Session K39: Evolutionary Design Principles of Bio-networks

Metabolic interactions and dynamics in microbial communities

Metabolism, in addition to being the engine of every living cell, plays a major role in the cell-cell and cell-environment relations that shape the dynamics and evolution of microbial communities, e.g. by mediating competition and cross-feeding interactions between different species. Despite the increasing availability of metagenomic sequencing data for numerous microbial ecosystems, fundamental aspects of these communities, such as the unculturability of many isolates, and the conditions necessary for taxonomic or functional stability, are still poorly understood. We are developing mechanistic computational approaches for studying the interactions between different organisms based on the knowledge of their entire metabolic networks. In particular, we have recently built an open source platform for the Computation of Microbial Ecosystems in Time and Space (COMETS), which combines metabolic models with convection-diffusion equations to simulate the spatio-temporal dynamics of metabolism in microbial communities. COMETS has been experimentally tested on small artificial communities, and is scalable to hundreds of species in complex environments. I will discuss recent developments and challenges towards the implementation of models for microbiomes and synthetic microbial communities.   

Gene Expression Noise, Fitness Landscapes, and Evolution

The stochastic (or noisy) process of gene expression can have fitness consequences for living organisms. For example, gene expression noise facilitates the development of drug resistance by increasing the time scale at which beneficial phenotypic states can be maintained [1]. The present work investigates the relationship between gene expression noise and the fitness landscape [2]. By incorporating the costs and benefits of gene expression, we track how the fluctuation magnitude and timescale of expression noise evolve in simulations of cell populations under stress. We find that properties of expression noise evolve to maximize fitness on the fitness landscape, and that low levels of expression noise emerge when the fitness benefits of gene expression exceed the fitness costs (and that high levels of noise emerge when the costs of expression exceed the benefits). The findings from our theoretical/computational work offer new hypotheses on the development of drug resistance, some of which are now being investigated in evolution experiments in our laboratory using well-characterized synthetic gene regulatory networks in budding yeast. [1] D.A. Charlebois, N. Abdennur, M. Kaern, Gene expression noise facilitates adaptation and drug resistance independently of mutation, Physical Review Letters, 107, 218101 (2011). [2] D.A. Charlebois, Effect and Evolution of Gene Expression Noise on the Fitness Landscape, Physical Review E, 022713 (2015).   

Modularity Enhances the Rate of Evolution in a Rugged Fitness Landscape

Biological systems are modular, and this modularity affects the evolution of biological systems over time and in different environments. We here develop a theory for the dynamics of evolution in a rugged, modular fitness landscape. We show analytically how horizontal gene transfer couples to the modularity in the system and leads to more rapid rates of evolution at short times. The model, in general, analytically demonstrates a selective pressure for the prevalence of modularity in biology. We use this model to show how the evolution of the influenza virus is affected by the modularity of the proteins that are recognized by the human immune system. A modular model of the fitness landscape of the virus better fits the observed virus evolution data.   

In-silico studies of neutral drift for functional protein interaction networks

We have developed a minimal physically-motivated model of protein-protein interaction networks. Our system consists of two classes of enzymes, activators (e.g. kinases) and deactivators (e.g. phosphatases), and the enzyme-mediated activation/deactivation rates are determined by sequence-dependent binding strengths between enzymes and their targets. The network is evolved by introducing random point mutations in the binding sequences where we assume that each new mutation is either fixed or entirely lost. We apply this model to studies of neutral drift in networks that yield oscillatory dynamics, where we start, for example, with a relatively simple network and allow it to evolve by adding nodes and connections while requiring that dynamics be conserved. Our studies demonstrate both the importance of employing a sequence-based evolutionary scheme and the relative rapidity (in evolutionary time) for the redistribution of function over new nodes via neutral drift. Surprisingly, in addition to this redistribution time we discovered another much slower timescale for network evolution, reflecting hidden order in sequence space that we interpret in terms of sparsely connected domains.   

