Axon/ Demyelination

[jboer]

I am trying to make a fairly simple myelinated axon model. Whereby controlling the myelin, demyelination and later the ion channel density and distribution. How would you recommend doing this. Should I start with a fairly simple model of alterating axon segments (internodes) connected to nodes, and then individually format each segment? Would you use the IClamp to stimulate the axon at one end and record at different nodes on the axon or use an external electrode to start an action potential. What I am not sure of how NEURON takes the myelin sheath into account and to adjust it's properties and what would then be remyelination? I can expand on what I am trying to do, but think this is a good starting point. Ironically it is a similar model to http://neuronsinaction.com/tutorials/partial

I have unsuccessfully tried to modify ModelDB files: Moore78, brill77, and shrager91
into the situation described above. But each attribute is a uniform variable over the length of the axon, while I am trying for more segmental variables.

Thanks

[ted]

Should I start with a fairly simple model of alterating axon segments (internodes) connected to nodes

Yes.

and then individually format each segment?

More or less.
Decide on a model architecture. Then use SectionLists to help automate specification of properties. Example: in these cartoons, nodes are represented by ., myelinated internodes by =, and demyelinated or partially myelinated regions by -.
You might start by making a completely normal axon.
The number of nodes and internodes should be about twice what is necessary for conduction velocity through the middle three internodes to be nearly identical, and for spike waveforms at the middle 4 nodes to be nearly identical (i.e. avoid initialization and sealed-end artifacts).
For the sake of a specific example, let's say you need 6 nodes and 5 internodes (you'll probably need at least 2x that). So the normal axon would be like this
.=.=.=.=.=.
The lesioned axon would look like so
.=.=.-.=.=.
A recovering axon would differ from a freshly lesioned axon in ways that you can easily discover from published articles (emergence of channels in the unmyelinated region, decreased length of 1 or more neighboring internodes upstream and downstream from the lesion, eventual appearance of a thin layer of myelin in the formerly demyelinated region--but verify all of these by digging into the literature).

The smart way to specify the properties of normal nodes and internodes is to append them to SectionLists, then iterate over the SectionLists. Read about the SectionList class in the Programmer's Reference. Nodes and internodes affected by the acute lesion and later recovery can be handled in a similar way. Use procedures to help keep things organized.

Would you use the IClamp to stimulate the axon at one end and record at different nodes on the axon or use an external electrode to start an action potential.

The former. The latter requires too much additional model complexity, and is arguably fundamentally unnatural (unless one is struck by lightning).

What I am not sure of how NEURON takes the myelin sheath into account and to adjust its properties

NEURON doesn't. You decide how to represent the effects of myelination. The simplest representation is to decrease cm and forgo the insertion of any ion channels into internodes. More faithful representations would insert voltage-gated k channels into internodes, and use the extracellular mechanism to represent the electrical properties of myelin, which are essentially those of a leaky dielectric. There is at least one article that describes the equivalent resistance and capacitance of myelin on a "per layer" basis (n layers would have R*n resistance and Cm/n capacitance); relevant articles date back to the 1980s or maybe earlier.

what would then be remyelination?

There is an experimental literature that describes what occurs; the hints given above are merely suggestions to help guide your search. Much of this work was done in the 1980s, but I am sure that there are also more recent articles. Authors to look for include Waxman and/or Kocsis.