Mechanistic studies of highly enantio-and diastereoselective aza-petasis-ferrier rearrangement catalyzed by chiral phosphoric acid

Article abstract of DOI:10.1021/ja5017206

The precise mechanism of the highly anti-and enantioselective aza-Petasis-Ferrier (APF) rearrangement of hemiaminal vinyl ethers catalyzed by a chiral phosphoric acid was investigated by undertaking experimental and theoretical studies. The APF rearrangem

Full text of DOI:10.1021/ja5017206

Article
pubs.acs.org/JACS
Mechanistic Studies of Highly Enantio- and Diastereoselective Aza-
Petasis−Ferrier Rearrangement Catalyzed by Chiral Phosphoric Acid
Masahiro Terada,*,†,‡ Takazumi Komuro,† Yasunori Toda,†,⊥ and Toshinobu Korenaga§
†Department of Chemistry, Graduate School of Science, Tohoku University, Aoba-ku, Sendai 980-8578, Japan
‡Research and Analytical Center for Giant Molecules, Graduate School of Science, Tohoku University, Aoba-ku, Sendai 980-8578,
Japan
§Department of Chemistry and Bioengineering, Graduate School of Engineering, Iwate University, 4-3-5 Ueda, Morioka, Iwate-ken
020-8551, Japan
S
* Supporting Information
ABSTRACT: The precise mechanism of the highly anti- and
enantioselective aza-Petasis−Ferrier (APF) rearrangement of
hemiaminal vinyl ethers catalyzed by a chiral phosphoric acid
was investigated by undertaking experimental and theoretical
studies. The APF rearrangement is characterized by the
following unique mechanistic features: (i) efficient optical
kinetic resolution of the starting racemic hemiaminal vinyl
ether, (ii) enantioconvergent process from racemic hemiaminal
vinyl ethers to optically active β-amino aldehyde products, and
(iii) anomalous temperature effects on the enantioselectivity
(enantioselectivity increases as reaction temperature increases).
The following experiments were conducted to elucidate the
unique mechanistic features as well as to uncover the overall scheme of the present rearrangement: (A) X-ray crystallographic
analysis of the recovered hemiaminal vinyl ether to determine its absolute configuration, (B) rearrangements of enantiomerically
pure hemiaminal vinyl ethers to validate the stereochemical relationship between the hemiaminal vinyl ethers and β-amino
aldehydes, (C) theoretical studies on the transition states of the C−O bond cleavage and C−C bond formation steps to gain an
insight into the optical kinetic resolution of the hemiaminal vinyl ether and the origin of the stereoselectivity, as well as to
elucidate the overall scheme of the present rearrangement, and (D) crossover experiments of two hemiaminal vinyl ethers having
different vinyl ether and aliphatic substituents to comprehend the mechanism of the anomalous temperature effect and the
enantioconvergent process. The results of experiments and theoretical studies fully support the proposed mechanism of the
present anti- and enantioselective APF rearrangement.
high diastereo- and enantioselectivities. At the beginning of the
research, N-p-methoxyphenyl (PMP)-protected imines were
widely employed as the electrophilic component.3,4 However,
strong oxidizing agents were usually required to remove the
PMP group on the nitrogen atom of the Mannich products.
Hence, the development of a Mannich reaction of aldimines
having a readily removable protective group was highly awaited
to enhance the synthetic utility of the organocatalytic Mannich
reactions. In 2007, the research groups of List5 and Cordova6
independently reported the proline-catalyzed syn-selective
direct asymmetric Mannich reaction of aldehydes and imines
protected with a readily removable tert-butoxycarbonyl (Boc)
group. Motivated by those successes, excellent enantioselective
methods that use aldimines possessing readily removable
protective groups, such as N-Boc or N-acyl derivatives,7
emerged, including anti-selective direct asymmetric Mannich
reactions.
INTRODUCTION
■
Optically active β-amino carbonyl compounds are highly
attractive chiral building blocks for biologically active natural
products and synthetic pharmaceutical agents.1 Among the β-
amino carbonyl compounds, β-amino aldehydes represent an
important subclass of compounds as they are useful and
versatile precursors of γ-amino alcohols, amino acids, β-lactams,
and amino sugars. Considerable effort has been devoted to the
development of efficient methods for the synthesis of β-amino
aldehydes, particularly α-branched aldehydes, in a highly
diastereo- and enantioselective manner.
The asymmetric Mannich reaction is the most powerful and
efficient method for synthesizing optically active β-amino
aldehydes.2 In the course of the development of the asymmetric
Mannich reaction, a number of efficient approaches that use
chiral metal catalysts have been established. Meanwhile, in the
past decade, a large number of chiral secondary amine catalysts,
such as proline and its derivatives, have emerged as a powerful
tool to accelerate the Mannich reaction of N-protected
aldimines with aldehydes to afford β-amino aldehydes with
Received: February 22, 2014
© XXXX American Chemical Society
A
dx.doi.org/10.1021/ja5017206 | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
However, one critical drawback inherent in the method-
ologies reported to date is that aromatic or glyoxylate-derived
aldimines are employed as adaptive substrates in most cases.7c
The enantioselective direct Mannich reaction of aliphatic
aldimines has remained largely unexploited because those
aldimines readily isomerize to their enamine forms, in
particular, N-Boc- or N-acyl-protected aldimines having
proton(s) at the α-position. To overcome this intrinsic
problem, we have developed an alternative strategy that
involves a combination of two catalytic reactions to furnish
optically active β-amino aldehydes having an aliphatic
substituent (R) at the β-position, (Figure 1).8 The sequence
involves an initial metal-catalyzed isomerization of a double
bond followed by a chiral Brønsted acid-catalyzed aza-Petasis−
Ferrier (APF) rearrangement9 using readily available allyl
hemiaminal ethers as the substrate.
Figure 2. Enantio- and diastereoselective APF rearrangement
catalyzed by chiral phosphoric acid.
of the transition states involved in this fascinating 1,3-
rearrangement.
RESULTS AND DISCUSSION
■
Survey of Preliminary Results of Enantio- and
Diastereoselective Aza-Petasis−Ferrier Rearrangement
and Addressing Issues To Be Clarified. In our preliminary
communication,8 we demonstrated the sequential transforma-
tion of racemic allyl hemiaminal ethers into optically active β-
amino-β-alkyl aldehydes 2 via intermediary vinyl ethers 1 by a
combination of two catalytic reactions, as shown in Figure 1.
The latter step, APF rearrangement mediated by chiral
phosphoric acid 3, is essential for the enantio- and
diastereoselective synthesis of 2 (Figure 3). The enantiose-
lective rearrangement catalyzed by (R)-3 proceeds through the
stereoablative loss19 of a vinyloxy group from the racemic
mixture of hemiaminal 1 to generate achiral intermediates,
aliphatic aldimine and enol form of the aldehyde, followed by
bond recombination through the Mannich-type reaction to
form the C−C bond and the generation of two new stereogenic
centers at the α- and β-positions under the influence of chiral
phosphoric acid catalyst 3.
Figure 1. Catalytic sequence for providing optically active β-amino
aldehydes having an alkyl substituent at the β-position.
The APF rearrangement proceeds through C−O bond
cleavage of the hemiaminal ether moiety by an acid catalyst.10
Subsequent recombination with C−C bond formation between
an imine intermediate and an enol form of the aldehyde results
in rearranged products, β-amino aldehydes.11−15 In 1993,
Frauenrath and co-workers pioneered the diastereoselective
APF rearrangement of hemiaminal vinyl ethers promoted by a
substoichiometric amount of TMSOTf, an achiral Lewis acid.12
After that initial report, however, no report by other researchers
followed until 2008,13 despite the wide applicability of β-amino
aldehyde products to the synthesis of a diverse array of
nitrogen-containing compounds. Fifteen years after the first
discovery, Tayama and co-workers reported a regioselective
APF rearrangement of hemiaminal dienyl ethers under the
influence of a catalytic amount of pyridinium p-toluenesulfo-
nate (PPTS), an achiral Brønsted acid.14 In 2010, Chmielewski
and co-workers reported a diastereofacial selective APF
rearrangement of the β-lactam framework using chiral 4-
vinyloxy-β-lactam as the substrate and a catalytic amount of
Brønsted acid.15 To the best of our knowledge, there is no
report that describes enantioselective catalysis in the APF
rearrangement even when chiral metal complexes are taken into
consideration, with the exception of our initial report in 2009.8
At the beginning of this article, we briefly survey our previous
studies on the enantio- and diastereoselective APF rearrange-
ment of hemiaminal vinyl ethers 1, where the highly
stereoselective synthesis of β-amino aldehydes 2 having an
aliphatic substituent (R) at the β-position was accomplished
using chiral phosphoric acid 316,17 as the chiral Brønsted acid
catalyst18 (Figure 2). In the established catalytic system, the
racemic mixture of hemiaminal vinyl ethers 1 is efficiently
converted into enantioenriched β-amino aldehydes 2 through
an enantioconvergent process19,20 mediated by 3. Herein, we
report in detail mechanistic investigations of the present highly
intriguing enantioconvergent process as well as a rationale for
the stereochemical outcome on the basis of theoretical studies
Figure 3. Plausible reaction pathway of the enantioconvergent process.
