A thio-Diels-Alder route to the azocine ring system. Total synthesis of (±)-otonecine

Article abstract of DOI:10.1021/ja973464e

A sulfur-based strategy for synthesis of otonecine is described. Key steps include the thio-Diels-Alder trapping of thioketone 8 by the Danishefsky diene, followed by conversion into the enone 27 and internal Michael addition to afford bicyclic thioaminal 28. Selective C-S bond cleavage was achieved after conversion to alcohol 36 or its derivatives 38, resulting in the azocine ring system. The successful route proceeded from 40b via 49a and sulfoxide elimination to the alkene 50a. The final conversions to otonecine were accomplished via low-temperature osmylation of 56, a crucial OsO₄-mediated oxidation of diol 58 to ketol 60/61, and Burgess elimination to 68/69. Several of the intermediates in late stages of the synthesis exist largely in the bicyclic valence bond tautomer form that is characteristic of the otonecine ring system.

Full text of DOI:10.1021/ja973464e

J. Am. Chem. Soc. 1998, 120, 3613-3622
3613
A Thio-Diels-Alder Route to the Azocine Ring System. Total
Synthesis of (()-Otonecine
Edwin Vedejs,* Rocco J. Galante, and Peter G. Goekjian1
Contribution from the Chemistry Department, UniVersity of Wisconsin, Madison, Wisconsin 53706
ReceiVed October 3, 1997
Abstract: A sulfur-based strategy for synthesis of otonecine is described. Key steps include the thio-Diels-
Alder trapping of thioketone 8 by the Danishefsky diene, followed by conversion into the enone 27 and internal
Michael addition to afford bicyclic thioaminal 28. Selective C-S bond cleavage was achieved after conversion
to alcohol 36 or its derivatives 38, resulting in the azocine ring system. The successful route proceeded from
40b via 49a and sulfoxide elimination to the alkene 50a. The final conversions to otonecine were accomplished
via low-temperature osmylation of 56, a crucial OsO4-mediated oxidation of diol 58 to ketol 60/61, and Burgess
elimination to 68/69. Several of the intermediates in late stages of the synthesis exist largely in the bicyclic
valence bond tautomer form that is characteristic of the otonecine ring system.
We describe synthetic studies targeted at the eight-membered
ring system of the pyrrolizidine alkaloid otonecine (1).2 There
has been some interest in the biological activity of this family
of cytotoxic agents,3 and a total synthesis of otonecine has been
described by Yamada et al.4 Our own investigation was initiated
with the primary goal of exploring the thioketone cycloaddition
approach to dihydrothiopyrans that could serve as precursors
of functionalized azocine derivatives (Scheme 1). The plan
depends on internal 1,4 addition of a tethered amine to an enone
acceptor, as in the conversion of 4 into the thioaminal 3,
Scheme 1
followed by controlled C-S cleavage and elimination.
A
similar approach was described in a preliminary report from
our laboratory, but the desulfurization sequence was not
sufficiently versatile for applications in the otonecine series.5
Another goal of the current study was to explore functional
group compatibility in the highly substituted azocine ring.
Otonecine experiences strong interactions between basic nitrogen
and ketone carbonyl,2b described by the equilibrium between
the valence bond tautomers 1a and 1b. Other unusual transan-
nular effects involving nitrogen were encountered, and some
of these resulted in unexpected synthetic challenges in this
deceptively simple ring system. Satisfactory solutions are
reported for several problems encountered in the conversion of
(1) Current address: Department of Chemistry, Box 9593, Mississippi
State University.
organosulfur precursors into the azocine ring system, as well
as in the eventual elaboration into racemic otonecine.
(2) (a) Zhdanovich, E. S.; Men’shikov, G. P. Zh. Obsch. Khim. 1941,
11, 835. Leonard, N. J. Rec. Chem. Prog. 1956, 17, 243. Bu¨rgi, H. B.;
Dunitz, J. D. Acc. Chem. Res. 1983, 16, 153. (b) Wunderlich, J. A. Acta
Crystallogr. 1967, 23, 846. Culvenor, C. C. J.; O’Donovan, G. M.; Smith,
L. W. Aust. J. Chem. 1967, 20, 801. Bernbaum, G. I. J. Am. Chem. Soc.
1974, 96, 6165. Per´ez-Salazar, A.; Cano, F. H.; Fayos, J.; Martinez-Carrera,
S.; Garcia-Blanco, S. Acta Crystallogr. 1977, B33, 3525.
(3) (a) Robins, D. J. Nat. Prod. Rep. 1993, 10, 487. Niwa, H.; Ogawa,
T.; Okamoto, O.; Yamada, K. Tetrahedron 1992, 48, 10531. (b) Mattocks,
A. R. Chemistry and Toxicology of Pyrrolizidine Alkaloids; Academic
Press: London, 1986. (c) Furuya, T.; Hikichi, M.; Iitaka, Y. Chem. Pharm.
Bull. 1976, 24, 1120. Hikichi, M.; Furuya, T. Tetrahedron Lett. 1974, 3657.
Mattocks, A. R.; White, I. N. H. Chem. Biol. Interact. 1971, 3, 383.
Culvenor, C. C. J.; Edgar, J. A.; Smith, L. W.; Jago, M. V.; Peterson, J. E.
Nature (London) New Biol. 1971, 229, 255. Schoental, R. Nature (London)
1970, 227, 401. Mattocks, A. R. Toxicol. Lett. 1982, 14, 111.
Our plans for the synthesis of the dihydrothiopyranone 4
assumed that one of the thioketone dienophiles 7-9 might
undergo [2 + 4] cycloaddition with the Danishefsky diene
derivatives 10 to give adducts having the regiochemistry of 5.
At the outset of our work, there was little precedent for the
regiochemistry of thioketone Diels-Alder reactions containing
donor alkyl as well as π-acceptor substituents as in 7 or 9,6 and
no confident prediction could be made whether these dienophiles
would react with useful selectivity for the desired regioisomer
5 rather than 6. On the other hand, a number of analogous
reactions of thioaldehydes had been studied, and the general
trends were shown to correlate with the FMO approximation.7,8
Much of the data could be explained by assuming that
regiochemistry is controlled by polarization in the thiocarbonyl
(4) Yamada, K.; Niwa, H.; Uosaki, Y. Tetrahedron Lett. 1983, 24, 5731.
Niwa, H.; Sakata, T.; Yamada, K. Bull. Chem. Soc. Jpn. 1994, 67, 2345.
(5) Vedejs, E.; Stults, J. S. J. Org. Chem. 1988, 53, 2226.
S0002-7863(97)03464-1 CCC: $15.00 © 1998 American Chemical Society
Published on Web 04/02/1998

3614 J. Am. Chem. Soc., Vol. 120, No. 15, 1998
Vedejs et al.
Scheme 2
Figure 1. Thioaldehyde cycloaddition regiochemistry.
LUMO and diene HOMO.8b Carbon has the larger LUMO
coefficient in the CdS bond of thioformaldehyde, and LUMO
polarization is somewhat enhanced by replacing a thioformyl
hydrogen with a saturated alkyl group. The Diels-Alder
reaction of this class of thioaldehyde dienophiles with unsym-
metrical dienes follows the regiochemistry shown in eqs 1 and
2 (Figure 1) and corresponds to preferred interaction of
thioaldehyde carbon with the enol ether terminus of the
unsymmetrical diene, the site that has the larger HOMO
coefficient. On the other hand, the selectivity pattern is inverted
for thioaldehydes containing an unsaturated acceptor group at
the R-carbon (eq 3). In this case, LUMO polarization places
the larger coefficient on thioaldehyde sulfur.
On the basis of the above considerations, thioketones such
as 7 might be expected to react with low selectivity because
the ester and alkyl substituents have opposing directive effects
in the thioaldehyde examples. On the other hand, steric factors
should work against the undesired regiochemistry 6, depending
on the nature of the substituent Z, and might promote the desired
selectivity for 5. The alternative of using an unactivated
thioketone 8 would provide a more predictable solution to the
problem of controlling regiochemistry according to the pattern
of eq 1. However, lower dienophilic reactivity was expected
in the absence of an unsaturated electron acceptor group and
there was the added risk of competing thioenolization. A third
thioketone (9) was therefore included in the preliminary survey
of options. The relatively electron-rich lactam carbonyl group
should be less likely to dominate the regiochemistry of
cycloaddition but might still provide a reasonable level of
activation for the cycloaddition by lowering the thioketone
LUMO energy level compared to that of the saturated analogue
8. Our study began with a comparison of the regiochemical
preferences for these dienophiles.
Results and Discussion
Lactam thioketone 9 was the most accessible thioketone in
the series 7-9 and was generated following the thioaldehyde
precedents by the photochemical cleavage of the corresponding
phenacyl sulfide 13 (Scheme 2). The latter was easily made
from phenacylmercaptan 12 and R-chloro-N-methylbutyrolactam
11 using potassium carbonate as the base. Sun lamp irradiation
of 13 in the presence of the Danishefsky diene 10a9 followed
by workup with trifluoroacetic acid gave two enones 14 and
15 in 72% yield. The major isomer 14 was determined to have
the desired regiochemistry, based on characteristic chemical shift
data for the enone protons (14, δ 7.34, 6.23 ppm; 15, δ 6.63,
6.13 ppm). However, the isomer ratio (3:1 14:15) was deemed
too low for an early step in the intended synthetic application.
Another concern was that regiochemical control might be lost
if it became necessary to use more highly oxygenated Dan-
ishefsky diene analogues such as 10b or 10c to introduce the
C-8 oxygen of otonecine. These dienes contain terminal
substituents with opposing directive effects.
(6) (a) For reviews, see: Metzner, P. Synthesis 1992, 1185. Weinreb, S.
E.; Staib, R. R. Tetrahedron 1982, 38, 3087. (b) Middleton, W. J. J. Org.
Chem. 1965, 30, 1390. Yamada, K.; Yoshioka, M.; Sugiyama, N. J. Org.
