Daniel Rozsar1, Michele Formica1, Ken Yamazaki1,2, Trevor A Hamlin2, Darren J Dixon1. 1. Department of Chemistry, Chemistry Research Laboratory, University of Oxford, 12 Mansfield Road, OX1 3TA Oxford, United Kingdom. 2. Department 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.
Abstract
The first metal-free catalytic intermolecular enantioselective Michael addition to unactivated α,β-unsaturated amides is described. Consistently high enantiomeric excesses and yields were obtained over a wide range of alkyl thiol pronucleophiles and electrophiles under mild reaction conditions, enabled by a novel squaramide-based bifunctional iminophosphorane catalyst. Low catalyst loadings (2.0 mol %) were achieved on a decagram scale, demonstrating the scalability of the reaction. Computational analysis revealed the origin of the high enantiofacial selectivity via analysis of relevant transition structures and provided substantial support for specific noncovalent activation of the carbonyl group of the α,β-unsaturated amide by the catalyst.
The first metal-free catalytic intermolecular enantioselective Michael addition to unactivated α,β-unsaturated amides is described. Consistently high enantiomeric excesses and yields were obtained over a wide range of alkyl thiol pronucleophiles and electrophiles under mild reaction conditions, enabled by a novel squaramide-based bifunctional iminophosphorane catalyst. Low catalyst loadings (2.0 mol %) were achieved on a decagram scale, demonstrating the scalability of the reaction. Computational analysis revealed the origin of the high enantiofacial selectivity via analysis of relevant transition structures and provided substantial support for specific noncovalent activation of the carbonyl group of the α,β-unsaturated amide by the catalyst.
Conjugate
additions are among the most prevalent transformations
in organic chemistry due to their ability to quickly generate complexity
from simple starting materials with perfect atom economy.[1] Despite the maturity of the field, examples of
enantioselective conjugate additions to α,β-unsaturated
amides remain scarce. Contrary to other carboxylic acid derivatives,
the electron-withdrawing property of the carboxamide functionality
is greatly diminished.[2−5] Over the past two decades, multiple strategies relying on structural
modification of α,β-unsaturated amides have been disclosed,
enabling enantioselective conjugate additions. These, however, are
reliant on tailored activating groups, such as imides, N-acyl pyrroles, and thioamides among others, curtailing the synthetic
efficiency of these procedures.[6] To date,
only a handful of catalytic enantioselective methods have been described
featuring 1,4-additions to nonactivated α,β-unsaturated
amides. Pioneering studies by Kobayashi employed chiral crown ethers
in the presence of KHMDS to gain reactivity and enantiofacial control
in the conjugate addition between α,β-unsaturated amides
and carbon centered pronucleophiles,[7] while
Harutyunyan,[8] employed chiral bisphosphine
ligated copper(I) catalysis for the conjugate addition of alkyl Grignard
reagents to α,β-unsaturated amides. Most recently the
enantioselective copper(I)-catalyzed hydrophosphination and rhodium(I)-catalyzed
hydroboration of α,β-unsaturated amides in the presence
of chiral bisphosphine ligands were reported by Yin and Li, respectively.[9] While elegant, these methods required the use
of bespoke and sensitive ligated metal systems sometimes combined
with super stoichiometric activators. However, the enantioselective
addition of (pro)nucleophiles to unactivated α,β-unsaturated
amides under metal-free catalysis remains an unsolved problem. In
2013, our group disclosed a new class of superbasic catalysts, the
bifunctional iminophosphorane (BIMP), which has proven to be exceptionally
active in catalyzing challenging enantioselective conjugate additions.[10−14] Recognizing the limitations in enantioselective conjugate additions
to α,β-unsaturated amides and seeking the opportunity
to test the capabilities of new BIMP catalyst systems on conjugate
acceptors at the bottom end of Mayr’s electrophilicity scale
(Figure ),[5c] we sought to realize the first nonmetal-catalyzed
enantioselective conjugate addition reaction to α,β-unsaturated
amides. We chose to exemplify this with the sulfa-Michael addition
(SMA). Enantioselective carbon–sulfur bond-forming reactions
are prevalent transformations in organic chemistry due to the abundance
of sulfur atoms in biomolecules and pharmaceutical compounds.[15a] Recently developed β-thioamide Tyclopyrazflor,
for example, is a powerful pesticide.[15b−15f] If successful, our new methodology could
be applied to the rapid assembly of libraries of novel, otherwise
difficult-to-obtain enantiomerically enriched β-thioamides.
