Because of limitations in funding and expertise, most laboratories choose to become skilled in one particular technique, be it behavior, molecular biology or electrophysiology – the practice of recording electrical activity in neurons. But as neuroscientists get closer to resolving some of the most complex mysteries of the brain, some researchers find themselves increasingly reaching the limits of those chosen methods. A behavioral researcher might wonder what cellular processes mediate the performance of an animal on a learning task, while a scientist studying neurons in isolation can only speculate about what those microscopic observations mean in an intact organism.
“Really the major problem in neuroscience right now is defining what is the underlying cause,” said Daniel McGehee, associate professor of anesthesia and critical care (and, full disclosure, my former thesis advisor).
The solution to that problem is collaboration, said McGehee and Xiaoxi Zhuang, associate professor of neurobiology, and a recent publication by the two researchers is a vivid example. Published late last month in the Journal of Neuroscience, their study of how an intracellular signal expressed in only one region of the brain mediates certain types of learning could only have been done by combining the strengths of their two laboratories.
A graduate student in Zhuang’s behavioral genetics lab, Mazen Kheirbek, was studying reward and learning in a mouse strain lacking the gene for an enzyme called adenylyl cyclase 5, or AC5. In the brain, AC5 is seen in only one place: a region called the striatum, which is involved in movement and the response to rewards like food and addictive drugs. Kheirbek found that mice lacking the AC5 gene appeared normal in most situations, but were slow to learn certain tasks that required a specific type of learning called response learning.
The test worked like this: animals were placed in a “cross maze” shaped like a plus-sign and filled with water. Mice were placed in one end of the maze, and asked to choose either the left or right arm to seek out a platform that allowed them to climb out of the water (mice will swim if necessary, but prefer to be dry). The platform was always placed in the right arm relative to where the mouse was originally placed, but the orientation of the maze was switched such that the mouse learned to always turn right to find the platform, rather than using landmarks to swim the correct direction.
This type of learning, sometimes called “habit learning,” is put into action when you follow a familiar path without really thinking about it. Think about navigating the hallways to your office – if you enter the building through your usual door, you can pretty much get to your desk with your eyes closed. But if you came in a different door in the morning, you must rely upon landmarks to find your office, which utilizes “spatial learning.” Response learning is known to occur in the striatum – damage an animal’s striatum and they’re much worse at this type of learning – while spatial learning occurs in another brain region called the hippocampus.
In Kheirbek’s mice, the striatum was intact, but missing the key signal of AC5. And sure enough, those AC5 knockout mice were much slower to learning the maze task, taking 2-3 times as long to remember to always make the right turn to find the platform.
This suggested to Kheirbek and Zhuang that adenylyl cyclase was critical for response learning. But explaining exactly what role it plays inside neurons of the striatum was beyond the scope of behavioral experiments. So Zhuang contacted McGehee, whose laboratory records the electrical activity of neurons from a brain slice. Though most of McGehee’s experiments take place in rats (which are bigger, not to mention nicer), he agreed to look at the electrical properties of neurons from AC5 knockout mice to see if there was a difference at the functional level.
Graduate student Jon Britt set about running the experiments, which involved comparing straital neurons from AC5 knockout mice to their normal counterparts. In most aspects, the neurons were exactly the same. But when Britt ran a test to see whether long-term plasticity – a phenomenon, thought to underlie learning, where the communication between two neurons is changed – could be induced in the neurons, he found long-term depression in neurons from normal mice, but not from AC5 knockout mice.
“Motor learning has been correlated very nicely with a weakening of synaptic strength in dorsal striatum,” McGehee said. “Initially we had a relatively easy question: is there a change in plasticity? And the answer was unequivocally yes.”
But when Britt attempted to more artificially induce long-term depression by directly applying a type of drug called an endocannabinoid, he successfully caused plasticity in both normal and AC5 knockout mice.
“Apparently all the machinery is still in place to induce the synaptic change,” McGehee said. That suggests that the researchers had removed only a single piece of the chain required to induce synaptic plasticity by knocking out AC5. The reverse assumption: AC5 is important for plasticity in the striatum, and by extension to behavior, that plasticity underlies response learning.
Zhuang and McGehee emphasize that such a conclusion remains unproven, as the electrophysiological and behavioral data can only be taken as a correlation. To truly show that AC5-mediated plasticity is necessary for learning the maze task, recordings would have to be taken from mice as they swam the maze, an experiment that is technically implausible at this time. But the researchers said that the collaboration inspired them to share animals and expertise on future projects, shining a light on neurobiological questions left in the dark by methodological limitations.
“It’s been really rewarding for us, and I think it’s going to continue to be really productive,” McGehee said.