Improving protein yield from mammalian cells by manipulation of stress response pathways

Monoclonal antibodies are a class of therapeutic that is an expanding area of the lucrative biopharmaceutical industry. These complex proteins are predominantly produced from large cultures of mammalian cells; the industry standard cell line being Chinese Hamster Ovary (CHO) cells. A number of optimisation strategies have led to antibody titres from CHO cells increasing by a hundred-fold, and it has been proposed that a further bottleneck in biosynthesis is in protein folding and assembly within the secretory pathway. To alleviate this bottleneck, a CHO-derived host cell line was generated by researchers at the pharmaceutical company UCB that stably overexpressed two critical genes: XBP1, a transcription factor capable of expanding the endoplasmic reticulum and upregulating protein chaperones; and Ero1α, an oxidase that replenishes the machinery of disulphide bond formation. This host cell line, named CHO-S XE, was confirmed to have a high yield of secreted antibody. The work presented in this thesis further characterises CHO-S XE, with the aim of using the information gained to lead the generation of novel host cell lines with more optimal characteristics than CHO-S XE. In addition to antibodies, it was found that CHO-S XE had improved production of two other secreted proteins: one with a simple tertiary structure and one complex multi-domain protein; and higher levels of a number of endogenous protein chaperones. As a more controlled system of gene expression to unravel the specific roles of XBP1 and Ero1α in the secretory properties of CHO-S XE, CHO cells with inducible overexpression of XBP1, Ero1α, or a third gene involved in the Unfolded Protein Response, GADD34, were generated. From these cell lines, it was shown that more antibody was secreted by cells with induced overexpression of XBP1; however, Ero1α and GADD34 overexpression did not improve antibody yield. Further investigation revealed that endogenous XBP1 splicing was downregulated in the presence of an abundance of the active form of XBP1. This result indicated a novel aspect of the regulation of the activity of IRE1, the stress-induced endoribonuclease responsible for XBP1 splicing. Overall, the work described in this thesis confirms that the overexpression of XBP1 has an enhancing effect on the secretory properties of CHO cells; information which could contribute to the development of host cells with a greater capacity for antibody production.

Host cellular regulatory networks in dengue virus-human interactions

Dengue fever is one of the most important mosquito-borne diseases worldwide and is caused by infection with dengue virus (DENV). The disease is endemic in tropical and sub-tropical regions and has increased remarkably in the last few decades. At present, there is no antiviral or approved vaccine against the virus. Treatment of dengue patients is usually supportive, through oral or intravenous rehydration, or by blood transfusion for more severe dengue cases. Infection of DENV in humans and mosquitoes involves a complex interplay between the virus and host factors. This results in regulation of numerous intracellular processes, such as signal transduction and gene transcription which leads to progression of disease. To understand the mechanisms underlying the disease, the study of virus and host factors is therefore essential and could lead to the identification of human proteins modulating an essential step in the virus life cycle. Knowledge of these human proteins could lead to the discovery of potential new drug targets and disease control strategies in the future. Recent advances of high throughput screening technologies have provided researchers with molecular tools to carry out investigations on a large scale. Several studies have focused on determination of the host factors during DENV infection in human and mosquito cells. For instance, a genome-wide RNA interference (RNAi) screen has identified host factors that potentially play an important role in both DENV and West Nile virus replication (Krishnan et al. 2008). In the present study, a high-throughput yeast two-hybrid screen has been utilised in order to identify human factors interacting with DENV non-structural proteins. From the screen, 94 potential human interactors were identified. These include proteins involved in immune signalling regulation, potassium voltage-gated channels, transcriptional regulators, protein transporters and endoplasmic reticulum-associated proteins. Validation of fifteen of these human interactions revealed twelve of them strongly interacted with DENV proteins. Two proteins of particular interest were selected for further investigations of functional biological systems at the molecular level. These proteins, including a nuclear-associated protein BANP and a voltage-gated potassium channel Kv1.3, both have been identified through interaction with the DENV NS2A. BANP is known to be involved in NF-kB immune signalling pathway, whereas, Kv1.3 is known to play an important role in regulating passive flow of potassium ions upon changes in the cell transmembrane potential. This study also initiated a construction of an Aedes aegypti cDNA library for use with DENV proteins in Y2H screen. However, several issues were encountered during the study which made the library unsuitable for protein interaction analysis. In parallel, innate immune signalling was also optimised for downstream analysis. Overall, the work presented in this thesis, in particular the Y2H screen provides a number of human factors potentially targeted by DENV during infection. Nonetheless, more work is required to be done in order to validate these proteins and determine their functional properties, as well as testing them with infectious DENV to establish a biological significance. In the long term, data from this study will be useful for investigating potential human factors for development of antiviral strategies against dengue.

