Seeing Shifting Cells

A fluffy black mouse twitches his nose. He sees a seed on the other side of a narrow window and he wants to eat it. He reaches his right paw toward the food. If he can just get his fingers through the gap, the seed will be his.

Inside the mouse’s brain, cells called neurons are hard at work. The neurons send tentacle-like feelers called dendrites throughout the brain to inspect what’s going on and report back to the cell body. As the mouse tries to grab the seed, the dendrites seek connections with other neurons that will help the mouse succeed. The dendrites form spines, mushroom-shaped protrusions that touch other neurons to get new information. As the mouse practices grabbing his seed, these spines change shape, new spines pop up, and unhelpful ones disappear—all while making a robust network of connections in his brain.

Neuroscientist Yi Zuo at UC Santa Cruz wants to watch these spines in action while mice learn new skills. By seeing changes in how neurons connect, Zuo will understand what happens when mice learn, perhaps giving insights into how the human brain learns. The problem is that it’s hard to see into the depths of a brain in a live mouse.

Now, a new microscopy technique promises to help biologists take images deep within thick tissues, such as brains or tumors. The technique borrows heavily from adaptive optics, which helps astronomers see clearly through Earth’s atmosphere. A similar approach could help biologists see through the blur of life, too. But progress has been slow. It’s been hard to troubleshoot problems because Zuo and her collaborators, the engineers who design adaptive optics, often don’t understand each other.

Collaborations among biologists and other scientists can help teams address more challenging questions. But as with Zuo and her project, communication is often a problem. “The physicists, engineers and biologists typically don’t even speak the same language,” says Nobel Laureate biochemist Thomas Cech, director of the BioFrontiers Institute at the University of Colorado Boulder.

The first step, Cech suggests, is finding projects that are both satisfying to the engineers and helpful for the biologists. Because adaptive optics for microscopy has a complicated optical setup, and because it promises to clarify the depths of biology, this project has the potential to make engineers and biologists drool—if only Zuo and her engineering counterparts can bridge their communication gap.

An illuminating tool

Neuroscientists use microscopes to see what’s going on inside cells. They want to spy on their favorite proteins and watch what happens when cells do something interesting, like talk to their neighbors. But with simple light microscopes, like those in a high school classroom, it’s hard to make out what’s what. Everything in the cell looks the same because there’s not enough visual contrast within the cellular soup. So scientists use tricks to make their proteins pop out.

Molecules called fluorescent dyes stick to a protein of interest and glow when biologists shine light on their sample. Then, “you have something highlighted on a dark background and you’re able to follow the specific molecules you’re interested in,” says biochemist Nathan Shaner at the Scintillon Institute in San Diego, a research consortium dedicated to advancing biological imaging. However, dyes can be tricky to get into cells.

In the late 1990s, biologists were all aflutter about green fluorescent protein, which originally came from a jellyfish called Aequorea victoria. GFP shelters a series of molecules within its cylindrical core. When scientists shine blue light on GFP, those core molecules release green light—and the whole protein glows. “GFP was a big deal because it was the first time you could genetically encode a fluorescent dye,” says Shaner. By inserting the GFP gene into a cell’s DNA, scientists could skip the dyes and engineer cells with fluorescently labeled proteins. Shaner and other protein engineers have now designed fluorescent proteins that span the colors of the rainbow, such as one Zuo uses to make some of her neurons glow yellow.

Fluorescence microscopes allow biologists to track dyes or proteins inside cells. To see GFP in a sample, for example, researchers shine blue light through the microscope’s objective lens. The light illuminates the GFP proteins, which glow green. The objective lens then sends the green light into a camera or eyepiece so researchers can locate their protein of interest.

Zuo wants to use fluorescence microscopy not just on isolated cells, but to see yellow neurons inside the brains of live mice as they learn new tasks. But when she looks deeper and deeper into mouse brains, the tissue surrounding her target neurons distorts their fluorescent signals and makes them harder to detect. Now, adaptive optics promises to clear things up.

Erasing the blur of life

Astronomers developed adaptive optics to sharpen their views of celestial objects like galaxies and planets through the swirling gases that make up our atmosphere. Light from distant bodies bends when it zips through the atmosphere, which makes the objects appear blurry through a telescope. Just as the rippling water in a stream makes it hard to see the bottom clearly, the shifting air overhead obscures the fine details of light from deep space.

