 Well, we certainly have a terrific turnout this afternoon. I think the reason for it was because you all heard rumors that a Neanderthal Institute director was going to introduce the speaker. I am that Neanderthal Institute director, Eric Green, director of the National Human Genome Research Institute, and I want to welcome you to actually a special talk that the institute is putting on, and shortly you'll be hearing from our special guest Dr. Svante Pabo. Dr. Pabo received his PhD from the University of Uppsala in Sweden, and he conducted his post-doctoral research at the Institute for Molecular Biology at the University of Zurich in Switzerland, and also in the Department of Biochemistry at the University of California at Berkeley. Now, if you fast forward, he is now the director of the Max Plant Institute for Evolutionary Anthropology in Germany. He's also an honorary professor of genetics and evolutionary biology at the University of Leipzig in Germany, and a guest professor of comparative genomics at the University of Uppsala in Sweden. And previously, he was full professor of general biology at the University of Munich. Now, Dr. Pabo is truly an authority on the study of ancient DNA. He was the first to successfully clone DNA from a mummy. His DNA sequencing efforts have determined that ancient bones come up from a previously unknown species of hominid called denisovins, which he's going to talk about. In 2009, he completed the sequencing of the Neanderthal genome in an effort that included a collaboration with Dr. Jim Mulligan, the director of the NIH Intramural Sequencing Center at NHGRI. And recently, his research has shown that Neanderthals interbred with modern humans some 50 to 80,000 years ago. This year, Dr. Pabo had a book come out entitled Neanderthal Man in Search of Lost Genome. And in this book, he explores the concept of what we can learn from the genes of our closest evolutionary relatives. It details his impressive portfolio of genomics research from the early 1980s to the present day, and also will notably explores the origins of modern humans and our relationships with Neanderthals. In February, his book received a glowing review from the journal Nature. In addition to his book, he's co-authored over 250 publications and his list of international awards and accolades is impressive. As an example, 2007, he was named by Time Magazine the list of 100 most influential people in the world. And various other awards decorate his very impressive CV. He was also recently elected to a member of the Royal Swedish Academy of Sciences, and he was also a Gruber Prize winner in genetics. It's been just a pleasure to have Svante here this week. He actually, two nights ago, down at the Smithsonian's National Museum of Natural History, gave a public presentation put on by the Smithsonian Associates involving a partnership between NHGRI and the Smithsonian Institution associated with our exhibition, Genome Unlocking Life's Code. And associated with that are a series of evening public events. And Svante was featured two nights ago as one of those events and it was really terrific. And he's even been more generous with his time. He spent yesterday at the Smithsonian giving a couple of talks, talking to their scientists. And then he's spending all day today here visiting NIH, giving this talk and visiting with NIH scientists. And then back tomorrow to the Smithsonian for more meetings before he gets to fly home to be with his children this weekend. So it's truly a pleasure to have Svante here joining us at NIH today. So please join me in welcoming a gifted scientist, a valued colleague and a good friend, Dr. Svante Pebo. Thank you very much. It's really a great, great pleasure to be here. I think it's 25 years ago when I was last visiting physically at the NIH, although we have collaborated a lot. And it's impressive the changes and growth that has been. Although it's much harder to get into the facilities now than it was 25 years ago. So what I wanted to do then is talk a little bit about the technical efforts in retrieving ancient DNA and sequencing ancient genomes. And then focus on what we have learned by studying genomes from our very closest extinct relatives, Neanderthals and relatives of Neanderthals. And for the last third of the time or so, talk about where we go from here, how we can use these genomes to try to get to functional differences between us and other organisms, particularly them, Neanderthals. But just as a means of introduction, I want to remind you about the fact that most of you are very well aware of that. If you look at genetic variation among present-day people, you find most of that variation in Africa. Whereas although there are many, many more people living outside Africa, we actually have less variation there. And not only that, there is almost all more common haplotypes anyway that you find outside Africa have closely related sequences inside Africa, but there is a component of the variation, so to say, in Africa that you don't find outside Africa. And the interpretation of this is that modern humans evolved in Africa, accumulated variation there. And a part of that variation went out and colonized the rest of the world. And by different tricks, looking at linkage disequilibrium, for example, in these haplotypes outside here relative to inside, you can estimate when that exodus out of Africa happened. And it's in the order of 100,000 years ago or less. So this is the basis of the recent African origin model for the emergence of modern humans. But there is, if you like, a problem with this model, and that is that 100,000 years ago, there were many other forms of humans around. Most famously, then Neanderthals in Europe and in Asia, other less well-described forms of humans. So here is a reconstructive skeleton of a Neanderthal compared to present day human. The Neanderthals were these sort of robust forms of hominins. They appear in the fossil record three, 400,000 years ago. And disappear again around 30,000 years ago. Generally, rather contemporaneously with that modern humans appear in an area. So there are two major ideas around in paleontology of how modern humans relate to Neanderthals and other extinct forms when they come out of Africa. One in this total replacement model says modern humans comes out, replaces these other forms with no mixing whatsoever. Another idea is that these modern humans that come replace the Neanderthals, but also mix with them and accumulate genetic variation from them to present day people in Europe and with other forms in Asia to present day Asians. So the first, so you could regard this as sort of a sliding