US20070166811A1

US20070166811A1 - Deracemisation of Amines

Info

Publication number: US20070166811A1
Application number: US11/685,939
Authority: US
Inventors: Marina Alexeeva; Alexis Enright; Mahmoud Mahmoudian; Nicholas Turner
Original assignee: Glaxo Group Ltd
Current assignee: Glaxo Group Ltd
Priority date: 2002-03-19
Filing date: 2007-03-14
Publication date: 2007-07-19
Also published as: GB0206415D0; US8178324B2; WO2003080855A2; JP4965680B2; WO2003080855A3; US20060257978A1; US7208302B2; JP4648631B2; AU2003216848A1; US20110065154A1; AU2003216848A8; JP2010178747A; JP2005520553A; EP2105507A1; EP1485496A2

Abstract

Methods are described for the deracemisation or chiral inversion of chiral amines by enzymatic treatment, such as by a stereoselective enzymatic conversion and either a non-selective or partially selective chemical or enzymatic conversion, simultaneously or sequentially. Also described are methods for selecting a suitable enzyme, particularly a suitable amine oxidase, and for the generation of novel enzymes suitable for use in the deracemisation method.

Description

The present invention relates to a method for the deracemisation or chiral inversion of chiral amines by enzymatic treatment of a mixture of enantiomers. The method employs a stereoselective enzymatic conversion and either a non-selective or partially selective chemical or enzymatic conversion, simultaneously or sequentially. The invention also provides a method for selecting a suitable enzyme, particularly a suitable amine oxidase, and for the generation of novel enzymes suitable for use in the deracemisation method.
Enantiomerically pure chiral amines are valuable synthetic intermediates, particularly for the preparation of pharmaceutical target molecules. Traditionally, chiral amines have been obtained by separation methods such as diastereomeric crystallisation using a chiral acid to form a salt of one of the enantiomers, or by kinetic resolution of a racemate using an enzyme to selectively react one enantiomer allowing easier separation by physical methods such as solvent partitioning or chromatography (1). Whilst such methods can achieve high enantiomeric excess (e.e.), they can yield only a maximum of 50% of the racemic starting material as the required enantiomer. As with many chiral compounds, there is an increasing desire to develop synthetic strategies for amines that involve either asymmetric approaches or which combine resolution with racemisation of the undesired enantiomer, both of which can in principle deliver the product in 100% yield and 100% e.e. Asymmetric methods suggested to date include the use of transaminases for conversion of ketones to chiral amines (2, 3, 4). Furthermore the kinetic resolution of amines using lipases such as Burkholderia plantarii lipase (5, 6) or Candida antarctica lipase (7) has been combined with racemisation of the unreacted amine either by formation of an imine (5,6) or by transfer hydrogenation with Pd/C as the catalyst (7).
An alternative approach, which has been termed deracemisation, involves the stereoinversion of one enantiomer to the other e.g. using a cyclic oxidation-reduction sequence. To date it has been shown that such a system can be applied to the preparation of L-α-amino acids by the use of an enantioselective D-amino acid oxidase in combination with a non-selective reducing agent. The original work (8) reported the stereoinversion of D- to L-alanine, albeit in low yield, using sodium borohydride as the reducing agent. The instability of sodium borohydride at pH 7 precludes its use on a practical scale, and recently we have shown that deracemisation of amino acids can be made more efficient by the use of more suitable reducing agents including sodium cyanoborohydride (13), ammonium formate with Pd/C and also borohydride complexes or amine:boranes (14). However, no-one to date has successfully applied a deracemisation method to amines.

SUMMARY OF THE INVENTION

The present invention provides a method for the deracemisation or chiral inversion (generally referred to herein as enantiomeric conversion) of chiral amines by treatment of a mixture of amine enantiomers with an enzyme capable of catalysing oxidation of the amine in a stereoselective manner and, subsequently or simultaneously, treating the mixture with a reducing agent. The method is applicable to mixtures of enantiomers in varying proportions, including racemic mixtures, and to conversion (epimerisation) of one single enantiomer to the other. For example the method is applicable to mixtures of R and S forms of an amine in a ratio of 1:1, 1:2, 1:5, 1:10, 2:1, 5:1, 10:1, 100:1 or other ratios. The product of the enantiomeric conversion is enriched in the desired enantiomer over the starting material i.e. the desired enantiomer is in enantiomeric excess. Preferably the product comprises a substantially pure single enantiomer. Thus, in preferred embodiments the enantiomeric conversion process of the invention is employed to convert a mixture of amine enantiomers into a composition consisting essentially of a single enantiomer, or is employed to convert one substantially pure amine enantiomer into the other, again in enantiomerically pure form.
The reducing agent may be partially stereoselective or non-stereoselective and may be a chemical reducing agent. Alternatively the reduction may be enzymatically catalysed. If a chemical reducing agent is to be employed, this may advantageously be selected from sodium borohydride, sodium cyanoborohydride, amine:borane complexes or a transfer hydrogenation reagent such as ammonium formate with Pd/C. If the stereoselective oxidation and non-stereoselective (or partially selective) reduction are performed sequentially, in an oxidation-reduction cycle, the cycle may be performed a plurality of times until the desired enantiomeric excess is achieved.
The enzyme capable of catalysing oxidation of the amine in a stereoselective manner may be a monoamine oxidase (MAO), particularly a microbial monoamine oxidase, but any amine oxidase enzyme may be employed. One MAO which may advantageously be employed in the method of the present invention is the Aspergillus niger monoamine oxidase or a variant thereof, for example a variant in which the enzyme differs from wild-type A. niger MAO by incorporation of one or more mutations, especially in the region of amino acids 25-265 and 334-350, particularly prefere4d are enzymes having a mutation at one or more of amino acids 259, 260, 336 and 348, more particularly the mutation N336S or the double mutation N336S, M348K.
The present invention also provides a method of directing the evolution of an originator enzyme by: a) mutating the originator enzyme to create at least one enzyme variant; b) screening said enzyme variant for activity against a homochiral substrate; and c) selecting one or more enzyme variants which show greater activity toward the homochiral substrate than does the originator enzyme. Optionally steps a), b) and c) may be repeated, using the enzyme variant selected in step c) as an originator enzyme. At appropriate stages, the enzyme variant(s) may be assayed against the opposite enantiomer of the substrate, or against a mixture of the substrate enantiomers, to confirm enantioselectivity. In particular embodiments, the originator enzyme is an oxidase which shows activity against amines, for example an amine oxidase, especially a monoamine oxidase. The substrate may be any chiral amine which can be oxidised to an imine, including cyclic secondary amines, for example amines of Formula I:

In which:
a) R is H or C_1-4alkyl; R1 and R2 are independently selected from substituted or unsubstituted C_1-10alkyl, C_1-10alkenyl, C_1-10cycloalkyl, C_1-10heterocycle, C_1-10aryl, C_1-10heteroaryl, C_1-4alkyl-aryl, C_1-4alkyl-heteroaryl, C_1-4alkyl-C_1-6cycloalkyl and C_1-4alkyl-C_1-6heterocycle; or
b) R is H or C_1-4alkyl, R1 and R2 together form a substituted or unsubstituted C_1-10cycloalkyl ring system or C_1-10aryl ring containing one or more heteroatoms; or
c) R and R1 together form a substituted or unsubstituted C_1-10cycloalkyl or C_1-10aryl ring system which may contain one or more heteroatoms and R2 is defined as in a) above.
As used herein, the terms “halo” or “halogen” refer to fluorine, chlorine, bromine and iodine.
As used herein, the term “alkyl” refers to a straight or branched hydrocarbon chain containing the specified number of carbon atoms. For example, C₁-C₃alkyl means a straight or branched hydrocarbon chain containing at least 1 and at most 3 carbon atoms. Examples of alkyl as used herein include, but are not limited to; methyl, ethyl, n-propyl, i-propyl.
As used herein, the term “cycloalkyl” refers to a fully saturated hydrocarbon ring containing the specified number of carbon atoms. The term “cycloalkyl” encompasses single and bicyclic ring structures. Examples of cycloalkyl as used herein include, but are not limited to cyclohexyl, cyclopropyl.
As used herein, the term “aryl” refers to an unsaturated which may be saturated or unsaturated hydrocarbon ring containing the specified number of carbon atoms. The term “aryl” encompasses single and bicyclic ring structures. Examples of cycloalkyl as used herein include, but are not limited to phenyl, naphthyl.
Where cycloalkyl or aryl ring systems contain one or more heteroatoms, these are selected from N, S or O, preferably N. Thus, the terms “heterocyclic” and “heteroaryl” refer to cycloalkyl and aryl groups, respectively, which contain up to three heteroatoms selected from N, S or O, preferably N.
Where one or more of R, R1 and R2, or a ring formed therebetween, are substituted, one to three substituents may be present and are selected from halogen, hydroxy, C_1-3alkyl, C_1-3alkenyl, C_1-3alkoxy, nitro, nitrile and CONH₂.
Conveniently, the target substrate can be used for evolution of enzymes with improved activity and enantioselectivity for that particular substrate.
The invention therefore provides a method of improving the catalytic activity, and optionally also the enantioselectivity, of an amine oxidase, especially a monoamine oxidase, by directed evolution comprising mutation of the enzyme and selection of a mutant having improved activity against a homochiral substrate. The use of an enzyme variant, selected by such a method, in a method for the deracemisation or epimerisation of amines is also provided.
The wild-type amino acid sequence of A. niger MAO is set out in SEQ. ID No. 1. In one embodiment the invention provides a variant of the A. niger MAO in which the enzyme differs from wild-type A. niger MAO in the region of amino acids 334-350, particularly amino acid 336 and/or amino acid 348, more particularly by incorporation of the mutation N336S or the double mutation N336S, M348K. The invention thus provides a variant of the A. niger MAO having the amino acid sequence set out in SEQ. ID No. 2 and, in a further embodiment, a variant of the A. niger MAO having the amino acid sequence set out in SEQ. ID No. 3.
Throughout the specification and the claims which follow, unless the context requires otherwise, the word ‘comprise’, and variations such as ‘comprises’ and ‘comprising’, will be understood to imply the inclusion of a stated integer or step or group of integers but not to the exclusion of any other integer or step or group of integers or steps.
A chiral compound will have two or more enantiomers which are stereochemically dissimilar. A composition which contains more than one enantiomer of a chiral compound is referred to as a “racemic mixture” if it contains the enantiomers in equal or substantially equal amounts. By contrast a composition is “homochiral”, “enantiomerically pure” or a “substantially pure single enantiomer” if it contains a single enantiomer, substantially free of the corresponding enantiomer, or consists essentially of one enantiomer in the absence of the other. By ‘substantially free’ is meant no more than about 5% w/w of the corresponding enantiomer, particularly no more than about 3% w/w, and more particularly less than about 1% w/w is present. “Enantioselective” and “stereoselective” are used herein interchangeably and refer to the tendency of a reaction to favour one enantiomer of a chiral compound over the other. “Partially stereoselective” (or “partially enantioselective”), “non-enantioselective” etc. shall be understood accordingly.
The invention will now be described in more detail, with reference to the accompanying drawings and sequence listings, in which:
FIG. 1 shows the reaction scheme for the oxidation-reduction cycle which results in deracemisation;
FIG. 2 shows diagrammatically the assay used to detect enzymes having the desired enantioselective amine oxidase activity;
FIG. 3 shows the results of 27 enzymes selected from the detection assay of FIG. 2, assayed against L-AMBA and D-AMBA;
FIG. 4 shows the substrate specificity of the N336S, M348K mutant A. niger monoamine oxidase;
FIG. 5 shows the enantioselectivity of the N336S, M348K mutant A. niger monoamine oxidase;
SEQ ID NO:1 shows the amino acid sequence of the wild-type Aspergillus niger monoamine oxidase enzyme (NB we found the sequence to differ by 4 amino acids from that reported by Schilling & Lerch (10,11). These changes most likely represent errors in the original DNA sequencing; two of these differences are also noted by Sablin (12).);
SEQ ID NO:2 shows the amino acid sequence of a variant monoamine oxidase enzyme generated by directed evolution in the following Examples. The mutation N336S is shown in bold face type and underlined;
SEQ ID NO: 3 shows the amino acid sequence of a variant monoamine oxidase enzyme generated by site directed mutagenesis of the N336S mutant enzyme of SEQ ID NO: 2. The mutations N336S and M348K are shown in bold face type and underlined.
Herein we report a significant new extension of deracemisation, by applying the method for the first time to the deracemisation of chiral amines. The reaction follows the general scheme shown in FIG. 1. One enantiomer is converted to the other because the enzyme reacts preferentially with one of the two chiral forms, leaving the other unreacted. The enzyme reaction produces the achiral imine, which does not have a chiral centre at the nitrogen-bearing carbon. Non-selective reduction of the imine results in the creation of a 1:1 mixture of amine enantiomers (if a partially selective reducing agent were employed, an unequal mixture of amine enantiomers would be formed). As this process is allowed to undergo a number of cycles, the amine approaches enantiomeric purity (assuming a sufficiently enantioselective amine oxidase). The yield can approach 100% If the imine can be efficiently reduced before undergoing hydrolysis to the ketone.
