Synthesis of 1,1-disubstituted alkenes via a Ru-catalyzed addition

Article abstract of DOI:10.1021/ja012009m

The synthesis of 1,1-disubstituted alkenes typically involves reactions that lack atom economy such as olefination protocols. The use of various ruthenium complexes to effect the addition of terminal alkynes to alkenes is explored as an atom economical st

Full text of DOI:10.1021/ja012009m

12504
J. Am. Chem. Soc. 2001, 123, 12504-12509
Synthesis of 1,1-Disubstituted Alkenes via a Ru-Catalyzed Addition
Barry M. Trost,* Anthony B. Pinkerton, F. Dean Toste, and Martin Sperrle
Contribution from the Department of Chemistry, Stanford UniVersity, Stanford, California 94305-5080
ReceiVed August 20, 2001
Abstract: The synthesis of 1,1-disubstituted alkenes typically involves reactions that lack atom economy such
as olefination protocols. The use of various ruthenium complexes to effect the addition of terminal alkynes to
alkenes is explored as an atom economical strategy. Two new ruthenium complexes have been discovered that
effect this reaction at ambient temperature, cyclopentadienylruthenium (triphenylphosphine) camphorsulfonate
and cyclopentadienylruthenium tris(acetonitrile) hexafluorophosphate. Using these complexes as catalysts,
reactions proceed at ambient temperature in acetone or DMF, respectively. Regioselectivity favoring the
formation of a 1,1-disubstituted over a 1,2-disubstituted alkene typically ranges from 9:1 to >25:1. The reaction
demonstrates extraordinary chemoselectivityseven di- and trisubstituted alkenes such as present in the products
do not compete with the starting monosubstituted alkene. Free hydroxyl groups as well as silyl and PMB
ethers are tolerated as are ketones, esters, and amides. The mechanism of the reaction is believed to invoke
formation of a metallacyclopentene. To account for the chemo- and regioselectivity, the initial formation of
the metallacycle is believed to be reversible. While formation of the 2,5-disubstituted ruthenacyclopentene,
which produces the linear product, is believed to be kinetically preferred, the rate of â-hydrogen elimination
from the 2,4-disubstituted ruthenacyclopentene, which produces the branched product, is believed to be faster.
Thus, the competition between the rate of â-hydrogen elimination and cycloreversion rationalizes the results.
Introduction
2.4,5 This strategy is highly atom economical, but it suffers with
The development of new synthetic methods that are
atom economical is an important goal.1 Ideally, reactions in-
volve additions that proceed chemo- and regioselectively
to generate complex products from simple building blocks.
The synthesis of 1,1-disubstituted alkenes illustrates the
issue (eq 1). The strategy that almost immediately jumps
respect to the issue of regioselectivity. Branched-to-linear ratios
(b/l) typically ranged from 3 to 6:1. In this paper, we report
two new ruthenium complexes competent to effect this reaction
and the remarkable influences of ligand and, in some cases,
alkene substrate on the regioselectivity of this process.6 With
regioselectivities typically 9 to >25:1 b/l, an excellent approach
to 1,1-disubstituted alkenes has resulted.
Catalysts. Scheme 1 outlines the working hypothesis for the
mechanism of the Ru-catalyzed alkene/alkyne coupling. The
initial coordination (step 1) and the tautomerization of 1a/1b
to metallacycles 2a/2b (steps 2) are believed to be reversible,
at least to some extent. The product ratio then depends on the
conversion of 2 to 3. If step 3 is faster than the reversal of step
2, then the product ratio derives from the initial ratio of 2a/2b.
If step 3 is slower than the reversal of step 2, then a Curtin-
Hammett situation ensues. The product ratio will depend on
to mind is an olefination protocol.2 While chemoselec-
tivity issues do arise in such a strategy, a major defic-
iency is the poor atom economy of such a process (path a).
A better strategy employs the more atom economical
addition of an organometallic to a terminal alkyne.3 While
this route clearly still retains issues of atom economy, a
large issue is chemoselectivity. We have been developing
an atom economical alternative to 1,1-disubstituted alkenes
based upon a ruthenium-catalyzed addition as shown in eq
(1) Trost, B. M. Science 1991, 254, 1471. Trost, B. M. Angew. Chem.,
Int. Ed. Engl. 1995, 34, 259.
