So with that introduction to single particle analysis, now we're going to talk about some of the special issues that arise in sample preparation for this technique. So most single particle analysis projects begin with negative staining and we covered this earlier but just a quick review to negatively stain your sample you put a small drop of concentrated purified sample onto an Grid and then you add a little bit of negative stain for instance a solution of 2% urinal acetate. And then, you remove most of the fluid with filter paper away from the grid and let the rest dry. Ideally, the metal in the stain solution will settle and precipitate as the fluid dries, will settle all around your specimen and embed it in a layer of stain. So for instance, here's a picture of this being done on the yeast Rrp44-exosome complex. And so here in between the particles it's dark. That's because the stain is present there and many electrons are being scattered away out of the beam there. And where the proteins are, these positions appear white. And that's because there's a hole in the stain of less dense material, namely the protein. And so the electrons pass on through and they go ahead and they hit the camera or the film. And so what can we learn from a negative stain experiment? While you immediately get a sense for how homogeneous your sample is. This one looks beautiful in that there's lots of particles that all look approximately the same size. You see that they lay down with different orientations on the grid, which is a good thing. You can see that they're spreading out nicely so that the images of each one is separated from the next one and the next one, and yet there's plenty of them in the field of view. And so you can get a sense that overall, your sample is going to be amenable to single particle analysis. Now, a typical target for concentration of your protein or complex is, for instance, one mg per mil. When you prepare a negatively stained grid, what you find is that the drying can increase the number of particles that persist on your grid. And so, typically, you may only need a hundredth of a mg per mil, for instance, in your sample, to produce a nicely populated, negatively stained grid like this. In addition it's also important to be aware that negative stained grids are really finicky. What you'll find is that on a whole grid surface, there'll be some areas where you might have a beautiful area, beautifully negatively stained particles where the stain is just the right thickness and you have a large number of particles. And then you might find on the other side of the grid, there's no particles at all or the staining is really bad. So you typically prepare a number of grids and hope to find a good region somewhere, and if you have even one good region, you could collect enough particle images to produce an initial reconstruction and move on with your project. Now it turns out that the condition of the grid surface is key. Given a continuous layer of carbon, the experience in the field has shown that if that grid with the carbon on it has sat at room temperature in a room for many weeks, the surface becomes highly hydrophobic, and so it's hard to get particles to spread out nicely on the grid. So glow-discharging can help overcome that. Others have found that to get the best results, if they use a freshly prepared carbon surface, where the carbon is evaporated onto a mica sheet and floated off onto water like we've discussed before, and then picked up freshly onto a grid, that those are the carbon surfaces that will attract a good number of the particles and produce good negatively stained grids. Other people have prepared that grid's surface, with anhalamine for instance. Another variable that can be played with is which negative stain is used. The most common is uranyl acetate but you can also try Phosphotungstic acid and others. So preparing a good negatively stained grid is often trial and error. You try different kinds of grid surfaces. You try different kinds of buffers. You can try different kinds of negative stain. And hopefully in the end you'll find a good condition that gives you a nice clean picture like this. Now just as the second example to show you how effective negatively stained grids can be. Here's another picture this is of the HIV integrates enzyme bound on either side with a fab complex. And so in the end the integrates enzyme plus the two fabs is a very, it's a long skinny object and this is a class average we'll talk more about them. Here's another class average. As we'll discuss later. But you can see, in the negatively stained grid image, you can clearly see lots of these objects and tell immediately from the negatively stained grid that you have a nicely homogeneous sample, and that the antibodies in this case bound well, and that you have something that you can move forward with your processing. Now, as we've discussed previously, negatively stained grids can only go so far in resolution. Because what you're actually imaging is the stain around the particle, rather than the density of the particle itself. Furthermore, when the stain dries on the grid, the particle can collapse, and can be deformed somewhat, and so to move to higher resolution and to get a structure of a more authentic particle in a native state, we move on to plunge freezing. So, we've talked about plunge freezing before and this is a beautiful image of plunge frozen GroEL particles, that I just took from Wikipedia. And here you can see the GroEL, they're barrel shaped. And so, here's the top view where you see the round barrel shape and here's the side view where you see the four layers, the four rings of GroEL. And you can see even detail of the edges as they come out seven fold symmetric around that ring, and so these are the kinds of pictures that contain secondary structure and even single residue detail that will be required to build an atomic model. Now, compared to negative stain grids, to prepare a very nice frozen, hydrated grid, you should expect to need about a hundred-fold higher