Reductionist and Integrative approaches to explore the H.pylori genome

The reductionist approach of decomposing biological systems into their constituent parts has dominated molecular biology for half a century. Since organisms are composed solely of atoms and molecules without the participation of extraneous forces, it has been assumed that it should be possible to explain biological systems on the basis of the physico-chemical properties of their individual components, down to the atomic level. However, despite the remarkable success of methodological reductionism in analyzing individual cellular components, it is now generally accepted that the behavior of complex biological systems cannot be understood by studying their individual parts in isolation. To tackle the complexity inherent in understanding large networks of interacting biomolecules, the integrative viewpoint emphasizes cybernetic and systems theoretical methods, using a combination of mathematics, computation and empirical observation. Such an approach is beginning to become feasible in prokaryotes, combining an almost complete view of the genome and transcriptome with a reasonably extensive picture of the proteome. Pathogenic bacteria are undoubtedly the most investigated subjects among prokaryotes. A paradigmatic example is the the human pathogen H.pylori, a causative agent of severe gastroduodenal disorders that infects almost half of the world population. In this thesis, we investigated various aspects of Helicobacter pylori molecular physiology using both reductionist and integrative approaches. In Section I, we have employed a reductionist, bottom-up perspective in studying the Cysteine oxidised/reduced state and the disulphide bridge pattern of an unusual GroES homolog expressed by H.pylori, Heat Shock protein A (HspA). This protein possesses a high Cys content, is involved in nickel binding and exhibits an extended subcellular localization, ranging from cytoplasm to cell surface. We have produced and characterized a recombinant HspA and mutants Cys94Ala and C94A/C111A. The disulphide bridge pattern has been assigned by integrating biochemical methodologies with mass spectrometry. All Cys are engaged in disulphide bonds that force the C-term domain to assume a peculiar closed loop structure, prone to host nickel ions. This novel Ni binding structural arrangement can be related to the Ni uptake/delivery to the extracellular urease, essential for the bacterium survival. In Section II, we combined different computational methods with two main goals: 1) Analyze the H.pylori biomolecular interaction network in an attempt to select new molecular targets against H.pylori infection (Chapters 4 & 5); 2) Model and simulate the signaling perturbations induced by invading H.pylori proteins in the host ephitelial cells (Chapter 6). Chapter 4 explores the 'robust yet fragile' feature of the H.pylori cell, viewed as a complex system in which robustness in response to certain perturbation is inevitably associated with fragility in response to other perturbations. With this in mind, we developed a general strategy aimed at identify control points in bacterial metabolic networks, which could be targets for novel drugs. The methodology is implemented on Helicobacter pylori 26695. The entire metabolic network of the pathogen is analyzed to find biochemically critical points, e.g. enzymes which uniquely consume and/or produce a certain metabolite. Once identified, the list of critical enzymes is filtered in order to find candidate targets wich are non-homologous with the human enzymes. Finally, the essentiality of the identified targets is cross-validated by in silico deletion studies using flux-balance analysis (FBA) on a recent genome-scale metabolic model of H. pylori. Following this approach, we identified some enzymes which could be interesting targets for inhibition studies of H.pylori infection. The study reported in Chapter 5 extends the previously described approach in light of recent theoretical studies on biological networks. These studies suggested that multiple weak attacks on selected targets are inevitably more efficient than the knockout of a single target, thus providing a conceptual framework for the recent success of multi-target drugs. We used this concept to exploit H.pylori metabolic robustness through multiple weak attacks on selected enzymes, therefore directing us toward target-sets discovery for combinatorial therapies. We used the known metabolic and protein interaction data to build an integrated biomolecular network of the pathogen. The network was subsequently screened to find central elements of network communication, e.g. hubs, bridges with high betweenness centrality and overlaps of network communities. The selected enzymes were then classified on the basis of available data about cellular function and essentiality in an attempt to predict successful target-combinations. In order to evaluate the network effect triggered by the partial inactivation of candidate targets, robustness analysis was performed on small groups of selected enzymes using flux balance analysis (FBA) on a recent genome-scale metabolic model of H.pylori. In particular, the FBA simulation framework allowed to predict the growth phenotype associated to every partial inactivation set. The preliminary results obtained so far may help to restrict the initial target-pool in search of target-sets for novel combinatorial drugs against H.pylori persistence. However, our long-term goal is to better understand the indirect network effects that lie at the heart of multi-target drug action and, ultimately, how multiple weak hits can perturb complex biological systems. H.pylori produces various a cytotoxic protein, CagA, that interfere with a very important host signaling pathway, i.e. the epidermal growth factor receptor (EGFR) signaling network. EGFR signaling is one of the most extensively studied areas of signal transduction, since it regulates growth, survival, proliferation and differentiation in mammalian cells. In Chapter 6, we attempted to build an executable model of the EGFR-signaling core process using a process algebra approach. In the EGFR network, the core process is the heart of its underlying hour-glass architecture, as it plays a central role in downstream signaling cascades to gene expression through activation of multiple transcription factors. It consists in a dense array of molecules and interactions wich are tightly coupled to each other. In order to build the executable model, a small set of EGFR core molecules and their interactions is tentatively translated in a BetaWB model. BetaWB is a framework for modelling and simulating biological processes based on Beta-binders language and its stochastic extension. Once obtained, the computational model of the EGFR core process can be used to test and compare hypotheses regarding the principles of operation of the signaling network, i.e. how the EGFR network generates different responses for each set of combinatorial stimuli. In particular, probabilistic model checking can be used to explore the states and possible state changes of the computational model, whereas stochastic simulation (corresponding to the execution of the BetaWB model) may give quantitative insights into the dynamic behaviour of the system in response to different stimuli. Information from the above tecniques allows model validation through comparison within the experimental data available in the literature. The inherent compositionality of the process algebra modeling approach enables further expansion of the EGFR core model, as well as the study of its behavior under specific perturbations, such as invading H.pylori proteins. This latter aspect might be of great value for H.pylori pathogenesis research, as signaling through the EGF receptors is intricately involved in gastric cancer and in many other gastroduodenal diseases.