
dihydrogelsevirene, and 19-(R)-hydroxydihydrogelsevirine (4)
from Gelsemium elegans,8 and 21-oxogelsevirine, 19-(R)-
acetoxydihydrogelsevirine (5), and 19-(R)-hydroxydihydro-
gelsemine (6) from Gelsemium rankinii.7,9
Soon after its structure was secured, reports describing
synthetic approaches to gelsemine began to appear in the
literature.10-14 These studies resulted in the development of
X Abstract published in AdVance ACS Abstracts, June 15, 1997.
(1) For reviews of literature related to Gelsemium alkaloids, see: Liu,
Z.-J.; Lu, R.-R. In The Alkaloids; Manske, R. H. F., Ed.; Academic Press:
New York, 1988; Vol. 33, pp 83-140. Saxon, J. E. Nat. Prod. Rep. 1992,
393.
Figure 1.
(2) See: Saxton, J. E. In The Alkaloids; Manske, R. H. F., Ed.; Academic
Press: New York, 1965; Vol. 8, pp 93-117.
much interesting chemistry and synthetic methodology, and in
1994 two successful syntheses were finally described.15,16 These
successes were followed shortly by two other syntheses of
(3) Gerard, A. Pharm. J. 1883, 13, 641.
(4) Moore, C. W. J. Chem. Soc. 1911, 99, 1231.
(5) Conroy, H.; Chakrabarti, J. K. Tetrahedron Lett. 1959, 6. For
revisions of the NMR spectroscopic properties of gelsemine, see: Schun,
Y.; Cordell, G. A. J. Nat. Prod. 1985, 48, 969.
(6) Lovell, F. M.; Pepinsky, R.; Wilson, A. J. C. Tetrahedron Lett. 1959,
1.
(7) For isolation and structure of 21-oxogelsemine, see: Nikiforov, A.
J.; Latzel, J.; Varmuza, K.; Wichtl, M. Monatsh. Chem. 1974, 105, 1292.
For spectroscopic characterization of 21-oxogelsemine, see: Schun, Y.;
Cordell, G. A.; Garland, M. J. Nat. Prod. 1986, 49, 483.
(8) Ponglux, D.; Wongseripipatana, S.; Subhadhirasakul, S.; Takayama,
H.; Yokota, M.; Ogata, K.; Phisalaphong, C.; Aimi, N.; Sakai, S.-I.
Tetrahedron 1988, 44, 5075.
(13) Fleming, I.; Loreto, M. A.; Michael, J. P.; Wallace, I. H. M.
Tetrahedron Lett. 1982, 23, 2053. Fleming, I.; Loreto, M. A.; Wallace, I.
H. M.; Michael, J. P. J. Chem. Soc., Perkin Trans. 1 1986, 349. Clarke, C.;
Fleming, I.; Fortunak, M. D.; Gallagher, P. T.; Honan, M. C.; Mann, A.;
Nubling, C. O.; Raithby, P. R.; Wolff, J. J. Tetrahedron 1988, 44, 3931.
Fleming, I.; Moses, R. C.; Tercel, M.; Ziv, J. J. Chem. Soc., Perkin Trans.
1 1991, 617.
(14) Abelman, M. M.; Oh, T.; Overman, L. E. J. Org. Chem. 1987, 52,
4133. Madin, A.; Overman, L. E. Tetrahedron Lett. 1992, 33, 4859. Early,
W. G.; Jacobsen, E. J.; Meier, G. P.; Oh, T.; Overman, L. E. Tetrahedron
Lett. 1988, 29, 3781. Earley, W. G.; Oh, T.; Overman, L. E. Tetrahedron
Lett. 1988, 29, 3785. Overman, L. E.; Sharp, M. J. J. Org. Chem. 1992,
57, 1035.
(9) Lin, L.-Z.; Yeh, S.; Cordell, G. A.; Ni, C.-H.; Clardy, J. Phytochem-
istry 1991, 30, 679.
(10) Autrey, R. L., Tahk, F. C. Tetrahedron 1967, 23, 901. Autrey, R.
L.; Tahk, F. C. Tetrahedron 1968, 24, 3337.
(11) Johnson, R. S.; Lovett, T. O.; Stevens, T. S. J. Chem. Soc. (C) 1970,
796.
(12) Stork, G.; Krafft, M. E.; Biller, S. A. Tetrahedron Lett. 1987, 28,
1035.
(15) Sheikh, Z.; Steel, R.; Tasker, A. S.; Johnson, A. P. J. Chem. Soc.,
Chem. Commun. 1994, 763. Dutton, J. K.; Steel, R. W.; Tasker, A. S.;
Popsavin, V.; Johnson, A. P. J. Chem. Soc., Chem. Commun. 1994, 765.
(16) Hiemstra, H.; Vijn, R. J.; Speckamp, W. N. J. Org. Chem. 1988,
53, 3884. Newcombe, N. J.; Ya, F.; Vijn, R. J.; Hiemstra, H.; Speckamp,
W. N. J. Chem. Soc., Chem. Commun. 1994, 767.
S0002-7863(97)00089-9 CCC: $14.00 © 1997 American Chemical Society

Free Radical Cyclizations in Alkaloid Total Synthesis
J. Am. Chem. Soc., Vol. 119, No. 27, 1997 6227
to allylic strain present in conformation 8 but absent in the
conformation derived from a 180° rotation around the C(15)sC-
(16) bond.21 This stereochemical mistake required development
of a correction plan, which will be described in due course.
We next examined the radical derived from sulfide 16.22 This
substrate contained a full complement of substituents needed
to attempt a synthesis of gelsemine. Although we worried that
the ketal might discourage cyclization for steric reasons, this
was fortunately not the case as tri-n-butyltin hydride mediated
cyclization of 16 provided 17 in 87% yield. Subsequent ketal
hydrolysis afforded ketone 18 (85%).
As a final cyclization substrate, we examined sulfide 19. We
hoped that the o-fluorophenyl group could be used to install
the oxindole portion of gelsemine using methodology developed
by Fleming.13 The stereochemistry at C(7) in 19 required that
the aryl group occupy an axial site if the free radical cyclization
occurred with the six-membered ring in a chair conformation.
In spite of this demand, the radical derived from 19 cyclized to
provide 20 in 40% yield along with 45% of the lactam derived
from simple reduction of 19. The structure of 20 was proven
by X-ray crystallography.23 The crystal structure interestingly
revealed that the cyclohexane substructure of 20 adopts a
boatlike conformation rather than the chairlike structure shown
in Figure 2.24 This boatlike conformation relieves steric
interactions between the aryl group and the C(20)-methoxy-
methyl group and also the endo-oxygen of the ketal and the
C(5) and C(16) hydrogens. Based on this result, we imagine
that the radical derived from 19 cyclizes with the cyclohexane
in a boatlike conformation. Treatment of 20 with excess
phenyllithium followed by p-toluenesulfonic acid gave ketone
21 (74%) in which ketone hydrolysis was accompanied by
epimerization at C(7). The preparation of 20 and 21 provided
some insight into stereochemical aspects of the aforementioned
cyclizations but did not ultimately lead to gelsemine, as several
yields were too low and an inadequate supply of material
prevented thorough investigation of its chemistry. Based on
these problems and the serendipitous observation ketones of type
18 tend to enolize toward C(7) rather than C(14),20 we focused
on installation of the oxindole after construction of the tricyclic
core of gelsemine. Preparation of a tricyclic core that was
eventually used to prepare gelsemine is described below.
Figure 2.
gelsemine,17,18 and it is likely that other quite different ap-
proaches will be completed before the close of this century.
Our own efforts toward gelsemine were initiated in 1984 to
see if R-acylamino radical chemistry developed in our group
could be used to solve what at the time was a classical unsolved
problem in alkaloid synthesis.19 Our synthetic plan was based
on the belief that a tricyclic structure of type 7 could be
converted into gelsemine if X, Y, and Z were designed to allow
incorporation of the oxindole, the vinyl group, and the tetrahy-
dropyran substructures of gelsemine, respectively (Figure 1). It
was hoped that 7 might be prepared by 5-hexenyl radical
cyclization of an R-acylamino radical of type 8. One question
associated with this plan was would the rate of cyclization of
radical 8 be fast enough to compete with unproductive bimo-
lecular processes that could occur from 9, the expected lowest
energy conformation of this free radical. Another question was
would the cyclization provide the proper stereochemistry at what
would become C(16) of gelsemine.
Initial Studies. A series of model studies revealed the path
that was eventually followed to achieve a synthesis of gelsemine.
The details of many of these studies have been described in an
earlier full paper.20 To provide continuity, however, a summary
of these studies will be presented here (Figure 2). Initial work
focused on the generation of 8 where X, Y, and Z were hydrogen
atoms. This was accomplished by preparing sulfide 10 and
treating it with tri-n-butyltin hydride and AIBN. Not surpris-
ingly, this resulted in the formation of only reduction product
11 (90%), even under high dilution reaction conditions. It was
hoped that rates of cyclization of more electron deficient olefins
would be fast enough to compete with reduction. Thus, radical
precursors 12 and 13 were prepared and subjected to typical
free radical cyclization conditions. This plan was successful
as cyclization products 14 and 15 were obtained in 68% and
92% yields, respectively. It is notable that these cyclizations
provided the wrong stereochemistry at C(16), most likely due
Preparation of Tricyclic Core Structures. Although neither
were destined to be intermediates in the synthesis of gelsemine,
tricyclic ketones 42 and 43 played important roles in ac-
complishing the synthesis. The preparation of these compounds
is described in Schemes 1 and 2. The first key step in the
synthesis of 42 and 43 was a Diels-Alder reaction between
N-methylmaleimide and diene 26. The diene was prepared on
a 100-g scale from commercially available alcohol 22 in four
steps (54% overall) as shown in Scheme 1 without comment.25
Diene 26 was warmed with N-methymaleimide in toluene to
provide cycloadduct 27.26 In early studies, the silyl enol ether
and tetrahydropyranyl ether groups in cycloadduct 27 were
hydrolyzed to give the corresponding keto alcohol, which was
(21) RajanBabu, T. V. Acc. Chem. Res. 1991, 24, 139.
(22) Experimental details for the preparation and cyclization of sulfide
19 are provided in the Supporting Information.
(23) The structures of 20, 46, 65a, 70, 74, 90, and 114 were determined
by Dr. Judith C. Gallucci at the OSU Deparment of Chemistry Crystal-
lography Facility. Crystallographic details for structures 46, 114, and two
structures prepared en route to 20 are provided in the Supporting
Information. Details for the other compounds will be reported elsewhere.24
(24) A comparison of solid state and solution conformational preferences
of several compounds reported herein will be reported elsewhere.
(25) Although 3-buten-1-ol is commercially available, we found it
convenient to prepare it using a variation of the following procedure: House,
H. O.; Liang, W. C.; Weeks, P. D. J. Org. Chem. 1974, 39, 3104.
(17) 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.
(18) Fukuyama, T.; Liu, G. J. Am. Chem. Soc. 1996, 118, 7426.
(19) Hart, D. J.; Tsai, Y.-M. J. Am. Chem. Soc. 1982, 104, 1430.
(20) Choi, J.-K.; Ha, D.-C.; Hart, D. J.; Lee, C.-S.; Ramesh, S.; Wu, S.
J. Org. Chem. 1989, 54, 279.

6228 J. Am. Chem. Soc., Vol. 119, No. 27, 1997
Atarashi et al.
Scheme 1
Figure 3.
manner, ozonolysis of 35 provided 37 (76%), which was
subjected to Wittig olefination to give ester 39 in 82% yield.
Ethoxy-thiophenoxy exchange using thiophenol and catalytic
p-toluenesulfonic acid in dichloromethane converted 38 and 39
into free radical precursors 16 (89%) and 40 (76%), respectively.
The free radical cyclizations of 16 and 40 proceeded smoothly
under standard tri-n-butyltin hydride-AIBN conditions to afford
17 and 41 in 87% and 64% yields, respectively. Finally,
treatment of 17 and 41 with excess phenylmagnesium bromide
gave tertiary alcohols that dehydrated upon treatment with
p-toluenesulfonic acid in acetone, accompanied by ketal hy-
drolysis, to give tricyclic ketones 42 (54%) and 43 (87%).
At this point several tasks had to be accomplished to complete
a synthesis of gelsemine. These included correction of the
stereochemical mistake at C(16) and construction of the
tetrahydropyran substructure, incorporation of the oxindole, and
conversion of the C(20) substituent to a vinyl group. Model
studies for accomplishing the first two tasks, using ketone 42
as a point of departure, have already been described (42 f 44).20
Thus, reduction of the ketone using sodium borohydride
provided the expected endo alcohol. Ozonolysis of the olefin
provided a hydroxy aldehyde which epimerized to a tetracyclic
hemiacetal upon treatment with DBU in dichloromethane.
Reduction of the hemiacetal with triethylsilane-trifluoroacetic
acid gave tetrahydropyran 44. This epimerization-trapping
strategy was developed fairly early in our studies, but it turned
out that installation of the oxindole and applying the stereo-
chemical correction protocol in the presence of the oxindole
was quite difficult. Before continuing with the synthesis,
however, several observations made during the course of
preparing 42 and 43 will be described.
Some Observations. One notable step in Scheme 2 is the
regioselective reduction of imide 30. The reason for regiose-
lectivity in this reduction is most likely due to stereoelectronic
effects (Felkin-Ahn arguements) as nicely suggested by
Speckamp and Hiemstra,33 whose synthesis of gelsemine also
used such a regioselective reduction. As part of our own studies
designed to uncover the reasons for this selectivity, we prepared
imide 45 and examined its reduction using sodium borohydride
in methanol. This reduction was also highly regioselective,
providing carbinol lactam 46 in 85% yield.23,24 This result is
consistent with the explanation advanced by Speckamp and
Hiemstra and provides another example of stereoelectronic
effects explaining regioselectivity in what to us was initially a
non-obvious situation.
then converted to ketal 28. On a large scale, however, the
hydrolysis step was accompanied by formation of significant
amounts of ketal 29. Thus, for operational reasons, it was easier
to directly treat the crude cycloadduct 27 with 2 equiv of 2,2-
dimethylpropan-1,3-diol and a catalytic amount of p-toluene-
sulfonic acid to provide 28 directly in 39% yield.
Conversion of 28 to 42 and 43 is described in Scheme 2.
Formal dehydration of 28 following the Grieco protocol gave
olefin 30 in 79% yield.27 Reduction of the imide with sodium
borohydride in methanol gave carbinol lactam 31 in 80% yield.28
Initially, a hydroxy-ethoxy exchange in acidic ethanol was used
to convert crude 31 to 32. Ketal hydrolysis, however, caused
problems when this procedure was conducted on a large scale.
This problem was avoided by treating 31 with sodium hydride
and iodoethane in tetrahydrofuran to afford a mixture of isomeric
ethoxy lactams 32 (71%) and 33 (11%). Alkylation of the
lithium enolate of 32 with either chloromethyl methyl ether or
benzyl chloromethyl ether gave 34 (88%) and 35 (96%),
respectively.29 As previously reported, Johnson-Lemieux
oxidation conditions were initially used to convert olefin 34 to
aldehyde 36.30 This reaction proved to be capricious on a large
scale, however, as epimerization of the aldehyde was often a
problem. A workable alternative to the Johnson-Lemieux
proved to be ozonolysis, followed by a reductive workup with
dimethyl sulfide.31 This provided crude 36 which was treated
with ethyl (triphenylphosphoranylidene)acetate32 to give crystal-
line unsaturated ester 38 in 96% overall yield. In a similar
(26) Although N-methylmaleimide is commercially available, we found
it convenient to prepare for large scale work: Piutti, A.; Giustiniani, E.
Gazz. Chim. Ital. 1896, 26, 431. Cava, M. P.; Deana, A. A.; Muth, K.;
Mitchell, M. J. Org. Synth. 1961, 41, 93.
(27) Grieco, P. A.; Gilman, S.; Nishizawa, M. J. Org. Chem. 1975, 30,
3760. Sharpless, B. K.; Young, M. W. J. Org. Chem. 1975, 40, 947.
(28) Hubert, J. C.; Wijnberg, J. B. P. A.; Speckamp, W. N. Tetrahedron
1975, 31, 1437. Chamberlin, A. R.; Chung, J. Y. L. Tetrahedron Lett. 1982,
23, 2619. Hart, D. J.; Yang, T.-K. J. Chem. Soc., Chem. Commun. 1983,
135.
Another observation recorded while executing Scheme 2 was
that the free radical cyclization of 41 gave compound 47 in
4-8% yields. The formation of this product can be explained
by a sequence of events involving initial generation of an
R-acylamino radical, 1,6-hydrogen atom transfer to provide a
(29) For a preparation of benzyl chloromethyl ether, see: Connor, D.
S.; Klein, G. W.; Taylor, G. N. Org. Synth. 1972, 52, 16.
(30) Pappo, R.; Allen, D. S.; Lemieux, R. U.; Johnson, W. S. J. Org.
Chem. 1956, 21, 478.
(31) For a convenient method for controlling selective ozonolysis of
olefins see Veysoglu, T.; Mitscher, L. A.; Swayze, J. K. Synthesis 1980,
807.
(32) For the preparation of ethyl (triphenylphosphoranylidene)acetate,
see: Denney, D. B.; Ross, S. T. J. Org. Chem. 1962, 27, 998.
(33) Vijn, R. J.; Hiemstra, H.; Kok, J. J.; Knotter, M.; Speckamp, W. N.
Tetrahedron 1987, 43, 5019.

