Enantioselective enolate protonation with chiral anilines: Scope, structural requirements, and mechanistic implications

Article abstract of DOI:10.1021/ja994437m

High enantioselectivity has been demonstrated in the protonation of N,N- diisopropyl amides (Table, 1, entries 1-4, 7, and 10-13) derived from certain β,γ unsaturated acids. Depending on double bond geometry and the degree of substitution at the γ-carbon, γ-protonation can be a competing reaction in the case of the aliphatic substrates 12, 14b, 14d, and 18. The evidence is most consistent with a mechanism that involves proton transfer from 1a to a mixed aggregate consisting of enolate 4a and the lithiated amide 5, but direct proton transfer from 1a to the enolate is not ruled out.

Full text of DOI:10.1021/ja994437m

4602
J. Am. Chem. Soc. 2000, 122, 4602-4607
Enantioselective Enolate Protonation with Chiral Anilines: Scope,
Structural Requirements, and Mechanistic Implications
E. Vedejs,* A. W. Kruger, N. Lee, S. T. Sakata, M. Stec, and E. Suna
Contribution from the Chemistry Department, UniVersity of Wisconsin, Madison, Wisconsin 53706, and the
Department of Chemistry, UniVersity of Michigan, Ann Arbor, Michigan 48109
ReceiVed December 20, 1999
Abstract: High enantioselectivity has been demonstrated in the protonation of N,N-diisopropyl amides (Table
1, entries 1-4, 7, and 10-13) derived from certain â,γ unsaturated acids. Depending on double bond geometry
and the degree of substitution at the γ-carbon, γ-protonation can be a competing reaction in the case of the
aliphatic substrates 12, 14b, 14d, and 18. The evidence is most consistent with a mechanism that involves
proton transfer from 1a to a mixed aggregate consisting of enolate 4a and the lithiated amide 5, but direct
proton transfer from 1a to the enolate is not ruled out.
Some years ago, we reported that the commercially available
reaction variables is complex. The current report summarizes
our efforts to address related issues in the protonation of amide
enolates. One goal of the studies was to define the essential
structural features of the chiral acids and the enolate substrates
required for high enantioselectivity. Another goal was to gain
insight into the mechanistic aspects of the process. The
clarification of transition state preferences has proved to be
difficult, but some progress has been made as described below.
chiral aniline 1a can be used for the enantioselective protonation
of hindered amide enolates 4.1 Subsequent work has established
that the process can be carried out under catalytic as well as
stoichiometric conditions,2 and that there is a qualitative
relationship between the optimum pKa value of the “chiral acid”
and the pKa of the carbonyl substrate.3 Thus, 1a is superior to
more acidic (1c) or less acidic (1b) analogues for the enanti-
oselective protonation of the strongly basic enolate 4. Only one
enantiomer of 1a is sold, but Noyori hydrogenation can be used
to prepare a quasi-enantiomeric diamine 2a that reacts with
complementary enantioselectivity.4 These chiral “acids” proto-
nate the enolates of hindered amides 3 with enantioselectivities
in the range of 90% ee or better. A variety of other chiral proton
donors are now known that can be used with certain prochiral
ketone enolates,5 and some of the best results have been reported
using a specific â,γ-unsaturated ester enolate as the substrate.6
Many of these studies have reached excellent levels of enan-
tioselectivity above 95% ee, but usually the range of substrates
is limited, and the relationship between enantioselectivity and
Results
Most of the detailed optimization experiments have been
performed using the naproxen amide rac-3a as the substrate.
The hindered amide requires somewhat forcing enolization
conditions to ensure >98% conversion to the enolate 4, but
treatment with s-BuLi at -78 °C is sufficient, and the process
is easily reproducible if 1.75 equiv of the base is used. When
the resulting enolate 4 is quenched with TMSCl, careful removal
of solvent followed by NMR assay affords a 24:1 ratio of enol
silane isomers in the best experiments. According to NOE
evidence,7 the major enol silane is derived from enolate 4-Z,
so this is the isomer that is largely responsible for the
enantioselective protonation.
* Address correspondence to this author at the University of Michigan.
(1) Vedejs, E.; Lee, N.; Sakata, S. T. J. Am. Chem. Soc. 1994, 116, 2175.
(2) Vedejs, E.; Kruger, A. W. J. Org. Chem. 1998, 63, 2792.
In the first experiments, the enolate solution from rac-3a was
treated with 2 equiv of 1a at -78 °C, followed by quenching
at the same temperature after 30 min. Similar results were
obtained using a Lewis acid (BF3/Et2O) or a Brønsted acid (CF3-
CO2D) to quench the enolate, and the product (R)-3a was
obtained with 85% ee in both experiments. In the latter case,
the level of deuterium incorporation was less than 20% by NMR
assay, suggesting that the R-proton in (R)-3a comes from 1a
rather than from the quenching agent, but the issue was not
probed beyond the qualitative level at this stage of work.
