Real cell-computer hybrid system

As part of our activity within the ARC Centre of Excellence for Integrated Brain Function our laboratory at Biology Node of the CfNE is developing new methods for modelling neurons to drive our study of disease mechanisms and in silico drug discovery in genetic epilepsy. We collaborate with members of the Blue Brain Project at EPFL in Geneva where we plan to incorporate our new models of disease ion channels and receptors into some of the most biophysically realistic neuronal and networks models in existence.  Here, we describe one of our key foundation activities which is the development of a real cell-computer hybrid system to enable real-time modelling of conductance models.  This real-time system enables accurate models to be built on as little as 1 second of recording data; compared to earlier voltage clamp methods the time saving is 2–3 orders of magnitude with no loss in accuracy.

Our method incorporates the dynamic clamp; an electrophysiological method that enables “wetware in the loop” analysis for real-time interaction of our biological system with an in silico computer model.  By enabling interaction normal neurophysiological behaviours such as bifurcations are allowed to occur providing information rich data.  This data is then used off-line to fit to either Hodgkin Huxley (HH) or Markov models of ion channels and in as little as one day at the bench and on the computer an accurate model can be developed.  The real strength is that this method allows for diagnostic prediction of the phenotypic effect of ion channel mutations or a mechanistic understanding of drugs to be rapidly undertaken and then staged for incorporation into large scale neuronal network models.  We plan to develop more efficient algorithms and faster computation approaches to provide not only phenotypic output data but parameters for HH and Markov models in real-time.

The computational models of neurons play a crucial role in studying neurological diseases to explain and predict complex behaviours of neurons otherwise impossible to do so. Therefore, developing faster and accurate approaches to model neurons based on dynamic clamp will improve the efficacy of understanding epileptic mechanisms and developing drugs leading way to precision medicine.

The “dynamic clamp” implementation we use for the realtime characterisation of neuronal systems in health and disease.
Kinetic modelling of NaV 1.2 ion channels based on the data from dynamic clamp analysis.  Data was acquired for 1 sec in panel a) for production of model that fit the real data with unprecedented accuracy as shown in b), c) and d).  Note in d) that only one action potential mismatch (marked by the missing paired dot abave the trace) was detected.

Yadeesha Deerasooriya, Geza Berecki, Saman Halgamuge and Steven Petrou.