Chem. 1968, 33, 1240. Ohno, A.; Ohnishi, Y.; Tsuchihashi, G. Tetrahedron
1969, 25, 871. Katada, T.; Eguchi, S.; Esaki, T.; Sasaki, T. J. Chem. Soc.,
Perkin Trans. 1 1984, 1869. Katada, T.; Eguchi, S.; Esaki, T.; Sasaki, T. J.
Chem. Soc., Perkin Trans. 1 1984, 2649. Bonini, B.; Mazzanti, G.; Zani,
P.; Maccagnani, G. J. Chem. Soc., Perkin Trans. 1 1989, 2083. Schatz, J.;
Sauer, J. Tetrahedron Lett. 1994, 35, 4767. (c) Pinto, I. L.; Buckle, D. R.;
Rami, H. K.; Smith, D. G. Tetrahedron Lett. 1992, 37, 7597.
(7) Kirby, G. W.; Lochead, A. W. J. Chem. Soc., Chem. Commun. 1983,
1325. Bladon, C. M.; Ferguson, I. E. G.; Kirby, G. W.; Lochead, A. W.;
McDougall, D. C. J. Chem. Soc., Perkin Trans. 1 1985, 1541.
(8) (a) Vedejs, E.; Eberlein, T. H.; Mazur, D. J.; McClure, C. K.; Perry,
D. A.; Ruggeri, R.; Schwartz, E.; Stults, J. S.; Varie, D. L.; Wilde, R. G.;
Wittenberger, S. J. Org. Chem. 1986, 51, 1556. (b) Vedejs, E.; Perry, D.
A.; Houk, K. N.; Rondan, N. G. J. Am. Chem. Soc. 1983, 105, 6999. (c)
Chandrasekhar, J.; Ramamurthy, V.; Rao, V. P. J. Chem. Soc., Perkin Trans.
2 1988, 647. Barone, V.; Arnaud, R. Chem. Phys. Lett. 1996, 251, 393.
The modest regiochemical preference in the reaction of 9 with
10a suggested that a scheme based on the ester thioketone 7
would not be viable. The corresponding phenacyl sulfide (16b)
was relatively difficult to make, so the regiochemistry issue was
probed in a model study using 16a for generation of methyl
thionopyruvate 17. Larsen et al. have reported several examples
of the reaction of 17 with unsymmetrical dienes using a different
method to generate the thionopyruvate,10 and a relevant [2 +
4] cycloaddition of the arylthionoglyoxylate m-CF3C6H4C(S)CO2-
Me with 2-(trimethylsiloxy)butadiene is also known.6c Our
regiochemical results are consistent with these earlier reports.
(9) Danishefsky, S.; Kitahara, T.; Schuda, P. F. Org. Synth. 1983, 61,
147.
(10) Larsen, S. D.; Fisher, P. V.; Libby, B. E.; Jensen, R. M.; Mizsak,
S. A.; Watt, W.; Ronk, W. R.; Hill, S. T. J. Org. Chem. 1996, 61, 4725.

Total Synthesis of (()-Otonecine
J. Am. Chem. Soc., Vol. 120, No. 15, 1998 3615
Scheme 3
the corresponding mercaptan was generated using methanolic
potassium carbonate and was trapped in situ with phenacyl
chloride to afford the key phenacyl sulfide 25b (65% overall
yield from 24a). Initial experiments to generate thione 8 by
photolytic fragmentation of 25b succeeded, but thione trapping
with the Danishefsky diene was difficult to control and
complications were encountered upon attempted scale-up.
However, thioenolization of 8 was not the problem, despite our
initial concerns. Instead, the reaction was found to be critically
dependent on the quality of solvents and reagents in the
photolysis step. Careful deoxygenation of the benzene solvent
was the key to obtaining good yields on gram scale, and it was
also important to use a large excess of the Danishefsky diene
(ca. 20:1 weight ratio, 10a:25b). When these precautions were
taken, the desired adduct 26 was formed, and distillation of
volatiles followed by brief treatment with aqueous HCl gave
the dihydropyranone 27 in 84% overall yield. Excess Dan-
ishefsky diene 10a was recovered by distillation and was
recycled multiple times in the photochemical step without
affecting the yield of thioketone trapping products. None of
the regioisomeric cycloadduct was detected in the trapping
experiments of 8 with 10a, and an acceptable solution to the
regiochemistry problem was in hand. It was also found that
crude 26 could be converted directly into the bicyclic thioaminal
28 by treatment with CF3CO2H, 82% based on 25b (77% after
recrystallization). This procedure is suitable for multigram scale
experiments and involves deprotection of the initially formed
27 followed by spontaneous cyclization of the resulting N-
benzylamine intermediate during workup.
The next problem was to determine the best stage for intro-
duction of the C-8 oxygen functionality of otonecine. Experi-
ments designed to introduce oxygen via enolate oxidation from
27 failed due to competing â-elimination of sulfur, but conver-
sion to the enol silane 29 was possible using TBSOTf/Et3N
(TBSOTf ) tert-butyldimethylsilyl triflate). However, attempts
to oxidize this substrate were foiled by the presence of the
relatively unhindered and easily oxidized vinyl sulfide subunit.
We therefore explored the possibility of thioketone trapping
using the more highly oxygenated dienes 10b and 10c. Either
of these structures might afford analogues of 4 with Z ) oxygen
if the regiochemistry of cycloaddition could be controlled.
Attempts to use 10b14 in place of 10a in the photolytic
generation of thioketone 8 were frustrated because this diene
proved to be unstable to sun lamp irradiation. The cyclic
analogue 10c was easily made by reaction of 2-formyl-1,4-
dioxane15 with the Horner-Emmons reagent 3016 as a mixture
of E/Z isomers (63% isolated), but it too decomposed upon
photolysis. This prompted an investigation of thermal genera-
tion7 of the thioketone 8 from the cyclopentadiene adduct 31
(available in 63% yield from the photolysis of 25b in the
presence of excess cyclopentadiene). When 31 was heated with
excess 10c at 130 °C, the reaction produced one major
cycloadduct that was tentatively assigned the pyran structure
Thus, sun lamp irradiation of 16a in the presence of the
Danishefsky diene (10a) afforded a mixture of labile adducts
18 + 19 (multiple CH3O signals detected by 1H NMR assay of
the crude product). Acidic workup gave two enones 20 and 21
in 65% combined yield based on 16a. The minor product (7%
isolated; 20) has the characteristic downfield 1H NMR chemical
shifts for the vinylic hydrogens of the desired isomer (δ 7.29,
6.17 ppm in 20; δ 6.74, 6.05 in 21). In FMO terminology, the
preference for 21 indicates that the LUMO of 17 has the larger
coefficient at sulfur, similar to the thioaldehyde example of eq
3, while the polarization in 9 is inverted and resembles that of
thioformaldehyde. On the basis of the above findings and the
literature analogies,10,6c there was little reason to expect useful
results with the more complex ester thione 7.
The unactivated thione 8 was the logical alternative for control
of cycloaddition regiochemistry, and studies were initiated to
determine whether 8 can be intercepted in practical yield. A
synthesis of the phenacyl sulfide precursor of 8 is outlined in
Scheme 3. Thus, reaction of allyl benzyl ether with nitrone 22
(from N-benzylhydroxylamine and formaldehyde)11 gave the [2
+ 3] cycloadduct 23 (71%) and reductive N-O cleavage using
1
32 (49% isolated) on the basis of H and 13C NMR chemical
shift data and the structural analogy to 29. However, attempts
to hydrolyze 32 produced complex mixtures, and oxidation
chemistry was not promising in view of the experience with
29. These problems were not unexpected, and several alterna-
tives had been considered where the C-8 oxidation would be
performed after removal of sulfur. Efficient routes could be
12
LiAlH4/NiCl2 afforded the alcohol 24a. After conventional
protection and Mitsunobu conversion to the thioacetate 25a,13
(11) Townsend, C. A.; Krol, W. J.; Mao, S.; Steele, D. L. J. Org. Chem.
1991, 56, 728.
(12) Tufariello, J. J.; Meckler, H.; Pushpananda, K.; Senaratne, A.
Tetrahedron 1985, 41, 3447.
(14) Danishefsky, S. J.; Maring, C. J. J. Am. Chem. Soc. 1985, 107, 1269.
(15) Shostakovskii, M. F.; Kuznetsov, N. V.; Che-min, Y. Dokl. Akad.
Nauk SSSR 1961, 9, 1570.
(16) Warren, S.; Earnshaw, C.; Wallis, C. J. J. Chem. Soc., Perkin Trans.
1 1979, 3099.
(13) Volante, R. P. Tetrahedron Lett. 1981, 22, 3119.

3616 J. Am. Chem. Soc., Vol. 120, No. 15, 1998
Vedejs et al.
Scheme 4
azocine methyl sulfides 40a and 40b, respectively, in 78-80%
overall yield from 36.
Several additional techniques for thioaminal desulfurization
were explored, and two of these gave interesting results. Caserio
et al. have reported a method for hydrolysis of thioaminals using
the dimethyl disulfide derived sulfonium salt 41 as an electro-
philic sulfenylating agent that selectively activates sulfur in the
presence of basic nitrogen.17 Attempts to use this method to
activate 28 for cyanoborohydride reduction resulted in undesired
C-S elimination to an enamine and further sulfenylation,5 but
the corresponding experiment with alcohol 36 produced an
intermediate salt 42 that was converted into the azocine disulfide
43 (79%) by NaBH3CN reduction in acetonitrile. This poten-
tially useful conversion was not exploited because elimination
of sulfur from derivatives of 40 using conventional methods
proved to be more efficient than from 43. However, the desired
selectivity for sulfur activation was demonstrated and efficient
reductive thioaminal C-S cleavage was observed. Another
result worth mentioning was encountered during efforts to
control Raney nickel desulfurizations. These experiments
afforded complex mixtures from 28, but an attempt to desulfu-
rize the alcohol 36 resulted in a sulfur-free product with
promising changes in the 1H NMR spectrum. However, exact
mass determination established that the product has a molecular
weight 2 amu below that expected for the azocine 44. This
observation was not pursued, but the tentative structure 45 (two
1
diastereomers according to H NMR) is in accord with the
absence of vinylic hydrogen signals in the NMR spectrum.18
Elimination of sulfur was the next obstacle. Selective
conversion of 40a or 40b to the sulfoxides 46a or 46b (Scheme
5) was readily achieved using the tetrabutylammonium Oxone
(TBA Oxone) reagent19 without interference by basic amine
nitrogen. Subsequent sulfoxide elimination in refluxing xylene
produced mixtures of alkene regioisomers, ca. 2.5-3:1 47:48.
