CNS*2020 Online has ended
Welcome to the Sched instance for CNS*2020 Online! Please read the instruction document on detailed information on CNS*2020.
Back To Schedule
Sunday, July 19 • 9:00pm - 10:00pm
P109: Investigating Water Transport Mechanisms in Astrocytes with High Dimensional Parameter Estimation

Log in to save this to your schedule, view media, leave feedback and see who's attending!

Feedback form is now closed.
Pierre-Louis Gagnon, Kaspar Rothenfusser, Nicolas Doyon, Pierre Marquet, François Laviolette

To join : https://meet.google.com/hha-rzff-pzz

Holographic microscopy allows one to measure subtle volume changes of cells submitted to challenges such as an osmotic shock or sudden increase in extracellular potassium. Interpreting volumetric data however remains a challenge. Specifically, relating the amplitude of volume changes to the biophysical properties of cells such as passive permeability to water or the rate of water transport by cation chloride cotransporter is a difficult but important task. Indeed, mechanisms of volume regulation are key for cell resilience and survival. Experimentally, the second author measured the volume response as well as the change in sodium concentration of astrocytes submitted to bath applied: hypo-osmotic solutions, solutions with high potassium concentration or solutions containing glutamate. Overall, he measured the time course of the response of over 2000 astrocytes. In order to interpret this rich data, we developed a mathematical model based on our biophysical knowledge of astrocytes. This model relates on the one hand the experimental perturbations of the extracellular medium and on the other the properties of the cell such as its various conductances or strengths of transporters to its responses in terms of volume change, changes in ionic concentrations and in membrane potential. Determining the biophysical properties of cells thus boils down to a problem of model calibration. This presentation is mainly focused on the work of the first author who designed and implemented a gradient-based optimization algorithm, to estimate model parameters and find the values of the parameters which best explain the data coming from distinct modalities and astrocytes.

A first computational challenge is to combine data from different modalities. In some experiments, the sodium response is measured while in others, the volume response is inferred from phase measurements. We also take advantage of the fact that expert knowledge provides information on variables which are not measured. For example, even if membrane potential is not measured, we impose that it is between -100 mV and -50 mV at equilibrium. Combining these different information sources translate into a complex loss function. Furthermore, using a priori knowledge on the value of parameters, we developed a Bayesian approach. Another challenge comes from the fact that different measurements come from different cells. Our goal is thus not to infer a single set of parameters but rather to infer how biophysical parameters are distributed within the population of cells. This was achieved by using a Tikhonov approach which penalizes parameter values laying far from the average of the distribution.

With our algorithm, we were able to infer the strength of the sodium potassium ATPase pump in each cell with a good precision. This could be useful in identifying cells which are more vulnerable. Parameters related to water transport such the passive membrane permeability to water or the rate of water transport through cation chloride cotransporters are elusive and cannot be determined by conventional methods. Our inference algorithms provided information on these values. Finally, our algorithm is flexible enough to adapt rapidly to take advantage of new experiment type or new data modality.

avatar for Pierre-Louis Gagnon

Pierre-Louis Gagnon

PhD Candidate, Département d'Informatique et Génie Logiciel, Université Laval

Sunday July 19, 2020 9:00pm - 10:00pm CEST
Slot 01