the default value of extracellular mechanism

[melissamou]

Hi Ted,

Happy new year!

I have submitted a paper and now get the comment from the reviewer, there is one question about my neuron model that I do not know how to answer.

Firstly, let me introduce my neuron model.

The model is a myelinated axion. I use extracellular mechanism in the model, and the extracellular potential is generated from COMSOL.(we just calculate the electric-field distribution along the axon. And add these voltage to the node of Ranvier )

the comment of the Reviewer is

"The standard NEURON distribution includes an extracellular mechanism (in hoc: "insert extracellular") with default parameter values {xraxial=1e+09 xg=1e+09 xc=0 e_extracellular=0}. In the authors' model, the extracellular voltage is calculated in COMSOL and is used as a fixed input to NEURON. Thus they assume that the transmembrane currents do not appreciably affect the extracellular voltage, but it is not clear why this should be so. Combining the extracellular applied voltage and the transmembrane voltage in a single NEURON model would allow for these variables to affect each other in both directions. The authors should explain why it is only necessary to consider one way effects, or show evidence that replacing COMSOL with the extracellular mechanism within
NEURON gives indistinguishable results, or else less accurate results."


As I remember, you have saie" The default values of extracellular's xraxial, gc, and xc are such that vext, the potential immediately adjacent to the outer surface of a section, will be equal to e_extracellular."

So, do you have any idea to answer the reviewer's question?

Thanks a lot. wish you a happy day.

[ted]

It can be difficult to discern exactly what someone means--and in this case, I'm not entirely sure about your own statements, or those of the reviewer. So the following comments are based not only on the contents of your post, but also the concerns that I would have if I were reviewing such a paper. You don't have to reply through the Forum, since that might expose confidential matters to public view (after all, your paper has not yet been accepted for publication).

This sententence

I use extracellular mechanism in the model, and the extracellular potential is generated from COMSOL.(we just calculate the electric-field distribution along the axon. And add these voltage to the node of Ranvier )

leaves me with some uncertainty about how the model represents the effects of myelination and the effects of extracellular stimulation.

Was extracellular inserted into both the nodes and the internodes? Was myelination represented by simply inserting pas into the internodes, setting g_pas to a small value, and reducing cm? Or did the membrane of the internodal sections have voltage-gated K channels and "normal" cm, plus the extracellular mechanism with xg and xc set to small values (not 1e-9 but something that would emulate the "imperfect insulator" properties of real myelin)? And if internodal sections did have the extracellular mechanism, was xraxial set to some value that allowed a bit of longitudinal current to flow adjacent to the axolemma (as it would in the case of a real axon), or was it so large (1e9) as to make longitudinal current spread negligible?

What happened to e_extracellular in the internodes--was it driven by extracellular potential values calculated for those locations with COMSOL (which is what should have happened), or did it remain fixed at 0?

Presumably the xg and xg[1] values were large (1e9), so that the potential immediately adjacent to the outer surface of the axolemma would be identical to e_extracellular. This is what I think the reviewer is most concerned about. This would only be an issue at nodes of Ranvier, where current density might be high enough--and extracellular resistivity high enough--to cause a significant change in the potential immediately adjacent to the external surface of the axon. Are there experimental or theoretical estimates of extracellular resistivity at nodes of Ranvier, which would provide a rational basis for assigning some nontrivial value to xg or xg[1]?

Are you reusing an existing model that was developed by someone else? This might allow you to invoke their rationale for details of the model design. Or is it possible to invoke the "reasonable first approximation" argument--i.e. say that it is reasonable that many/most/all of these issues can be ignored "as a first approximation," so that only a very simple model is required to address the question that was under study, but that these other issues can be considered in the Discussion section (and left to be investigated in "future work")?

If further, more specific, discussion would be useful to you, perhaps it should be done by email since your paper has not yet been published. You may reach me at
ted dot carnevale at yale dot edu