Biological networks - from signal processing in DNA-coded protein networks to non-local information processing in the cortex

This thesis focuses on systems biology methods to understand signal processing in biological networks. These systems exist in a complex, inherently interconnected space that is not readily quantifiable. To order the complexity of different systems, I use an approach grounded in the theory of emergence, which posits that a system's evolution can exceed the mere sum of its individual parts through interaction and communication, leading to increased processing power and environmental representation. Our strategy involves setting up a mathematical representation of biological systems using networks composed of nodes representing entities like nucleic acids, proteins, and neurons and edges representing interactions between them. These networks then facilitate systemic analyses integrating insights from various scientific fields, including molecular biology, neurobiology, physics, and computer science, thus fostering a deeper understanding of biological signal processing. During my thesis work, I established a fundamental approach for creating and evolving models of different biological networks. The initial step involves setting up a network model of interacting components (“topology”) based on the biological system’s morphology, followed by analyzing the reconstructed network topology (e.g., where are central components, modulatory inputs, key signal pathways, and integration points). The topology then informs space-time-dependent rules for dynamic simulations (who activates whom and whether a process is fast or slow) that are validated against observed biology. This iterative alignment process between simulation and experiment gradually refines the model, fostering a more accurate representation of the biological system in question. Importantly, a model improves understanding while enhancing the efficiency and efficacy of experiments, which has practical implications such as resource conservation and reduced reliance on model animals. The three chapters of the thesis focus on information processing at the DNA, protein, and cortical level. These processing domains are interconnected. DNA encodes information for RNA, which governs the signaling networks of proteins. At this level, we primarily studied the functions and behavior at the intracellular level and discussed the implications of system defects, pathology, and disease on information processing. However, the intracellular processing integrates into broader intercellular networks, such as cell-cell interaction and neural networks, eventually forming correlated units in cortical columns. The columns form the basis of the cortical network architecture associated with cognition, higher brain functions, and consciousness. As an outcome of our modeling approaches, we can study the system at different levels: zoom in to fine-tuned differences and zoom out to detect major drives. In detail, we could deduce and publish substantial and precise findings by integrating diverse information sources into large-scale network models and resulting simulations of biological systems (two first authors, two book chapters, five coauthorships, two preprints). Discoveries include identifying a silent SNP in the Ambra1 gene influencing RNA folding, which disturbed splicing, thus altering its expression level and protein function. However, large-scale system effects on protein networks, neuronal, and behavior levels remain obscured by emergent complexity. To comprehend the causality of the DNA effects, we need an understanding of the processing of the following emergent levels. Therefore, we established a framework for predicting and validating protein interactions within species and at species intersection. Using this framework for reconstructing fungal interactions during host invasion and evasion, I could elucidate key interactions during the immune evasion of the fungi, A. fumigatus, within the phagolysosome. We got insights into how the fungi manipulate the host, e.g., prevent phagolysosome maturation and the host’s counteractions. Furthermore, we have identified drugs potentially disrupting the critical interactions between the host and the pathogen, thereby expanding our arsenal of antifungal strategies. We were validating the pivotal interactions in subsequent experiments and are presently assessing the efficacy of the drugs in infection models. Moreover, I show that the modeling framework, together with the large-scale integration of proteomics and transcriptomics, is effective for elucidating significant differences in platelet signaling of mice and men. This holds considerable importance as mice serve as invaluable model organisms for thrombosis (causes of stroke and heart attack) and inflammatory research. Beyond that, the thesis elucidates the dynamic functionalities of blood platelets, offering insights into platelet-related diseases and showcasing how platelets make rapid decisions in complex environments. Likewise, I explored the network dynamics underlying megakaryocyte (mother cells of platelets) behavior during transendothelial migration and its implications for conditions such as thrombocytopenia. Investigating the reconstructed networks and dynamics allows for obtaining fundamental mathematical principles of how cells solve environmental complexity problems in general. One such observation was that fast-changing and highly variable nodes within feedback loops are indispensable for dynamic behavior such as cell migration or cell-cell communication. Here, fast dynamics might give rise to oscillatory signal processing, which demands different decoding strategies but offers more degrees of freedom and the possibility of establishing a new information processing paradigm. We explored this new domain of information processing by establishing a network model built on the principles of feedback loops and oscillators, aiming to simulate cortical dynamics. I could show that this oscillatory network model adds a new emergence level to local, modular information processing as information is processed non-locally and is represented holistically within a wave interference pattern. By adapting the model to cortical scales, we could show that it faithfully replicates electrophysiological characteristics of healthy and diseased brains and promises to elevate our understanding of disorders like schizophrenia, Alzheimer's, and epilepsy. Here, the model provides a solid theoretical framework for understanding the fundamental processes of neural tissues. Further analyses of the cortical model show that it could be suitable for maximizing information integration, a defining property that might be important for facilitating the development of consciousness. These profound insights into cortical systems present a fresh perspective on cortical processing and its evolution (1st author paper submitted). They agree with electrophysiological data and should now be followed up further by even more experiments. In conclusion, this thesis outlines the general principles of biological information processing and underlines the vital role of information integration in the intricate dance of life, environment, adaptive processing, and cognition. The analysis of the different emergent levels reveals a viable path for the evolution of networks, and increasing information integration appears to be a key signature. Biological networks may facilitate survival through increasing information integration to overcome inherent system limitations in interfacing with complex environments to improve the correspondence between sensing and actions.