Action potential backpropagation in a model thalamocortical relay cell
Introduction
Thalamocortical (TC) cells are the basic components that relay almost all sensory information, with the exception of olfactory, to the neocortex [12]. This gives the thalamus a vital role that is not yet fully understood. One interesting fact is that thalamocortical relay cells do not receive most of their synaptic inputs from sensory afferents [15]. The cortical innervation of the thalamus is a much greater source of input, with TC cells in sensory nuclei receiving feedback from the same areas of cortex to which they project [5]. Furthermore, there is a clear spatial segregation of these inputs in the dendritic arbour of TC cells, with cortical inputs located mainly in distal dendrites and sensory inputs mainly in proximal dendrites [10]. Knowledge of the way that TC cells integrate these inputs is crucial for accurate modelling of thalamocortical circuits and hence to understanding the role of TC cells in these circuits.
It has been shown that the shunting of EPSPs by backpropagating action potentials (APs), which was investigated in a recent paper by Häusser et al. [6], can play a part in synaptic integration. Their results indicated that due to this shunting, distal synapses may contribute to synaptic integration to a far greater extent than proximal synapses. If the same mechanisms are at play in TC cells, then the spatial segregation of afferents will be of functional relevance.
This paper investigates action potential backpropagation in a multi-compartment model of a TC cell as described by Destexhe et al. [3]. The results from simulations of this model are compared with recent experimental results of Williams and Stuart [17]. In [17], simultaneous whole-cell current-clamp recordings were made from the dendrites and somata of TC neurons, in slices of rat dorsal lateral geniculate nucleus (LGN), which preserve the dendritic morphology, and the amplitude of backpropagated action potentials through successive branch points of a single dendrite were measured. The APs demonstrated a significant attenuation of amplitude. They then measured AP amplitudes at various dendritic locations throughout the cell. By considering stem dendrites and higher order dendrites separately, they showed a clear dependence of AP amplitude on the dendritic recording location.
Consistent with other studies [16] these results indicated that there is a strong influence of dendritic morphology on the action potential propagation. This means that APs may fail to propagate into the most distal parts of the dendritic tree. Further results in [17] indicated that AP backpropagation is an active process, dependent on the activation of dendritic sodium channels, of which there is a non-uniform distribution throughout the extent of the TC cell. The results also pointed to an on-average uniform density of potassium channels across the cell.
In this paper, we show that this model does not reproduce Williams and Stuart's [17] results well. We then apply a number of biophysically plausible changes to the model and show that these result in a closer match between the behaviour of the model and the experimental results. Finally, we investigate the conditions under which burst and tonic firing is evoked in the model, with and without the changes, and compare our results with those reported in [4].
Section snippets
The model
The original cell morphology [3] used in these simulations was obtained from a TC cell from the rat ventrobasal nucleus. The cell was stained, reconstructed and incorporated into NEURON [7]. The model has a total of 206 compartments, which includes 11 dendrites with varying numbers of sections. The model contains Hodgkin–Huxley type sodium and potassium currents located solely in the soma, a passive leak current and IT, the low-threshold Ca2+ current to which burst activity is attributed [2].
Results in the unchanged cell
Williams and Stuart stimulated their cells using somatic current injection, and this was replicated in the model cell. Measuring the AP amplitude decrease through branchpoints, Williams and Stuart report an average decrease of 27.4% for an average electrode separation of (1.66% per μm). In the model cell, it was found that there is a 0.50±0.04% decrease per μm, demonstrating a clear divergence between the results.
The relationship between the AP amplitude and dendritic location is plotted
Discussion
The attenuation of backpropagated action potentials was investigated in a multi-compartment model of a thalamocortical cell [3]. We have shown that this model does not accurately reproduce the experimental results of [17], which show that the attenuation of the amplitude of APs varies with respect to dendritic location and is generally at a higher level than in the model cell.
A number of changes were made to the model cell, and this resulted in a better replication of the results in [17]. These
Nada Yousif studied physics at Imperial College, London and graduated in 2001. Since 2002 she has been studying thalamocortical networks using computational modelling techniques, in the Centre for Theoretical and Computational Neuroscience at the University of Plymouth, England.
References (17)
- et al.
Thalamic bursting mechanism: an inward slow current revealed by membrane hyperpolarization
Brain Res.
(1982) - et al.
Paying attention to the thalamic reticular nucleus
Trends Neurosci.
(1998) Branching dendritic trees and motoneuron membrane resistivity
Exp. Neurol.
(1959)- et al.
Dendritic current flow in relay cells and interneurons of the cat's lateral geniculate nucleus
Proc. Nat. Acad. Sci. USA
(1989) - et al.
Dendritic low threshold Ca currents in thalamic relay cells
J. Neurosci.
(1998) - et al.
The initiation of bursts in thalamic neurons and the cortical control of thalamic sensitivity
Philos. Trans. Roy. Soc. Lond. Series B
(2002) - et al.
Differential shunting of EPSPs by action potentials
Science
(2001) - et al.
The NEURON simulation environment
Neural Comput.
(1997)
Cited by (0)
Nada Yousif studied physics at Imperial College, London and graduated in 2001. Since 2002 she has been studying thalamocortical networks using computational modelling techniques, in the Centre for Theoretical and Computational Neuroscience at the University of Plymouth, England.
Mike Denham obtained his PhD in Mathematical Systems and Control Theory in 1972 from Imperial College, London. After five years as a postdoc and then lecturer at Imperial College, he joined Kingston University and from 1984 to 1988 he served as Head of the School of Computing. He joined the University of Plymouth as a Research Professor in 1988. In 1991 he set up the Centre for Neural and Adaptive Systems, and has led the Centre since that time. A new University Research Centre in Theoretical and Computational Neuroscience was established in October 2003 under his leadership. His research interests are in understanding the fundamental principles and mechanisms of information processing in neurons and neuronal networks, using a combination of mathematical analysis and computational modelling.