The purification of monoclonal antibodies and Fab’ antibody fragments are of central importance to the pharmaceutical industry. In 2008, 29 new therapies based on such molecules were approved for the US market. Traditionally, a multistep process achieves purification with the majority of steps being packed bed chromatography. Chromatography is the major contributor to the unit operation costs in terms of initial capital expenditure for packing and recurrent replacement costs. When considering the demand for biopharmaceuticals, it becomes necessary to consider alternative process strategies to improve the economics of purification of such proteins. To address this issue, this thesis investigates precipitation to selectively isolate Fab’ or remove protein impurities to assist the initial purification process. The hypothesis tested was that the combination of two or more precipitating agents will alter the solubility profile of Fab’ or protein impurities through synergistic multimodal effects. This principle was investigated through combinations of polyethylene glycol (PEG) with ammonium sulphate, sodium citrate and sodium chloride at different ratios in a novel multimodal approach. A high throughput system utilising automated robotic handling was developed in microwells at 1 mL scale per well to enable the rapid screening of a large number of variables in parallel using a Design of Experiments (DoE) approach to statistically design studies in a two stage process, based on Quality by Design principles. In the first stage, Fab’ precipitation using PEG was investigated using a screening study in the form of a full two level factorial DoE to investigate a large design space. This was followed by a second more focused central composite face centred DoE to find optimal experimental conditions to deliver a high Fab’ yield and purification factor in the range investigated. A design space comprised of the responses percentage Fab’ yield and purification factor was created to give a robust region where Fab’ yield was ≥ 90% with a maximum purification factor of 1.7. A normal operating range (NOR) was defined within this design space for operational simplicity when working at process scale. A confirmatory run was performed within the NOR with PEG 12000 15% w/v pH 7.4, which delivered a Fab’ yield and purification factor of 93% and 1.5 respectively. In the second stage, optimum conditions from the first study were used in a central composite face centred DoE incorporating multimodal conditions combining PEG with three salts from the Hofmeister series namely, ammonium sulphate, sodium citrate and sodium chloride. It was found that 90% Fab’ yield with a purification factor of 1.9 was achievable with PEG 12000 15% w/v/0.30 M sodium citrate/0.15 M ammonium sulphate pH 7.4. This was an improvement of 26% relative to the use of 15% w/v PEG 12000 pH 7.4 in single mode. However an alternative precipitation strategy to precipitate ~20% of protein impurities whilst Fab’ remained soluble using PEG 12000 6.25% w/v/0.4 M sodium citrate pH 7.4 was proposed instead. The advantage of this approach at process scale is the potential ease of processing due to removal of a solubilisation step and the significantly reduced viscosity of the precipitating agent relative to that of high concentrations of PEG. It was shown that this system could mimic process scale, which was verified at laboratory scale (50 mL stirred tank reactor (STR)) and pilot process scale (5 L STR). A process run through was performed using a 1 mL SP Sepharose Hi Trap pre packed bed column (GE Healthcare, Uppsala, Sweden) to capture Fab’ from homogenate (control), multimodal (PEG 12000 6.25% w/v/0.4 M sodium citrate pH 7.4) and single mode (PEG 12000 15% w/v pH 7.4) feedstreams. The final process purification factor for the three feedstreams were 2.5, 4.4 and 3.5 respectively. The use of multimodal precipitated impurities prior to a packed bed step had improved process performance by a purification factor of 1.9. This underlines the importance of assessing the interaction of individual processing steps, and the implementation of appropriate scale down models as a means of achieving process parameter ranging understanding. The impact of which has the further potential to improve the longevity of chromatography resins and reducing overall downstream purification cost.

