Ag-catalyzed diastereo- and enantioselective vinylogous mannich reactions of α-ketoimine esters. Development of a method and investigation of its mechanism

Article abstract of DOI:10.1021/ja8062315

An efficient diastereo- and enantioselective Ag-catalyzed method for additions of a commercially available siloxyfuran to α-ketoimine esters is disclosed. Catalytic transformations require an inexpensive metal salt (AgOAc) and an air stable chiral ligand

Full text of DOI:10.1021/ja8062315

Published on Web 11/04/2008
Ag-Catalyzed Diastereo- and Enantioselective Vinylogous
Mannich Reactions of r-Ketoimine Esters. Development of a
Method and Investigation of its Mechanism
Laura C. Wieland, Erika M. Vieira, Marc L. Snapper,* and Amir H. Hoveyda*
Department of Chemistry, Merkert Chemistry Center, Boston College,
Chestnut Hill, Massachusetts 02467
Received August 7, 2008; E-mail: amir.hoveyda@bc.edu
Abstract: An efficient diastereo- and enantioselective Ag-catalyzed method for additions of a commercially
available siloxyfuran to R-ketoimine esters is disclosed. Catalytic transformations require an inexpensive
metal salt (AgOAc) and an air stable chiral ligand that is prepared in three steps from commercially available
materials in 42% overall yield. Aryl- as well as heterocyclic substituted ketoimines can be used effectively
in the Ag-catalyzed process. Additionally, two examples regarding reactions of alkyl-substituted ketoimines
are presented. An electronically modified N-aryl group is introduced that is responsible for high reaction
efficiency (>98% conversion, 72-95% yields after purification) as well as diastereo- (up to >98:2 dr) and
enantioselectivity (up to 97:3 er or 94% ee). The new N-aryl unit is crucial for conversion of the asymmetric
vinylogous Mannich (AVM) products to the unprotected amines in high yields. Spectroscopic and X-ray
data are among the physical evidence provided that shed light on the identity of the Ag-based chiral catalysts
and some of the mechanistic subtleties of this class of enantioselective C-C bond forming processes.
1. Introduction
can be easily prepared and stored in air for an extended period
of time. Transformations are highly site-, diastereo-, and
enantioselective, require a commercially available siloxyfuran
and readily accessible substrates, and furnish products that bear
an R-quaternary amino ester.1 We have identified an electroni-
cally modified class of N-aryl ketoimines that readily undergo
AVM, affording products that can be converted to the unpro-
tected amines efficiently. Through the studies described below,
we offer insight regarding the identity of this emerging class
of Ag-based chiral catalysts for asymmetric Mannich reactions.7-9
Catalytic enantioselective additions of C-based nucleophiles
to ketoimines offer direct access to nonracemic N-substituted
quaternary carbon stereogenic centers, components of several
biologically active molecules.1 Ketoimines, however, are rela-
tively unreactive,2 often exist as a mixture of E and Z isomers,
and contain a sterically congested CdN bond that carries
difficult-to-differentiate substituents.3 There are only two re-
ported studies of Mannich-type additions4 to ketoimines. One
case corresponds to additions of aldehyde-derived enols to
geometrically constrained ketone-derived cyclic imines;5 another
disclosure outlines reactions of simple ketene acetals with
ketoimines that are mostly generated from methyl-substituted
ketones.6 Herein, we disclose a method for catalytic asymmetric
vinylogous Mannich (AVM) reactions of ketoimines; additions
are performed in the presence of a chiral Ag complex, which
(3) For a review on catalytic asymmetric reactions of ketoimines and
ketones, see: Riant, O.; Hannedouche, J. Org. Biomol. Chem. 2007,
5, 873–888.
(4) For a recent review of Mannich reactions of aldimines, see: (a)
Co´rdova, A. Acc. Chem. Res. 2004, 37, 102–112. (b) For select recent
reports of catalytic asymmetric Mannich reactions (not AVM) involv-
ing aldimines, see: Hamada, T.; Manabe, K.; Kobayashi, S. J. Am.
Chem. Soc. 2004, 126, 7768–7769. (c) Matsunaga, S.; Yoshida, T.;
Morimoto, H.; Kumagai, N.; Shibasaki, M. J. Am. Chem. Soc. 2004,
126, 8777–8785. (d) Hamashima, Y.; Sasamoto, N.; Hotta, D.; Somei,
H.; Umebayashi, N.; Sodeoka, M. Angew. Chem., Int. Ed. 2005, 44,
1525–1529. (e) Uraguchi, D.; Sorimachi, K.; Terada, M. J. Am. Chem.