Initially, the BF3 quenching procedure was used for com-
parison with an alternative enolate protonation method that relies
on BF3 complexation of chiral amine ligands to force internal
proton return (ipr).7 The ipr method is very sensitive to
stoichiometry and order of mixing, and requires a mixed
aggregate intermediate derived from 1 equiv of a chiral amine
as well as 1 equiv of the corresponding lithium amide per
equivalent of the enolate for optimum enantioselectivity.7
However, the reaction of 4a with 1a proved to be relatively
fast, and proton transfer was >80% complete before the
(3) Vedejs, E.; Kruger, A. W.; Suna, E. J. Org. Chem. 1999, 64, 7863.
(4) Vedejs, E.; Trapencieris, P.; Suna, E. J. Org. Chem. 1999, 64, 6724.
(5) (a) Reviews: Fehr, C. Angew. Chem., Int. Ed. Engl. 1996, 35, 2567.
Yanagisawa, A.; Ishihara, K.; Yamamoto, H. Synlett 1997, 411. (b)
Nakamura, Y.; Takeuchi, S.; Ohgo, Y.; Yamaoka, M.; Yoshida, A.; Mikami,
K. Tetrahedron 1999, 55, 4595. Asensio, G.; Cuenca, A.; Gavina, P.; Medio-
Simon, M. Tetrahedron Lett. 1999, 40, 3939. Asensio, G.; Aleman, P.;
Cuenca, A.; Gil, J.; Medio-Simon, M. Tetrahedron: Asymmetry 1998, 9,
4073. Takeuchi, S.; Nakamura, Y.; Ohgo, Y.; Curran, D. P. Tetrahedron
Lett. 1998, 39, 8691. Asensio, G.; Aleman, P.; Gil, J.; Domingo, L. R.;
Medio-Simon, M. J. Org. Chem. 1998, 63, 9342. Yanagisawa, A.; Kikuchi,
T.; Kuribayashi, T.; Yamamoto, H. Tetrahedron 1998, 54, 10253. Yanag-
isawa, A.; Inanami, H.; Yamamoto, H. J. Chem. Soc., Chem. Commun.
1998, 1573. Aboulhoda, S. J.; Reiners, I.; Wilken, J.; Henin, F.; Martens,
J.; Muzart, J. Tetrahedron: Asymmetry 1998, 9, 1847. Mikami, K.;
Yamaoka, M.; Yoshida, A. Synlett 1998, 607. Yanagisawa, A.; Ishihara,
K.; Yamamoto, H. Synlett 1997, 411. Martin, J.; Lasne, M.-C.; Plaquevent,
J.-C.; Duhamel, L. Tetrahedron Lett. 1997, 38, 7181. (c) Piva, O.; Mortezaei,
R.; Henin, F.; Muzart, J.; Pete, J. P. J. Am. Chem. Soc. 1990, 112, 9263.
Piva, O.; Pete, J. P. Tetrahedron Lett. 1990, 36, 5157. Henin, F.; Muzart,
J.; Pete, J. P.; M’boungou-M’passi, A.; Rau, H. Angew. Chem., Int. Ed.
Engl. 1991, 30, 416. Muzart, J.; Henin, F.; Pete, J. P.; M’Boungou-M’Passi,
A. Tetrahedron: Asymmetry 1993, 4, 2531.
(6) (a) Fehr, C.; Galindo, J. J. Am. Chem. Soc. 1988, 110, 6909. (b)
Fehr, C.; Guntern, O. HelV. Chim. Acta 1992, 75, 1023. (c) Fehr, C.;
Galindo, J. HelV. Chim. Acta 1995, 78, 539. (d) Fehr, C.; Galindo, J. Angew.
Chem., Int. Ed. Engl. 1994, 33, 1888.
(7) Vedejs, E.; Lee, N. J. Am. Chem. Soc. 1995, 117, 891.
10.1021/ja994437m CCC: $19.00 © 2000 American Chemical Society
Published on Web 04/27/2000

EnantioselectiVe Enolate Protonation with Chiral Anilines
Table 1. Quenching of Amide Enolates with 1aa
J. Am. Chem. Soc., Vol. 122, No. 19, 2000 4603
and 73% ee, respectively, from the corresponding enolates.
