Computational methods and modeling of cellular function for the optimization of protein production and the study of cellular disease states

Systems biology is a multidisciplinary field that weaves together all the basic sciences through the use of computational and bioinformatics tools, to provide a more integrative view of the complex molecular interactions taking place within and among cells. The successes in the development and improvement of techniques for high throughput âomics has rapidly increased the amount of available data. The complexity of the underlying biological system it describes requires the development of tools to properly process it and analyze it. Computational models mathematically describe the systems interactions, allowing intrinsic properties of the data to emerge that would otherwise be overlooked. These models provide context to the data and are used to make predictions about the behavior of the system and to simulate a broader landscape of hypothesis, saving the time and cost of performing numerous experiments. However, the number of required parameters to mathematically formulate the system increases with the model complexity to integrate the available data. Thus, the development of estimation procedures and workflows to retrieve these values from literature and databases become crucial. In this work, we developed a set of computational models, analysis tools, and pipelines to support the study of two biological systems crucial to the cell survival: metabolism and protein synthesis. Metabolism is responsible for the production of most cell biomass, including proteins. Together, these two systems balance the renewal of protein in the cell, where metabolism provides the amino acids obtained from protein breakdown to the mRNA translation machinery. Deregulation of these systems is known to cause multiple disorders, such as neurodegenerative diseases and cancer. In the study of protein synthesis, we employ a combination of deterministic and stochastic modeling approaches to understand its intrinsic mechanistic properties and its rate-limiting steps. A better understanding of the system properties can have a profound impact on the development of drug targets and, in particular, in the optimization of heterologous protein production. Our studies revealed that more than one factor plays a role in the speed of translation: competition for tRNA resources and the type of cognate binding interaction between tRNA and the mRNA-ribosome complex. We also derived an equation that, given the knowledge about certain intracellular parameters pertaining to the host organism of interest, can assist in the design of transcripts for optimizing heterologous protein production. For the study of human metabolism, we established a pipeline to generate tissue-specific reduced metabolic models that can be used to study the metabolic reprogramming of different cancers and compare it with the metabolic phenotype of a healthy cell type. Despite being presented herein for human models, this pipeline is general and can be applied to the models of any organism. Starting from a human genome scale metabolic model, the pipeline improves compound annotation, identification, thermodynamics parameter retrieval, and facilitates data integration through the connection of several compound databases in a semi-automatized fashion. This work sets a standard for metabolic model assessment and curation and improves on existing tools to generate the first thermodynamically feasible reduced model of human metabolism, which is specifically tailored to the physiology and conditions under study.

Identification and characterization of proteins supporting dehalorespiration in Desulfitobacterium hafniense strain TCE1

