Understanding divergent substrate stereoselectivity in the isothiourea-catalysed conjugate addition of cyclic α-substituted β-ketoesters to α,β-unsaturated aryl esters

The development of enantioselective synthetic methods capable of generating vicinal stereogenic centres, where one is tetrasubstituted (such as either an all-carbon quaternary centre or where one or more substituents are heteroatoms), is a recognised synthetic challenge. Herein, the enantioselective conjugate addition of a range of carbo- and heterocyclic α-substituted β-ketoesters to α,β-unsaturated aryl esters using the isothiourea HyperBTM as a Lewis base catalyst is demonstrated. Notably, divergent diastereoselectivity is observed through the use of either cyclopentanone-derived or indanone-derived substituted β-ketoesters with both generating the desired stereodefined products with high selectivity (>95 : 5 dr, up to 99 : 1 er). The scope and limitations of these processes are demonstrated, alongside application on gram scale. The origin of the divergent substrate selectivity has been probed through the use of DFT-analysis, with preferential orientation driven by dual stabilising CH⋯O interactions. The importance of solvation with strongly polar transition-states is highlighted and the SMD solvation model is demonstrated to capture solvation effects reliably.


Introduction
The ability to selectively control the relative and absolute conguration in compounds containing vicinal stereogenic centres, where one is tetrasubstituted (dened herein as a stereogenic carbon that does not have a proton as a substituent) is of signicant interest.Despite many advances in the development of synthetic methods in this area, 1,2 this still remains a meaningful challenge due to the generally low reactivity of suitable precursors caused by steric encumbrance.Among methods developed in this area, the enantioselective conjugate addition of a-substituted b-ketoesters has been widely explored based on the ability to generate up to two contiguous stereocentres, one of them quaternary (Scheme 1A). 3,4As representative examples, this strategy has been applied to the stereoselective addition to nitro-olens, [5][6][7] di-t-butyl azodicarboxylate, 8,9 vinyl ketones, 10 and propargyl alcohols. 11For example, Sodeoka et al. showed that enantioselective addition of various a-substituted b-ketoesters to enones proceeded in good yields with excellent enantioselectivity using Pd-BINAP derived enolates (Scheme 1B). 12Alternatively, the Pd-catalysed allylic alkylation of Morita-Baylis-Hillman adducts has also been developed with b-ketocarbonyls. 13Despite these precedents, stereoselective conjugate additions of a-substituted bketoesters to a,b-unsaturated esters is rare due to the low inherent electrophilicity of these Michael acceptors.
The use of isothioureas as Lewis basic organocatalysts has expanded remarkably over the last een years.Building upon the pioneering work of Birman who demonstrated their rst use as acyl transfer catalysts for kinetic resolution processes, these readily prepared chiral Lewis bases have since been harnessed for a plethora of transformations. 14This broad range of reactivity incorporates the generation and reactivity of acyl ammonium I, 14 a,b-unsaturated acyl ammonium II 15 and C(1)ammonium enolate III 16 intermediates that exploit a key 1,5 S/O chalcogen bonding interaction (n O to s * S-C ) to achieve stereocontrol (Scheme 2A).Of relevance to this work, the generation and exploitation of a,b-unsaturated acyl ammonium intermediates was initially achieved from acid chloride and anhydride starting materials.For example, isothioureas (2S,3R)-HyperBTM 1 and (S)-HBTM 2 have been employed in cascade processes with b-ketoesters 3 17 and g-ketomalonates 4, 18 that involve initial Michael addition followed by intramolecular cyclisation to promote catalyst turnover, generating dand blactones 5 and 6 (Scheme 2B).Related conjugate addition/ cyclisation processes with acyl benzazoles, 17,19 aminothiophenols, 20 heterocyclic nucleophiles, 21 as well as Rh-metallacycles has been reported. 22Joint synthetic and computational analysis of the factors leading to stereocontrol and chemoselectivity in acyl benzazole addition has led to enhanced understanding of these processes. 23n recent work, the use of a,b-unsaturated electron-decient aryl esters has been shown to offer a potential (albeit limited) solution to the recognised recalcitrance of a,b-unsaturated esters to enantioselective conjugate addition (Scheme 2C).Acylation of these species with an isothiourea generates an a,b-unsaturated acyl ammonium ion pair, with the enhanced electrophilicity of this intermediate species (compared to the parent ester) allowing for effective catalysis.Furthermore, the electron-decient phenoxides generated in situ can be exploited to promote catalytic turnover, while their multifunctional nature can be exploited through acting as both Brønsted base and Lewis base. 24Initially demonstrated in the enantioselective conjugate addition of nitroalkanes, 19 it has since been exploited using heterocyclic pronucleophiles, 21,25 malonates, 26 benzophenone imines, 27 as well as co-operative Pd-isothiourea promoted cascade reactions. 28onjugate addition and lactonisation processes using acylbenzothiazole and benzoxazole pronucleophile derivatives with a,b-unsaturated para-nitrophenyl esters has also been investigated synthetically. 29 Building upon these previous studies, in this manuscript the enantioselective conjugate addition of a range of carbo-and heterocyclic a-substituted b-ketoesters to a,b-unsaturated aryl esters using the isothiourea HyperBTM 1 as a Lewis base catalyst is reported.Interestingly, divergent diastereoselectivity is observed with the use of either cyclopentanone-derived or indanone-derived substituted b-ketoesters, with both generating the desired stereodened products with high selectivity (>95 : 5 dr, up to 99 : 1 er, Scheme 2D).The origin of the divergent substrate selectivity has been probed using DFT-analysis at the M06-2X SMD /def2-TZVP//M06-2X SMD /def2-SVP level.

