For more than a century, science hasn’t been able to explain even the most obvious feature that distinguishes the human brain from those of other primates—it’s conspicuously large size. And while we know that the human cerebral cortex, the seat of higher cognitive function, underwent a rapid expansion over the past two million years, we still don’t understand how it happened.
So how did our species evolve to become what it is today? What makes the human brain so unique?
Neuroscientists appear to be on the verge of answering these questions, and the key is genetics. We now have the technical means to pinpoint the genetic changes that enabled the expansion of the cerebral cortex, and the human traits such as abstract thought and complex language that it supports. We also have better tools to manipulate genomes in cells and animals to test how these genetic changes affect the way the brain is built and, ultimately, the way it functions.
Some are calling this the start of a “golden age” of human evolutionary genomics. Still, the task of understanding what’s uniquely human about the human brain remains enormously challenging, not least because we don’t have a good surrogate to study. Three leading researchers on the genetics of the human brain spoke with The Kavli Foundation, and shared their thoughts on the progress and pitfalls of studying its evolution.
- JAMES SIKELA, Ph.D. - is Professor of Biochemistry and Molecular Genetics and Human Medical Genetics and Neuroscience Programs at the University of Colorado School of Medicine in Aurora, CO.
- JAMES NOONAN, Ph.D. - is Associate Professor of Genetics and a member of the Kavli Institute for Neuroscience at Yale University in New Haven, CT.
- DANIEL GESCHWIND, M.D., Ph.D. - is Gordon and Virginia McDonald Distinguished Professor of Human Genetics, a professor of neurology and psychiatry at the David Geffen School of Medicine at the University of California, Los Angeles (UCLA), director of the Center for Autism Research and Treatment and co-director of the Center for Neurobehavioral Genetics, also at UCLA.
The following is an edited transcript of a roundtable discussion, which took place via teleconference on May 15, 2014. The participants have been provided the opportunity to amend or edit their remarks.
THE KAVLI FOUNDATION: Since Charles Darwin, scientists have been comparing the human brain to those of other species to understand how it differs—and how those differences are translated into uniquely human traits. Now, geneticists are also tackling this problem. Why is this such a good time for you to be involved?
JAMES SIKELA: Compared to 10 or 20 years ago, we have so much genomic data now, especially DNA sequence information, that one can compare at an unprecedented level of detail the genomes of humans with several other primate species to see what's really different. We call this human evolutionary genomics. That's a huge new element to the whole story that allows one to find specific genetic changes that are unique to humans, and go from there to explore the impact of those changes on the brain, behavior and cognition.
JAMES NOONAN: The biggest limitation in understanding evolution, particularly human evolution, is that we have no idea what the genetic drivers are. Until recently, we didn’t know where the important changes in the human genome were, and we didn't really have any guideposts to find them. So, in addition to having so much genomic data at our disposal, what's really changed in the last few years has been the ability to put filters on the genome to identify the changes that are most likely to be important.
DANIEL GESCHWIND: I think what both Jims articulate is really the power of genetics and genomics to take a totally unbiased approach to discover what's driving the evolution of the human brain. What I mean by ‘unbiased’ is that we’re looking at the whole genome without any preconceived ideas about what genes are important for a given trait, say, cognition or language, and what ones aren’t. As a result, we can identify genes we’d never dream would be involved in some aspects of brain function, as well as genes with no known function.
One of the fundamental issues with the brain is that in contrast to, say, the heart, we don’t really know how it does what it does. Without a model of how the brain works, it's very hard to put some of these genomic changes into context. Is this one important? Is that one important? How might it change the brain? That's where genomics comes in, because we can ask in a very unbiased way what has really changed as humans have evolved, and then try to connect those changes to specific traits.
NOONAN: There are also new genome engineering technologies that allow you to take a cell of an animal and modify its genetic code. It is then possible to look at a lot of potential drivers of human-specific traits, or phenotypes, in experimental systems. Those could be cell-based systems, in which you can take a non-human primate cell and try to ‘humanize’ it genetically, or a mouse that carries human genes.
TKF: Dan, you wrote in a commentary that comparative anatomy—the traditional approach to understand human brain evolution—and comparative genomics rarely meet in the middle. Why is that, and is it changing?
GESCHWIND: The difficulty is that the human is not an experimental organism, and in many ways, our closest ancestors like the chimpanzee aren't either. So the biological impact of the changes we’re finding in the human genome has to be extrapolated from experiments in model systems, which poses many challenges. Most evolutionary genomics depends on statistical analyses far removed from historical events. That said, it's definitely changing. We’re in a realm where we can actually test hypotheses based on comparative genomics in experimental systems, and that's a new frontier. So the study of brain evolution is becoming a bench science that assesses changing mechanisms, instead of a more computational science.
NOONAN: We need to have multiple convergent methods, technologies, and ways of thinking about the question of how the human brain evolved. Historically, the people that did comparative anatomy of the brain across species, and the people that do genetics and genomics, are not the same people. There are people like me who do genomic analysis to try to identify individual changes in genomes that might affect some molecular function that we can test. But we're not experts in brain development. So we need to talk to people who are. And that’s happening more and more. For example, I’m collaborating with Pasko Rakic, director of the Kavli Institute at Yale, and one of the world’s leading experts on the development of the human cortex.
