Ubiquitination is an essential post-translational modification that regulates signalling and protein turnover in eukaryotic cells. However, many ubiquitin E3 ligases remain poorly understood. The mammalian muskelin/RanBP9/CTLH complex contains eight proteins, five of which, RanBP9 (RanBPM), TWA1, Maea, Rmnd5a and muskelin, share striking similarities of domain organisation. In Saccharomyces cerevisiae, the related GID complex includes the Rmnd5a homologue GID2 which has E3 ubiquitin ligase activity and down-regulates gluconeogenesis. E3 ubiquitin ligase activity of mammalian Rmnd5a has not been reported. To better understand the large mammalian complex a major goal of this thesis was to analyse its evolution as a multi-protein system. Bioinformatic studies identify that TWA1, Rmnd5 and Maea are conserved throughout five eukaryotic supergroups. RanBPM is absent from excavates and from some lineages within other super-groups, and muskelin is present only in opisthokonts. Phylogenetic analysis based on the shared sequence regions that correspond to the lissencephaly-l homology (LisH) and C-terminal to LisH (CTLH) domains revealed closer relationships between Rmnd5 and MAEA, and TWAl and RanBPM, respectively. In-depth sequence analyses confirmed the greater similarity of the LisH/ CTLH domains of Rmnd5 and MAEA vs. TWAl and RanBPM, respectively, and id~ntified unique signatures of conserved residues within the LisH and CTLH domains of each protein. ~ further goal was to purify and express Rmnd5a and TWAl for laboratory experiments. Bacterially expressed Rmnd5a exhibits E3 ubiquitin ligase activity in Escherichia coli BL21lysates but not as a purified protein. Bacterial expression and purification of TWAl enabled biophysical characterisation of TWAl as an all a-helical, natively-dimerised protein. TWAl crystals were produced. When optimized, crystals diffracted to 3.5A, though a 3D structure was not resolved. Threaded structure predictions of Rmnd5a and TWAl agreed with secondary structure prediction algorithms. These studies advance knowledge of structural! functional relationships of proteins in this poorly-understood complex.

Caspases are cysteine-dependent aspartate-directed proteases responsible for the proteolysis of a plethora of substrates during programmed cell death. These include structural proteins of the cytoplasm and nucleus, components of the DNA repair machinery, protein kinases, signalling proteins and regulatory proteins. Caspases are synthesised as relatively inactive zymogens, that become activated by scaffold-mediated transactivation or <i>via</i> cleavage by upstream proteases in an intracellular cascade. The resulting heterotetrameric enzymes possess a unique absolute requirement for aspartate at the substrate cleavage site, and recognise a tetrameric sequence within the substrate. In order to assess the role of caspases in apoptotic execution, I set out to evaluate the synthesis of novel caspase inhibitors, which would enable the detection of active caspases from apoptotic whole cell extracts. First a 2,4-dinitrophenyl probe was designed for the affinity tagging of caspases in two-dimensional gel electrophoresis. Second, biotinylated peptidylaldehydes were prepared which will enable the affinity purification of caspases from apoptotic cytosolic extracts under non-denaturing conditions. To enable biochemical studies of caspases, I developed a method, which permits the affinity purification of caspases. Apoptotic chicken hepatoma cell-line extracts were purified over an avidin column using a biotinylated probe. Finally, to permit the appraisal of the caspase proteomic variability between different cell types, and methods of apoptotic induction, and the identification of post-translational modifications of caspases, I developed a reproducible system for the identification of caspases by two-dimensional gel electrophoresis.