Laure Prat

Respiration is a fundamental catalytic process in the aerobic and anaerobic energy metabolism of many prokaryotic and most eukaryotic organisms. The major difference between these organisms is that various organic and inorganic substrates can be used to donate or accept electrons in the bacterial and archaeal kingdoms, in various ranges of redox potentials in order to drive aerobic (with oxygen as electron acceptor) and anaerobic respiratory pathways. During the last few decades, several bacterial strains have been reported to use chlorinated compounds as terminal electron acceptor under anaerobic conditions, in a process called dehalorespiration. Among more than thirty dehalorespiring bacterial strains isolated to date, the Dehalococcoides group and the Gram-positive genus Desulfitobacterium comprise the largest number of reductively dehalogenating pure cultures. While most Dehalococcoides isolates appeared as highly specialized bacteria that depend strictly on dehalorespiration for growth, members of the genus Desulfitobacterium are very versatile microorganisms with regards to the electron donors and acceptors utilized. Globally, this remarkable prokaryotic respiratory flexibility is guaranteed by an elaborate reservoir of complex redox enzymes. Regarding the specific electron transport chain involved in dehalorespiration, rather little knowledge has been obtained so far with the exception of the key enzyme, the reductive dehalogenase, well described in several dehalorespiring bacteria, and of little information on b-type cytochromes and menaquinones. The overall objective of this thesis was to identify new possible components of the electron transport chain by comparative proteomic analysis, to investigate the presence/absence of a regulatory pathway leading to the biosynthesis of the reductive dehalogenase, and finally to characterize newly identified proteins that might support the dehalorespiration process. A comparative 2D proteomic analysis was performed on Desulfitobacterium hafniense strain TCE1, a bacterium capable of using tetrachloroethene (PCE) as electron acceptor. Soluble and membrane-associated proteins were analyzed from cells cultivated on different couples of electron donors and acceptors, lactate – fumarate (LF), lactate – PCE (LP), hydrogen – fumarate (HF), and hydrogen – PCE (HP). Cells growing on LP or HP revealed 72 and 93 protein spots in membrane fraction, and 49 and 12 spots in soluble fraction that correspond to increasingly expressed proteins compared to their fumarate-grown counterparts. More than 80 proteins were identified by mass spectrometry analysis, among which the key enzyme in the dehalorespiration process, the PCE reductive dehalogenase (PceA), as well as interesting candidates which could support dehalorespiration. The possible role of these proteins was analyzed in further details. Involved in energy and carbon metabolisms, electron transfer flavoproteins (ETF) and proteins related to the Wood-Ljungdahl pathway of CO2 fixation were identified in HP growth conditions. Proteins associated to stress response such as the phage shock protein PspA, and to regulatory pathways, the transcriptional repressor CodY, were also observed and discussed. Although only slightly induced by PCE, an additional protein displaying homology to sulfur-transferases was found in a large abundance in the fraction of membrane-associated proteins. This protein was named PhsE, as the corresponding gene belongs to a gene operon with homology to the molybdoenzyme thiosulfate/polysulfide reductase operons (phs/psr) found in Salmonella typhimurium or Wolinella succinogenes. PhsE protein architecture resembles to the class of tandem-domain rhodaneses, with the particularity to contain two active-site cysteines. It is also characterized by the presence of a typical lipoprotein signal peptide (LSP), which anchors PhsE on the outside of the cytoplasmic membrane. In the genome sequence of D. hafniense strain Y51, phsE (DSY3892) is part of an apparent operon encoding one out of about sixty different members of the complex iron-sulfur molybdoenzyme (CISM) family. PhsA, the catalytic molybdoenzyme, displays significant sequence homology with the thiosulfate/polysulfide reductase (PhsA/PsrA) subfamily. So far no other CISM operon has been described with a sulfur-transferase encoding gene. Despite the numerous functions proposed for rhodanese in cyanide detoxification, Fe/S cluster formation, oxidative stress protection, sulfur mobilization and involvement in general sulfur metabolism, the question of the physiological role of PhsE remains open. The expression of the Phs complex appeared to be influenced by the addition of sulfide in the culture medium, generally used as reducing agent in anaerobic cultures. Sulfide is known to affect the intracellular redox status. Therefore, it is hypothesized that PhsE might be involved in the control of the redox balance of the cell, notably by scavenging potentially toxic reduced sulfur species. To support data obtained at the protein level during proteomic analysis, genes and the corresponding messenger RNA from D. hafniense strain TCE1 were studied. While in silico DNA sequence analysis did not reveal any obvious features regulating the transcription of pceA, puzzling results were obtained in proteomic analysis showing an almost complete absence of PceA in cells grown on fumarate. To characterize this phenomenon, starting from a routinely PCE-grown inoculum, strain TCE1 was repeatedly transferred on medium containing fumarate as electron acceptor. The pool of pceA gene in the total culture DNA decreased significantly already after fifteen successive sub-cultures on fumarate, suggesting a shift in strain TCE1 population upon relief of the PCE selection pressure. Consequently the level of pceA mRNA and PceA protein also followed the same trend. Secondly, a PCE pulse experiment was performed in an anaerobic continuous culture of strain TCE1 on fumarate. It revealed that PCE itself has no power of induction on genetic level, except for the positive effect observed on pspA transcripts which encodes a protein involved in cell membrane integrity. Finally the expression study of the phsFABCDE gene cluster revealed that all phs genes are co-transcribed, and most likely build an operon. Thus phsE seemed to be strictly coregulated with the phs operon, and not under the influence of another promoter. In conclusion, proteomic analysis enabled to investigate the general physiological status of Desulfitobacterium hafniense strain TCE1 during dehalorespiration. A tentative model of the dehalorespiration pathway is proposed in summarizing recent data obtained on the topology of the PCE reductive dehalogenase and on sequence similarity with proteins involved in other anaerobic respiratory pathways. Regarding the possible support of PhsE towards the dehalorespiration, the complicated chemistry of sulfur does not allow to clearly attribute a function to the Phs complex, but there are indications that it might build a 'sulfur' oxidoreductase which could catalyze the conversion of sulfide to polysulfide with concomitant electron transfer to the quinone pool. The sulfide oxidation activity might be useful to D. hafniense to scavenge toxic sulfur species, avoiding that they disturb the redox balance of the cell, and redox-sensitive enzymatic complexes.