scale from total replacement, zero percent contribution from these extinct forms of humans to more and more contribution up to total continuity that I think no one sort of advocates at least in 15 years or so. So the first chance to test this with genetic means then came in the mid 90s when we got access to this specimen, which is not just any Neanderthal, it's the Neanderthal from Neanderthal that was found in 1856 and gave its name to this group of hominins. And it was actually the first time I realized that there had been other forms of humans here before the present form of humans. So we then extract DNA from such specimens under sort of clean room conditions to avoid contamination from ourselves into the experiments. And at that time with the technology that was there with PCR was focused on a particular valuable part of the mitochondrial genome. Canberra's only amplifying it, cloning pieces, believing the substitutions that were consistently there. And if you then reconstruct a phylogenetic tree for this mitochondrial genome, you will find that all present day mitochondrial genomes go back to common ancestor a hundred, two hundred thousand years ago. And much further back, half a million years ago or more, is there a common ancestor shared also with the Neanderthal mitochondrial genome. So since then, we and others have sequenced many other mitochondrial genomes from Neanderthals, they all fall outside the variation of present day humans. So it's quite clear from this, there is no people walking around on the planet today with a mitochondrial genome from Neanderthals. So in this scheme of things, for the mitochondrial genome, it's total replacement. We of course learned something else from this too, and that was this age of around half a million years for the split of the mitochondrial genome of Neanderthals from present day humans. That suggests that the population split between what became Neanderthals and modern humans later, happened sometime after half a million years ago. So that of course already means that the Neanderthals cannot be very different from us. Because if I just compare two pieces of DNA into genomes today, the average time back to common ancestor from those two pieces is somewhere around half a million years, with a large variance around it of course. So there is no problem to find some pieces in the genome where I am someone in order to differ by a million years. So that means that also if there was no mixture whatsoever for the nuclear genome, it would be the case that for many parts of the genome, someone today would be closer to Neanderthal than someone else in the audience would be. And if we moved to another part of the genome, the situation would be different of course. But this then doesn't mean that the Neanderthal genome would be uninteresting to study, because there will of course still be parts of the genome where all modern humans sort of fall outside the variation of Neanderthals. And the opportunity to go after the Neanderthal genome then came with high throughput DNA sequencing, where you could leave this sort of trying to retrieve a particular little piece of DNA, but simply go and extract all the DNA you have in a fossil, process it, a sequence it all to exhaustion, create a little data bank, and start seeing what pieces of DNA here look human-like, so they might be from the Neanderthal, what is bacterial, and so on. So this worked for the first time in the cave site in southern Europe, in the cave in Croatia, this bone that is 38,000 years old. And the first thing you will see if you look at sequences from such a bone is that there are indeed short pieces of DNA here, 50, 60 base pairs, hardly anything approaching 200 base pairs. Also, it's only a tiny fraction of the DNA that actually counts on the Neanderthal. The vast majority, in most cases, 99% or more of the DNA comes from soil bacteria and fungi that have lived in a bone when it was deposited in the cave. So we started the project and we're very happy to get funding for an effort then to sequence the entire, the first sort of overview version of the Neanderthal genome. And during that time, we worked a lot on making the methods more efficient in how we come from a bone to a sequencing library that we can put into the sequencing machines. The machines in that time got more efficient in terms of how many molecules they could sequence. We looked through many, many sites, many bones, to find the ones with the most Neanderthal DNA in them. And it was in the end, three bones from that site in Croatia, from three different Neanderthal individuals that generated data for the first version of the Neanderthal genome. So we sequenced a bit over a billion DNA fragments, sort of mapped them to the human genome, having algorithm to take into account that there are typical types of errors in these sequences that you have to take sort of into consideration. And we had a one-fold coverage of the genome so that we had seen just a little over half of the genome once. But that gave us the first chance to start asking questions. And the first question that we were interested in was this, what happened when modern humans came out of Africa and met Neanderthals? Did they mix or not? So to address that, we did that in several different ways because this is a very controversial question. There were really sort of wars in paleontology going on about this. So I had the feeling that I had to get it really right. But one very direct approach to ask that was to say that if there was a contribution from the Neanderthals to people in Europe, we would expect people in Europe to share more alleles with a Neanderthal than people in Africa where there had never been Neanderthals. So it's this idea here. If there is no mixture whatsoever, then the Neanderthal is equally far from people in Africa as in Europe today, whereas here it would on average be closer to Europeans. So we went out at the time and sequenced five genomes. We would have exactly the same error rates, et cetera. One person from Europe, for us, of course, the archetypical European is a French person, so this is a French genome. Two African genomes, one from China, one from Papua New Guinea. And we did a very simple analysis. We are saying, if you compare now these two African genomes, find positions where they differ from each other, and then we take the Neanderthal genome and say, how often does it match one African or the other African? And since Neanderthals have never been in Africa, so there's no reason for a Neanderthal to be closer to one African than the other, it should be 50-50, which is the case, statistically speaking. However, if we