(2) Kelly, S. E. In ComprehensiVe Organic Synthesis; Trost, B. M.,
Fleming, I., Schreiber, S. L., Eds.; Pergamon: Oxford, U.K., 1991; Vol. 1,
pp 729-818.
(3) Knochel, P. In ComprehensiVe Organic Synthesis; Trost, B. M.,
Fleming, I., Schreiber, S. L., Eds.; Pergamon: Oxford, U.K., 1991; Vol. 4,
pp 865-911.
(4) Trost, B. M.; Indolese, A. F.; Mu¨ller, T. J. J.; Treptow, B. J. Am.
Chem. Soc. 1995, 117, 615. Trost, B. M.; Indolese, A. F. J. Am. Chem.
Soc. 1993, 115, 4361.
(5) For a review, see: Trost, B. M.; Krische, M. J. Synlett 1998, 1.
(6) For a preliminary report of a portion of this work, see: Trost, B. M.;
Toste, F. D. Tetrahedron Lett. 1999, 40, 7739.
10.1021/ja012009m CCC: $20.00 © 2001 American Chemical Society
Published on Web 11/17/2001

Ru-Catalyzed Synthesis of 1,1-Disubstituted Alkenes
J. Am. Chem. Soc., Vol. 123, No. 50, 2001 12505
Scheme 1. Mechanistic Hypothesis for the Ru-Catalyzed Alkene-Alkyne Additiona
a Any open coordination site(s) in these complexes would be anticipated to be occupied by some ligand present including solvent.
Chart 1
the relative rate of the â-hydrogen elimination of 2a versus 2b.
In this catalytic cycle, it should be noted that at least one open
coordination site on Ru exists and, in some cases, two. Thus,
the regioselectivity may be affected by the occupants of the
open coordination site.
the reaction, this negative rate effect would be countermanded
by the cationic nature of the complex. Complexes 68 and 79 are
readily accessed according to eqs 3 and 4, the former according
Our first-generation catalyst 57 possesses a chloride, which
may serve as a ligand at any stage of the catalytic cycle. Such
an anionic ligand is believed to be detrimental to the rate of
the reaction. Cationic complexes were envisioned to be kineti-
cally more competent and would allow for a greater diversity
in the choice of a ligand. Placing a bulky substituent on Ru
might favor 2b over 2a and therefore enhance the regioselec-
tivity favoring the branched isomer. Our second-generation
cationic catalysts 6s and 7s (Cs ) camphorsulfonate; Chart 1)
would derive by protonation of the methallyl group of 6 and 7
with camphorsulfonic acid. Although the phosphine might slow
to a literature procedure.8 The quality of the Grignard reagent
was crucial, and preactivation of the magnesium by the method
of Brown et al.10 was preferred. Removal of the liberated
triphenylphosphine by sublimation was preferred over distilla-
tion. An X-ray structure of 7 (Figure 1) reveals a transoid
relationship of the central carbon of the allyl group and
phosphorus. This structure remains constant in solution as
revealed by its NMR spectra, which are similar to those reported
for 6. Anticipating a rate retardation by the presence of a
phosphine ligand, a cationic ruthenium complex lacking this
substituent was sought. The tris-acetonitrile complex 811 nicely
meets this requirement. Although it possesses a sterically small
(7) Albers, M. O.; Robinson, D. J.; Shaver, A.; Singleton, E. Organo-
metallics 1986, 5, 2199. For a review, see: Davies, S. G.; McNally, J. P.;
Smallridge, A. J. AdV. Organomet. Chem. 1990, 30, 1.
(8) Lehmkuhl, H.; Mauermann, H.; Benn, R. Liebigs Ann. Chem. 1980,
754.
(9) For the precursor, see: Trost, B. M.; Vidal, B.; Thommen, M. Chem.
Eur. J. 1999, 5, 1055.
(10) Baker, K. V.; Brown, J. M.; Hughes, N.; Skarnulis, A. J.; Sexton,
A. J. Org. Chem. 1991, 56, 698.
(11) Gill, T. P.; Mann, K. R. Organometallics 1982, 1, 485.
(12) Cf. Trost, B. M.; Krause, L.; Portnoy, M. J. Am. Chem. Soc. 1997,
119, 11319. Trost, B. M.; Portnoy, M.; Kurihara, H. J. Am. Chem. Soc.
1997, 119, 836.

