Scheme 4. Completion of the Synthesis of 2
Figure 2. Steric hindrance in 2 relative to 13.
our cyclizations. After considerable experimentation, we
discovered that TMSOTf (2 equiv) in CH2Cl2 (-10 °C, 15
min) facilitated cyclization and minimized enol ether hy-
drolysis, delivering 13 in 50% yield along with a minor
amount of 14 (11%). The necessity for an excess of TMSOTf
is presumably due to the abundance of Lewis basic functional
groups in 3.
carbon quaternary stereocenter vicinal to another congested
quaternary carbon.16
Surprisingly, all spectral data (1H NMR, 13C NMR, COSY,
HMQC) for 13 are consistent with the enol tautomer shown
rather than the expected keto form illustrated for 2 (Scheme
1). Molecular modeling (MacSpartan Pro, semiempirical
PM3) suggests that enol 13 is more stable than ketone 2 by
2.9 kcal/mol. In contrast, 14 and acutumine both exist as
the keto tautomers in solution.18 We believe that two
structural features are responsible for the stability of 13
relative to 2. First, the enol moiety in 13 is stabilized by an
intramolecular hydrogen bond to the neighboring methoxy
group. Second, the steric hindrance between a ketone
R-hydrogen in 2 and one of the geminal methyl groups
destabilizes this structure with respect to 13 (Figure 2).
Compound 14 possesses the same steric hindrance as 2;
however, it already contains an enol that presumably forms
an intramolecular hydrogen bond to the methoxy group.
Thus, the bis-enol tautomer of 14 cannot derive any
additional benefit from intramolecular hydrogen bonding,
and the molecule exists in solution as shown in Scheme 4.
In fact, PM3 calculations indicate that 14 is 6.5 kcal/mol
more stable than the aforementioned bis-enol. The enol of 1
could be stabilized by intramolecular hydrogen bonding.
Nevertheless, the presence of the spirocyclic cyclopentenone
instead of the geminal dimethyl groups of 13 affects the
geometry at the spirocyclic carbon, presumably attenuating
the strain of the keto tautomer and causing it to be more
stable than the enol. Again, molecular modeling supports
this analysis, as the PM3-minimized structure of 1 is 2.0
kcal/mol lower in energy than its enol tautomer.
To complete the tricyclic core of 1, it remained for us to
construct the pyrrolidine ring via C-N bond formation. Thus,
the terminal olefin of 12 was converted into a secondary
amine by a two-step sequence of ozonolysis followed by
reductive amination with MeNH2 (Scheme 4). While the
reductive amination proceeded without complication, the
ozonolysis was challenging as a result of competitive
oxidation of the tetrasubstituted alkene. The most efficient
protocol entailed conducting the ozonolysis in EtOAc and
halting the reaction prior to completion.17 The resulting
mixture of 12 and the derived aldehyde (ca. 1:1) was free of
overoxidized compounds and could be subjected to reductive
amination without purification. Subsequent chromatography
afforded pure 3 (37% over 2 steps; 73% based on recovered
12) and 12 (49%), which could be resubjected to ozonolysis.
The alternative method of dihydroxylation-oxidative cleav-
age was plagued by the sluggishness of the oxidative
cleavage step. The diol obtained from dihydroxylation of the
terminal alkene of 12 was unaffected by exposure to NaIO4
and reacted too slowly with Pb(OAc)4 for this route to be of
value.
We planned to convert 3 into our target structure 2 via
Lewis acid promoted cyclization of the secondary amine onto
the R,â-unsaturated ketal. Matsumoto has recently disclosed
a similar transformation;5 however, the substrate in this study
was devoid of the sensitive enol ether present in 3. Predict-
ably, the enol ether was difficult to retain, as we obtained
varying amounts of demethylated pyrrolidine 14 in all of
In conclusion, we have synthesized tricyclic compound
13, representative of the core of the bioactive natural product
acutumine. In the context of this synthesis, we developed a
strategy for the construction of an all-carbon quaternary
center and an adjacent amine-bearing quaternary carbon that
relies on an anionic oxy-Cope rearrangement followed by a
Lewis acid mediated Michael-type cyclization. Additionally,
(15) Allylmagnesium chloride was purchased from Aldrich, whereas
allylmagnesium bromide was synthesized in our laboratory. Thus, we believe
that unreacted Mg present in the homemade Grignard reagent solution may
reduce 4 to the corresponding radical anion. Loss of methoxide and hydrogen
atom abstraction by the incipient aryloxy radical from THF or another
adventitious hydrogen atom source would provide 5.
(16) For an oxy-Cope/Claisen/ene reaction cascade that results in the
formation of vicinal quaternary stereocenters, see: (a) Sauer, E. L. O.;
Barriault, L. Org. Lett. 2004, 6, 3329. (b) Sauer, E. L. O.; Barriault, L. J.
Am. Chem. Soc. 2004, 126, 8569.
(17) If the reaction was allowed to proceed to more than 50% conversion,
overoxidized byproducts began to emerge. Ozonolyses conducted in CH2-
Cl2, MeOH, or mixtures of these solvents were characterized by the
predominance of such byproducts. We plan to further optimize this
transformation on intermediates relevant to the total synthesis rather than
in this model system.
(18) (a) Sugimoto, Y.; Babiker, H. A. A.; Saisho, T.; Furumoto, T.;
Inanaga, S.; Kato, M. J. Org. Chem. 2001, 66, 3299. (b) Sugimoto, Y.;
Inanaga, S.; Kato, M.; Shimizu, T.; Hakoshima, T.; Isogai, A. Phytochem-
istry 1998, 49, 1293. The NMR spectra of 1 were acquired in pyridine-d5.
Our data were obtained in benzene-d6; however, a 1H NMR spectrum of
13 acquired in pyridine-d5 showed only the enol form.
Org. Lett., Vol. 7, No. 6, 2005
1091
Reeder, Matthew D.
