Osteoarthritis (OA) is a degenerative joint disease that affects an estimated 9.6 % of men and 18 % of women over 60 years of age. However, there is no chondroprotective agent that is approved for clinical application. We showed that LOXL2 is elevated in the regenerative response during fracture healing in mice and has a critical role in chondrogenic differentiation. Indeed, LOXL2 is an anabolic effector that attenuates pro-inflammatory signaling in OA cartilage of the TMJ and knee joint, induces chondroprotective and regenerative responses. The goal of the present study was to evaluate if LOXL2 adenovirus in vivo promotes protective and anabolic responses in the knee and TMJ-OA cartilage. We employed (Cho/+) OA mouse model to assess knee and TMJ-OA, which is a genetic model that develops progressive TMJ-OA due to a mutation in the Col11a1 gene. Intraperitoneal injection every 2 weeks of Adv-RFP-LOXL2 in four-month-old Cho/+ male and female mice for 16 weeks upregulated Sox9, aggrecan (Acan) and other anabolic genes in the knee and TMJ. Next, to evaluate if LOXL2 expression occurred in degenerative lesions in human clinical TMJ-OA, immunofluorescence and confocal image analysis were performed. LOXL2 has the potential to induce anabolic gene expression in the progressive knee and TMJ-OA mouse model. We showed for the first time that LOXL2-related functions could be useful for anabolic therapies for Cho/+ mice progressive knee and TMJ-OA. Thus, these data show that LOXL2 could have clinical translational applications in the future for OA-related anabolic therapies.

Fc-fusion proteins (e.g. EPO-Fc) are the most often created fusion proteins due to their beneficial biological and pharmacological properties. The economic success of Fc-fusion proteins and other biopharmaceuticals production however, greatly depends on a robust, low-cost and highly effective protein mammalian cell extraction system . Understanding of how cells respond to a protein production environment based on metabolic profiles provides new goals for bioengineering of cell lines for best performance in biomanufacturing. Furthermore, insights on how individual cell metabolism and therefore phenotype, respond to cell microenvironment allows the underlying biological mechanisms to be explored in greater detail. This study focused on the application of mass spectrometry (MS) technologies, combining the analysis of metabolic profiles of cells extracts by GC-MS and MALDI-MS and spatial visualisation and distribution of metabolites within cells producing the fusion protein by MALDI-MSI and SIMS imaging. The analysis of external and internal metabolome profiles of cells producing the protein showed an extended effect of EPO-Fc fusion protein production on cell metabolism. The findings indicate that changes observed in EPO-Fc producing cells are related to enhanced protein and lipid synthesis highlighting that these cells are in a state of increased metabolic activity with the protein exocytosis into growth medium. Moreover, the composition of lipid bilayer of induced cells seemed to be different to non-induced cells. These findings were confirmed with the analysis of EPO-Fc induced cells using MS metabolic imaging. Multivariate analysis highlighted a number of metabolites that were significantly influenced by the protein expression when compared to control cells. The major metabolic changes in induced cells were those related to lipid metabolism. The information about metabolic changes in tetracycline-induced cells obtained from the analysis of cell populations was further supported with the analysis based on single-cell studies. Single-cell based studies also proved that investigations of individual cells provide additional insights about changes in metabolism of induced cells that can be referred to a unique, single cell and its phenotype. The analysis of CHO cells revealed a high level of heterogeneity within a cell population. Different cell phenotype and hence, metabolite content allowed for correlation between cell locations and their metabolite characteristics, specific for each type of cells. This project has successfully shown combination of bio-analytical techniques to investigate external and internal metabolome changes related to a fusion protein production in mammalian cells. Additionally, the significance of single cell approaches in metabolomics has also been highlighted, providing insights into the sub-cellular distribution of metabolites in cells producing EPO-Fc and information on the level of heterogeneity within a cell population. A multidimensional approach for metabolic profiling and future technological improvements of single-cell platforms are required to provide improved data acquisition and data analysis in order to better understand unknown processes involved in cell metabolism.