12506 J. Am. Chem. Soc., Vol. 123, No. 50, 2001
Trost et al.
selectivity increased to 67% and a 26:1.0 b/l ratio. Thus, an
excellent selectivity for the 1,1-disubstituted alkene was now
in hand.
Third-Generation Catalyst Reactions. Initial studies with
our third generation catalyst 8 under conditions similar to those
above (30% CSA, acetone, 60 °C) gave a 78% yield of 10 and
11 but a disappointing 1:1.1 ratio. Using alternative additives
such as indium triflate, HMPA, and TMU did not have a bene-
ficial effect. In the absence of any additives at room tempera-
ture in acetone, a 73% yield of a 1.2:1 b/l ratio was obtained.
In contrast to our second-generation catalyst, using DMF as
solvent had a dramatic effect on the b/l ratio favoring b. The
reaction still proceeded at room temperature within 2 h to give
an 82% yield of a 9:1 b/l ratio. With this excellent result, a
range of substrates was examined and is summarized in the
Table.
The examples in the table clearly show good to excellent
regioselectivity. The lower regioselectivities of entries 1 and 2
are somewhat baffling given the examples in the rest of the
table. At the same time, it is difficult to see trends as to what
factors contribute to higher b/l ratios. Entry 3 comes closest to
a totally unbiased system, yet a 7.2:1 b/l ratio was still observed.
Curiously, the unfunctionalized 1-decene gave regioselectivities
as high as 15.3:1 (entry 15). More remarkable, methyl 10-
undecenoate, an alkene substrate whose ester functional group
is so far removed from the alkene that it should function more
like a simple hydrocarbon, reacted with some alkynes to give
the branched isomer as the only detectable one (entries 21 and
22). There does appear to be some role for substituents on the
alkyne influencing regioselectivity. Steric bulk or coordinating
groups proximal to the alkyne may be helpful. However, the
steric hindrance, if at the propargylic position, can reverse the
regioselectivity. Indeed, as shown in eq 7, the linear isomer 18
Figure 1. ORTEP plot of 7.
ligand, acetonitrile, it offers the prospect of exchanging it for
other ligands as desired.
Second-Generation Catalyst Reactions. Equation 5 illus-
trates the initial reaction examined using complexes 6 and 7 as
precatalysts. Using the typical conditions4 for complex 5 (3:1
DMF/water, 100 °C), a 41% yield of a 5:1 b/l ratio of adducts
10 and 11 was isolated. Treating complex 6 (10 mol %) with
camphorsulfonic acid (CSA, 15 mol %) gave the camphorsul-
fonate (Cs-) salt 6s. Peforming the reaction at 100 °C in DMF
gave only a 30% yield of a 3.7:1 ratio of b/l adducts. In acetone
at reflux, a 48% yield of a 3.4:1 ratio of b/l isomers was isolated.
Interestingly, the reaction even proceeded at ambient temperature
wherein the b/l ratio improved to 5.0:1.0. The presence of water
in acetone had a dramatic effect on the reaction. Using 5:1 to
2:1 acetone/water ratios at ambient temperature gave 76-66%
yields of 5.9:1.0-5.2:1.0 b/l ratios, respectively. In the case of
5:1 acetone/water, raising the temperature to reflux gave the
same 76% yield but the ratio dropped to 3.8:1.0 from 5.9:1.0.
Varying the amount of CSA had little effect on the reaction.
However, addition of indium triflate as a cocatalyst had a
significant effect.12 Adding anywhere from 10 to 40 mol % gave
little change in yield (70-74%) but significantly reduced the
regioselectivity to 1.8:1.0 b/l. Switching to complex 7 as the
precatalyst gave similar results.
was virtually the exclusive product of the reaction with alkyne
16. On the other hand, moving the quaternary center one atom
away from the alkyne and the regioselectivity completely
reverses giving only the branched product (Table, entry 21).