On the other hand, a single dominant (>20:1) isomer 50a (89%)
was obtained starting from the N-benzyloxycarbonyl (N-Cbz)
derivative 49a (from 40a by treatment with ClCO2CH2C6H5)21
in a similar oxidation-thermolysis sequence. The reasons for
this dramatic improvement in selectivity may be related to
electrostatic effects in the transition state for sulfoxide elimina-
tion,20 but a detailed rationale is not warranted because the
sulfoxide stereochemistry is not known.
Initial experiments for introduction of the C-8 oxygen of
otonecine focused on the m-chloroperoxybenzoic acid (MCPBA)
or OsO4 oxidations of the acetoxy alkene 50b, prepared from
47b and ClCO2CH2C6H5. It was anticipated that the tertiary
C-O bond in the epoxide 51 or the diol 52 might undergo
elimination to the desired trisubstituted alkene 53 corresponding
to the substitution pattern of otonecine. However, exposure of
51 to conditions for epoxide elimination afforded complex
product mixtures. In one of the more promising experiments,
treatment of 51 with the TMSOTf/2,6-lutidine/DBU reagent22
gave 29% of an enal (55) as the major product, presumably
formed via the exocyclic enol ether 54. Similarly, 55 was
obtained along with unidentified products from the diol 52 (1:1
envisioned from 28 if thioaminal C-S cleavage to 34 (Scheme
4) and subsequent sulfur elimination could be controlled to
selectively form the unsaturated ketone 35.
Reductive C-S bond cleavage from 28 to 34 proved to be a
difficult problem. A number of literature desulfurization proce-
dures including nickel boride, Raney Ni, and triphenyltin hy-
dride/2,2′-azobisisobutyronitrile (AIBN) failed to give the
desired azocine product. Sodium cyanoborohydride treatment
did produce small amounts of a mercaptan alcohol (37), but 34
was never detected as an intermediate. Control experiments
established that the reductive C-S cleavage is considerably
faster after the ketone has been reduced to the alcohol 36. This
rate difference is probably due to the unfavorable dipole inter-
action in the keto iminium salt 33a derived from 28 compared
to the alcohol iminium salt 33b. Efficient reductive trapping
of 33b by cyanoborohydride is essential for the conversion to
the azocine mercaptan 37. Thus, treatment of 28 with lithium
tri(sec-butyl)borohydride produced 36 as a single diastereomer
(relative stereochemistry is shown; racemic product), with
stereochemistry tentatively assigned assuming least hindered
attack syn to the compact sulfur bridge. Subsequent reduction
of 36 with NaBH3CN gave 37 in 57% yield. For preparative
purposes, the best results were obtained from 36 after O-
benzylation (NaH,PhCH2Br) or O-acetylation [Ac2O/Et3N/
DMAP(4-(dimethylamino)pyridine)]. The resulting benzyl ether
38a or acetate 38b was reduced cleanly using NaBH3CN in
trifluoroethanol-acetic acid, and S-methylation with CH3I/DBU
(DBU ) 1,8-diazabicyclo[5.4.0]undec-7-ene) produced the
(17) Caserio, M. C.; Kim, J. K.; Souma, Y.; Beutow, N.; Ibbeson, C. J.
Org. Chem. 1989, 54, 1714.
(18) Analogies: Pettit, G. R.; van Tamelen, E. E. Org. React. (N.Y.)
1962, 12, 356. Guziec, F.; Sanfilippo, L. J. Tetrahedron 1988, 44, 6241.
de Mayo, P.; Nicholson, A. A. Isr. J. Chem. 1972, 10, 341. Omote, Y.;
Yoshioka, M.; Yamada, K.; Sugiyama, N. J. Org. Chem. 1967, 32, 3676.
(19) Trost, B. M.; Braslau, R. J. Org. Chem. 1988, 53, 532.
(20) Trost, B. M.; Leung, K. K. Tetrahedron Lett. 1975, 16, 4197.
(21) Cooley, J. H.; Evain, E. J. Synthesis 1989, 1.
(22) Noyori, R.; Murata, S.; Suzuki, M. Bull. Chem. Soc. Jpn. 1982, 55,
247.

Total Synthesis of (()-Otonecine
J. Am. Chem. Soc., Vol. 120, No. 15, 1998 3617
Scheme 5
Scheme 6
This procedure gave an improved 30:1 ratio of 58:57 (81%), a
result that acquired special significance later, when it was found
that only the major diastereomer 58 is useful for the eventual
oxidation at C-8 as well as regiospecific elimination of the
tertiary C-1 OH group. The stereochemistry of 58 is assigned
by comparison of the H-C-7-C-8-H coupling constants in
the corresponding acetonides (acetonide from 58, J7,8 ) 7.5 Hz;
acetonide from 57, J7,8 < 2 Hz) and by evaluation of the relevant
conformational options. The stereochemistry also corresponds
to that expected from the precedented transition state where
OsO4 interacts with the more available alkene face of a local
conformer having the allylic ether substituent in a pseudoequa-
torial orientation as shown.25 The alternative of internal delivery
of OsO4 via coordination to the ring nitrogen is conceivable,
but this would be unlikely under the TMEDA-accelerated
conditions because the combination of alkene and bidentate
amine ligands occupies the available coordination sites at
osmium in the transition state.26 The stereochemical assignment
is not critical because neither C-1 nor C-8 is stereogenic at the
otonecine stage, but a knowledge of the geometry of 58 was
helpful in understanding subsequent transformations.
diastereomer mixture from catalytic osmylation of 50b) upon
attempted dehydration of the tertiary alcohol with the Burgess
reagent.23 Good results were eventually obtained after conver-
sion of the N-Cbz group to the otonecine N-methyl substitution
pattern by LiAlH4 reduction and introduction of the C-8 carbonyl
group of otonecine. This sequence was conveniently performed
in the O-benzyl series starting from 50a (Scheme 6). The
corresponding N-methyl derivative 56 was obtained in 96% yield
upon treatment of 50a with LiAlH4, and the necessary oxidation
pattern was introduced using an osmylation-elimination se-
quence as described below.
It was now necessary to oxidize the secondary C-8 hydroxyl
of 58. The oxidant must tolerate basic nitrogen and must oxidize
the diol to a ketol without C-1-C-8 bond cleavage, a situation
where Dess-Martin or Swern oxidations are generally used.
However, attempts to oxidize either diol diastereomer 58 or 57
using these methods did not result in loss of the C-8 hydrogen.
Instead, rapid conversion was observed to polar products having
the N-methyl signal at δ ca. 3.2 ppm, compared to a shift of δ
2.33 ppm in the starting diol. The formation of an ammonium
salt 59 by transannular nitrogen participation appears likely and
is supported by molecular weight determination using FAB mass
spectroscopy.
Osmylation of 56 using stoichiometric OsO4 at room tem-
perature gave an 18:1 mixture of 58:57, but the reaction also
produced a side product having an unusual downfield signal in
the 1H NMR spectrum at δ 5.07 ppm and lacking the C-3 NCH2
signals of 56.24a Fortunately, the complication could be avoided
under TMEDA-accelerated osmylation conditions at -78 °C.24b
(23) Burgess, E. M.; Penton, H. R., Jr.; Taylor, E. A. J. Org. Chem.
1973, 38, 26. Burgess, E. M.; Penton, H. R., Jr.; Taylor, E. A.; Williams,
W. M. Org. Synth. 1977, 56, 40.
(24) (a) Structure i is tentatively assigned to the major sideproduct, based
on 1H NMR data indicating a hemiaminal CHCH2 ABX pattern [δ 5.06
(1H, dd, J ) 8.7, 4.5 Hz), 2.08 (1H, dd, 15.3, 8.7 Hz), 1.98 (1H, 15.3, 4.5
Hz)].
In view of the limitations imposed by substrate functionality,
an OsO4-mediated oxidation of the diol 58 to the ketol 60 was
investigated. Osmium tetroxide had already been used suc-
cessfully to introduce the diol functionality, and Mehrotra et
(25) Vedejs, E.; Dent, W. H., III.; Gapinski, D. M.; McClure, C. K. J.
Am. Chem. Soc. 1987, 109, 5437.
(26) Corey, E. J.; Sarshar, S.; Azimioara, Newbold, R. C.; Noe, M. C.
J. Am. Chem. Soc. 1996, 118, 7851.
(b) The low-temperature osmylation procedure was based on the following
precedent where a chiral diamine is the accelerating ligand: Tomioka, K.;
Nakajuma, M.; Koga, K. J. Am. Chem. Soc. 1987, 109, 6213.

3618 J. Am. Chem. Soc., Vol. 120, No. 15, 1998
Vedejs et al.
Scheme 7
temperature dependent, and several signals including the ring
protons between δ 1.6-2.8 ppm were extensively broadened
at room temperature. The signals sharpened upon warming or
upon addition of HCl to the NMR sample. In the latter case,
some of the 1H NMR and 13C chemical shifts were also
displaced substantially to lower field due to conversion into the
bicyclic 63, and a typical ammonium N-methyl signal was
observed at δ 2.89 ppm in CD3OH/HCl. Under these condi-
tions, the carbonyl carbon in the 13C spectrum was replaced by
a new singlet at δ 117.5 ppm, corresponding to the bridgehead
carbon in 63. Related transannular bonding interactions proved
to be important in subsequent chemistry.
The dehydration of the ketol 60/61 to dibenzylotonecine (68/
69) was especially interesting. Treatment of 60/61 with the
Burgess reagent (Et3NSO2NCO2Me, 62)23 in C6D6 at room
temperature produced changes in the chemical shift of the CO2-
CH3 subunit derived from 62 as well as a downfield shift of
the N-methyl group consistent with the conversion of 60/61 into
a bicyclic structure having a formal positive charge at nitrogen.
