Search

This thesis describes multiple synthetic applications of the recombinantly expressed (in Pichia pastoris host) Agrocybe aegerita unspecific peroxygenase (rAaeUPO). Chapter One gives a description of the existing enzymatic and chemocatalytic methodologies, which allow the direct oxidation of non-activated sp3 C-H positions. In Chapter Two, the optimal operating conditions of rAaeUPO-catalysed reactions were studied, including the effect of substrate concentration, temperature, mode of H2O2 delivery and enzyme concentration. Building on these benchmarking studies, the oxidation of a variety of N�heterocycles were optimised for improved yields, mostly focusing on analogues of pyridine and tetrahydroquinolines. N-heterocyclic benzylic alcohols were generally yielded with high enantioselectivities. The optimised conditions also allowed the transformations to be performed on synthetically useful scales (few hundred milligrams). In Chapter Three, the potential carbene and nitrene insertion activities of rAaeUPO were explored. In our hands, this enzyme failed to support any of these transformations under the conditions employed. The possible reasons of this are described and discussed. Chapter Four describes the application of rAaeUPO for catalysing the oxidation of furan analogues. Direct oxidation of alkyl-substituted furans, followed by the Achmatowicz reaction of the product formed was not possible under the conditions used. On the other hand, the rAaeUPO-mediated oxidative ring expansion of a wide range of α-furylcarbinols and α�furylamine analogues was possible with high yields. The enzyme was also shown to tolerate high substrate concentrations, thereby making the approach easy to scale. In Chapter Five, the oxidation of ethylbenzenes, toluenes and the corresponding secondary and primary alcohols was studied first to understand the electronic, steric, and energetic factors influencing these oxidations. Whilst rAaeUPO is known to mediate benzylic oxidation with high enantioselectivity, the enzyme did not discriminate strongly between the secondary alcohol enantiomers. The Chapter also describes the oxidation of more complex benzylic and allylic aliphatic alcohols.

Chiral amines are fundamental building blocks in synthetic chemistry routinely used in the production of pharmaceuticals, agrochemicals or in the food industry. In recent years, given the need for more renewable and sustainable approaches to the synthesis of these chiral amines, biocatalysts have been investigated. The consideration of biocatalytic approaches has largely led to enantioselective routes to produce chiral amines often with better yields and less harsh reaction conditions and components. More recently, a class of enzymes, found by sequence driven searches of metagenomic databases and samples, known as native Amine Dehydrogenases (nat-AmDHs) have been used for the asymmetric reductive amination of a range of carbonyl compounds. Herein a selection of nat-AmDHs have been structurally and biochemically studied using a range of techniques such as X-ray crystallography, UV-vis spectrometry, and GC-FID analysis. The canonical structural functionalities of these enzymes were found to be consistent throughout. However, some aspects of these previously uncharacterised nat-AmDHs displayed interesting new properties. Namely the increased active site space in the binding pocket of MATOUAmDH2 which permitted the rational engineering of this enzyme to bind larger, more sterically hindered, substrates to produce pharmaceutically relevant amines such as 2-aminonorbornane. These results provide new knowledge and protein scaffolds for the further engineering of nat-AmDHs for the production of primary and secondary chiral amines of both biological and chemical significance.

The oxyfunctionalisation of inert bonds in synthetic chemistry is a very desirable yet challenging process. Methods for this activation, especially for asymmetric products, often require harsh conditions, use valuable metals or need complex catalyst synthesis. Biocatalysis may provide a more favourable approach because reactions can be carried out under mild conditions and can be highly selective. A group of heme-thiolate proteins called unspecific peroxygenases (UPOs), can carry out a diverse range of different oxyfunctionalisations using only hydrogen peroxide as a cosubstrate. The reactions carried out by UPOs are regio-, chemo- and enantioselective and they have a large substrate scope. In this work two UPOs were chosen for further investigation: an artificial UPO (artUPO) based on the UPO found in Marasmius rotula (MroUPO), and the UPO derived from Coprinus cinerea (CciUPO). The genes for artUPO and CciUPO were cloned for Pichia pastoris expression but only artUPO resulted in successful purified recombinant protein. The gene for artUPO was also successfully expressed in Escherichia coli. Both forms of artUPO were evaluated by UV/visible activity assays which displayed a small reduction in activity compared to MroUPO. The structure of both artUPO forms were determined by x-ray crystallography which developed our understanding of the mutations present in artUPO and enhanced understanding of general UPO structural biology. Substrate screens identified artUPO as effective biocatalyst for the asymmetric oxygenation of alkyl benzenes, styrenes and thioethers. With the latter, artUPO gives the complementary series of (S)-enantiomers to those obtained with the UPO from Agrocybe aegerita (AaeUPO), broadening the scope for application of the enzymes.

