Lin
Zhang
a,
Ken
Yamazaki
a,
Jamie A.
Leitch
a,
Ruben
Manzano
a,
Victoria A. M.
Atkinson
a,
Trevor A.
Hamlin
*b and
Darren J.
Dixon
*a
aDepartment of Chemistry, Chemistry Research Laboratory, University of Oxford, 12 Mansfield Road, Oxford, UK
bDepartment of Theoretical Chemistry, Amsterdam Institute of Molecular and Life Sciences (AIMMS), Amsterdam Center for Multiscale Modeling (ACMM), Vrije Universiteit Amsterdam, De Boelelaan 1083, 1081 HV Amsterdam, The Netherlands. E-mail: t.a.hamlin@vu.nl
First published on 24th June 2020
The construction of enantioenriched azabicyclo[3.3.1]nonan-6-one heterocycles via an enantioselective desymmetrization of allene-linked cyclohexanones, enabled through a dual catalytic system, that provides synchronous activation of the cyclohexanone with a chiral prolinamide and the allene with a copper(I) co-catalyst to deliver the stereodefined bicyclic core, is described. Successful application to oxygen analogues was also achieved, thereby providing a new enantioselective synthetic entry to architecturally complex bicyclic ethereal frameworks. The mechanistic pathway and the origin of enantio- and diastereoselectivities has been uncovered using density functional theory (DFT) calculations.
![]() | ||
Scheme 1 The design of an allene-linked cyclohexanone desymmetrization platform. En = enamine catalysis. |
The enantioselective desymmetrization of structurally simple molecules is widely considered as one of the most powerful and elegant strategies to create new stereocenters.5 In this context, the desymmetrization of carbonyl compounds through intramolecular cyclization manifolds has proven to be an excellent access point to bridged and fused bicyclic rings. Through elegant contributions from the Jia,6 Lu,7 and Zhou8 groups amongst many others – including our own9 – significant progress has been made in this field. In 2017, our group demonstrated that the silver and chiral amine co-catalyzed intramolecular desymmetrization/cyclization of alkyne-linked cyclohexanones afforded the morphan core with high enantioselectivity.10 Recently, Dong and co-workers described an intramolecular Pd-catalyzed α-allylic alkylation of alkyne-linked cyclohexanone derivatives, notably with the ability to access both exo and endo-diastereomers of the [3.2.1] morphan analogue, albeit with insignificant enantioinduction11 (Scheme 1, middle). For downstream application in complex molecule total synthesis, we were drawn towards developing a new desymmetrization methodology for the construction of the 4-vinyl-morphan and their analogous oxygenated architectures. This would require the simultaneous stereocontrolled creation of three new stereogenic centres in a bicyclic core possessing multiple key groups and functionalities.
To date, only a handful of examples of enantioselective intramolecular cyclization of allene-linked carbonyl compounds have been reported.12 Inspired by this prior art, and building on our own findings, we reasoned that a suitable dual catalyst system could be identified to furnish such target morphan structures. We envisioned that an intramolecular cyclization of allene-tethered cyclohexanones could take place via enamine catalysis and allene activation by an appropriate metal catalyst (Scheme 1, bottom), and herein we wish to report our findings.
Entry | [Cu] cat. | Prolinamide | 2a (%) | drc | erd |
---|---|---|---|---|---|
a General conditions: 1a (0.2 mmol), [Cu] catalyst (0.02 mmol, 10 mol%), 4-bromobenzoic acid (0.1 mmol, 50 mol%), prolinamide catalyst (0.06 mmol, 30 mol%), in CPME (0.1 M, 2 mL) at 120 °C, under an argon atmosphere for 48 h. b Isolated yield. c dr calculated via1H NMR analysis of the crude reaction mixture. d er value was determined via chiral HPLC analysis of the pure product. e Reaction concentration [0.02 M]. f P6 (30 mol%) was used as prolinamide catalyst. | |||||
1 | Cu(OTf)2 | P1 | 76 | 8![]() ![]() |
82![]() ![]() |
2 | CuI | P1 | 44 | >20![]() ![]() |
72![]() ![]() |
3 | Cu(OAc)2 | P1 | 40 | 7![]() ![]() |
70![]() ![]() |
4 | Cu(MeCN)4PF6 | P1 | 98 | 12![]() ![]() |
87![]() ![]() |
5 | Cu(MeCN)4BF4 | P1 | 76 | 11![]() ![]() |
78![]() ![]() |
6 | Cu(acac)2 | P1 | 93 | 10![]() ![]() |
80![]() ![]() |
7 | Cu(MeCN)4PF6 | P2 | 99 | >20![]() ![]() |
82![]() ![]() |
8 | Cu(MeCN)4PF6 | P3 | 90 | 18![]() ![]() |
80![]() ![]() |
9 | Cu(MeCN)4PF6 | P4 | 99 | >20![]() ![]() |
89.5![]() ![]() |
10 | Cu(MeCN)4PF6 | P5 | 90 | 18![]() ![]() |
89![]() ![]() |
![]() |
|||||
Change from entry 9 | |||||
11 | Reaction run at 100 °C | — | nd | nd | |
12 | Reaction concentration [0.04 M] | 91 | >20![]() ![]() |
90![]() ![]() |
|
13 | Reaction concentration [0.02 M] | 93 | >20![]() ![]() |
91![]() ![]() |
|
14e | P6 used as organocatalyst | 82 | >20![]() ![]() |