Microalgae are playing an increasingly important role in biotechnology because of their natural abundance of valuable products such as proteins, pigments and bioactive compounds. Intensive research is also being applied to biofuel production from algal biomass owing to their fast reproduction rates and high lipid content. This thesis explored several main steps in algal biotechnology including screening suitable algal strains; optimizing algal growth either by controlling cultivation conditions or using non-motile mutants instead of motile wild type strains; algal biomass harvesting through flocculation using chitosan as the flocculant; and the use of algal biomass for engine combustion. The effect of algal cell motility on growth rate and biomass productivity was explored using Chlamydomonas reinhardtii as the model species. The motility of a non-motile mutant bld1 was recovered by expressing the BLD1 cDNA in the nuclear genome, and the growth of these two strains with equivalent genetic backgrounds was compared. The superior cell growth of the bld1 mutant indicates that the use of non-motile mutants instead of their motile wild types for algal biomass production may help improve the productivity while using the same energy input. Flocculation with chitosan to harvest algal biomass of several algal species was explored in this study. Chitosan proved to be highly efficient in the induction of cell flocculation of green alga Chlorella sorokiniana with the clarification efficiency reaching above 99% below pH 7 at optimal dosage. Influencing factors of flocculation efficiency were explored in detail. Finally, to reduce the complicated and costive processes of algal biofuel production, this work explored an alternative way of utilizing energy from algal cells by blending algal slurry into fossil diesel using specific combinations of surfactants: the aim being to reduce the consumption of fossil diesel, and at the same time improve engine performance compared with standard diesel/water emulsions.

Current industrial trends encourage the development of sustainable, environmentally friendly processes with reduced energy and raw material consumption. Meanwhile, the increasing market demand as well as the tight regulations in product quality, necessitate efficient operating procedures that guarantee products of high purity. In this direction, process intensification via continuous operation paves the way for the development of novel, eco-friendly processes, characterized by higher productivity compared to batch (Nicoud, 2014). The shift towards continuous operation could advance the market of high value biologics, such as monoclonal antibodies (mAbs), as it would lead to shorter production times, decreased costs, as well as significantly less energy consumption (Konstantinov and Cooney, 2015, Xenopoulos, 2015). In particular, mAb production comprises two main steps: the culturing of the cells (upstream) and the purification of the targeted product (downstream). Both processes are highly complex and their performance depends on various parameters. In particular, the efficiency of the upstream depends highly on cell growth and the longevity of the culture, while product quality can be jeopardized in case the culture is not terminated timely. Similarly, downstream processing, whose main step is the chromatographic separation, relies highly on the setup configuration, as well as on the composition of the upstream mixture. Therefore, it is necessary to understand and optimize both processes prior to their integration. In this direction, the design of intelligent computational tools becomes eminent. Such tools can form a solid basis for the: (i) execution of cost-free comparisons of various operating strategies, (ii) design of optimal operation profiles and (iii) development of advanced, intelligent control systems that can maintain the process under optimal operation, rejecting disturbances. In this context, this work focuses on the development of advanced computational tools for the improvement of the performance of: (a) chromatographic separation processes and (b) cell culture systems, following the systematic PAROC framework and software platform (Pistikopoulos et al., 2015). In particular we develop model-based controllers for single- and multi-column chromatographic setups based on the operating principles of an industrially relevant separation process. The presented strategies are immunized against variations in the feed stream and can successfully compensate for time delays caused due to the column residence time. Issues regarding the points of integration in multi-column systems are also discussed. Moreover, we design and test in silico model-based control strategies for a cell culture system, aiming to increase the culture productivity and drive the system towards continuous operation. Challenges and potential solutions for the seamless integration of the examined bioprocess are also investigated at the end of this thesis.

