Pichia pastoris (P. pastoris) expression systems are gaining increased interest in industry due to their ability to achieve high cell densities and production levels. This thesis aims to characterise the following critical elements of P. pastoris upstream bioprocessing: cell growth, cell productivity and bioprocess monitoring. Methanol is the principle carbon source and gene expression induction agent in most P. pastoris fermentation strategies. Monitoring of methanol levels during fermentation enables cell growth and productivity to be optimised, whilst methanol toxicity is avoided. A novel approach to at-line methanol monitoring was investigated, seeking to exploit the chromogenic reactivity of dehydroascorbic acid (DHAA) with methanol. Recombinant variants of the human reporter enzyme, placental alkaline phosphatase (PLAP) were used as a model fermentation product, and the performance of two mechanistically distinct commercial assays was compared for their applicability to high cell density process streams. In total four new P. pastoris strains expressing PLAP variants were successfully created and the impact of methanol carbon source feed strategies on the strain performance investigated in terms of growth rate and recombinant protein production. A 2-fold increase in rate of methanol feed resulted in a final biomass increase of about 40% and in a volumetric productivity decrease of about 75%. A novel method, combining phases of low and high methanol feed rates, matched but did not exceed conventional methanol feed rate strategies.

Microalgae are considered as attractive platforms for the synthesis of high-value heterologous proteins due to their many beneficial attributes including ease of cultivation, lack of pathogenic agents, and low-cost downstream processing. However, recombinant protein levels are low compared to microbial platforms and commercial production is not a reality yet. Promising research using the model microalga Chlamydomonas reinhardtii has highlighted this potential, particularly for transgene expression in the chloroplast. The objective of this research was to study different strategies for expression in the C. reinhardtii chloroplast aimed at increasing growth rate and recombinant protein production, by means of cell engineering and bioprocess optimisation. Two main approaches were considered for this purpose. Firstly, the insertion of the light-harvesting protein proteorhodopsin (PR) into the cell membrane, that has been reported to increase growth rate and culture lifespan in bacterial systems. This has resulted in PR accumulation to low yet detectable levels providing a moderate increase (12 %) of specific cell growth. Noteworthy, this is the first demonstration of an integral membrane recombinant protein expressed in the chloroplast. The second approach was to express a gene for a novel fluorescent protein (VFP) under the control of different regulatory elements. Detectable VFP levels were produced, with increased protein levels when using the psaA promoter/5’UTR element, and with co-expression of a gene for the Spy chaperone. These strains were used to study the effect of temperature, media and light intensity on recombinant protein production and cell growth. Protein levels and fluorescence allowed determining improved cultivation conditions as 30 °C under mixotrophic mode, and these conditions were tested for the accumulation of a therapeutic protein (Cpl-1 endolysin). In conclusion, protein productivity was observed to be protein-specific and improved conditions that increase protein levels for one protein cannot necessarily be extrapolated to the accumulation of a different protein.

Human embryonic stem cells (HESC) are a promising source of retinal pigment epithelium (RPE) for the treatment of common and incurable forms of blindness such as age-related macular degeneration (AMD). Whilst most HESC lines will produce some pigmented RPE cells when allowed to overgrow and spontaneously differentiate for 30-60 days, the efficiency of this process is highly variable and the critical factors which determine target cell yield remain largely uncharacterised. This will prove problematic in the large-scale production of RPE cells needed for cell therapy. In this project the aim was to identify and minimise sources of variability during differentiation and to develop an efficient and scalable HESC-RPE differentiation protocol. Using a novel imaging platform in combination with quantitative gene expression analysis and immunocytochemistry, the relative differentiation efficiency in two new cell lines, Shef6 and Shef3 was characterised. It was found that the age of the starting HESC population, the cell seeding density and the passaging method used have a strong influence on RPE yields. It was also demonstrated that RPE can be generated from HESC following single cell dissociation and in the absence of feeder cells, thereby significantly simplifying cell culture logistics and reducing variability. In addition it was shown that the lower yielding cell line, Shef3, has a reduced innate propensity for neuroectoderm conversion and that by directing this process with a small molecule, dorsomorphin, efficiency can be significantly improved. Overall this novel protocol increased RPE foci yields per cm2 by 5 fold in Shef6 and 4 fold in Shef3 compared to traditional mouse embryonic fibroblast (MEFs) co-culture systems. Since HESC-derived RPE are now entering clinical trials, it has become increasingly important to optimise manufacturing; this study identifies several critical parameters that could help develop a robust, scalable and cost-effective strategy for HESC-RPE manufacturing.

