Transition-state charge stabilization through multiple non-covalent interactions in the guanidinium-catalyzed enantioselective claisen rearrangement

Article abstract of DOI:10.1021/ja110842s

The mechanism by which chiral arylpyrrole-substituted guanidinium ions promote the Claisen rearrangement of O-allyl α-ketoesters and induce enantioselectivity was investigated by experimental and computational methods. In addition to stabilization of the developing negative charge on the oxallyl fragment of the rearrangement transition state by hydrogen-bond donation, evidence was obtained for a secondary attractive interaction between the π-system of a catalyst aromatic substituent and the cationic allyl fragment. Across a series of substituted arylpyrrole derivatives, enantioselectivity was observed to vary predictably according to this proposal. This mechanistic analysis led to the development of a new p-dimethylaminophenyl-substituted catalyst, which afforded improvements in enantioselectivity relative to the parent phenyl catalyst for a representative set of substrates.

Full text of DOI:10.1021/ja110842s

ARTICLE
pubs.acs.org/JACS
Transition-State Charge Stabilization through Multiple Non-covalent
Interactions in the Guanidinium-Catalyzed Enantioselective
Claisen Rearrangement
Christopher Uyeda and Eric N. Jacobsen*
Department of Chemistry & Chemical Biology, Harvard University, Cambridge, Massachusetts 02138, United States
S Supporting Information
b
ABSTRACT: The mechanism by which chiral arylpyrrole-
substituted guanidinium ions promote the Claisen rearrange-
ment of O-allyl R-ketoesters and induce enantioselectivity was
investigated by experimental and computational methods. In
addition to stabilization of the developing negative charge on
the oxallyl fragment of the rearrangement transition state by
hydrogen-bond donation, evidence was obtained for a second-
ary attractive interaction between the π-system of a catalyst
aromatic substituent and the cationic allyl fragment. Across a
series of substituted arylpyrrole derivatives, enantioselectivity was observed to vary predictably according to this proposal. This
mechanistic analysis led to the development of a new p-dimethylaminophenyl-substituted catalyst, which afforded improvements in
enantioselectivity relative to the parent phenyl catalyst for a representative set of substrates.
Scheme 1. Hydrogen-Bonding Interactions in the Structure
of Bacillus subtilis Chorismate Mutase Bound to the
Oxabicyclic Transition-State Analogue 1
’ INTRODUCTION
Since its initial report nearly a century ago, the [3,3]-sigma-
tropic rearrangement of allyl vinyl ethers—the Claisen rearran-
gement—has been applied extensively in the synthesis of
structurally and stereochemically complex organic molecules.1
A principal feature of the Claisen rearrangement that underlies its
synthetic utility is the high and predictable diastereoselectivity
imparted by the pericyclic mechanism, allowing R- and β-
stereogenic carbonyl compounds of either the syn or anti relative
configuration to be prepared from precursors bearing the appro-
priate alkene geometries. While early efforts to obtain enantioen-
riched Claisen rearrangement products were focused on the use
of chiral substrates, particularly those derived from secondary
allylic alcohols, asymmetric methods involving metal-based cata-
lysts have recently been developed.2-4 Limitations in the scope
of these reactions persist, however, due to challenges associated
with competing background rearrangement and the strong
binding affinity of the products to the catalysts. Allyl vinyl ether
substrates are also susceptible to fragmentation in the presence of
catalysts that are either strongly Lewis acidic or promote the
formation of π-allyl metal species, and the dissociated inter-
mediates are often observed to recombine to form mixtures
of regioisomeric [1,3]- and diastereomeric [3,3]-rearrangement
products.
analogue 16 have led to the identification of arginine and/or
lysine residues in the active site that are positioned to interact
with the core heteroatom of the allyl vinyl ether system as well as
the pendant carboxylate functional groups (Scheme 1). While
the relative contributions of selective transition-state stabilization
Chorismate mutases accelerate the Claisen rearrangement of
chorismate to prephenate on the order of a million-fold by a
mechanism that involves the formation of multiple non-covalent
interactions between the enzyme and substrate. X-ray structures
of Bacillus subtilis5a,b (BsCM) and Escherichia coli5c chorismate
mutases co-crystallized with the oxa-bicyclic transition-state
Received: December 14, 2010
Published: March 10, 2011
r
2011 American Chemical Society
5062
dx.doi.org/10.1021/ja110842s J. Am. Chem. Soc. 2011, 133, 5062–5075
|

Journal of the American Chemical Society
ARTICLE
and substrate conformational effects in the mechanism of
catalysis have not been definitively elucidated,7 mutagenesis
studies have established the critical importance of these cationic
hydrogen-bond-donor residues. Several BsCM mutants that
incorporate lysine at position 90 or the nearby position 88 are
also catalytically competent;8a however, the Arg90Ala single-
point mutant exhibits no chorismate mutase activity.8b Further-
more, the replacement of Arg 90 with citrulline, an isosteric urea-
containing residue that is charge-neutral, results in a >10 000-fold
decrease in rate, while ground-state binding to 1 is minimally
disrupted.9
Scheme 2. Enantioselective Claisen Rearrangement of
O-Allyl R-Ketoesters Catalyzed by Pyrrole-Substituted
Guanidinium Catalysts
Electrostatic stabilization of the developing positive charge on
the allyl fragment of the rearrangement transition state has also
been proposed as a complementary mechanism of catalysis by
chorismate mutases. The active site of BsCM in particular
contains a phenylalanine residue, which is potentially oriented
to provide π-stabilization, at a 3.59 Å C-C distance from the
transition-state analogue.5b
Despite the fact that valuable mechanistic insight has been
gleaned from structural studies of chorismate mutases bound to
inhibitors that are geometric mimics of the rearrangement
transition state, these analogues possess neither the charge
distribution nor the dissociated structure of the actual pericyclic
transition state. Catalytic antibodies developed using such in-
hibitors display modest activity compared to the wild-type
enzyme,10 an observation that has been attributed to poor
electrostatic stabilization of the dipolar transition state.11 The
active site of the 1F7 antibody, for example, contains only a single
cationic hydrogen-bond donor , which is likely occupied in a salt
bridge with a carboxylate group.12
As a complement to these studies of Claisen rearrangements
mediated by biological macromolecules, we have investigated
non-covalent catalyst-transition state interactions in the context
of small-molecule hydrogen-bond donors that have the advan-
tages of being readily accessible by synthesis and amenable to
modeling using high-level computational methods. Guided in
part by the proposed mechanism of substrate activation by
chorismate mutases, we identified simple guanidinium ion deri-
vatives as effective catalysts for the [3,3]-sigmatropic rearrange-
ment of a variety of substrates in non-polar organic solvents.13
Allyl vinyl ethers bearing substituents that promote dipolar
transition structures were found to be particularly amenable to
catalysis by such hydrogen-bond donors.14,15 We subsequently
identified and optimized chiral, C2-symmetric guanidinium ion
derivatives as catalysts for enantioselective rearrangements of
chorismate analogues with carboxyl substitution on the vinyl
group (Scheme 2). Pyrrolo-trans-diaminocyclohexane-derived
guanidinium ions bearing aryl substituents at the 2-position of
the pyrrole group (e.g., 2) were found to be particularly effective,
displaying significantly higher reactivity and enantioselectivity
relative to pyrrole derivatives such as 3 that bear aliphatic substit-
uents. This empirical observation pointed to the intriguing possi-
bility of a secondary stabilizingrole of catalyst aromatic substituents
in the Claisen rearrangement transition state. Such an interaction
would be analogous to that proposed for an active-site phenylala-
nine residue in BsCM5b and is also precedented in the association
of both ground-state and transition-state cations to aromatic
π-systems in other well-characterized protein complexes.16
In an effort to elucidate the non-covalent interactions that are
responsible for rate acceleration and asymmetric induction in
guanidinium-catalyzed rearrangements of O-allyl R-ketoesters,
we have carried out kinetic analyses and quantitative catalyst
Table 1. Dependence of Conversion and Product Enantio-
meric Excess on Catalyst Structure and Solvent
entry
catalyst
solvent
conversion (%)a
ee (%)b
1
2
3
4
5
6
(R,R)-2
(R,R)-3
(R,R)-2
(R,R)-2
(R,R)-2
(R,R)-2
hexanes
hexanes
toluene
CH2Cl2
CDCl3
TBME
85
59
82
83
79
16
73
41
72
65
66
19
1
a Conversions were determined from crude reaction mixtures by H
NMR signal integration. All rearrangements afforded product 6 with a
>20:1 dr. b Enantiomeric excesses of purified products were determined
by GC analysis using commercial chiral columns.
structure-enantioselectivity relationship studies in combination
with computational transition-state modeling. A mechanistic
picture emerges in which hydrogen-bonding interactions with
the Lewis-basic heteroatoms of the substrate operate coopera-
tively with π-stabilization of the cationic charge developing on
the allyl fragment in the energetically favored rearrangement
transition state.
’ RESULTS AND DISCUSSION
Kinetic Studies of Guanidinium-Catalyzed Rearrange-
ments. The model O-crotyl 2-oxobutyrate substrate 5 was
observed to undergo a [3,3]-sigmatropic rearrangement cata-
lyzed by 20 mol % (R,R)-2 to afford 6 in 73% enantiomeric excess
(ee) and a >20:1 diastereomeric ratio (dr). The methyl-sub-
stitued catalyst (R,R)-3 exhibited both measurably decreased
activity and enantioselectivity (Table 1, entry 2).17 The product
stereochemistry was established as (S,S) by X-ray analysis of a
crystalline iodoether derivative, the relative configuration being
consistent with rearrangement through a chair-like six-membered
transition structure. In hexanes, the catalyst is completely
5063
dx.doi.org/10.1021/ja110842s |J. Am. Chem. Soc. 2011, 133, 5062–5075

