Geminal dicarboxylates as carbonyl surrogates for asymmetric synthesis. Part I. Asymmetric addition of malonate nucleophiles

Article abstract of DOI:10.1021/ja003774o

Asymmetric alkylations of allylic geminal dicarboxylates with dialkyl malonates have been investigated. The requisite allylic geminal dicarboxylates are prepared in good yields and high isomeric purities by two catalytic methods, ferric chloride-catalyzed addition of acid anhydrides to α,β-unsaturated aldehydes and palladium-catalyzed isomerization and addition reactions of propargylic acetates. The complex of palladium(0) and the chiral ligand derived from the diamide of trans-1,2-diaminocyclohexane and 2-diphenylphosphinobenzoic acid most efficiently catalyzed the asymmetric process to provide allylic carboxylate esters with high ee. By systematic optimization studies, factors affecting the enantioselectivity of the reaction have been probed. In general, higher ee's have been achieved with those conditions which facilitate kinetic capture of the incipient π-allylpalladium intermediate. These conditions also proved effective for achieving high regioselectivities. The minor regioisomeric product was formed when reactive substrates or achiral ligands were employed for the reaction, and could be minimized through the use of the chiral ligand. Under the established conditions, the alkylation of various gem-dicarboxylates afforded monoalkylated products in high yields with greater than 90% ee. The process constitutes the equivalent of an addition of a stabilized nucleophile to a carbonyl group with high asymmetric induction.

Full text of DOI:10.1021/ja003774o

J. Am. Chem. Soc. 2001, 123, 3671-3686
3671
Geminal Dicarboxylates as Carbonyl Surrogates for Asymmetric
Synthesis. Part I. Asymmetric Addition of Malonate Nucleophiles
Barry M. Trost* and Chul Bom Lee
Contribution from the Department of Chemistry, Stanford UniVersity, Stanford, California 94305
ReceiVed October 24, 2000
Abstract: Asymmetric alkylations of allylic geminal dicarboxylates with dialkyl malonates have been
investigated. The requisite allylic geminal dicarboxylates are prepared in good yields and high isomeric purities
by two catalytic methods, ferric chloride-catalyzed addition of acid anhydrides to R,â-unsaturated aldehydes
and palladium-catalyzed isomerization and addition reactions of propargylic acetates. The complex of palladium-
(0) and the chiral ligand derived from the diamide of trans-1,2-diaminocyclohexane and 2-diphenylphosphi-
nobenzoic acid most efficiently catalyzed the asymmetric process to provide allylic carboxylate esters with
high ee. By systematic optimization studies, factors affecting the enantioselectivity of the reaction have been
probed. In general, higher ee’s have been achieved with those conditions which facilitate kinetic capture of
the incipient π-allylpalladium intermediate. These conditions also proved effective for achieving high
regioselectivities. The minor regioisomeric product was formed when reactive substrates or achiral ligands
were employed for the reaction, and could be minimized through the use of the chiral ligand. Under the
established conditions, the alkylation of various gem-dicarboxylates afforded monoalkylated products in high
yields with greater than 90% ee. The process constitutes the equivalent of an addition of a stabilized nucleophile
to a carbonyl group with high asymmetric induction.
Introduction
Strategies to effect such transformations derive from recognition
of the stereochemical courses in each step of the catalytic cycle
and analysis of symmetry elements imparted in the substrates
or intermediates. The current strategies involve differentiation
of (1) the enantiotopic leaving groups of meso-compounds,14-16
(2) the allylic termini of symmetric intermediates,17-20 (3) two
interconverting intermediates,21-24 (4) the enantioface of
alkenes,25-27 and (5) the enantiofaces of nucleophiles.28-33
Another strategy is represented by the desymmetrization
transformation in eq 1 wherein the two enantiotopic leaving
groups located on the same carbon atom of an achiral substrate
are differentiated. The asymmetric induction of this system,
The palladium-catalyzed allylic alkylation reaction is one of
the most powerful tools for the controlled introduction of
carbon-carbon and carbon-heteroatom bonds into organic
compounds.1-6 Recent employment of chiral ligands in this
process has provided an opportunity to perform the reaction in
an enantioselective fashion, thereby greatly increasing the utility
of this method.7-12 Unlike most transition metal-catalyzed
processes, asymmetric allylic alkylations offer multiple mech-
anisms as a source of asymmetry because the enantio-dis-
criminating event can arise in any one of many steps in the
catalytic cycle. Thus, this unique feature of the asymmetric
allylic alkylation (AAA) reaction allows for the conversion of
starting materials of various types, such as racemic, meso-, and
achiral compounds, into enantiomerically enriched material.13
(14) Trost, B. M.; Patterson, D. E. J. Org. Chem. 1998, 63, 1339.
(15) Muchow, G.; Brunel, J. M.; Maffei, M.; Pardigon, O.; Buono, G.
Tetrahedron 1998, 54, 10435.
(16) Hayashi, T.; Yamamoto, A.; Ito, Y. Tetrahedron Lett. 1987, 28,
4837.
(1) Godleski, S. A. In ComprehensiVe Organic Synthesis; Trost, B. M.,
Fleming, I., Eds.; Pergamon Press: New York, 1991; Vol. 4, pp 585-661.
(2) Tsuji, J. Tetrahedron 1986, 42, 4361.
(3) Trost, B. M.; Verhoeven, T. R. In ComprehensiVe Organometallic
Chemistry; Wilkinson, G., Ed.; Pergamon Press: Oxford, 1982; Vol. 8,
Chapter 57.
(17) Trost, B. M.; Bunt, R. C. J. Am. Chem. Soc. 1994, 116, 9.
(18) von Matt, P.; Pfaltz, A. Angew. Chem., Int. Ed. Engl. 1993, 32,
566.
(19) Dawson, G. J.; Frost, C. G.; Williams, J. M. J. Tetrahedron Lett.
1993, 34, 3149.
(20) Sprinz, J.; Helmchen, G. Tetrahedron Lett. 1993, 34, 1769.
(21) Trost, B. M.; McEachern, E. J. J. Am. Chem. Soc. 1999, 121, 8649.
(22) Trost, B. M.; Toste, D. F. J. Am. Chem. Soc. 1999, 121, 3543.
(23) Pretot, R.; Pfaltz, A. Angew. Chem., Int. Ed. 1998, 37, 323.
(24) Auburn, P. R.; Mackenzie, P. B.; Bosnich, B. J. Am. Chem. Soc.
1985, 107, 2033.
(4) Trost, B. M. Acc. Chem. Res. 1980, 13, 385.
(5) Tsuji, J. Palladium Reagents and Catalysts; Wiley: New York, 1996.
(6) Heck, R. F. Palladium Reagents in Organic Syntheses; Academic
Press: New York, 1985.
(7) Trost, B. M.; Lee, C. B. In Catalytic Asymmetric Synthesis, 2nd ed.;
Ojima, I., Ed.; Wiley-VCH: New York, 2000; Chapter 8E, pp 593-649.
(8) Pfaltz, A.; Lautens, M. In CompresnesiVe Asymmeric Catalysis;
Jacobsen, E. N., Pfaltz, A., Yamamoto, H., Eds.; Springer-Verlag: Heidel-
berg, 1999; Vol. 2, Chapter 24, pp 833-884.
(9) Trost, B. M.; Van Vranken, D. L. Chem. ReV. 1996, 96, 395.
(10) Lu¨bbers, T.; Metz, P. In StereoselectiVe Synthesis; Helmchen, G.,
Hoffmann, R. W., Mulzer, J., Schaumann, E., Eds.; Thieme: Stuttgart, 1996;
Vol. 4, pp 2371-2473.
(25) Trost, B. M.; Krische, M. J.; Radinov, R.; Zanoni, G. J. Am. Chem.
Soc. 1996, 118, 6297.
(26) Fiaud, J. C.; Legros, J. Y. J. Org. Chem. 1990, 55, 4840.
(27) Fiaud, J. C.; Legros, J. Y. Tetrahedron Lett. 1988, 29, 2959.
(28) Trost, B. M.; Schroeder, G. M. J. Org. Chem. 2000, 65, 1569.
(29) Kuwano, R.; Ito, Y. J. Am. Chem. Soc. 1999, 121, 3236.
(30) Trost, B. M.; Schroeder, G. M. J. Am. Chem. Soc. 1999, 121, 6759.
(31) Trost, B. M.; Ariza, X. Angew. Chem., Int. Ed. Engl. 1997, 36, 2635.
(32) Trost, B. M.; Radinov, R.; Grenzer, E. M. J. Am. Chem. Soc. 1997,
119, 7879.
(11) Hayashi, T. In Catalytic Asymmetric Synthesis; Ojima, I., Ed.;
VCH: New York, 1993; Chapter 7.1, pp 325-363.
(12) Consiglio, G.; Waymouth, R. M. Chem. ReV. 1989, 89, 257.
(13) Trost, B. M. Acc. Chem. Res. 1996, 29, 355.
(33) Sawamura, M.; Sudo, M.; Ito, Y. J. Am. Chem. Soc. 1996, 118,
3309.
10.1021/ja003774o CCC: $20.00 © 2001 American Chemical Society
Published on Web 03/28/2001