This PhD project explored the role of tricarboxylic acid (TCA) cycle enzymes in regulating Chinese hamster ovary (CHO) cell metabolism with respect to growth and recombinant protein expression. It was hypothesised that regulation of central metabolism can influence CHO cell productivity in terms of biomass and protein production. Thus the aim of these studies was to identify the key metabolic reactions of the TCA cycle associated with growth and protein expression in CHO cells. The gene expression of all TCA cycle genes was independently knocked-down using RNAi technology. The small interfering RNA (siRNA) mediated silencing of 11 TCA cycle genes significantly reduced cellular growth along with a decline in adenylate energy charges and an increase in catabolic reduction charges. The gene profiling of glucose and amino acid metabolism (not targeted by siRNA) suggested siRNA mediated knock-down of targeted TCA cycle genes led to cellular stress along with an enhanced rate of glycolysis leading to channelling of glucose for the generation of pyruvate. For the purpose of estimating intracellular metabolites, quenching and extraction method using ammonium bicarbonate and methanol was optimised to use with UCB CHO-K1 cell line and static transient siRNA transfections. A gas chromatography-mass spectrometry (GC-MS) analysis post-silencing of the aconitase gene, which catalyses the conversion of citrate to isocitrate in the TCA cycle, yielded higher MS peak intensities of at least four metabolites (gluconic acid, lysine, threonine and leucine) 72 h post-transfection in comparison to the controls. Transient knock-down of gene expression of seven TCA cycle genes in a recombinant stable cell line (expressing a rabbit monoclonal antibody) reduced cellular growth and altered the energy charges leading to a decline in antibody expression. Although silencing of the pyruvate dehydrogenase E1 gene, which is the component of the pyruvate dehydrogenase complex connecting glycolysis to the TCA cycle, did not affect cell viability, a reduction in antibody expression was recorded. Seven TCA cycle genes which demonstrated the most significant effect on cellular growth and energy charges were transiently over-expressed along with a monoclonal antibody in CHO-K1 cells with addition of their corresponding preceding intermediates. No differences in protein expression and cell specific productivity were observed compared to the control transfections. These results could be due to limitations of the effects of transient transfections for enhancing the metabolic activity of CHO cells. The aconitase gene demonstrated the most significant effect on CHO cell growth and proliferation in this study, therefore this gene was proposed as a novel selection marker for a metabolic selection system for the generation of recombinant therapeutics. This PhD project also established metabolite analysis tools and siRNA protocols for future metabolomic studies for investigating the intracellular CHO metabolism. The findings validated the hypothesis that TCA cycle plays an important role in CHO cell growth and recombinant protein production. The key metabolic genes affecting cellular growth and altering energy metabolism can be further explored for generation of an energy efficient CHO host-cell line (by over-expression of key TCA cycle genes) for enhanced recombinant protein production.

Transcriptional regulation is a determinant of stability of recombinant protein production in CHO cells. Fundamental studies of recombinant gene transcription in relation to chromatin environment and promoter regulation are important for CHO cell line development and selection. This study has developed a methodology based on a cell/vector system to study recombinant transcription and expression stability of different promoters and/or proteins in the similar genomic environment. The CHO-FRT mini-pools developed in this project were mini-pools of CHO-S cell lines containing Flp Recombination Target (FRT) sites with ß-galactosidase gene, under the influence of a SV40 promoter. Continuous culture of these mini-pools for 8 weeks using a robotic system demonstrated that 20% of the mini-pools studied revealed an unstable profile (with 30% loss of protein expression). Two of these mini-pools with different characteristics, CHO-FRT 1 (low producer/unstable) and CHO-FRT 108 (high producer/stable), were selected to be used on the study of influence of SV40 and CMV promoters in long-term recombinant expression. Genes encoding fluorescent proteins were integrated in a site-directed manner under the influence of SV40 or CMV promoters. A sub-clonal population of the top 10% yellow fluorescent protein (YFP) expressing cells of each mini-pool/promoter combination was selected by cell sorting and cultured for 4 weeks. During this period protein expression was monitored by flow cytometry and compared between both promoters. The results revealed that both SV40 and CMV promoters had an unstable expression with different degrees of instability and long-term expressing behaviours. For CMV, instability was considerably high displaying a long-term logarithmic loss of 50-80% of productivity while for SV40 the loss of productivity observed was only 40-45% with a linear behaviour during long-term culture. The vector system generated contained an MS2-RNA tag sequence cloned 3'- of the recombinant gene to track the recombinant mRNA by using the MS2/MCP-GFP system. This study showed the development of a protocol to measure the transcriptional output of recombinant promoters in CHO cells. The results showed background signal in CHO cells that requires further optimisation studies to allow the direct live cell image quantification of the transcriptional activity of recombinant promoters. Although not yet optimised, the successful combination of site-directed integration with recombinant mRNA tagging method has the potential to become a valuable tool to study the mechanisms of transcriptional activity and stability of transcription driven by different promoters in CHO cells.