The next generation of monoclonal antibody (mAb) therapies are under increasing pressure from healthcare providers to offer cost effective treatments in the face of intensified competition from rival manufacturers and the looming loss of patent exclusivity for a number of blockbusters. To remain completive in such a challenging environment companies are looking to reduce R&D and manufacturing costs by improving their manufacturing platform processes whilst maintaining flexibility and product quality. As a result companies are now exploring whether they should choose conventional batch technologies or invest in novel continuous technologies, which may lead to lower production costs. This thesis explores the creation of a dynamic tool as part of a decision-support framework that is capable of simulating and optimising continuous monoclonal antibody manufacturing strategies to assist decision-making in this challenging environment. The decision-support framework is able to tackle the complex problem domain found in biopharmaceutical manufacturing, through holistic technology evaluations employing deterministic discrete-event simulation, Monte Carlo simulation and multi-attribute decision-making techniques. The hierarchal nature of the framework (including a unique sixth hierarchal layer; sub-batches) made it possible to simulate multiple continuous manufacturing scenarios on a number of levels of detail, ranging from high-level process performance metrics to low-level ancillary task estimates. The framework is therefore capable of capturing the impact of future titres, multiple scales of operation and key decisional drivers on manufacturing strategies linking multiple continuous unit operations (perfusion cell culture & semi-continuous chromatography). The work in this thesis demonstrates that the framework is a powerful test bed for assessing the potential of novel continuous technologies and manufacturing strategies, via integrated techno-economic evaluations that take proof-of-concept experimental evaluations to complete life-cycle performance evaluations.

A major challenge facing the development of effective cell therapies is the efficient differentiation of pluripotent stem cells (PSCs) into pure populations. This is caused, in part, by the heterogeneous presence of functionally distinct subpopulations in undifferentiated PSCs. These can exhibit variable developmental potential, suggesting that there will be a heterogeneous response to differentiation cues, and a low yield of the target cell type. The importance of recapitulating the in vivo stem cell niche during stem cell process development is now widely acknowledged, and thus, manipulation of microenvironmental factors will be invaluable in engineering optimal in vitro conditions for cell culture. Lowering oxygen tension to physiological levels can affect both the expansion and differentiation stages. However, to date, there are no studies investigating the consequences of culturing PSCs under hypoxic conditions on subsequent lineage commitment at ambient oxygen levels. Mouse Embryonic Stem Cells (mESCs) were passaged three times at 2% O2 before allowing cells to spontaneously differentiate as embryoid bodies (EBs) in normoxic (20% O2) conditions. Maintenance of mESCs under hypoxia was associated with a population shift away from a Rex1-positive ICM-like (Inner Cell Mass-like) state, to a primed epiblast-like state exhibiting a significant increase in the expression of early differentiation markers FGF5 and Eomes. Conversely, expression of these committed markers was decreased in human induced pluripotent stem cells (hiPSCs) following the same hypoxic step, and was instead associated with a significant increase in Rex1 expression. Hypoxic preconditioning primed mESCs for their subsequent differentiation into mesodermal and endodermal lineages, as confirmed by increased gene expression of Eomes, Goosecoid, Brachyury, AFP, Sox17, FoxA2, and protein expression of Brachyury, Eomes, Sox17, FoxA2, relative to normoxic cultures. The effects extended to the subsequent formation of more mature mesodermal lineages. A significant upregulation of cardiomyocyte marker Nkx2.5 was also observed, and critically a decrease in the number of contaminant pluripotent cells after 12 days using a directed cardiomyocyte protocol. However, the impact of hypoxic preconditioning was to preferentially prime human cells for ectodermal lineage commitment during subsequent EB differentiation, with significant upregulation of Nestin and β3-tubulin. The research discussed in this thesis demonstrates the importance of oxygen-tension control during cell maintenance on the subsequent differentiation of both mouse and human PSCs, and highlights the differential effects. The reported results have indicated that further to focusing on the differentiation stage itself, it will be critical to consider the prior maintenance of PSCs to fully optimise future stem cell processes. Furthermore, the work has highlighted the importance of considering each stage of bioprocess development individually, to best recapitulate the transient conditions in vivo.

