Session Y35: Principles of Cell-to-Cell Communication

Collective synchronization of divisions in Drosophila development.

Mitoses in the early development of most metazoans are rapid and synchronized across the entire embryo. While diffusion is too slow, in vitro experiments have shown that waves of the cell-cycle regulator Cdk1 can transfer information rapidly across hundreds of microns. However, the signaling dynamics and the physical properties of chemical waves during embryonic development remain unclear. We develop FRET biosensors for the activity of Cdk1 and the checkpoint kinase Chk1 in Drosophila embryos and exploit them to measure waves in vivo. We demonstrate that Cdk1 chemical waves control mitotic waves and that their speed is regulated by the activity of Cdk1 during the S-phase (and not mitosis). We quantify the progressive slowdown of the waves with developmental cycles and identify its underlying control mechanism by the DNA replication checkpoint through the Chk1/Wee1 pathway. The global dynamics of the mitotic signaling network illustrates a novel control principle: the S-phase activity of Cdk1 regulates the speed of the mitotic wave, while the Cdk1 positive feedback ensures an invariantly rapid onset of mitosis. Mathematical modeling captures the speed of the waves and predicts a fundamental distinction between the S-phase Cdk1 trigger waves and the mitotic phase waves, which is illustrated by embryonic ablation experiments.   

Collective Calcium Dynamics in Networks of Communicating Cells

Cells can sense and encode information about their environment with remarkable precision. These properties have been studied extensively for single cells, but intercellular communication is known to be important for both single- and multicellular organisms. Here, we examine calcium dynamics of fibroblast cells exposed to external ATP stimuli, and the effects of communication and stimulus strength on cells' response. Experimental results show that increasing communication strength induces a greater fraction of cells to exhibit oscillatory calcium dynamics, but the frequencies of oscillation do not systematically shift with ATP strength. We developed a model of calcium signaling by adding noise, communication, and cell-to-cell variability to the model of Tang and Othmer \footnote{Tang, Y. and Othmer, H. G., \textbf{Proc. Natl. Acad. Sci.} 92, 1995.}. This model reproduces cells' increased tendency to oscillate as a function of communication strength, and frequency encoding is nearly removed at the global level. Our model therefore suggests that the propensity of cells to oscillate, rather than frequency encoding, determines the response to external ATP. These results suggest that the system lies near a critical boundary separating non-oscillatory and oscillatory calcium dynamics.   

Precision of multicellular gradient sensing with cell-cell communication

Gradient sensing underlies diverse biological processes. In principle, bigger ``detectors’’ (cells or groups of cells) make better sensors, since then concentrations measured at the front and back of a detector are more different, and the gradient can be determined with higher precision. Indeed, experiments have shown that populations of cells detect gradients more precisely than single cells. However, this argument neglects the fact that information must be communicated between different parts of the detector, and the communication process introduces its own noise. Here we derive the fundamental limits to the precision of gradient sensing with cell-cell communication and temporal integration. We find that communication imposes its own sensory length scale, beyond which the precision cannot increase no matter how large the cell population grows. We also find that temporal integration couples the internal communication with the external signal diffusion, imposing an additional limit on the precision. We discuss how these limits can be improved by a strategy with two communicated molecular species, which we term ``regional excitation—global inhibition’’. We compare our findings to experiments with communicating epithelial cells, and infer a sensor length scale of about 4 cells.   

Collective synchronization of self/non-self discrimination in T cell activation, across multiple spatio-temporal scales

The immune system is a collection of cells whose function is to eradicate pathogenic infections and malignant tumors while protecting healthy tissues. Recent work has delineated key molecular and cellular mechanisms associated with the ability to discriminate self from non-self agents. For example, structural studies have quantified the biophysical characteristics of antigenic molecules (those prone to trigger lymphocyte activation and a subsequent immune response). However, such molecular mechanisms were found to be highly unreliable at the individual cellular level. We will present recent efforts to build experimentally validated computational models of the immune responses at the collective cell level. Such models have become critical to delineate how higher-level integration through nonlinear amplification in signal transduction, dynamic feedback in lymphocyte differentiation and cell-to-cell communication allows the immune system to enforce reliable self/non-self discrimination at the organism level. In particular, we will present recent results demonstrating how T cells tune their antigen discrimination according to cytokine cues, and how competition for cytokine within polyclonal populations of cells shape the repertoire of responding clones. Additionally, we will present recent theoretical and experimental results demonstrating how competition between diffusion and consumption of cytokines determine the range of cell-cell communications within lymphoid organs. Finally, we will discuss how biochemically explicit models, combined with quantitative experimental validation, unravel the relevance of new feedbacks for immune regulations across multiple spatial and temporal scales.   