Soc. 2004, 126, 11804–11805. (f) Cozzi, P. G.; Rivalta, E. Angew.
Chem., Int. Ed. 2005, 44, 3600–3603. (g) Song, J.; Wang, Y.; Deng,
L. J. Am. Chem. Soc. 2006, 128, 6048–6049. (h) Chi, Y.; Gellman,
S. H. J. Am. Chem. Soc. 2006, 128, 6804–6805. (i) Chen, Z.; Yakura,
K.; Matsunaga, S.; Shibasaki, M. Org. Lett. 2008, 10, 3239–3242.
(5) (a) Saaby, S.; Nakama, K.; Alstrup Lie, M.; Hazell, R. G.; Jørgensen,
K. A. Chem.sEur. J. 2003, 9, 6145–6154. (b) Zhuang, W.; Saaby,
S.; Jørgensen, K. A. Angew. Chem., Int. Ed. 2004, 43, 4476–4478.
(6) Suto, Y.; Kanai, M.; Shibasaki, M. J. Am. Chem. Soc. 2007, 129, 500–
501.
(1) For stereoselective synthesis of N-substituted quaternary carbon
stereogenic centers and quaternary R-amino acids, see: Cativiela, C.;
D´ıaz-de-Villegas, M. D. Tetrahedron: Asymmetry 2007, 18, 569–623.
(2) For catalytic enantioselective additions of C-based nucleophiles to
ketoimines, see: (CN additions) (a) Funabashi, K.; Ratni, H.; Kanai,
M.; Shibasaki, M. J. Am. Chem. Soc. 2001, 123, 10784–10785. (b)
Vachal, P.; Jacobsen, E. N. J. Am. Chem. Soc. 2002, 124, 10012–
10014. (c) Masumoto, S.; Usuda, H.; Suzuki, M.; Kanai, M.; Shibasaki,
M. J. Am. Chem. Soc. 2003, 125, 5634–5635. (d) Wang, J.; Hu, X.;
Jiang, J.; Gou, S.; Huang, X.; Liu, X.; Feng, X. Angew. Chem., Int.
Ed. 2007, 46, 8468–8470. (e) Rueping, M.; Sugiono, E.; Moreth, S. A.
AdV. Synth. Catal. 2007, 349, 759-764. (Alkylmetal additions) (f)
Lauzon, C.; Charette, A. B. Org. Lett. 2006, 8, 2743–2745. (g) Fu,
P.; Snapper, M. L.; Hoveyda, A. H. J. Am. Chem. Soc. 2008, 130,
5530-5541. (Allylmetal additions) (h) Wada, R.; Shibuguchi, T.;
Makino, S.; Oisaki, K.; Kanai, M.; Shibasaki, M. J. Am. Chem. Soc.
2006, 128, 7687–7691.
(7) For Ag-catalyzed Mannich-type reactions to aldimines (not AVM) in
the presence of 1, see: (a) Josephsohn, N. S.; Snapper, M. L.; Hoveyda,
A. H. J. Am. Chem. Soc. 2003, 125, 4018–4019. (b) Josephsohn, N. S.;
Carswell, E. L.; Snapper, M. L.; Hoveyda, A. H. Org. Lett. 2005, 7,
2711–2713.
9
570 J. AM. CHEM. SOC. 2009, 131, 570–576
10.1021/ja8062315 CCC: $40.75
2009 American Chemical Society

Vinylogous Mannich Reactions of R-Ketoimine Esters
A R T I C L E S
in 72% yield (entry 1, Table 2).11 The finding in entry 6 (Table
1) indicates that the o-methoxy group is beneficial to diastereo-
and enantioselectivity; it is the electron-withdrawing unit para
to the N-aryl, however, that is largely responsible for the
enhanced efficiency. The observation summarized in entry 7 of
Table 1 suggests that the positive effect of the o-methoxy group
is not simply due to steric effects. Reaction with a substrate
that bears an o-methyl-substituted N-aryl group (6) proceeds
to only 43% conversion and results in the formation of anti-7a
in 55% ee (77.5:22.5 er); it should be noted, however, that such
processes proceed with relatively high diastereoselectivity in
favor of the syn product isomer (anti/syn ) 8:92).
Table 1. Examination of Ketoimines Bearing Different N-Aryl
Groupsa
2.1.2. Range of Aryl-Substituted r-Ketoimine Esters. As the
data summarized in Table 2 indicates, aryl-substituted R-ke-
toimine esters undergo AVM reactions to furnish anti-7a-h
diastereoselectively (75:25-93:7 dr), in 81-92% ee (90.5:9.5
to 96:4 er) and 51-78% yield; minor diastereomers (syn-8a-h)
are typically obtained in 40-60% ee. (Data involving NaHCO3
quench are shown in the left column; reactions with acid quench,
the results of which are illustrated in the right-hand columns,
are discussed below).
