 Good. So we were funded in the second round of SEG's proposal, so we're one of the 14 or 15 lines here that have been funded during this program. That means I've had a chance not only to participate in the program, but to watch the development of lots of other SEG centers over the last 10 years. What I'll try to do is emphasize the science that we've done in the Stanford SEGs as an example of a SEG center and then come to some general issues at the end. So our SEG's has been focused on the genetic basis of vertebrate diversity. We all know animals differ in all sorts of interesting ways, and we've been trying to tackle really hard old questions about the genetic and the genomic basis of those differences. So we know animals differ. How many genetic changes actually produce those differences? What types of genes underlie the interesting differences you see between naturally occurring species? What types of mutations have occurred? Are there fewer lots of ways of evolving new traits? Those are obviously hard, ambitious, old questions. We decided to try to take a genetic and genomic approach to studying them. What we really wanted to do was cross different species with the idea that that would make it possible to map the genetic and the genomic basis of evolutionary change in some well-studied model system. We chose the small fish, the three-spine stickleback, because it's undergone very recent and repeated evolution. The marine fish are like salmon. They live in the ocean, but they migrate into freshwater streams and lakes to breed every spring. That migratory life cycle set off a huge evolutionary radiation at the end of the last ice age when all sorts of new lakes and streams got created. These marine fish migrated in and have since had 10,000 years or 10,000 generations to evolve a whole series of interesting new traits. So that's happened so recently that even though these fish show physical, physiological, and behavioral differences as large as you would see between different generative animals, it's evolved in the last 10,000 years and you can overcome the reproductive barriers between the different species using artificial fertilization to overcome the largely behavioral and mechanical incompatibilities between the recently evolved forms. That makes it possible to raise fertile F1 hybrids, raise large families, and treat as a genetic mapping problem. How many chromosome regions are there that control the interesting differences? What are the genes and the mutations that actually underlie the interesting traits? Now, when we first got interested in the system, Stichobacs had great biology, thousands of papers and full-length textbooks that had been written about them, but they weren't a traditional model organism. So, for example, there were no GenBank entries for three-spine Stichobacs. There were none of the kinds of resources that you take for granted in a model organism. And one of the things that our SEG Center is really focused on is building a complete genetic and genomic toolkit to carry out the genetic analysis of evolutionary change. So to do that has required combining the expertise of lots of different laboratories, including my laboratory that had a background in positional cloning of classical morphological traits in mice. We collaborated with Rick Myers and his group, originally at the Stanford Human Genome Center, with lots of expertise in high-throughput sequencing and genotyping. We also collaborated with Will Talbot, whose background is in zebrafish mapping and mutants and transgenics, and he was key for importing many of the techniques into a new fish system. And finally, one of the other co-PIs was Dmitri Petrov, whose background is in population genetics from the biology department at Stanford. We've also had, through SEG's, great collaborations with a variety of other groups, including groups with a lot of expertise in building key genomic resources like back libraries. We nominated three-spine Stichobacs for genome-wide sequencing, and the organism got picked up by the Broad Institute. We've had a very productive collaboration with Sherston Lindblad-Tow and her group at Broad, developing both a reference genome and re-sequencing lots of populations around the world. And then finally, and really importantly, we've also had numerous collaborations with labs whose background is actually in the fish, ecological genetics. They know a lot about the interesting traits that have evolved in particular populations around the world. Many of our genetic crosses got set up by building-owned fish that had long been studied by Dolph and Mike and Tom and Bjornie. Okay, as a result of all of those groups coming together, today there's a pretty good set of genetic and genomic resources for Stichobacs, including dense genetic and physical maps, genome-wide arrays and genotyping. We've developed transgenic and knockout approaches. There's a high-quality reference genome, lots of re-sequencing, and it has been possible to track multiple evolutionary traits down to particular chromosome regions and genes. One other thing I'll point out is that SEG's has a training component, and this was actually a very important part of our center. We set up a summer Stichobac molecular genetics lab course, modeled after the sort of intense laboratory courses that Cold Spring Harbor used in the founding of the research communities that work on lots