Neil Shubin (U. Chicago): The Evolution of Limbs from Fins


My name is Neil Shubin from the University of Chicago, and we’re going to talk today about organogenesis in deep time. In particular, we’re going to look at this: we’re going to try to compare a fish to a human. How do you compare a fish fin to a human limb? How did the limb come about from fins? And what are the different ways we pull together different types of data? Let’s think of it this way, this is a nice starting point. Think about comparing, like I say, a human arm to a fin of a fish, shown on the left here. They look very different. Right, if you look at the bones, shown in black, there doesn’t really seem to be a whole lot of correspondence. Fish have lots of bones, we have the one bone, two bone, little bone, finger pattern seen in chickens and whales and everything with limbs. Also, fish have fine rays, which we don’t have. Big differences, yet we can bridge these gaps when we look at fossils. If we were to fill this diagram with some of the fossils, what you see here are you start to see lots of finned creatures, creatures with fin webbing, fin rays, but also having the one bone, two bone, little bone pattern as well. So what this means is, if we want to bridge the gap between fins and limbs, what we need to do is to have expeditions targeted to key parts of the tree of life. That is, we can target certain time periods to find fossils, and some of those fossils will start to bridge the gap between fins and limbs. But that’s not the only thing that’s important here. When we start to have these fossils, we can start to compare living creatures in different ways. That is, we can start to compare a human arm to the fin of a fish by seeing correspondences that would have been absent to us without the fossil evidence. What that enables us to do is to design experiments, that is, we can design new experiments based on our paleontological understandings on the developmental genetics of all kinds of different kinds of fish, non-model organisms. And in fact it works all ways. Once we have these experiments based on non-model organisms, we can begin to target new parts of the tree of life where we may be missing fossil data. So, the central idea here is that fossils enable us to bridge gaps in the record, the anatomical record of the tree of life, that enables us to design experiments in developmental genetics on living creatures, and the more we understand about developmental genetics of living creatures, the more we understand about what gaps exist in the fossil record and where we need to lead the next expeditions. So, really, the fossil and genetic data work hand in hand. So, let’s work through an example here. Well, our work obviously begins with the origin of tetrapods, the transition, say, of something like a fish on top to a limbed animal on the bottom, and we can design expeditions that bridge this gap, and what we do is we look for places in the world that have rocks of the right age, rocks of the right type, and rocks that are exposed to the surface for us to find fossils. Using that tool kit, we can begin to bridge this gap. It turns out to understand the origin of tetrapods we need to focus on environments like this, near-shore marine environments like ancient seaways, but likewise ancient highlands, so delta systems turn out to be really perfect for us, because when we have a system like this we can sample ancient seas, ancient estuaries, ancient rivers and streams, you know, the whole enchilada, as they say. So, really, it became clear very early in our study that the best places that had these kind of delta systems of the right age, in the late Devonian period, were centered in three general places in North America — this is a slide actually taken from an undergraduate college geology textbook, which helped us launch a number of expeditions — but it became very clear to us that two of these areas that were seen in this diagram were known by scientists before. We had previously worked on the so-called Catskill rocks of eastern Pennsylvania. Other colleagues had worked in East Greenland, these are very well studied rocks. If in this diagram you can see what led us to the Arctic in the first place is rocks of the right age, rocks of the right type, rocks exposed across the surface, in an area of the Arctic that was completed explored by vertebrate paleontologists. So we had ideal geology, but really very few of our colleagues had worked on these rocks. So off we went. Anytime you talk about a fossil expedition you’re talking about teamwork, and I just want to give credit where credit’s due. My graduate mentor, Ferris Jenkins, he and I have these big smiles on our faces — the reason why is something that’s in this plaster jacket, here. My good friend and colleague Ted Daeschler, shown in the upper left, he’s been a partner in these expeditions for decades. Likewise, all the field crews that we’ve had over several decades as well as the lab team as well. This is a team effort, discovering fossils. We don’t go out there alone, we go out there in teams of very talented people. So, we started these expeditions in 1999, based on this kind of map, and what you see on this map are the islands of the Canadian Arctic, and surrounded in red are where the Devonian-age rocks are exposed. And the first set of them that we did, we had to get to by these helicopters and planes because it’s pretty far away. This sort of dictates the kind of science that we can do. Using this, spending several hours on a helicopter or a plane, we got to the western part of the Canadian Arctic, shown on the arrow here. This area was, you know, ideal for exposures. What you see is a vista, a plain of Devonian-age rocks all across this landscape, here, but this was the wrong fossil environment to hold the kinds of creatures we were interested in. This was an ancient marine system, this was a system that had ancient deep-water sediments, so we weren’t finding the kind of critters we were on the hunt for, which is, say, a flat-headed fish with fins. So we had to retool a little bit, so we used the geological understandings here. This is an ancient delta system… we were in the ancient seaway… we needed to move upstream. To move upstream in the ancient geological rocks meant moving east. So, we went east, and then the next year, you can see here in 2000, where the arrow is, we went to southern Ellesmere Island. This is what it looked like; it’s a really marvelous place. Moraine with, you know, with red rocks. This contained ancient rivers and streams that held a number of lobe-finned fish critters. We homed in on a particular valley that had a layer of fossil fish that were