As represented by the example in entry 9, ketoimines with
an o-substituted aryl group are not transformed with high
enantioselectivity (<40% ee); such processes are efficient
(>98% conv), however, and occur diastereoselectively, although
it is the syn isomer that is generated predominantly (anti/syn )
<2:>98). The outcome of the transformation shown in entry 9
of Table 2 is not as a result of electronic effects imposed by
the presence of the halide substituent; catalytic AVM reactions
of the corresponding R-ketoimine ester bearing an o-methylphe-
nyl moiety exhibits the same selectivity and reactivity profile.
It is worth noting that a similar diastereoselectivity preference
is observed when the N-aryl unit of the ketoimine bears an ortho
substituent (see entry 7, Table 1).
2.1.3. Effect of Workup Conditions on AVM Efficiency as
well as Diastereo- and Enantioselectivity. We next wondered
whether the lower diastereomeric ratios (data in left columns,
Table 2) and moderate yields of anti isomers could be improved
by a better understanding of the origin of the selectivity reversal
in reactions of 3 versus 4a (entries 1-3 and 5, Table 1). Such
considerations led us to determine that when the reaction of 2
and 4a (-78 °C) is quenched with aqueous NaHCO3, there is
71% conversion after only 30 min; syn-8a, however, is formed
predominantly (20:80 anti-7a/syn-8a vs 93:7 after 15 h), and
enantioselectivity for both isomers is low (<25% ee). After 6 h,
there is >98% conversion and anti-7a is obtained in 90% ee as
the major diastereomer of a 91:9 mixture.
a Reactions performed under N2 atmosphere. b Conversion levels and
diastereomeric ratios were determined by analysis of 400 MHz 1H NMR
spectra of unpurified products. c Enantiomeric ratio (er) values were
determined by HPLC analysis; see Supporting Information for details.
nd ) not determined.
2. Results and Discussion
2.1. Development of Ag-Catalyzed AVM of r-Ketoimine
Esters. 2.1.1. Enhancement of Reactivity through Electronic
Alteration of the N-Activating Group. We began by investigat-
ing the ability of 1, a small-molecule ligand (molecular weight
) 508.6 g/mol) used to promote Ag-catalyzed enolsilane
additions to o-anisidyl aldimines,7,8 to initiate the reaction of 2
with R-ketoimine ester 3 (see Table 1).10 Although high
diastereoselectivity is observed (>98:2 syn/anti; entries 1-3,
Table 1), reactions are inefficient (20-68% conv, 15 h) and
enantioselectivity is low (<35% ee); at lower temperatures (-78
°C), the desired products are not formed (entry 4).
To enhance AVM efficiency, we chose to alter the N-aryl
unit of the ketoimine by installing a nitro group para to the
imine. Such a modification allows AVM of 2 with 4a (entry 5,
Table 1) to proceed to >98% conversion at -78 °C, affording
anti-7a in 91% ee (95.5:4.5 er). The sense of diastereoselectivity
is reversed (vs 3), as the anti product isomer is now selectively
generated (dr ) 93:7). Pure anti-7a is obtained after purification
A plausible interpretation of the above data is that the addition
of aqueous NaHCO3 (Table 2) might not result in complete
quenching of the reaction at -78 °C (i.e., excess 2 is not
completely proto-desilylated);12 the syn-selective transformation
might occur as the solution is allowed to warm up. When the
reaction is quenched after extended periods of time (i.e., 10-
15 h), anti-7a is generated predominantly, since most of the
ketoimine has been consumed. Moreover, examination of the
unpurified mixture pointed to the formation of a byproduct,
derived from Michael addition of the unreacted enolsilane 2
(8) For Ag-catalyzed AVM reactions to aldimines in the presence of ligand
1, see: (a) Carswell, E. L.; Snapper, M. L.; Hoveyda, A. H. Angew.
Chem., Int. Ed. 2006, 45, 7230–7233. (b) For related investigations,
see: Uraguchi, D.; Sorimachi, K.; Terada, M. J. Am. Chem. Soc. 2004,
126, 11804-11805. (c) Yamaguchi, A.; Matsunaga, S.; Shibasaki, M.
Org. Lett. 2008, 10, 2319–2322. (d) Akiyama, T.; Honma, Y.; Itoh,
J.; Fuchibe, K. AdV. Synth. Catal. 2008, 350, 399–402.
(9) (a) For a review regarding Ag-catalyzed enantioselective reactions,
see: Naodovic, M.; Yamamoto, H. Chem. ReV. 2008, 108, 3132–3148.