concentration of your specimen. For instance, a mg per mil of protein or more, in order to get a nice distribution of the particles. Now, you'd like each particle to be slightly separated from its neighbors, because in the processing you'll pick them individually. But you also want them to be as frequent as possible or as densely packed as possible so you get more particles in each image that you record. Now some of the challenges to getting a really nice looking plunge frozen grid of a purified complex include that sometimes the proteins will preferentially move to the carbon film rather than in the holes. Usually you get the best images of the particles within the hole in the carbon film because it has reduced background. But some protein complexes seem to have a preference for the carbon. And you'll find that if you play around with the salts, and maybe add a little magnesium or something, then you can change the dynamics and push more of the particles into the whole. Other particles have been problematic because they seem to have an affinity for the fibers of the filter paper. And so when you blot the grid right before plunging it into liquid ethane, most of the sample is gone after the blotting process and, presumably, it absorbed to the fibers of the filter paper. Finally, some protein complexes are very sensitive to an air-water interface, so they find themselves in a very thin film of buffer. In the time that it takes between blotting and plunging into the ethane, each of the particles will diffuse and touch an air-water interface. And sometimes they'll denature badly when they hit the air-water interface. And we'll talk more about strategies to overcome these problems in a second. Now, notice that in a plunge-frozen sample, the background is now lighter than the proteins, because the background is a dilute buffer of approximately 1 gram per millilitre. And the protein, however, on average proteins have a density of 1.3 grams per millilitre. And so, the proteins stand out as dark against a lighter background. But before we go on, let's talk about a hybrid strategy somewhere in between negatively staining and plunge-freezing samples. It's a technique that's been called cryo-negative staining. So far, the most commonly used stain is ammonium molybdate. And the particles are mixed with the stain and then that mixture is spread into a thin film on the Grid and plunge-frozen this time, instead of being allowed to dry. And what happens is that the particles are then embedded or dissolved in an electron-dense, highly scattering buffer, and so here's a picture of GroEL particles cryo-negatively stained. And you see the background is again dark because it's full of ammonium molybdate and scattering electrons strongly. And the protein particles are cavities in that dense solution. And so they appear whiter because the electrons are passing through there. So the advantages of this procedure are first that it increases contrast. Now, the protein particles stand out stronger against the background because the background is a dense stain. Another advantage over negative staining is that the particles haven't actually dried. They're still in solution, so they're more likely to have a closer to native state. The disadvantage is, of course, that contrast is dominated by this stain, and so it suffers some of the same limitations as traditional negative staining. In that the stain is dominating the image rather than the protein densities. Cryo-negative staining can help, for instance, in discovering whether or not your sample is homogeneous. For example, here's another example image of a cryo-negative staining experiment with RNA polymerase II, which is a highly flexible dynamic molecular machine. And so, here, because of the cryo-negative staining, the increased contrast allows one to actually see more clearly the different conformational states that exist in the RNA polymerase and can help characterized particles one by one. Now, as we mentioned, one of the challenges in single particle analysis sample prep can be that particles have a large number of different confirmations. And they can also be unstable once they're concentrated and spread into a thin film next to an air water interface. And so, in some cases, the particle images don't look crisp and homogeneous. Here's an example of splicesomes, and here if you can see the individual splicesomes, the images aren't very crisp and they don't look so homogeneous. And in fact, if you do a class average, we'll talk about class averages in a second, you get a result something like this. It was because of this that Holger Stark's group invented a method that they called GraFix purification and stabilization. And the basic idea was to take a tube with a glycerol gradient going down the tube, so here's the glycerol gradient going from low concentrations of glycerol to high concentrations of glycerol. Laced within it a gradient of a chemical fixative like glutaraldehyde, so you have low concentrations at the top and high concentrations at the bottom. And so, this is all embedded in this tube. And then, you add to it your sample up at the top and perhaps a cushion and then you start to centrifuge a sample. As they move through the gradient, they're first lightly fixed and then fixed more and more completely as they move down and they separate from other populations of particles that aren't exactly like them. Then at the end of the spin, you can separate what you have in the tube according to its layer. And in the case of the splicesomes, there was a much cleaner result. Here, the GraFix treated particles. Each image is much crisper now. The edges are clearly defined in this image, and the class average now also has crisp, well-defined edges and well-defined structure, certainly compared to this. So the GraFix strategy can help hold particular complexes together through the procedure of preparing the single particle grid. And also, select for a more homogeneous population of particles. Now, speaking of how to stabilize particles, one very important class of particles