A thorough screening of several parameters of the present
rearrangement, including the geometry of the vinyl subunit, the
reaction solvents, the substituent of catalyst 3, and the reaction
temperature, enabled us to establish a highly anti- and
enantioselective rearrangement. The features of the present
catalytic rearrangement are summarized as follows:
B
dx.doi.org/10.1021/ja5017206 | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(i) The geometrical purity of 1 affects not only anti-
selectivity but also enantioselectivity. The use of vinyl
ether bearing (Z)-geometry is the key structural factor to
achieving a highly diastereo- and enantioselective
formation of anti-β-amino-β-alkyl aldehydes 2. In
contrast, the reaction of the (E)-isomer resulted in the
alternative syn-isomer as the major product albeit with
lower diastereo- and enantioselectivities.
corresponding alcohols 5 for the determination of the
stereochemical outcome, as shown in Table 1. Branched and
Table 1. Substrate Scope of Sequential Transformation of 4
into β-Amino Aldehydes 2 via 1 Followed by Reduction of
a
Formyl Group
(ii) The solvents used markedly affected the enantioselectiv-
ity, whereas high anti-selectivity was maintained. Among
the solvents examined, relatively basic solvents, such as
ethyl acetate and acetone, were the best in terms of
chemical yield and enantioselectivity.
(iii) Chiral phosphoric acids 3a and 3b having an isopropyl
group and a tert-butyl group, respectively, at the 4-
position of the phenyl ring had a beneficial effect on the
enantioselectivity, although 3b resulted in a slightly
higher enantioselectivity of the product than 3a.
(iv) An anomalous temperature effect was observed; a higher
temperature resulted in a higher enantioselectivity
despite the fact that in a typical enantioselective reaction,
the enantioselectivities of the products increase with
decreasing reaction temperature.
(v) Efficient optical kinetic resolution of racemic (Z)-1a by
(R)-3b was observed at the C−O bond cleavage step
(Scheme 1).21 (+)-(Z)-1a was recovered in an
enantiomerically pure form at 0 °C, even with
approximately 50% conversion of racemic ( )-(Z)-1a.
Thus, (−)-(Z)-1a was consumed much more rapidly
than (+)-(Z)-1a by (R)-3b, indicating that chiral
phosphoric acid (R)-3b functioned as an efficient
resolving catalyst of ( )-(Z)-1a, although the absolute
configuration of recovered 1a had not been determined
in our preliminary studies.
c
1 (Z)/
(E)
yield (%)
ee
d
b
entry
1
4 (R)
4a: Bn
conditions
5
(anti /syn) (%)
e
98:2
acetone, 40
5a
70 (95:5)
86 (91:9)
76 (93:7)
55 (95:5)
75 (95:5)
87 (93:7)
94
76
97
97
°C, 7 h
AcOEt, 40
°C, 2 h
f
2
3
4
5
6
4b: Me
97:3
5b
5c
5d
5e
5f
f
4c: n-penthyl
4d: i-butyl
4e: c-hexyl
4f: Ph
96:4
acetone, 40
°C, 7 h
AcOEt, 40
°C, 3 h
AcOEt, 40
°C, 3 h
AcOEt, 40
°C, 3 h
f
96:4
f
95:5
>99
98
f
96:4
a
All reactions were carried out using 0.25 mmol of allyl hemiaminal
ether 4. Reaction conditions for enantioselective APF rearrangement.
Combined yield of anti/syn-5 from 4 (three steps). For major anti-5.
Determined by chiral stationary phase HPLC analysis. Absolute
stereochemistry of anti-5a was determined to be 2R,3R-configuration.
See Supporting Information for details. Isomerization of allyl
hemiaminal ether 4 using NiI2(p-tolbiphep). Isomerization of allyl
b
c
d
e
f
hemiaminal ether 4 using NiI2(dppf).
linear aliphatic substituents were well tolerated and the
sequential transformation enabled to afford corresponding γ-
amino-γ-alkyl alcohols 5 in high anti- and enantioselectivities
(up to 95% anti, >99% ee) (entries 1, 3−5), with the exception
of the less sterically hindered methyl substituent (entry 2).
Remarkably, the optimized sequential protocol that was
originally developed for aliphatic substrates was also applicable
to aromatic hemiaminal ethers, giving rise to γ-amino-γ-aryl
alcohol 5f with high anti- and enantioselectivities (entry 6).
In spite of those intriguing findings, however, some issues
related to the stereoselective APF rearrangement catalyzed by
chiral phosphoric acid remain, including
Scheme 1. Optical Kinetic Resolution of Racemic
Hemiaminal Vinyl Ether ( )-(Z)-1a by Chiral Phosphoric
Acid Catalyst (R)-3b
a
(a) Determination of the absolute configuration of recovered
hemiaminal vinyl ether 1 under the influence of chiral
phosphoric acid catalyst (R)-3 to elucidate the
mechanisms of the optical kinetic resolution of racemic
hemiaminal vinyl ether 1.
(b) Mechanistic investigations of the optical kinetic reso-
lution of racemic hemiaminal vinyl ether 1 by chiral
phosphoric acid (R)-3, where experiments using
enantiomerically pure hemiaminal vinyl ether 1 would
be helpful to validate the stereochemical relationship
between hemiaminal vinyl ethers 1 and β-amino-β-alkyl
aldehydes 2.
(c) Detailed experimental studies to justify the anomalous
temperature effect, that is, the increase in enantioselec-
tivity with increasing reaction temperature.
(d) Theoretical studies on the transition states to rationalize
the stereochemical outcome of entire process consisting
of the C−O bond cleavage and the C−C bond formation
steps.
The % ee of 2a was determined after reduction of formyl group to
alcohol.
As mentioned at the beginning of this section, a sequential
protocol that combines the two catalytic reactions and abolishes
the need to separate the geometrical isomers of 1 was also
demonstrated in our preliminary studies. In order to establish
the sequential protocol, we first developed a (Z)-selective
double bond isomerization of allyl hemiaminal ether 4 using a
NiI2 complex having a bidentate phosphine ligand, dppf, or p-
tolbiphep.8,13,22 The combination of the developed (Z)-
selective isomerization with an enantioselective APF rearrange-
ment allowed us to accomplish the sequential transformation
that does not require the separation of the geometrical isomers
of 1. The present sequential protocol is applicable to a variety
of allyl hemiaminal ethers 4 to afford corresponding β-amino-β-
alkyl aldehydes 2 in good yields. Although 2 could be easily
isolated, the formyl group of 2 was reduced to yield
C
dx.doi.org/10.1021/ja5017206 | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(e) Comprehensive mechanistic considerations to under-
stand the enantioconvergent process in which racemic
hemiaminal vinyl ether 1 was efficiently transformed into
optically active β-amino-β-alkyl aldehydes 2 under the
influence of chiral phosphoric acid catalyst (R)-3.
Thus, recovered optically active hemiaminal vinyl ether
(+)-(Z)-1g (99% ee) was recrystallized from ether under
pentane vapor to give an X-ray grade single crystal. X-ray
crystallographic analysis of recovered 1g indicated that the
absolute stereochemistry of (+)-1g was the (S)-configuration
(Figure 4).23 The confirmed stereochemical outcome using
(Z)-1g should be applicable to that observed in the optical
kinetic resolution of racemic (Z)-1a (Scheme 1). Under the
influence of (R)-3b, (R)-(−)-(Z)-1a reacted much more rapidly
than (S)-(+)-(Z)-1a that was recovered in an enantiomerically
pure form.
In order to address those issues, we focused our efforts on
the elucidation of the overall scheme of the stereochemical
relationship in the present fascinating rearrangement by
conducting detailed and extensive studies on the mechanisms
of anti- and enantioselective APF rearrangement, including (A)
X-ray crystallographic analysis of the recovered hemiaminal
vinyl ether to determine the absolute stereochemistry, (B)
experiments of enantiomerically pure hemiaminal vinyl ethers
to understand the stereochemical outcome, (C) transition state
analysis by computational studies, and (D) crossover experi-
ments of two hemiaminal vinyl ethers having different vinyl
ether and aliphatic (R) substituents.
Experiment A: Determination of the Absolute
Configuration of Recovered Hemiaminal Ether. As
shown in Scheme 1, of particular interest is the efficient optical
kinetic resolution of racemic hemiaminal ether (Z)-1a at the
C−O bond cleavage step. At the outset of mechanistic studies
of the APF rearrangement, we began by determining the
absolute configuration of recovered hemiaminal vinyl ether (Z)-
1a. In order to determine the absolute stereochemistry of the
hemiaminal vinyl ether by X-ray crystallographic analysis, we
employed structurally similar hemiaminal vinyl ether (Z)-1g
that has a bromo substituent at the para-position of the phenyl
ring. To obtain enantiomerically pure (Z)-1g, we performed the
optical kinetic resolution of racemic (Z)-1g using 2 mol % of
phosphoric acid (R)-3a instead of (R)-3b (Scheme 2a) because
Figure 4. ORTEP drawing of (S)-(+)-(Z)-1g with probability
ellipsoids drawn at the 50% level.