3672 J. Am. Chem. Soc., Vol. 123, No. 16, 2001
Trost and Lee
however, is complicated by the fact that there are two prochiral
elements present in the substrate, the leaving groups and the
π-faces of the alkene. Since both the complexation and the
ionization steps are a potential source of enantioselectivity, the
AAA reaction of this type presents a formidable control problem.
In addition, due to the unsymmetric nature of the π-allylpalla-
dium intermediate, there is a regioselectivity problem depicted
in eq 1. Thus, differentiation of the two enantiotopic leaving
groups by a substitution process constitutes a rare example in
which asymmetric catalysis is achieved by chiral recognition
between two geminal sp3-sp3 bonds.34
system offers a good testing ground to examine the ability of a
chiral ligand to control other types of selectivities beyond
enantioselectivity. Herein, we report a full account of our efforts
in the development of asymmetric alkylation of gem-dicarboxy-
lates with malonate esters.42
Preparation of Allylic gem-Dicarboxylates. To evaluate the
efficacy of various acid catalysts, trans-2-hexenal (5a, R )
C3H7) was selected as a model substrate and reacted with acetic
anhydride (eq 2). The mixture of 5a and 2-5 equiv of acetic
anhydride neat or in methylene chloride was treated with various
Lewis acids (BF3OEt2,43 PCl3,44 H2SO4,45 I2,46 K-10 Clay,47
FeCl3,48 etc.). In all cases, the reaction was complete within 2
h at 0 °C, providing the desired trans-1,1-diacetoxy-2-hexene
(7a) as the major product in 60-95% yields. However, the
formation of minor isomers 8 (5-15% by GC) was also detected
before the complete consumption of the starting aldehyde. It
appeared that the Lewis acid catalyzed not only the formation
of allylic gem-diacetate 7 but also its rearrangement to vinyl
acetates 8. The allylic rearragement is particularly undesirable
The geminal dicarboxylates (4, X ) OCOR′ in 1), also known
as “acylals”,35,36 are a simple addition product of acid anhydrides
and aldehydes and are well suited as a substrate for reaction as
in eq 1. The seemingly labile gem-diester functionality has
served as an interesting protective group for aldehydes due to
its considerable stability to acid yet ease of removal under
very mildly basic conditions, complementing the commonly
used base stable acetal-protecting group.37 A wide variety of
Lewis and Bro¨nsted acids can catalyze this process to produce
the diesters of allylic 1,1-diols.38 Thus, the requisite allylic gem-
carboxylates are readily available from R,â-unsaturated alde-
hydes 5. Alternatively, they can be prepared from the corre-
sponding propargylic acetates 6 by a novel Pd-catalyzed redox-
isomerization and intermolecular addition reaction.39
and complicates the interpretation of the result because the
stereochemical outcome of the asymmetric alkylation is inevi-
tably affected by the presence of racemic chiral 8. Thus, it was
necessary to quench the reaction before a significant amount
of the desired product underwent rearrangement in order to
secure high isomeric purity at the expense of yield. Among the
various acid catalysts tested, ferric chloride gave the best result,
giving rise to desired gem-diacetate 7a (R ) C3H7) in 67% yield
and greater than 95% isomeric purity. Table 1 summarizes the
range of enals examined in the current study.
The gem-diesters possessing a carboxy moiety other than an
acetoxy group could also be prepared.48 Employing propionic
and isobutyric anhydride in the same manner as acetic anhydride
furnished the corresponding gem-dipropionates and gem-di-
isobutyrates (Table 2). The reaction of crotonaldehyde (11) with
acetic anhydride generated a mixture of desired gem-diacetate
12a as the major product and the isomeric vinyl acetates of
type 8 as minor products (entry 4). Similarly, both the gem-
dipropionate 12b and the gem-diisobutyrate 12c were prepared
in good yields albeit with somewhat lowered isomeric purity
of 89% (entries 5 and 6). Nevertheless, simple chromatographic
purification increased their isomeric purity to 94%.49
The Pd-catalyzed alkylation in eq 1 has been demonstrated
in a nonchiral fashion through the utilization of allylic gem-
diacetates (X ) OAc).40,41 In these studies, it was demonstrated
that gem-diacetates could undergo an allylic substitution reaction
with stabilized nucleophiles to afford products such as 2 and 3.
It is worth noting that the alkylation generated monoalkylated
products in good yields despite the presence of two leaving
groups in the starting gem-diacetate. Partial or complete control
of the product distribution was achieved, depending on the
nature of gem-diacetates and nucleophiles.
While employment of a chiral ligand could potentially induce
asymmetry for this process, it is of interest to see whether the
guidelines of chemo- and regioselectivities established within
the context of nonchiral reactions can be applicable to the
asymmetric process. In this regard, the AAA of the present
While most of the gem-diesters were prepared from their
corresponding R,â-unsaturated aldehydes, some were synthe-
(42) Trost, B. M.; Lee, C. B.; Weiss, J. M. J. Am. Chem. Soc. 1995,
117, 7247.
(43) Man, E. H.; Sanderson, J. J.; Hauser, C. R. J. Am. Chem. Soc. 1950,
72, 847.
(44) Michie, J. K.; Miller, J. A. Synthesis 1981, 824.
(45) Liebermann, S. V.; Connor, R. Organic Syntheses; Wiley & Sons:
New York, 1955; Collect. Vol. III, p 441.
(46) Deka, N.; Kalita, D. J.; Borah, R.; Sarma, J. C. J. Org. Chem. 1997,
62, 1563.
(34) Gomez-Bengoa, E.; Heron, N. M.; Didiuk, M. T.; Luchaco, C. A.;
Hoveyda, A. H. J. Am. Chem. Soc. 1998, 120, 7649.
(35) Wegschneider, R.; Spath, E. Monatsh. 1909, 30, 825.
(36) Hurd, C. D.; Cantor, S. M. J. Am. Chem. Soc. 1938, 60, 2677.
(37) Green, T. W.; Wuts, P. G. M. ProtectiVe Groups in Organic
Synthesis, 3rd ed.; Wiley: New York, 1999; p 306.
(38) Sydnes, L. K.; Soderberg, B. C. Tetrahedron 1997, 53, 12679.
(39) Trost, B. M.; Brieden, W.; Baringhaus, K. H. Angew. Chem., Int.
Ed. Engl. 1992, 31, 1335.
(40) Lu, X.; Huang, Y. J. Organomet. Chem. 1984, 268, 185.
(41) Trost, B. M.; Vercauteren, J. Tetrahedron Lett. 1985, 26, 131.
(47) Li, T. S.; Zhang, Z. H.; Fu, C. G. Tetrahedron Lett. 1997, 38, 3285.
(48) Kochhar; Kanwarpal, S. B.; Balkrishna, S. D.; Rajadhyaksha, R.
P.; Pinnick, S. N.; Harold, W. J. Org. Chem. 1983, 48, 1765.
(49) Commercially available 12a (Aldrich) was only 85% isomerically
pure.

AAA Using gem-Dicarboxylates, Part I
J. Am. Chem. Soc., Vol. 123, No. 16, 2001 3673
Table 1. Preparation of Allylic gem-1,1-Diacetates Using Ferric Chloride
1
a Isolated yield. b E:Z ) 10:1 by GC. c E:Z ) 8.3:1 by H NMR. d Isomeric purity measured by GC.
sized from propargylic esters by the palladium-catalyzed reaction
as summarized in Table 3. First, propargylic acetate 6a was
subjected to the previously developed conditions which typically
employed 5% Pd2dba3‚CHCl3, 50% triphenylphosphine, and 5
equiv of acetic acid (entry 1).39 When heated in toluene for 4
h, 6a was cleanly converted to gem-diacetate 13a as a single
isomer in 87% yield. Performing the same reaction with Pd-
(PPh3)4 as catalyst also proved effective, giving a yield of 89%
(entry 2). It should be noted that only 0.5% catalyst and 1.5
equiv of acetic acid were required in this case. Propargylic
acetates bearing a stereogenic center were reacted under these
modified conditions to give gem-diacetates without affecting
the stereochemical integrity (entries 3-5). While gem-diacetate
13b was obtained in 51% yield from the TBDMS-protected ester
6b, switching the ligand to the more electron-rich and sterically
demanding tri(o-anisyl)phosphine gave a somewhat better yield
of 61%. When the TBDMS protective group was replaced with
the bulky and robust TBDPS group, more significant improve-
ment was achieved as the yield of 13c increased to 81%.
Initial Alkylation Results. With the requisite gem-diesters
in hand, the investigation on their asymmetric alkylation
commenced with trans-cinnamyl 1,1-diacetate (7i) as a substrate.
The alkylation with sodium dimethyl alkylmalonates 14 and 15
was carried out with 2.5% π-allylpalladium chloride dimer (16),
and 7.5% chiral ligand 1750,51 (P:Pd ) 3:1) in THF at ambient
temperatures. The nucleophile was generated from 2.0 equiv
of sodium hydride and 2.5 equiv of dimethyl methyl- or
benzylmalonate. The reactions proceeded to completion in 2 h
to give the desired products 18 and 19 as single regioisomers
in 92 and 75% yields, respectively (eq 3). The reaction using
Pd2dba3‚CHCl3 as the palladium source gave rather capricious
results, leading to inconsistent yields and ee’s. The alkylation

