Page 1 of 1

increasing membrane time constant

Posted: Mon Feb 26, 2018 8:14 am
by theo92

I have a problem validating the membrane time constant of the principal neuron of my model.
Its value should be 27 ms and right now it is 8 ms.
I have already tried reducing the leakage conductance. However, then i would have very high input resistance.
The specific capacitance is already higher than that of other models (cm=1.5) so i cannot increase it any further.
In addition, i have already altered mod files of sigmoidal equations in hodgkin huxley mechanisms but i have not seen any significant progress.

Do you have any ideas?

Re: increasing membrane time constant

Posted: Mon Feb 26, 2018 10:56 am
by ted
If trying to fix membrane time constant by reducing leakage conductance makes input resistance too high, then it is clear that only small fractions of the voltage gated channels are open at resting potential. It follows that tinkering with their voltage dependence, in order to close even more of them, is going to have little effect on membrane resistance or time constant. You will have to increase membrane capacitance and/or reduce leak conductance.

If reducing leak conductance increases input resistance too much, then your model's surface area is too small. If the model is based on a detailed morphometric reconstruction, are you sure about the quality of that data? Was part of the cell amputated during sectioning? Was the entire cell stained or filled with dye/flurophore? Silver stains are notorious for partial staining. Dye-based stains tend to produce a low contrast gray appearance in the finer branches that makes them hard to see let alone measure (especially bad because most neurons have most of their membrane in the finer branches!).

Are you modeling a cell that has spines? Spines add a lot of surface area, especially to the smaller diameter branches where most of the cell membrane is located. Modelers use a variety of dodges to deal with this, ranging from explicit representation of each spine (increases model complexity and slows simulations) to multiplying specific conductance and specific capacitance of the spiny sections by a factor 1 + a where a is the ratio
(spine surface area)/bare neurite area
Typical value assumed for a in cells with lots of spines (1 spine per um dendrite length) is about 1.

Surface area of the neurites is itself more than a bit uncertain. Diameters based on light microscopy are (mis)interpreted as being the diameter of a cylinder or cone, but very few neural structures have a circular cross-section; most are quite irregular, which means that actual perimeter and surface area are larger, and volume is smaller, than the circular assumption predicts.

Another possible problem is current bottlenecks. Do any of your model's sections have extremely small or zero diameter? That will electrically isolate adjacent sections from each other, which prevents signals from spreading throughout the model and results in very high input resistance.

Re: increasing membrane time constant

Posted: Tue Feb 27, 2018 9:23 am
by theo92
Your advice is really helpful.
For now, I will apply different specific capacitances, as you mentioned, to validate the effect of the spines.
In addition, I will try increasing the size of the soma to counterbalance possible big changes in the input resistance.
Thx a lot