Traumatic brain injury (TBI) is a major cause of death and disability worldwide. The focus of my research project is the study of potassium dynamics in spreading depolarisation (SD) waves found in TBI. The SD waves occur following the injury and the ionic imbalance caused by these waves causes further brain damage and greater patient morbidity. The goal of my research is to detect these waves in real time and quantify their severity in order to help clinicians better understand and treat TBI quickly and effectively to improve patient outcomes. During my thesis project, I have first developed a miniaturised ion-selective electrode (ISE) for the detection of potassium (K+) transients associated with SDs. The average K+ ISE has a Nernstian sensitivity of 58.9 mV dec⁻¹ and temporal response of 5.1 seconds in the physiological range. The ISE, housed in a microfluidic flow cell with a sample volume of 70 nL, was applied in in vivo microdialysis studies to monitor real-time depolarisation waves. An SD wave causes the dialysate K+ to increase by 0.42 +/- 0.07 mM and 1.13 +/- 0.63 mM, from the physiological normal of 2.7 mM, in the animal model and in the human injured brains respectively. This dialysate SD marker has also been validated against tissue K+ level and cortical electrical activity which are currently the gold standards for SD detection. In continuous or single-phase laminar flow, Taylor dispersion prevails. The result is an attenuated measured concentration to that captured at the microdialysis probe. With the development of a droplet microfluidic system, the dialysis stream can be segmented into discrete droplets suspended in a carrier oil, preserving the original concentration character-istics of the sample as well as improving the temporal resolution by ten-fold. The sample droplets can be manipulated passively to fulfil a range of operations, such as fast mixing, merging and splitting, and to enable parallelisation of analysis. Lastly, the droplets, now the core unit of analysis, are also reliably detected using a newly developed miniaturised contactless device based on conductivity, removing the need of expensive optical equipment and interference with the chemistry of the droplet.

Operons are common in prokaryotic genomes but, except for Nematoda, are uncommon in eukaryotes. Nematoda operons comprise gene clusters, containing between two to eight genes, with short, approximately. 100 bp, intergenic regions. Operons are common in C. elegans with an estimated 17% of genes clustered in operons. The significance of Nematoda operons is unclear although evidence suggests they may have arisen during genome evolution and compaction - the so-called "easy come, slow go" theory. Conventionally a discrete nascent transcript, transcribed from a major promoter upstream of the first gene in the operon, undergoes frans-splicing and 5’ addition of one of two splice-leader sequences followed by ds-splicing of the resulting pre-mRNAs to form mature, discrete mRNAs. However evidence suggests operon gene transcription is likely to be more complex. For example, microarray data indicated that correlated operon gene expression is weaker with increasing intergenic distance suggesting expression is influenced by sequence elements located internally within the operon structure. Further evidence, derived from comparison of expression data generated with fluorescent protein (FP) transcriptional reporters driven by either the upstream "operon promoter" alone or by such potential intergenic "internal promoter" sequences, generated different expression patterns in different tissues. Such data indicated the presence of internal regulatory elements within a subset of operons which were subsequently termed "hybrid operons". An alternative method to further dissect the presence of such internal regulatory elements would be to utilize translational style reporters built directly from genomic clones in which an entire operon structure is contained within the insert sequence. Tagging each operon gene with a different FP would permit operon gene expression analysis within the genomic context of that operon.<br/>To facilitate construction of fosmid-based translational reporters via seamless counter-selection recombineering two resource sets were first constructed. The first of these contained FP coding sequences synthesized de novo and codon-optimized for expression in C. elegans. The second set comprised a series of constructs designed to streamline the recombineering method. A subset of these resources was used to insert FP coding sequences seamlessly into the 3’ ends of genes within three- (CEOP1312) and two- (CEOP1358) gene-containing operons each of which was located centrally on a different fosmid genomic clone insert. Comparative genomic analyses indicated that operon CEOP1312 was conserved across all available Caenorhabditis spp whereas, in contrast, operon CEOP1358 was not. Further investigation of the conservation of CEOP1312 within the Caenorhabditis spp lead tothe identification of a GC-rich non-coding sequence conserved in the intergenic region between the second and third genes in all the CEOP1312 orthologous operons. Making use of the available resource sets counter-selection recombineering was utilized to construct a series of fosmid-based translational reporters in which this conserved sequence was precisely, and seamlessly, manipulated. Although the absolute conservation of this non-coding sequence is strongly suggestive of functionality worms transgenic for these different constructs displayed essentially equivalent expression patterns for all CEOP1312 genes indicating that any such function may not, in this case, involve internal transcriptional regulation of one or more of the operon genes.

