RESOLUTION AND DESYMMETRISATION
Chapter I. ‘A Novel Racemate Resolution’. This describes a novel resolution strategy as applied to racemic α-amino acids in the solid state. The strategy is based on the possibility that second order asymmetric transformations (SOAT) would be more likely in the case of achiral molecules that form chiral crystals (i.e. a non- centrosymmetric space group).1 In such cases, a fundamental requirement of SOAT – that the molecules racemise in solution prior to crystallization – is obviated. Furthermore, the resulting enantiomerically-enriched crystals may be employed to effect a solid-state kinetic resolution of a different racemate (composed of chiral molecules). This strategy was explored with crystalline succinic anhydride (1, Scheme 1), which not only exists in a non-centrosymmetric space group (P212121) but also possesses reactive functionality to effect the resolution step.2
Thus, a finely-ground mixture of 1 (0.5 eqiv.) and a racemic α-amino acid (2, 1.0 eqiv.) was heated at ~ 70 oC over ~ 5 h without solvent. The resulting N-succinoyl derivative (3) was separated from the unreacted 2, which was found to possess significant levels of optical purity (typically ~ 70%). The strategy was applied to several common α-amino acids, the results being summarized in Table 1. These results, apart from establishing ‘proof-of-concept’ and the viability of the resolution strategy, indicate that crystalline succinic anhydride (1) is enantiomerically enriched as originally hypothesized.
Chapter II. ‘Enantiospecific Alkylation and Desymmetrisations’. This deals with two enolate-mediated strategies of asymmetric synthesis: one describes approaches towards the alkylation of the stereogenic centre in benzoin without loss of stereogenicity (Section A), and the other the desymmetrisation of a meso tartarate derivative with a chiral base catalyst (Section B).
Section A. This describes exploratory studies aimed at achieving the enantiospecific α-alkylation of optically-active benzoin (4, Scheme 2) via its enolate anion 5. The strategy depends on the possibility that 5 would exist in atropisomeric forms, because of steric interactions between the vicinal phenyl groups. (This is indicated in the crystal structure of the analogous enediol carbonate derived from racemic 4.)3 In such a case, remarkably, 5 would be chiral, despite its planar enediolate core! Thus, possibly, the configurational chirality in 4 (by virtue of the C2 stereogenic centre) would be transformed to the helical chirality in 5 (by virtue of the atropisomerism). Furthermore, enantioface-selective alkylation of 5 with achiral alkylating agents would, in principle, be possible.
Preliminary studies were then directed towards establishing that controlled deprotonation of optically-active 4, followed by the protonation of the resulting enediolate 5, leads back to the original 4. (+)-Benzoin (4) was prepared via resolution,4 and deprotonated with KH in THF.5 The resulting enediolate (5) was neutralized with acetic acid at -70 oC/THF to recover 4, but with insignificant levels of optical activity (e.e. ~ 12%). The results possibly indicate that ortho-substituted benzoin analogs may show greater retention of chirality upon deprotonation, as the racemisation of the enediolate atropisomers would be suppressed by steric hindrance between the aryl moities.
Section B. This describes studies directed towards the catalytic desymmetrisation of meso dimethyl tartarate (6, Scheme 3). The strategy involves the formation of the acetonide derivative 7 and its regioselective α-deprotonation with a chiral base catalyst. The enantioface-selective protonation of the resulting enolate (8) would lead to the chiral analog 9. The overall sequence offers a possible alternative to catalytic deracemisation, which is normally unviable for thermodynamic reasons.6
The above strategy hinges on the meso derivative 7 being thermodynamically less stable than the enantiomeric 9, which would thus be favoured at equilibrium. In fact, this is likely as the eclipsing interactions between the syn ester moieties in 7 would be relieved in 9, in which the ester moieties are anti.