Flavocytochrome <i>c₃ </i>is a soluble , periplasmic, fumarate reductase from <i>Shewanella frigidimarina </i>which consists of three domains. This thesis reports an investigation into the enzyme mechanism by site directed mutagenesis. Fumarate is bound in the active site via important interactions with the side chains of His504, His365, Thr377 and Arg544. The substitution of Thr377 by alanine causes a 13-fold decrease in k_cat and a 27-fold increase in K_M,consistent with a role purely in substrate binding. Substituting Arg544 with methionine dramatically lowers the k_cat by 10⁴-fold and raises the K_M 29-fold. This residue is involved in substrate binding but is likely to have an additional role polarising the substrate molecule, for hydride attack. A structural sodium ion is located close to both the active site and the FAD tail group. It is bound in approximately octahedral geometry by five backbone carbonyls (Thr506, Met507, Gly508, Thr536 and Glu534) and a water molecule. Substituting Glu534 results in an inability to retain FAD. His505 hydrogen bonds to the water molecule ligating the sodium and is also next to His504, which is one of the substrate binding residues. The pH-activity profile of wild-type fits to a single pK_a value of 7.5 ± 0.1 which is attributed to His504 stabilising the build up of charge in the reaction intermediate. This pK_a is raised to 8.2 ± 0.1 by the substitution H505Y and to 9.0 ± 0.2 by H505A. The k_cat value is lowered 2-fold and 20-fold for these mutants respectively. In the mutant enzymes His504 has become a weaker acid and is less able to enhance the rate at low pH. The role of His505 may be to moderate the effect of the negative charge of Glu534 on His504. The active site acid, Arg402, has a dual role as both a Lewis acid (stabilising the build up of charge after hydride transfer) and a Brønsted acid (delivering a proton to the substrate C3). The structure of the mutant enzyme R402A revealed a water molecule at the active site (Mowat; 2001), but this is too far from the substrate C3 to act as the acid catalyst and the enzyme is completely inactive (Doherty; 2000). The double mutant Q363F/R402A, however, is active but at a level 10⁴-fold lower than wild-type. The structure shows that the water molecule is now close enough to the substrate C3 to act as an acid catalyst. Wild-type has an overall solvent isotope effect k_H/k_D, of 8.2 ± 0.4 and its proton inventory fits to a model for multiple exchangeable hydrogenic sites, consistent with a complex transition state involving a proton pathway. Q363F/R402A has double the solvent isotope effect of wild-type and the proton inventory indicates that the transition state remains complex. So the active site water is trapped and requires reprotonation by the proton pathway. Substituting Arg402 by glutamine lowers the k_cat 10⁵-fold. Not only is glutamine a poor Lewis acid but the structure shows that the shorter side chain results in an increased proton transfer distance. The mutant R402F is completely inactive as phenylalanine is unable to protonate the substrate.

The function of an enzyme is often dependent on a few key functional residues and the principal objective of this project was to develop a novel function prediction system which takes advantage of this by comparing the conserved amino acids in known enzyme families to those in a putative enzyme. Multiple sequence alignments of well characterised enzyme families (with an E.C. number assigned) are used to create unordered sets of conserved functional residues, termed <i>Treads</i>.  Comparison of a query proteins <i>Tread </i> to the reference <i>Treads</i> is undertaken by projecting them in multidimensional space and measuring distance between them. A major advantage of this prediction strategy implemented in DAROGAN is that it should be able to recognise similarities in the functions of enzymes that are not similar in structure or sequence. The method has been tested with regard to its ability to predict cofactor-dependencies toward pyridoxal-5’-phosphate, thiamine, glutathione and folic acid utilising enzymes. An area of application for DAROGAN is the prediction of previously described enzyme functions in organisms with completed genomes to which no gene and protein sequence could be assigned though the standard annotation processes. Investigations were made into the potential of utilising the DAROGAN method to propose candidates for the missing pyridoxal-5’-phosphate utilising enzymes in the <i>E. coli</i> genome according to EcoCyc. Candidates are proposed by assessing the 511 sequences from the GeneQuiz project, to which there are homologues in other species, but with uncertain functions. The assessment takes the form of using the DAROGAN method to determine the similarities of each of the sequences to the reference <i>Treads.</i>

The methyl-CpG-binding domain protein MBD4 can bind to methylated DNA and has glycosylase activity against thymine or uracil mismatched with guanine, preferentially within a CpG site (Hendrich <i>et al.,</i> 1999). This particular combination of activities indicates that MBD4 may be involved in the recognition and repair of deamination-induced mismatches at CpG sites. A prediction of this model is that the number of C:G-T:A mutations would be elevated in the absence of MBD4. This hypothesis has been tested by the use of an <i>in vivo</i> mouse mutational assay to compare wild-type and MBD4-deficient animals. Analysis of the mutational status of these mice shows that MBD4-deficiency markedly increases the incidence of C:G-T:A mutations at CpG sites <i>in vivo</i>, thus demonstrating a role for MBD4 in the initiation of repair at deamination-induced G-T mismatches. Since a DNA repair defect can predispose to tumour formation, the consequences of MBD4 deficiency in relation to tumour formation were also examined. The data indicate that MBD4 acts as a tumour supressor in the intestine. A separate avenue of investigation has been to examine whether MBD4 acts alone or with partner proteins. Two-hybrid screening in yeast identified the mismatch repair protein MLH1, the kinase ZIP, and a novel protein as potential partners of MBD4. The possible biological roles of the interaction of MBD4 with these proteins have been probed using biochemical assays. In summary, the work presented in this thesis demonstrates that MBD4 acts <i>in vivo</i> to initiate repair at deamination-induced mismatches. In addition, the interactions of MBD4 with ZIP kinase, as well as its role in tumour suppression in the intestine indicate that this protein may have other, previously undescribed functions.