Glycoproteins have many important roles in cell signalling and diseases including cancer, neurodegeneration, and pathogenic infections. Metabolic oligosaccharide engineering has previously been used to incorporate unnatural sugar derivatives into glycoproteins and detect them with bioorthogonal labelling. Additional methods of labelling that work orthogonally to current methods, and are reversible, would allow the simultaneous investigation of multiple glycoproteins and their properties. Sortase is a transpeptidase found in Gram-positive bacteria that has been previously engineered to perform ligations of glycine- and LPXTG- bearing biomolecules. This project sought to use sortase to specifically label functional groups carried by sugar derivatives incorporated into glycoproteins. Initial incorporation of sugar derivatives bearing glycine-azide showed successful labelling with click chemistry. Subsequent attempts to label incorporated sugar derivatives bearing glycine-amine with the commercial sortase Srt5M failed. An alternative mannosamine-thiazolidine sugar was tolerated in HEK cells up to 300 μM. Live cells that had incorporated this sugar were successfully labelled using the organocatalyst-mediated protein aldol ligation. Labelling of intracellular glycoproteins was also tested but requires further optimisation. To expand the currently available sortases, 22 novel sortases were cloned and expressed from a metagenome database at the industrial partner Prozomix. Testing for the ability to ligate a range of peptides in vitro revealed two functioning enzymes, which were named Srt021 (class C) and Srt025 (class D). The two sortases were shown to function well at physiological temperature and pH, and in a range of buffers including without calcium. This represents the second class C sortase made to function in vitro, and the first class D sortase in vitro. Further characterisation and directed evolution of these two sortases should result in a wider range of tools for specific labelling of biomolecules.