that you may find yourself studying is membrane proteins. And so let's talk for a second about how to stabilize membrane proteins for cryoEM single particle analysis. So the first method, by far the most common way to stabilize and purify membrane proteins is detergent solubalization. So the idea is, suppose you have a membrane protein, here this object in yellow, and normally, it exists within a lipid bilayer, which the individual lipids are drawn here. And if you simply add detergent molecules to such a sample, the detergent molecules will replace those lipids and coat the hydrophobic surfaces of the membrane protein here. And so, these are individual detergent molecules that form some configuration around here to protect the hydrophobic surfaces of the protein and yet the hydrophilic head groups are exposed to the solution. So this is what's called detergent solubilization, and it allows you to purify membrane proteins in a native state and separate them one by one. Now, a second way to take care of membrane proteins, is to embed them in nanodiscs. Nanodiscs are hockey puck-shaped structures with lipids in the middle. And then, around those lipids are proteins that wrap around it to form a ring. And so, you can embed membrane proteins into these nanodiscs Structures. In this case, there's an entire ribosome that is embedded through the SecY protein into the nano disc. And it was this kind of a particle that was used for some very high resolution cryEM, a single particle analysis. A third idea is to embed that membrane protein into it's own liposome, and here's a picture. This is a technique we'll talk about in the next slide, also, that was developed by Fred Sigworth and his colleagues, to embed a membrane protein into a spherical liposome and then image the liposome in its entirety. And this approach is referred to as spherically constrained single particle reconstruction. This schematic illustrates that the membrane protein of interest is embedded in the spherical liposome. Then, the spherical liposome rests on the grid surface in this case. They used a two dimensional crystal here, which then sat on a lipid layer. Which then sat on a thin film of carbon. And then, here, is the grid bar supporting the entire grid. The wavy lines just mean it's a cut away from the rest of the grid behind it. And if you were to look at the density profile of just this liposome, you would see that the density due to the liposome rises quickly and then has two peaks for the two layers of the hydrophilic head groups and the expected density profile of the liposome can be calculated in great detail. And so, the grid might look like this, where here we have one, two, three liposomes of different diameters, with the protein of interest embedded within them. And a liposome of this size, one can calculate in great detail what its theoretical image would look like, so here is a calculation of the density of that liposome with a protein model embedded in certain locations. And what one can do is then subtract the predicted density due to the liposome from the original image. And obtain an image of theoretically just the proteins themselves get suspended within the buffer, as if there had been no liposome there. And so, here there are boxes around the membrane protein itself, in the positions where it was found in those liposomes. Now, as we'll talk more later, one of the critical steps in single particle analysis is to determine relative orientation of each of the proteins with respect to the others. So one of the advantages of this approach, is that if you assume that the particles are going to be pouring into perpendicular to the surface of the liposome, that constrains two out of three of the oiler angles that need to be found, and the only one left to be found is the relative rotation about that axis. The axis perpendicular to the surface of the liposome. And this is why it's called spherically constrained single particle reconstruction. But this is a third way to prepare membrane proteins for single particle analysis. Now, one of the advantages of single particle analysis is that the sample doesn't have to be perfectly purified. It can be heterogeneous and that heterogeneity can be dealt with computationally in the images. And this allows an interesting idea that's been called affinity grid capture. Now, in this approach, developed by Deborah Kelly and Tom Waltz, on the grid support is deposited some kind of surface that will have specific affinity for the target of interest. For instance, one surface can be a monolayer of lipids that are fused to a Ni-NTA moiety, and so that on the grid's surface you have a layer of lipids with Ni-NTA attached. And if one were to add on top of that surface, a solution of his-tagged proteins. So for instance, in the example shown is a 40S ribosomal subunit with one of the proteins linked to a his-tag, then the his-tags might bind the Ni-NTA and be attracted and bind to the grid's surface. And everything else can be washed out. In fact, in the extreme one can imagine licing a cell on top of an infinity grid and washing away everything out of the cell except that specific molecule of interest that has that tag that binds to the surface of the grid. This has the potential of purifying something that's much more nearly native and purifying all of the, perhaps, fragile complexes that a protein of interest is involved in inside the cell. And here's an example image of small ribosomal subunits bound to an affinity grid. And so, you can see individual protein complexes bound here. In summary, for cryoEM based single particle analysis, the sample first has to be purified and concentrated. This can be done by affinity capture, for instance, or other means. Sometimes stabilization requires detergent solubilization or some other special method to handle membrane proteins. Then once it's purified, it may need to be stabilized, for instant by light cross-linking. Then, one typically begins with a negatively stained grid. Followed by plunge freezing and imaging in the frozen hydrated state.