compare a European to an African, we found significantly more matching to the European individual than the African individual. When we went to China versus Africa, we again find the same thing. And even Papua New Guinea, we see the same thing. So this was very surprising to us at the time, because if we would see a contribution, we would have expected it in Europe when Neanderthals had lived, not in Papua New Guinea and China, where they had not existed. So the question was, how could this be? And the sort of model that we had time suggested as the most plausible model that has since been born out by much other work, also from other groups, is to say that when modern humans come out of Africa, between 50 and 100,000 years ago, they're presumably passed by the Middle East, and we know they lived Neanderthals there. So if those modern humans mixed with Neanderthals and then became the ancestors of everybody outside Africa, they would sort of carry with them this Neanderthal contribution out into the world to the extent that people outside Africa today have something between one and 2% of the genome from Neanderthals. There has been, we now begin to know a little more about this admixture process also because it has been observed that in East Asia, there is actually more Neanderthal DNA, about 20% more, so say instead of 1%, 1.2%, than there is in Europe. And Josh Ake published a paper in Science in January where they modeled this contribution from Neanderthals once to the common ancestor of all Africans, and later admixture went affecting only the ancestors of Asians, and that fits the data significantly better. So there seems to have been at least two events of admixture here, one to everyone, one perhaps in Central Asia or something like that, that affect the ancestors of Asians. You may also ask, when did this happen? That was also important to exclude other scenarios. And an approach to estimate that came primarily from David Reis' lab in Boster, which takes advantage of this fact that if you have two populations like Neanderthals and modern humans mixing, the first generation hybrids will of course have one chromosome from Neanderthal, one from a modern human. If these hybrids then continue to live in the modern human population and mix with them, you will have recombination in each generation, of course. So this Neanderthal DNA will be broken down to smaller and smaller pieces as generations go by. So you can then use the size of such Neanderthal chunks to get an approximate age of when they entered the human gene pool. So you're looking at things like this. If you have European sequences up here, the variants are sort of indicated in yellow, the Neanderthal down there, you will find that some Europeans here are almost identical to the Neanderthal over some tens of thousands of base pairs and quite different from everyone else. So you look at the size distribution of these things, make some assumptions about mutation rates, generation times, and genetic recombination landscape and arrive at a date somewhere between 37 and 86,000 years ago, which fits with when modern humans come out of Africa. Now, this relies on a lot of assumptions, this modeling, and I'm, of course, always a friend of real direct data. So there is, of course, another way to go after when this happened, and that would be to go back to really early modern humans and see if they had mixed with Neanderthals or not. And it's beginning to become possible to do that, I think. This is a place in Western Siberia called a little town called Ostishim, where two years ago one found a bone on the riverbank here and this is this bone here that we have just recently analyzed. So this is not published yet. But we have sequenced this genome now to 42x coverage from this bone, and we've also carbon dated it. And to our delight, this bone turned out to be 45,000 years old, so it's clearly the earliest modern human outside Africa and the Middle East that is known and we have a good genome of it now. So if we now just look as an example here on one chromosome, chromosome 12, and plot here what comes from Neanderthals, in Europeans and Asians, you can almost see that there is slightly more Neanderthal contribution in Asia than in Europe, actually, when you look at this. And the question is this, now this 45,000 year old individual have that one mixed with Neanderthals or not? And the answer is that indeed it has, but we see a lot bigger chunks, of course, as we would expect back there. So we can now go and look at sort of correlations with genetic distance in present day humans. This is the data that leads to this date of 37,000 to 86,000 years ago. This is this 45,000 year old early modern human. And this leads to an estimate that it happened around 300 generation before this guy lived. So that takes us to somewhere in the order of 50,000 to 60,000 years ago. So we're sort of beginning to narrow in on the time point when this happened. Something else that we're interested in is of course other extinct forms of humans than Neanderthals. And we are very lucky to work together with Professor Derev Yanko in Novosibirsk and his colleagues, but Professor Shunkov, that excavated many places in Siberia, but particularly at this cave site there, in southern Siberia, the Nisova Cave, on the border to China and Mongolia and Kazakhstan and Russia where the four countries meet. So it's at this place here in the Altai Mountains, where they in 2008 found what they very skillfully recognized might be a bone from a human. It's a fragment of the last phalanx of a pinky of a child. So we sort of got this bone and extracted DNA from it and it turned out to be surprisingly well preserved. So whereas our very best Neanderthal bones have 4% Neanderthal DNA, this one had 70% endogenous DNA. So we went on and sequenced the first sort of overview of this genome. And to our surprise found that this Nisova finger has a common, the genome has a common ancestry shared with Neanderthals, but Neanderthals since have a long independent history. So the divergence between this individual and Neanderthals is substantially deeper than any divergence among present-day people here. So we sort of then named this sort of group of extinct hominins, denissivants after the first site where they were then found, just as Neanderthals are called Neanderthals after Neanderthals were first discovered. It's the first time then that an extinct group of humans is described just from a genome sequence. What we have happened since is that we have improved our methods quite dramatically for retrieving small amounts of damaged DNA. There are many little tricks to this, but the