Key words
DFT calculation - heterocycles - piperidines - epoxidation - stereoselective synthesis
Nitrogen heterocycles are prevalent motifs in natural products and pharmaceutical
agents that constitute important synthetic targets for drug discovery.[1]
[2] Efficient methods of accessing functionalized nitrogen heterocycles with high levels
of stereocontrol are therefore highly sought after.[3] The ability of hydroxyl groups to direct the stereoselectivity of alkene epoxidations
through hydrogen-bonding interactions has been well-documented and investigated.[4]
[5] The role of torsional effects (‘torsional steering’) has also been established for
epoxidations and related oxidations.[6] Although examples of stereoselective epoxidations have been reported that are hypothesized
to be directed by nitrogen-containing groups,[7] no computational studies have aimed to elucidate the origins of these selectivities.[8]
In an effort to access multiple functionalized piperidines, the Ellman group[9] sought to exploit the hydrogen-bonding ability of nitrogen-containing groups to
enforce the diastereoselective epoxidation of tetrahydropyridines 3 (Scheme [1, a]). The experimental protocol involves the treatment of tetrafluorophthalic anhydride
1 with hydrogen peroxide to yield bifunctional epoxidation agent 2. In addition to the oxidizing peracid functionality, 2 also possesses a carboxylic acid group capable of forming strong hydrogen bonds with
the basic amine nitrogen in 3 either before or after proton transfer. It was hypothesized that the tetrafluoroarene
portion serves as a rigid covalent tether, forcing the epoxidation event to occur
at the hydrogen-bonded π-face of the molecule. Interestingly, for tetrahydropyridines
3, contrasteric delivery of the epoxide oxygen to the more hindered π-face is observed
(Scheme [1, b]). Epoxidation agents without the tether feature, such as m-CPBA and trifluoroperacetic acid, resulted in low diastereoselectivities (Scheme
[1, a]).
Scheme 1 (a) Diastereoselective epoxidations of tetrahydropyridines 3 by the in situ generated bifunctional epoxidation agent 2-carboperoxy-3,4,5,6-tetrafluorobenzoic
acid (2). (b) Design principle of the bifunctional epoxidation agent 2.
Given this remarkable ability of this transformation to override steric bias, we undertook
a computational investigation into the origins of diastereoselectivity for the epoxidation
of tetrahydropyridines shown in Scheme [1] (a).
Computations were performed with Gaussian 09.[10] Molecular geometries were optimized using the M06-2X[11] functional and the 6-31+G(d) basis set. The effect of solvation on molecular geometries
was accounted for during optimizations using the SMD[12] solvation model with tetrahydrofuran as the solvent. Frequency calculations were
carried out at the same level of theory as that used for geometry optimization to
characterize the stationary points as either minima (no imaginary frequencies) or
saddle points (one imaginary frequency) on the potential energy surface and to obtain
thermal corrections to the Gibbs free energies. Intrinsic reaction coordinate (IRC)
calculations were performed to ensure that the saddle points found were true transition
states connecting the reactants and the products. Single-point energies were calculated
with the M06-2X functional and the 6-311++G(3df,2pd) basis set, and solvation effects
were modeled using the SMD solvation model with tetrahydrofuran as the solvent. For
fragment distortion analysis, single-point energies of the molecular fragments were
computed at the M06-2X/6-311++G(3df,2pd) level using the M06-2X/6-31+G(d) geometries
obtained in solution. Molecular structures were visualized using CYLview.[13] Monte Carlo conformational searches were performed with the Merck molecular force
field (MMFF) implemented in Spartan ’16 to ensure that the lowest energy conformations
of intermediates and transition states are presented in the manuscript. Geometries
and energies of alternative conformers explored in this study can be found in the
Supporting Information.
We first calculated the hydrogen-bonded and nonhydrogen-bonded epoxidation transition
states for the epoxidation of unsubstituted 1,2,5,6-tetrahydropyridine 3a by 2 (Scheme [2, a]). A free energy preference of 3.8 kcal/mol was found in favor of the hydrogen-bonded
transition state TS-1, in which the O–H···N distance is 1.73 Å (Scheme [2, b]). The strong preference for a hydrogen-bonded TS is particularly striking considering
that the hydrogen bond has the effect of deactivating the alkene double bond toward
electrophilic epoxidation by making the alkene moiety less nucleophilic.
Scheme 2 (a) Epoxidation of 3a by bifunctional agent 2. (b) Calculated transition states for the epoxidation of 3a by 2. Energies are denoted in kcal/mol, and interatomic distances are shown in Ångströms.
(c) Side-on view of TS-1 and TS-2 showing alignment of the peracid and alkene moieties.