3674 J. Am. Chem. Soc., Vol. 123, No. 16, 2001
Table 2. Preparation of Gem-Dicarboxylates
Trost and Lee
a Isolated yields. b Isomeric purity of the crude product measured by GC.
Table 3. Preparation of gem-Diacetates from Propargylic Acetatesa
a All reactions were run in toluene at 110 °C. b Isolated yields.
products could be readily isolated in analytically pure forms
by typical flash chromatography on a silica gel column.
were greater than 95%. To confirm this conclusion, the
alkylation products were derivatized to the O-methylmandelate
esters.53 Due to a facile retro-aldol process upon hydrolysis of
the acetyl group, 18 and 19 were first reduced with LAH to
triols 20 and 21 and subsequently condensed with both (R)-
and (S)-O-methylmandelic acid using DCC independently to
provide the corresponding esters 22 and 23 in good yields
(Scheme 1). The signals for the two vinylic protons and
1
quaternary alkyl groups were completely resolved in the H
NMR spectra of the two diastereomers from (R)- and (S)-O-
methylmandelic acid. The appearance of only one set of signals
from 22 and 23 clearly indicated that the de was >95%, and
thus the ee of 18 and 19 was >95%.
(50) Trost, B. M.; Van Vranken, D. L. Angew. Chem., Int. Ed. Engl.
1992, 31, 228.
(51) Trost, B. M.; Van Vranken, D. L.; Bingel, C. J. Am. Chem. Soc.
1992, 114, 9327.
(52) Sullivan, G. E. Top. Stereochem. 1978, 10, 287.
(53) Trost, B. M.; Belletire, J. L.; Godleski, S.; McDougal, P. G.;
Balkovec, J. M.; Baldwin, J. J.; Christy, M. E.; Ponticello, G. S.; Varga, S.
L.; Springer, J. P. J. Org. Chem. 1986, 51, 2370.
The ee of the products was determined by chiral shift studies
using (+)-Eu(hfc)3 in CDCl3.52 Only a single set of signals was
observed in both cases, suggesting that the ee’s of these products

AAA Using gem-Dicarboxylates, Part I
J. Am. Chem. Soc., Vol. 123, No. 16, 2001 3675
Scheme 1. Derivatization to O-Methylmandelate and Absolute Configuration
Table 4. Comparison of Ligands in Alkylations of 10aa
entry ligand time (h) %yieldb of 24 %yieldb of 26 % eec of 24
The O-methylmandelate method to obtain diastereomers also
allows for assignment of the absolute stereochemistry of the
alkylation reaction using R,R-ligand 17.53 The 1H NMR signals
of the two vinylic protons Ha and Hb in (S)-mandelate 22
appeared at δ 5.74 and 5.58, whereas δ 6.39 and 5.98 were
observed for the corresponding chemical shifts in the (R)-
mandelate. The methyl group on the quaternary carbon also
followed the well-established pattern of change in chemical
shifts: δ 0.64 for the (S)-mandelate and δ 0.53 for the (R)-
mandelate. This shielding-deshielding pattern was equally
pronounced for the chemical shift (shown in parentheses) of
1d
2
PPh3
dppp
dppp
17
27a
27b
27c
27d
27e
27f
2
0.5
18
2
4
80
33
0
68
86
54
68
0
57
86
0
0
0
-
-
3e
4
-
89
68
77
5
6
7
8
9
10
2
12
24
24
1
5
∼0
no reaction
no reaction
70
-
-
0
-
-
(-)13f
(-)10f
-
1
11
12
13
14
27g
27h
27i
4
9
4
5
82
0
-
0
benzyl series 23. On the basis of these H NMR correlation
results, the absolute configuration of the carbinol chiral centers
could be assigned to be “R” as depicted.
no reaction
88
55
70
27j
0
(-)19f
Ligand Comparison. With the excellent regio- and enantio-
selectivities obtained from the alkylation of the cinnamyl system,
the investigation was then extended to other substrates. The
straight-chain derivative 10a was subjected to the standard
alkylation conditions (eq 4). The reaction with 14 proceeded
smoothly to generate acetate 24 as a single product in 68% yield.
An ee of 89% was obtained from chiral shift experiments on
desilylated alcohol 25 which was prepared in a quantitative yield
by treatment of 24 with TBAF. Performing the alkylation
reaction at 0 °C increased the yield to 85% and the ee to 91%.
Through a number of repeated runs, it was found that the amount
of the catalyst could be lowered to 0.5% of 16 and 1.5% of 17
with no detrimental effect on the yield or enantioselectivity.
a All reactions were performed in THF (0.2 M in substrate) with
1.5 equiv of nucleophile in the presence of 2.5% of 16 and 7.5% of
the ligand unless otherwise noted. b Isolated yields. c Enantiomeric
excess determined by chiral shift studies on the desilylated alcohol 25.
d 20% of the ligand was used. e 3.0 equiv of nucleophile were employed.
f The ee of the opposite enantiomer, (ent)-24.
by various ligands was expected to exert differential chiral
recognition in the enantio-discriminating ionization step due to
the different steric and electronic properties of each chiral ligand.
Thus, the analysis of the outcome in terms of similarities and
dissimilarities of these ligands would provide a basis for
evaluating the structural and electronic aspects of the ligand
effect. The ligands 27a-27e share the same chiral scaffold,
while they differ in the structure of linkers and metal binding
sites. Other ligands containing anthracenyl (27f and 27g),
diphenyl (27h), tartrate- (27i), or sugar-derived (27j) chiral
scaffolds were also tested for the alkylation. The C2-symmetry
present in the standard ligand 17 no longer exists in ligands
27b and 27d.
Although many ligands did generate the alkylation product
24, none of these ligands gave a better result than the standard
ligand 17. A modest level of ee’s was still obtained with
naphthyl ligand 27a54 and non-C2-symmetric ligand 27b. On
the other hand, despite the identity of the chiral scaffold, ligand
27c possessing a sulfonamide linker generated a mixture of
monoalkylated 24 and dialkylated 26 in which the monoalky-
lated product was almost racemic (entry 7). Changing the metal
binding site from a phosphine to a nitrogen(s) resulted in no
reaction (entries 8 and 9). Ligands 27f and 27g, based on an
anthracenyl scaffold, gave results inferior to those seen with
the standard conditions. The “invertomer” ligand 27g in which
the amide linkage was inverted gave a similar result as the
“normal” ligand 27f.55 Surprisingly, diphenyl derivative 27h,
Having established a protocol for ee measurement, the
reaction of eq 4 was carried out with various achiral and chiral
ligands as summarized in Table 4. The alkylation was first run
with achiral ligands. The use of triphenylphosphine as the ligand
generated monoalkylated product (rac)-24 in 80% yield (entry
1). In contrast, the reaction employing dppp went to completion
in only 0.5 h to give a mixture of mono- and dialkylated products
(entry 2). In the presence of excess nucleophile, the same
reaction produced only the dialkylated product (rac)-26 as a
single regioisomer in 86% yield (entry 3). A variety of chiral
ligands synthesized by a modular method (Figure 1) were tested
for the reaction (entries 4-14).51 The chiral pocket provided
(54) Trost, B. M.; Bunt, R. C. Angew. Chem., Int. Ed. Engl. 1996, 35,
99.

3676 J. Am. Chem. Soc., Vol. 123, No. 16, 2001
Trost and Lee
Figure 1. Chiral ligands synthesized by a modular method.
Table 5. Solvent Effect in Alkylations of 10a with Dimethyl
Table 6. Effect of Base and Additive in Alkylations of 10a with
Methylmalonatea
Dimethyl Methylmalonatea
entry
solvent
THF
benzene
CH2Cl2
1,4-dioxane
DME
DMF
DMSO
THF
base
NaH
NaH
NaH
NaH
NaH
NaH
NaH
Cs2CO3
Cs2CO3
Cs2CO3
Cs2CO3
time (h)
% yieldb
% eec
time
(h)
%
%
entry base
solvent
additive
yieldb
eec
1
2
3
2
4
3.5
2
2
1
68
61
65
65
93
91
89
92
80
68
81
89
66
65
81
88
1
2
3
n-BuLi THF-hexane
-
-
-
-
-
-
-
4
2
5
6
6
90
68
57
64
58
69
89
73
86
48
39
-
83
87
65
60
28
-
NaH
KH
THF
THF
THF
4
4d KH
5
6d
7d
8
6
5
6
7
8
9
t-BuOK THF
Cs2CO3 THF
TMGe THF
NaH
NaH
(-)5e
39
10 92
1
10 no reactionf
10
12
10
5
THF
THF
CH2Cl2
CH2Cl2
CH2Cl2
CH2Cl2
Me4NBr
20% 18-C-6
-
Me4NBr
Hex4NBr
-
2
3
3
3
3
86
88
65
79
83
9
10
11
CH2Cl2
CH3CN
DMF
12
(-)12e
(-)13e
10 NaH
11 NaH
12 NaH
13 DBU
a All reactions were performed with 2.0 equiv of base and 2.5 equiv
of nucleophile at room temperature in the presence of 1.0% 16 and
3.0% of ligand 17 unless otherwise noted. b Isolated yields. c Deter-
mined by chiral shift studies on the desilylated alcohol 25. d The
reactions were run at 0 °C. e ee of the opposite enantiomer, (ent)-25.
12 no reactionf
a All reactions were run with 2.0 equiv of base and 2.5 equiv of
nucleophile at room temperature in the presence of 1.0% 16 and 3.0%
of ligand 17 unless otherwise noted. b Isolated yields. c Determined by
chiral shift studies on alcohol 25. d The reaction was run with 3.8% 16
which showed the same efficacy as 17 in a number of
cases,51,56 was found to be absolutely ineffective for the reaction.
While tartrate derived ligand 27i gave a moderate ee, mannitol
derived ligand 27j containing a sterically demanding metal
binding site proved much less effective. It is noteworthy that
the configuration of the major enantiomer can be predicted by
a sterochemical mnemonic which provides a structural and
physical link between the ligand chirality and the mode of
enantioselection.13,42,51 On the basis of this ligand screening
result, the standard ligand, 17, was used for all of the following
studies.
Solvent Effect. The solvent has an effect on both the
ionization and nucleophilic addition steps in the catalytic cycle,
thereby potentially influencing the ee.57 As summarized in Table
5, the use of solvents other than THF for the reaction of eq 4
provided inferior results relative to the standard reaction. In the
cases of benzene and methylene chloride, the low solubility of
the nucleophile in these solvents led to the formation of a
heterogeneous reaction mixture and a longer reaction time.
While a modest range of ee’s was observed for these nonpolar
solvents (entries 2 and 3), ethereal solvents gave significantly
better results (entries 4 and 5). In particular, the alkylation using
DME afforded an almost identical ee as the standard reaction.
Surprisingly, the use of polar solvents resulted in a dramatic
drop of the ee despite the rapid conversion and good yields
and 11% 17 at 0 °C. e 1,1,3,3-Tetramethylguanidine. Unreacted starting
f
material 10a and aldehyde 9 were recovered.
(entries 6 and 7). When cesium carbonate was used instead of
sodium hydride, the effect of solvent polarity was more striking
(entries 8-11). The major enantiomer from polar solvents had
the opposite stereochemistry although the ee’s in this series were
much lower.
Counterion Effect. The dependence of the ee on the solvent
leads to speculation that counterions may also have an effect
since their role in the nucleophilic addition step should be
different in a given solvent due to the differential solvation
energy. Thus, a series of alkali metal hydride and carbonate
bases were tested for the alkylation (Table 6). Due to the kinetic
inefficiency of lithium hydride as base, the lithium enolate was
generated by the addition of a hexane solution of n-BuLi to a
THF solution of dimethyl methylmalonate at -78 °C. Indeed,
different bases had a strong effect on the enantioselectivity of
the reaction. Interestingly, the relationship between the ee and
the size of cations exhibited a nonlinear behavior wherein
sodium proved to be most effective in contrast to our earlier
work involving nucleophilic attack on the π-allyl intermediate
as the enantiodiscriminating event.17 The use of sodium hydride
provided a homogeneous reaction mixture and the fastest
reaction. In the case of lithium, an aggregated insoluble reaction
mixture, which remained as a heterogeneous suspension through-
out the course of the reaction, gave an ee of 69% (entry 1).
Large differences in the ee were noted within a category of the
same cation (entries 3-5). Whereas the ee and yield were
(55) Trost, B. M.; Breit, B.; Peukert, S.; Zambrano, J.; Ziller, J. W.
Angew. Chem., Int. Ed. Engl. 1995, 34, 2386.
(56) Trost, B. M.; Shi, Z. J. Am. Chem. Soc. 1996, 118, 3039.
(57) Reichardt, C. SolVents and SolVent Effects in Organic Chemistry,
2nd ed.; VCH: Weinheim, Germany, 1988.