The sense of enantioselection was confirmed by synthesis of
(R)-6a,b from the known acids, but attempts to cleave the
amides by hydrolysis encountered partial racemization. To avoid
this complication, a derivative of the R-arylpropionamide family
was sought that could be prepared with high enantioselectivity
from the enolate by protonation with 1a, and that could also be
cleaved to the corresponding acid with retention of stereochem-
istry. A solution to the problem was found with the hydrazide
7, available from N,N-dimethyl-N′-isopropylhydrazine and the
acid chloride from 8a and oxaloyl chloride. The hydrazide 7 is
isostructural with 3b, and the corresponding enolate reacts with
1a with nearly the same enantioselectivity (95% ee). Subsequent
hydrolytic cleavage was possible under oxidative conditions
using ceric ammonium nitrate (CAN) in HOAc/MeCN/water,9
and afforded the acid 8a in 64% yield. Conversion to the ester
entry
amide
ee, %
entry
amide
ee, %
1
2
3
4
5
6
7
8
9
3a
3b
3c
3d
6a
6b
7
97
10
11
12
13
14
15
16
17
10
97
97b
97
12
95e
97
14a
14b
14c
17a
17b
18
95c
79
95f
53
73
40g
32g
73g
95d
71
9a
9b
60
a Enolate generation in THF using s-BuLi at -78 °C; >90% recovery
after NH4Cl quench, 0 °C unless noted; assay by hplc analysis, chiral
stationary phase; (R)-configuration major unless noted. b The ee value
was determined after conversion to (R)-11. c Quench at -25 °C.
d Recovery: 83% after chromatography. e The isomer 13 was the major
product, 66% yield; 18% of 12 was obtained. f The product of
γ-protonation (15b) was isolated in 30% yield in addition to 14b (60%).
g Absolute configuration not assigned.
10
8b with Me3SiCHdN2 occurred smoothly in methanol and
quenching agent was added (NMR assay).7 The issue of
stoichiometry was therefore addressed, and it was found that
there is no specific requirement for the use of 2 equiv of the
chiral amine or a substantial excess of base. A careful experi-
ment using 1.1 equiv of sec-butyllithium and 1.0 equiv of 1a
followed by quenching with BF3/Et2O at -78 °C gave nearly
the same result (81% ee for (R)-3a) as in the previous
experiments. More recently, the corresponding conversion has
been performed under catalytic conditions using as little as 5
mol % of 1a together with excess sec-butyllithium and an achiral
proton source (PhCH2CO2Et), resulting in (R)-3a with >90%
ee.2 These observations have a number of ramifications to be
discussed later, but they prove that the protonation of 4a involves
direct proton transfer from 1a, and not the ipr process. On the
other hand, the results were more easily reproducible if excess
s-BuLi (1.75 equiv) was used with 2 equiv of 1a for enolate
quenching, so this procedure was adopted in subsequent
experiments.
subsequent hplc assay and comparison with authentic material
confirmed the (R) configuration with 93% ee. Thus, it is possible
to carry out the enantioselective enolate protonation using a
substrate that is readily cleaved to the carboxylic acid with
minimal loss of enantiomeric excess.11 This result was not
unexpected, but the similarity in ee for the hydrazide 7 vs the
corresponding amide 3b (Table 1, entry 2) is significant in the
context of transition state characteristics. Presumably, the
presence of an additional nitrogen in the enolate corresponding
to 7 could influence enolate stability and geometry because there
are additional options for lithium complexation. Since no change
in enantioselectivity was observed, it appears that the steric
similarity between 3b and 7 may be the dominant factor.
Certainly, the N,N-diisopropyl substituents in 3b are important.
This is apparent in the lower enantioselectivity seen with the
corresponding N,N-dimethyl amides 9a and 9b in the usual
experiment (Table 1, entries 8 and 9, 71% ee and 60% ee,
respectively).
Further optimization work established that the proton-transfer
process does not go to completion unless the mixture of 1a +
4a is warmed prior to quenching. In the case of rac-3a as the
substrate, warming to 0 °C followed by NH4Cl/H2O quench
gave ee values near 90%. The final improvement was realized
when the commercial material (the tartrate salt of 1a) was
recrystallized prior to neutralization with aqueous NaOH, and
the resulting 1a was recrystallized twice. With maximum
precautions taken, the reaction of 4a with 1a gave (R)-3a in
>95% yield and with 97% ee. If desired, 1a could be recovered
by simple acid/base extraction, recrystallized, and reused in
subsequent enolate protonation experiments.
When the same conditions were used with structurally related
amides rac-3b or rac-3c, essentially identical results were
obtained (Table 1, see entries 1-4). However, the flurbiprofen
amide rac-3d behaved differently, and gave (R)-3d with only
16% ee. The problem was traced to partial equilibrium between
the relatively acidic 3d and the corresponding enolate 4d in
the presence of lithiated aniline 5, and this was confirmed by a
control experiment.8 Fortunately, the resulting racemization of
3d could be suppressed by lowering the quenching temperature
from 0 to -20 °C. This adjustment in the usual procedure gave
(R)-3d with 95% ee.