A functional group two or three atoms away from the
alkyne appears helpful. A comparison of entries 8-11 indi-
cates that an aromatic ring may be sufficient. Hydroxyl
groups two atoms away also play this role. Conversion of
the hydroxyl group to its PMB ether significantly reduces
the regioselectivity (entry 2 vs 13). Running the reaction of
entry 13 in acetone as solvent gave the adduct in 46% yield
as a 2:1 b/l ratio, again showing the superiority of DMF as
solvent. On the other hand, a hydroxyl group three atoms
away leads to a 2:1 adduct in addition to the desired pro-
duct, both with low b/l ratios (eq 8). In this case, it appears
A more striking example is illustrated in eq 6. Using our first-
generation catalyst 5 in methanol at reflux gave a 50% yield of
a 3.8:1.0 ratio of 14/15.4 With our second-generation catalyst
6s in 5:1 acetone/water at room temperature, the yield and regio-

Ru-Catalyzed Synthesis of 1,1-Disubstituted Alkenes
J. Am. Chem. Soc., Vol. 123, No. 50, 2001 12507
Table 1. A Synthesis of 1,1-Disubstituted Alkenes via a Ru Catalyzed Additiona
a All reactions were performed with 10 mol % 8, 1.2 equiv of alkene, and 1.0 equiv of alkyne at 0.5 M in DMF at room temperature.
that cycloisomerization of the alkynol to 2-methylenetetrahy-
drofuran also occurs, the latter then derivatizing the primary
alcohol to form an acetal of either the starting alkynol or the
adduct, either of which may account for the final product acetal.

12508 J. Am. Chem. Soc., Vol. 123, No. 50, 2001
Trost et al.
There were several alkene substrates that gave low regiose-
lectivities (eqs 9-11). In the reaction of safrole (eq 9), the
interactions between the two carbons involved in C-C bond
formation of the metallacycle normally dominate and lead to
preference of transition state Il over Ib to favor metallacycle IIl
(see Scheme 2).15 Thus, to the extent that the rate of â-hydrogen
elimination is faster than reversal of metallacycle formation,
then linear products should dominate. Indeed, to the extent that
â-hydrogen elimination is made faster, the amount of the linear
product increases. Thus, when the â-hydrogen in IIb or IIl is
benzylic or allylic which, by decreasing the C-H bond strength
should increase the rate of â-H elimination, the amount of the
linear product increases and may even become dominant, albeit
slightly.
However, that circumstance appears to be more the exception
than the rule. In most cases, the â-hydrogen elimination is slower
than reversal of metallacycle formationsa situation that brings
the reaction toward Curtin-Hammett control. The solvent ef-
fects are in accord with this analysis. The better coordinating
solvent, DMF, favors branched over linear product with complex
8. Since â-hydrogen elimination requires open coordination sites
on the metal,16 to the extent such sites are occupied by solvent,
this reaction is slowed. Thus, DMF more effectively retards
â-hydrogen elimination in both cases, thus making reversal of
metallacycle formation faster than â-hydrogen elimination. In
the case of complex 6 as catalyst, no solvent effect on the
regioselectivity was noted since triphenylphosphine occupies
one of the open coordination sites in the metallacycle, which
makes it difficult for any other ligand to enter the coordination
sphere of the Ru(+4).
In the domain wherein metallacycle equilibration is faster than
â-hydrogen eliminationsi.e., kb, k-b, kl, and k-l are larger than
k2b and k2lsthen the product ratio depends on k2b/k2l. Steric
hindrance associated with the transition state leading to IIIl
makes k2b > k2l and thus favors the branched product. To the
extent that potential coordinating groups are present in R, any
such coordination would disfavor â-hydrogen elimination in the
case of IIl due to saturation of the metal, therefore further in-
creasing the ratio of k2b/k2l and the amount of branched product.
Among the catalysts explored to date, the tris-acetonitrile
complex 8 represents the most practical and general. Reactions
proceed readily at room temperature normally within a few
hours. The mild conditions undoubtedly also contribute to the
excellent selectivity. The reaction has excellent chemoselectivity.
It is not sensitive to water or oxygen although we do perform
the reactions under an inert atmosphere. A broad range of
functionality is compatible. It is important to note that the
products, which are alkenes, do not react further under the
reaction conditionssa fact that indicates a monosubstituted
alkene is a much better substrate than a disubstituted one. In
comparison to other methods to form 1,1-disubstituted alkenes,
this simple addition has the benefit of being highly atom
economical and very simple to perform. The formation of 1,4-
dienes constitutes a bonus since it provides a second alkene for
elaboration as well. For example, the formation of allyl alcohols
as in the table, entries 18 and 19, sets the stage for further
reactions such as allylic alkylations, Claisen rearrangements,
etc. This reaction should prove to be a valuable addition to the
arsenal of atom economic C-C bond forming reactions.
reaction proceeded readily at room temperature in both DMF
and acetone. While the regioselectivity was rather poor in both,
the major regioisomer in acetone was the linear one and in DMF
the branched one. For comparison, the same reaction catalyzed
by the first-generation catalyst 5 required methanol at reflux
and only gave a 46% yield of a 1:1 mixture of 19 and 20.