Journal of the American Chemical Society
ARTICLE
Figure 1. Dependence of (A) conversion and (B) product 6 ee on catalyst (R,R)-2 loading in hexanes and CDCl3. Reactions were conducted at 40 °C
for 14 h.
insoluble even in the presence of the substrate, suggesting that
reactions in this medium occur in the precipitated catalyst phase.
Despite the aggregated state of the catalyst, no evidence for
diastereomeric interactions between multiple guanidinium ions
was found in either the ground state or the rearrangement
transition state, as reactions conducted with scalemic mixtures
of catalyst 2 displayed a strictly linear dependence of product ee
on catalyst ee and nearly identical rates.18,19
In other non-polar organic solvents (Table 1, entries 3-5),
where the catalyst is either partially or completely soluble, rearra-
ngements proceeded with slightly diminished enantioselectivity.
The uncatalyzed rate was found to be significantly higher in these
solvents, however, suggesting that the lower product ee is pri-
marily a consequence of competing background rearrangement.
In accord with this hypothesis, enantioselectivities for reactions
conducted in hexanes and CDCl3 were observed to converge as
the catalyst loading was increased (Figure 1). In more Lewis-
basic solvents such as tert-butyl methyl ether (TBME), which is
capable of binding to the guanidinium ion in competition with
the substrate, low levels of catalysis were observed. The similarity
in enantioselectivity across a range of non-polar, non-coordinat-
ing solvents and under both heterogeneous and homogeneous
conditions suggests a common mechanism of catalysis and asym-
metric induction that is unlikely to involve the explicit participa-
tion of solvent molecules in the rearrangement transition state.
In order to establish the stoichiometry of the catalyzed rearra-
ngement transition state and identify catalyst resting states for
the reaction of 5 catalyzed by (R,R)-2, kinetic studies were
performed under fully homogeneous conditions in CDCl3.20,21
We conducted a series of experiments using a constant initial
substrate 5 concentration of 0.05 M and variable loadings of (R,
R)-2 from 0 to 30 mol %, monitoring the concentrations of both
Figure 2. Rate profiles for reactions with various loadings of catalyst (R,
R)-2 ([(R,R)-2]tot = 0.005-0.02 M; [5]i = 0.05 M; 55 °C, CDCl3). Each
set of points is the average rate determined from two individual kinetics
experiments, with the error bars representing the range of measure-
ments. The curves are best fits of the rate vs concentration data to eq 1.
sensitivity of rate to solvent polarity has been observed for other
pericyclic reactions that proceed through transition states that
are substantially more polarized than the ground state.14,24
The rate data for rearrangements catalyzed by (R,R)-2 are
consistent with first-order dependence on the total catalyst
concentration and saturation behavior in the substrate.
Rearrangements were also conducted at a constant (R,R)-2
total concentration of 0.01 M and initial concentrations of
substrate 5 ranging from 0.05 to 0.2 M (Figure 3). The lack of
overlay between the rate curves suggests substantial inhibition by
product 6, which accumulates over the course of the reaction.20
The contributions of various catalyst resting states to reaction
rate were quantified by fitting the data from all 10 kinetics
experiments to a rate law of the general form shown in eq 1 (R2 =
0.989), which contains terms for an uncatalyzed unimolecular
rearrangement and a catalyst-mediated rearrangement.
1
the substrate and the product by H NMR (Figure 2).22 All
rearrangements proceeded without the formation of detectable
side products, and no decomposition of the catalyst was detect-
able spectroscopically over the course of the reaction.
d½5ꢁ
dt
a½ðR, RÞ-2ꢁtot½5ꢁ
1 þ b½5ꢁ þ c½6ꢁ
The uncatalyzed rearrangement exhibits simple first-order rate
rate ¼ -
¼
þ kuncat½5ꢁ
ð1Þ
dependence on [5], with a rate constant (kuncat) of1.42ꢀ 10-5 -1
s .
By application of the Eyring equation, an activation free energy
(ΔG328q) of 26.5 kcal/mol was calculated.23 Using initial reac-
tion rates measured over the first 10% conversion, kuncat in
hexanes was determined to be 2.11 ꢀ 10-6 s-1 at the same
temperature, corresponding to ΔG328q = 27.8 kcal/mol. Similar
Exchange between unbound and various bound states of the
catalyst was found to be fast on the 1H NMR time scale at 55 °C,
indicating that the rates of catalyst association and dissociation
processes are faster by several orders of magnitude than the rate
5064
dx.doi.org/10.1021/ja110842s |J. Am. Chem. Soc. 2011, 133, 5062–5075