major trick is really a library preparation method where we use not double-stranded DNA that you add adapters to, but start by denaturing the DNA, by single-strand ligation add-on adapters with a biotin on them so we can immobilize them, synthesize the other strand, and then get something you can sequence. So this means that each double-stranded molecule has two chances to enter the library, so to say. So whenever you have some chemical modification on one strand that makes it impossible to replicate the sequence of molecule, the other strand still has a chance to enter the library. And when we sequence deeply enough, we actually see that we get both strands from molecules. So we've now been able then, we're then able to go and get high-coverage genomes from these things. So I have a 30x sequence from this finger bone here, where we then have very good coverage of the part of the genome we can map short fragments to, so say 1.8 gigabases or so of the genome. So what you can then do is, for example, begin to look at the time back to common ancestry for particular parts of the genome. You can look at the two alleles and see how far back they coalesce. And you can do that for hundreds and thousands of parts of the genome and look at the distribution of times back to common ancestry here. And with a method that Hengli and Richard Durbin have developed, you can then estimate effective population size back through time. And obviously, where you have more of these coalescent events is when the population have been smaller. So in this example, a dip here. So from looking at a single individual, you can estimate the population size history of the population that the individual comes from. So if you do this now for present day people from Africa, Europe, and Asia, you will see that everyone here shares, if this is a past, everyone shares a decrease in population size, an increase in population size. And only quite recently, since 50, 100,000 years, do we see a difference where Africans have a bigger effective size than non-Africans. And now we can look at this Denisova individual and look how that population history looks. And what you will then see is that if anything, it's likely bigger here, but then as it crashes down and goes extinct. So it's quite fascinating to me that the population history of this individual is dramatically different than anybody else that's around on the planet today. What you can also do when you have old, but high quality genomes is to begin to look for what we could call missing mutations. So an individual that died tens of thousands of years ago have had less time to accumulate mutations, of course. So in this case, if we say how many mutations do we miss here relative to the common ancestor to the chimp, it's a little over 1%. So if we now claim that this is 6.5 million years, which is very uncertain, but let's say that, this would mean that this bone is 60 to 80,000 years old. And this bone is, of course, so small so you cannot carbon-date it and you can actually not date it by any other means. But when we haven't, it's sort of, there are many caveats with this, I should say. We, of course, have different error rates in these present-day human genomes. So you actually see that the apparent age of these present-day humans vary by about 20% of this. There's something like 0.2% variance here back to this. But in the future, when we have even more accurate sequences, I think one can actually then date bones from genome sequences by this approach. Even bones that are then sometimes so small so you cannot use other physical methods for dating. You can ask for these denissivance, as the ponandatos have they contributed to present-day humans? And quite surprisingly, they have done so, but not very much in mainland Asia, but out in the Pacific. So Papua New Guinea, Australians, Fiji, and so on. So this suggests, presumably, that they were more widespread in the past so early modern humans mixed with them when they came to Southeast Asia, the ancestors of what is today the people in the Pacific. So I think this is sort of an indication of what will happen much more in archaeology in the future, that from tiny little bones, you can reconstruct a lot of the population history of individuals you find. But very frustrating, they have no idea how the skeleton of this individual looked, what stone tools they made, et cetera. But to just summarize before we come to the last part of the talk, about what we then think about the origin of neanderthals and denissivance first, we think they have an origin in Africa and come out of Africa and evolve in Western Eurasia to what we call neanderthals and in Eastern Eurasia to what we call denissivance. This is not to say that they were this widespread at any one point in time, nor that it was the only forms of extinct humans there. We don't really know what the border with neanderthals were, but at some time in these altimounties of neanderthals, at some other time, denissivance. Then modern humans emerge in Africa, come out of Africa, mix with neanderthals, presumably in the Middle East, and around 50,000 years ago, start spreading seriously over the world. They mix once more with neanderthals and perhaps in Central Asia, and they mix with denissivance, somewhere in Southeast Asia, and continue out into the Pacific. And then these forms become extinct, but they live on a little bit then, so to say, in people today, in that something like up to a maximum 2.5% perhaps of people's DNA come from neanderthals and you add another 5% here from denissivance in the Pacific. Now we sequence two extinct genomes and found two cases of admixture. I wouldn't be surprised if we find more, for example, in China, in the future, other forms that could also contribute to the bit, nor if there would have been contribution in Africa. So I don't think there's necessarily an absolute difference between Africans and non-Africans in that only non-Africans have this archaic contribution. But so we have clearly rejected this total replacement idea. We have a contribution from these extinct forms, but the major picture is of course still an exodus out of Africa, and most of our variation comes out of Africa. So the sort of term for this, I like to sort of suggest something like leaky replacement or something like that. There is a replacement, but a bit of genetic contribution from these other forms. So what's next then in this field? Well, one thing that was actually now just achieved a few weeks ago is that we now have a good Neanderthal genome also that we published actually two months ago now. Quite curiously comes from the Altai mountain and the very same