Amine oxidases have been classified into two groups, namely Type I (Cu/TOPA dependent) and Type II (flavin dependent) (15). In the catalytic cycle of the Type I enzyme, the intermediate imine remains covalently bound to the protein, which mitigates against intervention at this stage in the reaction to reduce the imine back to the amine. The Type II enzymes have been extensively studied from mammalian sources (9), however microbial sources of Type II enzymes are poorly documented and indeed at the outset of our work there were no reports of enantioselective transformations. Schilling et al., (10, 11) reported the cloning and expression of a Type II monoamine oxidase from Aspergillus niger (MAO-N) and subsequently Sablin et al., (12) purified the enzyme to homogeneity and carried out substrate specificity and kinetic studies. The enzyme was reported to have high activity towards simple aliphatic amines (e.g. amylamine, butylamine) but was also active, albeit at a lower rate, towards benzylamine.
Our studies have revealed that the A. niger MAO-N enzyme possesses very low, but measurable activity towards L-α-methylbenzylamine with even slower oxidation of the D-enantiomer. Thus the enzyme is partially enantioselective. We have carried out in vitro evolution (16, 17) to generate new amine oxidase enzymes having improved catalytic activity and enantioselectivity over the wild-type A. niger MAO-N enzyme. We have also demonstrated the applicability of the evolved mutants in deracemisation reactions.
To generate a large library of enzyme variants we used a mutator strain for random mutation of a plasmid containing an insert encoding the wild-type A. niger amine oxidase enzyme. The E. coli XL1-Red mutator strain (Stratagene) has been employed previously for in vitro evolution experiments and has the advantage that all parts of the plasmid are subject to mutation (cf. error prone PCR where only the gene of interest is mutated) which can result in improved production of enzyme as well as changes to enzyme activity and/or specificity. By carrying out multiple cycles with the mutator strain followed by transformation of the plasmid library into E. coli BL21 cells, we were able to generate a library of around 10⁶variants.
Having generated the mutant enzymes, we assayed them for amine oxidase activity. Amine oxidases are typical of all members of the oxidase family in that they evolve hydrogen peroxide as a by-product. Many of the reported assays for oxidase activity exploit the production of hydrogen peroxide, especially coupled with peroxidase in the presence of a substrate that upon oxidation yields a highly coloured product. We first used aminoantipyrine/tribromohydroxybenzoic acid as the substrate for the peroxidase, which yielded distinct pinkish-red colonies which could be easily visualised. However the coloured product is relatively soluble and hence colour from the active colonies rapidly diffused giving high background intensity. This problem was overcome by switching to 3,3′-diaminobenzadine (DAB) as the substrate which gave rise to a dark pink, insoluble product resulting in both very high definition and contrast of the active colonies. It should be noted that this high-throughput screen should be generally applicable to other oxidase enzymes in addition to amine oxidases.
The library was plated out directly onto nitro-cellulose filters on agar-plates (around 3,000 colonies per plate) and a sub-set of the plates taken through the screening protocol (see FIG. 2). Each filter was transferred to a petri-dish and stored at −20° C. for 24-72 h in order to partially lyse the cells. Thereafter, each plate was treated with a cocktail containing both the assay mixture and also 2% agarose at 60° C. The plates were incubated at 37° C. for 24-48 h after which they was inspected for positive clones which were then removed and plated at higher dilution to isolate pure colonies.
The result of screening a subset (ca. 150,000 clones) of the initial library led to the identification of 35 clones with improved activity towards L-AMBA compared to the wild-type enzyme. Each of these clones was grown on a small scale and assayed against both L- and D-AMBA as cell-free extracts resulting in the best 27 clones which were selected for further study. Each of these 27 were assayed against L-AMBA and D-AMBA, the results of which are presented in FIG. 3. Finally, the mutant enzyme clone which showed greatest improvement over the wild-type enzyme in terms of its selectivity, and activity, towards L-versus D-AMBA was selected for more detailed examination. Further mutation was then introduced to boost expression.
It is known that the codons used in the DNA and RNA to encode amino acids show a certain redundancy. Thus, a given amino acid may be coded for by more than one 3-base codon. However, a given organism may more commonly use one of the alternative codons than the other(s) for a given amino acid. This has implications if a heterogeneous nucleic acid is introduced into a host organism for expression. Sometimes the codons used by the originator organism are the “less preferred” codons for the host, which can result in difficulties in expression such as a reduced level of expression. Similarly, certain amino acids may be less commonly represented in the proteome of one organism when compared to another. It may sometimes possible to boost the levels of heterologous expression in a host cell by using an alternative nucleic acid sequence in which codons and/or amino acids which are less preferred by the host are exchanged for more preferred alternatives. The present invention provides an isolated nucleic acid which encodes a monoamine oxidase enzyme which is a variant of the A. niger MAO in which the codons for R259 and R260 are optimised for expression in a heterogeneous host cell. More particularly, when the expression of the MAO is intended to take place in E. Coli, the arginine amino acids at these positions are replaced by alternative amino acids of a similar charge and size. These mutations may in addition to mutations which alter the catalytic activity and/or enantioselectivity of the mutant enzyme. In a further embodiment, we provide an enantioselective monoamine oxidase enzyme which is a variant of the A. niger MAO and which differs from wild-type A. niger monoamine oxidase by incorporation of the mutations N336S and one or more of M348K, R259L and R260L.
The mutant N336S identified below may be taken as the basis for “hotspot” mutation in which further mutations are introduced in the amino acids surrounding position 336. Mutations may be made by site-directed mutagenesis, by the construction of chimeric recombinant enzymes using a “cut and paste” methodology employing restriction enzymes, or by other means well known to workers skilled in the art. This method is used to identify further mutants with enhanced activity over the N336S mutant, such as the N336S, M348K double mutant identified herein.