AAA Using gem-Dicarboxylates, Part I
J. Am. Chem. Soc., Vol. 123, No. 16, 2001 3677
Table 8. Alkylation of 10a with Dimethyl Benzylmalonate
Table 7. Concentration and Temperature Effects in the Alkylation
of 10a
16
17
temp equiv concn time
%
%
entry (mol %) (mol %) (°C) of Nu (M)
(h) yielda eeb
16
17
temp
(°C)
equiv concn time
%
%
entry (mol %) (mol %)
of Nu (M) (h) yielda eeb
1
2
3
4
2.5
2.5
2.0
0.5
7.5
7.0
6.0
1.5
25
0
0
1.2
2.0
4.0
1.2
0.20
0.24
0.20
0.35
2.5
2
2
58
86
98
76
80
90
90
90
1
2
3
4
2.5
2.5
2.0
1.0
2.5
1.0
0.5
8.0
7.0
6.0
3.0
7.5
3.0
1.5
25
0
0
1.6
1.2
1.5
0.50 1.5
68
76
85
87
99
89
91
91
90
90
0.29
0.14
0.55
0.35
0.20
0.16
2
2
6
8
3
3
0
3
0 to rt 1.5
5
-40
1.8
2.5
2.0
a Isolated yields. b Chiral shift studies on the desilylated alcohol 29
using (+)-Eu(hfc)3 in CDCl3.
6c,d
7d
0
94 (-)93e
-5
87
93
protocol, the same ee could still be obtained when the catalyst
loading was reduced to 0.5% 16 and 1.5% 17.
a Isolated yields of 24. b Determined by chiral shift studies on the
desilylated alcohol 25. c (S,S)-17 was employed. d Slow addition pro-
cedure. e Opposite enantiomer, (ent)-24.
In the initial studies with cinnamyl derivative 7i, the use of
15 as the nucleophile gave the same high level of ee (>95%)
as dimethyl methylmalonate (14). By contrast, the alkylation
of 10a with 15 provided a much lower ee of 80% under the
standard conditions (eq 5 and Table 8). While the ee jumped to
90% by simply running the reaction with 2.0 equiv of the
nucleophile at 0 °C, the additional increase of the nucleophile
to 4.0 equiv did not lead to further improvement. The application
of the slow addition protocol to this system proved beneficial
again as a comparable result could be obtained with only 1.2
equiv of the nucleophile and a much smaller amount of the
catalyst (entry 4).
improved with higher catalyst loading and lower reaction
temperatures, the enolate generated by potassium tert-butoxide
resulted in a significantly lower ee. A much lower ee and a
longer reaction time resulted when cesium carbonate was
employed (entry 6). An attempted reaction with a nonmetal
organic base such as TMG and DBU failed to produce the
desired product (entries 7 and 13).
The notion that the low solubility of the nucleophile might
be responsible for low ee’s led to the exchange of the counterion
for a more soluble alkylammonium or crown ether complexed
cation. The addition of tetramethylammonium bromide caused
a slight lowering in ee; whereas, the ee obtained with a crown
ether was indistinguishable with that of the standard reaction
(entries 8 and 9). The deterioration of ee by addition of alkyl-
ammonium salts was equally observed when methylene chloride
was used as a solvent (entries 10-13). In particular, as the size
of the cation increased from tetramethyl- to tetrahexylammo-
nium, the ee further decreasedsopposite to the case wherein
nucleophilic attack on a π-allyl constitutes the enantiodiscrimi-
nating event.17
Regiochemistry. The ionization of allylic gem-dicarboxylates
by a Pd(0) catalyst generates π-allylpalladium complex 30 in
which the allyl unit is differentially substituted by a carboxy
group and an alkyl group (Scheme 2). Due to this unsymmetric
structure, the subsequent substitution can occur at either allylic
termini to generate regioisomers, the “proximal” and “distal”
products with respect to the acyloxy substituent. However, the
alkylations of 7i (R ) Ph) and 10a (R ) TBDPSOCH2)
uniformly gave single regioisomers of type 31 which were
derived from nucleophilic attack at the carbon bearing the
acetoxy group (path “a”).
To investigate the factors affecting regioselectivity, substrates
with different substituents were examined (eq 6 and Table 9).
An unsubstituted system (R ) H) was first subjected to the
alkylation with dimethyl methylmalonate (entries 1-3). The
previous work showed that the use of N,O-bis(trimethylsilyl)-
acetamide (BSA) as base and PPh3 as ligand generated 34a
arising from distal attack exclusively.41 In contrast, the alkylation
using sodium hydride as base and dppp as ligand reoriented
the regioselectivity to the proximal attack to give 33a as the
major product. The alkylation was completed in only 0.5 h to
provide 79% of monoalkylated products as a 1.4:1 mixture of
inseparable regioisomers and 7% of dialkylation product 35a.
Using chiral ligand 17, the alkylation under standard conditions
generated a 5.5:1 mixture of regioisomers in 81% yield. The
major regioisomer 33a was derived from the proximal attack,
and no dialkylated product 35a was isolated. However, major
isomer 33a had only 13% ee, presumably due to a rapid
enantioface interconversion process. The possibility of a back-
ground reaction was excluded by a control reaction in the
absence of the palladium catalyst which proceeded very slowly
Temperature and Concentration Effect. The large change
in the enantioselectivity with different solvents and counterions
clearly indicates the strong influence of the reaction parameters.
Other factors of kinetic importance are the temperature and
concentration. In early studies, there was a small but discernible
effect of the temperature on the ee. Furthermore, the observed
reaction rate, though not rigorously measured, indicated that
faster reactions generally gave better ee’s. On the basis of these
observations, it was anticipated that the ee could be improved
by performing the reaction at higher concentrations or by using
excess nucleophile. However, attempts to carry out the alkylation
at high concentrations (>1.0 M) in THF met with difficulty
due to the limited solubility of the nucleophile. The use of a
large excess of nucleophile was not desirable because subsequent
isolation and analysis of the alkylation products became
problematic. To avoid these experimental difficulties, high
concentration of nucleophile was achieved by a slow addition
procedure in which substrates were slowly added to a mixture
of the nucleophile and the catalyst.
The results of alkylations of 10a at various temperatures and
concentrations are summarized in Table 7. As described before,
an ee of 89% obtained from the reaction at ambient temperature
was increased to 91% when the same reaction was performed
at 0 °C (entries 1-4). Further lowering of the reaction
temperature to -40 °C did not improve the ee (entry 5). No
reaction occurred when the alkylation was attempted at -78
°C. Interestingly, the ee was rather insensitive to the amount of
the catalyst and the final concentration. On the other hand, the
ee was increased to 93% by the slow addition protocol (entries
6 and 7). It should be noted that using this slow addition

