Researchers studying locust olfaction have uncovered a system of learning and memory previously thought to exist only in vertebrates.

Although insect and vertebrate olfactory systems were known to have a number of anatomical and physiological parallels, the mechanisms underlying insect olfaction were not well understood. Neuroscientist Gilles Laurent, at the California Institute of Technology in Pasadena, chose the locust as a model to study the neural mechanisms insects use to learn odours.

Laurent first began studying insect olfaction because it is a good way of tracing circuit function and neural computation in the brain. In this work, he and his graduate student, Stijn Cassenaer, turned their attention to the electrophysiological mechanisms at work in locust olfaction.

Odour representations are initially processed in an insect's antennal lobe, a structure analogous to the mammalian olfactory bulb. From there, they are projected to very selective neurons called Kenyon cells (KCs), which are located in a brain structure known as the mushroom body. This is the presumed centre of learning and memory in insects. Laurent and Cassenaer focused on a third layer of odour-processing cells — a population known as β-lobe neurons (β-LNs), found at the output of the mushroom body. “We hypothesized that if learning occurs in this structure, it's most likely at the synapse between the KCs and β-LNs,” says Laurent.

Although the locust is an unconventional model organism and little is known about its genetics, Laurent recognized it as having several advantages. First, it is a large, sturdy insect, making it ideal for electrophysiological experiments that require embedded brain probes and lengthy recording sessions. Second, many of its neuronal populations can be identified by their electrophysiological signatures, which makes interpretation much easier.

The two researchers presented locusts with odours typically encountered in the field, and recorded activity in pairs of β-LNs. They discovered that β-LNs are tightly synchronized, which indicated that timing is important in learning odours (see page 709).

Cassenaer then devised a way to make paired electrophysiological recordings of KCs and β-LNs in vivo. This was no small feat, but his ingenuity and patience paid off, and during one recording session he and Laurent observed a KC and a β-LN fire at the same time. In the next trial, only 10 seconds later, they noticed that the connection strength between the two neurons had been considerably enhanced. This form of synaptic learning and strengthening of nerve-cell connections, known as spike-timing-dependent plasticity (STDP), was thought to exist only in vertebrates.

The two then hypothesized that information travelling through several layers of neurons would inevitably result in a progressive decrease in timing precision. For synaptic learning to take place, timing is everything. The β-LN synchronization necessary to 'learn' an odour is dictated by the change in synapse strength that results from the temporal relationship of pre- and post-synaptic spikes. “In retrospect, it is obvious that mechanisms must exist to correct timing errors, given its importance here,” says Laurent.

To test the idea, they came up with a way of artificially modifying the timing of β-LN spikes evoked by an odour in an intact animal in real time. They found STDP to be adaptive — delaying or advancing spike time for each neuron cycle, and thus fine-tuning the β-LN spikes to work as a timing-regulation system. “Theorists had come up with the idea of a timing-regulation system without proof from the biological side,” says Laurent. “This is the proof.”