The 1:1 adduct structure 64 is consistent with this evidence and
with the presence of a quaternary carbon signal at 116.5 ppm
in the 13C spectrum. Upon heating to 80 °C, 64 reverted to the
starting ketol 60/61 without detectable elimination to alkene
products. In a separate experiment, the solution containing the
initial 1:1 adduct 64 and excess Burgess reagent was heated to
35-40 °C, resulting in the slow formation of a 2:1 adduct
believed to have the structure 65. Heating 65 to 80 °C followed
by hydrolysis in refluxing aqueous THF gave dibenzyl otonecine
68/69 in 50% yield. Further investigation revealed that the crude
product prior to hydrolysis contains four distinct precursors of
the desired 68/69. Chromatographic separation afforded samples
that could be partially characterized by 1H NMR, IR, and low-
resolution FAB mass spectroscopy. Two of these products are
assigned the diastereomeric sulfate structures 67 (two cis-fused
diastereomers) because the same NMR signals were produced
by treating dibenzylotonecine 68/69 with the pyridine-SO3
complex. The other two products contain methoxy signals in
the 1H NMR spectrum as well as other signals that are consistent
with the bicyclic valence bond tautomer form of the otonecine
nucleus, as in the sulfamate structure 66 (pair of diastereomers).
The molecular weights corresponding to 66 and 67 were detected
in the corresponding diastereomer mixtures using FAB mass
spectroscopy.
al. have reported diol oxidations using K3Fe(CN)6 and catalytic
OsO4 under strongly basic conditions.27 Related osmium
reagents have also been used to convert alkenes directly into
ketols.28 Attempts to follow the latter precedent for the con-
version of the alkene 56 into 60 gave low yields of ketol at
room temperature, but this approach was not explored in detail
because of concerns about oxidation R to amine nitrogen24a as
well as the possibility of diastereomer formation at the higher
temperature. On the other hand, a reagent made by the
substitution of CaO for K2CO3 in the Yamamoto conditions for
catalytic olefin hydroxylation with K3Fe(CN)6 as the stoichio-
metric oxidant29 converted the diol 58 to the ketol 60 in modest,
yet functional 53% yield at room temperature. Among the
oxidation conditions investigated, the osmium-based methods
were the only ones to produce detectable quantities of 60.
Ketol 60 is the first intermediate in the synthetic route that
contains basic azocine nitrogen as well as the otonecine carbonyl
oxidation state at C-8. A similar environment is responsible
for strong transannular interactions in otonecine (represented
by the equilibrium of 1a with 1b in Scheme 1),2b resulting in
unusual spectroscopic properties. Analogous behavior was seen
in 60, including broad carbonyl absorptions in the infrared
spectrum. A carbonyl carbon was also apparent at δ 200 ppm
in the 13C NMR spectrum in CD3CN, suggesting that both 60
Adduct 64 cannot easily undergo Burgess elimination via ion
pair formation due to the presence of positively charged
nitrogen, but elimination is possible from the bis-adduct 67.
The elimination precursors as well as the products are in the
bicyclic form, and the same is true of dibenzylotonecine 68/69
when the 13C NMR spectrum is recorded in acidic methanol (δ
1
118.6 ppm for C-8). Under neutral conditions, the H NMR
spectrum of 68/69 shows sharper signals at room temperature
compared to 60/61 and warming to 55 °C results only in a
relatively small improvement in resolution. Presumably, two
diastereomers of the bicyclic form interconvert under the higher
temperature conditions in 68/69 as well as 60/61. The 13C
spectrum (CDCl3) contains no clear carbonyl signal in the region
between δ 195 and 205 ppm, suggesting that the bicyclic valence
bond tautomer 69 is dominant in this case.
1
and 61 are present (Scheme 7). The H NMR spectrum was
Deprotection of dibenzyl otonecine 68/69 was best achieved
under Lewis acid catalyzed conditions for nucleophilic deben-
zylation. Treatment of 68/69 with TMSOTf in the presence of
thioanisole in acetonitrile at 0 °C gave crude otonecine as the
triflic acid salt and neutralization using ion exchange chroma-
tography resulted in otonecine free base. The product (ca. 80%
(27) Kappor, R. C.; Mehrotra, R. N.; Vajpai, S. K.; Chaudhary, P. Trans.
Met. Chem. 1991, 16, 65.
(28) VanRheenen, V.; Kelly, R. C.; Chan, D. Y. Tetrahedron Lett. 1976,
17, 1973. Lohray, B. B.; Bhushan, V.; Kumar, R. K. J. Org. Chem. 1994,
59, 1375.
(29) Minato, M.; Yamamoto, K.; Tsuji, J. J. Org. Chem. 1990, 55, 766.

Total Synthesis of (()-Otonecine
J. Am. Chem. Soc., Vol. 120, No. 15, 1998 3619
(CDCl3, ppm) δ 7.36-7.24 (10H, m), 4.56 (2H, s), 4.08-3.96 (1H,
m), 3.78 (2H, AB q, JAB ) 13.1 Hz), 3.44 (2H, dd, J ) 3.7, 1.8 Hz),
3.05-2.74 (2H, m), 1.80-1.60 (2H, m).
isolated) was identified by comparison of NMR chemical shifts
with those of authentic material.
Conclusion. Access to the otonecine ring system has been
demonstrated by a combination of thioketone Diels-Alder
addition and desulfurization technology. The sulfur-free azocine
50a was produced in 15 steps, 20% overall yield from
benzylhydroxylamine and allyl benzyl ether, the starting materi-
als for Scheme 3. The final steps to otonecine followed a more
conventional strategy, but they proved to be more difficult
because of the challenging oxidation-elimination sequence (ca.
30% yield from 50a to racemic otonecine 1, 5 steps). Regio-
chemical problems in the thioketone Diels-Alder step were
solved by using the nonactivated thione 8, and the cycloaddition
was surprisingly efficient. Methods for selective C-S bond
cleavage at the thioaminal stage were developed, and regio-
chemical complications in the sulfur elimination step were
solved by modifying the electronic environment at azocine
nitrogen. All of the chemistry involving intermediates with C-8
at the carbonyl oxidation state (Scheme 6) was dominated by
the transannular interactions between C-8 and the basic ring
nitrogen that is characteristic of the otonecine ring sytem.2b
These interactions complicate the use of electrophilic reagents
because the substrate is likely to exist in the cationic, bicyclic
valence bond tautomer form. Nevertheless, it was possible to
manipulate the molecule under electrophilic conditions at several
stages, including the key oxidation to 60/61, the Burgess
elimination to 68/69, and the final deprotection to otonecine 1.
N-Protected Alcohol 24b. Di-tert-butyl dicarbonate (Aldrich; 3.34
g, 15.3 mmol) was added to a solution of amino alcohol 24a (3.53 g,
12.4 mmol) and anhydrous potassium carbonate (2.6 g, 18.5 mmol) in
75 mL of 2:1 dioxane/water. This was allowed to stir overnight. The
mixture was concentrated to remove dioxane, and the solution was
saturated with sodium chloride. Extraction with ethyl acetate followed
by drying (MgSO4) gave crude 24b. The residue was purified by flash
chromatography on silica gel (3 cm × 45 cm), 1:1 EtOAc/hexane eluent,
to give 4.73 g (99%) of 24b: analytical TLC on silica gel, 1:1 EtOAc/
hexane, Rf ) 0.52; M+ calcd for C19H23NO4 329.16266, found m/e )
329.1617 (M - 56), error ) 3 ppm, base peak ) 91 amu; IR (CDCl3,
cm-1) 3600 O-H, 3440 O-H; 200 MHz NMR (CDCl3, ppm) δ 7.48-
7.13 (10H, br m), 4.70-4.15 (4H, br m), 3.93-3.01 (6H, br m), 1.85-
1.47 (2H, br m), 1.47 (9H, br s).
Thiol Acetate 25a. The procedure of Volante was used.13 Diiso-
propylazodicarboxylate (Aldrich; 1.5 mL, 7.4 mmol) was added to
triphenylphosphine (Aldrich; 1.95 g, 7.46 mmol) in 35 mL of THF. A
white precipitate formed after several minutes. After 1 h, a mixture
of alcohol 24b (1.42 g, 3.70 mmol) and thiolacetic acid (Aldrich; 0.5
mL, 7.40 mmol) in 15 mL of THF was added dropwise, and the
resulting solution was allowed to stir overnight. The resulting clear
yellow solution was concentrated, the residue was diluted in ether to
precipitate triphenylphosphine oxide and filtered, and the solution was
then concentrated to an oil. Flash chromatography on silica gel (3 cm
× 20 cm), 1:4 EtOAc/hexane eluent, gave thiolacetate 25a (1.7 g)
contaminated with diisopropylazodicarboxylate byproducts. Pure
individual fractions were collected for characterization. The crude
thiolacetate 25a was routinely used as is in the next step. For 25:
analytical TLC on silica gel, 1:4 EtOAc/hexane, Rf ) 0.43; M+ calcd
for C25H33NO4S 443.21307, found m/e ) 387.1531 (M - 56); calcd
for C21H25NO4S 387.15045, error ) 7 ppm, base peak ) 91 amu; IR
(CDCl3, cm-1) 1685 CdO; 200 MHz NMR (CDCl3, ppm) δ 7.35-
7.22 (10H, m), 4.55 (2H, s), 4.50-4.43 (2H, m), 3.80-2.90 (5H, m),
2.34 (3H, s), 2.26-1.98 (1H, m), 1.95-1.70 (1H, m), 1.50 (9H, s).
