Scientists have long sought to understand the biological basis of thought. In the second century A.D., physician and philosopher Claudius Galen held that the brain was a gland that secreted fluids to the body via the nerves—a view that went unchallenged for centuries. In the late 1800s clinical researchers tied specific brain areas to dedicated functions by correlating anatomical abnormalities in the brain after death with behavioral or cognitive impairments. French surgeon Pierre Paul Broca, for example, found that a region on the brain’s left side controls speech. In the first half of the 20th century, neurosurgeon Wilder Penfield mapped the brain’s functions by electrically stimulating different places in conscious patients during neurosurgery, triggering vivid memories, localized body sensations, or movement of an arm or toe.
In recent years new noninvasive ways of viewing the human brain in action have helped neuroscientists trace the anatomy of thought and behavior. Using functional magnetic resonance imaging, for instance, researchers can see which areas of the brain “light up” when people perform simple movements such as lifting a finger or more complex mental leaps such as recognizing someone or making a moral judgment. These images reveal not only how the brain is divided functionally but also how the different areas work together while people go about their daily activities. Some investigators are using the technology in an attempt to detect lies and even to predict what kinds of items people will buy; others are seeking to understand the brain alterations that occur in disorders such as depression, schizophrenia, autism and dementia.
But such studies reveal only the relative activity levels in different regions of the brain and the ways they change under certain circumstances. They do not disclose the biological underpinnings of these alterations in brain activity. To truly understand how autism unfolds, say, or how best to treat depression, scientists want to know what is happening inside the cells that control brain activity.
Genes, of course, provide the instructions for the molecular machinery inside a cell. Thus, biologists have long been engaged in a complementary effort to connect certain genes to defined disorders to get a molecular handle on those diseases. Indeed, researchers have now connected more than 500 genes to Parkinson’s disease, more than 600 to multiple sclerosis and more than 900 to schizophrenia. The ever expanding list of gene candidates is both a blessing and a curse. Although somewhere in this genetic stockpile lies the key to understanding these disorders, as the gene lists grow scientists have to laboriously sift through an increasing number of candidates and their interactions.
Now our team at the Allen Institute for Brain Science has developed a high-tech bridge between brain anatomy and genetics: an online interactive atlas of the human brain showing the activity of the more than 20,000 human genes. Preceded by a similar atlas of the mouse brain, the Allen Human Brain Atlas, which will expand in coming years, was launched in May with an initial body of data that already provides the most detailed view of gene activity in the human brain ever created. Now, for example, scientists can quickly determine where in the brain genes that encode specific proteins are active, including proteins that are likely to be affected by a new drug. Such information may help predict a new medication’s therapeutic effects, as well as its side effects. With the same ease, a researcher can zoom in on a particular brain structure—say, a region that brain scans have shown to be altered in schizophrenia—and find out which genes are at work there, in an attempt to discover the molecular footprint of the disorder. They also can gain molecular clues to ordinary brain functions such as memory, attention, motor coordination, hunger, and perhaps emotions such as happiness or anxiety.
The idea to create a brain atlas grew out of a series of think tank–like meetings convened by philanthropist Paul G. Allen, co-founder of Microsoft, beginning in 2001. Allen, who had been drawn to the mystery of how the brain works, had brought together some of the world’s top scientists in biology, genomics and neuroscience to consider the question: What can be done to propel neuroscience to the next level?
During these discussions, one project began to garner the most converts: a three-dimensional map of gene activity throughout the brain for all known genes. Such a map, if publicly accessible online, would enable scientists interested in the role of a particular gene or set of genes in, for example, depression, to bypass the tedious and expensive laboratory work needed to examine possible molecular culprits of the disease one at a time. Instead researchers could search the atlas to see where in the brain the genes are active as well as what other genes, active in the same regions, may be involved. In this way, researchers can identify the best candidate genes quickly and cheaply in silico.
The idea appealed to Allen because it was a big science project, along the lines of the Human Genome Project, that went beyond the capabilities of most labs and could greatly accelerate scientific discovery. So, in 2003, with a generous seed gift of $100 million, he launched the Allen Institute for Brain Science in Seattle.
To lay the groundwork for such an immense project, we first decided to create a map of the mouse brain. The mouse brain is significantly smaller and less complex than the human brain, so it would be a good inaugural project. In addition, such a map would be useful: many researchers test their theories about human behavior and disease in the mouse, because its brain is so similar to ours. The mouse and human brains share, for example, much of the same basic organization and physiology. What is more, 90 percent of the genes in the mouse have a counterpart in our genetic blueprint.
Our first challenge was to figure out how to efficiently map the approximately 20,000 genes in the mouse genome (the mouse has a similar number of genes as a human, suggesting that the complexity of the brain has more to do with its size than its genomic ingredients). At the time, a scientist working in a traditional research laboratory needed about five years to map the activity of just 10 genes throughout the mouse brain. But we saw that the scientific and technological landscape was rapidly changing. First, the results of the human genome effort and the imminent deciphering of the mouse genome would provide us with the molecular codes for the genes we would be mapping. Second, advances in automation had created high-throughput lab machines capable of working around the clock that could complete in hours tasks that would otherwise require weeks of human labor. We believed we could adapt this technology to perform the procedures our project required.