of other organisms like phage and yeast. That course had the structure of bringing in 16 students in a variety of guest faculty every summer and working through the sorts of methods that are being developed by the SEG's, including hands-on micro-injection and morpholeno gene expression mapping. We had students start with phenotypes and go all the way through genotyping and QTL mapping of traits during the ten days of the course. That's been a great way to disseminate the methods being developed by the SEG's and to get a whole cohort of graduate students and post-docs and faculty members incorporating these into their own research or developing Stichobac research programs of their own. And since 2003, we've had 200 participants from 15 countries that have come through that course. Okay, so let me give you examples of the kinds of methods that we've been using to study evolutionary differences. One of the cool things about the fish is a whole range of different traits have evolved in these natural populations so the very same infrastructure can be used to look at the molecular base of a whole range of interesting characteristics. Giving an example of a skeletal trait, limb changes of course have evolved repeatedly in different animals including one of the most dramatic changes the complete loss of particular limbs like the hind limb reduction that has evolved in whales and manatees. That same trait has also evolved in Stichobacs. This is a marine Stichobac and like most fish and land animals, Stichobacs have two sets of fins, four fins or arms and pelvic fins, the equivalent of legs, that's what's shaded here. The hind fin or pelvic apparatus of the Stichobac consists of this bony spine that articulates with an underlying pelvic structure. The fish can raise and lower that spine as a defense against soft mouth predators which they do all the time in the ocean. However many of the marine fish have colonized new freshwater locations where there are no predatory fish. Instead they're shallow and they're filled with insects and the way insects prey on Stichobacs is actually to grab on to the pelvic and the dorsal spine. So under a defined set of ecological conditions it's actually an adaptive trait to lose the pelvic hind fin and that trait has evolved repeatedly in freshwater populations. So what's the genetic architecture of completely losing a vertebrate limb under a full range of fitness constraints in the wild? Well, you can cross a marine fish with a robust pelvis to a freshwater population that's completely lost to pelvis, raise large families, isolate DNA, type them with a genome-wide set of linkage markers developed by the center. What we find when we do that is pelvic reduction is not a simple Mendelian trait but it's a very approachable trait. There's a single chromosome region that controls about two-thirds of the variation pelvic size in the cross as well as a series of unlinked modifiers. Different trait, interior posterior patterning of armor plates and marine fish they cover from head to tail. Freshwater fish have evolved greater body flexibility and higher burst swimming speed. They typically only have plates here at the interior end and if you do exactly the same sort of crossing experiment, marine by fresh, turns out there's a single chromosome region that controls three-quarters of the variation and armor plate number in the cross as well as a series of unlinked modifier genes. We've also looked at some non-skeletal traits like skin pigmentation which evolves. The fish basically background color match, dark sticklebacks in dark water, light sticklebacks in clear water environments and if you cross the dark ones and the light ones, there's a single chromosome region that controls about 50% of the variation in the pigment score in particular body regions as well as unlinked modifiers. So one of the important results from the center then is that evolutionary differences that have evolved in natural populations can actually be mapped to major loci. They're not Mendelian traits, but for armor plates and for pelvic loss and for these pigmentation changes, we see major QTLs controlling half or more of the variance as well as unlinked modifiers. Well, what are those major genes that control a lot of the phenotypes that have evolved in these fish? A lot of the work that we've done has been to track those major genes down to particular molecules using positional cloning approaches, back walking, wild type marine versus freshwater sequencing and ultimately transgenic rescue. And what we found for all three of these traits is that the major genes that control armor and pelvic loss and pigmentation turn out to be key developmental regulators, either homeodomain transcription factors in the case of pelvic loss or two different signaling molecules in the case of plate patterning and pigmentation. This is the well-known stem cell factor that actually controls the formation of lots of different tissues. In fact, that's actually a theme. All three of these major developmental regulators are required for the formation of multiple tissue types. They're well-known in the mouse and human disease communities because no mutations in those genes actually cause a whole range of phenotypes because of the key role those genes play in the formation of lots of different tissues. In fact, a mouse that was simultaneously carrying no mutations in those three