92.5![]() ![]() |
|
15 , | TFA used instead of 4-BrPhCO 2 H | 81 |
>20![]() ![]() |
96![]() ![]() |
With the optimal conditions established, the scope of N-tethered substrates was explored (Scheme 2). Aryl sulfonamide-tethered allenic cyclohexanones with either electron-donating or electron-withdrawing groups at the para-position of the arene delivered morphan products in good yields with high enantioselectivity and diastereoselectivity (2b–2f, 94:
6 to 96
:
4 er). Phenyl substituted compound 1g afforded 2g in 78% yield and 94
:
6 er. Importantly the absolute configuration of 2g was confirmed as 1S,4R,5S by single-crystal X-ray diffraction analysis.14Ortho-methylated and multiply-substituted arene substrates were also well-tolerated, albeit with a slight drop in diastereoselectivity (2h and 2i). Naphthyl derivative 1j reacted smoothly to afford 2j in 86% yield with 96
:
4 er. Substrates bearing heteroaromatic 3-pyridinyl (1k) and 2-thiophenyl (1l) substituents led to products 2k and 2l with great efficiency. Pleasingly, N-mesyl protected (1m) amines afforded the morphan bicycle with high enantioselectivity (94
:
6 er). Unfortunately, no product was observed when phenyl-substituted derivative (1o) was subjected to the optimized conditions, most likely due to excessive steric hindrance preventing the allene access to the enamine intermediate of the ketone moiety. Notably, the internal allene structure 1p also reacted smoothly under the optimized conditions, providing 2p in 89% yield with 91
:
9 er.
Following the success of the newly discovered dual catalytic system for the desymmetrization of nitrogen-linked allenes, our attention turned to their oxygen-substituted counterparts to construct the analogous 2-oxabicyclo[3.3.1]nonane heterocycle (oxamorphan for simplicity, herein). To the best of our knowledge, no enantioselective syntheses of such constructs via desymmetrization approaches or other stereoselective methodologies have been reported to date.15 Accordingly, application of this dual catalytic protocol would find added value in the synthesis of complex oxygen-containing heterocyclic constructs.2 Despite our first attempted application of the dual catalytic methodology to the cyclization of an oxygen-linked substrate (3a) being unsuccessful, installation of substitution at the α-position to the oxygen atom recovered reactivity, delivering the phenyl-substituted oxamorphan derivative (4b) as a single diastereomer in excellent yield and good enantioselectivity (92:
8 er).16 All further examples of the oxamorphan core were also shown to proceed with exquisite diastereoselectivity (>20
:
1 dr). Substitution on the aromatic ring was shown to have little deleterious effect on the reactivity of substrates (4c–4f, 84–98% yields and 90
:
10 to 94
:
6 er). Use of a protected catechol-derived substrate, however, led to a decrease in both the yield and enantioselectivity (4g, 67% yield and 81
:
19 er). Pleasingly, structures with an alkyl group (ethyl, benzyl and cyclohexyl) functioned efficiently, affording 4i–4k in high yields with good enantioselectivity. Moreover, allyl substitution was well-tolerated in this desymmetrization methodology (4l).
Control and deletion studies were carried out in order to probe the mechanistic pathway of our dual-catalytic entry to enantioenriched vinyl-morphan structures. A possible mechanistic pathway could proceed via an alkyne intermediate. To test this hypothesis, alkyne derivative (5a) was synthesized and subjected to the reaction conditions. However, no cyclized product was observed, hence ruling out a pathway via such an intermediate (Scheme 3A). During our preliminary scouting studies, prolinamide catalysts without an N–H bond (e.g. a tertiary amide) were found to be inferior in comparison to their secondary amide counterparts. In order to study this further, a methylated derivative of our optimized catalyst was employed under standard reaction conditions, and complete suppression of reactivity was indeed observed (Scheme 3B). Our protocol was then conducted whilst omitting each of the co-catalysts sequentially. In the absence the copper catalyst, no formation of the cyclized product was observed (Scheme 3C), whilst without the prolinamide catalyst, the morphan structure was still observed, albeit in only 16% yield and with no diastereoselectivity (Scheme 3D). This latter result points to an alternative minor pathway to racemic product presumably via an enol tautomer, which is operating under the reaction conditions. Accordingly, such background reactivity without the prolinamide catalyst is likely responsible for a slight erosion of enantioselectivity observed experimentally vs. that predicted from our computational calculations (vide infra).
