Thanks for the link and your comment.

The FI curve is important for characterizing the integrative properties of neurons. Using photo-stimulation instead of current injection we are able to investigate these properties under circumstances that was otherwise impossible or very difficult, such as during behavior. 

An important consideration here is whether you will be stimulating ChR2-positive presynaptic axons/neurons and measuring post-synaptic spiking, or whether you are measuring spikes in ChR2-positive neurons that are directly stimulated by light.

I think a before-and-after optical f-I curve as described above may work (and sounds neat), but I would caution against any comparison of f-I curves obtained via somatic current injection vs those obtained by optogenetic activation.

Another consideration: how will you modulate light? By duration? Or by intensity? 

Yes its a good point regarding your distinction between activating pre-synaptic potentials versus postsynaptic ChR2 activation. If the sole purpose of is to characterize the input/output properties of a neuron under different circumstances, a distinction between those two types of activation may not be so important as long as the activation remain the same under the different situations.

Regarding modulation of light: I was thinking either pulsing with different frequency or changing the intensity. I'm leaning more towards the latter, given the reference William provided, it seems that the FI-curve is sigmoidally shaped when the light intensity is increasing on a logarithmic scale. It is difficult to change intensity over several orders of magnitude by only changing the pulse frequency... 

Hi guys -

Just a couple of additional point related to this interesting discussion. If light is used instead of injected current for stimulation of a neuron, then we should talk about F-Lx function, not F-I curve. Due to many reasons, F-Lx can not be compared to F-I curve (see some below). Also, direct stimulation of neurons with current implies injection of a definite quantity of electrical charge, while with optical stimulation, this becomes quite complicated. Moreover, the composition of currents (sub-threshold) driving neurons into firing mode are quite different (passive membrane depolarization when injected with current vs. active conductance from the resting state activated with light). 

Next, if current is injected directly into a neuron, a fair assumption is made that its synaptic inputs remain more or less unaffected (although this is not always the case), whereas if neuron is stimulated with light and synaptic inputs are not blocked, the later will become activated and affect the characteristics of evoked responses (as mentioned above by Benjamin) as well as the slope of F-Lx.  

And a final point, while the idea of using extracellular recordings for  assessment of electrical responsiveness of neurons to stimulation is attractive (and is less invasive than whole cell current clamp recording), in light of above mentioned points with likely activation of synaptic inputs, I think this would make sense only if carried out under the complete blockade of synaptic currents activated by (ChR2), which is possible in brain slices but very challenging in vivo...   

Hi Saak

I don't get why its a problem activating pre-synaptic input. If we want to probe the integrative properties of a neuron, we want to give it input and measure its output. Whether the input is via ligand gated ion channels activated from a pre-synaptic neuron or whether it is from a photosensitive channel in the membrane doesn't make a difference, in my mind.  There could be a problem if these input would change under other circumstances, but the most fair assumption is that the pre-synapses have on average the same sensitivity. You could even argue that synaptic input is more close to real physiologic input than the ChR2-generated currents...

Perhaps we should be clear as to whether we are talking about input-output relation (neuronal firing generated by activation of synaptic inputs) or F - I curve (which is a measure of the responsiveness of neuronal membrane to direct stimulation with depolarizing current). While both independently can provide very useful information about the integrative properties of neurons, if combined together, the readout will become complicated. I guess this is exactly what is likely to happen when a neuron is stimulated with light, which in addition to post-synaptic membrane (direct activation) stimulates also presynaptic terminals (synaptic inputs). If somehow you manage to stimulate with light only synaptic inputs, then with extracellular recordings from soma you will be closer to physiological activation of exclamatory synaptic inputs, and on this you are right.

Adding one more point regarding direct/synaptic inputs - I believe that strong direct activation of a ChR2-positive neuron can lower its input resistance substantially, but I don't have a reference handy ... and unlike somatic current injection, in most cases this effect will be distributed throughout the dendrites as well.

Which brings me to another point: depending on your choice of objective lens (specifically, magnification), together with brain area and cell type, the exact positioning of your field-of-illumination can have a substantial effect on the effective stimulation:

In slices you'd care whether you are illuminating the somatic compartment only, or the entire dendritic arbor, or a subset of the dendritic arbor. Furthermore, since you will likely have multiple ChR2+ neurons in your slice, you would see a combination of direct and synaptic excitation - the ratio depending on how many other neurons (and their processes) are within your field-of-illumination. In slice, one way to achieve consistent illumination is to use an inverse Koehler configuration to stimulate via the epifluorescence path of your microscope - by using a low magnification objective lens and adjusting stage position, you can ensure even illumination across the full dendritic arbor (and i.e. cortical layers).

In vivo it can be more difficult to achieve even, repeatable illumination. If you illuminate the surface of the brain, most of the activated ChR2 will be near the cortical surface (mostly in L1, some in L2/3, very little below that) because of the steep decrease in intensity of blue light at greater depths (see for example this openoptogenetics.org wiki page for a collection of publications).

http://www.openoptogenetics.org/index.php?title=Transmission_Through_Brain_Tissue

A lot of factors need to be considered when generating input/output curve with presynaptic terminal stimulation as already discussed by others; ChR expression level/spread at the presynaptic brain region is very important as this can affect the density of ChR-expressing fibers at the postsynaptic site. ChR expression time is also critical.