An increasing number of diseases are associated with protein misfolding, one type of which results in the formation of amyloid fibrils. This research has addressed the hypothesis that all proteins can form amyloid fibrils and investigates what factors protect proteins from forming these macromolecular assemblies. Most analyses of the aggregation propensity of proteins have been limited to the properties of the amino acid sequence, thus fail to consider the roles that higher levels of organisation play in protecting polypeptides from misfolding. The (α/β)8 barrels are a common class of proteins and have never been shown to form amyloid fibrils. This thesis aims to elucidate the characteristics that prevent (α/β)8 barrels from misfolding using Escherichia coli dihydrodipicolinate synthase (DHDPS), a homotetrameric (α/β)8 barrel protein, as a model. It is widely accepted that the precursor of amyloid fibrils is a partially folded species. It is hypothesised in this thesis that the (α/β)8 barrel fold protects the protein against this partial unfolding. This was tested by generating a catalogue of site-directed mutants of DHDPS and screening each of these in a range of pHs and ionic strengths. Amorphous aggregation propensity was assessed by monitoring light scattering at 340 nm and β-sheet specific aggregation was assessed using ThT. Thermal stability was monitored using DSF and CD spectroscopy. Crystallography was used to assess tertiary and quaternary structures and in the cases where crystal structures were not obtained, kinetics was used as a proxy indicator of correct folding and monomeric association. CD spectroscopy was also used to investigate the secondary structure of the DHDPS variants and analytical gel permeation liquid chromatography and AUC were used to confirm quaternary structure. The stability and aggregation propensity of DHDPS and its variants were assessed under a range of pH and salt conditions. It was established through the characterisation of the wild-type protein that the predominant determinant of stability was, unsurprisingly, pH. This was a trend observed for all the variants described. Affinity tags were used during the course of this research to facilitate and expedite the production of the protein variants. The introduction of tags containing a polyhistidine motif to DHDPS significantly altered some biophysical properties. Whilst the secondary and quaternary structures were found to be similar to the wild-type enzyme, the catalytic properties were changed. In addition to this, the propensity to aggregate was altered. The full-length polyhistidine tags increased the propensity of DHDPS to form β-sheet-specific aggregate, although this did not result in the formation of amyloid fibrils for most of the variants. The Zyggregator algorithm was used to predict amino acid substitutions that would increase the aggregation propensity of DHDPS. It identified several amino acids, three of which were chosen for mutation and two of which were expressed in sufficient quantity for further study. DHDPS Q90L and A207V were characterised and the amino acid substitutions did not significantly alter the kinetic parameters of the enzyme. The crystal structure of A207V was solved and confirmed the results of the kinetic analysis demonstrating unchanged tertiary and quaternary structures. Both variants exhibited tertiary and quaternary structures similar to the wild-type enzyme, although Q90L contained more disorder than the wild-type enzyme. The thermal denaturation temperatures and aggregation propensities were also similar to wild-type, although the propensity for both variants to form β-sheet-specific aggregates was reduced. The combinatorial effects of Q90L, A207V and the polyhistidine tags were assessed. This revealed that whilst most biophysical properties were unaffected, the β-sheet-specific aggregation propensity for pET M11 and pET 151/D-TOPO DHDPS Q90L and pET M11 DHDPS A207V, were significantly increased compared to the wild-type enzyme. The evolutionary forces driving the association of the monomeric and dimeric subunits of DHDPS are undetermined. Investigation of two quaternary structure mutants (DHDPS Y107W and L197Y) revealed that the tetrameric nature of E. coli DHDPS is important for protein activity, stability and the prevention of aggregation. The combinatorial affects of the disrupted quaternary structure and the polyhistidine tags further increased the predisposition of DHDPS to form β-sheet-specific aggregates, resulting in the formation of linear aggregates with some characteristics of amyloid fibrils. The additive affect of Q90L, Y107W and a polyhistidine tag was assessed and revealed that the major determinant in protein stability and prevention of amorphous and β-sheet specific aggregation is the quaternary structure. This study demonstrates that the destabilisation of the quaternary structure of DHDPS can result in the formation of amyloid-like aggregates by an (α/β)8 barrel, the first example of an (α/β)8 barrel misfolding in such a way. This finding supports the assertion that all proteins can form amyloid fibrils.