This work investigated the biocompatibility of soluble phosphate based glasses as scaffolds for supporting the in vitro morphogenesis the hard-soft tissue interface (enthesis) as an approach dealing with ligaments and tendons clinical problems. The short term response of human oral osteoblasts (HOB), oral fibroblasts (HOF) and flexor tendon fibroblasts (HTF) was assessed on glass discs of various compositions, and different dissolution rates, of the generic ternary form (CaO)o.ox-(Na2O)o.oy-(P20s)o5 through evaluating the maintenance of the seeded cell attachment, survival, proliferation and phenotype by using SEM, immunocytochemistry and the CyQUANT cell density kit. Subsequently, the most biocompatible ternary glass compositions were utilised for fibre production. The effect of fibre diameter of cell adhesion and survival was determined and quaternary glass fibres, of the generic composition (CaO)o.46-(Na20)o.ox-(Fe203)o.oy-(P2C>5)o.5o, where characterised in terms of diameter and solubility profiles. Three-dimensional scaffolds were produced from these fibres and the long term viability, morphology and population growth of the seeded HOB and HOF cells were determined using immunocytochemistry and direct cell count. This was coupled with application of qPCR experiments to evaluate the maintenance of differentiation of the seeded cell population. The role of extrinsic factor inclusion in enhancing in vitro, scaffold associated, tissue morphogenesis was also investigated by stimulating osteogenic differentiation in the seeded HOB cell population. An open lamellar flow bioreactor providing nutrients, oxygen and waste perfusion to the cell-scaffold culture was deigned and assessed. The feasibility of simulating the anatomical architecture of the enthesis has also been addressed as cells were seeded in co-culture on a continuous fibre arrangement. In this study, quaternary phosphate based glass fibre scaffolds containing 3 mol% iron oxide (Fe203), of approximately 30 microm in diameter, and of the composition (CaO)o.46-(Na20)o.oi-(Fe203)o.o3-(P205)o.5o have been shown to support HOB and HOF attachment, well spread morphology, survival and proliferation with no negative impact of cell differentiation. Induction of osteogenesis in the scaffold culture has resulted in up-regulating HOB related gene transcription and the flow culture system, at certain flow rates, has been verified for future use. The Co-culture system design has been successfully implemented as HOB and HOF cells were seeded with an acellular separation zone across the fibre scaffold arrangement.

The ultra scale-down technology (USD) is used in this body of work to mimic and study pharmaceutical manufacturing process steps with reduced material requirements. The production of a novel type of domain antibody in recombinant E. coli is used as a context in which the technology is studied. The main focus areas are fermentation, centrifugation, depth-filtration and membrane filtration. Process characterisation was achieved through the detailed study and establishment of sets of critical parameters, such as process material physical properties (e.g. viscosity, solids volume fraction, product quantity and quality) that affect the efficiency of each these process steps, and discoveries from this activity were fed back towards the development of more complete USD models. An understanding of the synergy between the fermentation process and subsequent primary recovery was discovered, which lead to the formulation of a series of trade-offs between product quantity and subsequent ease of recovery. The effect of the cell broth pre-treatment, such as exposure to hydrodynamic shear, homogenization or flocculation, on the efficiency of separation by centrifugation, and the impact on subsequent process steps was also studied. These were linked to theoretical models which allowed the evolution of a prediction for processability based on process material quality. Finally, a methodology was developed to discover the best processing conditions (in terms of flux and shear rate) during membrane microfiltration and a roadmap was developed to establish the rules for accurate scale-up. This work provides improvements and insights into how the USD technology can be used to increase the rate and quality of bioprocess discoveries during drug development. In all process steps studied, the USD technology has shown that a much wider range of conditions than would be possible at large scale can be investigated in a reduced amount of time, which allows a large window of operation to be used during scale-up decisions.