Free Radical Cyclizations in Alkaloid Total Synthesis
J. Am. Chem. Soc., Vol. 119, No. 27, 1997 6229
Scheme 2a
a (a) o-NO2C6H4SeCN, n-Bu3P, THF; H2O2. (b) NaBH4, MeOH, -23 °C. (c) NaH, THF, Etl. (d) LDA, THF; CH3OCH2Cl. (e) LDA, THF;
PhCH2OCH2Cl. (f) O3, CH2Cl2, MeOH; Me2S. (g) Ph3PdCHCO2Et. (h) PhSH, p-TsOH (cat), CH2Cl2. (i) n-Bu3SnH, AIBN (cat), PhH. (j) PhMgBr.
(k) p-TsOH (cat), acetone, ∆.
Scheme 3
Scheme 4
benzylic radical (via a conformation of type 9), and intramo-
lecular addition of the benzylic radical to the unsaturated ester.
It is interesting that a similar product was not observed in the
cyclization of 16, in which case a 1,6-hydrogen atom transfer
would not have provided as stable a free radical.
Introduction of the Oxindole: Initial Studies. As men-
tioned above, attempts to carry a portion of the oxindole through
from the beginning of the synthesis were abandoned due to yield
problems. A break came when it was observed that ketone 18
gave a good yield of the ∆3,7-O-methyl enol ether upon treatment
with methanol, trimethylorthoformate, and catalytic p-toluene-
sulfonic acid. This result suggested that ketones 42 and 43
might also have a propensity for enolization toward C(7). Thus,
strategies for introducing the oxindole that relied on such an
enolization preference were investigated. Several unsuccessful
routes were examined before achieving success. One of the
more interesting dead-end routes is described in Scheme 4. We
imagined that a Fischer indole synthesis would introduce a
masked o-aminophenyl group at C(7).35 If this could be
followed by introduction of an appropriate one-carbon unit using
a directed electrophilic aromatic substitution reaction, we
thought the resulting indoline might be rearranged to a useful
oxindole. This process was realized, in part, in a model system.
For example, Fischer indole conditions converted ketone 5236
to indole 53 (80%). Reduction of the ester with lithium
aluminum hydride gave 54 (91%) and treatment of 54 with
dimethoxymethane and ethylaluminum dichloride produced
A final interesting observation was recorded while trying to
extend the epimerization-trapping procedure (42 f 44) to keto
aldehyde 48. For reasons that will become apparent, we hoped
that treatment of 48 (prepared in 65% yield by ozonolysis of
42) with methanol under acidic or basic conditions might lead
to epimerization of the aldehyde followed by hemiacetal
formation and attack at the ketone carbonyl group to form a
cyclic hemiacetal at C(3). Unfortunately, treatment of 48 with
acidic methanol gave only the dimethyl acetal of the aldehyde.
Treatment of 48 with DBU in the presence of methanol,
however, resulted in formation of a hydroxy ester in 58% yield.
We initially thought this compound might have structure 50,
resulting from formation of 49 followed by an intramolecular
hydride transfer (Scheme 3). This would have been fortunate,
as reduction of the carbomethoxy group would have then
provided a compound with a C(16)-hydroxymethyl group
disposed in the manner needed to complete tetrahydropyran
construction. However, NOE experiments revealed that the
C(16) methyl ester had undergone a second isomerization and
that the product was 51. For example, irradiation of H(3) (δ
4.10) gave a 21% enhancement of the signal due to H(16) (δ
3.04). Obviously, adjustment of C(16) stereochemistry without
a trap is not a favorable process.
(34) We thank Dr. Jeffrey Dener and Mr. Jonathan Dawson for
performing these experiments.
(35) Fischer, E.; Jourdan, F. Chem. Ber. 1883, 16, 2241. Sindwell, H.
G.; van Orden, R. B. Chem. ReV. 1942, 30, 78.
(36) Dreiding, A.; Tomasewske, A. J. J. Am. Chem. Soc. 1955, 77, 411.

6230 J. Am. Chem. Soc., Vol. 119, No. 27, 1997
Atarashi et al.
Scheme 5
Scheme 6
indoline 55 (59%). The first two steps of this process also
worked well in a system relevant to gelsemine.37 Thus,
treatment of 42 with phenylhydrazine and boron trifluoride
etherate in acetic acid gave the desired indole 56 (72%) and
demethylation using boron tribromide gave 57 (88%). However,
attempts to use the hydroxymethyl group to direct electrophilic
aromatic substitutions with 56 met with failure, and this route
was eventually abandoned.22
The plan that ultimately led to completion of the synthesis is
shown in Scheme 5. It was imagined that free radical
cyclization substrates of type 58 would be accessible from
ketones 42 or 43. We anticipated that application of a variant
of an oxindole synthesis developed by Jones might then provide
oxindole 60 via intermediate radical 59.39 It was originally
thought that the success or failure of this plan would hinge upon
obtaining the correct stereochemistry at C(7) in the cyclization
of an aryl radical derived from 58 and at C(3) in the reduction
of free radical 59. We imagined that once a compound of type
60 was in hand, the chemistry developed to convert 42 to 44
would be used to complete the synthesis.
Aryl bromide 64 (58 where R1 ) MOM, R2 ) R3 ) Me),
the initial substrate selected for study, was prepared as outlined
in Scheme 6. Treatment of 42 with potassium hydride and
dimethyl carbonate gave 61 in 78% yield.40 Conversion of 61
to vinylogous carbonate 62 was accomplished with use of
potassium hydride and iodomethane or more conveniently by
addition of hexamethylphosphorictriamide and dimethyl sulfate
to the enolate formed in the acylation of 42. The latter
procedure gave 62 in 74% overall yield from 42. Treatment of
62 with the amide derived from o-bromoaniline and trimethy-
laluminum,41 followed by N-alkylation of the resulting amide
63 (83%) using lithium diisopropylamide and chloromethyl
methyl ether, gave vinylogous urethane 64 (71%). When 64
was treated with 1.2 equiv of tri-n-butyltin hydride and 0.5 equiv
of AIBN under high dilution conditions, oxindole 65a was
obtained in 46% yield along with two other unidentified products
that contained oxindole substructures by NMR. The structure
of 65a was initially assigned on the basis of spectroscopic data.
For example, the stereochemistry at C(3) was based on the
appearance of H(3) as a doublet of doublets (J ) 12, 6 Hz), an
indication of an axial disposition for this proton. The stereo-
chemistry of the olefin was assigned on the basis of NOE
experiments that established the proximity of H(5) and H(17),
and the stereochemistry at C(7) was based on NOE experiments
that established the proximity of H(9) and the C(20)-meth-
oxymethyl group. This structure was also confirmed by X-ray
crystallography.23 It is probable that 65a was derived from
cyclization of an aryl radical, followed by 1,4-hydrogen transfer
to form a stabilized allylic radical, and subsequent reduction of
that radical. In accord with this proposal, treatment of 64 with
tri-n-butyltin deuteride gave only 65b with no deuterium
incorporation at C(3). Finally, one of the two unidentified
products, which was clearly a 1:1 coupling product of 64 with
an isobutyronitrile radical (C40H43N3O5 by HRMS), was ten-
tatively assigned structure 66 (18%) based on spectroscopic data.
The results shown in Scheme 6 were simultaneously encour-
aging and discouraging. On the up side, an oxindole had been
installed. On the down side, the stereochemistry at C(3) and
C(7) was incorrect, and the handle for correcting the stereo-
chemical mistake at C(16) had been destroyed. Nonetheless,
it was hoped that the new stereochemical mistake at C(7) might
be corrected by generating an alcohol at C(3) followed by a
retroaldol-aldol condensation across the C(3)sC(7) bond. Thus,
aryl bromide 69 (58 where R1 ) R2 ) MOM, R3 ) Bn) was
selected as the next substrate for study. The choice of protecting
group at C(3) was based on the aforementioned need to generate
an alcohol at C(3), and the change in protecting group in the
C(20) substituent was based on the notion that a benzyl group
would be more versatile than a methyl group. A streamlined
approach to 69 is shown in Scheme 7. Acylation of 43 with
o-bromophenylisocyanate42 in the presence of potassium hydride
gave 67 in 85% yield.43 Proton NMR spectroscopy indicated
this material existed completely as the enol tautomer in
deuteriochloroform. Alkylation of 67 with chloromethyl methyl
ether in the presence of Hunig’s base in N,N-dimethylformamide
first gave 68 (89%), which was followed by N-alkylation to
give 69 (85%). The most interesting results with 69 were
obtained when it was treated with an excess of tri-n-butyltin
hydride and AIBN in a minimal amount of benzene. Under
(37) Hart, D. J.; Wu, S. C. Heterocycles 1993, 35, 135.
(38) Cronyn, M. W. J. Org. Chem. 1949, 14, 1013.
(39) Jones, K.; Thompson, M.; Wright, C. J. Chem. Soc., Chem. Commun.
1986, 115. Wright, C.; Shulkind, M.; Jones, K.; Thompson, M. Tetrahedron
Lett. 1987, 28, 6389. Jones, K.; McCarthy, C. Tetrahedron Lett. 1989, 30,
2657.
(42) For a procedure for the preparation of 2-bromophenyl isocyanate
using triphosgene, see: Eckert, H.; Forster, B. Angew. Chem. 1987, 99,
922.
(40) Ruest, L.; Blouin, G.; Delongchamps, P. Synth. Commun. 1976, 6,
169.
(41) Basha, A.; Lipton, M.; Weinreb, S. M. Tetrahedron Lett. 1977, 18,
4171.
(43) Hendi, S. B.; Hendi, M. S.; Wolfe, J. F. Synth. Commun. 1987, 17,
13. Brown, C. A. J. Org. Chem. 1974, 39, 1324.

Free Radical Cyclizations in Alkaloid Total Synthesis
J. Am. Chem. Soc., Vol. 119, No. 27, 1997 6231
Scheme 7
Scheme 8
nolysis of 74 in 59% yield) might isomerize to hemiacetal 76
under similar conditions. Unfortunately, treatment of 75 with
DBU in dichloromethane gave only a complex mixture of
products (Scheme 8). This result suggested that an equatorial
C(3) hydroxyl group was not going to be useful for construction
of the tetrahydropyran.
these conditions, the R-alkoxy radicals derived from cyclization
of an initially formed aryl radical were intercepted by the tri-
n-butyltin hydride prior to 1,4-hydrogen atom transfer. This
resulted in formation of isomeric oxindoles 70 and 71 in 28%
and 9% yields, respectively. The structure of 70 was initially
based on decoupling and NOE experiments and eventually
confirmed by X-ray crystallography.23 It is notable that the
cyclohexane ring in 70 adopts a boatlike conformation both in
the solid state and in solution.24 The latter was evident from
NOE studies that revealed the proximity of H(3) and one of
the diastereotopic C(19) hydrogens. The structure of 71 was
based on NMR spectroscopy and was proven by chemical
correlation with 70 (Vide infra). It is notable that H(5), H(9),
and H(16) are shifted dramatically upfield in structure 71
compared to structure 70. For example, H(9) in 71 is a doublet
at 6.38 ppm, whereas there are no aromatic protons upfield of
6.88 ppm in 70. This trend was general and proved useful in
quickly determining oxindole stereochemistry for analogs of 70
and 71.
Retroaldol-Aldol Approach to C(7) Stereochemistry.
Although the yields of 70 and 71 were too low to be useful in
a total synthesis, it was useful to know that reduction of the
R-alkoxy radicals at C(3) gave the required stereochemistry at
that center. This study also enabled examination of a retroaldol-
aldol approach to adjusting the oxindole stereochemistry.44 Thus,
removal of the MOM protecting groups of 70 and 71 with
aqueous hydrochloric acid in methanol-dimethoxyethane gave
alcohols 72 (48%) and 73 (32%), respectively. When treated
with DBU in dichloromethane at reflux for only 20 min, both
72 and 73 afforded isomeric alcohol 74 in 85% and 48% yields,
respectively. The structure of 74 was consistent with spectro-
scopic data and was confirmed by X-ray crystallography.23 This
experiment showed that thermodynamics could be used to set
the oxindole stereochemistry. It was also noted that warming
70 with hydrochloric acid in aqueous methanol at reflux for 15
h gave a mixture of 72 and 74. Thus, it was also possible to
effect the retroaldol-aldol process under acidic conditions,
although the rate was slower than the base-induced isomeriza-
tion.45
Since the retroaldol-aldol was slow in acid, it was decided
to examine tetrahydropyran construction under such conditions.
Ozonolysis of olefin 70 gave aldehyde 77 (65%). Treatment
of 77 with hydrochloric acid accomplished hydrolysis of the
methyoxymethyl ether and isomerization of the C(16) aldehyde
to afford hemiacetal 78 in 54% yield. It is significant that
hemiacetal formation occurred without isomerization at C(3)
and C(7). The stereochemistry of the oxindole was confirmed
by NOE studies. For example irradiation of H(9) showed 2%,
5%, and 9% enhancements at H(3), the axial H(14) proton, and
H(19), respectively. Finally, treatment of 78 with triethylsilane
and trifluoroacetic acid in dichloromethane gave tetrahydropyran
79 in 63% yield.46 Reduction of the N-methoxymethyl protect-
ing group was observed under these conditions, suggesting that
the oxindole protecting group should be changed in subsequent
studies or that the protecting group would have to be removed
prior to hemiacetal reduction.
Observation of a Cyclization-Fragmentation. The cy-
clization of 69 did not supply enough 70 and 71 to make
progress toward gelsemine. The major problem with the
cyclization was the 1,4-hydrogen atom transfer (Scheme 6)
which required that the cyclization be run at high concentrations,
conditions not conducive to high yields. To deal with this
problem, it was decided to postpone introduction of the
diphenylethylene double bond until after the cyclization, hoping
that this change would slow the rate of 1,4-hydrogen atom
transfer. This plan was executed as shown in Scheme 9.
Treatment of ester 41 with phenylmagnesium bromide gave
carbinol 80 (87%). Alkylation of 80 using iodomethane and
sodium hydride in dimethylsulfoxide afforded methyl ether 81
in quantitative yield. Ketal hydrolysis (81 f 82) was ac-
complished in 94% yield without elimination of methanol using
p-toluenesulfonic acid in wet acetone. Ketone 82 was acylated
with o-bromophenylisocyanate to give 83 (80%) which was
converted sequentially to 84 and 85 in 84% overall yield.
Tri-n-butyltin hydride mediated cyclization of 85 proceeded
smoothly under high dilution conditions to afford a 55:45
mixture of oxindoles 86 and 87 in 83% combined yield. Pure
samples of 86 and 87 could be obtained in diminished yield
Since the C(7) stereochemical correction noted above was
accomplished under the conditions used to correct the C(16)
stereochemistry, it was hoped aldehyde 75 (prepared by ozo-
(44) For another use of a retroaldol-aldol strategy to adjust stereochem-
istry, see: Hamelin, O.; Depres, J.-P.; Greene, A. E.; Tinant, B.; Declercq,
J.-P. J. Am. Chem. Soc. 1996, 118, 9992.
(45) For model studies, see: Wu, S. C. Ph.D. Thesis, The Ohio State
University, 1991.
(46) Kursanov, D. N.; Parnes, Z. N.; Loim, N. M. Synthesis 1974, 633.

6232 J. Am. Chem. Soc., Vol. 119, No. 27, 1997
Atarashi et al.
Scheme 9
only after a difficult separation on silica gel followed by a
recrystallization. The increased yield in the cyclization was
welcome, but formation of a C(3) ketone was a surprise. This
product undoubtedly was the result of fragmentation of an
intermediate R-alkoxy radical.47 Thus, while the rate of 1,4-
hydrogen atom transfer was slowed, another process reared its
head. Several attempts to avoid the fragmentation failed. For
example, benzyl ether 88 also underwent cyclization-fragmen-
tation to provide a 1:1 mixture of 86 and 87 in 78% yield, and
attempts to prepare silyl enol ether 89, which was unlikely to
undergo fragmentation, were unsuccessful.48 It was possible,
however, to apply the knowledge gained in the retroaldol-aldol
studies to keto oxindoles 86 and 87. Treatment of pure samples
of 86 and 87 with p-toluenesulfonic acid in dichloromethane
gave olefins 90 and 91 in 88% and 98% yields, respectively.
The structure of 90 was established by X-ray crystallography,23
and the structure of 91 was assigned on the basis of spectral
data. Isomerization of 90 to 91 could be accomplished in 85%
yield using a catalytic amount of potassium cyanide in N,N-
dimethylformamide, presumably via one or the other of two
retroacylation-acylation pathways. Operationally, the easiest
way to prepare 90 was to convert the radical cyclization products
(86 + 87) to a mixture of olefins (90 + 91) and then subject
this material to the potassium cyanide mediated isomerization.
In this manner, 91 could be prepared in 71% overall yield from
85.
example, NaBH3CN at pH4 or NaBH4/AcOH] or free radical
conditions [for example (TMS)3SiH-AIBN] failed to produce
the desired products. In addition, aldehyde 92, prepared in 67%
yield by ozonolysis of the C(20)-methoxymethyl analog of 91,
did not undergo chemistry related to the conversion of aldehyde
48 to hydroxy ester 51 upon treatment with DBU in methanol.
Only a complex mixture of unidentified products was obtained.
Eventually this promising route was abandoned when another
approach, being pursued simultaneously, proved successful.
Synthesis of the Gelsemine Cage. To overcome the problem
of fragmentation of the C(3) R-alkoxy radical generated in the
aryl radical cyclization, benzyl ether 88 and silyl enol ether 89
were selected for study. As mentioned above, 88 also underwent
cyclization-fragmentation, and we were unable to prepare 89.
Promising results were eventually achieved using substrate 93,
which was prepared in 98% yield by treating 83 with acetic
anhydride and triethylamine in the presence of 4-(dimethylami-
no)pyridine as shown in Scheme 10. Free radical cyclization
of 93 provided three oxindoles, 94 (9%), 95 (7%), and 96
(42%).49 The stereochemical assignments were based on
treatment of pure samples of 94-96 with p-toluenesulfonic acid
in dichloromethane followed by NOE studies on the resulting
olefins 97 (44%), 98 (40%), and 99 (83%). Although we are
still at a loss to explain why the various cyclization substrates
gave different stereochemical results at the oxindole center, it
was gratifying to find that 93 provided mainly the stereochem-
istry needed to proceed with the synthesis of gelsemine.
It was felt that if ketone 91 could be reduced to alcohol 73,
it might be possible to complete the construction of the
gelsemine cage by sequential ozonolysis followed by epimer-
ization-trapping of the resulting aldehyde. Unfortunately we
were unable to accomplish the desired reduction. It was not
surprising that reduction of 91 with sodium borohydride gave
only equatorial alcohol 74. It is probable that 73 was formed
initially but simply isomerized to 74 under the basic reaction
conditions. Attempts to reduce 91 under acidic conditions [for
The synthesis of gelsemine was eventually accomplished from
oxindole 96. Treatment of this compound with p-toluene-
sulfonic acid in dichloromethane, followed by addition of
methanol to the reaction mixture, effected sequential production
of 99 and 100 in 93% overall yield. Ozonolysis of olefin 100
followed by a reductive workup gave aldehyde 101 in 65%
yield, along with a compound resulting from epoxidation of 100,
as a single diastereomer, in 15% yield. Next, treatment of 101
with aqueous hydrochloric acid in dimethoxyethane at 48 °C
(47) Hart, D. J.; Kuzmich, D. J. Chinese Chem. Soc. 1995, 42, 873.
(48) The problems with silyl enol ethers were a surprise given our success
with the preparation and cyclization of such compounds in a decalin model
system.45
(49) Steglich, W.; Neises, B. Angew. Chem., Int. Ed. Engl. 1978, 17,
522.

