(636c) A Multiscale Systems Model For Platelet Signaling And Activation

Purvis, J., University of Pennsylvania
Diamond, S. L., University of Pennsylvania

Upon activation, platelets release Ca2+ from intracellular stores and facilitate Ca2+ influx from the extracellular environment, triggering a host of chemical and morphological changes. To gain a better understanding of this process, we have constructed a multiscale, compartmentalized systems model of Ca2+ balance in the resting (steady state) and activated (time-dependent) platelet. The model incorporates multiple time and length scales, deterministic and stochastic kinetic simulators, and spatially-resolved depictions of the platelet environment to address specific quantitative questions about platelet function. The resting model comprises five isotropic compartments, 20 chemical species, 27 parameters, and four classes of ODE-based reaction equations: release, sequestration, influx, and extrusion (21 total reactions). Balance between compartments was achieved by the action of a Ca2+ ATPase, which counteracts the steady leakage of ions through open IP3 receptor channels by pumping them back against the concentration gradient. Free parameters corresponding to the number of channels and channel activation dynamics were calibrated with experimental data. The system was found to be stable under mild fluctuation, including a 100 nM spike in cytosolic Ca2+, but could be activated by a sufficient increase in key activators, such as extracellular ATP or intracellular IP3. Using standard electrochemical calculations and measurements for single-channel Ca2+ flux, we were able to derive an estimate for the number of Ca2+ channels per platelet, which has eluded experimental measurement. Our calculations suggest that as few as 10 open channels per platelet are sufficient to cause the observed burst in intracellular Ca2+. Likewise, we observed that less than 20 IP3 molecules are sufficient to trigger the same response. These calculations support the view that the heterogeneous signaling behavior observed among platelet populations is partly a stochastic feature due to low molecule copy number. To model signal transduction at the platelet plasma membrane, we developed a spatially-resolved kinetic Monte Carlo procedure which tracks individual interactions among the molecules localized to the lipid bilayer. As a pilot system, we simulated prothrombinase assembly on an 1850 nm2 patch of the plasma membrane. Explicit modeling of the 2D platelet surface allows allows us to extrapolate physical parameters (e.g., ?sticking? coefficients, surface mobilities) that are difficult or impossible to measure experimentally. The pilot surface model comprises seven species and six reactions, including adsorption, desorption, surface association/dissociation, and diffusion. Using fura-2-loaded human platelets and high-throughput robotics, we can explore the combinatorial effect of multiple platelet activators, such as thrombin, ADP and convulxin, as they converge on Ca2+ signaling pathways. These data will be used to calibrate a time-dependent model of platelet activation that is consistent with the steady state model. Machine learning will be used to assess the relative contributions of these activation pathways and propose new functional relationships among the network of signaling molecules.