TKF: Jim Noonan, you mentioned that we can now "put filters on the genome" to figure out which genetic changes are evolutionarily important and which ones aren’t. Can you give me an example?
NOONAN: Sure. In our laboratory, we study changes in gene regulation, not changes in genes that code for proteins. There are two filters we can use to identify regulatory sequences in the genome. One is to use conservation across genomes. So if a sequence is important for controlling gene expression, it's probably going to have been maintained by evolution for a while, in multiple species, and you can see that when you compare genomes to each other. The second is to do an experiment to look for certain biochemical signatures—an added layer of information superimposed on the genome—that are predictive of regulatory function.
These filters narrow the search base down considerably, and my lab has been able to use them to find some regulatory sequences that are only active in humans. We’re now using experimental tools to investigate the biological role of these regulatory elements and whether they affect human-specific traits such as brain size and complexity.
SIKELA: My lab’s approach to studying human and primate evolution has been to look for extreme genomic changes such as highly duplicated and deleted sequences of the genome, in other words, extreme copy number variants. That’s been our filter.
We found about 134 human genes that were elevated in copy number compared to the other primates we looked at. One, called DUF1220, which is actually a protein domain not a whole gene, had many, many more copies in humans: 270 copies compared to 120 in chimp and gorilla, 30 in monkey and fewer than 10 in some mammals. That’s a dramatic change to a protein-coding region of the genome, and there may be an evolutionary payoff that went along with that, related to brain size.
On the flip side, that payoff seems to have come at a major expense to humans. The chromosomal segment containing DUF1220 has been linked to 12 different diseases, including schizophrenia, autism, microcephaly and macrocephaly, and heart disease.
TKF: Duplications are just one kind of variation in the genome that provides the raw material for evolution. Does one type seem to be more important than others in the brain?
SIKELA: I think all of them are on the table as being important. One of the most important things to emerge from looking at human and primate genomes is that the genomic changes between species are more numerous and diverse than we ever expected. So we now know that the degree of difference between our genome and a chimpanzee’s really depends on where we look and how hard we look. So there’s lots of fertile ground in the human genome from which enhanced cognition and other traits could have emerged.
GESCHWIND: I would add that many of us are taking the stance right now that most of the findings are going to be in regions that don't actually code for proteins but have to do with the regulation of genes: when and where genes are turned on and how much they're turned on.
TKF: You’ve anticipated my next question, for Jim Noonan. Most people think of genes, or the protein-coding parts of the genome, as this currency of inheritance. Why is there growing interest in the role of non-coding regions—what used to be referred to as ‘junk DNA’—in evolution, especially the on/off switches that control gene expression?
NOONAN: Think about the evolution of the brain. There have been a lot of changes in brain development and structure in humans, but a lot of the genes that affect brain development are very well conserved across species. So what you effectively have is a modification on an ancient developmental program. There's not a lot of room for evolution to really tinker with those sorts of genes. If you start disrupting their functions, lots of things go wrong.
Regulatory sequences—those on/off switches you referred to—provide some space for evolution to work. They tend to be tissue-specific, meaning they act in a restricted place during development, so they can tolerate modification to some extent. And that lends plasticity to the system, some room for variations to emerge.
TKF: Dan, instead of looking at individual genes, your laboratory and others are studying how networks of genes work together in specific brain regions and at specific times during development. What are the advantages of studying these networks?
GESCHWIND: Well, we started our network analysis with a very simple question: Could it be that genes do not simply go up and down, but that their relationship with each other actually changes on the human lineage compared with other primates?
Let's say you find that a gene is expressed more highly in humans than in chimpanzees, showing that the gene is ‘upregulated’ on the human lineage. If you don’t have any other data, you don't actually know whether that change is meaningful—whether it’s part of the adaptive evolution of the human brain. But if you take a network approach, it puts genes in a true functional context.
Some genes are very central in networks, just as an airline hub such as Chicago is for United Airlines. You shut down Chicago, the whole international United network starts to experience delays. So shutting down the hubs has a big effect on the whole system. In an analogous way, if we see that a gene's position changes within a network, let's say from being a hub in one species to being peripheral in another, that is evidence that that gene has actually changed function. Without that kind of network context, there's no way to assign an unbiased functional role to changes in the genome.
TKF: There’s great excitement about ancient DNA, both among the scientific community and the general public. We saw the reference genome for Neanderthal published in January, and the genome of another extinct hominin, a “Denisovan,” published late last fall. What can we learn about the brain’s evolution from these human ancestors?
NOONAN: One of the most surprising things that came out of these studies is what a small number of actual differences there are between a Neanderthal genome or Denisovan genome and a modern human’s. The number of protein-coding changes is something like 84 in the entire genome. They're basically human. So I'm not sure exactly what they're going to tell us about brain evolution because, overall, I suspect the Neanderthal brain was pretty similar to ours.