Free Radical Cyclizations in Alkaloid Total Synthesis
J. Am. Chem. Soc., Vol. 119, No. 27, 1997 6233
Scheme 10
Scheme 11
103. Also possible, hemiacetal 102 could afford 106, which
could undergo enol ether hydrolysis followed by ketalization
to afford 103. Although the experiment was not performed,
conversion of 102 to 103 could be confirmed by resubjecting
pure 102 to the reaction conditions. It is noted that by thin-
layer chromatography, it appeared that formation of cyclic acetal
103 occurred only after formation of 102. Fortunately, the
formation of 103 could be minimized with careful control of
the temperature during the hydrolysis of 101. With hemiacetal
102 in hand, treatment with triethylsilane and trifluoroacetic
acid in dichloromethane afforded 107 in 83% yield.46 A 15%
NOE at H(5) upon irradiation of H(9), amongst other difference
NOE experiments, clearly established that 107 was the intact
gelsemine cage with the proper stereochemistry at C(7).
Introduction of the Vinyl Group and Completion of the
Synthesis. The original plan for introduction of the C(20) vinyl
group was to remove the benzyl ether protecting group, oxidize
the resulting alcohol to the aldehyde, and conduct a Wittig
olefination. The viability of a Wittig reaction as a final step in
a synthesis of gelsemine had been demonstrated by Landeryon.50
He had shown that protection of the oxindole portion of
gelsemine as the N-benzyl derivative followed by oxidation of
the C(20) vinyl group gave the noraldehyde of Na-benzyl-
gelsemine and a Wittig reaction with (methylidene)triph-
enylphosphorane gave Na-benzylgelsemine in 46% yield. We
first examined this approach on cage structure 44 (Scheme 12).
Treatment of 44 with boron tribromide in dichloromethane gave
alcohol 108 in 75% yield.38 A Swern oxidation proceeded
smoothly to afford aldehyde 109 in 92% yield.51 Alternatively,
the Dess-Martin periodinane gave aldehyde 109 in 80% yield.52
Unfortunately, Wittig olefination of 109 was unsuccessful (Vide
infra), and, thus, other methods were explored. For example
aldehyde 109 was treated with methylmagnesium bromide in
diethyl ether to provide the corresponding secondary alcohol
in 85% yield as a mixture of diastereomers. Dehydration of
the resulting alcohol with Martin’s sulfurane gave olefin 110
in quantitative yield.53 Unfortunately, this addition-elimination
sequence also failed to provide any 21-oxogelsemine when
for 18 h accomplished acetate hydrolysis and isomerization of
the aldehyde to afford diastereomeric hemiacetals 102 (65%)
and cyclic acetal 103 (14%). Initially this transformation was
accomplished with tetrahydrofuran as the cosolvent. Due to
the long reaction time required to hydrolyze the O-acetate,
however, conversion of tetrahydrofuran to 4-chlorobutanol also
occurred. The yield of 102 under these conditions was typically
poor and may be attributed to 4-chlorobutanol reacting with
either aldehyde 101 or hemiacetal 102. Cyclic acetal 103 was
an intriguing side product. Its structure was based on spectral
analysis and mechanistic considerations. The mass spectrum
indicated a parent ion at m/z 446, isomeric with hemiacetal 102.
One possible structure would be the product of acetate hydrolysis
prior to formation of the hemiacetal; however, this structure
was quickly ruled out since both the H and 13C NMR spectra
1
indicated the absence of an aldehyde. The 1H NMR also
indicated a single isomer, and comparison of the 13C NMR of
103 with hemiacetal 102 indicated a new triplet and one less
doublet. These and other NMR data were consistent with
proposed structure 103. Furthermore, mechanistic consider-
ations suggest that 103 might be formed under the reaction
conditions. Construction of the C(5)sC(16) bond of gelsemine
has been accomplished via acyliminium ion chemistry,16 and
thus, it seems reasonable that structure 103 could be derived
from either 101 or 102 as outlined in Scheme 11. Acetate
hydrolysis of 101 could be followed by a retro-Mannich reaction
of the resulting alcohol 104 to give a structure of type 105,
which in turn could undergo intramolecular ketalization to afford
(50) Landeryon, V. A. Ph.D. Thesis, University of Rochester, 1965; Diss.
Abstr. 1965, 26, 2477.
(51) Mancuso, A. J.; Swern, E. Synthesis 1981, 165.
(52) Dess, D. B.; Martin, J. C. J. Org. Chem. 1983, 48, 4155. Ireland,
R. E.; Liu, L. J. Org. Chem. 1993, 58, 2899.

6234 J. Am. Chem. Soc., Vol. 119, No. 27, 1997
Atarashi et al.
Scheme 12
confirms that free radical cyclization strategies can be applied
successfully to complex problems in alkaloid synthesis but also
shows that our understanding of the stereochemical course of
such reactions and competing radical reactions is not yet
sufficient to preclude empirical examination of substituents in
attempts to achieve a desired (or useful) reaction outcome. It is
also hoped that several of the transformations uncovered herein
(the conditions for direct carbamoylation of a ketone, the radical
cyclization-fragmentation process, the epimerization strategies
used to construct the tetrahydropyran and adjust oxindole
stereochemistry) may find use elsewhere in complex molecule
synthesis. Finally, we note that the chemistry of the gelsemine
cage uncovered in this research suggests several shorter but high
risk routes to gelsemine that might be interesting to explore.
Experimental Section
All melting points are uncorrected as are all boiling points. Proton
nuclear magnetic resonance spectra are recorded in parts per million
from internal chloroform, benzene, or dimethylsulfoxide on the δ scale
and are reported as follows: chemical shift [multiplicity (s ) singlet,
d ) doublet, t ) triplet, q ) quartet, qu ) quintet, m ) multiplet),
coupling constants in hertz, integration]. 13C NMR data are reported
as follows: chemical shift (multiplicity determined using DEPT
experiments). Mass spectra were using an ionization energy of 70 eV
unless stated otherwise. Solvents and reagents were dried and purified
prior to use when deemed necessary; benzene, diethyl ether, THF, and
toluene were distilled from sodium metal; dichloromethane was distilled
from calcium hydride or passed through activity I alumina; and
diisopropylamine was distilled from calcium hydride. Reactions
requiring an inert atmosphere were run under nitrogen. Column
chromatography was performed over EM laboratories silica gel (70-
230 or 230-400 mesh). All organometallic reagents (Grignard, organic
lithiums) were titrated prior to use with menthol using 1,10-phenan-
throline as the indicator.56 The order of experimental procedures follow
their order of appearance in the text. For ozonolyses, the ozone was
delivered using a Welsbach ozone generator.
(E)-6-[(Tetrahydro-2H-pyran-2-yl)oxy]-3-hexene-2-one (25).
Through a solution of 50 g (320 mmol) of olefin 23 in 400 mL of
dichloromethane cooled in a dry ice-acetone bath was passed a stream
of ozone until a pale blue color persisted. The excess ozone was purged
from the mixture by passing nitrogen through the solution until it
became clear. To the mixture was then added 132 g (413 mmol) of
(2-oxopropylidene)triphenylphosphorane57 in several portions. The cold
bath was then removed, and the mixture was warmed to room
temperature. A mild exotherm occurred at 20-25 °C, and the mixture
was cooled to 15 °C. The mixture was then warmed to room
temperature, stirred for 48 h, and concentrated in vacuo. The residue
was diluted with ether-hexane and filtered, and the filtrate was
concentrated in vacuo. The residue was dissolved in EtOAc-hexanes
(15:85) and passed through 450 g of silica gel (EtOAc-hexanes, 1:9).
The eluent was concentrated in vacuo, and the residue was distilled
under reduced pressure (118-124 °C, 4 Torr) to afford 39 g (61%) of
enone 25 as a colorless oil. Spectroscopic data for this compound have
been reported elsewhere.20
applied to the relevant system (Vide infra). It was suspected
that either steric hindrance and/or neighboring group participa-
tion of the oxindole carbonyl group was causing the problems.
Yet another olefination was examined using 109. Treatment
of 109 with bis(cyclopentadienyl)dimethyltitanium (Petasis’
reagent)54 in tetrahydrofuran at reflux gave 110 in 94% yield.
At this point we proceeded cautiously and examined this
olefination with a substrate derived from oxindole 79. This
substrate is sterically congested, but the oxindole is disposed
such that neighboring group participation is not possible.
Deprotection of the benzyl ether with boron tribromide gave
alcohol 111 (72%), and Swern oxidation of the alcohol provided
aldehyde 112 (86%). Olefination of 112 with dimethylti-
tanocene was successful as olefin 113 was obtained in 60%
yield.
Application of this reaction sequence to 107 was successful.
Treatment of the benzyl ether with boron tribromide gave
alcohol 114 (95%) whose structure was confirmed by X-ray
crystallography. Swern oxidation of 114 gave aldehyde 115 in
only 50% yield, but this oxidation proceeded in 71% yield when
the Dess-Martin periodinane was used. Finally treatment of
115 with Petasis’ reagent in tetrahydrofuran at reflux gave (()-
21-oxogelsemine (2) in 87% yield. The 1H and 13C NMR of 2
were in agreement with reported data.7 Furthermore a direct
1
comparison with an authentic sample gave identical H NMR
and TLC results.55 Since (()-21-oxogelsemine has been
converted to (()-gelsemine (1) in a single step by the Johnson
and Speckamp groups,15,16 this synthesis of 2 also constitutes a
synthesis of racemic gelsemine. Finally, we note that the seven-
step reaction sequence for conversion of 96 to 2 was also used
to prepared 7-epi-21-oxogelsemine (116) from oxindole 94.
In summary, this route to 21-oxogelsemine (2) described
herein requires 27 steps from 3-buten-1-ol, and the overall yield
is 0.25%. These numbers compare favorably with the Johnson
(27 steps from 2-phenylethanol and 0.26%) and Fukuyama (31
steps from methyl acetoacetate and 0.73%) syntheses but fall
short of the Speckamp synthesis (19 steps from 3,5-hexadien-
1-ol and 1.1%). Certainly more important, this research
Trimethyl[[(E)-1-methylene-5-[(tetrahydro-2H-pyran-2-yl)oxy]-
2-pentenyl]oxy]silane (26). To a solution of 65 mL (467 mmol) of
diisopropylamine in 500 mL of THF cooled in a dry ice-acetone bath
was added 200 mL (400 mmol) of 2.0 M n-BuLi in hexane. The
mixture was stirred for 30 min and then warmed to -40 °C. After 10
min, the mixture was cooled to -78 °C, and 75 g (378 mmol) of enone
25 in 50 mL of THF was added over a 20-min period. The mixture
was stirred for 20 min, and then 59 mL (467 mmol) of chlorotrimeth-
ylsilane in 30 mL of THF was added over a 20-min period. After 30
min, the cold bath was removed, and the mixture was stirred at room
temperature for 4 h. The mixture was filtered through Celite, the
filtercake was washed with hexane, and the filtrate was concentrated
(53) Arhart, R. J.; Martin, J. C. J. Am. Chem. Soc. 1972, 94, 5003. Martin,
J. C.; Franz, J. A. J. Am. Chem. Soc. 1975, 97, 6137.
(54) Petasis, N. A.; Bzowej, E. I. J. Am. Chem. Soc. 1990, 112, 6392.
Petasis, N. A.; Lu, S.-P. Tetrahedron Lett. 1995, 36, 2393.
(55) We thank Professor G. Cordell for kindly supplying a sample of
authentic 21-oxogelsemine.
(56) Watson, S. C.; Eastham, J. F. J. Organometallic Chem. 1967, 9,
165.
(57) For the preparation of (2-oxopropylidene)triphenylphosphorane,
see: Ramirez, F.; Dershowitz, S. J. Org. Chem. 1957, 22, 41.

