This thesis details the development of a system of microfabricated devices for the adherent culture of stem cells. The multipotency and self renewal of stem cells make them a potentially abundant source of valuable human cells, for both drug screening and regenerative medicine. However, processing stem cells is challenging due to the complexity of whole cell products, the number of process parameters, and the typical use of adherent culture. It is hypothesised that a microfabricated adherent culture system could facilitate process development with minimal use of resources. Furthermore, microfluidic systems offer advantages in spatial and temporal control over the microenvironment that would benefit process development. An existing prototype culture system is critically evaluated by: assessing the design, modelling fluid flow and dissolved oxygen, and successfully co-culturing human embryonic stem cells, on inactivated mouse embryonic fibroblasts, under perfused conditions. The utilisation of reversible seals facilitates the use of standard tissue-culture polystyrene culture surfaces and manual seeding techniques. The evaluation of the prototype system is used to inform improvements to the design, making it easier to use, increasing the robustness, allowing monitoring of whole culture chambers by microscopy, and improving control over mean pericellular dissolved oxygen. Modelling shows the improved culture system also achieves more uniform distribution of both pericellular dissolved oxygen and fluid velocity. The improved culture system shows similar mouse embryonic stem cell seeding behaviour to tissue culture flasks, but, with medium perfused at 300 μl.h 1, mouse embryonic stem cells reach full confluency in less than 48 h, compared with 72 hours for cells maintained statically in flasks. There is also inconclusive data suggesting that the growth rate is limited by pericellular dissolved oxygen and is, therefore, increased and made more uniform by the inclusion of a gas permeable lid system. The reliability, ease of use, comparability with traditional culture systems, and control over process parameters of the improved system should make it a useful tool for stem cell process development.

Autologous hMSCs are a promising therapeutic candidate for ischemic tissue injury such as that resulting from myocardial infarction. In recognition of the poor functional characterisation of hMSCs used in clinic and lack of priming methods to enhance cell function, the work presented in this thesis describes several ex vivo strategies designed to assess and enhance hMSC responses. Firstly, growth kinetics of hMSCs isolated from 10 healthy donors were mapped over 5 passages and inter-donor variability was determined. A process operating window was identified between passage 2 and passage 3 where different donors displayed no PDL and cell density differences. Expansion potential at the end of passage 2 was 17.29x106 ± 5.19x106 and at the end of passage 3 was 50.40x106 ± 8.59x106. Next, functional effects of sJag1 and sDll4 pre-conditioning on hMSCs attachment and migration responses were studied. The effect of low oxygen was also assessed, to more closely reflect the physiologic environment. In low oxygen conditions, sJag1 pre-conditioning significantly increased hMSC attachment to fibronectin after just 20 minutes compared with unconditioned cells. However, in ambient oxygen, the stimulatory effect was not seen until 40 minutes attachment time. sJag1 pre-conditioning also significantly enhanced hMSC chemotaxis through fibronectin in response to SDF-1α in low oxygen. Comparable effects were not detected when using a type I collagen substrate. Finally, sJag1 also enhanced the production of angiogenic factors VEGF-A165 and HGF by hMSCs on fibronectin. We assessed the effects of sJag1 and sDll4 on vascular support capacity of hMSCs using in vitro co-culture assays with endothelial cells. Pre-conditioning hMSCs with sJag1 resulted in significantly increased branching within co-culture networks and enhanced vessel-like network formation and structural maintenance over 96 hours in the low oxygen environment. Finally, the requirement for functional Nrp1 in MSCs was studied using Nrp1 cytoKO mice. Although cell attachment was impaired in MSCs lacking fully functional Nrp1, sJag1 was still able to enhance cell attachment. Since hMSCs are widely used in clinical trials with limited positive outcomes, pre-conditioning with sJag1 may be a tool for priming immature cells to improve retention and therapeutic efficacy in ischemic tissues.

