Based on the comments in this post and also this chat. For discussions and speculations please comment in the chat.

The basic question is what is the advantage of having a single long axon such as that of sciatic nerve (~1m) compared to a series of neurons?

The primary advantage is surely the conduction speed which will get affected because of chemical synapses. However, gap junction synapses can reduce that delay. The larger volume of the soma of the post synaptic neuron may also reduce the conduction speed, but signal strengthening can also happen. Cell architecture can be adjusted to minimize the loss of speed.

The obvious disadvantage of having a single large cell is that there would be tremendous burden on the nucleus for the cellular maintenance. There would be delays in transfer of non-electrical signals such as biomolecules from axon termini to soma and vice versa. Long axons also means continuous transport along them which in turn demands large number of molecular motors and therefore ATP. Moreover, a small injury can disable the entire conduction channel. It can be imagined that this trait (long axons) can also be selected against; injury leading to inability to move and thereby causing perishment.

EDIT (Based on comments to Anne's answer)

Time is of course a valid advantage for having a single axon as an information conduction channel. However, the energy or the maintenance cost is the factor that I am actually interested in. Maintenance cost would increase with cell volume (in fact surface area) and having serial neurons would demand more energy (cumulative) than a single axon (including costs of maintaining a nucleus). However a long cell would need much higher number of molecular motors to maintain the traffic flow rate. All responses that require transcriptional control would be slow (such as response to injury). Moreover a single soma also imposes a limit on the number of mitochondria. There should be a limit on how long an axon can be. I am especially interested in case of big animals (with long hindlimbs) that also have a good reflex (perhaps camels, even horses).

Has someone tried to analyse the tradeoffs of having a long axon vs series of neurons connected by gap junctions? If not is this problem worth modeling or quite trivial?

Saw this in Quora:

My first thought mostly agreed with the question details, except I'm pretty sure that the dorsal root ganglion (DRG) has a longer axon than the motor information carried in the sciatic nerve (which is the longest nerve in the body, but not axon).

Does this mean that sciatic nerve has serial neurons?

  • 1
    $\begingroup$ Forget conduction speed. What about signal fidelity? How much noise does each extra gap junction add to a signal? $\endgroup$ – tel Jan 6 '15 at 4:32
  • 1
    $\begingroup$ @tel I am not sure about the noise at the gap junction (GJ). Can you tell me how is noise generated at the GJ? $\endgroup$ – WYSIWYG Jan 6 '15 at 5:27
  • $\begingroup$ I'm not a neruo guy, so I can't give you any specifics, but in terms of general information theory/2nd law considerations, every time you convert a signal from one form to another (axonal current to neurotransmitter gradient, say) your signal/noise ratio gets a little worse. You can recover the original signal somewhat, but it always involves an expense of free energy. $\endgroup$ – tel Jan 6 '15 at 16:51
  • 1
    $\begingroup$ @tel Oh well.. but in the gap junctions the wave can move to another cell. There are no neurotransmitters involved. $\endgroup$ – WYSIWYG Jan 6 '15 at 17:18
  • 1
    $\begingroup$ If you want to reduce the variance in signal transmission with n serial steps, then you need to add more steps. Therefore to achieve signal transmission with really low variance, you either want a single axon (case A) or really lot of them connected in series (case B). May be overall, case A is not so bad compared to case B not to be selected for. $\endgroup$ – Dilawar Jan 9 '15 at 9:26

First, I must clarify that a nerve is not the same as a neuron. A nerve is a collection of axons (with each axon a part of an individual neuron) in the peripheral nervous system. Thus, when you mention “single long axon such as the sciatic nerve”, this is technically an incorrect phrase as the sciatic nerve is a collection of many long axons. If one of these axons is nonfunctional, this would not severely affect the function of the entire sciatic nerve. However, you are right that if the entire sciatic nerve is severed, then there would be serious consequences downstream.

Second, the primary advantage of conduction velocity is less dependent on length and more dependent on presence and degree of myelination (directly proportional to fiber diameter) and presence of nodes of Ranvier.

With regards to the content of your question, the advantage of length is exactly what one might think--- greater distance. The sciatic nerve is the longest nerve in the body in order for our nervous system to innervate as far as the tip of our toes. Having hundreds of side-by-side series of neurons compared to hundreds of side-by-side axons seems like a far less efficient system in a couple of ways in exactly the two ways you mention: energy and time.

Energy: Having serial neurons would require an increase the number of neurons needed to transmit a message. Having more neurons requires more Na+, K+ and Na/K pumps and individual channels, requiring more energy.

Time: Having serial neurons would also require an increase in the number of synapses between neurons— either electrical or chemical. If the synapses are chemical, then time is increased many-many fold. If the synapses are electrical, then time is slightly increased by 10-fold per synapse compared to chemical synapses, but this would still take more time compared to having a single axon with myelination and nodes of Ranvier which by about 100-fold faster than a single electrical synapse.

In summary, having longer nerves made up of axons rather than serial neurons is beneficial by improving energy and time efficiency, which allows our bodies to effectively innervate structures farther away.


Regarding the ATPase absolute number increase in serial vs one long axon, I assume this will depend on the number of neurons in series. However, I read this paper, which says that increased length between each node of Ranvier has been physiologically associated with increased conduction velocity. This suggests fewer needed nodes, despite concentrated ATPases at these locations, and thus fewer ATPases. I haven't been able to find studies which compare ATPase concentration in the soma compared to nodes of Ranvier within axons.

Regarding energy use in soma vs axon, mitochondria are not only made in the soma but are concentrated in pre- and post-synaptic membranes (and growth cones). The more of these structures there are (as in serial neurons), the more presumed mitochondria. See mitochondria and neuronal activity.

Finally, my literature search could not find studies which showed longest possible energetically feasible length for myelinated or unmyelinated neurons.

  • $\begingroup$ Why would multiple neurons need more ATPases? The only difference is more cell bodies? Could you provide a citation for the gap junction transduction velocity? I would think it would be about as fast as myelinated conduction. If I am wrong than this is a likely answer! $\endgroup$ – AliceD Jan 9 '15 at 5:00
  • 2
    $\begingroup$ @ChrisStronks I will add references tomorrow, I used quite a few so it will take time. $\endgroup$ – Anne Jan 9 '15 at 5:19
  • $\begingroup$ @ChrisStronks Multiple neruons need more ATPs as they will have other soma related function to which they have to cater to. More soma=more protoplasm=More volume of space where electrolyte balance need to be maintained=More channels active=More ATP $\endgroup$ – One Face Jan 9 '15 at 5:40
  • 1
    $\begingroup$ @CRags Requirement of ATP is more or less related to total volume of the cell which would be slightly more in case of serial neurons but not overwhelmingly high. Also a long cell would need much higher number of molecular motors to maintain the traffic flow rate. All responses that require transcriptional control would be slow (such as response to injury). $\endgroup$ – WYSIWYG Jan 9 '15 at 5:42
  • 2
    $\begingroup$ @WYSIWYG With neurons, diameter plays a much larger role than length. Nernst Potential and Nerve Conduction Speed $\endgroup$ – Anne Jan 9 '15 at 6:25

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.