cave as the Denisova finger, where deeper down in the stratigraphy in 2010, they found a toe bone. And that toe bone we obviously thought would be Denisovan. We went and sequenced the genome, but it turned out to be far away from the Denisova and close to other Neanderthals. So this is a Neanderthal, and we went on and sequenced the genome to 50X. So we can now look at that genome too. So if you just look at heterosagosity, for example, how much variation is there between the two versions of the genome here? If we compare present day Africans to non-Africans, we see this that non-Africans are less variation. And here are then the Denisovan and Neanderthal. So they have very little variation, extremely little. And not only that, if we look at this Neanderthal genome, we found something else. We found that there were huge chunks on the chromosomes that were identical between the two parents. So an example here of chromosome 21, whereas we don't see such long pieces saying the Denisova genome. This obviously indicates that the parents of these individuals were very closely related. So you can then model the relationships that these individuals' parents must have had. It must have been one of these four relationships. They were half-sib, say, or grandfather, granddaughter, or these other relationships that are complicated. I can't even describe them. But it's sort of quite interesting that we can get to some of the social things that went on in that cave sort of 60,000 years ago in Siberia. And it will be very interesting when we have high-caverage genomes from other Neanderthals to see if this is a general pattern in Neanderthal or something special for this population. We do see evidence of closely related ancestors also further back in the pedigree of this individual. It's not just in the last generation. So we then have these two good genomes here. We have one Neanderthal, low-caverage genomes from the Caucasus and the one from Croatia. So we can now begin to look at contributions not only to inter-presenter humans, but between these extinct forms. So if we do that here, modern humans, Neanderthals, Denisovans, we see this contribution from Neanderthals to non-Africans and from Denisovans to people in the Pacific. We now also, with the high-quality Denisovan genomes, see a tiny contribution to mainland Asia. So people in China have something like 0.2% Denisovan contribution. But we can also see that there have been genes flow from the Siberian Neanderthals into Denisovans. And quite interestingly, we find a quite old component in the Denisovan genome that we don't see in the Neanderthals that comes from something else, something that diverged earlier from the modern human lineage between one and four million years ago. So it's very tempting to suggest that this is homorectus in Asia, or something like that, that contributes to Denisovans. So it's very clear, I think, all these human groups have mixed with each other, but generally it's mixture of low magnitude. We don't haven't yet found something where 30, 40% have come from some other group. What you can then also do now is start to look more careful in present-day human genomes, in the 1000 Genomes data, for example, for contributions from Neanderthals. So there were two papers that appeared in January in Nature and Science, one from David Rives' group, where we were involved, and one from Josh Ake's group in Seattle, where one had looked at the frequency of integration in the 1000 Genomes data of Neanderthals pieces. And you will see that there are some things that actually reach high frequencies in humans, 70, 80%. So if you first ask what are those things that have risen to high frequency, what genes do we find there? The only functional group we consistently find are keratin's inherent skin. So something like that seems to come over and influence the phenotype today, perhaps. Already earlier, it had been shown from a group of Stanford, from Peter Parham's group, that MHC elites had come over from Neanderthals and from Denisovans, and sometimes reached high frequencies, presumably from selection, perhaps from pathogens or so. And in December, there was a paper published in Nature from the Sigma Consortium that David Altschuler is organizing, where they found a new risk allele for type two diabetes in Mexican Native Americans, a lipid transporter, but the risk allele carries four amino acid differences from the non-risk allele. And this risk allele occurs in East Asia and sometimes up to 25% or so native Americans. And if you make a tree of the risk alleles and non-risk alleles and put in the Neanderthals here, you will find that the Neanderthals fall right into that. So this is obviously an allele that had come over from Neanderthals into ancestors of Native Americans in Asia somewhere and risen to high frequency, perhaps because it conferred some advantage in starvation or something like that, and now causes type two diabetes. So some of these mixtures, where these archaic forms actually have, seems to have some phenotypic consequences. What you can also look for here is of course regions where you don't find contributions from Neanderthals. So you can look in present day genomes where for regions where you would statistically expect the Neanderthal contribution, yet you don't see it. So it seems to be negative selection there. And these are just some examples of that on the X chromosome and X elsewhere. And if we ask what genes are located in such deserts, as we would like to call them, of Neanderthal contribution, where are those genes expressed? The only tissue that stands out with genes expressed in such regions are genes expressed in the male urn line in testicles. So it's very tempting to suggest that maybe the hybrids had some problem with male fertility, which of course is not uncommon when close related populations mix with each other, be that horse and donkey or what have you, that the male offspring is sterile and the females are fertile. So there may have been something like that going on. But to end then, I would like to spend a little bit of time on what I perhaps find as the most exciting sort of future development of this. And that if we can get to sort of functional things that are unique to modern humans relative to our closest relatives. So things that have changed here, which we can now identify when we have the high quality Neanderthal and the Nisman genomes that have changed here, yet spread to everybody today, no matter where we live on the planet. So things that happened here and became fixed in humans very recently in our history. And why are those sort of changes particularly interesting? Well, I