The homotetrameric enzyme dihydrodipicolinate synthase (DHDPS, E.C. 4.2.1.52) from Escherichia coli was used as a model for probing oligomeric structure in enzymes. Dimeric mutants of this enzyme have been found in previous work to be largely inactive, due to the trapping of a covalent adduct. Partial restoration of catalytic activity has been achieved by incubation in the presence of the substrate pyruvate to displace the adduct. It was hypothesized that the buttressing of dimeric units against one another in the wildtype tetrameric form of DHDPS provides stability in the dimer interface, necessary to maintain optimum catalytic performance and substrate specificity. We hypothesized that buttressing a dimeric DHDPS mutant against a surface would result in restoration of catalytic activity by mimicking the buttressing proposed to occur in the tetrameric structure. To test this hypothesis, dimeric DHDPS mutants were immobilised against an agarose support and the immobilised enzymes characterised. Three DHDPS mutants were prepared, the double mutant DHDPS-C20S/L167C was produced by mutagenesis and a crystal structure obtained in collaboration with Dr Renwick Dobson. Two other mutants, DHDPS-Ll67C and DHDPS-Ll97Y were also over expressed and purified. The quaternary structures of the three mutants were characterised in solution, DHDPS-Ll67C was determined to be tetrameric, DHDPS-C20S-Ll67C was found to equilibrate between tetramer and dimer and DHDPS-Ll97Y was confirmed as a dimer, consistent with previous findings. Modification experiments indicated that the sulfhydryl groups of DHDPS-C20S/L167C were available for immobilisation. Activation experiments indicated that both DHDPS-Ll67C and DHDPS-Ll97Y activated. These results were in accord with those of others in indicating that the displacement of an a-ketoglutarate adduct from the active site was responsible for the activation of mutant DHDPS enzymes. Wild-type DHDPS and the mutants were immobilised through amine and sulfhydryl groups. The free and immobilised enzymes were rigorously characterised, with thermal stability, pH optima, kinetic and lysine inhibition properties determined and compared to wild-type DHDPS. Following immobilisation, substrate affinity was found to decrease for wild-type and mutant enzymes, wild-type KmPyr = 0.26 mM free, 0.8-1.2 mM immobilised, Km(S)-ASA = 0.10 mM free, 1.5-2.5 mM immobilised. Lysine inhibition was determined to be largely unaffected by immobilisation. The largest change in K, was an increase to double that of the free enzyme. Restoration of some catalytic activity was found following the immobilisation of dimeric DHDPS-Ll97Y, the immobilised enzyme was 31 ± 12% more active than free DHDPS-Ll97Y. DHDPS-C20S/L167C was also found to immobilise as a dimer. Comparison ofthe immobilised DHDPS-C20S/L167C dimer with a derivatised free dimeric form ofthis enzyme indicated that an increase from 3% to 9% of wild-type activity had resulted from immobilisation. These results supported the hypothesis that buttressing of a dimeric mutant of DHDPS against a support surface would increase catalytic activity and that buttressing across the dimerdimer interface is essential for optimal catalytic activity in DHDPS enzymes.