3678 J. Am. Chem. Soc., Vol. 123, No. 16, 2001
Trost and Lee
Scheme 2. Regiochemistry of the Nucleophilic Addition to a π-Allyl Complex
Table 9. Alkylation of gem-Diacetate with Dimethyl
Methylmalonatea
ratioc
%
time % yieldb
yieldb
%
entry
R
ligand (h) 33 + 34 33:34 (E:Z)c 35
eed
1e
2
3
4
5
6
7
8
H
H
H
CH3
CH3
CH3
PPh3
dppp
17
PPh3
dppp
17
11
0.5
0.5
0.5
2
2
4
1
5
70
79
81
71
69
86
69
62
73
0:1 (20:1)
1.4:1 (9.5:1)
5.5:1 (1.4:1)
2.7:1 (20:1)
0
7
0
0
-
-
13 (-)
-
2.9:1 (12:1) 24
-
nucleophile, the alkylation of 12a gave a 67% yield of two
allylic isomers 37 and 38 in a much lowered 2.3:1 regioselec-
tivity (eq 8). In contrast to the alkylation with 14 (see Table 9,
entry 6), the minor product possessed only (E)-geometry.
Surprisingly, a discrepancy between the ee of the major (37,
90% ee) and the minor (38, 83% ee) isomers was noted.
11:1 (6.3:1)
2.5:1 (1.7:1)
5.7:1 (2.7:1)
12:1 (2.0:1)
0
0
0
0
92 (43)f
-
(CH2)2CH3 PPh3
(CH2)2CH3 dppp
-
9g (CH2)2CH3 (S,S)-17
93h (nd)
a All reactions were run at 0 °C in THF with 2.0 equiv of 14 in the
presence of 2.5% 16 and 7.5% ligand except for entries 4 and 7 where
5% Pd(PPh3)4 was used as catalyst. b Isolated yields. c (E)- and (Z)-
isomeric ratio of the vinyl acetates 34 measured by GC. d The ee of
the major regioisomer 33 determined by the 1H NMR integration ratio
of the derivatized tris-O-methylmandelate esters. e Reference 41. f The
ee of the 11:1 mixture of (E)- and (Z)-34b determined by GC ratio of
the derivatized (S)-R-methylbenzylamides. g Slow addition procedure
was applied. h The ee of (ent)-33c.
to give only vinyl acetates 34a as the product in less than 5%
yield at room temperature after 24 h.
While chiral shift studies revealed the ee of the major
regioisomers, their absolute configuration was determined by
derivatization to the corresponding tris-O-methylmandelates in
the same manner as Scheme 1. On the other hand, analysis of
the minor products was carried out by derivatization to (S)-R-
methylbenzylamides 41 as shown in Scheme 3.58 The sequence
involving acetalization, oxidation, and amide formation reactions
uneventfully furnished the amide derivatives. GC integration
of 41 readily revealed the de and thus the ee of 34b and 38.
Leaving Group. The key element of chiral induction in the
present system is differentiation of the two enantiotopic leaving
groups in the ionization step where the two leaving groups
experience a chiral recognition process. It was anticipated,
therefore, that bulkier leaving groups might give better enan-
tioselection. Along these lines, two series of geminal dipropi-
onates and diisobutyrates were subjected to the alkylation, and
their results were compared with those from the gem-diacetates
(eq 9 and Table 10). Under the standard conditions, gem-
dipropionate 12b was reacted with dimethyl methylmalonate
to produce a 5.5:1 mixture of isomers 42b and 44b in almost
quantitative yield (entry 2). The ee’s of major isomer 42b and
the mixture of minor products 44b were determined to be 92
and 51%, respectively. On the other hand, the alkylation of gem-
diisobutyrate 12c, containing the bulkiest leaving group within
the series, gave a much lower regio- and enantioselectivity (entry
3). In the series of gem-dicarboxylates 10, all of the experiments
were performed under the standard conditions and duplicated
by using (S,S)-17 as the ligand (entries 4-9). While the
Alkyl substituted gem-diacetates were next examined (entries
4-9). The regiochemical preference for the “proximal” attack
was more clearly manifested regardless of the ligand employed.
Within the achiral reactions, dppp generally provided a better
regioselectivity than triphenylphosphine. Unlike the unsubsti-
tuted system, the alkylation of these alkyl substituted systems
with ligand 17 induced high enantioselectivities, giving 33b and
33c in 92% and 93% ee, respectively. Interestingly, the
asymmetric reactions exhibited higher regioselectivities than
achiral reactions in accord with the results from the unsubstituted
system. When the alkyl substituent became significantly bulky
as in eq 7, complete regioselection could be achieved. The
reaction of isopropyl substituted 7b furnished 36 as a single
product in 95% ee. However, with sterically more congested
substrates such as 7d, 7e, 7k, and 13a, the alkylation did not
take place even at an elevated reaction temperature (65 °C). In
these cases, the starting gem-diacetates or the corresponding
R,â-unsaturated aldehydes were recovered.
The nucleophile also had an influence on the regioselectivity.
Using less reactive dimethyl sodiobenzylmalonate (15) as the
(58) Hoye, T. R.; Koltun, D. O. J. Am. Chem. Soc. 1998, 120, 4638 and
references therein.

AAA Using gem-Dicarboxylates, Part I
J. Am. Chem. Soc., Vol. 123, No. 16, 2001 3679
Scheme 3. Derivatization of the Vinyl Acetates to (S)-R-Methylbenzylamides
Table 10. Alkylation of gem-Dicarboxylate with Dimethyl Methylmalonatea
entry
R
R′
CH3
C2H5
CH(CH3)2
CH3
CH3
C2H5
C2H5
CH(CH3)2
CH(CH3)2
ligand
product
% yieldb
ratioc
% ee
1
2
3
4
5
6
7
8
9
CH3
CH3
CH3
(S,S)-17
(S,S)-17
(R,R)-17
(R,R)-17
(S,S)-17
(R,R)-17
(S,S)-17
(R,R)-17
(S,S)-17
(ent)-42a + (ent)-44a
(ent)-42b + (ent)-44b
42c + 44c
43a
86
99
93
85
76
95
97
87
83
11:1
5.5:1
1.6:1
-
92d (43)e
92d (51)e
70d (nd)
91f
TBDPSOCH2
TBDPSOCH2
TBDPSOCH2
TBDPSOCH2
TBDPSOCH2
TBDPSOCH2
(ent)-43a
-
89f
43b
-
91f
(ent)-43b
43c
(ent)-43c
-
89f
-
75f
-
74f
a All reactions were run in THF at 0 °C with 2.5% 16 and 7.5% ligand. b Isolated yields. c Isomeric ratio determined by GC. d Determined by 1H
NMR integration of the tris-O-methylmandelate. e The ee of the minor isomer (E:Z ) 6.8:1) determined by GC ratio of the (S)-R-methylbenzylamide.
f Chiral shift studies with (+)-Eu(hfc)3 on the desilylated alcohols.
alkylation in this series gave only one regioisomer as the
alkylation product, the ee showed a similar pattern as the methyl-
substituted substrates. Namely, the propionate-leaving group
gave a comparable ee, and the isobutyrate-leaving group led to
a precipitous decrease in ee (entries 8 and 9).
Diastereoselective Alkylation. So far, all of the substrates
employed in these asymmetric reactions had been achiral.
However, substrates possessing a stereogenic center adjacent
to the allylic alkylation site may experience effects on their
diastereoselectivity, thereby creating an issue of reagent versus
substrate stereocontrol. To address this issue, two chiral
substrates were subjected to the alkylation with dimethyl
sodiomethylmalonate (eq 11, Table 11). The alkylation of
d-mannitol derived 7c exhibited rather strong stereochemical
bias when achiral triphenylphosphine was used as ligand. The
use of (R,R)-17 reinforced the preference increasing the dr to
96:4. On the other hand, the formation of a 1:1 diastereomeric
mixture was observed with (S,S)-17 which apparently caused a
mismatched event. Similarly, acyclic chiral substrate 13c was
also examined (entries 4-7). The alkylation using dppp ligand
occurred in a nonselective fashion resulting in the formation of
the two diastereomers in equal amounts. Employment of
triphenylphosphine revealed a small stereochemical bias of the
substrate. The reversal of intrinsic bias, albeit to a small degree,
was realized by the use of (R,R)-17. However, the reaction
suffered from poor conversion as the starting material was
mostly recovered after prolonged reaction time. The slight
preference for 48b by triphenylphosphine ligand was switched
to complete control of diastereoselectivity by chiral ligand (S,S)-
17, which in this case represented a matched ligand.
The adverse effect of a larger leaving group was more clearly
noted when the alkylation was carried out with 15. With this
nucleophile, even the propionate-leaving group no longer
provided a result comparable to that of the acetate (eq 10). The
alkylation of dipropionate 12b with 15 gave rise to a mixture
of regioisomers in which the major isomer 44 had only 67%
ee. Furthermore, the reaction occurred more slowly, giving a
much lowered 2.3:1 regioselectivity. This result stands in
contrast to those from the alkylation of 12a with 14 (Table 9
entry 6) and the alkylation of 12a with 15 (eq 8).

