Hi. My name is Tom Pollard, I'm a professor at Yale University, and I'm here to share with you some of the exciting things we've learned about the molecular mechanism of cellular motility. Here we see on the screen the first frame in a video of an amoeba. And you'll see this amoeba crawling around on the microscope slide looking for somebody to eat, and here it's found a fission yeast cell that's it's going to take in by phagocytosis. About a billion and a half years ago, single-celled organisms like this amoeba developed the machinery that allowed them to move around and capture prey, as you can see here in this movie. Today, our cells use exactly the same molecules to move about in the body. For example, white blood cells looking for bacteria that have infected the human body. Perhaps the most remarkable example of cellular motility in the human body is the movement of growth cones of nerve cells, and at the bottom you can see a series of micrographs of the tiny tip of a nerve cell that's growing toward its target. Your brain has 1 million miles of connections that have been laid down by this process of cellular motility, which you can see illustrated here in this movie. A spectacular example of cellular movements. If you look at a cell moving over time, here in this series of micrographs, you can see that this cell is moving toward the northeast. You can remove the nucleus and then most of the organelles from the cell, creating something called the cytoplast, which you can see up here at the top. The cytoplast moves along just like the whole cell, so the organelles, including the nucleus, and including most of the microtubules in the cells, are not required for this movement. The action takes place at the leading edge of the cell. You can see here in the middle in this micrograph of a cell stained for actin filaments, and you can see the bright yellow distribution of actin, which is concentrated along the leading edge of the cell. Svitkina and Borisy worked out methods of fixing cells while they were moving, and then removed the plasma membrane, and shadowed the internal contents with platinum so they could see the machinery in the electron microscope. And up above here, you can see a whole cell prepared in this way, and down below you can see a tiny region at the front end of the cell. Well, what they discovered was that the filaments at the leading edge of the cell are all part of a branched network. Deeper in the cytoplasm of the cell, back here, the branches disappear, and the filaments are replaced... the branched filaments are replaced by long unbranched filaments. The pace of the action here at the front end of the cell is pretty spectacular. This part of the cell was only a micrometer or so further to the left a few seconds before this cell was prepared for microscopy, so all of these branched filaments are formed over the time span of just a couple of seconds. Experiments by many folks, including Watanabe and Mitchison, showed that the assembly of these filaments takes place at the leading edge of the cell. The red curve at the bottom here shows the intensity of polymerization. The blue curve shows the turnover of these actin filaments. So, there's very vigorous assembly of filaments near the front end of the cell, and deeper in the cell there's some turnover. This process was studied by a strategy which most biologists had been using for the last hundred years, and that is a reductionist strategy. You could come up with a good biological question, and this biological question is, "How do cells move?", something that I got excited about when I was in high school already. Secondly, you need to find the parts, the molecular parts that make up the machinery to carry out the biological process. In this particular case, virtually all of the proteins required for cellular motility were discovered by biochemists, who were working in cold rooms and purified these molecules. Once you have the inventory of molecules in hand, then there are three big agendas that you have to carry out. One is a structural agenda, using light microscopy, electron microscopy, X-ray crystallography, and, now, molecular dynamics simulations with computers, to study the structure of the molecules. And knowing the structures are important for two reasons. One is the structures are absolutely essential to understand molecular mechanisms, but, secondly, knowing the structures, you can design much better biological experiments. Secondly, there's a biochemical agenda over here on the right. Biochemists study the interactions of the proteins and measure the rates of the reactions and the thermodynamics, coming up finally with a list of rate constants and equilibrium constants for these reactions. A desirable thing to do at that point is to make a mathematical model of the biochemical mechanism that you are studying to see if it will reproduce the results of your biochemical experiments. The third agenda is down here at the bottom, and that's the cellular agenda. At the top of my list here I have measuring the concentrations of the molecules in the cell. This is actually rarely done, but it's absolutely important because these processes are just complicated chemical and physical reactions, and the concentrations are essential to understanding the dynamics of the system. Then one has to document the dynamics of the molecules in live cells, and test the ability of simulations of your hypothesis to reproduce the behavior you observed in the live cells. Now, typically, when you do that you actually disprove your hypothesis, and I'll give you some examples of that in this talk. And, having disproved your hypothesis, then you have to... your hypothesis would be the ability of the model to carry out the biology... having disproved your hypothesis, then you have to double back and see whether you're missing some parts, or whether some of your biochemical measurements or cellular measurements are inadequate to describe the process. But, by iterating around this part of the process down here, eventually you'll home in on a mechanism that can not only explain your observations but