Non-invasive techniques capable of visualising and quantifying blood flow in-vivo are highly desirable in studying a wide range of cardiovascular diseases. Although existing ultrasound imaging techniques have been widely used clinically to visualise and quantify blood flow, they have various limitations in terms of field of view, temporal and spatial resolution, imaging sensitivity, and beam-flow angle dependence. In this thesis, our aim is to develop flow quantification tools capable of non-invasively measuring the flow velocity, wall shear stress (WSS) as well as intraluminal mixing. Firstly, a high frame-rate ultrasound imaging velocimetry (UIV) system was developed based on tracking the speckle patterns of microbubbble contrast agents in contrast-enhanced ultrasound image sequences acquired from a plane wave imaging system. Initial evaluation of the system demonstrated the potential of the new system as a flow velocity mapping tool capable of tracking fast and dynamic flow and we improved our flow velocity measurement technique by introducing an incoherent ensemble correlation approach in the UIV tracking algorithm. Such a modified UIV technique avoids the motion artifact which could potentially affect the velocity measurement as compounded plane wave images are not coherently summed during the compounded plane wave image formation. Ultrasound flow simulations were conducted to fully evaluate our new modified-UIV technique. Together with some in-vitro experiments on physiologically relevant flow phantoms, we demonstrated the capability of our system to provide robust, angle independent, sensitive, and accurate two-dimensional velocity measurements. Secondly, as studies have revealed strong correlation between WSS and the initiation and development of atherosclerosis, we extended our UIV technique to the derive spatio-temporal wall shear rate from the velocity flow profile. The performance of the system to provide wall shear stress distributions was initially evaluated in simulation and demonstrated in-vitro using physiologically relevant flow phantoms. Thirdly, a novel approach which uses the high frame rate system and controlled microbubble destruction for flow visualisation and intraluminal mixing quantification was also proposed. Three different model vessel geometries: straight, planar curved and helical, with known effects on the flow field and mixing were evaluated against computational fluid dynamics (CFD) results. The findings indicated the technique is not only capable of visualising the secondary flows, but also able to quantify the degree of mixing in the different configurations. Finally, real time processing of the image formation and flow quantification technique were explored due to the large amount of data generated from the high frame rate ultrasound system. Initial development of a graphic processing unit (GPU) accelerated plane wave UIV system was demonstrated with the potential for real time measurements.

Production of microalgal derived biofuels and bioproducts have recently been the focus of significant academic and commercial research, with the aim of replacing prevailing finite fossil fuel derived alternatives. Much of these efforts are focused on increasing algal and lipid productivity in order to reduce costs to an economically sustainable level, while maintaining environmental sustainability. In this research project, a series of investigations were carried out to determine the influence of organic carbon supplementation on the carbon metabolism of green algae and how this impacted both growth rate and lipid productivity.

Biochemical reaction networks involve many chemical species and are inherently stochastic and complex in nature. Reliable and organised functioning of such systems in varied environments requires that their behaviour is robust with respect to certain parameters while sensitive to other variations, and that they exhibit specific responses to various stimuli. There is a continuous need for improved models and methodologies to unravel the complex behaviour of the dynamics of such systems. In this thesis, we apply ideas from information theory to develop novel methods to study properties of biochemical networks. In the first part of the thesis, a framework for the study of parametric sensitivity in stochastic models of biochemical networks using entropies and mutual information is developed. The concept of noise entropy is introduced and its interplay with parametric sensitivity is studied as the system becomes more stochastic. Using the methodology for gene expression models, it is shown that noise can change the sensitivities of the system at var- ious orders of parameter interaction. An approximate and computationally more efficient way of calculating the sensitivities is also developed using unscented transform. Finally, the methodology is applied to a circadian clock model, illustrating the applicability of the approach to more complex systems. In the second part of the thesis, a novel method for specificity quantification in a receptor-ligand binding system is proposed in terms of mutual information estimates be- tween appropriate stimulus and system response. The maximum specificity of 2 × 2 affinity matrices in a parametric setup is theoretically studied. Parameter optimisation methodology and specificity upper bounds are presented for maximum specificity estimates of a given affinity matrix. The quantification framework is then applied to experimental data from T-Cell signalling. Finally, generalisation of the scheme for stochastic systems is discussed.