A generic spatial-stochastic framework for quantifying noisy information flow in multicellular systems

Spatio-temporal protein signals play a crucial role in communicating information within and between cells. However, their ability to convey signals robustly is hampered by noise in gene regulation and biochemical transport, occuring at low copy numbers. While we increasingly understand distinct strategies of biochemical noise control, it remains unclear how nature orchestrates them to maximize information flow. Our recent work extends our information-theoretic framework for gene regulation to an explicitly spatial setting. We constructed a stochastic model enabling fast calculation of local means and variances in a spatially coupled gene regulatory system, which we use for rigorous quantification of information flow in an ensemble of units sensing a spatially distributed input and exchanging information via diffusion. By applying our framework to the paradigmatic Bcd-Hbk system in early fly development, we demonstrate that diffusive coupling can be of substantial benefit in encoding positional information, and uncover a novel optimal regulatory strategy relying on spatial coupling. Thanks to the generic methodology employed, our framework is universally applicable for realistic predictive modeling and data-driven inference of multicellular systems engaging in noisy communication.   

Increased dimensionality of cell-cell communication can decrease the precision of gradient sensing

Gradient sensing is a biological computation that involves comparison of concentrations measured in at least two different locations. As such, the pre- cision of gradient sensing is limited by the intrinsic stochasticity in the com- munication that brings such distributed information to the same location. We have recently analyzed such limitations experimentally and theoretically in multicellular gradient sensing in mammary epithelial cell organoids. For 1d chains of collectively sensing cells, the communication noise puts a se- vere constraint on how the accuracy of gradient sensing increases with the number of cells in the sensor. A question remains as to whether the effect of the noise can be mitigated by the extra spatial averaging allowed in sensing by 2d and 3d cellular organoids. Here we show using computer simulations that, counterintuitively, such spatial averaging decreases gradient sensitiv- ity (while it increases concentration sensitivity). We explain the findings analytically and propose that a recently introduced Regional Excitation - Global Inhibition model of gradient sensing can overcome this limitation and use 2d or 3d spatial averaging to improve the sensing accuracy.   

Synthetic Quorum Sensing and Induced Aggregation in Model Microcapsule Colonies with Repressilator Feedback

We model a system of synthetic microcapsules that communicate chemically by releasing nanoparticles or signaling molecules. These signaling species bind to receptors on the shells of capsules and modulate the target shell’s permeability, thereby controlling nanoparticle release from the target capsule. Using the repressilator regulatory network motif, whereby three species suppress the production of the next in a cyclic fashion, we show that large amplitude oscillations in nanoparticle release can emerge when many capsules are close together. This exemplifies quorum sensing, which is the ability of cells to gauge their population density and collectively initiate a new behavior once a critical density is reached. We present a physically realizable model in which the oscillations exhibited in crowded populations induce aggregation of the microcapsules, mimicking complex biological behavior of the slime mold \textit{Dictyostelium discoideum} with only simple, synthetic components. We also show that the clusters can be dispersed and reformed repeatedly and controllably by addition of chemical stimuli, demonstrating possible applications in creating reconfigurable or programmable materials.   

Limits to collaborative concentration sensing in cell populations

Cells sense chemical concentrations with a precision that approaches the physical limit set by molecular diffusion. Recent experiments have vividly shown that cells can beat this limit when they communicate. We derive the physical limits to concentration sensing for cells that communicate over short distances by directly exchanging small molecules across their membranes (juxtacrine signaling), and over long distances by secreting and absorbing a diffusive messenger molecule (paracrine signaling). In the latter case, we find that the cell spacing that optimizes precision can be large, due to a tradeoff between maintaining communication strength and reducing signal cross-correlations. This leads to the surprising result that paracrine signaling allows more precise sensing than juxtacrine signaling for sufficiently large populations, even though this means that the cells are spaced far apart. We compare our results to recent experiments.