Phenacyl Sulfide 25b. The crude thiolacetate 25a (1.7 g) was
dissolved in 20 mL of methanol and cooled to 0 °C. Anhydrous
potassium carbonate (1.31 g, 9.50 mmol) followed by R-chloroac-
etophenone (Aldrich; 715 mg, 4.63 mmol) in 25 mL of methanol was
added. After stirring overnight, the mixture was concentrated and
diluted in methylene chloride, washed once with saturated NaHCO3,
and dried (MgSO4). After removal of solvent (aspirator), the residue
was purified by flash chromatography on silica gel (3 cm × 45 cm),
gradient elution using 50 mL of 3:17, 50 mL of 1:5, and 200 mL of
3:7 EtOAc/hexane as the eluents, to give 1.25 g (65% from 24b) of
25b as an oil: analytical TLC on silica gel, 1:4 EtOAc/hexane, Rf )
0.35; M+ calcd for C31H37NO4S 519.24432, found m/e ) 446.1803 (M
- 73); calcd for C27H28NO3S 446.17902, error ) 3 ppm, base peak )
91 amu; IR (CH2Cl2, cm-1) 1684 CdO, 3595 N-H, 3475 O-H; 200
MHz NMR (CDCl3, ppm) δ 7.99-7.15 (15H, m), 4.49 (2H, s), 4.48-
4.25 (2H, m), 3.98-3.75 (2H, br m), 3.66-3.05 (4H, br m), 3.00-
2.60 (1H, br m), 2.15-1.85 (1H, br m), 1.80-1.50 (1H, br m), 1.47
(9H, s), 2.30-2.17 (1H, br m), 2.00-1.80 (1H, br m), 1.47 (9H, br s).
Photolysis of 25b with Danishefsky’s Diene: Preparation of 27
and 28 (5-(Benzyloxymethyl)-2-aza-9-thiabicyclo[3.3.1]octan-7-one).
Photograde benzene was prepared by washing 3 L of benzene with a
mixture of KMnO4 (100 mL, saturated aqueous) and H2SO4 (10 mL,
concentrated) four times. Caution! Saturated KMnO4 and concentrated
sulfuric acid should not be premixed, as solid KMnO4 may detonate in
the presence of H2SO4. The benzene was then washed with water (100
mL), H2SO4 (8 × 50 mL, concentrated), water (100 mL), NaHCO3 (2
× 100 mL), and brine (2 × 100 mL), dried over MgSO4, and distilled
from CaH2.
Experimental Section
General Procedures. Flash chromatography on ammonia-impreg-
nated silica gel (designated in the Experimental Section by silica gel-
NH3) was performed by packing a dry column with silica gel (EM
Silica Gel 60 230-400 mesh) and eluting with 4% NH4OH in THF
until ammonia vapors could be detected in the eluent (wet pH paper).
The column was then equilibrated with NH4OH-saturated CHCl3,
followed by the chromatography solvent. High- and low-resolution
FAB mass spectra were obtained on a VG Analytical AutoSpec using
nitrobenzyl alcohol as matrix (high resolution, PEG standard, 30-35
KV Cs ions (SIMS), 10K resolution).
2-Benzyl-4-(benzyloxymethyl)oxazolidine (23). Nitrone 2211 (11.38
g, 84.2 mmol) and allyl benzyl ether31 (11.16 g, 75.3 mmol) were
dissolved in 100 mL of benzene and heated to reflux for 25 h.
Concentration followed by distillation to remove excess allyl benzyl
ether gave crude isoxazolidine 23. The residue was purified by column
chromatography on silica gel (4.5 cm × 45 cm), 1:1 EtOAc/hexane
eluent, to give 15.9 g (74%) of pure 23: analytical TLC on silica gel,
1:1 EtOAc/hexane, Rf ) 0.33; molecular ion (M+) calcd for C18H21-
NO2 283.15723, found m/e ) 283.1549, error ) 8 ppm, base peak )
91 amu; IR (CDCl3, cm-1) 2875 C-H, 1100 C-O, 1660 CdO; 200
MHz NMR (CDCl3, ppm) δ 7.58-7.12 (10H, m), 4.59 (2H, s), 4.50-
4.26 (1H, m), 3.71-2.22 (2H, m), 3.68-3.41 (2H, m), 3.29-2.55 (2H,
m), 2.50-2.23 (1H, m), 2.16-1.93 (1H, m).
Amino Alcohol 24a. The procedure of Tufariello was used.12
Isoxazolidine 23 (16 mg, 56 mmol) and nickel(II) chloride (20.1 g,
155 mmol) were stirred in THF at 0 °C. Lithium aluminum hydride
(Aldrich; 6 g, 155 mmol) was added in portions, and the mixture was
warmed slowly to room temperature. After 10 h of stirring, anhydrous
potassium carbonate was added, and the solution was stirred for 0.5 h.
The mixture was filtered through a pad of Celite and washed extensively
with ether. Drying (MgSO4) and concentration gave the amino alcohol
24a (16 g, 72%) which was used without further purification: M+ calcd
for C18H23NO2 285.17288, found m/e ) 285.1673, base peak ) 91
amu; IR (CDCl3, cm-1) 3585 O-H, 3220 N-H; 200 MHz NMR
Phenacyl sulfide 25b (395 mg, 0.76 mmol) was divided into four
parts in 50 mL round-bottom flasks with 1-methoxy-3-(trimethylsily-
loxy)-1,3-butadiene9 (2 mL, 10.3 mmol), and photograde benzene (7.5
mL; degassed by repeated freeze-thaw cycles under vacuum and argon)
was photolyzed (Sun Lamp through a water-cooled solution of CuSO4;
saturated, ca. 4 cm layer, 22-24 °C) with magnetic stirring and a gentle
(30) Fuji, N.; Otaka, A.; Sugiyama, M.; Hatano, M.; Yajima, H. Chem.
Pharm. Bull. 1987, 35, 3880.
(31) Von Braun, J. Chem. Ber. 1910, 43, 1350.
(32) Monzingo, R. Organic Syntheses; Wiley: New York, 1955; Collect.
Vol. III, p 181.

3620 J. Am. Chem. Soc., Vol. 120, No. 15, 1998
Vedejs et al.
argon sparge. After 8 h of photolysis, the mixtures were combined
and concentrated. The excess diene was removed under vacuum with
gentle warming. The residue was diluted in THF and treated with 1 N
HCl for 5 min. The solution was neutralized with saturated NaHCO3
and concentrated to remove the THF. The solution was extracted with
ether and dried (MgSO4). After removal of solvent (aspirator), the
residue was purified by flash chromatography on silica gel (2.2 cm ×
30 cm), gradient elution using first 25 mL of 1:5 then 3:7 EtOAc/
hexane as the eluents, to give 300 mg (84%) of enone 27: analytical
TLC on silica gel, 1:4 EtOAc/hexane, Rf ) 0.22; M+ calcd for C27H33-
NO4S 467.21307, found m/e ) 467.2138, error ) 2 ppm, base peak )
91 amu; IR (CH2Cl2, cm-1) 1685 CdO, 1665 CdO, 1125 C-O; 200
MHz NMR (CDCl3, ppm) δ 7.36-7.22 (11H, m), 6.10 (1H, d, J )
10.2 Hz), 4.48 (2H, br s), 4.37 (2H, br s), 3.57-3.09 (4H, br m), 2.83
(1H, br d, J ) 15.7 Hz), 2.63 (1H, br d, J ) 15.7 Hz), 2.08-1.78 (2H,
br m), 1.46 (9H, br s). Bicyclic thioaminal 28: A solution of the
phenacyl sulfide (2.57 g, 4.94 mmol) and Danishefsky’s diene (50 mL)
in photograde benzene (115 mL) was photolyzed as described above
for 12 h. The solvent was removed (aspirator), and the excess diene
was distilled off under vacuum (95 °C, ca. 1 Torr) and recycled. The
reaction mixture was heated under vacuum (1 Torr, 110 °C, 3 h), and
the resulting oil was taken up in CH2Cl2 (100 mL), cooled to 0 °C,
and treated with trifluoroacetic acid (6.5 mL, dropwise). The mixture
was then stirred at room temperature for 5 h and concentrated
(aspirator). The resulting black oil was redissolved in CH2Cl2, extracted
once with 2 N NaOH and once with brine, and dried (Na2SO4), and
the solvent was evaporated (aspirator). Filtration through a plug of
silica gel with ether gave an eluant that was concentrated and purified
by flash chromatography on silica gel, 15% EtOAc/hexane eluent, to
yield the bicyclic ketone 28 as a pale yellow solid (1.48 g, 4.03 mmol,
82% yield). Crystallization from EtOAc/hexane gave 1.40 g of slightly
yellow crystals (mp 88.5-89.5 °C, 3.81 mmol, 77% yield). An
analytical sample was obtained by recrystallization from EtOAc/hexane
(mp 90.0-90.5 °C): M+ calcd for C22H25NO2S 367.16064, found m/e
) 367.1605, error ) 0 ppm; IR (CDCl3, cm-1) 1700 CdO, 1090 C-O;
200 MHz NMR (CDCl3, ppm) δ 7.37-7.24 (10H, m), 4.60 (2H, s),
4.37-4.34 (1H, m), 4.07 (1H, d, J ) 13.5 Hz), 3.86 (1H, d, J ) 13.5
Hz), 3.43 (1H, d, J ) 9.3 Hz), 3.35 (1H, d, J ) 9.3 Hz), 3.18-2.68
(6H, m), 2.09 (1H, dddd, J ) 1.0, 4.9, 13.5, 13.5 Hz), 1.28 (1H, ddd,
J ) 2.5, 2.5, 13.5 Hz). Anal. Calcd: C, 71.89; H, 6.87. Found: C,
71.78; H, 6.90.
Bicyclic Thioaminal Alcohol 36 (5-(Benzyloxymethyl)-2-aza-9-
thiabicyclo[3.3.1]octan-7-ol). Ketone 28 (226 mg, 0.61 mmol) and
L-Selectride (Aldrich; 1.2 mL, 1.2 mmol, 1M) were stirred in THF at
room temperature for 3 h. After removal of solvent (aspirator), the
residue was purified by flash chromatography on silica gel (2 cm ×
30 cm), 1:1 EtOAc/hexane eluent, to give 221 mg (97%) of 36:
analytical TLC on silica gel, 1:1 EtOAc/hexane, Rf ) 0.16; M+ calcd
for C22H27NO2S 369.17630, found m/e ) 369.1776, error ) 4 ppm,
base peak ) 91 amu; IR (CDCl3, cm-1) 3580 O-H; 200 MHz NMR
(CDCl3, ppm) δ 7.36-7.26 (10H, m), 4.58 (2H, s), 4.21-4.04
(2H, m), 3.85 (2H, AB q, JAB ) 16.6 Hz), 3.30 (2H, s), 3.33-3.12
(1H, m), 2.85 (1H, dt, J ) 14.6, 5.2 Hz), 2.36-2.17 (2H, m), 2.05-
1.62 (4H, m).
