I spent a semester in college as a teaching assistant in an animal behavior lab. One of the very most frustrating failed experiments was the mating assay: we put two flies, a male and a female, in Eppendorf tubes with an air hole, then waited to see if strain 1 and strain 2 were as likely to cross-mate as to mate with members of their own strain. These particular flies had a loooooong latency to mate, and as you can imagine, the roomful of undergrads made a whole lot of jokes about mood music. When they finally began courtship behavior, the observers would often lean closer to track those tiny movements better. Bad idea. It turns out that for flies, a looming object is the biggest mood-killer there is.
We would have known that if we’d talked to Tom Clandinin at Stanford. His lab studies, among other parts of fly vision, the neurons that let the fly respond to looming objects. I got the opportunity to hear Dr Clandinin speak about his work recently, and I think the study, in Current Biology earlier this month, is so cool that I’m going to try summarizing it in the style of Scicurious and Ed Yong.
Dr Clandinin started out interested in how vision and behavior are connected in the fly brain. This kind of problem is much easier to study if you can identify smaller components, so in an earlier paper, the lab identified a group of neurons with something in common separate from other neurons. Using the knowledge that similar cells have similar patterns of gene expression, and a genetic tool called enhancer trapping, they inserted a reporter gene at random into the genome in a large group of flies, then looked for changes in behavior. This let them identify flies where a change in only a few cells caused a major change in behavioral responses to visual stimuli.
In particular, they found five neurons (shown in green in the image) in a connective region between the visual lobe and the rest of the brain, which they dubbed Foma1 cells (it stands for “failure of motion assay 1” and means that the flies responded weirdly to visual stimuli). The location of the cells is interesting, because the connective region is a chokepoint between two areas that can communicate lots of information, but relatively little information can get through. Think of it as an undersea cable between phone networks on different continents; some kind of integration or discarding of information must occur there, so that only important stuff gets through.
But how could they figure out what kind of information was passing through these neurons? What made the mutated flies behave differently than their wild-type brethren? They used an ingenious set-up: put a miniaturized TV screen in front of a fly, project simple, abstract moving images, and then use electrophysiology to listen in on individual neurons. They didn’t see any action in these five neurons when they presented a series of two-dimensional stimuli: rolls, yaws, up-and-down movement, or anything else that the fly could conceivably see while staying in one place and rotating in space (here’s an idea of the kind of image they used). Instead, the neurons fired like crazy in response to a looming stimulus: a square that got bigger and bigger on the screen. The lab used slight changes in the projected image to characterize the neurons further, finding that they fired more often when the looming object seemed to be approaching faster; they responded the same way to objects coming from any part of the visual field and in any color; and they didn’t respond to a general change in screen brightness without the looming illusion. This combination of features made the Foma1 neurons loom detectors; that is, what they recognized was the looming, and they seemed to respond to anything that loomed.
Figuring out what could stimulate the Foma1 neurons was a cool piece of neuroscientific sleuthing. The next question was, were they behaviorally important? The team knew that in response to the illusion of a looming object, 92% of flies would raise their wings and look like they’d fly away; about 75% would take off before the looming object was due to hit them. But when the group silenced Foma1 neuron activity by expressing an inactive ion channel selectively in the Foma cells, only 30% of flies escaped. The only problem with this test was that it couldn’t tell the difference between what the fly knows (i.e., loom detection) and how the fly responds. Maybe the entire visual lobe was yelling, “Mayday! Mayday!” and the Foma1 neurons were only important in passing the message along.
So to eliminate any confusion they used another cool technology: optogenetic stimulation. Again, this involved selective ion channel expression only in the foma neurons; they expressed channelrhodopsin, which is sensitive to light, and then blinded the fly and shone a bright light on it. This set-up activates only the Foma neurons and anything downstream of them, and it allowed them to show that activation of Foma1 neurons is enough to make the fly fly away.
They did one more interesting experiment, which isn’t in the paper; if the fly is already flying (glued to a stick, so that it has the illusion it’s supporting itself but also is held in one place), then instead of flying to escape a looming stimulus, it throws up its legs in a landing pattern. This led them to infer that there was a downstream decision point: if the fly was not flying, a looming object meant it should fly away (for instance, a fly swatter or an attentive student of animal behavior was approaching). If it is flying, the looming object means it should land—in other words, watch out for that… TREE!
“Loom-sensitive neurons link computation to action in the Drosophila Visual System” Current Biology March 6 2012 http://www.cell.com/current-biology/abstract/S0960-9822%2812%2900008-5