predict the results of experiments you have never done. Now, to do this sort of mechanistic work on something like cellular motility, it's important to choose the right experimental organisms to answer mechanistic questions. And one very important tool for doing this is the ability of the organism to undergo homologous recombination, so you can introduce foreign DNA any place into the genome where you would like, so you can modify genes and tag genes in order to carry out your experiments. So, let me give you a little background about how this field got started. The original work in this field was done trying to understand the molecular mechanism of muscle contraction. In the 1940s, Albert Szent-Gyorgyi and his colleagues in Hungary discovered two important proteins. One is actin, and the other is myosin. Both of them formed filaments in muscle cells. They discovered that when they took the two purified proteins and mixed them together in the presence of ATP they could make threads that would contract, a spectacular reconstitution of the process they were trying to reproduce. In the 1950s, H.E. Huxley used electron microscopy to study the structure of skeletal muscle, and showed that there were two types of filaments. The thicker filaments are myosin filaments and the thinner filaments are the actin filaments. The myosin is the motor. It uses ATP to pull on the actin filaments, and he showed that the actin filaments slide past the myosin filaments during a contraction of muscle. In the 1960s, we did not know that actin and myosin existed outside of muscle, until these four gentlemen came along. We have here Ed Taylor, a professor at University of Chicago, with his student Mark Adelman, and Fumio Oosawa, a professor at Nagoya University with his student Shigashi Hatano. And these two groups, working independently, on a slime mold called Physarum down here, were able to isolate actin and myosin from this non-muscle cell, showing that myosin and actin were not unique to muscle, but in fact that they are likely to be universal components of eukaryotic cells. Here we can see the structure of an actin molecule. This is a ribbon diagram from the work of Kabsch and Holmes. You can see that in the center of the protein there is a small structure, which is ATP, which is clamped between the two sides of the protein. On the bottom, you can see a space filling model of the actin. Actin is an amazing protein, doing many things in biology, but one of its most remarkable features is that it's one of the top two or three most abundant proteins on the surface of the earth. And in fact, in human beings, it's one of the two or three most abundant proteins in our bodies. Probably only collagen, which is an extracellular matrix protein, is present in greater numbers. So, if one takes purified actin and simply adds physiological concentrations of salt, the actin polymerizes into filaments which you can observe by electron microscopy. On the right over here you can see a model of the actin filament. There are two strands of actin subunits that are wrapped around each other in this polymer. The filament has two different ends, because each of the subunits along the filament are oriented the same direction, perhaps as you can see in this middle diagram. In 1975, a student, Diane Woodrom, working with me, showed that the actin filaments grow at different rates at the two different ends. What Diane did was to take an actin filament and decorate it with some myosin, forming these arrowhead-shaped complexes that I hope you can see along the length of the filament. Then she added actin subunits to these decorated filaments and observed that the filament grew rapidly at this end, where the barbs on the arrowheads are located, and grew slowly at the other end, where the points of the arrowheads are located. And she decided we should call these the barbed and pointed ends. A decade later, I used electron microscopy to measure the rate constants for the addition of ATP-actin and ADP-actin to the two ends of the filaments. There are eight numbers here. These are the rate constants for the reactions, and these numbers are the foundation for all future work... have been the foundation for all future studies of actin assembly. The ATP-actin adds to the barbed end of the filament much faster than it does to the pointed end of the filament up here. Nowadays, we can actually visualize this very easily by light microscopy. Using total internal reflection microscopy, shown over here on the right, you can see a small actin filament, and in the buffer there's lot of fluorescent actin molecules as well, but you can't see them because this illumination system only looks at the surface of the slide. So, if I turn on this movie you'll see that the filament is growing rapidly at one end, up here are the top, and slowly at the other end, down here at the bottom. So, this is the barbed end up at the top and the pointed end down here at the bottom, and you can actually measure the same sort of chemical constants for these reactions by light microscopy, now. Now, I mentioned before that actin has ATP bound in the cleft between the two halves of the molecule. After the actin molecule is incorporated into the filament, then ATP is hydrolyzed. The half time for this reaction is about 2 seconds, and then the dissociation of the gamma phosphate is much slower, on the order of 6 minutes, over here. Now, once the gamma phosphate has dissociated, then the properties of the filament change, considerably, and there are different rate constants for the dissociation and association of ADP-actin than ATP-actin. And, in particular, the rate of dissociation of ADP-actin is much faster than the rate of dissociation of ATP-actin from the barbed end of the filament. Now, if you just had a tube full of actin and polymerized it with physiological salt solutions, virtually all the actin would polymerize. But, in the cell, things are controlled, and they are controlled by regulatory proteins. So, after the discovery of the actin and myosin in non-muscle cells in the 1960s, then the field spent the next few decades