In this thesis, an online oxygen monitoring system has been presented to quantify the dissolved oxygen concentration levels in a microfluidic device for the adherent culture of pluripotent stem cells. Oxygen is a critical environmental cue regulating stem cell fate. Therefore an online monitoring system, combined with the greater control over the soluble microenvironment provided by microfluidic systems, is of interest to assist the development of robust bioprocessing approaches for regenerative medicine therapies. An oxygen monitoring system consisting of optical sensors have been designed and integrated with the microfabricated modular culture device technology previously established by the Szita Group. The monitoring system enables non-invasive measurements of bulk and peri-cellular oxygen concentration levels in an adherent culture device. A platform for the operation of the device and monitoring system has been established to meet the requirements for adherent stem cell culture. It supports the adherent culture of mouse embryonic stem cells for up to one week under continuous perfusion and integrates automated monitoring systems for dissolved oxygen and culture confluency. Standard operating procedures for the microfluidic culture have been defined to ensure the robustness and repeatability of the process. The seamless integration of online oxygen monitoring with confluency monitoring has enabled for the first time to quantify non-invasively oxygen uptake rates of adherent stem cell cultures.

The two traditional approaches to modelling can be characterised as the development of mechanistic models from 'first principles' and the fitting of statistical models to data. The so-called 'hybrid approach' combines both elements within a single overall model and is thus composed of a set of mass balance constraints and a set of kinetic functions. This thesis considers methodologies for building hybrid models of bioprocesses. Two methodologies were developed, evaluated and demonstrated on a range of systems of simulated and experimental systems. A method for inferring models from data using support vector machines was developed and demonstrated on 3 experimental systems a Murine hybridoma shake flask cell culture, a Saccharopolyspora erythraea shake flask cultivation and a 42L Streptomyces clavuligerus batch cultivation. On the latter system the method produced models of similar accuracy to previously published hybrid modelling work. While support vector machines have been widely applied in other contexts this method is novel in the sense that there are no previously published papers on the use of support vector machines for kinetic modeling of bioprocesses. On 50 randomly created dynamical systems it was shown that the hybrid models produced using the support vector machine methodology were generally more accurate than those developed using feed forward neural networks and that could not be distinguished from models produced using a more computationally demanding method based round genetic programming. Additionally a Bayesian framework for hybrid modelling was developed and demonstrated on simple simulated systems. The Bayesian approach requires no interpolation of data, can cope with missing initial conditions and provides a principled framework for incorporating a priori beliefs. These features are likely to be useful in practical situations where high quality experimental data is difficult to produce.

Optimisation of a mammalian cell culture process requires the testing of many process parameters. High yielding processes can result in reduced batches, hence bringing the product to market quicker and increasing manufacturing capacity. To reduce the cost and duration of process optimisation a novel miniaturised stirred bioreactor system (MBR), the BioXplorer™, a prototype of a commercial MBR system initially developed for microbial fermentations is described here. The system enables the operation of 4-16, 500 mL, independently controlled bioreactors in parallel. Each bioreactor is a scale down model of a lab-scale stirred tank bioreactor (STR) and constructed from the same materials. Agitation of the bioreactor can be via a magnetically driven 4 blade marine impeller or a directly driven 3 blade marine impeller. Aeration can be achieved through a variety of sparger designs directly into the culture or via the headspace at a maximum flow rate of 200 mL/min. A detailed characterisation of the key engineering parameters has been conducted focusing on power input and the power to volume ratio (P/V), mixing time and the overall volumetric mass transfer coefficient (kLa). Successful scale comparison studies were conducted to 5L scale using constant P/V and mixing time, employing an industrially relevant GS-CHO cell line producing an IgG antibody. The growth kinetics and product titres compared favourably in both systems when conducting fed-batch operations. μ-max in the MBR was 0.024 h-1 and the maximum viable cell concentration was 10.4 x 106 cells/mL while in the 5L STR μ¬max was 0.029 h-1 and the maximum viable cell concentration was 9.8 x106 viable cells/mL. The product titres were also very similar in both the MBR (1.07 g/L) and the 5L STR (1.05 g/L). It has also been shown that the MBR can conduct continuous feeding using built-in peristaltic pumps, maintaining the glucose concentration in the culture at approximately 2.0 g/L after initiation of feeding. The MBR described here potentially provides a valuable and effective tool for process optimisation and is capable of performing complex feeding strategies.