preserved one on top of the other. These fish were very well preserved and it really wasn’t until 2004 that one of my colleagues removed a rock from this layer… Steve Gates, who is a professor at Brown University here on the left, removed a rock here and he saw a V. And he called us over and he said, “What’s this bone here?” You can barely see it in this picture, but it was beautiful, because what it is is it’s a snout of a fish, and not just any fish, it’s a snout of a flat-headed fish. And one of the big transitions is going from a conical head to a flat-headed animal. Here I had a flat-headed fish looking right at me. So we bring these things move, it turns out we found four of them this first year, they come home and the preparers begin to work on them, removing the rock grain by grain, and you could see what’s emerging here is a flat head with eyes on top. Several months later, you could see this thing exposing even more. You could see the head revealing itself. You can even see the shoulder girdle, here, revealing itself, and maybe this creature even has a neck. Remember what this quest is all about: sort of bridging this gap between lobe-finned fish and a limbed animal. Maybe finding a flat-headed fish with fins… this is what the expedition led to, a flat-headed fish with fins. Like a fish, it has scales on its back and fins with fin webbing. Like a tetrapod, it has a flat head with eyes on top, a neck, and, when we cracked open the fin, w e found bones that correspond to upper arm, forearm, even parts of a wrist. Here’s a CT scan of the fin and you can see what it has is a humerus, and then two bones here, a radius and an ulna, and shown in blue are the fin rays, so it’s a real mix of characteristics. Shown on the left, this is the work of Justin Lemberg, a graduate student in my laboratory, shows the joints of this animal. In a you see the shoulder of this animal, the socket of the shoulder on the left and the ball of the humerus on the right. This is a fish with an elbow, you can see the elbow in b, and there are even two parts of a wrist, a proximal carpus and a distal carpus. This is a fish with components of our own anatomy inside. And we can use CT scanning, as you can see in the image here… we can begin to dissect the skull using CT scanning and begin to see the individuals bones and how they suture together. It turns out that when we use living animals… this is an alligator gar shown on the bottom right, here, and the alligator gar will bite animals in the water, but as it does so the bones of the skull show cranial kinesis. They move in particular ways relative to one another, and when we analyze Tiktaalik’s skull, here, which is this fossil creature I’m showing you, we can begin to see that the joints of this animal’s skull can actually move. It has cranial kinesis, much like a living alligator gar. So, what I’m saying is when we find these fossils, the discovery is really only the beginning, because then we can start to work on their anatomy, compare them to other creatures, and begin to assess their biomechanics — how they ate, how they walked, and how they lived in these aquatic environments, in Devonian streams. So, we have this creature, Tiktaalik roseae, it’s an animal that has lungs and gills, it has fins that have components of limbs inside, it has a neck… it really has a mix of characteristics. And when we map this in the phylogenetic tree, what we see is it holds a relatively special place. That is, you can see the fish on the bottom and the limbed animals on top. Tiktaalik sits right here in the middle. It shows us the sequence of the acquisition of tetrapod characteristics, whether it’s necks, fingers, wrists, toes, and so forth. Well, how is this relevant to developmental biology? Well, remember what we’re saying is when we have fossils like Tiktaalik we can compare the arms of, say, people and chickens and mice to the fins of fish in novel ways. What creatures like Tiktaalik are showing us is that fish, back in the Devonian, had wrists. They had components of the distal, the terminal ends of the appendage, such as seen in our own limbs. So what that means is, if the fossil should be read at face value, is that sometime in the distant past, and maybe even in living fish, there should be the machinery by which limbs and toes and fingers and wrists and ankles are developed. So, let’s get back to this comparison here. If you look at a zebrafish, say, the fish on the left, and a human here on the right, you know, the bones of the fins don’t look very similar. Where the similarities start to emerge is when we compare them to the fossils, like I just showed you, but also when we compare their development. See, what we have here is a chicken limb in its development, taken from a textbook, and you can see the limb bud shown on the left. It develops this little bud, sticks out of the body, and as it develops the cartilage skeleton begins to form. Now, what’s driving the development of that cartilage skeleton are a set of interactions among signaling centers, like this region here, the AER, and another one seen at the bottom here, the ZPA, but as well as other factors, genes and proteins, that are turned on and off, driving the patterns of development and pattern formation that are so characteristic of limbs. Really, the comparison we want to make is not just between the structures of fins and limbs, but the developmental mechanisms by which the skeletal patterns of fins and limbs emerge. How similar are they and how different are they? So, to do that, we focus on a variety of different signaling systems, as well as transcription factors. One of the transcription factors that’s been incredible important to us are the Hox genes. The Hox genes have been shown to be important in a variety of processes of development from hindbrains to the axial skeleton to the limbs. To give you an example of why they’re considered so important, here’s a wild type mouse limb. If you knock out some of the Hox genes what are known as the 13 paralog groups, Hoxd13 and Hoxa13, you can develop a mouse limb that has no fingers, toes, or wrists or ankle bones. And these are segment-specific modifications of the appendage. And if you knock out elements of the 11 cognate groups, you’re missing the middle segment of the appendage. So, these are genes that are really involved with the specification of different components of our appendages. They take a very, very special role in our understanding of the origin of digits from fish fins. The question is, how have these patterns of expression and the patterns of activity of these genes evolved? Are they present in fish? Are they doing similar things in fish? What’s involved