Experiment B: Aza-Petasis−Ferrier Rearrangement of
Enantiomerically Pure Hemiaminal Ethers. In order to
gain a precise understanding of the stereochemical relationship
between starting (Z)-1a and rearranged product 2a, we
performed the APF rearrangement using both enantiomers of
(Z)-1a in an enantiomerically pure form. The enantiomers of
(Z)-1a were separated by chiral stationary phase HPLC to
afford each enantiomer in a pure form. Thus, prepared (S)-(Z)-
1a and (R)-(Z)-1a were subjected to the APF rearrangement
individually in the presence of (R)-3a. The rearrangement was
performed at 40 and 0 °C in ethyl acetate using 2 mol % of (R)-
3a and was followed by the reduction of 2a to give rise to
corresponding alcohol (2R,3R)-anti-5a as the major stereo-
isomer in good yields in all cases (Table 2). The results
obtained by using the racemic mixture of (Z)-1a are also
displayed for reference (entries 3 and 6). As expected from the
efficient optical kinetic resolution observed in the present
rearrangement (Schemes 1 and 2), the reaction of (R)-(Z)-1a
proceeded much more rapidly than that of (S)-(Z)-1a at both
40 and 0 °C (entries 2 vs 1 and 5 vs 4). Although (S)-(Z)-1a
underwent the rearrangement sluggishly, the enantioselectiv-
ities were the highest compared with those observed for the
opposite enantiomer (R)-(Z)-1a and racemic ( )-(Z)-1a
(entry 1 vs 2 and 3 and entry 4 vs 5 and 6). More importantly,
in the reaction of (S)-(Z)-1a, the enantioselectivities increased
with decreasing reaction temperature as observed in typical
catalytic enantioselective reactions (entry 1 vs 4). Indeed,
nearly enantiomerically pure 5a was afforded at 0 °C (entry 4).
In sharp contrast, the reaction of (R)-(Z)-1a resulted in a
considerable decrease in enantioselectivity when the reaction
temperature was lowered from 40 to 0 °C (entry 2 vs 5). This
anomalous temperature effect was detected in our preliminary
studies on the APF rearrangement of ( )-(Z)-1a (entry 3 vs
6).8 It can be readily expected that the enantioselectivity
observed in the reaction of ( )-(Z)-1a is the intermediate of
those observed in the reaction of (S)-(Z)-1a and (R)-(Z)-1a
Scheme 2. Optical Kinetic Resolution of Hemiaminal Vinyl
Ethers (Z)-1a and (Z)-1g Using Catalyst (R)-3a
a
The % ee of 2a was determined after reduction of formyl group to
alcohol.
of the availability of catalyst 3a. As shown in Scheme 2b,
efficient optical kinetic resolution of (Z)-1a was also achieved
by using phosphoric acid (R)-3a as the resolving catalyst,
although lowering of the reaction temperature was required to
obtain enantiomerically pure (Z)-1a in good yield. The optical
kinetic resolution of racemic (Z)-1g was conducted at −5 °C
for 6 h. (Z)-1g was converted into 2g in 51% yield and
recovered in 28% isolated yield (48% NMR yield) in a nearly
enantiomerically pure form (99% ee) after preparative HPLC
purification.
D
dx.doi.org/10.1021/ja5017206 | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Table 2. APF Rearrangement of Enantiomerically Pure and
a
Racemic (Z)-1a Catalyzed by (R)-3a
Figure 5. Plausible mechanism of APF rearrangement catalyzed by
chiral phosphoric acid (R)-3.
cleavage rapidly to generate intermediate B. However, the
subsequent C−C bond formation of intermediate B via a
retentive manner is retarded by (R)-3. Accordingly, the chiral
information on (R)-(Z)-1a is unfavorable for the C−C bond
formation under the influence of (R)-3 (C−C mismatched),
and hence, the rearrangement would proceed via an inversion
process. The low enantioselectivities of product anti-5a
obtained from (R)-(Z)-1a imply that the unfavorable retentive
process from intermediate B partially participates in the
rearrangement. The anomalous temperature effect observed
in the reactions of (R)-(Z)-1a and ( )-(Z)-1a may be due to
the loss of unfavorable chiral information at higher temperature
because the enantioselectivities increase in those cases.
a
All reactions were carried out using 0.1 mmol of hemiaminal vinyl
b
ether (Z)-1a. Reaction conditions for enantioselective APF rearrange-
c
d
ment. Combined yield of anti/syn-5a from 1a (two steps). For major
anti-5a. Determined by chiral stationary phase HPLC analysis.
Absolute stereochemistry of anti-5a was determined to be 2R,3R-
configuration. See Supporting Information for details. Enantiomeric
purity of (Z)-1a in >99% ee.
e
Experiment C: Computational Studies of Transition
States. The significant differences in the reaction rates between
enantiomers (Z)-1a and the intriguing stereochemical relation-
ship between starting (Z)-1a and rearranged product 2a
prompted us to further study the mechanism of the catalytic
enantioselective APF rearrangement in detail. In the previous
section, we proposed a plausible mechanism of the present
rearrangement. However, (i) the efficient optical kinetic
resolution, (ii) the predominant formation of the anti-isomer
having the (2R,3R)-configuration from the (Z)-substrate, and
(iii) the overall scheme of the present enantioconvergent
process remain to be elucidated. Therefore, we theoretically
studied the mechanism of the present APF rearrangement in
combination with the above experimental observations.
Catalyst (R)-3b and substrate (Z)-1a were used for the
theoretical studies.
(entries 3 vs 1 and 2, and 6 vs 4 and 5) because ( )-(Z)-1a is a
mixture of equal amounts of (S)- and (R)-(Z)-1a.
The stereochemical relationship between starting (Z)-1a and
corresponding product 5a (derived from 2a) should be noted
in the present APF rearrangement. The stereogenic center of
starting (Z)-1a is lost during the course of the rearrangement.
The newly generated stereogenic center of product 5a has
(2R,3R)-configuration in all cases, despite the fact that the
absolute configuration of enantiomeric (Z)-1a is the opposite.
The reaction of (S)-(Z)-1a proceeded with retention at the
stereogenic center bearing an N-Boc amino group, whereas
(R)-(Z)-1a exhibited inversion at that stereogenic center. From
experimental evidence obtained using enantiomerically pure
substrates (Z)-1a (Table 2), a plausible mechanism of the
present rearrangement is proposed, as illustrated in Figure 5.
Under the influence of (R)-3, (S)-(Z)-1a undergoes C−O
cleavage sluggishly to generate intermediate A. However, the
subsequent C−C bond formation of intermediate A is favored
in the presence of (R)-3, because the highest enantioselectivity
was observed in the reaction of (S)-(Z)-1a. It can be considered
that the combination of (R)-3 and intermediate A generated
from (S)-(Z)-1a is a matched pair of their chiralities for C−C
bond formation (C−C matched). More importantly, during the
course of the rearrangement, C−O bond cleavage and C−C
bond formation proceed on the same side of the enantiotopic
face, that is, the si face, of the imine fragment in intermediate A.
Consequently, the retention process dominates the overall
scheme of the present rearrangement despite the slow C−O
bond cleavage. In contrast, (R)-(Z)-1a undergoes C−O bond
All calculations were performed with the Gaussian 09
program.24 Geometries were fully optimized and characterized
by frequency calculation using the ONIOM method,25
combining the B3LYP (Becke’s three-parameter hybrid
method26a using the Lee−Yang−Parr correlation functional26b
)
) density functional theory (DFT) with the 6-31G(d) basis
set27 in the low-level layer and the Møller−Plesset perturbation
theory second-order correction (MP2)28 with the 6-31+G(d,p)
basis set29 in the high-level layer. The atom distribution in
those two layers is shown in Figure 8, where atoms in the high
level layer are represented by a “ball & bond type” model and
atoms in the low level layer by a “tube” model. Free energies
(298.15 K, 1 atm) were initially computed for the gas phase.
The geometrically optimized structures were further calculated
E
dx.doi.org/10.1021/ja5017206 | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
for the continuum solvation model, where the single-point
energy calculation with the self-consistent reaction field
(SCRF) method based on the polarizable continuum model
(CPCM,30 ϵ = 5.9867 for ethyl acetate) was carried out at the
same level as the one used for geometry optimization. The free
energies in ethyl acetate were calculated from the sum of the
single-point energies in ethyl acetate and the value of thermal
correction to Gibbs free energy in gas phase.
As postulated in Figure 5, it seems that the rearrangement is
composed of two transformation steps: C−O bond cleavage
and C−C bond formation. Therefore, we calculated the
transition states of those two steps individually. The calculation
was started at the C−O bond cleavage step, where the first issue
of efficient optical kinetic resolution is the main subject to be
addressed. On the basis of the acid/base dual function of
phosphoric acid catalyst 3,21f,g,31 it can be considered that the
C−O bond cleavage is mediated by the protonation of the
ether oxygen atom of hemiaminal vinyl ether (Z)-1a by the
Brønsted acidic proton of (R)-3b, whereas the Lewis basic
phosphoryl oxygen interacts with the N−H proton of the
hemiaminal, forming two parallel hydrogen bonds between
(Z)-1a and (R)-3b, as illustrated in Figure 6. Eight initial TS
structures are generated from the two absolute configurations
of (Z)-1a (R and S; TScor and TScos), the two protonation sites
at the ether oxygen atom of (Z)-1a, and the two orientations of
the vinyl ether moiety.32
even with approximately 50% conversion of (Z)-1a (Scheme
1).
