α,β-epoxy vinyl triflates in Pd-catalyzed reactions

Article abstract of DOI:10.1021/ol000064e

(matrix presented) Reactions of steroidal αβ-epoxy vinyl triflates in Pd-catalyzed reactions are described. Oxidative insertion of Pd⁰ into the C-O bond, giving vinylpalladium 12, is faster than formation of the π-allyl derivative from the vinyl epoxide. Although 12 can be trapped under certain conditions, it eventually rearranges to palladium alkoxide 14, which is in equilibrium with 15 and/or 10.

Full text of DOI:10.1021/ol000064e

ORGANIC
LETTERS
2000
Vol. 2, No. 12
1709-1712
r,â-Epoxy Vinyl Triflates in
Pd-Catalyzed Reactions
Kana Yamamoto and Clayton H. Heathcock*
Department of Chemistry, UniVersity of California, Berkeley,
Berkeley, California, 94720-1460
chh@steroid.cchem.berkeley.edu
Received March 23, 2000
ABSTRACT
Reactions of steroidal r,â-epoxy vinyl triflates in Pd-catalyzed reactions are described. Oxidative insertion of Pd0 into the C−O bond, giving
vinylpalladium 12, is faster than formation of the π-allyl derivative from the vinyl epoxide. Although 12 can be trapped under certain conditions,
it eventually rearranges to palladium alkoxide 14, which is in equilibrium with 15 and/or 10.
As part of a project directed toward the synthesis of
cephalostatins,1 we wanted to prepare lactone 1 from epoxide
Scheme 1
2a (Scheme 1).2 Lactone 1 can be envisioned to arise from
either 3 or 4a. To this end, triflate 5a appeared to be an
attractive intermediate since it would allow access to both
compounds by Pd-mediated carbonylation or reduction.3
Conversion of 2a to vinyl triflate 5a was achieved by the
use of Comins’ procedure.4 Triflate 5a possesses two
possible sites for reaction with Pd0 (insertion into either the
vinyl triflate C-O bond or the allyl epoxide C-O bond),
and it was uncertain which functionality would be more
reactive.5 In addition, the desired products (3 or 4a) could
(1) Petit, G. R. Y.; Williams, M. D.; Boyd, M. M. R. J. Nat. Prod. 1998,
61, 953 and references therein.
(2) Epoxide 2a was synthesized from the commercially available steroid,
hecogenin in seven steps. The detailed experimental procedure is shown in
the Supporting Information.
(3) Methods for the Pd-mediated reduction: (a) With Et3SiH, Kotsuki,
H.; Datta, P. K.; Hayakawa, H.; Suenaga, H. Synthesis 1995, 1349. (b)
With Bu3N/HCO2H, Cacchi, S.; Morera, E.; Ortar, G. Tetrahedron Lett.
1984, 25, 42, 4821. (c) With Bu3SnH, see ref 7. (d) With Et2NH/BH3,
Lipshutz, B, H,; Buzard, D. J.; Vivian, W. Tetrahedron Lett. 1999, 40, 6871.
(4) (a) Comins, D. L.; Dehghani, A.; Foti, C. J.; Joseph, S. P. Org. Synth.
1997, 74, 77. (b) Comins, D. L.; Dehghani, A. Tetrahedron Lett. 1992, 33,
6299.
(5) It was reported that in the case of 1-acetoxy-2-bromo-2-alkenes, the
bromo substituent dramatically reduces the reactivity of the olefin in Pd-
catalyzed substitutions, and that coupling between bromoalkenes and
terminal alkynes proceeds without side reactions with acetoxy substituents.
(a) Nwokogu, G. C. Tetrahedron Lett. 1984, 25, 31, 3263. (b) Nwokogu,
also undergo Pd0 insertion to give π-allylpalladium com-
plexes. However, in any event, it was found that carbonyl-
ation of 5a provides ester 3 in good yield.6
G. C. J. Org. Chem. 1985, 50, 3900.
10.1021/ol000064e CCC: $19.00 © 2000 American Chemical Society
Published on Web 05/17/2000

