Systems Biology of Protein Secretion in Human Cells: Multi-omics Analysis and Modeling of the Protein Secretion Process in Human Cells and its Application.
Doktorsavhandling, 2021
Since the emergence of modern biotechnology, the production of recombinant pharmaceutical proteins has been an expanding field with high demand from industry. Pharmaceutical proteins have constituted the majority of top-selling drugs in the pharma industry during recent years. Many of these proteins require post-translational modifications and are therefore produced using mammalian cells such as Chinese Hamster Ovary cells. Despite frequent improvements in developing efficient cell factories for producing recombinant proteins, the natural complexity of the protein secretion process still poses serious challenges for the production of some proteins at the desired quantity and accepted quality. These challenges have been intensified by the growing demands of the pharma industry to produce novel products with greater structural complexity, as well as increasing expectations from regulatory authorities in the form of new quality control criteria to guarantee product safety.
This thesis focuses on different aspects of the protein secretion process, including its engineering for cell factory development and analysis in diseases associated with its deregulation. A major part of this thesis involved the use of HEK293 cells as a human model cell-line for investigating the protein secretion process by generating different types of omics data and developing a computational model of the human protein secretion pathway. We compared the transcriptomic profile of cell lines producing erythropoietin (EPO; as a model secretory protein) at different rates to identify key genes that potentially contributed to higher rates of protein secretion. Moreover, by performing a transcriptomic comparison of cells producing green fluorescent protein (GFP; as a model non-secretory protein) with EPO producers, we captured differences that specifically relate to secretory protein production. We sought to further investigate the factors contributing to increased recombinant protein production by analyzing additional omic layers such as proteomics and metabolomics in cells that exhibited different rates of EPO production. Moreover, we developed a toolbox (HumanSec) to extend the reference human genome-scale metabolic model (Human1) to encompass protein-specific reactions for each secretory protein detected in our proteomics dataset. By generating cell-line specific protein secretion models and constraining the models using metabolomics data, we could predict the top host cell proteins (HCPs) that compete with EPO for metabolic and energetic resources. Finally, based on the detected patterns of changes in our multi-omics investigations combined with a protein secretion sensitivity analysis using the metabolic model, we identified a list of genes and pathways that potentially play a key role in recombinant protein production and could serve as promising candidates for targeted cell factory design.
In another part of the thesis, we studied the link between the expression profiles of genes involved in the protein secretory pathway (PSP) and various hallmarks of cancer. By implementing a dual approach involving differential expression analysis and eight different machine learning algorithms, we investigated the expression changes in secretory pathway components across different cancer types to identify PSP genes whose expression was associated with tumor characteristics. We demonstrated that a combined machine learning and differential expression approach have a complementary nature and could highlight key PSP components relevant to features of tumor pathophysiology that may constitute potential therapeutic targets.
This thesis focuses on different aspects of the protein secretion process, including its engineering for cell factory development and analysis in diseases associated with its deregulation. A major part of this thesis involved the use of HEK293 cells as a human model cell-line for investigating the protein secretion process by generating different types of omics data and developing a computational model of the human protein secretion pathway. We compared the transcriptomic profile of cell lines producing erythropoietin (EPO; as a model secretory protein) at different rates to identify key genes that potentially contributed to higher rates of protein secretion. Moreover, by performing a transcriptomic comparison of cells producing green fluorescent protein (GFP; as a model non-secretory protein) with EPO producers, we captured differences that specifically relate to secretory protein production. We sought to further investigate the factors contributing to increased recombinant protein production by analyzing additional omic layers such as proteomics and metabolomics in cells that exhibited different rates of EPO production. Moreover, we developed a toolbox (HumanSec) to extend the reference human genome-scale metabolic model (Human1) to encompass protein-specific reactions for each secretory protein detected in our proteomics dataset. By generating cell-line specific protein secretion models and constraining the models using metabolomics data, we could predict the top host cell proteins (HCPs) that compete with EPO for metabolic and energetic resources. Finally, based on the detected patterns of changes in our multi-omics investigations combined with a protein secretion sensitivity analysis using the metabolic model, we identified a list of genes and pathways that potentially play a key role in recombinant protein production and could serve as promising candidates for targeted cell factory design.
In another part of the thesis, we studied the link between the expression profiles of genes involved in the protein secretory pathway (PSP) and various hallmarks of cancer. By implementing a dual approach involving differential expression analysis and eight different machine learning algorithms, we investigated the expression changes in secretory pathway components across different cancer types to identify PSP genes whose expression was associated with tumor characteristics. We demonstrated that a combined machine learning and differential expression approach have a complementary nature and could highlight key PSP components relevant to features of tumor pathophysiology that may constitute potential therapeutic targets.
Erythropoietin
cancer protein secretory pathway
protein secretion modeling
HEK293
protein secretion
Integrative omics analysis
genome-scale modeling
To be held online through Zoom
Opponent: Prof. Lars Keld Nielsen, The University of Queensland, Australia
Författare
Rasool Saghaleyni
Chalmers, Biologi och bioteknik, Systembiologi
Machine learning-based investigation of the cancer protein secretory pathway
PLoS Computational Biology,;Vol. 17(2021)
Artikel i vetenskaplig tidskrift
Evolution from adherent to suspension: systems biology of HEK293 cell line development
Scientific Reports,;Vol. 10(2020)
Artikel i vetenskaplig tidskrift
Ämneskategorier
Biologiska vetenskaper
Infrastruktur
C3SE (Chalmers Centre for Computational Science and Engineering)
ISBN
978-91-7905-517-2
Doktorsavhandlingar vid Chalmers tekniska högskola. Ny serie: 4984
Utgivare
Chalmers
To be held online through Zoom
Opponent: Prof. Lars Keld Nielsen, The University of Queensland, Australia