Journal of the American Chemical Society
Article
substitution of one diastereomer over another. Two lowest-
energy diastereomeric transition states calculated for the
substitution step are presented in Figure 5A. Aside from the
stacking interaction existing in both diastereomeric transition
states, the diastereomeric transition state whose leaving
hydroquinone is located on the top of the imine plane benefits
from a 1.58 Å hydrogen bonding with the morpholine moiety
(TS-major), while the other has a much weaker C−H···N
interaction of 2.21 Å with the incoming indole placed at the
top (TS-minor). This critical interaction again plays the
decisive role by introducing the chirality of the aminocatalyst
to the substitution process, leading to satisfying enantiose-
lectivity. Accordingly, a 1.5 kcal/mol difference in enthalpy is
obtained between the two diastereomeric transition states.
Figure 5A shows both the ΔΔH⧧ and ΔΔG⧧ values, since we
found that the E and H at various levels always give a
preference for TS-major, while the ΔG values are near zero and
favor the major or the minor TS at different computational
levels. The kinetic preference does agree with the observed
enantioselectivity; however, complete consumption of starting
material and yields greater than 50% from a racemate rule out a
kinetic resolution regime.8,19,26,27 These key attributes imply a
stereomutative transformation prior to indole substitution.
More specifically, enantiomeric excess as a function of time
appears marginally insensitive to conversion dependence and
most closely resembles a DYKAT (dynamic kinetic asymmetric
transformation) type 1 scenario, as shown in Figure 5B.26
One baffling aspect in this process is that since the indole
substitution step is rate-determining and highly exergonic, the
stereomutative transformation could not be feasible unless it is
much faster and highly reversible, requiring a much better
nucleophile than indole in the reaction system. We investigated
the possibility of epimerization of the stereocenter driven by
the self-substitution of hydroquinone.8 DFT calculations were
performed to test this hypothesis (Figure 6A). Indeed, an R-
zwitterionic intermediate is found to be in a fast equilibrium
with its S-diastereomer. The indistinguishable energies of these
diastereomers predict that they will be a 1:1 mixture at
equilibrium; one of these can then undergo the indole
substitution at a faster rate, resulting in the major observed
enantiomer. A transition state of self-substitution is located
with a 20.3 kcal/mol energy barrier and should then be faster
than substitution with indole (∼30 kcal/mol, Figure 3). In this
interconversion process, the morpholine moiety on the
aminocatalyst subtly directs the incoming hydroquinone: the
basic tertiary amine deprotonates the hydroquinone and forms
a hydrogen bond during the self-substitution. At the same time,
the negative charge is prone to concentrate on the attacking
oxygen atom of hydroquinone. The hydroquinone hence
exhibits an anionic property in the transition state, making it a
better nucleophile relative to indole. The unexpected SN2-DKR
nicely explains the behavior of enantiomeric excess and
conversion rate, though violating the common intuition that
Walden inversion occurs on a definitive stereocenter during SN2. In
fact, this process is an exemplar model of the Curtin−Hammett
principle.
racemic mixture of product at low catalyst concentrations is
consistent with an unselective oxidation and a diminished rate
of interconversion prior to indole substitution.
Additional studies were performed to confirm the vital role
of the tertiary amine moiety. The protonation of the
morpholine moiety on the aminocatalyst plays multiple roles
throughout the substitution: that is, it promotes the hydro-
quinone-mediated stereomutative interconversion, facilitating
indole substitution by stabilizing the transition state and
effectuating the stereoinduction from the chiral aminocatalyst.
To further expose the function of this molecular fragment, we
have conducted an intrinsic reaction coordinate (IRC)
calculation for the substitution process (Figure 6C). Bond
lengths of the cleaving C−O bond and the critical hydrogen
bond are recorded at each point along the IRC pathway.
Starting with the plots that represent the primary structures of
the reaction at the left top, the hydrogen bond length changes
from ∼2.15 Å to 2.00 Å as the C−O bond cleaves, suggesting
that this hydrogen bonding initiates the “abstraction” of the
C−O bond by strengthening the attractive force of O···H
interaction. Subsequently, a remarkable linear correlation
appears as the C−O bond is cleaving, ranging from 1.6 to
3.1 Å, clearly corresponding to the stabilization effect of the
hydrogen bond. The transition state is located at the late stage
of this region. After the leaving group has been completely
substituted by the nucleophile, the hydrogen bonding is
fortified to a higher degree, neutralizing the negative charge on
the leaving group.
From the IRC in Figure 6C, it was found that the N−H of
the protonated morpholino moiety in the catalyst clearly plays
an important role. To experimentally investigate this, a catalyst
without the amine moiety was prepared. The amine-bearing
morpholine group in catalyst 3a was substituted for the isostere
cyclohexyl for catalyst 3b (eq 4 and eq 5). As a result,
differences in reactivity can be ascribed to the lack of H-
bonding interactions. Remarkably, a racemic mixture of
products was obtained when this isostere was used, despite
the aminocatalyst bearing an identical stereocenter.41 This
emphasizes the pivotal role of the H-bonding interactions
deduced in the IRC to achieve high enantioselectivity.
In placing these empirical studies in the greater context of
the density functional theory (DFT) calculations, these results
affirm a Curtin−Hammett regime initiated by the tertiary
amine moiety of the catalyst. Notably, the dynamic self-
substitution pathway describing the racemization of the DDQ
adduct is reminiscent of the Walden cycle.1−4,6 In analogy,
both enantiomers of the reactive DDQ adduct are accessible by
consecutive SN2 reactions as depicted in Figure 7A. An
enantiomer of the DDQ adduct is condensed with the chiral
Trifunctional Role of the Aminocatalyst. To indirectly
probe the occurrence of the stereomutative transformation
prior to the rate-determining indole substitution, the
enantiomeric excess was evaluated as a function of amino-
catalyst loading. As shown in Figure 6B, the observed
correlation affirms aminocatalyst involvement in the preceding
interconversion step of the reaction mechanism. A near
7517
J. Am. Chem. Soc. 2021, 143, 7509−7520
Rezayee, Nomaan M.