Substitutions of the phylogenetically conserved phenylalanine residue F393 were constructed in flavocytochrome P450 BM3 from <i>Bacillus megaterium. </i>The high degree of conservation of this residue in the P450 superfamily and its proximity to the heme (and its ligand Cys400) implies an essential role in P450 activity. Extensive kinetic and thermodynamic characterisation of mutant enzymes F393A, F393H and F393Y highlighted significant differences from wild-type P450 BM3. All enzymes expressed to high levels and contained their full complement of heme. Whilst the reduction and subsequent treatment of the mutant P450s with carbon monoxide led to the formation of the characteristic P450 spectra in all cases, the absolute position of the Soret absorption varied across the series WT/F393Y (449 nm), F393H (445 nm) and F393A (444 nm). Steady-state turnover rates with both laurate and arachidonate showed the trend WT>F393Y>>F393H>F393A. Conversely, the trend in the pre-steady-state flavin-to-heme electron transfer was the reverse of the steady-state scenario, with F393A>F393H>>F393Y>wild-type. These data are consistent with the more positive substrate-free (-312 mV (F393A), -332 mV (F393H)) and substrate-bound (-151 mV (F393A), -176 mV (F393H)) reduction potentials of F393A and F393H heme domains, favouring the stabilisation of the ferrous-form in the mutant P450s relative to wild-type. Elevation of the heme iron reduction potential in the F393A and F393H mutants facilitates faster electron transfer to the heme. This results in a decrease in the driving force for oxygen reduction by the ferrous heme iron, hence an overall decline in the rate of turnover of the mutant P450s. We postulate that the nature of the residue at position 393 is important in controlling the delicate equilibrium observed in P450s, whereby a trade-off is established between the rate of heme reduction and the rate at which the ferrous heme can bind and, subsequently, reduce molecular oxygen. The structural and spectroscopic characterisation of the mutants reveals the probable role of the phenylalanine to be to preserve the hydrophobic environment of the heme. This prevents unfavourable inter- and intra-molecular interactions with the heme or heme-ligand, which have been demonstrated to perturb this delicate thermodynamic balance. This study highlights the electronic influence the cysteine ligand has in dictating the characteristic reactivity of P450s, and its sensitivity to its chemical environment. Modulating the electron density of the Fe-S bond (introduction of intramolecular hydrogen bonds) has a direct effect on the ability of P450s to bind and activate molecular oxygen. This point is key to the high degree of conservation of this residue throughout the P450 superfamily ?—monooxygenation requires oxygen activation. This phenylalanine is conserved in every member which needs to perform this activation, for the small number that do not, the nature of this residue varies.

Cytochrome P-450 monoxygenases catalyse the insertion of molecular oxygen into unactivated carbon-hydrogen bonds. This reaction is difficult to reproduce synthetically, therefore cytochrome P-450 enzymes are of great interest as potential biocatalysts. Of all the cytochrome P-450 systems investigated to-date cytochrome P-450cam form <i>Pseudomonas putida </i>is probably the best characterized. The use of cytochrome P-450cm, as a biocatalyst, is hampered by the narrow substrate range. This was overcome by use of the cytochrome P-450cam Y96A mutant. To identify the substrate range of the enzyme a model of the cytochrome P-450cam binding site was generated and potential substrates were modelled. Discrepancies between predicted and observed results led to a more rigorous study of the mutant binding site. Conditions were developed and the cytochrome P-450cam mutant enzyme was crystallized in the presence of a number of substrates. Six different substrate bound structures were then solved by X-ray crystallography. With a crystal structure of the binding site it was then possible to perform more rigorous molecular modelling and this was achieved with modelling program DOCK. Biotransformations and binding studies were performed with a range of substrates to test the validity of the modelling and to characterize the substrate specificity of the cytochrome P-450cam Y96A mutant.

