Recent advances in enzymatic and chemical deracemisation of racemic compounds.

Deracemisation of racemic compounds is still the most important strategy to produce optically pure compounds despite many recent advances in asymmetric synthesis. Especially deracemisation approaches that give rise to single enantiomers are preferred, which can be achieved either by invoking organocatalysts, metal complexes or enzymes - leading to dynamic kinetic resolution - or by applying other deracemisation techniques based on stereoinversion or enantioconvergence. In this tutorial review, we will provide an overview of new, recently developed approaches that lead to the important goal of creating organic compounds of single chirality.


Introduction
The synthesis of enantiomerically pure compounds remains a 20 challenge in modern organic chemistry due to the importance of chirality in biological processes. The majority of biomolecules existing in living organisms is chiral and consist of only one of the two possible enantiomeric forms. As a result, molecular interactions of a racemic substrate with these biomolecules will 25 always be different for one enantiomer than the other. This plays a particularly important role in the pharmaceutical industry, but also in other industrial sectors such as the food, and flavours and fragrances industry. As a result, it is crucial for those industries to produce many of their products (e.g. drugs, food additives, 30 vitamins) as single enantiomers.
The first person being able to separate a racemic mixture into the two enantiomers was Louis Pasteur, who as early as in 1849 physically separated tartaric acid crystals of opposite handedness. Evcer since, the separation of a racemate into the two 35 enantiomers -commonly known as resolution -has been the most prominent way to separate two enantiomers in numerous applications. This resolution is often performed through addition of an enantiopure (auxiliary) reagent leading to the preferred formation of one diastereoisomer over the other, which on the 40 basis of different physical properties can then be separated. Removal of the auxiliary compound produces the desired enantiopure product. Another commonly used technique is treatment of a racemate with an enzyme, which recognizes one enantiomer and converts it into a different product, while the 45 other enantiomer remains untouched. At this point, the two enantiomeric forms have different physical properties and can be readily separated. These kinetic resolutions can be carried out in small scale laboratory experiments, but also in large scale industrial processes. The limitation of such types of 'classic' 50 kinetic resolutions (KR), however, are that both enantiomers can only be obtained in a maximum theoretical yield of 50%. The wish of the chemical industry to reduce costs and waste in the production of chiral building blocks led chemists to develop novel resolution procedures of racemic mixtures that proceed 55 beyond the 50%. This is called deracemisation, which is defined as the conversion of a racemate into one of the corresponding enantiomers in theoretically 100% chemical yield and 100% optical purity without isolation of intermediates.
The existing different deracemisation techniques have been 60 classified by the mechanism of action as proposed by Faber (Scheme 1). 1 Inspired by the fact that deracemisation strategies continue to be discovered, we herewith provide an overview of deracemisation methods, based on the classification by Faber, with a strong focus on new and innovative approaches that have 65 appeared in recent literature.

Dynamic kinetic resolution (DKR)
Dynamic kinetic resolution (DKR) is the most widely used technique for deracemisation. It is a combination of 'classic' 5 kinetic resolution, in which a racemate (SS and SR) is resolved by chemical transformation of one single enantiomer into a product (PR), with in situ racemisation of the 'unwanted' enantiomer (SS) of the starting material (Scheme 1). Many examples have been revealed combining metal-catalysed racemisation with enzymatic 10 kinetic resolution. Since such chemoenzymatic DKRs have already been extensively reviewed, 2,3 in this review, we will focus on DKR processes employing either organocatalysis, transition metal catalysis or enzymatic transformations. DKR of atropoisomers and so-called attrition-enhanced deracemisation 15 will be also discussed.