Free Radical Cyclizations in Alkaloid Total Synthesis
J. Am. Chem. Soc., Vol. 119, No. 27, 1997 6235
in vacuo. The residue was diluted with 500 mL of hexane and filtered,
and the filtrate was concentrated in vacuo to provide 102 g (99%) of
diene 26 as a yellow oil which was used without further purification.
Spectroscopic data for this compound have been reported elsewhere.20
evolution ceased. The mixture was cooled to 30 °C, and 15 mL (0.18
mmol) of ethyl iodide was added over a 5-min period. The mixture
was stirred at 30 °C for 1 h and then at 45 °C for 4 h. The mixture
was then cooled to 0 °C, and the excess NaH was cautiously destroyed
with EtOH. The mixture was partitioned between 500 mL of brine
and 200 mL of EtOAc. The organic layer was separated and the
aqueous phase was extracted with two 200-mL portions of EtOAc. The
combined EtOAc layers were washed with three 200-mL portions of
brine and dried (MgSO4), and decolorizing carbon (Norit A) was added.
The mixture was filtered through Celite, and the filtrate was concen-
trated in vacuo causing a precipitate to form. Hexane was added to
the precipitate, and the resulting solid was collected to afford 21 g
(71%) of 32 (115-121 °C). The concentrated filtrate (6.9 g) was
chromatographed over 60 g of silica gel (EtOAc-hexane, 3:7 f 1:1)
and triturated with hexane to afford an additional 2.4 g (8%) of 32.
The mother liquor gave 3.4 g (11%) of additional material that was a
(()-(8R*,9S*,10S*)-10-(2-Hydroxyethyl)-N,3,3-trimethyl-1,5-
dioxaspiro[5,5]undecane-8,9-dicarboximide (28). A mixture of 102
g (0.38 mol) of diene 26 and 50 g (0.45 mol) of N-methylmaleimide
in 800 mL of toluene was heated at reflux for 15 h. An additional 12
g (0.108 mmol) of N-methylmaleimide was added, and the mixture
was stirred at reflux for 1 h. The mixture was then cooled to room
temperature, and 106 g (1.019 mol) of 2,2-dimethyl-1,3-propanediol
and 3 g of p-toluenesulfonic acid monohydrate were added. The
apparatus was fitted with a Dean-Stark trap and warmed under reflux
with removal of water for 16 h. The mixture was cooled, concentrated
in vacuo to a volume of 550 mL, and stirred over 25 g of potassium
carbonate for 30 min. The mixture was passed through a 600 g pad of
silica gel (EtOAc-hexanes, 25:75 f 1:1 f 75:25 f EtOAc). The
eluent was concentrated in vacuo and placed on the Kugelrhor (75 °C,
0.1 Torr) to remove excess N-methylmaleimide and 2,2-dimethyl-1,3-
propanediol to provide 46 g (39%) of imide 28 which hardened to an
amber resin upon cooling. Spectroscopic data for this compound have
been reported elsewhere.20
1
1:1 mixture of diastereomers 32 and 33 by H NMR. Spectroscopic
data for this compound have been reported elsewhere.20
(()-(3′R*,3′aS*,7′S*,7′aR*)-7′a-[(Benzyloxy)methyl]-3′-ethoxytet-
rahydro-2′,5,5-trimethyl-7′-vinylspiro[m-dioxane-2,5′(4′H)-isoin-
dolin]-1′-one (35). To a solution of 8.70 mL (62.1 mmol) of
diisopropylamine in 50 mL of THF at -78 °C was added 20.80 mL of
2.5 M n-BuLi in hexane. The mixture was stirred for 30 min, and
13.5 g (41.7 mmol) of 32 in 30 mL of THF was added over a 15-min
period. The mixture was stirred at -78 °C for 15 min and was then
warmed to -20 °C. After 10 min, the mixture was again cooled to
-78 °C, and 7.2 g (46.19 mmol) of benzyl chloromethyl ether was
added. The mixture was stirred for 15 min at -78 °C, warmed to
room temperature, stirred for 1 h, and partitioned between 200 mL of
EtOAc and 300 mL of water. The organic layer was separated, and
the aqueous phase was extracted with three 75-mL portions of EtOAc.
The combined EtOAc layers were washed with four 200-mL portions
of brine, dried (MgSO4), treated with carbon (Norit A), filtered through
Celite, and concentrated in vacuo. The residue was chromatographed
over 95 g of silica gel (EtOAc-hexane, 1:4) to afford 17.8 g (96%) of
35 as a viscous oil: IR (neat) 1697 cm-1; 1H NMR (300 MHz, CDCl3)
δ 0.85 (s, 3H), 1.02 (s, 3H), 1.15 (t, J ) 6.6 Hz, 3H), 1.37 (dd, J )
13.7, 13.3 Hz, 1H), 1.82 (dd, J ) 15.3, 6.4 Hz, 1H), 1.98 (dd, J )
14.4, 6.2 Hz, 1H), 2.09 (dd, J ) 14.1, 3.3 Hz, 1H), 2.69-2.69 (m,
2H), 2.79 (s, 3H), 3.38-3.52 (m, 6H), 3.60 (ABq, J ) 9.0 Hz, 2H),
4.44 and 4.49 (ABq, J ) 12.2 Hz, 2H), 4.59 (d, J ) 2.4 Hz, 1H), 4.99
(dd, J ) 9.0, 2.0 Hz, 1H), 5.03 (s, 1H), 6.21 (ddd, J ) 16.6, 10.4, 9.5,
Hz, 1H), 7.2-7.29 (m, 5H); 13C NMR (75.5 MHz, CDCl3) δ 15.3 (q),
22.2 (q), 22.6 (q), 27.4 (q), 29.9 (s), 33.7 (t), 34.6 (t), 38.5 (d), 39.5
(d), 51.4 (s), 62.4 (t), 69.8 (t), 69.9 (t), 72.2 (t), 73.1 (t), 93.7 (d), 97.8
(s), 115.6 (t), 127.2 (d), 127.3 (d), 128.0 (d) 138.3 (d), 174.0 (s), one
aromatic singlet was not resolved; exact mass calcd for C26H37NO5
m/z 443.2672, found m/z 443.2682.
(()-(8R*,9S*,10R*)-N,3,3-Trimethyl-10-vinyl-1,5-dioxaspiro[5,5]-
undecane-8,9-dicarboximide (30). To a solution of 29.9 g (93.9
mmol) of alcohol 28 and 25.5 g (112.3 mmol) of o-nitrophenylsele-
nocyanate in 300 mL of THF cooled in an ice bath was added 30 mL
(121.3 mmol) of tri-n-butylphosphine over a 5-min period. The mixture
was warmed to room temperature and stirred for 2.25 h. The mixture
was cooled in an ice bath, and 52 g (196 mmol) of anhydrous Na2-
HPO4 was added. After 5 min, 144 mL of 30% aqueous hydrogen
peroxide was added while maintaining the temperature below 0 °C.
The mixture was stirred at 40 °C for 1 h and at room temperature for
12 h. The solution was concentrated in vacuo, and the residue was
partitioned between 200 mL of EtOAc and 300 mL of water. The
organic layer was separated, and the aqueous phase was extracted with
three 200-mL portions of EtOAc. The combined EtOAc extracts were
washed with three 100-mL portions of brine, dried (MgSO4), filtered,
and concentrated in vacuo. The residue was diluted with ether-hexane
(1:1), filtered, and concentrated in vacuo. The residue was chromato-
graphed over 240 g of silica gel (EtOAc-hexanes, 1:19 f 1:9 f 2:8)
to afford 21.9 g (79%) of alkene 30 as a viscous red oil. Spectroscopic
data for this compound have been reported elsewhere.20
(()-(3′R*,3′aS*,7′S*,7′aR*)-Tetrahydro-3′-hydroxy-2′,5,5-tri-
methyl-7′-vinylspiro[m-dioxane-2,5′(4′H)-isoindolin]-1′-one (31). To
a solution of 10.3 g (35.2 mmol) of imide 30 in 120 mL of MeOH
cooled to -23 °C was added 4.0 g (105.7 mmol) of sodium borohydride
over a 5-min period in several portions. The mixture was stirred at
-23 °C for 30 min and then at room temperature for 2 h and then
concentrated in vacuo. The residue was partitioned between 100 mL
of EtOAc and 100 mL of saturated aqueous sodium bicarbonate. The
basic aqueous layer was separated and extracted with five 50-mL
portions of EtOAc. The combined EtOAc layers were dried (MgSO4),
filtered, and concentrated in vacuo. The residual orange oil was
chromatographed over 340 g of silica gel (ether; then EtOAc-hexane,
8:2) to provide 8.4 g (80%) of carbinol 31 as a viscous oil which
solidified to an off white solid under high vacuum: mp 119-120 °C;
IR (KBr) 3416, 1667 cm-1; 1H NMR (300 MHz, C6D6) δ 0.50 (s, 3H),
0.80 (s, 3H), 1.52-1.66 (m, 2H), 2.21-2.25 (m, 3H), 2.31-2.37 (m,
1H), 2.51-2.62 (m, 1H), 2.64 (s, 3H), 3.12-3.26 (m, 4H), 3.61 (broad,
1H, OH), 4.53 (m, 1H), 5.06 (m, 1H), 5.10 (s, 1H), 6.93-7.05 (m,
1H); 13C NMR (75.5 MHz, C6D6) δ 22.1 (q), 22.2 (q), 25.6 (q), 28.5
(t), 29.6 (s), 34.8 (t), 37.8 (d), 38.3 (d), 45.6 (d), 69.5 (t), 69.6 (t), 83.3
(d), 97.6 (s), 113.5 (t), 141.26 (d), 172.7 (s); exact mass calcd for C16H25-
NO4 m/z 295.1784, found m/z 295.1775. This compound decomposed
when NMR data were collected in CDCl3.
(()-(3′R*,3′aS*,7′R*,7′aS*)-7′a-[(Benzyloxy)methyl]-3′-ethoxytet-
rahydro-2′,5, 5-trimethyl-1′-oxospiro[m-dioxane-2,5′(4′H)-isoindo-
line]-7′-carboxaldehyde (37). Through a solution of 10.0 g (22.5
mmol) of olefin 35 in 200 mL of CH2Cl2-MeOH (95:5) cooled to -86
°C was passed ozone gas until a pale blue color was evident, and starting
material was no longer present by thin-layer chromatography (EtOAc-
hexane, 3:7). Excess ozone was then purged from the mixture using
a stream of nitrogen, and then 20 mL of dimethyl sulfide was added in
one portion. The cold bath was then removed, and the mixture was
stirred at room temperature for 10 h. The mixture was diluted with
200 mL of brine, and the organic phase was separated. The aqueous
phase was extracted with two 30-mL portions of CH2Cl2. The combined
CH2Cl2 layers were washed with three 50-mL portions of brine and
dried (MgSO4), decolorizing carbon was added, and the mixture was
filtered through Celite and concentrated. The residue was triturated
with ether-hexane to provide 7.8 g (78%) of aldehyde 37 which was
used without further purification. An analytically pure sample was
obtained from a single recrystallization from ether-hexane to afford
aldehyde 37 as a crystalline white solid: mp 117-119 °C; IR (KBr)
1711 cm-1; 1H NMR (300 MHz, CDCl3) δ 0.83 (s, 3H), 1.06 (s, 3H),
1.16 (t, J ) 7.0 Hz, 3H), 1.23 (dd, J ) 14.2, 12.6 Hz, 1H), 1.36 (dd,
J ) 13.8, 10.1 Hz, 1H), 2.11 (ddd, J ) 13.8, 6.6, 2.1 Hz, 1H), 2.60-
2.68 (m, 2H), 2.84 (s, 3H), 2.88 (dd, J ) 12.5, 4.2 Hz, 1H), 3.31-
3.56 (m, 5H), 3.63 (d, J ) 11.5 Hz, 1H), 3.81 (d, J ) 9.4 Hz, 1H),
(()-(3′R*,3′aS*,7′R*,7′aR*)-3′-Ethoxytetrahydro-2′,5,5-trimethyl-
7′-vinylspiro[m-dioxane-2,5′(4′H)-isoindolin]-1′-one (32). To a sus-
pension of 4.8 g (0.12 mmol) of a 60% NaH dispersion in oil (washed
with three 10-mL portions of hexane) in 250 mL of THF warmed to
45 °C was added 26.9 g (91.2 mmol) of hydroxylactam 31 in several
portions. After the addition, the mixture was stirred until hydrogen

6236 J. Am. Chem. Soc., Vol. 119, No. 27, 1997
Atarashi et al.
of 41 as a white solid: mp 108-110 °C; IR (KBr) 1726, 1687 cm-1
3.94 (d, J ) 9.4 Hz, 1H), 4.35 (d, J ) 0.7 Hz, 1H), 4.53 and 4.58
(ABq, J ) 12.3 Hz, 2H), 7.23-7.34 (m, 5H), 10.12 (s, 1H); 13C NMR
(75.5 MHz, CDCl3) δ 15.2 (q), 22.1 (q), 22.6 (q), 26.8 (t), 28.2 (q),
29.9 (s), 35.4 (t), 39.1 (d), 46.5 (d), 51.3 (s), 63.6 (t), 69.9 (t), 70.0 (t),
72.4 (t), 73.2 (t), 95.4 (d), 96.6 (s), 127.3 (d), 127.5 (d), 128.2 (d),
137.8 (s), 174.1 (s), 202.0 (d); exact mass calcd for C25H35NO6 m/z
445.2465, found m/z 445.2472. Anal. Calcd for C25H35NO6: C, 67.39;
H, 7.92. Found: C, 67.30; H, 7.89.
;
1H NMR (300 MHz, CDCl3) δ 0.83 (s, 3H), 1.03 (s, 3H), 1.24 (t, J )
7.1 Hz, 3H), 1.65 (m, 1H), 1.81 (dd, J ) 14.6, 2.8 Hz, 1H), 2.05 (m,
2H), 2.17 (dd, J ) 16.5, 9.3 Hz, 1H), 2. 23 (dd, J ) 16.5, 6.2 Hz,
1H), 2.42 (m, 1H), 2.61 (dd, J ) 14.6, 3.8 Hz, 1H), 2.88 (s, 3H), 3.10
(m, 1H), 3.27 (dd, J ) 11.4, 1.6 Hz, 1H), 3.40 (d, J ) 11.5 Hz, 2H),
3.5 (t, J ) 2.2 Hz, 1H), 3.66 (d, J ) 11.5 Hz, 1H), 3.71 (d, J ) 9.7
Hz, 1H), 3.90 (d, J ) 9.7 Hz, 1H), 4.11 (q, J ) 7.1 Hz, 2H), 4.58 and
4.64 (ABq, J ) 12.4 Hz, 2H), 7.20-7.38 (m, 5H); 13C NMR (75.5
MHz, CDCl3) δ 14.0 (q), 22.1 (q), 22.6 (q), 29.6 (s), 30.3 (q), 33.4 (t),
35.3 (t), 35.9 (t), 39.1 (d), 41.1 (d), 48.6 (d), 55.4 (s), 60.2 (t), 65.0 (t),
65.6 (d), 69.4 (t), 70.4 (t), 73.4 (t), 96.3 (s), 127.1 (d), 127.4 (d), 128.0
(d), 138.5 (s), 172.2 (s), 176.0 (s); mass spectrum m/z (rel intensity)
231 (24), 91 (93), 57 (100); no parent peak was observed. Anal. Calcd
for C27H37NO6: C, 68.77; H, 7.91. Found: C, 68.68; H, 7.93. The
filtrate was then concentrated to 100 mL, passed through a 110 g pad
of silica gel (ether-hexane; 1:1) and recrystallized from ether-hexane
Ethyl (()-(rE,3′R*,3′aS*,7′S*,7′R*)-7′a-[(Benzyloxy)methyl]-3′-
ethoxytetrahydro-2′,5,5-trimethyl-1′-oxospiro[m-dioxane-2,5′(4′H)-
isoindoline]-7′-acrylate (39). A mixture of 5.0 g (11.2 mmol) of 37
and 5.9 g (16.8 mmol) of (carbethoxymethylidene)triphenylphosphorane
in 55 mL of CH2Cl2 was heated under reflux for 42 h. An additional
1.3 g (3.7 mmol) of (carbethoxymethylidene)triphenylphosphorane was
then added, and the mixture was stirred under reflux for 54 h. The
mixture was then concentrated in vacuo, and the residue was chro-
matographed over 100 g of silica (EtOAc-petroleum ether, 2:7 then,
3:7) to provide 4.2 g (73%) of ester 39 and 1.45 g of mixed fractions.
The mixed fractions were chromatographed a second time to afford an
1
to provide 700 mg (4%) of 47 as a white solid: mp 155-156 °C; H
NMR (300 MHz, CDCl3) δ 0.85 (s, 3H), 1.05 (s, 3H), 1.22 (m, 1H),
1.22 (t, J ) 7.1 Hz, 3H), 1.35 (dd, J ) 13.3, 11.7 Hz, 1H), 1.96-2.05
(m, 3H), 2.19-2.32 (m, 2H), 2.50 (broad d, J ) 13.3 Hz, 1H), 2.66
(d, J ) 10.0 Hz, 1H), 2.87 (s, 3H), 3.34-3.64 (m, 7H), 3.97-4.10 (m,
3H), 4.46 (d, J ) 10.3 Hz, 1H), 7.23-7.34 (m, 3H), 7.48-7.51 (m,
2H); 13C NMR (75.5 MHz, CDCl3) δ 14.1 (q), 22.3 (q), 22.7 (q), 29.2
(t), 29.8 (s), 30.0 (q), 33.3 (t), 34.5 (d), 37.0 (d), 37.1 (t), 39.3 (d),
47.7 (s), 51.8 (t), 60.0 (t), 69.73 (t), 69.77 (t), 72.9 (t), 84.0 (d), 96.9
(s), 127.8 (d), 128.0 (d), 128.1 (d), 140.5 (s), 171.9 (s), 174.2 (s); mass
spectrum, m/z (rel intensity) 471 (M, 100), 425 (68), 383 (66), 317
(71), 57 (82); exact mass calcd for C27H37NO6 m/z 471.2620, found
m/z 471.2617.
1
additional 540 mg (9%) of 39: IR (neat) 1698 cm-1; H NMR (300
MHz, CDCl3) δ 0.84 (s, 3H), 1.04 (s, 3H), 1.15 (t, J ) 6.6 Hz, 3H),
1.27 (t, J ) 7.1 Hz, 3H), 1.41 (t, J ) 13.3 Hz, 1H), 1.69 (dd, J ) 14.1,
7.5 Hz, 1H), 2.05 (dd, J ) 14.2, 6.3 Hz, 1H), 2.16 (dd, J ) 13.9, 3.1
Hz, 1H), 2.64 (m, 1H), 2.80 (s, 3H), 2.88 (m, 1H), 3.37-3.62 (m, 8H),
4.16 (dq, J ) 7.2, 1.1 Hz, 2H), 4.47 (s, 2H), 4.50 (d, J ) 1.5 Hz), 5.77
(d, J ) 15.6 Hz, 1H), 7.21-7.33 (m, 5H), 7.38 (dd, J ) 15.6, 9.5 Hz,
1H); 13C NMR (62.2 MHz, CDCl3) δ 14.1 (q), 15.2 (q), 22.2 (q), 22.5
(q), 27.6 (q), 29.8 (s), 33.4 (t), 37.7 (d), 38.5 (d), 49.5 (s), 51.5 (s),
59.8 (t), 62.9 (t), 69.8 (t), 69.8 (t), 71.9 (t), 73.1 (t), 94.1 (d), 97.0 (s),
122.0 (d), 127.2 (s), 127.3 (d), 128.1 (d), 138.0 (s), 148.4 (d), 166.0
(s), 173.5 (s); exact mass calcd for C29H41NO7 m/z 515.2884, found
m/z 515.2865.
Ethyl (()-(rE,3′R*,3′aS*,7′S*,7′aR*)-7′a-[(Benzyloxy)methyl]tet-
rahydro-2′,5,5-trimethyl-1′-oxo-3′-(phenylthio)spiro[m-dioxane-2,5′-
(4′H)isoindoline]-7′-acrylate (40). A mixture of 4.7 g (9.1 mmol) of
ethoxylactam 39, 1.5 g (13.7 mmol) of thiophenol, 50 mg of
p-toluenesulfonic acid monohydrate, and 5 g of 4 Å molecular sieves
in 65 mL of CH2Cl2 was stirred for 3 h. The mixture was diluted with
100 mL of saturated aqueous sodium bicarbonate, and the CH2Cl2 layer
was separated. The basic aqueous phase was extracted with three 40-
mL portions of CH2Cl2. The combined CH2Cl2 layers were washed
with two 50-mL portions of brine, dried (MgSO4), filtered, and
concentrated in vacuo. The residue was chromatographed over 65 g
of silica gel (EtOAc-hexane, first 2:8, then 3:7) to afford 4.1 g (76%)
of 40 as a viscous oil: IR (neat) 1717, 1691 cm-1; 1H NMR (300 MHz,
CDCl3) δ 0.87 (s, 3H), 0.99 (s, 3H), 1.25 (t, J ) 7.1 Hz, 3H), 1.52 (dd,
J ) 14.2, 12.0 Hz, 1H), 1.82 (dd, J ) 14.7, 5.7 Hz, 1H), 2.03 (dd, J
) 14.3, 3.2 Hz, 1H), 2.34 (dd, J ) 14.7, 3.6 Hz, 1H), 2.66-2.72 (m,
2H), 2.78 (d, J ) 9.2 Hz, 1H), 2.89 (s, 3H), 3.37-3.48 (m, 5H), 4.13
(dq, J ) 6.4 Hz, 2H), 4.25 and 4.31 (ABq, J ) 12.1 Hz, 2H), 4.81 (d,
J ) 5.7 Hz, 1H), 5.70 (d, J ) 15.6 Hz, 1H), 7.09 (dd, J ) 15.6, 9.5
Hz, 1H,), 7.17-7.40 (m, 10H); 13C NMR (75.5 MHz, CDCl3) δ 14.1
(q), 22.3 (q), 22.4 (q), 27.9 (q), 29.8 (s), 31.5 (t), 34.4 (t), 38.0 (d),
40.1 (d), 51.4 (s), 59.9 (t), 69.8 (t), 69.9 (t), 71.7 (t), 72.1 (d), 73.0 (t),
98.0 (s), 122.4 (d), 127.3 (d), 127.4 (d), 128.1 (d), 128.4 (d), 129.0
(d), 131.7 (s), 134.4 (d), 137.8 (s), 147.6 (d), 165.8 (s), 172.5 (s); mass
spectrum m/z (rel intensity) 579 (M, 0.01), 472 (30), 470 (100), 379
(19), 91 (72); exact mass calcd for C27H36NO6 (M - SC6H5) m/z
470.2381, found m/z 470.2515.
(()-(1′R*,3′aS*,4′R*,7′aR*,8′R*)-3′a-[(Benzyloxy)methyl]tetrahy-
dro-8′-(2-hydroxy-2,2-diphenylethyl)-2′,5,5-trimethylspiro[m-diox-
ane-2,6′(5′H)-[1,4]methanoisoindoline]-3′-one (80). To a solution of
4.0 g (8.5 mmol) of ester 41 in 80 mL of THF cooled to -20 °C was
added a solution of PhMgBr [prepared from 8.18 g (52.1 mmol) of
bromobenzene and 1.10 g (45.2 mmol) of magnesium turnings in 35
mL of anhydrous THF] over a 35-min period. The mixture was stirred
at 10 °C for 5 h and at room temperature for 1 h. The mixture was
then cooled to -20 °C, quenched with 150 mL of saturated ammonium
chloride, and extracted with three 150-mL portions of EtOAc. The
combined EtOAc layers were washed with three 250-mL portions of
brine, dried (MgSO4), filtered, and concentrated in vacuo. The oily
residue was dissolved in 150 mL of ether, and 50 mL of hexanes was
added. The mixture was allowed to stand for 18 h, and then an
additional 200 mL of hexane was added. After 1 h, the resulting white
solid was collected and washed with EtOAc-hexane (1:3) to provide
4.3 g (87%) of alcohol 80 as a white solid: mp 187-188 °C; IR (neat)
1
3399, 1685 cm-1; H NMR (300 MHz, CDCl3) δ 0.73 (s, 3H, CH3),
0.99 (s, 3H), 1.55 (dd, J ) 14.4, 2.7 Hz, 1H), 1.75 (broad s, 1H), 1.90-
1.92 (m, 2H), 2.12 (dd, J ) 14.1, 7.5 Hz, 1H), 2.20 (dd, J ) 14.1, 7.5
Hz, 1H), 2.24 (s, 1H), 2.31 (broad, 1H), 2.42 (dd, J ) 14.1, 4.5 Hz,
1H), 2.75 (broad, 1H), 2.93 (dd, J ) 11.7, 1.8 Hz, 1H), 2.97 (s, 3H),
3.16 (dd, J ) 11.4, 1.8 Hz, 1H), 3.22 (d, J ) 11.6 Hz, 1H), 3.30 (d, J
) 11.4, 1H), 3.38 (t, J ) 2.4 Hz, 1H), 3.65 (d, J ) 9.7 Hz, 1H), 3.86
(d, J ) 9.7 Hz, 1H), 4.56 and 4.61 (ABq, J ) 12.6 Hz, 2H), 7.18-
7.39 (m, 15H); 13C NMR (75.5 MHz, CDCl3) δ 22.1 (q), 22.6 (q),
29.4 (s), 31.1 (q), 31.5 (t), 36.5 (t), 40.4 (d), 40.8 (d), 43.6 (t), 48.6
(d), 55.6 (s), 65.4 (t), 67.8 (d), 69.4 (t), 70.5 (t), 73.4 (t), 78.0 (s), 96.4
(s), 125.9 (d), 126.0 (d), 126.6 (d), 126.7 (d), 127.1 (d), 127.3 (d),
128.01 (d), 128.05 (d), 128.1 (d), 138.7 (s), 147.1 (s), 147.9 (s), 176.5
(s); exact mass calcd for C37H43NO5 m/z 581.3141, found m/z 581.3147.
Anal. Calcd for C37H43NO5: C, 76.39; H, 7.45. Found: C, 76.45; H,
7.49.
Ethyl (()-(1′R*,3′aS*,4′R*,7′aR*,8′R*)-3′a-[(Benzyloxy)methyl]-
tetrahydro-2′,5,5-trimethyl-3′-oxospiro[m-dioxane-2,6′(5′H)-[1,4]-
methanoisoindoline]-8′-acetate (41) and Ethyl (()-(3′R*,4′R*,4′aS*,-
7′aR*,10′aS*)-Octahydro-5,5,9′-tri methyl-10′-oxo-3′-phenylspiro-
[m-dioxane-2,6′(7′H)-[1H]pyrano[3,4-d]isoindole]-4′-acetate (47). To
a solution of 21.7 g (36.8 mmol) of sulfide 40 in 900 mL of benzene
heated at reflux was added a mixture of 11.6 g (40.1 mmol) of tri-n-
butyltin hydride and 200 mg of AIBN in 110 mL of benzene over a
44-h period using a syringe pump. The mixture was cooled to room
temperature and concentrated in vacuo, and the residue was diluted
with one part ether and then nine parts hexane. The mixture was chilled
to facilitate crystal growth, and the resulting solid was collected by
filtration and washed with ether-hexane (1:1) to afford 11.3 g (65%)
(()-(1′R*,3′aS*,4′R*,7′aR*,8′R*)-3′a-[(Benzyloxy)methyl]tetrahy-
dro-8′-(2-methoxy-2,2-diphenylethyl)-2′,5,5-trimethylspiro[m-diox-
ane-2,6′(5′H)-[1,4]methanoisoindoline]-3′-one (81). To a suspension
of 605 mg (15.12 mmol) of 60% NaH in mineral oil (washed with
three 10-mL portions of hexane) in 65 mL of DMSO warmed to 35 °C
was added 4.4 g (7.57 mmol) of alcohol 80 in several portions. The
mixture was stirred for 15 min and cooled to room temperature, and
2.14 g (15.13 mmol) of iodomethane was added. The mixture was
stirred for 45 min and then poured into 200 mL of ice cold saturated