in their regulation and their activity? How is this assembled going from fish to limbed animals? Well, there have been a number of studies of the expression of these genes in diverse limbed animals, and interestingly they follow two phases of expression. Look at the limbed skeleton on the left. That has three components. It has a top component consisting of one bone, it has a central segment composed of two bones, and then it has a distal segment composed of multiple bones. It turns out there are two phases in the expression of Hox genes that are involved with the specification of these components of the appendage. The earliest phase, shown on top here, shows the different genes of the Hox system expressed within one another, so these are like nested sets of expression of one set of genes in the domain of expression of another — think of Russian dolls. This phase of expression acts early in limb development and is involved in the specification of the first two segments of the appendage. Coming on later is a late phase pattern of these same genes. This involves expression across the entire distal domain of what will become the digits and wrist of the limb, and activity of the late phase is what’s driving specification of the distal component, that component which includes the wrist and finger bones of tetrapods. It’s an open question: To what extent is the origin of the tetrapod limb based on the origin of a novel, late phase pattern of Hox expression? How did this come about? Is this something we see in fish? Is this something that comes about with tetrapods? How is it assembled over evolutionary time? If you take creatures like Tiktaalik at their word, the fossils, it would suggest that perhaps late phase expression already existed in fish fins, and maybe it’s doing something else. Let’s look at that. So, the question is really, when did late phase Hox expression come about? Did it come about… is it unique to tetrapods or is it something that we see, primitively, in fish. So, one of my graduate student, Marcus Davis, started this quest to understand patterns of activity of Hox genes in limbs and fins, and he started by looking at paddlefish, and you wonder, why paddlefish? Well, here’s a paddlefish. Paddlefish, it turns out… you can get a lot of embryos of these things and they have big, fleshy fins. As you can see in blue, here, this is the cartilage of the fin. It’s big, fleshy cartilage, so these are really relatively easy to analyze. Furthermore, these critters have a phylogenetic position that’s very relevant. They’re very basal ray-finned fish, so they’re sort of close to the branch point of creatures like Tiktaalik, so it gives us a window into that. So, in looking at Hox expression, Marcus found looking at early expression, he found they have an early phase pattern of Hox expression. If you look later on, they have a late phase pattern of Hox expression. So, it really does appear they have two phases of Hox expression, and that late phase Hox expression correlates to just like a distal strip of cells that you see in the distal terminus of the appendage. The real question here is if we look at these two phases of Hox expression… if you look at a mouse, early phase Hox expression is one [side] of the chromosome, on the telomeric phase of the chromosome, and late phase expression, that expression that’s driven in the wrists and digits and so forth, is on the centromeric side of the chromosome. So there’s a real structural organization to the enhancers and regulatory apparatus, the architecture, that drives these patterns of late and early phase activity. And this is well known from mouse from the work of Denis Duboule’s laboratory. So what we thought we would ask is, how is this pattern of regulation generated? Is it present in fish and what is it doing? And the problem is we don’t know much about fish, so we really had to assemble those data. But here’s the problem: if you look at late phase expression, the potential in some of these enhancers that are present in fish, you actually have some of the late phase enhancers, such as this one here, CsB, and it turns out if you make a reporter of the fish elements, say from fugu, and put it in a mouse reporter, you don’t get any activity in the limb. So the earliest analyses seem to suggest that late phase enhancers, regulatory apparatus, are present in fish genomes, but they’re not active in limbs, that they’re not capable of driving late phase expression. This kind of analysis suggested that late phase expression is unique to digits, unique to tetrapods, and that regulatory apparatus is there, but not functional in the same way. Well, it turns out if you look at this, it seems to be maybe we’re not relying on the right animals for comparison. So, let’s take this area here. Here is a chromosome, you can see the Hox cluster on the left, and on the right, shown in green, are early phase enhancers. If you look at a VISTA plot of this enhancer, comparing human through fish, through zebrafish and pufferfish, and you compare the similarities of these regions, what you’ll see is… you have curves on the top, the human and the chicken, which suggest they are very similar, that they have this early phase enhancer. But if you look at the zebrafish and the pufferfish, no lines whatsoever — there’s no conservation at all. So, this kind of conservation analysis would suggest that these enhancers aren’t even present in fish fins. Well, it turns out we might not be comparing the right animals, and the reason for this is that fish have a whole genome duplication. And there are three people from my lab who have been working on this particular problem: Andrew Gehrke is a graduate student, our colleague and collaborator José Luis Gómez-Skarmeta, and Tetsuya Nakamura, a postdoc in my laboratory. And they’ve been interested in this whole genome duplication as perhaps a reason for maybe why we’re not seeing these enhancers in certain kinds of fish. Look at it this way: if we look at the Hox clusters, in humans there are four Hox clusters, you can see it in the middle here… if you look at basal creatures such as Amphioxus or jawless fish, what you’ll see is there is a set up duplications that go from the single Hox cluster shown in Amphioxus to the four clusters that are shown in humans. But if we look at living fish, the ones that have been the basis for the comparisons we’ve already talked about, you can see they’re taken this duplication one step further. Zebrafish have eight of these clusters, in fact even salmon have sixteen of them. So how do you know what to compare? Maybe functions have been shuffled between them. So the idea of Andrew in the laboratory is, maybe, what if we