To identify a cell, researchers often have to abuse it—rip it from its home, douse it with toxic fixatives, doctor its DNA, or coerce it into making exotic proteins that could upset its biochemistry. Even if the cell survives, it may never be the same again. But a strong yet gentle beam of light could one day allow researchers to classify cells while leaving them unharmed and alive for additional study.
A team led by biophysicist Cynthia McMurray and physicist Michael Martin of Lawrence Berkeley National Laboratory (LBNL) in California has found that by scanning cells with an intense beam of infrared radiation produced by a synchrotron, a type of particle accelerator, they can capture a biochemical signature that reveals cells' identities.
The researchers presented early results from the method in June at a meeting in the United Kingdom, and they are now evaluating it with a 1-year pilot grant from the Chan Zuckerberg Initiative (CZI). If it works, the team's spectral phenotyping technique could provide a tool for another endeavor backed by CZI: the Human Cell Atlas, an international project that aims to chart the type and location of every cell in the body. And if the synchrotron-driven method can be adapted to more modest infrared instruments available to other labs and hospitals, spectral phenotyping might one day also help diagnose illnesses, probe the cellular changes that lead to disease, and delve into embryonic development. "The tools we are putting together will blow open this field," McMurray predicts.
Scientists who are familiar with the still unpublished results call the approach promising. "I'm looking forward to seeing the research that's going to come out," says spectroscopist Peter Gardner of The University of Manchester in the United Kingdom. Chemical physicist Hugh Byrne of the Dublin Institute of Technology is impressed by how thoroughly the group is testing its approach. "It's a concerted program to demonstrate the capabilities of the technique."
Martin and McMurray like to contrast their approach with a widely used cell-identification technique: fluorescent labeling. To spur cells of a specific type to produce a label such as green fluorescent protein (GFP), scientists have to equip them with the molecule's gene. The techniques for adding DNA can alter the cells, and because GFP is foreign to them—it's originally from a jellyfish—it could also modify their physiology. Moreover, McMurray notes, researchers typically have to zap fluorescent labels with a laser to induce them to light up, which can harm or kill cells. Other techniques are no less invasive. "If you are doing labeling or staining, you are changing the true chemistry" of cells, Martin says. "We want to explore what the chemistry is, not alter it to do the measurements."
That's where infrared spectroscopy comes in. "Infrared is not invasive, so it can be used on intact tissues and living cells," McMurray says. When a sample is exposed to different wavelengths of infrared radiation, how much light of each wavelength it absorbs indicates the kinds of chemical groups it contains. Unlike fluorescent labeling, the absorption pattern usually can't reveal whether a cell is producing a specific molecule—for example, the immune receptors CD4 or CD8, which are often used to define two classes of T cells. But a cell's infrared spectrum does reveal broad types of molecules—such as fats and proteins—providing a biochemical fingerprint. As a result, "You get a much more holistic picture of the cell," Byrne says.
Martin and McMurray say standard infrared sources don't provide the sensitivity they needed, so the team turned to LBNL's Advanced Light Source synchrotron, whose infrared beam is one of the brightest in the world. It "allows us to get better resolution and fidelity," Martin says. At the June SPEC2018 conference in Glasgow, U.K., McMurray and Martin revealed they could discriminate two types of brain cells—neurons and astrocytes—in slices of brain from mice. In brain tissue from rodents with a condition mimicking Huntington disease, they could also detect an increase in lipids that indicates degeneration. In the future, the researchers plan to automate cell identification by enlisting machine learning algorithms to pick out distinguishing features of each cell type.
McMurray and her colleagues still need to determine whether a cell's spectral signature remains constant or varies with its location in the body. For potential medical uses, they also want to find out whether a human cell's infrared signature changes when a person becomes ill. So far, however, the researchers have analyzed only mouse tissues. "We wanted to make sure the method is robust," McMurray says.
One limitation of the new technique is obvious—synchrotrons are huge, expensive, and rare, and often have monthslong waiting lists. "You aren't going to be taking your synchrotron into the hospital," Gardner says. But lab machines are rapidly approaching the infrared-generating power of particle accelerators, he notes. McMurray adds that after using the synchrotron to pinpoint distinctive spectral patterns for a variety of cell types, the researchers plan to publish a catalog that would allow other scientists to compare results from their own samples, even ones captured with less discerning lab devices.
Gardner expects the project to have an impact. "They have the tools, the expertise, and the personnel to accelerate this work," he says.