Information Processing in Multicellular Networks

Graduation date: 2017The ability for individual cells and multicellular networks to process \ud information contained in perceived stimuli is of vital importance to their proper \ud functioning and ultimately their cellular fate. The primary mechanism employed by \ud cells to this end is the use of biochemical signaling in which a series of molecular events \ud results in response to the stimuli. These series of molecular events can be highly \ud dynamic in their behavior and recent work has suggested that these dynamics are \ud important in the processing of information and resulting cellular response. Further, it \ud has been shown that multicellular systems themselves exhibit dynamic, coordinated \ud behavior that may act as an additional dimension to information processing in these \ud systems. \ud To explore and make rigorous the information processing abilities of individual \ud cells and multicellular networks, we focus on the store-operated calcium pathway in \ud mouse fibroblast and human breast cancer cells stimulated by ATP. This pathway is \ud well-known in its ability to create both spatial (blips, puffs, and waves) and temporal \ud (oscillations) cellular calcium dynamics via a nonlinear feedback system. Using \ud \ud microfluidics, fluorescence and confocal microscopy, stochastic modeling, and \ud information theoretic techniques we probe how individual cells and multicellular \ud networks utilize these dynamics to encode and decode environmental cues. \ud In the case of individual cells, we explore possible sampling strategies \ud employed by cells to process dynamic biochemical signals. We find that sampling as \ud frequently as possible leads to increased information but that this is a costly strategy \ud and leads to increased redundant information. Instead an infrequent and well-separated \ud sampling strategy arises as the optimal strategy, showing that cells cannot easily \ud distinguish fast dynamics and thus act as a low-pass filter. Further, we find that \ud information is more affected by extrinsic noise than intrinsic noise in the case of multi\ud time point or vector encoding. \ud For the case of multicellular networks, we probe how ATP stimuli are encoded \ud in the calcium dynamics of varied network architectures. Our results suggest that the \ud strength of stimuli is encoded in the magnitude of the calcium response in addition to \ud the oscillation propensity of cells in the network. Modeling confirms this behavior and \ud predicts that communication via gap junctions is additionally a vital part of encoding. \ud We find that modulating communication of the network by adjusting cell density \ud influences the oscillation propensity at moderate ATP levels. Further, modulating the \ud communication of the network by increasing the fraction of cancer cells produces the \ud same result at moderate ATP levels. We conclude that stimuli information is \ud multiplexed in the oscillation propensity of the multicellular system, which is \ud potentially beneficial in responding with specificity to a wide array of stimuli