Structural analysis of the most energetically favorable
transition states for each substrate, TScor1 and TScos1 (Figure
8), allowed us to identify the major factors contributing to the
efficient optical kinetic resolution. In the most favorable TScor1
(Figure 8a), the imine fragment of substrate (R)-(Z)-1a is
almost parallel to the aryl plane of the catalyst substituent (G:
4-tert-butyl-2,6-diisopropylphenyl) to avoid steric congestion.
In contrast, the imine fragment of substrate (S)-(Z)-1a in
TScos1 is oriented perpendicularly to the aryl plane of the
catalyst substituent (G) (Figure 8b), where both the N-Boc
group and the benzyl moiety of substrate (S)-(Z)-1a are located
close to the aryl substituent (G). This leads to steric repulsion
between the substrate and the catalyst (indicated by red curved
double-ended arrows), which would be responsible for the
destabilization of TScos1.
Figure 6. Two-point binding model for the C−O bond cleavage step
through the formation of parallel hydrogen bonds between (Z)-1a and
(R)-3b.
Among the possible transition structures searched under the
influence of (R)-3b, the energetically favorable transition
structures of TScor for (R)-(Z)-1a and TScos for (S)-(Z)-1a
are shown in Figure 7.33 TScor1 is the most favorable transition
state as it is 2.9 kcal/mol lower in energy than TScos1. This
result is consistent with the experimental result that the optical
kinetic resolution of racemic (Z)-1a gave rise to enantiomeri-
cally pure (S)-(Z)-1a as the recovery of the starting substrate
Figure 7. Schematic representation models of energetically favorable
TScor and TScos. Relative free energies (kcal/mol) obtained by single-
point energy calculations with the SCRF method based on CPCM (ϵ
= 5.9867 for ethyl acetate) are shown. Relative free energies (kcal/
mol) of the optimized structures in the gas phase are shown in
parentheses.
Figure 8. 3D structures of the most energetically favorable transition
states for the C−O bond cleavage step are shown (the high level layer
and the low level layer are represented by “ball & bond type” and
“tube” models, respectively): (a) (R)-(Z)-1a for TScor1. (b) (S)-(Z)-
1a for TScos1. Relative free energies (kcal/mol) in the solution phase
are also shown.
F
dx.doi.org/10.1021/ja5017206 | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 9. 3D structures and schematic representation models of four transition states for the C−C bond formation step. The 3D structures of each
fragment are represented as follows: imine fragment and phosphoric acid subunit, “tube” model; enol fragment, “ball & bond type” model; and
binaphthyl backbone and substituent (G), “wire” model. Relative free energies (kcal/mol) obtained by single-point energy calculations with the
SCRF method based on CPCM (ϵ = 5.9867 for ethyl acetate) are shown. In parentheses are the relative free energies (kcal/mol) of the optimized
structures in the gas phase. (a) TSccsi-anti: Bond formation between si face of imine and si face of enol to give (2R,3R)-anti-2a. (b) TSccsi-syn: Bond
formation between si face of imine and re face of enol to give (2S,3R)-syn-2a. (c) TSccre-anti: Bond formation between re face of imine and re face of
enol to give (2S,3S)-anti-2a. (d) TSccre-syn: Bond formation between re face of imine and si face of enol to give (2R,3S)-syn-2a.
Further calculations were carried out to acquire a mechanistic
insight into the C−C bond formation step, where the second
issue of the predominant formation of the (2R,3R)-anti-isomer
is the main subject to be addressed. On the basis of the acid/
base dual function of phosphoric acid catalyst 3,31 the imine
and the enol, generated in situ via the C−O bond cleavage,
interact with chiral phosphoric acid (R)-3b through two
hydrogen bonds, as illustrated in Figure 5. As shown in Figure
9, a thorough search of the transition structures afforded four
possible stereoisomers of 2a and resulted in the successful
optimization of the four transition states. Among the four
diastereomeric transition states displayed, the most energeti-
cally favorable transition state is TSccsi-anti, which affords
(2R,3R)-anti-2a (Figure 9a). The formation of the (2R,3R)-
isomer is consistent with the experimental result that the
rearrangement under the influence of (R)-3 resulted in the
production of (2R,3R)-anti-2a as the major stereoisomer,
irrespective of the chirality of starting (Z)-1a (see Table 2).
The preferential formation of (2R,3R)-anti-2a over the syn-
isomer is well rationalized by the structural analysis of the
transition states. In particular, when the si face of the imine
reacts with enol, TSccsi-anti (Figure 9a) is 6.0 kcal/mol lower
in energy than TSccsi-syn (Figure 9b), despite the structural
resemblance of those two transition states with the exception of
the enol orientation.34 The relative positions of the methyl
substituent of the enol and the benzyl substituent of the imine
are responsible for the marked energy difference between
TSccsi-anti and TSccsi-syn (indicated by a red curved double-
ended arrow). Indeed, the methyl and benzyl groups assume an
anti-conformation in TSccsi-anti, whereas the gauche inter-
action between the methyl and benzyl groups results in the
destabilization of TSccsi-syn.
Detailed structural analysis of TSccsi-anti and TSccre-anti
allowed us to identify the origin of the high enantioselectivity
(Figure 10). As shown in Figure 10a, in TSccsi-anti, the imine
fragment is positioned vertically and shows no detrimental
repulsive interaction with the catalyst substituent (G), affording
(2R,3R)-anti-2a predominantly. In contrast, the imine fragment
in TSccre-anti is horizontally oriented (Figure 10b), leading to
repulsive interactions between the catalyst substituent (G) and
both the N-Boc and benzyl groups of the imine fragment
(indicated by red curved double-ended arrows), and thereby
destabilizing TSccre-anti and preventing the formation of
(2S,3S)-anti-2a.
Using the energy profiles of the C−O bond cleavage and C−
C bond formation steps in the present APF rearrangement
summarized in Figure 11, the third issue of the clarification of
the enantioconvergent process is addressed. It is worthy of note
G
dx.doi.org/10.1021/ja5017206 | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 10. 3D structures of transition states for C−C bond formation step giving anti-2a. (a) TSccsi-anti: si face/si face giving (2R,3R)-anti-2a. (b)
TSccre-anti: re face/re face giving (2S,3S)-anti-2a. The 3D structures are represented in accordance with Figure 9. Relative free energies (kcal/mol)
in the solution phase are also shown.
Figure 11. Energy profiles of C−O bond cleavage and C−C bond formation steps. The relative free energy (kcal/mol) of the sum of (Z)-1a and
(R)-3b is set to zero. The results of single-point energy calculations with the SCRF method based on CPCM (ϵ = 5.9867 for ethyl acetate) are
shown. Activation energies are shown in italic.
that the proposed reaction mechanism depicted in Figure 5 can
be rationalized by the present energy profile without acute
contradiction. (S)-(Z)-1a undergoes C−O bond cleavage
through the higher energy transition state TScos1 under the
influence of (R)-3b. In fact, the C−O bond cleavage of (S)-(Z)-
1a proceeds sluggishly compared with that of (R)-(Z)-1a, even
if efficient optical kinetic resolution is achieved. However, in the
subsequent C−C bond formation step, the si face of the imine
fragment reacts with enol via the energetically lowest transition
state TSccsi-anti under the influence of (R)-3b, giving rise to
(2R,3R)-anti-2a exclusively. The si face of the imine fragment
generated from (S)-(Z)-1a is the same enantiotopic face where
the C−O bond cleavage takes place. Consequently, (S)-(Z)-1a
undergoes the rearrangement in a retentive manner at the
stereogenic center bearing the N-Boc amino group, where the
initial chiral information on (S)-(Z)-1a is favorable for the
subsequent C−C bond formation (C−C matched), albeit the
sluggish rearrangement. In contrast, the C−O bond cleavage of
(R)-(Z)-1a proceeds more rapidly than that of (S)-(Z)-1a
because of the energetically lower transition state of TScor1 (or
TScor2) for (R)-(Z)-1a under the influence of (R)-3b.
However, the subsequent C−C bond formation at the re face
of imine (TSccre-syn or TSccre-anti), which originates from the
same enantiotopic face as that of the C−O bond cleavage of the
H
dx.doi.org/10.1021/ja5017206 | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(R)-substrate, is energetically unfavorable (C−C mismatched).
Hence, the C−C bond formation proceeds through the
energetically lowest transition state TSccsi-anti, affording
(2R,3R)-anti-2a as the major stereoisomer, where the inversion
at the stereogenic center bearing the N-Boc amino group is
enhanced by (R)-3b at the C−C bond formation step. As
mentioned above, the intriguing enantioconvergent process
observed in this rearrangement can be rationalized on the basis
of the present energy profiles. In addition, it seems that the
present two-step sequential transformation is rate-determined
by C−O bond cleavage of both enantiomers of (Z)-1a.