Organocatalysed DKR
Organocatalysis has gained much interest from organic chemists in recent years. Catalysis proceeds without metals using small organic molecules instead, thereby contributing to the sustainable 20 (green) nature of the given conversions. Not unimportantly, the small chiral molecules that are generally employed as organocatalysts are often readily available at relatively low cost.
In recent years the group of Birman reported several examples of DKR catalysed by small organic molecules. In 2010, they 25 disclosed the use of benzotetramisole (BTM) as an enantioselective acyl transfer catalyst in the DKR of azlactones. 4 In initial experiments, different amidine-based catalysts were screened in the alcoholysis of compound 1 (R = Me) using an equimolar proportion of catalyst and benzoic acid. Benzoic acid 30 appeared an essential promotor in this reaction, since no conversion was observed in its absence. The best results in terms of conversion and ee were obtained using (S)-BTM 3 as catalyst (Scheme 2). Under these slightly acidic conditions, the azlactone 1 racemizes via the corresponding enol, while the chiral acyl 35 transfer catalyst causes the sterically congested alcohol nucleophile (4) to react faster with the (S)-than with the (R)-enantiomer. With less bulky nucleophiles, the acylation was less selective. 40 Scheme 2 DKR of azlactones via BTM catalysis Further variation of reaction conditions showed that catalyst loadings as low as 2 mol% were still effective though requiring longer reaction times. Lower temperatures proved to be detrimental for enantioselectivity , while elevated ones resulted in 45 higher rates but with the same enantioselectivity. The applicability of this system to other azlactones was studied using a series of different R groups under optimal conditions. Good yields and ee`s (around 90%) were obtained except for isopropyland 2,4-dimethoxyphenyl-substituted substrates which proved to 50 be resistant to alcoholysis. One year later, the same group published DKR of azlactones using chiral Brønsted acids. 5 Perhaps inspired by the necessity of a protic acid in the previous example, they reasoned that a strong chiral Brønsted acid might also be able to sufficiently activate 55 azlactones to promote direct alcoholysis without intermediacy of a nucleophilic catalyst. Several 3,3`-disubstituted BINOL-derived phosphoric acids were screened as catalysts for the reaction of 2,5-diphenylazlactone with benzyl alcohol giving the most promising results (91% conversion, 63% ee) with 9-anthryl- 60 substituted BINOL 6. This most likely proceeds through protonation of one of the the azlactone enantiomers with the chiral phosphate acting as the counterion, followed by nucelophilic attack of the alcohol. Use of the more bulky 3,5dimethoxyphenyl group led to substantial improvement in ee 90% conversion, 100% ee and 91% conversion, 90% ee, respectively. DKR of substituted 4-aryl azlactones under optimized conditions yielded the corresponding substituted αarylglycine derivatives in 86-93% ee and 81-90% yield. The most recent report of Birman`s group describes the DKR of 10 α-arylthio-and α-alkylthioalkanoic acids. 6 This latest example was inspired by the serendipitous finding that α-arylthioesters are readily racemised by the previously mentioned amidine-based catalysts, thereby opening up the possibility of DKR. In this process the alkanoic acid 10 was first reacted with an activator 15 resulting in the formation of the corresponding anhydride in the presence of base (iPr2NEt) (Scheme 4).

Scheme 4 DKR of α-thioalkanoic acids 20
With α-phenylthiopropanoic acid 10 (R 1 = Me, R 2 = Ph) as a substrate, the use of DCC as an activator resulted in slow anhydride formation accompanied by the formation of side products. The use of benzoic anhydride as an activator and 1naphthylmethanol as the nucleophile gave the ester 12 in 91% 25 yield and 91% ee. After optimisation, (S)-homobenzotetramisole (HBTM, 11) appeared the best acylation catalyst. Finally, ester 12 was reduced using lithium aluminium hydride (95% yield) without any loss of optical purity.
Deracemisation by means of an enantioselective acylation 30 catalyst was also achieved by Qu et al. 7 In this case, the meso-1,2-diol 13 was non-selectively mono-acylated, after which via DKR selective acylation was performed using an enantiopure organicatalyst to selectively acylate the remaining free alcohol. A base was used to catalyse the racemisation of the starting material 35 (Scheme 5).

