Images from brain scans and new microscopy techniques are offering a strikingly clear glimpse of what’s going on underneath the bumpy surface of our skulls.
New technologies—and the innovative ways in which scientists have harnessed them— have driven advances in neural imaging beyond what any expert predicted 10 years ago. Ever more sophisticated images from brain scans and new microscopy techniques are offering a strikingly clear glimpse of what’s going on underneath the bumpy surface of our skulls.
Some of the greatest excitement in neural imaging right now surrounds the fast-emerging frontier of optical imaging, or, more precisely, two-photon excitation microscopy combined with fluorescent dyes that label individual molecules in living tissue. Scientists are applying these tools to track brain function in living animals in real time, right down to the level of synaptic connections and beyond.
A recent meeting on neural imaging at Cold Spring Harbor Laboratory in New York was testimony to the results possible using so-called “light microscopy” approaches. A few dozen of the world’s leading experts in imaging gathered to compare notes, debate technical hurdles, and share some of the most remarkable video and still images of mammalian brains in action.
Two-photon microscopes use longer wavelengths of light, supplied by lasers, to penetrate tissue more deeply and with less damage than other optical imaging modes. A critical advance that pushed the field forward was the identification and cloning of the gene for green fluorescent protein (GFP), reported in a landmark Science paper in 1994. GFP is a naturally occurring protein that essentially makes cell tissue light up like a neon sign when viewed with light microscopy.
Roger Tsien at the University of California, San Diego, and others have developed a whole spectrum of GFP variants, a broad pallet of colors that neuroscientists worldwide are now using to characterize neuronal structure and activity to a degree never before possible.
Among alternative methods, “there’s no competition for optical imaging,” says Karel Svoboda of Cold Spring Harbor, who co-chaired the meeting. “Using genetic tricks with GFP and its dozens of variants, you can now put into neurons fluorescent markers of structure, of specific molecules, or of cellular function. This has enabled a better understanding not only of the structural biology of the brain at the level of synaptic circuits, but also has begun to help us learn about the function of populations of neurons in the intact brain.”
“The genetics has gotten to the point where you can target cells pretty precisely with a fluorescent protein,” says David Kleinfeld, a University of California, San Diego, neurophysicist who attended the meeting. “You can go back to the same cell every time and determine its functional identity. Does the cell report the same sensory features time and time again, or does its role in a circuit evolve with experience? You can now study the same animals over time, which is particularly critical when you’re studying brain development.”
Grappling with Circuits
These new methods are also making a huge impact on systems neuroscience, which seeks to construct “wiring diagrams” that correlate brain activity to specific behaviors. Much of the Cold Spring Harbor meeting focused on the challenge of understanding neural circuits, Svoboda says.
“In the mammalian brain, you have a million upon millions of neurons,” he says. “If you think of it from an engineering standpoint, the brain is an electrical signaling device, and neurons are the signaling units. Any engineer will tell you that if you want to understand a circuit, you need to have a circuit diagram. You not only need a list of components, but you also need to know how neurons connect with one another and with what probability.” There has been good progress on the “parts list,” but understanding the connection matrix is still at a “primitive stage,” he says.
One reason: until now, the standard technique for constructing a diagram of a neural circuit had changed little since the late 1800s, when Spanish anatomist Santiago Ramon e Cajal pioneered it. The technique essentially involves staining single neurons, identifying where axons and dendrites overlap, and marking those junctures as synapses.
A problem with this approach, Svoboda says, is that “there’s no functional context. You don’t really know whether or not and to what extent these neurons ‘synapse’ onto one another.” While electrical recording studies can measure activity across synaptic connections, Kleinfeld says optical imaging now makes it possible “to observe how different sensory and motor patterns sculpt and resculpt the connectivity.”
Some of the most cutting-edge work with light microscopy involves finding ways to identify neuronal function in a manner that is rational, quantifiable, and reproducible. The ultimate goal is to use different types of GFP-based indicators of neuronal function in various types of neurons in order to understand how they interconnect and influence one another.
“What you’re really after is to record something that tells you about the state of the neuron: is it sending an output signal or not; what are its input signals like; what is its sense of history?” says Kleinfeld. Each of these states can be understood by looking at specific physiological indicators that can now be visualized with optical imaging.
Josh Sanes, a neurobiologist and head of Harvard’s new Center for Systems Neuroscience, describes his dream scenario: “to label 10, 20, 30 different neuronal types with different colors, and do it such a way that when the neuron fires it would change color.” This would make it possible to track neural activity throughout the circuit, with different cell types and functional characteristics clearly demarcated. Then, individual cells or even genes could be turned off or on to understand their roles in the circuit.
Such studies are just beginning, and many technical hurdles remain. Still, Svoboda says, “We’ve made remarkable progress.” For example, his group has pioneered in vivo imaging of neurons over long periods, even months at a time, something that was “just a pipe dream 10 years ago.”
