Session B47: Physics of Multicellular Information Processing

A physical sciences approach to understanding tumor dormancy

This abstract not available.   

Signal transmission in a heterogeneous bacterial population

Biological systems such as tissues or bacterial communities often require reliable signal transmission among cells to coordinate actions at a distance. One of the key obstacles for such signal propagation is the spatial heterogeneity that arises when only a fraction of cells contributes to signal transmission. This cell-to-cell heterogeneity can cause signal propagation to die out before reaching its desired target. Motivated by electrochemical signaling within bacterial biofilms, in which only a fraction of cells participates in the signaling, we develop a model of signal propagation in a heterogeneous community. We integrate percolation theory, which describes the structure of the heterogeneity, with the FitzHugh-Nagumo model, which describes the excitable dynamics of signaling at the single-cell level. We find that the transition between signal propagation and signal failure is determined not only by the structural properties (e.g., the percolation threshold), but also by the dynamic properties (e.g., the excitation threshold) of the model. Our integrated model provides predictions that we test using gene-deletion strains that modify the fraction of participating cells in the biofilm.   

Disrupting Microbial Communication Leads to a 2D Percolation Transition.

Bacteria often converse with each other to coordinate macro-scale expressions such as pathogenic infections, production of biofilms, production of antibiotics and bioluminescence. Empirical evidence suggests that bacteria are capable of conducting these conversations at length scales far exceeding the size of themselves. We observe that by disrupting this large-scale communication, the conversations within these microbes follow a pattern similar to a percolation transition. In experiments, we have one bacterial species that is capable of communication (sender) and another species that is capable of producing an enzyme capable of disrupting this communication (degrader). We spatially distribute the senders and gradually increase the amount of degraders to observe a critical density at which the large-scale communication is suppressed. Using experimental image analysis and reaction-diffusion modeling, we were able to calculate critical exponents and confirmed that this transition is a 2D percolation universality class. Our results suggest that, even for systems as complex and diverse as a heterogeneous microbial community, statistical mechanical models can still be a powerful tool for extracting robust, universal quantitative features.   

Collective Creativity and Survival Algorithms of Bacteria

We are constructing physical puzzles using nano and microfabrication that the bacteria must solve to access food, but we are making the puzzles so computationally complex that only a very small percentage of the bacteria, if they use the normal biased random chemoattractive algorithm and if they do not collectively share information on where food is, will survive. We ask this fundamental question: do the bacteria collaborate to come up with clever solutions to the puzzles we pose and find food in the puzzle by sharing information? If they do, and we think they do based on our early experiments, then there are signs of creativity at a very simple level of life, and this would point to a collective creativity.   

Efficient transmission of unique cell identities via correlated fluctuations, emergent discreteness, and error-correcting codes

In a developing embryo, information about the position of cells is carried by the concentrations of specific molecules. In the fruit fly, these molecules are known, and form a network with several layers—maternal inputs, gap genes, and pair-rule genes whose expression forms a blueprint for the segments of the developed organism. Recent work shows that the concentrations of the gap gene products point to positions with an accuracy comparable to the spacing between neighboring cells. This is nearly enough to specify unique cellular identities. We make three observations which suggest theoretical mechanisms for precisely and efficiently encoding cellular identities. (1) Correlations in the fluctuations of signaling molecule concentrations enhance the transmission of information about positions along the embryo.(2) Although the positional information is carried by continuous concentration gradients, maximum information transmission is achieved by a discrete set of cell identities. (3) Inference of these discrete identities corresponds to a statistical physics problem. Noise acts as a random field and its correlations couple neighboring variables. We consider the possibility of an ordered phase which is stable against the random field, and hence defines an error-correcting code.   

How Nonuniform Contact Profiles of T Cell Receptors Modulate Thymic Selection Outcomes

T cell receptors (TCRs) bind foreign/self peptides presented by APCs, with a strength that may induce T cell activation. TCRs form through stochastic gene rearrangement, and are selected in the thymus prior to T cell maturation by screening against a number of self-peptides. If a TCR binds too strongly to a self-peptide, or not strongly enough to any, the T cell dies; this shapes the post-selection TCR repertoire to recognize diverse foreign peptides yet avoid self-reactivity. Past work mapped thymic selection to an extreme value problem, and analyzed the statistical properties of the post-selection repertoire as a function of selection parameters. Here, motivated by recent experiments showing that amino acids at certain positions of the TCR sequence are important for inducing self-reactive T cells, we develop a method to incorporate information about TCR structure through its nonuniform contact profile into our model of thymic selection. We find that statistical enrichment or depletion of amino acids is now greatly enhanced at positions making larger contacts, and intriguingly, this effect depends nontrivially on how nonuniform contacts are mediated during thymic selection.   