3680 J. Am. Chem. Soc., Vol. 123, No. 16, 2001
Table 11. Diastereoselective Alkylation
Trost and Lee
1
a Isolated yields. b Diastereomeric ratio determined by H NMR integration.
Scheme 4. Mode of Enantioselective Ioinzation and Stereochemical Mnemonic
Discussion
series supports this interpretation. Nevertheless, for simplicity,
the cartoon depicts the time-average situation. Considering the
effective C2-symmetry of the ligand and the geometry of the
allyl unit (M or W shape), there are four possible transition
states for the ionization. First, two different trajectories, exo
and endo, for the ionization arise due to the structure of the
π-allyl complex in which the metal cants toward the anti
substituents to allow a better overlap between the metal and
the allyl unit. These two pathways differ in their orientation
relative to the π-allyl. The exo pathway involves departure of
the leaving group from the side of the syn substituent whereas
in the endo pathway the same event takes place from the side
of the anti substituent. Within these two pathways, the exo
pathway should be preferred since this trajectory is closer to
the optimum angle with respect to the newly forming pal-
ladium-carbon bond. Given this stereoelectronic argument, the
additional consideration of a steric interaction provided by the
chiral ligand suggests that the expulsion of the leaving group
through the raised flap should be favored. It is conceivable that
the difference in the transition-state energy between ionization
of this fashion and that of other modes becomes larger since
the development of an anionic charge on the leaving group is
accompanied by solvation in the transition state. This transition
state argument, to a first-order approximation, is equally appli-
cable to the nucleophilic addition step based on the principle
of microscopic reversibility (Xδ- ) Nuδ- in Figure 2).
The results of the present investigation demonstrate the
feasibility of discriminating two enantiotopic geminal leaving
groups by the AAA reaction. The allylic substitution reactions
using ligand 17 give only the mono-alkylated product with high
regio- and enantioselectivity. It is worth noting that the
corresponding desymmetrization by enzymatic hydrolysis or
acylation is not feasible with these 1,1-diol derivatives. More-
over, the absolute stereochemistry of the product can be correctly
predicted on the basis of a simple stereochemical mnemonic
which provides a structural and physical link between the ligand
chirality and the mode of ionization.13,42 High ee’s of the
alkylation product suggest that the clockwise or counterclock-
wise rotation of the ligand with respect to the substrate occur
in a highly selective fashion (see Scheme 4 and earlier work51
for a fuller description of this type of model). Apparently, this
ionization process accompanies control of the syn- and anti-
orientation of the allylic substituent within one mode of
ionization since the syn- and anti-π-allyl complexes would lead
to the formation of enantiomeric products.
The selection processes in the ionization and alkylation steps
can be understood by a simple model as depicted in Figure 2.59
This cartoon model retains the predictive power of the simple
clockwise or counterclockwise rotation-based mnemonic, and
more significantly, imparts physical interpretations. While the
model adopts a representation involving C2 symmetry, this
representation does not likely describe the instantaneous struc-
ture. The ligand bound to the metal likely adopts conformations
leading to non-C2 symmetrical structures whose dynamic
behavior creates a functional equivalent of C2 symmetry. The
observed memory effects60 and an X-ray structure55 in a related
In the exo ionization event that occurs through the raised flap,
the substrate can ionize through another transition state, leading
to the formation of a syn,anti-π-allyl complex (Figure 3).
However, this transition state involves the development of
(60) Trost, B. M.; Bunt, R. C. J. Am. Chem. Soc. 1996, 118, 235. Also
see: Lloyd-Jones, G. G.; Sephen, S. C. Chem. Eur. J. 1998, 4, 2539; Poli,
G.; Scolastico, C. Chemtracts: Org. Chem. 1999, 12, 837.
(59) Trost, B. M.; Toste, F. D. J. Am. Chem. Soc. 1999, 121, 4545.

AAA Using gem-Dicarboxylates, Part I
J. Am. Chem. Soc., Vol. 123, No. 16, 2001 3681
Figure 2. Transition state models for ionization (and nucleophilic addition).
Figure 3. Transition states for syn,syn- and syn,anti-π-allyl complex formation.
unfavorable A1,3-allylic strain between the anti-substituent and
the anti-hydrogen, and the steric hindrance between the anti-
substituent and the wall of the chiral pocket. Furthermore, the
preference for a syn,syn-geometry by ligand 17 is consistent
with all of the results from other acyclic systems such as the
1,3-dimethylallyl system.61,62 Thus, the high enantioselectivity
obtained from the present reaction is believed to stem initially
from the preference for the formation of the diastereomeric
syn,syn-π-allyl intermediate. Although this intermediate is not
necessarily the most stable π-allyl species, it may well be the
most reactive intermediate since the nucleophilic addition
process undergoes a reverse mechanistic pathway of ionization
with respect to the interactions between the chiral ligand and
the substrate. Therefore, this model can be applied to both the
ionization and nucleophilic addition steps to fully explain the
observed pattern of enantio- and regioselectivities.
of formation of the dialkylated product 26 is important in
interpretating the resulting enantioselectivity due to the kinetic
picture shown in Scheme 5. In the first alkylation, because (R,R)-
17 favors the ionization of the pro-S leaving group, reaction
“a” should be faster than reaction “b”, leading to an excess of
enantiomer 24 over (ent)-24. On the other hand, this same ligand
may promote the second alkylation via the ionization of (ent)-
24 to a greater extent than 24. As a consequence, the ee of 24
can be enriched by over-reaction at the expense of the yield.
Since the dialkylation was never observed in the reactions using
chiral ligand 17, the obtained ee should reflect the true chiral
discrimination. From the ligand-screening experiments, the
strong dependence of the enantioselectivity on all of the
components of the ligand is readily noted. Although it is rather
difficult at this point to formulate the requirements for a good
ligand in this reaction, the results clearly support that all of the
structural and electronic elements of the parent ligand 17 are
significant for the successful alkylation.
In contrast to the alkylations using achiral ligands, asymmetric
reactions produce only the monoalkylated product. The dialky-
lation seems to be a result of the rapid second ionization by the
sterically less demanding palladium(0)-dppp complex. The lack
Although solvents participate in many different aspects of
the reaction, often making interpretation of their specific role
very difficult, they have an effect on the ionization and
nucleophilic addition steps in the catalytic cycle which involve
charge development and neutralization processes.57 The degree
(61) Trost, B. M.; Krueger, A. C.; Bunt, R. C.; Zambrano, J. J. Am.
Chem. Soc. 1996, 118, 6520.
(62) Trost, B. M.; Radinov, R. J. Am. Chem. Soc. 1997, 119, 5962.

3682 J. Am. Chem. Soc., Vol. 123, No. 16, 2001
Trost and Lee
Scheme 5. Kinetic Amplification of Enantioselectivity by Dialkylation
of chiral recognition can vary due to the differential solvation
The results from the effect of temperature and concentration
of the incipient ionic π-allyl complex in the enantio-discriminat-
ing ionization step. Similarly, the rate of nucleophilic addition
is affected by solvation of the electrophile and the nucleophile,
both of which are charged species. The observations from the
solvent-screening experiments indicate that the origin of the
effect on the ee lies both in the polarity of solvents and their
coordinating ability. Ethereal solvents of intermediate polarity
(e.g. THF, DME, and 1,4-dioxane) provide the best results.
Nonpolar solvents appear to be a medium for good chiral
recognition although the subsequent nucleophilic addition is
sluggish due to their poor ability to solvate the nucleophile. To
the extent that the ionization is reversible, the slow rate of the
nucleophilic addition fails to reflect the initial enantiodiscrimi-
nation because the “mismatched” ionization becomes more
competitive in this situation. The opposite may be true for cases
with polar solvents. The good ee’s obtained with ethereal
solvents suggest that an optimal compromise may be reached
between the degree of enantioselective ionization and the rate
of the nucleophilic addition owing to the relatively nonpolar,
yet coordinating nature of these solvents.
The nonlinear relationship between ee and the ionic radius
of the cation attests to the complex nature of these alkylations.
Given that the enantio-discriminating event occurs in the
ionization step, the nature of the nucleophile should, a priori,
not affect the initial asymmetric induction process. Therefore,
the observed variance of ee appears to be the consequence of
an indirect effect derived from the rate of the nucleophilic
addition.62 Since the different counterions can cause the dif-
ferential solvation and aggregation of the nucleophile, it is
conceivable that the efficiency of quenching the π-allyl inter-
mediate by a nucleophile can vary, depending on the counterion.
In cases where the nucleophile is slow to react, reversal of the
“matched” ionization occurs and ultimately leads to significant
competition of the “mismatched” ionization. Since 0.5-2.5%
of (η3-C3H5PdCl)2 (16) was used as a palladium source, the
reaction mixture always contains 1.0-5.0% chloride ion.
Therefore, it is likely that the additive effects reside mostly in
the cation rather than anion.63 The addition of alkylammonium
salts did promote solvation of the nucleophile, providing
homogeneous reaction mixtures. However, this increased solu-
bility did not translate into an increase in the reaction rate.
Instead, nucleophiles with an alkylammonium cation generally
retarded the reaction, with the retardation increasing with
increasing size,64 and decreased the ee. Interestingly, the smaller
alkylammonium cations give higher ee’s in accord with the trend
observed in the solvent-screening experiment; faster reactions
give higher ee’s. This effect is again opposite to that observed
when nucleophilic attack is the enatiodiscriminating event.17
lead to a simple conclusion that a lower reaction temperature
and a higher nucleophile concentration give a better enantiose-
lectivity. The increase in ee observed in cooling from 25 °C to
0 °C seems to arise from better chiral recognition at a lower
reaction temperature. However, no improvement of the ee by
further lowering the reaction temperature suggests that the good
enantio-discrimination achieved in the ionization step should
be followed by a fast nucleophilic addition. Given the fact that
the enantioselectivity of these alkylations is rather insensitive
to the amount of the catalyst, the direct metal-metal displace-
ment mechanism is unlikely to be the source of lowering the
ee. The slow addition procedure, which would make occurrence
of this process more likely due to the increased catalyst
concentration with respect to that of substrate, generally gives
better ee’s instead.
The regioselectivity of the alkylation is primarily reflective
of the strong electronic effect of an oxygen atom that stabilizes
the R-cation through resonance favoring nucleophilic attack at
that carbon. This preference is reinforced by the effect of the
substituents as in the case of 7i (R ) Ph) and 10a (R )
TBDPSOCH2). While the aromatic ring favors formation of the
regioisomers in which the resultant double bond is conjugated
with the phenyl group (path “a” in Scheme 2), the silyloxy-
methyl substituent in 10a disfavors nucleophilic attack at the
adjacent carbon (path “b”) due to its destabilizing effect on a
partial cationic charge, and thus leads to a “proximal” alkylation
product. In addition to these substrate originated factors, the
ameliorative influence of the chiral ligand on the regioselectivity
is readily apparent from the results. The higher regiocontrol in
the asymmetric reaction relative to achiral reactions stems from
the chiral recognition by ligand 17. If the chiral ligand favors
a certain mode of ionization, the reverse pathway, which
involves a nucleophilic attack at carbon where the leaving group
was placed, should also be favored.
The formation of (E)- and (Z)-alkenes in the minor regio-
isomer is indicative of the presence of the π-σ-π isomerization
among the π-allyl intermediates prior to the nucleophilic addition
(Scheme 6). This isomerization represents an enantioface
interconversion process in the case of R ) H, presumably
leading to the poor enantioselectivity of the unsubstituted
system. It is important to note that (Z)-alkene was never
observed as the major isomer. This fact suggests that the
anti,syn-complex is not formed or that other π-allyl complexes
are more reactive. It is also possible that the anti,syn-complex
undergoes nucleophilic attack exclusively at the distal carbon.
The much lower ee’s of the minor products relative to the major
isomer is in part derived from the fact that the ee was determined
from a mixture of (E)- and (Z)-isomers. The configuration of
the distal carbon in the syn,syn- and syn,anti -isomers would
be opposite to each other because the π-σ-π mechanism
entails a π-facial interconversion process, leading to a change
(63) Burckhardt, U.; Baumann, M.; Togni, A. Tetrahedron: Asymmetry
1997, 8, 155.
(64) Trost, B. M.; Bunt, R. C. J. Am. Chem. Soc. 1998, 120, 70.

