Mice can get information about their environment by detecting subtle changes in odors. (AP Photo/Patrick Semansky)
A new brain probe small enough to fit on a mouse without impeding its movements that can be left in place for weeks at a time could be a significant technological advance in neuroscience, providing reams of previously unavailable data on brain activity, according to the scientists who developed the device.
Introduced in a paper published Thursday in Science, the probe — known as Neuropixels 2.0 — is the miniaturized, next-generation version of a device that has been used in research settings for the past few years. The latest model can be left on the skulls of rodent subjects for weeks or months, allowing for long-term observations and, most critically, tracking groups of neurons with a high degree of accuracy.
"The big-deal scientific advance in this paper is the number of cells that you're confident you can follow over a long period of time," said Timothy D. Harris, a research professor at Howard Hughes Medical Institute in Maryland, and one of the study's authors. "There's just no precedent for that."
The device's development, some 10 years in the making, was a response to "really inadequate technology in a neurological research environment," Harris said. Most neurological researchers were working with twisted-wire arrays inserted in a rodent's brain or early micro-fabricated devices with 32 electrodes.
Electrodes are conductors that contact a nonmetallic part of a circuit. In this context, they are metal bumps along a slender piece of silicone known as the shank that is inserted into a rodent's brain, and act as electronic pickups for brain signals. In other words, electrodes are the microphones of neuroimaging. And a probe with 32 electrodes just isn't cutting it for Nicholas A. Steinmetz, an assistant professor in the Department of Biological Structure at the University of Washington in Seattle.
"We are looking at mouse brains that have 100 million neurons in them that have incredibly complex patterns of activity," said Steinmetz, who was a co-lead author of the paper. "We can't even observe that right now — that's just where we're at. Imagine your screen: If you have only a few pixels, you aren't going to be able to follow the plot of the movie, right? The more information we can get, the better our picture will be of what's happening in the brain."
Launching the project in 2012, Harris and his colleagues started developing a probe with 380 electrodes. Neuropixels 1.0, as the researchers now refer to the device, was presented in a 2017 paper published in Science. It works well on rat subjects, but has proven too bulky for mice, Harris said. They had to go smaller.
This meant their next challenge was developing a probe that could simultaneously measure two areas of a mouse's brain — about the size of a human's little fingernail — without limiting the animal's movement. They also wanted to increase the number of electrodes from 378 to 6,000. "You want the device to be so small that you can implant it and cement it onto the mouse's skull, and they can actually walk around while you're recording," Steinmetz said.
The latest device is about three times smaller than its predecessor. At about 1.1 grams, Neuropixels 2.0 isn't much of a burden for a 20-gram mouse. Though tiny, it's equipped with more targeted sensing capacity, allowing researchers to hone in on specific areas of the brain. "If your interest is in the cortex, you get four times as much cortex data as you did with the old probe," Harris said. "That doesn't sound like a lot to anybody who doesn't do this, but it is a lot."
It also has a feature a bit like the jitter-correction feature in a digital camera that accounts for the subjects' movements. "If you have a freely moving mouse, every time their body shifts, it's like their brain is sloshing around inside the skull, the spinal cord is tugging at it," Steinmetz said. "If you have a probe that is fixed relative to the skull, the neurons are going to move. The question is, 'How do you detect the same signals before and after the brain moved, and how do you know which neurons those signals came from?' They look different now that you're picking them up on different microphones."
The solution was borrowing a concept from image registration from computer vision processing — if you're seeing the same image but from different sensors due to a shift in movement, built-in computer software can estimate how big the shift was and correct the recording. "That gives you a stable-over-time version of the movie that you're seeing of brain activity," Steinmetz explained.
The free-motion aspect could potentially allow researchers to observe more ethological behavior, or the natural activity of rodents. But the device isn't designed with just rodents in mind. Mice are most often used in neuroscience experiments because they're small and easy to handle — and they're mammals with all the same major brain structures as humans.
"Broadly speaking, they're just such a close match for the neuro-human anatomy," Steinmetz said. "When we're understanding the patterns of neuron activity, the way different brain structures are communicating with each other — those computations happening in a mouse's brain — we really think, by and large, we're also seeing what happens in the human brain."
"We really, really don't know how brains work," Harris emphasized. "We don't know very basic things like, 'What changes in your brain when you learn?' We learn; mice learn, but we don't have a detailed appreciation of how the brain accomplishes the task of learning. We don't have that information now. We're just at the very beginning of understanding what it means when a big collection of neurons learns something."
It could be up to a year before the Neuropixels 2.0 probe is widely available to neuroscientists, Harris said. If the team meets its loftiest five-year goal, it will develop technology capable of tracking tens of thousands of neurons — still less than a tenth of one percent of a full sample of the neurons in a mouse's brain. "Now, at least we can follow the same neurons through time in a way that we've never been able," he added. "That's a big deal."
The study, "Neuropixels 2.0: A miniaturized high-density probe for stable, long-term brain recordings," published in Science on April 15, was authored by Nicholas A. Steinmetz, University College London and University of Washington; Cagatay Aydin and Martijn Broux, Neuroelectronics Research Flanders; Sebastian Haesler, Neuroelectronics Research Flanders and Vlaams Instituut voor Biotechnologie; Anna Lebedeva, Marius Bauza, Maxime Beau, Jai Bhagat, Dimitar Kostadinov, Kenneth D. Harris, John O'Keefe, Michael Häusser and Matteo Carandini, University College London; Michael Okun, University of Leicester and University College London; Rik J.J. van Daal, ATLAS Neuroengineering, Neuroelectronics Research Flanders and KU Leuven; Zhiwen Ye, University of Washington; Fabian Kloosterman, Neuroelectronics Research Flanders, Interuniversity Microelectronics Centre (IMEC), Vlaams Instituut voor Biotechnologie and KU Leuven; Marius Pachitariu, Claudia Böhm, Susu Chen, Jennifer Colonell, Bill Karsh, Junchol Park, Britton Sauerbrei, Joshua Dudman, Adam W. Hantman, Albert K. Lee, Karel Svoboda, and Timothy D. Harris, Howard Hughes Medical Institute; Richard J. Gardner, Abraham Z. Vollan and Edvard I. Moser, Norwegian University of Science and Technology; Carolina Mora-Lopez, John O'Callaghan, Jan Putzeys, Shiwei Wang, Marleen Welkenhuysen and Barundeb Dutta, IMEC; and Alfonso Renart, Champalimaud Centre for the Unknown.