To gain further insight regarding geometric factors that
influence enantioselectivity, a number of â,γ-unsaturated amide
enolates were compared as substrates for the asymmetric
protonation (Scheme 2). The cyclohexenyl analogue 1013 was
expected to mimic 3b in terms of geometry, and the expected
high enantioselectivity (97% ee) was confirmed (Table 1, entry
10). The reaction proceeded in excellent yield, sufficient to allow
the use of the resulting (R)-10 as the starting material for the
synthesis of chiral butenolides via diastereoselective osmylation
and intramolecular acyl transfer.14 To confirm the absolute
configuration of (R)-10, a chemical correlation was performed
with 3b. This sequence was based on the Birch reduction of 3b
to 11, followed by partial hydrogenation using the Wilkinson
(9) De Oliveira Baptista, M. J. V.; Barrett A. G. M.; Barton, D. H. R.;
Girijavallubhan, M.; Jennings, R. C.; Kelly, J.; Papadimitriou, V. J.; Turner,
J. V.; Usher, N. A. J. Chem. Soc., Perkin Trans. 1 1977, 1477.
(10) Hashimoto, N.; Aoyama, T.; Shioiri, T. Chem. Pharm. Bull. 1981,
29, 1475. Aoyama, T.; Shioiri, T. Tetrahedron Lett. 1990, 31, 5507.
(11) A hydrazide corresponding to 7, but derived from N,N′-diisopro-
pylhydrazine could also be deracemized. Because of the presence of an
N-H subunit, a temporary blocking procedure was used as follows: N-H
lithiation with 1 equiv of n-BuLi; N-silylation with TMSCl (1 equiv) at
-78 °C; enolate formation with 2 equiv of n-BuLi; protonation with 1a,
and quenching as usual. The recovered hydrazide after aqueous workup
was formed with 69-72% ee.
(12) Prepared from the known carboxylic acids (ref 13) via the acid
chloride and diisopropylamine; see the Experimental Section.
(13) (a) 10: Kon, G. A. R.; Nargund, K. S. J. Chem. Soc. C 1932, 2461.
(b) 12: Black, T. H.; Eisenbeis, S. A.; McDermott, T. S.; Maluleka, S. L.;
Tetrahedron 1990, 46, 2307. (c) 14c: Henin, F.; Mortezaei, R.; Muzart, J.;
Pete, J. P.; Piva, O. Tetrahedron 1989, 45, 6171. (d) Pirkle, W. H.; Welch,
C. J. J. Liq. Chromatogr. 1991, 14, 3387.
Variation of the R-alkyl substituent resulted in decreased
enantioselectivity. Thus 6a and 6b were obtained with 79% ee
(8) In a control experiment, (R)-3d (95% ee) was treated with the same
ratio of diamine 1a + lithioamide 5 that would be present after enolate
protonation in THF prior to workup. After 10 min at 0 °C, the mixture was
allowed to warm to 20 °C and was then quenched in the usual way to give
4c with 31% ee, indicating partial racemization.
(14) Vedejs, E.; Kruger, A. W. J. Org. Chem. 1999, 64, 4790.

4604 J. Am. Chem. Soc., Vol. 122, No. 19, 2000
Vedejs et al.
Scheme 1
Scheme 2
catalyst. It proved very difficult to separate unreacted 3b and
11 from 10, but sufficient material was obtained from the best
chromatography fractions to confirm that the absolute config-
uration of (R)-3b is the same as in the enolate protonation
experiment to form rac-10.
A cyclopentenyl analogue 12 was investigated next, and an
unexpected result was obtained. Enolate generation with s-BuLi
proceeded normally, and enolate quenching with NH4Cl/H2O
returned the starting rac-12. However, the corresponding
experiment using 1a to quench the enolate resulted in 13 as the
major product (79%) together with (R)-12 (18%; 95% ee). In
other respects, the reaction was typical in that the bright yellow
color of the enolate was discharged upon the addition of 1a at
-78 °C, and was replaced by the pink-orange color that is
characteristic of the N-lithiated derivative 5.
Similar regiochemical results were encountered with certain
other â,γ-unsaturated amides, depending on the details of
substitution at the γ-carbon. Thus, 14a15 reacted normally upon
treatment with s-BuLi followed by 1a to give (R)-14, 97% ee,16
with nearly quantitative recovery. No definitive evidence was
found for the isomeric amide 15a that would be formed by
γ-protonation of the conjugated enolate, although traces of 14b
were detected due to contamination in the starting material 14a.
In contrast, the same experiment starting from the Z-trisubsti-
tuted alkene isomer rac-14b afforded a substantial amount (30%)
of 15b (Z-isomer) in addition to (R)-14b, (95% ee), while rac-
14d gave 15d with >10:1 selectivity. The corresponding
reaction of the γ,γ-dimethyl amide rac-14c produced only the
desired amide, but the enantiomeric purity was significantly
lower (50% ee). Presumably, the absence of a branch point at
the â-carbon in 14c is the reason for the modest enantioselec-
tivity compared to 14b.