Microorganisms have been viewed as planktonic, free living single cells but predominantly they exist as sessile multispecies communities in the natural environment forming ‘biofilms’. Biofilms are beneficial for organisms to survive in natural environment as well as for biotechnological applications such as microbial fuel cells and bioremediation. However, biofilms are associated with disease persistence and biofouling and are comprised of adhered microbes within a hydrated matrix rich in polysaccharides, proteins and extracellular DNA (eDNA). eDNA is an important structural component and its degradation by deoxyribonucleases may be a novel approach to eradicate biofilm related problems. The present work was undertaken in this context to discover and produce microbial nucleolytic enzymes for applications for the control of harmful biofilms. Eighty six out of 260 bacterial isolates which included thermophilic and psychrophilic strains, showed deoxyribonuclease activity. The diversity and function of extracellular nucleases was also investigated throughout the microbial world using bioinformatics tools. Sequence driven analysis suggested that major bacterial lineages contain diverse extracellular nucleases with biological function related to nitrogen, phosphate and carbohydrate metabolism, protection, survival and virulence. Production optimisation for one specific extracellular nuclease, NucB, from Bacillus licheniformis EI-34-6 was carried out. This enzyme was previously known to cleave eDNA causing biofilm dispersal, and may therefore be used commercially to remove biofilms. The understanding of B. licheniformis physiology was applied in order to enhance NucB production 10-fold. For further characterisation of the enzyme and to Abstract iv understand its biological mechanism in breaking down biofilms, NucB was expressed in the SURE expression host B. subtilis NZ8900. This allowed a 68-fold increase in protein yield. NucB protein has been purified to high degree purity with specific nuclease activity of 15000 U/mg of protein. Biophysical characterisation showed that the protein was thermally stable and could reversibly refold. Statistical optimisation of extracellular nucleases production in diverse bacteria grown at different temperatures was demonstrated as a promising methodology for enhancing key enzyme secretion. The effectiveness of biofilm disruption by NucB was successfully tested with different single species biofilms grown on polystyrene, glass, and stainless steel surfaces. Biofilm dispersal efficiency of other microbial nucleases ranged between 60 – 95 % of removal after 1 h. The results presented in this thesis demonstrate that bacteria were able to produce nucleases across broad temperature range. In context to biofilm dispersal, bioinformatic analysis speculates the ecological implication of secreted diverse microbial extracellular nuclease-like genes were to decide the fate of eDNA and play pivotal role in nutrient cycling of the eco-system. Bioprocess development confirmed process optimisation can reliably produce functional and well-folded recombinant NucB at levels suitable for applications where biofilm removal is needed. Production optimisation of extracellular nucleases from diverse bacteria expanded the availability of different nucleases with wide range of anti-biofilm properties. Evidence is also presented to show that extracellular nucleases can disperse preformed microbial biofilms on different substrata. Microbial extracellular nucleases therefore appear to be a rich and unexplored source of anti-biofilm enzymes.

The future success of the cell therapy sector depends largely upon the ability to produce large numbers of high quality and efficacious cells. Within the production of living cells, the product and process become inseparable, thus, the selection and implementation of a large scale commercial production process becomes crucial to the success of the therapy. Traditional cell therapy lifecycles from the discovery phase through to patient administration utilise various unit operations, necessary for expansion, primary recovery, purification and final formulation. It is the induction of variation within cell populations caused whilst processed within these processes that can potentially render the product ineffective and unusable, creating major concern within the healthcare sector. This body of work casts focus upon the recovery of human cells intended for therapeutic use by centrifugation and sediment resuspension, a commonly adopted process which facilitates primary recovery, cell washing and purification processes, as well as the concentration of cells to a final administration specification. With the creation of ultra scale-down techniques coupled with a Design of Experiments approach, the impact of key unit associated parameters such as centrifugation spin time and intensity, along with the technique adopted for sediment resuspension has been characterised for three different human cell lines intended for therapy. A potential developmental roadmap has been created for the production and evaluation of cells for therapies. A range of cell quality attributes were evaluated forming the characterisation criteria, encompassing the analysis of structural integrity, surface marker profiling, cell death induction, cell proliferation capacity, a change in cell size and morphology, the induction of cell signalling in the form of cytokine release and crucially, the potency of the therapy. This facilitated an indepth view into the response of cells when processed, exemplifying the somewhat robust capacity of human cells. This work has aided the formation of a mechanistic hypothesis concerning cell damage as a consequence of the process. Traditional centrifugation protocols operate at the lower extremes of the design space investigated, so characterisation can inform as to the limits of manufacturing at large scale. Critical process effects affecting the recovery of viable cells included (i) the cell preparation with long hold times leading to greater loss during centrifugation and (ii) the method of cell pellet resuspension with the use of few passes at high velocities resulting in greater loss than many passes at low velocity. The mechanisms behind these and other process induced damage are explored.