took a fish that didn’t have this whole genome duplication, say a gar, and use that as the basis of comparison? Maybe having the right fish system would allow us to pick up these enhancers which we’re not seeing in other fish. The good news for us is working with John Postlethwait and Ingo Braasch from Oregon, the genome of a spotted gar is now available, and this was very fortunate for us, we were able to apply this genome. So, when we take the gar and put it in this comparison… you recall, what we showed before is… here is the human and the chicken shown on the baseline comparison to a mouse. You can see there’s lots of similarity. Remember, the take-home message before was zebrafish and pufferfish don’t show any similarity. When we take the gar as a unit of comparison, the story changes dramatically. Here, you have the human and the chicken, but look, the gar now shows this enhancer peak conserved, and even having the gar as an intermediate taxa enabled us to pull out small peaks for both the pufferfish and the zebrafish, which were invisible to us before. Now the question is, is this chromatin accessible at the right stage of development? Are these functioning like real enhancers? For that we used a new technique known as ATAC-seq, which shows us the accessibility of the chromatin at the right stage. We can ask the question, by looking at this, is this enhancer, CNS65, accessible? And you can see here those large peaks you see in whole-body, 24-hours post-fertilization, they show that that chromatin is accessible, functioning likely as an enhancer. Now, when we take that enhancer region and we put it in a fish with the reporter, here’s how it drives expression. It drives expression throughout the fin in 31 hours post-fertilization and then knocks out at 60 hours post-fertilization. When we take the fish element and put it in a mouse we get the same pattern. Early in mouse limb development it’s driving expression throughout the forelimb, and in late development it begins to knock out in the area that will form the distal part of the appendage. So, the fish element in mouse is functioning just as it should for an early phase enhancer. So, this is a case where having the right model organism allowed us to find an enhancer present in fish that has a conserved function with mouse. Now, the real question we’re interested in is not just these early phase telomeric enhancers; what we’re interested in is those centromeric enhancers all the way on the right, because these are the ones that are driving digit expression. So here we’re asking the question, do fish have the genetic apparatus that drives Hox activity which drives the formation of digits? And here’s the whole Hox cluster… just to make a long story short, there’s the Hox cluster itself, these are the early phase enhancers, here, on the left in yellow are the late phase enhancers. And what I just showed you is early phase enhancers are present in fish, this is what I just showed you, and you can see they report both in mouse and in gar, and in very similar ways, and they function as early phase enhancers… you’ll notice how the expression is knocked down in the distal fin. Now when we look at the late phase enhancers, indeed they are present in fish fins when you add the gar to the comparison. They report in very similar ways in mouse and in gar, you’ll see the expression activity driven by the gar element in a mouse, it’s very similar to that driven by a mouse element in a mouse. And indeed, when we look at their expression in the fin, what they do is they drive expression across the entire fin in early development, but only in the distal fin in later development, and the same is true for other late phase enhancers seen in mouse. They drive activity both endogenously in mouse, but the gar element also drives activity within the mouse, and you can see all the way on the left, here, just like a late phase enhancer should, it drives activity throughout the fin in early development and just in a distal strip of tissue in late development. What’s interesting here is when we take the gar and put it in mouse and the endogenous mouse, they are very similar. Yet the zebrafish, an animal with that duplicated genome, barely even reports in the mouse genome. So, this is a case to show, when you have the right genetic model, the right genomic model, you can see hidden similarities that would be hidden to you otherwise. The zebrafish doesn’t report easily in mouse, probably because it has that duplicated genome, whereas the gar, which has the unduplicated genome, reports very much like a mouse, it behaves very much like a mouse. So, just to give you a sense, looking at other Hox genes, I just want to show this for one simple point… when you look at the endogenous activity of these enhancers in a fin, what you’ll notice is they drive expression in a distal strip of tissue, just like fish, late phase expression of Hox genes. These same elements in mouse drive distal expression across the mouse paddle, which becomes the digits and the wrist. So, the idea here is a developmental and genomic equivalency between that distal strip of tissue you see on the right of the fish fin and the entire distal paddle of a mouse limb. So, this leads us to the evolutionary comparison, supported by Tiktaalik, fossils like Tiktaalik, and supported by the developmental biology, is that fish indeed do have wrists, and if we take the developmental data at face value, it seems like the distal region of a fish fin, which consists of those little blobs shown in yellow, on the left, correspond to the wrists of humans. So, the take-home message here is we can leverage multiple lines of data to understand evolutionary history. Look, I’m a paleontologist, I don’t find enhancers buried in rocks, but what I have is the means to compare the enhancers of living creatures that are separated by huge phylogenetic distances. The way we do it is first start with fossils to bridge the gaps and then devise experiments which help really bridge those gap s in a mechanistic way. So it’s really an exciting time for science because we can begin to analyze evolutionary transformations using both genetic and molecular data, as well as classic paleontological data. To do an analysis like this takes amazingly talented people, including the artist who drew this lovely diagram, as well as my members of my laboratory, my good colleagues, who have co-led the Tiktaalik expeditions with me, the Inuit community and Canadian government, which has supported our work for several decades, my molecular colleagues and the colleagues who have provided access to the gar genome, and of course the funders of our work. Thank you very much.