Journal of the American Chemical Society
ARTICLE
Figure 3. Rate profiles for reactions with various initial concentrations
of 5 ([(R,R)-2]tot = 0.01 M; [5]i = 0.05-0.2 M; 55 °C, CDCl3). Each set
of points is the average rate determined from two individual kinetics
experiments, with the error bars representing the range of measure-
ments. The curves are best fits of the rate vs concentration data to eq 1.
Scheme 3. Rate and Equilibrium Constants for Catalyzed
and Uncatalyzed Rearrangements of 5 in CDCl3
Figure 4. Rate profiles for the rearrangement of 5, catalyzed by (R,R)-2
(red circles), (R,R)-3 (blue squares), and 4 (green diamonds). [cat]tot
=
0.01 M; [5]i = 0.05 M; 55 °C, CDCl3.
structures and relative energies of Claisen rearrangement transi-
tion states was further investigated by computational methods
using Gaussian 03.25 We first examined the mechanism of the
uncatalyzed rearrangement of 5 at the B3LYP/6-31G(d) level of
density functional theory (DFT), which has been utilized
extensively to model pericyclic reactions and has been validated
against experimental kinetic isotope effect and activation energy
data.26,27 Stationary points were located for the substrate,
chair-like rearrangement transition state, and product with both
s-trans and s-cis stereochemical relationships between the vinyl
ether and ester groups (Figure 5).28,29 For the substrate and
transition state, very small energetic differences of <1 kcal/mol
were calculated between the two conformations; however, a
more significant difference in energy of 1.8 kcal/mol between the
s-trans and s-cis conformations of the product was determined.30
An activation energy (ΔEq) of 27.4 kcal/mol was calculated
for the uncatalyzed rearrangement; from frequency calculations,
ΔGq298 was estimated to be 28.2 kcal/mol. Boat-like transition
structures were also optimized and are approximately 4 kcal/mol
higher in energy, consistent with the high levels of diastereo-
selectivity observed in these rearrangements.
In order to establish plausible modes of interaction between
the guanidinium functional group and the substrate, a simplified
catalyst, N,N0-dimethylguanidinium ion (7), was modeled. Cat-
alyst complexes of the s-cis conformational series are consistently
lower in energy than for the s-trans series by >4 kcal/mol,
indicating that hydrogen-bonding to the ester carbonyl is en-
ergetically favored. The calculated activation energy for the
rearrangement is lowered by 4.4 kcal/mol in the guanidinium-
catalyzed pathway relative to the uncatalyzed pathway. In the
catalyst-bound transition-state complex, the length of the hydro-
gen bond between the catalyst and the ether oxygen is decreased
by 0.08 Å relative to that in the ground state.31 This shortening of
the hydrogen-bonding distance can be rationalized by greater
of rearrangement. It is, therefore, possible to apply an equilibrium
approximation to the data in order to estimate binding constants
for the kinetically relevant catalyst complexes (Scheme 3). From
the kinetic parameters, the association constant (Ka) of the
substrate-catalyst complex (R,R)-2 5 was calculated to be 1.7
3
times greater than the value for the product-catalyst complex
(R,R)-2 6. The kcat/kuncat deduced from these data is 37, which
3
corresponds to a 2.3 kcal/mol lowering of the activation free
energy at 328 K. In hexanes, kcat/kuncat = 250 was approximated,
under the assumption that the rate constant for the catalyzed
rearrangement is independent of solvent; at high catalyst load-
ings, reactions in CDCl3 and hexanes proceed to similar levels of
conversion at 40 °C after 14 h (Figure 1).
The rate profile for catalyst (R,R)-2 was compared to that of
the methyl-substituted catalyst (R,R)-3 as well as the N,N0-dicy-
clohexylguanidinium catalyst 4 (Figure 4). Both of these guanidi-
nium ions exhibit significantly lower catalytic activity than (R,R)-2,
suggesting that while the presence of the pyrrole functional group
does not affect the rate, the phenyl substituent plays a significant
role in lowering the activation barrier for the rearrangement.
Computational Studies with a Simplified Model Guanidi-
nium Catalyst. The effect of guanidinium ion catalysis on the
5065
dx.doi.org/10.1021/ja110842s |J. Am. Chem. Soc. 2011, 133, 5062–5075

Journal of the American Chemical Society
ARTICLE
Figure 5. Energy diagram for the uncatalyzed (top pathway) and N,N0-dimethylguanidinium (7)-catalyzed (bottom pathway) rearrangement of 5 to 6.
All stationary points are fully optimized at the B3LYP/6-31G(d) level of theory and verified by frequency analysis. Uncorrected electronic energies in
kcal/mol are relative to the lowest-energy structure of the substrate or catalyst-substrate complex. Distances for hydrogen-bonding interactions are
shown in angstroms.
electrostatic stabilization of the developing negative charge in the
transition state. By contrast, a negligible geometric change
(0.01 Å) is observed in the interaction with the ester carbonyl.
Optimized geometries for thermal and N,N0-dimethylguan-
idinium-catalyzed rearrangement transition structures are shown
in Figure 6. Compared to the transition structure for unsubsti-
tuted allyl vinyl ether, s-cis-TS-5 is substantially more dissociated,
with longer breaking C-O and forming C-C bond distances.
Estimates of the partial charges on the allyl and oxallyl frag-
ments using the Mulliken, Natural Bond Orbital (NBO),32 and
CHELPG33 methods of population analysis consistently indicate
increased dipolar character as a consequence of ester substitu-
tion. In the presence of the guanidinium ion, the partial C-O
and C-C bond distances are further lengthened by approxi-
mately 0.1 Å, and a fraction of the guanidinium ion positive
charge is delocalized primarily into the allyl fragment of the
rearrangement transition state.
as the pyrrole-cyclohexane C-N bond. In the lowest-energy
structure, shown in Figure 8, the guanidinium functional group is
disposed in a (Z,Z)-conformation, consistent with both an X-ray
structure that was previously obtained13 for (R,R)-2 (Figure 7A)
and ROESY cross-peaks that were observed for a catalyst
solution in CDCl3 (Figure 7B).
The guanidinium ion NH2 hydrogens reside in close contact
with the π-faces of the pyrrole rings: there is a 3.23 Å distance
between the guanidinium nitrogen and nearest carbon atom of
the pyrrole in the calculated structure—these distances are 3.22
and 3.19 Å in the crystal structure.34 This intramolecular inter-
action influences the degree to which the cyclohexane ring is canted
with respect to the plane of the guanidinium ion and places the
phenyl groups in proximity to the substrate binding site. The
sensitivity of the energy of this structure to deviations from the
ground-state geometry was probed by computationally scanning
the dihedral angle defined by the C-NH2 bond of the guanidinium
group and the axial C-H bond of the cyclohexane ring, highlighted
in red in Figure 8. We performed constrained optimizations at 5°
dihedral increments, and the relative energies were calculated for
both (R,R)-2 and N,N0-dicyclohexylguanidinium ion 4. It is evident
from the comparison of these two scans that, for (R,R)-2, rotation
of the dihedral angle in the negative direction is hindered by
repulsive interactions between the guanidinium ion and the pyrrole,
and rotation in the positive direction is disfavored due toweakening
of the guanidinium-pyrrole interaction.
As a point of comparison, the lowest-energy transition struc-
ture for the N,N0-3,5-bis(trifluoromethyl)phenylthiourea (8)-
catalyzed rearrangement is also shown in Figure 6. In accord
with earlier observations that N-aryl urea and thiourea derivatives
such as 8 display little catalytic activity in the rearrangement of O-
allyl R-ketoesters,13,15b the calculated activation energy as well as
distance and charge metrics for 8 s-cis-TS-5 are intermediate
3
between those for the uncatalyzed (s-cis-TS-5) and N,N0-di-
methylguanidinium-catalyzed (7 s-cis-TS-5) transition states.
3
Structure and Conformations of Catalyst 2. Having estab-
lished a basic model for the hydrogen-bonding interactions
between a simplified guanidinium ion, 7, and the substrate, we
turned our attention to studies involving the chiral catalyst 2. The
geometry of the cation was optimized computationally at the
B3LYP/6-31G(d) level of theory, and minima were located for
various rotamers about the guanidinium ion C-N bonds as well
Ground-State Binding Interactions. Kinetic data presented
above were consistent with an initial, reversible binding of the
substrate to the catalyst prior to the rate-limiting sigmatropic
rearrangement. At 22 °C, the rearrangement of 5 in the presence
of (R,R)-2 was sufficiently slow to allow the substrate-catalyst
complex to be studied directly by 1H NMR. A selected region of
the spectra for a series of equimolar solutions of 5 and (R,R)-2 in
5066
dx.doi.org/10.1021/ja110842s |J. Am. Chem. Soc. 2011, 133, 5062–5075