However, deprotonation of 7 at the other α position would compete to varying extents, depending on the selectivity induced by the chiral base. At total equilibrium, the sequence would occur via deprotonation at both α sites at equal rates, and no net optical induction would be observed. (This is a thermodynamic requirement via the principle of microscopic reversibility.) Thus, the success of the above strategy depends on stalling the deprotonation-protonation sequence at a quasi-equilibrium stage involving only one of the enantiomers (9).6
The other operational requirement was the compatibility of the pKa’s of 7 and the chiral base employed: too low a pKa of the base would result in inefficient deprotonation and slow overall rate, but a high pKa would generate a large quantity of the enolate 8 at equilibrium. After due consideration, the lithiated chiral fluorene derivative 11 (pKa ~ 22) was chosen as the chiral base catalyst [11 was prepared from fluorene (10) as indicated].
Treating 7 with 0.2 equivalent of 10 in THF at -65 oC over 2 h, led to the formation of a mixture of 7 and 9 in a 45:55 ratio (isolated in 85% total yield). Chromatographic separation of the mixture led to the isolation of pure (+)-9, which was identified spectrally; it was found to possess [α]D24 = +21.84 (c 1.0, CHCl3), corresponding to e.e. = 64%. (This implies the indicated (4S, 5S) configuration for 1, 3-dioxolane 9, as previously reported.)7 These results, despite the moderate e.e. levels obtained, indicate the viability of the above catalytic desymmetrisation strategy, bearing in mind the mechanistic ambiguities mentioned above.
SYNTHESES OF ALDEHYDES AND AMINO ACIDS
Chapter III. ‘An Asymmetric Synthesis of Aldehydes’. This describes an oxazoline approach to the synthesis of chiral aldehydes. The oxazoline methodology for the synthesis of homochiral α-alkylated carboxylic acids is well known,8 and it was of interest to adapt this to the synthesis of the corresponding aldehydes. Essentially, it was envisaged that the reaction sequence could be diverted towards aldehydes via reduction of the alkylated oxazoline intermediate (Scheme 4).
Thus, 2-ethyl-4(S)-methoxymethyl-5(R)-phenyl-1,3-oxazoline (12) was deprotonated with lithium diisopropylamide in THF, and the resulting anion treated with various alkyl halides, in the reported manner.8 The resulting alkylated product (13) was N-methylated with MeI in refluxing MeNO2 over 6 h, to obtain the quaternary salt 14. This was reduced with NaBH4 in MeOH to obtain the expected N- methyl oxazolidine 15, which was hydrolyzed in refluxing aqueous oxalic acid to the free aldehydes 16. These were isolated in moderate yields and e.e. values as shown.
Chapter IV. ‘A Darzens Route to α-Amino Acids’. This describes a novel route to α-amino acids, based on the classical Darzens glycidic ester synthesis.9 In this approach (Scheme 5), the glycidic ester (19) was prepared from benzaldehyde (17) and t-butyl bromoformate (18), with KOH in THF as base, and tetrabutylammonium bromide (TBAB) as phase transfer catalyst.9b The oxirane ring in 19 was cleaved via nucleophilic attack with an amine (20), to furnish the two regio-isomeric hydroxy- amino acids (21) and (22). Generally, the β-hydroxy-α-amino acid product (21) predominated over the α-hydroxy-β-amino acid product (22), the two being separated chromatographically. The hydroxyl group in 21 was reductively cleaved via its xanthate derivative (23), by refluxing it in toluene with AIBN (10 mol %) over 4 h. The resulting α-amino acid derivatives (24) were obtained in moderate yields (< 60
%) upon chromatographic purification. (The β-amino analog 22, would lead to the corresponding β-amino acid, but this was not pursued further.)
This strategy lends itself to creating structural diversity at the β-centre in the α- amino acid, drawing upon the wide scope of the well-established Darzens condensation reaction. Also, the introduction of the amino moiety is facilitated by the enhanced reactivity at the α-centre of the oxirane ring in the glycidic ester (19), presumably for both electronic and steric reasons.||en_US