Small Changes, Enormous Differences
NOVA: What is the big question you're addressing with your research?
Katie Pollard: I've had a long-standing interest in human origins and in anthropology, and I've been curious for a long time in this very broad question of what makes us human. Ever since our species began to walk around on the globe, we've probably been wondering about this.
What we're trying to do in my lab is to bring a new type of data, genetic data, and a new tool, which is high-powered computing, to bear on this age-old and fundamental question.
When you were a student a few decades ago, could you imagine having these tools to look at human evolution?
No. When I was first studying human evolution, most of what we could use were things like bones and artifacts, things that we dug out of the ground, as well as what we observed looking around the Earth today—the behavior of humans and behavior of non-human primates. These were the tools of the trade.
I also had an interest in math back then, but I saw very little connection between that and my interest in human origins. Fortunately, in the next decade, we had a human genome project and then a chimp genome project, and there were major advances in computing. Suddenly this question of what makes us human had a whole new set of data and a whole new set of tools. And I was in a great position to jump into that field.
We were able to take a task that would have run for 35 years on a desktop computer and do it in one afternoon.
Why were you interested in comparing the DNA of humans and chimps?
Since chimps are our closest living relative on the tree of life, we can start to figure out what's special about us if we compare our DNA to their DNA. If we can find the little pieces of our genome that are unique, somewhere hiding in there are messages about what makes humans human.
How similar are chimps and humans in genetic makeup?
When we finished the chimpanzee genome in 2005 and lined it up next to the human genome, we made the amazing discovery that our DNA is almost 99 percent identical. That seems like a bit of a paradox given all the differences we see between ourselves and chimps. But, in fact, it's not a paradox, because all it takes to make a new species, all it takes to make a human, is a few changes in just the right places.
How did you begin to hone in on those "right places"?
Given that our DNA sequence has about three billion "letters" in it, one percent is still a pretty vast territory to search. There are about 15 million human-specific letters that have changed in the last six million years, since humans and chimps had a common ancestor.
No individual person is going to sit and look through that list. So tackling this question of which of those 15 million made a difference required writing computer programs.
And what did you find when you ran your programs?
After months of programming and debugging and running my computer code on a massive computer cluster, I finally had some results. The top hit was a sequence that was 118 letters long. And, amazingly, 18 out of those letters were different between human and chimp. To put that in perspective, the same sequence is in the chicken genome and has only two differences between chimp and chicken. There have been hundreds of millions of years of evolution separating chicken and chimp, and yet we only see two changes. Whereas in six million years, a blink of the eye in evolution's time, we see 18 changes in this stretch of the human genome.
[Editor's note: Such acceleration in the rate of change in the genome is a hallmark of what biologists call "positive selection," in which mutations that help an organism survive and reproduce are more likely to be passed on. The sequence of 118 letters Pollard found became known as human accelerated region 1 (HAR1).]
Why was this sequence particularly exciting?
This fast-evolving sequence was exciting in and of itself. But when I looked around on the Internet and databases that talk about things that people know about little pieces of our genome, lo and behold, it turned out this sequence is active in the human brain. At that moment, I just thought, "This is awesome." This is pretty much a home run when it comes to doing science.
You've honed in on other regions as well. How would you characterize them?
It turns out that the vast majority of these fast-evolving sequences are not genes, the parts of our genome that encode proteins. The pieces that have changed the most in our DNA look like they are switches, switches that turn nearby genes on and off. So what makes a human different from a chimp isn't that we're made up of different building blocks, different genes, but instead that we're using those pieces in different ways.
Another thing that was really amazing was that more than half of these fast-evolving switches are near genes that are active in the brain, either in the developing embryo or in the adult brain. In a way, this makes perfect sense, because our brains are essential to a lot of things that make us human: our speech, our culture, religion, even our ability to do science.
How many fast-evolving sequences have you now identified?
We've zoomed down now to about 200 sequences, most of which are switches. And the next step is to figure out exactly what genes those switches are operating and how the switches work. We understand how they work a lot less well than we understand how genes work.
We are able to get back to questions that people have been thinking about for hundreds of years.
These switches are part of what scientists used to call "junk DNA." How has the view of junk DNA changed in the past decade?
When I first started working in genetics, the focus was largely on genes. But we've learned that genes only make up a little less than two percent of our DNA. The vast majority of these three billion letters in our DNA aren't sending signals about how to make proteins.
At first, the thought was that there was just a bunch of junk in there. But within less than a decade, the way that we think about this so-called junk DNA has really changed. Encoded in those pieces is information about how to turn genes on and off—these switches—as well as features that affect the structure of the DNA. And the structure of the DNA is tightly linked to its function—you can't copy a gene or turn a gene on if it's closed in a tight structure.
Twenty years ago, did most scientists think that more complex creatures had more genes?
Yes. The prevailing view was that to make a more complex organism, you needed to have more genes. We can think about genes as building blocks. The idea was that if you have a lot more blocks to work with, you can make much more complex structures. But as soon as we started sequencing genomes, we quickly realized that even something like a brewer's yeast that we use to make beer or bread, a single-celled organism, really doesn't have a lot fewer genes than a human.
After realizing that a simple, single-celled organism such as yeast has about 6,000 genes and that a human only needs maybe three or four times that many genes to survive and function, we had to let go of the idea that to have a multicellular and complex being, you needed to have more genes.
So how do you get complex organisms? We realized two things: First, each gene, each place in the genome that we call a gene, doesn't just make one protein. You can make a lot of different kinds of proteins by using different parts of a gene. And the other thing we realized, which is perhaps the most important, is that it isn't how many building blocks you have, but it's how you turn them on and off. Part of what makes us humans more complex is we have a much more complex switching system.
Could changes in the switches account largely for how our species evolved?
Yes. By flipping a bunch of different switches, you can have a very profound effect. And if this happens in the developing embryo, you can end up with totally different structures, different kinds of hands, different-sized brains, more or less hair on the body—all kinds of things that can make a different species.
What sort of computer power does your work take? Would this sort of comparative genomics have been possible a decade ago?
This genome-wide scan for fast-evolving regions of the human genome would not have been possible 10 or 15 years ago, without advances in technology, in particular in cluster computing, which means that you basically take a bunch of computer hard drives and you stack them up together. Effectively, I chopped up the genome into little pieces and put a chunk on each one of these nodes. We were able to take a task that would have run for 35 years on a desktop computer and do it in one afternoon.
Would Darwin have been astonished?
Yes. Based on some really keen observations, Darwin was able to get the basic idea of how evolution works, which is really amazing considering the tools he had in his day. But fast-forward to today, and we're in a completely different ballpark. We've got these massive computer clusters, we've got DNA sequences to fill thousands and thousands of phonebooks—more data than we can even imagine—that we parse through with these computers.
We are able to get back to questions that people have been thinking about for hundreds of years: How does evolution work? What makes us human? And we now, finally, have some of the tools to get to the bottom of this.
So do you love your work?
What I love about my work is that on a daily basis I get to do something I think is really fun, which is geeking out on a computer and writing programs and thinking about biology. And that in doing this, I'm actually working on something that not just scientists care about, but really every human being can relate to and cares profoundly about, and that's what makes us human.