Similarly, the allylic substrates 21 and 24 also participate but
also give low regioselectivity. The formation of 1,3-dienes is
quite interesting. The latter reaction (eq 11) required methanol
at reflux with catalyst 5, but the results were bettersa 54%
yield of a 4:1 ratio of 25 and 26. While these three examples
demonstrate the scope and limitation for good regioselectivity,
they help provide insight into the mechanism of this process.
Discussion
The mechanistic hypothesis outlined in Scheme 1 generally
accounts for the observations to date. The ability to form a ru-
thenacyclopentene (from an alkyne and alkene) compared to a
ruthenacyclopentadiene (from two alkynes) or a ruthenacyclo-
pentane (from two alkenes) can be understood by envisioning
metallacycle formation being reversible. The much higher coor-
dinating affinity of an alkyne compared to an alkene should
lead to the rate of metallacycle formation decreasing in the order
metallacyclopentadiene > metallacyclopentene . metallacyclo-
pentane.14 The current results can be understood in the con-
text of this scheme if metallacyclopentene formation is reversible
and the relative rates of â-hydrogen elimination and metallacycle
formation are competitive. Significant data suggest that steric
(13) For a few other reactions involving a ruthenacyclopentene inter-
mediate, see: Mitsudo, T.; Kokuryo, K.; Shinsugi, T.; Nakagawa, Y.;
Watanabe, Y.; Takegami, Y. J. Org. Chem. 1979, 44, 4492. Trost, B. M.;
Imi, K.; Indolese, A. F. J. Am. Chem. Soc. 1993, 115, 8831. Warrener, R.
N.; Abhenants, A.; Kennard, C. H. L. J. Am. Chem. Soc. 1994, 116, 3645.
Mitsudo, T.; Naruse, H.; Kondo, T.; Ozaki, Y.; Watanabe, Y. Angew. Chem.,
Int. Ed. Engl. 1994, 33, 580. Marimoto, T.; Chatani, N.; Fukumoto, Y.;
Murai, S. J. Org. Chem. 1997, 62, 3762. Kondo, T.; Ozaki, Y.; Watanabe,
Y. Angew. Chem., Int. Ed. Engl. 1994, 33, 580. Marimoto, T.; Chatani, N.;
Fukumoto, Y.; Murai, S. J. Org. Chem. 1997, 62, 3762. Kondo, T.; Suzuki,
N.; Okada, T.; Mitsudo, T. J. Am. Chem. Soc. 1997, 119, 6187. Matsushima,
Y.; Kikuchi, H.; Uno, M.; Takahashi, S. Bull. Chem. Soc. Jpn. 1999, 72,
2475. De´rien, S.; Ropartz, L.; Le Paih, J.; Dixneuf, P. H. J. Org. Chem.
1999, 64, 3524.
Experimental Section
Ru-Catalyzed Reaction with Complex 6: 7-Methylene-4-tridecen-
2-one (14). To a mixture of 9.6 mg (0.02 mmol) of 6 and 18.6 mg
(0.08 mmol) of CSA under nitrogen was added 1.5 mL of dried acetone
(15) Vollhardt, K. P. C. Angew. Chem., Int. Ed. Engl. 1984, 23, 539.
(16) Cross, R. J. In The Chemistry of the Metal-Carbon Bond; Hartley,
R. F., Patai, S., Eds.; Wiley: New York, 1985; Vol. 2, Chapter 8. Also see:
Schmidt, G. F.; Brookhart, M. J. Am. Chem. Soc. 1985, 107, 1443.
(14) Trost, B. M.; Toste, F. D.; Pinkerton, A. B. Chem. ReV. 2001, 101,
2067-2096.