Comments 10

  • I've been a huge fan of Shubin's ever since the release of the PBS Evolution series. My advanced biology class reads Your Inner Fish, followed by a viewing of two episodes from the series. This video provides a great succinct overview for my younger students.

    I'm such a huge Shubin fan, that I named my first child Tiktaallic. Joke.

  • I'd be even more interested in the evolution from limbs to fins, in the case of water-dwelling mammals =P

  • Neil Shubin, I swear your's and everybody else's work can't be appreciated enough. Every time I think of the awe-striking and breath-taking view of the cosmos and the life within it that scientist struggle diligently to uncover, I find myself reflecting on a presentation just like this one where that discovery is showcased in all of its glory. Awesome story and don't stop exploring the natural world.

  • 9:50 Are we looking at ancent fish/tetrapods (pre-amphibians)? So, did ancient fish have skeletal structure which resemble humans, or did that resemblance only come about during the transition with Tiktaalik? If the latter, then did ancient fish branch off in at least two directions, one resulting in the line to amphibians, mammals, primates, humans, and another to…present day fish? If so, it makes our line branching to humans seem extraordinary, and the branch which leads to present day fish seem mundane. Am I nearly grasping these concepts? It's hard to pay close attention when I'm often squinting, looking away, and remarking, "Ew, gross."

  • This is my first viewing of anything by Dr. Shubin. I'm happy and impressed to see it, but although I have the biology background to digest 95% of it, I think a bit more watered down explanations are needed if this sort of strong evidence for evolutionary development is going to convince creationists. Good job, all! Keep it up.

  • I Love Neil Shubin. But i havent read any of his books. Has he written any?

  • 11:05 What textbook is that picture taken from?

  • Dear Dr. Shubin, it would be great an updated edition of your inner fish

  • fish turning into people, what a lunatic
    maybe learn basic biology what a psychopath

  • Ok, but where are the transitional fossils?

Leave a Reply

Your email address will not be published. Required fields are marked *