Lights, Camera, Action

“We are now doing experiments on mammals that used to only be possible in insects,” says David Kleinfeld, a neurophysicist at the University of California, San Diego. “It’s a fun time.”

With living models come new challenges. Researchers also need to be surgeons, exposing the brains of mice with dexterity and choosing the approach that is the least likely to damage neurons. And the live models become patients, requiring imaging techniques that keep them healthy and comfortable in a setting that is as natural as possible.

The Scientist consulted with experts in live-rodent brain imaging to bring you a guide for peering into the working brain. Employing these techniques, researchers can return to the same glowing neuron over time, noting structural changes resulting from experience or drug treatment. And they can use indicator molecules that light up when neurons fire to reveal a continuously updating map of neural activity in the brain.

Let there be light

The first step is to light up the neurons of interest. The molecule of choice depends on the question under study. To examine the physical shape of individual neurons and their synapses, you’ll need to enable the neurons to express a fluorescent molecule, such as green fluorescent protein (GFP).

To follow a single neuron over time, cells should be sparsely labeled, says Anthony Holtmaat of the University of Geneva. Too many fluorescent neurons and you create “a forest where you have to identify a tree that you were looking at the day before,” he says. “But if you have just a few trees, your forest is not so dense.” A classic mouse line, called THY- GFP-M, has only a few neurons labeled throughout the brain.

To look at only certain types of neurons, there are two options. First, you can engineer a line of mice using the well-established CRE site-specific recombination system. This method works well if there’s a genetic marker that is expressed only in your cell of interest, says Robby Weimer, a senior scientist at Genentech. Researchers can use this gene’s regulatory region to target CRE expression to these cells. A fluorescent reporter gene inserted close to the regulatory region switches on only in the presence of CRE, thus restricting expression to those cell types. But CRE engineering involves complex crosses between mice, which take time.

CORTICAL CLOSE-UP: Images shot through a cranial window over a period of 8 days show details of the same presynaptic axon and postsynaptic dendrite expressing green fluorescent protein in a mouse cortex. The asterisks denote signal-receiving dendritic spines that are changing over time. COURTESY OF JEROME RANDALL AND ANTHONY HOLTMAAT

A second option is to introduce the gene encoding a fluorescent molecule into mouse brains in utero by injecting a virus or circular (plasmid) DNA. This approach allows researchers to target general brain regions, but not specific cell types. It’s possible to label more than one brain region with this method by injecting a combination of fluorescent molecules at the same time, says Weimer.

To track neuronal activity, rather than neuronal anatomy, researchers mostly rely on calcium indicators, called GCaMPs. These are GFP molecules altered so that they fluoresce only in the presence of calcium, which rushes into neurons as they fire. As with GFP, GCaMPs can be either engineered or injected into cells.

Getting in

Once you’ve labeled your neurons of interest, you’ll need to open a window into the brain. There are three possible approaches: replacing a piece of the skull with glass; shaving the skull so that it becomes paper thin; or reaching deeper into the brain with clear tubes. Each method has advantages and disadvantages, depending on the experiment at hand.

With the cranial-window technique, researchers remove a circular section of the skull about 3 mm in diameter, leaving the underlying protective membrane layer (the dura) intact. The excised skull piece is replaced with a custom-made cover slip, held in place with dental acrylic. The advantage of this technique is that it is relatively permanent: a cranial window remains stable for up to seven months.

Mice need to recover from the surgery before imaging begins, but researchers can then follow them over time without having to operate before every session. After about seven months, however, the bone starts regrowing and cannot be re-cut.

The primary concern with the cranial-window technique is that removing a piece of the skull is invasive and can cause inflammation if the dura is damaged, says Holtmaat, who published a protocol for the procedure (Nat Protoc, 4:1128-44, 2009). It’s important to wait at least 10 days after the surgery before imaging, he says.

LOOKING THROUGH THE SKULL: Affixing a glass cover over a surgically thinned area of the skull allows the study of changes in uninflamed, undamaged brain blood vessels over time. This image shows the thinned skull (blue) covering the cortex of a mouse. Green fluorescent protein (GFP) labels the cortical pyramidal neurons and a red fluorescent dye courses through the surrounding blood vessels.DATA ACQUIRED BY ANDY SHIH AND PROJECTED BY PHILBERT TSAITo image on scales longer than seven months, the skull-thinning approach may be preferable. The technique involves shaving a region thin enough that light can pass through it. Because the procedure is less invasive and does not carry the risk of contacting the dura, it is less likely to lead to inflammation. But thinning the skull must be done before each imaging session and causes pain to the mouse, which must be anesthetized, says Holtmaat. Researchers are also restricted to a smaller region of the brain, as thinning too much skull makes the brain “wobbly.”

A third protocol, which is a fusion of these two methods, addresses some of the drawbacks of each. This method fuses glass to the top of a thinned skull, stabilizing the opening and keeping it clear for months (Nat Methods, 7:981-84, 2010). Kleinfeld established the protocol to look at blood vessels, which are damaged by opening up the skull.

All of these techniques, however, reveal only the outermost layers of the brain. “Neuroscience is getting very biased by what’s happening in the upper layers of the cortex,” says Kleinfeld.

More recently, researchers have begun inserting small glass tubes directly into the brain to shine a light on previously inaccessible brain regions such as the hippocampus (Nat Med, 17:223-28, 2011). These tubes, developed by Mark Schnitzer’s laboratory at Stanford University and available commercially from Inscopix, are paired with optical probes that reach into the tube and are connected to a microscope. The probes can image tissues such as the hippocampus or the striatum, which are more than 1 mm deep inside the mouse brain.