Amide bond formation is one of the most important reactions in synthetic medicinal chemistry, however current methods rely on toxic or hazardous reagents and suffer from poor atom economy. In search of biocatalytic alternatives for amide bond synthesis, this work investigated a recently discovered class of ATP-dependent amide bond synthetases (ABSs), typified by McbA from Marinactinospora thermotolerans. A number of homologs, including AcABS from Actinoalloteichus cyanogriseus and ShABS from Streptalloteichus hindustanus, were identified and heterologously expressed in E. coli and purified. These homologs have demonstrated diverse activity, catalysing the formation of pharmaceutically relevant amides from β-carboline carboxylic acids and equimolar amine precursors. In addition to demonstrating an expanded substrate tolerance within families of parent acid and amine partners, these couplings can be achieved in high yield on a preparative scale. McbA showed good activity with much simpler acids, including indole-, naphthyl-, and benzoic acids, as well as simple aromatic and aliphatic amines, such as aniline, propargylamine and methylamine. Crystallisation studies of McbA revealed a large 149° conformational change, around the hinge residue Q394, following formation of the AMP adenylate intermediate. These two conformations were annotated as 'adenylation' and 'amidation' conformers, mirroring the 'adenylation' and 'thiolation' conformations, respectively, of other adenylating enzymes such as the Carboxylic Acid Reductases (CARs) and CoA ligases. The acid binding site is comprised of hydrophobic residues, including F301 and L202, which form pi-interactions with the 1-acetyl-β-carboline carboxylic acid, and residue Y294 which may serve to limit the size of the active site. There are also interactions between the acetyl oxygen and pyridine nitrogen of the native acid with the main chain NH and C=O of residue G295. The limited number of interactions of active site residues with the native 1-acetyl-β-carboline carboxylic acid provide a possible explanation for the observed broad acid substrate tolerance. The nucleotide binding site is comprised of residues D377, R392 and R407, which form stabilising interactions with the ribose hydroxyls of ATP. Residue K483 has also been implicated in the stabilisation of the transition state on adenylate formation, through structural analysis and mutagenesis. McbA (K483A) mutant resulted in a 6-fold reduction in the Vmax and almost 5-fold increase in the KM compared to wt-McbA, using a fluorimetric assay to detect pyrophosphate release. Docking studies with the 1-acetyl-β-carboline adenylate and 2-phenylethylamine revealed a hydrophobic amine binding site, comprised of residues including I197, L202, F241 and W246. The model placed the amine in close proximity to a catalytic aspartate residue, D201, proposed to act as a base in activating the amine prior to nucleophilic attack into the activated acyl adenylate. Mutagenesis of this aspartate to alanine, D201A, resulted in a significantly reduced conversion for the coupling of 1-acetyl-β-carboline acid and propargylamine, with wt-McbA and McbA (D201A) producing the amide product in 26% and 0.7% conversion, respectively, over 4 h. This provided evidence for the involvement of D201 in catalysis. Substrate screens have highlighted application of these enzymes to the synthesis of a number of pharmaceutical molecules and, through the use of an ATP recycling system, this can be achieved using sub-stoichiometric concentrations of ATP. Two polyphosphate kinases, AjPPK2 and SmPPK2, were used to convert AMP to ADP and ADP to ATP respectively, using polyphosphoric acid as the phosphate source. This system was applied to the McbA-catalysed synthesis of the anti-depressant drug Moclobemide, achieving a conversion of 63%, highlighting these ABSs as promising biocatalysts for the sustainable synthesis of amide pharmaceuticals.

Previous work at York demonstrated that fragment molecules can increase the activity of the glycoside hydrolase, BtGH84. The initial aim of this project was to use fragment-based discovery methods to identify activators of several enzymes used in cellulose degradation where low activity is one of the limiting steps in the industrial process. This was successful for one enzyme, the fungal glycoside hydrolase, TrBgl2. The characterisation of the mechanism of activation for this enzyme is the main focus of this thesis. A fragment screen of a library of 560 commercially available fragments using a kinetic assay identified a small molecule activator of TrBgl2. An analogue by catalogue approach and detailed kinetic analysis identified compounds that behaved as nonessential activators with up to a 2-fold increase in maximum activation. The compounds did not activate the related bacterial glycoside hydrolase CcBglA demonstrating specificity. Interestingly, an analogue of the initial fragment inhibits both TrBgl2 and CcBglA, apparently through a mixed-model mechanism. Although it was not possible to determine crystal structures of activator binding to 55 kDa TrBgl2, solution NMR experiments demonstrated a specific binding site for the activator. A partial assignment of the NMR spectrum gave the identity of the amino acids at this site, allowing a model for TrBgl2 activation to be built. The activator binds at the entrance of the substrate binding site, stabilizing the enzyme-substrate complex.

The objective of this work is the investigation and structural characterisation of pyridoxal phosphate (PLP)-dependent racemases. In particular α-amino-ε-caprolactam racemases (ACLRs), such as the enzyme from Achromobacter obae (AoACLR) are of interest for this work, as they have been found to racemise amino acid derivatives. This can be beneficial for the use of enzymes for the application for the dynamic kinetic resolution (DKR) of amino acid derivatives. In this work focus was put on the racemisation of the model substrate phenylalanine methylester due to the potential of a biocatalytic DKR of amino acid esters to the corresponding amides. In addition to AoACLR, two further enzymes were identified in a homology search. Racemases from Rhizobum freirei (RfACLR) and Ochrobactrum anthropi (OaACLR) were identified as targets for the evaluation of PLP-dependent racemases. These enzymes were heterologous expressed in Escherichia coli and purified. The purified ACLRs were subjected to crystallisation trials and especially RfACLR was found to crystallise well. Intensive structural analysis of RfACLR gave structures of different reaction intermediates of the racemisation reaction. A reaction mechanism proceeding via the formation of a geminal diamine and an achiral quinonoid was proposed based on these results. Furthermore, the catalytic residues D210 and K267 were identified by mutational analysis. The purified enzymes were subjected to activity assay to evaluate their substrate spectrum. Only OaACLR was found to be active towards phenylalanine methylester. With information obtained from the structure of OaACLR, structure-guided engineering of OaACLR resulted in 3.4-fold activity-improved variant. This variant, OaACLR-L293C, was characterised in detail and found to be an overall improved variant of OaACLR. OaACLR-L293C is therefore a promising biocatalyst for the application DKR of amino acid esters.