The homotetrameric enzyme dihydrodipicolinate synthase (DHDPS, E.C. 4.2.1.52) from Escherichia coli was used as a model for probing oligomeric structure in enzymes. Dimeric mutants of this enzyme have been found in previous work to be largely inactive, due to the trapping of a covalent adduct. Partial restoration of catalytic activity has been achieved by incubation in the presence of the substrate pyruvate to displace the adduct. It was hypothesized that the buttressing of dimeric units against one another in the wildtype tetrameric form of DHDPS provides stability in the dimer interface, necessary to maintain optimum catalytic performance and substrate specificity. We hypothesized that buttressing a dimeric DHDPS mutant against a surface would result in restoration of catalytic activity by mimicking the buttressing proposed to occur in the tetrameric structure. To test this hypothesis, dimeric DHDPS mutants were immobilised against an agarose support and the immobilised enzymes characterised. Three DHDPS mutants were prepared, the double mutant DHDPS-C20S/L167C was produced by mutagenesis and a crystal structure obtained in collaboration with Dr Renwick Dobson. Two other mutants, DHDPS-Ll67C and DHDPS-Ll97Y were also over expressed and purified. The quaternary structures of the three mutants were characterised in solution, DHDPS-Ll67C was determined to be tetrameric, DHDPS-C20S-Ll67C was found to equilibrate between tetramer and dimer and DHDPS-Ll97Y was confirmed as a dimer, consistent with previous findings. Modification experiments indicated that the sulfhydryl groups of DHDPS-C20S/L167C were available for immobilisation. Activation experiments indicated that both DHDPS-Ll67C and DHDPS-Ll97Y activated. These results were in accord with those of others in indicating that the displacement of an a-ketoglutarate adduct from the active site was responsible for the activation of mutant DHDPS enzymes. Wild-type DHDPS and the mutants were immobilised through amine and sulfhydryl groups. The free and immobilised enzymes were rigorously characterised, with thermal stability, pH optima, kinetic and lysine inhibition properties determined and compared to wild-type DHDPS. Following immobilisation, substrate affinity was found to decrease for wild-type and mutant enzymes, wild-type KmPyr = 0.26 mM free, 0.8-1.2 mM immobilised, Km(S)-ASA = 0.10 mM free, 1.5-2.5 mM immobilised. Lysine inhibition was determined to be largely unaffected by immobilisation. The largest change in K, was an increase to double that of the free enzyme. Restoration of some catalytic activity was found following the immobilisation of dimeric DHDPS-Ll97Y, the immobilised enzyme was 31 ± 12% more active than free DHDPS-Ll97Y. DHDPS-C20S/L167C was also found to immobilise as a dimer. Comparison ofthe immobilised DHDPS-C20S/L167C dimer with a derivatised free dimeric form ofthis enzyme indicated that an increase from 3% to 9% of wild-type activity had resulted from immobilisation. These results supported the hypothesis that buttressing of a dimeric mutant of DHDPS against a support surface would increase catalytic activity and that buttressing across the dimerdimer interface is essential for optimal catalytic activity in DHDPS enzymes.

Dihydrodipicolinate Synthase (EC 4.3.3.7; DHDPS), the product of the dapA gene, is an enzyme that catalyzes the condensation of pyruvate and S-aspartate-β-semialdehyde (ASA) into dihydrodipicolinate via an unstable heterocyclic intermediate, (4S)-hydroxy-2,3,4,5-tetrahydro-(2S)-dipicolinic acid. DHDPS catalyzes the first committed step in the biosynthesis of ʟ-lysine and meso-diaminopimelate; each of which is a necessary cross-linking component between peptidoglycan heteropolysacharide chains of bacterial cell walls. Therefore, strong inhibition of DHDPS would result in disruption of meso-diaminopimelate and ʟ-lysine biosynthesis in bacteria leading to decreased bacterial growth and cell lysis. Much attention has been given to targeting the active site for inhibition; however DHDPS is subject to natural feedback inhibition by ʟ-lysine at an allosteric site. In DHDPS from Campylobacter jejuni ʟ-lysine is known to act as a partial uncompetitive inhibitor with respect to pyruvate and a partial mixed inhibitor with respect to ASA. Little is known about how the protein structure facilitates the natural inhibition mechanism and mode of allosteric signal transduction. This work presents ten high resolution crystal structures of Cj-DHDPS and the mutant Y110F-DHDPS with various substrates and inhibitors, including the first reported structure of DHDPS with ASA bound to the active site. As a body of work these structures reveal residues and conformational changes which contribute to the inhibition of the enzyme. Understanding these structure function relationships will be valuable for the design of future antibiotic lead compounds. When an inhibitor binds to the allosteric site there is meaningful shrinkage in the solvent accessible volume between 33% and 49% proportional to the strength of inhibition. Meanwhile at the active site the solvent accessible volume increases between 5% and 35% proportional to the strength of inhibition. Furthermore, inhibitor binding at the allosteric site consistently alters the distance between hydroxyls of the catalytic triad (Y137-T47-Y111') which is likely to affect local pKa's. Changes in active site volume and modification of the catalytic triad would inhibit the enzyme during the binding and condensation of ASA. The residues H56, E88, R60 form a network of hydrogen bonds to close the allosteric site around the inhibitor and act as a lid. Comparison of ʟ-lysine and bislysine bound to wt-DHDPS and Y110F-DHDPS indicates that enhanced inhibition of bislysine is most likely due to increased binding strength rather than altering the mechanism of inhibition. When ASA binds to the active site the network of hydrogen bonds among H56, E88 and R60 is disrupted and the solvent accessible volume of the allosteric site expands by 46%. This observation provides some explanation for the reduced affinity of ʟ-lysine in high ASA concentrations. ʟ-Lysine, but not other inhibitors, is found to induce dynamic domain movements in the wt-DHDPS. These domain movements do not appear to be essential to the inhibition of the enzyme but may play a role in cooperativity between monomers or governing protein dynamics. The moving domain connects the allosteric site to the dimer-dimer interface. Several residues at the weak dimer interface have been identified as potentially involved in dimer-dimer communication including: I172, D173, V176, I194, Y196, S200, N201, K234, D238, Y241, N242 and K245. These residues are not among any previously identified as important for formation of the quaternary structure.