Birth of Modern Imaging
Such advances were unimaginable back in the 1970s, when the advent of computerized tomography (CT) scanning marked the beginning of the modern era of neural imaging. “CT was a remarkable advance, because it was the first time you could look into the brain of a living person,” says Arthur Toga, who heads the Laboratory of Neuro Imaging at the University of California, Los Angeles.
Magnetic resonance imaging (MRI) and positron emission tomography (PET) followed CT. These powerful tools have enabled an unprecedented look not only at the brain’s anatomical structure, but also, in the case of PET and functional MRI (fMRI), at the patterns of brain activity that underlie mental functions and pathological states. Such “whole-brain” imaging modalities have transformed neuroscience research and are increasingly influencing the clinical practice of neurology, psychiatry, and neurosurgery.
PET and fMRI are the elder statesmen of neuroimaging. Not only have they become indispensable for basic research, but their capacity to show changes in oxygen and glucose metabolism indicative of neural activity has driven the burgeoning field of cognitive neuroscience, which seeks to understand higher-order brain functions and psychological states.
At the same time, new PET imaging agents—the radioisotopes that zero in on specific brain chemicals— have extended the uses of PET, making it possible to identify changes in dopamine receptors in Parkinson’s disease, for example.
In terms of clinical practice, neuroimaging has undoubtedly had the greatest impact on neurosurgery. Brain scans are routinely used presurgically and, increasingly, during surgery to identify critical brain structures that must be avoided in the operation and to guide the surgeon’s scalpel to a tumor or vascular occlusion. But imaging is also playing a greater role in neurology and psychiatry clinical practices.
One sign of this progression is the government’s recent announcement that Medicare will cover the cost of PET scans in certain people suspected of having Alzheimer’s disease, a recognition of PET’s utility in differentiating Alzheimer’s from other types of dementia. In October, the NIH launched a five-year, 50-site study designed to identify biological markers for Alzheimer’s through brain imaging, with the ultimate goal of improving early diagnosis and intervention. (See “New Techniques Detect Alzheimer’s Before Symptoms Develop,” this issue.)
Integration Is Key
Toga encapsulates what he finds most exciting about the current state of neuroimaging in one word: integration. “We have made tremendous progress in terms of the technological advances to acquire images that describe one part of the brain or another,” he says. Examples include diffusion tensor imaging, which processes MRI scans in a way that enables researchers to see the white tracts of neuronal axons, and new approaches to looking at vascular architecture and blood flow changes in the brain.
“What’s now occurring is the application of complex computational strategies that extract more information out of the images that are acquired, giving you a much more comprehensive view of what’s happening in a normal brain and what’s going wrong in pathological conditions,” Toga says. “So now we can take that data, ‘massage’ it, compare it against statistical and imaging databases, and apply a variety of visualization algorithms to look at it new ways.”
Such progress would not have been possible without the integration of multiple disciplines, Toga says. “You have mathematicians, computer scientists, and related disciplines now working on these problems of imaging the brain. That’s relatively new.”
Coupled with technological advances, this unprecedented collaboration has spawned novel approaches to neural imaging and allowed scientists to look at age-old questions about the human mind in a whole new way.
“It’s the great quest,” Toga says. “The brain is the only organ in the body that makes us who we are, so we can’t help but want to see if we can get a handle on that.”
Imaging and Behavior: Reality Check
The visual appeal of imaging studies, coupled with their relevance to things people care about, such as memory and emotion, can leave the work open to less-than-critical interpretation. Experts say findings on imaging and human behavior come with a few caveats. Marcus Raichle of the Washington University School of Medicine acknowledges that by the time discussions of what imaging can show enter the public arena, brain function has begun to sound overly localized. A good example is the so-called fusiform face area in the temporal lobe, considered by some researchers to be specialized for recognizing faces and by others to be part of a complex distributed network. “By the time people read about this part of the brain, what they come away with is that it’s the ‘face area,’” Raichle observes, but that may be an oversimplification. It’s also important to remember that imaging is correlational. Ed Smith, cognitive neuroscientist at Columbia University, explains: “By themselves, imaging studies don’t prove that a particular area is the source of a given mental process, only that it’s active at the same time.” To prove cause and effect, scientists are increasingly backing up results with studies of patients who have damage in the same area; a lesser degree of activation in these patients adds weight to the argument that the region is necessary to the behavior in question. Smith also notes that time plays a role in imaging. Functional magnetic resonance imaging, or fMRI, for example, can distinguish events that occur a minimum of two seconds apart. Most of the mental tasks investigated in imaging studies take far less time. For example, naming a picture involves matching the picture to an internal memory bank, retrieving its name, and pronouncing the word. The entire process takes about half a second. More comprehensive views will take shape as whole-brain imaging studies, such as fMRI, are joined to “faster” measurements and with cellular studies in animals, such as two-photon microscopy.