![]() | ||
Scheme 3 Control experiments: (A) use of an alkyne substrate. (B) Use of methylated catalyst. (C) No copper catalyst. (D) No prolinamide catalyst. |
In order to paint a full mechanistic picture, density functional theory analysis of the reaction pathway was performed. All the calculations reported herein were performed using the Amsterdam Density Functional (ADF) software.17 Equilibrium structures and transition structure geometries were optimized using the BLYP functional18,19 and the TZ2P basis set.20 This approach was extensively tested against ab initio reference benchmarks from hierarchical series up until CCSD(T).21 Solvent effects of Et2O (as CPME was not available) were accounted for using the conductor-like screen model (COSMO) of solvation.22 Dispersion interactions were included using Grimme's DFT-D3 correction with Becke–Johnson damping.23 The zeroth-order regular approximation (ZORA) was used to account for scalar relativistic effects.24 This level is referred to as COSMO(Et2O)-ZORA-BLYP-D3(BJ)/TZ2P. All stationary points have been verified, through vibrational analysis, to be minima (zero imaginary frequency) or transition states (one imaginary frequency). The character of the normal mode associated with the imaginary frequency has been analyzed to ensure it resembles the reaction under consideration. Optimized structures were illustrated using CYLview.25 Potential energies were refined by means of single point calculations using the M06 functional.26 This level is denoted COSMO(Et2O)-ZORA-M06/TZ2P//COSMO(Et2O)-ZORA-BLYP-D3(BJ)/TZ2P.
To understand the origin of enantio- and diastereocontrol of the key C–C bond forming step, we quantitatively analyzed the two competing transition structures using the activation strain model (ASM)27 of reactivity. ASM involves decomposing the electronic energy of the transition structure ΔE‡ into the strain ΔE‡strain associated with the structural deformation of the reactants from their equilibrium geometry and the interaction ΔE‡int between the deformed reactants (eqn (1)).27 The ΔE‡strain is determined by the rigidity of the reactants and by the extent to which they must deform to achieve the geometry of the transition structure. The ΔE‡int is usually stabilizing and is related to the electronic structure of the reactants and how they are mutually oriented over the course of the reaction.
ΔE‡ = ΔE‡strain + ΔE‡int | (1) |
Our DFT studies uncovered that the enantio- and diastereoselectivities are determined during the key nucleophilic attack of an enamine28 to a copper-coordinated allene13 in the intramolecular cyclization step.
Coordination of copper to the allene along with the interaction between the prolinamide and trifluoroacetate (ligated to the copper species) play important roles in the kinetically preferred transition structure associated with selective C–C bond formation. Enantioselectivity is determined by stabilization from the hydrogen bond interaction between the prolinamide N–H bond and the O atom of trifluoroacetate. This stabilizing interaction is only accessible when the copper-coordinated allene is positioned in close proximity to the amide. Diastereoselectivity is determined by the strain energy caused by the larger dihedral angle of the enamine (φ) and the smaller angle of the allene (θ, Scheme 4, inset). Optimal geometries of the enamine and the allene are planar and linear, respectively, and the destabilizing repulsion between the allene moiety and the pyrrolidine ring causes a greater distortion that results in a higher energy barrier.
Reaction of the cyclohexanone substrate and the prolinamide catalyst forms either s-cis enamine A1 or s-transA2, and the small energy difference indicates a rapid equilibrium between these conformations. We hypothesized that the bulky substituent of the amide prevents reactivity from the top-surface of the enamine in each of the structures. The enamines and a Cu(I) species generate metal complexes and this process is highly exergonic. Reactant complexes B1 and B2 benefit from stabilizing hydrogen bonds between the amide N–H bond and an O atom on trifluoroacetate that are absent in B3 and B4. The key C–C bond formation viaTSB1-C1 proceeds with stabilizing hydrogen bonding interactions and no destabilizing steric clash between the prolinamide and allene, and is kinetically preferred by 2.2 kcal mol−1 compared to the next most favorable TSB2-C2. An activation strain analysis was performed on TSB1-C1 and TSB2-C2 to quantitatively understand the factors leading to the difference in computed reactivity and the results are summarized in (Scheme 4, inset). TSB1-C1 goes with a lower energy barrier due to a less destabilizing activation strain (ΔE‡strain = 27.1 kcal mol−1) compared to TSB2-C2 (ΔE‡strain = 29.0 kcal mol−1) that results from a smaller distortion of the enamine and the allene moieties. This step is exergonic by 10.4 kcal mol−1 and has a large Gibbs free energy barrier for the reverse reaction (28.4 kcal mol−1) that is likely irreversible. After the intramolecular cyclization, trifluoroacetic acid coordinates to the alkene of the alkylcopper(I) species C1 to form a complex D1. The protonation process through TSD1-E1 generates a vinyl group, which then regenerates Cu(TFA) and hydrolyzes the enamine to give the cyclized product E1.
Footnote |
† Electronic supplementary information (ESI) available. CCDC 2000237. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/d0sc02878a |
This journal is © The Royal Society of Chemistry 2020 |