Journal of the American Chemical Society
ARTICLE
Figure 7. (A) 50% probability ellipsoid representation of the X-ray
structure of (R,R)-2 co-crystallized with two isopropanol molecules. The
counterion is omitted for clarity. (B) Selected ROESY cross-peaks for
(R,R)-2 in CDCl3.
Figure 6. Calculated transition structures at the B3LYP/6-31G(d) level
of theory for (A) the uncatalyzed rearrangement of allyl vinyl ether, (B)
the uncatalyzed rearrangement of 5, (C) the rearrangement of 5
catalyzed by guanidinium ion 7, and (D) the rearrangement of 5
catalyzed by thiourea 8. Distances for the breaking C-O and forming
C-C bonds as well as hydrogen bonds are in angstroms. Mulliken
charges, NBO charges in parentheses, and CHELPG charges in square
brackets for the oxallyl and allyl fragments as well as the guanidinium ion
are shown in red.
a concentration range of 0.1-0.00026 M is shown in Figure 9A.
At the high-concentration limit of the experiment, the signal
corresponding to the guanidinium N-Ha protons is shifted
downfield by approximately 1.5 ppm relative to that in the free
catalyst, consistent with a binding event that involves a hydrogen-
bonding interaction to the substrate.35 A similar shift was
observed for dilution experiments performed with 80% ee (S,
S)-6 (Figure 9C). By comparison, the chemical shift of the N-Hb
protons of the guanidinium NH2 group at 3.5 ppm remains
relatively unchanged. The methylene protons (Hd) of 5 , which
appear as a doublet for the free substrate, become diastereotopic
and undergo a >1 ppm upfield shift upon complexation, suggest-
ing an intimate association with the chiral framework of the
catalyst.
Figure 8. (Top) (Z,Z)-geometry of (R,R)-2 optimized at the B3LYP/
6-31G(d) level of theory and (bottom) scan of the dihedral angle
between the C-NH2 bond of the guanidinium ion and the axial C-H
bond of the cyclohexane ring, highlighted in red.
closest contact between a substrate hydrogen atom and a carbon
atom of the phenyl ring is 3.0 Å. This geometry provides a ratio-
nale for the spectroscopically observed complexation-induced
upfield shift of the substrate signals corresponding to the methyl-
ene group protons, Hd.36
The chemical shift data for the methyl ester singlet (Hc) of 5
over the entire concentration range of the dilution experiment
provided a good fit to a 1:1 binding model (R2 = 0.996), and Ka =
218 M-1 was calculated.37 A 1:1 stoichiometry for the complex
was further established by the method of continuous variation
(Job’s plot).38,39 The same dilution procedure was repeated for
the complexes between 80% ee (S,S)-6 and each enantiomer of
catalyst 2 (Table 2). The two diastereomeric catalyst-product
complexes were thus found to exhibit nearly identical binding
More detailed structural insight into the (R,R)-2 5 complex
3
was obtained by computational optimization of its geometry. In
the lowest-energy structure, the substrate is in a pro-(S,S)
conformation, and the methylene group is located in proximity
to the π-faces of the catalyst phenyl substituents (Figure 10). The
5067
dx.doi.org/10.1021/ja110842s |J. Am. Chem. Soc. 2011, 133, 5062–5075

Journal of the American Chemical Society
ARTICLE
Figure 9. (Left) 1H NMR dilutions of (A) a 1:1 mixture of (R,R)-2 and 5 in CDCl3 (0.1-0.00026 M) and (B) the free substrate 5. (Right) 1H NMR
dilutions of (C) a 1:1 mixture of (R,R)-2 and 80% ee (S,S)-6 in CDCl3 (0.1-0.00031 M) and (D) the free product 6.
Table 2. Binding Constant Measurements
Figure 10. (R,R)-2 pro-(S,S)-5 complex optimized at the B3LYP/
3
6-31G(d) level of theory.
constants. The Ka for 5 was determined to be roughly twice the
value for 6, corresponding to a 0.41 kcal/mol energetic prefe-
rence for substrate binding over product binding. The relative
values of these binding free energies measured at 22 °C are con-
sistent with those extracted from the kinetic data at 55 °C.
Computational Model for the Enantioselective Rearran-
gement. Having established the basic stoichiometry of the
catalyzed rearrangement transition state from kinetics experi-
ments and examined ground-state binding interactions, we next
conducted computational modeling studies with the full struc-
ture of catalyst 2 in order to gain more detailed insight into the
origin of asymmetric induction. Of particular interest was a
rationale for the beneficial effect of the catalyst phenyl substi-
tuent on both rate and enantioselectivity. Geometries of the
catalyst-bound substrate, rearrangement transition state, and
product were fully optimized in the gas phase at the B3LYP/6-
31G(d) level of DFT. Relative energies for these structures,
leading to both the experimentally observed major (S,S) and
minor (R,R) enantiomers of product 6, are shown in Figure 11.
The energy difference between the two diastereomeric rear-
rangement transition-state complexes (ΔΔEq) was calculated to
be 2.99 kcal/mol, with the pro-(S,S) transition state being the
lowest in energy. Detailed representations of these structures
are shown in Figure 12. In both complexes, the oxallyl fragment
is roughly planar with respect to the guanidinium ion of the
entry
catalyst
substrate/product
Ka (M-1 a
)
1
2
3
(R,R)-2
(R,R)-2
(S,S)-2
5
218 ( 14
108 ( 4
107 ( 4
(S,S)-6 (80% ee)
(S,S)-6 (80% ee)
a Association constants for a 1:1 complex between the guanidinium ion 2
and 5 or 6. Uncertainties are standard errors of the curve fit.
catalyst, and the allyl fragment is projected toward either the
phenyl substituent of the pyrrole in the case of the major
transition state (Figure 12A) or the cyclohexanediamine back-
bone in the minor transition state (Figure 12B). The energetic
preference for interaction of the allyl fragment with the phenyl
group vs the cyclohexane ring provides a plausible explanation for
the calculated difference in transition-state energies. In the major
transition structure, the closest distance between a C-H of the
allyl fragment and the centroid of the phenyl ring is 2.98 Å,
placing it within an appropriate distance for an attractive
interaction.40
5068
dx.doi.org/10.1021/ja110842s |J. Am. Chem. Soc. 2011, 133, 5062–5075

Journal of the American Chemical Society
ARTICLE
Figure 11. Energy diagram for the asymmetric rearrangement of 5 to 6, catalyzed by (R,R)-2. All stationary points are fully optimized at the B3LYP/
6-31G(d) level of theory and verified by frequency analysis. Uncorrected electronic energies in kcal/mol are relative to (R,R)-2 pro-(S,S)-5. Distances
3
for hydrogen-bonding interactions are shown in angstroms.
Figure 12. Fully optimized diastereomeric transition structures for the rearrangement of 5 catalyzed by (R,R)-2, leading to the (A) major pro-(S,S) and
(B) minor pro-(R,R) enantiomers of the product (B3LYP/6-31G(d)). Key distances for non-covalent interactions are shown in angstroms.
5069
dx.doi.org/10.1021/ja110842s |J. Am. Chem. Soc. 2011, 133, 5062–5075

Journal of the American Chemical Society
ARTICLE
Table 3. Comparison of Computational Methods
computational method
ΔΔEq (kcal/mol)a
B3LYP/6-31G(d)
2.99 (2.89)
3.24
B3LYP/6-311þG(d,p)//B3LYP/6-31G(d)
MP2/6-31G(d)//B3LYP/6-31G(d)
M05-2X/6-31G(d)
2.69
2.44 (2.47)
M05-2X/6-31þG(d,p)//M05-2X/6-31G(d)
2.55
a Uncorrected differences in transition-state energies. Values in par-
entheses include an unscaled correction for zero-point vibrational
energy.
Table 4. Dependence of Enantioselectivity on Catalyst
Electronics
Figure 13. Experimental vs calculated B3LYP enantioselectivity. The
black line represents a least-squares fit to a linear function (intercept,
-3.41; slope, 5.33; R2 = 0.74).
catalyst
catalyst substituent
exptl era exptl ΔΔGq (kcal/mol)b
2
-
6.33 ( 0.05
5.05 ( 0.12
8.01 ( 0.20
2.61 ( 0.01
1.15 ( 0.01
1.01 ( 0.01
1.29 ( 0.02
0.597 ( 0.002
0.830 ( 0.001
1.21 ( 0.01
0.92 ( 0.01
9a
R = 4-fluoro
9b
9c
R = 4-dimethylamino
R = 3,4,5-trifluoro
Figure 14. Experimental vs calculated M05-2X enantioselectivity. The
black line represents a least-squares fit to a linear function (intercept,
-2.10; slope, 3.95; R2 = 0.88).
9d
10a
10b
R = 2,3,4,5,6-pentafluoro 3.80 ( 0.01
R0 = methyl
R0 = trifluoromethyl
7.03 ( 0.03
4.40 ( 0.05
2.29 kcal/mol at the B3LYP/6-31G(d) level of theory.41 From
the comparison of this value to that obtained for (R,R)-2, the
interaction of the phenyl substituent with the allyl fragment can
be estimated to provide approximately 0.7 kcal/mol of stabi-
lization to the major transition state. As an alternative explana-
tion, the lower enantioselectivity both observed and calculated
for catalyst (R,R)-3 might be rationalized on the basis of repulsive
non-bonding interactions with the methyl group in the major
transition state. However, the distance between the closest
hydrogen atoms on the methyl group and the allyl fragment in
the geometry-optimized structure is 2.56 Å, placing them outside
of van der Waals contact.42 Furthermore, such a steric transition-
state destabilization model would predict decreased reactivity for
the 2-methylpyrrole-bearing catalyst (R,R)-3; in fact, kinetic
studies revealed that (R,R)-3 catalyzed the rearrangement with
a rate similar to that of the N,N0-dicyclohexylguanidinium
catalyst 4.
Electronically Substituted Arylpyrrole Catalysts. In order
to devise an experimental test of the proposed stabilizing role of
the catalyst phenyl substituent in the lowest-energy diastereo-
meric transition state, we prepared and evaluated a series of
arylpyrrole catalysts bearing substitution that was expected to
perturb this interaction (9a-d, 10a,b, Table 4). Substituent
effects on the strength and geometry of ground-state cation-π
interactions have been modeled computationally and studied
a Enantiomeric ratios are averages of two experiments, with the error
bars representing the range of results. b Relative activation free energies
were estimated according to classical transition-state theory (ΔΔGq =
-RT ln([(S,S)-6]/[(R,R)-6]), T = 313.15 K).
Single-point energies for the two transition states were also
calculated using a larger basis set as well as the MP2 method in
order to establish that the model for selectivity is robust across
different levels of theory. The transition structures were also fully
optimized using the M05-2X functional. These results are sum-
marized in Table 3. All computational methods are in agreement
with respect to the sense of enantioinduction and accurately
predict the observed absolute configuration of the product. The
magnitudes of ΔΔEq, while narrowly distributed, consistently
overestimate the experimental enantioselectivity. These discre-
pancies might be due to neglect of entropic contributions and
medium effects in the computational model. Regardless of the
source of the calculated overestimation of ee, it is expected that
many of the errors associated with these approximations cancel
out in the analysis of selectivity trends across different catalyst
structures.
The geometries of the pro-(S,S) and pro-(R,R) transition
structures were also optimized for the methyl-substituted
pyrrole catalyst (R,R)-3, and the ΔΔEq was calculated to be
5070
dx.doi.org/10.1021/ja110842s |J. Am. Chem. Soc. 2011, 133, 5062–5075