Partially constrained amide analogues of 14 were explored
in an attempt to clarify transition state preferences. Thus, 17a
and 17b were prepared from the known acids.17 The corre-
sponding dienolates would be able to adopt E or Z geometries
with respect to the exocyclic enolate double bond, but the
cyclohexenyl ring enforces a transoid conjugated dienolate
geometry. On the other hand, the amide 1818 (exocyclic
ethylidene derivative) can adopt E or Z geometries in a cisoid
conjugated enolate. Enolate formation and protonation with 1a
proceeded smoothly with 17a,b, and no products of γ-proto-
nation were detected, but 17a and 17b were recovered with
(15) Prepared from the lithium enolate of N,N-diisopropylpropionamide
and E-2-bromo-2-butene (for 14a) or Z-2-bromo-2-butene (for 14b) in the
presence of NiBr2 by analogy to the following: Millard, A. A.; Rathke, M.
W. J. Am. Chem. Soc. 1977, 99, 4833.
(17) Van Bekkum, H.; Van den Bosch, C. B.; Van Minnen-Pathuis, G.;
De Mos, J. C.; Van Wijk, A. M. Recl. TraV. Chim. Pays-Bas 1971, 90,
137.
(16) The absolute configuration is assigned by analogy to (R)-10.
(18) Whitesell, J. K.; Helbling, A. M. J. Org. Chem. 1980, 45, 4135.

EnantioselectiVe Enolate Protonation with Chiral Anilines
J. Am. Chem. Soc., Vol. 122, No. 19, 2000 4605
modest ee (40% ee and 32% ee, respectively).19 These selectivi-
ties resemble the 50% ee value observed in the asymmetric
protonation to afford 14c, a case where the dienolate is likely
to prefer a transoid geometry to minimize nonbonded interac-
tions. A higher ee value (73% ee) was obtained with the
constrained cisoid amide 18, but this system gave substantial
amounts of a second product (1.25:1 ratio) that could be the
conjugated isomer. Isomer separation failed and the NMR
characteristics of the new isomer were largely obscured by
signals due to 18.
Scheme 3
One additional example of a constrained amide was explored
(19a). This substrate is constrained to have a Z-enolate geometry,
similar to 4-Z (Scheme 1), but with the enolate carbons as well
as the nitrogen substituents constrained to a nearly planar
arrangement within the five-membered ring. The dark red
enolate 21 was formed in the usual way, and addition of 1a
caused a color change to an intense orange. When this solution
was warmed to 0 °C and quenched with NH4Cl/H2O, the
resulting 19a was completely racemic according to hplc assay.
Because of the unusual color behavior, it was suspected that
the enolate may not be completely protonated by 1a due to the
expected decrease in the ion pair pKa of 19a, resulting from
the cyclic structure.20 To test this possibility, the experiment
was repeated to the stage of the intensely orange solution (19a
+ 2 equiv of s-BuLi at -78 °C, followed by 2.5 equiv of 1a;
warmed to -50 °C), and excess iodomethane was added. This
caused the color to fade as the temperature was allowed to reach
0 °C. After aqueous quench, the experiment returned less than
3% of 19a, and NMR signals were observed consistent with
the formation of 20. Evidently, the enolate 21 is in equilibrium
with 19a, the amine 1a, and the corresponding N-lithio derivative
5, and C-methylation of 21 can take place. For final confirma-
tion, a similar experiment was performed where 21 was
generated, combined with 1a as before, and the resulting mixture
was quenched with CF3CO2D. This returned a ca. 6:1 mixture
of 19a:19b, suggesting that a minimum of ca. 15% of enolate
21 is present in the equilibrium mixture. This 15% is a lower
limit for the presence of enolate because CF3CO2D is likely to
induce substantial internal proton return involving the N-H
protons of 1a in an enolate-diamine complex.21 Thus, the cyclic
lactam enolate 21 is not basic enough for complete and
irreversible protonation by 1a, as required in an asymmetric
protonation experiment. An attempt to protonate 21 with the
more acidic 1c also gave racemic 20a, so this series of
experiments was not pursued.
An effort to clarify the importance of enolate geometry in
the flexible substrates 3a, 6, and 9 was initiated in the hope
that enantioselectivity in the asymmetric protonation might be
optimized by optimizing the enolate Z:E ratio. As already
mentioned, the usual s-BuLi conditions converted 3a into 4a-Z
and 4a-E in a ratio of 24:1 according to a highly optimized
assay based on conversion to the enol silane. This isomer ratio
is qualitatively consistent with the 97% ee if the minor enolate
isomer affords mostly the minor product enantiomer. However,
numerous attempts to test this idea by preparing enolate mixtures
enriched in the E-enolate failed because the same dominant
Z-isomer 4a-Z was formed using a variety of deprotonation
conditions. Apparent support for the importance of enolate
geometry in the asymmetric protonation was obtained when it
was found that the R-cyclopentyl analogue 6b produces a 4.7:1
Z:E ratio of 22-Z:22-E (Scheme 3). This ratio corresponds
qualitatively to the enantiomer ratio (6.3:1) for the asymmetric
protonation of 6b (73% ee) with 1a. Presumably, the lower
enolate Z:E ratio from 6b reflects a steric effect (cyclopentyl
vs phenyl) on the population of amide rotamers in the transition
state for deprotonation, so the effect of decreasing the size of
amide N-alkyl substituents was probed using 9b as the substrate.
