Tools that allow cost-effective screening of the susceptibility of cell lines to operating conditions which may apply during full-scale processing are central to the rapid development of robust processes for cell-based therapies. In this study, an ultra scale-down (USD) device has been developed for the characterization of the response of human cell lines to membrane-based processing, using just a small quantity of cells that is often all that is available at the early discovery stage. Key operating conditions investigated were cross-membrane flow rate, cell age prior to processing and cell concentration (viscosity). The impact was evaluated by cell damage on completion of membrane processing as assessed by trypan blue exclusion and release of intracellular lactate dehydrogenase (LDH). Similar insight was gained from both methods and this allowed the extension of the use of the LDH measurements to examine cell damage as it occurs during processing by a combination of LDH appearance in the permeate and mass balancing of the overall operation. The main cell line studied was a clinically relevant human fibroblast. As expected, increased shear rates led to significant increases in rate and extent of cell damage. Cells aged (21°C hold for 24 hours) before processing led to a doubling of the extent of damage. Increased cell concentration from 1x106 to 100x106 cells mL-1 gave no change in the proportion of cells damaged. Preliminary studies showed that increased shear stress also led to morphological changes and the appearance of apoptotic cells post-processing. Two other human cell lines were also tested briefly for cell damage; a neural stem cell line and a prostate cancer cell line. These appear to be less robust than the fibroblasts with, for example ~0%, ~18% and ~42% damage being observed at the lowest shear stress (~44 Pa) conditions for fibroblasts, prostate cells and neural stem cells respectively. The effects of increasing shear rate, age of cells or concentration varied for each of the cell lines studied. Overall, this work suggests how membrane processing may be used for the recovery of human cells for therapy and how USD studies can speed the route to manufacture.

Mesenchymal stem cells (MSCs) in many adult tissues provide cell sources to sustain tissue growth and/or repair in vivo, yet MSCs are mainly studied based on their in vitro characteristics. One emerging population of such MSCs are from dental pulp mesenchymal tissue, termed dental pulp stem cells (DPSCs). For instance, the continuously growing rodent incisor model has recently provided the first in vivo evidence that the in vivo identities of MSCs are of multiple origins including from perivascular niches. However, little is known about the molecular mechanisms underlying MSC response to injury in vivo, including that in the context of tooth repair. We therefore compared processes involved in recruiting stem cells during injury repair, particularly cell migration of pulp cells isolated from distinct anatomical locations. We found pulp cells from the region containing putative stem cells showed the highest migration capacity and their migration ability could be stimulated by activating Wnt activity in vitro. Furthermore, following in vivo tooth injury on transgenic mice, Wnt/β-catenin was also found up-regulated close to the injury site, possibly regulating injury repair via promoting perivascular-associated stem cell accumulation in close proximity to the injury site. In addition, analysis of a novel injury experimental model- the incisor tip, that undergoes constant attrition/repair through natural feeding, confirmed that this rapid incisal tip repair is also facilitated by perivascular stem cells, similar to other experimental injury models, but at a far more striking level. Thus, future work will utilise this novel model to investigate regulatory mechanisms including Wnt signalling in mediating mesenchymal tissue repair. Taken together, we demonstrated that the Wnt pathway may play a crucial role in regulating MSCs during incisor injury repair in vitro and in vivo. Also, the naturally existing “incisal tip niche” is potentially a unique model for new insights into mesenchymal tissue repair in vivo.

