Session P39: Information Processing in Cellular Signaling and Gene Regulation

Towards a predictive theory for genetic regulatory networks

When cells respond to changes in the environment by regulating the expression levels of their genes, we often draw parallels between these biological processes and engineered information processing systems. One can go beyond this qualitative analogy, however, by analyzing information transmission in biochemical ``hardware'' using Shannon's information theory. Here, gene regulation is viewed as a transmission channel operating under restrictive constraints set by the resource costs and intracellular noise. We present a series of results demonstrating that a theory of information transmission in genetic regulatory circuits feasibly yields non-trivial, testable predictions. These predictions concern strategies by which individual gene regulatory elements, e.g., promoters or enhancers, read out their signals; as well as strategies by which small networks of genes, independently or in spatially coupled settings, respond to their inputs. These predictions can be quantitatively compared to the known regulatory networks and their function, and can elucidate how reproducible biological processes, such as embryonic development, can be orchestrated by networks built out of noisy components. Preliminary successes in the gap gene network of the fruit fly Drosophila indicate that a full ab initio theoretical prediction of a regulatory network is possible, a feat that has not yet been achieved for any real regulatory network. We end by describing open challenges on the path towards such a prediction.   

Theory of optimal information transmission in E. coli chemotaxis pathway

Bacteria live in complex microenvironments where they need to make critical decisions fast and reliably. These decisions are inherently affected by noise at all levels of the signaling pathway, and cells are often modeled as an input-output device that transmits extracellular stimuli (input) to internal proteins (channel), which determine the final behavior (output). Increasing the amount of transmitted information between input and output allows cells to better infer extracellular stimuli and respond accordingly. However, in contrast to electronic devices, the separation into input, channel, and output is not always clear in biological systems. Output might feed back into the input, and the channel, made by proteins, normally interacts with the input. Furthermore, a biological channel is affected by mutations and can change under evolutionary pressure. Here, we present a novel approach to maximize information transmission: given cell-external and internal noise, we analytically identify both input distributions and input-output relations that optimally transmit information. Using E. coli chemotaxis as an example, we conclude that its pathway is compatible with an optimal information transmission device despite the ultrasensitive rotary motors.   

BMP4 density gradient in disk-shaped confinement

We present a quantitative model that explains the scaling of BMP4 gradients during gastrulation and the recent experimental observation that geometric confinement of human embryonic stem cells is sufficient to recapitulate much of germ layer patterning. Based on a assumption that BMP4 diffusion rate is much smaller than the diffusion rate of it’s inhibitor molecules, our results confirm that the length-scale which defines germ layer territories does not depend on system size.   

Mechanical Feedback and Arrest in Gene Expression

The ability to watch biochemical events~at the single-molecule level has increasingly revealed that stochasticity plays a leading role in many biological phenomena. One important and well know example is the noisy, ``bursty'' manner of transcription. Recent experiments~have~revealed~relationships between the level and noise in gene expression hinting at deeper stochastic connections. In this talk we will discuss how the mechanical nature of transcription can explain this relationship and examine the limits that the physical aspects of transcription place on gene expression.~   

Information processing in multi-step signaling pathways

Information processing in complex signaling networks is limited by a high degree of variability in the abundance and activity of biochemical reactions (biological noise) operating in living cells. In this context, it is particularly surprising that many signaling pathways found in eukaryotic cells are composed of long chains of biochemical reactions, which are expected to be subject to accumulating noise and delayed signal processing. Here, we challenge the notion that signaling pathways are insulated chains, and rather view them as parts of extensively branched networks, which can benefit from a low degree of interference between signaling components. We further establish conditions under which this pathway organization would limit noise accumulation, and provide evidence for this type of signal processing in an experimental model of a calcium-activated MAPK cascade. These results address the long-standing problem of diverse organization and structure of signaling networks in live cells.   