This thesis, completed in collaboration with Eli Lilly & Co., aims to understand and assess the structural and mechanical changes that occur as agarose-based chromatography resins are exposed to different bioprocessing conditions in an attempt to explore the mechanisms by which different resins age. By understanding this, there is significant potential for facilitating timely and improved decisions in large-scale chromatographic operations, maximising resin lifetime whist maintaining acceptable column performance. Scanning electron microscopy (SEM) was used to image the structural properties of nine widely used agarose-based chromatography resins before use while pressure-flow analysis was used to characterise the mechanical properties of the same fresh resins. The results showed that the Capto family had the highest critical velocities (Capto Adhere- 492, Capto Q- 477 cm/hr), whilst Sepharose CL-6B, Sepharose 4 Fast Flow and Sepharose CL-4B had the lowest critical velocity values (283, 204, 149 cm/hr respectively). There were practical limitations in using the pressure-flow technique alone to for mechanical characterisation, including the large quantity of chromatography resin and buffers and the stringent criteria required to pack a column. Dynamic mechanical analysis (DMA) was therefore developed as a novel technique in this field to address these limitations and allowed for further mechanical characterisation based on the viscoelastic properties using 1ml of resin. The technique was applied to the nine studied resins and correlated with the results obtained using the pressure-flow technique. The same trends were observed – The Capto family showed the highest resistance to deformation (Capto Adhere- 2.7, Capto Q- 1.92 1/%min-1) through to Sepharose CL-6B, Sepharose 4 Fast Flow and Sepharose CL-4B which exhibited the lowest slurry resistances (0.59, 0.4, 0.3 1/%min-1 respectively). These results indicate that DMA can be used as a small volume, high-throughput technique, relative to pressure-flow analysis, for the mechanical characterisation of chromatography media. The structural imaging and mechanical testing tools developed in this study were then applied to measure changes in resins that had undergone lifetime studies. These studies expose the resins to repeated cycles of use to understand how they age in a particular bioprocess, enabling decision making about their use. The first set of experiments exposed the resins to the cleaning cycle only, whilst in the second set of experiments, the resins had been used for lifetime studies in the production of monoclonal antibodies (termed ‘aged’ resins). The results indicated that MabSelect (highly cross-linked protein A resin) and Q-Sepharose High Performance (anion exchange cross-linked resin) appeared to show similar mechanisms of aging. Their matrices showed agarose fibre breakage with increased exposure to process conditions. In the case of Capto Adhere (highly cross-linked multimodal anion exchange resin) and MabSelect Xtra (highly cross-linked protein A resin), the mechanism of aging appeared to be associated foulants coating the surface fibres. The results indicate that the interaction of CIP reagents and foulants (as opposed to CIP reagents alone) cause the greatest impact on the structural integrity of the resins. Pressure-flow and DMA characterisation were used to examine the mechanical properties of the cycled resins to provide the first systematic study of these issues. The results showed that fresh resins were consistently more robust than either of the cycled resins but the greatest mechanical differences were observed between fresh resins and aged resins, which corroborated the structural analysis data. Statistical analysis was performed with one way ANOVA to determine whether DMA could be independently used to assess the impact of process conditions on the mechanical properties of chromatography media and the results show a >80% certainty that DMA can be employed for this purpose.

Bone is one of the most common tissues to be transplanted, with over 2.2 million grafting procedures performed worldwide every year (Van der Stok et al., 2011). Autologous bone grafts, while considered the current gold standard, have inherent risks including limited donor tissue availability, donor site morbidity, surgical complications, and pain of procedure. Alternative approaches to treating bone tissue defects are required based on clinically effective bone graft substitutes that can be manufactured at a commercially relevant scale. Tissue engineering is an alternative strategy that uses biocompatible scaffolds in combination with cells as a bioactive implant to induce bone repair. In this thesis, microspherical bioactive glasses have been studied as a platform for scalable bone tissue engineering that has flexibility to address diverse geometric requirements with the aim of becoming a commercially available tool. Specifically, titanium-doped phosphate glass microspheres have been studied for their ability to support bone progenitor cells. Here, the microspheres (5 and 7 mol% TiO2) were assessed in their ability to support proliferation of osteoblast-like cell (MG63) and proliferation and osteogenic differentiation of human bone-marrow derived mesenchymal stem cells (hBM-MSCs) under static and dynamic agitation culture. Scalability was assessed using scalable dimensionless Froude number to scale microwell plate cultures to 125ml Erlenmeyer flask cultures using Froude as a tool to map mixing systems at both scales. MG63 and hBM-MSC proliferation was observed on the microspheres under all conditions studied as well as extracellular matrix protein secretion, confirming the biocompatibility of the materials tested. Similar growth kinetics was observed at both scales, where moderate agitation stimulated cell proliferation, but higher agitation was damaging to cells. Upregulation of key bone expression markers (COL1A1 and SPP1) was observed also at moderate orbital agitations, while on at high agitation rates this was largely absent, except for upregulation of SPP1 on the control microsphere, Synthemax. Furthermore, biomaterial resorption was observed upon differentiating mouse-derived monocytes into osteoclasts on the titanium-phosphate glass discs. In conclusion, large-scale culture using titanium-doped phosphate glass microspheres was achieved with hBM-MSCs, with the substrate effectively supporting cell proliferation and osteogenic differentiation. This research provides a stepping stone in understanding how biomaterials processed into microcarrier format can be utilised in a commercial environment to create clinically relevant quantities of tissue engineering bone.