However, the mode of reaction is dependent on the chirality of
substrate (Z)-1a. In the rearrangement of (S)-(Z)-1a, it is
reasonable to suggest that C−O bond cleavage is effectively
irreversible because C−C bond formation is lower in energy
than the backward pathway of C−O bond formation. On the
other hand, in the rearrangement of (R)-(Z)-1a, the
equilibrium of C−O bond cleavage and formation between
(R)-(Z)-1a and intermediate B would likely proceed during the
whole reaction process (Figure 5).
Experiment D: Crossover Experiments of Two Hemi-
aminal Vinyl Ethers. The theoretical studies strongly support
our mechanistic proposal for the present intriguing rearrange-
ment. However, whether the inversion observed in the
rearrangement of (R)-(Z)-1a proceeds in an intramolecular
manner is still unclear. To enhance our understanding of the
inversion process and to determine the overall scheme of the
present enantioconvergent process, we conducted crossover
experiments of (Z)-1a and (Z)-1h. Before starting the
crossover experiments, we collected reference data of the
APF rearrangement of ( )-(Z)-1a, ( )-(Z)-1h, ( )-(Z)-1ah,
and ( )-(Z)-1ha. The rearrangement of racemic substrates
(Z)-1 was performed using 2 mol % of (R)-3a at 40 °C for 1.5
h and at 0 °C for 30 h in ethyl acetate, followed by the
reduction of corresponding products 2 with sodium borohy-
dride to determine the stereoselectivities of the products. The
stereochemical outcomes of those reference reactions are
summarized in Table 3.
Table 3. APF Rearrangement of Racemic (Z)-1 Catalyzed by
(R)-3a as Reference for Crossover Experiments
a
yield
c
anti-5/
syn-5
ee
(%)
b
d
entry
1
conditions
5
(%)
1
2
3
4
5
6
7
8
( )-(Z)-1a
40 °C, 1.5 h 5a
0 °C, 30 h
40 °C, 1.5 h 5h
0 °C, 30 h
92
75
94
92
88
74
90
95
97:3
97:3
95:5
93:7
97:3
96:4
97:3
96:4
92
87
98
96
92
86
98
97
( )-(Z)-1h
( )-(Z)-1ah 40 °C, 1.5 h 5ah
0 °C, 30 h
( )-(Z)-1ha 40 °C, 1.5 h 5ha
0 °C, 30 h
a
All reactions were carried out using 0.1 mmol of hemiaminal vinyl
b
ether (Z)-1. Reaction conditions for enantioselective APF rearrange-
ment. Combined yield of anti/syn-5 from 1 (two steps). For major
anti-5. Determined by chiral stationary phase HPLC analysis.
c
d
As shown in Table 3, all substrates (Z)-1 exhibited a similar
reactivity to afford corresponding products 5 in good to
excellent yields with high anti-selectivities. The anomalous
temperature effect was also observed in all of the substrates,
that is, the enantioselectivities increased as the reaction
temperature increased. In addition, hemiaminal vinyl ethers
having a (4-methoxyphenyl)propyl group at the stereogenic
center, ( )-(Z)-1h and ( )-(Z)-1ha (entries 3, 4, 7, and 8),
exhibited higher enantioselectivities than the benzyl-substituted
ones, ( )-(Z)-1a and ( )-(Z)-1ah (entries 1, 2, 5, and 6). In
contrast, the substituent at the terminal of the vinyl ether
moiety had no influence on the enantioselectivity (entries 1 vs
5, 2 vs 6, 3 vs 7, and 4 vs 8) and nearly equal enantioselectivities
were obtained by changing the substituent from methyl to ethyl
[(Z)-1a → (Z)-1ah, (Z)-1ha → (Z)-1h] under each set of
reaction conditions.
In principle, in the crossover experiment between (Z)-1a and
(Z)-1h, 5a and 5h should be the exclusive products via the
intramolecular pathway. In contrast, the intermolecular
exchange process would afford not only cross-products 5ah
and 5ha but also normal-products 5a and 5h, as shown in
Figure 12. The estimated participation of the normal
intermolecular process during the whole transformation can
be statistically calculated to be twice that of the cross-products.
For instance, the cross- and normal-products would be
Figure 12. Projected product distributions from the crossover
experiments.
obtained in a ratio of 30 (5ah:5ha = 15:15) to 70 (5a:5h =
35:35), respectively, in which the percentage of the
intermolecular exchange process is estimated to be 60%, that
is, twice the 30% proportion of the cross-products. In addition,
to simplify our analysis of the crossover experiments, we
focused on the isomeric ratios of major anti-5 because less than
5% syn-5 is formed in most cases, as based on the reference
data.
Bearing the above considerations in mind, we began by
performing the crossover experiments of racemic substrates
( )-(Z)-1a and ( )-(Z)-1h under the same conditions as
those for the reference reaction using an equimolar mixture of
substrates ( )-(Z)-1. As shown in Scheme 3, considerable
I
dx.doi.org/10.1021/ja5017206 | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Scheme 3. Crossover Experiments of Racemic (Z)-1a and
(Z)-1h
Scheme 4. Crossover Experiments of Enantiomerically Pure
(S)-(Z)-1a and (S)-(Z)-1h
(Scheme 4 vs Scheme 3) because of the C−C matched
combination. It should be pointed out that the intramolecular
pathway tends to proceed in a retentive manner at the
stereogenic center. In contrast, the cross-products, in particular
5ah, displayed similar enantioselectivity to that observed in the
reaction of racemic (Z)-1 (Scheme 4 vs Scheme 3). More
importantly, the enantioselectivity of cross-product 5ah was
lower than that of normal-product 5a (Scheme 4). The
observed difference in enantioselectivity between 5ah and 5a
implies that chiral information on substrate (Z)-1 is lost during
the course of the intermolecular exchange process.
amounts of cross-products 5ah and 5ha were obtained (at 40
°C, 42%; at 0 °C, 30%), although normal-products 5a and 5h
were formed mainly.35 The results strongly suggest that the
rearrangement proceeds predominantly through an intermolec-
ular exchange process because normal-products 5a and 5h are
also formed in the intermolecular process, as shown in Figure
12. Here, the percentage of the intermolecular exchange
process can be estimated to be 84% (42 × 2) and 60% (30 × 2)
at 40 and 0 °C, respectively, on the basis of the above
considerations. By taking into account of estimated proportion
of intra/intermolecular processes (16:84) at high reaction
temperature, it is interesting to note that most of the
rearrangements took place via an intermolecular exchange
process. As expected, the enantioselectivities observed in both
normal- and cross-products 5 are nearly identical with those
obtained from individual reference reactions within exper-
imental error (Scheme 3 vs Table 3).
Further crossover experiments were conducted using
enantiomerically pure substrates to gain an insight into the
intermolecular exchange process as observed in the reaction of
the racemic substrates. The C−C matched combination of (S)-
(Z)-1 and (R)-3a was investigated first. Crossover experiments
of (S)-(Z)-1a and (S)-(Z)-1h were performed under the same
reaction conditions as those for the crossover experiments of
the racemic substrates. As shown in Scheme 4, the ratios of
cross-products 5ah and 5ha were markedly reduced in
comparison with those observed in the reaction of the racemic
substrates (Scheme 4 vs Scheme 3).36 At the low temperature
(0 °C), the participation of the intermolecular exchange
process was significantly suppressed and the intramolecular
pathway predominated in the C−C matched combination (the
estimated percentages of the intramolecular pathway are 70% at
0 °C and 32% at 40 °C). As expected, higher enantioselectiv-
ities were observed for normal-products 5a and 5h than for
those obtained in the crossover experiments of racemic (Z)-1
As mentioned above, it seems that the intermolecular
exchange process is primarily responsible for the loss of chiral
information on substrate (Z)-1. To validate this intriguing idea,
we continued the crossover experiments using the C−C
mismatched combination of (R)-(Z)-1 and (R)-3a. From the
results of crossover experiments of racemic substrate ( )-(Z)-1
and C−C matched substrate (S)-(Z)-1, we surmised that the
C−C mismatched combination undergoes the rearrangement
through the intermolecular exchange process exclusively. With
this in mind, crossover experiments of (R)-(Z)-1a and (R)-(Z)-
1h were conducted under the same reaction conditions as those
mentioned above. As shown in Scheme 5, indeed, the crossover
reactions of (R)-(Z)-1a and 1h proceeded smoothly to afford
an approximately 1:1 mixture of normal- and cross-products.
The estimated contribution of the intermolecular exchange
process to the whole transformation was calculated to be 100%
and 94% at 40 and 0 °C, respectively. It should be emphasized
that at the high temperature of 40 °C, the rearrangement
completely proceeded through the intermolecular exchange
process. As pointed out previously, (Table 2, entries 2 and 5),
enantiomerically pure substrate (R)-(Z)-1a predominantly
underwent the rearrangement through the inversion at the
stereogenic center and the inversion is the more dominant
process at the higher temperature (Table 2, entry 2 vs entry 5).