genes would have short hind limbs and it would be white and it would have dermal bone defects but it would also be dead about three times over. Sterile, craniofacial malformations, all sorts of severe anemia and other defects that obviously look like actually what it looks like is the standard objection that many evolutionary biologists had about the participation of major genes in the evolution of phenotypes and that is these key developmental regulators are so important that if you do something to them you're going to create deleterious problems and fitness defects. Nonetheless, these are exactly the genes that evolution is using. We have fish swimming around that have genetic variants in all three of those genes and that raises the issue of the type of mutation that's been selected repeatedly in these natural populations. So if you look at that, so for like the Heinlein reduction trait, this Pidex-1 homeodomain transcription factor, the sequence of that gene, the coding region of the gene, is identical in the marine fish and the pelvic reduced fish. However, if you look at the expression level, in the marine populations, the gene turns on in the head region in a little spot along the side of the body where the pelvic hind fin would normally develop. In the freshwater population, there's an expression in the head but you completely lose expression in that little spot along the side of the body which looks like a potential regulatory change that's altered the pattern of expression of the gene. We think then that what nature has done is to evolve new traits using key developmental regulators but by making a different kind of mutation than what is typically studied in a mouse knockout lab. The mouse labs knock out the coding region of the gene and get simultaneous defects in all the places where the gene is normally required. Thicklebacks have preserved the coding region of the gene. They preserved the information that causes expression in the head and the pituitary. They've simply lost expression at one particular anatomical site in the body. So we think that could be done by a regulatory mutation and is it actually possible to find that sort of proposed regulatory mutation? Well, we were able to use high-resolution mapping to define the candidate interval upstream of the PITX-1 gene that completely controls the macroscopic presence or absence of a pelvis in wild populations and to try to identify the postulated regulatory elements. We then used the transgenic approaches that had been developed by the Center to fuse sequences from this candidate region with GFP reporter genes, inject those reporters into fertilized thickleback eggs, and of course what we're looking for is some magic piece of DNA that will drive expression specifically at the site where the fish show changes. And they're there. The key genetic interval for pelvic reduction contains a short sequence that drives expression. I hope you can see this. It's a very specific expression pattern. It turns on specifically at the little site where the pelvis will normally develop along the side of the fish. Well, if that's really the right gene and that's really the right sequence, maybe it would be possible to do an even more ambitious experiment than that is to try to reverse evolutionary change. So for this experiment, we hooked the marine control information up not to a GFP gene, but injected that into the fertilized eggs from an evolved pelvicless population that would normally never form an external pelvic hind fin. And we were thrilled to see that the introduction of the marine information will actually put the pelvis back on the stickleback. Here's the original evolved population with the vestigial pelvic apparatus. Here's one of these transgenic fish where the introduction of the marine information has stimulated the formation of a robust pelvis of one of these serrated spines that articulates with the pelvis and the fish can raise and lower it just like a marine fish. So we really do think this is the right sequence, and then with that in hand, we were able to look at what's happened when the nice things about these fish is that the same phenotypes have evolved over and over again. What we found was that different pelvic-reduced populations are using this same mechanism over and over again. Independent pelvic-reduced fish have a few hundred to a few thousand base pairs that completely eliminate this pelvic-enhancer region. So this sort of regulatory evolution is playing out over and over again in nature. Okay, so I went through those as examples and just to come back to the kinds of questions that I mentioned at the beginning. For the genetic mapping of plates and pelvis and pigments and sticklebacks, we find it isn't Mendelian, but nonetheless it's manageable and striking that a few regions have very large effects on these evolutionary phenotypes. The genes that have these large effects turn out to be major developmental regulators that are required for the formation of lots of different tissues, and although those genes are essential for viability, nature has made regulatory mutations in those genes that produce a major alteration at a particular part in the body, but otherwise preserve the function in other places and actually confer a selective advantage that spreads through populations when the mutations occur. And finally, that mechanism works so well that when the same