Cell therapy has the potential to treat a wide variety of unmet therapeutic indications that affect a growing number of people globally. Many of these therapies require purification steps to separate specific cell types from heterogeneous populations. This thesis investigates current affinity purification platforms to isolate human pluripotent stem cell-derived progenitor photoreceptors for the treatment of retinal dystrophies, and introduces a novel purification technology which possess bioprocessing and clinical advantages over current techniques. Successful production of progenitors was achieved using both induced pluripotent stem cells (iPSC) and human embryonic stem cells (hESC). By controlling the cell aggregation step and other iterative improvements to the retinal differentiation, 35.7% of cells generated expressed Cone-Rod Homeobox (CRX)-positive – a key marker to define progenitor photoreceptors. The critical performance metrics of fluorescent-activated cell sorting (FACS) and magnetic-activated cell sorting (MACS) were then derived through experimentation. Sort purity, progenitor yield and viable cell recovery of CD73-positive cell populations – a surface marker shown to co-express with CRX - were measured, and demonstrated that high purity separations above 90% were attained. However, both methods suffered from low cell recoveries with over 30 or 40% of cells (for FACS and MACS respectively) lost through the numerous processing steps involved in labelling cells with either a fluorescent or paramagnetic tag, washing and sorting samples. Cell labelling also leaves the product with a bound cellular label, complicating additional processing and potentially causing toxic clinical affects. A novel purification technology was assessed with SpheriTech affinity beads that possess bioprocess and clinical advantages over current purification methods. Cells are unmodified through isolation, with the positively selected cell type remaining label-free after processing. Consequently, cells experience minimal process steps so the time, risk, cost burden of purification and cell loss is reduced. Comparable purity with all separations was observed, however progenitor yield was noted to be lower with SpheriTech affinity beads than for FACS and MACS. To assess the impact different purification technologies have upon the complete bioprocess in an iPSC-derived therapy, an economic cost-modelling tool was created. By inputting experimentally-derived data into an integrated model, the cost of goods (COG) per dose was evaluated when using each of the three affinity purification methods. FACS was found to be economically favourable only at small production scales due to throughout limitations, with MACS presenting the most cost effective technology at all other scales. However, if progenitor yield could be increased to improve process yields through further process development, SpheriTech would compete with MACS across all scales tested.

The development of therapeutic proteins is a driving force in the current manufacture of biopharmaceuticals. Freeze drying is widely used in the fabrication of final dosage forms of therapeutic proteins. Using a series of A33 Fab mutants, this thesis aimed to correlate their physicochemical properties to the outcomes of freeze-drying. Preliminary studies employed a homogeneous freeze-drying process on 96-well plates. It was found that K65M and K133M surface mutations, the use of acetate buffer, low pH, increased ionic strength, and the use of NaCl, caused the most monomer loss; whereas S75K, C226S, and L50K mutations, high pH, and the use of Na2SO4 caused the least monomer loss. Several in-silico modelling tools were used to design mutants for studying the impact of protein conformational stability. Rosetta software, RMSF and B-factor analyses were used to evaluate the mutant candidates and restrict the mutations mainly located in the flexible regions. Unstable mutants were prepared as controls to validate the prediction accuracy. In freeze-drying, most of the stabilising mutants had 20% less monomer loss than C226S, while the destabilising ones had 14-46% more monomer loss. Tm and ΔΔG estimated the monomer loss in freeze-drying with low degree of accuracy. Compared to freeze-drying, a more distinct difference was observed in the aqueous phase as all the destabilising mutants aggregated more than 5 times faster than C226S and the stabilising mutants did. Tm correlated well with the aggregation in aqueous phase, indicating conformational stability was more important in aqueous phase than that in freeze-drying. In addition, excipients barely exerted influence on the stable mutants but provided sufficient protection for the unstable ones, which was reflected by their correlations to Tm values. The rank-order of excipient effects for individual mutants, relative to that of wild type, became less similar as the mutant ΔTm magnitude increased.