Industrial Methods for the Production of Optically Active Intermediates

Authors


Abstract

Enantiomerically pure amino acids, amino alcohols, amines, alcohols, and epoxides play an increasingly important role as intermediates in the pharmaceutical industry and agrochemistry, where both a high degree of purity and large quantities of the compounds are required. The chemical industry has primarily relied upon established chemical methods for the synthesis of these intermediates, but is now turning more and more to enzymatic and biotechnological fermentation processes. For the industrial implementation of many transformations alternative methods are available. The advantages of the individual methods will be discussed herein and exemplified by syntheses of relevant compounds.

1. Introduction

The life-science industry is an important market for the chemical industry. The total revenue from sales in the pharmaceutical and agrochemical industries for the year 2000 is estimated to be in excess of €20 billion, of which the greatest share comes from the pharmaceutical industry. Optically active intermediates used as chemical building blocks, auxiliaries, or advanced intermediates have an estimated fraction of 15 % of the market.

Sales of special intermediates are increasing at about 7–8 % annually. Because of the increasing demand in the pharmaceutical industry for optically active intermediates, these compounds have a disproportionate share with 10 % annually. About 80 % of the active compounds that pharmaceutical companies have in the pipeline are chiral, and it is estimated that this fraction will increase, as the development of active compounds continues to be improved. The introduction of enantiopure active substances is also enforced through the ever stricter regulations of the US Food and Drug Administration (FDA). The authorities responsible for the registration of new active compounds increasingly demand the targeted synthesis of one stereoisomer. Since 1992 both the FDA and the European Committee for Proprietary Medicinal Products have stipulated that the physiological action of each enantiomer of a pharmaceutical product must be individually characterized, and since 1997 the fast-track, single-isomer program of the FDA has allowed pharmaceutical-product manufacturers shortened registration times with a so-called “chiral switch”. In addition to lower dosages and improved efficacy the extension of patent term is also a driving force for pharmaceutical companies to convert a racemic active compound into the enantiomerically pure form.

Enantiomerically pure active compounds are also being used increasingly in the agrochemicals industry. The reasons for this trend are similar to those in the pharmaceutical sector. The targeted synthesis of the respective active enantiomer can improve the economics of the process and lead to reduced quantities applied and thus to reduced environmental impact.

Optically active amines, alcohols, and carboxylic acids are important classes of compounds for the synthesis of many active pharmaceutical and agricultural products. However, the availability of industrial quantities of many of these compounds with high optical purity was previously very limited. In recent years the development of new technologies and their implementation in large-scale industrial processes have opened up new perspectives and economically more attractive methods for the production of active compounds.

2. Industrial Production of Amino Acids

2.1. Introduction

In 1999 the total market for proteinogenic amino acids was about 1.5 million tons, with an estimated revenue of around €3.5 billion. Amino acids have this considerable economic importance because of their broad spectrum of industrial applications. Their use in human nutrition and health makes up about 40 % of the market, and the animal-nutrition sector corresponds to about 55 %. L-Glutamic acid, L-phenylalanine, L-aspartic acid, and glycine are produced as additives for foodstuffs. L-Lysine, methionine, L-threonine, and L-tryptophan are important additives in modern animal feed, as conventional feed sources are deficient in these amino acids.

In respect of market volume L-glutamic acid is the most important amino acid. More than 650 000 tons are produced annually. Glutamic acid is mainly used in the form of its monosodium salt as a flavor enhancer in foodstuffs. The second-largest market volume is held by methionine with about 450 000 tons per year (t a−1). L-Phenylalanine and L-aspartic acid are also important compounds, at about 12 000 t a−1 each. They are required for the synthesis of the sweetener aspartame (α-L-aspartyl-L-phenylalanine methyl ester). In addition to the natural amino acids, nonnatural amino acids are also of industrial importance as building blocks for active compounds. Important examples are the D-amino acids used in the production of semisynthetic β-lactam antibiotics.

Three fundamentally different routes are used for the production of amino acids. In the extractive processes amino acids are isolated as components of natural protein-containing material. Alternative chemical and biological methods allow the targeted synthesis of amino acids.

2.2 Production of Amino Acids by Extraction

The oldest way of accessing amino acids is their extraction from protein-rich biological material, for which a wide variety of natural substances, such as hair, meat extracts, and plant hydrolysates, are used. These are mostly industrial by-products or waste, which are thus exploited for further utilization. An advantage of the extractive methods is the direct access to a series of different amino acids. Extractive isolation is particularly attractive for those amino acids for which demand is low. A feature of the extractive processes is the formation of salts as by-products, which can lead to considerable environmental impact. A number of more recent approaches attempt to avoid this disadvantage through biocatalytic processes. In particular, the use of extremophilic biocatalysts has already led to very promising results.1, 2 Extractive processes are currently used mainly for the production of L-cysteine. Keratin from hair and animal bristles is hydrolyzed by treatment with strong acids. The dimeric oxidation product cystine can be filtered from the hydrolysate because of its poor water solubility. Electrochemical reduction of the disulfide bridge in cystine yields cysteine.

A more recent method for the industrial production of cysteine has been developed by the Consortium für elektrochemische Industrie GmbH of Wacker Chemie.3 This is a fermentation process in which the metabolic regulation of Escherichia coli is modified by genetic manipulation in such a way that the production strain produces far more cysteine than it needs itself. The excess cysteine enters the fermentation medium, where it is oxidized to the poorly soluble cystine by atmospheric oxygen. The workup is analogous to that of the conventional process. The product prepared biocatalytically is of higher quality than cysteine obtained from biological residues. Cysteine is marketed in the USA by Wacker Chemie in alliance with Kyowa Hakko.

2.3. Chemical Processes for the Manufacture of Amino Acids

Although asymmetric syntheses of amino acids are known,4, 5 no economical process has yet emerged. In the industrial manufacture of racemic amino acids, resort is usually made to the Bucherer–Bergs variant of the Strecker synthesis.6 The respective α-aminonitrile is obtained from ammonia, hydrogen cyanide, and an aldehyde and is either hydrolyzed directly to the amino acid or converted into a hydantoin by reaction with CO2. The hydantoin is then hydrolyzed to the racemic amino acid in a basic medium. An alternative route to racemic amino acids is amidocarbonylation catalyzed by a transition metal. In a three-component reaction N-acyl α-amino acids are formed from an aldehyde, an amide, and CO.7

Even though there is no commercially viable chemical process for the production of enantiomerically pure amino acids, the synthesis of racemic amino acids is still of considerable importance because racemates can be converted readily into enantiomerically pure compounds by a number of biocatalytic methods. Chemical procedures that do not provide enantiomerically pure amino acids are used for just two products on the amino acid market. Glycine is achiral and the nonnatural enantiomer D-methionine is converted in adults and animals into the biologically active L-enantiomer through a transamination reaction.

2.3.1. Methionine Synthesis

The chemical D,L-methionine synthesis developed by Degussa is especially elegant.8 In this variant of the classical amino acid synthesis, in contrast to other processes, the formation of large quantities of salts is avoided (Scheme 1).

Scheme 1.

Methionine synthesis by the Degussa process.

3-Methylmercaptopropionaldehyde (2) is prepared from acrolein (1) and methanethiol. Aldehyde 2 then reacts with hydrogen cyanide and ammonium hydrogen carbonate to form 5-(2-methylthioethyl)hydantoin (3), which is hydrolyzed at 170 ¤C in an alkaline solution with an excess of potassium carbonate. The product potassium d,l-methioninate (4) is formed along with carbon dioxide and ammonia, from which ammonium hydrogen carbonate is recovered for the second step. The potassium methioninate solution is then acidified under CO2 pressure to produce D,L-methionine (D,L-5) and potassium hydrogen carbonate. The latter undergoes decomposition upon heating and concentration into CO2 and potassium carbonate, which is recycled by use in the hydantoin hydrolysis.

2.4. Production of Amino Acids by Biocatalysis

In biocatalytic methods enzymes are employed as catalysts. Therefore biocatalysis is mostly carried out under moderate temperature and pressure, and frequently, if not exclusively, in aqueous media. Biocatalytic methods can be broken down into the categories biotransformations and fermentation methods.

2.4.1 Fermentation Methods

In fermentation methods the synthesis apparatus of living microorganisms is used to produce useful compounds from simple, cheap raw materials. In principle all proteinogenic amino acids are accessible by fermentation methods. In the production of the high-turnover, optically active products glutamic acid, lysine, and threonine fermentation processes have become established for economic reasons. In 1999 the world market for these amino acids was 650 000 (glutamic acid), 40 000 (lysine), and 6000 t a−1 (threonine).

In 1957 the bacterium Corynebacterium glutamicum was isolated. Organisms of this genus9 were developed through a number of improvements to yield high-performance strains, which synthesize glutamic acid and lysine in remarkable space–time yields. With modern production strains amino acid concentrations of more than 160 g L−1 are reached.10 Originally the explanation for this extraordinary synthetic efficiency relative to that of other microorganisms was that the cell membrane of the bacteria is permeable to amino acids, and that the compounds formed intracellularly can diffuse passively into the medium.11 However, more recent work has shown that active transport across the cell membrane is responsible for the secretion of the amino acids.12

A detailed understanding of amino acid biosynthesis is required for improvement to be made in production strains. It was possible to take advantage of the results from the work on Corynebacterium glutamicum in this way. The first steps in the biosynthesis of glutamic acid, lysine, and threonine in Corynebacterium are catalyzed by the same enzymes. Thus, knowledge of regulatory mechanisms may be transferred to production strains for different amino acids. For example, by overproduction of the enzyme dihydropicolinate synthase the metabolism in Corynebacterium can be directed in favor of lysine production. In contrast, deletion of the appropriate gene leads to the biosynthesis of L-threonine.13

The economical fermentation production of L-threonine was first established in the 1980s when efficient E. coli mutants were developed. End concentrations of L-threonine of 100 g L−1 can be achieved with genetically modified E. coli or Serratia marcescens.14, 15 For the industrial production of L-threonine E. coli strains which contain an increased copy number of the threonine operon appear to be the most suitable.16

2.4.2 Biotransformations

Whereas living microorganisms are used as biocatalysts in the production of amino acids by fermentation, the catalysts in biotransformations are isolated enzymes or metabolically inactive cells. Depending on the specificity of the biocatalyst used, compounds that do not occur in natural metabolic pathways may also be transformed. Consequently biotransformations are the methods of choice for the production of enantiomerically pure D-amino acids and other nonnatural amino acids. In the production of natural amino acids biotransformations are used predominantly for those products whose market volume is less than 10 000 t a−1.16, 17

2.4.3. Lyases as Biocatalysts in Amino Acid Synthesis

A well-established application of lyases is the production of L-aspartic acid (L-6) from fumaric acid (7) and ammonia with the enzyme L-aspartate ammonia lyase (Scheme 2). Variations of this biotransformation have been used since 1958. The synthesis of L-phenylalanine (L-8) from (E)-cinnamic acid (9) proceeds analogously with the enzyme phenylalanine ammonia lyase as the catalyst.18 However, on an industrial scale this method has been replaced by a more economical fermentation process with phenylalanine overproducers.16

Scheme 2.

Lyase reactions.

L-Alanine (l-10) can also be produced through catalysis by a lyase. However, the lyase in question, L-aspartate β-decarboxylase, does not mediate the formation of a C[BOND]N bond as in the synthesis of L-phenylalanine or L-aspartic acid, but instead catalyzes the decarboxylation of L-aspartic acid (L-6) to L-10 (Scheme 3). The catalysts used industrially are immobilized cells that contain L-aspartate β-decarboxylase.19

Scheme 3.

L-Alanine synthesis.