Our hope was to identify a suitable BIMP superbase catalyst capable
of significant activation of the α,β-unsaturated amide
electrophile and simultaneous deprotonation of high pKa alkyl thiol pronucleophile, and here we wish to report
our findings.
Figure 1
Mayr’s electrophilicity scale (bottom). Previous
BIMP catalysts
for SMAs to unsaturated carboxylic acid derivatives and this work
(top).
Mayr’s electrophilicity scale (bottom). Previous
BIMP catalysts
for SMAs to unsaturated carboxylic acid derivatives and this work
(top).
Results and Discussion
Readily available
(E)-N,N-dibenzyl
crotonamide 1a, being sterically
and electronically unbiased, was selected as the model substrate for
the enantioselective SMA.[5e] A preliminary
performance investigation of catalysts (at 10 mol %) was carried out
in THF at room temperature in the presence of 3.0 equiv of 1-propanethiol 2a (Table ). Initial experiments revealed that cinchona-derived catalyst C was essentially inactive in the transformation, resulting
in <3% product 3a formation after more than 1 week
reaction time (entry 1). First-generation thiourea and amide-containing
BIMP catalysts B1 and B2 bearing a single
stereocenter provided 3a in high yield, albeit with 37%
and 55% ee, respectively (entries 2 and 3). Diastereomeric second-generation
catalysts B3 and B4 bearing an additional
stereocenter flanking the hydrogen-bond-donor group efficiently furnished
product 3a in 25% and 51% ee, respectively (entries 4
and 5), demonstrating enantiocontrol was arising from both stereogenic
centers. Further architectural fine-tuning of catalyst B4 did not allow for significantly higher enantiocontrol, thus we turned
our attention to the nature of the hydrogen-bond-donor moiety of the
catalyst. Due to the inherently high Lewis basicity of carboxamides,
we speculated that a hydrogen-bond donor with an increased Brønsted
acidity could offer enhanced binding and thus better stabilization
of the transition structure (TS). Based on this reasoning, and inspired
by the pioneering work of Rawal and Jacobsen, a squaramide-containing
catalyst appeared to be a rational choice, due to its enhanced hydrogen-bonding
properties.[16,17] To our delight, switching to
squaramide-based catalyst B5 and the solvent to toluene
afforded 3a in 90% yield and 66% ee (entry 6). In a bid
to boost enantiocontrol, we introduced an additional stereocenter
on the distal side of the squaramide motif to give third-generation
BIMP catalyst, B6. We were pleased to find this structural
modification provided 70% ee and 82% yield (entry 7). Changing the
catalyst to one bearing two anti-configured tert-butyl groups, and switching the solvent to EtOAc, boosted
the ee to 85% (entry 8). The convenient late-stage formation of the
iminophosphorane moiety then allowed for facile tuning of the BIMP
catalyst by simply varying the phosphine component of the Staudinger
reaction. This systematic structural variation revealed the importance
of peripheral, bulky, and electron-donating groups,
leading to catalyst B8, which provided 3a in 95% ee and 88% isolated yield (entry 9).[18] Additionally, the inclusion of air in the reaction vessel did not
change the outcome of the transformation, and with the optimized conditions
in hand, the scope of the protocol was then explored (Scheme ). Initially the effects of
substituents on the amide nitrogen were evaluated (3a–3s). Pleasingly, switching one benzyl group
on 1a to a methyl group was well tolerated, and corresponding
product 3b was formed with 83% yield and 92% ee. Dimethylamine
derivative 1c afforded product 3c in 87%
ee, and no major change in reactivity was observed. Dibutylamine derivative
product 3d was obtained in 84% yield and 94% ee. Product 3e and unprotected indole derivative 3f were
formed smoothly under the optimized conditions in 94% ee and 95% ee,
respectively. Substrates bearing cyclic N-substituents afforded products 3g–3i with excellent enantioselectivities
and moderate to high yields. Amides 1j and 1k proved to be exceptionally reactive, and consequently cooling to
−20 °C was exploited to enhance enantioselectivity and
control, and products 3j and 3k were yielded
with 83% ee and 74% ee, respectively. Pharmaceutically relevant[19] isoquinoline derivative 3l and
thienopyridine derivative 3m were both compatible with
our method providing nearly quantitative yield and high ee. Even N-methylaniline-derived substrate 1n was well
tolerated, furnishing 3n with 88% yield and 81% ee. Encouraged
by the high degree of tolerance of heterocyclic moieties, we explored
the enantioselective SMA to α,β-unsaturated amide 1o. The reaction proceeded smoothly under the optimized conditions,
furnishing 3o, an analogue of Tyclopyrazoflor[15b−15e] with 90% ee and 90% isolated yield. Secondary amide 1p was also a competent substrate in this reaction, albeit a solvent
switch to toluene was required to better solubilize the starting material.