Scope and limitations
To decipher these results, competition experiment 1 (Scheme 6C) took a 50 : 50 ratio of nucleophiles 18 and 7 in the addition to fumarate 26, giving an 86 : 14 ratio of products 19 and 13.On the assumption of irreversible nucleophilic addition, this is consistent with the rate of addition of methyl 1-acetyl-3-oxoindolinone-2carboxylate 18 being greater than that of ethyl 2-oxocyclopentane-1-carboxylate 7.In addition, we considered previous work that showed both isothioureas and NBu 4 OPNP can promote (Z)-to (E)-Scheme 6 Effect of (Z)-or (E)-enoate configuration on stereoselectivity.All ers determined by HPLC analysis on a chiral stationary phase.All product drs determined by 1 H NMR analysis of the crude reaction mixture.Yields are isolated yields after purification.
enoate isomerisation, 19,28a,34 and that isomerisation of CF 3substituted (Z)-enoates occurred at a signicantly faster rate than that of maleates.Application to this case (Scheme 5D) would involve N-acylation of HyperBTM 1 with (Z)-24 or (Z)-25 to give the corresponding (Z)-a,b-unsaturated acyl ammonium ion pair 27.Subsequent reversible conjugate addition of para-nitrophenolate (or HyperBTM 1), followed by bond rotation and elimination, will lead to the thermodynamically favoured (E)-enoate via the corresponding (E)-a,b-unsaturated acyl ammonium ion pair 27 (Scheme 6D).Bringing this together, the observed experimental results in Scheme 6A and B can be understood based on the relative rates of nucleophilic addition of the b-ketoesters 18 and 7 to the (E)-or (Z)-a,bunsaturated acyl ammonium intermediate 28 or 27, alongside competitive (Z)-to (E)-enoate isomerisation.Using ethyl 2oxocyclopentane-1-carboxylate 7 the rate of conjugate addition is slow relative to (Z)-to (E)-isomerisation, resulting in the same product enantiomer using either (E)-or (Z)-enoate as starting material.However, using methyl 1-acetyl-3-oxoindolinone-2carboxylate 18, the conjugate addition to an intermediate (Z)-a,bunsaturated acyl ammonium species must occur at a faster rate than isomerisation of maleate 24, giving preferentially the product antipode (2R,2 ′ R)-23 compared to the reaction with fumarate 26.Using this argument, the effectively racemic nature of product 19 from CF 3 -substituted (Z)-enoate 25 and 18 is assumed to arise from approximately identical rates of (Z)-to (E)-isomerisation and conjugate addition to the (Z)-a,b-unsaturated acyl ammonium species.To test this hypothesis, competition experiment 2 was designed and carried out (Scheme 6E).Using a 50 : 50 mixture of CF 3 -substituted (E)-enoate 8 and (Z)-enoate 25 in the reaction of 18 was predicted to give an approximate 75 : 25 mixture of product (2S,2 ′ S): (2R,2 ′ R)enantiomers; in practice product 19 was isolated in 55% yield (>95 : 5 dr, 74 : 26 er), supporting this hypothesis.