AAA Using gem-Dicarboxylates, Part I
J. Am. Chem. Soc., Vol. 123, No. 16, 2001 3683
Scheme 6. Isomerization of π-Allyl Complexes via π-σ-π
leaving groups become diastereotopic. While the leaving group
anti to the η2-palladium-olefin complex is ionized, the rotation
about the C1-C2 bond can change the orientation of the leaving
group. To the extent that a preequilibrium between these two
species is established via this bond rotation and the reversible
complexation, the ionization favors the formation of (syn)-50
over (anti)-50. If the R′ group is sterically demanding, steric
interactions between the carboxylate in 49s and the pocket might
begin to disfavor 49s relative to 49a. The fact that 1,3-
disubstituted allyl systems bearing sterically bulky substituents
are poor substrates with these catalysts support this contention.65
As a consequence, the degree of differentiation between the two
species becomes smaller, and thus competitive formation of
(ent)-2 lowers the ee. Alternatively, it is possible that the
preference of the remaining carboxy group to assume the anti
position becomes larger as the steric demand increases. Indeed,
such change of syn versus anti preference by chiral ligand 17
has been utilized in the stereoselective synthesis of a (Z)-
alkene.66
Mechanism
in both the configuration of the stereogenic center and the
geometry of the olefin. Therefore, the ee obtained from the (E)
and (Z) mixture of 34 should reflect the consequence of the
isomerization process to give a much lower ee than that of the
major isomer 33.
However, the lower ee of 38 relative to 37 raises an interesting
mechanistic question as to the origin of the minor product. The
factors that lead to higher ee’s (i.e. high concentration of the
nucleophile or slow addition of the substrate) suggest the
existence of a mechanism eroding the ee before the nucleophilic
addition step. Although the background reaction could form
vinyl acetate 38 in a racemic fashion, the slow rate of this
process in the control experiment and the exclusive (E)-geometry
make this possibility less likely to be responsible for the lower
ee. Instead, a simple rationalization evolves from the possibility
that the generated π-allylpalladium intermediates may have
differential reactivities toward the nucleophile. As is shown in
Scheme 7, ligand 17 generates the two diastereomeric π-al-
lylpalladium intermediates 47 and 48 with different rates. In
the case of the matched ionization, the nucleophilic addition
leading to the formation of allylic acetate 37 should be favored
since it constitutes another matched process (i.e. microscopic
reverse of the ionization). When the subsequent nucleophilic
addition is slow, the return of the π-allylpalladium intermediate
to starting diacetate 12a competes with alkylation. With the less
reactive nucleophile 15, more significant competition between
the matched and mismatched ionization would cause an erosion
of the ee. Furthermore, “distal attack” of the nucleophile to the
mismatched intermediate 48 becomes a matched process leading
to preferential formation of enantiomeric vinyl acetate (ent)-
38. Consequently, the ee of minor isomer 38 and the regiose-
lectivity of the overall alkylation deteriorate. Therefore, for the
maximum chiral induction and regioselection, the kinetically
preferred π-allyl intermediate 47 must be captured by the
nucleophile before it undergoes any kind of selectivity-lowering
process.
The similar acidity (pKa) of acetic (4.75), propionic (4.86),
and isobutyric (4.84) acids suggests that the unexpected inverse
relationship between the ee and the size of the leaving group
may be mostly steric in origin. In particular, precipitous drops
of the ee’s by changing leaving groups from a linear propionate
to a branched isobutyrate indicates that chiral ligand 17 has a
well defined chiral pocket to differentiate the branching in the
leaving group. However, the complicated nature of the chiral
recognition process, which originates from the two prochiral
elements, the two faces of the double bond, and the two
enantiotopic leaving groups, makes it rather difficult to rational-
ize these results. Nevertheless, a simplified picture wherein only
counterclockwise ionization is considered, may explain the
observed results (Scheme 8). Upon coordination of the pal-
ladium(0)-17 complex to the double bond, the two enantiotopic
The results of the diastereoselective alkylations attest to the
presence of the matched and mismatched ionizations. As is
evident from the results with achiral ligands, the two substrates
have a stereochemical bias. It is interesting to note that the
structural feature of these systems permits an opportunity for
1,4-asymmetric induction, and thus the alkylation product is
equivalent to a vinylogous Felkin-Anh adduct. The bias is
believed to arise from the conformational preference of the
allylic systems upon coordination to Pd(0) (Scheme 9). In the
case of achiral ligands, the initial facial selectivity in the
complexation more or less determines the de of the reaction.
On the other hand, a certain mode of ionization (clockwise vs
counterclockwise) is favored by the chiral ligands. Since the
coordination is reversible, this two-stage selection process allows
the chiral ligands to amplify the inherent stereochemical bias
of the substrates in the matched cases.
Considering the factors discussed thus far, a unified kinetic
picture that outlines the pathways for enantio- and regioselective
alkylation of gem-dicarboxylates can be formulated (Scheme
10). In this scheme, all of the favored pathways are drawn as a
solid line. With ligand (R,R)-17, the initial ionization that occurs
in a counterclockwise motion and generates a syn,syn-π-allyl
intermediate is kinetically preferred. While this intermediate can
undergo an isomerization process via a π-σ-π mechanism to
generate isomeric π-allyl complexes, the mode of the nucleo-
philic addition to each complex is again influenced by the chiral
pocket, providing control of regioselectivity. The major product
is derived from a series of kinetically favored events, each of
which constitutes a matched process. It is remarkable that the
alkylation using ligand 17 produces the major product out of
numerous kinetic possibilities in a highly enantio- and regiose-
lective fashion.
Conclusions
In summary, a new strategy in the enantioselective allylic
alkylation reaction has been developed using allylic gem-
dicarboxylates as substrates and alkylmalonates as nucleophiles.
The complex derived from palladium(0) and ligand 17 was most
effective, giving rise to the alkylation product in high enantio-
and regioselectivities. The sense of induced stereochemistry can
be correctly predicted on the basis of a simple stereochemical
mnemonic. Simple comparison of the results from the achiral
(65) Trost, B. M.; Oslob, J. D. J. Am. Chem. Soc. 1999, 121, 3057.
(66) Trost, B. M.; Heinemann, C.; Ariza, X.; Weigand, S. J. Am. Chem.
Soc. 1999, 121, 8667.