Unlike the other techniques, the tubes penetrate the brain and can cause an immune response in their immediate vicinity. Schnitzer advises waiting at least two to four weeks after inserting the tubes before imaging.

Placing probes into the brain probably does not disrupt its function, but detailed microscopy of the surrounding cells should be done to be absolutely sure, says Kleinfeld. Whichever method you choose, the key is to be adept at surgery, which minimizes artifacts caused by inflammation or by destroying cells, Kleinfeld says. “Your preparation is only as good as your surgery. You have to become a very good surgeon. It’s an art form.”

Obtaining images

FOLLOWING BEHAVIOR: Miniature microscopes affixed to the heads of mice allow researchers to track how their brains respond to social contact.COURTESY OF INSCOPIXOnce you’ve perfected your technique, and your mouse is prepped and ready, it’s time to place the animal under a microscope. The instrument of choice is the two-photon microscope, which uses a laser to penetrate more than 1 mm into tissue and results in minimal photobleaching and phototoxicity. To accommodate a mouse, the microscope needs extra space between the lens and the stage. When researchers first began to use two-photon microscopy for in vivo mouse imaging, most built their own microscopes. Now, companies such as Prairie Technologies (now part of Bruker Corporation) and Zeiss sell ready-made versions, which are priced around $500,000 to $800,000, depending on the peripherals.

Following a particular neuron among the millions in the mouse brain across multiple imaging sessions isn’t as hard as it sounds, says Holtmaat. Researchers affix a small bar to the top of the mouse’s head, which clips onto the microscope, returning the mouse to the same orientation each time. The brain’s blood vessels, which are very stable, “act as a Michelin guide to the cells that you are interested in,” he says. Using the vessel pattern as a guide, one can locate the general region where the cell of interest resides and then return to the magnification and coordinates used before to pick it out of the crowd.

Be careful not to look too long, however. “It’s very tempting to keep looking at your neurons,” says Holtmaat. “It’s beautiful when you go in and see all these neurons with their synapses in their full glory.” But if you photobleach the fluorescent molecules—although the cell will express more to shine another day—they may become toxic to the neuron.

Instead of bringing the mouse to the microscope, new technologies allow researchers to bring the microscope to the mouse. Schnitzer and colleagues have developed a one-photon system that makes use of miniature cell-phone cameras and weighs only two grams (Nat Methods, 8:871-78, 2011). Inscopix sells the scopes for about $100,000 each. As studies of naturally interacting mice may require the simultaneous use of a number of mini microscopes, Inscopix discounts the price for buyers of multiple scopes. (See “Brainspotting,” The Scientist, December 2011 and “Top 10 Innovations of 2013,” December 2013.)

The one-photon system has a wide field of view and can image as many as a thousand neurons at the same time. It may also be combined with the guide tubes and optical probes to look at brain regions below the cortex.

STED SUPERRESOLUTION: The image shows a spine-studded stretch along the dendrite of a neuron located in the visual cortex of a live mouse. Stimulated emission depletion (STED) allows superior resolution of cup-shaped dendritic spines (white box), which receive signals transmitted across a synapse.COURTESY OF STEFAN HELLOther techniques peer deeper than ever before. Two-photon microscopes are limited by the ability to differentiate fluorescent molecules, and so can only resolve structures around to 200 to 300 nm in size. A new method, called stimulated emission depletion, or STED, microscopy “smashes that barrier,” says Stefan Hell, a professor at the Max Planck Institute for Biophysical Chemistry in Göttingen, Germany, who developed the system (Science, 335:551, 2012). (See “Next Generation: Dynamic Nanoscale GFP,” The Scientist, February 12, 2013.)

A STED microscope simultaneously illuminates a small region of tissue and quenches fluorescence in the surrounding area, preventing neighboring fluorescent molecules from bleeding together. As a result, a researcher can see details as small as 50 nm in size. If you want to see a dendritic spine—a mushroom-shaped body on the neuron that receives signals—and even discern how its shape changes in response to chemical or visual cues, STED microscopy is the way to go. Commercially available setups are costly, however, ranging from about $400,000 to $600,000. Researchers experienced with optics may be able to adapt their own system, says Hell.

Managing your models

With mini microscopes, researchers can observe neuronal activity as a mouse freely explores its environment. By contrast, traditional microscopes require a mouse’s head to be fixed in place during the imaging. But this doesn’t prevent researchers from engaging the mice in behavior and watching how their brains react.

David Tank of Princeton University has developed a rotating Styrofoam ball suspended above pressurized air that acts as a treadmill (Neuron, 56:43-57, 2007). The mouse’s head stays clipped to the microscope, immobilized, while stimuli from a virtual-reality system allows the animal to experience and react to multiple environments.

Tank has also developed a system in which rats freely explore a novel environment, occasionally clipping themselves at will into a two-photon microscope (Neuron, 80:371-84, 2013). The rats wear skull caps outfitted with attachments that look like wings, and are conditioned to insert the attachments into a slot to get a drink of water. This movement engages a device that positions the rat’s head precisely, to within one micron, each time.

“The rat [withdraws] its head, walks around, does its thing, comes back for a new trial, and every trial you can image the same neurons,” says Tank.

Such techniques allow researchers to image the brains of awake, behaving animals. Some types of experiments, however, such as photographing fine structural details, require mice to be immobile. One possibility is anesthetizing them. Alternatively, you may be able to train your mice to fall asleep under the microscope. Mice trained to expect a treat when they return to their cages eventually relax during the experiment. Imaging during the day (when mice normally sleep) also encourages them to take a nap. “Milk also works, just like for little kids,” adds Kleinfeld. “It helps put them to sleep.”