Journal of the American Chemical Society
ARTICLE
experimentally by the measurement of gas-phase interactions
energies and solution-phase binding constants; however, few
reports have described the systematic characterization of these
effects for transition states with cationic character.43-45
Enantioselectivities were determined for reactions performed
in hexanes at 40 °C using 20 mol % catalyst loading (Table 4).
For all catalysts, rearrangements under these conditions proce-
eded to high levels of conversion relative to the thermal rear-
rangement conducted in the absence of catalyst, indicating that
the trends that were observed are related to the intrinsic enan-
tioselectivity of the catalyzed pathway rather than variable
amounts of competing racemic background reaction. Catalyst
9b, which contains an electron-donating 4-dimethylamino sub-
stituent, provided higher levels of enantioselectivity than 2, while
catalyst 9a, with an inductively withdrawing 4-fluoro substituent,
exhibited the opposite effect. Polyfluorinated catalysts 9c,d
afforded particularly low enantioselectivities compared to the
parent catalyst. Direct substitution of the pyrrole ring was also
explored with either a donating methyl or withdrawing trifluoro-
methyl group (10a,b).
Enantioselectivities were also determined computationally at
the B3LYP/6-31G(d) level of theory from the energy difference
between diastereomeric transition states of the general structure
shown in Figure 12.46 Although the overall trend in enantio-
selectivity was reproduced, poor quantitative correspondence was
observed between calculated and experimental results (Figure 13).47
Because of the well-established limitations of the B3LYP func-
tional in accurately reproducing the energy of weak non-covalent
interactions such as cation-π interactions, transition structures
for all catalysts were also fully optimized using the M05-2X
functional, which has been specifically parametrized for such
purposes.48 Using the latter method, significantly higher corre-
lation (R2 = 0.88) with experimental data was observed
(Figure 14).
Electrostatic potential maps were generated for a representa-
tive sample of arylpyrrole structures in order to provide a
qualitative model for the observed trend in enantioselectivity
(Figure 15). Dougherty has shown that variations in ground-state
binding energies between alkali metal cations and substituted
arenes can be largely correlated with the electrostatic component
of the interaction.43 The most selective catalyst, 9b, has signifi-
cantly greater negative potential above the π-face of the arene,
while the fluorinated catalysts have significantly less negative
potential. For the pentafluorophenyl catalyst 9d, the electrostatic
component of the interaction between the cationic allyl fragment
and the π-system is expected to be repulsive.
A comparison of the optimized geometries of the major pro-(S,
S) transition structures for the phenyl catalyst 2 and the penta-
fluorophenyl catalyst 9d at the M05-2X level of DFT is shown in
Figure 16. For 9d, the distance between the closest hydrogen
atom of the allyl fragment and the centroid of the arene is
significantly lengthened to 3.26 Å, compared to 2.59 Å in the
phenyl-substituted catalyst 2. A second pro-(S,S) transition
structure of lower energy, by 0.6 kcal/mol, was located in which
the oxallyl group of the substrate is nearly perpendicular relative
Figure 15. Electrostatic potential maps for fully optimized structures
(B3LYP/6-31G(d)) of (A) the rearrangement transition state for 5 and
N-methyl (B) 2-phenylpyrrole, (C) 2-(4-(dimethylamino)phenyl)pyr-
role, (D) 2-(4-fluorophenyl)pyrrole, and (E) 2-pentafluorophenylpyr-
role. Negative potentials are shown in red and positive potentials in blue.
Figure 16. pro-(S,S) Transition structures for (A) catalyst 2 and for the pentafluoro-substituted catalyst 9d highlighting interactions of the cationic allyl
fragment with (B) the π-face of the arene and (C) the meta-fluorine substituent (M05-2X/6-31G(d)). The relative energy of structures (B) and (C)
is shown.
5071
dx.doi.org/10.1021/ja110842s |J. Am. Chem. Soc. 2011, 133, 5062–5075

Journal of the American Chemical Society
ARTICLE
Table 5. Substrate Scope and Comparison of Catalysts 2 and 9ba
a Reactions run on a 0.1 mmol scale in 2 mL of hexanes using a 20 mol % loading of catalyst (R,R)-2 or (R,R)-9b. b Isolated yields following purification by
silica gel chromatography. c Diastereomeric ratios determined from 1H NMR spectra of the crude reaction mixture. d Enantiomeric excesses determined
by GC or HPLC analysis using commercial chiral columns (see Supporting Information).
to the plane of the guanidinium ion (Figure 16B). Although the
hydrogen-bond angles are far from ideal in this transition
structure, the electrostatic interaction between the allyl group
and the fluoroarene is more favorable, with a 2.27 Å distance
between a hydrogen atom on the allyl group and a meta-fluorine
substituent. Similar geometries for interactions between highly
fluorinated arenes and both early-transition-metal cations49 and
arene C-H bonds50 have been observed crystallographically.
Substrate Scope with the Dimethylamino-Substituted
Catalyst 9b. Having established the importance of catalyst
electronic effects on the enantioselective rearrangement of model
substrate 5, we compared (R,R)-2 and the dimethylamino-substi-
tuted catalyst (R,R)-9b for a representative set of rearrangements.
Small increases in enantioselectivity, corresponding to an average free
energy of 0.19 ( 0.08 kcal/mol for entries in Table 5, were
consistently observed across a range of O-allyl R-ketoesters with
different olefin substitution patterns. Substrates were selected that
form products with R-stereogenic centers of different steric demands
(entries 1 and 2), vicinal tertiary stereogenic centers of both the syn
and anti relative stereochemistry (entries 3-5), and β-quaternary
stereogenic centers (entries 6 and 7).
rearrangement of 5, as compared to the thermal rearrangement in
hexanes, corresponding to a rate acceleration of approximately
250-fold. In computational models, guanidinium catalysts are
seen to interact with the allyl vinyl ether substrate through
hydrogen bonds with both the ether oxygen atom and the
pendant ester group. This interaction allows stabilization of the
developing negative charge in the transition state. For rearrange-
ments catalyzed by 2, a secondary interaction is evident in the
major diastereomeric transition state between the π-system of
the catalyst phenyl substituent and the cationic allyl fragment of
the substrate. This proposal is supported by the experimental
observation that 2 is both more enantioselective and more active
than 3, which lacks an appropriately positioned Lewis-basic
functional group. Furthermore, the strength of this interaction
is rationally tunable through substitution of the arene. Thus,
catalyst 9b, which possesses a dimethylamino substituent, is
more selective than 2 for a range of substrates with different
olefin substitutions.
While the mechanistic model proposed here for rearrange-
ments promoted by a synthetic small-molecule catalyst bears
striking analogy to the enzymatic rearrangement of chorismate,
important differences are worth noting. The active sites of
chorismate mutases are sufficiently recessed within the protein
structure to allow these enzymes to extract substrates from an
aqueous environment and engage them in a large number of non-
covalent interactions in order achieve high levels of catalytic
’ CONCLUSIONS
The phenylpyrrole-substituted guanidinium catalyst 2 induces
a 3.6 kcal/mol lowering of the activation free energy for the
5072
dx.doi.org/10.1021/ja110842s |J. Am. Chem. Soc. 2011, 133, 5062–5075

