www.neuron.yale.edu

Hi,

I have a question regardng use of either i_cap vs dvdt*capacitance for calculating ‘total membrane current’. My intent is to explore the relevance of I-V curves (using SEClamp) in a single compartment with HH conductances, to neuronal behaviour under ‘normal’ conditions. I have had a browse through these threads (viewtopic.php?f=12&t=2437, http://www.neuron.yale.edu/phpbb/viewto ... =16&t=2146) but can’t quite seem to find an answer.

In a single compartment with area=300um2, cm=1uF/cm2 (cell capacitance 3e-6uF) I calculated total membrane current as dvdt*3e-6. I compared this to total ‘i_cap’ (calculating over surface area) which produced quite different results. Am I doing something wrong here?

Finally, is either of the above currents in any sense ‘equivalent’ to current generated by SEClamp during a voltage-step protocol? Ie, as channels are activated SEClamp produces an equal/opposite current to the ‘total membrane current’, which can be estimated by either i_cap or dvdt*C

Any help much appreciated

Fact: integral of membrane current over the surface of a cell is always 0 unless you are using a sharp or patch electrode to inject current into the cell, in which case injecting a positive current will produce a total membrane current that is positive and equal to the total injected current. This is true regardless of whether the cell is electrotonically compact or has an extended architecture.

Perhaps you were thinking of total capacitive current, or total ionic current (which is equal to total capacitive current but opposite in sign). Focussing on capacitive current, consider:
100 um2 area with 1uf/cm2 specific capacitance has total capacitance of 1 pF (1e-12 F).
1 A charging 1 F makes dV/dt 1 V/s
1 A charging 1 pF makes dV/dt 1e12 V/s = 1e12 mV/ms
1 nA charging 1 pF makes dV/dt 1e3 V/s = 1e3 mV/ms
1 nA charging 3 pF makes dV/dt = (1/3)e3 mV/ms

Hi Ted, thanks for you reply.
Yes apologies for the sloppy(incorrect) terminology - I was referring to total ionic current and capacitive current.
I think the differences I see when calculating the above using either dvdt (eg your calculations above) or i_cap are numerical, as I am looking at values during AP up and downstroke. As an example, peak i_cap vs the maximal value taken from dvdt.
If I intend to maximise precision of recorded current over this period, is it best to use cvode specified with a small DT (eg viewtopic.php?f=8&t=1127)?

A practical approach would be to run the simulation with adaptive integration, but instead of recording at all time points (which may span both very short and extremely long intervals), use Vector.record's Dt argument to capture time and simulation results at some reasonable fixed interval. The recorded values will be interpolated from the values at the actual solution times. For example, to capture a variable at 25 us intervals,
vdest.record(&var, 0.025)
Read about this in the Programmer's Reference entry on Vector record.

Thanks Ted, much appreciated. Works perfectly with consistent results now.

Hi Ted, I have a related question I am hoping you could help with. It is touched on here (viewtopic.php?f=12&t=2437) but I would like to make sure I understand correctly.
I want to determine the contribution of different mechanisms to the capacitive current (i_cap) recorded at the soma of a multi-compartment cell. For arguments sake with 1 dendrite, H-H Na/K channels and an electrode with constant current stimulus.
If I record i_na, i_k, axial dendritic current to the soma (converting from nA to mA/cm2) and electrode current (same conversion) should this sum to i_cap assuming no other mechanisms exist in the soma?

asb wrote:I want to determine the contribution of different mechanisms to the capacitive current (i_cap) recorded at the soma of a multi-compartment cell . . . should this sum to i_cap assuming no other mechanisms exist in the soma?

Yes.