Free Radical Cyclizations in Alkaloid Total Synthesis
J. Am. Chem. Soc., Vol. 119, No. 27, 1997 6237
aqueous ammonium chloride. The white solid was collected and
washed with 1 L of water to provide 4.55 g (100%) of methyl ether 81
which was used without further purification: mp 74-79 °C; IR (KBr)
1699 cm-1; 1H NMR (300 MHz, CDCl3) δ 0.80 (s, 3H, CH3), 1.05 (s,
3H, CH3), 1.40 (m, 1H), 1.45 (dd, J ) 17.1, 2.8 Hz, 1H), 1.91 (m,
2H), 2.01 (dd, J ) 14.3, 3.6 Hz, 1H), 2.20 (dd, J ) 14.3, 7.6 Hz, 1H),
2.29 (m, 1H), 2.44 (dd, J ) 14.3, 3.6 Hz, 1H), 2.80 (m, 1H), 3.01 (s,
3H), 3.02 (s, 3H), 3.12 (dd, J ) 2.8, 2.1 Hz, 1H), 3.16-3.33 (m, 3H),
3.51 (d, J ) 11.5 Hz, 1H), 3.61 (d, J ) 9.7 Hz, 1H), 3.83 (d, J ) 9.7
Hz, 1H), 4.55 and 4.59 (AB q, J ) 13.0 Hz, 2H), 7.15-7.35 (m, 15H);
13C NMR (75.5 MHz, CDCl3) δ 22.2 (q), 22.6 (q), 29.5 (s), 31.5 (q),
31.8 (t), 36.3 (t), 37.7 (t), 39.7 (d), 40.0 (d), 48.6 (d), 50.5 (q), 55.7
(s), 65.2 (t), 67.7 (d), 69.6 (t), 70.5 (t), 73.4 (t), 82.2 (s), 96.3 (s), 126.7
(d), 126.9 (d), 127.0 (d), 127.1 (d), 127.3 (d), 127.8 (d), 127.9 (d),
128.0 (d), 138.6 (s), 144.6 (s), 145.1 (s), 176.5 (s), one aromatic doublet
was not resolved; mass spectrum, m/z (rel intensity) 595 (M, 8), 489
(53), 474 (14), 398 (100), 197 (84), 91 (56); exact mass calcd for C38H45-
NO5 m/z 595.3297, found m/z 595.3294.
(10%) of 83 (mp 171-176 °C): IR (neat) 3409, 1697 cm-1; 1H NMR
(CDCl3) δ 1.43 (broad s, 1H), 1.77 (dd, J ) 18.9, 2.3 Hz, 1H), 2.12-
2.19 (m, 2H), 2.31 (dd, J ) 14.2, 7.7 Hz, 1H), 2.56 (dd, J ) 14.2, 3.3
Hz, 1H), 2.94 (broad s, 1H), 3.00 (s, 3H), 3.09 (s, 3H), 3.35 (d, J )
9.7 Hz, 1H), 3.52 (t, J ) 1.9 Hz, 1H), 3.79 (d, J ) 9.7 Hz, 1H), 4.47
(d, J ) 12.6 Hz, 1H), 4.61 (d, J ) 12.6 Hz, 1H), 7.01 (td, J ) 7.9, 1.5
Hz, 1H), 7.17-7.37 (m, 16H), 7.55 (dd, J ) 8.2, 1.5 Hz, 1H), 7.70 (s,
1H), 8.28 (dd, J ) 8.2, 1.5 Hz, 1H), 13.79 (s, 1H); 13C NMR (300
MHz, CDCl3) δ 31.5 (q), 34.6 (t), 37.4 (t), 39.4 (d), 45.1 (d), 47.9 (d),
50.6 (q), 55.9 (s), 64.1 (t), 69.6 (d), 73.3 (t), 82.1 (s), 96.9 (s), 114.4
(s), 122.1 (d), 125.4 (d), 127.02 (d), 127.06 (d), 127.1 (d), 127.3 (d),
128.0 (d), 128.1 (d), 128.3 (d), 132.2 (d), 134.8 (s), 138.1 (s), 144.1
(s), 144.4 (s), 169.3 (s), 173.3 (s), 174.8 (s), three aromatic doublets
were not resolved; mass spectrum, m/z (rel intensity) 403 (20), 197
(100), 173 (60), 171 (62), 91 (45); exact mass calcd for C40H39BrN2O5
m/z 706.2042, found m/z 706.2047. An analytically pure sample was
prepared by recrystallization from EtOH. Anal. Calcd for C40H39-
BrN2O5: C, 67.97; H, 5.57. Found: C, 67.70; H, 5.65.
(()-(1R*,3aS*,4R*,7aS*,8R*)-3a-[(Benzyloxy)methyl]-2′-bromo-
3a,4,5,7a-t etrahydro-8-(2-methoxy-2,2-diphenylethyl)-6-(meth-
oxymethoxy)-2-methyl-3-oxo-1,4-methanoisoindoline-7-carboxanil-
ide (84). To a solution of 200 mg (0.28 mmol) of â-ketoanilide 83 in
3 mL of DMSO was added 286 mg (2.82 mmol) of triethylamine
followed by 113 mg (1.41 mmol) of chloromethyl methyl ether. The
mixture was stirred for 1 h, poured into 40 mL of saturated aqueous
sodium bicarbonate, and extracted with three 40-mL portions of EtOAc.
The combined organic layers were washed with three 40-mL portions
of brine, dried (MgSO4), filtered, and concentrated to provide 212 mg
(100%) of â-ketoanilide 84 as a white solid which was used without
further purification: mp 172-177 °C. The following data were
obtained from a sample recrystallized from EtOAc-hexane: mp 191-
193 °C; IR (film) 3342, 1702, 1663 cm-1; 1H NMR (300 MHz, CDCl3)
δ 1.46 (broad, 1H), 1.74 (dd, J ) 18.4, 2.6 Hz, 1H), 1.99 (broad d, J
) 8.4 Hz, 1H), 2.17 (dd, J ) 18.4, 3.7 Hz, 1H), 2.26 (dd, J ) 14.4,
8.4 Hz, 1H), 2.56 (dd, J ) 14.3, 2.3 Hz, 1H), 2.98 (s, 3H), 3.12 (s,
3H), 3.27 (s, 3H), 3.32 (d, J ) 9.8 Hz, 1H), 3.47 (t, J ) 2.1, 1H), 3.64
(d, J ) 9.8 Hz, 1H), 3.66 (t, J ) 2.8 Hz, 1H), 4.44 (d, J ) 12.8 Hz,
1H), 4.61 (d, J ) 12.8 Hz, 1H), 4.97 (s, 2H), 6.95 (ddd, J ) 9.0, 7.5,
1.6 Hz, 1H), 7.12-7.34 (m, 16H), 7.55 (dd, J ) 8.0, 1.5 Hz, 1H), 8.64
(dd, J ) 8.4, 1.5 Hz, 1H), 10.04 (s, 1H); 13C NMR (75.5 MHz, CDCl3)
δ 31.0 (t), 31.3 (q), 37.0 (t), 40.2 (d), 44.3 (d), 49.3 (d), 50.5 (q), 54.0
(s), 56.7 (q), 64.4 (t), 70.7 (d), 73.1 (t), 82.0 (s), 92.8 (t), 109.6 (s),
112.4 (s), 121.4 (d), 124.2 (d), 126.8 (d), 126.9 (d), 127.0 (d), 127.1
(d), 127.3 (d), 127.8 (d), 128.0 (d), 128.1 (d), 128.2 (d), 132.2 (d),
137.1 (s), 138.5 (s), 144.3 (s), 144.6 (s), 158.7 (s), 163.8 (s), 175.4 (s),
one aromatic doublet was not resolved; mass spectrum, m/z (rel
intensity) 752 (M, 4), 750 (M, 4), 382 (34), 197 (100), 91 (85); Anal.
Calcd for C42H43BrN2O6: C, 67.18; H, 5.78. Found: C, 67.23; H, 5.76.
(()-(1R*,3aS*,4R*,7aR*,8R*)-3a-[(Benzyloxy)methyl]tetrahydro-
8-(2-methoxy-2,2-diphenylethyl)-2-methyl-1,4-methanoisoindoline-
3,6(5H)-dione (82). To a solution of 4.15 g (6.97 mmol) of ketal 81
in 750 mL of acetone cooled in an ice bath was added 750 mg of
p-toluenesulfonic acid monohydrate. The mixture was stirred for 8 h,
the cold bath was then removed, and the mixture was stirred at room
temperature for 16 h. The mixture was then basified with solid sodium
bicarbonate, diluted with 100 mL of saturated aqueous sodium
bicarbonate, and concentrated in vacuo. The residue was partitioned
between 100 mL of EtOAc and 200 mL of water. The organic layer
was separated, and the aqueous phase was extracted with two 100-mL
portions of EtOAc. The combined organic layers were washed with
three 100-mL portions of saturated aqueous sodium bicarbonate and
three 100-mL portions of brine, dried (MgSO4), filtered, and concen-
trated in vacuo. The oily residue was diluted with 125 mL of ether
followed by 50 mL of hexane and cooled to 0 °C. The resulting solid
was collected by filtration and washed with ether-hexane (1:1) to
provide 3.35 g (94%) of ketone 82 as a white solid: mp (CH2Cl2-
ether) 165-167 °C; IR (KBr) 1703 cm-1; 1H NMR (300 MHz, CDCl3)
δ 1.45 (broad, 1H), 1.91-2.09 (m, 3H), 2.28-2.54 (m, 5H), 2.96 (s,
3H), 3.00 (s, 3H), 3.12 (t, J ) 2.0 Hz, 1H), 3.68 (d, J ) 10.4 Hz, 1H),
3.92 (d, J ) 10.4 Hz, 1H), 4.44 (d, J ) 12.2 Hz, 1H), 4.51 (d, J )
12.2 Hz, 1H) 7.15-7.30 (m, 15H); 13C NMR (75.5 MHz, CDCl3) δ
31.1 (q), 38.0 (t), 39.2 (t), 41.3 (d), 42.7 (t), 44.9 (d), 49.1 (d), 50.6
(q), 54.7 (s), 66.6 (t), 67.7 (d), 73.6 (t), 82.3 (s), 126.9 (d), 127.0 (d),
127.1 (d), 127.4 (d), 127.5 (d), 128.0 (d), 128.2 (d), 137.7 (s), 143.8
(s), 144.3 (s), 175.7 (s), 208.8 (s), two aromatic doublets were not
resolved; exact mass calcd for C33H35NO4 m/z 509.2565, found m/z
509.2535. Anal. Calcd for C33H35NO4: C, 77.76; H, 6.93. Found:
C, 77.69; H, 6.96.
(()-(1R*,3aS*,4R*,7aS*,8R*)-3a-[(Benzyloxy)methyl]-2′-bromo-
3a,4,5,7a-tetrahydro-8-(2-methoxy-2,2-diphenylethyl)-6-(meth-
oxymethoxy)-N-(methoxymethyl)-2-methyl-3-oxo-1,4-methanoisoin-
doline-7-carboxanilide (85). To a solution of 212 mg (0.28 mmol)
of amide 84 in 5 mL of DMSO was added 56 mg (1.39 mmol) of a
60% dispersion of NaH in mineral oil. The mixture was stirred for 5
min, and then 89 mg (1.12 mmol) of chloromethyl methyl ether was
added. The mixture was stirred for 1 h, and an additional 100 mg (2.5
mmol) of 60% NaH and 183 mg (2.29 mmol) of chloromethyl methyl
ether were added. The mixture was stirred for 30 min, and an additional
54 mg of chloromethyl methyl ether was added. The mixture was
stirred for 1 h, then poured into 40 mL of saturated aqueous sodium
bicarbonate, and extracted with three 30-mL portions of EtOAc. The
combined organic layers were washed with four 40-mL portions of
brine, dried (MgSO4), filtered, and concentrated in vacuo. The residue
was chromatographed over 12 g of silica gel (first CH2Cl2; then
EtOAc-hexane, 3:7 f 1:1 f 6:4 f 7:3) to provide 209 mg (84%) of
85 as an oil which when concentrated from ether-hexane and afforded
(()-(1R*,3aS*,4R*,7aS*,8R*)-3a-[(Benzyloxy)methyl]-2′-bromo-
3a,4,5,7a-tetrahydro-6-hydroxy-8-(2-methoxy-2,2-diphenylethyl)-2-
methyl-3-oxo-1,4-methanoisoindoline-7-carboxanilide (83). To a
suspension of 1.24 g (31.00 mmol) of 60% NaH in mineral oil (washed
with three 10-mL portions of hexane) and 145 mg (3.62 mmol) of KH
in 50 mL of THF was added 3.52 g (6.94 mmol) of ketone 82 followed
by 7.2 g (36.36 mmol) of 2-bromophenylisocyanate.42 The mixture
was warmed under reflux for 5 h and then an additional 1.0 g (5.05
mmol) of 2-bromophenylisocyanate was added, and the mixture
continued to reflux for 1.5 h. The mixture was cooled to -10 °C, and
the excess NaH-KH was cautiously destroyed with EtOH. The mixture
was stirred for 20 min and then partitioned between 250 mL of ice
cold saturated aqueous ammonium chloride and 150 mL of EtOAc.
The organic layer was separated, and the aqueous phase was extracted
with three 80-mL portions of EtOAc. The combined organic layers
were washed with three 80-mL portions of brine, dried (MgSO4),
filtered, and concentrated in vacuo. The residue was chromatographed
over 65 g of silica gel (first CH2Cl2; then MeOH-CH2Cl2, 2:98) and
then recrystallized from ether-hexane to afford, in two crops, 3.42 g
(70%) of â-ketoanilide 83 as a white solid (mp 174-177 °C). The
mother liquor was chromatographed over 7 g of silica gel (first CH2-
Cl2; then MeOH-CH2Cl2, 2:98), and the material from the column
was recrystallized from ether-hexane to afford an additional 491 mg
1
85 as a white powder: IR (KBr) 1701 cm-1; H NMR (300 MHz,
CDCl3): At 303 K, this material gave a complicated spectrum due to
the presence of geometrical isomers; mass spectrum, m/z (rel intensity)
796 (M, 1), 794 (M, 1), 548 (9), 197 (60), 91 (100), 45 (70); exact
mass calcd for C44H47BrN2O7 m/z 796.2548 and 794.2568, found m/z
796.2558 and 794.2552.

