Neuron to Glia Synapses on Axons?

I posted a couple months ago about neuron to glia (in this case oligodendrocyte) synapses in the hippocampus, and how researchers had shown that these synapses were capable of LTP. This was an example of two themes 1) the brain is a tricky business -- particularly with respect to information processing and 2) glia are much more important than we thought they would be.

Here is another set of articles in that vein. Publishing independently in Nature Neuroscience, Kukley et al. and Ziskin et al. has shown that there are neuron to glia synapses in the corpus callosum AND that these synapses involve the release of neurotransmitter from the axon.

First some background.

When we think of synapses we usually think of the chemical junctions between two neurons. An action potential is transmitted down the axon to the axon terminal. In the axon terminal vesicles fuse with the membrane in response to the action potential; this releases neurotransmitters into the synaptic cleft. The neurotransmitters bind receptors on the dendrites of the post-synaptic neuron, triggering electrical potentials in that neuron which can in some cases result in an action potential in the next neuron down the line.

All very canonical, that.

In general, we do not think of the middle of the axon itself as the place where neurotransmitters are released. The axon is the place where the electrical signal -- the action potential is transmitted -- and it is evolved to do this very quickly. One of the evolutionary advances that is present in vertebrates is that of myelin.

Myelin is a fatty membrane wrapped around the axon that functions as an electrical insulator. It is formed by a cell called an oligodendrocyte (in the central nervous system). The myelin speeds electrical transmission of the action potential by preventing the electrical potential from leaking out of the axon. One of the really big highways of axons in the vertebrate brain, by the way, is called the corpus callosum. It is an area in the middle of the brain that has lots of axons and myelin because it connects the two hemispheres together.

Typically, neurons are no thought to divide to make new neurons, although there has been considerable evidence over the last several years that they do in some cases. Oligodendrocytes, on the other hand, are manufactured continuously. An oligodendrocyte precursor cell -- one that is in the process of being made but is not mature -- is identified by a particular marker called NG2+ -- hence they are called NG2+ positive cells. NG2+ cells will divide and then go seek out an axon to myelinate by sending out processes. (I posted a video of this last year.)

This bring us to the studies at hand.

The authors show that neurotransmitters on released from the middle of the axons result in electrical stimulation of NG2+ cells in the corpus callosum. (Remember that the corpus callosum is where axons and myelin live. We did not believe that there would be any synapses there.) They show that:

1) Action potentials in the axon result in quantal release of glutamate (a neurotransmitter) that is inhibited by inhibitors of vesicular fusion.

2) That this quantal release results in measurable electrical activity in NG2+ positive cells through its action on AMPA receptors.

3) These axonal-glial synapses are ultrastructurally very similar to normal synapses.

Below is an electron micrograph of that part (click to enlarge). On the left is a synapse from the cortex (the Ss are dendrites). Note the postsynaptic density -- the thick dark line on the post-synaptic side. On the right is the synapses they observed in the corpus callosum. The DsRed cell is an NG2+ cell. Note how it has the same density.

This is really interesting stuff for a variety of reasons.

1) The means of axonal release of glutamate is exactly like at a synapse, but I am not aware of other studies that have shown it happening in the middle of an axon. That is just a really weird spot.

2) The significance of the glutamate to the NG2+ cell is unclear. It is possible that it is regulating myelination of the axon. Lots of glutamate would indicate to the NG2+ cell that it is a really busy axon that should be myelinated.

Here is a summary of the importance of this work in a News and Views by Gallo in the same issue:

Both groups assembled an impressive amount of data to support the unexpected existence of functional axon-OPC synapses in the corpus callosum. Patch-clamp recordings from NG2+ cells showed that action potentials traveling through callosal axons triggered release of glutamate from terminals at discrete sites along these axons. Ultrastructural analysis demonstrated the presence of specialized synaptic junctions, and immunohistochemistry showed the expression of several synaptic vesicle proteins. As in neuron-neuron synapses, the authors found that the neurotoxin alpha-latrotoxin, which promotes Ca2+-dependent vesicle fusion in glutamatergic terminals, increased the frequency of synaptic events recorded in NG2+ cells. Furthermore, confocal laser scanning microscopy on rat callosal axons filled with a calcium indicator revealed calcium transients dependent on the generation of action potentials. Finally, when refilling of neurotransmitter vesicles was inhibited, or activation of voltage-dependent Ca2+ channels in the axon terminals was blocked, the postsynaptic response decreased significantly. Thus, plenty of evidence supports the release of glutamate from presynaptic terminals of the axon-OPC synapses through a Ca2+-dependent vesicular mechanism.

In many ways, the newly identified synapses found on NG2+ cells in the corpus callosum of the rat and mouse resembled glutamatergic synapses between neurons in the gray matter. Synaptic facilitation occurred after paired stimuli were applied to callosal axons, and increasing stimulus intensity progressively enhanced glutamatergic responses in NG2+ cells, implying that each OPC receives several inputs. Yet axon-OPC synapses also had distinguishing characteristics. In rat corpus callosum, the intercellular cleft showed an irregular morphology, varying in width and lacking extracellular matrix, indicating that these axon-OPC synapses might be 'looser' than their neuronal counterparts3. Consistent with this, NG2+ OPCs in the corpus callosum could sense lower levels of extracellular glutamate than postsynaptic neurons in gray matter, which would explain the smaller amplitude of AMPAR-mediated currents in OPCs. However, although biochemical studies show that NG2 interacts with AMPAR subunits through association with the postsynaptic protein GRIP1, the role of NG2 in establishing or maintaining axon-OPC synapses is unclear.

...

The function of neuronal-glial signaling in white matter remains somewhat speculative, but previous work offers some clues. In various in vitro culture systems, activation of AMPARs in OPCs reduces proliferation and inhibits differentiation11. The studies of Kukley et al. and Ziskin et al. indicate that the vast majority of NG2+ OPCs are synaptically connected in the corpus callosum. Therefore, axon-OPC glutamate-mediated signaling may have significant developmental implications and represent a major physiological cue for OPCs to sense the presence of an axon and stop dividing. If this is true, other signals promoting OPC differentiation would need to overcome glutamate-induced inhibition to generate myelinating oligodendrocytes in the vicinity of glutamate-releasing axons. We should continue to identify these signals and elucidate the intracellular mechanisms that promote OPC development in an unfavorable glutamatergic environment. (Emphasis mine. Citations removed.)

Really cool stuff.