Antibody engineering is an innovative field of research that has generated a wide range of novel antibody-based formats that both exploit and improve natural antibody properties. Novel format antibodies have the potential to offer significant advantages over natural antibodies when used as biopharmaceuticals, however these non-natural structures often pose a great challenge to the host cell used for their manufacture. Protein expression is a highly regulated process, and quality control mechanisms at each stage can result in a block, or "bottleneck" in expression. This can impact product yield, cost of goods and entry into the clinical pipeline. The molecular determinants that govern novel-format expression in host cells are poorly defined, however there is growing evidence that limited variations in both nucleotide and amino acid sequence can have a severe impact on antibody expression. Therefore this Thesis aims to investigate the consequences of sequence variation on the expression of a novel antibody format (mAbdAb) in mammalian host cells in order to determine the molecular mechanisms that govern their expression. A diverse panel of mAbdAbs with sequence variations limited to the dAb domain were generated through phage display and cloning technologies. It was determined that amino acid variations located within the CDRs of the dAb results in a range of expression titres in both transient HEK and stable CHO expression platforms. In vitro translation of mAbdAb heavy chain proteins in rabbit reticulocyte lysates (RRL) showed no difference in expression between sequence variants, therefore cell-free translation was suggested as a potential expression platform. Examination of each stage of expression in stable CHO cells revealed that the amount of mRNA was not limiting to expression and distinct expression profiles were observed at the protein level. The majority of mAbdAb constructs showed little evidence of intracellular heavy chain polypeptide which was not altered through chemical inhibition of proteolytic degradation pathways, indicating that degradation was not responsible for poor expression. This led to the hypothesis that low titres were related to how the CHO cell utilises the heavy chain message.

Animal cell culture is widely used as a platform for the production of a variety of biopharmaceuticals. The development of an efficient and productive cell culture requires a deep understanding of intra-cellular mechanisms as well as extra-cellular conditions for optimal synthesis. Mathematical modeling can be an effective strategy to predict, control, and optimize cell performance under different culture conditions. This research presents the evaluation of Chinese hamster ovary (CHO) cell culture secreting recombinant anti-RhD monoclonal antibody (MAb) through different processing modes, namely batch, fed-batch and perfusion operations. The ultimate objective of this study was to establish a comprehensive dynamic model which may be used for model-based optimization of the cell culture for MAb production in both batch, fed-batch or perfusion systems. In analyzing process performance, the key potential cause of cell growth inhibition was attributed to lowering of pH in the culture possibly due to the accumulation of dissolved carbon dioxide. The most important finding in this regard was the significantly different observed maximum total viable cell density in two identical cultures differing in culture volume only (250mL and 500mL). However, the other byproduct metabolites such as lactate and ammonia and glucose depletion were also capable affecting growth adversely causing growth arrest, viability reduction, apoptosis initiation and progress. Employing the experimental results of nutrient consumption, metabolite and biomass production, a metabolic flux based methodology was developed for modeling the metabolism of a CHO cell line. The elimination of insignificant fluxes resulted in a simplified metabolic network which was the basis for modeling the significant extracellular metabolites. Using kinetic rate expressions for growing and non-growing subpopulations, a logistic model was first developed for cell growth and dynamic models were formulated to describe culture composition and monoclonal antibody (MAb) secretion. The viable cell population was assumed to consist of normal growing, normal non-growing and apoptotic cell subpopulations. The rate of apoptotic cell formation was assumed to have a second order dependence on the normal cell concentration. The proposed mathematical model for metabolites included distinct terms that reflected the metabolic rates of growing and non-growing cell populations. The model was validated for a range of glutamine and glucose concentrations. Good agreement was obtained between model predictions and experimental data. In subsequent steps the attempt was to correlate the growth kinetics to significant variables of the culture. The regulatory effects identified through each culture condition were combined for a rational design of a dynamic model constructed for the viable cell subpopulation. A Tessier-based model was applied for defining the fraction of growing cells as a function of a growth inhibitor, presumably dissolved carbon dioxide. Although only few variables appeared in the biomass model, all equations were solved simultaneously. The parameters were estimated using the Metropolis-Hastings algorithm and the fmincon function in MATLAB. The final model adequately predicted the effect of significant variables on the metabolic behavior of CHO cells in batch, fed-batch and perfusion systems.