Based on the above, a direct synthesis of L-10 from fumarate 7 and ammonia would appear feasible. However, as the optimal pH values for L-aspartate ammonia lyase and L-aspartate β-decarboxylase are very different, the procedure has to be carried out in two steps.20 The synthesis of L-10 from fumarate 7 and ammonia is the first two-step process based on biocatalysts. In the decarboxylation of L-6 stoichiometric amounts of CO2 are released. The formation of gas bubbles leads to inhomogeneity in the immobilization bed. To avoid this problem reactions are carried out in a special pressure reactor, in which a biotransformation can be carried out at 10 bar.21 The increased pressure prevents the release of gaseous CO2 and has no effect upon enzyme activity.

2.4.4 Amino Acid Dehydrogenases

Enantioselective biotransformations are also possible on an industrial scale with amino acid dehydrogenases (deaminating amino acid oxidoreductases). As a number of these enzymes are characterized by very low substrate specificity, nonnatural compounds can also be transformed. In addition to ammonia and the respective 2-oxocarboxylic acid, amino acid dehydrogenases need a cosubstrate, which supplies the hydride ions for the reduction of the intermediate Schiff bases.22 For example, nicotinamide adenine dinucleotide (NADH) is a suitable cosubstrate. The oxidized cosubstrates (e.g. NAD+) must be regenerated in a further redox reaction because of their high cost.

One established regeneration system involves a formate dehydrogenase from Candia boidinii, which oxidizes formate irreversibly to CO2 and thereby reduces the cosubstrate NAD+ to NADH.23 A combination of L-leucine dehydrogenase, formate dehydrogenase, and cosubstrate is used by Degussa to produce L-tert-leucine (L-12) on a ton scale from ammonium formate and trimethylpyruvic acid (11) (Scheme 4).24

Scheme 4.

Production of L-tert-leucine (L-12) through reductive amination with leucine dehydrogenase.

2.4.5 Aminotransferases

The use of aminotransferases allows the problem of cosubstrate regeneration to be circumvented. These enzymes transfer amino groups of amino acids to 2-oxocarboxylic acids. An illustrative example of their industrial application is the synthesis of L-2-aminobutyric acid (L-13) by Great Lakes Chemicals.25 Genetically modified E. coli is used as the catalyst in this biotransformation (Scheme 5). L-Aspartic acid (L-6) functions as the amino-group donor. It is converted into 2-oxalylacetic acid (14) by aspartate aminotransferase. Intermediate 14 undergoes spontaneous decarboxylation to give pyruvate 15 and CO2. The amino group of L-6 is transferred to 2-oxobutyric acid (16). As 16 is not readily accessible, the gene for L-threonine deaminase is incorporated into the cells. In this way 16 can be prepared in situ from inexpensive L-threonine.

Scheme 5.

Production of D- and L-aminobutyrate (D,L-13) from 2-oxobutyric acid (16) and L-aspartic acid (L-6) through enantioselective catalysis by aminotransferases.

Because of an undesired side reaction of aminotransferases alanine (10) can be formed from 15. To prevent this course two molecules of 15 are converted into acetolactate 17 by the enzyme acetolactate synthase. The acetolactate synthase gene from a Bacillus strain was transferred into E. coli. Compound 17 undergoes spontaneous decarboxylation to 2-hydroxybutane-3-one (acetoin, 18), which can be removed from the reaction mixture by distillation. The removal of 18 displaces the reaction equilibrium in favor of the desired products and leads to a significant decrease in the formation of the by-product L-10. Thus, the ratio of the main product L-13 to L-10 is increased from 2.4:1 to 22.5:1, and end concentrations of L-13 of 27 g L−1 can be obtained.25

Depending on the choice of amino acid transferase, both enantiomers can be obtained. The L-tyrosine aminotransferase from E. coli is so indiscriminate that it can be used for the production of L-13. Strains that contain a D-amino acid transferase from Bacillus sphaericus accordingly form D-13.

The company Mercian (Japan) uses a combination of dehydrogenase and aminotransferase in an elegant whole-cell method for the production of L-piperidine-2-carboxylic acid (S-20) from natural L-lysine (19; Scheme 6). The method is based on a biotransformation with recombinant E. coli bacteria. L-Lysine (19) enters the cells with the assistance of the overexpressed L-lysine permease transport system and is there deaminated to the aldehyde–amino acid 21 by the lysine aminotransferase from Flavobacterium lutescens. Amino acid 21 is in equilibrium with 2,3,4,5-tetrahydropyridine-2-carboxylic acid (22) and is reduced to (S)-20 with the intrinsic pyrroline-5-carboxylate reductase of E. coli and NADPH.26 The conversion observed is greater than 90 % with respect to L-lysine, and (S)-20 is obtained from the culture filtrate in 70 % yield after purification. However, to reach high end concentrations of 50 g L−1, long fermentation times (150 h) are necessary. (S)-20 serves as a building block for a number of pharmaceutical agents, for example, the local anesthetic bupivacaine (AstraZeneca). An alternative access to enantiomerically pure 20 is acylase-catalyzed racemate resolution (Section 2.4.8.3).

Scheme 6.

2.4.6. D-Amino Acids from Fermentation Processes

Monsanto has patented a fermentation process for the production of D-amino acids by genetically modified microorganisms.2 For example, D-phenylalanine (D-8) can be synthesized with these biocatalysts (Scheme 7). The metabolism of the microorganisms is modified in such a way that no functional D-amino acid deaminases are produced. However, the strains contain additional genes for L-amino acid deaminases, D-amino acid aminotransferases, and a specific racemase. The growth medium of the cells is supplemented with an L-amino acid (L-23) as an amino-group donor, which may be L-alanine, L-glutamine, or L-aspartic acid.

Scheme 7.

Production of D-phenylalanine (D-8) by fermentation with genetically modified microorganisms (according to reference 27). Alanine, glutamine, or aspartic acid are added as the amino-group donor (L-23); L-8 is formed by endogenous overproduction. The α-keto acid 25 is metabolized to CO2 and H2O.

The carbon skeleton of D-8 originates either from endogenously produced L-8 or from L-8 added to the medium, which is deaminated to phenylpyruvate 24 by L-amino acid deaminase. The amino-group donor L-23 is racemized by the microorganisms to give D-23, a substrate for D-amino acid aminotranserases. The amino group of D-23 is transferred to 24, whereby D-8 and the 2-oxocarboxylic acid 25 are formed. Carboxylic acid 25 is then degraded by the microbial metabolism with displacement of the equilibrium of the D-amino acid aminotransferase reaction toward D-8 formation. As both L-amino acid deaminases and D-amino acid aminotransferases are highly nonspecific, this approach is applicable to the production of a wide variety of D-amino acids from the corresponding L-amino acids.27

2.4.7. Chemoenzymatic Synthesis of D-Amino Acids

There is also an elegant chemoenzymatic option for these amino acid syntheses.28 In this case the L-amino acid is oxidized by an L-amino acid oxidase (Scheme 8). The intermediate imine is reduced by Pd-C in an ammonium formate buffer. Only the L-enantiomer of the resulting racemic mixture is utilized by the oxidase, and the D-enantiomer accumulates. The enantiomeric form of the amino acid produced depends on the specificity of the oxidase. Yields of 90 % and optical purities of 99 % ee have been observed for this process.

Scheme 8.

Chemoenzymatic synthesis of D-amino acids.

2.4.8. Preparation of Enantiomerically Pure Amino Acids by Racemate Resolution

2.4.8.1. Preparation of L- and D-Amino Acids with the Hydantoinase–Carbamoylase System

Racemic hydantoins 26 are readily accessible by chemical synthesis. A series of specific natural hydrolases have been isolated for their enantioselective hydrolysis to the corresponding carbamoyl derivatives. Specific hydrolases also exist for these compounds so that the preparation of optically active amino acids from the racemic hydantoins presents no difficulty.29

This method is particularly important for the preparation of D-amino acids. D-Phenylglycine and D-hydroxyphenylglycine, which are incorporated in the side chains of semisynthetic β-lactam antibiotics, are produced in this way.30 As hydantoins racemize spontaneously at pH>8, the complete conversion of the racemate into one enantiomer is possible. The rate of racemization is depends on side chain R. If the racemization of the hydantoin is the rate-determining step of the reaction, the reaction can be accelerated significantly by the addition of a racemase (Scheme 9).

Scheme 9.

Production of optically active amino acids with enantioselective hydantoinases and carbamoylases.

Whereas the biocatalytic production of D-amino acids from d,l-hydantoins is a well-established process, the synthesis of L-amino acids with this system is not so well developed technically, mainly as a result of the poor stereoselectivity of the known L-hydantoinases. However, improvements in this field are ongoing. Thus, it was shown that the enantioselectivity of a D-hydantoinase from Arthrobacter sp. DSM 9771 could be reversed through evolutionary enzyme development (“directed evolution”), with a simultaneous increase in activity.31L-tert-Leucine can also be synthesized with the hydantoinase–carbamoylase system. Degussa currently produces this compound on a 100-kg scale through a novel biotransformation in a pilot process. In this process a recombinant microorganism that contains the aforementioned modified hydantoinase, a carboxylase, and the respective racemase is used.32

2.4.8.2. Preparation of L-Cysteine by Racemate Resolution

As early as the 1970s a biocatalytic procedure was developed as an alternative to the extraction of cysteine (27) from human hair, feathers, or animal bristles. The reaction starts from d,l-2-amino-Δ2-thiazoline-4-carboxylic acid (d,l-28), which may be obtained by chemical synthesis from α-chloroacrylic acid (29) and thiourea (30; Scheme 10). The bacterium Pseudomonas thiazolinophilum transforms d,l-28 into cysteine. Thus, L-28 is cleaved to L-cysteine (L-27) by two specific hydrolases via (S)-carbamoyl-L-cysteine (L-31). A racemase converts D-28 into the corresponding L-enantiomer, so that complete conversion of the racemate is possible. The process takes place in yields of 95 % to give L-27 in concentrations of 30 g L−1.33 It was implemented by Ajinomoto in 1982.

Scheme 10.

Enantioselective synthesis of L-cysteine (L-27).

2.4.8.3. Racemate Resolution with Acylases and Amidases

As well as hydantoinases, acylases and amidases are used for racemate resolution. L-Acylases catalyze the resolution of acyl d,l-amino acids to L-amino acids and N-acyl D-amino acids. After separation of the products, the N-acyl D-amino acid is racemized in a second step and returned to the enzyme reaction. One of the first industrial applications of an amino acid acylase was introduced by the Tanabe Seiyaku (Japan) in 1969.34

At Degussa L-methionine (5) is produced for use in foodstuffs through racemate resolution catalyzed by an acylase (Scheme 11). The process is carried out continuously in a stirred reactor. The acylase, which is not immobilized, is retained by ultrafiltration in a hollow fiber module.

Scheme 11.

Acylase-catalyzed racemate resolution of D,L-N-acetyl-5 (Degussa).

The amidase process developed by DSM operates by the same principle, but with a different biocatalyst. D,L-Amino acid amides are hydrolyzed enantioselectively with amidases. After separation of the L-amino acid and the D-amino acid amide, the latter can either be racemized or converted into the corresponding D-amino acid by chemical hydrolysis.35

Nonnatural cyclic amino acids, such as piperidine and proline derivatives, are also accessible through the use of enantioselective hydrolases. Often both enantiomers are of industrial interest. Examples are piperidine-2-carboxylic acids (20), which are incorporated into various active compounds, and N-CBZ-protected proline (N-CBZ-32). Researchers at Lonza have developed biocatalytic methods that allow access to this class of compounds through reactions with hydrolases.36 Scheme 12 illustrates the synthesis of CBZ-protected proline derivatives.

Scheme 12.

Proline acylase process by Lonza for the production of enantiomerically pure N-Cbz-proline (N-Cbz-32).