3684 J. Am. Chem. Soc., Vol. 123, No. 16, 2001
Trost and Lee
Scheme 7. Matched and Mis-matched Ionization and Nucleophilic Addition
Scheme 8. Effect of Leaving Group and Mechanism
Scheme 9. Matched and Mis-matched Coordination and Ionization
reactions and their asymmetric version reveals that chiral ligand
17 not only performs the alkylation in a highly enantioselective
fashion but also provides excellent chemo- and regioselectivities.
By systematic studies, various factors affecting the stereochem-
ical outcome of the reaction have been probed and analyzed.
Proper control of reaction conditions proves critical for obtaining
high ee’s (>90%). Among reaction parameters, the solvent and
the concentration of the nucleophile are most important. The
observed solvent and concentration effects suggest that two
conditions must be met in order to obtain good enantio- and
regiocontrol of the reaction: (1) a good level of chiral
recognition in the ionization step and (2) the effective capture
of the kinetic π-allyl intermediate in the nucleophilic addition.
A comparison of the results from reactions of structurally
different geminal diesters such as diacetates, dipropionates, and
diisobutyrates provides further mechanistic insight into the chiral
recognition process and the kinetic picture of the reaction. The
mode of asymmetric induction achieved in the present study
represents a novel process of differentiation between two
geminal carbon-oxygen bonds, a difficult transformation by
enzymatic methods. Synthetically, the present reaction consti-
tutes the equivalent of a stabilized nucleophile addition to a
carbonyl group with high asymmetric induction.
Experimental Section
1-Methoxy-1-(3′,3′-diacetoxy-1′(E)-propen-1′-yl)-cyclopentane (13a).
Procedure A. Freshly distilled acetic acid (0.70 mL, 12.2 mmol) was
added to a solution of Pd2dba3‚CHCl3 (41.4 mg, 0.040 mmol) and
triphenylphosphine (105 mg, 0.400 mmol) in degassed toluene (30 mL)
at room temperature. After the solution stirred for 5 min, a solution of
6a (0.785 g, 4.00 mmol) in toluene (20 mL) was added, and the resultant
mixture was heated at 110 °C for 4 h. The reaction mixture was cooled,
concentrated, and purified by flash chromatography on a silica gel
column (hexanes/ethyl acetate 15:1) to yield diacetate 13a as a clear
oil (0.897 g, 87%).
Procedure B. Acetic acid (0.43 mL, 7.52 mmol) was added to a
solution of Pd(PPh3)4 (29 mg, 0.025 mmol) in degassed toluene (20

AAA Using gem-Dicarboxylates, Part I
J. Am. Chem. Soc., Vol. 123, No. 16, 2001 3685
Scheme 10. Reaction Pathways for Asymmetric Alkylation of gem-Dicarboxylates
mL) at room temperature. After the solution stirred for 5 min, acetate
6a (0.985 g, 5.02 mmol) in toluene (10 mL) was added, and the resultant
mixture was heated at 110 °C for 12 h. The reaction mixture was cooled,
concentrated, and purified by flash chromatography on a silica gel
column (hexanes/ethyl acetate 15:1) to yield diacetate 13a (1.150 g,
89%): tr ) 7.94 min (GC); IR (film) νmax 1766, 1676, 1436, 1373,
and the progress was monitored by TLC or GC. After complete
consumption of the starting material or cessation of the reaction, the
reaction mixture was poured into 10% aqueous NaHSO4 and extracted
with ether or ethyl acetate. The combined organic extracts were washed
with brine, dried over anhydrous MgSO4, concentrated, and purified
by flash chromatography on a silica gel column with ethyl acetate/
hexanes (or petroleum ether) mixture as an eluent.
1
1242, 1204, 1069, 1009, 961 cm-1; H NMR (300 MHz, CDCl3) δ
7.15 (dd, J ) 6.0, 0.9 Hz, 1H), 6.02 (dd, J ) 16.0, 0.9 Hz, 1H), 5.65
(dd, J ) 16.0, 6.0 Hz, 1H), 3.11 (s, 3H), 2.09 (s, 6H), 1.89-1.83 (m,
2H), 1.80-1.57 (m, 6H); 13C NMR (75 MHz, CDCl3) δ 168.8, 139.7,
122.9, 89.3, 86.1, 50.9, 35.7, 23.0, 20.8; Anal. Calcd for C13H20O5: C,
60.92; H, 7.87. Found: C, 60.82; H, 7.71.
B. Slow Addition Procedure. The nucleophile (14 or 15) was
generated in the same manner as described in the standard procedure.
A solution of the palladium catalyst and the ligand in THF was added
to the nucleophile solution via a cannula. To this mixture was added a
solution of the geminal dicarboxylate in THF by a syringe pump over
an indicated period of time (3-5 h). After the addition was complete,
the reaction mixture was stirred for an additional hour and worked up
in the same manner as the standard procedure.
(S)-1,1-Diacetoxy-4-(tert-butyldimethylsilyloxy)-pent-2(E)-ene (13b).
Freshly distilled acetic acid (0.028 mL, 0.490 mmol) was added to a
solution of Pd2(dba)3‚CHCl3 (2.5 mg, 0.0022 mmol) and tri(o-
methoxyphenyl)phosphine (6.8 mg, 0.019 mmol) in degassed toluene
(1.5 mL) at room temperature. A solution of acetate 6b (62 mg, 0.242
mmol) in toluene (1 mL) was added, and the resultant mixture was
heated at 110 °C for 4 h. After cooling to room temperature, the reaction
mixture was concentrated and chromatographed on a silica gel column
(hexanes/ethyl acetate 15:1) to yield diacetate as 13b a clear oil (46.5
mg, 61%): tr ) 8.02 min (GC); [R]D +0.11 (c 2.37, CHCl3); IR (film)
νmax 1766, 1682, 1473, 1442, 1372, 1247, 1205, 1142, 1094, 1051,
Dimethyl (2′E,1′R)-2-[1′-Acetoxy-4′-(tert-butyldiphenylsilyloxy)-
2′-buten-1′-yl]-2-methylmalonate (24). Following the standard pro-
cedure, the reaction of diacetate 10a (0.325 g, 0.762 mmol) with
dimethyl sodiomethylmalonate (14), prepared from sodium hydride
(60% dispersion, 50 mg, 1.25 mmol) and dimethyl methylmalonate
(0.200 g, 1.37 mmol), was performed in THF (1.5 mL) at room
temperature for 2 h in the presence of π-allylpalladium chloride dimer
16 (7.0 mg, 0.019 mmol) and (R,R)-17 (42 mg, 0.061 mmol).
Purification by flash chromatography on a silica gel column (hexanes/
ethyl acetate 15:1) afforded acetate 24 as a colorless oil (0.265 g, 68%,
89% ee). Following the slow addition procedure, 10a (0.410 g, 0.964
mmol) in THF (3 mL) was reacted with 14 by adding to a mixture of
sodium hydride (48.7 mg, 1.93 mmol), dimethyl methylmalonate (0.325
g, 2.41 mmol), 16 (1.8 mg, 0.0049 mmol), and 17 (10.0 mg, 0.0145
mmol) in THF (2 mL) via a pump driven syringe at -5 °C over 3 h.
After stirring for an additional hour followed by a standard workup,
flash chromatography on a silica gel column (hexanes/ethyl acetate 15:
1) gave 24 as a colorless oil (0.429 g, 87%, 93% ee): [R]D -8.95 (c
1.05, CHCl3); IR (film) νmax 1748, 1590, 1429, 1372, 1227, 1112, 1022,
1
1007, 964 cm-1; H NMR (300 MHz, CDCl3) δ 7.14 (d, J ) 6.3 Hz,
1H), 6.02 (ddd, J ) 15.5, 4.4, 0.8 Hz, 1H), 5.70 (ddd, J ) 15.5, 6.3,
1.1 Hz, 1H), 4.39-4.30 (m, 1H), 2.08 (s, 6H), 1.21 (d, J ) 6.4 Hz,
3H), 0.89 (s, 9H), 0.05 (s, 3H), 0.04 (s, 3H); 13C NMR (75 MHz, CDCl3)
δ 168.82, 166.79, 141.3, 121.4, 89.2, 67.7, 25.7, 23.7, 20.7, 18.1, -4.9,
-5.0; HRMS Calcd for C13H25O3Si (M+ - C2H3O2) 257.1573, Found
257.1576.
Pd-Catalyzed Asymmetric Allylic Alkylation. General Proce-
dures. A. Standard Procedure. To a suspension of sodium hydride
in THF (or an indicated solvent) was added dimethyl methyl- or
benzylmalonate at 0 °C. The mixture was stirred until evolution of
hydrogen ceased. To this mixture was added a solution of π-allylpal-
ladium chloride dimer (16), ligand 17 (or an indicated ligand), and the
geminal dicarboxylate in THF (or an indicated solvent) via cannula.
The resulting orange solution was stirred at 0 °C (or room temperature),
1
824 cm-1; H NMR (300 MHz, CDCl3) δ 7.67-7.63 (m, 4H), 7.43-
7.35 (m, 6H), 6.02-5.99 (m, 1H), 5.84-5.83 (m, 2H), 4.19 (s, 2H),
3.71 (s, 3H), 3.70 (s, 3H), 2.05 (s, 3H), 1.47 (s, 3H), 1.04 (s, 9H); 13
C
NMR (100 MHz, CDCl3) δ 170.1, 169.7, 169.4, 135.6, 135.5, 134.7,

3686 J. Am. Chem. Soc., Vol. 123, No. 16, 2001
Trost and Lee
133.5, 133.4, 129.7, 127.7, 127.6, 122.9, 74.1, 63.2, 57.6, 52.6, 52.5,
26.6, 20.8, 19.0, 15.6; Anal. Calcd for C28H36O7Si: C, 65.60; H, 7.08.
Found: C, 65.79; H, 6.86.
the University of California-San Francisco, supported by the
NIH Division of Research Resources.
Supporting Information Available: Experimental proce-
dures for the preparation of all new compounds as well as
characterization data are included (PDF). This material is
available free of charge via the Internet at http://pubs.acs.org.
Acknowledgment. We thank the National Science Founda-
tion and the National Institute of Health, General Medical
Sciences, for their generous support of our programs. C.B.L.
thanks the Veatch Foundation for a fellowship. Mass spectra
were provided by the Mass Spectrometry Regional Center of
JA003774O