Proposed mechanism
Based on previous studies and the observations reported herein, 19,26,28a,29,35 a general and simplied catalytic cycle for this transformation can be proposed (Scheme 7).7][38][39][40][41][42][43] Deprotonation of the b-ketoester by the released p-nitrophenoxide, followed by subsequent stereoselective conjugate addition to the a,b-unsaturated acyl isothiouronium intermediate 28 in the assumed stereodetermining step gives enolate 29.Subsequent protonation, presumably by the generated p-nitrophenol, gives acyl-isothiouronium species 30.The aryloxide subsequently effects catalyst turnover to afford pnitrophenyl ester 31, which upon treatment with benzylamine gives the corresponding amide product.
While this general mechanism accounts for the connectivity observed within the products, the divergent stereoselectivity observed with choice of pronucleophile was further investigated by DFT analysis.

DFT computation
5][46][47][48][49][50] Following the pioneering computational work by Cheong et al. 23 and Wang et al. 30 who identied the Michael addition (C-C bond formation) between a cationic HyperBTM-N-acyl complex and a C nucleophile (Scheme 6) as the stereodetermining step in isothioureacatalysed conjugate addition cyclisation reactions, our computational modelling concentrated on that step.An extensive transition-state (TS) search was conducted, exploring the conformational freedom of the approaching nucleophile (see ESI †).The diastereoselectivity for the reaction was evaluated as the difference in free energy (DD ‡ G) of the lowest TS leading to each of the diastereoisomeric products.From this extensive TS search, we found that the facial selectivity imposed by the catalyst was consistent with previous work, driven by the 1,5 S/O chalcogen interaction (n O to s * S-C ) between the acyl O and catalyst-derived S atom that provides a conformational lock, with reaction through the s-cis conformation. 23,41This places the phenyl substituent perpendicular to the plane of the catalyst, with conjugate addition anti-to this unit to the si-face of the a,b-unsaturated isothiouronium intermediate, consistent with the high enantioselectivity and absolute conguration observed.
Between the possible approaches towards the accessible siface of the isothiouronium ion intermediate, substrate facial selectivity is driven by a variety of non-covalent interactions between the nucleophile and a,b-unsaturated isothiouronium intermediate.For the addition of 7 to give preferentially 9, dual non-classical CH/O stabilising interactions between the acidic + NC-H derived from the catalyst and the ketoester of the nucleophile are observed, accounting for the preferential orientation of the major and minor TS (86 : 41 dr, Fig. 1).Experimentally, benzannulation of 7 to give ethyl 1-oxo-2,3- with an observed reduction in dr (72 : 28 dr).This is qualitatively reproduced computationally (55 : 45 dr), with the major TS again stabilised by CH/O interactions between the benzylic + NC-H derived from the catalyst and nucleophile.In this case, the minor diastereoisomeric TS for 16 does not contain CH/O interactions, but instead exhibits a stabilising cation-p interaction due to the orientation of the benzannulated ring of the nucleophile with the isothiouronium ion within the acylated catalyst (Fig. 1a and b).This interaction slightly stabilises the minor TS, reducing the energy difference between the two diastereoisomeric TS and lowering the dr.This subtle interplay between cation-p and CH/O interactions is fully in line with ndings by Wang et al. for a related Michael addition reaction. 30he divergent selectivity observed with indoline derived substrate 19 was also investigated.The computed major TS aligns with experiment, with the minor TS lacking in CH/O interactions.Instead, the minor TS features p-type interactions between the substrate acyl group and the isothiouronium acylated catalyst, analogous to the cation-p interaction of 16 (see Fig. 1b).
Importantly, the choice of implicit solvent model was found to dramatically impact the ability of the computational method to reproduce experimental results.These models allow for simple representation of the solvent as a continuous dielectric but can become unreliable in the case of large dipole moments or charges.The chosen transition states are strongly zwitterionic composed of an anion derived from deprotonation of the b-ketoester and a cationic isothiouronium derived Michael acceptor.
