Wendy Mitchinson

Transcription

>> Sean Sanders: Our final speaker for this webinar is going to be Dr. Clare Waterman, Director of the Cell Biology and Physiology Center and Chief of the Laboratory of Cell and Tissue Morphodynamics at the National Heart, Lung, and Blood Institute. She will talk on quantitative fluorescent speckle microscopy -- a technique pioneered in her lab, which allows quantitative analysis of the dynamics of and interactions between proteins within macromolecular assemblies. Welcome, Dr. Waterman. >> Clare Waterman: Thank you for the opportunity, AAAS, to represent the Intramural Program here at the NIH. And I am going to tell you about some work over the years on fluorescent speckle microscopy that was developed in my lab in collaboration with Ted Salmon years ago, and some of our applications. So what we are interested in is the dynamic process of cell migration. And here is a movie that epitomizes the dynamics of this process. So this is from Paul Martin's lab, it is wound-healing in a zebrafish. And what you see is the fin of the zebrafish and a big wound at the bottom that has been impinged with a needle. And this wound exudes chemokines and diffusible molecules that are received by white blood cells that are flowing through the circulatory system above. And white blood cells adhere to the endothelial cell wall, squeeze through the endothelium, and undergo a directed motion towards the site of this wound. This is a very dynamic process that is mediated by the assembly and disassembly of macromolecular assemblies in specific locations, at specific times, within these cells so that they have a sense of direction and are doing this when they are asked to and not when they are not asked to. So the key here -- point that I want to make is what a dynamic process cell migration is and how can we study these dynamic processes with a light microscope. So, light microscopy is the key to studying protein dynamics in living cells in space and time because it is a non-perturbing tool and as Hari pointed out, these days with multiple colors of fluorescent proteins that have been developed and the ability to tag them onto any molecule of interest we can study the dynamics of multiple molecules making up cellular machines or cellular macromolecular ensembles, how these machines interact with each other and how these correlate with the dynamics of overall cell behavior. So, light microscopy and fluorescent protein tagging afford dynamics and specificity that allow us to understand how inanimate molecules self-assemble into dynamic structures to basically animate life. So the problem, as has been pointed out, is that light microscopy is limited in its resolution to about a quarter of the wavelength of light, or about 200 to 250 nanometers. And when these diffraction-limited image regions or point-spread functions get close together you can't resolve them anymore. And this can be pointed out in this example of the actin cytoskeleton that I am going to be using throughout this talk. So to understand the problem, we see a single -- over on the right we see a single actin filament. This is made up of globular actin proteins that are about 8 nanometers in size. These assemble into polymers that are dynamic, assembling and disassembling, and it is the assembly and motion of these polymers that animate the cell and drive cell migration So you can imagine now, you take an 8 nanometer fluorescent actin molecule, assemble a string of fluorescent actin molecules and now, when this gets assembled into a structure, this gray-scale image, in the bottom right, at a high density of filaments, what you are going to see, in the top left, is basically sort of evenly fluorescent label along the leading edge of this migrating cell. The problem is if molecules are assembling and disassembling into those actin filaments, we can't see those because basically they are below the limit of resolution of the light microscope and the image will just look evenly fluorescent, in spite of the fact that there are dynamics going on in there. So we have developed this technique of speckle microscopy that allows you to see dynamics of molecules that, at the level of the light microscope would look homogeneous if they were labeled evenly with the molecules. So, we see the full actin cytoskeleton labeled with fluorescent phalloidin. This cell is expressing a very low level of fluorescent labeled actin in a different color, and when that actin assembles into the actin cytoskeleton, you can that, in the red dots in the lower right, hypothetical fluorophore distribution. When that is imaged with the light microscope, because of the uneven assembly of the fluorescent subunits with the non-fluorescent subunits, what you end up with is a speckled image at the level of the light microscope, so that would be the zoomed FSM image and the FSM image. So what good do speckles do you? Well, the speckles -- I like to say that a speckle acts as a local probe of biochemistry and physics in a living cell. So each one of those speckles, so now what you are seeing is a speckle microscopy movie of the actin cytoskeleton at the leading edge of a migrating cell. This is just the leading edge, about 20 by 25 microns, and the time clock that you are seeing is minutes and seconds over time. So every single one of those speckles represent the change in intensity of those speckles -- so they are all sparkling -- the change in intensity represents the rate of assembly and disassembly of actin monomers into those actin filaments within that tiny diffraction-limited region. So every single one of those speckles is encoding a rate of binding and dissociation, that is biochemistry, and then the motion of those speckles encodes the trajectory, the velocity, or even the material properties of the actin cytoskeleton, and that is the physics. You can collect these images with a sensitive camera these days on the time scale of about 10 milliseconds and you can track them with techniques that Hari talked about fitting point-spread functions, or estimating the center of point-spread functions on the order of about 10 nanometers. The information content is massive. Each one of these images has about 50,000 speckles and over time there are going to be hundreds of thousands of speckles. And if you are a graduate student that was act to track these by hand, you are going to hara-kiri. So in order to overcome this problem, we collaborated with a computer vision scientist, Gaudenze Danuser, now at Harvard Medical School, who developed image analysis algorithms that will allow you to locate each one of those speckles, track the positions of those speckles individually over time, or using