Making Genes Appear
What does it mean to make genes visible in the brain? First, some background: gene activity, also called gene expression, occurs when a gene gets “read”—a complicated process involving, among other actors, a molecular transcript called messenger RNA (mRNA) and ending in the assembly of a protein. Although all of a person’s genes are present in every cell, they become active only in certain tissues or at certain times, and it is then that their RNA transcripts and proteins can be “seen.” Proteins are critical building blocks and workhorses of every cell in our bodies. In the brain, proteins help with such tasks as building the connections in neural circuits, driving chemical signaling and doing the cellular housekeeping necessary for brain health. Alterations in a gene, known as mutations, create malformed proteins, which in turn can lead to diseases such as Huntington’s. Additionally, changes in the regulation of gene expression can lead to too many, too few or misplaced proteins, thereby interfering with normal physiology. Such changes have been implicated in neurodegenerative and neurodevelopmental disorders, for example.
To see gene expression in the mouse brain, we cut frozen brain tissue into slices thinner than a human hair and bathed each slice in a solution of molecular probes that bind specifically to the mRNA from a single gene. Next, we started a chemical reaction that stained the probes purple, marking their locations within the slice and thus indicating which cells expressed that gene. We then used robotic microscopes to photograph one million of these slices—enough to survey all 20,000 genes, one per slice—and shuttled the resulting image data into a computer database. We transformed that information into a 3-D digital reconstruction of the brain with its gene expression patterns and made it available online. The entire process took just three years.
The completed atlas revealed that at least 80 percent of mouse genes are expressed in the animal’s brain. That percentage is much higher than previous studies had indicated, perhaps because our method detected mRNA in nooks and crannies that other techniques can miss. That so many genes are active is testament to the brain’s complexity. More practically, this finding suggests that many drugs designed to affect proteins in other tissues, such as the liver or kidney, for example, may alter brain function as well.
The vast majority of the genes are expressed in very specific brain regions, representing the specialized function of these areas. These gene expression patterns create identifiable molecular signatures that, for instance, distinguish the cells in the striatum, a deep brain structure involved in basic control of movement, from cells in the cortex, which is involved in higher-level information analysis. Within the cortex, the genes active in the somatosensory area, which processes information about touch, differ from those expressed in the visual cortex.
In general, the structures revealed by the gene expression patterns reflect those that have been worked out by classical neuroanatomists who have been looking at brain sections with their microscopes for more than 100 years. In some cases, however, our technique for visualizing gene expression revealed finer subdivisions within structures than had been seen before. For example, we saw previously undiscovered compartments within the hippocampus, a structure deep inside the brain that plays a key role in memory and learning. We do not yet know what the cells in these compartments do, but the identification of these new subdivisions may help us better understand how the hippocampus works and thus, perhaps, identify where and how best to intervene to combat memory impairment in disorders such as Alzheimer’s disease.
Working in Concert
One of the most powerful features of the Allen Mouse Brain Atlas, as well as the atlases created later, is the ability to look at the expression patterns of many genes across the brain one at a time, in groups and in different combinations. Previously scientists often could study only one or a few genes at once because of the lab work involved. As a result, many current conceptions of the brain circuits governing complex behaviors may tell just part of the story.
Now, however, scientists have learned that brain wiring and biochemical pathways are sometimes more complex than originally thought. For example, neuroscientists are interested in the circuits that regulate eating and drinking, which hold the key to solving problems such as obesity and anorexia. These circuits, which must integrate internal signals such as hunger and thirst with environmental cues, also provide clues to the function of analogous brain networks.
In the past, scientists explained food and drink consumption by focusing on single gene products, such as the hunger-stimulating hormone ghrelin, or single brain centers implicated in hunger, satiety or thirst. But in a study published in 2008 obesity specialist Pawel K. Olszewski and his colleagues at the University of Minnesota revealed a more complex reality after they used the Allen Mouse Brain Atlas to assess the expression patterns of 42 genes in eight brain structures that had been implicated in the regulation of eating behavior. The researchers found that supposed hunger centers actually contain a mixture of genes, some of which increase appetite and others that diminish it. The results indicate that assigning individual brain regions single functions could be a mistake. They also may help explain the failure of antiobesity drugs that target single proteins, suggesting instead that successful treatments are likely to act on multiple molecules.
The atlas has already yielded insights into the genetic roots of cognitive differences among individuals. In a study published in 2006 neurogenomicist Andreas Papassotiropoulos and human geneticist Dietrich Stephan of the Translational Genomics Research Institute in Phoenix and their colleagues identified a human gene called KIBRA with variants that correlate with differences in a person’s ability to perform memory tasks, such as trying to remember a list of words after five minutes and then again after 24 hours. The variants were also associated with differences in brain activity in the hippocampus while subjects performed these tasks. By looking up the gene in the mouse brain atlas, the researchers found that the gene is expressed in the hippocampus—sewing up the case that the gene plays a direct role in short-term memory.