When six- and seven-membered α,β-unsaturated lactams 1q and 1r were used as substrates, products 3q and 3r were obtained in 93% yield and 87%
ee and 40% yield and 76% ee, respectively. Product 3s, bearing an α-stereogenic center, could also be obtained (via
enantioselective reprotonation) employing the same catalyst in 35%
yield and 80% ee. We then turned our attention to the β-substituents
on the enoyl backbone. Particularly unreactive[5c] cinnamide derivative 3t was obtained with
excellent enantioselectivity (95% ee) but moderate yield. Product 3u, on the contrary, was easily obtained, likely due to the
phenyl groups present on the amide moiety, twisting the N atom out
of conjugation.[5e] Product 3v was obtained with moderate yield and ee and was used to determine
absolute stereochemical configuration.[18] Cinnamides 1w and 1x bearing electron-withdrawing
groups were smoothly converted to the corresponding thioethers with
high levels of selectivity and reactivity. Pyridine-containing derivative 1y was well tolerated, furnishing 3w in near
quantitative yield and 85% ee. The introduction of a longer alkyl
chain in substrate 1z was well tolerated, albeit with
a slight decrease in reactivity. Substrate 1aa bearing
an additional methyl group compared to 1z exhibited decreased
reactivity (most probably due to additional steric encumbrance), and
therefore more forcing conditions had to be applied to afford product 3aa in 58% yield and 74% ee. Finally, a thorough assessment
of the nucleophile scope was performed using primary, secondary alkyl,
and benzyl-substituted thiols. Primary alkyl thiols were broadly tolerated,
affording the corresponding thioethers with high enantioselectivity
and reactivity (3ab–3af). Notably,
a thiol bearing an ester functionality afforded product 3af in 66% yield and 90% ee. A decrease in ee was observed in the case
of 3ag which contained a proximal CF3 group.
Secondary alkyl thiols provided products with high enantioselectivities
albeit with slightly diminished reactivity (3ah–3ai), while benzylic thiols underwent the transformation with
high levels of reactivity at a modest expense of ee (3aj–3am).[20]
Table 1
Selected Reaction Optimization (0.1
mmol Scale)
Isolated yield.
NMR yield after 7 days. Ee determined
by HPLC on a chiral stationary phase. PMP: p-methoxyphenyl.
Scheme 1
Reaction Scope for the BIMP B8-Catalyzed Enantioselective
SMA to α,β-Unsaturated Amides (0.1 mmol Scale)
Reactions were conducted under
air. Reaction carried out using: a10.0 equiv, b4.0 equiv, c3.0 equiv thiol. Reaction carried out at: d–20 °C and e40 °C. fReaction carried out in toluene. g0.5 M concentration.
*Absolute stereochemical configuration of product 3v was
determined by chemical correlation.[18]
Isolated yield.NMR yield after 7 days. Ee determined
by HPLC on a chiral stationary phase. PMP: p-methoxyphenyl.
Reaction Scope for the BIMP B8-Catalyzed Enantioselective
SMA to α,β-Unsaturated Amides (0.1 mmol Scale)
Reactions were conducted under
air. Reaction carried out using: a10.0 equiv, b4.0 equiv, c3.0 equiv thiol. Reaction carried out at: d–20 °C and e40 °C. fReaction carried out in toluene. g0.5 M concentration.
*Absolute stereochemical configuration of product 3v was
determined by chemical correlation.[18]After establishing the scope and limitations
of this new methodology,
we wanted to demonstrate its scalability using model substrate 1a and 1-propanethiol 2a. Doubling the reaction
concentration and reducing catalyst loading to 2.0 mol % allowed a
400-fold (40.0 mmol) scale-up of the model reaction, and product 3aS was obtained in 96% isolated yield (13.2 g) and 93% ee
(Scheme ). Next, a
series of transformations were performed using 3aS to
showcase the synthetic utility of this product. When treated with
lithium aluminum hydride, aminosulfide 4a was obtained
in 79% isolated yield and was subsequently debenzylated in the presence
of CbzCl in PhMe at 60 °C to afford protected secondary amine 4b in 50% yield and 92% ee. A cyclopropane motif could be
installed via the Kulinkovich–de Meijere reaction[21] using ethylmagnesium bromide and titanium(IV)
isopropoxide. Aminocyclopropane 4c was obtained in 43%
yield and 92% ee. Oxidation in the presence of oxone provided sulfone 4d in 87% yield with no loss of optical purity. Oxidation
with mCPBA furnished sulfoxides 4e and 4e′ in 91% yield but low diastereoselectivity; however,
the two diastereoisomers could be separated, providing single diastereoisomers
with practically no erosion in ee.