Preliminary calculations using the Polarisable Continuum Model (PCM) 51,52 employed in previous studies 26,50 led to the incorrect prediction of the minor diastereoisomer of 16 (33 : 64 dr), arguably due to the poor treatment of solvation.Two alternate approaches identied the correct major diastereomer with PCM: explicit micro-solvation (85 : 15 dr, Fig. 1b) and increased scaling of the solvent cavity ($1.15 van der Waals radii, Fig. 1c).Micro-solvation adds an explicit solvent molecule to the DFT calculation, allowing for more accurate interaction with the solvent than just an implicit model.We nd stabilising p-type interactions with the conjugated DMA solvent (Fig. 1c) and calculate an improved 85 : 15 dr, in closer agreement with the experimental dr (72 : 28 dr).Similarly, the dr can be improved by scaling the solvent cavity, which in turn is constructed from the van der Waals radii (Fig. 1d).With a larger cavity, there is a greater separation between the large molecular dipole and the dielectric medium which leads to a reduction in coulombic interaction.This is larger for the minor TS than the major TS due to the larger dipole moment (m = 29.54 and 19.84 D, respectively).Both approaches introduce a greater separation between the highly charged or dipolar regions of the TS and the dielectric medium and reduce the relative overstabilisation of the minor diastereomeric TS.Unfortunately, both approaches are undesirable from a methodological point of view.Explicit microsolvation requires vast exploration of solvation sites and conformers, ultimately calling for extensive molecular dynamics simulations to sample the relevant minima in an ensemble, and articially tweaking parameters of a solvent model may reduce its predictive power and general applicability.
Pleasingly, the Solvation Model based on Density (SMD) correctly reproduces the major diastereoisomer without scaling the solvent cavity or introducing micro-solvation.This model includes parameterisation to account for some of the dispersive interactions between the solute and solvent, mimicking the increased accuracy brought about by micro-solvation, whilst using a smaller solvent cavity than PCM.This model appears to help in stabilising the major diastereoisomer, presumably due to the parameterised dispersive interactions in its construction even with a reduced solvent cavity compared to PCM.

Conclusions
In conclusion, the enantioselective conjugate addition of a range of carbo-and heterocyclic a-substituted b-ketoesters to a,b-unsaturated aryl esters using the isothiourea HyperBTM 1 as a Lewis base catalyst has been demonstrated.Divergent diastereoselectivity is observed through judicious choice of pronucleophile, with either cyclopentanone-derived or indanonederived substituted b-ketoesters generating divergent stereo-dened products with high stereoselectivity (>95 : 5 dr, up to 99 : 1 er).The scope and limitations of these processes are demonstrated, alongside application on gram scale.Competitive isomerisation alongside conjugate addition is observed with (Z)-enoates.The origin of the divergent stereoselectivity has been studied using DFT and can be attributed to a variety of weak non-covalent interactions between the a,b-unsaturated acyl isothiouronium intermediate and the incoming nucleophile.These weak attractive interactions modulate the steric repulsion from the stereodirecting C(2)-phenyl substituent within the catalyst-substrate a,b-unsaturated acyl isothiouronium intermediate and need to be accounted for by the computational methodology.Because the key TSs are highly polar, an adequate treatment of solvation effects in the model is instrumental.Building upon this work and the insights gained from DFT analysis, further applications of related reaction processes are currently under investigation in this laboratory. 53 Scheme 2 (A) Isothioureas and common intermediates derived from N-acylation; (B) established isothiourea-promoted Michael additionlactonisation; (C) isothiourea-catalysed addition to a,b-unsaturated aryl esters; (D) this work: divergent substrate dependent selectivity in addition to a,b-unsaturated aryl esters.