trait evolves over and over again, exactly the same thing, the same gene, the same regulatory elements are hit repeatedly. Okay, in the early phase of our SIG Center, we focused on these case histories and working out the techniques and technologies to get all the way from a trait down to genes. We've also been very interested in trying to collect enough examples to establish general patterns, and this has been a major theme of the recent work at the SIG Center. Here we've been using really the biological trick of the fish. They've evolved the same traits over and over again. We know from the case histories that the same genes tend to be used over and over again. And so we've carried out a genome-wide search for all of the loci in the genome that the fish are using repeatedly, using whole genome sequencing and collaboration with the Broad Institute. So the idea here is to establish a reference genome and then do comparative sequencing of multiple examples of marine and freshwater fish that have evolved over and over. Window the genome and look for those places where all of the freshwater fish are consistently different from the marine fish. And if you do that, simply hold genome sequence comparison can actually take you to exactly the same genes that we had found by positional cloning in five years of chromosome walking and transgenic rescue, et cetera. In addition to the ones we already knew about, it gives a genome-wide set of regions a whole set of areas that are being used over and over again in different populations. So the genome-wide analysis, this was published earlier this year, defines 84 different regions because it's coming from a whole bunch of different populations. Each population actually provides informative breakpoints to try to narrow the region. And the size of those intervals defined by the pattern of biological replication is relatively small, about four kilobases. And that's actually really important for us because it then makes it possible to try to address a really old, long-standing and important question, and that's whether adaptive evolution occurs primarily by coding or regulatory changes, not just in the few case histories that you've worked out, but in a genome-wide set of loci. So for those well-defined small regions that have come from the genome-wide comparative sequencing, we can put them in the categories. It turns out that 17% of the loci show consistent amino acid differences between marine and freshwater fish, but the overwhelming majority of the regions either map entirely to the intergenic areas in between coding exons, or those regions contain both intergenic and coding regions, but the only consistent differences that we see between the marine and the freshwater fish are in the non-coding regions rather than the coding regions. Regulatory contributes to stickleback evolution, but the overwhelming set is regulatory. I think it's interesting to compare those results to recent surveys that have also been identifying loci that have been subject to strong positive selection in the human genome, right? So comparison of a whole series of molecular signatures in different populations around the world has identified a set of loci that also looked like they've been subject to positive selection during recent human migrations out of Africa and to new environments around the world. Nice paper by Partis Abedi using the signals from a variety of different tests to narrow a set of 178 loci. And I went through her set and just put them in the same categories that I just illustrated for the sticklebacks. I think it's really striking that in these two genome-wide sets of adaptive loci, the fractions that you see are remarkably similar for the fraction of changes that are coding or regulatory as sticklebacks migrate out of the ocean into fresh water, or as humans migrate out of Africa into new environments around the world. And that really suggests that regulatory mutations play a really important role in creating the sort of adaptive changes that underlie the evolution of at least vertebrate species. Okay, so finally, I'll just end with some broader questions and issues about SEGs. This question already got asked, what are the best metrics for success of a SEG? And I think that's a hard question. You can list a whole series of different criteria, including did you meet your specific aim? How many papers did you publish? How many citations? How much is your data being used? We've got a stickleback web browser that is now maintained by the SEG Center. It gets about 30,000 hits a year from people that are looking up issues and doing comparative genomics. To what extent has it been possible to use the SEGs to get other grants and awards, patents and technology transfer? And then finally, I mentioned this one earlier, training and education is really one of the mandates of the SEGs program, and you can see the important role that that played in our own center. I think the mix of these metrics is going to vary depending on what the project is. I also think probably the most important criteria in the end is whether a center changed what people think can do. And we were pleased in 2005 when the progress on stickleback evolution was cited as part of the scientific breakthrough of the year for evolution and action by Science Magazine. I have to tell you, I was even more pleased when my son came home from his high school