Elaboration of ester 3 to lactone 1 was found to be
problematic, and we therefore turned our attention to the
alternative route. Initial attempts to reduce the triflate 5a with
Bu3SnH7 following Stille’s procedure8 gave a number of
products depending on reaction conditions. While we were
able to obtain moderate amounts of the desired allyl epoxide
4a,9 the interesting structures of other isolated products
prompted us to further investigate these reactions. Triflate
5b, prepared from the known ketone 6,10 was also utilized
in these studies (Scheme 2).
as a mixture of C16 epimers (eq 3).17 A small amount of
allenone 10 was also obtained when 5a was used in this
reaction. These results suggest that allenic alcohol 7 is formed
by direct rearrangement of the vinylpalladium intermediate.18
(13) It is speculated that the steric repulsion between C12 TBS group of
14a suppresses its formation, resulting in the lower ratio.
Scheme 2
(14) The C17 configuration of the homoallyl alcohol 8a was determined
as follows. Hydrogenation of 8a gave saturated alcohol 16, which was
oxidized to ketone 17. Allyl alcohol 9a was oxidized to enone 18, which
was hydrogenated to obtain 17. Since it is known that the ∆17(20) olefin in
steroids is hydrogenated from the R-face, the C17 configuration in 17 is R,
as shown.
Some representative examples of the reactions are shown
in Scheme 3. Reaction of 5a with Pd(OAc)2, PPh3, LiCl,
and 1 equiv of Bu3SnH gave a mixture of vinyl epoxide 4a
and allenic alcohol 7a along with recovered starting mate-
rial11,12 in a ratio of 4:1:2 (eq 1). Subjecting 5b to the same
reaction conditions also gave the corresponding epoxide 4b,
allenic alcohol 7b, and starting material, but in a ratio of
2:2:1.13
(15) The high regioselectivity in the reduction of 4a compared to 4b
can be explained by the relative stability of the isomeric π-allyl palladium
species in each case. It is likely that 13 is more stable than 12 because of
less steric repulsion between the steroidal core. However, sterics between
Pd ligand and bulky TBS protecting group at C12 (see 13a) could invert
the relative stability. Transmetalation and reductive elimination gives the
reduced compounds, of which the ratio should reflect the energy difference
between 12 and 13.
We speculated that 7 might have arisen from allyl epoxide
4 by oxidative addition and subsequent â-elimination.
However, resubjecting 4 to the reaction conditions resulted
in reduction instead of â-elimination, giving a mixture of
homoallyl alcohol 814 and allyl alcohol 9 as a mixture of E
and Z isomers (eq 2).15
On the other hand, treatment of 5 with a stoichiometric
amount of Pd0 led to the formation16 of 7, but surprisingly,
(6) Similarly, it was recently reported that Pd-catalyzed carbonylation
of 1-acetoxy-2-bromo-2-alkenes proceeds without involvement of the allylic
ester. Trost, B. M.; Oslob, J. D. J. Am. Chem. Soc. 1999, 121, 3057.
(7) Alternative methods gave unsatisfactory results: Et3SiH procedure3a
resulted in very low conversion. Bu3N/HCO2H procedure3b gave a mixture
of unidentifiable products, presumably due to acid sensitivity of vinyl
epoxide.
(8) (a) Scott, W. J.; Stille, J. K. J. Am. Chem. Soc. 1986, 108, 3033. For
leading references for the Stille reaction, see (b) Stille, J. K. Angew. Chem.,
Int. Ed. Engl. 1986, 25, 508. (c) Farina, V.; Krishnamurthy, V.; Scott, W.
J. The Stille Reaction; John Wiley and Sons, Inc.: New York, 1998.
(9) Because of the difficulty in separation from starting material, 4a was
not isolated in most cases. An alternative way to make this compound is
described in the Supporting Information.
(16) Addition of LiCl resulted in no reaction, presumably because of
the stablization of PdII by chloride. Tsuji, J. Palladium Reagents and
Catalysis, InnoVations in Organic Synthesis; John Wiley and Sons, Inc.:
New York, 1995; pp 19-20.
(17) To determine the stereochemistry of isomeric alcohols 7a, the
mixture of isomers was oxidized to allenone 14 and reduced with NaBH4.
It is assumed that the major isomer is the one with the â-configuration at
C16, 7a.
(10) (a) Julian, P. L.; Meyer, E. W.; Ryden, I. J. Am. Chem. Soc. 1950,
72, 367. (b) Julian, P. L.; Karpel, W. J. J. Am. Chem. Soc. 1950, 72, 362.
(c) Julian, P. L.; Meyer, E. W.; Karpel, W. J.; Waller, I. R. J. Am. Chem.
Soc. 1950, 72, 5145.
(11) This reaction does not go to completion with 1 equiv of Bu3SnH.
It is known that triorganostannyl hydrides are converted into ditins by Pd
catalyst,12 although Pd-catalyzed reduction has also been described. In our
case, addition of more than 1 equiv of the hydride resulted in a mixture of
products resulting from over-reduction of the allyl epoxide, along with 4,
7, and 5. Addition of additional Pd catalyst had no effect on the reaction.
(12) Mitchell, T. N. Synthesis 1992, 9, 803.
(18) It is also possible that vinylpalladium species transmetalates with
(Bu3Sn)2 (generated from 2 equiv of Bu3SnH)12 to give vinyltin species
before the rearrangement. However, the use of (Bu3Sn)2 in place of Bu3-
SnH gave different products depending on substrates.
1710
Org. Lett., Vol. 2, No. 12, 2000

