Virus Structures
Transcript of Part 3: Non-Enveloped Virus Entry
00:00:03.22 Hello, I'm Stephen Harrison of Harvard Medical School, 00:00:07.26 Children's Hospital, and the Howard Hughes Medical 00:00:10.09 Institute. Welcome to Part 3 of this series of lectures on 00:00:15.13 virus structure. In this part, we'll talk about the structure of 00:00:20.07 a non-enveloped virus particle and its implications for the 00:00:26.12 mechanism by which this sort of virus particle gets into 00:00:29.26 cells. As you'll recall, viruses come in two major flavors: 00:00:37.27 enveloped viruses with lipid bilayers, like influenza virus 00:00:42.01 that was the subject of Part 2; and non-enveloped 00:00:47.05 viruses, viruses that have a tightly fitting protein coat to 00:00:52.29 protect the nucleic acid, but no lipid bilayer, such as 00:00:58.15 rotovirus, which we'll be talking about largely today. 00:01:02.21 Rotoviruses are the cause of childhood diarrhea, they're a 00:01:08.04 virus that grows in the small intestine, and is the major 00:01:12.05 source of infantile dehydrating diarrhea, which is 00:01:17.12 particularly serious in developing countries. There's a 00:01:20.29 recent introduction of a vaccine that may ameliorate the 00:01:26.09 spread of this virus, which has been, in recent years, 00:01:29.09 responsible for as many as half a million childhood deaths 00:01:34.24 each year. The virus particle is shown here in an electron 00:01:43.27 micrograph. You'll notice that the contrast appears to be 00:01:50.19 much less than in the micrograph that I show of influenza 00:01:55.18 virus because this is a micrograph taken with a 00:02:02.05 cryopreserved specimen with no stain, rather than a 00:02:05.19 micrograph contrasted with negative stain. We're going to 00:02:09.23 be talking a little bit in the course of looking at the 00:02:13.15 structure of rotovirus about how one uses electron 00:02:18.13 cryomicroscopy to get three-dimensional structures of 00:02:23.24 large macromolecular assemblies such as this one. I also 00:02:28.17 point out on this slide that enveloped viruses enter cells, 00:02:36.11 penetrate cells, by membrane fusion, as we discussed in 00:02:39.22 great detail in the last part, and non-enveloped viruses 00:02:46.01 need to get in by some sort of perforation process, since 00:02:50.07 they don't have a membrane of their own. We want to 00:02:53.29 talk a little bit about what we know about the mechanism 00:02:56.26 of this process in the second half of today's talk. So 00:03:02.15 here's an introduction to the rotovirus particle. It's 00:03:05.06 sometimes called a triple-layered particle because it has 00:03:08.14 three protein layers, an inner blue layer, an intermediate 00:03:13.10 green layer, and an outer yellow and red layer, composed 00:03:18.29 of proteins known as viral protein 2, rather unimaginative 00:03:23.02 nomenclature, VP2, viral protein 6 (the green one), and 00:03:28.18 viral protein 7 and 4 (the yellow and red). The red protein 00:03:33.20 is cleaved in a step that we're going to talk about a little 00:03:36.28 bit, to two fragments known as VP8 and VP5, and those 00:03:43.09 of you who have followed the previous part may begin to 00:03:48.22 see similarities as we go forward between this sort of 00:03:52.20 cleavage, and the kind of cleavage that activates the 00:03:56.06 viral fusion proteins, such as flu hemagglutinin. The 00:04:00.27 particle, by the way, is about 800 angstroms in diameter, 00:04:06.24 so it's almost as large as the 1000 angstrom diameter 00:04:11.25 influenza virus particle. It packages a double-strand RNA 00:04:16.10 genome. In this animation, I show you that the virus 00:04:22.04 particle, as it originally leaves a cell, does not have these 00:04:29.13 extended spikes, but a cleavage, that cleavage from VP4 00:04:35.10 to VP5 and 8, erects the spikes. The spike protein is quite 00:04:40.26 an unusual structure, we're going to talk about it. It's 00:04:43.26 anchored in the inner layer by the yellow outer layer 00:04:50.07 protein. It's the job of the outer layer to deliver the inner 00:05:01.02 particle into the cytoplasm. The inner particle never 00:05:05.15 uncoats, it has a polymerase and a capping enzyme. 