do think that there are some things that have happened when modern humans came that have really set us off on a totally different historic trajectory, of course. So if you look at stone tools, for example, that the Neanderthals made 300,000 years ago when they appear and when they disappear 30,000 years ago, those stone tools look pretty much identical to each other. All technology across the whole range of Neanderthals is also pretty similar. When modern humans come, all that changes, technology starts changing rapidly and becomes regionalized in different regions. Modern humans make different tools and we have, after all, existed only a third of the time that Neanderthals have existed on the planet. And I think we agree that our technology today is quite different as 50,000 years ago. There are other things such as figurative art, art that really depicts things, comes only with modern humans and, of course, this thing that we spread across the whole globe. These earlier forms of humans lived for two million years in Africa, Europe, and Asia, never crossed water where we don't see land on the other side. Then come modern humans and in 50,000 years, one colonized not only America's, Australia, but every little speck of land in the Pacific. So the hope would, of course, be that some of this sort of reason for this very special thing that happened in human history would be hidden in this catalog of genetic changes that characterize present day humans. So if we now, just to illustrate this, take a very strict definition and say, what are the changes that happen here that to our best of our knowledge is in 100% of all humans today when we look at all databases we have access to, yet Neanderthals and the Nisvans look like the apes. Then that catalog is not a very big catalog of things. It's 31,000 substitutional differences, some insertions and deletions, and so on. So you can scroll through them in the computer in a few hours. So it's, for example, only around 3,000 regulatory regions. Now it's, of course, how you define regulatory regions, but if you take one definition of them, it's something like 3,000 changes. It's 96 amino acid changes, and I just focus on them for a second to sort of illustrate how you can go after this. This is not to say that regulatory changes are not perhaps even more important. So these 96 are actually located in just 87 proteins because some have several changes. And you can, of course, now start saying, could some of these be important? Hopefully there will be many experts on these genes that will sort of find interesting things here. We are, of course, biased to things that's saying things in the brain might be particularly interested. So we were quite interested when the Allen Brain Institute actually looked in different parts of the developing human brain and looked at these genes that have amino acid differences, how many of them are expressed there, and there's a control genes with silent changes that don't change amino acids, but otherwise are resentened in exactly the same way, fixed in humans, ancestral inandotoxin, and so on, and found that only in the proliferative zone where the stem cells divide to make new neurons, do we see significantly more of these amino acid proteins that have changed amino acids expressed? And these are, of course, signals that rely on few proteins. It's actually relies on five proteins, that signal, and quite strikingly, three of those five proteins are involved in the kinetic core and in the spindle. That surprised me a lot, but there are, of course, indications that say the mitotic cleavage planes of the stem cells in this epithelium when neurons are born determine, say, how many cell divisions you have and what types of neurons you make. So this is just sort of idle speculation that perhaps these three proteins should be particularly interested in studies in brain development, what these changes meant. But that brings us to the last question that I had, and that is this question, how should we go after these types of changes that are essentially fixed in humans, yet are perhaps involved in human-specific traits where we, per definition, have no sort of animal model? And I've gone around for years making jokes in the talks like this and saying what we want to do is take the Neanderthal allele and put in a transgenic human and the human allele in a transgenic chimp and said that we can't do that, of course, or we have to think about something else. But now it's suddenly not so ridiculous anymore because there is people like George Church, a professor at Harvard, no less, that go around and it's even more extreme like this and say we should clone the Neanderthals, we should put in all these changes in our catalog in a stem cell and make an individual. And I don't know why we are sort of forced to discuss this, I think it's technically impossible and ethically unthinkable, of course. But it doesn't sort of solve the problem, of course, we sort of, how will we go after these things? And I think that one thing that will happen soon, but people here are much better positioned than me to think and discuss this, is to say that I think in the future we'll be able to find back mutations in humans when millions of people will have a genome sequence when they go into the doctor's office. We'll, of course, know that all mutations compatible with human life exist in the human population because the genome is small enough and we're enough people there. But I think it will probably become ethical possible to find ways to go out and sequence and study these individuals, but that's a better way. You can engineer these into IPS cells, of course, and we and others are in the process of doing that to study differentiation of cells in vitro. And what I want to make a little bit of propaganda for the last three, four minutes, is then that one could perhaps after all use model organisms such as the mouse. And I want to illustrate that with one gene that we have studied since nine years now, it's a gene called FoxP2 on chromosome seven here. And it caught our interest because it encodes a transcription factor that when one copy is mutated, it's not only to severe language and speech problem in human families. It's also a very conserved protein that's encoded. So if you look in the mouse, it has only one amino acid difference here all the way to the chip. But interestingly, two amino acid differences on the human lineage there. They are close to each other in one exon. So it seemed very interesting first of all to look in the Neanderthal genome and Denisovagino to see if these changes are shared with them. And indeed it turns out it