Scheme 5 DKR of meso-1,2-diols 13
In initial experiments with monochloroacetate 13 (R = CH2Cl) 40 and iPr2NEt as a base in CCl4 racemisation was slower than acylation resulting in decrease of ee during the reaction. Racemisation was faster with more electron-withdrawing groups on the acetate rendering it more susceptible to nucleophilic attack. However, the presence of these groups makes the acetate a 45 worse hydrogen bond acceptor thereby lowering the enantioselectivity of the reaction. Several halogen substituents were tested and the best results were obtained with dichloroacetylated diol 13 (R = CHCl2). Optimisation of the reaction conditions proved that DABCO was the best base (e.r. 50 85:15) in CCl4 at -20 °C. Scheidt et al. reported that also N-heterocyclic carbenes can be used as organocatalysts in enantioselective acylation-based DKR to synthesize highly substituted β-lactones. 8  With β-ketoester 16 (R = Et) as a substrate different catalysts were screened and NHC 17 was found to give the best results (72% yield, d.r. of 5:1 and 90% ee). This reaction proceeds through nucleophilic attack of the carbene onto to aldehyde, which gives rise to an umpolung so that the resulting enol, 5 directed by the chiral NHC, diastereoselectively attacks the ketone, which then continues to form the β-lactone. The alkoxide that is released in the process liberates the NHC to give product 18. Using the same catalyst, the R-group in the substrate was varied changing also the acidity of the α-position and thus the rate 10 of racemisation. The use of a trifluoroethyl ester led to an increase of diastereoselectivity (8:1 d.r.) with excellent ee (99%), albeit in lower yield (64%). Further optimisation of the conditions revealed that 1,2-dichloroethane and cesium carbonate were the best solvent and base, respectively. Under these conditions 15 substrate 16 was transformed into the desired lactone in high yield (86%), good d.r. (6:1) and excellent ee (98%).

Transition metal-catalysed DKR
Considering the asymmetric transformations that are catalyzed by 20 transition metals, asymmetric hydrogenation stands out as a general ang highly practical way to convert olefins, ketones, or imines into the corresponding enantiopure reduced products. This approach was discovered as early as in 1995, when the Japanese chemist Ryoji Noyori developed a ruthenium-based catalyst for 25 asymmetric transfer hydrogenation of ketones and imines. 9 As a result, in 2001 he was awarded the Nobel Prize for his work on asymmetric catalytic hydrogenation. Asymmetric hydrogenation of a racemate, however, will give rise to the formation of diastereoisomers. Several groups have addressed this issue, where 30 in situ racemisation will give rise to the formation of single, enantiomerically pure products.
In 2010 the group of Ratovelomanana-Vidal and Genêt reported a ruthenium-catalysed DKR in the synthesis of Taxotere ® , an anti-mitotic chemotherapy drug. 10 In this process, 35 asymmetric hydrogenation was initially used for resolution of αchloro-β-ketophosphonates 19 and esters 21. Racemisation proceeded via tautomerisation of the ketophosphonate to the corresponding enol. First, different ruthenium complexes were employed in the DKR of racemic 19 (R = Me, Scheme 7).  approach to ketoester 21, the envisioned precursor in the Taxotere ® synthesis, was also described (Scheme 8).

Scheme 8 DKR of α-chloro-β-ketoesters
The best selectivity (96% de, 92% ee) was achieved using (R)-55 FLUOROPHOS at 50 °C, albeit in moderate conversion (58% after 48 h). In order to increase the conversion several solvents were screened with DCM giving the best compromise between conversion and selectivity.

65
Scheme 9 DKR of α-alkoxy-β-ketoesters Diluting the reaction mixture from 1 to 0.2 M improved the d.r. to 97:3. Under optimised conditions DKR of a broad range of 70 substrates was performed with very good conversions (80-99%) and excellent selectivities (d.r. 97:3, >99% ee). In 2010, related DKR of α-alkyl-β-ketoamides via rutheniummediated asymmetric transfer hydrogenation was also reported. 12 In initial studies compound 26 was used as a test substrate with 75 ruthenium complex 27 as the catalyst and a mixture of formic acid and triethylamine as hydrogen source (Scheme 10).
In 2012, Johnson et al. published a DKR of α-ketoesters also via asymmetric transfer hydrogenation. 14 A broad range of 30 catalysts were screened on substrate 35 (R = Ph) in DMF at 75 °C using a 5:2 mixture of formic acid and triethylamine as hydrogen source (Scheme 13). Scheme 13 DKR of α-ketoesters 35 The best results were obtained using ligand 36, which gave the desired product in 92% yield, excellent d.r. (> 20:1) and e.r.
(94:6). Substrates incorporating electron-rich, electron-poor and heteroaryl substituents gave all products in high yields (around 90%) and enantioselectivities (typical e.r. of 95:5). 40 On a somwhat different note, in the same year Fu et al. In this case, Ru-catalysts was responsible for racemisation of the secondary alcohols (via dehydrogenation/hydrogenation through hydrogen transfer), while in a kinetic resolution only one of the enantiomeric alcohols reacts faster with the in situ formed 50 enantiopure activated acyl-DMAP intermediate. The best results were obtained with acetyl isopropyl carbonate as the acylating agent in a 1:1 mixture of tert-amyl alcohol and toluene at 10 °C. Then, various aryl alkyl carbinols were used as substrates to give successful results (yields generally exceeding 90% and ee`s around 90%). 5