This proved difficult because the corresponding enol silanes
23b-Z and 23b-E were exceptionally sensitive to hydrolytic
cleavage back to the starting 9b. The highest ratio (2.1:1) was
seen under conditions where hydrolytic cleavage was minimal
(<5%), so the 2.1:1 ratio probably corresponds to the enolate
ratio. If this is correct, then enolate geometry cannot be a
dominant factor in the case of 9b because the enantioselectivity
(71% ee) indicates an enantiomer ratio that is considerably
higher (5.9:1).
Evaluation of enantioselectivity for the individual enolate
isomers would require access to the same enolate, but with a
substantially different Z:E ratio. To this end, an independent
method for generation of the enolate 24 was investigated
following an analogy for making ester enolates from ketenes.22
Thus, distilled methyl phenyl ketene was added to LiNMe2 in
THF at -78 °C, and the resulting 24 was converted to 23b
with TMSCl. The isomer ratio was ca. 1.8:1 23b-Z:23b-E in
the best experiment, nearly the same as in the deprotonation
experiments starting from 9a, so this approach was not
investigated further. However, the result was useful in that it
supports the assignment of enolate geometry for 23b. Originally,
this was based on chemical shift analogies,23 but the assignment
is also consistent with least hindered attack by LiNMe2 in the
ketene trapping experiment.
(22) Baigrie, L. M.; Seiklay, H. R.; Tidwell, T. T. J. Am. Chem. Soc.
1985, 107, 5391. Kresz, G.; Runge, E.; Ruch, E. Justus Liebigs Ann. Chem.
1972, 756, 112.
(23) Preliminary assignment of the Z-enol silane geometry was made
based on chemical shift precedents fom ref 22 (the Z-enol silane SiMe3
signal upfield relative to the E-isomer). The assignment was confirmed in
the case of the enol silanes derived from 4a by an NOE experiment. In
contrast to the major enol silanes derived from 4a and from 9b, the major
enol silane isomer from 9a has a downfield chemical shift (δ 0.253 ppm,
major, vs -0.122 ppm, minor), suggesting that this may be the E-enol silane
23a-E. A definitive assignment could not be made using NOE methods.
(19) Absolute configuration not assigned.
(20) Fachetti, A.; Streitwieser, A. J. Org. Chem. 1999, 64, 2281.
(21) Seebach, D.; Boes, M.; Naef, R.; Schweizer, W. B. J. Am. Chem.
Soc. 1983, 105, 5390. Hogeveen, H.; Zwart, L. Tetrahedron Lett. 1982,
23, 105.

4606 J. Am. Chem. Soc., Vol. 122, No. 19, 2000
Vedejs et al.
Scheme 4
of NMR detection, and (R)-3a was isolated with 93% ee. These
tests prove that the R-proton in (R)-3a is derived from 1a.
The next series of experiments was designed to probe the
details of enantioselective proton transfer from chiral aniline
diamines to the enolate 4a. The initial purpose was to learn
whether the conversion from 4a to (R)-3a occurs in a single
step (direct H transfer to the amide R-carbon) or whether the
reaction occurs via an enol intermediate 27. Direct H-transfer
provides a clear role for the chiral diamine 1a in the enanti-
oselectivity determining step, and some of the evidence points
to such a scenario. For example, the qualitative correlation of
chiral acid pKa with enantioselectivity (maximum ee when the
pKa of the chiral aniline is ca. 3 units lower than the pKa of 3)3
is easy to understand in this situation. More strongly acidic
derivatives of 1 would react faster and less selectively, and this
is the empirical result. On the other hand, the apparent cor-
relation with aniline pKa values might be accidental. If the enol
27 is the kinetic product, as in the case of ketone enolate
protonation with a variety of acids,26 then diamine 1a or the
lithiated 5 might serve as chiral proton carriers that mediate
the conversion from enol to carbonyl. A similar process occurs
in the enantioselective ketonization of dienols,5c and a correlation
betwen aniline pKa and the ability of the aniline to act as a
proton transfer agent is plausible. Ketone enols having similar,
bulky substituents can be relatively long-lived,27a but we could
find no data directly relevant to the lifetimes of amide-derived
enols. Accordingly, the reaction of 4a with 1a was studied at
low temperatures using the REACT-IR system to monitor the
process vs time and temperature. Conversion from 3a to 4a
could be followed by the disappearance of amide carbonyl
signals at 1644 cm-1 and the appearance of a new absorption
at 1548 cm-1, tentatively assigned to the enolate CdC (OLi)
double bond. The latter signal vanished upon addition of 1a,
and the signals of 3a reappeared. However, no OH absorptions
of a possible enol intermediate 27 were detected, nor was there
any other indication of a transient intermediate other than 4a
over the range of temperatures from -78 to 0 °C. This evidence
argues against a long-lived enol intermediate, but the negative
result is not decisive.