The development of new biocatalytic processes is hindered by the number of factors that must be investigated and optimised in order to create a robust and reproducible process, particularly where a novel enzyme is involved. It is therefore advantageous to perform process development experiments at micro scale, in order to reduce the material requirement for experimentation and increase experimental throughput by parallelisation. The initial focus of the thesis is on the design of a microfluidic tangential flow filter to test downstream processing conditions. The device was designed for reversible clamp sealing, allowing the simple integration of different filtration membranes, and the seal achieved with the device was shown to be robust up to internal pressures of 100 psi. The filtration device was applied to the recovery of L-erythrulose (ERY) from a synthesis reaction performed using transketolase (TK), where it was demonstrated that the enzyme could be fully retained using a commercially-available membrane, while ERY was able to permeate the membrane freely. The filtration device was joined in-line to the output of a T-junction reactor with a staggered herringbone mixer, used to perform the synthesis reaction. The filter was capable of continuously separating the ERY from the lysate mixture exiting the reactor over the course of several hours, producing 3.6 mg h-1 of pure ERY. A novel multi-input reactor (MIR) is also demonstrated for the purpose of intensifying ERY output, designed to overcome the effect of substrate inhibition on the TK enzyme by mimicking a fed-batch reactor. Feeding strategies were designed for the conversion of various concentrations of the less inhibiting substrate hydroxypyruvate (HPA) and tested in the MIR. A 4.5-fold increase in output concentration and a 5-fold increase in throughput were achieved compared with the single input reactor used in previous experiments. However, conversion in the MIR was reduced at higher concentrations, suggesting that the reaction in the MIR was being inhibited by the evolution of carbon dioxide.

The pharmaceutical industry is suffering from declining R&D productivity and yet biopharmaceutical firms have been attracting increasing venture capital investment. Effective R&D portfolio management can deliver above average returns under increasing costs of drug development and the high risk of clinical trial failure. This points to the need for advanced decisional tools that facilitate decision-making in R&D portfolio management by efficiently identifying optimal solutions while accounting for resource constraints such as budgets and uncertainties such as attrition rates. This thesis presents the development of such tools and their application to typical industrial portfolio management scenarios. A drug development lifecycle cost model was designed to simulate the clinical and non-clinical activities in the drug development process from the pre-clinical stage through to market approval. The model was formulated using activity-based object-oriented programming that allows the activity-specific information to be collected and summarized. The model provides the decision-maker with the ability to forecast future cash flows and their distribution across clinical trial, manufacturing, and process development activities. The evaluation model was applied to case studies to analyse the non-clinical budgets needed at each phase of development for process development and manufacturing to ensure a market success each year. These cost benchmarking case studies focused on distinct product categories, namely pharmaceutical, biopharmaceutical, and cell therapy products, under different attrition rates. A stochastic optimization tool was built that extended the drug development lifecycle cost evaluation model and linked it to combinatorial optimization algorithms to support biopharmaceutical portfolio management decision-making. The tool made use of the Monte Carlo simulation technique to capture the impact of uncertainties inherent in the drug development process. Dynamic simulation mechanisms were designed to model the progression of activities and allocation of resources. A bespoke multi-objective evolutionary algorithm was developed to locate optimal portfolio management solutions from a large decision space of possible permutations. The functionality of the tool was demonstrated using case studies with various budget and capacity constraints. Analysis of the optimization results highlighted the cash flow breakdowns across both activity categories and development stages. This work contributed to the effort of providing quantitative support to portfolio management decision-making and illustrated the benefits of combining cost evaluation with portfolio optimization to enhance process understanding and achieve better performance.