The regulation of AMP-activated protein kinase by vascular endothelial growth factor

The function of the vascular endothelium is to maintain vascular homeostasis, by providing an anti-thrombotic, anti-inflammatory and vasodilatory interface between circulating blood and the vessel wall, meanwhile facilitating the selective passage of blood components such as signaling molecules and immune cells. Dysfunction of the vascular endothelium is implicated in a number of pathological states including atherosclerosis and hypertension, and is thought to precede atherogenesis by a number of years. Vascular endothelial growth factor A (VEGF) is a crucial mitogenic signaling molecule, not only essential for embryonic development, but also in the adult for regulating both physiological and pathological angiogenesis. Previous studies by our laboratory have demonstrated that VEGF-A activates AMP-activated protein kinase (AMPK), the downstream component of a signaling cascade important in the regulation of whole body and cellular energy status. Furthermore, studies in our laboratory have indicated that AMPK is essential for VEGF-A-stimulated vascular endothelial cell proliferation. AMPK activation typically stimulates anabolic processes and inhibits catabolic processes including cell proliferation, with the ultimate aim of redressing energy imbalance, and as such is an attractive therapeutic target for the treatment of obesity, metabolic syndromes, and type 2 diabetes. Metabolic diseases are associated with adverse cardiovascular outcomes and AMPK activation is reported to have beneficial effects on the vascular endothelium. The mechanism by which VEGF-A stimulates AMPK, and the functional consequences of VEGF-A-stimulated AMPK activation remain uncertain. The present study therefore aimed to identify the specific mechanism(s) by which VEGF-A regulates the activity of AMPK in endothelial cells, and how this might differ from the activation of AMPK by other agents. Furthermore, the role of AMPK in the pro-proliferative actions of VEGF-A was further examined. Human aortic and umbilical vein endothelial cells were therefore used as a model system to characterise the specific effect(s) of VEGF-A stimulation on AMPK activation. The present study reports that AMPK α1 containing AMPK complexes account for the vast majority of both basal and VEGF-A-stimulated AMPK activity. Furthermore, AMPK α1 is localized to the endoplasmic reticulum when sub-confluent, but translocated to the Golgi apparatus when cells are cultured to confluence. AMPK α2 appears to be associated with a structural cellular component, but neither α1 nor α2 complexes appear to translocate in response to VEGF-A stimulation. The present study confirms previous reports that when measured using the MTS cell proliferation assay, AMPK is required for VEGF-A-stimulated endothelial cell proliferation. However, parallel experiments measuring cell proliferation using the Real-Time Cell Analyzer xCELLigence system, do not agree with these previous reports, suggesting that AMPK may in fact be required for an aspect of mitochondrial metabolism which is enhanced by VEGF-A. Studies into the mitochondrial activity of endothelial cells have proved inconclusive at this time, but further studies into this are warranted. During previous studies in our laboratory, it was suggested that VEGF-A-stimulated AMPK activation may be mediated via the diacylglycerol (DAG)-sensitive transient receptor potential cation channel (TRPCs -3, -6 or -7) family of ion channels. The present study can neither confirm, nor exclude the expression of TRPCs in vascular endothelial cells, nor rule out their involvement in VEGF-A-stimulated AMPK activation; more specific investigative tools are required in order to characterise their involvement. Furthermore, nicotinic acid adenine dinucleotide phosphate (NAADP)-stimulated Ca2+ release from acidic intracellular organelles is not required for AMPK activation by VEGF-A. Despite what is known about the mechanisms by which AMPK is activated, far less is known concerning the downregulation of AMPK activity, as observed in human and animal models of metabolic disease. Phosphorylation of AMPK α1 Ser485 (α2 Ser491) has recently been characterised as a mechanism by which the activity of AMPK is negatively regulated. We report here for the first time that VEGF-A stimulates AMPK α1 Ser485 phosphorylation independently of the previously reported AMPK α1 Ser485 kinases Akt (protein kinase B) and ERK1/2 (extracellular signal-regulated kinase 1/2). Furthermore, inhibition of protein kinase C (PKC), the activity of which is reported to be elevated in metabolic disease, attenuates VEGF-A- and phorbol 12-myristate 13-acetate (PMA)-stimulated AMPK α1 Ser485 phosphorylation, and increases basal AMPK activity. In contrast to this, PKC activation reduces AMPK activity in human vascular endothelial cells. Attempts to identify the PKC isoform responsible for inhibiting AMPK activity suggest that it is one (or more) of the Ca2+-regulated DAG-sensitive isoforms of PKC, however cross regulation of PKC isoform expression has limited the present study. Furthermore, AMPK α1 Ser485 phosphorylation was inversely correlated with human muscle insulin sensitivity. As such, enhanced AMPK α1 Ser485 phosphorylation, potentially mediated by increased PKC activation may help explain some of the reduced AMPK activity observed in metabolic disease.