To clear up the sky with adaptive optics, astronomers first must measure how the light gets distorted. They focus on a “guide star,” which is often a nearby star. If there is no star near the object, astronomers shine a yellow laser into the sky. The laser hits a band of sodium atoms in the atmosphere and makes a spot to focus on. The light from either a real star or a laser-generated star should look like a point. To sharpen the blurry image, an adaptive optics system bounces the starlight off of small mirrors that rapidly deform to erase the light’s distortions. When the star comes into crisp focus, it brings a clearer image of its celestial neighbor with it.

Adaptive optics took off in the early 2000s. By 2005, engineers such as UCSC’s Joel Kubby got to work designing deformable mirrors for new telescopes. Then Kubby saw other engineers using adaptive optics with fluorescence microscopes to image slices of tissue, like mouse intestines. Their idea was that biological tissues blur fluorescent signals from cells, just as the atmosphere blurs light from space.

Unlike in astronomy, there’s no “guide star” in biology—no glowing point-like object that researchers can use to directly measure how tissue blurs the sample. To circumvent this problem, many engineers estimate the biological light distortion. They skip the deformable mirrors and instead “fix” each ray of light until their image becomes clear, like finding the clearest sound on a radio dial. But Kubby knew how well the guide star method worked for astronomy. So he developed his own adaptive optics system.

Part of our initial working together was like a tower of Babel.

To make a microscope that used a telescope’s strategy, Kubby needed to figure out what to use as a guide star. He started by injecting fluorescent beads, small plastic spheres the size of a bacterium that contain fluorescent dyes, into tissues. Kubby tried to use deformable mirrors to correct the light from the beads until they appeared as perfect circles, bringing the neighboring tissue into focus. But he couldn’t accurately get the beads next to whatever he wanted to image. “It was pretty crude,” he admits. So he had to develop a more sophisticated idea.

Now Kubby uses fluorescent regions in the tissues themselves to make guide stars. When a biologist brings him a fluorescently labeled sample—it doesn’t matter what color—he and his team illuminate a tiny circular point in the sample so that it glows. Then they can correct it into a perfect circle with adaptive optics, clarifying the whole image of the target tissue.

After making progress on these guide stars, Kubby had another problem to tackle. In astronomy, light from faraway objects travels in one direction: from space to Earth. But in biology, light travels both into and out of a sample with fluorescence microscopy. Depending on the microscope, adaptive optics must correct light on the way to the sample, on the way out of the sample, or both. By 2013, Kubby had developed systems to work with a variety of microscopes, depending on what best suited the biologists he’d teamed up with. It was time to show off what adaptive optics could do.

“I approached biologists and asked if they’d be interested in improving the imaging of their samples,” he says. That’s how he met Zuo.

Clarifying communication

Zuo originally wanted to become a medical doctor, she says. Then she realized that neurological disorders are hard to treat because we don’t know how normal brains work. So she went to graduate school instead to study neuroscience.

Now she’s on a quest to understand how animals learn physical tasks by watching how their brains change as they perform. “I’m teaching mice to grab something,” she says, demonstrating this action by snatching an orange off her desk. She’s training her mice to grasp food through slits or off pedestals. Then she watches what happens deep in the brain as the mice learn and get faster at grabbing.

Zuo is investigating the spines that jut from the spidery dendrites of nerve cells in the brain’s motor cortex, which helps coordinate our movements. As the dendrites receive messages from other neurons, the protruding spines pop up, disappear, and change shape rapidly. As her mice practice snatching food, Zuo sees these neural structures shift in as quickly as 20 minutes. She hopes that by looking at spine dynamics while animals are learning, she will understand how neurons change their connections to improve the creatures’ reaching skills.

To watch this happen in live mice, Zuo has to perform surgery by either thinning the skulls of the mice or opening a hole. Even so, she can monitor only the dendrites that reach the top of the motor cortex. Her microscope can’t image the deeper layers; there’s too much tissue in the way in a live brain. Until Zuo can see more deeply, she doesn’t know how learning affects the neurons hiding below.