2.5. Improvements in Biocatalytic Amino Acid Synthesis

2.5.1 Improvements in Fermentation Processes

Classical improvement of production strains is based on undirected mutations. The genome of the production organisms is altered at random, and more efficient mutants are identified in a subsequent selection process. The mutants are characterized by, for example, increased membrane permeability,37 regulation defects, or biosynthesis enzymes with altered kinetic characteristics.38

In the 1980s it was recognized that a fundamental understanding of the microbial amino acid metabolism would be necessary for further increases in productivity. Detailed analysis allows quantification of the flow of metabolites in a fermentation as a function of time (metabolic flux analysis). In this way the addition of nutrients can be optimized, and yields can be increased as a result. Metabolic flux analysis also makes it possible to model the metabolism of a given production strain (metabolic modeling). Moreover, conclusions drawn from metabolic flux analysis can then form the basis for targeted intervention in the metabolism by genetic methods.39

Another important requirement for the targeted improvement of amino acid production strains is knowledge about the genome of the microorganism concerned. The genome of Corynebacterium glutamicum has already been sequenced by at least three important amino acid producers.8, 40, 41 It is merely a question of time before the results of genome sequencing and “metabolic modeling” lead to significantly improved amino acid production.

2.5.2. Improvements in Biotransformations

Not only fermentation processes, but also biotransformations are undergoing continual improvement. As with chemical processes, in addition to selectivity, productivity (expressed in terms of the total turnover number) is the critical measure of the quality of the catalyst.

A variety of strategies can be implemented to establish or improve a biocatalytic process. In the ideal situation a suitable biocatalyst is already available, and only the technical process itself must be elaborated. However, this situation rarely arises, and most frequently the development of a new catalyst is necessary. In wide-ranging screening programs, attempts are being made to identify novel enzymes. These are no longer targeted solely at isolating whole microorganisms from the biodiversity, as in classical screening, but instead at accessing genetic information directly. DNA is isolated from samples and transferred into suitable expression systems. The desired activity is then identified through high-throughput screening.42 In this way direct access should be possible to the gene pool of nonculturable microorganisms, which by far outnumber culturable microorganisms.43

Of course, specific characteristics of known enzyme systems can also be improved.44 In evolutionary enzyme optimization, also known as “directed evolution”, natural evolution is mimicked in vitro. Mutations are incorporated randomly into the genes of the biocatalyst to generate an abundance of variants, which are then screened for those mutations with improved catalytic properties. The work of May et al. on the D-hydantoinase from Arthrobacter illustrates the potential of this methodology.31 As well as untargeted modification, the recombination of genetic material plays a critical role in evolution. These mechanisms may also be copied and carried out in vitro.45

The development of biocatalysts for industrial applications is dominated by smaller companies, which offer their services to larger partners. Of course, fermentation processes for the production of amino acids can also be improved by means of evolutionary enzyme optimization. The optimized biocatalysts can also be transferred into production strains to avoid bottle necks in amino acid biosynthesis.

3. Production of Carboxylic Acids

As for amino acids, there are in principle three possible methods for the production optically active carboxylic acids. Naturally occurring carboxylic acids can be isolated from suitable raw materials. However, most carboxylic acids are produced in chemical and biocatalytic processes. In the case of natural carboxylic acids microbial fermentation processes are available, especially for quantities >1000 t a−1. Glucose or cheap industrial products, such as molasses, are used as the carbon source, which is metabolized by the microorganisms into biomass (growth), metabolic by-products, such as CO2, and the desired carboxylic acid.

In this section examples of the different technologies are presented. The rapid development of enzyme-catalyzed syntheses has already been summarized in the form of reviews to which reference will be made where appropriate.

3.1. Isolation from Natural Sources (Chiral Pool)

Natural chiral compounds (from the chiral pool) often offer an alternative to the synthesis of enantiomerically pure products.46 One chiral carboxylic acid isolated from natural sources is L-(+)-tartaric acid (33)

Thumbnail image of

. This isomeric form of tartaric acid separates out as tartar (potassium hydrogen tartrate) during the fermentation of grape must. Upon its reaction with calcium chloride or calcium hydroxide and sulfuric acid, 33 is released; yeast residues and gypsum occur as by-products. An alternative process has been established recently by Lurgi Life Science, whereby the tartar is transformed with potassium hydroxide into the water soluble dipotassium salt. After filtration and ion-exchange chromatography crystalline tartaric acid and feed yeast are obtained as by-products. Potassium sulfate and ammonium sulfate are formed as by-products during regeneration of the ion exchanger. In addition to numerous applications in the textile and food industries, 33 and derivatives, such as (−)-dibenzoyl-L-tartaric acid, are used as racemate resolving agents and for asymmetric synthesis, for example, of 2-aminotetralines47 and methyl (R)-3-hydroxybutyrate.48

Natural carbohydrate building blocks have been used for decades for the preparation of enantiomerically pure sugar acids.49 In industrial syntheses saccharides are oxidized selectively. During the middle of the last century oxidations with hypochlorite50 and electrochemical oxidations49 were developed for the preparation of D-gluconic acid (34) from D-glucose (35). These have been replaced by biotechnological processes in which the monosaccharide is selectively oxidized with microorganisms (mainly Aspergillus niger and Gluconobacter sp.) in a submersed aerobic process (Scheme 13).

Scheme 13.

Selective enzymatic oxidation of D-glucose (35) to D-gluconic acid (34).

A new method has been devised by DSM/Gist Brocades for use in the food industry. Glucose oxidase transforms 35 into the lactone 36, which is then hydrolyzed to 34 and H2O2 Scheme 14). Catalase decomposes the H2O2 formed to regenerate oxygen for the oxidation step and prevent deactivation of the enzyme by side reactions. The process is characterized by the high purity of the product 34 and the near-quantitative yield.51

Scheme 14.

Alternative synthesis of D-gluconic acid (34) of high enantiomeric purity through the oxidation of D-glucose (35) (DSM/Gist Brocades).

D-Glyceric acid can be produced either by stepwise chemical oxidation of D-glucose with anthraquinone-2-sulfonate/H2O2 and hypochlorite or by enzymatic decarboxylation of L-tartrate.52, 53

3.2. Classical Chemical Synthesis

3.2.1. Crystallization with Enantiomerically Pure Amines

The enantiomers of racemic carboxylic acids are separated by fractional crystallization of the diastereoisomeric salts that they form with enantiomerically pure amines (Scheme 15; see also Section 4.1.1). (R)-Mandelic acid ((R)-37) is produced by Yamakawa Chemical Industry Ltd (Japan) by classical racemate resolution from (R,S)-37 with (R)-1-phenylethylamine ((R)-38; Scheme 16). The enantiomerically pure (R)-38 can be recovered and reused several times as the resolving agent.96

Scheme 15.

Racemate resolution of optically active carboxylic acids through salt formation with enantiomerically pure amines and fractional crystallization.

Scheme 16.

The enantiomerically pure thiazolidinone carboxylic acid (S)-39, an intermediate in the synthesis of the new calcium antagonist CP-060-S ((S)-40), is isolated by resolution of the racemate with (S)-N-benzyl-1-phenylethylamine ((S)-41; Scheme 17). Although the salt-forming carboxy functionality is four bonds away from the stereogenic center, (S)-39 was isolated with 99.8 % ee.54

Scheme 17.

Racemate resolution with (S)-N-benzyl-1-phenylethylamine ((S)-41) leads tp enantiomerically pure (S)-39 for the synthesis of the calcium antagonist CP-060-S ((S)-40).

The chiral amino acid (S)-20 can be synthesized in a similar way through fractional crystallization of diastereomeric salts with 2-phenoxypropionic acid (42). The less soluble salt (S)-20–(S)-42 crystallizes with high diastereomeric purity from water/alcohol mixtures (Scheme 18).55

Scheme 18.

“Dutch resolution”, a variant of diastereoisomeric crystallization in which carboxylic acid racemates are resolved through the use of amine mixtures, will be discussed in Section 4.1.1.

3.2.2. Asymmetric C[BOND]C Coupling

Recently new (salen)titanium and (salen)vanadium complexes 43

Thumbnail image of

(CACHy-catalysts),56 which mediate the asymmetric addition of trimethylsilyl cyanide to aromatic and aliphatic aldehydes 4457 and ketones58, have been described for the preparation of substituted α-hydroxycarboxylic acids (Scheme 19, Table 1). This method has been extended to the use of potassium cyanide as a cheaper nucleophile.59 The α-hydroxycarboxylic acids 46 are available from the cyanohydrin trimethylsilyl ethers 45 by acid hydrolysis. As these reactions are in competition with the enzymatic processes described below, they can only be applied in a few special syntheses.

Scheme 19.

CACHy catalysts 43 mediate the enantioselective addition of trimethylsilyl cyanide to aromatic and aliphatic aldehydes 44. The products 45 are hydrolyzed to enantiomerically pure α-hydroxycarboxylic acids 46.

Table 1. Addition of trimethylsilyl cyanide to aldehydes with CACHy catalysts 43 (Scheme 19).[a]
RCatalyst 43 (M=Ti)Catalyst 43 (M=V)
 ee [%]
  1. [a] Catalyst: 0.1 mol %; reaction time: M=Ti: <5 min, M=V: 18 h; quantitative conversion; see reference 57.

Ph8294
p-OMe-Ph8490
o-Me-Ph7690
m-Me-Ph9095
p-Me-Ph8794
p-NO2-Ph5073
Et5277
tBu6668

3.3. Biotechnological Synthesis

3.3.1. Asymmetric Syntheses

3.3.1.1. Oxidations

In some hydroxylation processes bacterial mono- or dioxygenases are used to catalyze bond formation between one or both oxygen atoms of O2, respectively, with nonactivated carbon atoms. These hydroxylations are usually regio- but not stereoselective.60, 61 Many microbial hydroxylations to provide hydroxycarboxylic acids involve sequential α,β-dehydration and hydrolysis steps. The preparation of L-carnitine (L-47) from γ-butyrobetaine (48) by the Lonza process (>100 t a−1) is an example of such a process. The biochemical transformation takes place via butyrobetainyl-CoA (49) and crotonobetainyl-CoA (50; Scheme 20).62 The Kaneka process proceeds analogously through the β-hydroxylation of 2-methylpropionate (51) with Candida rugosa to give (R)-52, a building block for the synthesis of ACE inhibitors, such as captopril (Scheme 21).63

Scheme 20.

Synthesis of L-Carnitin (L-47) according to the Lonza process. 1. Thioester synthetase, 2. dehydrogenase, 3. hydrolase, 4. thioesterase.

Scheme 21.

β-Hydroxylation of 2-methylpropionate (51).

3.3.1.2. Reduction

In principle, the enantioselective biocatalytic reduction of keto acids is a possible route to enantiomerically pure hydroxycarboxylic acids. One current disadvantage is the dependence upon expensive cofactors such as NADH. A variety of approaches have been described for the regeneration or even the replacement of these reagents (see Section 2.4.4). Further examples are described in Section 5.

3.3.1.3. Hydrolysis

The enzyme-catalyzed enantioselective addition of water to the double bond of prochiral fumaric acid (7) is used in the preparation of L-malic acid (L-53) with immobilized brevibacteria or fungi (Tanabe process; Scheme 22).64, 65L-Malic acid is used above all for the stabilization of fruit juices in food technology.

Scheme 22.

Production of L-malic acid (L-53) through the enantioselective addition of water to 7 (Tanabe process).

3.3.1.4 C[BOND]C Coupling (Hydroxynitrile Lyase Process)

Hydroxynitrile lyases (HNL) catalyze the stereoselective addition of hydrogen cyanide to aldehydes and ketones or, depending on the reaction conditions, the cleavage of cyanhydrins.66 The α-hydroxycarboxylic acids are obtained by hydrolysis of the chiral cyanhydrins with hydrochloric acid. In this efficient synthesis quantitative conversion of the aldehyde into the product is possible. HNL processes have recently been implemented in industrial syntheses of some chiral aromatic α-hydroxycarboxylic acids.