6238 J. Am. Chem. Soc., Vol. 119, No. 27, 1997
Atarashi et al.
(()-(1′R*,3S*,3′aS*,4′R*,7′aS*,8′R*)-3′a-[(Benzyloxymethyl]-
3′a,7′a-dihydro-8′-(2-methoxy-2,2-diphenylethyl)-1-(methoxymethyl)-
2′-methylspiro[indoline-3,7′(4′H)-[1,4]methanoisoindoline]-2,3′,6′-
(5′H)-trione (86) and (()-(1′R*,3R*,3′aS*,4′R*,7′aS*,8′R*)-3′a-
[(Benzyloxymethyl]-3′a,7′a-dihydro-8′-(2-methoxy-2,2-diphenylethyl)-
1-(methoxymethyl)-2′-methylspiro[indoline-3,7′(4′H)-[1,4]meth-
anoisoindoline]-2,3′,6′(5′H)-trione (87). A. From MOM Ether 85.
To a solution of 185 mg (0.23 mmol) of 85 in 20 mL of benzene heated
at reflux was added a mixture of 129 mg (0.44 mmol) of tri-n-butyltin
hydride and 4 mg of AIBN in 6 mL of benzene over a 19-h period
using a syringe pump. The mixture was then concentrated in vacuo.
The residue was diluted with 25 mL of ether, and DBU was added
dropwise until the mixture no longer became cloudy. The mixture was
stirred for 18 h, then filtered through Celite, and concentrated in vacuo
to near dryness. The residue was twice chromatographed over 12 g of
silica gel (ether for the first column, and for the second column, CH2-
Cl2, then EtOAc-CH2Cl2, 5:95, then 10:90) to provide 129 mg (83%)
CH2Cl2 was added 100 mg of p-toluenesulfonic acid monohydrate. The
mixture was stirred for 48 h and then passed through a 6 g pad of
silica gel (EtOAc-CH2Cl2, 1:9). The eluent was concentrated in vacuo,
and the residue was recrystallized from ether-hexane to afford 158
mg (88%) of olefin 90 as a white solid: mp 178-179 °C; IR (CCl4)
1
1731, 1708 cm-1; H NMR (300 MHz, CDCl3) δ 2.15 (broad s, 1H),
2.41 (dd, J ) 19.2, 2.8 Hz, 1H), 2.71 (dd, J ) 19.3, 3.3 Hz, 1H), 2.81
(broad s, 1H), 3.03 (s, 3H), 3.31 (s, 3H), 3.75 (broad d, J ) 9.4 Hz,
1H), 4.06 (d, J ) 10.4 Hz, 1H), 4.13 (broad s, 1H), 4.32 (d, J ) 10.4
Hz, 1H), 4.70 and 4.77 (ABq, J ) 11.9 Hz, 2H), 5.06 (s, 2H), 5.86 (d,
J ) 9.4 Hz, 1H), 6.94-7.00 (m, 2H), 7.18-7.46 (m, 17H); 13C NMR
(75.5 MHz, CDCl3) δ 30.9 (q), 41.1 (d), 42.2 (t), 49.7 (d), 55.1 (d),
56.1 (q), 57.7 (s), 63.5 (s), 65.6 (d), 66.3 (t), 71.2 (t), 74.1 (t), 109.6
(d), 123.1 (d), 124.5 (d), 127.0 (d), 127.1 (d), 127.43 (d), 127.48 (d),
127.7 (d), 127.9 (d), 128.1 (d), 128.2 (d), 128.4 (d), 128.9 (d), 129.6
(d), 130.5 (s), 137.5 (s), 139.0 (s), 141.7 (s), 141.8 (s), 144.6 (s), 173.3
(s), 174.3 (s), 204.2 (s); mass spectrum, m/z (rel intensity) 638 (M,
48), 498 (83), 252 (100), 91 (92); exact mass calcd for C41H38N2O5
m/z 638.2780, found m/z 638.2795. Anal. Calcd for C41H38N2O5: C,
77.08; H, 6.00. Found: C, 77.15; H, 6.08. The structure of 90 was
confirmed by X-ray crystallography.
1
of a 1.2:1 mixture of oxindoles 86 and 87, respectively, by H NMR.
Typically, the mixture was used in the next step without separation,
since the isomers were easier to separate at that point. Analytically
pure samples of oxindoles 86 and 87 could be obtained in diminished
yield, 35% and 20% respectively, after repeated chromatography
(EtOAc-CH2Cl2) and a single recrystallization from ether-hexane.
Pure oxindole 86 was a white solid: mp (ether-hexane) 191-193 °C;
IR (KBr) 1707, 1611 cm-1; 1H NMR (300 MHz, CDCl3) δ 1.33 (broad
s, 1H), 2.12 (dd, J ) 19.2, 2.9 Hz, 1H), 2.25 (dd, J ) 19.6, 3.0 Hz,
1H), 2.33 (dd, J ) 14.0, 9.4 Hz, 1H), 2.68 (broad s, 1H), 2.74 (dd, J
) 13.9, 1.0 Hz, 1H), 2.99 (broad d, J ) 7.9 Hz, 1H), 3.10 (s, 3H),
3.15 (s, 3H), 3.34 (s, 3H), 3.86 (d, J ) 10.4 Hz, 1H), 4.04 (broad s,
1H), 4.10 (d, J ) 10.4 Hz, 1H), 4.60 (d, J ) 12.0 Hz, 1H), 4.68 (d, J
) 12.0 Hz, 1H), 5.08 and 5.12 (ABq, J ) 10.9 Hz, 2H), 6.90-7.08
(m, 3H), 7.13-7.41 (m, 16H); 13C NMR (62.9 MHz, CDCl3) δ 31.7
(q), 38.6 (t), 39.3 (d), 42.6 (t), 44.5 (d), 50.4 (q), 54.6 (d), 56.1 (q),
57.9 (s), 63.7 (s), 66.1 (d), 66.4 (t), 71.3 (t), 74.0 (t), 81.6 (s), 109.7
(d), 123.1 (d), 124.6 (d), 126.6 (d), 126.8 (d), 126.9 (d), 127.1 (d),
127.6 (d), 127.9 (d), 128.0 (d), 128.26 (d), 128.28 (d), 128.8 (d), 130.6
(s), 137.7 (s), 141.8 (s), 144.7 (s), 145.2 (s), 173.8 (s), 174.4 (s), 205.1
(s); mass spectrum, m/z (rel intensity) 670 (M, 18), 548 (15), 197 (17),
43 (100); exact mass calcd for C42H42N2O6 m/z 670.3042, found m/z
670.3050. Anal. Calcd for C42H42N2O6: C, 75.19; H, 6.31. Found:
C, 75.26; H, 6.32. Pure oxindole 87 was also a white solid: mp
(recrystallized from diethyl ether-hexane) 158-161 °C; IR (KBr) 1730,
(()-(1′R*,3R*,3′aS*,4′R*,7′aS*,8′R*)-3′a-[(Benzyloxy)methyl]-8′-
(2,2-diphenylvinyl)-3′a,7′a-dihydro-1-(methoxymethyl)-2′-methylspiro-
[indoline-3,7′(4′H)-[1,4]methanoisoindoline]-2,3′,6′(5′H)-trione (91).
Method A. To a solution of 90 mg (0.13 mmol) of 87 in 15 mL of
CH2Cl2 was added 86 mg of p-toluenesulfonic acid monohydrate. The
mixture was stirred for 48 h and then passed through a 6 g pad of
silica gel (EtOAc-CH2Cl2, 15:85). The eluent was concentrated, and
the residue was recrystallized from MeOH. The solid was collected
by filtration and washed with ether to provide 84 mg (98%) of olefin
91 as a white solid: mp (after recrystallization from MeOH) 187-188
1
°C; IR (CCl4) 1733, 1715 cm-1; H NMR (300 MHz, CDCl3) δ 2.35
(broad s, 1H), 2.69 (broad s, 1H), 2.72 (dd partially obscured, J )
3.65 Hz, 1H), 2.99 (s, 3H), 3.00 (m, 2H), 3.14 (s, 3H), 3.55 (broad s,
1H), 4.41 and 4.45 (ABq, J ) 9.1 Hz, 2H), 4.66 and 4.74 (ABq, J )
12.1 Hz, 2H), 4.93 and 5.02 (ABq, J ) 11.0 Hz, 2H), 5.78 (d, J ) 7.4
Hz, 1H), 5.96 (d, J ) 8.7 Hz, 1H), 6.83 (t, J ) 7.5 Hz, 1H), 6.97 (d,
J ) 7.8 Hz, 1H), 7.22-7.59 (m, 16H); 13C NMR (75.5 MHz, CDCl3)
δ 30.6 (q), 39.9 (d), 42.6 (t), 49.5 (d), 54.6 (d), 55.8 (q), 58.2 (s), 63.6
(s), 65.6 (t), 65.8 (d), 71.1 (t), 73.0 (t), 109.9 (d), 123.1 (d), 123.8 (d),
126.2 (d), 126.4 (d), 127.0 (d), 127.2 (d), 127.4 (s), 127.7 (d), 127.9
(d), 128.3 (d), 129.0 (d), 129.1 (d), 138.7 (s), 139.3 (s), 139.8 (s), 142.4
(s), 144.7 (s), 174.0 (s), 174.7 (s), 204.5 (s), two aromatic doublets
were not resolved; mass spectrum, m/z 638 (M, 16), 607 (49), 91 (100);
exact mass calcd for C41H38N2O5 m/z 638.2780, found m/z 638.2774.
Method B: Via Isomerization of Oxindole 90. A mixture of 58 mg
(0.09 mmol) of oxindole 90 and 1 mg of potassium cyanide in 5 mL
of DMF was warmed to 50 °C for 68 h. The mixture was then cooled,
diluted with 20 mL of saturated aqueous ammonium chloride, and
extracted with three 20-mL portions of EtOAc. The combined organic
layers were washed with four 30-mL portions of brine, dried (MgSO4),
filtered, and concentrated in vacuo to afford 55 mg of a 85:15 mixture
1
1709 cm-1; H NMR (300 MHz, CDCl3) δ 1.99 (broad s, 1H), 2.22
(broad, 1H), 2.51-2.60 (m, 4H), 2.76 (dd, J ) 18.6, 2.0 Hz, 1H), 3.03
(s, 6H), 3.14 (s, 3H), 3.38 (broad s, 1H), 4.28 and 4.34 (ABq, J ) 9.1
Hz, 2H), 4.61 and 4.70 (ABq, J ) 12.2 Hz, 2H), 4.94 and 5.01 (AB q,
J ) 10.9 Hz, 2H), 6.10 (d, J ) 9.5 Hz, 1H), 6.86 (td, J ) 7.6, 0.8 Hz,
1H), 6.97 (d, J ) 7.6 Hz, 1H), 7.19-7.40 (m, 16H); 13C NMR (75.5
MHz, CDCl3) δ 31.4 (q), 38.6 (t), 40.3 (d), 42.7 (t), 45.2 (d), 51.1 (q),
54.7 (d), 55.9 (q), 57.9 (s), 63.7 (s), 65.7 (t), 66.0 (d), 71.1 (t), 73.0 (t),
83.1 (s), 110.0 (d), 123.0 (d), 123.9 (d), 126.7 (d), 127.0 (d), 127.2
(d), 127.4 (d), 127.5 (d), 127.7 (s), 127.9 (d), 128.3 (d), 128.3 (d),
129.1 (d), 138.9 (s), 142.6 (s), 143.0 (s), 144.7 (s), 174.3 (s), 174.9
(s), 205.4 (s), one aromatic doublet was not resolved; mass spectrum,
m/z (rel intensity) 670 (M, 50), 548 (61), 197 (100), 91 (80); exact
mass calcd for C42H42N2O6 m/z 670.3042, found m/z 670.3025. Anal.
Calcd for C42H42N2O6: C, 75.19; H, 6.31. Found: C, 74.93; H, 6.68.
B. From 88: To a solution of 111 mg (0.13 mmol) of 88 in 7 mL of
benzene heated at reflux was added a mixture of 76 mg (0.26 mmol)
of tri-n-butyltin hydride and 3 mg of AIBN in 3 mL of benzene over
a 21-h period using a syringe pump. The mixture was then concentrated
in vacuo, and the residue was partitioned between 10 mL of acetonitrile
and 10 mL of hexane. The acetonitrile layer was separated, washed
with 12 5-mL portions of hexane, and concentrated in vacuo to provide
90 mg of a 1:1 mixture of crude oxindoles 86 and 87, respectively, by
1H NMR. The mixture of isomers was chromatographed over 3 g of
silica gel (dichloromethane; then EtOAc-dichloromethane, 1:9) to
provide 69 mg (78%) of a mixture of oxindoles 86 and 87.
1
of oxindoles 91 and 90, respectively, by H NMR. The mixture was
recrystallized from MeOH to afford 49 mg (84%) of pure oxindole 91
whose 1H NMR spectrum was identical to material obtained via method
A.
(()-(1R*,3aS*,4R*,7aS*,8R*)-N-Acetyl-3a-[(benzyloxy)methyl]-
2′-bromo-3a,4,5,7a-tetrahydro-6-hydroxy-8-(2-methoxy-2,2-diphe-
nylethyl)-2-methyl-3-oxo-1,4-methanoisoindoline-7-carboxanilide
Acetate (Ester) (93). To a solution of 2.6 g (3.67 mmol) of
â-ketoanilide 83, 3.7 g (36.6 mmol) of triethylamine, and 20 mg of
4-(dimethylamino)pyridine cooled to 5 °C was added a solution of 1.9
g (18.6 mmol) of acetic anhydride in 5 mL of DMF over a 2-min period.
The mixture was stirred at room temperature for 30 min and then poured
onto a slurry of crushed ice in 100 mL of saturated aqueous sodium
bicarbonate and 300 mL of water. The white solid was collected and
washed with water. The solid was dried to afford 2.8 g (98%) of 93
which was used without further purification: IR (film) 1760, 1698 cm-1
.
(()-(1′R*,3S*,3′aS*,4′R*,7′aS*,8′R*)-3′a-[(Benzyloxy)methyl]-8′-
(2,2-diphenyl vinyl)-3′a,7′a-dihydro-1-(methoxymethyl)-2′-methyl-
spiro[indoline-3,7′(4′H)-[1,4]methanoisoindoline]-2,3′,6′(5′H)-tri-
one (90). To a solution of 189 mg (0.282 mmol) of 86 in 15 mL of
The 1H NMR spectrum was complex due to the presence of geometrical
isomers. 13C NMR (75.5 MHz, DMSO-d6, 373 K) δ 20.7 (q), 25.0
(q), 30.9 (q), 33.2 (t), 36.8 (t), 39.3 (d), 50.3 (d), 54.9 (s), 65.2 (t),
70.8 (d), 73.1 (t), 82.1 (s), 120.6 (s), 123.5 (s), 126.6 (d), 126.7 (d),

