What's the oldest DNA sequence we can read?

Now our sausage DNA is “fresh”, but we can also use similar techniques to extract and read the genetic sequence of DNA samples that are far older. Eske Willerslev from the University of Cambridge is a pioneer in this area and explained to Chris Smith how far back in time DNA technology can take us...

Eske - I think my team still have the record for the oldest genome and that would be from a 700 thousand-year-old horse. But my prediction will be that we can even go beyond that, beyond 1 million years in some cases.

Chris - What sort of condition is the DNA in in things that are more than half a million to a million years old then?

Eske - It’s extremely fragmented so when you retrieve DNA from these ancient samples it’s very often that it’s no longer than a hundred base pairs. In comparison, if you take DNA from you and me for example, we would see stretches that would be millions of base pairs long. So it’s getting highly fragmented and there’s also other modifications to the DNA that can result in incorporation of the wrong bases during amplification that you just described.

Chris - Can one think of it a bit like a jigsaw, I suppose? And in me the jigsaw is intact, you can see the whole picture, but in a 700 thousand-year-old bone remnant or something, it’s a bit like someone’s shaken the jigsaw box and broken up the jigsaw into all the little pieces, so in assembling the genome back together you’d have to try and work out where each of the individual pieces go to make that picture?

Eske - Exactly. It’s like a major puzzle right. So you have these very short fragments of DNA that you then have to assemble into genome and to do this you normally use a reference genome. If you’re sequencing DNA from an ancient human, for example, using one of the human reference genomes and then putting piece by piece these small fragments that you have sequenced in place.

Chris - And you’re doing that painstaking working out where each of the fragments go, you’re doing that with a computer because it can basically look at millions of possible arrangements and combinations at a time and do what would take us millions of years to do?

Eske - Exactly, it’s through computer programming. But it’s quite surprising that with fragments down to 35 base pairs, I think it’s in the range of 60% of the human genome you can map uniquely with these small fragments.

Chris - How do you actually go about getting DNA out of these ancient specimens and why have we only gone back a million years? We’ve got lots and lots of things which are much older than that sitting in museums, haven’t we?

Eske - Yeah we do. One thing we discovered more recently is that there are certain parts of the skeleton, if it’s a skeleton that you’re looking at, where the DNA is better preserved. For example, the petrous bone which is part of the inner ear is the most dense bone in the body, and the DNA preservation seems to be very good there compared to other places. Also in teeth, you also find good DNA preservation compared to other places. But, of course, with time and also very dependent on what type of environment you’re dealing with, the DNA would finally get so fragmented that it’s actually impossible to retrieve any information out of it.

Chris - Are we at a stage now where with people like you able to get DNA from smaller and smaller specimens that are older and older, that sometimes we find the genetic image of a fossil, something that existed historically before we actually find the thing itself? So we know we’ve got a genetic signature for it but now we’ve got to go and find the thing that we’ve found the genetics for?

Eske - It’s correct - that’s not uncommon. You can say that we find genetic sequence that doesn’t match any existing sequence and, therefore, it derives most likely from some kind of extinct species that maybe hasn’t been described yet, or at least hasn’t been sequenced yet. But in some cases you also have material that has been described morphologically but hasn’t been sequenced yet and, therefore, you get a sequence that doesn’t match anything. It’s hard to say whether it’s one or the other.

Chris - What can we learn thinking about humans and our origins? What can we now learn and what’s emerging by using the sorts of techniques that you’re developing and applying them to human evolution?

Eske - You can study. You can say the biological history of humans. There’s a number of things that have emerged over the last few years but one thing that has become increasingly clear is that modern humans have migrated over long distances, even from the very early times. Spread all over the landscape and have met each other, mixing with each other at various timepoints. It also makes, for example, the whole idea of races kind of ridiculous because groups have been meeting, separating, meeting again since as far back in time as we can measure.

Chris - Can you infer anything about the social structure of humans because we’re a very social species aren’t we? Can the association of certain clusters of genes or gene sequences, which might be in certain populations, tell us something about who was mating with whom back in evolutionary time?

Eske - Exactly. We actually had a paper in Science this week on this issue here where we tried to understand did early modern humans actually understand the concept of inbreeding and outbreeding? Of course, if you get inbred you get all kinds of diseases and it’s very bad for the population. In order to do this we sequenced a number of individuals from the same locality 34,000 years back in time and, to our great surprise, these individuals which are coming from a very small group of people doesn’t seem to be inbred. They don’t seem to be as closely related as most people would have suggested. This really tells that at least in the upper palaeolithic, 34,000 years ago, people had an understanding of the need of getting mates from outside the group itself, in that sense avoiding inbreeding...