The pharmaceutical industry is under increasing pressure to deliver new medicines quickly and cost effectively; traditional small molecule product pipelines have dried up and companies are increasingly investing into biopharmaceuticals. To date, the most successful biopharmaceuticals have been monoclonal antibodies. The ability to construct common manufacturing platforms for a range of antibody products has underpinned this interest. Antibodies are most often produced as heterologous proteins at large scale in stirred tank reactors. However, at manufacturing scale there is limited opportunity to undertake process development and optimisation. If a manufacturing process can be ‘scaled down’ experiments could be carried out at much greater throughput and occur in parallel throughout the entire product lifecycle. In creating a small scale model, the fundamental challenge lies in accurately recreating the engineering environment experienced at large scale in order to yield process relevant data. In this thesis a miniature, single use, 24-well shaken bioreactor platform was investigated as a small scale cell culture device. This plate format can operate either using direct (REG plate) or headspace sparging (PERC plate) i.e. with either the presence or absence of a dispersed gas phase. Initial work involved the experimental and theoretical characterisation of the novel, miniature bioreactor (7 mL) and the conventional stirred bioreactors (1.5 L), themselves mimics of pilot scale GSK cell culture processes. Under typical operating conditions in the miniature bioreactor, measured mixing times were 0.8 – 13 s and apparent kLa values in the range 5 – 50 hr-1. Based on these findings, cell culture kinetics were investigated. A methodology for consistent, parallel cell cultures was first established and then used to determine the impact of the dispersed gas phase on culture kinetics of a model CHO cell line. Cultures performed with head space aeration showed the highest viable cell density (15.2 × 106 cells mL-1) and antibody titre (1.58 g L-1). Final cell density in the PERC plate was nearly 40 % greater than shake flask cultures due to the improved control of process conditions. In contrast, cultures performed with direct gas sparging showed a 25 – 45% reduction in cell growth and 40 – 70 % reduction in antibody titre. The platform nature of the system was confirmed with similar findings obtained using a second antibody and cell line cultured under different conditions. The miniature bioreactor was then investigated for use as an early stage, cell line selection tool. A strong positive correlation between PERC and shake flask data was found (0.88), indicating the suitability of the platform for this application. In contrast, selection results in the REG plate format differed notably, highlighting the fact that the presence of a dispersed gas phase can significantly alter cell culture kinetics; and potentially cell line selection. A panel of four CHO clones was then investigated alongside bench scale bioreactors, operating at matched mixing times; the REG plate format provided the most comparable match in terms of cell growth and product titre. Primary recovery studies investigated use of a small scale depth filtration tool to analyse material generated previously with regards to ease of processing. Data showed that cells cultures in the presence of a dispersed gas phase yielded the most accurate prediction of primary recovery data. Subsequently, detailed product quality analysis confirmed consistent product quality attributes across the different cell culture formats. In summary, this work shows the utility of miniature bioreactor systems for high throughput strain selection under process relevant conditions.

The aim of this project is to study the use of Staphylococcus carnosus as an alternative host in whole-cell biocatalysis, using the NADPH-dependent conversion of cyclic ketones by cyclohexanone monooxygenase (CHMO) as the model system. The use of whole cells in industrial biocatalysis has been actively researched due to their promise as biological factories that can perform complex reactions in environments where the use of isolated enzymes would not be feasible. However, the majority of whole-cell processes rely on a small range of organisms that, while been extensively characterized and easily engineering to express enzyme biocatalysts, have some inherent limitations that hinder their application to many industrial processes. Thus. There is a need to find novel microorganisms that can fill in the shortcoming of those conventional strains. In this project, a flexible shuttle vector was constructed to allow for the cloning and expression of CHMO in both E. coli and S. carnosus by using a dual promoter system in which one of the promoters could be replaced. Higher levels of CHMO expression were achieved in E. coli strains cloned with the shuttle vector compared to an expression vector used in previous studies. No biocatalyst expression was detected in S. carnosus. The shuttle vector was subsequently modified to allow for the quantitative characterization of different synthetic promoters based on the gemone of S. carnosus. A novel in-silico promoter selection methodology based on the codon bias was developped to allow for the rational selection of potential genomics promoters. Two synthetic promoters based on sequences upstream of ribosomal proteins rplK and rplJ were subsequently shown to allow for protein expression in both E. coli and S. carnosus. The stronger of these two promoter was inserted into the CHMO expression vector, but the resulting construct did not manage to express detectable levels of the biocatalyst in S. carnosus. It was thus concluded that S. carnosus was not a suitable host for biocatalysis with CHMO, and further studies with other enzymes would have to be conducted to access its suitability as a general biocatalytic host.

