 Thank you, Roya. And thank you, Roya, Christian, and Rudy for organizing this really wonderful meeting and for including me in it. I'm going to be talking about joint work with Chuck Nobler, who had hoped very much to be at the meeting. And I'm going to be talking about the work of several former students and current students. Avi already mentioned Jago Pal. I'll be showing some of the work of Ruben Kedena-Nava, who's professor now in Mexico. Mauricio Comasco is here. Alan Rhine is lab at the NIH now. Reese Garmin is here at Harvard, and Vinayana Harnes' lab. Christian Byrne is here. Current graduate student, Adesia Zizgalshani is a PhD student and postdoc in the group. I have a little square that pops up on my monitor. One out of three times, I turn on my computer, and it'll be there throughout the talk. Sorry about it. I'm glad it's not a bigger one. So Chuck and I have appreciated from the beginning that we've had the great benefit of a brave and talented group of graduate students who have to learn molecular biology and biochemistry on their own and who are willing to try out our ideas about the several viruses we're working on. Avi already mentioned that we've been doing joint theoretical work almost 40 years now. And he mentioned these particular theory students. So one of the ways we started trying to understand viruses from the beginning was to contrast RNA viruses with DNA viruses. When I talk about RNA, I will mean single-stranded RNA even though I won't say it explicitly. Similarly, when I talk about DNA, it's double-stranded DNA. So the question we found helpful to ask was to understand the difference between life cycles and many features of RNA versus DNA viruses. Wouldn't it be useful to think about the differences between RNA and DNA, the genomes themselves, as physical objects? And we've already heard a lot about how DNA, double-stranded DNA, is strongly confined in viruses. It requires a lot of work to package the genome. The release or delivery of the genome is spontaneous. RNA viruses, the packaging of the genome, the formation of the virus particle is spontaneous. You're going to a lower free energy state. Suggests you'd have to do work to get the genome out. We want to understand these things in terms of the extent to which an RNA gene is a different physical object from a DNA gene. I'll show you some examples of how explicitly I mean that. I want to emphasize that we find it useful to talk about the genotype and the phenotype in the same breath, namely the RNA genome corresponding as it does to a certain sequence of nucleotides and therefore coding for certain proteins. Clearly that molecule is the genotype, but its size or particular secondary structure features that bind proteins in certain ways, that's the phenotype of the virus. You don't go looking for what the color of the virus's eyes are. That's the usual phenotypic characterization. In this case, you're asking, what does the genome look like as a physical object? So that's the phenotype. We've seen this picture several times. I show it because I want to contrast it with the situation you have for RNA viruses. Here the whole point is that DNA is strongly confined by being packaged in its protective capsid. The size of the capsid is small compared to the size of the genome. The DNA contour length is orders of magnitude larger. Significantly, the persistence length of DNA is smaller, of course, than the contour length. That's why this behaves on a larger length scale like a semi-flexible polymer. But the persistence length is larger than the radius of the capsid, which means the DNA is strongly bent all along its length in addition to being strongly crowded on itself. And that's what gives rise to the high forces, pressures, and the large amount of work you have to do to package the DNA. In the case of RNA viruses, again, you've seen several introductions to this. Avi just mentioned, and I think in previous talks, this example was given. There's something special about 3,000 nucleotides for this virus, namely, that's the typical gene content or nucleotide number associated with each one of the identical capsids that protects different parts of the genome. This virus could have evolved to have all its genes in one molecule in a bigger capsid, but there are many important reasons why the plant viruses need to be small in at least one dimension. And so you commonly have multiportide genomes and protect pieces of the genome one at a time. We've just heard about what allows this packaging to occur spontaneously, namely that the RNA, unlike DNA, is already compactified before you add capsid protein and try to protect it. This is a cryoelectron micrograph from a Jacob Paul's work. To make the RNA molecules that we package, we make them by in vitro transcription from the cDNA. And so when we want to work with the viral RNA molecule of CCMV that's a little over 2,000 nucleotides long, we transcribe it from a piece of DNA that length. And so conveniently, you can ask, how differently do these two molecules look? They have exactly the same genetic information, but one is in DNA form. This is the 2117 base pair long template from which we make the 2117 nucleotide long single-stranded RNA. And you see the RNA molecules are branched. They're compactified by the branching. And you have this amount of genetic information in very efficient form. DNA is a less efficient way from a space point of view to store genetic information. DNA, double-stranded DNA, has many advantages, obviously, chemically and otherwise over single-stranded RNA. But as physical objects, we see the difference in the phenotype of this genetic information. Here is a picture of one of the three molecules of the virus, CCMV, RNA 2. It's almost 3,000 nucleotides long. We have a