SIKELA: That’s true but there are regions of the Neanderthal genome that are dramatically different. When the Neanderthal draft sequence came out, some researchers did a copy number comparison and, for the DUF1220 protein domain that my lab is interested in, there were something like 50 extra copies in the Neanderthal genome compared to human. We’ve got to take that with a grain of salt because only one Neanderthal individual had been sequenced. But it was striking, especially because we’ve known for decades that the Neanderthal brain case is on average much larger than our own, and DUF1220 has been linked to brain size.
In general, with the archaic genomes, there are still limitations in terms of genome quality, though it’s improving. So there may be pockets of the genome that are under-examined but are evolutionarily quite dynamic and important. The other limitation is that it’s difficult to study brain function in archaic species because we don’t know much about what they were capable of. We still don’t know why the Neanderthals became extinct. There are many different theories about what could have happened. While some relate to cognition, others are unrelated to cognitive ability, such as some type of immune scenario or climate change.
TKF: We’ve talked about how researchers like yourselves can now mine the genome for changes that may be important in evolution like never before. Now that we’ve got an ever-growing list of changes, what’s next? Where’s the field going?
GESCHWIND: We’ve got to move it to the functional level where we say, okay, this genetic mutation seems to be under positive selection, meaning it provides some sort of evolutionary advantage to the organism and has spread through the population. Now, let’s assess which organ it’s related to, and if it’s the brain whether it actually has an effect on brain function. That’s a very challenging problem.
SIKELA: I totally agree with Dan. Probably the biggest challenge to the field is to assign some function or non-function to candidate genes or genomic changes. It’s daunting because it's very hard to model the human brain in species that can be studied by researchers, such as mice. So a big part of where the field is right now is trying to find the appropriate functional assays with which we can tease out the potential importance of human-specific genomic changes.
TKF: Are we getting closer?
GESCHWIND: Yes, but there still aren’t many examples because it’s extremely difficult to do.
There have been a couple studies on a protein that plays a role at the synapse, the junction between neurons, by Evan Eichler at the University of Washington and Frank Polleux at Columbia University’s Kavli Institute for Brain Science. They found that over the course of evolution there have been gene duplication events that changed a gene called SRGAP2. The altered form of the gene is found in modern humans and Neanderthal, as well as Denisovans, another human ancestor, but not in the great apes, suggesting that it has a new role in human brain function. The researchers were able to show, in vitro and in vivo, that the altered version of the gene leads to a higher density of dendritic spines, the long, thin branches on neurons that receive electrical signals from other cells, that actually are more human-like than a macaque’s or a chimpanzee’s. That’s really remarkable.
Similarly innovative work on the evolution of the NMDA receptor, which binds the brain signaling chemical glutamate, in mammals has been published by Seth Grant at the University of Edinburgh and colleagues. And Jim Noonan’s work with enhancers, the things that turn genes on and off in certain brain regions at certain developmental times, is the same kind of work. But again, most of this work has to be done in mice, which lack the complex cognitive and social abilities of primates. Maybe down the road we'll be doing some of these experiments with marmosets.
TKF: Why marmosets?
GESCHWIND: There are a number of labs that are trying to make transgenic marmosets because they are one of the smaller primates, and therefore easier to handle, yet they have very high-level social interactions.
People are also developing in vitro systems that model aspects of human cortical development in a dish based on various forms of stem cells and organoid cultures.
TKF: Dan, you study autism, so you must face the issue of how to study a uniquely human disorder in mice and other experimental organisms all the time. Some have hypothesized that autism and schizophrenia are a consequence of human brain evolution—that the genes and developmental processes that support cognition and language are also the ones that when disrupted lead to these disorders. What are you learning?
GESCHWIND: A very tough question. There are basically two major hypotheses that one can entertain about autism in regard to this question. One is that the genes that are critical to human brain evolution, to human cognition and behavior, are those that are disrupted in autism. The other is that most of the genes that are mutated in autism aren't those genes, but interact with those pathways.
So far, most of the genes that have been linked to autism are pretty highly conserved across species, even down to flies. In humans, these genes may be disrupting language. In mouse or in flies, they're obviously not disrupting language, but in many cases they are disrupting the same biochemical pathways. So the fundamental signaling and biochemical processes that we currently understand seem to be conserved across species. And if you’re searching for a treatment, that’s really where drugs act.
SIKELA: My lab studies autism as well. Earlier this year, we published a paper that links increases in DUF1220 copy number to increasing severity of the three primary symptoms of autism: social and communication impairments and repetitive behaviors. So it suggests that the same gene family that may be involved in human brain evolution is also involved in autism severity.
TKF: Do you think we can learn about the evolution of the human brain by studying a disorder like autism?
GESCHWIND: Yes. Some neurodevelopment conditions are shedding light on fundamental processes of brain function. Take the Fragile X protein, FMRP, which causes 1 percent of autism, as well as intellectual disability syndrome. The gene that codes for it is highly conserved from flies to mouse to humans. But to cause the deficits in social cognition and language in humans that it does, it needs to intersect somehow with the development of those processes and circuits. And we’re beginning to understand how it does that.