 which HIV-1 Gag protein interacts with RNA. So just a couple of facts to get us all on the same page as it were. The Gag protein is the building block of the retrovirus particle expression of this protein in a mammalian cell, even in the absence of other viral proteins or of the viral RNA is sufficient for efficient production and release of virus particles. Secondly, after the particle is released from the cell, Gag is cleaved into at least three cleavage products by the viral protease. This is what's called virus maturation. The three products always include from N to C terminus matrix capsid and nucleic capsid. And this maturation event is really a global reorganization of the structure of the particle and is, of course, absolutely required for infectivity. What is the function of these different? So these, of course, are domains within the Gag polyprotein, which is what forms the immature particle. And to a first approximation, the function of the matrix domain of Gag is in interaction of the Gag protein with the plasma membrane of the host of the virus producing cell. But in fact, it also binds RNA. This is significant in a couple of ways. I will be talking about this a little. The capsid domain does most, if not all, of the protein-protein interaction that's involved in virus assembly. And the nucleic capsid domain is what does much of the interaction with RNA. It contains two zinc fingers that are crucial in the interactions with RNA and is also quite basic. I will be talking about that. The cleavage pattern profile of HIV Gag is a little more complex than what I just showed you. The generic version I just showed you, there are two cleavages, not one, between capsid and nucleic capsid, resulting in the release of a 14 residue, so-called spacer, SP1. And there's two cleavages and one more domain, downstream of NC, which I will not mention at all. Okay, so what I wanna tell you today is about these not one, not two, but three different ways that Gag interacts with RNAs. All these ways are important for virus replication. Gag is a nucleic acid chaperone. Gag uses RNA to, it depends on RNA in constructing the virus particle. And of course, it must select the genomic RNA for incorporation into the particle, despite the fact that that RNA is swimming in a sea of cellular RNAs. So, first of all, as a nucleic acid chaperone, what is a nucleic acid chaperone? A nucleic acid chaperone acts just like an enzyme. It catalyzes, that is, lowers energy barriers, the rearrangement of nucleic acids into the most thermodynamically favorable configurations, which in general is the configuration with the maximum number of base pairs. Again, this is just getting nucleic acids out of kinetic traps. No ATP is involved. There is actually a sizable literature about the activity of nucleic-capsid protein as a nucleic acid chaperone and its significance for the virus. So briefly, this is nucleic-capsid protein. It's only 55 amino acids. As you see, it is quite basic and has these two zinc fingers, which are crucial in a couple of ways. I will come back to that. So nucleic acid chaperones essentially promote breathing of nucleic acids, transiently breaking existing base pairs, and thus enabling nucleic acid strands to separate from their partners and search for new partners, but very gently. Whoops. The mechanisms or the properties of nucleic-capsid protein, which give it this activity, have been studied in quite a bit of detail, largely by Carine Musier Forsythe, Yulia, who is speaking after me, and Mark Williams, and at the moment, our understanding of how nucleic-capsid does this depends on the fact that it is a basic protein that is a polycane ion helping to bring nucleic acid molecules close together. It is also a weak, and I emphasize weak destabilizer of base pairs, again, permitting breathing, and its interactions with nucleic acids involve very rapid on and off rates. This chaperone activity of nucleic-capsid is crucial during reverse transcription, that is the synthesis of DNA from the viral RNA as the first step in infection. This synthesis of DNA from RNA involves several strand transfer, that is annealing events where a DNA sequence is produced and then must find an anneal to a hybridized to a complementary DNA sequence elsewhere. So nucleic-capsid is essential for the virus in this way, and gag is also a chaperone. Of course, gag contains nucleic-capsid as a domain within it. I assume that its chaperone activity represents the chaperone activity of its nucleic-capsid domain. So just to give you a little idea of what I'm talking about here, this was an experiment in which we showed that either nucleic-capsid or gag could cause complementary short DNA oligonucleotides to anneal to each other, and in fact, gag and nucleic-capsid had the same activity on a molar basis. This activity, I believe, is essential for virus replication. For the following reason, it seems to be gag which is responsible for annealing a transfer RNA molecule to an 18 base complementary stretch on viral RNA where it must be there to serve as a primer for reverse transcription. So in case there are physicists in the audience, I'm afraid there may be, tRNA molecules are highly structured, compact molecules, they play an essential role in protein synthesis, and they're highly structured, that is, a large fraction of their bases are paired intramolecularly and are thus not available for pairing with a complementary nucleic acid. Many pre-existing base pairs must be broken before the tRNA bases can be paired with bases in the viral RNA. Now, in the lab, it's easy, we just heat it up. Heating it up breaks the base pairs and now the strands can wander around and when we cool it down, they can look for new complementary partners. But retroviruses do it at 37 degrees. And they do it because of the nucleic acid activity, chaperone activity of gag protein. So this was an experiment which just showed that we could anneal free transfer RNA to an RNA representing the viral genome in the presence of gag protein. And if the viral genome was the correct viral genome which had the complementary sequence for the tRNA used here. That's all I'm gonna say about the chaperone activity. Gag also interacts with RNA in constructing the virus particle. Now, I've been working on this a long time and some of you have probably heard me tell this story several times now. And so I am gonna go over it pretty fast this time to get to the last part. But just in a few words and a few slides. So we make recombinant gag protein in bacteria. It lacks a fatty acid modification and is a soluble protein in our hands. And the way we get it to assemble into virus like particles in vitro is we add a nucleic acid. So here, total pellet, supernatant. If no nucleic was added, you see that the gag was in the supernatant fraction. But if any of a variety of nucleic acids was added, including a short, stupid DNA oligonucleotide, the gag is converted into, transferred into a large structure which can be pelleted. That is virus like particles. I was really very surprised 20 years ago