purified sample of it. We take pictures of it and try to sort out branching characteristics and so on. But we can put in a scale bar, determined in the usual way from the microscope. This is the best scale bar of all. We throw in a little bit of the purified virus. We know that that's 28 nanometers. And you see that indeed, as you've seen in previous slides, the RNA that is packaged by protected by the capsid protein you add to a solution of the purified RNA under assembly conditions, it's already about the right size. In contrast to the situation for DNA, I want to speak briefly about the two-step in-vitro self-assembly protocol that was worked out by Ruben Briceau in ultimately clarifying it as follows. We mix the purified capsid protein and the RNA of interest. It can be viral RNA or non-viral RNA, ranging from hundreds to up to 10,000 nucleotides. Mix them in physiological buffer and then dialyze against lower pH. What happens at neutral pH is that the RNA is saturated by capsid protein, and you lower the pH heading towards the isoelectric point of the protein, and you enhance the lateral interactions between the proteins that are bound on the RNA. And then you have the formation of capsids. And you analyze the assembly products with gel, sucrose gradients, or looking directly at what you have. These are pictures from Brice's work where he wanted to check that indeed you have to lower the pH to at least six from neutral pH before the protein-bound RNA complexes form proper variants or virus-like particles. Happens in negative stain, lectin microscopy. By adding urinal acetate, you're lowering the pH by several units. So you can take the neutral pH assembly mix where we know from cryoAM that you don't have properly formed capsids. You just have RNA saturated by protein. When you prepare your EM sample, you are actually lowering the pH. And that's why you see capsids. What we learned, again, through the work of Ruben Cadena, Mauricio Comas-Carzi, and Ries Garment, is that if you have a sufficient excess of capsid a super stoichiometric one. So for 3,000 nucleotide long molecule, instead of needing only 180 proteins per RNA molecule to give you a T equals 3 capsid containing that molecule, to get complete packaging of the RNA, you need more like 300. So corresponds to a 6 to 1 mass ratio of protein to RNA. And we call it the magic ratio because it allows complete packaging no matter what the RNA is if you have that mass ratio. And it happens to correspond, not coincidentally, to the mass ratio that allows the cationic charge on the n-termini of the proteins to be equal to the anionic charge on the RNA. And if the RNA isn't too short or too long, you get complete packaging into a wild type T equals 3 capsid. If it's shorter than that, you get smaller capsids. If it's much smaller than that, you get many copies of RNA per capsid. If it's longer than 4,000 nucleotides, you get the wild type structure, the T equals 3, but you get two or more capsids sharing a single RNA or multiple. Learning how much capsid protein you need to completely package an RNA allowed ratio to do well-defined competition experiments where you take equal masses of two RNA molecules whose relative packaging efficiencies you would like to determine, you take equal masses of them and add six times as much mass of protein as you have mass of either one of the RNAs. So now you have a 3 to 1 mass ratio of protein to total RNA, it's enough to completely package one of them, but not both of them. So now you have proper grounds for a competition and you go through the same two-step packaging protocol. Perhaps I didn't mention before, you add RNAs to get rid of any unpackaged RNA. And then you analyze your assembly products and see what fraction of the virus-like particles of one RNA and what fraction of the other. Through a lot of work of that kind, we learned that indeed, viral sequences are packaged more efficiently than equal-length non-viral sequences. And we interpret that, understand that in terms of the viral sequences giving a more compact structure, which can be more easily accommodated by the protein that prefers a certain curvature and T equals T structure. We've heard from Raiden about the importance of packaging signals. Obviously, the viral sequences have them, non-viral sequences don't. So after sorting things out along these lines, we thought we should go outside our comfort zone and ask, what happens when you take an unnatural RNA, like a single-letter RNA, that has no possibility of forming secondary structure? It has no self-complementarity. We picked Polly U because of out of all the single-letter RNAs, it's the one that has the weakest base stacking so you don't even get any helical ordering of the single-stranded polymer. We guessed that for all the reasons I've been talking about, it would not be an efficiently packaged molecule and certainly couldn't compete with normal composition RNA, little viral RNA. And it turned out to be a very interesting experiment because we got nothing of what we were expecting and we're still trying to understand it. It's hard to prepare monodisperse samples of Polly U. See, it is synthesized, you have a broad distribution of sizes, you can fractionate your sample into very roughly monodisperse plus or minus 50% samples. So instead of talking about 2,000 nucleotide long Polly U, we talk about Polly U in this range, this range, this range, and so on. And we take each of those samples. This is the work of Christian Baron. He doesn't have a poster on it. He has a poster on 3D reconstructions of CCMV and BMV RNA and capsids, actually just BMV for the reconstructions. We take each one of those samples, carry out the in vitro packaging, and find our first surprise that we don't see any wild type T equals 