Enzymatic DKR
In addition to enantiopure organocatalysts and transition metal complexes, enzymes can be powerful catalysts to induce enantioselective reactions due to their high substrate selectivity. In order to overcome the 50% yield barrier in enzymatic 10 resolutions of racemic mixtures, the DKR principle has also been recently applied to a variety of enzymatic reactions.
In 2011 the group of Asano reported enzyme-promoted DKR of α-aminonitriles leading to optically pure α-amino acids. 16 This process consisted of three steps: (i) hydrolysis of racemic α- 15 aminonitrile by a non-selective nitrile hydratase (NHase), (ii) stereoselective hydrolysis of the resulting α-amino acid amide by an amino acid amide hydrolase, and (iii) simultaneous racemisation of the α-amino acid amide by ACL racemase (Scheme 15  Several substrates were tested using this system and both (R)-and (S)-amino acids could be obtained in excellent yields and ee`s (>99%). Such an approach, however, will always be dependent 25 on the availability of different enzymes that are able to convert bothe enantiomers. The group of Ramström used a lipase from Burkholderia cepacia (a group of Gram-negative bacteria) to catalyse the reaction between an acyl donor (phenyl acetate) and an amine to 30 give the acetylated amine. 17 In initial experiments KR of αaminonitrile 38 was investigated (Scheme 16). The conversion into product (R)-39 was recorded at 60% with an enantiomeric purity of 88% ee. The conversion exceeding the theoretical 50% maximum of KR and the observation that the remaining starting 35 material 38 was racemic, clearly indicated that a DKR-type process took place. Further mechanistic studies led to the conclusion that racemisation of the α-aminonitrile was catalysed by lipase via a retro-Strecker mechanism through the same active site that catalyses transacylation. This method was demonstrated to give good yields (87-90%) and ee`s (85-88%), but only after prolonged reaction times (7-13 days). 50 In 2010 Kroutil et al. published a DKR of racemic mandelic acid 40 to yield the corresponding enantiopure α-amino acid Lphenylglycine (42). 18 The α-hydroxy acid was first oxidised to the α-ketoacid 41 using D-mandelate dehydrogenase (D-MDH), after which it was converted into the α-amino acid via 55 asymmetric reductive amination with L-amino acid dehydrogenase (L-AADH) (Scheme 18).

Scheme 18 DKR of mandelic acid 40
Because oxidation and reduction took place simultaneously, the 60 NADH/NAD + cofactor was recycled in this system. In the screening process, 10 out of 24 tested AADHs gave excellent results (>97% conversion, >97% ee), all favouring L-42.
D`Arrigo et al. reported in 2011 an enzymatic DKR of N-Bocamino acid thioesters. 19 In this process, the authors take 65 advantage of the fact that thioesters, as opposed to the resulting amides, are prone to racemisation under basic conditions. The formation of the amide bond can be catalysed in an enantioselective manner by proteases. Subtilisin, commercially available under the name alcalase, is known to be stable in 70 organic solvents (an aqeuous reaction will lead to hydrolysis of the thioester) and in the presence of bases. In this system, ammonia or amines can be used as nucleophiles (Scheme 19).
In 2010, Gotor et al. published two reports about DKR using Baeyer-Villiger monooxygenases (BVMOs). In one of them, a DKR of 2-alkyl-1-indanones to form the corresponding dihydroisocoumarins was described. 20 An enzymatic DKR of 10 racemic 2-methyl-1-indanone 45a was performed in a biphasic system of Tris-HCl buffer with 5% hexane, in which racemisation most likely proceeds via the corresponding enol. The ring expansion was carried out using a single mutant of thermostable phenylacetone monooxygenase (M446G PAMO) from 15 Thermobifida fusca in the presence of glucose-6-phosphate dehydrogenase (G6PDH) to recycle NADP + (Scheme 20).

