Setting RMP according to experimental measurements.

[alexandrapierri]

Hello NEURON community

I set the RMP of my model cell to a certain value, i.e at -69mV but as soon as the simulation starts the RMP becomes hyperpolized. I realize this can be an effect of ion channels combined with morpholgical aspects causing the RMP to drop at a lower value than what I set it at originally. Namely, I set the RMP at -69mV and a few ms later it sets around -85mV.

I need to fit the RMP to expeimental measurements. What is the orthodox way of achieving that? Considering that seting the value to an x amount in my hoc script won't dictate where the RMP sits eventually, I am not sure what the proper way it is to define it.

thank you,
Alex

[ted]

NEURON allows users to include so much complexity into their models that there's no single initialization strategy that works (let alone "works best") for all models. Particularly difficult are models that
1. have one or more state variables (other than membrane potential) with slow dynamics, e.g. concentrations of ions or second messengers
or
2. involve some kind of use-dependent plasticity (e.g. synaptic plasticity, activity-dependent regulation of ion channel density)
or
3. generate spontaneous chaotic activity

So tell me more about your model, and I'll try to point you in the right direction.

[alexandrapierri]

I use this model by Aberra et al 2018. https://senselab.med.yale.edu/ModelDB/S ... del=241165

These are L23 cortical pyramidal cells that are driven by synaptic events from other cortical layers.
I have replaced the morphologies with morphogies of L23 pyramidal cells given to me by my experimental collaborators and I use the ephys measurements that they measure.
For example although in Aberra et al the pyramidal cells have an RMP set at -75mV, my collaborators have measured the RMP to be at -69mV. I set the RMP at -69mV, run the simulation and I see that my cell hyperpolarizes to -85mV.

I believe these models do not have plasticity and don't generate spontaneous chaotic activity either.

I am looking for best strategies to initialize the model.

thank you,
Alex

[ted]

Do these model cells have any mechanisms that affect concentrations of ions or second messengers? Examine the NEURON blocks of the mod files. Do you see any USEION statements that WRITE a concentration?

[alexandrapierri]

yes, many mechanisms do USEION and then WRITE concentrations of ions such as those describing calcium concentrations, Na and potassium channels.

[ted]

If a model cell is not spontanously active (not firing spikes or having spontaneous subthreshold fluctuations of membrane potential), then

1. If the model cell's concentrations and reversal potentials are all constant, it is fairly straightforward and easy to ensure that the model will have whatever resting potential you like.

2. If concentrations and/or reversal potentials are state variables that can change in the course of a simulation, it is fairly easy to initialize the model to steady state. BUT the membrane potential will almost certainly NOT be what you want. It may not even be uniform throughout the entire cell. For some models, it MIGHT be possible to devise and implement a strategy that makes the model have a particular resting potential at one location, or even throughout the entire cell, but there is no guarantee that this can be done, and even if it can be done, significant conceptual effort and programming effort will be required.

Statement 2 applies to every cell class in your model that uses one or more MOD files that WRITE one or more ionic concentrations. Expertise in Python (or hoc, if your existing model is implemented in hoc), NMODL, and debugging will be required. Will you be able to muster the programming resources necessary to deal with this nontrivial task?