The evaluation and optimisation of microalgae cultivation process for biomass, lipid and high value chemicals production requires experimental investigation of several interacting variables. This thesis addresses the development of a range of small-scale photobioreactor technologies and shows how they can be applied for rapid, early stage evaluation and scale-up of microalgae cultivation processes. In particular, the work focuses on the engineering evaluation of a novel shaken miniature photobioreactor (mPBr) and a single-use photobioreactor (SUPBr) that can be adapted for both phototrophic and heterotrophic cultivation. A prototype twin-well mPBr was initially designed and fabricated with light provided from cool white light emitting diodes (LED). This was scaled-out to a 24-well mPBr system (4 mL working volume) on a novel shaken platform. High power warm white LEDs provided a maximum light intensity of 2000 µmolm‾²s‾¹. In both systems, surface aeration (via a semipermeable membrane) and mixing were provided by orbital shaking. Real-time control of temperature, relative humidity and CO2 levels was achieved via incubator level control. Amongst the tested geometries of the mPBr, round base and pyramid base gave the best performance. The mass transfer coefficient (kLa) values in the 24-well were measured between 20 – 88 h‾¹ and visual observation of fluid hydrodynamics showed an increase in total surface area with increased shaking frequency. Negligible evaporation was observed at 90% relative humidity for light intensity of < 400 µmolm‾²s‾¹ and at 32 °C, while light intensity variation across the platform is in the range ± 20 µmolm‾²s‾¹. Evaluation of phototrophic culture kinetics of Chlorella sorokiniana in both mPBr designs showed good reproducibility between wells. The best culture performance occurred at 380 µmolm‾²s‾¹, 300 rpm and 5% CO2, where final biomass concentration and total lipid concentration achieved were 9 ± 0.2 gL‾¹ and 55% w/w respectively. The SUPBr comprised a transparent polymeric CultiBagTM operated on the illuminated rotary shaken platform described above. Mixing time values were determined over the range 40 - 220 rpm and were generally less than 40 s. Hydrodynamic studies showed three distinct flow regimes at various shaking frequencies: in-phase, transitional and out-of-phase. Under optimal flow regime, the highest cell concentrations achieved was 6.7 gL‾¹ ± 0.3. Doubling the total working volume resulted in 35 - 40% reduction in biomass concentration due to an increase in the light path length. Phototrophic scale-up criteria from mPBr to SUPBr was successfully achieved based on light–path length and kLa values. Comparison of final biomass concentrations showed similar performance of 6 ± 0.2 gL‾¹ and comparable total lipid production of 25 – 30% by weight at a light intensity of 180 ± 20 µmolm‾²s‾¹. Furthermore, application of the shaken 24-well system for heterotrophic cultivation of microalgae and scale-up to a 7.5 L stirred tank bioreactor was also shown. Cells were cultured in 24 parallel wells, shake flasks and a 7.5 L bioreactor with working volumes of 4 mL, 100 mL and 4000 mL respectively using glucose (10 gL‾¹) as the main carbon source. Constant k(L)a was chosen as scale-up criteria and the values range between 30 – 60 h‾¹. Final biomass concentrations showed good agreement in the range of 4.5 ± 0.5 gL‾¹ and total lipid production of 43 – 50% by weight for the three systems. Overall, the results show the utility of the mPBr and SUPBr technologies for the rapid evaluation and scale-up of both phototrophic and heterotrophic microalgae cultivation conditions.