Scheme 20 DKR of 2-alkyl-1-indanones
It was found that longer reaction times (7 days) are beneficial for 20 conversions (84%, compared to 33% after 3 days), but ee`s decreased (82% in comparison with 92% after 3 days) due to partial racemisation of the lactone under the reaction conditions.
A second report of the Gotor group described DKR of benzyl ketones catalysed by BVMOs. 21 Racemic substrate 47 (R 1 = R 2 = 25 Me) was subjected to DKR catalysed by 4-hydroxyacetophenone monooxygenase (HAPMO) from Pseudomonas fluorescens in the presence of G6PDH to regenerate NADPH from NADP + . Racemisation of the substrate occurred spontaneously under basic conditions, and was enhanced by the presence of an anionic 30 exchange resin (Scheme 21). Several resins were tested and the best results (86% conversion, 84% ee after 116 hours) were obtained for (R)-48 using Dowex MWA-1, a weak anionic exchange resin. 35

Scheme 21 DKR of benzyl ketones
With differently substituted aromatic rings (X = m-Me, m-CF3, p-Cl) conversions varied from 69 to 88% with ee`s of the product varying from 58 to 81%. Clearly, the results using this system are strongly dependent on the substrate structure. 40 In 2011, Yang et al. reported a DKR of secondary aromatic alcohols using an acidic resin known as CD8604, which is known to racemise benzylic alcohols, presumably via formation of the benzylic cation and attack again from water. In addition, Novozyme 435 (a commercially available immobilised 45 preparation of lipase B from Candida antarctica) was used for the kinetic resolution. 22 Different acyl donors were investigated using this system (Scheme 22). It was found that longer chain acyl donors (R = Bu, Pent, etc) strongly inhibited non-selective transesterification catalysed by the resin due to steric effect of the substrates, as well as to formation of micromicelles in the pores of the resin. After screening of acyl donors with varying chain length (R = Me to 55 octyl), 4-chlorophenyl valerate (R = Bu) was found to be the most appropriate in DKR of phenylethanol (yield >99%, ee >99% after 6 hours). Employing this acyl donor in DKR of 1-phenylethanols with various substitution patterns in the phenyl ring resulted in yields and ee`s up to 99% in almost all cases. Finally, the 60 reusability of this system was examined up to 20 times without any loss in yield or ee. DKR of the racemic Boc-protected vicinal diamine 51 catalysed by lipase B from Candida antarctica (CAL-B) was described by Rebolledo et al. 23 First, KR of (±)-51 was performed to find that the use of 1-phenylethyl acetate as acyl donor and methyl tert-butyl ether as solvent at 28  The group of Giacomini described a DKR via reduction of arylpropionic aldehydes to (2S)-2-arylpropanols promoted by alcohol dehydrogenases (ADHs). 25 Several racemic substituted 2arylpropanals 58 (readily racemisation at higher pH values via 40 enolisation) were tested and after optimisation of reaction conditions (pH, cosolvent) for every substrate, good yields and enantiomeric ratios were obtained (Scheme 26). A noteworthy result was obtained for ketoprofenal 60 which bears two carbonyl groups. HLADH exhibited a high degree of selectivity, since no 45 traces of possible ketone reduction by-products were detected.

50
Two special cases of DKR are discussed in this section, namely, (1) DKR of atropisomers, compounds being chiral due to hindered rotation around a single bond, and (2) so-called attritionenhanced deracemisation, a process employing the property of some compounds to crystallise as a conglomerate in DKR 55 processes.
In 2010, an example of DKR of atropisomers was reported by Miller et al. 26 Compound 61 was triply brominated in reaction with NBS in the presence of a catalytic amount of Nmethylpyrrolidine (Scheme 27). This reaction gave rise to 60 compound 62, which due to the increased steric bulk at one of the aromatic rings, showed the occurrence of two atropisomers.

Scheme 27 Triple bromination using NBS
The enantiopure tripeptide 63, which catalyses aromatic 65 electrophilic substitution, was selected as the best catalyst for further study with NBP as a bromine source (Scheme 28). In this case, the catalyst kinetically favours bromination of one biaryl enantiomer over the other, which prior to the bromination are in equilibrium, but are after the bromination are not. In this way the racemic mixture is resolved.