Free Radical Cyclizations in Alkaloid Total Synthesis
J. Am. Chem. Soc., Vol. 119, No. 27, 1997 6239
126.9 (d), 127.0 (d), 127.2 (d), 127.34 (d), 127.36 (d), 127.8 (d), 128.0
(d), 128.1 (d), 129.0 (d), 130.8 (d), 131.0 (d), 132.6 (s), 133.5 (d),
137.6 (s), 139.1 (s), 145.0 (s), 145.2 (s), 167.5 (s), 167.6 (s), 171.5 (s),
174.1 (s), one aliphatic quartet was not resolved. During the course
of acquiring 1H and 13C NMR data above 363 K minor peaks began to
appear in the spectra due to decomposition; mass spectra, m/z (rel
intensity) 535 (10), 533 (10), 197 (30) 91 (100). No parent ion was
observed.
128.1 (d), 128.2 (d), 137.9 (s), 139.6 (s), 144.6 (s), 144.8 (s), 170.5
(s), 170.9 (s), 174.8 (s), 178.4 (s), one aliphatic doublet was not resolved
and three too many aromatic doublets are present; mass spectra, m/z
(rel intensity) 712 (M, 0.6), 574 (10), 197 (100), 91 (89); exact mass
calcd for C44H44N2O7 m/z 712.3148, found m/z 712.3131. Purification
of oxindole 94: All of the remaining fractions and filtrates containing
oxindole 94 were combined and chromatographed over 70 g of silica
gel (CH2Cl2 to load the sample; then EtOAc-hexane, 20:80, then 25:
75) to afford 127 mg (9%) of pure oxindole 94 as a white powder
when concentrated from ether-hexane: mp 104-113 °C; IR (neat)
(()-(1′R*,3S*,3′aS*,4′R*,6′R*,7′aS*,8′R*)-1-Acetyl-3′a-[(ben-
zyloxy)methyl]-3′a, 5′,6′,7′a-tetrahydro-6′-hydroxy-8′-(2-methoxy-
2,2-diphenylethyl)-2′-methylspiro[indoline-3,7′(4′H)-[1,4]meth-
anoisoindoline]-2,3′-dione Acetate (Ester) (94), (()-(1′R*,3S*,3′aS*,-
4′R*,-6′S*,7′aS*,8′R*)-1-Acetyl-3′a-[(benzyloxy)methyl]3′a,5′,6′,7′a-
tetrahydro-6′-hydroxy-8′-(2-methoxy-2,2-diphenylethyl)-2′-methyl-
spiro[indoline-3,7′(4′H)-[1,4]methanoisoindoline]-2,3′-dione Acetate
(Ester) (95) and (()-(1′R*,3R*,3′aS*,4′R*,6′R*,7′aS*,8′R*)-1-Acetyl-
3′a-[(benzyloxy)methyl]-3′a,5′,6′,7′a-tetrahydro-6′-hydroxy-8′-(2-
methoxy-2,2-diphenylethyl)-2′-methylspiro[indoline-3,7′(4′H)-[1,4]-
methanoisoindoline ]-2,3′-dione Acetate (Ester) (96). A solution of
1.5 g (1.89 mmol) of bromide 93, 1.29 g (4.46 mmol) of tri-n-butyltin
hydride, and 12 mg of AIBN in 500 mL of benzene cooled to 10 °C
was irradiated with a mercury arc lamp for 1.5 h. An additional 541
mg (1.85 mmol) of tri-n-butyltin hydride was then added, and the
mixture was irradiated for an additional 40 min. The mixture was then
concentrated in vacuo, and the residue was partitioned between 100
mL of acetonitrile and 100 mL of hexane. The acetonitrile layer was
separated, and the hexane layer was extracted with two 50-mL portions
of acetonitrile. The combined acetonitrile layers were washed with
eight 100-mL portions of hexane and concentrated in vacuo to provide
1.4 g of crude oxindoles which by 1H NMR indicated that the mixture
contained approximately 41% of oxindole 96. The residue was
chromatographed over 60 g of silica gel (CH2Cl2 to load the sample;
then EtOAc-CH2Cl2, 5:95) to afford, in the following order, partially
purified oxindoles 94, 95, and 96. Purification of oxindole 96: All
of the fractions containing oxindole 96 were combined and recrystallized
from ether-hexane to afford 516 mg (38%) of pure oxindole 96 as a
1
1748, 1707 cm-1; H NMR (300 MHz, CDCl3) δ 1.21 (broad d, J )
5.6 Hz, 1H), 1.56 (dd, J ) 14.4, 7.4 Hz, 1H), 1.83 (s, 3H), 2.13 (m,
1H), 2.21 (dd, J ) 14.1, 8.5 Hz, 1H), 2.48 (t, J ) 1.7 Hz, 1H), 2.59
(dd, J ) 14.3, 2.5 Hz, 1H), 2.67 (s, 3H), 2.99 (s, 3H), 3.16 (s, 3H),
3.37 (dd, J ) 2.7, 1.8 Hz, 1H), 3.60 (broad d, J ) 8.4 Hz, 1H), 3.82
(d, J ) 10.5 Hz, 1H), 4.13 (d, J ) 10.5 Hz, 1H), 4.65 and 4.72 (AB q,
J ) 11.8 Hz, 2H), 5.28 (dd, J ) 9.2, 7.5 Hz, 1H), 6.98 (td, J ) 7.6,
1.0 Hz, 1H), 7.15-7.40 (m, 16H), 7.50 (dd, J ) 7.7, 0.8 Hz, 1H), 8.09
(dd, J ) 8.1, 0.6 Hz, 1H); 13C NMR (75.5 MHz, CDCl3) δ 20.6 (q),
26.7 (q), 31.8 (q), 33.2 (t), 38.5 (t), 38.7 (d), 44.8 (d), 50.3 (q), 53.8
(s), 55.9 (d), 58.7 (s), 65.7 (d), 67.8 (t), 71.8 (d), 73.9 (t), 81.9 (s),
115.3 (d), 124.1 (d), 125.6 (d), 126.7 (d), 126.8 (d), 126.9 (d), 127.1
(d), 127.5 (d), 127.7 (d), 127.8 (d), 128.0 (d), 128.2 (d), 128.4 (d),
133.5 (s), 137.7 (s), 138.1 (s), 144.9 (s), 145.1 (s), 169.5 (s), 170.5 (s),
174.5 (s), 175.1 (s); mass spectrum, m/z (rel intensity) 712 (M, 1), 515
(13), 197 (85), 91 (100), 43 (31); exact mass calcd for C44H44N2O7 m/z
712.3148, found m/z 712.3167.
(()-(1′R*,3R*,3′aS*,4′R*,6′R*,7′aS*,8′R*)-1-Acetyl-3′a-[(ben-
zyloxy)methyl]-8′-(2,2-diphenylvinyl)-3′a,5′,6′,7′a-tetrahydro-6′-hy-
droxy-2′-methylspiro[indoline-3,7′(4′H)-[1,4]methanoisoindoline]-
2,3′-dione Acetate (Ester) (99). To a solution of 68 mg (0.09 mmol)
of oxindole 96 in 20 mL of CH2Cl2 was added 20 mg of p-
toluenesulfonic acid monohydrate. The mixture was warmed to reflux
for 20 min, then cooled, and concentrated in vacuo. The residue was
chromatographed over 7 g of silica gel (CH2Cl2 to load the sample;
then EtOAc-hexane, 25:75). The material from the column was
recrystallized from EtOAc-hexane to afford 54 mg (83%) of olefin
1
white solid: mp 132-135 °C; IR (film) 1759, 1738, 1711 cm-1; H
99 as a crystalline solid: mp 234-235 °C, IR (film) 1755, 1711 cm-1
;
NMR (300 MHz, CDCl3) δ 1.68 (s, 3H), 1.74 (broad s, 1H), 1.94 (broad
d, J ) 15.1 Hz, 1H), 2.26 (broad s, 1H), 2.45 (s, 3H), 2.56 (m, 4H),
2.92 (broad s, 1H), 2.96 (s, 3H), 3.07 (s, 3H), 3.90 (d, J ) 9.2 Hz,
1H), 4.06 (d, J ) 9.2 Hz, 1H), 4.59 (d, J ) 12.3 Hz, 1H), 4.67 (d, J
) 12.3 Hz, 1H), 5.29 (d, J ) 7.9 Hz, 1H), 6.39 (d, J ) 7.5 Hz, 1H),
6.92 (t, J ) 7.3 Hz, 1H), 7.20-7.47 (m, 16H), 8.18 (d, J ) 7.9 Hz,
1H); 13C NMR (75.5 MHz, CDCl3) δ 20.5 (q), 26.0 (q), 31.4 (q), 31.8
(t), 37.8 (t), 39.5 (d), 42.3 (d), 51.0 (q), 52.8 (d), 54.3 (s), 56.7 (s),
65.8 (d), 65.8 (t), 66.2 (d), 72.9 (t), 82.9 (s), 116.5 (d), 123.7 (d), 125.2
(d), 126.3 (s), 126.7 (d), 126.8 (d), 127.0 (d), 127.1 (d), 127.3 (d),
128.0 (d), 128.1 (d), 128.2 (d), 128.5 (d), 138.7 (s), 139.9 (s), 143.5
(s), 145.3 (s), 169.8 (s), 170.7 (s), 175.2 (s), 179.1 (s), one aromatic
doublet was not resolved; mass spectrum, m/z (rel intensity) 712 (M,
0.4), 589 (4), 531 (8), 197 (78), 91 (100), 43 (57); exact mass calcd
for C44H44N2O7 m/z 712.3148, found m/z 712.3115. Purification of
oxindole 95: The filtrate from the recrystallization of oxindole 96 was
combined with appropriate mixed fractions from the first column and
chromatographed over 30 g of silica gel (CH2Cl2 to load the sample;
then EtOAc-hexane, 20:80 f 30:70). Fractions containing oxindole
96 were recrystallized from ether-hexane to afford an additional 63
mg (4%) of pure oxindole 96. The mother liquor was enriched with
oxindole 95. This material was combined with fractions containing
oxindole 95 and twice recrystallized from ether-hexane to afford 99
mg (7%) of oxindole 95 as a white solid: mp 211-213 °C; IR (KBr)
1H NMR (300 MHz, CDCl3) δ 1.51 (s, 3H), 2.03 (m, partially obscured
by broad s at d 2.05, 1H), 2.05 (broad s, 1H), 2.31 (broad s, 1H), 2.45
(s, 3H), 2.57 (m, 1H), 2.90 (s, 3H), 3.15 (dd, J ) 2.6, 1.8 Hz, 1H),
3.36 (broad dd, J ) 9.2, 2.5 Hz, 1H), 4.04 (d, J ) 9.1 Hz, 1H), 4.14
(d, J ) 9.1 Hz, 1H), 4.60 (d, J ) 12.2 Hz, 1H), 4.68 (d, J ) 12.2 Hz,
1H), 5.29 (d, J ) 7.9 Hz, 1H), 5.86 (d, J ) 9.3 Hz, 1H), 6.16 (dd, J
) 7.6, 0.7 Hz, 1H), 6.91 (td, J ) 7.6, 1.0 Hz, 1H), 7.18-7.40 (m,
13H), 7.48-7.62 (m, 3H), 8.18 (dd, J ) 8.2, 0.7 Hz, 1H); 13C NMR
(75.5 MHz, CDCl3) δ 20.5 (q), 26.0 (q), 30.6 (q), 31.8 (t), 39.3 (d),
46.6 (d), 52.7 (d), 53.8 (s), 57.2 (s), 65.5 (t), 65.7 (d), 65.7 (d), 72.9
(t), 116.5 (d), 123.8 (d), 125.2 (d), 126.4 (s), 126.5 (d), 127.1 (d), 127.38
(d), 127.40 (d), 127.6 (d), 128.0 (d), 128.2 (d), 128.6 (d), 128.8 (d),
129.6 (d), 138.6 (s), 139.7 (s), 140.1 (s), 140.2 (s), 143.9 (s), 169.4
(s), 170.6 (s), 175.1 (s), 178.8 (s), one aromatic doublet was not
resolved; mass spectrum, m/z (rel intensity) 680 (M, 57), 193 (79),
167 (24), 91 (100), 43 (19). Anal. Calcd for C43H40N2O6: C, 75.85;
H, 5.93. Found: C, 75.76; H, 5.95.
(()-(1′R*,3R*,3′aS*,4′R*,6′R*,7′aS*,8′R*)-3′a-[(Benzyloxy)methyl]-
8′-(2,2-diphenylvinyl)-3′a,5′,6′,7′a-tetrahydro-6′-hydroxy-2′-meth-
ylspiro[indoline-3,7′(4 ′H)-[1,4]methanoisoindoline]-2,3′-dione Ac-
etate (Ester) (100). To a solution of 216 mg (0.303 mmol) of oxindole
99 in 30 mL of CH2Cl2 was added 102 mg of p-toluenesulfonic acid.
The reaction was monitored by thin-layer chromatography (EtOAc-
hexane, 1:1), and once starting material was no longer present, 1 mL
of MeOH was added to facilitate cleavage of the N-acetyl group. The
mixture was stirred for an additional 24 h and then diluted with 50 mL
1
1754, 1710 cm-1; H NMR (300 MHz, CDCl3) δ 1.28 (broad s, 1H),
1.45 (s, 3H), 1.70 (broad dd, 2H), 2.18 (dd, J ) 14.2, 8.2 Hz, 1H),
2.28 (t, J ) 1.6 Hz, 1H), 2.58 (dd, J ) 14.2, 2.3 Hz, 1H), 2.72 (s, 3H),
2.84 (broad, 1H), 2.99 (s, 3H), 3.14 (s, 3H), 3.66 (dd, J ) 2.74, 1.8
Hz, 1H), 3.86 (d, J ) 9.8 Hz, 1H), 4.02 (d, J ) 9.8 Hz, 1H), 4.44 (s,
2H), 5.14 (t, J ) 8.4 Hz, 1H), 6.98-7.41 (m, 18H), 8.16 (dd, J ) 8.2,
0.6 Hz, 1H); 13C NMR (75.5 MHz, CDCl3) δ 19.9 (q), 26.7 (q), 29.7
(t), 31.8 (q), 38.0 (t), 39.3 (d), 40.1 (d), 50.3 (q), 54.2 (s), 56.7 (d),
57.0 (s), 65.8 (t), 69.7 (d), 73.1 (t), 81.7 (s), 116.1 (d), 123.3 (d), 126.6
(d), 126.77 (d), 126.79 (d), 126.8 (d), 127.08 (d), 127.10 (d), 127.16
(d), 127.3 (d), 127.7 (d), 127.80 (s), 127.88 (d), 127.9 (d), 128.0 (d),
of saturated aqueous sodium bicarbonate. The CH
2Cl2 layer was
separated and washed with 50 mL of saturated aqueous sodium
bicarbonate and 25 mL of brine, dried (MgSO4), filtered, and
concentrated in vacuo. The residue was recrystallized from EtOAc-
hexane to afford 180 mg (93%) of olefin 100 as a white solid: mp
210-211 °C; IR (KBr) 1729, 1617 cm-1; 1H NMR (300 MHz, CDCl3)
δ 1.47 (s, 3H), 1.98 (dd, J ) 15.0, 2.6 Hz, 1H), 2.10 (broad, 1H), 2.24
(broad, 1H), 2.52 (ddd, J ) 15.6, 8.8, 2.7 Hz, 1H), 2.90 (s, 3H), 3.24