From Mouse to Man
Given the stream of findings emerging from the mouse brain map, we hoped that a similar atlas of the human brain would yield even more fruitful insights into diseases and behaviors that may differ between humans and mice. Such discoveries may enable better predictions about, say, which new medicines tested in animals will really work in people. To build this bigger map, however, we needed a different approach. Given the size of the human brain, analyzing gene expression in brain sections one gene at a time would take decades. Our streamlined method involved the use of specialized gene chips, also called DNA microarrays, to measure the activity of all genes simultaneously in each of about 1,000 distinct brain areas. These regions would be represented by tissue samples ranging from the size of a pea (for larger, more uniform brain areas) to that of a pinhead (for smaller, more intricate structures).
Developed in the mid-1990s, a DNA microarray is dotted with numerous microscopic DNA segments, called probes, each of which binds the mRNA for a specifically matched gene and “lights up” to reveal both the presence and level of that gene’s expression. Some gene chips contain tens of thousands of probes, enough to test for the presence of all human genes in a single experiment. Although it cannot provide the same fine, cellular-level detail as the mouse brain atlas, the microarray strategy is fast and yields numerical data—as opposed to images of slices of mouse brain—that are much easier to analyze, enabling scientists to draw correlations between different patterns of gene activity that might elude the human eye.
In March 2009, after nearly two years of planning, we were ready to begin making our human brain map. One hurdle remained: we needed a brain. The brain had to be free of disease or other abnormalities. It had to be whole and fresh and obtained and quickly frozen within 24 hours of death—or the mRNA we were looking for would degrade and we could not detect gene expression. Such brains are rare, and when they are available other organs must be collected first for people in need of an organ transplant. Only if our 24-hour window had not closed by then and surviving family members gave their consent could we have the brain.
Nevertheless, we received the first of several brains needed for the atlas in July 2009, kicking off the 10-month process required to complete data generation for one brain. We scanned the brain using MRI, creating a three-dimensional digital image onto which we mapped all the microarray data, along with that from tissue sections stained to reveal the brain’s cellular architecture. This spring’s debut of the atlas contained an almost complete data set for that first brain, including nearly 50 million gene expression measurements.
Neuroscientists hope that the Allen Human Brain Atlas can help them explain, on a deeper level, some of the more tantalizing results from human brain-imaging experiments. For example, fMRI results suggest that the fusiform face area of the brain’s temporal lobe, which is involved in the recognition of faces, tends to be underactive in children with autism. Other research suggests that, in people with certain genes, areas of the brain affected by Alzheimer’s are hyperactive when they perform memory tasks, a finding that may help predict their risk of acquiring the disease. And individuals with schizophrenia exhibit hyperactivity in the hippocampus and dorsal lateral prefrontal cortex, which may reflect a loss of inhibitory neuron function that contributes to their symptoms.
Deciphering the biology underlying these changes is essential to understanding these disorders. What changes in the brain cells of the autistic child cause these facial perception areas to be hypoactive? How do the genes that confer an increased risk of Alzheimer’s affect the function of the memory centers of the brain? What is going on—on the molecular level—within the neurons of the hippocampus and prefrontal cortex of an individual with schizophrenia? Now scientists working on these and other topics can begin to match the brain areas they are identifying with gene expression data in the atlas. From these data, biologists can start to construct the molecular processes underlying the activity revealed by fMRI and other imaging techniques.
In the coming years the atlas will be expanded and enhanced. More data will be generated from additional brains, enabling analyses across individuals that may reveal which features of brain anatomy and chemistry are shared and where individual variation may occur. In addition, we will incorporate more sophisticated data-search and visualization tools to help investigators more rapidly sift through the immense store of information and home in on those findings that are most relevant to their research programs.
In addition, future upgrades for the atlas will include more gene expression data for key brain structures, such as the hippocampus and hypothalamus, providing the degree of cellular detail available in our original mouse brain atlas and thus further insight into the cellular underpinnings of brain function. The entire atlas is scheduled to be completed by 2013.
Adding to this genetic treasure trove are several other Allen Institute resources. For example, the Allen Developing Mouse Brain Atlas maps changes in gene activity throughout the rodent brain as it grows from an embryo into an adult. It reveals clues to how brain structures form and forge ties during development and how those processes might go awry in developmental disorders such as autism, dyslexia and schizophrenia. The Allen Institute is also spawning smaller-scale maps, including an analysis of the genes involved in human glioblastoma, a devastating form of brain cancer [see “New Hope for Battling Brain Cancer,” by Gregory Foltz; Scientific American Mind, March/April 2010]. More than 20,000 visitors explore the atlases and other data every month.
With such resources, along with a growing number of gene databases being compiled by others around the world, we may soon be able to answer some very basic questions about human brain function in health and disease. Perhaps someday these tools may give us a handle on more fundamental and long-standing curiosities such as: How do we think and feel? What is consciousness? And what makes us human?