biology class and said that they had just spent the day talking about sticklebacks as an example of detailed worked out molecular basis of vertebrate evolution. And this is the kind of thing that is now in textbooks. And then finally, we've been pleased that I think one of the accomplishments of tackling and trying to build a whole set of genetic and genomic resources for a new organism is that many others have decided that it is possible to try to attack biological questions outside of the standard set of laboratory model organisms. And so there's been a lot of work developing cichlids and guppies and kayfish and kellyfish and pigeons and paramiscus and other systems. Many of whom will directly cite an inspiration coming from watching the progress of stickleback research. Just since January, there were two nice papers published in Science and Nature, one on pigeons and the evolution of morphological traits by artificial selection in humans. That paper was done by Mike Shapiro, one of the leads on the PIDX-1 work in sticklebacks who took that approach and then it went off and applied it to birds. And Hopi Hoxter's paper on behavioral evolution and paramiscus, and she also launched the paramiscus system inspired in part by watching the progress of genetic and genomics in stickleback. So size and duration of funding, I think one of the key things about the SEG Center is it's targeted at a medium scale, right, somewhere in between individual RO1s and very large sequencing centers. One SEG Center is about the size of 6 to 8 RO1s. They have a rigorous review of both the new and the established centers. I think of the first 10 SEG Centers that were funded, only half of them were renewed for the second five years. They all have a finite lifetime, all of the centers end after a maximum of one renewal. And I think to me the really unique and powerful thing about the SEG's program is particularly for an institute like NHGRI, this medium scale is very well matched to genome-wide initiatives. Genome science is usually bigger than RO1 scale science, and I think you can see at least in the story I've told you today that the development of a whole set of genetic and genomic and transgenic resources for a new organism is something we would never would have been able to do as a RO1 scale science project. But by bringing together a lot of different people with different expertise in genomics and ecology and population genetics, it has been possible to do that. The other key feature is that the proposals are bottom up rather than top down, right? NHGRI never issued an RFA for new proposals in stickleback genetics and genomics. This is a program that was available at this medium scale but solicited ideas from the community. And I think that this is one of the only places, not only in NHGRI but across the rest of NIH, where investigator-initiated interdisciplinary genome-scale research is currently funded. And I think that's one of the real achievements of the SEGs program. Okay, so I will stop there. Dede is also here to talk about another SEGs center, and Jeff, we can either take questions individually or together at the end. Particularly if you have questions about the science, why don't we ask those now? And then if there are questions about the program, if you're opposed to Dede and David, we can do that together with them up there at the end. Yeah. Are there any features in and around these regulatory elements that would make a higher chance of recombination occurring and having them deleted? That's the first question. The second question is that if that's the mutational mechanism, then that would mean that the evolution is unidirectional. Once you lose that, I don't see. You can pick it up from someplace else that you wouldn't be able to put that back in. Yep. Do you want to comment on that? Sure. The first question is something we're quite interested in because that mutation spectrum is unusual. It is deletions over and over again, and they don't actually have the same endpoint. So there are staggered ends in different populations. There's a variety of molecular signatures that suggest the area may actually be fragile. It shares some sequence relationships to the things that have been found in fragile chromosomes in humans. We've actually now cloned the region into yeast artificial chromosome clones to major breakage rates, and the marine sequence is fragile, and the mutation process eliminates the fragility. So if you look at the alleles that have spun out in these freshwater populations, they've eliminated some feature that makes it prone to breakage. Despite that, when the deletions occur, the chromosome on which they occur then shows molecular signatures of positive selection. So we don't think the region is just decaying. We think that this area is prone to throw off adaptive alleles that are then subject to positive selection in these populations. And you're right, once you've thrown the thing away, that should be an irreversible change if it's only present in one copy in the initial populations, and that's actually something we're also starting to address now because of some other features which suggest there may be a lot of variability in the region. What happens when you normalize the target size of coding region versus non-coding? I mean, you're saying it's 89 versus, what was it, you know, as much smaller percent for coding, but the target size