Glutathione transferases (GSTs) are soluble enzymes with activity towards a wide range of xenobiotic and endogenous compounds. The Arabidopsis genome encoded 54 GSTs, which have been classified into eight classes including Tau and Phi. Members of these Tau and Phi clades are strongly upregulated in response to abiotic stresses such as xenobiotics and biotic stresses including pathogen attack. The most characterised activity of GSTs is the transfer of glutathione to an electrophilic centre to form a polar glutathionylated conjugate. However, increasing number of research demonstrated a non-catalytic activity plants GSTs, especially in the transportation of flavonoids from the cytosol to the vacuole. Despite the wealth of investigations into GSTs, and probably as a result of overlapping substrate specificities, the endogenous roles for the vast majority of these plant GSTs remains unknown. Here, the binding interaction of camalexin, indole-3-aldehyde, quercetrin and quercetin to GSTF2 in a non-catalytic fashion was observed in three different sites; two identical sites of L1 and one L2 from the X-ray crystallography data. Mutagenesis of the active residues, Q73L, H77A, Y97A and R154A were performed and using isothermal calorimetry (ITC) techniques, lower binding affinities were observed for all mutants towards all ligands except for Y97A and Q73L which showed higher binding affinities with indole-3-aldehyde. This unexpected finding was likely due to the conformational change of the mutant compared to the wild type, as observed in the structure of mutant Y97A. On elucidating the catalytic activity of GSTs, the structure of GSTU25 in complex with disulphide glutathione was obtained. The GSTU25 has been recently identified to catalyse the denitration of TNT to form 2-glutathionyl- 4,6-dinitrotoluene, a potentially more amenable product for subsequent degradation. This structure complex provides insights of GSTU25 folding upon substrate binding. The involvement of GSTU25 in the detoxification of TNT was further analysed using CRISPR/Cas9 technology. Subclades of GSTU25, including GSTU24, GSTU21, GSTU19 were knockout with the aim to remove overlapping substrate specificities and to finally reveal TNT-specific phenotypes. Only gstU25 Cas9 segregated Arabidopsis were obtained from the experiment that could potentially be optimised in the future study.

Hydroxylases introduce hydroxyl groups with excellent regio- and enantioselectivity making them of significant interest for use in the production of pharmaceutical intermediates and drug metabolites. 2-oxoglutarate dependent oxygenases (2OGDOs) are non-haem dependent Fe(II) containing enzymes that catalyse various oxidation reactions, including the hydroxylation of free amino acids. Unlike the more studied cytochromes P450, these enzymes only require molecular oxygen, Fe(II) and 2-oxoglutarate for catalysis, circumventing the need for a costly cofactor regeneration system. The targets of this work were three proline hydroxylases: a trans-4-proline hydroxylase from Dactylsporangium sp. RH1 (DOGDH), a cis-3-proline-hydoxylase from Streptomyces sp. (StP3H) and a cis-4-proline-hydroxylase from Mesorhizobium loti (MlC4H). Genes encoding all three were cloned into the pET-YSBLIC3C (and pET22b for DOGDH) expression vectors, expressed in Escherichia coli, and produced and purified by chromatography for use in crystallisation studies and enzymatic transformations. Extensive crystallisation trials were attempted for DOGDH including enzymatic, chemical and mutagenic modification with little success. A homology model was therefore constructed in order to identify catalytic residues within the active site that could be manipulated for enhancing the function of DOGDH. A precolumn derivatisation assay using FMOC-Cl was developed for the analysis of proline and its hydroxylated equivalents by HPLC and LC-MS. Biotransformations were performed with L-proline using the three hydroxylases with whole cell reaction conditions deemed optimal due to the multi-component nature of the enzymes, with the cell providing machinery for the recycling of cofactors. Reactions were scaled from shake flasks to stirred tank vessels with the flow of air into the vessel and stirring rate deemed key parameters for optimal function. Finally, a high-throughput substrate screening method using a BioLecter micro-bioreactor was successfully developed and trialled with the three hydroxylases with a panel of substrates providing a platform for future investigations.