EXAMPLE 1—PREPARATION OF MUTATED PLASMID DNA LIBRARY BY E. COLI XL1-RED MUTATOR STRAIN

The MAO gene from Aspergillus niger cloned in pET3a was obtained from B. Schilling (11). The gene was amplified by pfu Turbo DNA polymerase (Stratagene) using primers designed according to E. coli codon usage. The PCR product was subcloned into the pET16b vector (Novagen). This construct (MAOpET16b) was submitted for mutagenesis by the E. coli XL1-Red mutator strain.
E. coli XL1-Red mutator strain competent cells were obtained from Stratagene. Competent cells were transformed as described in the Stratagene protocol. 700 μl of transformed cell suspension (Transformation 1) was inoculated in 20 ml of LB medium (tryptone, 10 g/L; yeast extract, 5 g/L; NaCl, 10 g/L; pH 7) with ampicillin (100 μg/ml; LB Amp) and grown for 18 h in a 50 ml Falcon tube in an incubator shaker at 37° C. These growing conditions were used throughout the experiment.
20 μl of growing culture was inoculated in 10 ml of fresh LB Amp and grown for 24 h. The plasmid was purified (Qiagen Plasmid DNA Miniprep Kit) from 1 ml culture (pMAO2) and used for the second transformation (Transformation II) of the mutator strain. The total transformed cell suspension was inoculated in 10 ml of LB Amp and grown for 24 h. The plasmid was purified from 1 ml of culture (pMAOretr1.1). 100 μl of Transformation II growing culture was used to inoculate 10 ml LB Amp. The culture was grown for 24 h, the plasmid purified (pMAOretr1.2) and used for Transformation III. Total transformed cell suspension (1 ml) was inoculated in 10 ml LB Amp, grown for 24 h and the plasmid purified (pMAOretr2.1). 100 μl of Transformation III growing culture was used to inoculate the next 10 ml of LB Amp, grown for 24 h and the plasmid purified from 1 ml culture (pMAOretr2.2) and used for Transformation IV. Total suspension of transformed cells (1 ml) was used to inoculate 10 ml of LB Amp, grown for 24 h and the plasmid purified from 1 ml culture (pMAOretr3.1). 100 μl of Transformation IV growing culture was used to inoculate fresh 10 ml LB Amp, grown for 24 h and the plasmid purified (pMAOretr3.2).
Collected pools of mutated plasmid DNA (pMAOretr2.2, pMAOretr3.1, pMAOretr3.2) were used to transform E. coli BL21 (DE3) to express mutated MAO genes and detect activity towards L-α-methylbenzylamine (L-AMBA) oxidation.

