Autapses are synapses that a neuron makes on itself. Although there is structural evidence for the existence of autapses in the brain, it is not clear whether and how they affect neuronal function. Now, new data in The Journal of Neuroscience indicate that a subset of interneurons make autaptic connections that help to sculpt inhibition in the neocortex.

Bacci et al. recorded from different types of interneuron in cortical slices and found that fast-spiking interneurons formed inhibitory synapses on themselves. Depolarizing a fast-spiking interneuron led to the appearance of a GABA (γ-aminobutyric acid)-mediated synaptic current on the same cell shortly after the action potential. But is this current the result of true autaptic activity? Although the short latency of the current provides a good indication of its autaptic nature, the authors reasoned that if the GABA-mediated current was elicited by autapses, then interfering with transmitter release in the same neuron should readily block it. They therefore injected the calcium chelator BAPTA into fast-spiking interneurons, causing the currents to disappear. By contrast, this manipulation did not affect GABA-mediated currents elicited by neighbouring interneurons.

What is the function of these autapses? One clue came from experiments in which Bacci et al. depolarized a fast-spiking interneuron twice, in quick succession, and asked whether the efficacy of the second stimulus to elicit an action potential changed when autapses were active. They found that if they blocked autaptic transmission with a GABA antagonist, the second depolarization was more effective in eliciting an action potential, indicating that autapses might provide a form of feedback inhibition in the neocortex. They also found that autaptic activity had an inhibitory effect on repetitive interneuron firing; blocking GABA-mediated transmission while the fast-spiking interneuron was made to fire repeatedly led to an increase in firing frequency.

These results show that autapses are not mere artifacts but actually shape inhibitory activity. It will now be important to investigate the implications of this new form of feedback inhibition for rhythmic activity, network oscillations and other cortical patterns of firing that involve the activity of interneurons.