Chiral mandelic acid derivatives are used as building blocks in diverse syntheses of active compounds and also as racemate resolving agents (Section 4.1.1). Chemical and biocatalytic synthetic routes have already been described on several occasions,67 so we will restrict our discussion to a few more recent examples. HNL-based production processes have been developed for (R)-o-chloromandelic acid (54), an intermediate for the synthesis of the antidepressant and platelet-aggregation inhibitor clopidogrel,68 by DSM Chemie Linz, Nippon Shokubai, and Clariant (Scheme 23).69 Hydrocyanic acid adds enantioselectively to o-chlorobenzaldehyde (55) in microaqueous or biphasic70 systems. Mandelonitrile lyase (PaHNL), for example, in the form of almond-flour extract or immobilized on Avicel microcrystalline cellulose, is used as the catalyst. Depending on the solvent employed, the enzyme can be used for several months. The (R)-o-chloromandelonitrile ((R)-56) thus formed is converted into the corresponding carboxylic acid by hydrolysis with concentrated HCl, without undergoing racemization. The advantage of this process over racemate resolution through the crystallization of diastereomeric ammonium salts or through enzymatic acylation/deacylation or transesterification, is that a 100 % yield is theoretically possible.

Scheme 23.

HNL-catalyzed synthesis of (R)-o-chloromandelic acid ((R)-54).

In recent years processes for the production of S cyanohydrins and S hydroxycarboxylic acids catalyzed by HNLs from Hevea brasiliensis and Manihot esculenta have been made economically attractive by recombinant enzyme preparation in E. coli.71 One recent example is the production of (S)-m-phenoxybenzaldehyde cyanohydrin ((S)-57) in a biphasic process by DSM Chemie Linz and Nippon Shokubai.72 The cyanohydrin is an intermediate in the production of pyrethroids 58 (Scheme 24).

Scheme 24.

(S)-m-Phenoxybenzaldehyde cyanohydrin ((S)-57) is an intermediate in the synthesis of pyrethroids, such as 58.

3.3.1.5 Ester Hydrolysis and Transacylation

Researchers at DSM have developed an enzymatic process for the production of the captopril intermediate D-59, whereby the enantiomers of methyl 3-chloro-2-methylpropionate (d,l-60) are separated with a lipase.73 All lipases investigated hydrolyze L-60, and the unreacted D-60 is converted into D-59 (Scheme 25). Up to 98 % ee at 64 % conversion is observed with the lipase from Candida cylindracea.

Scheme 25.

Synthesis of the ACE antagonist captopril.

The resolution of d,l-pantolactone (d,l-61) by enantioselective pantolactone hydrolases can be regarded as a special case of ester hydrolysis. D-Pantolactone (D-61) is a precursor of D-pantothenic acid (D-62, vitamin B5), which is required as a feed additive in pig and poultry breeding (Scheme 26). In the 1990s researchers at the Daiichi subsidiary Fuji Chemical Industries (now Daiichi Fine Chemical Co.) developed a process in cooperation with Yamada and co-workers74 in which d,l-61 is resolved into sodium D-pantoate (D-63) and L-61 by a fungal aldonolactonase (capacity: 3500 t a−1).75 The enzyme is highly selective and extremely stable in the form of immobilized fungal mycelia; >90 % activity is observed after 180 runs (or 180 days).

Scheme 26.

Production of D-pantothenic acid (D-62) from D,L-pantolactone (D,L-61). Sodium D-pantoate (D-63) is formed in the enantioselective ester hydrolysis and is then converted into D-61. The enantiomer L-61 is racemized to D,L-61 and resubjected to the reaction sequence.

3.3.1.6. Dehalogenase Processes

The enantioselective dehalogenation of (R)-2-chloropropionic acid ((R)-64) occurs with inversion of stereochemistry to give (S)-64 and (S)-lactic acid ((S)-65; Scheme 27). This transformation is used in the industrial synthesis of (R)-α-phenoxypropionic acid ((R)-42).76

Scheme 27.

Industrial synthesis of (R)-α-phenoxypropionic acid ((R)-42, R=H, Ar=Ph) starting from (R)-o-chloropropionic acid ((R)-64).

3.3.1.7. Enzymatic Hydrolysis of Nitriles

The hydrolysis of nitriles by nitrilases or nitrile hydratase/amidase systems in combination with hydrocyanic acid chemistry offers useful access to α-hydroxycarboxylic acids in a manner analogous to HNL processes.77 In recent years a number of nitrilase processes have become established, for example, at Lonza (Section 3.3.1). BASF and Mitsubishi Rayon also use nitrilase processes for the production of (R)-mandelic acid ((R)-37) and its derivatives on a multiton scale.78 (R)-Mandelic acid ((R)-37) is isolated by crystallization from the ammonium salt formed in the biocatalytic reaction. Because of the in situ racemization of the cyanohydrin 66, this process represents a dynamic kinetic resolution (Scheme 28). If the rate of interconversion of the enantiomers of 66 is faster than the enzymatic hydrolysis, 67 can be converted completely into (R)-37. An advantage of the nitrilase process is that no organic solvents are required. Moreover, the synthesis of a nitrile is avoided, as the cyanohydrin 66 is formed from 67 and HCN under the reaction conditions.79

Scheme 28.

Industrial synthesis of (R)-mandelic acid ((R)-37) through the enantioselective hydrolysis of the cyanohydrin 66 (nitrilase process).

In principle nitrile hydratase/amidase systems, in which nitrile hydrolysis occurs via the free amide, also provide access to chiral carboxylic acids.80 However, only a small number of enantioselective nitrile hydratases are known,81, 82 for example, the nitrile hydratase for the resolution of (R,S)-2-(p-chlorophenyl)-3-methylbutyronitrile (68) to (S)-69 (Scheme 29).83 The nonselective nitrile hydratase from Rhodococcus rhodochrous J1 is used on an industrial scale for acrylamide production (Nitto process, >20 000 t a−1)84 and nicotinamide synthesis (Lonza, 3000 t a−1), and regioselective nitrile hydratase/amidase systems are used for the hydrolysis of dinitriles.85 In most cases the racemic amides are resolved, so that a maximum yield of 50 % is possible. An example of such a process is the production by Lonza of (S)-piperazinecarboxylic acid ((S)-70), an intermediate in the synthesis of the anti-HIV drug crixivan, whereby a Klebsiella amidase is used to resolve racemic (R,S)-piperazine-2-carboxamide ((R,S)-71; Scheme 30).86

Scheme 29.
Scheme 30.

Selective conversion of racemic (R,S)-piperazine-2-carboxamide ((R,S)-71) into (S)-piperazinecarboxylic acid ((S)-70).

3.3.2. Fermentation

Owing to the advances made in the areas of strain development and metabolic engineering, microbial syntheses have become possible and economically attractive for many compounds produced on a large scale. These include amino acids, as well as lactic acid (Lactobacilli fermentation), ascorbic acid (72, vitamin C), and pantothenic acid (62, vitamin B5). In terms of the pharmaceutical market, penicillins must also be mentioned.

The world-wide production of vitamin C (72; 80 000 t a−1) still relies on the Reichstein process.87 Biotechnological improvements are based in the first instance on the fermentation production of 2-keto-L-gulonic acid (73), an open-chain precursor of 72, from D-glucose (D-35) (Scheme 31). These processes were compared recently by Hancock and Viola.88

Scheme 31.

Production of vitamin C (72) from D-glucose (D-35).

In the 1990s researchers at Takeda Chemical Ind. developed a fermentation process in E. coli for the production of D-pantothenic acid (D-62), whereby high end concentrations were observed upon feeding with β-alanine (74; >60 g L−1 of D-62 after a fermentation time of 72 h).89 Researchers at Degussa, in cooperation with the research groups of Pühler and Sahm, are currently working on a method based on this process for the synthesis of pantothenic acid by fermentation in E. coli90 and Corynebacterium glutamicum,91 respectively. The genes relevant for pantothenic acid biosynthesis were over-expressed, and the production of the precursor α-ketoisovalerate (75) was increased by ilvBNCD over-expression and ilvA deletion. For the workup of secreted D-62, the cells are separated from the fermentation broth. The pantothenic acid present as a mixed salt is converted into calcium D-pantothenate, for example, by addition of calcium hydroxide, concentration, and spray drying (Scheme 32).92

Scheme 32.

Biosynthesis of deprotonated D-pantothenic acid (D-62). MTHF= 5-methyltetrahydrofolate, THF=tetrahydrofolate.

4. Production of Amines

4.1. Classical Chemical Processes

4.1.1. Crystallization with Chiral Carboxylic Acids

The crystallization of diastereomeric salts of chiral carboxylic acids with chiral amines is still of considerable importance today in the isolation of enantiomerically pure amines (Scheme 33).93 Thus, (R)- or (S)-1-phenylethylamine ((R)- or (S)-38) can be obtained on an industrial scale by crystallization either with (S)-malic acid ((S)-53)94 or (R)-mandelic acid ((R)-37),96 or with (R,R)-tartaric acid ((R,R)-33),95 respectively (Scheme 34). Racemate resolution with (R)-37 is of particular commercial interest. During process optimization it was found that one equivalent of (R)-37 could be replaced by acetic acid (76) (Scheme 35).96a During the process the R-38R-37 is formed with d.r.≈97.5:2.5.

Scheme 33.

Classical racemate resolution through fractional crystallization of salts of racemic amines and enantiomerically pure carboxylic acids.

Scheme 34.

Racemate resolution of (R,S)-1-phenylethylamine ((R,S)-38) through fractional crystallization.

Scheme 35.

Mandelic acid (37) has proved to be a convenient resolving agent for a large number of amines. Racemates of substituted aryl alkyl amines, such as 1-(m-methoxyphenyl)ethylamine (77)97 and 1-(o,p-dichlorophenyl)ethylamine (78),98 can also be resolved with 37 (Scheme 36).

Scheme 36.

Racemate resolution with (R)- and (S)-mandelic acid ((R)- and (S)-38).

Another versatile resolving agent for the production of chiral amines is L-phenylcarbamoyllactic acid (L-81), which can be prepared readily and cheaply from the commercially available starting materials ethyl L-lactate (L-79) and phenylisocyanate (80; Scheme 37).99 For example, (R)-1-phenylethylamine ((R)-38) is isolated with >99 % ee upon treatment of (R,S)-38 with L-81.99b Furthermore, L-81 is suitable for the resolution of 1-(p-chlorophenyl)ethylamine (82), a key intermediate in the synthesis of a chiral fungicide,100 as well as 1-(1-naphthyl)ethylamine (83)99b and ephedrine (rac-84; Scheme 38).99b

Scheme 37.

Synthesis of L-phenylcarbamoyllactic acid (L-81), a reagent for the separation of the enantiomers of chiral amines.

Scheme 38.

Examples of racemate resolution with L-81.

Derivatives of L-81 are also accessible by treatment of L-lactic acid (L-79) with substituted aryl isocyanates (Scheme 37).101 However, only L-79 is available inexpensively, and therefore usually only one enantiomer of the amine can be obtained.

4.1.1.1. Dutch Resolution

A notable variant of classical racemate resolution is the so-called “Dutch resolution” developed by researchers at DSM and Syncom BV.102 In an attempt to shorten the search for a suitable resolving agent for an amine through a combinatorial approach a mixture of several optically active acids was used, whereby salt precipitated that contained not one, but several acid anions.

In a typical experiment racemic 1-(3-methylphenyl)ethylamine ((R,S)-85) is treated with one equivalent of an equimolar mixture of the chiral phosphoric acid esters 86, 87, and 88. Upon mixing, an amine salt that contains the amine (S)-85 in nearly enantiomerically pure form precipitates immediately (Scheme 39); the phosphates 86, 87, and 88 (5:5:1) act as counterions. Even after repeated recrystallization of the precipitated salt the counterion ratio does not change.

Scheme 39.

Dutch resolution: The precipitation of salts of enantiomerically pure amines that contain a mixture of chiral acid anions.

The method is highly versatile, and various mixtures for the resolution of racemic amines and carboxylic acids have become since established (Scheme 40). However, as complex mixtures of the resolving agents are present in the mother liquor of the precipitated salt, the reuse of this auxiliary is problematic.

Scheme 40.