Examination of TS-1 and TS-2 reveals several important geometrical consequences of the hydrogen bonding. In the
absence of a hydrogen bond, the peracid and the alkene in TS-2 assumes the expected spiro geometry (Scheme [2, c]), which maximizes the stabilizing secondary orbital interaction between the oxygen
lone pair and the alkene π* orbital.[14] This interaction is weaker in TS-1, presumably because a spiro geometry cannot be achieved due to geometrical constraints
imposed by the hydrogen bond. The same geometrical constraints also force the peracid
moiety to adopt a nonplanar arrangement. This nonplanarity weakens the hydrogen bond
between O1 and the peracid proton (2.12 Å, compared to 1.84 Å in TS-2). In TS-2, the formation of the two C–O bonds is only slightly asynchronous at 2.16 and 2.22
Å, respectively, whereas TS-1 is much more unsymmetrical (2.20 and 2.02 Å).
Having established the favorability of the hydrogen-bonded TS, we next investigated
the epoxidation of hexasubstituted tetrahydropyridine scaffolds, represented by tetrahydropyridine
1b. These tetrahydropyridines are observed to undergo epoxidation at the more sterically
encumbered π-face.
Computed transition states for the epoxidation of 3b by 2 are shown in Scheme [3]. For each product diastereomer, a hydrogen-bonded TS and a nonhydrogen-bonded TS
can be envisioned. For TS-4 and TS-6, the two transition states without OH···N hydrogen bonding, calculated free energies
of activation indicate that the reaction should be 2.0 kcal/mol in favor of epoxidation
at the less sterically hindered face. With OH···N hydrogen bonding, however, the preference
is 7.1 kcal/mol in favor of contrasteric epoxidation (TS-3 and TS-5). Geometrically, the undirected transition states TS-4 and TS-6 are spiro and highly symmetrical, similar to TS-2. The hydrogen-bonded TS-3 and TS-5 are both distorted away from a spiro geometry, with highly asynchronous formation
of the two C–O bonds. Overall, the hydrogen-bonded TS-3 is 1.7 kcal/mol lower in energy than TS-6, which explains the experimentally observed selectivity for diastereomer 4b-A. This result confirms that the OH···N hydrogen bond is strong enough to override
the effects of steric congestion. Considering that the amine moiety is likely to be
partially protonated under reaction conditions, which would deactivate the C=C bond
toward electrophilic epoxidation, undirected epoxidation is likely to have barriers
that are higher than those calculated for TS-4 and TS-6, as these transition states do not take the effect of amine protonation into account.
Scheme 3 Four calculated transition states for the epoxidation of 3b by 2. Energies are denoted in kcal/mol, and interatomic distances are shown in Ångströms.
Hydrogen atoms attached to carbons are omitted for clarity.
To further explain the severe energetic unfavorability of TS-5, we calculated the distortion energies (Edist
‡) of the tetrahydropyridine fragment in epoxidation transition states TS 3–6 (Figure [1]). These Edist
‡ values are obtained by comparing gas-phase electronic energies of transition-state
geometries of the tetrahydropyridine to that of the corresponding ground-state geometry.[15] The tetrahydropyridine fragment in TS-5 has a high Edist
‡ value of 13.6 kcal/mol, which is 8.8 kcal/mol higher than the corresponding Edist
‡ value in TS-3. This difference in distortion energy is comparable in magnitude to the ΔΔG‡ value of 7.1 kcal/mol between TS-3 and TS-5, indicating that the free energy distinction can be mostly attributed to the distortion
of the tetrahydropyridine substrate. Geometrically, the tetrahydropyridine fragment
in TS-5 is deformed into a twist-boat-like structure to allow the nitrogen lone pair to occupy
the pseudo-axial position needed for hydrogen bonding. Unfavorable near-eclipsing
interactions in the twist-boat-like structure (Figure [1]) result in a high Edist
‡ value.
Figure 1 Geometries of the tetrahydropyridine fragment in epoxidation transition states TS-3–6, and distortion energies compared to the ground-state reactant. Newman projection
insets show near-eclipsing interactions between ring substituents. Energies are denoted
in kcal/mol. Hydrogen atoms are omitted for clarity.
In conclusion, we used DFT calculations to reveal the origins of the contrasteric
π-facial stereoselectivity in the epoxidation of densely substituted tetrahydropyridines
by a bifunctional peracid reagent 2. We show that epoxidation of 1,2,3,6-tetrahydropyridines with 2 have a strong preference to proceed through OH···N hydrogen-bonded transition states.
Even for hexasubstituted tetrahydropyridines that are highly congested at one π-face,
the OH···N hydrogen-bonding interaction is strong enough to overcome the steric disadvantage
and deliver the contrasteric epoxide product.