00:05:11.11 There are 11 segments of double-strand RNA wound 00:05:15.15 inside, and the polymerase can transcribe that RNA, the 00:05:22.20 capping enzyme can cap it, and the message is extruded 00:05:26.15 from this so-called double-layer particle, or "DLP." Now 00:05:33.00 this animation is not pure fantasy, it's based on detailed 00:05:36.28 structural data: x-ray crystal structures of various proteins 00:05:42.05 and their fragments; an x-ray crystal structure of the intact 00:05:46.00 DLP; and, as we'll talk about in a little bit more detail, a 00:05:53.01 three-dimensional image reconstruction, or several 00:05:55.05 different three-dimensional image reconstructions, from 00:05:58.00 electron cryomicroscopy, or "cryoEM" for short. This slide 00:06:08.24 shows you that recent advances in electron 00:06:13.12 cryomicroscopy mean that we can now obtain density maps 00:06:20.12 representing the structure with essentially the same detail 00:06:25.20 that we've been able to get from x-ray crystallography of 00:06:28.16 large assemblies hitherto. And so, here's a comparison of 00:06:37.01 the x-ray crystal structure, or the density map obtained 00:06:41.07 from that analysis for the double-layered particle, one 00:06:45.02 particular little bit of it, and the similar density map from 00:06:51.27 cryoEM. This was done in collaboration with a colleague 00:06:56.12 at Brandeis named Niko Grigorieff, and these two 00:07:01.08 members, Xing Zhang and Ethan Settembre, of our 00:07:05.20 laboratories. Now, the process by which this sort of 00:07:13.10 analysis is carried out depends on being able to suspend 00:07:20.13 the particles you wish to analyze in a very thin film of 00:07:25.21 vitreous ice, ice or a solution suspension frozen so rapidly 00:07:32.10 that the ice doesn't form crystalline ice, and hence 00:07:34.29 doesn't expand in volume and distort the solute. In the 00:07:42.05 electron microscope, one is seeing a projection of each 00:07:46.14 particle, and by combining data from literally thousands of 00:07:53.09 such images, it's possible to obtain the sort of 00:07:57.29 reconstructed view that I showed you. The single 00:08:01.05 particles are randomly oriented in the vitreous ice, so one 00:08:04.19 has ever conceivable view, and mathematical algorithms 00:08:07.24 had been worked out (you might be familiar with some of 00:08:11.05 them from CAT scans) to determine the relative orientation 00:08:17.28 of all those views, and to combine the data into a three- 00:08:26.27 dimensional picture of the object in question. This 00:08:29.18 obviously depends on the fact that all of the particles are 00:08:33.13 identical. And so cryoEM images of biological structures 00:08:38.12 are possible when those particles are very uniform. But as 00:08:42.16 I've emphasized, the images themselves are very noisy 00:08:47.03 because of the very low electron dose that's required to 00:08:50.12 avoid specimen damage. If you were to try to get a less 00:08:55.02 noisy image, one with higher signal to noise, then you'd fry 00:08:59.24 the specimen. And so, you depend on the fact that the 00:09:04.03 particles are very uniform, and in the case of virus 00:09:07.08 particles, the additional huge advantage of their high 00:09:10.05 symmetry, to make it possible to reach the molecular 00:09:13.05 resolution that I've suggested. So what we're going to 00:09:18.29 focus on for today are the outer layer proteins and their 00:09:24.15 role in delivering the double-layered particle into the cell. 00:09:29.19 Remember that the outer layer proteins are VP4 (which 00:09:33.12 gets cleaved when this erected conformation is 00:09:38.25 established, to VP8 and VP5) and the yellow protein, as I 00:09:43.29 called it, VP7, which locks everything in place. Now VP4, 00:09:51.11 the spike protein, is actually, as you probably noticed, a 00:09:55.06 very curious structure indeed. I've colored the VP8 part in 00:10:04.25 magenta and the three different (it's a trimer, as