Using shaken microwell and stirred miniature bioreactors in early stage bioprocess development requires understanding such devices from an engineering perspective. One particular area which has not to this date been fully explored is the suspension of solid particles in multi-phase operations in such devices. In order to broaden our understanding of this phenomena, a novel correlation is described for predicting the minimum shaking frequency required to suspend at least 90% of solid particles (0.81 gml−1 < ρp < 1.33 gml−1) present in a bioprocess involving shaken microwell operations. It was found that predictions made by this correlation are in good general agreement with empirical data with an uncertainty figure of just ±17%. Suitability of Zwietering’s correlation for predicting the minimum stirring frequency required to suspend solid particles in a miniature and a laboratory scale stirred tank was examined by comparing its predictions to empirical data. It was found that the correlation overestimates the required agitation rate, sometimes by as much as ≈ 70%, for the miniature bioreator. Flow inside the two stirred tank bioreactors were also numerically simulated and the results found to be in general agreement with the empirical data.

Lipoprotein VLPs, known also as lipid-envelope VLPs, are expressed intracellularly in yeast, localised on the host endoplasmic reticulum (ER). The main challenge in production lies in the complexity of the recovery and purification process. Using the Hepatitis B surface antigen (HBsAg) as the VLP model, strategies for exploiting the full potential of primary recovery operations to raise the overall process efficiency and throughput were investigated. Detergent-mediated liberation of HBsAg from the host endoplasmic reticulum is a critical step which influences product quality and yield and defines the characteristics of the resulting process stream. Screening studies established that Triton X-100 gave superior performance if concentrations were maintained within the range of 0.2% v/v to 0.5% v/v. Concentrations above this threshold led to HBsAg delipidation and loss of antigenicity. Increased levels of co-liberated host protein and of lipid contamination which resulted in poorer ultrafiltration performance were also observed. PEG and ammonium sulphate precipitations were investigated to reduce the level of host protein and lipid contaminants. In each case, although significant product enrichment was achieved, product loss of up to 50% was incurred owing to the difficulty of PEG precipitate recovery and resolubilisation and the lack of selectivity of the ammonium sulphate agent. A novel selective product recovery methodology was developed in which an additional centrifugation step was introduced post-cell disruption but prior to the addition of detergent. HBsAg associated with the ER was pelleted allowing bulk cytosolic contaminants in the supernatant to be eliminated. This approach was further enhanced by the use of moderate homogenisation pressures in the region of 400 bar. Recovery of active HBsAg improved by ~20% under lower shear conditions and a further ~40% reduction in the level of host lipid contaminants was observed. The benefits of the proposed process on a downstream chromatography step include better product capture and improved step yield. The scalability of the selective recovery method relies on the dewatering and clarification performance. A CARR tubular type centrifuge was identified to be the most suitable equipment type for this purpose. Ultra scale-down centrifugation experiments were employed to predict suitable operating conditions for a pilot-scale process which was subsequently carried out and successfully verified that the selective recovery methodology remained feasible upon scale-up. Overall, this study has demonstrated how the adoption of a rational approach to the design of primary purification operations can lead to significant improvements of downstream purification performance.

The carbon-carbon bond forming ability of transketolase (TK), along with its broad substrate specificity, makes it very attractive as a biocatalyst in industrial organic synthesis. Through the production of saturation mutagenesis libraries focused on individual active site residues, several variants of TK have been discovered with enhanced activities on non-natural substrates. We have used computational and bioinformatics tools to increase our understanding of TK and to guide engineering of the enzyme for further improvements in activity. Computational automated docking is a powerful technique with the potential to identify transient structures along an enzyme reaction pathway that are difficult to obtain by experimental structure determination. We have used the AutoDock algorithm to dock a series of known ketol donor and aldehyde acceptor substrates into the active site of E. coli TK, both in the presence and the absence of reactive intermediates. Comparison of docked conformations with available crystal structure complexes allows us to propose a more complete mechanism at a level of detail not currently possible by experimental structure determination alone. Statistical coupling analysis (SCA) utilises evolutionary sequence data present within multiple sequence alignments to identify energetically coupled networks of residues within protein structures. Using this technique we have identified several coupled networks within the TK enzyme which we have targeted for mutagenesis in multiple mutant variant libraries. Screening of these libraries for increased activity on the non-natural substrate propionaldehyde (PA) has identified combinations of mutations that act synergistically on enzyme activity. Notably, a double variant has been discovered with a 20-fold improvement in kcat relative to wild type on the PA reaction, this is higher than any other TK variant discovered to date.