is. So these changes are carried also by Neanderthals and Denisovans. So they happen back here. So they are not a modern human specific one, but they are still interesting of course if they have something to do with speech and language with human specific traits. So what we have done already some years ago is to engineer these two changes into the human, into the chimp FoxP2 gene. So since this is such a conserved protein, we now have a mouse that essentially makes the human protein from its endogenous FoxP2 gene with one other very conserved change over there. So you then have this mouse and you have the problem that you want to study human specific traits, speech and language. So we tried to speak to the mouse, but it didn't lead to anywhere. So instead we said this is a very conserved protein. So it could be anything in the phenotype really. So we were very lucky to work together with this mouse clinic in Southern Germany that do a very comprehensive screen of then the homozygous knock-in mice with their litter mates that are wild type. So do look at very many phenotypes as to give you a feeling for it to sort of going through them here. Hematology, what have you, eye function. All in all, they looked at 323 phenotypic traits. Not at all all independent of each other. Many are of course highly correlated, but it's still a very comprehensive phenotypic screen. We had two independent knock-in lines we made and there are only two phenotypes that were significantly different between knock-ins and litter mates. The first one is that the humanized mice are slightly more cautious in a new environment. Measured so that you have a social group here, you open doors and the mice can go in and explore an open area. Mice feel more safe along the walls. So the humanized mice stay at the walls, whereas the other ones do mostly that, but also venture in the middle here. But it's just a transitory phase. It's just the first few minutes are slower to explore the middle of the open field than the wild type mice. I have no clue what this means whatsoever. The other phenotype though arose my interest and that is that they vocalize differently. There's no way to say they vocalize in a more human-like way, but they measure that by taking two weeks old pups out of their nests and the peep in the ultrasound area here and the mound comes and brings them back into the nest. And there are a number of features in these sonograms where the humanized differ from the wild type. Sometimes it looks like sort of a dominant trait, sometimes co-dominant. So this is very interesting. It's sort of support that this changes might have something to do with muscle control and the oral pharynx or something like that. But the frustrating thing is that there is sort of hardly anything known about the neurobiology of mouse vocalization, of course. But you, after all, now have an animal model. So you can take out neuronal precursor cells, for example. You can differentiate them in vitro. And you will find here that if you do that from the stratum, they actually grow out longer neurites. So things that would become dendrites and axons. And in the live animals, you can also see bigger dendritic trees in the stratum. You can do electrophysiology in slices from these things and you can see that long-term depression where sort of the neuronal excitability goes down after a high frequency stimulation, sort of model for learning and memory is much stronger in the stratum in the humanized mice than in the non-humanized mice. So what we then know about this, it alters vocalization, you have more dendrites, you have this increased genetic plasticity and you have decreased dopamine levels that I haven't mentioned. So it sort of generates a hypothesis. The reasonable hypothesis is that we would have something to do with corticobasal ganglase circuits. Where the basal ganglase receives input and gives output to the cortex and the brain stem that has to do with motor learning often. And there are other sort of things that support this, the readings in the brain where we see this bigger dendritic trees. Just very recently then, a previous graduate student have worked with Anne Greyville at MIT and looking at motor learning in these humanized mice. So they do different tests where the mouse have to sort of learn to go towards a signal to get some reward that can be a tactile signal here or a visual signal. And under certain paradigms, these mice learn in seven to eight days what the wild type mice will take 11, 12 days to learn. And by doing various variations of this, this is not my field, I just relate what they have found. They have sort of made it likely that this has to do with a switch from sort of what they call a declarative phase of learning where the animal needs the cue, nose, it goes towards the lamp. And then after a while, you can stop turning on the lamp and it will know it will go to the left. It will sort of come to a procedure phase where it's sort of endogenously know I just turned left to get my reward. So they illustrate this to me by saying if you think about how you learn to bicycle in the beginning you think about how you bicycle and you're very bad at it. And then after a while you sort of automated it and you get very good at bicycling. And this correlates with activity shifts from the medial part of the striatum to the lateral part of the striatum. So this generated sort of hypothesis that we would see some difference in synaptic plasticity medially and laterally. So we then went back and did the electrophysiology in the mice and in the lateral part of the striatum we see this increased synaptic plasticity, increased LTD whereas when we then look medially we see no effect at all or if anything an effect in the opposite direction but no significant such effect. So this then seems to sort of support the current hypothesis that these amino acid changes have something to do with changing corticobase or granular circuits to make for faster proceduralization sort of automation of learning and perhaps then with aspects of learning speech and language. And you can of course make a nice sort of speculation and say what we learn to do when we learn to speak is that we automate coordination of muscle movements in oral pharynx to produce articulate speech. And it's probably the most sophisticated muscle coordination that we actually do in our life. And a form of thing that say apes are not able to do. I should accelerate a little bit here. I should just say in the FoxB2 gene we also find a regulatory change it seems that is very conserved all the way to frogs yet changed in modern humans. And it affects the PAL-3F2 site where