UCLA

The intrinsic cardiac nervous system (ICNS), termed as the hearts little brain, is the final point of neural regulation of cardiac function. Studying the dynamic behaviour of these ICNS neurons via multiscale neuronal computer models has been limited by the sparsity of electrophysiological data. We developed and analysed a computational library of neuronal electrophysiological models based on single neuron transcriptomic data obtained from ICNS neurons. Each neuronal genotype was characterized by a unique combination of ion channels identified from the transcriptomic data, using a cycle threshold cutoff that ensured the electrical excitability of the neuronal models. The parameters of the ion channel models were grounded based on passive properties (resting membrane potential, input impedance and rheobase) to avoid biasing the dynamic behaviour of the model. Consistent with experimental observations, the emergent model dynamics showed phasic activity in response to the current clamp stimulus in a majority of neuronal genotypes (61%). Additionally, 24% of the ICNS neurons showed a tonic response, 11% were phasic-to-tonic with increasing current stimulation and 3% showed tonic-to-phasic behaviour. The computational approach and the library of models bridge the gap between widely available molecular-level gene expression and sparse cellular-level electrophysiology for studying the functional role of the ICNS in cardiac regulation and pathology. KEY POINTS: Computational models were developed of neuron electrophysiology from single-cell transcriptomic data from neurons in the hearts little brain: the intrinsic cardiac nervous system. The single-cell transcriptomic data were thresholded to select the ion channel combinations in each neuronal model. The library of neuronal models was constrained by the passive electrical properties of the neurons and predicted a distribution of phasic and tonic responses that aligns with experimental observations. The ratios of model-predicted conductance values are correlated with the gene expression ratios from transcriptomic data. These neuron models are a first step towards connecting single-cell transcriptomic data to dynamic, predictive physiology-based models.