EPFL2009

O6-alkylguanine-DNA alkyltransferase as a new tool for functional proteomics

Anke Arnold

Recently the sequencing project of the human genome has been completed. The interest of the post-sequencing-era is shifting nowadays towards the investigation of the biological function and localization of the encoded proteins. Proteins are essential parts of all living organisms and participate in every process in the living cell. Despite their importance, the detailed function of the majority of all proteins remains still unknown. Therefore high-throughput methods are needed for the systematic examination of proteins. The first part of this thesis describes the development of a novel functional protein microarray, taking advantage of the highly selective reaction of the human O6-alkylguanine-DNA alkyltransferase (AGT) with O6-benzylguanine derivatives. By functionalization of otherwise bioinert glass surfaces with the substrate O6-benzylguanine, AGT fusion proteins can be selectively and covalently immobilized on a solid surface (Figure S1). Due to the high selectivity of the covalent immobilization the main advantage of this AGT based approach is that fusion proteins can be immobilized directly out of cell extracts without prior purification. By analyzing a set of literature known protein-protein interactions, protein-small molecule interactions and posttranslational modifications we demonstrate the feasibility of this approach (Chapter 3). Figure S1: Principle of an AGT-based protein microarray in which x can be any protein of interest Furthermore, the development of surface materials suitable for functional protein microarrays is described. The generated surfaces allow immobilization of proteins in defined orientation and density. Intrinsically bioinert polymer brushes were generated in collaboration with the Laboratory of Polymers at the EPFL. For the production of these bioinert brush-surfaces, surface-initiated atom transfer radical polymerization (SI-ATRP) was used (Chapter 4). Additionally we addressed the importance of the orientation of immobilized proteins on solid surfaces. The immobilization of fusion proteins with random orientation on aldehyde-coated glass slides was compared with immobilization in a defined orientation on O6-benzylguanine functionalyzed polymer brushes. Subsequently, retention of activity of the proteins upon immobilization on the two different surfaces was analyzed. We demonstrated that a defined orientation of immobilized proteins is advantageous over random immobilization due to the fact that proteins, immobilized using the AGT approach, are attached in a defined orientation with their active site facing the solution phase, whereas random immobilization of proteins can result in hidden active sites (Chapter 5). In the second part of this thesis the problem of insolubility of recombinant expressed proteins is addressed. We analyzed the protein expression of a set of proteins, which were prior determined to be insoluble when expressed in E. coli. The proteins were expressed as AGT fusion proteins either in E. coli or M. smegmatis, with the goal to identify a suitable host for soluble protein expression with high yield. The AGT approach was exploited to screen for expression of soluble proteins and for the determination of the expression yields. We showed that fusion to AGT does not influence the solubility of the expressed proteins and that the expression yield in E. coli does not significantly differ from the yield in M. smegmatis, where proteins were expressed with less degradation. However, we assessed E. coli. to be the more suitable host for multi-parallel protein expression due to its easy handling and fast growth (Chapter 6). The gained knowledge of this project was transferred to the investigation of proteins from Mycobacterium tuberculosis, which is presented in the last part of this thesis. A better understanding of M. tuberculosis, with the goal of unveiling and validating new therapeutic targets, is an imperative need to improve the control and treatment of tuberculosis disease. We focused in particular on the bacterial two component adaptation systems (TCS). Since these systems do not exist in mammals they present attractive candidates for new drug targets. In order to gain a better understanding of these adaptation systems we focused on one specific TCS, SenX/RegX, one of the first identified and well established TCS, which is believed to be implicated in virulence. A radioactive in vitro phosphorylation assay was developed, which allowed the identification of two new in vitro substrates for histidine kinase SenX, namely DevR and NarL. These results confirm the hypothesis that TCS might not only consist of two components, but that cross-talk between different TCS exist in order to adapt rapidly to environmental changes.

EPFL2007

A genome-scale metabolic model of Saccharomyces cerevisiae that integrates expression constraints and reaction thermodynamics

Vassily Hatzimanikatis, Maria Masid Barcon, Ljubisa Miskovic, Mohammad Omid Oftadeh, Pierre Guy Rémy Salvy

Eukaryotic organisms play an important role in industrial biotechnology, from the production of fuels and commodity chemicals to therapeutic proteins. To optimize these industrial systems, a mathematical approach can be used to integrate the description of multiple biological networks into a single model for cell analysis and engineering. One of the most accurate models of biological systems include Expression and Thermodynamics FLux (ETFL), which efficiently integrates RNA and protein synthesis with traditional genome-scale metabolic models. However, ETFL is so far only applicable for E. coli. To adapt this model for Saccharomyces cerevisiae, we developed yETFL, in which we augmented the original formulation with additional considerations for biomass composition, the compartmentalized cellular expression system, and the energetic costs of biological processes. We demonstrated the ability of yETFL to predict maximum growth rate, essential genes, and the phenotype of overflow metabolism. We envision that the presented formulation can be extended to a wide range of eukaryotic organisms to the benefit of academic and industrial research. Formulating metabolic networks mathematically can help researchers study metabolic diseases and optimize the production of industrially important molecules. Here, the authors propose a framework that allows to model eukaryotic metabolism considering gene expression and thermodynamic constraints.