Chiral amines are industrially useful chemicals found in the pharmaceutical, chemical and agrochemical industries. While many excellent methods have been developed for their synthesis using abiotic catalysis, these are often reliant on precious metals the global supply of which is becoming increasingly scarce. Additionally, those catalysts often require harsh, non-environmentally friendly reaction conditions such as high temperatures and pressures. Biocatalytic imine reduction, which involves the asymmetric reduction of imines to form chiral amines using imine reductases (IREDs), offers an efficient and sustainable synthesis of chiral amines which overcomes many of the limitations encountered in abiotic catalysis. Two NADPH-dependent IREDs were expressed and purified; SkR-IRED from Streptomyces kanamyceticus and SS-IRED from Streptomyces sp. GF3546, which catalyse the reduction of the model imine 2-methylpyrroline to (R)¬- and (S)- amine products respectively. The SkR¬-IRED monomer consists of an N-terminal Rossman fold motif and a C-terminal helical domain. SkR-IRED exists as a homodimer of two monomers which are linked by an unusual reciprocal domain sharing arrangement. The structure of SkR-IRED was used as a basis to study mechanism using mutagenesis experiments whichsuggested that residue Asp187 may be important for catalysis. The structure of SS-IRED was solved to a resolution of 3.2 Å in its apo form and revealed that the enzyme shares the structural features of SkR-IRED. Mutagenesis experiments suggested that residue Tyr169 is necessary for catalysis. Last, a novel IRED from the moss Physcomitrella patens (Pp-IRED) was expressed and purified. Pp-IRED was active towards the amine N-methyl-1-quinolin-6-ylmethanamine in the oxidative direction, the first IRED active towards this substrate. The structure of Pp-IRED was solved in both its apo form and in complex with NADPH (2.5 Å resolution). The structure was distinct from other IREDs as it did not display the reciprocal domain sharing arrangement seen in SkR-IRED and SS-IRED.

Current abiotic synthetic methods of obtaining chiral amines often involve conditions that are not considered environmentally friendly. One of the biggest concerns for chemists is the diminishing supply of precious metals that are often essential for organometallic catalysts in the production of chiral compounds including chiral amines. Imine reductases (IREDs) and reductive aminases (RedAms) are biocatalysts that have the potential for replacing these organometallic catalysts as IREDs are capable of reducing preformed imines to afford chiral amines and RedAms are capable of performing reductive aminations. Both IREDs and RedAms have a wide substrate spectrum and are often highly enantioselective. RedAms are even more remarkable in that they can catalyse reductive aminations utilising a carbonyl-to-amine substrate ratio of 1:1.

In 2004, unspecific peroxygenases (UPOs) from fungi were first identified and showed to catalyse selective oxygenation reactions, with high turnovers, good stability and a broad reaction scope, providing a valuable alternative to established biocatalytic hydroxylation systems such as cytochrome P450s. At the start of this project, access to UPOs was limited to expression in their native fungi, hence studies were performed looking into heterologous expression of the enzymes from Agrocybe aegerita and Agaricus bisporus, followed by characterisation, and application of the enzymes.