(10) Ketoimine substrates exist predominantly as an E isomer (2.5-20:1
E/Z); there appears to be no correlation between diastereo- or
enantioselectivity and such ratios. For example, 4b (93% ee) is a 5:1
E/Z ratio, whereas 4h (94% ee) is nearly exclusively E (95:5).
Spectroscopic studies indicate that the ketoimine isomers can readily
interconvert at 22 °C.
(11) The absolute stereochemistry of anti product diastereomers in Table
2 is based on an X-ray crystal structure analysis. See Supporting
Information for details.
(12) Use of 2 equiv of enolsilane is required for efficient AVM (to achieve
>98% conv).
9
J. AM. CHEM. SOC. VOL. 131, NO. 2, 2009 571

A R T I C L E S
Wieland et al.
Table 2. Ag-Catalyzed AVM of Aryl-Substituted R-Ketoimine Estersa
a Reactions performed under N2 atmosphere; >98% conversion in all cases (substrate consumption based on internal standard). b Diastereomeric ratios
were determined by analysis of 400 MHz 1H NMR spectra of the unpurified reaction mixtures. c Yields are of the purified anti-7. d Enantiomer ratio (er)
values were determined by HPLC analysis; see Supporting Information for details. e Data for syn-8i as the major isomer.
with unsaturated lactones anti-7 and syn-8 (1H NMR and mass
gram scale, in three simple steps and 42% overall yield from
commercially available starting materials; the amino-acid-based
phosphine ligand can be purified by a single crystallization.
Moreover, the requisite AgOAc and siloxyfuran are com-
mercially available.
Ag-catalyzed AVM reactions can be easily performed on
significant amounts of R-ketoimine esters. As an example,
reaction with 2.0 g of 4a affords 2.24 g of purified anti-7a (88%
yield) with high selectivity (96:4 er). Furthermore, although
t-Leu-containing ligand 1 is optimal, the corresponding phos-
phine 9, bearing the less expensive valine moiety, promotes
AVM efficiently and in lowersbut usefulsselectivity; the
example in eq 1 is illustrative.
spectroscopic analysis).13
The above observations suggested that a quenching procedure
that more efficiently removes unreacted enolsilane at -78 °C
could enhance diastereo- and enantioselection (i.e., minimal
Mannich addition at elevated temperatures) as well as afford
higher yields (i.e., minimal Michael addition). Indeed, as
illustrated in entries 1-8 of Table 2 (right column), when
reaction mixtures are treated with HOAc in MeOH (vs aq.
NaHCO3), anti-7a-h are formed in 89:11 to >98:2 dr (vs 75:
25-93:7 dr), 93.5:6.5-97:3 er (vs 90.5:9.5-96:4 er), and
72-95% yield (vs 51-78% yield). As the data in entry 9
indicate, the modification in quenching conditions exerts little
or no effect on the AVM of o-substituted R-ketoimine ester 4i.
It should be noted that ketoimine substrates are required for
adventitious syn-selective AVM to occur at higher temperatures
during workup. Control experiments (internal reference) il-
lustrate that, after 15 h, >98% of the substrate is converted to
AVM products. It is therefore likely that retro-Mannich reactions
take place to a limited extent during the basic NaHCO3 quench,
generating the small amount of ketoimine.
As illustrated by the enantioselective synthesis of amino esters
7j and 7k, respectively, transformations can be performed with
heteroaromatic R-ketoimine esters. The lower selectivities
obtained for 7j versus 7k may be due to competitive chelation
of the Lewis basic furyl oxygen (vs OMe of the N-aryl) with
the Lewis acidic Ag-based complex.