we can show that ancestral version actually binds less of the dimer and also it's more efficient in driving expression in the model system. But we don't know anything more about that but it somehow seems that we have these amino acid changes and perhaps a regulatory change then there. So this is a lot of work over several, several years of course and in the end we sort of have to study these changes I think in this systematic way back into the organisms to then hopefully maybe in the end understand something of what set humans on such a very special track among all other primates. So I hope I sort of convinced you that if you're interested then in human evolution and recent changes in humans it's very good to have the genomes or closest relatives because you can sort of see what have happened on this line here in the future also here when we have more genomes here but it will not be enough to have the genomes. The genomes are just a two locals what we have to do in the end is sort of functional work and I think the most reasonable well forward there is after all a humanized mouse. So there are many people involved in this many more than I can mention I'd already mentioned that in the analysis consortium Jim Mulligan and his group here played an important role. I also want to mention just one person Mattias Meyer and the group that came up with this single stranded library protocol that really made it possible to generate these high-caverage genomes from these extinct forms of humans. And with that I thank you for your attention. We have microphones in each aisle and so if people have questions I'd ask them to please go to a microphone. Yes? You mentioned at the beginning that mitochondrial DNA did not support a mixture hypothesis. And later you said that wouldn't this imply that the DNA from the Neanderthal was male? Yes, so that would be one explanation for it. It's also compatible with that. This is just stochastic variation. The mitochondrial genome is inherited as just one unit. So it's rather easily lost if a female has no female offspring it's sort of her lineage is lost. So the more boring explanation is that. We also see on the X chromosome reduced contribution from Neanderthals. When we first saw that I said that's clear. It has to be male gene flow because males transmit only to have their children that transmit an X chromosome. The other ones get a Y. But then we've seen these sort of deserts in other chromosomes too. So that made us shy away from saying that this is male gene flow. But if I would guess I would say it's mainly males. Okay, and doesn't this conflict with what you said later on that Neanderthals had less, probably less fertility? Well, the hybrids would have had less fertility. So I think in the hybrids this would suggest that the female hybrids may have been fertile but the male hybrids may have reduced fertility. In the analysis that led to the testes specific genes standing out in speciation and model systems heterochromatic regions can be very important. So do you see any signals pointing you towards those non-genic regions? The only analysis we have done is that in these deserts what genes are enriched there. So there might well be, we would not have seen it as we have analyzed it. Much of the heterochromatic things are perhaps also repeated or so there we have really mapping problem from our short little pieces. So we can only statistically say how many copies of a repeat is there. We cannot really reconstruct the array of repeats. So Svante, it seems that you're trying to identify humans with back mutations as you called them is not a crazy idea. I mean it's absolutely gonna be probably feasible in only a matter of years if we can get all the data aggregated in a way you can look specifically. But when you, how are you imagining then taking that knowledge to sort of the next study? Are you, I mean if you had a person in front of you with some of these, I mean are you gonna study the individual, you're gonna study their cells, you are imagining just immediately going to mice but it would seem to me you'd want to have the opportunity of taking advantage of having a person. I would after all hope that one would find this in cohorts where we recall possibility. And what would you do with those individuals? Well that of course depends on what gene or what regulatory element is affected where you would get a phenotype. So you'd want to win that with some functional insights. Yes, I would imagine that, yes. And before we get to that, I would propagate that on genotyping arrays these are just 31,000 SNPs. Why don't we add them to genotyping arrays today so we can pick them up quicker actually? I'm a little troubled to understand that your statement that Neanderthals didn't exist in Africa at all. They came out of Africa 300,000 years before none of them stayed or they only developed or they radiated, they couldn't walk back to Africa. So it's my sloppy language I guess. It's the ancestors of Neanderthals that come out of Africa. So there are things that Neanderthals we call homohydral burgansis that you see both in Africa and outside Africa. And the idea is that they evolve into Neanderthals. But the morphology we use to define Neanderthals is never seen, that's something that appears in Western Eurasia. And we now have a little bit of DNA sequence from the homohydral burgansis at 400,000 years old. So we begin to be able to reconstruct this perhaps. The DNA evidence is, I can see that link being there, but the fact that we haven't found a Neanderthal in Africa, it could be found tomorrow. Yes, but I would still be surprised if one did that because yes, there has been, after all, quite a lot of things found also in Africa by now. Okay, well, we have one more question. Last question. Just quickly, the Fox P2 story is so compelling. Are there any other kind of specific gene function stories that are evolving in your lab or elsewhere? Well, so we are now trying to put in a few others of these more recent changes into mice. Three of them that we sort of bet on will have some phenotype. There are other studies in mice that have come along that has been this Idar gene that has to do with sweat glands and hair structure that's a variant among humans today that have been put into mice and gave a sort of reasonable expected phenotype and even expectation of phenotypes that they then could go out and check in the human population and find them. So I think it's not unreasonable to suggest that mice can be for some things a good model, but it has of course its limitations. Okay, well, please join me in thanking Svante for just a terrific presentation. Thank you.