discovering regulatory proteins. One of the first was the protein profilin, discovered by Tilney and by Lindberg and his colleagues. A second one was the protein capping protein, discovered by Gerhard Isenberg when he was a postdoc in my lab. Now, together, the profilin and the capping protein allow the cell to maintain a large pool of unpolymerized actin. So, let me show you how that works. At the left, here, you can see a diagram showing the structure of profilin. It's a small protein with 125 amino acids, and these red and yellow residues, which you can see here in the diagram, interact with the barbed end of the actin filament. So, in our diagram, which you can see here, you can see the profilin binding to the actin and binding to the barbed of the actin subunit. When it's bound in this position it affects the actin assembly reactions. Normally, actin filaments get started by actin monomers getting together to make actin dimers and then adding another subunit to make an actin trimer, and then all other longer filaments elongate as I've described in a previous slide. When profilin is bound to the actin, it blocks the formation of dimers and trimers, so it's a very effective inhibitor of spontaneous nucleation, and the profilin also blocks the binding of actin to the pointed end of the filament. However, we discovered in the 1980s that profilin does not actually interfere with the binding of [actin] to the barbed end of the filament. So, profilin alone could not prevent the polymerization of the actin subunits, and this would not allow the cell to maintain a high concentration of unpolymerized actin. However, capping protein binds with a very high affinity to the barbed end of the actin filament, and so the combination of capping protein and profilin block all of the reactions shown here on this line of the diagram. And that allows the cell to maintain a very high concentration of unpolymerized actin, on the order of 500 times higher than you would expect if you just had actin by itself in a test tube. Another important protein, cofilin, ADF/cofilin, was discovered by Jim Bamburg. He originally called it actin depolymerizing factor, or ADF, because it seemed to depolymerize the actin filaments in his biochemical experiments. We now know that the main function of the cofilin protein is to sever actin filaments. At the lower left you can see a ribbon diagram of cofilin from fission yeast, and in this panel you can see the first frame of a time-lapse movie. This is another total internal reflection fluorescence micrograph; you can see the actin filaments here. Now, if you add cofilin to this sample, you will see that the cofilin severs the actin filaments up into small bits quite rapidly. In the 1990s, Laura Machesky, shown down here at the right, who was a student in my lab, discovered a complex of proteins which we call the Arp2/3 complex. And, a few years later, when she was a postdoc, she connected the Arp2/3 complex to a family of proteins called the Wiskott-Aldrich syndrome proteins. So, here is a gel from Laurea Machesky's original work. She purified from amoebas a complex of seven proteins. We actually didn't know what they were doing until we got some microsequence information about the top two subunits, up here. And it turned out they were actin-related proteins number 3 and number 2, which had been discovered previously in some sequencing efforts, but no one knew what they were doing. There were additional subunits. This one was related to a WD40 protein, and the other four subunits were all novel. It turned out they're all part of a stable complex of these polypeptides. A few years later, Dyche Mullins, who was a postdoc in my lab, and my friend John Heuser, got together and showed that the Arp2/3 complex would make branches on the sides of preexisting actin filaments. We can see that beautifully, now, by light microscopy. This is a mixture of purified actin, Arp2/3 complex, and a nucleation promoting factor, and what you can see in this movie, as it loops around, is that successive generations of branches are forming on the sides of the filaments. This led to what has now become an iconic diagram, which Dyche Mullins made when he was in my lab, and the idea here has been called the dendritic nucleation hypothesis. Let me take a few minutes to walk you through this. The discovery of the branching capacity of the Arp2/3 complex allowed us to put together about 25 years of biochemistry, and a reasonable pathway for how actin assembles and disassembles in the cell. And so, it was the discovery of the Arp2/3 complex and its characterization which made this possible. Up at the top, we have the plasma membrane of the cell, and over here on your left you can see a stimulus coming in through a receptor from outside the cell. That could be a 7-helix receptor or a receptor tyrosine kinase. Downstream from these receptors are signaling molecules such as Rho-family GTPases and polyphosphoinositides. These signaling molecules activate WASp/Scar proteins, turning them from an inactive state, shown in light blue, to the dark blue active state up at the top. That allows them to interact with Arp2/3 complex, shown here in green, and actin monomers, and finally, for that complex to bind to the side of a preexisting actin filament, which completes the activation of the Arp2/3 complex and allows it to nucleate the formation of a branch, such as the one you can see right here. These branches grow rapidly because of the high concentration of actin/profilin in the cytoplasm of the cell, on the order of about 200 subunits per second, or a half a micrometer per second, and as they grow they push against the inside of the plasma membrane. The filaments actually produce a lot of force. They're almost 100 percent thermodynamically efficient, and each one produces a few picoNewtons of force. And, collectively, all those filaments provide enough force to move the plasma membrane forward. Capping protein, shown here in blue, terminates the growth of these filaments within a short period of time. And so, at the leading edge of the cell, the filaments are on the