<b>The Memory Hole</b><br/><br/>

The Memory Hole

After earning bachelor's and Master's degrees from Tsinghua University in Beijing, China, Wen-Biao Gan came to the United States for doctoral work at Columbia University (New York, NY). He then went to Washington University School of Medicine for postdoctoral studies under Jeff Lichtman, where he worked with a new line of transgenic mice that express green fluorescent protein (GFP) in neurons of the cerebral cortex. He joined the faculty at NYU's Skirball Institute in 1999.

Gan has been a prolific researcher since graduate school, with more than two dozen publications under his belt. He has also won several awards, including the Peter Sajovic Memorial Prize for outstanding work in Biology, a Whitehead Fellowship for Junior Faculty in Biomedical Science, and a New Scholar Award from the Ellison Medical Foundation.

Alan Dove, Ph.D. Whether we're learning a skill, meeting someone for the first time, or tasting a new flavor of ice cream, our brains record memories by rewiring the synapses between neurons. This remodeling occurs rapidly and commonly. Or perhaps very slowly and rarely. Or maybe somewhere in between.

Distinguishing those possibilities is important for everything from basic neuroscience to the development of new treatments for Alzheimer's disease, but researchers who study synaptic remodeling have long faced a fundamental anatomical problem: mammalian skulls are opaque. In simple, transparent organisms, scientists can label neurons fluorescently, then watch the synapses remodel in real time in response to learning. With higher organisms, researchers have largely been confined to studying fixed brains and reasoning by analogy.

"You cannot get information about dynamics from fixed tissue," says Wen-Biao Gan, an assistant professor of physiology and neuroscience at New York University's Skirball Institute (New York, NY).

During his postdoctoral work, Gan was among the first researchers with access to a line of transgenic mice expressing green fluorescent protein (GFP) in a subset of neurons in the brain. The mice were useful, but Gan was still frustrated by his inability to image the cells in the intact, active brains. He wanted to look inside the skull of a live mouse.

A close shave

Specifically, he wanted to use two-photon microscopy to image the GFP-expressing cerebral neurons in three dimensions. Two-photon microscopy works by focusing a laser beam into a fluorescently labeled sample. The laser excites fluorescence only in the plane of focus, allowing the investigator to image "slices" of the sample at specific depths, then construct a 3-D view.

Two-photon microscopy is a powerful, cutting-edge technique, but the scientists still had to figure out a way to make a mouse skull transparent. For that, they went decidedly low-tech.

"I was thinking that we could thin the [skull] bone to a very thin layer if we are careful enough, and not actually penetrate the bone," says Gan. It was a tricky procedure. "We tried many ways. A drill ... didn't work so well, then we tried sandpaper, and that doesn't work ... so eventually we used a small surgical razor blade just to scrape the bone," he says.

Once they've thinned a small patch of the skull to about 15 micrometers, the two-photon microscope can image and map the synapses underneath. Because synapse remodeling takes place over a period of days or weeks, during which time the thinned skull begins to thicken and heal, the team must re-scrape the same patch at different time points, locate the same neurons again, and then map the changes in their synapses. It's tedious, but it works.

Don't do windows

Pruning Synapses along a dendrite as a young mouse ages.

Indeed, the method has already uncovered some major surprises. One of the team's first discoveries was that the synapses in the mouse brain are remarkably stable, with 74% remaining intact over most of an adult mouse's lifespan. That contradicted results from an alternative system, in which researchers had actually drilled holes all the way through the mouse skull and installed glass windows to do the imaging. In the mice with windows, synaptic turnover was rapid and widespread, with about 20% turnover per day. "So this was a big controversy," says Gan.

The team decided to test the two methods side-by-side, comparing not only synapse remodeling but also the general health of the animals' brains. "If you remove the skull [patch] and place a cover glass, you see dramatic changes of the brain turnover of synapses, and at the same time we also find all these immune cells, like microglial cells and astrocytes. Clearly by doing that you cause a lot of changes in the brain," says Gan. Thinning the skull does not cause the same immune response.

Gan's technique has caught on quickly, and he estimates that about two dozen labs are now doing two-photon microscopy through thinned skull patches. Several of those groups have visited NYU to learn the difficult technique, but Gan cautions that it still has its limitations.

One problem is that the two-photon microscope can only penetrate a few hundred microns into the cerebral cortex. "To look at subcortical structures I think is still very challenging," says Gan.

Nonetheless, the surface of the cerebral cortex provides a lot of interesting territory for neuroscientists to explore. In one recent study, for example, Gan and his colleagues looked at a mouse model of Alzheimer's disease. In these mice, amyloid plaques appear to cause nearby synapses to turn over faster, and also cause adjacent neuronal dendrites to shrivel and disappear. Drugs that inhibit that response might eventually help treat this debilitating disease. Not bad for a razor blade.