In 2011, Zuo jumped at the opportunity to work with Kubby to try to see spines changing in the depth of the brain. But as their partnership began, they found that it was hard to communicate. “Part of our initial working together was kind of like a tower of Babel,” Kubby recalls.

For example, Zuo and Kubby had different expectations. Kubby wanted to prove his technique worked, while Zuo wanted a tool to help her see deeper into her mouse brains. Zuo recalls asking Kubby what color fluorescent protein she should use for the guide star. She says he told her, “Anything is possible,” because he knew he could make his system work with any color. But Zuo wanted specific instructions.

“I need to design a protein that will get into the animal’s brain,” she says. “For them it’s a trial, [but] if it fails, then months of my work is gone.”

One way to solve this communication clash is to be “a biologist-wannabe,” which is how engineer Na Ji describes herself. “I don’t get up every morning saying, ‘I want to develop adaptive optical techniques,’” says Ji, who is moving from the Howard Hughes Medical Institute Janelia Research Campus in Ashburn, Virginia, to UC Berkeley. “Every morning, I think, ‘I want to understand the brain. So what can I do to make that happen?’” Ji built her own adaptive optics system at Janelia. Unlike Kubby’s system, Ji’s measures distortions indirectly, correcting each ray of light as it exits the sample until she gets a crisp image.

In 2015, Ji’s lab published a paper demonstrating that adaptive optics could uncover new information about how we perceive edges. Her group looked deep into the brains of awake mice to see how signals from the eye are sent through a relay center into a brain’s visual cortex.

Previously, researchers thought the neurons in our visual cortex processed images from our eyes to determine where things begin and end. But Ji’s team wondered if neurons in the relay center could start to process images before they reached the visual cortex. The group focused on specific regions of these relay center neurons, called boutons, which send information to the neurons in the visual cortex. Ji’s lab imaged 28,000 individual boutons deep in the brain. Without adaptive optics, they would have been a blur. The team showed that some signals leaving the boutons are already processed, contradicting the theory that the visual cortex alone determines edges.

Kubby views Ji’s paper as a success for adaptive optics. “We’re not really competitors,” he says. “We’re both trying to advance the field. She just has a different approach.” Both Kubby and Ji hope people will see the results and want to try adaptive optics for themselves.

Less of a boutique technique

Ten years after the first adaptive optics papers for microscopy were published, we are just starting to see fascinating results. That’s common for interdisciplinary work, says UCSC astrophysicist Claire Max, who was instrumental in creating artificial guide stars for astronomy.

Max is a member of an elite team of scientists, called JASON, which advises the U.S. government on science and technology. Her first task with this group was to make adaptive optics technology to help the Air Force scan for spy satellites. She shined a yellow laser into the sky that mimicked a real star and brought satellites above the atmosphere into focus.

After the project ended, Max spent almost 10 years redeveloping this technology for astronomy at Lick Observatory in San Jose, California. Then she needed to make it useful broadly, not just for what she calls “boutique topics.”

“It seemed like adaptive optics in 1999 was still used by only a few people, and they never talked to each other,” she says. Astronomers knew adaptive optics might help their research, but they weren’t sure how to use it.

So Max and the late astronomer Jerry Nelson started the Center for Adaptive Optics at UCSC using money from a National Science Foundation grant. The center held workshops where astronomers, physicists, and engineers worked together to show that adaptive optics was useful in solving astronomy problems. “Otherwise,” Max says, “who’s going to bother?”

For now, Kubby says, adaptive optics in microscopy is still an experiment, suggesting that the technique is still in its “boutique” phase. But with Ji’s ability to think like both a biologist and an engineer, and with Kubby and Zuo learning how to bridge their communication gap, adaptive optics may be well on its way to becoming a product.

Back in Zuo’s lab, the mouse has gotten quite good at grabbing his seed through the window. Zuo hopes the information she learns from her studies will tell her something about how humans learn, too. “Maybe you play basketball,” she says. “Sure you improve, but what is changing in the brain? We don’t know.” Adaptive optics may soon be able to clear that up.

© 2017 Sarah McQuate / UC Santa Cruz Science Communication Program