Modeling the axon: how much detail?

[christina]

To Ted, Michael and the NEURON Forum:

I am developing a model of a neuron (from the goldfish hindbrain) using detailed 3D soma and dendrite morphology reconstructed with Neurolucida. We are studying how dendritic morphology and active channels contribute to somatic firing patterns. Before incorporating an axon into the model, I am looking for some more information. How much detail is commonly used to model axons? Although the answer is probably, “it depends on what you want to do with itâ€

[ted]

First let me mention a couple of my favorites:
Vabnick, I., Trimmer, J.S., Schwarz, T.L., Levinson, S.R., Risal, D., and Shrager, P.
Dynamic potassium channel distributions during axonal development prevent aberrant firing patterns.
Journal of Neuroscience 19:747-758, 1999.
Alle, H. and Geiger, J.R.P.
Combined analog and action potential coding in hippocampal mossy fibers.
Science 311:1290-1293, 2006.

Neither of these is in ModelDB, but the former is described in sufficient detail that it might be
possible to replicate if you feel so inclined (or you might just ask Peter Schrager for the source
code). And a qualitative replication of the 2nd paper is quite doable without any more information
than what is available from the few published articles that describe the anatomy of
dentate granule cell axons, plus what might reasonably be surmised about their
biophysical properties.

If you are concerned about axonal ion channel density and spike initiation, don't stop with
Mainen et al. 1995--check out this too:
Colbert, C.M. and Pan, E.H.
Ion channel properties underlying axonal action potential initiation in pyramidal neurons.
Nature Neuroscience 5:533-538, 2002.


It is no indulgence in fashionable relativism to say that what one includes in a model should
be guided by what one intends to do with the model. Given the problem you are addressing
"how dendritic morphology and active channels contribute to somatic firing patterns"
and the evidence that the site of spike initiation may depend on the nature of the stimulus
(weak stimuli tend to favor axonal initiation, stronger inputs tend to elicit somatic or even
distal dendritic initation), you do need to include a model axon, and probably should
preserve significant anatomical and biophysical properties to the extent that they are
known. If they are not known, a reasonable course might be to assume something
simple, and do most of your modeling with that. Once you have hit upon a set of
conclusions with key simulations that demonstrate them, run a few more simulations
using a model that _lacks_ an axon just to see whether this produces a qualitatively
significant change in the results. The standard practice of "bracketing assumptions to
see if conclusions are robust despite parameter uncertainties."

[christina]

Thanks Ted.

All three of these references look interesting and certainly relevant. Particularly the Colbert and Pan paper demonstrating that the kinetics, not the densities, of axonal channels may be critical to AP initiation and propagation.

The proximal region of the axons were reconstructed in these cells, but little is known of the biophysics. I will use these data to constrain the axonal morphology in our model, but plan to keep the rest of the axon relatively simple (e.g. likely I'll omit internode/nodes of Ranvier, and will use published data to constrain kinetics).

Thanks again for your advice. I'll post again if I've got further questions/comments.

-Christina