6240 J. Am. Chem. Soc., Vol. 119, No. 27, 1997
Atarashi et al.
(dd, J ) 2.8, 1.8 Hz, 1H) 3.38 (ddd, J ) 9.2, 2.4, 2.4 Hz, 1H), 4.09 (d,
J ) 8.9 Hz, 1H), 4.18 (d, J ) 9.2 Hz, 1H), 4.57 (d, J ) 12.4 Hz, 1H),
4.74 (d, J ) 12.3 Hz, 1H), 5.22 (d, J ) 7.7 Hz, 1H), 5.89 (d, J ) 9.2
Hz, 1H), 6.04 (d, J ) 7.4 Hz, 1H), 6.72 (m, 2H), 7.13 (t, J ) 7.8 Hz,
1H), 7.16-7.61 (m, 15H), the NH proton was not resolved; 13C NMR
(75.5 MHz, CDCl3) δ 20.7 (q), 30.6 (q), 31.7 (t), 39.6 (d), 45.8 (d),
52.5 (d), 53.9 (s), 57.2 (s), 65.6 (d), 66.0 (t), 66.3 (d), 73.1 (t), 110.2
(d), 120.9 (d), 126.1 (d), 126.5 (d), 126.9 (d), 127.3 (d), 127.4 (d),
127.8 (s), 128.0 (d), 128.2 (d), 128.3 (d), 128.8 (d), 129.6 (d), 138.7
(s), 140.1 (s), 140.3 (s), 141.5 (s), 143.6 (s), 169.3 (s), 176.0 (s), 180.7
(s), two aromatic doublets were not resolved; mass spectrum, m/z (rel
intensity) 638 (M, 14), 578 (9), 470 (16), 193 (51), 91 (100), 43 (25);
Anal. Calcd for C41H38N2O5: C, 77.09; H, 5.99. Found: C, 77.31;
H, 6.05.
2 mL of dimethoxyethane was added 7 mL of 6 N aqueous HCl. The
mixture was warmed at 48 °C for 16 h, cooled to room temperature,
and then made basic with sodium bicarbonate. The mixture was diluted
with 20 mL of water and extracted with six 25-mL portions of EtOAc.
The combined organic layers were dried (MgSO4), filtered, and
concentrated in vacuo. The residue was chromatographed over 5 g of
silica gel (CH2Cl2; then THF-CH2Cl2, 2.5:97.5 f 5:95) to afford 8
mg (14%) of cyclic acetal 103 as a white solid: mp 275-310 °C (dec);
1
IR (film) 3223, 1714 cm-1; H NMR (300 MHz, CDCl3) δ 1.74 (d,
1H), 1.90 (m, 1H), 2.10 (m, 2H), 2.88 (s, 3H), 3.15 (m, 2H), 3.89
(broad s, 1H), 3.97 (d, 1H), 4.46 (d, 1H), 4.49 (s, 2H), 4.96 (d, 1H),
5.45 (broad d, 1H), 6.87 (d, 1H), 7.08 (t, 1H), 7.28 (m, 6H), 7.89 (broad
s, 1H), 8.32 (d, 1H); 13C NMR (75.5 MHz, CDCl3) δ 25.2 (t), 26.9
(q), 28.6 (t), 32.4 (t), 40.5 (d), 51.1 (s), 53.9 (s), 69.6 (d), 69.7 (t), 72.8
(t), 88.2 (d), 96.0 (d), 109.0 (d), 122.4 (d), 127.1 (d), 127.3 (d), 127.9
(d), 128.6 (s), 128.6 (d), 129.7 (d), 138.5 (s), 140.1 (s), 177.1 (s), 177.9
(s); mass spectrum, m/z (rel intensity) 446 (M, 5), 355 (100), 240 (25),
91 (92); exact mass calcd for C26H26N2O5 m/z 446.1841, found m/z
446.1847. Continued elution (THF-CH2Cl2, 7.5:92.5, then 25:75)
afforded 46 mg (65%) of hemiacetal 102 as a mixture of diastereo-
(()-(1′R*,3R*,3′aS*,4′R*,6′R*,7′aS*,8′S*)-3′a-[(Benzyloxy)methyl]-
8′-(1,2-epoxy-2,2-diphenylethyl)-3′a,5′,6′,7′a-tetrahydro-6′-hydroxy-
2′-methylspiro[indoline-3,7′(4′H)-[1,4]methanoisoindoline]-2,3′-di-
one Acetate (Ester) and (()-(1′R*,3S*,3′aR,4′S*,-6′S*,7′aR*,8′S*)-
3′a-[(Benzyloxy)methyl]-3′a,5′,6′,7′a-tetrahydro-6′-hydroxy-2′-methyl-
2,3′-dioxospiro[indoline-3,7′(4′H]methanoisoindoline]-8′-car-
boxaldehyde Acetate (Ester) (101). To a solution of 400 mg (0.626
mmol) of olefin 100 in 15 mL of CH2Cl2-MeOH (4:1) chilled to -78
°C was added 60 mL of a 0.094 M solution of ozone gas, prepared by
passing ozone through 75 mL of CH2Cl2-MeOH (4:1) at a rate of 1.1
mmol/min. The addition of ozone was monitored by thin-layer
chromatography (EtOAc-hexane, 1:1) and stopped once starting
material was no longer evident. The mixture was then purged of ozone
by passing nitrogen gas through the solution, and then 3 mL of dimethyl
sulfide was added. The cold bath was removed, and the mixture was
stirred at room temperature for 18 h and then concentrated in vacuo.
The residue was chromatographed over 15 g of silica gel (first CH2-
Cl2; then EtOAc-CH2Cl2, 1:9) to afford 62 mg (15%) of the epoxide
derived from 100 as a white solid: mp 178-183 °C; IR (neat) 1726
cm-1; 1H NMR (300 MHz, CDCl3) δ 1.52 (s, 3H), 2.02 (dd, J ) 15.7,
3.5 Hz, 1H), 2.17 (t, J ) 1.5 Hz, 1H), 2.29 (broad d, J ) 8.4 Hz, 1H),
2.36 (broad s, 1H), 2.55 (ddd, J ) 15.7, 8.1, 2.1 Hz, 1H), 2.82 (s, 3H),
3.19 (d, J ) 8.5 Hz, 1H), 3.38 (dd, J ) 2.6, 2.0 Hz, 1H), 4.17 (s, 2H),
4.54 (d, J ) 12.3 Hz, 1H), 4.69 (d, J ) 12.3 Hz, 1H), 5.18 (d, J ) 7.7
Hz, 1H), 5.59 (d, J ) 7.7 Hz, 1H), 6.63-6.71 (m, 2H), 7.10 (td, J )
7.7, 0.9 Hz, 1H), 7.16-7.40 (m, 10H), 7.54-7.66 (m, 5H), 7.98 (broad
s, 1H); 13C NMR (75.5 MHz, CDCl3) δ 20.7 (q), 30.0 (q), 32.1 (t),
37.5 (d), 46.7 (d), 52.4 (d), 53.5 (s), 57.1 (s), 62.1 (d), 65.5 (d), 65.7
(t), 66.0 (s), 67.8 (d), 73.1 (t), 109.9 (d), 121.3 (d), 126.1 (d), 126.5
(d), 127.0 (d), 127.5 (d), 127.9 (d), 128.0 (d), 128.1 (s), 128.2 (d),
128.4 (d), 128.5 (d), 128.6 (d), 128.8 (d), 137.1 (s), 138.8 (s), 139.0
(s), 140.9 (s), 169.0 (s), 175.7 (s), 179.9 (s); mass spectrum, m/z (rel
intensity) 654 (M, 0.1), 563 (36), 167 (73), 91 (100); exact mass calcd
for C41H38N2O6 m/z 654.2729, found m/z 654.2768. Continued elution
(EtOAc-CH2Cl2, 2:8 f 3:7 f 1:1 f EtOAc) gave partially purified
aldehyde which was recrystallized from EtOAc-hexane to afford 186
mg (61%) of 101. A second recrystallization from EtOAc-hexane
gave 154 mg (51%) of pure 101 as a white solid: mp 235-241 °C; IR
mers: IR (film) 3252 (broad), 1713 cm-1 13C NMR (mixture of
;
diastereomers, 75.5 MHz, CDCl3) δ 23.1 (t), 23.2 (t), 27.4 (q), 27.5
(q), 29.3 (d), 31.1 (d), 46.7 (d), 48.0 (d), 49.7 (d), 50.0 (d), 53.2 (s),
53.5 (s), 58.6 (s), 59.0 (s), 60.9 (d), 61.9 (d), 65.9 (t), 66.1 (t), 68.3
(d), 70.8 (d), 72.6 (t), 89.7 (d), 91.2 (d), 109.3 (d), 109.7 (d), 121.7
(d), 122.0 (d), 126.8 (d), 126.9 (d), 127.2 (d), 127.8 (d), 128.2 (d),
128.4 (d), 128.6 (d), 130.3 (s), 130.4 (s), 139.0 (s), 140.1 (s), 140.4
(s), 177.6 (s), 177.7 (s), 178.5 (s), 178.7 (s), one aliphatic triplet, three
aromatic doublets and one aromatic singlet were not resolved; mass
spectrum, m/z (rel intensity) 446 (M, 1), 355 (68), 91 (100); exact mass
calcd for C26H26N2O5 m/z 446.1841, found m/z 446.1854. The 1H NMR
spectrum of this material is not reported here due to its complexity.
(()-(3R*,3′R*,4aR*,5S*,8S*,8aS*,9S*)-5-[Benzyloxy)methyl]-
1,3,4,4a,5,7, 8,8a-octahydro-7-methylspiro[3,5,8-ethanylylidene-6H-
pyrano[3,4-c]pyridine-10,3′-indoline]-2′,6-dione (107). To a solution
of 46 mg (0.103 mmol) of hemiacetal 102 in 20 mL of CH2Cl2 was
added 0.5 mL of triethylsilane followed by 0.5 mL of trifluoroacetic
acid. The mixture was stirred for 24 h at reflux, cooled to room
temperature, and then diluted with 25 mL of saturated aqueous sodium
bicarbonate. The CH2Cl2 layer was separated, and the aqueous phase
was extracted with three 25-mL portions of CH2Cl2. The combined
organic layers were dried (MgSO4), filtered, and concentrated in vacuo.
The residue was chromatographed over 3 g of silica gel (CH2Cl2; then
THF-CH2Cl2, 15:85). The combined fractions from the column
containing product were concentrated in vacuo, and the residue was
recrystallized from EtOAc-hexane to provide 36 mg (83%) of
tetrahydropyran 107 as a white solid: mp 187-189 °C; IR (CCl4) 3199,
1
1715 cm-1; H NMR (300 MHz, CDCl3) δ 2.16 (m, 3H), 2.34 (broad
s, 1H), 2.76 (s, 3H), 2.81 (dd, J ) 13.6, 2.9 Hz, 1H), 3.82 (d, J ) 1.3
Hz, 1H), 3.89 (broad s, 1H), 3.98 (dd, J ) 11.4, 0.9 Hz, 1H), 4.13 (m,
2H), 4.45 (d, J ) 8.6 Hz, 1H), 4.51 and 4.56 (ABq, J ) 11.8 Hz, 2H),
6.82 (d, J ) 7.7 Hz, 1H), 7.03 (td, J ) 7.6, 1.1 Hz, 1H), 7.17-7.32
1
(KBr) 3426, 1720 cm-1; H NMR (300 MHz, CDCl3) δ 1.77 (s, 3H),
1
(m, 6H), 7.38 (d, J ) 7.6 Hz, 1H), 7.70 (broad s, 1H); H NMR (300
2.06 (dd, J ) 17.5, 3.2 Hz, 1H,), 2.34 (broad s, 1H), 2.70 (m, 2H)
2.71 (s, 3H), 3.61 (broad s, 1H), 3.88 (dd, J ) 2.9, 1.9 Hz, 1H), 4.21
(s, 2H), 4.57 (d, J ) 12.2 Hz, 1H), 4.71 (d, J ) 12.2 Hz, 1H), 5.37 (d,
J ) 7.7 Hz, 1H), 6.84 (d, J ) 7.7 Hz, 1H), 7.01 (t, J ) 7.5 Hz, 1H),
7.15-7.26 (m, 5H), 7.29-7.40 (m, 2H), 8.14 (broad, 1H), 9.74 (d, J
) 1.6 Hz, 1H); 13C NMR (75.5 MHz, CDCl3) δ 20.7 (q), 30.1 (q),
31.9 (t), 32.8 (d), 52.8 (d), 53.8 (s), 57.4 (s), 58.6 (d), 63.3 (d), 65.6
(t), 65.6 (d), 73.1 (t), 110.7 (d), 121.3 (d), 125.8 (d), 127.0 (d), 127.4
(d), 127.7 (s), 128.0 (d), 129.0 (d), 138.6 (s), 141.4 (s), 169.1 (s), 175.2
(s), 180.0 (s), 200.6 (d), the overlapping doublet and triplet at 65.6
ppm was assigned from the DEPT spectrum; mass spectrum, m/z (rel
intensity) 488 (M, 0.4), 397 (55), 91 (100), 43 (41); exact mass calcd
for C28H28N2O6 m/z 488.1947, found m/z 488.1948.
(()-(3R*,3′R*,3aS*,5R*,7R*,9R*,9aS*)-9a-[(Benzyloxy)methyl]-
3,3a,7,8,9,9a-hexahydro-2-methylspiro[3,7-epoxy-5,9-methanooxo-
cino[4,5-c]pyrrole-4(5H),3′-indoline]-1(2H),2′-dione (103) and (()-
(3R*,3′R*,4aR*,5S*,8S*,8aS*,9S*)-5-[(Benzyloxy)methyl]-
1,3,4,4a,5,7,8,8a-octahydro-1-hydroxy-7-methylspiro[3,5,8-
ethanylylidene-6H-pyrano[3,4-c]pyridine-10,3′-indoline]-2′,6-
dione (102). To a solution of 62 mg (0.127 mmol) of aldehyde 101 in
MHz, C6D6) δ 1.42 (broad d, 1H), 1.64-1.74 (m, 2H), 2.26 (s, 3H),
2.30 (broad s, 1H), 2.70 (dd, J ) 14.0, 3.2 Hz, 1H), 3.30 (d, J ) 1.4
Hz, 1H), 3.34 (dd, J ) 11.4, 2.0 Hz, 1H), 3.55 (dd, J ) 11.4, 2.2 Hz,
1H), 3.77 (broad s, 1H), 3.38 (d, J ) 8.2 Hz, 1H), 4.62 (d, J ) 12.6
Hz, 1H), 4.63 (d, J ) 8.1 Hz, 1H), 4.73 (d, J ) 12.0 Hz, 1H), 6.30
(dd, J ) 7.7, 0.8 Hz, 1H), 6.86 (td, J ) 7.5, 1.2 Hz, 1H), 6.96 (td, J
) 7.7, 1.3 Hz, 1H), 7.03-7.08 (m, 2H), 7.16-7.18 (m, 1H), 7.28 (ddd,
J ) 7.4, 1.0, 0.4 Hz, 1H), 7.41-7.44 (m, 2H), the NH was not resolved;
13C NMR (75.5 MHz, CDCl3) δ 23.2 (t), 27.5 (q), 30.9 (d), 43.0 (d),
50.0 (d), 53.7 (s), 58.6 (s), 60.6 (t), 65.8 (d), 66.0 (t), 68.7 (d), 72.7 (t),
109.4 (d), 121.7 (d), 126.8 (d), 127.0 (d), 127.6 (d), 127.8 (d), 128.2
(d), 130.7 (s), 138.9 (s), 140.4 (s), 177.4 (s), 178.6 (s); mass spectrum,
m/z (rel intensity) 430 (M, 0.2), 339 (65), 91 (100); exact mass calcd
for C26H26N2O4 m/z 430.1892, found m/z 430.1886.
(()-(3R*,3′R*,4aR*,5S*,8S*,8aS*,9S*)-1,3,4,4a,5,7,8,8a-Octahy-
dro-5-(hydroxymethyl)-7-methylspiro[3,5,8-ethanylylidene-6H-pyrano-
[3,4-c]pyridine-10,3′-indoline]-2′6-dione (114). To a solution of 31
mg (0.072 mmol) of benzyl ether 107 in 20 mL of CH2Cl2 cooled to
-78 °C was added 216 µL (0.216 mmol) of 1 M boron tribromide in

Free Radical Cyclizations in Alkaloid Total Synthesis
J. Am. Chem. Soc., Vol. 119, No. 27, 1997 6241
CH2Cl2. The mixture was stirred at -78 °C for 30 min and then at
-20 °C for 45 min. The mixture was quenched with 10 mL of saturated
aqueous sodium bicarbonate and stirred for 3 h. The CH2Cl2 layer
was separated, and the aqueous layer was extracted with four 25-mL
portions of CH2Cl2. The combined organic layers were dried (MgSO4),
filtered, and concentrated in vacuo. The solid was triturated with two
2-mL portions of ether-hexane (1:1) to provide 23.5 mg (95%) of
alcohol 114: mp (EtOH-hexane) 303-309 °C; IR (KBr) 3464, 1725,
1696 cm-1; 1H NMR (300 MHz, DMSO-d6) δ 1.90 (broad s, 1H), 1.97-
2.08 (m, 2H), 2.15 (broad d, J ) 7.6 Hz, 1H), 2.52 (dd, J ) 3.1 Hz,
second coupling constant obscured by DMSO peak, 1H), 2.64 (s, 3H),
3.65 (broad s, 1H), 3.92-4.09 (m, 5H), 4.32 (dd, J ) 10.3, 4.0 Hz,
1H), 6.80 (d, J ) 7.6 Hz, 1H), 6.94 (td, J ) 7.6, 0.9 Hz, 1H), 7.20 (td,
intensity) 338 (M, 13), 228 (20), 188 (40), 152 (100); exact mass calcd
for C19H18N2O4 m/z 338.1266, found m/z 338.1270.
(()-(3R*,3′R*,4aR*,5R*,8S*,8aS*,9S*)-1,3,4,4a,5,7,8,8a-Octahy-
dro-7-methyl-5-vinylspiro[3,5,8-ethanylylidene-6H-pyrano[3,4-c]py-
ridine-10,3′-indoline]-2′6-dione [(()-21-Oxogelsemine] (2). To so-
lution of 16.5 mg (0.048 mmol) of aldehyde 115 in 750 µL of THF
was added 500 µL (0.25 mmol) of 0.5 M bis(cyclopentadienyl)-
dimethyltitanium54 in THF. The mixture was warmed under reflux for
24 h and reduced in volume to 100 µL, and an additional 1 mL (0.50
mmol) of a 0.5 M solution of bis(cyclopentadienyl)dimethyltitanium
in THF was added. The mixture was stirred at reflux for an additional
24 h, then cooled to room temperature, diluted with 50 mL of ether,
filtered through Celite, and concentrated in vacuo. The residue was
chromatographed over 1 g of silica gel (CH2Cl2; then EtOAc-CH2-
Cl2, 1:9 f 2:8 f 3:7 f 4:6) to provide 14.3 mg (87%) of
21-oxogelsemine (2): mp 155-159 °C [lit.2 (natural product) 148-
150 °C]; IR (film) 3222, 1715, 1616 cm-1; 1H NMR (300 MHz, CDCl3)
δ 2.06 (t, J ) 1.3 Hz, 1H), 2.17 (ddd, J ) 14.5, 5.6, 2.7 Hz, 1H), 2.22
(m, 1H), 2.49 (dd, J ) 7.4, 3.0 Hz, 1H), 2.78 (s, 3H), 2.97 (dd, J )
14.5, 3.0 Hz, 1H), 3.86 (broad s, 2H), 4.00 (dd, J ) 11.5, 2.0 Hz, 1H),
4.15 (dd, J ) 11.5, 2.2 Hz, 1H), 5.21 (dd, J ) 17.8, 1.1 Hz, 1H), 5.49
(dd, J ) 11.1, 1.1 Hz, 1H), 6.06 (dd, J ) 17.8, 11.1 Hz, 1H), 6.85 (dd,
J ) 7.7, 0.6 Hz, 1H), 7.05 (td, J ) 7.6, 1.1 Hz, 1H), 7.25 (td, J ) 7.8,
1.2 Hz, 1H), 7.39 (d, J ) 7.6 Hz, 1H), 7.86 (s, 1H); 13C NMR (75.5
MHz, CDCl3) δ 23.0 (t), 27.8 (q), 31.6 (d), 42.5 (d), 53.2 (s), 53.7 (d),
60.4 (s), 60.6 (t), 66.1 (d), 68.9 (d), 109.4 (d), 117.2 (t), 121.9 (d),
127.8 (d), 128.4 (d), 130.3 (s), 133.1 (d), 140.2 (d), 176.8 (s), 177.4
(s); mass spectrum, m/z (rel intensity) 336 (M, 100), 254 (13), 122
(69); exact mass calcd for C20H20N2O3 m/z 336.1475, found m/z
336.1485.
J ) 7.6, 0.6 Hz, 1H), 7.38 (d, J ) 7.5 Hz, 1H), 10.35 (broad, 1H); 13
C
NMR (75.5 MHz, DMSO-d6) δ 22.7 (t), 27.2 (q), 30.2 (d), 42.4 (d),
49.6 (d), 53.7 (s), 57.7 (t), 59.3 (s), 60.1 (t), 65.2 (d), 68.3 (d), 109.1
(d), 120.9 (d), 128.2 (d), 128.4 (d), 130.8 (s), 141.6 (s), 178.0 (s), 178.1
(s); mass spectrum, m/z (rel intensity) 340 (M, 24), 322 (79), 304 (24),
240 (100), 214 (57), 132 (70), 98 (77); exact mass calcd for C19H20N2O4
m/z 340.1423, found m/z 340.1423.
(()-(3R*,3′R*,4aR*,5S*,8S*,8aS*,9S*)-4,4a,6,7,8,8a-Hexahydro-
7-methyl-2 ′,6-dioxospiro[3,5,8-ethanylylidene-1H-pyrano[3,4-c]py-
ridine-10,3′-indoline]-5(3H)-carboxaldehyde (115). To 23.5 mg
(0.069 mmol) of alcohol 114 was added 20 mL of CH2Cl2-acetonitrile
(1:1). The suspension was warmed with a heat gun to coax the alcohol
into solution. Once dissolved, the solution was cooled to room
temperature, and 77 mg (0.182 mmol) of the Dess-Martin periodinane52
was added. The mixture was stirred at room temperature for 2 h,
concentrated to 5 mL in vacuo, diluted with 15 mL of ether, filtered
through Celite, and concentrated in vacuo. The residue was chromato-
graphed over 1 g of silica gel (CH2Cl2; then EtOAc-CH2Cl2, 1:9 f
2:8 f 3:7) to provide 23 mg (98%) of aldehyde 115 which was
recrystallized from EtOAc-hexane to afford 17 mg (71%) of pure
Acknowledgment. We thank the National Institutes of
Health for support of this research, Dr. Kurt Loening for help
with nomenclature, and the Campus Chemical Instrumentation
Center at the Ohio State University for technical support.
aldehyde as a white solid: mp 278-280 °C; IR (KBr) 1723, 1696 cm-1
;
1H NMR (300 MHz, CDCl3) δ 2.21 (ddd, J ) 14.9, 5.5, 2.6 Hz, 1H),
2.28 (d, J ) 8.4 Hz, 1H), 2.54 (broad s, 1H), 2.73 (dd, J ) 17.9, 2.8
Hz, 1H), 2.81 (s, 3H), 2.87 (m, 1H), 3.74 (broad s, 1H), 3.99 (m, 2H),
4.15 (dd, J ) 11.6, 2.2 Hz, 1H), 6.93 (d, J ) 7.7 Hz, 1H), 7.09 (td, J
) 7.7. 1.0 Hz, 1H), 7.30 (td, J ) 7.7, 1.1 Hz, 1H), 7.38 (d, J ) 7.6
Hz, 1H), 8.98 (s, 1H), 9.70 (s, 1H); 13C NMR (75.5 MHz, CDCl3) δ
22.6 (t), 27.5 (q), 29.5 (d), 42.2 (d), 51.9 (d), 53.1 (s), 60.3 (t), 67.0
(d), 67.4 (s), 68.7 (d), 110.0 (d), 122.3 (d), 127.6 (d), 128.8 (d), 129.5
(s), 140.0 (s), 173.4 (s), 178.0 (s), 194.0 (d); mass spectrum, m/z (rel
Supporting Information Available: Experimental proce-
dures not provided in the Experimental Section and crystal-
lographic data for 46, 114, and two structures prepared en route
to 20 (87 pages). See any current masthead page for ordering
and Internet access instructions.
JA970089H