Azocine Mercaptan Alcohol 37. Bicyclic thioaminal alcohol 36
(40 mg, 0.11 mmol), sodium cyanoborohydride (Aldrich; 31 mg, 0.50
mmol), and acetic acid (17 µL, 0.30 mmol) in 2,2,2-trifluoroethanol
(Aldrich) were stirred at room temperature for 2 days. The mixture
was quenched with 5% NaOH, extracted with CH2Cl2, and dried
(MgSO4). After removal of solvent (aspirator), the residue was purified
by flash chromatography on silica gel (1.2 cm × 10 cm), 3:2 EtOAc/
hexane eluent, to give 21 mg (57%) of azocine 37: analytical TLC on
silica gel, 3:2 EtOAc/hexane, Rf ) 0.37; M+ calcd for C22H29NO2S
371.19196, found m/e ) 371.1936, error ) 4 ppm, base peak ) 338
(M - SH) amu; IR (CH2Cl2, cm-1) 3690 O-H; 200 MHz NMR
(CDCl3, ppm) δ 7.36-7.24 (10H, m), 4.50 (1H, d, J ) 12.1 Hz), 4.43
(1H, d, J ) 12.1 Hz), 4.12-4.02 (1H, m), 3.55 (1H, d, J ) 12.8 Hz),
3.46 (1H, d, J ) 12.8 Hz), 3.40 (1H, d, J ) 9.2 Hz), 3.35 (1H, d, J )
9.2 Hz), 2.74-2.33 (5H, m), 2.19 (1H, br s), 2.10-1.47 (6H, m).
5-(Benzyloxymethyl)-7-(benzyloxy)-2-aza-9-thiabicyclo[3.3.1]-
octane (38a). A solution of the alcohol 36 (286 mg, 0.774 mmol) in
5:1 THF/DMF (25 mL) at 0 °C under nitrogen was treated with an
excess of hexane-washed NaH (175 mg) and a catalytic amount of
imidazole. Benzyl bromide (500 µL, 4.20 mmol) was added, followed
by tetrabutylammonium iodide (45 mg). The mixture was stirred for
1 h at 0 °C and 10 h at room temperature. The reaction was quenched
with methanol and stirred overnight. Aqueous workup (CH2Cl2, NH4-
Cl) and flash chromatography on silica gel, 10% to 20% EtOAc/hexane
eluent, gave the benzyl ether 38a as a clear colorless oil (314 mg, 0.684
mmol, 88% yield): analytical TLC on silica gel, 20% EtOAc/hexane,
Rf ) 0.36; M+ calcd for C29H33NO2S 459.22327, found m/e ) 459.2209,
error ) 5 ppm: (M - S) 427.2524, error ) 3 ppm; IR (neat, cm-1
)
3028 dC-H, 1095 C-O, 1073 C-O; 200 MHz NMR (CDCl3, ppm)
δ 7.40-7.15 (15H, m), 4.56 (2H, s), 4.53 (2H, s), 4.23 (1H, d, J )
2.8, 9.2 Hz), 4.03 (1H, d, J ) 13.7 Hz), 3.96-3.73 (1H, m), 3.85 (1H,
d, J ) 13.7 Hz), 3.58 (1H, ddd, J ) 3.5, 13.8, 14.3 Hz), 3.28 (2H, s),
2.81-2.57 (2H, m), 2.36 (1H, dd, J ) 6.5, 13.5 Hz), 2.05 (1H, ddd, J
) 4.6, 13.1, 13.8 Hz), 1.79 (1H, ddd, J ) 2.8, 10.0, 14.0 Hz), 1.59
(1H, dd, J ) 9.9, 13.5 Hz), 1.15 (1H, ddd, J ) 13.1, 3.5, 3.5 Hz).
Methylthioazocine (40a). A solution of 38a (263 mg, 0.573 mmol)
in trifluoroethanol (10 mL) at room temperature under nitrogen was
treated with NaBH3CN (280 mg, 4.45 mmol) and acetic acid (275 µL,
4.80 mmol). The resulting suspension was stirred at room temperature
for 48 h. The reaction mixture was concentrated, quenched with NH4-
Cl, and extracted with degassed methylene chloride. The combined
organic layers were dried over Na2SO4 and concentrated to yield a
cloudy colorless oil of the crude mercaptan 39a. A solution of 39a
and DBU (500 µL) in benzene (20 mL) was treated with methyl iodide
(200 µL), and the mixture was stirred at room temperature for 14 h.
After aqueous workup (ether, NH4Cl, NaHCO3, brine), the residue was
purified by flash chromatography on silica gel, 10% to 20% EtOAc/
hexane eluent, to yield 40a as a clear colorless oil (232 mg 0.488 mmol).
In addition, a mixture of starting material 38a and product were
recovered which were resubmitted to the same reaction conditions to
yield an additional 11 mg of the desired product (combined yield 243
mg, 0.511 mmol, 89% yield): M+ calcd for C30H38NO2S 476.2623,
found [HRFAB] m/e ) 276.2615, error ) 2 ppm; (M - SCH3)
428.2590, error ) 0 ppm; 500 MHz NMR (CDCl3, ppm) δ 7.40-7.17
(15H, m), 4.54 (1H, d, J ) 11.8 Hz), 4.48 (1H, d, J ) 11.8 Hz), 4.34
(2H, AB q, J ) 7.4 Hz, J ) 8.3 Hz), 3.72 (1H, ddd, J ) 3.6, 7.9, 10.6
Hz), 3.48 (1H, d, J ) 12.7 Hz), 3.42 (1H, d, J ) 12.7 Hz), 3.33 (1H,
d, J ) 9.7 Hz), 3.27 (1H, d, J ) 9.7 Hz), 2.66 (1H, ddd, J ) 4.1, 4.1,
14.9 Hz), 2.62-2.53 (2H, m), 2.49 (1H, ddd, J ) 1.8, 12.7, 14.4 Hz),
2.28 (1H, ddd, J ) 3.4, 3.5, 12.2 Hz), 2.02 (3H, s), 1.87-1.78 (3H,
m), 1.67 (1H, ddd, J ) 4.2, 12.9, 14.9 Hz), 1.53 (1H, dddd, J ) 4.2,
12.0, 12.5, 12.7 Hz); 13C NMR (128 MHz, DEPT, CDCl3, ppm), δ
139.6 s, 139.0 s, 138.2 s, 129.1 d, 128.1 d, 128.0 d, 127.9 d, 127.5 d,
127.4 d, 127.3 d, 127.1 d, 126.8 d, 75.6 d, 74.8 t, 73.0 t, 69.8 t, 63.7
t, 52.0 t, 49.5 t, 49.4 s, 39.9 t, 36.8 t, 30.6 t, 11.2 q.
Sulfenylation of 36. Disulfide 43. Bicyclic sulfide alcohol 36 (57
mg, 0.15 mmol) in 7 mL of acetonitrile was treated with dimethyl-
(thiomethyl)sulfonium tetrafluoroborate17 (110 mg, 0.56 mmol) and
stirred at room temperature for 1 h. Sodium cyanoborohydride (Aldrich;
189 mg, 3.00 mmol) was added, and the mixture was allowed to stir
for 2 days. The mixture was quenched with saturated NH4Cl, extracted
with CH2Cl2, and dried (MgSO4). After removal of solvent (aspirator),
the residue was purified by flash chromatography on silica gel (1.2 cm
× 13 cm), 1:5 EtOAc/hexane eluent, to give 50 mg (79%) of disulfide
43: analytical TLC on silica gel, 1:1 EtOAc/hexane, Rf ) 0.38; M+
calcd for C22H29NO2S (M - CH2S) 371.1920, found m/e ) 371.1903,
error ) 4 ppm, base peak ) 91 amu; IR (CH2Cl2, cm-1) 3600 O-H;
200 MHz NMR (CDCl3, ppm) δ 7.35-7.15 (10H, m), 4.33 (2H, s),
3.96-3.85 (1H, br m), 3.49 (1H, d, J ) 12.8 Hz), 3.40 (1H, d, J )
12.8 Hz), 3.25 (1H, d, J ) 9.8 Hz), 3.21 (1H, d, J ) 9.8 Hz), 2.70-
2.34 (5H, m), 2.30 (3H, s), 2.09-1.80 (3H, m), 1.78-1.44 (3H, m).
N-Carbobenzylmethylthioazocine (49a). A solution of the N-
benzyl azocine 40a (837 mg, 1.76 mmol) in dry THF (50 mL) at -78
°C under nitrogen was treated with benzyl chloroformate (2.5 mL,
dropwise). The mixture was warmed to room temperature and stirred
for 16 h. After aqueous workup (CH2Cl2, NaHCO3), excess reagent
was removed under vacuum (1 Torr). The residue was purified by
flash chromatography on silica gel, 10% to 30% EtOAc/hexane eluent,

Total Synthesis of (()-Otonecine
J. Am. Chem. Soc., Vol. 120, No. 15, 1998 3621
to give 49a as a colorless oil (872 mg, 95%): M+ calcd for C31H37-
NO4S 519.24432, found m/e ) 519.2443, error ) 0 ppm, base peak )
320 amu; IR (neat, cm-1) 1697 CdO, 1235 C(O)-N, 1097 C-O; 500
MHz NMR (CDCl3, ppm) δ 7.38-7.21 (15H, m), 5.19 (0.6H, d, J )
12.4 Hz), 5.15 (0.4H, d, J ) 12.2 Hz), 5.08 (0.6H, d, J ) 12.4 Hz),
5.02 (0.4H, d, J ) 12.2 Hz), 4.63 (0.6H, d, J ) 12.1 Hz), 4.55 (0.6H,
d, J ) 11.6 Hz), 4.55 (0.6H, d, J ) 12.1 Hz), 4.50 (0.6H, d, J ) 11.6
Hz), 4.53 (0.8H, s), 4.39 (0.8H, AB q, J ) 8.4 Hz), 4.05 (4.6H, ddd,
J ) 2.7, 12.2, 14.8 Hz), 3.88-3.74 (2H, m), 3.89 (0.4H, ddd, J ) 2.7,
12.1, 14.8 Hz), 3.59 (0.6H, d, J ) 9.8 Hz), 3.46 (0.6H, d, J ) 9.8 Hz),
3.32 (0.4H, d, J ) 9.8 Hz), 3.39 (0.4H, d, J ) 9.8 Hz), 3.05-2.93
(2H, m), 2.01 (1.2H, s), 2.04 (1.8H, s), 2.05-1.70 (6H, m); 13C NMR
(128 MHz, DEPT, CDCl3, ppm) δ 155.8 s, 155.0 s, 138.8 s, 138.7 s,
138.2 s, 138.1 s, 136.8 s, 136.6 s, 128.4-127.3 (13×) d, 75.2 t, 74.8
d, 74.7 d, 70.1 t, 70.1 t, 67.1 t, 66.9 t, 49.1 s, 44.0 t, 43.8 t, 43.7 t, 42.9
t, 40.2 t, 34.6 t, 33.3 t, 31.5 t, 30.8 t, 11.2 q, 11.1 q.