you'll see 00:10:08.29 in a minute) VP5 parts in various shades of sort of red and 00:10:15.01 orange. And of course the first thing you probably noticed, 00:10:19.22 even with the first image I showed you, was this thing 00:10:23.07 looks like a dimer. There are two lobes sticking up and 00:10:27.27 two ears sticking off of them, and yet I've just said to you 00:10:31.11 it's a trimer, what's going on? Well, turns out that this is an 00:10:35.01 extremely unusual sort of asymmetric arrangement of three 00:10:41.09 proteins. The bottom part, we call it the foot, as you'll see, 00:10:47.06 is perfectly trimeric. The outer projecting spike has a very 00:10:56.08 nice twofold axis. And one is adapted to the other, by this 00:11:03.26 sort of diagonal cantilever. So if you now look, you'll see 00:11:09.14 that the beginning of each protein subunit (remember that 00:11:16.04 VP8 is magenta, and I'm about to show you it's the N- 00:11:20.24 terminal part) is down here, and there are three of them. 00:11:26.21 Polypeptide chain comes up, one of them "quits," and the 00:11:30.10 other two come on up and form these ears. Polypeptide 00:11:34.17 chain then resumes (the cleavage is between VP8 and 00:11:39.13 VP5) as this bean-shaped domain of VP5, and continues 00:11:45.18 on back down. The third bean-shaped domain is here, 00:11:50.27 and so we've lost one ear, it's almost certainly been 00:11:57.02 cleaved by a cleavage here and a cleavage here, as I'll 00:12:00.13 explain in a minute. And that third bean-shaped domain 00:12:05.04 forms this diagonal cantilever that supports the twofold 00:12:10.14 clustered spike of the remaining two. Now already this is a 00:12:16.05 pretty unusual conformation, but I'm about to show you 00:12:21.15 some other gyrations that this protein appears to be able 00:12:25.18 to go through. So as I said, the tryptic cleavage that 00:12:31.09 separates VP8 and VP5... and that probably occurs in the 00:12:37.25 gut as the virus emerges by lysis of the intestinal cells that 00:12:43.17 it has infected, or emerges by some other secretory 00:12:51.27 method, the actual emergence is a bit unclear. The tryptic 00:13:02.00 cleavage then allows the rearrangement of this spike. 00:13:07.27 The three protein subunits on the outer parts of them are 00:13:12.26 probably more disordered if the protein has not been 00:13:20.22 cleaved, and if you look in the electron microscope, you 00:13:23.13 don't see any ordered parts of the outer assembly. And 00:13:28.01 since I've said you have to average many images in order 00:13:31.17 to get a decent three-dimensional representation, then if 00:13:38.23 the outer part is disordered, the averaging will basically 00:13:42.05 blur out all of those elements, and all you'll see is the 00:13:47.23 ordered foot. So if we compare now these spike regions, 00:13:55.08 which in the electron micrographs, or in the three- 00:13:57.15 dimensional reconstruction from electron micrographs, are 00:14:00.21 not quite as well ordered as the parts I showed you, 00:14:03.23 because of some slight flexibility as these structures stick 00:14:10.01 out from the virus. But we can sit them very well with 00:14:13.29 known x-ray structures. If we compare them, see that an 00:14:17.05 x-ray structure of the beam-shaped domain shows a nice 00:14:24.00 dimer like this. It has some hydrophobic loops at the tips 00:14:28.15 that we're going to talk about, and then a separate x-ray 00:14:31.12 structure of this region yields a lectin-like domain that 00:14:39.16 binds sialic acid, which is a receptor for rhesus rotovirus, 00:14:45.07 from which these proteins in our experiments were 00:14:49.00 derived. Now, we therefore believe that the protein, 00:14:58.08 which is synthesized as a monomer, combines with the 00:15:03.18 double-layered particle, three of them for each of the 00:15:07.21 positions on which it sits (you may have noticed that there 00:15:10.04 were 60 spikes sticking out, corresponding to the 00:15:15.20 icosahedral symmetry), and inserts and is locked in by 00:15:21.28 VP7, but is flexible until tryptic cleavage occurs, in