S. erythraea is the main producing organism of erythromycin, an important broad spectrum antibiotic. Its yield is low because the carbon source utilisation, product kinetics and the regulatory pathways are not yet fully understood. This work provides a detailed characterisation of the carbon source metabolism of S. erythraea. The organism was cultured under various conditions and the growth and product kinetics investigated. In batch bioreactor fermentations, S. erythraea was cultured on both glucose and gluconate as sole carbon and energy sources. The culture was subsequently evaluated in bolus feed addition fermentations using phosphoenolpyruvate, oxaloacetate, propionate and methyl oleate with glucose as main carbon source. On both glucose and gluconate, erythromycin production depended on the nature of the carbon source and the growth rate. On glucose, growth was fast with erythromycin production commencing after the growth phase. On gluconate, growth was subdued and erythromycin production was growth-related. Two phases were distinguished for both carbon sources: 1) erythromycin synthesis and 2) precursor accumulation phase. Erythromycin production was enhanced by increased activity of the pentose phosphate and the anaplerotic pathway. Growth rate and the carbon source uptake rate were found to have a major effect on erythromycin production. The split in pathway activities at key branch points were found to be dependent on the growth rate and the rate of carbon source uptake. During bolus feed addition fermentation in shake flasks, erythromycin levels were increased by the addition of oxaloacetate, propionate and methyl oleate respectively. Then effect of metabolite supplementation on erythromycin levels depended on the phase during which feeding is done. In bioreactor culture, levels of erythromycin were enhanced following addition of PEP or methyl oleate. The pyruvate metabolite node was found to be flexible responding to supplementation and the nature of the sole carbon and energy source. Carbon allocation to pyruvate was 1.7 % and 5.4 % for growth on glucose and gluconate respectively. This metabolite node is, therefore, critical to improving the biosynthesis of erythromycin. However, the phosphoenolpyruvate node appears rigid and addition of PEP was excreted as 7-Orhamnosyl flaviolin (red pigment). The ratio of the rate of carbon source consumption and oxygen uptake is concluded to be a critical parameter in erythromycin biosynthesis.

Orthopaedic implants function to replace or support damaged or diseased bone. Due to a global rise in demand, there is a need to prolong the service life of these implants. The current work focuses on crystallised hydroxyapatite (HA)-coated titanium (Ti) implants. One specific problem during the annealing of as-deposited amorphous HA, to induce crystallisation, is the formation of unwanted titanium oxide (Ti-O) species at the HA/Ti interface that leads to HA layer disruption. This necessitates the introduction of a diffusion barrier layer at the HA/Ti interface. Another reason for implant failure is bone disintegration around the implant; therefore enhancing early bone cell growth on the implant may be beneficial. The current work investigates the radio-frequency magnetron co-sputtering (RF-McoS) method to coat silicon-substituted hydroxyapatite (Si:HA) on commercially pure Ti (CpTi) substrates, with and without a titanium nitride (TiN) barrier coating. Processed coatings were designated according to the sputtering targets used, namely HA/TiN, Si:HA, Si:HA/TiN and Si+O2:HA/TiN (i.e. reactively sputtered Si:HA/TiN in a mixed Argon, Ar, and oxygen, O2, atmosphere). The as-deposited amorphous HA layers were crystallised by annealing at 700oC for 2 hours and 4 hours in Ar/trace O2 (sample set 1) or for 2 hours in pure Ar (sample set 2). Coatings were assessed using a combination of complementary analytical bulk and near-surface techniques, combined with in vitro biological tests using an MG63 cell line to appraise the early stages of osseointegration. In both sample sets, coatings containing TiN were generally more effective at retarding the development of rutile (TiO2). Coatings annealed in trace levels of O2 (set 1) were more prone to delamination due to the development of Ti2N. Better process control was achieved following annealing in pure argon (sample set 2), with Si:HA/TiN showing significantly improved cell proliferation, whilst most closely resembling stoichiometric HA. Cell trials demonstrated both sample sets to be biocompatible, despite variations in coating morphologies, crystallinities and stoichiometries reflecting the issues of process control. Suggestions on improving the process control of these orthopaedic implant coatings are discussed.