Biotin protein ligase (BPL) catalyzes the formation of biotinyl-5’-AMP from biotin and ATP and the succeeding biotinylation of the biotin carboxyl carrier protein (BCCP). The genes encoding the BPL from the hyperthermophyle <i>Aquifex aeolicus </i>were overexpressed in <i>E. coli</i> and the recombinant protein <i>Aa</i>BPL was purified to homogeneity. <i>Aa</i>BPL was purified to homogeneity.  <i>Aa</i>BPL was proven to be catalytically active and to biotinylate specifically the C-terminal biotinyl domain of BCCP (BCCPΔ67). It was determined by isothermal titration calorimetry experiments that in the class I <i>Aa</i>BPL, the presence of biotin is not required for ATP binding in absence of Mg²⁺ ions and the binding of biotin and ATP has been determined to occur <i>via </i>a random but cooperative process. In the second step of the enzymatic reaction, BPL has been suggested to form a BPL:BCCP complex. This complex was characterized in <i>A. aeolicus </i>by chemical cross-linking and mutational studies have identified a salt bridge between <i>Aa</i>BPL and BCCPΔ67 which is important for heterodimerisation. The structures at 2.4 Å resolution of <i>Aa</i>BPL in the apo-form and in complex with biotin and ATP were determined. These are the first crystal structures of a BPL complex with biotin and ATP and also of an ATP-bound BPL. The adenylate binding loop is ordered in the structure of apo-<i>Aa</i>BPL and the ATP binding pocket is well defined. The solvent-exposed β- and γ-phosphates of ATP are located in the inter-subunit cavity formed by the N- and C-terminal domains. The Arg40 residue from the conserved GXGRXG motif is shown to interact with the carboxyl group of biotin and to stabilise the α- and β- phosphates of the nucleotide.

The results presented herein represent a programme focussed towards producing a deracemisation system for hydroxylamines. A number of different avenues have been explored. The use of a number of biocatalytic systems was investigated in an attempt to construct a one-pot system. These included a novel application of monoamine N and laccase enzymes. The enantioselective oxidation of hydroxylamine ethers by MAO-N was a novel discovery. A further development is the engineering of a MAO-N variant which displayed improved catalytic activity for a model hydroxylamine ether substrate compared with the parental enzyme. A second research programme lead to the discovery of the applicability of a laccase:mediator system to the oxidation of hydroxylamines. Investigations into oxime reduction systems with reaction conditions that were complementary to the biocatalytic systems were performed.

The cuticle of <I>C. elegans</I> is extensively cross-linked by covalent disulphide bridges, tyrosine bonds and possibly glutamyl-lysine bonds. Four genes predicted to be involved in the formation of tyrosine bonds have been identified in <I>C. elegans. </I>tyr<I>-1</I> and <I>tyr-2</I> map to chromosome III, <I>tyr-3</I> and <I>tyr-4</I> map to chromosome I. These encode tyrosinase-like enzymes. The tyrosinase genes are very similar in structure: all genes have two Cu active sites (CuA and CuB), predicted secretory leader peptides and sxc domains (found in other proteins from <I>C. elegans </I>and <I>Toxocara canis</I>). <I>tyr-1</I> has an additional polyglutamine region which may be involved in protein-protein interactions. A set of cDNAs prepared from a synchronous population of worms, harvested at two hour intervals through the lifecycle, starting shortly after hatching, was used in semi-quantitative fluorescent PCR. Steady state levels of <I>tyr-1, -2</I> and <I>-4</I> genes are upregulated at each moult, suggesting their involvement in the synthesis of the new cuticle. <I>tyr-3</I> transcripts could not be detected in this set of cDNAs, however it was isolated from a population of worms enriched in males. Studies using lacZ- and GFP-reporter genes driven by promoter fragments of the tyrosinase genes showed that <I>tyr-4</I> and <I>tyr-1</I> are expressed in specific subsets of hypodermal cells. In addition <I>tyr-1</I> is expressed in the vulval cells. <I>tyr-2</I> was found to be expressed only faintly in hypodermal cells, and showed strong expression in the uterine cells. No expression of <I>tyr-3</I> was observed. These data imply that tyrosinases are not only involved of cross-linking of cuticular proteins, but are probably also involved in the generation of the egg shell.