which 00:15:28.12 case, one of the VP8s is excised, and the rest of the 00:15:35.15 structure reorganizes to the unusual-looking spike that 00:15:43.05 I've shown you. Now, as if this weren't odd enough, if you 00:15:50.14 make a piece of the protein that's a bit longer than the 00:15:54.24 one that gave that dimer I showed you, you get this 00:15:59.20 trimeric structure, in which a segment of the protein that 00:16:06.09 was missing from the dimeric construct is present, and it 00:16:10.07 has formed a beautiful, trimeric, coiled coil, and the 00:16:15.28 hydrophobic loops of the bean-shaped domain are now 00:16:18.25 pointing, if you wish, down rather than up. In other words, 00:16:23.19 it's as if the structure has gone from this sort of 00:16:28.17 arrangement to this one, and those of you who were 00:16:34.08 following some of the conformational changes of 00:16:39.04 enveloped virus fusion proteins may find that familiar, 00:16:43.22 since it's essentially the same kind of conformational 00:16:48.04 change we've seen there. And just anticipating, we 00:16:51.29 believe that indeed it is that conformational change that, 00:16:55.22 in this case, drives not fusion (there's no membrane on 00:16:58.19 the virus) but drives the membrane disruption that will get 00:17:04.29 the double-layer particle into the cytoplasm. And so, our 00:17:11.24 scheme based on this information so far is that this spike- 00:17:17.27 like structure (and the tryptic cleavage is essential for viral 00:17:22.06 infectivity)... this spike-like structure attaches through 00:17:28.06 sialic acid, some suitable trigger (and we actually don't yet 00:17:34.17 know what that is) allows the two VP8s that are attached 00:17:44.06 to separate enough that these bean-shaped domains can 00:17:50.14 insert or interact with the target cell membrane through 00:17:56.00 those hydrophobic loops, and then this umbrella-like 00:18:01.18 folding back leads, in a way that we hope to understand 00:18:05.19 but don't fully yet, to a disruption event that can allow the 00:18:11.02 double-layer particle to translocate into the cytosol. Now, 00:18:17.27 this list basically tells you what we're thinking. An 00:18:23.21 extended intermediate forms, hydrophobic loops contact 00:18:27.08 the membrane, the protein folds back to the umbrella-like 00:18:30.10 conformation, and there's a coupling of the fold back and 00:18:34.08 the membrane, to perforate the bilayer. To test these 00:18:40.05 notions, we can take advantage of an extremely 00:18:43.22 interesting property of double-strand RNA viruses like 00:18:47.08 rotovirus, which I like to call functional recoating. If you'd 00:18:52.19 like to study the entry of a virus by genetic methods, by 00:18:59.11 trying to making mutations that would impair entry, you 00:19:03.03 have a problem on your hands, because how are you 00:19:07.09 going to get enough of an entry-incompetent virus to 00:19:10.25 study in the first place? You can get around that problem 00:19:15.22 with these viruses in quite a clever way. If you take 00:19:20.16 double-layered particles prepared by removing the outer 00:19:25.12 shell, VP7 and VP4 (or VP8 plus VP5), from infectious 00:19:34.26 virus particles, the double-layer particle's no longer 00:19:38.21 infectious because it can't get in. But if you recoat it with 00:19:45.02 recombinant proteins, recombinant VP4 and recombinant 00:19:50.04 VP7, and then treat with trypsin to cleave and activate 00:19:54.27 the VP4, you get perfectly infectious particles. This is a 00:20:00.07 process worked out (mimicking some experiments 00:20:06.26 originally done by Kartik Chandran and Max Nibert on 00:20:10.10 rheovirus) by Shane Trask and Phil Dormitzer. As a result, 00:20:16.05 we can use this trick to do two things: First, we can 00:20:19.13 mutate the hydrophobic loops and ask does the virus get 00:20:23.08 in or infect, and if it doesn't, what sort of properties does it 00:20:28.04 have? And second, as you'll see, we can use this 00:20:33.28 scheme actually to watch virus particles getting into a 00:20:37.07 cell. And so, some