Dutch resolution: Frequently used reagents.

4.1.2. Reduction of C[DOUBLE BOND]N Bonds

The production of optically active amines by reduction of a prochiral precursor 89 containing a C[BOND]N double bond has been treated in detail in several review articles (Scheme 41).103 From an industrial viewpoint the problems generally lie in the synthesis of the precursor 89 and in the cleavage of the auxiliary group X from 90 to give the free amine 91 (for example, cleavage of the N[BOND]N bond in optically active hydrazines with SmI2).104 The diastereoselective hydrogenation of Schiff bases 92 derived from a ketone and (R)-38105a,b to secondary amines 93 can also be carried out on a large scale (Scheme 42).

Scheme 41.

Prochiral precursors 89 with a C[BOND]N double bond for the enantioselective synthesis of amines.

Scheme 42.

The hydrogenation of the optically active Schiff base 95 from the electron-rich aryl ketone 94 gives a secondary amine 96 (Scheme 43). The two benzyl positions in 96 can be differentiated, so that selective cleavage of one benzylic C[BOND]N bond leads to (R)-97.105c,d The diastereoselectivity in the hydrogenation of 95 is critically dependent upon the reaction conditions: At increased temperatures the selectivity observed is higher than at room temperature! The primary amine (R)-97 can be obtained in greater than 99 % ee through a single precipitation of the hydrochloride.105c

Scheme 43.

The three-step method developed by Burk and co-workers for the synthesis of optically active amides 98 by enantioselective catalytic hydrogenation of 99 with RhI / DuPHOS(104) is also interesting in terms of technical applications.106 The enamides 99 are available in turn by reduction of the oximes 100 with iron (Scheme 44). The amide 103 is obtained from pinacolone oxime (101) via the enamine 102 in nearly enantiomerically pure form in the presence of just 0.1 mol % of the catalyst.

Scheme 44.

Enantioselective catalytic hydrogenation with RhI / DuPHOS(104) to give optically active amides.

Researchers at Solvias 107 and Avecia Ltd.108 take advantage of the facile cleavage of P[BOND]N bonds in their syntheses of optically active amines. The method of Solvias starts from the ketone oximes 105, which are readily converted into diphenylphosphorylketimines 107 by treatment with chlorodiphenylphosphane (106; Scheme 45).109 The asymmetric catalytic hydrogenation in the presence of a RhI / josiphos(108) catalyst system110 results in the formation of optically active phosphonamides 109, from which the corresponding chiral amines (R)-110 are released readily by hydrolysis under acidic conditions. Under optimal conditions just 0.2 mol % of the RhI / 108 catalyst is sufficient for formation of the product 109 with >99 % ee at complete conversion of 107. When para-substituted acetophenones are used, the amines 110 are produced with significantly lower enantioselectivities (30–97 % ee).

Scheme 45.

The asymmetric hydrogenation of diphenylphosphorylketimines 107 with a RhI / josiphos(108) catalyst is the key step in the production of chiral amines (R)-110.

Avecia Ltd. uses a very similar approach in the synthesis of (R)-1-(1-naphthyl)ethylamine ((R)-83). Starting from ketone 111 a phosphorylated imine 113 is produced with diphenylphosphoramidate (112). Catalytic asymmetric transfer hydrogenation (CATHy; see Section 3.2.2) of 113 with the Rh complex 114 leads to the optically active phosphorylated amine 115, from which the amine (R)-83 is released by hydrolysis (Scheme 46). With 0.1 mol % of 114 and substrate concentrations of about 0.5 mol L−1 the products are obtained with >99 % ee at satisfactory conversions.108b

Scheme 46.

Synthesis of enantiomerically pure (R)-1-(1-naphthyl)ethylamine ((R)-83; Avecia Ltd.).

4.2. Biotechnological Processes

4.2.1 Transaminations

At the end of the 1980s a new biotechnology process for the production of optically active amines was devised at Celgene.111 In this process transaminases are used, which are able to convert carbonyl compounds, such as ketones or α-keto acids 117 or 118, into amines 119 and 116, respectively. The enantioselective transfer of an amino group is dependent upon the cofactor pyridoxal phosphate (120; Scheme 47). To suppress the reverse reaction and shift the equilibrium in favor of the desired product, the reagents are normally used in different stoichiometries. The concentration of lipophilic substrates in the aqueous phase can also be kept low by using a two-phase system with an organic phase.

Scheme 47.

Enantioselective transamination with pyridoxal phosphate (120) as the cofactor.

As researchers at Celgene have developed both S- and R-selective transaminases, both enantiomers of an amine 119 are accessible with this technology. Furthermore, the reaction can be carried out within the context of a synthesis or of a kinetic resolution (Scheme 48). In a synthesis the prochiral ketone 121 accepts an amino group from the amino-group donor isopropylamine (122). The amine (S)-123 and acetone (124) are formed in more than 90 % yield. It is possible to resolve the amine (R,S)-123 with the same transaminase. In this case (S)-123 is converted into the ketone 121. As amine-group acceptors low-molecular-weight aldehydes are used, such as propionaldehyde (125), from which propylamine (126) is obtained. When α-ketocarboxylic acids are used as the amino-group acceptor amino acids are formed.

Scheme 48.

Examples of S-transaminase-catalyzed asymmetric amino-group transfer (Celgene).

The Celgene technology has already been used on a 2.5-m3 scale. The transaminases accept many different aliphatic and aromatic ketones and amines as substrates.111d. However, one disadvantage is that in principle the reaction can only be carried out either in an aqueous solution or in mixtures of water and an organic solvent. For this reason only low product concentrations can usually be attained with hydrophobic substrates, and product mixtures from racemate resolution (Scheme 48) can only be separated by cumbersome methods.

4.2.2 Kinetic Resolution

At the beginning of the 1990s initial work on the enantioselective hydrolysis of racemic amides was carried out at Shell.112 Both enantiomerically pure (S)-1-phenylethylamine ((S)-38) and, after extended reaction times, enantiomerically pure (R)-127 were thus produced through the selective, enzyme-mediated hydrolysis of racemic N-1-phenylethylacetamide ((R,S)-127; Scheme 49). The use of whole cells proved disadvantageous, as did the long reaction times required as a result of the low selectivity to provide the enantiomerically pure R amide.

Scheme 49.

Enantioselective hydrolysis of racemic N-(1-phenylethyl)acetamide ((R,S)-127).

In the middle of the 1990s methods for the selective hydrolysis of the racemic acetamide (R,S)-128 of 1-(p-chlorophenyl)ethylamine (82) were developed at Bayer.113 It was found that lipase B from Candida antarctica was a suitable catalyst for the selective hydrolysis of (R)-128 to the free amine (R)-82 (Scheme 50). As high concentrations of the catalyst are apparently required for high space–time yields to be observed, this process is not yet exploited industrially.

Scheme 50.

Selective hydrolysis of (R,S)-128 (Bayer).

Alternatively, the kinetic racemate resolution of amines can be carried out in a single synthetic step. In this case the racemic amine (R,S)-116 is treated with an acylating agent 129 in the presence of a catalyst. The enantiomer (R)-116 undergoes acylation to the amide (R)-130, whereas (S)-116 remains unreacted as the free amine (Scheme 51). Fu and co-workers recently described the successful use of planar-chiral DMAP derivatives, such as 131, as acylation catalysts in combination with oxazoles of type 132 in the kinetic resolution of amines (Scheme 52).114 Enantiomeric enrichment by a factor of 10–30 was observed in the formation of the amide.

Scheme 51.
Scheme 52.

Enantioselective amide formation in the presence of the ferrocene derivative 131 as the catalyst; selectivity: 10–30.

However, until now many results obtained through the use of biocatalysts in combination with acylating agents could not be surpassed (Table 2). An example is the selective acylation of (R)-1-phenylethylamine ((R)-38) in a process patented by BASF: In a single step (S)-38 is produced in nearly enantiomerically pure form along with the amide (R)-133 (Scheme 53).116 After separation of (S)-38 by distillation or extraction (R)-38 is released through basic hydrolysis of the amide. In certain solvent mixtures the hydrolysis proceeds in nearly quantitative yield without racemization.116, 125 A remarkable feature of this process is the broad substrate tolerance of the catalyst. Thus, a wide variety of amines have been resolved, in some cases on a multiton scale (Scheme 54). The versatility of the process with regard to possible substrates, the ability to recycle undesired enantiomers by racemization,126 and the recovery of the acylating agent 127 are the main advantages of the process. At the beginning of 2002 a new tailor-made, cGMP-compliant (cGMP: current Good Manufacturing Practices) plant went on stream at BASF, which can produce more than 1000 t a−1 of optically active amines. In another BASF plant 2500 t a−1 of (S)-(1-methoxy)-2-propylamine ((S)-164) can be produced by this process (see Section 4.3.2).

Scheme 53.

Lipase-catalyzed enantioselective amide formation from (R,S)-38 to give (S)-38 and (R)-133. The hydrolysis of (R)-133 gives (R)-38 (BASF).

Scheme 54.

The resolution of many chiral amines is possible by the BASF process (Scheme 53). Some compounds are resolved on a multiton scale.

Table 2. Biocatalysts and acylating agents used in the kinetic resolution of racemic amines.
BiocatalystAcylating agentReference
Subtillisin carlsbergThumbnail image of 115
Burkolderia plantariiThumbnail image of 116a
Candida antarctica lipase BThumbnail image of 117
 Thumbnail image of 118
 Thumbnail image of  
 R=C3H7119a
 R=C7H15119b
 Thumbnail image of 120
 Thumbnail image of  
 X=OCH3116c, 121a
 X=Cl 121a
 X=CN 121a
 X=COOR121b
 Thumbnail image of 122
 Thumbnail image of 123
Penicillin acylaseThumbnail image of 124

4.3. Production of Optically Active Amino Alcohols

4.3.1. Classical Chemical Processes

4.3.1.1. Crystallization with Chiral Carboxylic Acids

(S)-2-Aminobutanol (S-134) is one of the most important amino alcohol intermediates. It is required for the synthesis of the tuberculostatic ethambutol (S,S)-135; Scheme 55). This drug must be administered in its enantiomerically pure form, as the (R,R)-135 causes blindness128. Enantiomerically pure (S)-134 is obtained from the racemate by crystallization with L-tartaric acid (L-33).129

Scheme 55.

The synthesis of the tuberculostatic ethambutol ((S,S)-135) proceeds via (S)-2-aminobutanol ((S)-134).

With the development of protease inhibitors for the treatment of AIDS130 there has been intense demand for optically active intermediates since the beginning of the 1990s. Thus, during the course of the development of the drug indinavir (136) at Merck & Co considerable efforts were made world-wide toward the synthesis of (1S,2R)-1-aminoindan-2-ol ((1S,2R)-137; Scheme 56). However, no chemical131 or biocatalytic132 asymmetric syntheses can compare with the classical method, in which the racemate rac-137 is obtained through an elegant synthetic sequence from indene oxide (138) and in a second step resolved with (R,R)-33 (Scheme 57). The crystallization of the tartrate as a hydrate is important:133 The tartrate is first precipitated from methanol and then recrystallized from water to give enantiomerically pure (1S,2R)-137.

Scheme 56.
Scheme 57.

Synthesis of enantiomerically pure (1S,2R)-1-aminoindan-2-ol ((1S,2R)-137), an intermediate in the synthesis of the anti-HIV agent indinavir (136, Scheme 56; Merck & Co.)

4.3.1.2 Reduction of Amino Acids

Researchers at Degussa have opened up an important route to amino alcohols 139 with a primary hydroxy functionality based on substrates from the chiral pool. Amino acids 140 are reduced with NaBH4 in the presence of an activator (I2134a or H2SO4134b), which transforms the NaBH4 into B2H6 in situ (Scheme 58). The reduction takes place with retention of the configuration of the stereogenic center, and the products are isolated in good yields. L-Valinol (141), L-tert-leucinol (142), L-phenylglycinol (143), L-phenylalaninol (144), and L-neopentylglycinol (145) are prepared on a ton scale by this method (Scheme 59).135

Scheme 58.

