Coming from the Rockefeller University in NY, Matthieu Louis leads the Sensory Systems and Behaviour group at the CRG, the only lab in Barcelona, and one of the few in Spain, investigating Drosophila neuroscience. His team comprises eight people with backgrounds in molecular biology, engineering and physics. Their aim is to correlate neural circuit function with behaviour using fruit fly larvae. “The Drosophila larva has a repertoire of complex behaviours and key cognitive functions. Yet its nervous system has 10 million neurones fewer than humans”, explains the physicist.
The group tries to understand how odours are encoded by the olfactory system. Features such as quality, “Does this smell like banana?”, and intensity, “Is this a morsel of banana or a bunch?”, are efficiently represented by only 21 olfactory sensory neurones, so that the larva can distinguish between hundreds of food-related odours. The researcher says that there must be a combinatorial code, yet it does not seem to be as trivial as the activation of different combinations of neurones by distinct odours. “We have evidence that the nature and the intensity of an odour is represented not only by the identity of the sensory neurones it activates, but also the way each one is activated”, he explains.
From information processing to chemotaxis
Once a smell has been encoded, it has to be processed. To find the higher-order neurones involved in the integration of olfactory information, the group is undertaking a large behavioural screen. They test thousands of fly lines in which subsets of neurones are inhibited or over-activated. They then characterise how these perturbations affect chemotaxis, the orientation behaviour observed in response to an odour gradient. To decide whether to go straight ahead or turn, the larva monitors information about odour concentration changes. When a wild-type animal detects an intensity increase of an attractive odour, it keeps going forwards, but, as the group has recently described, if the odour intensity decreases the larva reorients through an active-sampling mechanism: much like rats and dogs, the larva sweeps its head laterally to check intensity levels on either side.
With their screen, the researchers are looking for mutants showing reorientation defects. To that end, they have developed their own computer-vision software. “We needed an algorithm to quantify subtle movements of the head and body posture with a high space-time accuracy. As no tool like this existed, we spent a year developing one”, says the head of the group.
“If we understand the neural logic of larval chemotaxis well enough, we should be able to synthetically produce predictable behavioural sequences”. Such a model could be useful for robotics. “Currently, dogs are trained to find mines. We could design robots that navigate spatially, searching for the chemical compounds present in explosives”.
Many questions remain to be answered before this becomes reality. How does the larva integrate a series of stimuli (touch, light, heat, smell) that are received simultaneously before deciding what to do next? And how is sensory input converted into motor output? “There are many challenges ahead. But it is thrilling to witness the genesis of a decision in a minibrain, from elementary spikes in sensory neurones down to the coordinated contraction of dozen of muscles. Flies have much to teach us about the function of our own brain”, concludes the Belgian researcher.
This article was published in the El·lipse publication of the PRBB.