An alternative approach was considered for testing the
possible involvement of the enol 27 in the enantioselectivity-
determining step. If 27 is an intermediate, then a chiral environ-
ment must be maintained until 27 has been converted to (R)-
3a. This is possible if 5 or unreacted 1a participates in the proton
transfer. A second possibility is that the enol 27 itself may be
chiral due to restricted rotation.27b If it is generated with a >98%
preference for one enantiomer in the protonation step, then 27
might be converted into (R)-3a without further participation by
1a. The latter possibility is suspect for several reasons, but it
can be discounted according to the evidence described below.
A qualitative test to probe the possible involvement of 5 or
1a in proton delivery at the enolate R-carbon was performed
using an analogue 29 designed to serve as a “suicide acid” in
the reaction with 4a (Scheme 5). Conversion to the N-lithiated
30 must occur if the enol 27 is formed in this experiment.
Attention was turned to the evaluation of the structural
features of 1a that are important for its use as an effective chiral
proton donor (Scheme 4). From the experimentally determined
DMSO pKa value of 27.7,24 it appeared likely that the aniline
proton is involved in enolate protonation. To test this supposi-
tion, 25a was prepared from 1a via the lithiated diamine 5a by
N-alkylation with iodomethane. As expected, 25a was not
effective in the asymmetric protonation of 4a. In fact, the
minimal 4% ee in the product 3a favored the (S)-enantiomer,
in contrast to the behavior of 1a. It was less clear whether the
piperidine N-H would be important, so two derivatives 26a
(from 1a and ClCO2Bu) and 26b (from 25b via reductive
methylation and hydrogenolysis over Pd/C) were compared, but
neither was effective (<5% ee with 4a). Thus, both of the N-H
subunits of 1a are essential for the enantioselective protonation
of amide enolates, presumably because both nitrogens participate
in complexation of lithium in the enolate.25 The result is
consistent with transfer of the aniline N-H proton to the enolate
carbon, either directly (one-step mechanism) or via the enol 27
and subsequent intramolecular proton transfer (two-step mech-
anism). The latter pathway could be mediated by the neutral
diamine 1a (present in excess) or by the N-lithio derivative 5
(two-step mechanism) acting as proton-transfer agents. A third
possibility is that proton transfer occurs during aqueous workup
and that the proton comes from the quenching agent. To choose
among these options, the deuterated diamine 28 was prepared
by dilithiation of 1a with excess s-BuLi, followed by quenching
with D2O. Best results were obtained by minimal handling of
28, so the unpurified diamine was used in the standard
enantioselective protonation experiment with 4a. The resulting
(R)-3a(D) was found to be >80% deuterated at the R-carbon
by NMR integration, and the product was obtained with 92%
ee. Attempts to purify the diamine 28 resulted in a lower percent
deuterium incorporation. A second experiment was performed
using the standard conditions for the reaction of 4a with 1a,
but the aqueous quench procedure was carried out with NH4Cl
in D2O. This gave no deuterium incorporation within the limits
(26) Zimmerman, H. E. Acc. Chem. Res. 1987, 20, 263.
(27) (a) Kresge, A. J. Chem. Soc. ReV. 1996, 25, 275. Kresge, A. J. Acc.
Chem. Res. 1990, 23, 43. (b) Seebach, D.; Wasmuth, D. Angew. Chem.
1981, 93, 1007. Kawabata, T.; Yahiro, K.; Fuji, K. J. Am. Chem. Soc. 1991,
113, 9694. Fuji, K.; Kawabata, T. Chem. Eur. J. 1998, 4, 373. Hughes, A.
D.; Price, D. A.; Shishkin, O.; Simpkins, N. S. Tetrahedron Lett. 1996, 37,
7607.
(24) This value was determined by Bordwell and Satish; see ref 7.
(25) Seebach, D. Angew. Chem., Int. Ed. Engl. 1988, 27, 1624. Juaristi,
E.; Beck, A. K.; Hansen, J.; Matt, T.; Mukhopadhayay, T.; Simson, M.;
Seebach, D. Synthesis 1993, 1271.
(28) Ott, H.; Hardtmann, G.; Denzer, M.; Frey, A. J.; Gogerty, J. H.;
Leslie, G. H.; Trapold, J. H. J. Med. Chem. 1968, 11, 7.
(29) Zinner, G.; Kliegel, W.; Ritter, W.; Bo¨lke, H. Chem. Ber. 1966,
99, 1678.