3 capsids. We only see T equals 2 sized capsids. Bogdan, looking at you, I remember years ago, when we mentioned T equals 2 structures, we saw, I'll come to them in a moment, when we're looking at short RNA molecules with normal composition going up to longer ones, generating doublets, triplets, quadruplets, and so on, showing that when RNA is shorter than 2,500 nucleotides, you get smaller than T equals 3 structures. We call them T equals 2. You said you don't know that until you do a 3D reconstruction. We know from Christian trying to do a 3D reconstruction that we have T equals 2 sized particles, but they're not T equals 2. They're trying very hard to be, but they don't manage. That's somewhat polymorphic, a little anisotropic. But again, every one of these colors is a different size of polyu, and every one of them gives a distribution centered around T equals 2 size in stark contrast to the wild type virus, or we would get the same size distribution for a normal length, say 3,000 nucleotide, normal composition RNA. We again see a succession of multiplets, and they set in at the same lengths as for normal composition RNA, but you never see bigger capsids than the T equals 2 size capsids. Then Christian did competition experiments between polyu and viral RNA, DMV, RNA1, which happens to be the most efficiently packaged molecule when you use CCMV capsid protein. It's more efficient than CCMV RNA1. But it loses in a competition against polyu. Here it's easy to analyze the assembly products. You don't have to carefully put different fluorescent labels on the RNA molecules because you're getting different sized particles, and you just count the relative numbers of the different sized particles, and you find that there are more T equals 2 size particles than T equals 3. And you confirm with the usual labeling experiments. So not only do you have the surprise that you get a non-wild type virus like particle, not at T equals 3, but it turns out that polyu is more efficiently packaged than viral RNA. Is it really packaged or it's a cluster? It is packaged in terms of what it looks like in an electron micrograph, and it's protecting the polyu against RNAs. Those are the criteria for packaging into a virus like particle. So it raises the question, what determines the size of capsid that will be formed when you take a certain capsid protein and add RNA of different lengths and sequences to it? And now, of course, we're taking a broad range of lengths. We're also taking a dramatically broad range of sequences, including polyu. So I'm going to be contrasting a single-letter RNA, like polyu, where you have no possibility of sucking the structural formation, with what we call normal composition RNA, whether it's viral sequence or not, you have comparable numbers of each of the four letters. It's easy to answer this question, what determines the curvature. You just think of all things that might determine the curvature, and you start listing them. It's the intrinsic or spontaneous curvature of the capsid protein. So it's the nature of the protein interactions. But we also know that it depends on how the conformation of the protein is affected by interaction with RNA. So here's another factor. We also know it's determined by the size of RNA, because when we go from short RNA to long RNA, we go from T equals 1 to T equals 2 to T equals 3 type structures, bigger and bigger capsids. But I remind you, we get different scenarios for different kinds of RNA and different kinds of anionic polymers. For normal composition RNA, you go through this progression of sizes, and then to multiplets involving the biggest size. The first work we did was with polystyrene sulfonate before we started packaging RNA with CCMV capsid protein. And again, we found there a progression from smaller than T equals 3, T equals 2 sized VLPs to T equals 3. With polyu, we find a very different result. You never get to the wild type T equals 3 structure, but you do go through the progression from singlets to multiplets. Kristin is presently looking at poly A and sees something similar to poly U, but we have to look more carefully at it. So here is where I want to go back to normal composition RNA and use what we've learned about packaging of RNA by capsid protein that's capable of in vitro self-assembly from purified components, namely BMV or CCMV capsid protein. I want to go back to what we've learned and ask how can we use it to develop the best gene delivery system. And so I want to go over some of the requirements for a good gene delivery system and how the answer to the question, what's the best system, will come from looking at the differences between RNA and DNA genes. RNA genes, in contrast to DNA, can be packaged spontaneously if we use a promiscuous indiscriminate capsid protein like CCMV or BMV, which doesn't care about whether it's packaging viral RNA or heterologous RNA or even synthetic anionic polymer. Now we get into a little biology. RNA genes, most of the RNA viruses that have been mentioned so far, get to HIV soon, and that's very different. They're not just RNA viruses, similar standard RNA viruses. They are positive sense RNA genomes, namely there are messenger RNA molecules that are directly translated by ribosomes in the cytoplasm of the infected cell. So to a significant extent, the nucleus of the infected cell is never involved. The viral life cycle is much shorter because as soon as the genome is made available to the ribosomal machinery, you have gene expression and viral replication. And then another thing that's special about RNA is that it can be self-replicating. And I'll explain what I mean by that and how we take advantage of it with the DNA virus. DNA gets into the nucleus. You have transcription of the viral genes. You have