The N-aryl group can be removed by a two-step procedure
that affords the desired amines in >80% yield; an example is
presented in Scheme 1. Several points regarding deprotection
of the N-aryl unit are worth noting: (a) Removal of the o-anisidyl
group in a corresponding AVM product (e.g., 3, entry 1, Table
1) is substantially less efficient (<25% yield). That is, the
sterically unencumbered amine of the N-aryl unit, generated
from reduction of the NO2 group, is likely to react more readily
with PhI(OAc)2 than the more sterically encumbered heteroatom
2.1.4. Issues of Practicality and Removal of the N-Acti-
vating Group. The Ag-catalyzed process is performed in the
presence of an air stable chiral ligand that can be prepared, on
9
572 J. AM. CHEM. SOC. VOL. 131, NO. 2, 2009

Vinylogous Mannich Reactions of R-Ketoimine Esters
Scheme 1. Removal of the N-Aryl Unit
A R T I C L E S
Table 3. Dependence of Ag-Catalyzed AVM Selectivity on
Temperaturea
entry
temp (°C)
7a/8ab
yield (%)c
7a er; ee (%)d
8a er; ee (%)d
1
2
3
4
5
-78
-50
-30
-15
+4
93:7
88
76
58
32
41
95.5:4.5; 91
85:15; 70
58:42; 16
63:37; 26
<55:45; <10
82.5:17.5; 65
85:15; 70
81:19; 62
79.5:20.5; 59
77:23; 54
33:66
14:86
10:90
8:92
a Reactions performed under N2 atmosphere; all reactions were
quenched at the specified temperature and kept at that temperature for
3 h before allowing to warm to 22 °C. All conversions ) >98%
(substrate consumption based on internal standard). b Diastereomeric
ratios were determined by 400 MHz 1H NMR analyses of product
mixtures prior to purification. c Yields are of purified products; with the
exception of entry 1 (yield of only anti-7a); total yields of isomeric
mixtures are shown. d Enantiomer ratio (er) values were determined by
HPLC analysis; see Supporting Information for details.
of the quaternary carbon. (b) Attempts to unmask the amine
through oxidative procedures involving DDQ (e.g., 1.5 equiv
of oxidant, 20:1 CH2Cl2/H2O, 22 °C, 2 h) results in <2%
conversion. (c) When ceric ammonium nitrate (CAN) is used,
unprotected 11 can be obtained in one step but only in 45%
yield.14 (c) Subjection of 7a to catalytic hydrogenation condi-
tions (1 atm of H2, 10% w/w Pd(C), EtOAc, 22 °C, 4 h) leads
to isolation of the saturated lactone bearing the p-nitroaryl
moiety in 79% yield along with 10% of the saturated aniline.
2.2. Structures of Chiral Ag-Based Complexes and Related
Mechanistic Issues. 2.2.1. Dependence of Selectivity on Tem-
perature. We verified the strong dependence of diastereoselec-
tivity on reaction temperature through systematic investigations;
the results of these studies are summarized in Table 3. Such
findings have notable mechanistic implications. The reversal of
selectivity (see entries 1 and 5, Table 3) implies a change in
the energetics of the catalytic cycle; alterations in the structure
of the chiral catalyst may also be responsible for different
stereochemical outcomes. To gain additional insight, we ex-
plored the identity of the active chiral complex.
2.2.2. Identity and Catalytic Activity of Ag-Based Chiral
Complexes. Treatment of 1 with AgOAc (1.0 equiv, 22 °C, THF)
for 5 min followed by filtration (to remove uncomplexed
AgOAc, as determined by IR analysis) and solvent removal
results in isolation of an air stable white powder that can serve
as an effective AVM catalyst. The transformations in entries
1-2 and 7 of Table 2 are initiated by 10 mol % (based on ligand
1; -78 °C, 15 h) of a 1 year old sample of the powder, affording
anti-7a-b and 7g with nearly identical reactivity and selectivity.
For example, in the presence of the aforementioned powder
sample, 7g is isolated in 95:5 dr, 97:3 er, and 89% yield. When
a solution of the powder (d8-THF) is allowed to cool to -78
Figure 1. 1H NMR spectra (400 MHz, d8-THF) of chiral Ag-based
complexes of 1.
when the sample is allowed to warm to 22 °C.15 These data
indicate that the powder consists of two complexes that rapidly
interconvert at ambient temperature.
When petroleum ether is added to a THF solution of the
powder, a crystalline solid is formed, which, as confirmed by
X-ray analysis, is the 2:1 complex 12 (Figure 2). In contrast to
the powder form, 12 is not an effective catalyst at -78 °C (10
mol %, 15 h: 48% conv) and delivers products with low
1
°C, there is a notable change in the H NMR spectrum (see
spectra a-e, Figure 1): while the signal for the amide proton
shifts downfield from δ 10.09 ppm (spectrum a, Figure 1, 22
°C), it splits into two peaks of approximately equal intensity at
δ 10.87 and δ 10.78 ppm at -78 °C (spectrum e, Figure 1).15
Signals recoalesce and return to their original chemical shifts
selectivity (anti-7a/syn-8a ) 33:67 dr; eranti ) 74:26; ersyn
)
83:17).
1
(13) Control experiments indicate that Michael addition is Ag-catalyzed
but does not stereoselectively consume one product diastereomer. Thus,
selective catalytic Michael reaction is likely not the cause of the
difference in dr values observed with different quenching procedures.
(14) Removal of the N-aryl group in 7a in the presence of CAN was
performed in the following manner: 3 equiv of CAN, 1:1 MeCN/
H2O, 0 °C, 10 min; 10 equiv of H2SO4, 0 °C, 10 min; Na2CO3 (pH )
10), wash with CH2Cl2.