Scheme 3
When 5 was subjected to salt-free conditions, the steroidal
to the vinyl triflate moiety proceeds faster than that of vinyl
epoxides.6 Presumably, in the presence of LiCl, 12 quickly
exchanges ligand to give 13.20 The transmetalation step then
becomes relatively faster than the rearrangement,21 resulting
in predominant formation of 4 upon reductive elimination
(eq 1). Under salt-free conditions (eq 4), the transmetalation
proceeds slower and rearrangement to alkoxide 14 becomes
competitive. It is of note that carbonylation of 5a (Scheme
D ring fragmented to give aldehyde 11 along with trace of
7 (eq 4). Overall, this reaction (from epoxy-ketone to alkyne-
aldehyde) is equivalent to Eschenmoser’s fragmentation.19
The foregoing experimental results can be explained as
follows (Figure 1). As mentioned, there are three possible
reactive sites (vinyl triflate of 5 or allyl epoxide of 4 and 5)
for Pd0 under these conditions. However, oxidative addition
Figure 1.
Org. Lett., Vol. 2, No. 12, 2000
1711

1) under salt-free conditions gave no rearrangement, indicat-
ing that CO migratory insertion also proceeds faster than
the rearrangement.22
epimerization at C16. This mechanism is also attractive since
isolation of 11 can be explained. It is possible that both
pathways operate competitively.28 With 5a, it is likely that
the ring-fragmentation route would be disfavored because
the TBS protecting group at C12 would hinder the coordina-
tion of PdII with the allene. The major pathway for this
substrate would then likely be the allenone-pathway, which
could explain the isolation of 10 from this substrate.
To summarize, it was shown that oxidative addition of
Pd0 to the vinyl triflate in 5 is faster than that to the vinyl
epoxide, in accordance with a previous report with a related
substrate.6 The resulting vinylpalladium 12 can be trapped
if the subsequent step is faster than rearrangement (i.e.,
carbonylation or transmetalation in the presence of chloride
ligand). However, it eventually rearranges to alkoxide 14,
which is in equilibrium with 10 or 15.
The mechanism for the C16 epimerization in eq 3 is
unclear. One possibility is that Pd alkoxide 14 undergoes
â-elimination to give allenone 10.23 Readdition of the
palladium hydride24 from the R-face of the C16 ketone would
result in overall epimerization of this stereocenter. The
isolation of a small amount of 10 supports this idea.
Alternatively, 14 might be in equilibrium with ring-
fragmented intermediate 15, formed either by inter- or
intramolecular coordination of PdII. This could close back25-27
from either face of the aldehyde, resulting in overall
(19) (a) Tanabe, M.; Crowe, D. F.; Dehn, R. L. Tetrahedron Lett. 1967,
40, 3943. (b) Muller, R. K.; Felix, D.; Schreiber, J.; Eschenmoser, A. HelV.
Chim. Acta 1970, 53, 1479.
(20) Farina, V.; Krishnan, B.; Marshall, D. R.; Roth, G. P. J. Org. Chem.
1993, 58, 5434.
Acknowledgment. This research was supported by a
research grant from the National Institute of Heath (GM46057).
K.Y. acknowledges the University of California, Berkeley,
and Ishizaka foundation for graduate fellowships.
(21) It was originally reported that LiCl is required in the Stille reaction
for vinyl or aryl sulfonates.8a However, it was later found that the effect of
LiCl is more complicated and sensitive to ligands and solvents.8c
(22) It has been recognized that the rate-limiting step for the Stille
reaction is the transmetalation step, whereas that for carbonylation is
oxidative addition.8c
(23) For â-elimination of Pd-methoxide to form “Pd-H”, (a) Grushin,
V. V. Chem. ReV. 1996, 96, 2011. (b) Elsevier: C. J.; Toth, I. Organo-
metallics 1994, 13, 2118. (c) Milstein, D.; Frolow, F.; Portnoy, M.
Organometallics 1991, 10, 3960. (d) Milstein, D.; Portnoy, M. Organo-
metallics 1994, 13, 600.
(24) Although “Pd-H” insertion into CO, CO2, and CS2 are well
documented,23a the corresponding reaction with aldehydes is not precedented.
(25) While allyl- or propargylpalladium species are electrophilic in
nature,26 corresponding nickel species are known to react with aldehydes.27
We therefore speculate that the intramolecular version of this transformation
may be possible.
Supporting Information Available: Experimental pro-
cedures and characterization data for new compounds. This
material is available free of charge via the Internet at
http://pubs.acs.org.
OL000064E
(27) Hegedus, L. S. Transition Metals in the Synthesis of Complex
Organic Molecules; University Science Books: California, 1994; p 320.
(28) When the reaction is carried out with stoichiometric amount of Pd0,
11 is not formed (Scheme 3, eq 3). This observation might be due to a
stable bidentate chelate of 14 that does not readily undergo transmetalation
with Bu3SnH.
(26) A bis-π-allylpalladium species was reported to have a nucleophilic
character: (a) Nakamura, H.; Asao, N.; Yamamoto, Y. J. Chem. Soc., Chem.
Commun. 1995, 1273. (b) Nakamura, H.; Shim, J. G.; Yamamoto, Y. J.
Am. Chem. Soc. 1997, 119, 8113.
1712
Org. Lett., Vol. 2, No. 12, 2000