the needed viral gene products like capsid protein and so on. How do you replicate the genome? You use the DNA polymerase of the host. DNA replication is not an exotic thing for the host cell. But RNA replication is. So genomes of positive sense RNA viruses code for an enzyme that wouldn't be found in the cell otherwise, namely an RNA-dependent RNA polymerase. It's an RNA replicase. And that replicase messenger RNA codes for the enzyme that replicates it. The whole genome is replicated by the RNA-dependent RNA polymerase gene product. That's one of the few genes of the RNA virus. So here's our strategy. You recognize this genome? Arguably, the simplest genome organization of an virus is this one, where you have two open reading frames. Alpha viruses have this structure. One open reading frame for all the nonstructural genes. So most importantly, those genes include the genes for RNA replication. And then one open reading frame for the structural genes. It would be capsid protein for a naked virus and capsid protein plus membrane proteins for an envelope virus like the alpha viruses. What's special here is that this genome is directly translated. There's a stop code on here so that the only gene products that are directly synthesized are the enzymes that will make up to a million copies of the genome. It's a very powerful strategy for a virus to make a million copies of itself before anything else happens. Significantly, to get structural protein gene products, you need to have RNA replication. So when you make minus strands of the full length genome, from those minus strands, you make full length plus strands. Those are the copies of the genome you need for the next generation of virus. But you also start transcribing the minus strand of the full genome at this junction here. And you make millions of copies of the messenger RNA for the structural proteins, of which you need a huge number. So I argue it's the most powerful way to get brought to evolution by these viruses to get high levels of expression of protein. And that's what we want to take advantage of. Basically, we want to, an RNA viral genome like this, cut out the structural genes, replace them by genes of interest, which could be, at first, reported genes to quantify what's happening, and then viral vaccines, cancer vaccines, anything you like. We're basically inserting into the viral replicon a gene of interest in taking out the structural genes of the virus so that you don't have an infectious reagent. We need to find an RNA virus whose replicase genes are short enough to be packaged by CC and B-capsid protein. That's our next requirement. And we found an insect virus, notomura, very closely related to flockhouse virus, whose genome has a related but different structure to what I just showed you. Namely, it's a two molecule genome. And here's the open reading frame for the replicase genes and here for the capsid protein, molecule 2. So the two molecules are co-packaged. And you make the replicase, which makes a large number of copies of both molecules. And this is the replicon. It's only 3,000 nucleotides, which gives us 1,000 to 1,500 nucleotides to work with for inserting genes of interest. And that's just what we do. So the RNA is dependent RNA polymerase, which is protein A in the case of notomura. We leave intact. We add a gene of interest. We add sequence for self-cleaving peptide because we want the RDRP and the gene of interest to be functional. We make them as they are made as a polyprotein and then cleaved. And then these are pictures and analyses by Rhys Garmin. With the first construct we made of this kind, while it was still at UCLA, where the reporter gene is an EYFP, the total length of the viral replicon plus inserted gene is about 4,000 nucleotides. And you get nice RNAs resistant, thermally stable, aggregation stable, T equals 3 VLPs. Remember, we've got the replicase gene from an insect virus. We're packaging it in the capsid from a plant virus. And we want everything to work in mammalian cells. One of the reasons we picked notomura was not just that it had the right replicon length, but it had been shown to be active in mammalian cells. And we checked that. This is work out of C. Aziz-Kolshani. By transfecting these in vitro reconstituted particles with the fluorescent protein reporter gene into mammalian cells and seeing high levels of fluorescence. We've checked it since in many ways in many different cell lines because we want to use this as a gene delivery system. Five minutes? One minute. OK. Thank you. Just checking, not only that you get a high level. In this case, it's a Luciferase reporter gene. Because of the RNA amplification, you get expression for many days. This is the messenger RNA form of Luciferase. This is the replicon form of Luciferase. So not only are you amplifying the messenger RNA for Luciferase, you're doing it over the course of many days. And that was work of Adam Biddlecombe. So this is, in fact, my last slide. And I want to summarize what we've learned and where we are. We choose genes in RNA form because we can in vitro package them, which means we can protect our genes that we want to deliver, taking advantage of the flexibility and compactness of RNA genes versus DNA genes. We use the fact that certain plant viruses like CCMV have risk in discriminant capsid proteins that will completely efficiently package the RNA replicon gene. We use the self-amplification strategy of positive sense RNA viruses and take advantage of the fact that those particular replicases work in mammalian cells. So we haven't created an artificial virus, but if you like a hybrid virus, for the reasons I talked about, I'll stop now and I hope there are questions and we can discuss these things further. Thank you very much.