EXAMPLE 2—SCREENING FOR MAO MUTANTS

The plate assay method (FIG. 2) was used to identify MAO mutants with activity towards L-AMBA. More specifically, E. coli BL21 (DE3) transformants (2500 colonies per plate) were plated on HiBond-C Extra (Amersham Pharmacia) membrane placed on an LB Amp agar plates and incubated for 24 h at 37° C. Membranes containing the colonies were pulled from the plates, kept at −20° C. for 24 h and incubated with assay mixture at room temperature for 24 h.
Assay Mixture:

- 1 tablet of DAB (3,3′-diaminobenzidine, Sigma, D-4418)
- 1 ml of K phosphate buffer (1M, pH 7.6)
- 30 μl L-AMBA (10 mM)
- 30 μl horseradish peroxidase (Sigma) 1 mg/ml
- 10 ml 2% agarose (Bio-Rad)
- Water up to 20 ml

Positive clones were subjected to a second round of screening (100-200 colonies per plate) to confirm activity.

EXAMPLE 3—ACTIVITY STUDIES

a) Clone Selection
27 positive clones identified in the plate assay as having improved activity towards L-AMBA compared to the wild-type enzyme were grown on a small scale and assayed against both L- and D-AMBA. 10 ml LB Amp was inoculated with a single colony of E. coli BL21 (DE3) transformed with the protein expression vector pET16b harbouring the gene of interest and cultured in a 50 ml Falcon tube with agitation for 24 h. At the end of incubation 1 ml of cell culture was centrifuged, the pellet was resuspended in 1 ml of 25 mM potassium phosphate buffer pH 7.6 and 0.1 ml was used to perform a hydrogen peroxide formation assay using both L- and D-AMBA as substrates.
Assay Mixture:

- (Manfred Braun et al, Applied Microbiology and Biotechnology (1992) 37:594-598)
- 5 ml phosphate buffer (1M, pH 7.6)
- 500 μl 2,4,6-tribromohydrobenzoic acid (2% in DMSO),
- 37.5 μl 4-aminoantipurine (1M)
- 32.5 μl L or D-AMBA (final concn 5 mM)
- Water up to 50 ml
  To 895 ul Assay Mixture was Added:
- 5 μl HRP (Sigma, P6782) (1 mg/ml)

Sample (100 μl) was added and absorbance at 510 nm measured after 24 hours against a control without sample. The absorbance results are shown in FIG. 3.
Several of these 27 clones appear to be expression mutants, as they demonstrate increased protein expression when visualised on polyacrylamide gels (data not shown). The plasmid of the best mutant identified, which appeared not to have increased expression but which showed greatest improvement over the wild-type enzyme in terms of its selectivity, and activity, towards L-versus D-AMBA, was grown on a larger scale and the enzyme purified. The wild-type enzyme was also purified by the same protocol and the two enzymes compared for substrate specificity and enantioselectivity.
The mutated gene was also sequenced and the sequence is shown in SEQ ID NO: 2. There was a single amino acid change from the wild-type enzyme with serine replacing asparagine at position 336.
b) Growth and Comparison of Mutant and Wild-Type Enzymes
Growth and Purification of MAO Mutant Expressed in E. coli BL21 (DE3)
LB medium (6×300 ml) containing ampicillin (100 μg·ml⁻¹) in 1 L baffled flasks was inoculated with a single colony of monoamine oxidase mutant from an LB agar plate. Cultures were incubated at 30° C. for 22 hours (OD₆₀₀˜3.6) then harvested and washed with phosphate buffer (50 mM, pH 8) to yield a yellow-brown pellet (11.2 g).
The pellet was resuspended in Tris/HCl buffer (25 mM, pH 7.8, 30 ml) and sonicated on ice (30 s on, 30 s off for 10 minutes). The suspension was then centrifuged (20,000 rpm, 4° C.) until clear supernatant was obtained and the supernatant dialysed against Tris/HCl buffer (25 mM, pH 7.8). The cell free extract was filtered through 0.45 μm sterile membrane and chromatographed on a QSepharose anion exchange column. Fractions were assayed using colorimetric hydrogen peroxide based assay and active fractions were stored at −80° C. The active fraction has a protein content of 1 mg·ml⁻¹, and a specific activity of 0.193 U·mg⁻¹against amylamine.
Chromatography Conditions:

- Column=HiFlow QSepharose 26/10
- Buffer A=Tris/HCl (25 mM, pH 7.8)
- Buffer B=Tris/HCl (25 mM, pH 7.8)+1 M NaCl
- Flow rate=4 ml·min⁻¹
- Fraction collect=10 ml
- Column wash=2 CV 100% buffer A
- Elution=10 CV 100% buffer A to 100% buffer B
- Column clean=4 CV 100% buffer B
  Assay Mixture:
- 5 ml phosphate buffer (1 M, pH 7.6)
- 500 μl 2,4,6-tribromo-3-hydroxybenzoic acid (2% in DMSO)
- 37.5 μl 4-aminoantipyrine (1.5 M)
- 30 μl amine substrate (final concentration 0.015-5 mM)
- 44.4 ml water
  Assay Conditions:
- 990 μl assay mixture
- 5 μl horse radish peroxidase (1 mg·ml⁻¹)
- 10 μl enzyme

The spectrophotometer was blanked against assay mixture and HRP. Enzyme was added and the absorbance at 500 nm measured at 3 s intervals for a 10 minute period.
Results:

Wild-type Enzyme Mutant Enzyme

Substrate K_mmM k_catmin⁻¹ K_mmM k_catmin⁻¹

L-AMBA ND 0.17 0.4 8.0

D-AMBA ND 0.01 ND 0.08

benzylamine ND 371 ND 196

amylamine — 1000 0.4 116
The K_mvalues were calculated using ‘KaleidaGraph for Windows’ (Synergy Software). For calculation of the k_cat, the active enzyme concentration was determined by estimating the FAD content from the absorbance at 458 nm using an extinction coefficient of 11 mM⁻¹cm⁻¹(see ref 12).
The data reveal that the activity of the mutant amine oxidase towards L-AMBA (kcat=8.0 min−1) is 47 fold higher than the wild-type (kcat=0.17 min−1). Moreover the selectivity of the mutant for the L-enantiomer versus D-AMBA (ca. 100:1) has also been increased relative to the wild type enzyme (ca. 17:1). Thus the outcome of the in vitro evolution experiments has been to simultaneously improve both the enantioselectivity and catalytic activity of the enzyme. For comparison, the activity towards the best substrate for the wild-type enzyme, namely amylamine, and also benzylamine, is presented. The substantial improvement in activity and selectivity of the mutant was confirmed by chiral HPLC (Chiralcel CrownPak CR+) in which after 24 h complete oxidation of the L-enantiomer was apparent whereas there was no detectable conversion of the D-enantiomer.

EXAMPLE 4—DERACEMISATION REACTION

Using DL-AMBA as the substrate, in the presence of the mutant MAO-N, a range of reducing agents were screened (sodium borohydride, catalytic transfer hydrogenation, amine:boranes). This screen identified ammonia:borane as the optimal reagent which gave a 77% yield of D-AMBA with an enantiomeric excess=93%. Greater optical purities of the product (up to 99% e.e.) could be achieved although at the expense of yield.
More specifically, MAO mutant N336S (100 μl of 0.193 U·ml⁻¹=0.02 U) was added to a solution of DL-AMBA (0.13 μl, final concentration=0.8 mM) and ammonia-borane complex (10 μl of 4M solution, final concentration=80 mM, 100 eq) in phosphate buffer (400 μl, 20 mM, pH 8). A 10 μl aliquot was diluted in 990 μl perchloric acid, pH 1.5 and analysed by HPLC. The reaction mixture was incubated at 30° C. and the reaction monitored by HPLC at regular intervals until no further reaction was observed.
Yield: D-AMBA=77%, e.e.=93%
Analytical Conditions:

- Column=Chiralcel CrownPak CR+
- Mobile phase=100% perchloric acid, pH 1.5
- Flow rate=0.8 ml·min⁻¹
- Detection=200 nm
- Temperature=25° C.
- r.t. (L-AMBA)=12.8 min
- r.t. (D-AMBA)=16.5 min

EXAMPLE 5—STEREOINVERSION REACTION

We also carried out the stereoinversion of L- to D-AMBA (18% yield, 99% e.e.) and showed that under identical conditions there was no conversion of D- to L-AMBA.
More Specifically, MAO mutant N336S (100 μl of 0.193 U·ml⁻¹=0.02 U) was added to a solution of L-AMBA (0.13 μl, final concentration=0.4 mM) and ammonia-borane complex (10 μl of 4M solution, final concentration=80 mM, 200 eq) in phosphate buffer (400 μl, 20 mM, pH 8). A 10 μl aliquot was diluted in 990 μl perchloric acid, pH 1.5 and analysed by HPLC. The reaction mixture was incubated at 30° C. and the reaction monitored by HPLC at regular intervals until no further reaction was observed.
Yield D-AMBA=18%, e.e. >99%:
NB: no reaction was observed after 24 hours when D-AMBA was used as substrate under identical reaction conditions.
Analytical Conditions:

Analysis of the HPLC chromatograms suggests that the yield in the deracemisation reactions is prevented from reaching 100% due to the formation a by-product with longer retention time as the reaction progresses. We are currently optimising the deracemisation protocol to achieve the exquisite levels of yield and selectivity previously demonstrated for the deracemisation of α-amino acids.
Abbreviations:
MAO—monoamine oxidase.
AMBA—α-methylbenzylamine
TBHBA—2,4,6-tribromo-3-hydroxybenzoic acid
AAP—4-aminoantipyrine
HRP—horse radish peroxidase type VI from bovine liver
LB—Lubria Bertani
r.t.—retention time

EXAMPLE 6—SUBSTRATE SPECIFICITY OF THE N336S MUTANT ENZYME

The activity of the mutant enzyme described above was studied in relation to a variety of amine substrates. Assay conditions were essentially as set out in Example 3. The substrates tested were:

Results:



		Relative
	Sample	Rate

	(S)-α Methylbenzylamine	1
	(R)-α Methylbenzylamine	*
	(rac)-α Methylbenzylamine	*
	(S)-4-Methylphenylethylamine	0.28
	(R)-4-Methylphenylethylamine	*
	(S)-α-Methyl-4-nitrobenzylamine	0.91
	(R)-α-Methyl-4-nitrobenzylamine	0.06
	(S)-4-Bromo-α-phenylethylamine	0.24
	(R)-4-Bromo-α-phenylethylamine	*
	(rac)-4-Bromo-a-phenylethylamine	*
	(S)-1-4-Methoxyphenylethylamine	0.81
	(R)-1-4-Methoxyphenylethylamine	0.13
	(S)-MTQ	0.67
	(R)-MTQ	*
	(rac)-MTQ	0.13
	(S)-3-Methyl-2-butylamine	5.4
	(R)-3-Methyl-2-butylamine	*
	(S)-3-3-Dimethyl-2-butylamine	1.4
	(R)-3-3-Dimethyl-2-butylamine	*
	2-Methylcyclohexylamine	0.43
	N-heptylamine	12.1
	N-amylamine	25.2
	Hexylamine	18.6
	Bis(α-methyl)benzylamine	*
	1,3-Dimethylbutylamine	1.8
	1,2,3,4-Tetrahydro-1-naphthylamine	0.13