In this thesis I present the magnetic haemofilter - a novel medical device designed to remove magnetic materials directly from a patient's bloodstream. The haemofilter is a high gradient magnetic separator incorporated into an extra-corporeal loop. A patient's blood is constantly circulated through the device, which magnetically captures and retains target agents while the rest of the blood returns to the patient unharmed. Clinical versions of the device were conceived, designed and modelled. Small scale versions, designed to mimic the performance of the clinical designs, were manufactured using 3D printing and tested in benchtop in vitro experiments. Abstract Many potential applications for the device are envisioned, and justified with an extensive literary review of magnetic labelling, the process of binding magnetic particles to specific targets to enable their separation. The device has significant potential as a platform technology enabling these varied applications. In this project, however, malaria was chosen as the primary application. Malaria infected red blood cells are paramagnetic, so their separation does not require magnetic labelling - the haemofilter simply exploits their naturally occurring magnetic properties. Abstract The haemofilter was tested using samples of malaria infected blood filtered at a variety of flow rates, with a reduction in parasitaemia observed in every experiment at throughputs orders of magnitude higher than any previously reported clinical magnetic haemofiltration device. The results demonstrate that even without further optimisation, the clinical version of the device could halve a child's parasitaemia in less than 90 minutes. The flow rates used in the experiments, and those that could be used in the clinical versions, are orders of magnitude higher than any previously reported clinical magnetic haemofilter. Experiments on samples donated by malaria patients demonstrated that no other blood components are affected by the process. Abstract A commercial evaluation of the haemofilter as a medical device to treat malaria was conducted. The market analysis showed a large potential total addressable market, segmented into three principal patient populations. The product development and manufacturing costs were estimated and shown to be reasonable, while a financial analysis showed that a margin could be earned while still saving customers money. A route-to-market and commercialisation strategy is presented. Abstract I conclude that the magnetic haemofilter has the potential to deliver significant clinical benefits to a wide variety of malaria patients, saving lives in serious cases, providing a treatment option for currently untreatable patients, and speeding up recovery in uncomplicated cases, while improving the efficacy and eliminating the side-effects of pharmaceutical drugs.

Naphthenic acids (NAs) and high molecular weight polycyclic aromatic hydrocarbons (HMW-PAHs) are both natural components of fossil fuels that cause a number of environmental issues. Using enrichment cultures derived from oil sands-process affected water (05PW), the biodegradation of three HMW-PAHs that differed in number of aromatic rings was investigated. All HMW-PAHs were degraded by up to 75% over 33 days. Bacterial communities were dominated by the genera Pseudomonasl Bacillus and Microbacterium and fungal communities by Cladosporium and Penicllium. Pseudomonas putida KT2440 has previously been shown to transform both NAs and other hydrocarbons and so was selected for co-culture biodegradation experiments with a NA-tolerant culture of Chlorella vulgaris Nr 211/12. Co-cultures demonstrated biodegradation of recalcitrant components of commercial NAs within 14 days and degraded significantly more adamantane-1-carboxylic acid (A1CA) than individual cultures of either P. putida or C. vulgaris within 33 days. Enrichment cultures derived from 05PW were used to further investigate the biodegradation of A1CA and 3EA. Both A1CA and 3EA were significantly degraded after 33 days and mass spectral analysis demonstrated the production of hydroxyadamantane intermediates with a concurrent decrease in toxicity. Comparisons of 165 rRNA gene sequences at days 0 and 33, during biodegradation showed an increase in abundance at day 33 of the genera Pseudomonas, Acidovorax, Flavobacterium and Methylobacterium.

Characterisation and understanding of genetic components is a key part of both synthetic biology and systems biology. Quantitative knowledge of how DNA parts encode function allows parts to be predictably constructed into synthetic gene circuits. Less understood is how the expression of a synthetic gene circuit can have a detrimental effect on its host cell (the chassis) and how these effects can feed back to the behaviour of the circuit. In this thesis, we investigate how synthetic circuits use cellular resources (e.g. DNA polymerase, RNA polymerase, ribosomes, tRNA, etc.) to replicate and express and we quantify these effects and model gene expression in a way that accounts for this. This is done by considering this shared 'resource pool' as an interface between the host cell and the synthetic circuit. Through genetic engineering and synthetic biology, we have created a system that monitors the availability of shared resources in E. coli, thus enabling the quantification of the burden a synthetic circuit places on the cell's resources. We then measure the burden of a combinatorial library of different designs to examine how different genetic components influence the magnitude of burden. This is accompanied by a mathematical model. Through this method we work towards a system that will enable the prediction of how to optimise the design of a synthetic circuit with regards to its output and the levels of burden it places on a cell.