Scheme 2
Decagram-Scale
Enantioselective SMA
(I) Azide A8 on
the left, phosphine P19 on the right; (II) crude catalyst B8 after removal of THF; (III) substrate 1a;
(IV) product 3aS (left) and product derivatization (right).
Unintentionally, substrate
binding/activation of the new catalyst
system was effectively revealed using N,N-dibenzyl 4-nitrocinnamide 1x and thiol 2a. Substrate 1x can undergo nucleophilic addition reactions
to the conjugated alkene at either the α or β position
with respect to the amide functionality, and thus regioselectivity
of the addition to this dual-mode Michael acceptor can be used to
probe catalyst function (Table ). Performing the reaction under the optimized conditions
using an achiral organic superbase bearing no hydrogen-bond donor
(BEMP) revealed that the inherent reactivity of 1v is
governed by the 4-nitrostyrene moiety, providing a 1:10 mixture of 3x:3x′, implying that this functionality
is indeed more electron withdrawing than the amide moiety (entry 1).
However, running the reaction using catalyst B8 under
the same conditions reversed the regioselectivity, furnishing products 3x and 3x′ in a 4:1 ratio, thus providing
convincing evidence for the activation of the amide moiety by the
BIMP catalyst (entry 2). Subjecting the 1:1 mixture of B8 catalyst’s azide precursor A8 and BEMP to the
optimal reaction conditions afforded a 1.0:1.6 ratio of 3x:3x′ and a significantly lower ee of 3x compared to B8-catalyzed reaction, accentuating the
importance of the chiral tether between the iminophosphorane superbase
and hydrogen-bond donor in B8 (entry 3).
Table 2
Mechanistic Investigation Employing
a Dual-mode Michael Acceptor#
Yields were determined by quantitative 1H
NMR (0.1 mmol scale).
Decagram-Scale
Enantioselective SMA
(I) Azide A8 on
the left, phosphine P19 on the right; (II) crude catalyst B8 after removal of THF; (III) substrate 1a;
(IV) product 3aS (left) and product derivatization (right).Yields were determined by quantitative 1H
NMR (0.1 mmol scale).
Computational
Study
A DFT study using the ADF program[22] was
performed to elucidate the origins of stereocontrol in the BIMP-catalyzed
SMA to α,β-unsaturated amides (Figure ).[23] The reaction
proceeds via the sequential complexation between
the catalyst and substrates, leading to a TS, a subsequent conjugate
addition, and finally an irreversible protonation of the enolate.
In this study, we have focused on the stereoselectivity-determining
conjugate addition step, as no kinetic isotope effect (KIE = 1.0)
or a change in ee was observed when performing the reaction with propanethiol-d1, suggesting that the protonation step is neither
rate- nor enantio-determining.[18] Initially,
the iminophosphorane moiety of the BIMP catalyst deprotonates the
thiol and forms zwitterionic intermediate RC1, stabilized
by the two hydrogen-bond donors of the squaramide, and the protonated
iminophosphorane. Prior to the conjugate addition step, the amide
electrophile coordinates to RC1 and generates intermediate RC2. Due to the conformational freedom and the existence of
two potential activation modes of the BIMP catalyst, we computed and
compared all the possible TSs during the enantio-determining conjugate
addition step, involving amide 1c and methyl thiol as
the model nucleophile (Figure S9).[11k,12c,14f,14h] Both side chains of the BIMP squaramide catalyst can freely rotate
and the two most stable conformations for its “left
arm” (LA) and “right
arm” (RA) were explored.[18] As originally hypothesized by Pápai,
there are two modes (mode A and B) in which the catalyst may bind
to the substrate in the TSs.[18,24] In the case of mode
A, the electrophile is activated by the hydrogen-bond-donor moiety,
while the nucleophile is bound to the protonated iminophosphorane.
In the case of mode B, the nucleophile is coordinated to the hydrogen-bond
donor and the electrophile is activated by the protonated iminophosphorane.