EnantioselectiVe Enolate Protonation with Chiral Anilines
J. Am. Chem. Soc., Vol. 122, No. 19, 2000 4607
Scheme 5
7, and 10-13) derived from certain â,γ-unsaturated acids. A
flexible, cisoid dienolate geometry appears necessary for the
highest selectivity, as indicated by the low selectivities with
the constrained transoid dienolates from 17a or 17b. Enolate
geometry may be important for the hindered N,N-diisopropyl
amides, but enantioselectivity with the N,N-dimethyl analogues
9a or 9b (Table 1, entries 8 and 9) is higher than expected if
each enolate isomer reacts with high and opposite enantiospeci-
ficity. Depending on double bond geometry and the degree of
substitution at the γ-carbon, γ-protonation can be a competing
reaction in the case of the aliphatic substrates 12, 14b, 14d,
and 18. The undesired protonation pathway is dominant in the
case of 12 and 14d, and results in the formation of the R,â-
unsaturated amides 13 and 15b, respectively. Fehr and Galindo
have reported analogous observations regarding the regioselec-
tivity of protonation in their study of cyclogeranate ester
dienolates.6c
The proton-transfer step is not sensitive to the presence of
lithium halide, in contrast to some of the other enantioselective
enolate protonation techniques, and the stoichiometry is not
critical if sufficient base is used for complete conversion to the
enolate. Most of the mechanistic evidence is consistent with
direct proton transfer from the aniline nitrogen of 1a to the
enolate carbon. This would account for the labeling results using
28, as well as the qualitative pKa trends. However, there is one
data point that argues against the simplest version of a direct
proton-transfer process. The experiment where 4a is treated with
29 to give racemic 3 cannot easily be understood if the transition
state for enantioselective protonation involves only the chiral
diamine and 4a. Formal rejection of this mechanistic possibility
is not warranted because the test experiment produces a
“negative” result. On the other hand, 29 is the only derivative
of 1 or 2 to give racemic 3a among many analogues that were
tested containing an alkyl group as well as a proton at aniline
nitrogen. We regard the formation of racemic 3a as mechanisti-
cally significant, partly because of the analogies (for example,
32 + 4a to give (S)-3a, 78% ee), but mainly because there is a
mechanism that is consistent with all of the evidence. This
mechanism involves proton transfer from 1a to a mixed
aggregate consisting of 4a and the lithiated amide 5. If this is
correct, then the protonation as well as the formation of the
aggregate must be fast compared to the direct proton transfer
from 1a to 4a to explain the high enantioselectivity under
catalytic as well as stoichiometric conditions. However, all three
components are present by the nature of the experiment, and
their involvement in the selectivity-determining step cannot be
ruled out. Fehr et al. have reached similar conclusions.6 The
complexity of the components in such a mechanism precludes
discussion of transition state models for the enantioselective
proton-transfer process, so one of the goals stated at the outset
cannot be addressed pending further study. However, the scope
of the enantioselective protonation of amide enolates is now
clear, and the reaction with 1a is successful with a relatively
broad range of enolate substrates. Further studies will be needed
to address the remaining mechanistic questions.
However, 30 should cyclize to an aziridine 31. If 27 is formed
in a chiral geometry that is converted into (R)-3a, then the
formation of 31 should be irrelevant, and the ee of the process
should be comparable to that using an isostructural proton donor
32. The latter was prepared from 334 via lithiation (n-BuLi)
and alkylation with n-PrI, and was shown to serve as an effective
chiral acid for the protonation of 3a, 78% ee for (S)-3a (32 is
quasienantiomeric compared to 1a). The synthesis of 29 was
challenging due to the reactivity of the “suicide” chloroethy-
lamino side chain, but a successful route was realized by
sequential N′-protection and N-chloroacetylation of 33 to give
34, followed by borane reduction and deprotection under
hydrogenolysis conditions. The resulting 29 was stable enough
for rapid chromatographic purification, and treatment with
s-BuLi at -78 °C gave the suicide product, the aziridine 31 as
required to use 29 as a mechanistic probe in the enantioselective
protonation. The key experiment was carried out in the usual
way by adding 29 to preformed 4a at -78 °C, followed by
aqueous NH4Cl quench at 0 °C. Aziridine 31 was formed as
expected, and 3a was isolated from the neutral products in good
yield. According to hplc assay, the product was completely
racemic. One added control experiment was necessary to show
that the LiCl byproduct of aziridine formation would not
interfere with the usual enantioselective protonation process.
The control was performed with the highly selective reaction
of 1a with 4a, and the product (R)-3a was obtained with 96%
ee in the presence of an equivalent of LiCl, generated in situ
under anhydrous conditions by reaction of TMSCl with s-BuLi.
Thus, formation of 27 in a chiral conformation that determines
enantioselectivity in a subsequent proton transfer is unlikely.
Furthermore, the experiment suggests that the chiral proton
donor or its N-lithiated derivative must be present in the
transition state for enantioselective protonation of the enolate
4a for high selectivity
Acknowledgment. This work was supported by the National
Institutes of Health (GM44724).
Supporting Information Available: Experimental proce-
dures and characterization (PDF). This material is available free
of charge via the Internet at http://pubs.acs.org.
Summary
Excellent enantioselectivity has been demonstrated in the
protonation of N,N-diisopropyl amides (Table 1, entries 1-4,
JA994437M