Together, the results strongly suggest that the inversion
originates from the intermolecular exchange process, although
catalyst (R)-3a promotes the inversion at the stereogenic
center. In other words, the involvement of the intermolecular
exchange process is the key to achieving the present
enantioconvergent process.
J
dx.doi.org/10.1021/ja5017206 | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Scheme 5. Crossover Experiments of Enantiomerically Pure
(R)-(Z)-1a and (R)-(Z)-1h
Scheme 6. Crossover Experiments of Racemic (Z)-1a and
(Z)-1h in Toluene
different absolute configurations from each other. The
rearrangement rates of those substrates differed markedly and
hence, it can be anticipated that the normal-products
overwhelm the cross-ones. However, if the hemiaminal vinyl
ether as it stands undergoes intermolecular exchange
irrespective of the rearrangement, a considerable amount of
cross-products would be formed by taking into consideration
the previous crossover experiments. Crossover experiments of
(S)-(Z)-1a and (R)-(Z)-1h were performed under the same
reaction conditions as those mentioned above. As shown in
Scheme 7, the formation of cross-products 5ah and 5ha was
In contrast, at the low temperature of 0 °C, normal-product
5a (0 °C: 70% ee) was obtained with low enantioselectivity
compared with that obtained at the high temperature (40 °C:
90% ee). The low enantioselectivity of normal-product 5a at
the low temperature is rationalized by considering the fact that
the intramolecular pathway proceeds in a retentive manner at
the stereogenic center. The estimated percentage of the
intramolecular pathway is 6% at 0 °C, and hence, normal-
product 5a was partially formed through the intramolecular
pathway at that temperature. The participation of the
intramolecular pathway, that is, the retentive process, leads to
a considerable reduction of the enantioselectivity of normal-
product 5a. Furthermore, the origin of the anomalous
temperature effect is also well rationalized on the basis of the
participation of the intramolecular process, because the
intramolecular pathway, that is, the retentive process, becomes
dominant with decreasing reaction temperature.
Scheme 7. Crossover Experiments of Enantiomerically Pure
(S)-(Z)-1a and (R)-(Z)-1h
As mentioned in the survey of our preliminary results, the
present APF rearrangement is sensitive to the solvent
employed. The solvents have a relatively basic nature; thus,
ethyl acetate and acetone are more suitable for achieving high
enantioselectivity than the less polar solvents, such as toluene
and dichloromethane, despite the fact that these less polar
solvents are commonly used in chiral phosphoric (Brønsted)
acid-catalyzed reactions. On the basis of a series of the
crossover experiments, it can be presumed that this intriguing
solvent effect is caused from changes in rate of the
intermolecular exchange process. Hence, we conducted the
crossover experiments of racemic substrates ( )-(Z)-1a and
( )-(Z)-1h under the same conditions as those mentioned
above with the exception of using the less polar solvent,
toluene. In fact, as shown in Scheme 6 (vs Scheme 3), the rate
of the intermolecular exchange process was markedly sup-
pressed in toluene (in ethyl acetate → in toluene: 84% → 70%
at 40 °C and 60% → 36% at 0 °C). In addition, the
enantioselectivity of the product, in particular 5a and 5ah, was
reduced as observed in our preliminary studies.8,37 It is obvious
that the significant decrease in the enantioselectivities observed
in toluene is primarily originated from the suppression of the
intermolecular exchange process in the less polar solvent.
Finally, we attempted to perform crossover experiments of
enantiomerically pure (S)-(Z)-1a and (R)-(Z)-1h having
significantly suppressed in this particular combination and
hence, most of substrates (S)-(Z)-1a and (R)-(Z)-1h under-
went the rearrangement independently. The results imply that
the intermolecular exchange of the hemiaminal vinyl ethers
without the formation of intermediates A and B may not be a
plausible pathway for the exchange process.
K
dx.doi.org/10.1021/ja5017206 | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
CONCLUSION
ASSOCIATED CONTENT
* Supporting Information
■
■
S
The precise mechanism of the highly anti- and enantioselective
aza-Petasis−Ferrier rearrangement of hemiaminal vinyl ethers
catalyzed by a chiral phosphoric acid was investigated. In order
to elucidate the mechanism of the present intriguing trans-
formation, we conducted the following experiments: (A) X-ray
crystallographic analysis of the recovered hemiaminal vinyl
ether, (B) rearrangements of enantiomerically pure hemiaminal
vinyl ethers, (C) theoretical studies on the transition states of
the C−O bond cleavage and C−C bond formation steps, and
(D) crossover experiments of two hemiaminal vinyl ethers
having different vinyl ether and aliphatic substituents. The
experimental results and the theoretical studies fully support
the mechanistic proposal for the optical kinetic resolution, the
enantioconvergent process, and the anomalous temperature
effect observed in the present anti- and enantioselective APF
rearrangement. The determination of the absolute config-
uration of the recovered hemiaminal vinyl ether by X-ray
crystallographic analysis enabled to validate the stereochemical
relationship between hemiaminal vinyl ethers and β-amino
aldehydes as well as to evaluate theoretical studies on the C−O
bond cleavage step of each enantiomer. The overall scheme of
the present stereoselective APF rearrangement was proposed
on the basis of experimental evidence from the rearrangements
that used enantiomerically pure hemiaminal vinyl ethers and
was well consistent with the energy profiles formulated by
theoretical studies of the C−O bond cleavage and C−C bond
formation steps. Indeed, the origin of the optical kinetic
resolution of the racemic hemiaminal vinyl ether was clarified
by theoretical studies on the C−O bond cleavage step. The
formulated energy profiles coupled with observations from
crossover experiments are beneficial to rationalize the
enantioconvergent process in which racemic hemiaminal vinyl
ethers are efficiently transformed into optically active β-amino
aldehydes in the presence of a chiral phosphoric acid catalyst.
Meanwhile, it became clear that the participation of
intermolecular exchange is indispensable to achieving the
present enantioconvergent process because the intermolecular
exchange is responsible for the inversion at the initial
stereogenic center of the hemiaminal vinyl ether. Furthermore,
the experimental results of the rearrangements using
enantiomerically pure hemiaminal vinyl ethers and the
crossover experiments are also helpful to clarify the anomalous
temperature effect, where the participation of the intra-
molecular pathway in the rearrangement is essential. The
intramolecular pathway, that is, the retentive process, becomes
dominant with decreasing reaction temperature and its
involvement leads to a considerable decrease in enantioselec-
tivity of the C−C mismatched transformation at the reduced
temperature. We have elucidated the overall scheme of the
present anti- and enantioselective APF rearrangement through
detailed experimental and theoretical studies. The present
method provides an efficient access to the highly anti- and
enantioselective β-amino aldehydes having an aliphatic
substituent at the β-position. Further studies on the application
of the present activation mode of hemiaminal derivatives using
chiral phosphoric acid catalysts are in progress with the aim of
developing even more efficient enantioselective transforma-
tions.
Representative experimental procedure, spectroscopic data for
hemiaminal vinyl ethers (1), β-amino aldehydes (2), and amino
alcohols (5), and determination of the absolute stereochemistry
of 1g and 5a; complete list of authors in the Gaussian 09
reference; Cartesian coordinates and energies of stationary
points and all transition structures. This material is available
free of charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
Corresponding Author
mterada@m.tohoku.ac.jp
■
Present Address
⊥Department of Chemistry (Professor Johnston’s group),
Vanderbilt University, Nashville, TN 37235, United States.
Author Contributions
All authors have given approval to the final version of the
manuscript.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
This work was partially supported by a Grant-in-Aid for
Scientific Research on Innovative Areas “Advanced Molecular
Transformations by Organocatalysts” from MEXT, Japan and
NOVARTIS Foundation (Japan) for the Promotion of Science.
We thank JSPS for a research fellowship for Young Scientists
(Y.T.). We also sincerely thank Prof. T. Iwamoto and Prof. S.
Ishida (Graduate School of Science, Tohoku University) for
conducting X-ray analysis of (Z)-1g and Mitsubishi Chemical
Co., Ltd. for supporting the HRMS analysis of new compounds
and the HPLC analysis in the crossover experiments.
REFERENCES
■
(1) For reviews, see: (a) Kobayashi, S.; Ishitani, H. Chem. Rev. 1999,
99, 1069−1094. (b) Liu, M.; Sibi, M. P. Tetrahedron 2002, 58, 7991−
8035. (c) Cordova, A. Acc. Chem. Res. 2004, 37, 102−112. (d) Weiner,
́
B.; Szymanski, W.; Janssen, D. B.; Minnaard, A. J.; Feringa, B. L. Chem.
Soc. Rev. 2010, 39, 1656−1691. (e) Burns, N. Z.; Jacobsen, E. N. In
Science of Synthesis, Stereoselective Synthesis; De Vries, J. G.; Molander,
G. A.; Evans, P. A., Eds.; Georg Thieme Verlag KG: New York, 2011;
Vol. 2, pp 785−834.
(2) For recent reviews, see: (a) Ting, A.; Schaus, S. E. Eur. J. Org.