Scheme 28 DKR of atropisomers via triple bromination with NBP
A range of differently substituted biaryl compounds were tested 5 and all the desired products could be obtained in good yields (70-85%) and moderate to good e.r.`s (85:15-97:3).
A different mechanochemical approach to DKR of atropisomers was reported in 2012 by Wiggins and Bielawski. 27 Enantiopure atropisomers were racemised by applying 10 mechanical force using poly(methyl acrylate) (PMA) chains on (R)-and (S)-BINOL in combination with ultrasound. Thus, by treatment of a racemic mixture of polymer-functionalised BINOL with deoxycholic acid and cholesterol esterase, esters of (S)-BINOL are selectively hydrolysed (Scheme 29). Upon subjecting 15 the remaining enantiopure starting material to ultrasound, the atropisomer racmises, ultimately producing only (S)-BINOL, which could be isolated in 90% yield and >98% ee. An entirely different, recently discovered resolution process is so-called attrition-enhanced deracemisation. This is a process that can only be applied on compounds which crystallise as a conglomerate, i.e. the enantiomers form separate solid phases. 25 Furthermore, compounds must be prone to deracemisation, e.g. by deprotonation under basis conditions. The driving force for deracemisation is a process called Ostwald ripening, meaning that dissolution of smaller crystals is favoured over larger ones. The resolution is not based on kinetics, as in a 'classical' DKR, but on 30 thermodynamics, and thus it can be called dynamic thermodynamic resolution (DTR).
Noorduin et al. reported in 2008 a deracemisation of phenylglycine derivative 64 using this technique. 28 The compound readily racemised under basic conditions making it a 35 potentially good substrate for attrition-enhanced deracemisation. Nearly racemic mixtures of 64 were suspended in MeOH and grinded by magnetic stirring in the presence of glass beads. Once the equilibrium between the solution and solid states had been established, DBU was added as a base to initiate racemisation 40 (Scheme 30).

Scheme 30 Equilibria during attrition-enhanced deracemisation
Because of the slight excess of one enantiomer, this enantiomer will have a slightly larger surface area for crystallisation and will 45 therefore crystallise faster, thus shifting all equilibria until the other enantiomer is depleted in the solid state. The ee increases exponentially during the reaction.
The same group reported in 2009 the use of this technique in the deracemisation of naproxen. 29 While naproxen itself crystallises as a racemic compound, its ethyl and methyl esters 65 and 66 both crystallise as conglomerates. Interestingly, the 5 solubility of 66 in methanol is lower than that of 65. This difference can be used to generate a supersaturated solution of 66 starting from a saturated solution of 65. To introduce a gradual feed of racemic 66, a mixture of rac-65 (92 mol%) and (S)-66 (8 mol%) was partially dissolved in MeOH/NaOMe and grinded at 10 700 rpm (Scheme 31).

Scheme 31 Attrition-enhanced deracemisation of naproxen
During the reaction all solid material 65 was converted into 66. Simultaneously, crystalline 66 was deracemised via attrition- 15 enhanced deracemisation, in favour of the (S)-enantiomer. It also appeared that the use of a gradual feed decreased the deracemisation time to 15 hours. 20 Cyclic deracemisation and stereoinversion are techniques involving chemically stable, achiral intermediates (Scheme 1). In cyclic deracemisation, the achiral intermediate (I) is nonselectively converted into a racemic mixture of the substrate (backreaction). The rates of conversion of the two enantiomers SR 25 and Ss into the intermediate I differ significantly, thus leading to accumulation of the slow-reacting enantiomer SS. In a stereoinversion process, a single enantiomer of the substrate (SS) is transformed into an achiral intermediate I, followed by irreversible conversion to the desired (opposite) enantiomer (SR). 30 Märkle and Lütz reported deracemisation of amino acids via a cyclic process in 2008 combining stereoselective enzymatic oxidation with non-selective electrochemical reduction. 30 In a test reaction, the electochemical reduction of 4-methyloxovalerate (69) in ammonia buffer at pH 10 (to establish an equilibrium 35 between keto acid 69 and imino acid 68) was pursued. The electrolysis potential was chosen such to only reduce the imino acid 68. So far, the keto acid could be converted into the racemate of leucine 67. A D-amino acid oxidase (D-AAO) was applied to selectively oxidise the D-amino acid (Scheme 32). 40

Scheme 32 Deracemisation of amino acid via cyclic deracemisation
Starting from racemic leucine, a yield of 80% of L-leucine (S)-67 could be obtained with 91% ee. In addition, starting from 45 enantiopure D-leucine a yield of 60% of (S)-67 with 77% ee was obtained.
Kroutil et al. reported deracemisation of α-chiral primary amines via stereoinversion. 31 ω-Transaminases (ω-ATAs) were used to catalyse stereoselective oxidation of one amine 50 enantiomer to the corresponding ketone (kinetic resolution). This ketone was then selectively reconverted to a single enantiomer of the desired amine (Scheme 33).

