Published as part of the 50 Years SYNTHESIS – Golden Anniversary Issue
Key words
allylic alkylation - acyclic stereocontrol - tertiary alcohols - α-tertiary amines
- quaternary carbon stereocenters
1
Introduction
Palladium-catalyzed asymmetric allylic alkylation (Pd-AAA) stands as a unique reaction
due to the high number of stereocenters that may be created by this process.[1] Arguably, few transition-metal-catalyzed reactions offer the synthetic chemist the
ability to form C–C, C–N, C–O, C–F, and C–S bonds by such a variety of mechanisms
for asymmetric induction (Scheme [1]). The development of Pd-AAA has been reviewed elsewhere; however, a specific aspect
of Pd-AAA has significantly matured over the past 20 years, namely the synthesis of
acyclic tetrasubstituted stereocenters.[2]
Pd-AAA has been shown to be successful for inducing asymmetry on both electrophilic
and nucleophilic partners. Consequently, this review is divided into these two classes
of reactions, with an emphasis on vinyl epoxides as electrophiles and enolates as
prochiral nucleophiles. We conclude with a discussion on cyclic pronucleophiles that
are readily converted into acyclic tetrasubstituted stereocenters subsequent to the
alkylation reaction, as well as some important advances made with other metals.
Scheme 1 Examples of stereocenters formed in Pd-AAA reactions
Stereocontrol on Prochiral Electrophiles
2
Stereocontrol on Prochiral Electrophiles
In the case of Pd-allylic alkylation, the most common mechanism operative involves
an initial coordination of a chiral Pd(0) catalyst with an olefin bearing a leaving
group at the allylic position (Scheme [2]). Ionization occurs with inversion, where the η2-π-allyl–Pd(II) complex is situated on the opposite face to the leaving group. Alkylation
of the η3-π-allyl–Pd(II) complex occurs with inversion of the stereochemistry about Pd(II),
with simultaneous regeneration of the chiral Pd(0) catalyst. Because both the ionization
and nucleophilic addition occur with inversion, the nucleophilic substitution occurs
with net retention of the stereochemistry of the leaving group. An outer-sphere attack
of a nucleophile is the most common pathway for soft nucleophiles (pK
a <25). For hard nucleophiles, the nucleophile attacks the metal directly and undergoes
an inner-sphere reductive elimination with retention of stereochemistry about Pd(II),
and overall net inversion of stereochemistry.
Scheme 2 Mechanism of regio- and stereospecific Pd-catalyzed AA
Nucleophiles such as ketone enolates alkylate at the α-carbon, producing a stereocenter
on the nucleophile. In this case, asymmetric induction can be difficult to achieve
since the C–C bond forming event most commonly occurs via an outer-sphere mechanism
and the nucleophile remains distal to the chiral information about the metal–ligand
environment. This mode of asymmetric induction will be discussed in Section 3.
Alkylation of allyl electrophiles bearing substituents, although typically much easier
substrates for obtaining asymmetric induction, operate via more complicated mechanisms
involving π–σ–π equilibration of the π-allyl intermediate (Scheme [3d]). Additionally, the enantiodetermining step of the reaction may occur during any
step of the catalytic cycle, save for dissociation of the alkylation product. In Pd-AAA,
only two mechanisms are likely to be operative in the synthesis of acyclic tetrasubstituted
stereocenters on the electrophile, namely differentiation of enantiotopic olefin faces
(Scheme [3b]) and π–σ–π equilibration (Scheme [3d]). In most cases, the chiral ligands employed in Pd-AAA are phosphines, and a variety
of scaffolds have arisen from the many creative approaches toward acyclic stereocontrol
in allylic alkylation (Figure [1]).
Scheme 3 Modes of asymmetric induction on electrophile
Figure 1 Structures of chiral ligands used in Pd-AAA strategies toward acyclic tetrasubstituted
stereocenters
2.1
Reactions of Isoprene Monoepoxide
The potential of achieving Pd-AAA with isoprene monoepoxide (1) was an attractive pursuit, due to the prospect of achieving high enantioinduction
for alkylation at the branched position (Scheme [4]). The Trost group has achieved branch-selective alkylation for carbon,[3] oxygen,[4]
[5] and nitrogen[6–8] nucleophiles, affording valuable chiral acyclic tetrasubstituted building blocks
containing orthogonal hydroxymethyl and vinyl functional group handles.
Scheme 4 Pd-AAA strategy for the synthesis of acyclic tetrasubstituted building blocks using
electrophilic isoprene monoepoxide
These reactions constitute a dynamic kinetic asymmetric transformation (DYKAT), whereby
racemic isoprene monoepoxide is ionized by a palladium catalyst, with the initially
formed π-allyl species bearing an opposite sense of chirality or opposite syn/anti configuration[9] (Scheme [5]). Since the unbranched terminus of the π-allyl contains two hydrogen substituents,
the η1-σ-Pd(II) complex allows for rotation from one stereoisomer isomer to the other without
syn/anti isomerization (Scheme [5a]). Although π–σ–π equilibration may occur at the branched carbon (Scheme [5b]), this pathway is unproductive since both the π-allyl stereochemistry and the syn/anti configuration have isomerized.
Scheme 5 Mechanism of dynamic kinetic asymmetric transformation (DYKAT) of isoprene monoepoxide
An important structural consideration for this reaction also involves the equilibrium
between syn and anti isomers (Scheme [6]). Upon ionization of butadiene monoepoxide (2), diastereomeric π-allyl complexes are formed. The hydroxymethylene substituent is
sterically more hindered than the hydrogen substituent; consequently, complex 2a is the most favored complex due to the least amount of steric clash with the ligand
environment. The regioselectivity of nucleophilic addition runs counter to that typically
observed for Pd-AAA, where the branched isomer is formed in high selectivity. The
‘wall and flap model’ shows the approximate chiral environment about each terminus
of the electrophile based on chelation of a C2-symmetric ligand bearing diarylphosphines
(i.e. Trost ligand, BINAP).
Scheme 6 Wall and flap model as a rationale for regio- and enantioselectivity in allylic alkylation
with butadiene monoepoxide
In the case of isoprene monoepoxide (1), the methyl and hydroxymethylene substituents are sterically very similar; however,
the presence of an alkoxide from the epoxide opening renders the two electronically
dissimilar (Scheme [7a]). Two factors dictate alkylation at the branched terminus, namely: 1. The alkoxide
leaving group often acts as a base for the pronucleophile and engages in hydrogen
bond directed alkylation of the nucleophile. 2. Ligand and additive effects can alter
the most favorable trajectory of the incoming nucleophile toward the branched product.[10] The outer-sphere nucleophilic attack occurs with inversion of the stereochemistry
of the π-allyl, and each enantiomer of product can be obtained from two of the stereoisomeric
complexes (Scheme [7b]).
Scheme 7 Mechanistic pathways for Pd-AAA of isoprene monoepoxide
Methanol was shown to alkylate at the branched carbon with high enantioselectivity
in the Pd-AAA of isoprene monoepoxide (Scheme [8]).[4] The use of catalytic borane was crucial for reactivity, as the reaction proceeded
with little conversion in its absence. Initially, stoichiometric trimethyl borate
[B(OMe)3] was used in the reaction, affording the product in 80% yield and 2% ee. It was envisioned
that the stoichiometric use of borane results in too rapid a reaction, such that equilibration
of isoprene monoepoxide cannot occur prior to nucleophilic attack. Use of catalytic
triethylborane in the presence of methanol produced diethylmethoxyborane as catalyst,
and these conditions proved optimal for enantioselectivity.
Scheme 8 Optimization of allylic etherification with isoprene monoepoxide
Because of the lower acidity of aliphatic alcohols, their use in Pd-AAA has remained
limited; however, the kinetic effect imparted by the boron additive renders this reaction
favorable in alkylation of vinyl epoxides (Scheme [9a]).[4] Additionally, the unimolecular mechanism accounts for the high selectivity for the
branched product. Primary alcohols 3 undergo the alkylation chemoselectively over secondary alcohols. Additionally, the
high yields indicate the product alcohol 6 is not competitive with primary alcohol substrate 3, presumably due to either a slow alkylation reaction or a small population as the
alkoxydiethylborane species 4 (Scheme [9b]).