A biochemical analysis into the co-translational folding of a G protein-coupled receptor; GPR35

The folding and targeting of membrane proteins poses a major challenge to the cell, as they must remain insertion competent while their highly hydrophobic transmembrane (TM) domains are transferred from the ribosome, through the aqueous cytosol and into the lipid bilayer. The biogenesis of a mature membrane protein takes place through the insertion and integration into the lipid bilayer. A number of TM proteins have been shown to gain some degree of secondary structure within the ribosome tunnel and to retain this conformation throughout maturation. Although studies into the folding and targeting of a number of membrane proteins have been carried out to date, there is little information on one of the largest class of eukaryotic membrane proteins; the G-protein-coupled receptors (GPCRs). This project studies the early folding events of the human ortholog of GPR35. To analyse the structure of the 1st TM domain, intermediates were generated and assessed by the biochemical method of pegylation (PEG-MAL). A structurally-similar microbial opsin (Bacterioopsin) was also used to investigate the differences in the early protein folding within eukaryotic and prokaryotic translation systems. Results showed that neither the 1st TM domain of GPR35 nor Bacterioopsin were capable of compacting in the ribosome tunnel before their N-terminus reached the ribosome exit point. The results for this assay remained consistent whether the proteins were translated in a eukaryotic or prokaryotic translation system. To examine the communication mechanism between the ribosome, the nascent chain and the protein targeting pathway, crosslinking experiments were carried out using the homobifunctional lysine cross-linker BS3. Specifically, the data generated here show that the nascent chain of GPR35 reaches the ribosomal protein uL23 in an extended conformation and interacts with the SRP protein as it exits the ribosome tunnel. This confirms the role of SRP in the co-translational targeting of GPR35. Using these methods insights into the early folding of GPCRs has been obtained. Further experiments using site-directed mutagenesis to reduce hydrophobicity in the 1st TM domain of GPR35, highlighted the mechanisms by which GPCRs are targeted to the endoplasmic reticulum. Confirming that hydrophobicity within the signal anchor sequence is essential of SRP-dependent targeting. Following the successful interaction of the nascent GPR35 and SRP, GPR35 is successfully targeted to ER membranes, shown here as dog pancreas microsomes (DPMs). Glycosylation of the GPR35 N-terminus was used to determine nascent chain structure as it is inserted into the ER membrane. These glycosylation experiments confirm that TM1 has obtained its compacted state whilst residing in the translocon. Finally, a site-specific cross-linking approach using the homobifunctional cysteine cross-linker, BMH, was used to study the lateral integration of GPR35 into the ER. Cross-linking of GPR35 TM1 and TM2 could be detected adjacent to a protein of ~45kDa, believed to be Sec61α. The loss of this adduct, as the nascent chain extends, showed the lateral movement of GPR35 TM1 from the translocon was dependent on the subsequent synthesis of TM2.

An investigation into the folding and assembly of engineered antibodies and novel antibody formats

Monoclonal antibodies and novel antibody formats are currently one of the principal therapeutic in the biopharmaceutical industry worldwide and are widely used in the treatment of autoimmune diseases and cancer. It is for this reason that the productivity and quality of antibody production requires improvement; specifically investigations into the engineering of antibodies and any issues that may arise from the production of these therapeutics. The work presented in this thesis describes an investigation into the folding and assembly of seven antibodies plus the novel antibody format FabFv. IgG is comprised of two identical HCs and two identical LCs. The folding process of immunoglobulin is controlled by the CH1 domain within the HC. The CH1 domain remains in a disordered state and is sequestered by BiP in the endoplasmic reticulum. Upon the addition of a folded CL domain, BiP is displaced, the CH1 domain is able to fold and the complete IgG protein can then be secreted from the cell. The results presented in this thesis however, have outlined an additional mechanism for the folding of the CH1 domain. We have shown that the CH1 domain is able to fold in the absence of LC resulting in the secretion of HC dimers in a VH dependent manner. The proposed mechanism for the secretion of HC dimers suggests that some VH domains can interact with each other in order to bring the CH1 domains in close proximity to enable folding to occur. As HC dimer secretion is a hindrance in antibody production, this result has highlighted an engineering target to improve antibody yield. Examination of the folding of IgG4 with the variable region A33 has revealed the inability to secrete LC dimers, cleavage of the HC during expression and secretion of HC dimers in the Fab, FabFv and full length forms. The attributes described have also been shown to be variable region dependent. This has introduced a new concept that the variable domain is important in determining the expression and secretion of antibodies and their individual chains. Pulse chase and 2D gel electrophoresis analysis of the novel antibody format FabFv has revealed that the folding and expression of the LC and HC causes multimeric species of FabFv to be secreted, as opposed to the monomeric form which is the desired therapeutic. Our hypothesis is that this process occurs via a LC dependent mechanism. The proposed hypothesis suggests that further engineering to the LC could diminish the formation and secretion of FabFv multimers. The results from these investigations can be applied to increase the productivity of therapeutics and increase the biological understanding of the domain interactions of IgG during folding, assembly and secretion.