Towards Predictive Modeling of Information Processing in Microbial Ecosystems With Quorum-Sensing Interactions

Bacteria communicate using external chemical signals in a process known as quorum sensing. However, the efficiency of this communication is reduced by both limitations on the rate of diffusion over long distances and potential interference from neighboring strains. Therefore, having a framework to quantitatively predict how spatial structure and biodiversity shape information processing in bacterial colonies is important, both for understanding the evolutionary dynamics of natural microbial ecosystems, and for the rational design of synthetic ecosystems with desired computational properties. As a first step towards these goals, we implement a reaction-diffusion model to study the dynamics of a LuxI/LuxR quorum sensing circuit in a growing bacterial population. The spatiotemporal concentration profile of acyl-homoserine lactone (AHL) signaling molecules is analyzed, and used to define a measure of physical and functional signaling network connectivity. From this, we systematically investigate how different initial distributions of bacterial populations influence the subsequent efficiency of collective long-range signal propagation in the population. We compare our results with known experimental data, and discuss limitations and extensions to our modeling framework.-/abstract-   

Thermodynamics of nuclear transport

Molecular transport across the nuclear envelope is important for eukaryotes for gene expression and signaling. Experimental studies have revealed that nuclear transport is inherently a nonequilibrium process and actively consumes energy. In this work we present a thermodynamics theory of nuclear transport for a major class of nuclear transporters that are mediated by the small GTPase Ran. We identify the molecular elements responsible for powering nuclear transport, which we term the ``Ran battery" and find that the efficiency of transport, measured by the cargo nuclear localization ratio, is limited by competition between cargo molecules and RanGTP to bind transport receptors, as well as the amount of NTF2 (i.e. RanGDP carrier) available to circulate the energy flow. This picture complements our current understanding of nuclear transport by providing a comprehensive thermodynamics framework to decipher the underlying biochemical machinery.   

Vector Encoding in Biochemical Networks

Encoding of environmental cues via biochemical signaling pathways is of vital importance in the transmission of information for cells in a network. The current literature assumes a single cell state is used to encode information, however, recent research suggests the optimal strategy utilizes a vector of cell states sampled at various time points. To elucidate the optimal sampling strategy for vector encoding, we take an information theoretic approach and determine the mutual information of the calcium signaling dynamics obtained from fibroblast cells perturbed with different concentrations of ATP. Specifically, we analyze the sampling strategies under the cases of fixed and non-fixed vector dimension as well as the efficiency of these strategies. Our results show that sampling with greater frequency is optimal in the case of non-fixed vector dimension but that, in general, a lower sampling frequency is best from both a fixed vector dimension and efficiency standpoint. Further, we find the use of a simple modified Ornstein-Uhlenbeck process as a model qualitatively captures many of our experimental results suggesting that sampling in biochemical networks is based on a few basic components.   

Deciphering the Minimal Algorithm for Development and Information-genesis.

During development, cells with identical genomes acquires different fates in a highly organized manner. In order to decipher the principles underlining development, we used C.elegans as the model organism. Based on a large set of microscopy imaging, we first constructed a ``standard worm'' in silico: from the single zygotic cell to about 500 cell stage, the lineage, position, cell-cell contact and gene expression dynamics are quantified for each cell in order to investigate principles underlining these intensive data. Next, we reverse-engineered the possible gene-gene/cell-cell interaction rules that are capable of running a dynamic model recapitulating the early fate decisions during C.elegans development. we further formulized the C.elegans embryogenesis in the language of information genesis. Analysis towards data and model uncovered the global landscape of development in the cell fate space, suggested possible gene regulatory architectures and cell signaling processes, revealed diversity and robustness as the essential trade-offs in development, and demonstrated general strategies in building multicellular organisms.   

Reliable Signal Transduction

Stochasticity inherent to biochemical reactions (intrinsic noise) and variability in cellular states (extrinsic noise) degrade information transmitted through signaling networks. We analyzed the ability of temporal signal modulation - that is dynamics - to reduce noise-induced information loss. In the extracellular signal-regulated kinase (ERK), calcium (Ca(2$+))$, and nuclear factor kappa-B (NF-$\kappa $B) pathways, response dynamics resulted in significantly greater information transmission capacities compared to nondynamic responses. Theoretical analysis demonstrated that signaling dynamics has a key role in overcoming extrinsic noise. Experimental measurements of information transmission in the ERK network under varying signal-to-noise levels confirmed our predictions and showed that signaling dynamics mitigate, and can potentially eliminate, extrinsic noise-induced information loss. By curbing the information-degrading effects of cell-to-cell variability, dynamic responses substantially increase the accuracy of biochemical signaling networks.