Animal cell culture is widely used as a platform for the production of a variety of biopharmaceuticals. The development of an efficient and productive cell culture requires a deep understanding of intra-cellular mechanisms as well as extra-cellular conditions for optimal synthesis. Mathematical modeling can be an effective strategy to predict, control, and optimize cell performance under different culture conditions. This research presents the evaluation of Chinese hamster ovary (CHO) cell culture secreting recombinant anti-RhD monoclonal antibody (MAb) through different processing modes, namely batch, fed-batch and perfusion operations. The ultimate objective of this study was to establish a comprehensive dynamic model which may be used for model-based optimization of the cell culture for MAb production in both batch, fed-batch or perfusion systems. In analyzing process performance, the key potential cause of cell growth inhibition was attributed to lowering of pH in the culture possibly due to the accumulation of dissolved carbon dioxide. The most important finding in this regard was the significantly different observed maximum total viable cell density in two identical cultures differing in culture volume only (250mL and 500mL). However, the other byproduct metabolites such as lactate and ammonia and glucose depletion were also capable affecting growth adversely causing growth arrest, viability reduction, apoptosis initiation and progress. Employing the experimental results of nutrient consumption, metabolite and biomass production, a metabolic flux based methodology was developed for modeling the metabolism of a CHO cell line. The elimination of insignificant fluxes resulted in a simplified metabolic network which was the basis for modeling the significant extracellular metabolites. Using kinetic rate expressions for growing and non-growing subpopulations, a logistic model was first developed for cell growth and dynamic models were formulated to describe culture composition and monoclonal antibody (MAb) secretion. The viable cell population was assumed to consist of normal growing, normal non-growing and apoptotic cell subpopulations. The rate of apoptotic cell formation was assumed to have a second order dependence on the normal cell concentration. The proposed mathematical model for metabolites included distinct terms that reflected the metabolic rates of growing and non-growing cell populations. The model was validated for a range of glutamine and glucose concentrations. Good agreement was obtained between model predictions and experimental data. In subsequent steps the attempt was to correlate the growth kinetics to significant variables of the culture. The regulatory effects identified through each culture condition were combined for a rational design of a dynamic model constructed for the viable cell subpopulation. A Tessier-based model was applied for defining the fraction of growing cells as a function of a growth inhibitor, presumably dissolved carbon dioxide. Although only few variables appeared in the biomass model, all equations were solved simultaneously. The parameters were estimated using the Metropolis-Hastings algorithm and the fmincon function in MATLAB. The final model adequately predicted the effect of significant variables on the metabolic behavior of CHO cells in batch, fed-batch and perfusion systems.

The creation of stable mammalian cell lines for biopharmaceutical production often require several months, and is unfavourable for the rapid production of multiple drug candidates for screening in the early stages of development. Biopharmaceutical production by transient transfection provides a possible alternative of quickly producing these early stage drug candidates. The Epi-CHO transient expression system, which consists of a Chinese hamster ovary (CHO) cell line (CHO-T) expressing the murine polyomavirus Large T-Antigen (LT), emonstrated enhanced transient recombinant protein production. The aim of this study was to prolong transient recombinant protein prod.Jction of the Epi-CHO expression system by creating a CHO cell line expressing both LT and EBNA1 (ECHO-T). The pEBNA1-LT expression vector encoding LT and EBNA1 was constructed and transfected into CHO-K1. A total of 20 clones were isolated from the antibioticresistant pool and screened for the expression of functional LT and EBNA1. PCR analysis showed 16 of the 20 clones was positive for EBNA1 and LT DNA. Of the 16 clones, six were positive for EBNA1 and LT expression by RT-PCR. Detection of LT and EBNA1 by immunofluorescence showed positive staining for the P7-G3 clone. Western blotting suggested the P7-G3 clone was: positive for EBNA1, and clones P3-C7 and P7-E2 were positive for LT. A plasmid replication assay confirmed the expression of functional LT in all six clones. Plasmid maintenance assay confirmed clone P7-G3 as the ECHO-T clones to express functional EBNA1. The P7-G3 clone demonstrated prolonged and sustained transient recombinant protein expression when compared to CHO-T. The P7-G3 clone achieved sustained transient protein expression for 32 days in the absence of selection, the longest currently reported for CHO cells.