Scheme 33 Deracemisation of primary amines via stereoinversion 55
In the second step, the equilibrium lies towards the substrates (ketone, alanine) rather than towards the products (amine, pyruvate). To shift the equilibrium towards the products, the pyruvate formed was removed by reduction using lactate dehydrogenase (LDH). With this system, enantiomers of a range 60 of primary amines were accessible, depending on the order of addition of ω-ATAs and were obtained in generally good conversions and excellent ee`s (> 99% for most cases).
In 2011, the group of Patel reported three enzymatic routes to (S)-amino acid 72 from racemic mixtures, two of which 65 proceeded via a stereoinversion mechanism, while the third one was a cyclic deracemisation. 32 The basic strategy was to employ an (R)-amino acid oxidase to convert the (R)-enantiomer of the amino acid into an achiral keto acid intermediate, in combination with another enzyme, to reduce the keto acid selectively to the 70 (S)-amino acid (Scheme 34). The oxidase was inhibited under the reductive amination conditions, so that a two-step procedure was required. Using this process, a 54% yield of (S)-72 was obtained in >99% ee. The third strategy described by this group was a cyclic deracemisation. The use of a Celite-immobilised (R)-amino acid oxidase was combined with chemical reduction of the imine with a borane-ammonia complex. This reaction occurred before the 20 enzyme-bound imine was hydrolysed to the keto acid (Scheme 36).

Scheme 36 Cyclic deracemisation of an amino acid
The reaction was performed in a phosphate buffer solution with 25 ammonium formate at different pH values. The best yields (76-79%) were obtained at pH 6.0-7.0 in >99.9% ee.
In 2012, Yun et al. described deracemisation of the unnatural amino acid homoalanine using D-amino acid oxidase (DAAO) and ω-transaminase (ω-TA). 33 DAAO stereospecifically catalyses 30 oxidation of D-amino acids into the corresponding imino acids, which are spontaneously hydrolysed to the corresponding keto acids and ammonia. The ω-TA from Vibrio fluvialis was used to catalyse a transfer of the amino group from the amino donor onto the carbonyl moiety of the amino acceptor, in this case keto acid 35 76 (Scheme 37). After shaking at 37 °C for 8 hours, a 99% conversion was reached with >99% ee. Scheme 37 Deracemisation of homoalanine via stereoinversion 4 Whole cell-mediated deracemisation 40 The use of whole cells as biocatalysts in enzymatic processes has the advantage that time-consuming and thus expensive cell lysis and/or protein purification procedures are not necessary. This makes that whole cell preparations are often used for production scale enzymatic process. A disadvantage, however, is that 45 because of the large amounts of different enzymes present in living cells, the exact mechanism of whole-cell-mediated deracemisation cannot always be completely elucidated. The best results were obtained using cells treated with 20 kHz ultrasound. A 100% conversion of racemic acetate 77 in favour of (S)-78 with 45% ee was achieved in one day. After 7 days, the ee 5 had increased to 89%. The effect of pH on the medium was also investigated and a pH of 6.0 in combination with ultrasonic homogenisation at 20 kHz gave the best ee (96%) and conversion (100%). Alcántara and Molinari also described a whole cell-mediated 10 deracemisation of benzoin to the (S)-enantiomer employing whole cells from the non-convenional yeast Pichia glucozyma. 35 The reaction showed successful deracemisation with (S)-78 being the only enantiomer present in 85% yield. Benzil 79 was detected (10-15%) in the organic phase during the reaction, indicating an 15 oxidation-reduction cascade taking place (Scheme 39). The reactions performed in highly hydrophobic solvents as hexane and isooctane took place with excellent ee`s (>98%) and yields (91% for isooctane). Kurina-Sanz and Orden et al. described deracemisation of secondary alcohols by a chemoenzymatic sequence of oxidation/reduction using plant cells. 37 The reactivity of six cell 40 lines towards the oxidation of racemic 1-phenylethanol (83a) was screened. Only whole cells from Gardenia jasminoides were able to resolve the racemic mixture providing the (R)-isomer with 90-92% ee after 6 days. In order to investigate the substrate scope for this whole cell catalyst, whole cell-mediated kinetic resolution of 45 various 1-phenylethanol derivatives was performed. Substrates 83c and d gave the best results and were therefore selected for a chemoenzymatic deracemisation using NaBH4 as non-selective reducing agent (Scheme 41). Scheme 42 Deracemisation of aromatic sulfoxides using E. coli cells 60 The enantioselective reduction of racemic methyl p-tolyl sulfoxide (MTSO, 85b) was repeated three times with fresh E. coli cells leading to 49.3% conversion and 97.3% ee of (S)-MTSO. The resulting thioether was subjected to chemical oxidation leading to 22% conversion. The resulting racemic 65 sulfoxide was subjected to a new deracemisation cycle. After three cycles, (S)-MTSO (85) was obtained in 56% yield and 98% ee.