Enzyme discovery today is proceeding at an enormous pace due to ever-growing technology development. As a result of this, more than 21000 cytochrome P450s have been identified in all kingdoms of life to date, making a wide range of enzyme resources with outstanding potential for biocatalytical implementation available. P450s from filamentous fungi, such as Beauveria bassiana, represent particularly compelling targets for the discovery of novel enzymes as these organisms have a long history of application in industrial hydroxylation reactions, many of which are believed to be P450-dependent. In addition, the genome sequence of B. bassiana has recently been completed revealing 83 putative P450s. In order to uncover new cytochrome P450-based biocatalysts from the fungus Beauveria bassiana extensive bioinformatics analysis of the Beauveria CYPome were performed. As a result 7 genes encoding for heme domains with possible alkane hydroxylase function and one encoding a naturally fused P450 with homology to P450foxy from Fusarium oxysporum could be identified for subsequent cloning, heterologous expression and characterization. Different expression hosts as well as various expression conditions have been investigated. Despite our efforts, delivery of active biocatalysts could not be realized. However, empirical data acquired in this project will be of value for future studies of fungal P450s. In addition, 23 cytochrome P450 heme domains from Rhodococcus jostii fused to the P450 reductase domain (RhfRED) of cytochrome P450Rhf from Rhodococcus sp. NCIMB 9784 have been investigated in a further strand of this work and provided a screening platform that could be applied for industrial purposes.

This thesis concerns the investigation into the structure and function of the 38.6 kDa FAD-containing flavoprotein Stenotrophomonas maltophilia flavin-containing monooxygenase, SMFMO, which was encoded from a gene from the marine bacterium Stenotrophomonas maltophilia. The enzyme was found to catalyse the asymmetric oxidation of prochiral sulfides and the regioselective Baeyer-Villiger oxidation of bicyclo[3.2.0]hept-2-en-6-one. SMFMO was unusual amongst FPMOs as it demonstrated an ability to employ either NADH or NADPH as nicotinamide cofactor in order to reduce the flavin for catalysis. In an effort to determine the residues responsible for the cofactor promiscuity of SMFMO the structure of SMFMO was determined and revealed that the cofactor promiscuity of SMFMO may be due to the substitution of an arginine residue, responsible for the recognition of the 2’-phosphate on the NADPH ribose in related NADPH dependent FMOs, with a glutamine residue in SMFMO. In an attempt to explore the cofactor determinants in SMFMO, the two residues Gln193 and His194 in the cofactor binding site of SMFMO were mutated in order to mimic the cofactor binding site of the NADPH-dependent FMO, mFMO, from Methylophaga aminisulfidivorans sp. SK1, in which structurally homologous residues Arg234 and Thr235 bind the 2’-phosphate on NADPH. mFMO possesses an asparagine residue which is thought to be involved in the stabilisation of the flavin hydroperoxide intermediate, in SMFMO this residue is replaced by Phe52. Mutation of the Phe52 residue revealed that this residue is a determinant in enantioselectivity. The natural variants of SMFMO, PFMO from Pseudomonas stutzeri and CFMO from Cellvibrio sp., also had the ability to use both nicotinamide cofactors equally to reduce the flavin. The structure of PFMO revealed that the residues Gln194 and Glu195, structurally homologous to Gln193 and His194 in SMFMO, were orientated away from the 2’-phosphate site and thus the Glu195 would not repel the negatively charged phosphate as originally thought.

The Baeyer-Villiger reaction is a key reaction in organic synthesis, due to the utility of the addition of an oxygen atom adjacent to a carbonyl group. This reaction is also useful in an industrial setting and Baeyer-Villiger monooxygenases are often capable of performing this reaction in an exceptionally regio- and enantio-selective manner. This remarkable selectivity means that they are excellent biocatalyst targets for a number of industrially relevant syntheses, including the stereoselective synthesis of lactones and sulfoxides and also the resolution of racemic species, including �-hydroxyketones. This PhD project focussed on the enzyme MO14, encoded by the gene ro03437 from the bacterium Rhodococcus jostii sp. RHA1. MO14 has previously demonstrated particularly high regioand enantioselectivity in the conversion of the model BVMO substrate, bicyclo[3.2.0]hept-2-en-6- one. This enzyme, along with several others from the same organism, was selected for study of the activity and MO14 in particular has been singled out due to its remarkable breadth of substrate scope and S-selective character. All of the selected enzymes were tested against a selection of industrially relevant targets, then focus concentrated on MO14, as it demonstrated the most interesting biocatalytic activities. A variety of purification strategies were examined for the purification of MO14, with several potential lines of enquiry identified for the full purification of this enzyme. A study of the transformation of bicyclo[3.2.0]hept-2-en-6-one was conducted, with several variables of the reaction assessed, followed by investigation of the ability of this enzyme to transform a range of prochiral sulfides. As a precursor to industrial applications, a series of scale-up reactions were conducted using MO14 to examine the potential for use on a scale much larger than standard laboratory investigation. Finally, a series of mutants were generated to examine the origin of the exceptional selectivity exhibited by this enzyme.