Reduction of the natural amino acids L-140 with NaBH4 to the enantiomerically pure amino alcohols 139.

Scheme 59.

The amino alcohols L-valinol (141), L-tert-leucinol (142), L-phenylglycinol (143), L-phenylalaninol (144), and L-neopentylglycinol (145) are produced industrially through reduction of the corresponding amino acids.

4.3.1.3. Opening of Epoxides

A very elegant method for the preparation of (1S,2R)-137 is based on the enantioselective epoxidation of indene by Jacobsen and co-workers131a (Scheme 80, Section 6.1.3). The indene oxide (1S,2R)-138 thus obtained with 84–86 % ee is transformed into the oxazoline 146 by treatment with acetonitrile, and subsequent hydrolysis affords the desired amino alcohol (1S,2R)-137. Through this reaction sequence the enantiomeric excess of the epoxide (1S,2R)-138 is transferred to the product (1S,2R)-137 (Scheme 60). However, for the synthesis of indinavir (136) the optical purity of (1S,2R)-137 must be improved by crystallization with (R,R)-tartaric acid ((R,R)-33) (Scheme 57).131a,b

Scheme 60.

The treatment of indene oxide ((1S,2R)-138) with acetonitrile, followed by hydrolysis, gives (1S,2R)-137. The enantiomeric excess of the starting epoxide, which is available through the Jacobsen epoxidation of indene, is transferred to the product in this process.

The optically active amino alcohol 148 is required for the synthesis of the HIV-protease inhibitor nelfinavir (147; Agouron, Roche; Scheme 61). This key intermediate is produced on an industrial scale from the meso epoxide 149. Upon treatment of 149 with (R)-1-phenylethylamine ((R)-38) in the absence of a catalyst the diastereomeric amino alcohols 150 and 151 are formed in equal amounts (Scheme 62). After separation of the desired diastereomer 150 by crystallization, the chiral auxiliary is removed by debenzylation to give 148 in an overall yield of 37–40 % (based upon 149).136 A significant advance was made with the addition of a catalytic amount of Ti(OiPr)4 and (S)-1,1′-bis-2-naphthol ((S)-152). Under these conditions the ring opening of the meso epoxide 149 takes place with higher diastereoselectivity, and 148 can be isolated in 91 % yield.137

Scheme 61.
Scheme 62.

Selective ring opening of the epoxide in 149 with Ti(OiPr)4 and (S)-1,1′-bi-2-naphthol ((S)-152) to give the diastereomeric amino alcohols 150 and 151 in a >9:1 ratio.137

4.3.2. Biotechnological Processes

4.3.2.1. Amino Acids from Hydroxy Acids

The known high stereoselectivity of the enzyme-catalyzed reduction of 1,3-dicarbonyl compounds is exploited by Biocatalytics Inc. in the production of optically active amino alcohols.138a,b In the first step a β-ketoester 153 is reduced to a β-hydroxyester by enzyme catalysis. Depending on the enzyme used the diastereomeric β-hydroxyesters 154 and 155 are isolated in nearly enantiomerically pure form (Scheme 63). Hydrazinolysis then affords the hydrazides 156 and 157, respectively, which are transformed into the amino alcohols 158 and 159, respectively, by the Curtius rearrangement.

Scheme 63.

Enzyme-catalyzed stereoselective reduction of the 1,3-dicarbonyl compound 153 (Biocatalytics Inc.).

4.3.2.2. Kinetic Resolution

Effective access to optically active amino alcohols is provided by an enzyme-catalyzed kinetic resolution established at BASF. To suppress nonspecific side reactions the hydroxy functionality is protected as an ether. Thus, resolution of trans-2-aminocyclopentanol (rac-160) gives the amine (S,S)-160 and the amide (R,R)-161, in each case with approximately 25 % ee. The reaction of the corresponding benzyl ether rac-162, however, proceeds with high enantioselectivity to give the amine (S,S)-162 and the amide (R,R)-163) with >99.5 % and 93 % ee, respectively (Scheme 64).138c,d This method can also be used with methyl ethers. For example, the enantiomers of racemic (1-methoxy)-2-propylamine ((R,S)-164) can be separated (Scheme 65). The resulting (S)-164 (“S-MOIPA”), formed along with the amide (R)-165, is required for the production of the chiral corn herbicide 166 (“Outlook”, BASF). BASF currently produce up to 2500 t a−1 of (S)-164 by enzyme-catalyzed kinetic resolution.

Scheme 64.

Enzyme-catalyzed kinetic resolution of racemic trans-2-aminocyclopentanol (rac-160) and its benzyl ether (rac-162; BASF).

Scheme 65.

Resolution of racemic (1-methoxy)-2-propylamine ((R,S)-164) to give (R)-165 und “S-MOIPA” ((S)-164), an intermediate in the synthesis of the chiral corn herbicide 166 (“Outlook”, BASF).

5. Production of Alcohols

5.1. Classical Chemical Processes

5.5.1. Asymmetric Hydrogenation of Ketones

The work of Noyori et al.139 played a decisive role in the development of the asymmetric hydrogenation of ketoesters 167 and ketones 168. Ruthenium complexes of binap140 and derivatives, such as tol-binap141 and segphos,142 are used as catalysts. Industrial development of these reactions up to the ton scale was advanced by Takasago Inc. in Japan. (R)-Propanediol (169) can be obtained in this way with up to 94 % ee (Scheme 66). Precursors of β-lactam antibiotics are also accessible by this route,143 sometimes by exploiting dynamic kinetic resolution. Researchers at Roche developed the biphemp ligands 170;144 the conversion of ketopantolactone (171) into (R)-pantolactone (172) was carried out in a pilot plant (Scheme 67).

Scheme 66.

RuII / tol-binap-catalyzed asymmetric hydrogenation of carbonyl compounds.

Scheme 67.

Catalytic asymmetric hydrogenation of ketopantolactone (171) with RhI / (R)-170 (Roche).

In more recent work Noyori et al. report efficient transfer hydrogenation catalysts145 with which formic acid or 2-propanol serve as hydrogen donors, rather than H2. An oxidative variant has also been described.146 The heterogeneous asymmetric hydrogenation of Blaser and Studer,147 which provides a precursor for ACE inhibitors, was operated on a ton scale at Ciba-Geigy. A cinchona-doped platinum catalyst was used for the hydrogenation of the ketoester 173 to the 4-phenylhydroxybutyrate 174 (Scheme 68).

Scheme 68.

5.1.2 CBS Borane Reductions

An efficient catalyst based on a chiral modified borane pioneered by Itsuno and co-workers148 has undergone considerable further development by Corey et al.149 Borane is used as an adduct with THF or dimethylsulfide as a stoichiometric reducing agent. The method has a broad scope and provides products with high ee values within short reaction times, not only with aryl alkyl ketones. Thus, substituted styrene oxides 176 can be prepared with high enantioselectivites in a two-step sequence from halogenated acetophenone derivatives 175 (Scheme 69).

Scheme 69.

CBS borane reductions of halogenated acetophenone derivatives 175.

Both enantiomers of the diphenylprolinol auxiliary are available and recoverable by extraction. The handling of relatively large quantities of the borane adduct is cumbersome and thus a disadvantage of this method. However, this disadvantage may be circumvented by generating borane in situ from sodium borohydride and trimethylchlorosilane150 or a Lewis acid. Chemists at Merck, USA and Sipsy have described industrial applications of the process. 151, 152 Commercially available borane adducts are frequently stabilized with sodium borohydride; this can strongly influence the enantioselectivity of the reaction.153 A review by Deloux and Srebnik provides a comprehensive overview of borane-catalyzed reactions.154

5.1.3. Kinetic Resolution by Epoxide Hydrolysis

Jacobsen and co-workers developed the (salen)cobalt complex (R,R)-(salen)CoOAc (177) as a catalyst for the enantioselective hydrolysis of racemic epoxides 178.155 Through epoxide ring opening the S diols 179 are formed with excellent selectivities, and the remaining R epoxides 180 are also isolated with high enantiomeric purities (Scheme 70; see also Section 6.1.4).

Scheme 70.

Enantioselective hydrolysis of racemic epoxides 178 under the catalysis of the (salen)cobalt complex 177.

Not only water but also phenols can be used as nucleophiles to open epoxide rings highly selectively.156 The resulting 1-phenoxy-2-alkanols are intermediates for the synthesis of β-blockers.157

5.1.4. Further Methods

5.1.4.1. Kinetic Resolution of Alcohols with Organic Catalysts

This process has already been discussed in review articles by Dalko and Moisan158 and by Jacobsen and co-workers,196 which provide an excellent overview of more recent developments, for example, chiral ferrocene derivatives as catalysts for nucleophilic acylation, kinetic resolution, resolution with peptide enzyme mimetics, and kinetic acylation. No industrial applications of these reactions are yet known.

5.2. Biotechnological Processes

5.2.1. Enzyme-Catalyzed Resolution

Enzymatic acylations for the resolution of racemic alcohols were developed as early as the beginning of the 1980s (Scheme 71)159 Racemic alcohols 181 are treated with an acylating agent 182 under enzyme catalysis: One enantiomer 183 remains unconverted, whereas the second enantiomer is esterified to 184. Bacterial and fungal lipases are normally used as biocatalysts.

Scheme 71.

Resolution of racemic alcohols through enzymatic acylation.

Esterases also play a part. A number of research groups have made use of the astonishing stability of enzymes in organic solvents, which offer various advantages over aqueous media, such as higher substrate concentrations and simpler workup. Moreover, it is easier to separate the enzyme and to develop an industrially interesting continuous process.

Enzymatic resolutions are usually carried out irreversibly, for example, with a vinyl ester as an acylating agent.160 The vinyl alcohol formed in the transesterifaction tautomerizes rapidly and irreversibly to acetaldehyde; the equilibrium is thus shifted in favor of the products. Many lipases are sensitive toward acetaldehyde, which forms Schiff bases with free amine groups. In a milder variant with propenyl esters, acetone is released.161 Anhydrides, such as succinic anhydride (186), and diketenes are also suitable as acylating agents. The uncoverted enantiomer is separated from the acylated enantiomer 187 by distillation or, when succinic anhydride is used, by acid–base extraction Scheme 72). This process has been developed on an industrial scale at BASF for the synthesis of styrene oxides 176.

Scheme 72.

The enantioselective synthesis of styrene oxides 176 through the enzymatic acylation of α-chloroalcohols.

In general the selectivities observed are high, but enzymes, too, have their weaknesses with substrates that are difficult to differentiate sterically, such as 2-butanol. Not only high selectivity, but the highest possible activity is also important, so that good space–time yields can be attained in industrial applications. These and other properties, such as solvent tolerance, substrate versatility, and temperature spectrum, can now be improved remarkably in a targeted manner by enzyme engineering.162 With modern genetic methods it is possible to produce several thousand mutants per day. To make it possible to screen such a large number of enzyme variants as quickly as possible, sophisticated screening systems must be developed.163

In most cases only one enantiomer of a specific intermediate in the synthesis of an active compound is required. For economic and ecological reasons the undesired enantiomer should be recycled, for example, by oxidation/reduction or by catalytic dehydrogenation/hydrogenation. More recent work by the research groups of Park,164 Bäckvall,165 Williams,166 and others involves dynamic, kinetic resolution, whereby the nonacylated enantiomer is racemized in situ by a transition-metal-catalyzed Meerwein–Ponndorf–Verley reduction or Oppenauer oxidation. The resistance of many lipases toward these harsh conditions is astonishing. Up to 10 mol % of KOtBu or KOH and temperatures of up to 70 °C are tolerated. As yet no industrial application of this process is known.