Sulfoxide Pyrolysis: 50a. A solution of 49a (240 mg, 0.462 mmol)
in CH2Cl2 at 0 °C under nitrogen was treated with a solution of
tetrabutylammonium oxone19 (3.2 mL, 150 mg/mL in CH2Cl2, 0.15 M
activity) while monitoring for complete conversion by TLC (CHCl3
eluant). The reaction mixture was stirred at 0 °C for 1.5 h, diluted
with EtOAc (30 mL), and concentrated (aspirator). The product was
taken up in EtOAc and filtered through Celite (1 in.). The solids were
rinsed thoroughly with EtOAc (35 mL), and the organic filtrates were
concentrated (aspirator). The crude mixture of sulfoxides 50 was
dissolved in xylene (20 mL) and was added dropwise to a refluxing
suspension of CaCO3 (750 mg) in xylene (20 mL) at such a rate as to
maintain a gentle reflux. The mixture was refluxed under nitrogen for
15 min. Flash chromatography on silica gel, 20% EtOAc/hexane eluent,
gave the N-Cbz azocine allyl ether as a clear colorless oil (194 mg,
0.411 mmol, 89% yield): M+ calcd for C30H34NO4 472.2489, found
[HRFAB] m/e ) 472.2487, error ) 1 ppm, base peak ) 472 amu; IR
(neat, cm-1) 2942 C-H, 1698 CdO, 1082 C-O; 500 MHz NMR
(CDCl3, ppm) δ 7.45-7.20 (15H, m), 5.71 (0.4H, d, J ) 7.9 Hz), 5.69
(0.6H, d, J ) 7.9 Hz), 5.14 (0.6H, d, J ) 12.5 Hz), 5.11 (0.4H, d, J )
12.5 Hz), 5.09 (0.6H, d, J ) 12.5 Hz), 5.06 (0.4H, d, J ) 12.5 Hz),
4.53 (0.6H, d, J ) 11.8 Hz), 4.55 (0.4H, d, J ) 11.7 Hz), 4.50 (1.2H,
s), 4.43 (0.4H, d, J ) 11.7 Hz), 4.46 (0.8H, s), 4.32 (0.6H, d, J ) 11.8
Hz), 4.17-4.09 (1H, m), 4.07-3.83 (3H, m), 3.65-3.53 (1H, m), 3.02-
2.84 (2H, m), 2.63-2.53 (0.4H, m), 2.46-2.14 (2.6H, m), 1.70-1.56
(1H, m); 13C NMR (128 MHz, DEPT, CDCl3, ppm) δ 156.2 s, 155.3
s, 138.3 s, 138.2 s, 138.0 s, 137.9 s, 137.2 s, 136.8 s, 136.6 s, 131.7 d,
130.9 d, 128.4-127.4 (11×) d, 74.4 d, 74.2 t, 74 d, 1t, 72.3 t, 72.1 t,
71.0 t, 70.9 t, 67.1 t, 66.8 t, 49.3 t, 49.2 t, 46.5 t, 45.7 t, 34.5 t, 33.1
t, 29.6 t, 29.4 t.
N-Methylazocine 56. A solution of 50a (331 mg, 0.702 mmol) in
THF (35 mL) was added dropwise to a suspension of LiAlH4 (120
mg, 3.16 mmol) in dry THF (25 mL) at 0 °C under nitrogen. The
mixture was warmed to room temperature and stirred for 8 h. The
reaction was quenched with EtOAc (50 mL), followed by Glauber’s
salt (Na2SO4‚10H2O), and the mixture was stirred vigorously overnight.
The mixture was filtered through Celite, and the pad was rinsed
thoroughly with EtOAc. After removal of solvent (aspirator), the
residue was purified by flash chromatography on silica gel-NH3 (see
General Experimental), 2:3 hexane/CHCl3 eluent, to yield the N-
methylazocine 56 as a clear colorless oil (238 mg, 0.676 mmol, 96%
yield): analytical TLC on silica gel, 1:9 methanol/CHCl3, Rf ) 0.50;
M+ calcd for C23H30NO2 352.2277, found [HRFAB] m/e ) 352.2287,
error ) 3 ppm, base peak ) 352 amu; IR (neat, cm-1) 3004 dC-H,
2791 NCH3, 1090 C-O; 500 MHz NMR (CDCl3, ppm) δ 7.40-7.27
(10H, m), 5.68 (1H, d, J ) 7.2 Hz), 4.61 (1H, d, J ) 11.7 Hz), 4.50
(2H, AB q, J ) 11.9 Hz), 4.45 (1H, d, J ) 11.7 Hz), 4.03 (1H, d, J )
12.2 Hz), 3.94 (1H, d, J ) 12.2 Hz), 2.67 (1H, ddd, J ) 3.9, 3.9, 12.4
Hz), 2.50-2.42 (2H, m), 2.57 (1H, ddd, J ) 4.4, 4.4, 14.6 Hz), 2.48-
2.41 (1H, m), 2.41-2.32 (1H, m), 2.37 (3H, s), 2.28-2.07 (2H, m),
1.47 (1H, dddd, J ) 3.8, 3.9, 11.4, 12.1 Hz); 13C NMR (125 MHz,
DEPT, CDCl3, ppm) δ 138.7 s, 138.3 s, 137.5 s, 130.7 d, 128.3 d,
128.3 d, 127.7 d, 127.6 d, 127.5 d, 127.4 d, 75.8 d, 74.0 t, 71.9 t, 71.0
t, 58.1 t, 56.9 q, 53.6 t, 36.3 t, 29.7 t.
TMEDA (60 µL, 0.40 mmol) and stirred at -78 °C for 10 min. A
solution of the N-methylazocine 56 (113 mg, 0.321 mmol) in THF (20
mL) was added dropwise, and the mixture was stirred at -70 °C for
42 h. The reaction was warmed to room temperature, quenched with
saturated aqueous NaHSO3, stirred at room temperature for 2 h, and
then refluxed under nitrogen for 4 h. The mixture was extracted with
EtOAc, and the organic layer was washed with Na2CO3 and brine. The
combined aqueous layers were taken to pH > 9 and extracted with
CHCl3. The organic layers were dried over Na2SO4/K2CO3. After
removal of solvent (aspirator), the residue was purified by flash
chromatography on silica gel (NH3, CHCl3 eluent), to give the diol as
a clear colorless oil, 30:1 mixture of diastereomers (99 mg, 0.26 mmol,
81%): M+ calcd for C23H32NO4 386.2331, found [HRFAB] m/e )
386.2336, error ) 1 ppm, base peak ) 386 amu; IR (neat, cm-1) 3493
O-H, 2857 C-H, 1069 C-O; 500 MHz NMR (CDCl3, ppm) δ 7.40-
7.20 (10H, m), 4.57 (2H, s), 4.52 (2H, AB q, ∆υ ) 0.0266 ppm, J )
11.7 Hz), 4.01 (1H, d, J ) 6.5 Hz), 3.95 (1H, ddd, J ) 3.9, 6.5, 7.3
Hz), 3.61 (1H, d, J ) 9.7 Hz), 3.56 (1H, d, J ) 9.7 Hz), 2.83 (1H,
ddd, J ) 5.2, 12.4, 12.5 Hz), 2.51 (1H, ddd, J ) 5.7, 8.7, 13.5 Hz),
2.58 (1H, ddd, J ) 5.1, 6.0, 13.5 Hz), 2.33 (3H, s), 2.34 (1H, ddd, J
) 2.8, 5.4, 12.5 Hz), 2.14 (1H, ddd, J ) 5.4, 12.4, 15.4 Hz), 2.04 (1H,
dddd, J ) 3.9, 5.7, 6.0, 15.3 Hz), 1.98 (1H, dddd, J ) 5.1, 7.3, 8.7,
15.3 Hz), 1.81 (1H, ddd, J ) 2.8, 5.2, 15.4 Hz); 13C NMR (125 MHz,
DEPT, CDCl3, ppm) δ 138.6 s, 138.2 s, 128.3 d, 128.2 d, 127.6 d,
127.5 d, 127.5 d, 127.3 d, 79.9 d, 76.4 t, 74.5 s, 73.5 t, 71.8 t, 52.2 t,
51.7 t, 46.5 q, 33.2 t, 30.7 t.