Scheme 9 Borane co-catalyst in Pd-catalyzed regio- and enantioselective allylic etherification
of isoprene monoepoxide
The reaction scope shows several primary alcohols 3 were competent nucleophilic partners in this reaction (Scheme [10]).[4] For substrates with low enantioselectivity using triethylborane as catalyst, it
was reasoned that etherification was occurring faster than the required π-allyl equilibration.
Use of the more bulky tri-sec-butylborane proved useful for solving this problem (6b, 6d, 6g). With a hampered rate of etherification, the products were formed with higher enantioselectivity.
Spontaneous cyclization to a hemiketal (6d) or lactone (6b) occurred when the nucleophilic alcohol contained a pendant ketone or methyl ester
respectively. The etherification product of 4-methoxybenzyl alcohol addition 6c is a particularly useful synthetic precursor due to the ubiquity of PMB-protected
alcohols in synthetic sequences.
Scheme 10 Select examples of tertiary ether synthesis with isoprene monoepoxide
A net addition of nucleophilic hydroxide was discovered for this reaction using a
mixed system of triethylborane and sodium hydrogen carbonate (Scheme [11]).[5] Upon ionization, a borinic ester intermediate is formed that directs the alkylation
of hydrogencarbonate to the branched position with high enantioselectivity. The unstable
carbonate product 7 undergoes decarboxylation to directly afford the tertiary alcohol product 8.
Scheme 11 Pd-AAA of isoprene monoepoxide with sodium hydrogen carbonate in the presence of
borane co-catalyst
In the absence of borane additive; however, the carbonate product 9 was isolated (Scheme [12]).[5] The difference in reaction outcomes was attributed to the attack of carbon dioxide
by the alkoxide leaving group. This system proved more challenging for optimization
due to the lability of 9 to racemization. In the presence of tetrabutylammonium chloride, the product was
isolated with high enantioselectivity at low conversion, but was nearly racemic upon
completion of the reaction. However, sodium hydrogen carbonate in phosphate buffer
resulted in high enantioselectivity at high conversion. Under the optimized conditions,
low catalyst loading in the absence of buffer proved effective for achieving high
enantioselectivity.
Scheme 12 Pd-AAA of isoprene monoepoxide with sodium hydrogen carbonate in the absence of borane
co-catalyst.
Carbon nucleophiles were successfully applied in this chemistry using nitromethane
(10) or β-keto esters (Scheme [13]).[3] These reactions were pioneering for the metal-catalyzed synthesis of acyclic all-carbon
quaternary stereocenters, which has been an active area of pursuit since the development
of this chemistry.[11] The product from nitromethane addition 11 is particularly valuable, as the resulting nitromethylene substituent serves as an
acyl anion equivalent, orthogonal to the two other functional group handles (Scheme
[13a]). In the case of β-keto ester 12, the hydroxymethyl group spontaneously cyclizes to form a hemiketal 13 upon allylic alkylation (Scheme [13b]). A retro-aldol reaction of 13 proceeds cleanly in the presence of tetrabutylammonium fluoride (TBAF) to furnish
acyclic 14.
Scheme 13 Pd-AAA of isoprene monoepoxide with carbon nucleophiles
Isoprene monoepoxide has been successfully alkylated by phthalimide (15) to afford the branched, α-tertiary imide 16 (Scheme [14]).[7] The regioselectivity is invoked to be the result of both hydrogen bonding and ligand-directed
effects. Achiral dppe afforded exclusively the linear product. The reaction with chiral
ligand L2 proved to be solvent, base, and temperature sensitive. High enantioselectivity could
be achieved without exogenous base; however, this resulted in exceedingly long reaction
times. Prolonged reaction time after completion of the reaction resulted in a higher
proportion of branched product, likely due to re-ionization of 16.
Scheme 14 Pd-AAA of isoprene monoepoxide with phthalimide
Enantiopure amino esters 17 were reacted with isoprene monoepoxide in a catalyst-controlled regio- and diastereoselective
allylic alkylation (Scheme [15]).[8] A subsequent cyclization by KCN afforded 2-oxomorpholines 18 with high diastereoselectivity. Although the products of the sequence are cyclic,
the product of the first step of the reaction sequence is an acyclic α-tertiary amine,
demonstrating that amines react in analogy to phthalimide with isoprene monoepoxide.
It was shown that catalyst control could dictate diastereodivergent pathways when
using the natural amino acids. Multiple ligands were employed, due to the presence
of match-mismatch effects unique to each amino ester. Since the enantiomeric amino
acids are commercially available, all four stereoisomeric products are accessible
by this route. This is a valuable feature, especially in diversity-oriented approaches
to compound libraries.[12]
Scheme 15 Pd-AAA of isoprene monoepoxide with enantiopure amino esters
2.2
Applications in Complex Molecule Synthesis
Isoprene monoepoxide was successfully used in the key steps in the synthesis of hyperolactone
C (26) and (+)-biyouyanagin A (29) (Scheme [16]).[13] The nucleophilic precursor 24 was accessed by dianion functionalization of methyl acetoacetate (21) with benzaldehyde. The necessary diazo precursor 23 was synthesized in two steps from 22, and the Pd-AAA pronucleophile 24 was accessed by rhodium carbene insertion into an enol O–H bond. The two vicinal
tetrasubstituted stereocenters of 25 were formed by a regio-, enantio-, and diastereoselective Pd-AAA between 24 and isoprene monoepoxide (1). The short reaction time proved crucial to the success of this reaction, as a host
of side reactions occurred upon completion of the reaction. The 59% isolated yield
of 25 is reflective of a modest 2.1:1 branched/linear ratio; however, the product was isolated
as a single enantiomer in 26:1 d.r.
Scheme 16 Pd-AAA of isoprene monoepoxide in the total synthesis of hyperolactone C and (+)-biyouyanagin
A
Lactonization of 25 catalyzed by 4-toluenesulfonic acid produced hyperolactone C (26) in a single step. For the synthesis of (+)-biyouyanagin A (29), ent-hyperolactone C was readily accessed by simply changing the enantiomer of Trost L2. The efficiency of the sequence is outstanding, with the synthesis of hyperolactone
C requiring a longest linear sequence of six steps, and (+)-biyouyanagin A synthesized
by an additional photocycloaddition with zingiberene (27) in the presence of 2′-acetonaphthone (28).
In the total synthesis of (–)-terpestacin[14] (Scheme [17]), the key Pd-AAA reaction between a diosphenol 30 and racemic isoprene monoepoxide (1) was observed to proceed rapidly and in high yield, but with low enantioselectivity
(~50% ee). It was reasoned that the low enantioselectivity potentially resulted from
a short lifetime of the π-allyl intermediate due to rapid nucleophilic attack. Addition
of halide additives has been shown to increase enantioselectivity by increasing the
rate of π–σ–π equilibration.[15] Increasing the concentration of tetrabutylammonium chloride displayed this effect;
however, slow addition of 30 proved most effectual for increasing the lifetime of the π-allyl intermediate. Under
optimized conditions, silylation of the product alcohol was performed in a single
operation, affording product 31 in 93–95% yield and 88–96% ee. The chiral information of the newly formed stereocenter
could be translated via a [3,3]-sigmatropic rearrangement, and direct oxidation of
the Claisen rearrangement product afforded the desired diketone 33 in 78% yield over the sequence, with 4:1 selectivity in the formation of the resultant
olefin. A Sakurai allylation of 33 installed an allyl functional group handle to form 34. This precursor was used in the construction of macrocycle 35. An additional catalyst-controlled Pd-AAA/sigmatropic rearrangement in the late stages
of the synthesis afforded 38, which was efficiently elaborated to the target.
Scheme 17 Pd-AAA of isoprene monoepoxide in the total synthesis of (–)-terpestacin
The synthesis of tipranavir was accomplished by setting the stereochemistry of a tetrasubstituted
and tertiary stereocenters by Pd-AAA (Scheme [18]) and Mo-AAA, respectively (Scheme [19]).[16] A propyl-substituted vinyl epoxide 41 was accessed in two steps from 1-chloropentan-2-one (40). After addition of vinylmagnesium bromide, the tertiary alcohol product was efficiently
cyclized to form 41 in 86% yield over two steps. A borane co-catalyzed Pd-AAA of the vinyl epoxide with
4-methoxybenzyl alcohol proved highly enantioselective in the construction of acyclic
tertiary ether 42. Synthesis of the phenethyl side chain was readily accomplished by Heck arylation
and hydrogenation of the vinyl functional group handle. The hydroxymethylene functional
group handle was homologated via an oxidation/Wittig olefination/hydroboration/oxidation
sequence.
Scheme 18 Synthesis of acyclic tertiary ether fragment in the total synthesis of tipranavir
The tertiary stereocenter was synthesized by a branch-selective cinnamylation of dimethyl
malonate sodium salt 48 and 3-nitrocinnamyl electrophile 47 under chiral molybdenum catalysis (Scheme [19]). Decarboxylation of a methyl ester afforded nucleophilic precursor 48, which was coupled with the previously prepared aldehyde 46 via an ester enolate aldol reaction. Oxidation of the resulting alcohol, followed
by PMB deprotection provided substrate 50 for a sodium hydroxide mediated lactonization. Hydrogenation conditions afforded
both nitro and olefin reduction, and the product 52 was completed by sulfonamide formation. A similar strategy was employed in the total
synthesis of (–)-malyngolide.[17] Additionally, synthetic studies on amphidinolide B1,[18] as well as (+)-pleuromutilin[19] have employed isoprene monoepoxide in Pd-AAA for the synthesis of early chiral building
blocks.
Scheme 19 A Pd- and Mo-AAA strategy for the synthesis of tipranavir
2.3
Other Acyclic Electrophiles
In the work of the Trost group on the prenylation of oxindoles, conditions were developed
to selectively afford both the branched and linear prenylation products (Scheme [20]).[20] This strategy culminated in a unified approach toward flustramine natural products.
For the reverse prenylated product 56, vicinal quaternary carbons are formed while setting the cyclic stereocenter in high
enantioselectivity. Formation of the linear product 55 using the Trost L1 was typical of most Pd-AAA; however, the unique structural features of Trost L2 provide the branched product. Additionally, tetrabutylammonium difluorotriphenylsilicate
(TBAT) as a halide additive improved the reaction outcome, due most likely to a required
rate increase in π–σ–π equilibration.
Scheme 20 Regioselective Pd-catalyzed prenylation of oxindoles
An extension of this chemistry allowed for the synthesis of an acyclic quaternary
carbon stereocenter when using geranyl-type electrophiles (Scheme [21]). When using racemic, branched electrophile 57, the linear and branched products were formed in equal amounts, with the branched
product 58 formed with high enantio- and diastereoselectivity.
Scheme 21 A regiodivergent Pd-AAA of oxindoles using racemic branched electrophile
It was determined from subsequent experiments that when using achiral (Z)-60, a regio-, enantio-, and diastereoselective synthesis of vicinal quaternary stereocenters
was achieved (Scheme [22]). These results indicate that ionization of the electrophile is enantiodetermining
with respect to the acyclic electrophile, and diastereoselective alkylation by the
oxindole occurs faster than π–σ–π equilibration. This process underscores both the
versatility in controlling stereochemical reaction outcomes and the importance of
understanding the underlying processes in Pd-AAA reaction optimization.
Scheme 22 Pd-catalyzed regio-, diastereo-, and enantioselective synthesis of vicinal quaternary
stereocenters
A Pd-catalyzed enantioselective allyl-allyl cross-coupling protocol was reported by
Morken and co-workers for the asymmetric synthesis of acyclic quaternary stereocenters
(Scheme [23]).[21] This process represents one of few Pd-catalyzed AAA reactions to occur with hard
nucleophiles with high enantioselectivity. It was shown that both (E) and (Z) linear electrophiles, as well as branched racemic electrophile 62 afforded the same enantiomer of the respective product in similar yields. This result
likely indicates that rapid equilibration of the π-allyl intermediate occurs before
reductive elimination, and it is ligand control of the π-allyl stereochemistry that
dictates the enantioselectivity. In the reaction optimization it was observed that
β-hydride elimination was a significant side product. Addition of fluoride in a THF/water
mixture was shown to minimize this side reaction by acceleration of allyl transmetalation.
Scheme 23 Catalytic enantioselective branch selective allyl-allyl cross coupling
The substrate scope revealed high enantioselectivities for the synthesis of acyclic
benzylic quaternary stereocenters. High enantioselectivity was observed when placing
branched aliphatic substituents in addition to the methyl group (64f, 64g, 64i). A modest, but appreciable enantioinduction was observed for a geranyl-type substrate
(64j).
The C–C bond-forming event is invoked to occur via a 7-membered 3,3′ inner-sphere
reductive elimination of bis(η1-allyl)Pd(II) complex 67 (Scheme [24]). In this scenario, palladium is bound to the unsubstituted terminus of the trisubstituted
allyl partner. The alternative mechanism would involve a more classical 3-membered
reductive elimination; however, calculations place this pathway at ~15 kcal/mol higher
in energy.[22] The source of enantioinduction in this reaction is not immediately clear. The stereocenter
is created in the reductive elimination step to form 68; however, the pre-equilibrium of the (E)- and (Z)-isomers of 66 may be enantiodetermining if the Re or Si pathway is significantly favored for both isomers.
Scheme 24 Mechanism of enantioselective allyl-allyl cross coupling
The branch-selective Pd-AAA reaction has been achieved using ligands developed by
the Hou group.[23] Racemic tertiary allylic acetates were alkylated with modest enantioselectivity
and branch/linear selectivity using dimethyl malonate as the pronucleophile (Scheme
[25]). Extensive optimization of ligand, solvent, additive, and base was required in
the process. This ligand class offers modulation of the BINOL, phosphorus, and oxazoline
chirality for achieving branch-selective Pd-AAA. The presence of high (E)-selectivity in the linear product 71 indicates that high syn/anti control is achieved in the π-allyl formation. Enantioinduction is achieved by facial
discrimination of the sterically differentiated methyl and aryl substituents.
Scheme 25 Pd-catalyzed branch-selective cinnamylation of dimethyl malonate for the synthesis
of acyclic quaternary stereocenters
Stereocontrol on Prochiral Nucleophiles
3
Stereocontrol on Prochiral Nucleophiles
In Pd-AAA, achieving high enantiocontrol on nucleophiles remains inherently more difficult
than achieving enantiocontrol on the electrophilic partner, due to the outer-sphere
mechanism for nucleophilic addition. In the alkylation event, the chiral information
about the metal–ligand sphere must be successfully relayed distal to the metal in
order to observe catalyst differentiation of enantiotopic faces of the nucleophile.
This problem is compounded by the presence of enantioconvergent and enantiodivergent
allylation pathways (Scheme [26]). In the case of enantioconvergent allylation (Scheme [26a]), the chiral catalyst recognizes the substituents at the site of reaction. Although
the restriction of controlling the enolate geometry has been removed, R1 and R2 are limited to sterically differentiable substituents (i.e. alkyl vs. aryl).
Scheme 26 Structural considerations for catalyst recognition of prochiral enolates
When an enantiodivergent pathway is operative (Scheme [26b]), the catalyst recognizes the substituents vicinal to the site of reaction. This
pathway is desirable if R1 and R2 are of limited steric differentiation; however, the enolate geometry must be strictly
controlled in order to observe high enantioselectivity. The enantiodivergent pathway
has an important advantage over the enantioconvergent pathway, since the identity
of X can be exploited as an achiral auxiliary and a tunable parameter for reaction
optimization.
Scheme 27 Potential mechanism for the erosion of enolate stereochemistry
As will be seen, both the α substituents and the enolate geometry are commonly recognized
by the catalyst in the allylation event. In this case, the presence of match-mismatch
effects will be apparent when both enolate isomers can be synthesized in geometrically
pure form. The stereofidelity of the enolate is assumed to remain intact during enantiodivergent
allylations; however, oxygen-to-carbon migration may provide a mechanism for erosion
of the enolate geometry if palladium enolates are formed as intermediates (Scheme
[27a]). Isomerization of palladium enolates has been exploited for the Pd-AAA DYKAT of
butenolide electrophiles with phenolic nucleophiles (Scheme [27b]).[24] In this scenario, the palladium enolate is formed on the butenolide electrophile
(Scheme [27c]), and the mechanism for isomerization of the π-allyl species involves carbon-to-oxygen
migration.
3.1
Intermolecular Alkylation of Prochiral Nucleophiles
Many of the first syntheses of acyclic tetrasubstituted stereocenters by reactions
of prochiral nucleophiles were discovered by the Ito group. In 1992, allylation of
an acyclic 1,3-diketone 81 (Scheme [28a]) was reported in modest enantioselectivity by employing a point and planar chiral
ferrocenyl ligand scaffold bearing an aza-crown ether.[25] In 1996, allylation of α-nitro ester 83 was achieved by a similar strategy (Scheme [28b]).[26] In order to enhance enantioselectivity upon nucleophilic attack of a π-allyl palladium(II)
complex, it was envisioned that linking a crown ether to chiral phosphine ligands
would allow for recognition of the nucleophile counterion. In this case, the chiral
information of the catalyst could be relayed distal the site of the reaction, thus
solving the problem of poor selectivity when using prochiral nucleophiles in Pd-AAA.
In both cases, significant optimization of the crown ether and counterion was required
to obtain the enantioselectivity.
Scheme 28 Asymmetric allylation of a 1,3-diketone and an α-nitro ester by counterion recognition
of a chiral phosphine ligand
A dual catalytic approach to Pd-AAA was achieved by exploiting chiral rhodium enolate
chemistry (Scheme [29]).[27] It had been shown that rhodium salts form a trans bis-phosphino complex that is bound to the nitrogen of cyano-derived enolates.[28] This chemistry had been previously applied to Michael additions and aldol reactions.[29] Using ferrocenyl ligands developed in the Ito group, they were able to achieve high
enantioselectivity in the allylation of cyano ester 85, cyanoamide 88, and α-cyanophosphonate 90 under π-allyl–Pd catalysis. Potentially, the ligand forms complexes with both rhodium
and palladium, and the chirality of each species is synergistic in the alkylation
event.
Scheme 29 Dual catalytic Pd-AAA of chiral cyano-rhodium enolates
Scheme 30 Pd/BINAP-catalyzed allylation of α-acetamido-β-keto esters
An early report on asymmetric induction utilizing acyclic nucleophiles was provided
by the Ito group[30] in their Pd(0)/BINAP-catalyzed Pd-AAA of α-acetamido-β-keto esters 92 (Scheme [30]). High levels of enantioselectivity were reported when using potassium tert-butoxide as base. Additionally, high geometric control of the resulting olefinic
products 94 was reported, even for commonly difficult aliphatic allylic acetates (94g, 94h, 94i). The highest enantioselectivities were observed for cinnamylation reactions (94c, 94d, 94e, 94f), although simple allylations proceeded in good enantioselectivity.
Curiously, the reaction outcome was largely independent of whether the branched 96 or linear (Z)-95 electrophilic substrate was used (Scheme [31]). It was proposed that any stereo- or regiochemistry present in the starting electrophiles
is lost due to a rapid π–σ–π equilibration relative to intermolecular nucleophilic
alkylation.
Scheme 31 Independence of electrophile regio- and stereochemistry on reaction outcome
A similar catalytic system was shown to provide allylation products for α-acetamido-β-ketophosphonate
nucleophiles 97 (Scheme [32]).[31] A possible explanation for the high enantioselectivity in this acyclic system is
control of the enolate geometry by chelation between the phosphonate and enolate oxygen
atoms. The highest enantioselectivity was observed for 3-substituted allylic electrophiles
93; allylation proceeded with only modest enantioselectivity.
Scheme 32 Pd-AAA of α-acetamido-β-ketophosphonates
Other work by the Ito group demonstrated acyclic control in the cinnamylation of 1,3-diketones
(Scheme [33]).[32] Two examples demonstrated cinnamylation of acyclic 1,3-diketones in good enantioselectivity.
Similar levels of asymmetric induction on cyclic substrates were also reported. Cryogenic
temperatures were required for high enantiocontrol in this reaction, however, in conjunction
with their work on α-acetamido-β-keto esters and α-acetamido-β-ketophosphonates, these
reactions serve as some of the earliest examples of Pd-AAA with stereocontrol on acyclic
nucleophiles under simple catalyst conditions.
Scheme 33 Examples of acyclic 1,3-diketone cinnamylation
Hou and co-workers have demonstrated Pd-AAA of acyclic diphenylamides[33] using the ferrocenyl-based ligands that they developed (Scheme [34]). In most cases, α-tertiary carbon centers were synthesized, however, the report
contains two examples of the asymmetric synthesis of acyclic tetrasubstituted stereocenters.
The N,N-diphenylamide was found to be optimal in the initial optimization. Subsequent variation
of the chirality on the oxazoline, phosphorus, and the BINOL moieties of the ligand
proved crucial for achieving high enantioselectivity.
Scheme 34 Synthesis of acyclic tetrasubstituted stereocenters by α-alkylation of N,N-diphenylamides
List and Jiang were able to demonstrate the asymmetric α-allylation of α-arylpropanals
107 under triple chiral acid/enamine/palladium catalysis (Scheme [35]).[34] An achiral palladium source was used in the reaction, and asymmetric induction was
achieved by achiral amine catalyst (e.g., 106) in the presence of (S)-TRIP phosphoric acid L15. Optimization of the amine catalyst showed benzhydrylamine (106) to be differential among other achiral benzylic amines. In addition to allyl alcohol,
more substituted alcohols 108 were effective in the reaction. Although the reaction showed impressive display of
control for asymmetric quaternary carbon synthesis, the method was limited to α-arylpropanals,
as an elongation of the aliphatic aldehyde side chain resulted in diminished enantioselectivity
(109f).
Scheme 35 The direct α-allylation of α-arylpropanals with allylic alcohols
Steric elements that result in high geometric control of the transiently formed chiral
enamine were invoked to account for the stereoselectivity observed (Scheme [36a]), and chiral information is transferred by hydrogen bonding by the enamine to the
chiral phosphate (Scheme [36b]). Since this species is counterion-paired with the cationic π-allylpalladium(II)
species, asymmetric induction can be realized in the alkylation event. Additionally,
it cannot be ruled out that the phosphoric acid acts as a ligand to the π-allylPd(II)
species. Interestingly, under the reaction conditions, allyl alcohols can be used
directly, due to the activation of the hydroxyl leaving group by the phosphoric acid
(Scheme [36c]).
Scheme 36 Proposed mechanistic feature of triple catalysis
Another dual-catalytic approach was showcased by Ooi and co-workers for the α-cinnamylation
of α-nitro esters (Scheme [37]).[35] Like the system developed by List and Jiang, the phosphine ligands employed in the
method are achiral. However, a linked cationic ammonium functional group allows for
self-assembly with a chiral BINOL counterion. A Pd/ligand ratio of 1:2 was used, as
it was believed that a C
2-symmetric palladium complex would form upon cis-complexation of the phosphine ligands. Excellent yields and enantioselectivities
were observed for the reaction when using cinnamyl carbonate electrophiles 111, although no enantioselectivity was observed for the simple allyl system (112d).
Scheme 37 Chiral counterion strategy for the cinnamylation of α-nitro esters
The use of dual chiral palladium/chiral boron catalysis was applied to the α-allylation
of carboxylic acids (Scheme [38]).[36] A chiral boron species is generated by chelation of a boron Lewis acid with the
chiral amino acid catalyst L18. Upon ionization of the allyl ester by the chiral palladium species (L17), a carboxylate ene-diolate species is readily formed, presumably due to the Lewis
acidity of the coordinated boron catalyst and the presence of DBU as base. Although
the enolates of carboxylic acids possess no geometry, the large steric differences
of the α-aryl substituents render the two enantiotopic faces of the nucleophile sufficiently
distinguishable. The ability to use two chiral catalysts enables excellent enantioselectivity
for both allyl and 3-substituted allyl systems. The scope of the reaction was largely
limited to α-aryl substrates, although and α-benzyl substrate gave 114b with 80% ee.
Scheme 38 Select examples in chiral Pd-AAA/chiral borane-catalyzed α-allylation of carboxylic
acids
The DAAA of cyclic α-fluoro-β-keto esters proceeds with high enantioselectivity.[37] Acyclic systems afforded products with high efficiency, although the enantioselectivities
were low. More promising enantioselectivities were obtained when generating geometrically
pure (Z)-lithium enolates (Scheme [39]).[38] These species were obtained by deprotonation of α-fluoropropiophenones by LiHMDS
at 0 °C, and the transient species were subjected to Pd-AAA conditions to provide
α-allyl-α-fluoropropiophenones with 60–90% ee.
Scheme 39 Pd-AAA of α-fluoropropiophenones
3.2
Decarboxylative Allylic Alkylation Strategies
Unlike intermolecular allylic alkylation reactions, Pd-catalyzed decarboxylative allylic
alkylations occur via a largely unimolecular reaction pathway (Scheme [40]).[39] That is, upon ionization of an allyl moiety, the Pd–π-allyl species is counterion-paired
to its pronucleophile. Upon a decarboxylation event, alkylation occurs via an outer-sphere
attack of the nucleophile on the reactive Pd–π-allyl species, effectively furnishing
the desired product and the regenerated chiral Pd(0) species.
Scheme 40 Mechanism of the Pd-catalyzed decarboxylative allylic alkylation
The question of whether this reaction occurs via an inner- or outer-sphere mechanism
is complicated by differences in the experimental outcomes of these reactions and
the computed lowest energy pathway.[40] It is worth noting that computational work is for phosphino-oxazoline (PHOX) ligands,
and experimental work is with structurally different bisphosphino Trost ligands. Experimental
work showed that enol carbonate 118 underwent Pd-DAAA with kinetic resolution of the enantiomeric starting material (Scheme
[41]). The alkylation product 119 was formed with net retention of the stereochemistry, indicating an outer-sphere
mechanism.
Scheme 41 Experimental evidence for an outer-sphere mechanism
Scheme 42 Enantiodivergent allylations of (E)- and (Z)-enol carbonates
It was observed by the Trost group in work on the α-allylation of acyclic ketones
that control of the enolate geometry was crucial for high asymmetric induction (Scheme
[42]).[41] It was shown that opposing enolate geometries diverged to opposite enantiomers of
product, with matched and mismatched cases. This problem was easily overcome in the
case of α-tertiary ketone formation, since either enolate isomer could be synthesized
by judicious choice of base. However, the synthesis of acyclic tetrasubstituted stereocenters
would prove more difficult, due in large part to the small steric differences in the
α-substituents.
The Trost group developed Pd-DAAA reactions for the synthesis of acyclic tetrasubstituted
stereocenters by the synthesis of protected α-tertiary hydroxyaldehydes.[42] The allyl carbonate precursors were synthesized with high geometric purity (Scheme
[43]). A highly flexible route to substrates allowed for α-hydroxy 122 or α-bromo ketones 124 to be used as substrates. Either regioisomer could be synthesized using hard or soft
enolization conditions. Under soft enolization using TBSOTf and triethylamine, acyl
transfer was suppressed, presumably due to the absence of any enolate formed, and
127a was isolated with high regioselectivity. Under hard enolization with sodium hexamethyldisilazanide,
acyl transfer occurs to form the more stable aldehyde enolate. Trapping with TBSCl
allows for synthesis of the opposite regioisomer 128a.
Scheme 43 An efficient regiodivergent synthesis of Pd-DAAA substrates
A DAAA reaction proceeds smoothly with high regio- and enantioselectivity when the
alcohol protecting group can undergo alkoxide transfer (Scheme [44]). The process was found to be regioconvergent; that is, either regioisomeric starting
material afforded the tertiary aldehyde product with nearly identical enantioselectivities.
When an alkoxide transfer is not required (substrate 127a), the product was formed in 15 minutes in 93% yield compared to a 1 hour reaction
for the regioisomeric enol carbonate 128a.
Scheme 44 A regioconvergent Pd-DAAA strategy toward α-tertiary hydroxyaldehydes
The substrate scope revealed aromatic, vinyl, and alkynyl substituents provided the
product with high enantioselectivity (Scheme [45]). High yields and a rapid reaction rate were achieved by employing substrates 127 with a tert-butyldimethylsiloxy group at the benzylic position, as these substrates undergo Pd-AAA
without silyl transfer. Additionally, decarboxylation likely occurs more rapidly with
these regioisomeric substrates, since the resulting aldehyde enolate is more stabilized
than the initially formed ketone enolate when using the regioisomeric substrate.
Scheme 45 A Pd-DAAA strategy toward α-tertiary hydroxyaldehydes
This strategy was applied to the synthesis of a protected α-hydroxy ketone 132 (Scheme [46]). Displacement of the alkyl bromide 130 occurred smoothly in 87% yield. Under hard enolization conditions, acyl transfer
was not observed due to the trans relationship of the oxygen atoms. This isomer 131 was observed to be the major one, and it was separated from the minor isomeric products
in 67% yield. The Pd-DAAA proceeded smoothly with regiospecificity in 94% yield and
80% ee.
Scheme 46 Synthesis and reaction of ketone-derived α-alkoxy enol carbonate
Additionally, electrophiles bearing stereochemistry could be subjected to the reaction
conditions to produce alkylation products with high diastereoselectivity. This strategy
was applied to the formal synthesis of (S)-oxybutynin (Scheme [47]). The sodium carbonate of 133 was formed by treatment with sodium hydride, followed by carbon dioxide capture.
The transiently formed species was then treated with 2-bromoacetophenone and was subsequently
converted into DAAA substrate 135 by regioselective silylation. Although the starting enol carbonate 135 existed as a mixture of enantiomers, the stereochemistry about the electrophile was
lost upon ionization, as the π-allyl species becomes symmetrical. The DAAA reaction
proceeded smoothly in 18 hours to afford the hydroxyaldehyde 136 in 99% yield and 11:1 d.r. For substrate 135, the tert-butyldimethylsilyl protecting group transfer occurred efficiently, as indicated by
the high yield in the DAAA reaction. Hydrogenation of the olefin afforded product
137 with 84% ee, reflecting the stereoselectivity about the tetrasubstituted carbon.
Upon desilylation of 137 and Pinnick oxidation, hydroxy acid 139 was obtained as a single isomer after a single recrystallization, completing the
formal synthesis of (S)-oxybutynin.
Scheme 47 Formal synthesis of (S)-oxybutynin by Pd-DAAA
A method for the synthesis of acyclic quaternary stereocenters by DAAA was investigated
by Tunge and Ariyarathna.[43] Racemic 141 was used with the hope that stereoinduction would be observed, even in the absence
of enolate control (Scheme [48]). In this system, the two α-substituents are of considerable steric difference,
and it could be imagined that this could result in stereocontrol of the formed enolate.
Under conditions of catalyst control, only modest enantioselectivity could be observed,
with Trost Ligand L4 affording the product with the highest enantioselectivity.
Scheme 48 Synthesis of quaternary stereocenters by Pd-DAAA of acyclic allyl β-oxo esters
Additionally, a series of chiral auxiliary containing substrates 143 were employed as mixtures of diastereomers at the α-carbonyl position (Scheme [49]). Products were formed with some level of stereoinduction observed, and one auxiliary
proved useful for recrystallization of the product in diastereomerically pure form.
Scheme 49 Auxiliary approach for the Pd-DAA of acyclic allyl β-oxo esters
A more elaborate method of synthesizing geometrically defined enolates was developed
by Marek and co-workers.[44] An auxiliary-containing ynamide was carbometallated and oxidized to give geometrically
pure enolates. This work was originally employed in aldol reactions; however, Starkov,
Marek, Stoltz, and co-workers demonstrated that the transiently formed enolate can
be successfully trapped as the enol carbonate (Scheme [50]).[45] Two protocols were developed for this reaction, due to limitations in substrate
scope using a one-pot method when employing acyclic carbamate substrates. In the second
method (Scheme [50b]), the iodo enamide 149 was lithiated with t-BuLi, and the enol carbonate 150 was isolated with geometric purity after oxidation and acylation.
Scheme 50 One-pot carbometalation/oxidation/acylation synthesis of tetrasubstituted amide enol
carbonates and two-pot carbometalation/iodination followed by lithiation/oxidation/acylation
Optimization with a variety of PHOX and Trost-style ligands revealed electron-deficient
Trost ligand L6 bearing the anthracene-9,10-diamine backbone provided the product with the highest
enantioselectivity (Scheme [51]). Substrate 151a was shown to provide 152a with higher enantioselectivity compared to substrates bearing an oxazolidinone.
Scheme 51 Optimized substrate for Pd-DAAA of tetrasubstituted amide enol carbonates
The subsequent Pd-DAAA reaction provided allylation products with high enantioselectivity,
even in cases where the two α-substituents are of similar steric bulk (Scheme [52]). This likely indicates that the chiral catalyst recognizes the steric and electronic
differences of the enolate oxygen and the achiral auxiliary as opposed to the differentiation
of the α-substituents. A two-step allylation/cross-metathesis protocol was employed
to better facilitate assay of the enantioselectivity. The acrylate moiety provided
products 153 with higher differentiability of substituents and better absorptivity for UV detection.
Scheme 52 Substrate scope of Pd-DAAA of tetrasubstituted amide enol carbonates
For substrates bearing two α-substituents with considerable steric difference (i.e.,
aryl vs. alkyl), high enantioselectivity can be achieved for aryl ketones (Scheme
[53]).[46] Using a protocol developed for the synthesis of analogous enol tosylates,[47] enol carbonate 155a was synthesized with (E) selectivity. Solvent optimization with an electron-deficient PHOX ligand L20 provided high enantioselectivity with a mixed hexane/toluene solvent system.
Scheme 53 A Pd-DAAA of stereodefined tetrasubstituted 1,2-diaryl enol carbonates
Scheme 54 Independence of enolate geometry on enantioselectivity for an electron-deficient
(t-Bu)-PHOX ligand
Although high geometric control could be achieved in the enol carbonate formation,
it was observed that this was unnecessary when employing the PHOX L20, as similar levels of enantioselectivity were observed for a mixture of enolate isomers
and β-keto ester 157 (Scheme [54]). However, high levels of enantioselectivity (86% ee) were observed when using Trost
ligand L4 for the geometrically defined enol carbonate under the optimized conditions for the
PHOX system, but selectivity (50% ee) was diminished when using a mixture of isomers.
It was speculated that rapid C–O equilibration of the palladium(II) enolate may account
for this effect when using the PHOX system. Potentially, optimization with the PHOX
ligand resulted in conditions where the facial differentiability of the α-substituents
dictates the stereochemical outcome of the reaction.
A reaction involving the DAAA of α-acetamido-β-keto esters 158 has been described for an acyclic system (Scheme [55]).[48] Interestingly, with little optimization, high levels of enantioselectivity were
realized in this reaction, albeit for a limited substrate scope. Potentially, there
is control of the enolate geometry due to hydrogen bonding between the amide hydrogen
and the transient enolate ion. Phenol and naphthol were found to improve the enantioselectivity,
although the origin of this effect is not well understood.
Scheme 55 Pd-DAAA of α-acetamido-β-keto esters
A Pd-AAA of nitroalkanes has been described by Shibasaki and co-workers (Scheme [56]).[49] Although most of the reactions described in the method involve stereocontrol on
the electrophile, a single example of Pd-AAA with a racemic secondary nitroalkane
160 and allyl carbonate 161 was reported, albeit with low enantioselectivity.
Scheme 56 Asymmetric allylation for the synthesis of α-tertiary nitroalkanes
Temporary Cyclic Pronucleophiles
4
Temporary Cyclic Pronucleophiles
4.1
Reactions of Azlactones
Alkylation of azlactones directly furnishes products with cyclic tertiary amino functionality,
and the subsequent hydrolysis of the products has proven useful for producing acyclic
α-tertiary amino acids (Scheme [57]). This provides an efficient method for the synthesis of unnatural amino acids,
as the synthesis and elaboration of azlactones can be readily accomplished using commercially
available amino acids. Additionally, the identity of the carboxylic acid template
may serve as a point for reaction optimization.
Scheme 57 Azlactones as precursors to acyclic α-tertiary amino acids
The Trost group has shown a breadth of Pd-AAA reactions with azlactones for a variety
of allyl electrophiles.[50] Impressively, some of the first asymmetric Pd-catalyzed benzylation reactions were
developed using azlactones as pronucleophiles.[51] For electron-deficient electrophiles 163, a diphenyl phosphate leaving group in conjunction with elevated temperatures allowed
for the difficult ionization to occur (Scheme [58]).
Scheme 58 Pd-catalyzed benzylation of azlactones
In the case of electron-rich benzyl electrophiles 166, the diethyl phosphate leaving group proved sufficient at ambient temperatures. This
methodology was used in a short asymmetric synthesis of α-methyl-d-DOPA (Scheme [59a]). Under conditions for the asymmetric benzylation of electron-rich benzylic phosphates,
the key alkylation step afforded α-tertiary azlactone 167 in 83% yield and 90% ee. The desired acyclic target 168 was synthesized in 96% yield by hydrolysis of the azlactone. Conveniently, the methylenedioxy
group was hydrolyzed under these conditions. Additionally, triflic acid hydrolysis
in the presence of methanol allows for access to the acyclic protected amino acids
171 (Scheme [59b]).
Scheme 59 Pd-catalyzed benzylation strategy for the synthesis of α-methyl-d-DOPA and benzylation strategy toward acyclic α-tertiary amino acids
Similarly, allyl electrophiles allow for the installation of a functional group handle
for synthetic modification (Scheme [60]).[50a] Interestingly, symmetrical allyl and 2-methallyl electrophiles yielded products
with low enantioselectivity. However, unsymmetrical 3-substituted electrophiles and
prenyl electrophiles afforded the α-tertiary alkylated products with high enantioselectivity.
It is important to note that prenyl and cinnamyl electrophiles produce alkylation
products with an opposite sense of chirality.
Scheme 60 Enantiodivergent alkylation with reverse prenyl and cinnamyl electrophiles
For unsymmetrical allyl systems, diastereomeric π-allyl complexes are possible (Scheme
[61]). This feature likely dictates a distinct mode of attack by the nucleophile, and
in the case of azlactones, this trajectory of approach results in better recognition
of the prochiral faces of the incoming nucleophile. When the linear product is favored,
it can be seen that by placing the 3-allyl substituent under the less hindered flap
(Scheme [61a]), nucleophilic attack occurs at the terminus where chiral information is better
relayed from the metal center.
Scheme 61 Rationale for higher enantioselectivity for alkylation of 3-substituted electrophiles
In addition to reactions that form stereochemistry on the nucleophile, the Trost group
has utilized this transformation for simultaneous stereoinduction on the electrophilic
partner (Scheme [62]).[50b] A Pd-AAA with cyclic electrophile 177 afforded the product 178 with high diastereo- and enantioselectivity. The remarkably high level of enantioselectivity
observed in this reaction is likely a result of statistical enrichment, since the
Pd(0)/chiral ligand system has been shown to induce high levels of asymmetric induction
on each partner independently.
Scheme 62 Diastereo- and enantioselective Pd-AAA of azlactones with cyclic prochiral electrophiles
A geminal diacetoxyallylic electrophile 179 was applied to the synthesis of sphingofungin F (Scheme [63]).[52] In this reaction, the stereochemistry on the electrophile results from selective
ionization of enantiotopic leaving groups independent of the presence of the nucleophile.
The stereochemistry on the azlactone in 180 was additionally obtained with high diastereoselectivity. The relative stereochemistry
of the amino and acetoxy functional groups was strategically inverted in subsequent
steps in the synthesis of sphingofungin F.
Scheme 63 A diastereo- and enantioselective synthesis of sphingofungin F
4.2
Cyclic α-Alkoxy-Bearing Substrates
A similar strategy as that used for α-tertiary amino acids could be envisioned for
asymmetric α-tertiary alcohol synthesis, whereby a cyclic template could serve to
control enolate geometry (Scheme [64]). The Stoltz group have shown that dioxanones 182 derived from 1,3-dihydroxyacetone can be used for this purpose.[53]
Scheme 64 Acyclic α-tertiary hydroxy ketones, acids, and esters from dioxanone
Alkylation of unsubstituted dioxanones, followed by silyl enol ether formation provided
substrates 184 for allylic alkylation. The trimethylsilyl enol ethers proved too unstable; however,
the regioisomeric triethylsilyl enol ethers 183 and 184 could be separated by silica gel chromatography. In the presence of TBAT as a desilylation
reagent, allylation proceeded in high yield and enantioselectivity using symmetrical
unsubstituted and 2-substituted allyl carbonates 185 as electrophiles. A 4-toluenesulfonic acid catalyzed hydrolysis of 186 in the presence of methanol gave the acyclic hydroxy ketone products 187. Periodate cleavage, followed by esterification provided α-tertiary hydroxyl esters
188. This method was used in an efficient synthesis of (+)-eucomic acid.[54]
A Pd-DAAA of 5- and 6-membered lactam precursors was developed by the Stoltz group
(Scheme [65]).[55] An α-hydroxy ester 191 was accessed by hydrolysis of the DAAA product 190 in the presence of catalytic methanol and sulfuric acid.[56] The method additionally was also performed using a sulfur-containing lactam, an
allyl 2-methyl-3-oxothiomorpholine-2-carboxylate, although derivatization of the product
to the acyclic precursor was not demonstrated.
Scheme 65 Lactam precursor to α-hydroxy tertiary esters
Likewise, thiopyranones were used for the synthesis of acyclic quaternary stereocenters
(Scheme [66]).[57] Acylation of thiopyranone and subsequent alkylation of the resulting β-keto ester
provides substrates 192 for Pd-DAAA. The reaction proceeds with high yield and enantioselectivity in most
cases. Direct extrusion of the sulfur atom was attempted, but hydrogenation of the
allyl olefin of 193a could not be avoided. A hydroboration/oxidation proved useful for avoiding this problem,
as Raney nickel in ethanol afforded the desired acyclic α-quaternary ketone 195 in 94% yield over the sequence. Although this method represents an application of
cyclic structure for creating acyclic stereocenters, it suffers from the drawback
of being limited to the synthesis of α-methyl ethyl ketones only.
Scheme 66 Thiopyranones as precursors of acyclic quaternary stereocenters
Allylic Alkylation with Other Metals
5
Allylic Alkylation with Other Metals
The first reactions within the area of allylic alkylation to demonstrate stereocontrol
in the synthesis of acyclic stereocenters were in Pd-AAA, followed by reactions in
Mo-catalyzed AAA; however, within the last few years, major advances have been made
employing other metals, most notably rhodium and iridium. Copper-catalyzed allylic
substitution reactions[58] nicely complement methods with other metals due to the unique SN2′ pathway and the ability to employ hard nucleophilic precursors. The structures
of ligands used in Mo-, Ir-, and Rh-catalyzed AAA are shown in Figure [2].
Figure 2 Structures of ligands used in Mo-, Ir-, and Rh-catalyzed AAA
5.1
Molybdenum-Catalyzed AAA
The ability to selectively access branched allylic alkylation products is an attractive
feature for creating stereochemistry on electrophilic partners. Among metals that
typically afford branched products selectively (Mo and Ir), selectivity on the nucleophilic
partner was first achieved in Mo-AAA for azlactones[59] and oxindoles.[60]
In the case of branch-selective azlactone alkylation,[58] the reaction proceeded in excellent yield as well as diastereo- and enantioselectivity
(Scheme [67]). Similar levels of yield and selectivity were observed when using a carbonate or
phosphate leaving group; however, the use of lithium base was found to afford higher
branch selectivity compared to sodium or potassium hexamethyldisilazanide. Additionally,
adopting a one-pot protocol for hydrolysis of the azlactone afforded 197 in higher yield compared to the stepwise process.
Scheme 67 One-pot Mo-catalyzed regio-, diastereo-, and enantioselective allylic alkylation
of azlactones followed by hydrolysis
The scope of the reaction showed the best results for cinnamyl methyl carbonate electrophile
(197a–e), with slightly diminished enantioselectivities for heteroaromatic as well as 2-substituted
phenyl systems (197f–j). High selectivity was achieved for aliphatic α-substituents with varied substitution
patterns. It is important to note that linear, achiral electrophiles 196 performed better than branched, racemic allylic carbonates, presumably due to slow
π–σ–π equilibration relative to nucleophilic attack. The mode of asymmetric induction
on the electrophile therefore likely results from differentiation of the enantiotopic
faces of the electrophile in olefin coordination and/or ionization.
With success using azlactones as precursors for acyclic α-tertiary amino acids, the
Trost group studied the isomeric oxalactims 198
[61] as precursors to acyclic α-tertiary alcohols (Scheme [68]).
Scheme 68 Mo-catalyzed regio-, diastereo-, and enantioselective allylic alkylation with oxalactims
Indeed, these pronucleophiles undergo Mo-AAA with regio-, diastereo-, and enantioselectivity.
In addition to the standard cinnamyl-type electrophiles, vinyl-substituted systems
perform as well (199c,d). An efficient hydrolysis of the products directly affords acyclic α-tertiary hydroxy
amides (Scheme [69]).
Scheme 69 Conversion of oxalactim into acyclic α-tertiary hydroxy amides
The synthesis of acyclic quaternary stereocenters was achieved using α-substituted
α-cyano esters 200 as pronucleophiles (Scheme [70]).[62] Interestingly, the process affords acyclic quaternary carbon products in excellent
enantioselectivity in cases where the allyl electrophile bears no stereochemistry
(201a,b). Both alkyl and aryl α-substituents resulted in high regio-, diastereo-, and enantioselectivity.
Additionally, α-tertiary ether product 201h was accessed with good diastereo- and enantioselectivity.
Scheme 70 Mo-catalyzed regio-, diastereo-, and enantioselective allylic alkylation of acyclic
α-cyano esters
5.2
Iridium-Catalyzed AAA
In iridium-catalyzed allylic alkylation, the ability to control stereochemistry on
the nucleophilic partner simultaneous to stereocontrol was first demonstrated by Takemoto
and co-workers[63] in 2003. Since that first report, enantio- and diastereoselective Ir-AAA reactions
have been reported for several systems. This topic has been reviewed elsewhere,[64] so we will focus on two systems that demonstrate the synthesis of acyclic quaternary
carbons on the nucleophilic and electrophilic partners, respectively.
In an impressive display of enantio- and diastereodivergence, Carreira and co-workers
(Scheme [71])[65] were able to demonstrate the ability to access all four stereoisomers in the α-cinnamylation
of aldehydes to afford vicinal acyclic quaternary and tertiary stereocenters. Borrowing
from the success of dual-catalysis in Pd-AAA, they employed cinchona alkaloids with
chiral phosphoramidite ligands. The pseudo-enantiomeric catalyst was employed to access
the other diastereomer. Despite a small match-mismatch effect, all four diastereomers
can be accessed in near perfect selectivity by this method.
The branch-selective Ir-AAA of trisubstituted cinnamyl electrophiles was demonstrated
by the Stoltz group (Scheme [72]).[66] In this reaction, acyl anion equivalent 205 is used to allow for easy access to α-aryl-α-vinyl carbonyl compounds 207. This method demonstrates a new landmark in Ir-AAA, as more highly substituted electrophiles
206 have proven difficult in Ir-AAA. In a single operation, the acyl anion was cleaved
to directly afford carboxylic acid 207. Additionally, conditions were developed to obtain esters and primary and secondary
amides directly from the allylic alkylation products.
Scheme 71 Regio-, diastereo-, and enantioselective α-cinnamylation of aldehydes by dual chiral
amine/chiral iridium catalysis
Scheme 72 Branch selective Ir-AAA of trisubstituted electrophiles by an acyl anion equivalent
nucleophile
The reaction was limited in electrophile scope, as replacement of the methyl group
with ethyl (207j), isopropyl, or butyl groups or the para- or mono-meta-substituted phenyl group with di-meta- (207i) or ortho-substituted phenyl groups (207l) resulted in lower, or no, reactivity, but not lower enantioselectivity. Additionally,
excess electrophile over the acyl anion equivalent was required for high yields. Triethylborane
proved to be crucial for reactivity, as no reaction proceeded in its absence, and
lower yields were observed for lithium bromide as an additive. Presumably, triethylborane
assists by Lewis acid activation in the otherwise difficult ionization reaction. The
mode of asymmetric induction of this reaction is selective ionization from enantiotopic
faces. This was demonstrated by observing no enantioselectivity when using the branched,
racemic electrophile.
5.3
Rhodium-Catalyzed AAA
Exploiting the stereospecificity of the Rh-allylic alkylation, the Evans group was
able to alkylate enantiopure tertiary allylic carbonates with regio- and stereospecificity
(Scheme [73]).[67] An electron-deficient phosphite ligand was found to afford the product with highest
enantiospecificity (denoted as es) compared to more electron-neutral phosphite ligands.[68] All ligands screened showed complete regiospecificity in the reaction. It was reasoned
that the electron deficiency of the ligand facilitated a more rapid intermolecular
alkylation event relative to π–σ–π equilibration. Additionally, more bulky ligands
resulted in significantly diminished enantiospecificity, presumably due to high coordinative
unsaturation of the Rh complex, thus favoring the η1-allyl species necessary for isomerization.
Scheme 73 Regio- and enantiospecific Rh-allylic alkylation of tertiary allylic carbonates with
a cyanohydrin acyl anion equivalent
Scheme 74 Scope of Rh-allylic alkylation of tertiary allylic carbonates with a cyanohydrin
acyl anion equivalent
The reaction scope revealed high regioselectivity in all cases (Scheme [74]). The stereospecificity of the reaction was high for most substrates, although disubstituted
aromatics for Ar1 resulted in diminished enantiospecificity. The electronics of Ar2 showed high stereospecificity for electron-withdrawing substituents at the 4-position,
and diminished specificity for electron-donating groups. This likely results from
a more electrophilic π-allyl species with electron-withdrawing substituents, favoring
a rapid intermolecular reaction with cyanohydrin nucleophile relative to isomerization.
This work was extended to vinyl acyl anion equivalent 215 (Scheme [75]).[69] The product 217 can be chemoselectively reduced to afford the formal product from an aliphatic acyl
anion equivalent addition, affording acyclic α-quaternary α-vinyl ketone 218. This reaction was highlighted for a tertiary allylic carbonate bearing two aliphatic
substituents, with the reaction proceeding with high regio- and stereospecificity.
Scheme 75 Rh-allylic alkylation of tertiary allylic carbonates with a vinyl cyanohydrin acyl
anion equivalent
The Evans group demonstrated the first direct asymmetric alkylation reaction of α-tertiary
nitriles 219 by Rh-catalyzed AAA (Scheme [76]).[70] The process improves step-economy in the α-functionalization of nitriles by generating
the reactive anion in situ, thus side-stepping the synthesis and isolation of racemic
N-silylketenimines. The reaction was shown to proceed smoothly in the absence of additives;
however, improved enantioselectivity was observed in the presence of a crown ether.
High enantioselectivity was demonstrated for substrates bearing methyl, isopropyl,
and benzyl substituents.
Scheme 76 Rh-catalyzed asymmetric alkylation of secondary benzylic nitriles
Additionally, the Evans group have shown a direct intermolecular Rh-catalyzed allylic
alkylation of α-aryl aldehydes (Scheme [77]).[71] Deprotonation with lithium hexamethyldisilazanide affords lithium enolates that
undergo allylic alkylation with allyl benzoate (222) with high enantioselectivity. Although the reaction requires a ligand loading of
40 mol%, ligand L25 is conveniently accessed in one step from BINOL.
Scheme 77 Rh-catalyzed asymmetric allylic alkylation of α-aryl aldehydes
Strangely, control experiments showed that both enolate isomers converge to the same
enantiomer of product (Scheme [78]). The respective silyl enol ethers are converted into the lithium enolates by treatment
with methyllithium. The enantioselectivity in the catalytic reaction is significantly
higher. It is worth noting that the presence of hexamethyldisilazanide is absent in
the control experiments. The fact that both enolate isomers converge to the same product
is all the more surprising considering the small steric differences between the α-substituents
for products 223b, 223c, 223d, 223h, 223i, 223j. It could be envisioned that there is electronic differentiation due to the strong
dependence of enantioselectivity on the 4-substitution of the aromatic substituent.
Scheme 78 Effect of enolate geometry on stereochemical outcome
6
Conclusions and Outlook
An understanding of the mechanistic features of Pd-AAA has allowed for a variety of
methods for the asymmetric synthesis of acyclic tetrasubstituted stereocenters including
tertiary alcohols and ethers, α-tertiary amines, and all-carbon quaternary carbons.
Isoprene monoepoxide and related electrophiles have served as excellent precursors
for forming difficult stereocenters, many of which appear in natural products. Additionally,
the presence of hydroxymethylene and vinyl functional group handles in the resulting
products allow for the synthesis of highly valuable sources of chirality. Stereocontrol
on other electrophiles has been demonstrated in the branch-selective reverse-prenylation
of oxindoles. This remains an exciting precedent in control of acyclic tetrasubstituted
stereocenters on electrophiles; however, reactions with other nucleophiles is still
underdeveloped in this area of research.
Despite the difficulty of achieving asymmetric induction on nucleophiles in Pd-AAA,
important advances have been made over the past 20 years. The ability to control acyclic
geometric elements of enol carbonates has been achieved by many well-designed substrate
syntheses. This strategy works well in conjunction with strategies employing cyclic
precursor such as azlactones, as evidenced by methods for the synthesis of tertiary
alcohols, α-tertiary amines, and all-carbon quaternary stereocenters. The addition
of dual and triple catalysis has enabled the chemist to introduce new modes of stereocontrol
as well as the ability to exploit multiple chiral catalysts within the same reaction
for enhanced stereoselectivity.
Following the success of Pd- and Mo-AAA strategies for stereocontrol in the synthesis
of acyclic stereocenters, recent advances in Ir- and Rh-AAA nicely demonstrate acyclic
stereocontrol on both nucleophilic and electrophilic reaction partners. Undoubtedly,
more efficient strategies in this broad area of research will continue to emerge,
further enriching the options of synthetic chemists for stereocontrol in asymmetric
synthesis.