At 22 °C, the chemical shifts of the signals in the H NMR
spectrum of the crystalline form are identical to those observed
for the powder.15 Upon cooling, an identical downfield shift of
the amide proton is observed; as shown in spectrum f in Figure
1, however, one of the singlets (δ 10.87 ppm at -78 °C) now
predominates (2:1 by peak integration; see below for more
details). The above observations imply that the powder consists
(15) For further details, see Supporting Information.
9
J. AM. CHEM. SOC. VOL. 131, NO. 2, 2009 573

A R T I C L E S
Wieland et al.
Second, peaks corresponding to the cationic 1:1 (calculated
m/z for 1•Ag+: 615.1331, found: 615.1327) and the cationic
2:1 complex (calculated m/z for 12•Ag+: 1123.3610, found:
1123.3625) are detected in the high-resolution mass spectra of
the powder as well as of the crystal.
Third, addition of excess AgOAc21 to a THF solution of 2:1
complex 12 causes an increase in the amount of the catalytically
more active system (1H NMR analysis);15 nearly identical results
are obtained from this sample as observed with the aforemen-
tioned powder.15 Similarly, addition of excess 1 to a solution
of the crystalline form (see spectrum f, Figure 1 and spectrum
c, Figure 3) affords a sample that is >98% 12 (<2% 1:1
complex).
Figure 2. X-ray structure of Ag-based complex 12 (Ar ) p-OMeC6H4).
The availability of a sample of 12 uncontaminated by the
1:1 complex has allowed us to substantiate that the former is a
relatively ineffective catalyst at -78 °C (∼25% conv, 15 h,
95:5 syn/anti, 76:24 er in favor of the syn diastereomer) and
that, at -50 °C, the 2:1 complex promotes a highly syn-selective
AVM with improved efficiency (∼70% conv, 15 h, 95:5 syn/
anti, 71:29 er). These observations indicate that the turnover in
diastereoselectivity as a function of temperature (see Tables 1
and 3) is due to higher activity of the syn-selective 2:1 complex
(12) at higher temperatures as well as a change in the
stereochemical preference in the reaction promoted by the 1:1
complex (e.g., due to a change in the identity of the stereo-
chemistry-determining step). At -78 °C, the 1:1 complex
preferably catalyzes the formation of the anti diastereomer but
favors more of the syn adduct at elevated temperatures.
2.2.3. Transition State Working Models. The above mecha-
nistic considerations support the scenarios put forth previously
for reactions of aldimines, involving a 1:1 Ag/ligand complex.8a
A model (endo mode of approach I, G ) H, Figure 4), wherein
the substrate binds in a manner to minimize interaction with
the bulky amino acid substituent (R in I) and ketoimine’s
sterically hindered aryl substituent is situated trans to the Si-
based nucleophile, may be proposed. Another noteworthy feature
is the Lewis base activation22 of the siloxyfuran by the amide
terminus. We have previously demonstrated the importance of
the chiral phosphine’s Lewis basic C-terminus to the facility
of this class of transformations; for example, we have shown8a
that replacement of the p-methoxyphenyl amide with a p-
trifluoromethylphenyl amide results in significant diminution in
AVM efficiency.
Figure 3. 31P NMR spectra (162 MHz, d8-THF) of the powder and crystal
forms of the chiral Ag-based complexes of 1 at various temperatures.
of a mixture of 12 and another complex (NH at δ 10.78 ppm at
-78 °C), which likely promotes reactions more readily and with
higher selectivity at -78 °C. Thus, in the crystalline sample,
the more catalytically active Ag-based complex is the minor
constituent.16
The findings summarized below indicate that the active
catalyst is likely the less sterically encumbered 1:1 ligand/Ag
complex (component represented by the more upfield signal in
spectra e-f, Figure 1):17
The o-methoxy group of the N-activating unit is required for
the anti-diastereomer to be strongly favored as well as high
enantioselectivity to be achieved (see entries 5-7, Table 1),
since the bidentate mode of coordination enhances transition
structure organization within the catalyst-substrate complex.
Thus, in spite of unfavorable steric repulsion caused by the
presence of the o-methoxy unit, it elevates catalytic activity
(compare entries 5-7, Table 1).