order of a half a micrometer long, which means they've grown for about a second or so. Now, that may seem counterproductive, to block the growing filaments which are pushing the cell forward, but that's actually important, because if the filaments grow longer than about a micrometer the force they produce is enough to cause them to buckle, and that prevents them from pushing effectively on the inside of the plasma membrane. And so, by capping the filaments and starting, continuously, new generations of branches, this allows you to conserve the fuel, which is the population of unpolymerized actin in the cytoplasm. Now, the actin filaments, as you've heard earlier, spontaneously hydrolyze their bound ATP and dissociate the phosphate, and so this is an aging process, which is shown here by this arrow, and the yellow is ADP/Pi/Actin and the red is ADP/actin. This is preparing the actin filaments for disassembly, because the ADF/cofilin proteins will attack the ADP/actin filaments and cut them up into short pieces, which subsequently depolymerize. This releases ADP/actin into the cytoplasm, and then profilin returns and catalyzes the exchange of ADP for ATP on the actin subunits, refilling the fuel tank for this process down here at the bottom. So, that's how it works. This was Dyche's diagram in 1998, and the details have actually held up extremely well to further experimentation over the last decade and a half. Now, let me speak briefly about the formation of branches. The branch formation requires actin monomers, it requires Arp2/3 complex, it requires a nucleation promoting factor like WASp or Scar, and a preexisting actin filament. Over on the left, here, you can see the ribbon diagram of our crystal structure of Arp2/3 complex. Up here in orange is Arp3 and down below in red and gray is Arp2. The other subunits are holding these two Arps physically apart; they're barely touching each other here. And this assures that, before activation, that the Arp2/3 complex is inactive and cannot start up the filament on its own. Over here on the right, in this space filling diagram, which is turning around so you can see all sides of this big complex, you can see the Arp3 up at the top and the Arp2 down at the bottom. This shows more clearly the fact that they're not positioned correctly to start up an actin filament. So, how do the various elements get together to make a branch. One important clue about this came from studying the structure of the branch junction by electron microscopy. This is a collaboration with the Hanein and Volkmann labs in San Diego. At the top, you can see there a 3-D reconstruction of a branch junction. This is form tomography and negatively-stained specimens, and although the resolution is low, on the order of 25 Angstroms, it's good enough to position accurately the Arp2/3 complex in the branch junction. Here you can see it in colors from several different angles. Now, this structure showed that, amazingly enough, all seven subunits of Arp2/3 complex are in contact with the mother filament. It also showed that there are substantial conformational changes in both the Arp2/3 complex and in the mother filament between the time that the Arp2/3 complex and the filaments get together to make the branch. So, for example, down here at the bottom, here's the crystal structure of the inactive Arp2/3 complex. Over on the right are models showing how Arp2 has to move about 30 Angstroms in order to come into close contact with Arp3 and to become the first two subunits in the daughter filament. You can appreciate that from the reconstruction. You can see here, Arp3, right here, and Arp2. You can see them end-on right here... Arp3 and Arp2, they're the first two subunits in the branch. So, how does this all happen. We have a partial explanation for this from kinetic and thermodynamic studies and from crystallography. It turns out that there are multiple binding sites for the nucleation promoting factors on the Arp2/3 complex. We're gonna call the nucleation promoting factors VCA here, because there is a V, a C, and an A motif in all of these nucleation promoting factors. There's a high affinity site. It's actually loacted on the Arp2/3 complex, as shown here from electron microscopy, in a position where it blocks interaction of Arp2/3 complex with the mother filament. So, this is inhibitory. We don't understand the biological significance of this. A lower affinity site is located on the backside of Arp3. Here you can see form Chris Jurgenson's crystal structure, you can see the three terminal amino acids of a VCA bound to the Arp3. It turns out that binding to this site is enhanced if the Arp2/3 complex is bound to a mother filament. So, a thermodynamic, detailed balance, that means that the binding of the VCA to the Arp2/3 complex will enhance its affinity for a filament 30-fold, which surely is an important part of the process. This is a space-filling model of Arp3. Up here in this corner you can see the C-terminus of the nucleation promoting factor bound to the Arp. In the middle, here, you can see a helix which corresponds to the C region of VCA. Now, when the nucleation promoting factor is bound to this subunit of the Arp2/3 complex, it's perfectly positioned to deliver the first subunit to the daughter filament, as shown by this diagram. Here you can see the Arp3 in orange. Here is the C-terminus, the A region, here is the C region, this helix right here, and here in grey is the first subunit of the daughter filament, which is attached to the V region of VCA. So, in addition to this binding site, we believe, from the work of the Dominguez and Rosen labs, that there is a third binding site on the other side of Arp2/3 complex that delivers the second subunit to the daughter filament, shown here in brown. So, here's the dendritic nucleation hypothesis again, and you might say, "Well, that's really nice biochemistry, but is this actually what's going on in cells?" In the second part of my talk, I'll tell you how we've been able to test this hypothesis in live cells and confirm that these are the reactions that are taking place. Thank you.