experiments carried out by Irene Kim, a 00:20:42.11 graduate student, showed that if you mutate the 00:20:45.12 hydrophobic residues, you lose infectivity, the virus 00:20:50.28 becomes engulfed in the cell, it's taken up by a process 00:20:55.07 I'm about to show you, but never infects. And you can 00:20:59.22 also show that is never disrupts a membrane, because 00:21:03.06 the virus can allow a toxin called α-sarcin to get into the 00:21:09.00 cell, sort of sweeps it in, so to speak, if it is actually 00:21:13.26 infectious, if it actually perforate some sort of endosomal 00:21:18.21 membrane, but the hydrophobic loop mutations not only 00:21:26.21 block infectivity, but block the capacity of the virus to 00:21:30.17 sweep this toxin into the cell along with it. And so, those 00:21:36.11 experiments are strongly consistent with the view that the 00:21:47.16 hydrophobic loops matter, but in order to understand more 00:21:52.12 of this process, we've really got to understand more about 00:21:56.02 the compartment that the virus arrives in when taken up 00:22:00.28 in a cell, about the kinetics of those events, and about 00:22:06.00 other characteristics, so that we can try to figure out how 00:22:10.03 to design experiments to look at these subsequent steps. 00:22:15.27 In order to describe the experiments that we've devised to 00:22:23.24 try to do this, I should mention the outer shell protein VP7, 00:22:28.03 the other partner here, which, as you see from these 00:22:32.01 images and from the animation early in this lecture, locks 00:22:40.08 the spike protein, which we believe to be the membrane 00:22:44.12 perforator, so to speak, into the particle. VP7 is a trimeric 00:22:50.11 protein, and it's held together by calcium, there are two 00:22:53.15 calcium ions that the x-ray crystal structure showed us 00:22:59.23 hold the protein subunits together. There are negatively 00:23:06.15 charged residues interacting with those calcium ions so 00:23:11.08 that the ligation of calcium is critical for the stability of this 00:23:15.20 trimer. As a result, we believe that some sort of calcium 00:23:20.27 withdrawal may be part of a triggering signal, but we have 00:23:26.10 as yet to define that more precisely. So, in order to try to 00:23:34.24 understand what's going on, we've taken advantage of 00:23:39.01 the recoating experiment in a different way. We found, or 00:23:45.23 Aliaa Abdelhakim in the laboratory has found, that one 00:23:50.27 can label with a fluorescent dye each of the components 00:23:58.15 of the particle before recoating, that is, the double-layered 00:24:03.17 particle, VP7, and VP4, with distinct fluorescent dyes, and 00:24:09.17 therefore follow not only the binding and entry of the 00:24:17.04 particle into the cell, but can detect the point at which the 00:24:21.11 double-layer particle is released into the cytosol, as I'll 00:24:25.02 show you. So here's an experiment with VP7 pseudo- 00:24:31.10 colored in red, the fluorophore on VP7, and the double- 00:24:35.14 layer particle in green. The experiment is done by looking 00:24:40.10 at the very thin edge of an epithelial cell in culture, so that 00:24:45.06 you can look at the top surface without much aberration, 00:24:48.12 and as you'll see, the particle in the circle will suddenly 00:24:53.27 release and fly around in the cytoplasm. If you look 00:24:59.27 closely, you can probably see a trace of red remaining in 00:25:04.01 the circle. This experiment in which the three components 00:25:12.11 had been labeled separately is a little easier to follow. I'll 00:25:16.23 go back and show you that this particle suddenly 00:25:22.29 releases, and if you look at the circles, you'll see that 00:25:26.13 there's still some purple, that is, VP4 and VP7 (or rather, 00:25:32.02 VP5/VP8 plus VP7), at the site of entry, and we stopped 00:25:41.01 the frame as the double-layer particle was diffusing 00:25:44.04 around rapidly in the cell after release. A summary of what 00:25:49.25 we think these images are telling us might be roughly as 00:25:53.23 follows. One has the particle that is activated