First, as illustrated in Figure 3 (spectrum b), analysis of the
31P NMR of a sample of the powder form indicates two distinct
sets of doublets, with coupling constants consistent with
complexes bearing a P-Ag-P (2:1 complex; JP-Ag ) 413 and
477 Hz)18 and a P-Ag (1:1 complex; JP-Ag ) 622 and 762
Hz) unit.19 In agreement with the 1H NMR data discussed above
(spectrum e, Figure 1), the 1:1 and 2:1 complexes exist in a
2:1 ratio in the powder form,20 as judged by the 31P NMR
spectrum b (Figure 3). Analysis of the 31P NMR spectrum
derived from 12 (spectrum c, Figure 3) indicates a 2:1 mixture
with the 1:1 complex being the minor component,20 a finding
The relatively lower asymmetric induction observed with
substrates bearing an o-substituted aryl unit (e.g., entry 9, Table
(18) For P-Ag-P coupling constants, see: (a) Ohkouchi, M.; Masui, D.;
Yamaguchi, M.; Yamagishi, T. J. Mol. Catal. A 2001, 170, 1–15.
(19) For P-Ag coupling constants, see: (a) Goel, R. G.; Pilon, P. Inorg.
Chem. 1978, 17, 2876–2879.
1
that is also consistent with H NMR spectrum f in Figure 1.
(20) The integration value obtained by analysis of the 1H and 31P NMR
spectra must be corrected since one complex bears two amide protons
and two P atoms, respectively.
(16) The lower selectivity in reactions with 12, containing a small amount
of the more active 1:1 complex, is consistent with the observations
that, at low catalyst loadings, AgOAc and 1 deliver inferior results.
(17) The minor amount of the 1:1 complex in the solution derived from
crystalline 12 is likely due to a small degree of phosphine dissociation.
Accordingly, addition of excess 1 leads to complete disappearance of
the 1:1 complex (NH at δ 10.78 ppm).
(21) The precise amount of excess AgOAc cannot be measured due to the
salt’s incomplete solubility in the reaction solvent.
(22) For a review regarding Lewis base activation of Lewis acids, see:
Denmark, S. E.; Beutner, G. L. Angew. Chem., Int. Ed. 2008, 47, 1560–
1638.
9
574 J. AM. CHEM. SOC. VOL. 131, NO. 2, 2009

Vinylogous Mannich Reactions of R-Ketoimine Esters
A R T I C L E S
Scheme 2. Ag-Catalyzed AVM Reactions of Methyl-Substituted
R-Ketoimine Esters and Electronically Unactivated N-Aryl Groups
Figure 4. Proposed working models for Ag-catalyzed AVM reactions of
aryl-substituted R-ketoimine esters, including Lewis acid activation of the
substrate and Lewis base activation of the enolsilane nucleophile.
2) could be attributed to the steric interactions caused by such
a structural modification, leading to a preference for AVM to
proceed through the exo catalyst-substrate complex II (G )
H, Figure 4). The suggested scenario accounts for the reversal
of diastereoselectivity in such cases as well (preferential
formation of syn isomers). The diminished enantioselectivity
might be attributed to the geometric constraints associated with
interaction of the Lewis basic amide terminus with the Si atom
of the siloxyfuran. Enantiofacial control may thus be largely
governed by the steric differentiation between the two modes
of attack, as illustrated in II (Figure 4); while the phenylphos-
phine partially hinders approach from one face of the ketoimine
in II, the sterically hindered t-Leu might block siloxyfuran from
the opposite enantiotopic face of the CdN bond. Such a model
implies that increasing the size of the Si-based nucleophile
should have a deleterious effect on enantioselectivity, giving
rise to preferable generation of the syn diastereomer. As shown
by the example in eq 2, Ag-catalyzed AVM of R-ketoimine
ester 4a (G ) Me, Figure 4) with 4-methyl-substituted siloxy-
furan 13 results in near exclusive formation of syn-14 (95:5
syn/anti) in 75.5:24.5 er (51% ee).23
in the identity of the turnover-limiting (stereochemistry-
determining) steps. With ketoimines carrying a p-nitroaryl group
(e.g., 4a-h, anti selective), substrate-catalyst association may
be relatively slow, since the electrophile is a ketoimine of
diminished Lewis basicity (more reluctant to associate with the
chiral Lewis acidic Ag). Thus, once I, which may be kinetically
favored, is generated, it might readily undergo AVM to afford
the anti diastereomer in high enantioselectivity. In contrast, with
substrates bearing an unactivated N-aryl moiety (e.g., 3 in Table
1 and 15 in Scheme 2), association with the Ag-based catalyst
might be relatively facile and reversible, while the addition step
is turnover-limiting. The above mechanistic scheme is consistent
with the proposal that reactions involving sterically bulky imines
or siloxyfurans proceed through complex II, allowing the
nucleophile to have less obstructed access to the catalyst-bound
ketoimine.