of the two is grossly different. It depends on how you count the target size. If the target size for the regulatory elements is surrogate, it would be the accumulated conserved sequences that are found outside of the coding exons. That would probably be, you know, roughly equal target size because it's only a small fraction of the genome that is either coding or conserved non-coding. And nonetheless, we see that we see the strong bias. You know, I will say that when we started this project, we did not have an axe to grind on the coding versus regulatory issue. The whole philosophy here was, let the fish tell us what's been important or, in the case of scale genome studies in humans, it's let the chips fall where they may for whatever the patterns of variation are in different human populations. I do think it's striking that when you compare those distributions, the loci that look like they underlie adaptive change have a huge fraction of regulatory change. David, can you say a few things about how the sort of the goals and sort of the strategic view of your SEGs changed over the 10 years of its lifespan? Because my memory is right. When you started, and I think I was actually there when this first grant was started, I remember it had a slightly different flavor and it really did evolve. I'm saying this in a good way. Absolutely. Because maybe you could also frame the answer as to the flexibility of the program or how the program of all allowed you to change over that decade. So I think, Eric, that's a great question. When this segment was originally funded, it was well-tailed but as the PI, sticklebacks as one of the specific aims, but the other key specific aims was looking at issues of gene duplication and sub-functionalization using zebrafish. And although we worked on that as part of the first five years of the SEGs, we felt at the end of the first five years that the stickleback stuff was going great and in fact in many ways better than the biggest skeptics imagined. And the zebrafish stuff was frankly going worse than we had originally hoped with the morpholino approach, which was the primary way of trying to parse out the functions of the different gene duplicates actually often revealing morpholino artifacts instead of the subtle differences between sub-functionalized genes. So at the five-year point, we decided for the second phase to focus entirely on stickleback evolution, phase the zebrafish out. And the one other thing that we added and I didn't have time to talk about today was then trying to apply the principles found from sticklebacks to other organisms. So the new aim that we also picked up in the second phase was trying to look for the sorts of patterns we see in sticklebacks in genes that underlie evolutionary change in other mammals. So I think we've been flexible. When we originally put the proposal together, we knew that some people would see the sticklebacks as very, very risky and it might not work at all. For those people, zebrafish was there as something that will probably give data for sure. As it turned out, the most exciting science we thought came from the part that was the riskiest and we did have the flexibility to adjust those priorities as the program went on. Can I ask one more question? Susan Rosenberg has published a lot on increased mutation rate in E. coli in response to a variety of external environmental stresses. And I'm wondering whether there's a relationship between shift in electrolytes and in sodium content of the external milieu on recombination and deletion. So that's a great question and again, something we're quite interested in because I'm intrigued by the possibility that the very act of migration out of the ocean environment, which is both buffered in temperature and salt levels into freshwater environments that have much more variable temperatures and a whole range of different salinities might actually influence the mutations that underlie some of the adaptations that are seen in freshwater. And that's one of the reasons we're working hard now to monitor the mutation mechanisms that underlie the deletion of this regulatory element because we'd also like to see whether it's sensitive to environmental conditions. I had a similar thought that I don't know much about how transposed on sequences are organized in the genome but certainly in some organisms you have those kinds of stress responses, either temperature or even salt, et cetera. You can activate movement and I'm wondering maybe that would be too high a frequency for those kinds of events that were required for this adaptive evolution. But I wonder about the regulatory sequences being related to transposed on sequences, whether they're new or ancient. I think that's an interesting possibility. There's actually lots of proposals in the evolutionary literature about the kinds of molecular mechanisms that might underlie adaptive change, including transposons and repeats and microRNAs and gene duplicates and all sorts of things. I think we're just at the beginning of being able to address those because of the case histories and the genome-wide set of loci. We have not yet seen that proposed mechanism where transposons are popping in as the mutational mechanism that's generated something in freshwater, although there's one trait that someone had just started to work on the lab that has a little bit of that smell right now. So we'll see.