This thesis examined the enzyme dihydrodipicolinate synthase (DHDPS, E.C. 4.2.1.52) from the pathogen Mycobacterium tuberculosis. DHDPS is a validated antibiotic target for which no potent inhibitor based on substrates, intermediates or product has been found. The importance of the homotetrameric quaternary structure in E. coli DHDPS has been demonstrated by the 100-fold decrease in activity observed in a dimeric variant, DHDPS-L197Y, created by site-directed mutagenesis. This suggested a new approach for inhibitor design: targeting the dimer-dimer interface and disrupting tetramer formation. DHDPS catalyzes the first committed step in the biosynthetic pathway of meso-diaminopimelic acid, a critical component of the mycobacterial cell wall. In this study, wild-type M. tuberculosis DHDPS was thoroughly characterized and compared with the E. coli enzyme. A coupled assay was used to obtain the kinetic parameters for M. tuberculosis DHDPS: KM(S) ASA = 0.43 (±0.02) mM, KMpyruvate = 0.17 (±0.01) mM, and kcat = 138 (±2) s 1. Biophysical techniques showed M. tuberculosis DHDPS to exist as a tetramer in solution. This is consistent with the crystal structure deposited as PDB entry 1XXX. The crystal structure of M. tuberculosis DHDPS showed active-site architecture analogous to E. coli DHDPS and a dimeric variant of M. tuberculosis DHDPS was predicted to have reduced enzyme activity. A dimeric variant of M. tuberculosis DHDPS was engineered through a rationally designed mutation to analyze the effect of disrupting quaternary structure on enzyme function. A single point mutation resulted in a variant, DHDPS-A204R, with disrupted quaternary structure, as determined by analytical ultracentrifugation and gel-filtration chromatography. DHDPS-A204R was found to exist in a concentration-dependent monomer-dimer equilibrium, shifted towards dimer by the presence of pyruvate, the first substrate that binds to the enzyme. The secondary and tertiary structure of DHDPS-A204R was analogous to wild-type M. tuberculosis DHDPS as judged by circular dichroism spectroscopy and X ray crystallography, respectively. Surprisingly, this disrupted interface mutant had similar activity to the wild type enzyme, with a kcat of 119 (±6) s-1; although, the affinity for its substrates were decreased: KM(S) ASA = 1.1 (±0.1) mM, KMpyruvate = 0.33 (±0.03) mM. These results indicated that disruption of tetramer formation does not provide an alternative direction for drug design for DHDPS from M. tuberculosis. Comparison with the recently discovered dimeric DHDPS from Staphylococcus aureus shed further light on the role of quaternary structure in DHDPS. In M. tuberculosis DHDPS-A204R and the naturally dimeric enzyme, the association of monomers into the dimer involves a greater buried surface area and number of residues than found in E. coli DHDPS-L197Y. This provides a framework to discriminate which DHDPS enzymes are likely to be inactive as dimers and will direct future work targeting the dimer-dimer interface of DHDPS as an approach for drug design.