texture-based tracking to track the flow over time, to give you maps of these parameters, motion, speed, assembly and disassembly rates that can give you mechanistic insight into specific problems in cell biology. So, this is an example now of the computational analysis of fluorescent speckle microscopy images. We have actin speckle microscopy and co-expression of a green myosin motor protein in the same cell. Now the image below the actin image is not a co-localization of myosin and actin it is actually a map of the rates of assembly and disassembly in the actin cytoskeleton that was generated by analyzing a movie like I showed you in the last slide. And then the image next to that shows, so the red is fast assembly and green is fast disassembly. And then the heat map next to it shows the rates of motion. So you have motion of the actin cytoskeleton very close to the edge of the cell, and then a slower rate a little further back. And by characterizing this in many cells and different cell types, we come up with the notion that the actin cytoskeleton is really building two distinct machines in the cell. At the leading edge of the cell there is a structure called the lamelipodium where there is rapid assembly and disassembly juxtaposed spatially and a fast motion of the cytoskeleton. And behind that where the myosin motors are organized there is a slower convergent motion of the actin cytoskeleton that is thought to generate the pulling forces associated with cell migration. So, how do you do speckle microscopy? So really the key to speckle microscopy is in the specimen. What you need is a very well labeled functional protein. So a brightly labeled functional protein. We have experimented a little bit with multiple GFPs in a chain attached to a single protein. You don't want a non-functional fluorescent protein diffusing, contributing to fluorescent backgrounds that degrade the contrast of these speckles. You need very, very low expression level in order to get the stochastic labeling that gives rise to the speckle. On the order of about 1% fluorescent labeled, so you need very low expression level of your fluorescent molecule co-expressed with your endogenous, unlabeled protein. There needs to be a difference in fluorescence intensity between adjacent diffraction-limited image regions, so you need high resolution imaging and the stability of these differences in adjacent image regions on the time scale of an image acquisition. So a diffusing molecule which is moving very fast in the cell is not going to contribute to a speckle. But something that is immobilized in the cytoskeleton on the time scale of 15 to 100 milliseconds will contribute to the intensity in a speckle. So this is just an example of how do you get low level expression and how does it compare to the total level of protein. This is fluorescent vinculin expressed in the adhesion complexes of a migrating cell, and on the right -- shows the low level expression driven by a crippled CMV promoter -- truncated CMV promoter -- and the left shows the image of the endogenous vinculin that is imaged by immunofluorescence. So what you see is a very small fraction of the actual molecules, but because of this you can see the dynamics. So, what about the hardware requirements? Well, the hardware requirements are not any bigger of a deal than just doing really good high resolution microscopy. So, you want to prevent photobleaching, because there are very few fluorophores in each speckle, you do that with illumination shutters. You want highly efficient photon collection. A good, high-quantum efficiency camera, getting extraneous optical elements out of the light path, focus stability and a high magnification and the highest resolution possible optics. The thing about speckle microscopy is that it is really not in the microscope, it is in the specimen. So any mode of fluorescence microscopy that can satisfy these simple hardware requirements is capable of being combined with speckle microscopy. And here is just an example of Total Internal Reflection Fluorescence combined with speckle microscopy, which allows you to see very high contrast at the surface -- fluorescence at the surface of the cover slip. So if you compare the wide field versus the TIRF speckles, what you can see is higher contrast of the speckles because of the ability of TIRF microscopy to give you a higher contrast image at that specific region of the specimen. You can do multi-color speckles, labeling with distinct fluorescent proteins on different proteins to see how these interact in cells, and what we did, together with the Danuser lab, was develop a way of correlating the dynamics of two differently colored speckles to give us some idea of how these molecules may be interacting in a living cell. So in this example we have a speckled focal adhesion protein and a speckled actin cytoskeleton. We did dual-wavelength Total Internal Reflection Speckle Microscopy. We track the flow of the actin and adhesion speckles and specifically within those adhesions we get a vector field of the flow of those two, and then we -- look at the correlation of the direction of the actin movement with the direction of the focal adhesion molecules as some indication of whether these motions are coupled or not. And this is just an example of that technology where we see that the adhesion protein vinculin, the cytoskeleton protein actin, and color encoded on the right is the result of computational analysis two vector fields, with a correlated direction being encoded in red. So, vinculin has hot spots of moving in the same direction of actin across these focal adhesions indicating that these might have some sort of transient coupling between these two molecules in living cells. So where are we going with this technology? Right now, we measure rates of assembly or disassembly; we don't know absolute numbers of molecules. So absolute numbers of molecules that are being assembled and disassembled into the cytoskeleton would be a great step forward. We want other people to use this technology. Looking at other interesting structures in the cells. 3-D is a direction that would be very important. This would rely on combining speckle microscopy probably with 3-D super-resolution technologies that allow shortening the point-spread function along the z-axis and allowing that high contrast between adjacent diffraction-limited regions along the z-axis. So I would like to thank my lab members and my great collaborators over the year who have contributed to my ability to do this work that I tell you about today. >> Sean Sanders: Great, thank you so much Dr. Waterman. And thank you to all of our speakers for the excellent presentations.