Journal of the American Chemical Society
ARTICLE
activity as well as exquisite substrate specificity: chorismate mutases
accelerate the rearrangement of chorismate over a million-fold;
however, modifications to the pendant carboxylate or alcohol
functionality of the substrate generally result in either a
significant or complete loss of activity.51 By contrast, the primary
catalytic functional group, the guanidinium ion, of 2 is largely
solvent-exposed, and in computationally optimized transition
structures, only the ester-substituted vinyl ether system and the
methylene group of the substrate are intimately associated with
the catalyst framework. As a consequence, these small-molecule
catalysts operate efficiently only in non-polar media where
desolvation energy is minimal and the strength of electrostatic
interactions is maximized. While limited contacts with the
substrate impose a constraint on rate acceleration, such catalysts
can accept a broader range of substrate structures.
In this study, experimentally validated computational models
for key enantioselectivity-determining steps have provided de-
tailed insight into the operative molecular recognition processes
in the guanidinium ion-catalyzed asymmetric Claisen rearrange-
ment.52Enantioselectivity was found to rely on multiple attractive
interactions to differentially stabilize a single transition structure.
Such cooperativity effects are emerging as a general principle in
small-molecule hydrogen-bond-donor catalysis.53
Chem. 1997, 62, 1449–1456. (d) Anderson, C. E.; Overman, L. E. J. Am.
Chem. Soc. 2003, 125, 12412–12413.
(4) Examples of enantioselective Claisen rearrangements with chiral
reagents: (a) Maruoka, K.; Banno, H.; Yamamoto, H. J. Am. Chem. Soc.
1990, 112, 7791–7793. (b) Maruoka, K.; Saito, S.; Yamamoto, H. J. Am.
Chem. Soc. 1995, 117, 1165–1166. (c) Corey, E. J.; Lee, D. H. J. Am.
Chem. Soc. 1991, 113, 4026–4028. (d) Ito, H.; Sato, A.; Taguchi, T.
Tetrahedron Lett. 1997, 38, 4815–4818. (e) Ito, H.; Sato, A.; Kobayashi,
T.; Taguchi, T. Chem. Commun. 1998, 2441–2442. (f) Yoon, T. P.;
MacMillan, D. W. C. J. Am. Chem. Soc. 2001, 123, 2911–2912.
(5) (a) Chook, Y.-M.; Ke, H.; Lipscomb, W. H. Proc. Natl. Acad. Sci.
U.S.A. 1993, 90, 8600–8603. (b) Chook, Y.-M.; Gray, J. V.; Lipscomb,
W. N. J. Mol. Biol. 1994, 240, 476–500. (c) Lee, A. Y.; Karplus, A.;
Ganem, B.; Clardy, J. J. Am. Chem. Soc. 1995, 117, 3627–3628.
(6) Bartlett, P. A.; Johnson, C. R. J. Am. Chem. Soc. 1985,
107, 7792–7793.
(7) Recent theoretical studies of the relative contribution of transi-
tion-state stabilization in the enzymatic rearrangement of chorismate
(see also references therein): (a) Ranaghan, K. E.; Ridder, L.; Szefczyk,
B.; Sokalski, W. A.; Hermann, J. C.; Mulholland, A. J. Org. Biomol. Chem.
2004, 2, 968–980. (b) Guimar~aes, C. R. W.; Repasky, M. P.; Chan-
drasekhar, J.; Tirado-Rives, J.; Jorgensen, W. L. J. Am. Chem. Soc. 2003,
ꢁ
125, 6892–6899. (c) Strajbl, M.; Shurki, A.; Kato, M.; Warshel, A. J. Am.
Chem. Soc. 2003, 125, 10228–10237. (d) Zhang, X.; Zhang, X.; Bruice,
T. C. Biochemistry 2005, 44, 10443–10448. (e) Khanjin, N. A.; Snyder,
J. P.; Menger, F. M. J. Am. Chem. Soc. 1999, 121, 11831–11846. (f) Guo,
H.; Cui, Q.; Lipscomb, W. N.; Karplus, M. Proc. Natl. Acad. Sci. U.S.A.
2001, 98, 9032–9037. (g) Guo, H.; Cui, Q.; Lipscomb, W. N.; Karplus,
M. Angew. Chem., Int. Ed. 2003, 42, 1508–1511. (h) Martí, S.; Andrꢀes, J.;
Moliner, V.; Silla, E.; Tu~nꢀon, I.; Bertrꢀan, J.; Field, M. J. J. Am. Chem. Soc.
2001, 123, 1709–1712. (i) Garcia-Viloca, M.; Gao, J.; Karplus, M.;
Truhlar, D. G. Science 2004, 303, 186–195.
(8) (a) Kast, P.; Asif-Ullah, M.; Jiang, N.; Hilvert, D. Proc. Natl. Acad.
Sci. U.S.A. 1996, 93, 5043–5048. (b) Cload, S. T.; Liu, D. R.; Pastor,
R. M.; Schultz, P. G. J. Am. Chem. Soc. 1996, 118, 1787–1788. (c) Liu,
D. R.; Cload, S. T.; Pastor, R. M.; Schultz, P. G. J. Am. Chem. Soc. 1996,
118, 1789–1790.
’ ASSOCIATED CONTENT
S
Supporting Information. Complete experimental pro-
b
cedures; characterization data including 1H and 13C NMR
spectra for catalysts, substrates, and products; chromatographic
traces for enantioselective rearrangements; data for kinetic and
binding studies; geometries and energies of all calculated sta-
tionary points; and complete ref 25. This material is available free
of charge via the Internet at http://pubs.acs.org.
(9) Kienh€ofer, A.; Kast, P.; Hilvert, D. J. Am. Chem. Soc. 2003,
125, 3206–3207.
’ AUTHOR INFORMATION
(10) Catalytic antibodies that accelerate the sigmatropic rearrange-
ment of chorismate on the order of 102-104-fold: (a) Jackson, D. Y.;
Jacobs, J. W.; Sugasawara, R.; Reich, S. H.; Bartlett, P. A.; Schultz, P. G.
J. Am. Chem. Soc. 1988, 110, 4841–4842. (b) Jackson, D. Y.; Liang,
M. N.; Bartlett, P. A.; Schultz, P. G. Angew. Chem., Int. Ed. 1992,
31, 182–183. (c) Hilvert, D.; Carpenter, S. H.; Nared, K. D.; Auditor,
M. T. Proc. Natl. Acad. Sci. U.S.A. 1988, 85, 4953–4955. (d) Hilvert, D.;
Nared, K. D. J. Am. Chem. Soc. 1998, 110, 5593–5594.
(11) (a) Wiest, O.; Houk, K. N. J. Org. Chem. 1994, 59, 7582–7584.
(b) Wiest, O.; Houk, K. N. J. Am. Chem. Soc. 1995, 117, 11628–11639.
(c) Mader, M. M.; Bartlett, P. A. Chem. Rev. 1997, 97, 1281–1301.
(12) Haynes, M. R.; Stura, E. A.; Hilvert, D.; Wilson, I. A. Science
1994, 646–652.
Corresponding Author
jacobsen@chemistry.harvard.edu
’ ACKNOWLEDGMENT
This work was supported by the NIH (GM 43214). We thank
Matthew Rienzo for experimental assistance and Dr. Shao-Liang
Zheng for X-ray analysis.
’ REFERENCES
(1) For reviews: (a) Ziegler, F. E. Chem. Rev. 1988, 88, 1423–1452.
(b) Castro, A. M. M. Chem. Rev. 2004, 104, 2939–3002.(c) Hiersemann,
M., Nubbemeyer, U., Eds.; The Claisen Rearrangement; Wiley-VCH:
Weinheim, Germany, 2007.
(13) (a) Uyeda, C.; Jacobsen, E. N. J. Am. Chem. 2008,
130, 9228–9229. (b) Uyeda, C.; R€otheli, A. R.; Jacobsen, E. N. Angew.
Chem., Int. Ed. 2010, 49, 9753–9756.
(14) Solvent effects in the Claisen rearrangement: (a) Coates, R. M.;
Rogers, B. D.; Hobbs, S. J.; Curran, D. P.; Peck, D. R. J. Am. Chem. Soc.
1987, 109, 1160–1170. (b) Brandes, E.; Grieco, P. A.; Gajewski, J. J.
J. Org. Chem. 1989, 54, 515–516. (c) White, W. N.; Wolfarth, E. F. J. Org.
Chem. 1970, 35, 2196–2199.
(15) Rate acceleration by ureas and thioureas: (a) Curran, D. P.;
Lung, H. K. Tetrahedron Lett. 1995, 36, 6647–6650. (b) Kirsten, M.;
Rehbein, J.; Hiersemann, M.; Strassner, T. J. Org. Chem. 2007,
72, 4001–4011.
(16) (a) Dougherty, D. A. Science 1996, 271, 163–168. (b) Zachar-
ias, N.; Dougherty, D. A. Trends Pharmacol. Sci. 2002, 23, 281–287. (c)
Thoma, R.; Schulz-Gasch, T.; D’Arcy, B.; Benz, J.; Aebi, J.; Dehmlow,
H.; Hennig, M.; Stihle, M.; Ruf, A. Nature 2004, 432, 118–122. (d)
(2) Lewis acid-catalyzed enantioselective rearrangements of ester-
substituted allyl vinyl ethers: (a) Abraham, L.; Czerwonka, R.; Hierse-
mann, M. Angew. Chem., Int. Ed. 2001, 40, 4700–4703. (b) Abraham, L.;
K€orner, M.; Hiersemann, M. Tetrahedron Lett. 2004, 45, 3647–3650.
(c) Linton, E. C.; Kozlowski, M. C. J. Am. Chem. Soc. 2008, 130,
16162–16163. (d) Mariꢀe, J.-C.; Xiong, Y.; Min, G. K.; Yeager, A. R.;
Taniguchi, T.; Berova, N.; Schaus, S. E.; Porco, J. A., Jr. J. Org. Chem.
2010, 75, 4584–4590.
(3) Catalytic enantioselective Claisen and related rearrangements
proposed to proceed by olefin activation mechanisms: (a) Akiyama, K.;
Mikami, K. Tetrahedron Lett. 2004, 45, 7217–7220. (b) Geherty, M. E.;
Dura, R. D.; Nelson, S. G. J. Am. Chem. Soc. 2010, 132, 11875–11877. (c)
Calter, M.; Hollis, T. K.; Overman, L. E.; Ziller, J.; Zipp, G. G. J. Org.
5073
dx.doi.org/10.1021/ja110842s |J. Am. Chem. Soc. 2011, 133, 5062–5075