Phase transitions in the evolution of gene regulatory networks

The role of gene regulatory networks is to respond to environmental conditions and optimize growth of the cell. A typical example is found in bacteria, where metabolic genes are activated in response to nutrient availability, and are subsequently turned off to conserve energy when their specific substrates are depleted. However, in fluctuating environmental conditions, regulatory networks could experience strong evolutionary pressures not only to turn the right genes on and off, but also to respond optimally under a wide spectrum of fluctuation timescales. The outcome of evolution is predicted by the long-term growth rate, which differentiates between optimal strategies. Here we present an analytic computation of the long-term growth rate in randomly fluctuating environments, by using mean-field and higher order expansion in the environmental history. We find that optimal strategies correspond to distinct regions in the phase space of fluctuations, separated by first and second order phase transitions. The statistics of environmental randomness are shown to dictate the possible evolutionary modes, which either change the structure of the regulatory network abruptly, or gradually modify and tune the interactions between its components.   

Genome-scale modeling of the evolutionary path to C4 photosynthesis

In C4 photosynthesis, plants maintain a high carbon dioxide level in specialized bundle sheath cells surrounding leaf veins and restrict CO$_2$ assimilation to those cells, favoring CO$_2$ over O$_2$ in competition for Rubisco active sites. In C3 plants, which do not possess such a carbon concentrating mechanism, CO$_2$ fixation is reduced due to this competition. Despite the complexity of the C4 system, it has evolved convergently from more than 60 independent origins in diverse families of plants around the world over the last 30 million years. We study the evolution of the C4 system in a genome-scale model of plant metabolism that describes interacting mesophyll and bundle sheath cells and enforces key nonlinear kinetic relationships. Adapting the zero-temperature string method for simulating transition paths in physics and chemistry, we find the highest-fitness paths connecting C3 and C4 positions in the model's high-dimensional parameter space, and show that they reproduce known aspects of the C3-C4 transition while making additional predictions about metabolic changes along the path. We explore the relationship between evolutionary history and C4 biochemical subtype, and the effects of atmospheric carbon dioxide levels.   

Precision matters for position decoding in the early fly embryo

Genetic networks can determine cell fates in multicellular organisms with precision that often reaches the physical limits of the system. However, it is unclear how the organism uses this precision and whether it has biological content. Here we address this question in the developing fly embryo, in which a genetic network of patterning genes reaches 1\% precision in positioning cells along the embryo axis. The network consists of three interconnected layers: an input layer of maternal gradients, a processing layer of gap genes, and an output layer of pair-rule genes with seven-striped patterns. From measurements of gap gene protein expression in hundreds of wild-type embryos we construct a ``decoder", which is a look-up table that determines cellular positions from the concentration means, variances and co-variances. When we apply the decoder to measurements in mutant embryos lacking various combinations of the maternal inputs, we predict quantitative changes in the output layer such as missing, altered or displaced stripes. We confirm these predictions by measuring pair-rule expression in the mutant embryos. Our results thereby show that the precision of the patterning network is biologically meaningful and a necessary feature for decoding cell positions in the early fly embryo.   

Enhancing Functional Robustness of Gene Regulatory Networks Based on Fitness Landscape Design

We aim to develop design principles for enhancing functional robustness of engineered cells using gene-network topology. We observed the effect of genetic regulation types (inhibition and activation) on robustness. Inhibition was much more stable than activation in E. coli. In the case of activation, if the upstream activator expression is shutdown by mutation, then its downstream expression is shut down as well. Without activation, the activator shutdown due to mutation will make its downstream expression “remains`` turned off. Thus, the change in the metabolic load is higher in the activation case. Therefore, the stronger activation, the less robust the circuits are. In the inhibition case, we found that the story becomes opposite. When an inhibitor expression is shut down by mutation, the downstream expression turns on because the inhibitor is not expressed. This compensates changes in the metabolic load that might have been decreased without the inhibition. This result presents potential significant roles of network topology on the robustness of engineered cellular networks. This also emphasizes that the concept of fitness landscape, where the local slope corresponds to the fitness difference between different genotypes, can be useful to design robust gene circuits.   

The ecology and evolution of microbial behavior in complex communities

Microbes form complex ecological communities with multiple species coexisting and interacting with each other. Often, the ecological interactions among these species are mediated by molecules that the microbes actively secrete to the outside world. A large number of microbes are decomposers, and thus particularly relevant examples of these secreted molecules are the enzymes that microbes use to break down complex organic matter (e.g. dead tissue) and extract nutrients from it. In this talk, I will present an overview of the work that we have done to understand the ecology and evolution of the genes responsible for the expression of these enzymes. In particular, I will discuss how by regulating the amount of investment in the production of extracellular enzymes, microbes may modulate ecological interactions and change the number and stability of equilibria in ecosystems.