Chem. 2007, 5797−5815. (b) Verkade, J. M. M.; van Hemert, L. J. C.;
Quaedflieg, P. J. L. M.; Rutjes, F. P. J. T. Chem. Soc. Rev. 2008, 37, 29−
41. (c) Bhadury, P. S.; Song, B.-A. Curr. Org. Chem. 2010, 14, 1989−
2006.
(3) For syn-selective asymmetric Mannich reactions catalyzed by
chiral amines, see: (a) Cordova, A.; Watanabe, S.; Tanaka, F.; Notz,
́
W.; Barbas, C. F., III J. Am. Chem. Soc. 2002, 124, 1866−1867.
(b) Hayashi, Y.; Tsuboi, W.; Ashimine, I.; Urushima, T.; Shoji, M.;
Sakai, K. Angew. Chem., Int. Ed. 2003, 42, 3677−3680. (c) Notz, W.;
Tanaka, F.; Watanabe, S.; Chowdari, N. S.; Turner, J. M.;
Thayumanavan, R.; Barbas, C. F., III J. Org. Chem. 2003, 68, 9624−
9634. (d) Co
(e) Chowdari, N. S.; Suri, J. T.; Barbas, C. F., III Org. Lett. 2004, 6,
2507−2510. (f) Ibrahem, I.; Cordova, A. Tetrahedron Lett. 2005, 46,
2839−2843. (g) Fustero, S.; Jimenez, D.; Sanz-Cervera, J. F.; Sanchez-
Rosello, M.; Esteban, E.; Simon-Fuentes, A. Org. Lett. 2005, 7, 3433−
rdova, A. Chem.Eur. J. 2004, 10, 1987−1997.
́
́
́
́
́
3436. (h) Hayashi, Y.; Urushima, T.; Shin, M.; Shoji, M. Tetrahedron
2005, 61, 11393−11404. (i) Cheong, P. H.-Y.; Zhang, H.;
Thayumanavan, R.; Tanaka, F.; Houk, K. N.; Barbas, C. F., III Org.
Lett. 2006, 8, 811−814.
L
dx.doi.org/10.1021/ja5017206 | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
(4) For anti-selective asymmetric Mannich reactions catalyzed by
Science of Synthesis, Asymmetric Organocatalysis 2, Brønsted Base and
Acid Catalysts, and Additional Topics; Maruoka, K., Ed.; Georg Thieme
Verlag KG: New York, 2012; Vol. 2, pp219−296.
(17) For seminal studies, see: (a) Akiyama, T.; Itoh, J.; Yokota, K.;
Fuchibe, K. Angew. Chem., Int. Ed. 2004, 43, 1566−1568. (b) Uraguchi,
D.; Terada, M. J. Am. Chem. Soc. 2004, 126, 5356−5357.
chiral amines, see: (a) Cor
́
dova, A.; Barbas, C. F., III Tetrahedron Lett.
2002, 43, 7749−7752. (b) Kano, T.; Yamaguchi, Y.; Tokuda, O.;
Maruoka, K. J. Am. Chem. Soc. 2005, 127, 16408−16409. (c) Franzen
́
,
J.; Marigo, M.; Fielenbach, D.; Wabnitz, T. C.; Kjærsgaard, A.;
Jørgensen, K. A. J. Am. Chem. Soc. 2005, 127, 18296−18304.
(d) Ibrahem, I.; Cordova, A. Chem. Commun. 2006, 1760−1762.
́
(18) For reviews on chiral Brønsted acid catalysis, see: (a) Akiyama,
T.; Itoh, J.; Fuchibe, K. Adv. Synth. Catal. 2006, 348, 999−1010.
(b) Taylor, M. S.; Jacobsen, E. N. Angew. Chem., Int. Ed. 2006, 45,
1520−1543. (c) Doyle, A. G.; Jacobsen, E. N. Chem. Rev. 2007, 107,
5713−5743. (d) Akiyama, T. Chem. Rev. 2007, 107, 5744−5758.
(e) Yamamoto, H.; Payette, N. In Hydrogen Bonding in Organic
Synthesis; Pihko, P. M., Ed.; Wiley-VCH: Weinheim, 2009; pp 73- 140.
(f) Kampen, D.; Reisinger, C. M.; List, B. Top. Curr. Chem. 2010, 291,
395−456.
(e) Zhang, H.; Mitsumori, S.; Utsumi, N.; Imai, M.; Garcia-Delgado,
N.; Mifsud, M.; Albertshofer, K.; Cheong, P. H.-Y.; Houk, K. N.;
Tanaka, F.; Barbas, C. F., III J. Am. Chem. Soc. 2008, 130, 875−886.
(f) Kano, T.; Hato, Y.; Yamamoto, A.; Maruoka, K. Tetrahedron 2008,
64, 1197−1203. (g) Pouliquen, M.; Blanchet, J.; Lasne, M.-C.;
Rouden, J. Org. Lett. 2008, 10, 1029−1032. (h) Kano, T.; Yamaguchi,
Y.; Maruoka, K. Chem.Eur. J. 2009, 15, 6678−6687. (i) Gomez-
Bengoa, E.; Maestro, M.; Mielgo, A.; Otazo, I.; Palomo, C.; Velilla, I.
Chem.Eur. J. 2010, 16, 5333−5342.
(19) Mohr, J. T.; Ebner, D. C.; Stoltz, B. M. Org. Biomol. Chem. 2007,
5, 3571−3576.
(5) Yang, J. W.; Stadler, M.; List, B. Angew. Chem., Int. Ed. 2007, 46,
609−611. List and co-workers demonstrated that proline can also
catalyze the direct asymmetric Mannich reaction of acetaldehyde and
N-Boc-protected imines to provide the simplest Mannich products as
versatile chiral building blocks, albeit in low to moderate yields. See
also: (a) Yang, J. W.; Chandler, C.; Stadler, M.; Kampen, D.; List, B.
Nature 2008, 452, 453−455. N-benzoyl aldimines, see: (b) Hayashi,
Y.; Okano, T.; Itoh, T.; Urushima, T.; Ishikawa, H.; Uchimaru, T.
Angew. Chem., Int. Ed. 2008, 47, 9053−9058.
(20) For a seminal study of dynamic kinetic asymmetric trans-
formation (DYKAT), see: (a) Trost, B. M.; Toste, F. D. J. Am. Chem.
Soc. 1999, 121, 3543−3544. (b) Trost, B. M.; Bunt, R. C.; Lemoine, R.
C.; Calkins, T. L. J. Am. Chem. Soc. 2000, 122, 5968−5976. See also:
(c) Trost, B. M.; Fandrick, D. R. Aldrichimica Acta 2007, 40, 59−72.
Selected examples of DYKAT, see: (d) Mohr, J. T.; Behenna, D. C.;
Harned, A. M.; Stoltz, b. M. Angew. Chem., Int. Ed. 2005, 44, 6924−
6927. (e) Roggen, M.; Carreira, E. M. Angew. Chem., Int. Ed. 2012, 51,
8652−8655. For direct enantio-convergent transformation (DET),
see: (f) Ito, H.; Kunii, S.; Sawamura, M. Nat. Chem. 2010, 2, 972−976.
(21) For optical kinetic resolution of racemic substrates by chiral
(6) Vesely, J.; Rios, R.; Ibrahem, I.; Cor
2007, 48, 421−425.
́
dova, A. Tetrahedron Lett.
(7) (a) Zhang, H.; Mitsumori, S.; Utsumi, N.; Imai, M.; Garcia-
Delgado, N.; Mifsud, M.; Albertshofer, K.; Cheong, P. H.-Y.; Houk, K.
N.; Tanaka, F.; Barbas, C. F., III J. Am. Chem. Soc. 2008, 130, 875−
886. (b) Gianelli, C.; Sambri, L.; Carlone, A.; Bartoli, G.; Melchiorre,
P. Angew. Chem., Int. Ed. 2008, 47, 8700−8702. (c) Dziedzic, P.;
̌
phosphoric acid catalysts, see: (a) Coric, I.; Muller, S.; List, B. J. Am.
́
̈
Chem. Soc. 2010, 132, 17370−17373. (b) Guojian, L.; Birman, V. B.
Org. Lett. 2011, 13, 356−358. (c) Wang, C.; Luo, H.-W.; Gong, L.-Z.
Synlett 2011, 992−994. (d) Mandai, H.; Murota, K.; Mitsudo, K.;
Suga, S. Org. Lett. 2012, 13, 3486−3489. (e) Qabaja, G.; Wilent, J. E.;
Benavides, A. R.; Bullard, G. E.; Petersen, K. S. Org. Lett. 2013, 15,
1266−1269. (f) Saito, K.; Shibata, Y.; Yamanaka, M.; Akiyama, T. J.
Am. Chem. Soc. 2013, 135, 11740−11743. (g) Harada, S.; Kuwano, S.;
Yamaoka, Y.; Yamada, K.; Takasu, K. Angew. Chem., Int. Ed. 2013, 52,
10227−10230.