Journal of the American Chemical Society
ARTICLE
Hoshino, T.; Sato, T. Chem. Commun. 2002, 291–301. (e) Wendt, W. K.;
Schultz, G. E.; Corey, E. J.; Liu, D. R. Angew. Chem., Int. Ed. 2000,
39, 2812–2833.
(17) A more comprehensive account of catalyst optimization studies
is included in the Supporting Information.
(35) Examples of complexation-induced downfield shifts of guani-
dinium ion protons in the presence of anionic guests: (a) Echavarren,
A.; Galan, A.; Lehn, J.-M.; de Mendoza, J. J. Am. Chem. Soc. 1989,
111, 4994–4995. (b) Dixon, R. P.; Geib, S. J.; Hamilton, A. D. J. Am.
Chem. Soc. 1992, 114, 365–366.
(18) A review of non-linear effects in asymmetric synthesis:Kagan,
H. B. Adv. Synth. Catal. 2001, 343, 227–233.
(19) Plots of product 6 ee and conversion as a function of catalyst (R,
R)-2 ee for rearrangements conducted in both hexanes and CDCl3 are
included in the Supporting Information.
(20) A review of reaction progress kinetic analysis conducted under
synthetically relevant conditions:Blackmond, D. G. Angew. Chem., Int.
Ed. 2005, 44, 4302–4320.
(21) Examples of kinetic investigations of asymmetric reactions
catalyzed by organic small molecules: (a) Zuend, S. J.; Jacobsen, E. N.
J. Am. Chem. Soc. 2007, 129, 15872–15883. (b) Zuend, S. J.; Jacobsen,
E. N. J. Am. Chem. Soc. 2009, 131, 15358–15374. (c) Zotova, N.;
Franzke, A.; Armstrong, A.; Blackmond, D. G. J. Am. Chem. Soc. 2007,
129, 15100–15101.
(36) Examples of similar complexation-induced upfield shifts for
aliphatic hydrogen atoms in proximity to arene π-faces in host-guest
complexes: (a) Kobayashi, K.; Asakawa, Y.; Kikuchi, Y.; Toi, H.; Aoyama,
Y. J. Am. Chem. Soc. 1993, 115, 2648–2654. (b) Ferrand, Y.; Crump, M. P.;
Davis, A. P. Science 2007, 218, 619–622. (c) Biros, S. M.; Bergman, R. G.;
Raymond, K. N. J. Am. Chem. Soc. 2007, 129, 12094–12095.
(37) Recent reviews (see also references therein): (a) Fielding, L.
Tetrahedron 2000, 56, 6151–6170. (b) Hirose, K. J. Inclusion Phenom.
Macrocyclic Chem. 2001, 39, 193–209.
(38) (a) Job, P. Ann. Chim. 1928, 9, 113–203. (b) MacCarthy, P.
Anal. Chem. 1978, 14, 2165.
(39) A plot of Δδ ꢀ mole fraction of 5 vs mole fraction of 5 reached
a maximum at 0.5 at a constant total concentration ([5] þ [(R,R)-2]) of
0.1 M. See Supporting Information for details.
(40) An average distance of 2.91 Å was measured for intermolecular
polar C-H to π interactions in a database study of organic crystal
structures:Umezawa, Y.; Tsuboyama, S.; Takahashi, H.; Uzawa, J.;
Nishio, M. Tetrahedron 1999, 55, 10047–10056.
(22) Detailed procedures for data acquisition and processing are
included in the Supporting Information. 1H NMR signal integration was
used to generate [5] data as a function of time. Rate vs concentration
plots were obtained from fits of the concentration vs time data to high-
order polynomial functions using the least-squares method, followed by
differentiation of that function with respect to time. These data were
parsed at 5% conversion intervals for the purposes of plotting and
derivation of the rate law.
(23) Rate constants and activation parameters for the rearrange-
ments of various ester-substituted allyl vinyl ethers: Rehbein, J.; Leick, S.;
Hiersemann, M. J. Org. Chem. 2009, 74, 1531–1540.
(24) Huang, Y.; Rawal, V. H. J. Am. Chem. Soc. 2002, 124, 9662–9663.
(25) Frisch, M. J.; et al. Gaussian 03, Revision E.01; Gaussian, Inc.:
Wallingford, CT, 2004.
(41) ΔΔE between the ground-state pro-(S,S) and pro-(R,R) con-
formations of the (R,R)-3 5 complex was calculated to be 2.57 kcal/mol.
3
(42) From crystallographic measurements, the van der Waals radius
of a hydrogen atom is estimated to be 1.09-1.20 Å: (a) Bondi, A. J. Phys.
Chem. 1964, 68, 441–451. (b) Rowland, R. S.; Taylor, R. J. Phys. Chem.
1996, 100, 7384–7391.
(43) Effect of arene substituents on the strength of cation-π
interactions: (a) Mecozzi, S.; West, A. P., Jr.; Dougherty, D. A. J. Am.
Chem. Soc. 1996, 118, 2307–2308. (b) Ma, J. C.; Dougherty, D. A. Chem.
Rev. 1997, 97, 1303–1324.
(26) (a) Meyer, M. P.; DelMonte, A. J.; Singleton, D. A. J. Am. Chem.
Soc. 1999, 121, 10865–10874. (b) Aviyente, V.; Yoo, H. Y.; Houk, K. N.
J. Org. Chem. 1997, 62, 6121–6128. (c) Guner, V.; Khuong, K. S.; Leach,
A. G.; Lee, P. S.; Bartberger, M. D.; Houk, K. N. J. Phys. Chem. A 2003,
107, 11445–11459.
(27) For computational studies of the thermal, aqueous, and thiour-
ea-catalyzed rearrangements of ester- and carboxylate-substituted allyl
vinyl ethers, see refs 10b, 15b, and the following: (a) Rehbein, J.;
Hiersemann, M. J. Org. Chem. 2009, 74, 4336–4342. (b) Cramer, C. J.;
Truhlar, D. G. J. Am. Chem. Soc. 1992, 114, 8794–8799.
(28) For consistency, s-trans is used here to describe an anti-
periplanar relationship between the vinyl ether and ester carbonyl
oxygen atoms about the intervening single bond, and s-cis describes a
syn-periplanar geometry.
(29) Zero-point vibrational energy corrections and free energies for
all stationary points are included in the Supporting Information.
(30) The energetic preference for R-ketoesters to adopt a dipole-
minimized, s-trans conformation, which precludes chelation of the
two carbonyl oxygen atoms by a catalyst, provides a possible rationale
for the general observation of catalytic turnover in both Lewis and
Brønsted acid-mediated Claisen rearrangements of O-allyl R-keto-
ester substrates.
(31) In ref 10b, an analogous shortening of an amidinium ion proton
to ether oxygen distance was observed in the transition state for the
Claisen rearrangement of a carboxylate-substituted substrate.
(32) Glendening, E. D.; Reed, A. E.; Carpenter, J. E.; Weinhold, F.
NBO, Version 3.1; Gaussian Inc.: Pittsburgh, PA, 1990.
(33) Breneman, C. M.; Wiberg, K. B. J. Comput. Chem. 1990,
11, 361–373.
(44) Examples of the use of unnatural amino acids to resolve specific
cation-π interactions in protein-ligand complexes: (a) Zhong, W.;
Gallivan, J. P.; Yinong, Z.; Li, L.; Lester, H. A.; Dougherty, D. A. Proc.
Natl. Acad. Sci. U.S.A. 1998, 95, 12088–12093. (b) Xiu, X.; Puskar, N. L.;
Shanata, J. A. P.; Lester, H. A.; Dougherty, D. A. Nature 2009,
458, 534–537.
(45) Studies of cation-π interactions in asymmetric catalysis: (a)
Cannizzaro, C. E.; Houk, K. N. J. Am. Chem. Soc. 2002, 124, 7163–7169.
(b) Wei, Y.; Held, I.; Zipse, H. Org. Biomol. Chem. 2006, 4, 4223–4230.
(c) Li, X.; Liu, P.; Houk, K. N.; Birman, V. B. J. Am. Chem. Soc. 2008,
130, 13836–13837. (d) Knowles, R. R.; Lin, S.; Jacobsen, E. N. J. Am.
Chem. Soc. 2010, 132, 5030–5032.
(46) Because of the large size of the systems being modeled and the
presence of multiple small vibrational frequencies, zero-point vibrational
energy corrections and free energies, derived from frequency calcula-
tions using the rigid-rotor/harmonic oscillator approximation, were not
used to estimate enantioselectivities.
(47) Selected examples of correlations between experimental and
calculated enantioselectivities in catalytic reactions, see ref 21b and the
following: (a) Schneebeli, S. T.; Hall, M. L.; Breslow, R.; Friesner, R.
J. Am. Chem. Soc. 2009, 131, 3965–3973. (b) Kozlowski, M. C.; Dixon,
S. L.; Panda, M.; Lauri, G. J. Am. Chem. Soc. 2003, 125, 6614–6615. (c)
Donoghue, P. J.; Helquist, P.; Norrby, P.-O.; Wiest, O. J. Am. Chem. Soc.
2009, 131, 410–411.
(48) For comparisons of functionals for systems involving weak
noncovalent interactions, see refs 21a ,21b, 45c, and the following: (a)
Zhao, Y.; Truhlar, D. G. J. Chem. Theory Comput. 2007, 3, 289–300. (b)
Zhao, Y.; Truhlar, D. G. Acc. Chem. Res. 2008, 41, 157–167.
(49) Examples of C-F M interactions between fluoroarenes and
3 3 3
(34) For the binding of ammonium cation to pyrrole, an interaction
energy of 22.05 kcal/mol and a nearest distance of 2.1 Å from a proton of
ammonium to a carbon atom of the pyrrole were calculated at the
B3LYP/6-31G(d,p) level of theory:Zhu, W.-L.; Tan, X.-J.; Puah, C. M.;
Gu, J.-D.; Jiang, H.-L.; Chen, K.-X.; Felder, C. E.; Silman, I.; Sussman,
J. L. J. Phys. Chem. A 2000, 104, 9573–9580.
metal cation interactions in crystal structures: (a) Plenio, H. Chem. Rev.
1997, 97, 3363–3384. (b) Kulawiec, R. J.; Crabtree, R. H. Coord. Chem.
Rev. 1990, 99, 89–115. (c) Bouwkamp, M. W.; Budzelaar, P. H. M.;
Gercama, J.; Morales, I. D. H. M.; de Wolf, J.; Meetsma, A.; Troyanov,
S. I.; Teuben, J. H.; Hessen, B. J. Am. Chem. Soc. 2005, 127,
14310–14319.
5074
dx.doi.org/10.1021/ja110842s |J. Am. Chem. Soc. 2011, 133, 5062–5075