Monoliths are an alternative stationary phase format to conventional particle based media for large biomolecules. Conventional resins suffer from limited capacities and flow rates when used for viruses, virus-like particles (VLP) and other nanoplex materials. Monoliths provide an open pore structure to improve pressure drops and mass transport via convective flow. The challenging capture of a VLP from clarified yeast homogenate was used to develop a new monolith separation which found hydrophobic interaction based separation using a hydroxyl derivatised monolith had the best performance. The monolith was then compared to a known beaded resin method, where the dynamic binding capacity increased three-fold for the monolith with 90% recovery of the VLP. Confocal microscopy was used to visualise lipid contaminants, deriving from the homogenised yeast. The lipid formed a layer on top of the column, even after column regeneration, resulting in increasing pressure drops over a number of cycles. Removal of 70% of the lipid pre-column by Amberlite/XAD-4 beads significantly reduced the fouling process. Applying a reduced lipid feed versus an untreated feed further increased the dynamic binding capacity of the monolith from 0.11 mg/mL column to 0.25mg/mL column. Control of chromatographic conditions can impact the product concentration during elution. Critical parameters which influenced the concentration of measureable VLP eluted included column contact time, salt concentration in mobile phase, and inclusion of lipid. The parameters were co-dependant with a crude lipid feed loaded at low salt and extra wash time of 40 minutes causing the largest decrease of 40%. Reducing the time of contact between the column and the VLP helped reduce such adverse effects. Increasing the flow rate in the column had no effect on the elution profile with crude or reduced lipid feeds. This informs process development strategies for the future use of monoliths in vaccine bioprocessing.

The efficient differentiation of retinal cells from human pluripotent stem cells remains a major challenge for the development of successful and cost-effective cellular therapies for various forms of blindness. Current differentiation strategies rely upon exposing pluripotent stem cells to soluble growth factors which play key roles during early development (such as DKK-1, Noggin and IGF-1) at 20% oxygen (O2). This O2 tension is however, considerably higher than O2 levels during organogenesis and may impair the differentiation process. In this study, we examined the effect of mimicking the physiological O2 tension (2%) on the generation of retinal progenitor cells (RPCs) from human induced pluripotent stem cells (iPSCs) and human embryonic stem cells (hESCs). Both cell types were induced to differentiate into RPCs at 20% and 2% O2. After three days in suspension culture as embryoid bodies (EBs), 2% O2 caused the activation of hypoxia inducible factor (HIF) responsive genes VEGF and LDHA and was accompanied by elevated expression levels of the early eye field genes Six3 and Lhx2. 21 days after plating EBs in an adherent culture, we observed more RPCs co-expressing Pax6 and Chx10 at 2% O2. qPCR analysis confirmed that lowering O2 tension had caused a rise in the expression of both genes compared to 20% O₂. Our results indicate that mimicking physiological O2 is a favorable condition for the efficient generation of RPCs from both hiPSCs and hESCs.

Cancer, Alzheimer’s and Parkinson’s are a few of the very complicated diseases facing our ageing population. While small chemically derived pharmaceuticals have a more major role in curing these illnesses, it is hoped that a range of new larger biopharmaceuticals, namely antibody based therapies, could be effective due to their ability to target sites for treatment. This next generation of complex proteins, such as fusions of macromolecules and antibodies are likely to be sensitive to the engineering environment. Hence new insight is needed on where in a bioprocess this damage may occur and how this can be avoided. This thesis focuses on the stability, in a bioprocess engineering environment, of antibody fusion proteins used to treat colorectal cancer with a novel drug delivery system called Antibody Directed Enzyme Prodrug Therapy (ADEPT). An ultra scale-down shear device was used to mimic bioprocess conditions using millilitre quantities of process solution. This allowed protein characterisation and identified new engineering parameters which resulted in the reduction of yield impurities. The amount of intact fusion protein was measured over time during a constant shear field. It was found that the presence of an air/liquid interface exacerbated the rate of fusion protein degradation. First order rate constants were used to characterise the fusion protein degradation for defined bioprocess environments. The use of a surface active agent was shown to protect the protein even at high shear rates and air/liquid interfaces. However, when this surface active agent was added to the large scale production, no discernable effects on the yield were observed. Nevertheless the results showed that the surfactant did improve the proportion of intact protein giving a less heterogeneous product after FPLC purification. The research presented in this thesis shows that detailed characterisation in a ultra scale-down shear device of millilitre quantities of dilute protein solutions can be used to identify problems, such as increased impurities levels in the final product generated during the large scale production.