Ketol 60/61. A vigorously stirred solution of diol 58 (53 mg, 0.14
mmol) in tert-butyl alcohol (1 mL) and water (1.5 mL) was treated
with CaO (155 mg), OsO4 (0.5 mL, 2.8 mg/mL in tert-butyl alcohol)
and K3Fe(CN)6 (350 mg, 1.06 mmol, all at once). The resulting bright
yellow suspension was stirred in the dark for 55 min. The mixture
was extracted with THF [Caution! OsO4 present]. The combined THF
extracts were quenched with a mixture of NaHSO3 and Na2SO3 (1.5 g
ea) in water (5 mL) and stirred at room temperature overnight. The
reaction was worked up by the same procedure as used for the previous
step (diol 58). The residue was purified by flash chromatography on
silica gel-NH3, 1:1 Hex/CHCl3 (saturated), CHCl3 (saturated), gradient
10% (0.4%) to 33% (1.3%) THF/CHCl3 (%NH4OH) eluent, to give
the ketol 60/61 as a colorless oil (27 mg, 0.071 mmol, 53% yield):
M+ calcd for C23H30NO4 384.2175, found [HRFAB] m/e ) 384.2178,
error ) 1 ppm, base peak ) 384 amu; IR (neat, cm-1) 3403 O-H,
1676 CdO, 1095 C-O. The NMR spectrum shows both acid- and
temperature-dependent phenomena. The NMR spectrum of the free
base is fluxional at room temperature and sharpens at elevated
temperature (70 °C, CD3CN). Free base: 500 MHz NMR (CD3CN,
70 °C, ppm) δ 7.40-7.23 (10H, m), 4.54 (1H, d, J ) 11.6 Hz), 4.44
(1H, d, J ) 11.9 Hz), 4.46 (1H, d, J ) 11.6 Hz), 4.48 (1H, d, J ) 11.9
Hz), 4.12 (1H, dd, J ) 3.5, 3.5 Hz), 3.89 (1H, d, J ) 9.6 Hz), 3.56
(1H, d, J ) 9.6 Hz), 2.74 (1H, ddd, J ) 3.5, 12.9, 13.2 Hz), 2.63-
2.50 (3H, m), 2.48 (1H, ddd, J ) 2.6, 4.8, 13.2 Hz), 2.25 (1H, dddd,
J ) 3.5, 4.8, 12.9, 14.1 Hz), 2.03 (3H, s), 2.06 (1H, dddd, J ) 2.6,
3.5, 3.5, 14.1 Hz), 1.73-1.67 (1H, m); 13C NMR (125 MHz, DEPT,
CD3CN, 70 °C, ppm) δ 200.4 s, 140.3 s, 139.3 s, 129.5 d, 129.4 d,
129.0 d, 128.9 d, 128.8 d, 128.7 d, 128.6 d, 84.1 br d, 80.3 s, 77.6 t,
74.5 t, 72.7 t, 51.3 t, 50.6 t, 42.2 q, 37.4 br t, 36.6 t. Protonated (63):
500 MHz NMR (CD3OD/HCl, ppm) δ 7.40-7.20 (10H, m), 4.64 (1H,
d, J ) 12.4 Hz), 4.57 (1H, d, J ) 12.4 Hz), 4.50 (1H, d, J ) 11.9 Hz),
4.56 (1H, d, J ) 11.9 Hz), 4.38 (1H, dd, J ) 2.6, 5.4 Hz), 3.75-3.65
(2H, m), 3.69 (1H, d, J ) 9.7 Hz), 3.52 (1H, d, J ) 9.7 Hz), 3.55 (1H,
ddd, J ) 2.0, 7.7, 11.8 Hz), 3.45 (1H, ddd, J ) 7.3, 9.0, 11.8 Hz),
2.89 (3H, s), 2.55-2.48 (1H, m), 2.30 (1H, ddd, J ) 7.8, 12.0, 13.8
Hz), 2.18-2.10 (1H, m), 2.06 (1H, ddd, J ) 1.0, 6.0, 14.1 Hz); 13C
NMR (125 MHz, DEPT, CD3OD/HCl, ppm) δ 139.3 s, 138.4 s, 129.5
d, 129.4 d, 129.4 d, 129.2 d, 129.0 d, 128.8 d, 117.5 s, 81.7 s, 74.8 d,
73.9 t, 73.8 t, 63.3 t, 62.6 t, 47.0 q, 36.1 t, 31.2 t.
Dibenzylotonecine 68/69. Ketol 60/61 (13 mg, 0.033 mmol) was
treated with excess Burgess reagent 223 (29 mg, recrystallized from
toluene, 0.12 mmol) and C6D6 (0.25 mL). After 15 min, the mixture
was concentrated to dryness (1 Torr), redissolved in C6D6 (0.25 mL),
and heated at 36 °C (water bath temperature) for 38 h. The reaction
mixture was diluted with C6D6 (1 mL) and added dropwise to refluxing
N-Methylazocine Diol 58. A solution of OsO4 (95.0 mg, 0.374
mmol) in THF (5 mL) at -78 °C under nitrogen was treated with

3622 J. Am. Chem. Soc., Vol. 120, No. 15, 1998
Vedejs et al.
benzene (15 mL). The tube was rinsed with C6H6 (2 × 1 mL) and
CHCl3 (1 mL). The mixture was refluxed for 10 min and quenched at
reflux with Na2CO3 (saturated aqueous). After aqueous workup (CHCl3,
Na2CO3), drying (Na2SO4), and evaporation, the crude product was
dissolved in wet THF (15 mL, 6.5 µL H2O + 5 µL additional H2O
after 20 min reflux), refluxed under nitrogen for 30 min, and quenched
at reflux with Na2CO3 (satd aq). Aqueous workup (CHCl3, Na2CO3)
and purification by flash chromatography on silica gel-NH3 (1 × 4
cm), 20%(0.6) to 100%(4) THF/CHCl3(% NH4OH) eluent, gave
dibenzylotonecine 68/69 as a clear colorless oil (6.0 mg, 0.0170 mmol,
50% yield): M+ calcd for C23H28NO3 366.2069, found [HRFAB] m/e
) 366.2067, error ) 1 ppm, base peak ) 366 amu; IR (neat, cm-1),
2806 NCH3, 1665 CdO, 1099 C-O; 500 MHz NMR (CDCl3, ppm) δ
7.38-7.23 (10H, m), 5.74 (1H, s), 4.87 (1H, br d, J ) 12.5 Hz), 4.56
(1H, d, J ) 12.5 Hz), 4.47 (2H, s), 4.19 (1H, dddd, J ) 1.8, 1.8, 1.8,
11.8 Hz), 4.05 (1H, d, J ) 11.8 Hz), 3.87 (1H, dd, J ) 3.9, 3.9 Hz),
3.20 (1H, d, J ) 18.0 Hz), 3.11 (1H, d, J ) 18.0 Hz), 2.83 (1H, ddd,
J ) 4.2, 11.7, 12.2 Hz), 2.62 (1H, ddd, J ) 3.9, 4.7, 12.2 Hz), 2.29
(3H, s), 2.15 (1H, dddd, J ) 3.9, 4.7, 11.4, 14.2 Hz), 1.96 (1H, dddd,
J ) 3.9, 3.9, 4.2, 14.2 Hz); 13C NMR (125 MHz, DEPT, CD3OD/HCl,
ppm) δ 138.5 s, 138.1 s, 136.9 s, 129.6 d, 129.5 d, 129.4 d, 129.3 d,
129.3 d, 129.2 d, 129.1 d, 129.0 d, 128.9 d, 128.6 d, 126.1 d, 126.1 d,
124.7 d, 118.6 s, 84.1 s, 77.8 d, 74.0 t, 74.0 t, 73.6 t, 73.2 t, 71.2 t,
70.4 t, 66.0 t, 65.9 t, 64.0 t, 63.5 t, 47.6 q, 47.3 q, 29.1 t, 29.0 t.
Otonecine (1). A solution of dibenzylotonecine 68/69 (9.7 mg) and
thioanisole (150 µL) in dry CD3CN (0.5 mL) at 0 °C was treated with
TMSOTf (30 µL). The mixture was shaken and allowed to stand at 0
°C for 30 min. The reaction was quenched with H2O (50 µL) and
concentrated under vacuum (aspirator, then <1 Torr). The mixture
was triturated with CHCl3, and the CHCl3 layers were filtered. The
product was recovered from the filter with CH3CN. Ion exchange
chromatography through a Bio-Rex 5 column (0.4 cm3, basic form),
onto a weakly acid column (Bio-Rex 70, 0.5 mL, H+ form) in 1:1
methanol/water (35 mL), followed by elution of the acid column with
a gradient of AcOH (0.6 mL) in methanol/water (5 mL), gave
otonecine‚HOAc salt as a pale yellow oil (4.6 mg). The product was
filtered through a basic ion-exchange resin (Bio-Rex 5, 0.5 cm3, basic
form) in 1:1 methanol/water, to yield otonecine as a pale yellow oil
(3.5 mg, ca. 73-77%). Minor signals of a contaminant were detected
(δ 5.9-5.8, 4.4-4.2, 3.70-3.55, 2.5-2.3; estd 5-10% by comparison
of the downfield signals). The other NMR signals in CD3OD (room
temperature) matched a spectrum of natural otonecine at 270 MHz
provided by Prof. K. Yamada according to chemical shift, but with
improved resolution due to the higher field. For 1: LRFAB 186.1;
500 MHz NMR (CD3OD, 60 °C, ppm) δ 5.71 (1H, br s), 4.11 (1H,
dddd, J ) 1.6, 1.6, 1.6, 14.1 Hz), 4.09 (1H, dddd, J ) 1.8, 1.8, 1.8,
14.1 Hz), 3.92 (1H, dd, J ) 3.2, 3.8 Hz), 3.50-3.0 (2H, m), 2.97 (1H,
ddd, J ) 1.6, 5.8, 12.2 Hz), 2.86 (1H, ddd, J ) 4.8, 12.2, 12.5 Hz),
2.29 (3H, s), 2.13 (1H, dddd, J ) 3.8, 5.8, 12.5, 14.4 Hz), 1.98 (1H,
dddd, J ) 1.6, 3.2, 4.8, 14.4 Hz).
Acknowledgment. We are indebted to Professor K. Yamada
1
of Nagoya University for kindly providing us with H NMR,
IR, and mass spectra of natural otonecine. This work was
supported by NIH (CA17918; postdoctoral fellowship GM
15045 to P.G.G.).
Supporting Information Available: Experimental details
and characterization for 10c, 11, 13-16a, 20, 21, 31, 32, 38b,
40b, 45, 47a, 48a, 47b, 48b, 50b, 51, 55, and 52 (13 pages,
print/PDF). See any current masthead page for ordering
information and Web access instructions.
JA973464E