Modeling Collective Cell Migration: Clusters and Monolayers

Collective migration is prevalent in nature, from migrating animals to groups of cells. At the level of the cells, motile collectives appear in the form of isolated clusters of different sizes, as well as within continuous tissues. The modeling of this phenomena will be presented, using analytic simplified models, as well as particle-based numerical simulations. We describe several examples where physical models shed light on the biological system: (i) Chemotaxis of 2D and 3D cellular clusters. By comparing the observed dynamics of cellular clusters to the theoretical models, we can extract the nature of the underlying traction forces, and the coordination between these forces. The phenomenon of spontaneous transitions between migration and rotation of these clusters is also elucidated. (ii) The collective motion at the edge of an expanding, continuous monolayer exhibits instabilities. We model this instability, which also explains the behavior of collective flows within geometric confinement.   

Repertoire-based approach to identifying sequence motifs specific to an effective vaccine

The T-cell repertoire response to the yellow fever vaccine builds near total immunity. To quantify the response of this high dimensional stochastic system we developed a statistical model of differential T-cell proliferation and used it as a basis for inference from high-throughput receptor sequencing data obtained from individuals pre and post vaccination. We used replicate data to infer parameters of the experimental variation of clone sizes, and pre and post vaccination data to quantify the response and identify candidate clones responsive to the vaccine by their posterior expansion probability. These candidates are experimentally validated and show above random sequence similarity.   

Generation Probabilities of T cell receptors: a story of Coarse Graining

The adaptive immune system operates by stochastically generating a large repertoire of unique receptors to tag foreign peptides through specific binding affinities. As a result, there is a general interest (both clinical and scientific) in being able to compute the probability of generating (Pgen) and observing a particular receptor or functionally equivalent group of receptors. Previous work has defined probabilistic models for the underlying V(D)J recombination events, however it had only been tractable to compute Pgen on the nucleotide level due to the exponential explosion of events leading to the same amino acid sequence of the receptor. Through the introduction of a novel algorithm, which leverages dynamic programming, it is now possible to compute the Pgen of amino acid sequences, as well as sequences comprised of functional motifs, quickly and efficiently. This not only provides a tool for computing the baseline generation probability of receptors, but allows statistical analysis of how the Pgen distribution changes as a result of coarse graining from recombinatorial event, to nucleotide sequence, and finally to receptor. The reduction in entropy stemming from this coarse graining suggests at a possible explanation of so-called 'public' receptors.   

Mesoscale Organization and Dynamics of T Cell Activation Proteins

Careful regulation of T cell activation is important to ensure that signals persist over controlled periods of time. If signals are too short then the immune system will not properly respond, but if signals persist for too long it could result in autoimmune diseases such as Lupus and Type 1 Diabetes. In this work we explore how the spatial structure, organization, and dynamics of a protein complex called the POLKADOTS signalosome allow it to serve as both a positive and negative regulator of the activation signal. Analysis of super-resolution images suggests that proper function of the signalosome involves intricate dynamics of spatial structures: self-assembly of the structural protein Bcl10 into a filamentous, rod-like shape; transport of Malt1, a protease that binds to and decorates the Bcl10 filaments; and autophagic degradation of the signalosome structure from the filament ends. We have incorporated these findings into a simple model of self-assembly and degradation to shed light on the underlying dynamic processes that guide signalosome formation, organization, and degradation.   

On public and private aspects of the adaptive immune system as revealed by statistical inference

The adaptive immune system can recognize unanticipated threats by maintaining a large diversity of T cells with different membrane receptors. The diversity is dynamic, with T cells constantly being produced with randomly-generated receptors. The randomness means that different individuals hold different repertoires of T cells, despite the fact they are able to defend against the same pathogens. An interesting aspect of this phenomenon is that certain receptors are “public”, appearing in almost any individual, while others are “private”, being unique to one individual.
We have learned how to infer probabilistic models that capture the statistics of the T cell ensemble carried by individuals. A key feature of these models is that individual immune cells are produced with different likelihoods, with the distribution of generation probabilities spanning many decades. The statistical power of our models enables us to compare samples from different individuals or compartments. We use the models to predict how many receptors (both in number and in specific receptor sequences) will be shared between any number of individuals, simply as a result of the biased generation statistics. We show results for mice and men and draw conclusions about the predictability of a stochastic immune system.