Schyman, P.; Kullberg, M.; Cor
́
dova, A. Chem.Eur. J. 2009, 15,
4044−4048. (d) Kano, T.; Yamaguchi, Y.; Maruoka, K. Angew. Chem.,
Int. Ed. 2009, 48, 1838−1840. (e) Galzerano, P.; Agostino, D.;
Bencivenni, G.; Sambri, L.; Bartoli, G.; Melchiorre, P. Chem.Eur. J.
2010, 16, 6069−6076.
(8) Terada, M.; Toda, Y. J. Am. Chem. Soc. 2009, 131, 6354−6355.
(9) For reviews on metal complex/organocatalyst combined systems,
see: (a) Park, Y. J.; Park, J.-W.; Jun, C.-H. Acc. Chem. Res. 2008, 41,
222−234. (b) Shao, Z.; Zhang, H. Chem. Soc. Rev. 2009, 38, 2745−
2755. (c) Zhou, J. Chem.Asian J. 2010, 5, 422−434. (d) Hashmi, A.
S. K.; Hubbert, C. Angew. Chem., Int. Ed. 2010, 49, 1010−1012. (e) de
Armas, P.; Tejedor, D.; García-Tellado, F. Angew. Chem., Int. Ed. 2010,
49, 1013−1016. (f) Zhong, C.; Shi, X. Eur. J. Org. Chem. 2010, 2999−
3025. (g) Rueping, M.; Koenigs, R. M.; Atodiresei, I. Chem.Eur. J.
2010, 16, 9350−9365. (h) Lv, F.; Liu, S.; Hu, W. Asian J. Org. Chem.
2013, 2, 824−836.
(10) Terada, M.; Machioka, K.; Sorimachi, K. Angew. Chem., Int. Ed.
2009, 48, 2553−2556.
(11) For a review on organocatalytic rearrangements, see: Moyano,
A.; El-Hamdouni, N.; Atlamsani, A. Chem.Eur. J. 2010, 16, 5260−
5273.
(22) Yamamoto, Y.; Kurihara, K.; Yamada, A.; Takahashi, M.;
Takahashi, Y.; Miyaura, N. Tetrahedron 2003, 59, 537−542.
(23) CCDC 985635 [(S)-(+)-(Z)-1g] contains the supplementary
crystallographic data for this paper. This data can be obtained free of
charge from The Cambridge Crystallographic Data Centre via www.
ccdc.cam.ac.uk/data_request/cif.
(24) Frisch, M. J. et al. Gaussian 09, Revision C.01; Gaussian, Inc.:
Wallingford, CT, 2010.
́
(25) Dapprich, S.; Komaromi, I.; Byun, K. S.; Morokuma, K.; Frisch,
M. J. J. Mol. Struct.: THEOCHEM 1999, 462, 1−21.
(26) (a) Becke, A. D. J. Chem. Phys. 1993, 98, 5648−5652. (b) Lee,
C.; Yang, W.; Parr, R. G. Phys. Rev. B 1988, 37, 785−789.
(27) For Gaussian basis sets: (a) Frisch, M. J.; Pople, J. A.; Binkley, J.
S. J. Chem. Phys. 1984, 80, 3265−3269. (b) Hehre, W. J.; Radom, L.;
Schleyer, P. v. R.; Pople, J. A. Ab initio Molecular Orbital Theory; John
Wiley: New York, 1986 , and references cited therein..
(12) Frauenrath, H.; Arenz, T.; Raabe, G.; Zorn, M. Angew. Chem.,
(28) Head-Gordon, M.; Pople, J. A.; Frisch, M. J. Chem. Phys. Lett.
1998, 153, 503−506.
Int. Ed. Engl. 1993, 32, 83−85.
(13) Wille, A.; Tomm, S.; Frauenrath, H. Synthesis 1998, 305−308.
(14) Tayama, E.; Otoyama, S.; Isaka, W. Chem. Commun. 2008,
4216−4218.
(29) For diffuse function: Clark, T.; Chandrasekhar, J.; Spitznagel, G.
W.; Schleyer, P. v. R. J. Comput. Chem. 1983, 4, 294−301.
(30) Barone, V.; Cossi, M. J. Phys. Chem. A 1998, 102, 1995−2001.
(31) For selected examples of theoretical studies on chiral phosphoric
acid catalysis including two-point binding mode, see: (a) Yamanaka,
M.; Itoh, J.; Fuchibe, K.; Akiyama, T. J. Am. Chem. Soc. 2007, 129,
6756−6764. (b) Yamanaka, M.; Hirata, T. J. Org. Chem. 2009, 74,
3266−3271. (c) Mori, K.; Katoh, T.; Suzuki, T.; Noji, T.; Yamanaka,
M.; Akiyama, T. Angew. Chem., Int. Ed. 2009, 48, 9652−9654.
(d) Marcelli, T.; Hammar, P.; Himo, F. Chem.Eur. J. 2008, 14,
(15) Koziol, A.; Grzeszczyk, B.; Koziol, A.; Staszewska-Krajewska, O.;
Furman, B.; Chmielewski, M. J. Org. Chem. 2010, 75, 6990−6993.
(16) For reviews on BINOL-derived phosphoric acids, see:
(a) Terada, M. Chem. Commun. 2008, 4097−4112. (b) Terada, M.
Synthesis 2010, 1929−1982. (c) Terada, M. Bull. Chem. Soc. Jpn. 2010,
101−119. (d) Zamfir, A.; Schenker, S.; Freund, M.; Tsogoeva, S. B.
Org. Biomol. Chem. 2010, 8, 5262−5276. (e) Terada, M. Curr. Org.
Chem. 2011, 15, 2227−2256. (f) Akiyama, T. In Science of Synthesis,
Asymmetric Organocatalysis 2, Brønsted Base and Acid Catalysts, and
Additional Topics; Maruoka, K., Ed.; Georg Thieme Verlag KG: New
York, 2012; Vol. 2, pp169−217. (g) Terada, M.; Momiyama, N., In
8562−8571. (e) Simon
́
, L.; Goodman, J. M. J. Am. Chem. Soc. 2008,
, L.; Goodman, J. M. J. Am. Chem. Soc.
130, 8741−8747. (f) Simon
́
2009, 131, 4070−4077. (g) Chen, X.-H.; Wei, Q.; Luo, S.-W.; Xiao,
M
dx.doi.org/10.1021/ja5017206 | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
H.; Gong, L.-Z. J. Am. Chem. Soc. 2009, 131, 13819−13825. (h) Li, N.;
Chen, X.-H.; Song, J.; Luo, S.-W.; Fan, W.; Gong, L.-Z. J. Am. Chem.
Soc. 2009, 131, 15301−15310. (i) Shi, F.-Q.; Song, B.-A. Org. Biomol.
́
Chem. 2009, 7, 1292−1298. (j) Simon, L.; Goodman, J. M. J. Org.
Chem. 2010, 75, 589−597. (k) Xu, S.; Wang, Z.; Li, Y.; Zhang, X.;
Wang, H.; Ding, K. Chem.Eur. J. 2010, 16, 3021−3035. (l) Cai, Q.-
A.; Zheng, C.; You, S.-L. Angew. Chem., Int. Ed. 2010, 49, 8666−8669.
́
(m) Simon, L.; Goodman, J. M. J. Org. Chem. 2011, 76, 1775−1788.
(n) Shi, F.; Xing, G.-J.; Tao, Z.-L.; Luo, S.-W.; Tu, S.-J.; Gong, L.-Z. J.
Org. Chem. 2012, 77, 6970−6979. (o) Mori, K.; Ichikawa, Y.;
Kobayashi, M.; Shibata, Y.; Yamanaka, M.; Akiyama, T. J. Am. Chem.
Soc. 2013, 135, 3964−3970. (p) Shibata, Y.; Yamanaka, M. J. Org.
Chem. 2013, 78, 3731−3736.
(32) The s-cis conformation of CNCO(Ot-Bu) is significantly
stable to neglect the s-trans conformation. The orientation of the
benzyl group is strictly restricted by the catalyst substituent introduced
at the 3,3′ positions of the binaphthyl backbone.
(33) Four transition states of (R)-(Z)-1a/(R)-3b, including TScor1
and TScor2, were obtained, whereas three transition states, including
TScos1 and TScos2, were optimized for (S)-(Z)-1a/(R)-3b. See
Supporting Information for details.
(34) TSccre-syn is also energetically unfavorable compared with
TSccre-anti in terms of activation energy (Figure 11).
(35) Crossover experiments were also conducted by changing the
concentrations of catalyst (R)-3a and racemic substrates ( )-(Z)-1a
and 1h. However, no significant concentration effect was observed in
all cases. See Supporting Information for details.
(36) A concerted pathway for the present rearrangement cannot be
completely ruled out. However, the concerted mechanism may not be
suitable for the major pathway of the present rearrangement because a
considerable amount of cross-products was obtained even in the CC
matched combination, as shown in Scheme 4.
(37) The reactions of ( )-(Z)-1a catalyzed by (R)-3a were
conducted in toluene at 40 and 0 °C to afford product 5a in 90%
yield with 98% anti and 89% ee (for anti-5a) and in 84% yield with
97% anti and 68% ee (for anti-5a), respectively.
N
dx.doi.org/10.1021/ja5017206 | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX