Louis C.
Morrill
and
Andrew D.
Smith
*
EastCHEM, School of Chemistry, University of St Andrews, North Haugh, St Andrews, Fife, UK KY16 9ST. E-mail: ads10@st-andrews.ac.uk
First published on 28th May 2014
This tutorial review highlights the organocatalytic Lewis base functionalisation of carboxylic acids, esters and anhydrides via C1-ammonium/azolium enolates. The generation and synthetic utility of these powerful intermediates is highlighted through their application in various methodologies including aldol-lactonisations, Michael-lactonisations/lactamisations and [2,3]-rearrangements.
Key learning pointsThis tutorial review will:(1) Define C1-ammonium and azolium enolates and place them in context with other related methodologies. (2) Outline how these enolates can be accessed from carboxylic acids, esters and anhydrides and why this is desirable. (3) Showcase the utility of these enolates in organic synthesis. (4) Provide clear explanations of the mechanisms of reactions and insights into the origin of stereocontrol. (5) Show the current scope and limitations of this area and define challenges for the future. |
C1-ammonium enolates have traditionally been accessed via the nucleophilic attack of a tertiary amine catalyst on either a pre-formed or in situ generated ketene formed from an acid chloride (Fig. 2).6,8 C1-azolium enolates can be formed in a similar manner using NHCs and ketenes. Alternatively, α-functionalised aldehydes or enals can be employed as precursors in redox catalysed transformations,7 while aliphatic aldehydes are useful C1-azolium enolate precursors in the presence of a stoichiometric oxidant.9 However, the development of alternative routes to C1-ammonium or azolium intermediates that do not employ unstable ketenes, their precursors or a redox transformation is an active field of research. To this end, this review exclusively concentrates on strategies for the generation of C1-ammonium or azolium intermediates from alternative, bench-stable starting materials at the carboxylic acid oxidation level, namely carboxylic acids, esters or anhydrides.
These strategies all rely upon the initial acylation of a tertiary amine or NHC Lewis base catalyst to form a transient acylammonium or acylazolium intermediate from a carboxylic acid, anhydride or ester starting material that can subsequently undergo deprotonation to afford the corresponding C(1)-enolate (Fig. 3). Despite the commercial availability, high stability and low cost of a variety of carboxylic acids, they have been sparsely used as starting materials in Lewis base organocatalysis. Direct reaction of an acid with a Lewis base results in non-productive salt formation, so acid substrates must firstly be derivatised in situ to generate an “activated” electrophilic species (such as a mixed anhydride) that can be readily intercepted by the Lewis basic tertiary amine or NHC, generating an acylammonium or acylazolium intermediate en route to the corresponding enolate. Alternatively, bench stable acid anhydrides or “activated” phenolic esters are directly susceptible to nucleophilic attack by a Lewis base to generate the desired acylammonium or acylazolium intermediate.
In this context, this review will describe the generation of C1-ammonium enolates from carboxylic acid, ester and anhydride starting materials. The use of various tertiary amine Lewis basic catalysts will be demonstrated, including Cinchona alkaloids, pyridine-derived nucleophiles and isothioureas. The subsequent utility of these enolates towards both intra- and intermolecular processes will be discussed and placed in context with their utility in natural product synthesis. The generation of C1-azolium enolates from carboxylic ester starting materials using NHC catalysts will also be described; to date the use of carboxylic acids and anhydrides to prepare azolium anolates has not been described but this clearly offers new opportunities for exploitation.
This process was rendered asymmetric through use of chiral tertiary amines. For example, using O-acetyl quinidine 4 as the Lewis base generates bicyclic β-lactone (1R,2S)-5 from 6 in high ee, providing evidence for a NCAL reaction pathway over a possible thermal [2+2] cycloaddition (Scheme 2).
The proposed mechanism involves initial derivatisation of carboxylic acid 6 with Mukaiyama's reagent and base to form activated ester 7in situ (Scheme 3). Two possible pathways are then proposed: (i) ketene 8 may form via elimination, which can be intercepted by the cinchona catalyst to generate ammonium enolate 9; or (ii) the catalyst may undergo acylation with the activated ester to form acyl ammonium 10 that gives ammonium enolate 9 upon deprotonation. It is likely that the acyl ammonium pathway (path ii) is more prevalent due to ketene 8 being present at very low concentrations at room temperature.
The origin of the stereoselectivity in this process can be explained by the aldehyde approaching the Re face of ammonium enolate 11, opposite to the steric bulk of the quinoline ring, generating 12. Nucleophilic attack on the Si face of the aldehyde gives cis-aldolate 13, which undergoes lactonisation to generate the cis-β-lactone 5 (Fig. 4).
Various quinidine derivatives were tested as catalysts in this NCAL protocol, with changing the nature of the functional group at C9 (a variety of esters, a carbamate and a carbonate) having minimal influence upon observed enantioselectivities (Scheme 4).13 Significantly, through the use of either pseudo enantiomeric O-acetyl quinine 14, or rigidified catalyst 15, the enantiomeric β-lactone (1S,2R)-5 can be prepared in high ee.
The ability of modified Mukaiyama reagents 16–18 to increase reaction yields for the catalytic asymmetric NCAL reaction whilst maintaining high levels of enantioselectivity have also been investigated (Fig. 5).14,15 This was primarily attributed to: (i) improved reagent solubility, allowing reactions to be carried out in less polar solvents such as THF and CH2Cl2 and minimising potential ring-opening reactions promoted by polar media; (ii) less nucleophilic counterions – triflate and tetrafluoroborate – reducing β-lactone decomposition.
In 2011, the Dikshit group demonstrated a catalytic asymmetric aza variant of the NCAL reaction (Scheme 5).16 Using modified Mukaiyama's reagent 19, amino acid derived aldehyde-acids 20 could be converted into either enantiomer of the corresponding bicyclic β-lactone 21 through use of either O-acetyl quinine 14 or O-acetyl quinidine 4, in good yields (51–67%) and high enantioselectivities (83–97% ee). This strategy provides evidence for the wider application of Romo's NCAL strategy towards alternative heterocyclic systems.
Building upon this work, Romo developed a diastereoselective NCAL reaction that allowed the conversion of enantiomerically enriched aldehyde-acid substrates bearing γ- and δ-substituents into bicyclic β-lactones in high diastereoselectivity.17 Amongst several examples, treatment of aldehyde-acid 22 (87% ee) under the standard NCAL conditions using Et3N as the nucleophilic promoter, gives a substrate controlled 2:1 mixture of diastereomeric anti- and syn-β-lactones 23 and 24 respectively (Table 1). Chiral catalysts were next examined to test if catalyst control could override the inherent substrate bias. Use of O-TMSQD results in a reversal of diastereoselectivity (1:7 anti:syn) representative of the mismatched case, while employing O-TMSQN gives exclusively the anti-β-lactone (>19:1 anti:syn) indicative of a matched situation.
The potential to access both tetrahydrofuran- and tetrahydropyran-fused β-lactones via the same NCAL process was also investigated, however this generally resulted in both poor diastereoselectivities and isolated yields.
In a significant advance, the NCAL methodology was extended towards less reactive keto-acid substrates 25.18 Using super-stoichiometric quantities of 4-pyrrolidino pyridine (PPY) 26, the ammonium enolate formed is sufficiently reactive to undergo bis-cyclisation, affording a variety of bicyclic and tricyclic β-lactones 27 in high yield and diastereoselectivity (Scheme 6).
To showcase the utility of this methodology, it was used as the key step in an enantioselective synthesis of (+)-dihydroplakevulin A 28, known to be a precursor to DNA polymerase inhibitor plakeuvulin A 29 (Scheme 7). Under optimised conditions, enantioenriched keto-acid 30 undergoes bis-cyclisation to give bicyclic β-lactone 31 in moderate yield and excellent dr, which is easily converted over two steps into (+)-dihydroplakevulin A 28 through ring-opening with MeOH followed by silyl deprotection.
Further applications of this methodology were disclosed in the synthesis of both (±)-cinnabaramide A and (±)-salinosporamide A 32 (Scheme 8).19 For example, bis-cyclisation of keto-acid 33 (1:1 dr), catalysed by PPY, gives β-lactone 34 in 34% yield and 2:1 dr with the major diastereoisomer subsequently elaborated to 32. The moderate yield of the key bis-cyclisation step is likely due to the difficult functionalisation of a sterically congested α,α-disubstituted carboxylic acid group and again highlights some of the current limitations of this methodology.
Romo subsequently demonstrated that (−)-salinosporamide A 32 could be obtained in enantiomerically enriched form (Scheme 9).20,21 This was achieved by synthesising enantio- and diastereomerically pure keto-acid 33 and utilising modified bis-cyclisation conditions that minimise racemisation of the starting material. Employing mesyl chloride as activating agent at −5 to −10 °C in toluene gave bicyclic β-lactone 35 in 53% yield, 5:1 dr and 90% ee on gram scale. Elaboration of 35 into (−)-salinosporamide A 33 was analogous to the previous synthesis.
In 2011, Romo and co-workers again demonstrated the utility of the NCAL reaction towards natural product synthesis. Bis-cyclisation of (R)-carvone derived keto-acid 35, using TsCl as activating agent and 4-PPY as the nucleophilic Lewis base catalyst, gave the desired β-lactone product 36 in 83% yield and >19:1 dr (Scheme 10).22 Importantly, in the context of natural product synthesis, this NCAL reaction can be carried out on greater than 10 g scale. β-Lactone 36 was elegantly elaborated into molecularly complex (+)-omphadiol 37 over 7 steps in an impressive 33% overall yield, representing the first total synthesis of this product.
In the key bis-cyclisation step towards the synthesis of (+)-omphadiol, Romo and co-workers noted that K2CO3 as base in combination with i-Pr2NEt as “shuttle base” was optimal for achieving a high yield of 36 in a significantly reduced reaction time of 2 h (typically 24 h). This led to the development of a more practical and scalable general procedure for the highly diastereoselective NCAL reaction of keto-acids 38 into bicyclic β-lactones 39 using commercially available p-TsCl as activating agent and DMAP 40 as nucleophilic catalyst (Scheme 11).23
In 2008, Romo and co-workers utilised the NCAL reaction to access a range of tricyclic β-lactones 41 from keto-acids 42 using their standard reaction conditions (Scheme 12).24 Interestingly, these products undergo an unusual dyotropic rearrangement via 1,2-acyl migration upon treatment with Lewis acidic Zn(OTf)2, giving bridged γ-lactones 43 with high stereospecificity.
As a proof of principle, Romo reported the first asymmetric bis-cyclisation of keto-acid 44 using stoichiometric quantities of commercially available (S)-tetramisole hydrochloride 45, forming 46 in 97% ee (Scheme 13). This experiment provided both the first direct evidence for nucleophile involvement in the stereodefining step of biscyclisations with keto-acid substrates and the first demonstration of isothioureas in ammonium enolate chemistry.
In a significant breakthrough, Romo extended this approach to the first catalytic asymmetric NCAL reaction of keto-acids.25 Isothiourea (S)-HBTM 47, developed by Birman as a highly efficient O-acyl transfer agent,26 catalyses the transformation of a range of keto-acids 48 into bi- and tricyclic β-lactones 49 (Scheme 14). Careful optimisation showed that that p-TsCl was the optimal carboxylic acid activating agent and that use of both 0.1 M concentrations and 1 eq. of LiCl Lewis acid co-catalyst were optimal for obtaining 49 in high isolated yields (71–93%) and in excellent enantioselectivities (84 to >98% ee).
A model is provided that explains the observed relative and absolute configuration (Fig. 6). Formation of the (Z)-ammonium enolate, with Li–S chelation results in a bicyclic chair-like transition state 50 that includes ketone activation by Li, rationalising the observed increase in yield.
In a powerful demonstration of utility, this catalytic asymmetric NCAL reaction was later used by the Romo group as a key step towards their asymmetric synthesis of (−)-curcumanolide A and (−)-curcumalactone.27 Desymmetrisation of dione 51 using (R)-HBTM 47 (20 mol%), p-TsCl as activating agent and LiCl as Lewis acid additive gave the desired enantiomer of tricyclic β-lactone 52 in 65% yield, >19:1 dr and 98% ee on gram scale (Scheme 15). This key building block was further elaborated into (−)-curcumalactone 53 and (−)-curcumanolide A 54 in 9 and 8 steps, respectively.
Building upon these precedents, the first application of carboxylic acid derived ammonium enolates in Michael addition processes was demonstrated by the Smith group.28 Using commercially available pivaloyl chloride as activating agent and (S)-tetramisole hydrochloride 45 as nucleophilic precatalyst, a range of enone-acids 55 undergo intramolecular Michael addition-lactonisation to give functionalised indenes 56 or dihydrofuran carboxylates 57 (after in situ ring-opening) with excellent diastereo- and enantiocontrol (up to 99:1 dr, up to 99% ee) (Scheme 16).
This approach was later applied to the synthesis of disubstituted pyrrolidines.29 Enone-acids 58, made in situ via either ozonolysis/Wittig olefination or cross metathesis from 59, were transformed into 2,3- or 3,4-syn-disubstituted pyrrolidines 60 under similar catalytic conditions in excellent diastereo- and enantiocontrol (up to 99:1 dr, up to 99% ee) (Scheme 17).
Notably, the diastereoselectivity of this transformation can be reversed through judicious choice of nucleophilic catalyst. Using modified Mukaiyama's reagent 61 as activating agent and OTMS-quinidine 62 as catalyst, enone-acid 63 gives 3,4-anti-disubstituted pyrrolidine 64 preferentially in modest diastereoselectivity (67:33 dr) with the major diastereoisomer formed in excellent enantioselectivity (99% ee) (Scheme 18).
In the tetramisole promoted cyclisations, the observed stereoselectivity can be explained via pre-transition state assembly 65 in which the enolate oxygen is orientated syn to the sulfur atom of the catalyst, allowing for a stabilising no to σ*C–S interaction (or electrostatic stabilisation). Michael addition then proceeds anti to the stereodirecting phenyl substituent via the enolate Si-face generating the syn-product (Fig. 7). For the OTMS-quinidine catalysed reaction, cyclisation proceeds preferentially via the enolate Re-face (pre-transition state assembly 66), which minimises steric clashes with the ethylene bridge within the quinidine skeleton, giving the anti-diastereoisomer as the major product in high ee.
Fig. 7 Stereochemical rationale for Lewis base promoted intramolecular Michael addition-lactonisations. |
This intermolecular protocol was extended towards alternative electron deficient Michael acceptors. Employing trifluoromethyl enones 71 under closely related reaction conditions generates trifluoromethyl bearing anti-dihydropyranones 72 with high stereocontrol (up to 95:5 dr, up to >99% ee) (Scheme 20).30 Notably, these heterocyclic products could be readily derivatised into those containing CF3-stereogenicity via highly diastereoselective methods including hydrogenation of 73 to form saturated lactone 74 or reduction with LiAlH4 to give lactol 75 with no erosion in enantiopurity.
Kinetic investigations indicated this process to be first order in mixed anhydride 76 and HBTM-2.1 67, whilst being zero order in trifluoromethyl enone 77 (Scheme 21). Increasing the stoichiometry of i-Pr2NEt (up to 8 equivalents) had a negligible effect upon reaction rate, consistent with the rate-determining transition structure being constructed from the catalyst and the mixed anhydride. Importantly, a primary kinetic isotope effect kobsH/kobsD = 3.8 is observed when the reaction is performed using α,α-di-deuterio 4-fluorophenylacetic acid, consistent with deprotonation being rate determining.
Based upon these kinetic observations, a catalytic cycle for the reaction has been proposed that proceeds via initial in situ formation of mixed anhydride 78, which is intercepted by HBTM-2.1 67 to form the corresponding acyl ammonium ion 79 (Scheme 22). Rate-determining deprotonation by pivalate generates the (Z)-ammonium enolate 80 which undergoes stereoselective Michael addition with trifluoromethyl enone 71, followed by intramolecular cyclisation to afford anti-dihydropyranone 72 with regeneration of the catalyst.
Evidence for the presence of an acyl isothiouronium ion such as 79 in the catalytic cycle was provided through its use as a precatalyst. 81 was prepared and isolated by reaction of HBTM 2.1 with the corresponding acid chloride, and used as precatalyst, giving anti-dihydropyranone 82 in identical diastereo- and enantiocontrol (90:10 dr and 99% ee) to that employing HBTM-2.1 67 directly in this protocol (Scheme 23).
Furthermore, the organocatalytic reaction was shown to be stereospecific, with the (Z)-enone 83 affording syn-dihydropyranone 84 in high diastereoselectivity (85:15 dr) with the major diastereoisomer formed in excellent enantioselectivity (99% ee) (Scheme 24). This useful experiment revealed that the either diastereoisomer of the product can be obtained via judicious choice of starting material configuration.
Models explaining the observed stereoselectivity involve the isothiouronium heterocycle adopting a half-chair type conformation with the C(2)-phenyl substituent pseudoaxial to minimise 1,2 steric interactions and the C(3)-i-Pr unit pseudoequatorial (Fig. 8).31 Within the (Z)-ammonium enolate, the oxygen atom preferentially lies syn to the sulfur atom within the isothiouronium ion, allowing a stabilising no to σ*C–S interaction. The reaction proceeds though the diastereomeric transition states shown, giving the observed anti-diastereoisomer (from (E)-enone) or syn-diastereoisomer (from (Z)-enone) in high enantiomeric excess.
In addition to carbon-carbon bond forming reactions, the catalytic asymmetric α-amination of carboxylic acids using isothioureas has been demonstrated.32 HBTM-2.1 67 efficiently catalyses the α-functionalisation of a range of carboxylic acids with N-aryl-N-aroyl diazenes 85 at low catalyst loadings (as low as 0.25 mol%), giving either 1,3,4-oxadiazin-6-ones 86 or N-protected α-amino acid derivatives 87 upon ring-opening with excellent enantiocontrol (up to >99% ee) (Scheme 25). Importantly, non-arylacetic acids could be incorporated in this protocol through the use of a highly electron deficient Michael acceptor and PS-BEMP as base. The N–N bond within the hydrazide products 88 could be readily cleaved using SmI2 to afford bespoke N-aryl-α-aryl glycine building blocks 89 in excellent enantioselectivity.
Carboxylic acid derived ammonium enolates have also been used in Michael addition-lactamisation processes.33 BTM 90 efficiently catalysed the asymmetric formal [4+2] cycloaddition between arylacetic acids 68 and ketimines 91, giving a range of anti-dihydropyridones 92 with good diastereocontrol (typically 85:15 dr) and excellent enantiocontrol (up to 99% ee) (Scheme 26).
Building upon this work, this Michael addition-lactamisation protocol was applied to the synthesis of functionalised pyridines.34 Achiral isothiourea DHPB 93 efficiently catalyses the Michael addition-lactamisation process between (phenylthio)acetic acid 94 and ketimines 95, initially giving dihydropyridones 96 (Scheme 27). These products undergo rapid elimination of thiophenol to give pyridones 97, followed by intramolecular N- to O-sulfonyl transfer upon heating, affording a range of 2,4,6-substituted pyridines 98 in acceptable yields (40–69%) over 3 steps in one-pot. A closely related one-pot protocol was later developed to access a range of trifluoromethyl substituted 2-pyrones.35
Of particular note in this chemistry is that the activating sulfonyl group on the ketimine is transformed into the synthetically useful functional handle (the 2-sulfonate group) in the resulting pyridines. A number of derivatisations showcasing the versatility of this group were demonstrated, including a host of cross-coupling methodologies. This methodology was also applied to the synthesis of 99 – a pyridine with known biological activity as a COX-2 inhibitor (Scheme 28). Pyridine 100 was accessed from ketimine 101 in 53% yield using the optimised reaction conditions and subsequent SNAr with cyclohexylamine afforded 99 in 92% yield.
In 2014, the first intermolecular formal [2+2] cycloaddition involving a carboxylic acid derived ammonium enolate was demonstrated.36 BTM 90 effectively catalyses the annulation of arylacetic acids 68 and N-sulfonyl imines 102, in the presence of tosyl chloride, giving β-lactams 103 with excellent diastereocontrol and high enantiocontrol (up to >95:5 dr, up to 88% ee) (Scheme 29). Alternatively, using HBTM-2.1 67 as catalyst, a range of β-amino esters 105 could be accessed after in situ ring-opening with n-BuLi/MeOH in excellent diastereo- and enantiocontrol (up to >95:5 dr, up to 99% ee).
The isothiourea-mediated functionalisation of 3-alkenoic acids proceeding via an ammonium dienolate has also been disclosed by Smith and co-workers.37 α-Functionalisation of the intermediate dienolate derived from 3-alkenoic acids 106 and HBTM-2.1 67 with N-tosyl aldimines 102 gives a range of β-lactams 107 with modest diastereocontrol (up to 82:18 dr), although each diastereoisomer is formed with excellent enantioselectivity (typically >95% ee) (Scheme 30).
The Chi group later expanded the scope of this reaction to include simple activated alkylacetic esters 111 through the use triazolium precatalyst 112 (30 mol%). In the presence of DBU, anti-dihydropyridines 113 are obtained in excellent yields and stereoselectivities (up to 18:1 dr, up to >99% ee) (Scheme 33).41 Further work by the same group highlighted the ability to organocatalytically functionalise the simplest activated acetic ester (R1 = H in Scheme 33) using similar reaction conditions.42
Chi and co-workers also applied ester-derived C1-azolium enolates towards the synthesis of spyrocyclic oxindoles.43 Achiral NHC precatalyst 114 (30 mol%), in the presence of base, promotes the reaction between activated arylacetic esters 110 and isatin-derived azadienes 115 giving a range of spyrocyclic oxindoles 116 in modest diastereoselectivities (63:37 to 72:28 dr) (Scheme 34). Attempts to render the process asymmetric were thwarted by the low reactivity of the azadienes. The best result was obtained using L-phenylalanine-derived triazolium salt 117, affording δ-lactam 118 in 45% yield with modest diastereo- and enantiocontrol (70:30 dr and 62% ee).
In 2013, Chi reported the first NHC-catalysed asymmetric γ-functionalisation of α,β-unsaturated esters via an azolium dienolate intermediate.44 Triazolium NHC precatalyst 119, in the presence of K2CO3, effectively promotes the γ-activation of α,β-unsaturated esters 120 that undergo addition to hydrazones 121 to give δ-lactam products 122 in high yield and with excellent levels of enantioselectivity (up to >98% ee) (Scheme 35). Using slightly modified reaction conditions, γ-substituted and β,β′-dialkyl substituted α,β-unsaturated esters could also be used in this protocol, giving the corresponding δ-lactams in moderate to good yields and enantioselectivity.
A proposed catalytic cycle for this process involves initial formation of acyl azolium 123, which undergoes γ-deprotonation to form the key C1-azolium dienolate 124 (Scheme 36). Subsequent stereoselective nucleophilic attack towards hydrazone 121 gives NHC-bound adduct 125 that rapidly collapses to give δ-lactam 122 with regeneration of the catalyst.
The only asymmetric use of C(1)-ammonium enolates from activated esters has been through the demonstration of an isothiourea-catalysed asymmetric [2,3]-rearrangement of allylic ammonium ylides.46 A range of allylic quaternary ammonium salts 129 (either isolated or made in situ), undergo BTM 90-catalysed [2,3]-rearrangement in the presence of i-Pr2NH as base and HOBt additive to afford a range of syn-α-amino acid derivatives 130 in excellent diastereo- and enantiocontrol (up to >95:5 dr, up to >99% ee) (Scheme 38).
This journal is © The Royal Society of Chemistry 2014 |