Our computational analysis utilizes the terminology discussed above
(Figure ).[18]
Figure 2
Possible binding modes and terminology used for computational
studies.
Possible binding modes and terminology used for computational
studies.All 16 possible TSs for the stereoselectivity-determining
conjugate
addition step were located (Figure S9).
The lowest-energy conjugate addition TS was TS1 that
forms the (S)-product, which is in agreement with
the experimentally confirmed absolute stereochemical outcome of the
reaction (Figure ).
The relatively low-energy barrier of 20.5 kcal mol–1 is furthermore consistent with the mild reaction conditions required
to perform the transformation. The lowest-energy TS, responsible for
the formation of the minor (R)-product, is TS1′, which proceeds through a higher energy barrier
than TS1 (ΔΔG⧧ = 4.1 kcal mol–1). The stereoselectivity for this
transformation originates from the TS geometry that benefits from
multiple inter- and intramolecular stabilizing interactions, including
hydrogen bonding, CH−π, and π–π interactions.
These stabilizing features are enhanced in TS1 compared
to TS1′. The intramolecular hydrogen bonding between
the O(squaramide)–H(amide) fixes the conformational freedom
of the “LA” of the BIMP catalyst, creating
a three-dimensionally defined pocket within which the α,β-unsaturated
amide can fit without considerable steric repulsion during the C–S
bond forming event. Furthermore, the thiolate anion interacts with
the aromatic ring of the iminophosphorane moiety for additional satbilization.
Analysis of noncovalent interaction (NCI) plots allows one to qualitatively
visualize these weak interactions between the catalyst and substrates
(Figure S12).[25] The observed NCIs were then quantified using energy decomposition
analysis (EDA) for the comparison of the stabilization effects between TS1 and TS1′ (Figures S13 and S14).[18,26] In addition, to understand why
both the reactivity and enantioselectivity were improved by the exchange
of a thiourea with a squaramide in the catalyst structure, a computational
comparison between these motifs was performed. Our findings indicate
that the squaramide BIMP catalyst can facilitate a faster conguate
addition (lower ΔG⧧) with
a greater degree of enantioselectivity (larger ΔΔG⧧ relative to the next most favorable
TS) compared to the thiourea BIMP catalyst. These findings were supported
by experimental studies comparing analogous squaramide and thiourea-based
BIMPs (Figure S11).[18] Overall, our state-of-the-art computations have enabled
us to understand the origin of the enantioselectivity for this transformation
and open up new avenues toward the rational design of novel catalysts.
Figure 3
Computed
potential energy surface (ΔG [kcal
mol–1]) of the BIMP squaramide-catalyzed sulfa-Michael
addition computed at COSMO(EtOAc)-ZORA-M06-2X/TZ2P//COSMO(EtOAc)-ZORA-BLYP-D3(BJ)/DZP.
Energies (kcal mol–1) and forming bond lengths (Å)
of the TS geometries are provided in the inset.
Computed
potential energy surface (ΔG [kcal
mol–1]) of the BIMP squaramide-catalyzed sulfa-Michael
addition computed at COSMO(EtOAc)-ZORA-M06-2X/TZ2P//COSMO(EtOAc)-ZORA-BLYP-D3(BJ)/DZP.
Energies (kcal mol–1) and forming bond lengths (Å)
of the TS geometries are provided in the inset.
Conclusion
Exemplified by the alkyl thiol SMA, the first metal-free catalytic
enantioselective intermolecular conjugate addition to unactivated
α,β-unsaturated amides has been developed. A thorough
investigation of substrate types revealed a general methodology that
furnishes a wide range of SMA products, including heterocyclic derivatives,
in high yields and ee. Computational and mechanistic studies revealed
the origins of selectivity and the important substrate/catalyst binding
modes. Efforts continue in our laboratories to uncover new BIMP designs
and to expand the range of BIMP-enabled transformations.
Authors: Branislav Ferko; Marián Zeman; Michele Formica; Sebastián Veselý; Jana Doháňošová; Ján Moncol; Petra Olejníková; Dušan Berkeš; Pavol Jakubec; Darren J Dixon; Ol'ga Caletková Journal: J Org Chem Date: 2019-05-23 Impact factor: 4.354
Authors: Jonathan C Golec; Eve M Carter; John W Ward; William G Whittingham; Luis Simón; Robert S Paton; Darren J Dixon Journal: Angew Chem Int Ed Engl Date: 2020-08-07 Impact factor: 16.823