5.2.2 Fermentation Processes

Microbial reactions with yeasts are amongst the oldest chemical processes used. As early as 1874 Dumas recognized the reductive effect of a Saccharomyces species.167 Csuk and Glänzer provide a comprehensive overview;168 industrially interesting biotransformations have been reviewed by Liese, Seelbach, and Wandrey.169

In fermentation processes for the production of optically active alcohols bacterial dehydrogenases in whole living cells are usually used. For example, a process at Daicel provides access to the hydroxycarboxylic acid 189, an intermediate in the synthesis of ACE inhibitors, through reduction of the α-keto acid 188 (Scheme 73).

Scheme 73.

Reduction of the α-ketocarboxylic acid 188 by a dehydrogenase (Daicel).

An interesting comparison of chemical and biocatalytic methods for the production of (R)-4-phenyl-2-hydroxybutyric acid, a key intermediate for ACE inhibitors, is presented by Blaser et al.170 A more detailed comparison of chemical and biocatalytic reductions has been made by Hage et al.171 It emerged that there is no sovereign route, but that depending on the nature of the substrate one route is more suitable than another. In practice economic and legal aspects play a critical role.

The reduction of the ketosulfone 190 to the corresponding alcohol 191 is the key step in the synthesis of Trusopt, a glaucoma remedy (Scheme 74). This reaction is carried out on a ton scale172 and gives 191 with >98 % ee.

Scheme 74.

A fermentation method is only economical if high substrate concentrations (>30 g L−1) can be reached and if the product can be separated from the biomass in a straightforward manner. The advantages lie in the generally high selectivity of these reactions and the fact that they can be scaled up readily. Fermentations on a 100-m3 scale can be carried out on an industrial scale without difficulty. Organic nutrients, such molasses or beet syrup, are available inexpensively in large amounts for the growth of the organisms.

Some examples of reductions with isolated dehydrogenases have also been described,173 but these reactions usually require expensive cofactors, such as NAD or FAD. Such processes are only economical when the cofactor can be regenerated in situ in a second catalytic cycle.174 Dehalogenating strains, such as Alcaligenes and Pseudomonas, are also used for the preparation of intermediates.175 Daiso (Japan) used this technology for the synthesis of a portfolio of enantiomerically pure C3 and C4 building blocks, such as 192 (Scheme 75; see also Section 6.2.1). By means of modern gene technology many properties of the strains used can be improved in a targeted manner.

Scheme 75.

Enantioselective dehalogenation of rac-192 (Daiso).

The production of the sympathomimetics ephedrine (193) and pseudoephedrine (194) was one of the first processes to combine biotechnological and chemical synthetic steps (Scheme 76). As early as 1920 it was found that fermenting yeast catalyzed the stereoselective acyloin condensation of benzaldehyde (195) and endogenous acetaldehyde to form (R)-1-hydroxy-1-phenylpropanone (196).176. A route to 193 and 194 by reductive methylamination of 196 was established at Knoll.177 Even today most synthetic ephedrine derivatives are produced by this method (world market 1997: >€50 million).

Scheme 76.

Synthesis of the sympathomimetics ephedrin (193) and pseudoephedrin (194), and an alternative biocatalytic synthesis of 194 through the dynamic kinetic resolution of ephedrone (197; Fuji).

More recent work by researchers at Fuji Chemical Industries illustrates an alternative synthetic route based on biocatalysis (Scheme 76):178 Chemically synthesized rac-2-methylamino-1-phenylpropanone (197, ephedrone) is reduced enantioselectively to pseudoephedrine (194) by a number of different microorganisms. The enantiomer that is not reduced racemizes in situ in this dynamic, kinetic resolution.

6. Production of Epoxides

6.1. Chemical Methods

6.1.1. Sharpless Asymmetric Epoxidation

One method for the direct epoxidation of primary allylic alcohols is the Sharpless epoxidation in which stoichiometric amounts of Ti(OiPr)4 and an enantiomerically pure dialkyl tartrate are used in the presence of an oxygen-atom donor, for example, cumene hydroperoxide (CHP) or tert-butyl hydroperoxide (TBHP).179 The breakthrough for industrial application came in 1986 with the discovery that upon addition of molecular sieves to the reaction mixture only catalytic amounts (5–10 mol %) of the titanium tartrate complex were required.180 The oxidation of allyl alcohol (198) to (R)- and (S)-glycidol (199) were the first Sharpless epoxidations to be carried out on an industrial scale (Arco Chemical; Scheme 77).181 With CHP as the oxygen-atom donor the products were obtained with 91 % ee.182 The isolation of the product was greatly simplified by derivatization of the water-soluble and labile glycidol in situ, for example, as the m-nitrobenzene sulfonate 200. After crystallization the sulfonate is obtained with >99 % ee.183 PPG-Sipsy uses the Sharpless asymmetric epoxidation for the production of 199 under license from Arco Chemical.184 Enantiomerically pure 199 and 200 are used as intermediates in the synthesis of cidofovir and indinavir (136).185 (2S,3R)-1,2-Epoxy-4-penten-3-ol (201), an intermediate in the synthesis of the immunosuppressant FK-506 (tacrolimus), can be synthesized by the Sharpless epoxidation of divinylmethanol (202; Scheme 78).186

Scheme 77.

Industrial application of the Sharpless epoxidation. CHP=cumene hydroperoxide, DIPT=diisopropyltartrate.

Scheme 78.

6.1.2. Sharpless Asymmetric Dihydroxylation

A further route to chiral epoxides is based on optically active diols, which can be converted into the corresponding oxiranes. In 1987 Sharpless and co-workers developed a catalytic procedure for the preparation of chiral vicinal diols. In this method a catalytic amount of a chiral tertiary-amine ligand (e.g. dihydroquinine) and OsO4 are used for the oxidation of an alkene. A stoichiometric oxidizing agent (typically an amine N-oxide or K3[Fe(CN)6]) is used to reoxidize the osmium complex.187

In the production of (R)-m-chlorostyrene oxide (207) on a 100–4000-L scale (Rhodia ChiRex) m-chlorostyrene (203) is first converted into the diol 205 with OsO4 in the presence of the phthalazine diquinine ligand 204 ((DHQD)2PHAL). The diol 205 is then transformed into 207 (>99 % ee) via the primary nitrobenzene sulphonate 206 (Scheme 79).188 By treatment with trimethoxyethane and acetyl bromide cis diols can also be converted into optically active epoxides with retention of configuration.189

Scheme 79.

(R)-m-Chlorostyrene oxide (207) is prepared on an industrial scale by the Sharpless dihydroxylation of 203, followed by cyclization (Rhodia ChiRex). pNS=p-nitrophenylsulfonyl.

To avoid effluent contamination, which occurs through the use of stoichiometric amounts of K3[Fe(CN)6], an electrocatalytic variant of the asymmetric Sharpless dihydroxylation has been developed at Sepracor and ChiRex.190 More recent work has shown that the osmium-catalyzed dihydroxylation can also be carried out at pH 10.4–12 with O2 as the stoichiometric oxidizing agent.191 This opens up new perspectives for the industrial synthesis of chiral diols and epoxides.

6.1.3. Jacobsen Asymmetric Epoxidation

The Jacobsen asymmetric epoxidation is based on a (salen)manganese(iii) precatalyst; hypochlorite is used as the stoichiometric oxidizing agent.192 Reactions of cis alkenes in the presence of 1 mol % of the catalyst afford epoxides with >98 % ee. Jacobsen epoxidations of trans, tetra-substituted, and terminal alkenes give lower yields and lower ee values. ChiRex has exclusive rights to this technology, and applies it on a multiton scale. An example is the epoxidation of indene (208) to (1S,2R)-indene oxide ((1S,2R)-138) with catalytic amounts of the (S,S)-(salen)manganese(III) complex 209 (Scheme 80). Compound (1S,2R)-138 is an intermediate in the synthesis of HIV-protease inhibitors, such as indinavir (136) by Merck (Section 4.3.1.1).186, 193

Scheme 80.

Jacobsen epoxidations: The (S,S)-(salen)manganese(III) complex (S,S)-209 and its R,R enantiomer (not shown) serve as catalysts.

A further example is the epoxidation of ethyl cinnamate (210) in 56 % yield to cis-ethyl-3-phenylglycidate (211; 95–97 % ee), an intermediate in the synthesis of paclitaxel (taxol) and docetaxel (Scheme 80).194 The ligand for the chiral catalyst (R,R)-209 is prepared from 1,2-diaminocyclohexane and 2,4-di-tert-butylphenol on a 100-kg scale.195

6.1.4. Jacobsen Hydrolytic Kinetic Resolution

In 1996 Jacobsen et al. developed a method for the resolution of racemic terminal epoxides by the enantioselective nucleophilic hydrolysis of one enantiomer (see Section 5.1.3, Scheme 70). With the (salen)cobalt(III) catalyst 177 enantioselectivities of >99 % ee and yields of 40–48 % are observed (Scheme 81), which corresponds to a selectivity of E=500. The reaction is particularly suitable for terminal epoxides196 and has been developed by Rhodia ChiRex,197 for example, for the resolution of styrene oxide (176), propylene oxide (180), epichlorohydrin (212), and ethyl 2,3-epoxyacrylate (213) on a ton scale.198

Scheme 81.

Selective hydrolysis of one enantiomer of the epoxide 178 with the (salen)cobalt(III) catalyst 177 (Scheme 70).

The epoxide opening proceeds by second-order kinetics with respect to the catalyst. Therefore it is assumed that a metal ion activates the epoxide as a Lewis acid, while a further metal center forms the counter ion of the nucleophile. As expected, oligomeric (salen)cobalt(III) complexes exhibit higher reactivities and selectivities than monomeric complexes, such as 177.199

Researchers at Synetix are developing a Jacobsen catalyst in which the salen complex is immobilized on zeoliths, which should simplify separation and recycling of the catalyst and lead to less contamination of the product.200 The company Daiso, which produces epichlorohydrin (212), glycidol (199), and chiral precursors by microbial resolution, recently announced the change to the Jacobsen technology in the production of 212. The license was obtained from ChiRex, which also supplies the catalyst. Production capacity is to be extended to 50 t a−1 by 2004.201

6.1.5. Other Methods

A further route to chiral epoxides is based on optically active α-halo alcohols, which are prepared through an enantioselective Darzens reaction.202 In the production of the α,β-epoxycarboxylic acid ester 214, an intermediate in the synthesis of docetaxel, the enolate of the N-α-bromoacyl oxazolidinone 215 reacts with benzaldehyde (67) in an aldol condensation.203 The resulting bromohydrin was then cyclized to the epoxide 214 (Scheme 82).

Scheme 82.

A further example is the addition of chloromethyllithium generated in situ to the chiral α-aminoaldehyde 216 (Scheme 83). The epoxide (2S,3S)-217 was isolated in 65 % yield with 70 % de after column chromatography.

Scheme 83.

6.2. Biotechnological Processes

6.2.1. Racemate Resolution

6.2.1.1. Lipases

By the hydrolysis of racemic glycidylbutyrate (218) with porcine pancreatic lipase Ladner and Whitesides obtained (R)-218 in 89 % of the theoretical yield and with 92 % ee (Scheme 84).204 This process was developed further by Andeno-DSM and implemented for the production of (R)-218 and (R)-glycidol ((R)-199) on a multiton scale.205

Scheme 84.

Synthesis of (R)-218 by porcine pancreatic lipase catalyzed hydrolysis (Andeno-DSM).

In the kinetic resolution of the epoxy alcohol 219 with porcine pancreatic lipase F and acetic anhydride the 2S,3R enantiomer is acylated preferentially. The unconverted (2R,3S)-219 can be isolated in 27 % yield and with 99 % ee (Scheme 85).206 This intermediate can be used in the synthesis of the moth pheromone disparlur ((7R,8S)-220). In the production of styrene oxides at BASF, the corresponding racemic styrene chlorohydrins are resolved with a lipase (Section 5.2.1, Scheme 70).

Scheme 85.

The epoxide (2R,3S)-219 is an intermediate in the synthesis of disparlur ((7R,8S)-220).