	* No measurable rate

EXAMPLE 7—CREATION OF A FURTHER MUTANT CLONE

The first published sequence of wild-type A. niger monoamine oxidase showed a lysine at position 348. Some of the mutant enzymes generated above and initially tested in Example 3(a) were thought to be double mutants because they had a methionine at this position. However, when the wild-type sequence was checked (resequenced), the wild-type amino acid at position 348 was shown to be methionine (as set out in SEQ ID NO:1). In order to elucidate the impact (if any) of a mutation at this site, site-directed mutagenesis was performed on the wild-type enzyme to introduce the M348K mutation. It was noted that the identity of the amino acid at this position influenced the efficiency of expression, with lysine giving rise to increased expression without altering the catalytic activity of the enzyme (absolute activity in U/ml is greater than wild type, but correction for the amount of protein gives a value of kt which is the same as for the wild-type enzyme). Site directed mutagenesis was then performed upon the N336S mutant described above to generate a second mutation (M348K), thus combining the catalytic enhancement of the N336S mutation with an increase in expression. The amino acid sequence of the double mutant is set out in SEQ ID NO: 3.

EXAMPLE 8—SUBSTRATE SPECIFICITY OF THE N336S, M348K MUTANT ENZYME

The activity of the double mutant enzyme described above was studied in relation to a variety of amine substrates. Assay conditions were essentially as set out in Example 3. The substrates tested and the relative rates of conversion (compared to L-AMBA set at 100) are shown in FIG. 4. For some substrates, both enantiomers were tested as substrates. The enzyme appears to demonstrate enantioselectivity in every case, with activity against the opposite enantiomer being undetectable in several cases (the enantioselectivity data is shown in FIG. 5).

EXAMPLE 9—APPLICATION OF EXPRESSION MUTANT R260K

The mutation R260K was found to be in low occurrence codon for arginine. The sequencing results revealed that all of the “expression” mutants had the same mutation at position 260 wherein arginine was replaced by lysine.
The wild type MAO-N amino acid sequence has two arginines at position 259 and 260 both encoded by the codon (AGG), which has a low occurrence in E. coli genes (4%). Lysine and arginine are both basic amino acids and hence replacement of one by the other should not affect the charge of the protein, however the replacement of a low frequency codon for Arg AGG (4%) by a high frequency codon for Lys AAG (22%) results in improved protein expression in E. coli. Thus we decided to replace the codons of both Arg 259 and Arg 260 with the codon CGT (38%) by site directed mutagenesis to evaluate the effect upon the expression level of MAO-N. An alternative approach would be to create a “silent” mutation by altering the AGG codon to a different codon which still encodes arginine (AGA, CGT, CGC, CGA, CGG).
MAO WT and MAOArg259/260 were partially purified from recombinant E. coli BL21(DE3) harbouring a pET16b carrying the corresponding mao-n gene. Cell free extracts were prepared by sonication and the specific activities of MAO WT and MAOArg259/260 towards M were measured for both the soluble and insoluble fractions. (Table 3)

TABLE 3

Specific activities towards AA in partially purified MAO WT and

MAOArg259/260 towards amylamine.

Protein Soluble Insoluble Total

concentration fraction fraction activity

Enzyme mg · ml⁻¹ U · mg⁻¹ U · mg⁻¹ U

MAO WT 1.18 0.23 0.33 0.57

MAOArg259/260 1.30 0.46 0.56 1.2

To confirm the increased level of the R260K mutant expression, Cell free extracts of MAO WT, mutant 4 (R260K) and MAOArg259/260 were obtained and assayed towards L-AMBA. This was achieved by incubating 50 μl of each sample for 240 minutes and specific activities were determined. (Table 4)

TABLE 4


		Soluble
	ΔAbs	fraction
Enzyme	510 nm	U · mg⁻¹

MAO WT	0.06	3.6 × 10⁻⁵
MAOArg259/260	0.12	7.0 × 10⁻⁵
MAO mutant	0.16	5.7 × 10⁻⁵
(R260K)

In conclusion, we have significantly extended the deracemisation strategy by applying the method for the first time to the deracemisation of chiral amines. In so doing we have successfully achieved the ‘directed evolution’ of an enzyme in order to meet the specific requirements of a novel biotransformation. Another interesting aspect of the present work is the identification of a highly enantioselective mutant by using a single enantiomer substrate in the screen (L-AMBA). There has been much discussion concerning the need for truly enantioselective screens in which racemates are used, thereby mimicking the real-life situation found in a kinetic resolution in which the two enantiomers compete for the enzyme. It may be that if one is able to screen truly large and diverse libraries of variant genes (e.g. 10⁶) then it is possible to select for enantioselectivity in the manner described herein, using inherently simpler screens.
The application of which this description and claims form part may be used as a basis for priority in respect of any subsequent application. The claims of such subsequent application may be directed to any feature or combination of features described herein. They may take the form of product, composition, process or use claims and may include, by way of example and without limitation, one or more of the following claims:

REFERENCES

(1) L. E. Iglesias, V. M. Sanchez, F. Rebolledo and V. Gotor, Tetrahedron: Asymmetry (1997) 8, 2675-2677.
(2) J. S. Shin, B. G. Kim, A. Liese and C. Wandrey, Biotechnol. Bioeng. (2001) 73 179-187;
(3) G. Matcham, M. Bhatia, W. Lang, C. Lewis, R. Nelson, A. Wang and W. Wu, Chimia (1999) 53, 584-589;
(4) J. S. Shin, B. G. Kim, Biotechnol. Bioeng., (1999) 65, 206-211.
(5) G. Hieber and K. Ditrich, Chimica Oggi (2001) 19, 16-20;
(6) F. Balkenhohl, K. Ditrich, B. Hauer and W. Ladner, J. Prakt. Chem. (1997) 339, 381.
(7) M. T. Reetz and K. Schimossek, Chimia, (1996), 50, 668-669.
(8) E. W. Hafner and D. Wellner, Proc. Nat. Acad. Sci. (1971) 68, 987.
(9) R. B. Silverman, J. M. Cesarone and X. Liu, J. Am. Chem. Soc. (1993) 115, 4955.
(10) B. Schilling and K. Lerch, Biochim. Biophys. Acta. (1995), 1243, 529.
(11) B. Schilling and K. Lerch, Mol. Gen. Genet. (1995) 247, 430.
(12) S. O. Sablin, V. Yankovskaya, S. Bernard, C. N. Cronin and T. P. Singer, Eur. J. Biochem. (1998) 253, 270.
(13) T Beard and N J Turner Chem. Commun. 2002 in press.
(14) F R Alexandre, D P Pantaleone, P P Taylor, I G Fotheringham, D J Ager and N J Turner (2002) Tetrahedron Lett. 43, 707-710.
(15) P L Dostert, M S Benedetti and K F Tipton (1989) Medicinal Research Reviews 9, 45-89.
(16) J D Sutherland (2000) Curr. Opinion. Chem. Biol. 263.
(17) F H Arnold (1998) Acc. Chem. Res. 31, 125.

Claims

1. A method for the enantiomeric conversion of amines comprising treating a homochiral amine or a mixture of amine enantiomers with an enzyme capable of catalysing oxidation of the amine in a stereoselective manner and, subsequently or simultaneously, treating with a reducing agent.

2. A method according to claim 1 which is a method of epimerisation of one single amine enantiomer to the other.

3. A method according to claim 1 in which the mixture of amine enantiomers is a racemic mixture.

4. A method according to claim 1 in which the reducing agent is partially enantioselective.

5. A method according to claim 1 in which the reducing agent is a chemical reducing agent and is non-enantioselective.

6. A method according to claim 5 in which the reducing agent is selected from sodium borohydride, sodium cyanoborohydride, an amine:borane complex or a transfer hydrogenation reagent.

7. A method according to claim 6 in which the transfer hydrogenation reagent is ammonium formate with Pd/C.

8. A method according to claim 6 in which the amine:borane complex is an ammonia:borane complex.

9. A method according to claim 1 comprising an oxidation-reduction cycle in which the steps of treating with an enzyme capable of catalysing oxidation of the amine and treating with a reducing agent are performed sequentially, a plurality of times.

10. A method of improving the activity, substrate specificity, or enantioselectivity of an amine oxidase enzyme by directed evolution, the method comprising mutation of the enzyme and selection of a mutant having improved activity against an homochiral substrate.

11. A method of directing the evolution of an amino oxidase enzyme comprising the steps of:

a) mutating the enzyme to create at least one enzyme variant;

b) screening said enzyme variant for activity against a homochiral substrate; and

c) selecting one or more enzyme variants which show greater activity toward the substrate than does the originator enzyme.

12. A method according to claim 11, further comprising repeating steps a), b) and c) one or more times, using the enzyme variant selected in step c) as the enzyme to be mutated.

13. A method according to claim 11 further comprising performing site directed mutagenesis on an enzyme selected in step c).

14. A method according to claim 11 in which the enzyme is a monoamine oxidase.

15. A method according to claim 14 in which the enzyme is a microbial monoamine oxidase.

16. A method according to claim 15 in which the enzyme is the Aspergillus niger monoamine oxidase, or a variant thereof.

17. A method according to claim 16 in which the enzyme is a variant of A. niger monoamine oxidase which differs from wild-type A. niger monoamine oxidase by mutation in the region of amino acids 334-350.

18. A method according to claim 14 in which the enzyme is a variant of A. niger monoamine oxidase which differs from wild type A. niger monoamine oxidase by mutation of the amino acid at position number 336.

19. A method according to claim 14 in which the enzyme is a variant of A. niger monoamine oxidase which differs from wild type A. niger monoamine oxidase by incorporation of the mutation: N336S.

20.-23. (canceled)

24. A method according to claim 1 in which said enzyme is an enantioselective monoamine oxidase (MAO) enzyme having the amino acid sequence of SEQ ID NO:2.

25. A method according to claim 1 in which said enzyme is an enantioselective monoamine oxidase enzyme which is a variant of wild-type A. niger MAO, where said variant differs from wild-type A. niger monoamine oxidase by incorporation of the mutation: N336S.

26. A method according to claim 1 in which said enzyme is an enantioselective monoamine oxidase enzyme which is a variant of wild-type A. niger MAO, where said variant differs from wild-type A. niger monoamine oxidase by incorporation of the mutations: N336S and M348K.

27. A method according to claim 1 in which said enzyme is an enantioselective monoamine oxidase enzyme which is a variant of wild-type A. niger MAO, where said variant differs from wild-type A. niger monoamine oxidase by incorporation of the mutations: N336S and one or more of M348K, R259L and R260L.

28. (canceled)