Ru-Catalyzed Synthesis of 1,1-Disubstituted Alkenes
J. Am. Chem. Soc., Vol. 123, No. 50, 2001 12509
Scheme 2. Competitive Ruthenacycle Formation and â-Hydrogen Elimination
to form an orange solution. 1-Octyne (24.2 mg, 0.22 mmol) and 5-
hexen-2-one (19.6 mg, 0.20 mmol) were added sequentially. The reac-
tion was complete after 3 h. It was evaporated and the residue directly
chromatographed (1:8 ethyl acetate/hexane) to give 27.8 mg (67% yield)
of 14. GC analysis revealed a 26:1 ratio of 14 (retention time 10.35
min) to 15 (retention time 10.59 min): IR (neat) 1720, 1645, 1430m
1358, 1156, 1018 cm-1; 1H NMR (300 MHz, CDCl3) δ 5.56-5.59 (m,
2H), 4.72 (s, 1H), 3.15 (d, J ) 4.6 Hz, 2H), 2.75 (d, J ) 4.2 Hz, 2H),
2.16 (s, 3H), 2.00 (t, J ) 7.6 Hz, 2H), 1.21-1.31 (m, 8H), 0.89 (t, J
) 6.8 Hz, 3H); 13C NMR (75 MHz, CDCl3) δ 208.01, 149.19, 133.42,
124.07, 110.25, 47.87, 39.69, 36.25, 31.98, 29.57, 29.25, 27.78, 22.82,
14.25; HRMS calcd for C14H24O 208.1827, found 208.1844 (4.3).
General Procedure with Ruthenium Complex 8. The alkyne (1
equiv) and alkene (1.2 equiv) were dissolved in the solvent (to give
0.5 M solution) and then added to CpRu(CH3CN)3PF6 (10 mol %) in
a test tube. The reaction was stirred under nitrogen for 2 h. The solvent
was then removed in vacuo and the crude mixture was analyzed by
proton NMR. The residue was then subjected to silica gel chromatog-
raphy.
A typical example is given as follows: 5-Hexyn-1-ol (24.5 mg, 0.25
mmol) and 1-decene (42.1 mg, 0.3 mmol) were dissolved in DMF (0.5
mL) and then added to CpRu(CH3CN)3PF6 (10.9 mg, 0.025 mmol) in
a test tube. The reaction was stirred under nitrogen for 2 h. The reaction
was then subjected to silica gel chromatography (2:1 petroleum ether/
ether) to give 37 mg of 9 (62%) as a 7.4:1 ratio of the branched to the
linear isomers, as determined by relative integration of the following
signals in the proton NMR: δ 5.48-5.38 (representing all four protons
from the linear isomer, and two protons from the branched isomer)
and two singlets at δ 4.76 and 4.75 (representing the branched isomer).
5-Methylene-pentadec-7-en-1-ol (9). Colorless oil. Rf ) 0.29 (2:1
petroleum ether/ether); IR (neat) 3331, 3078, 2926, 2860, 2358, 1642,
1450, 1375, 1156, 1062, 970, 891 cm-1; 1H NMR (500 MHz, CDCl3)
δ 5.48-5.38 (m, 2H), 4.76 (s, 1H), 4.75 (s, 1H), 3.68 (t, J ) 6.4 Hz,
2H), 2.71 (d, J ) 6.4 Hz, 2H), 20.8-1.99 (m, 4H), 1.62-1.51 (m,
4H), 1.40-1.29 (m, 11H), 0.90 (t, J ) 6.9, 3H); 13C NMR (125 MHz,
CDCl3) δ 148.9, 132.4, 127.5, 109.5, 62.9, 39.5, 35.5, 32.5, 32.4, 31.9,
29.5, 29.2, 29.1, 23.7, 22.7, 14.1. Additional peaks for linear isomer:
δ 131.2, 130.4, 192.2, 128.4, 35.6. Anal. Calcd for C16H30O: C, 80.61;
H, 12.68. Found: C, 80.48; H, 12.43.
Acknowledgment. We thank the National Science Founda-
tion and the National Institutes of Health, General Medical
Sciences, for their generous support of our programs. Mass
spectra were provided by the Mass Spectrometry Regional
Center of the University of CaliforniasSan Francisco, supported
by the NIH Division of Research Resources. A.B.P. thanks
Abbott Laboratories, Glaxo-Wellcome, the ACS Division of
Organic Chemistry (sponsored by Bristol-Myers Squibb), and
Eli Lilly for fellowship support. F.D.T. was a Stanford Graduate
Fellow. M.S. thanks the Alexander von Humboldt Foundation
for a Feodor Lynen Postdoctoral Fellowship.
Supporting Information Available: Experimental details
for all reactions as well as characterization data and X-ray data
(PDF). This material is available free of charge via the Internet
at http://pubs.acs.org.
JA012009M