Initial studies regarding Ag-catalyzed reactions of alkyl-
substituted ketoimines indicate that caution must be exercised
in extending the mechanistic conclusions described above to
other classes of related processes. As the preliminary data
summarized in Scheme 224 regarding transformations of alkyl-
substituted R-ketoimine esters suggest, AVM reactions of
varying types of electrophiles can be governed by different
mechanistic regimes and energetic nuances. With the smaller
methyl substituent in 15 (vs aryl groups in 4a-i), the presence
of an N-activating group bearing a p-nitro unit is no longer
required; this is a critical difference that, as indicated by the
data in Table 1, leads to the exclusive formation of the syn
It is plausible that the difference in the stereochemical
preferences of the two types of substrates discussed above lies
(23) The stereochemical identity of AVM product syn-14 is based on 1H
NMR analysis (comparison with other related syn and anti diastere-
omers, the structures of which have been determined by X-ray
analysis).
(24) The chiral ligand related to 16 but lacking the two methoxy substituents
at the N-terminus is only slightly less effective (>98% conv under
identical conditions; 91.5:8.5 er, 83% ee, 66% yield for 16).
9
J. AM. CHEM. SOC. VOL. 131, NO. 2, 2009 575

A R T I C L E S
Wieland et al.
isomer (>98:<2 syn/anti for 1725 and 18).26 Studies to develop
efficient catalytic AVM of other alkyl-substituted imines and
delineation of the aforementioned mechanistic differences are
subjects of ongoing studies.
class of substrates is a 1:1 complex that exhibits different
stereochemical preferences at different temperatures. We il-
lustrate that, depending on the reaction conditions, a competing
2:1 complex can prove competitive and, in at least certain cases,
cause diminution of stereoselectivity. The investigations de-
scribed herein should therefore prove crucial to future explora-
tions in catalyst design, as they suggest that structurally modified
chiral ligands that are less prone to generate the less active 2:1
complexes might prove to be more effective catalysts in at least
certain types of Ag-catalyzed asymmetric Mannich reactions.
Studies directed toward more efficient chiral catalysts for
enantioselective Mannich reactions and design of other sterically
and/or electronically modified N-aryl groups for reactions of
aldimines and ketoimines are in progress. Future investigations
will include development of catalytic AVM processes for a
wider range of substrates and additional Ag-catalyzed asym-
metric processes with ketoimines.
3. Conclusions
We have developed a Ag-catalyzed process for AVM of
R-ketoimine esters that involves a commercially available metal
salt (AgOAc) and nucleophile (siloxyfuran 2) and an easily
prepared chiral ligand (amino acid-based phosphines 1 or 9).
The products, generated in high diastereo- and enantioselectivity,
bear an N-substituted all-carbon quaternary stereogenic center
as well as an unsaturated lactone moiety that is amenable to
various functionalization procedures.8a These investigations
highlight one of the important aspects of N-aryl units as general
imine activating groups. The reactivity and, in certain instances,
the stereoselectivity of the corresponding catalytic reactions can
be controlled effectively by electronic modification of such
moieties. The p-nitro unit introduced in ketomine substrates
discussed above is not only crucial to obtaining high efficiency
in addition to diastereo- and enantioselectivity, the resulting
aniline (after reduction of the NO2 unit) is critical to the facility
with which the N-aryl group is removed (see the example in
eq 2).
Acknowledgment. Financial support was provided by the NIH
(GM-57212) and Novartis (fellowship to L.C.W.). We are grateful
to Dr. H. Mandai (Boston College) for valuable discussions. We
are indebted to Mr. S. J. Malcolmson (Boston College), Ms. K.
Mandai (Boston College), Dr. B. Li (Boston College), Dr. P. Mu¨ller
(MIT), and Dr. R. Staples (Harvard University) for their generous
assistance in securing X-ray structures. We thank Mr. A. T. Colucci
(Boston College) for experimental assistance. Mass spectrometry
facilities at Boston College are supported by the NSF (DBI-
0619576).
The studies detailed above demonstrate that the active chiral
Ag complex in catalytic Mannich reactions of this particular
(25) The relative and absolute stereochemistry of 17 has been established
by X-ray crystal structure analysis. See Supporting Information for
details.
(26) Attempts to prepare and study Ag-catalyzed AVM reactions of the
methyl-substituted ketoimine ester that bears a p-nitro-o-methoxyphe-
nylimine led to complex mixtures of products, presumably as a result
of enamine formation.
Supporting Information Available: Experimental procedures
and spectral data for substrates and products. This material is
available free of charge via the Internet at http://pubs.acs.org.
JA8062315
9
576 J. AM. CHEM. SOC. VOL. 131, NO. 2, 2009