Commercial pressures to reduce the costs and timelines involved in bringing new medicines to market are driving investment in methods for high throughput bioprocess development. The establishment of mammalian cell culture processes for therapeutic antibody production is an area of particular interest. This is an experimentally intensive procedure initially involving evaluation of a large number of clones followed by optimisation and scale-up of cell culture conditions. In this thesis a high throughput microwell platform is established for use in early stage cell line selection and cell culture process development. It was first demonstrated that shaken 24 SRW microwells were suitable for batch suspension culture of a commercially available CHO cell line, provided that appropriate culture conditions were selected. However, a high rate of evaporation from the microwells was identified as a potential limiting factor. Further work with an industrial GS-CHO cell line led to the development of a fed-batch method. This used a combination of diluted liquid feeds and a ‘sandwich lid’ to counteract microwell liquid losses by water replacement and reduction of evaporation to negligible levels. This led to comparable cell growth and product formation kinetics as well as similar metabolite utilization kinetics in 24 SRW plates and conventional shake flasks. Engineering characterisation of 24 SRW, shake flask and laboratory scale (5L) stirred tank systems was subsequently performed, encompassing the oxygen mass transfer coefficient (kLa), mean energy dissipation rate (P/V), and liquid phase mixing time (tm). Mixing times in particular showed a strong dependence on the speed and diameter of orbital shaking while in general kLa values were sufficient (>1h-1) for oxygen transfer not to be rate limiting. Consequently it was suggested that matched liquid phase mixing times could be a suitable scale translation parameter for the types of small scale bioreactor formats used in early stage cell culture process development. It was subsequently demonstrated that at mixing times that promoted a homogeneous culture environment at each scale (tm = 5s) similar culture performance in all three bioreactor formats could be achieved using this engineering basis for two distinct cell culture processes involving different GS-CHO cell lines and alternative feeding methodologies. Peak viable cell densities achieved in stirred tank, shake flask and 24 SRW formats were 5.9, 6.7 and 6.4 x 106 cells mL-1 respectively, while antibody titres were 1.23, 0.81 and 0.88 g L-1. The higher IgG concentration in the stirred tank was attributed to tighter on-line pH control. The utilization rates for key metabolites were also closely matched. Finally the industrial relevance of this methodology was demonstrated through the successful parallel fed-batch cultivation of more than 50 GS-CHO cell lines. The rank order of cells based on product titre agreed closely with existing development procedures using shake flasks and static microwell plates. The microwell cell culture process presented here thus offers the potential for a considerable increase in throughput during the early stages of biopharmaceutical product development.

Embryonic stem (ES) cells hold great promise as a new paradigm in medicine. The first clinical trials using ES cells for regenerative therapies in humans have recently begun. Despite this, much of the basic research into the mechanical properties and the response of ES cells to the extra-cellular force applied during their routine culture and bioprocessing remains to be characterised. Atomic force microscopy (AFM) was used to apply specific force to OCT4-GFP and E14Tg2A mES cells. The formation of mechanically-induced blebs was observed in round, pluripotent cells. Blebs occur during the detachment of the plasma membrane from the main body of the cell. They are known to occur spontaneously and were found to form upon the application of ≥5nN force. Cytoskeletal studies investigating the distribution of actin, tubulin and pERM confirmed that undifferentiated mES cells have a less well-developed cytoskeleton, leaving them more susceptible to force-induced damage than differentiating or differentiated cells. Shear forces are routinely exerted on a cell population throughout passaging, expansion and differentiation. Therefore, it is essential to understand the effect of such forces as their influence may have dramatic consequences on ES cell viability, integrity and fate decisions, thus impacting their utility for regenerative therapies. For this reason, exposure to transient shear was also investigated. Shef-3 and Shef-6 hES cell lines were adapted to enzymatic passaging, facilitating the generation of sufficient cells which can tolerate single cell dissociation, to study the impact of cell culture induced shear forces. Single-cell suspensions ensured all hES cells were exposed to equal forces. Transient shear exposure had no noticeable impact on cell viability or on the induction of apoptosis. Significant differences in fold expansion rates (P≤0.005) and gene expression of the pluripotency markers OCT4 and Nanog were detected (P≤0.005). Assessment by qPCR revealed that shear exposure can influence cell fate decisions. Current thinking indicates that for regenerative therapies to be successful, ES cells will need to be differentiated into the required cell types prior to transplantation, thus confirming the significance of the findings presented herein.

The concept of de novo metabolic engineering through novel synthetic pathways offers new directions for multi-step enzymatic synthesis of complex molecules. This has been complemented by recent progress in performing enzymatic reactions using immobilised enzyme microreactors (IEMR). This work is concerned with the construction of de novo designed enzyme pathways in a microreactor synthesising a chiral molecule. An interesting compound, commonly used as the building block in several pharmaceutical syntheses, is a single diastereoisomer of 2-amino-1,3,4-butanetriol (ABT). This chiral amino alcohol can be synthesised from simple achiral substrates using two enzymes, transketolase (TK) and ω-transaminase (TAm). This project involves the design and the development of an IEMR using His6-tagged TK and TAm immobilised onto Ni-NTA agarose beads and packed into tubes to enable multi-step enzyme reactions. The IEMR was first characterised based on the operational and storage stability. Furthermore, kinetic parameters of both enzymes were determined using single IEMRs evaluated by a kinetic model developed for packed bed reactors. For the multi-step enzyme reaction, two model systems were investigated. The first model investigated was the dual TK (pQR 791)-TAm (pQR 801) reaction. With initial 60 mM (HPA and GA each) and 6 mM (MBA) substrate concentration mixture, the coupled reaction reached approximately 83% conversion in 20 minutes at the lowest flow rate. On the other hand, the second model reaction comprises of three sequential enzyme reaction, TAm (pQR 1021)-TK (pQR 791)-TAm (pQR 1021). A 6% yield of ABT was produced from initial substrate mixture of 100 mM serine and 40 mM GA at flow rate of 0.5 μL/min. Further considerations to improve the system would allow for better yield of the target product and potentially make this IEMR system a powerful tool for construction and evaluation of de novo pathways as well as for rapid determination of various enzymes kinetics.