Journal of the American Chemical Society
ARTICLE
(50) Examples of C-F H-C contacts of 2.2-2.8 Å in fluoroar-
3 3 3
ene-arene crystal structures: (a) Shimoni, L.; Glusker, J. P. Struct.
Chem. 1994, 5, 383–397. (b) Ngola, S. M.; Dougherty, D. A. J. Org.
Chem. 1998, 63, 4566–4567. (c) Smith, C. E.; Smith, P. S.; Thomas,
R. L.; Robins, E. G.; Collings, J. C.; Dai, C.; Scott, A. J.; Borwick, S.;
Batsanov, A. S.; Watt, S. W.; Clark, S. J.; Viney, C.; Howard, J. A. K.;
Clegg, W.; Marder, T. B. J. Mater. Chem. 2004, 14, 413–420.
(51) (a) Pawlak, J.; Padykula, R. E.; Kronis, J. D.; Aleksejczyk, R. A.;
Berchtold, G. A. J. Am. Chem. Soc. 1989, 111, 3374–3381. (b) Galopin,
C. C.; Ganem, B. Bioorg. Med. Chem. Lett. 1997, 7, 2885–2886.
(52) For a broader discussion of computational approaches to the
elucidation of asymmetric catalytic mechanisms, see:Houk, K. N.;
Cheong, P. H.-Y. Nature 2008, 455, 309–313.
(53) (a) Taylor, M. S.; Jacobsen, E. N. Angew. Chem., Int. Ed. 2006,
45, 1520–1543. (b) Knowles, R. R.; Jacobsen, E. N. Proc. Natl. Acad. Sci.
U.S.A. 2010, 107, 20678–20685.
5075
dx.doi.org/10.1021/ja110842s |J. Am. Chem. Soc. 2011, 133, 5062–5075