The formation or cleavage of carbon-carbon bonds is often thermodynamically unfavourable and, using abiotic catalysts, only achieved under harsh reaction conditions. As an alternative, numerous enzymes are able to catalyse equivalent reactions and constitute versatile biocatalytic tools for the production of high value chemical building blocks. Three such catalysts have now been analysed with respect to their structure, function, reaction mechanism and/or substrate specificity. The Bacillus subtilis phenolic acid decarboxylase (BsPAD) is a carbon-carbon bond lyase. Despite the availability of an apo structure, the enzyme’s reaction mechanism remained speculative. Within this work, an active site mutant library was prepared and analysed and the structure of one of the variants could be determined in complex with its phenolic acid substrate. Using combined structural and kinetic data, a catalytic mechanism was proposed and confirmed previous modelling approaches. In addition, the codon optimised genes of two carbon-carbon bond hydrolases were subcloned for recombinant protein production. Phloretin hydrolase (Phy), a retro-Friedel Crafts hydrolase from Eubacterium ramulus, was purified and crystallised. Although the crystals’ diffraction quality was too poor for structure determination, a fluorescence spectrum revealed Phy to bind zinc. Together with a substrate screen and kinetic data from an active site mutant library, these findings gave novel insights into a unique metal dependent hydrolase. As a member of the β-diketone cleaving family of enzymes, the oxidised polyvinyl alcohol hydrolase (OPH) from Pseudomonas sp. VM15C was analysed. After extensive solubility screening and construct optimisation, OPH could, for the first time, be purified as a soluble fusion construct with an N-terminal GST tag. The fused protein crystallised and now awaits further structural and mechanistic analysis. It is hoped that the data obtained as part of this project will now advance the understanding of the characterised biocatalysts and create a basis for their optimisation and use in industrial applications.

To date more than 15000 Cytochromes P450 have been identified and named so far, with one third belonging to the plant kingdom. This is a key biochemical resource, providing a wealth of biocatalysts covering a diverse range of chemistries. Characterisation, however, has been greatly hindered by the poor solubility of many P450s, a result of the membrane anchoring region common to all plant P450s. Fusions of plant P450 heme domains to an appropriate reductase without the hydrophobic membrane anchor could provide the basis for developing robust, soluble plant enzyme systems for substrate screens to discover novel activities that are also of benefit to industry. In this project, the two predominantly expressed Arabidopsis reductases ATR1 and ATR2 have been cloned, without the membrane anchor, and expressed in Escherichia coli. These two truncated enzymes have been purified and assessed for activity with ATR2 found to be more active than ATR1. ATR2 was chosen for engineering into a novel plant P450 reductase vector platform for high throughput applications, whereby the P450s can be easily and quickly swapped using ligation independent cloning techniques. Four different plant P450s (CYP93C1, CYP73A5, CYP82E4 and CYP81D8) were selected to validate this technology, and activity for the fusions of CYP93C1 (Isoflavone synthase I from Glycine max) and CYP73A5 (cinnamate-4-hydroxylase from Arabidopsis) with ATR2 have been shown. The presence of CYP73A5 fused to ATR2 was verified through purification and further studies showed that it has to be membrane associated for activity. Additionally, CYP93C1 and CYP73A5 were also fused with the bacterial RhF reductase from Rhodococcus sp. and expressed in E. coli and compared to the plant P450 – plant reductase fusion protein. These novel plant-bacterial fusion P450 systems are the first example of active plant P450s fused to a reductase from a bacterium. This platform technology will provide the possibility for characterisation studies of eukaryotic P450s with unknown function and the discovery of new activities.