2. 2 . 1
Conjugate additions of ethyl 2-oxocyclopentane-1carboxylate to a,b-unsaturated p-nitrophenyl esters.Further work considered the scope and limitations of this process initially through changing the a,b-unsaturated para-nitrophenyl ester component of the reaction (Scheme 3).Under the developed conditions the incorporation of an electron withdrawing b-substituent was necessary to promote reactivity, with bperhalogenated substrates giving the corresponding CF 2 Cl, CF 2 Br and C 2 F 5 variants 10-12 respectively in synthetically useful yield and excellent enantioselectivity (>95 : 5 dr, 99 : 1 er).

aScheme 3
Scheme 3 Conjugate addition of ethyl 2-oxocyclopentane-1carboxylate.All product ers were determined by HPLC analysis on a chiral stationary phase.All product drs were determined by1 H NMR analysis of the crude reaction mixture.Following reaction conditions were applied unless stated otherwise: 0.2 mmol 7, 0.2 mmol pnitrophenol ester, 0.01 mmol 1 were stirred in 0.8 mL DMA at rt for 16 h.Subsequently, 0.2 mmol BnNH 2 was added and stirred at rt for 1 h.(a) Combined yield of inseparable diastereomers.

2 . 2 . 4
Scheme 4 Conjugate addition of carbo-and heterocyclic asubstituted b-ketoesters to a,b-unsaturated p-nitrophenyl esters.All product ers were determined by HPLC analysis on a chiral stationary phase.All product drs were determined by 1 H NMR analysis of the crude reaction mixture.Following reaction conditions were applied unless stated otherwise: 0.2 mmol ketoester, 0.2 mmol 8, 0.01 mmol 1 were stirred in 0.8 mL DMA at rt for 16 h.Subsequently, 0.2 mmol BnNH 2 was added and stirred at rt for 1 h.(a) 59 : 41 er of minor diastereomer, combined yield of inseparable diastereomers.(b) 97 : 3 er of minor diastereomer, combined yield of inseparable diastereomers.(c) 96 : 4 er of minor diastereomer, diastereomers isolated yields (major/minor): 51%/22%.(d) 95 : 5 er of minor diastereomer, combined yield of inseparable diastereomers.Scheme 5 Conjugate addition of methyl 1-acetyl-3-oxoindolinone-2-carboxylate to a,b-unsaturated p-nitrophenyl esters.All product ers were determined by HPLC analysis on a chiral stationary phase.All product drs determined by 1 H NMR analysis of the crude reaction mixture.Following reaction conditions were applied unless stated otherwise: 0.2 mmol 18, 0.2 mmol p-nitrophenol ester, 0.01 mmol 1 were stirred in 0.8 mL DMA at rt for 16 h.Subsequently, 0.2 mmol BnNH 2 was added and stirred at rt for 1 h.(a) Reaction on a gram scale with purification by crystallisation.(b) 76 : 24 er of minor diastereomer, combined yield of inseparable diastereomers.(c) 72 : 18 er of minor diastereomer, combined yield of inseparable diastereomers.

Fig. 1
Fig. 1 (a) DFT analysis of the most stable major and minor diastereomeric transition states.Important interactions are highlighted, with dual CH-O interactions present in all major TS.M06-2X SMD /def2-TZVP//M06-2X SMD /def2-SVP free energies (DG 298 ) are shown in kcal mol −1 and selected hydrogen atoms have been removed for clarity.(b) Non-covalent interaction of key cation-p stabilisation of minor diastereomeric TS of 16.(c) Explicit micro-solvation with p-type interaction with conjugated DMA solvent (with PCM solvation model).(d) Solvent cavity as a sum of van der Waals radii and graph of the selectivity (DDG 298 ) as a function of the scaling parameter.Asterisks indicate the default value of a for each solvent model.

Table 1
Initial optimisation a