Why the textbook account of spike initiation is wrong

In previous posts, I have tried to explain why spike initiation is “sharp” or abrupt, basically as in an integrate-and-fire model. The explanation relies on the fact that spikes are initiated in the axon, next to the soma. What may not be so obvious is how this differs from the textbook account of spike initiation. That textbook account would go as follows: first, the spike is initiated at a distal site in the axon, through the interplay between transmembrane currents, then the spike is propagated to the soma. On the course of its propagation, its shape changes and becomes sharper (see the explanation in this paper for example).

This account is wrong on several levels. But fundamentally, it has to do with the inadequacy of the transportation metaphor. Let us take a step back. What do we mean when we say that a spike propagates along an axon? In fact, not much actually moves physically along the axon. There is very little ion movement in the longitudinal direction. Ions move mostly through the membrane, in the radial direction. It produces a spatiotemporal pattern of electrical activity, and the reason we say a spike propagates is that in certain conditions that pattern is a solitary wave for which we can define a velocity. This is actually how Hodgkin and Huxley predicted spike velocity in the squid giant axon, by postulating that the potential V(x,t) can be written as U(x-v.t), and then looking for the velocity v that was consistent with their equations. This velocity has actually nothing to do with the velocity of the ions that carry the electrical charges. So basically we often describe electrical activity in axons in terms of transportation of spikes, but one should keep in mind that this is just a metaphor.

Now when we say that there is a spike that moves and gets transformed on its way, we should realize that the metaphor has reached its limits. As nothing actually gets transported, we are not saying anything else than there is an irregular spatiotemporal pattern of activity. That is, we are not saying anything at all. That is all the more true when we are talking of time scales shorter than the duration of a spike (as in the backprogation from initiation site to soma).

The transportation metaphor, unfortunately, is highly misleading. It leads to incorrect reasoning in this case. Here is the reasoning in question. The onset of a spike at the axonal initiation site is smooth. The onset of a spike at the soma is sharp. Therefore, the spike onset gets sharper through the propagation from initiation site to soma. On the surface, the reasoning seems flawless. But now here is a disturbing fact: from the spike shape at the axonal initiation site, I can make a quantitative prediction of onset rapidness at the soma, without knowing anything of what's in between (that's in a paper being submitted). Therefore, onset rapidness at the soma is determined by properties of spike initiation, not of propagation. How can that be?

The flaw is this: in the transportation metaphor, somatic and axonal spikes are implicitly seen as local events both in space and time, which can then be related by the transportation metaphor (object at place A and time t1 travels to place B at time t2). As mentioned above, the relevance of the transportation metaphor is questionable at this spatial and temporal scale, all the more when what is being transported changes. What I showed in my previous paper and review is precisely that spike initiation is not local. It is determined both by the properties of local Na channels and by resistive properties of the coupling between soma and axonal initiation site, which are not local. For example, if you moved the Na channels away from the soma, the neuron would become more excitable, even though local properties at the initiation site would be identical. Spike initiation is not a local phenomenon because of the proximity of the axonal initial segment to the big somatodendritic compartment.

Thus the sharpness of somatic spikes is actually determined not by propagation properties, contrary to what was claimed before, but by spike initiation properties. The catch is that those properties are not local but rather encompass the whole soma-initial segment system.

I had previously called this biophysical explanation the “compartmentalization hypothesis”. I realize now that this can be very misleading because physiologists tend to think in terms of initiation in one compartment and transportation from one compartment to the other. I will now use a different terminology: “critical resistive coupling”, which emphasizes that spike initiation relies on a system of coupled compartments.