A R T I C L E S
Yang et al.
possible solutions based on regiodirected epoxide ring cleavages
or on conventional allylic oxidation schemes proved to no avail,
it was made clear that this challenge had to be resolved in a
different manner. In 1948, Treibs briefly described his observa-
tions involving the allylic acetoxylation of olefins upon heating
with mercuric acetate in acetic acid.32 This process has been
adopted only infrequently since that time,33-35 seemingly
because the conditions have generally come to be regarded as
somewhat too forcing. At the mechanistic level, the mercuric
reagent presumably attacks the π bond to generate a mer-
curinium ion, the rearrangement of which to an allylic organo-
mercurial sets the stage for solvolysis with formation of Hg(0)
and one or both epimeric allylic acetates.33,36 On this basis, it
was anticipated that heating 43 with mercuric trifluoroacetate
in benzene followed by stirring with aqueous sodium bicarbonate
might possibly lead stereoselectively to 44 in a simple one-pot
operation. Indeed, these conditions led smoothly to the formation
of 44 in an isolated yield of 77%. We attribute the remarkably
good regioselectivity of this transformation to the unsymmetrical
nature of the putative mercurinium ion, whose biased “open
character” places a modicum of positive charge on the methyl-
substituted carbon (C6) and induces more facile deprotonation
at C5 as allylic mercurial character develops.37 No allylic alcohol
having an exocyclic double bond was noted, a feature that hints
at the possible operation of thermodynamic control.
sion of 47 to 48 was complete in 3 h (88%). This reaction time
is significantly longer than the 5 min required to perform the
quantitative saponification of 44 under identical conditions.41
This kinetic inequality can be traced to a more elevated level
of ring strain in 44. The 2-fold oxidation of 48 to jatrophatrione
(1) was mediated by IBX in dimethyl sulfoxide (DMSO).22 The
synthetic material produced in this manner exhibited a 500 MHz
1H NMR spectrum fully consistent with that of the natural
sample recorded earlier at lower field strength.42 However, the
13C NMR data originally reported differs from our measurements
made at 100 and 125 MHz in a systematic way (F ) 0.998).
As detailed elsewhere,43 a minor complication associated with
the dwell clock in their Fourier transform spectrometer of those
involved in the isolation of 1 is likely responsible for this
phenomenon.
Notwithstanding, we can confidently assert that the total
synthesis of racemic jatrophatrione has indeed been accom-
plished since the treatment of 1 prepared as detailed above with
m-chloroperbenzoic acid in CH2Cl2 at room temperature leads
unmistakably to citlalitrione (2, 22%) alongside a greater
proportion of the unnatural R-epoxide 49 (67%). In this instance,
1
the H and 13C NMR spectra of synthetic 2 were identical to
those registered for the natural sample, whose structural features
had previously been secured by means of X-ray crystallography.
Summary
With the preparation of 44 accomplished, oxidation with tetra-
n-propylammonium perruthenate (TPAP)38 was undertaken to
give 45, the benzyloxy group in which was cleaved with boron
trichloride in CH2Cl2.39 The last major hurdle, the regiodirected
dehydration of 46, now had to be addressed. Two considerations
held central importance as we assessed this step. First, a syn
elimination of the C3 hydroxyl was mandated in either of the
two regiochemical options. Also, pilot studies carried out on
less advanced analogues of 46 revealed a strong kinetic
preference for avoidance of the site of ring fusion. Since these
model systems lacked the C5-C6 double bond, our working
proposition was that the enone segment might well induce a
higher level of acidification at H4 relative to H2 despite the
nonplanarity of this chromophore.
We initially engaged 46 in xanthate formation, only to
discover that this hydroxy ketone is sensitive to basic reagents.
Its vinylogous aldol character is likely responsible for this
behavior. In contrast, the targeted dehydration could be brought
about without event simply by heating 46 with thiocarbonyl-
diimidazole in 1,2-dichlorobenzene.40 This tactic lent itself to
the formation of a two-component product mixture, chromato-
graphic separation of which furnished in 37% yield the target
dienone 47 and 10% of the ∆2,3-regioisomer.
The first total syntheses of two representative [5.9.5] tricyclic
diterpenoids have been accomplished. The several noteworthy
facets of this venture include an anionic oxy-Cope rearrangement
that conveniently sets in place a tetracyclic structural framework
(viz., 14) capable of ready chemical modification. With arrival
at 17 only seven steps later, it becomes possible to hydroxylate
C12 and C14 via intramolecular hydrosilylation and to protect
this pair of R-OHs as a cyclic carbonate. Application of the
Treibs reaction to 43 constituted a particularly productive step,
with the resulting allylic alcohol 44 serving as a reliable platform
for regiocontrolled dehydration as defined by 46 f 47. Finally,
the 1,3-diketone segment is introduced to deliver jatrophatrione
(1), subsequent peracid oxidation of which gives (()-citlalitrione
(2). The length of the linear sequence from the point of
convergence is 20 steps. We anticipate that the successful route
to 1 and 2 detailed herein will significantly facilitate the rational
synthesis of other diterpenoids of this class.
Acknowledgment. We thank Professor Robert Bates (The
University of Arizona) for his attempts to locate a sample and/
or the spectra of jatrophatrione and Dr. Howard Williams (Texas
1
A&M University) for the H and 13C NMR spectra of citlal-
itrione.
We could now undertake hydrolysis of the cyclic carbonate
ring. With K2CO3 in methanol at room temperature, the conver-
Supporting Information Available: Complete experimental
procedures and spectral data for all previously unreported
compounds described herein, including copies of the 1H NMR
spectra of synthetic jatrophatrione as well as natural and
synthetic citlalitrione (PDF). This material is available free of
(31) Tamao, K.; Ishida, N.; Kumada, M. J. Org. Chem. 1983, 48, 2120.
(32) Treibs, W. Naturwissenschaften 1948, 35, 125.
(33) Arzoumanian, H.; Metzger, J. Synthesis 1971, 527.
(34) Massiot, G.; Husson, H.-P.; Potier, P. Synthesis 1974, 722.
(35) Reischl, W.; Kalchhauser, H. Tetrahedron Lett. 1992, 33, 2451.
(36) Wiberg, K. B.; Nielsen, S. D. J. Org. Chem. 1964, 29, 3353.
(37) In THF, CH2Cl2, or HMPA, the allylic mercury intermediate was the only
product isolated. The generation of this species appears to be rapid in all
media. The ensuing step in which Hg(II) is reduced to Hg(0) is believed to
be the rate-determining step. If so, the role of nonpolar benzene is to
stabilize the transition state and accelerate the overall rate.
(38) (a) Griffith, W. P.; Ley, S. V. Aldrichim. Acta 1990, 23, 13. (b) Ley, S.
V.; Norman, J.; Griffith, W. P.; Marsden, S. P. Synthesis 1994, 639.
(39) Williams, D. R.; Brown, D. C.; Benbow, J. W. J. Am. Chem. Soc. 1989,
111, 1923.
JA021177R
(40) Williams, D. R.; Coleman, P. J.; Henry, S. S. J. Am. Chem. Soc. 1993,
115, 11654.
(41) Yang, J. Unpublished observations in this laboratory.
(42) In response to our request for a copy of the 1H NMR spectrum or a sample
of natural jatrophatrione, Professor Robert Bates responded (letter dated
December 15, 1995) that neither could be located at that time.
(43) Footnote 15 of ref 4.
9
1574 J. AM. CHEM. SOC. VOL. 125, NO. 6, 2003