Amino acid oxidases (AAOs) are enantioselective flavoenzymes that catalyse the oxidation of amino acids into imino acids, which spontaneously hydrolyse in water to form keto acids. AAOs have several potential applications, most notably as biocatalysts in the production of enantiomerically pure amino acids and keto acids, or in enzymatic biosensors. Therefore, there is a demand for a range of AAOs with specific activity against various substrates, encouraging the characterisation of less well understood oxidases. Seven putative oxidases were cloned into the pET-YSBLIC-3C expression vector, expressed in E. coli-DE3 expression strains and assayed for activity against all proteinogenic amino acids. Of these seven targets, the L-amino acid oxidase from Pseudomonas putida (PpLAAO) was found to be highly soluble, had detectable activity against L-aspartate and L-asparagine and had not been investigated in depth previously. The purified PpLAAO protein showed high substrate specificity against L-aspartate and lower activity with substrate inhibition against L-asparagine. Very low activity was also detected against L-glutamate. The purified protein had optimal activity around pH 7.5 and at temperatures between 4°C and 30°C. To investigate the role of residues in the active site area of the PpLAAO protein thirteen active site residues, determined by comparison with the structure of the L-aspartate oxidase from E. coli (L-AspO), were mutated to alanine. Eleven of these mutants were purified and assayed against L-aspartate, L-asparagine and L-glutamate. Results were largely consistent with knowledge of L-AspO. Ingenza Ltd. has an interest in potential applications of L-tyrosine and L-alanine oxidases. Because of this iterative combinatorial active site saturation testing, using the active structure of L-AspO was performed alongside a small scale epPCR mutagenesis in an attempt to introduce activity against L-alanine and L-tyrosine. L-homoserine was also targeted as part of a substrate walking approach towards L-alanine; however no novel activity was detected in any transformants.

The bacterial and archaeal Asp/Glu racemase enzyme superfamily contains a variety of catalytic functions that have great potential for use in industrial biocatalysis. Members of this superfamily include aspartate racemases (AspRs), glutamate racemases (GluRs), hydantoin racemases (HydRs), arylmalonate decarboxylases (AMDs) and maleate cis-trans isomerases (MIs). Despite their catalytic diversity, all characterised members share the same protein fold, catalytic cysteine residues and reaction intermediate. Attempts to exploit this evolutionary flexibility for new processes have had limited success so far, showing that the employed mechanisms are not yet fully understood. For example, the well-characterised Bordetella bronchiseptica AMD (BbAMD) enantiospecifically decarboxylates a range of arylmalonates was but is not able to decarboxylate alkylmalonates despite considerable efforts made by site directed mutagenesis. In this work an investigation of the sequence diversity of the superfamily was undertaken and a range of BbAMD sequence homologues was tested for both aryl- and alkylmalonate decarboxylation (Chapter 3). However, none of the homologues exhibited decarboxylation activity. Targeted mutation of active site residues in an attempt to introduce decarboxylase activity was also unsuccessful. In an alternative approach to identify new alkylmalonate decarboxylating enzymes, a range of bacterial strains capable of processing alkylmalonates was isolated using selective enrichment from soil samples (Appendix D). The only characterised superfamily enzymes without a described three dimensional protein structure are MIs. In order to illuminate the distinct mechanism of MIs, the activity of the superfamily member Nocardia farcinica MI (NfMI) was characterised (Chapter 4) and its structure was determined by X-ray crystallography (Chapter 5). A potent inhibitor and substrate analogue bromomaleate was found. Mutagenesis of the active site cysteine dyad confirmed its catalytic role and Cys76 was found to be more important than Cys194. The data support a mechanism initiated by nucleophilic attack by Cys76 on the double bond of maleate. Although alternative mechanisms cannot be excluded at present, these findings indicate that the mechanistic chemistry in the superfamily is more adaptable than previously thought.