Enantioconvergence
The last category mentioned in Scheme 1 to achieve deracemisation is enantioconvergence. In this approach deracemisation is accomplished by the use of enantioconvergent methods which transform the two enantiomers of the substrate (SS 5 and SR) into a single enantiomer of the product (PR). The key feature of this strategy is that each enantiomer of the substrate follows a different reaction pathway towards a single enantiomer of the product. In 2009, Matsumoto et al. reported deracemisation of 10 monotosylated diols combining enzymatic hydrolysis with a Mitsunobu esterification. 39 Racemic 2-acetoxyhexyl tosylate 87 (R = Bu) was chosen as a representative target compound. Its hydrolysis using lipase PS via a KR-type process had already been described. The Mitsunobu reaction of alcohol 88 into (S)-87 15 was achieved with 100% yield and 99% ee using solid-supported PPh3, DEAD and acetic acid in toluene at room temperature. Using these conditions, the two-step deracemisation of a series of four racemic 2-acetoxyalkyl tosylates was performed. Deracemisation of 87 (R = Bu, Me, CH2OPMB, CH2Cl) was 20 successful in all cases with good to excellent yields (73-91%) and ee`s (93-98%) (Scheme 43).
Scheme 43 Two-step deracemisation using an enantioconvergent strategy 25 Sawamura et al. reported in 2010 a direct enantioconvergent transformation of racemic allylic ethers. 40 These compounds have a 'robust' chiral center that cannot be efficiently racemised and therefore are not suitable substrates for DKR. The racemic substrate 89a was reacted with bis(pinacolato)diboron (94) in the 30 presence of Cu(OtBu) and chiral phosphine ligand (R,R)-QuinoxP* (91, Scheme 44).

Scheme 44 Deracemisation of allylic ethers via enantioconvergence
Complete conversion was reached within 24 hours to give the 35 corresponding allylboronate (S)-90 in high yield (98%) and excellent ee (97%). Substrates 89b and 89c were also employed to investigate details of this process. Both were successfully converted into the corresponding (S)-allylboronates with good yields (92 and 91%, respectively) and 92% ee. The 40 enantioconvergence has been rationalised by showing that the achiral in situ formed Cu-boron nucleophile reacts with the allylic ether in an anti-type SN2' substitution with one substrate enantiomer, while the chiral phosphine-coordinated Cu-boron nucleophile reacts in a syn-fashion with the other substrate 45 enantiomer.
In 2011, the group of Faber reported a deracemisation of secondary alcohols employing a stereoselective alkylsulfatase which catalyses the hydrolysis of a sulfate ester with inversion of configuration at the stereocentre. 41 The alkylsulfatase from 50 Pseudomonas sp. DSM 6611, termed Pisa1, showed excellent stereoselectivity in the hydrolysis of linear, branched and cyclic sec-alkyl sulfates. To show the feasibility of an enantioconvergent deracemisation procedure employing this newly discovered enzyme, enzymatic hydrolysis was combined 55 with chemical hydrolysis under acidic conditions (Scheme 45).