The increasingly stringent regulatory requirements on the quality and cost-effectiveness of biopharmaceuticals as well as the escalating upstream titre are placing considerable pressure on downstream bioprocessing. In particular, batch chromatography is fast becoming the process bottleneck due to its limited capacity and expensive operation. Recent developments in multi-column counter-current chromatography (MCC) for the separation of biomolecules have demonstrated the potential to overcome these limitations. However, the uptake of MCC processes by the biopharmaceutical industry is hindered by system complexity and the liable validation difficulties. This research provides a solution to the above challenges through the development of an integrated experimental and modelling design approach and a process analytical technology (PAT)-compliant controller. These tools were exemplified using protein A chromatography, because of its key role in purification and the associated adsorbent cost. Sequential multi-column chromatography (SMCC) was the MCC process of choice as it is universal to a wide range of chromatographic principles. The design approach was applied to both batch chromatography and SMCC for fair comparison of the two processes at their optimal performance. Significantly higher productivities (up to two-fold) were found for SMCC than for batch chromatography with varying improvements in each case. The integrated use of experimentation and modelling enabled sufficient process understanding to be gained quickly and efficiently, and thus providing a dramatic reduction in the time and costs to find the optimal design (up to an estimate of five-fold). The SMCC controller ensures robust and optimal process operation by controlling a set of characteristic points that were monitored within the system loop for various zones. Such a controller is simple to implement and offers quick actions (within 15 cycles for loading due to the multi-column effect and within 2 cycles for the other zones) against possible process disturbances and system uncertainties at system start-up.

Expanded bed adsorption (EBA) combines clarification, concentration and purification into a single processing step reducing processing time and increasing productivity. Much work has been conducted on model proteins but little attention has been paid to the emerging issues of tailoring the operational variables of matrix size, operating flowrate and ligand type so as to maximise process outputs. The aims of this study were to investigate the effects of matrix properties and operating parameters on breakthrough behaviour and productivity. Second generation matrices designed to operate at high flowrates were also examined. Operating at low velocities, beds formed from smaller particles had a shallower breakthrough than beds formed from large particles but the latter were more productive. At the higher velocities typically utilised in EBA, the behaviour of beds formed from either small or large particles was comparable in terms of breakthrough with the beds made up of small particles being slightly more productive than those containing large particles. A prototype 2nd generation EBA matrix consisting of a multi-modal ligand that could operate at high conductivity levels typical of fermentation broths was also examined. It was found that the productivity could be increased 5-fold as compared with commercially available matrices, with no loss of yield or purity. Conventional breakthrough curve analysis is not applicable in situations of variable particle size and operating velocity. Results using a novel dimensionless group to facilitate comparison between unlike systems is presented. The thesis presents an analysis of the relative productivity of a range of systems and highlights the gains in terms of productivity to be achieved with the development of 2nd generation EBA multi-modal adsorbents. The thesis concludes with an analysis of the commercial and regulatory aspects of EBA.

Japanese Encephalitis (JE) is the most common form of viral encephalitis in the world, caused by the Japanese Encephalitis virus (JEV), it is responsible for around 10,000 deaths a year whilst many more are left with long term neurological sequelae and disability. This work sought to use small-scale development techniques alongside high-throughput methodologies to explore and develop selected processing techniques. Formaldehyde inactivation of JEV was characterised and optimised through the use of Design of Experiments screening techniques where temperature, time and formaldehyde concentration were found to be key factors in antigen loss. Glycine and to a lesser extent sorbitol were found to have positive effects as stabilisers during inactivation at different stages of the process. Four anion exchange resins were screened at micro-scale, with the help of an ELISA method evaluated for high-throughput screening, for their potential to replace sucrose gradient purification as the principle purification step of the process. Although Q Sepharose FF was eventually chosen for scale-up studies, the transition of method and recovery rates from batch bind micro-plate studies to 1 mL column scale proved difficult. Yet it was observed that pre-treatment of feed with formaldehyde and glycine could increase JEV antigen recovery rates in flow-through mode chromatography, thought to be due to enhanced stability of the virus particles.