by trypsin, 00:25:58.11 that is probably even before it emerges from one host and 00:26:04.13 enters another. This is a fecal-oral transmission virus and 00:26:09.01 probably is already exposed to trypsin in the gut of the 00:26:14.02 initial infected individual. It binds to the surface of the cell 00:26:20.08 it's going to infect, in the case of the virus infecting a 00:26:27.00 person or a monkey, since this is a rhesus rotovirus, then 00:26:32.13 it will bind to a small intestinal epithelial cell. It's taken up 00:26:38.15 by a process I'm going to tell you a little bit more about, 00:26:43.09 that does not depend on familiar endocytic routes like 00:26:47.12 clathrin-mediated uptake. Something triggers the 00:26:52.26 conformational changes we've seen, so the membrane 00:26:55.19 perforates, the particle escapes and can start to make 00:27:00.10 RNA. And so, what can we learn from this sort of 00:27:08.04 experiment about this sequence of events? We're just in 00:27:13.21 the process of being able to try to do that, these 00:27:16.09 experiments aren't even yet published, but I think that 00:27:22.29 what I'd like to end on then is a series of questions that I 00:27:27.27 think this kind of experiment can answer. And I hope that 00:27:32.21 they will illustrate to you that, by combining structural data 00:27:37.09 with the kinds of, in this case, single virus particle imaging 00:27:41.27 on living cells that takes advantage of contemporary 00:27:46.20 fluorescence microscopy, we can begin to answer these 00:27:51.05 sorts of questions. So the first question is about the 00:27:55.03 engulfment. Is it simply what I like to call an 00:27:58.14 "autoendosome," that is, does the virus just wrap itself up 00:28:01.22 in membrane? There are various reasons why we think 00:28:04.02 that's at least a reasonable possibility. One is that the 00:28:07.22 process is quite rapid and that it doesn't depend on any of 00:28:11.06 the known cellular pathways. Within a minute or two, the 00:28:15.15 virus particle is sequestered from agents like EDTA, 00:28:23.15 which would take it apart by pulling calcium off, or 00:28:27.11 antibodies that might bind to it. A second question is 00:28:33.14 about the membrane disruption that then occurs. Is the 00:28:38.04 membrane disruption by VP5, this folding back that I 00:28:43.25 showed you, purely mechanochemical, if you wish, as in 00:28:47.26 fusion, or are there other more, in you wish, detergent-like 00:28:54.18 qualities of that interaction? Does it indeed do what I 00:29:02.02 pictured in the previous slide, is there an extended 00:29:05.02 intermediate? Does it remain anchored in the particle 00:29:08.22 when it undergoes this whole process? What triggers it, 00:29:13.09 and when does VP7 come off? But I think you can 00:29:16.12 imagine we can begin by combining the powers of 00:29:22.14 structure-based mutation and design of constrained or 00:29:29.21 altered proteins with recoating and with the direct 00:29:34.14 visualization, particle by particle, of this entry process, to 00:29:38.15 answer some of these questions. And so in conclusion, let 00:29:43.01 me simply acknowledge the large number of people who 00:29:46.09 have participated in this work. I should acknowledge both 00:29:52.17 people in my own laboratory, as well as in collaborating 00:29:55.14 laboratories. Let me just name the specific collaborators 00:30:00.02 with whom the work has been carried out: Phil Dormitzer, 00:30:04.22 who began as a member of our laboratory, and I then 00:30:07.16 have collaborated with him extensively since he 00:30:12.10 established an independent research program; Niko 00:30:16.05 Grigorieff at Brandeis, one of the pioneers of the sorts of 00:30:21.01 cryoEM analysis that I described briefly to you; Tom 00:30:26.00 Kirchhausen, whose work on live-cell imaging and whose 00:30:30.17 development of technologies has allowed us to do the 00:30:34.07 experiments I just described; and Dick Bellamy at the 00:30:37.18 University of Auckland, who first got me interested in 00:30:40.23 rotovirus in the first place. And thank you.