A pair of dye-coupled hippocampal neurons. Courtesy of D. Schmitz, University of California San Francisco, USA.

It seems that the more we find out about the nervous system, the more we have to overturn our old models in favour of newer, more complex ones. The simple idea of an axon as a fairly passive transmitter for the output of a neuron is gradually being eroded as evidence accumulates that axons have a more active role in signal processing and integration. Schmitz and co-workers add to the story with the demonstration that the axons of hippocampal neurons are electrically coupled by gap junctions.

Hippocampal pyramidal and granule cells show small, rapid somatic depolarizations known as 'spikelets'. These had been assumed to arise from attenuated dendritic action potentials, although more recently it was suggested that they were related to electrical coupling. Schmitz et al. carried out a detailed analysis of these spikelets and showed that stimulation of the axon of a neighbouring neuron often gives rise to spikelets in the soma of a pyramidal cell. Propagation of the spikelet to the soma depends on activation of fast Na+ channels in the axon of the recorded neuron, and the generation of spikelets in the recorded neuron by the action potential in the stimulated neuron is reduced by the use of carbenoxolone and other manipulations to prevent gap junction transmission.

The authors proposed that the signal was transmitted from one axon to the other through gap junctions linking them. Further evidence for this idea came from an experiment in which spikelets were recorded in the axons of pyramidal cells after nearby axonal stimulation of another neuron. These spikelets preceded those recorded from the cells' somata.

Schmitz et al. also looked for structural evidence of axo–axonal gap junctions. They found pairs of dye-coupled pyramidal cells with clearly separated somata, showing that the cells' somata could not be coupled directly. In a few instances it was possible to identify the location of coupling as the axon of the coupled neuron. In these cases, the axonal coupling distance was 50–120 μm from the soma, which is before myelination usually begins.

What is the function of this axo–axonal coupling? It would certainly enable very fast communication between hippocampal neurons. The authors suggest that axonal coupling might underlie high-frequency oscillations and the synchronization of neuronal activity across networks, and could contribute to the abnormal discharges seen in epilepsy. They also comment that neuromodulators could open and close the gap junctions, thereby altering the network connectivity under different conditions, which would provide the system with great flexibility.