Stuttgart, Univ., Diss., 2008.

Doctora en Ciencias de la Ingeniería, Mención Ingeniería Química y Biotecnología / Las células animales son uno de los principales sistemas para la producción de biofármacos, sin embargo la mayoría de las estrategias para mejorar su productividad no están respaldadas por conocimiento específico para las líneas celulares usadas en la industria. En este trabajo se reconstruye un modelo a escala genómica para células CHO para ampliar el conocimiento de procesos celulares asociados con productividad en la síntesis de biofármacos. Para lograr este objetivo analizamos el modelo a escala genómica de ratón iMM1415 usando herramientas desarrolladas para explorar el efecto de knockout de genes y determinación de flujos en crecimiento celular. Nuestros resultados muestran que esta red metabólica está dominada por metabolismo de lípidos. Adicionalmente, confirmamos que un enfoque de sampleo, donde se explora el espacio de soluciones en vez de imponer un objetivo de optimización, es más apropiado para el estudio del metabolismo en sistemas complejos como células animales. Posteriormente, se desarrolla en estudio comparativo de las herramientas desarrolladas para la generación automática de modelos preliminares, donde los resultados obtenidos utilizando tres algoritmos (modelSEED, Pantograph y Pathway tools) muestran que Pantograph es la herramienta más apropiada para la generación de un modelo a escala genómica de células CHO. Este algoritmo produce un modelo basándose en un modelo previo y ortología entre ambos organismos, produciendo un modelo preliminar que hereda características como asociaciones de genes distintas para distintos organelos celulares lo que es crucial para potenciales aplicaciones de modelos para eucariontes. El modelo iNJ1301 para células CHO es reconstruido de acuerdo a la metodología propuesta basándose en iMM1415 y el modelo humano Recon 1. iNJ1301 tiene 3,709 reacciones asociadas a 1,301 genes y fue validado con información experimental para esta línea celular prediciendo correctamente el crecimiento celular en un 88% de los casos simulados. Adicionalmente, este modelo es reducido imponiendo cambios en expresión de genes reportados para representar un metabolismo ineficiente del carbono caracterizado por síntesis de lactato y el shift metabólico observado en esta línea celular, mostrando el potencial de este enfoque para ser usado en la integración de datos transcriptómicos. Al utilizar un nuevo enfoque basado en ortología se pudieron encontrar nuevas asociaciones de genes no incluídas en reconstrucciones creadas utilizando metodologías clásicas para la generación de modelos, por lo que los resultados de este trabajo fueron incorporadas en el modelo consenso iCHO1766. La identificación de marcadores asociados a productividad mejorada en células CHO ha sido abordada desde la perspectiva genómica, transcriptómica y proteómica. En este trabajo integramos datos transcriptómicos para dos clones de células CHO que muestran distintos perfiles de productividad en el modelo iNJ1301. Datos de un alto (HP) y bajo (LP) productor de IgG son integrados al modelo usando iMAT (integrative metabolic analysis tool) obteniéndose modelos reducidos que presentan una alta conservación de vías metabólicas de glutatión y azúcares nucleotídicos. Este enfoque es luego acoplado con sampleo de las redes metabólicas encontrándose que ambos modelos presentan comportamientos distintos, donde el clon HP está enfocado en el uso del ciclo del TCA y la vía pentosas fosfato. Finalmente, concluimos que este enfoque que combina distintas herramientas utilizadas en biología de sistemas es una nueva herramienta que permite el estudio exhaustivo de sistemas biológicos complejos tales como líneas celulares animales.