Hydrolysis of the undesired enantiomer of the epoxycarboxylic acid ester 221 with a lipase from Candida sp. affords (2R,3S)-221 (Scheme 86). The carboxylic acid 222 formed decomposes spontaneously to CO2 and the aldehyde 223, which is a potent inhibitor of the lipase. Therefore, 223 is transformed into the bisulfite adduct 224 and separated.207 In 1993 the company Tanabe started operation of a hollow-fiber reactor for the production of (2R,3S)-221 with an immobilized lipase from Serratina marescens.208 The glycidate (2R,3S)-221 is used as an intermediate in the synthesis of the calcium antagonist diltiazem.209

Scheme 86.

The enantioselective hydrolysis of the epoxycarboxylic acid ester 221 affords (2R,3S)-221.

6.2.1.2. Epoxide Hydrolases

Archelas and co-workers have been able to prepare both enantiomers of 4-chlorostyrene oxide (225) from the racemate with epoxide hydrolases from Aspergillus niger (AnEH) and Solanum tuberosum (StEH; Scheme 87). AnEH catalyzes the epoxide-ring opening of (R)-225 through the attack of a water molecule at the C-1 position to yield the diol (R)-226 with retention of the stereochemistry. The S epoxide remains unconverted. In contrast, (S)-225 is hydrolyzed with StEH to the same diol (R)-226 through attack at the C-2 position and inversion of the stereochemistry, whereby (R)-225 remains unconverted and can be isolated. Thus, Archelas and co-workers did not only develop a route to enantiomerically pure (R)- and (S)-p-chlorostyrene oxide, but also an enantioselective synthesis of (R)-226, a possible intermediate in the synthesis of the glutamate antagonist eliprodil, in 93 % yield and with 94 % ee starting from racemic 225.210

Scheme 87.

Enantiomerically pure (R)- and (S)-p-chlorostyrene oxide ((R)- and (S)-225) are accessible through enzyme-catalyzed epoxide-ring opening. With both enzymes (R)-226 is obtained as a by-product.

At Merck an epoxide hydrolase from Diplodia gossipina has been used for the resolution of racemic indene oxide (138; Scheme 88). The enantiomer (1S,2R)-138, a precursor for HIV-protease inhibitors, is obtained in 14 % yield and with 100 % ee; indanediol (227) is also isolated (Scheme 88).211

Scheme 88.

Janssen and co-workers isolated (S)-p-nitrostyrene oxide ((S)-228; >99 % ee) by selective azidolysis of (R)-228 with a halohydrin dehalogenase from Agrobacterium radiobacter (Scheme 89). (R)-2-Azido-1-(p-nitrophenyl)ethanol ((R)-229) was obtained with 96 % ee.212

Scheme 89.

The selective azidolysis of rac-228 leads to (S)-228 and (R)-229.

6.2.2. Microbial Methods for Racemate Resolution

In 1994 researchers at Daiso developed a process for the preparation of (S)-2,3-dichloro-1-propanol ((S)-230) by selective assimilation of (R)-230 with Pseudomonas sp. (OS-K-29; Scheme 90).213 The enantiomer (R)-230 could be obtained by degradation of (S)-230 with Alcaligenes sp. (DS-K-S38).214 Under basic conditions (R)- and (S)-230 are converted into optically active epichlorohydrin ((S)- and (R)-212, respectively). Under similar fermentation conditions 3-chloropropane-1,2-diol (192) can also be produced stereoselectively.215 The enantiomers (R)- and (S)-192 are converted into glycidol ((R)- and (S)-199, respectively).

Scheme 90.

Biotechnological preparation of enantiomerically pure epichlorohydrin ((R)- or (S)-212) and glycidol ((S)- or (R)-199) by fermentation (Daiso).

6.2.3. Monooxygenases

Although monooxygenases of many microorganisms catalyze the stereoselective epoxidation of alkenes, their application is problematic for a number of reasons. First, the epoxides are frequently merely intermediates of a decomposition pathway and are not accumulated; second, the epoxides as alkylating agents act as toxins for the microorganisms, thus leading to product inhibition.

Nocardia corallina (Nippon Mining) can be used for the epoxidation of a broad spectrum of terminal and subterminal alkenes, such as styrenes (Table 3). Depending on the class of substrate three different procedures are used.216 In the oxidation of short-chain, gaseous alkenes (C3–C5) the highly toxic epoxide is continuously expelled by using a fast gas flow. In this way propylene oxide can be obtained with 83 % ee from propene in a space–time yield of 23 g L−1 day−1. Epoxides of long-chain alkenes (>C13) are less toxic and can be prepared with cells growing on glucose without product extraction. In the epoxidation of alkenes of medium chain length (C6–C12) and of styrenes the use of resting cells in the presence of a nontoxic organic phase (e.g. hexadecane) has proved effective.217 The organic solvent lowers the concentration of the inhibitory epoxide in the aqueous phase and facilitates continuous product extraction.

Table 3. Epoxidations with Nocardia sp.
 SubstrateConfiguration of the epoxideee [%]
C3–C51,2-propeneR83
 3-chloropropeneS81
C6–C121,2-hepteneR94
 1,2-octeneR91
 1,2-deceneR86
 1,2-dodeceneR87
 o-chlorostyreneR86
 m-chlorostyreneR82
 p-chlorostyreneR72
>C131,2-trideceneR92
 1,2-pentadeceneR81
 1,2-heptadeceneR81

At Shell (R)-phenylglycidyl ether ((R)-232) was obtained by oxidation of 3-phenoxypropene (231) with suspended Pseudomonas oleovorans cells. This process provides (R)-232, an intermediate in the synthesis of β-adrenoreceptor blockers, with >99.9 % ee (Scheme 91).218

Scheme 91.

Styrene monooxygenases from styrene-degrading Pseudomonads were expressed in E. coli cells and used in the selective oxidation of styrene derivatives 233 to the corresponding S styrene oxides (S)-176 (Scheme 92).219 In the presence of an organic phase (dioctylphthalate) yields of 11 g L−1 were observed.220

Scheme 92.

Styrene monooxygenases oxidize 233 selectively to the S styrene oxides (S)-176.

6.2.4. Dehydrogenases

Access to chiral epoxides is also possible from enantiomerically pure α-halo alcohols (Scheme 70), which can be produced from the corresponding chloroketones by enantioselective reduction. The company Kaneka uses dehydrogenases in the form of whole cells for the production of R and S styrene oxides on a pilot-plant scale.

Bristol Meyers Squibb prepared methyl (3S)-3,4-epoxybutyrate (234) from the chlorohydrin (S)-235. The latter compound is available with 99 % ee from the keto compound 236 on a multi-kg scale by enantioselective dehydrogenase reduction with whole cells from Geotrichium candidum (Scheme 93).221

Scheme 93.

7. Conclusions and Prospects

In comparing chemical processes with biotransformations no all-embracing judgment can be made with respect to economic efficiency and environmental impact. Advantages and disadvantages of individual methods must be assessed for specific cases. When the Sharpless epoxidation was compared with microbial epoxidation by life-cycle analysis (LCA), for example, neither process proved to be more environmentally friendly than the other.222 The specific problems of chemical routes are often solvent emission or the toxicity of certain compounds, whereas with microbial procedures low yields relative to the biomass used or low product concentrations cause problems.

Chiral technologies are still developing very rapidly. For their successful implementation on an industrial scale a number of factors are important. Above all, highly versatile technologies are sought. Many chiral intermediates are produced in small quantities (10–100 t), and there are often competing manufacturers. Therefore, methods are desired that have a broad substrate spectrum, do not require specialist equipment, and enable cost-effective access to a comprehensive range of products.

In the case of biotechnological processes substrate concentration is often the deciding factor; in catalytic chemical methods the substrate/catalyst ratio is of critical importance. Today robot-assisted optimization helps us to meet these goals in the shortest possible time.

Catalytic technologies will be the way of the future. Auxiliary-based methods are loosing ground because the quantities of material required in each cycle are too large. Classical racemate resolution still has its place in the rapid provision of enantiomerically pure building blocks. Although dynamic kinetic resolution is still in an early phase of development, it clearly has considerable potential.

However, there will always be scope for a variety of technologies. Which technology is ultimately implemented in a production process often also depends on the general conditions, such as the availability of precursors and equipment, in the individual companies.

Acknowledgements

We thank Dr. Gisela Hieber for her support in procuring market information. Our special thanks go to Tanja Jenak, through whose enthusiastic involvement this Review first took shape.

Biographical Information

Tilo Habicher, born in 1969, studied chemistry at the Universität Leipzig and the Ecole Européenne des Hautes Etudes des Industries Chimiques in Strasbourg. After completing his PhD in 1998 in supramolecular chemistry in the research group of F. Diederich at the ETH Zürich, he took up a fellowship of the Studienstiftung des deutschen Volkes (BASF-research fellowship) at the UC, Berkeley and at The Scripps Research Institute in La Jolla with P. Schultz. There he studied the in vitro evolution of catalytic antibodies on phage surfaces. Since 2000 he has been at BASF AG in the biocatalysis division.

original image

Biographical Information

Maria Keßeler, born in 1969, received a biology degree at the Universität Marburg in 1993 and then took up a scholarship of the Studienstiftung des deutschen Volkes and of the Fonds der Chemischen Industrie for a research stay at the University of the Witwatersrand in Johannesburg. After completing her PhD at the Universität Göttingen (1996) she began at BASF AG in Ludwigshafen in 1997, where she is a laboratory group leader in the biotransformations group of the fine chemicals division.

Biographical Information

Michael Breuer, born in 1965, studied biology at the Universität Bonn. From 1992 to 1993 he held a scholarship of the Boehringer-Ingelheim-Fonds to carry out research in the research group of H. G. Floss at the University of Washington. He completed his PhD at the Universität Bonn under the guidance of E. Leistner on ansamacrolide bioynthesis in Actinomyceten (1995) and has been a laboratory group leader in the fine chemicals and biocatalysis research division at BASF AG since 1996.

Biographical Information

Bernhard Hauer, born in 1955, studied biology at the Universität Hohenheim, where he completed his PhD in 1982 in the research group of F. Lingens on microbial alkaloid decomposition. During his postdoctoral research stay at the University of Chicago he investigated the transposon Tn7 under the guidance of J. A. Shapiro. In 1983 he joined BASF AG, where he is currently in charge of the research area biocatalysis as scientific director. He also completed his habilitation in 1996 at the Universität Heidelberg and is a professor of molecular biology there.

Biographical Information

Thomas Zelinski, born in 1965, completed his PhD in chemistry in 1995 in the research group of Prof. M.-R. Kula at the Universität Düsseldorf. The topic of his thesis was enzymatic reductions. From 1996 to 1997 he undertook postdoctoral research with H. Waldmann at the Universität Karlsruhe, where he investigated a topic in phosphopeptide synthesis. He has been working in the research division fine chemicals and biocatalysis at BASF AG since 1997.

Biographical Information

Klaus Ditrich, born in 1956, studied chemistry at the Universität Marburg and completed his PhD in 1986 on natural product synthesis in the research group of Prof. R. W. Hoffmann. He has been at BASF AG since 1987, first in the group active compound discovery (plant protection) and since 1992 in the division fine chemicals and biocatalysis. His main focus areas include asymmetric catalysis, enzymatic synthesis, and the development of production methods for optically active intermediates.

Biographical Information

Rainer Stürmer, born in 1963, studied chemistry at the Universität Marburg. In 1992 he completed his PhD in the research group of Prof. R. W. Hoffmann with the synthesis of a natural product. During a postdoctoral stay at Harvard University (1992–1993), funded with a fellowship from the Studienstiftung des deutschen Volkes, he carried out research on lanthanoid catalysis under the guidance of Prof. D. A. Evans. He has been a laboratory group leader in the fine chemicals division of BASF AG since 1993. There he is primarily concerned with asymmetric catalysis, enzymatic synthesis, and optically active intermediates.

Ancillary