by these observations. And I have puzzled for years to understand how nucleic acid contributes to virus assembly. And now I think at some level I understand it. I'm very pleased to say that some of you have even heard me tell that a couple of times. So again, in a few words, we analyzed assembly by a gag protein, a chimeric protein in which the nucleocapsid domain was replaced by a leucine zipper which is a dimerizing domain. And we worked on those proteins and fragments of those proteins a lot both in vivo, in cells, and in test tubes. And we came to the following conclusion. A gag decides to assemble. That is, undergoes changes which make it able, capable of assembly. For example, exposes new interfaces for gag-gag interaction when two or more gag molecules are brought close together at their C-termini which you can do either by binding nucleic acid or by attaching a leucine zipper instead. This juxtaposition of two gags near their C-termini, we showed induces a conformational change in SP1, that little spacer between capsid and nucleocapsid which we suggest leads to further changes in the capsid domain and it does the ability to assemble. We've recently found that assembly can also be induced by adding an acetalhexacys phosphate to gag in vitro. This compound is essentially a six-membered ring with a minus 12 charge. And as you see here, gag protein assembles into nice spherical VLPs when this compound is added. So in other words, we know three ways to induce gag to assemble. Add nucleic acid, add IP6, another highly charged polyanion, or replace the nucleocapsid domain with the leucine zipper. And we believe that all these agents are acting by bringing gags together and flipping a switch within SP1 as shown here in this artist's rendition. So in a free gag molecule solution, SP1 is unstructured but when gags come close together, like when they bind cooperatively to an RNA, that induces a helical, transition to a helical conformation of SP1, which we believe induces further changes leading to assembly, further changes elsewhere in gag, like in capsid leading to assembly. That's all I'm gonna say about that. Now, selecting the genomic RNA for incorporation into the particle, and I am going to show you quite a bit of data about this from a recent paper by Mauricio, which was just accepted in a good journal. And then I'm still not gonna do it justice and Mauricio will tell you more of the story in a short talk later. So this is the problem that gag faces. It needs to find the right RNA for incorporation into the virus particle. Well, there it is. And this is the kind of the paradox, the complication which is such a problem to understand. When gag is expressed in mammalian cells in the absence of viral RNA, it still assembles efficiently. The particles released from these cells contain normal amounts of RNA and the RNA in these particles is cellular messenger RNA. Now, genomic RNA is selectively packaged if it's present because it contains a packaging signal or Psi. It's selected very efficiently with very high fidelity. As I've been telling you, if it's not there, particle assembly is still efficient and cellular mRNAs are packaged. We were interested in the question, well, if gag can't find what it really loves, what's its next choice? And actually, there is very little selectivity in the packaging of cellular mRNAs. So this is the results of a micro array experiment. Actually, two of them, one with HIV, one with myriad leukemia virus. So what was done here was to compare the populations. I don't think you can understand this without an explanation. It's not self-explanatory. We compared the population of mRNAs in the viruses with the population of mRNAs in the cells that made the viruses. And what you see here is a fold change, a bell curve around one. One means that the representation of an mRNA in the virus simply, directly, linearly reflects its representation in the cell that made the virus. It was neither enriched nor de-enriched. And so that is the basic result of this experiment is the vast majority of mRNAs were packaged unselectively. Gag just didn't care. We've actually gone back and looked at this data with the help of a bioinformatician lately. And it turns out, I mean that first result is still correct, but there is a little selectivity. There is, that is, what you see here is the length of the three prime untranslated region for the mRNAs that were packaged, that were enriched just a little, enriched somewhat, the most enriched mRNAs and the not enriched at all, and the, or the not enriched at all, and the de-enriched a little mRNAs. And it turns out that actually Gag does prefer slightly mRNAs with a longer three prime untranslated region. This is not very impressive to look at, but the p-value for this correlation is about 10 to the minus sixteenths. And again, this is experiments with two different, quite different viruses giving the same result. We picked that property because we had no, in the original study, we did tabulate the five or 10 mRNAs that were most enriched. It wasn't very impressive. And we noticed, we looked at one in particular, I mean we started to study, does it have a packaging signal? But it has a long three prime untranslated region. So then went back and looked at the, this study involves between 10 and 20,000 mRNA species. Thanks for the question. tRNA, tRNA is different. I don't believe tRNA is mixed in this pool here, but of course retroviruses need to package tRNA because one of them is gonna prime DNA synthesis and they have mechanisms which differ from one virus to another. Mechanisms for incorporating tRNAs with significant selectivity into the virus because it needs them. And the viruses also have miscellaneous other small RNAs which are there and I don't know why. Okay, so coming back to the central problem, viral RNA is in competition with a very large excess of cellular RNA, mRNA for incorporation into the virus. Psi confers an advantage in this competition. What is the nature of this advantage? So you might imagine, many people have imagined that the packaging signal is a high affinity binding site for gag. That's a nice, simple explanation. Well, we've measured the affinities. This is not trivial. Obviously, you must do it under conditions in which mixing the gag with the RNA does not result in assembly of virus-like particles. That's no way to measure a binding constant. And we have used, we of course is Mauricio here. This is his data. We have used a fluorescence correlation spectroscopy set up for these measurements, although as you'll see the readout was not D, the result of fluorescence correlation spectroscopy, but quenching of the CY5 fluorophore at the three prime end of the RNA. Mauricio chose to study these three RNAs. This represents the HIV packaging signal. These are all just under 200 bases. The HIV packaging signal, another highly structured RNA from the interior of the HIV genome, and the murine leukemia virus packaging signal, which is not functional as a packaging signal for HIV. In turn, HIV Psi was studied in both monomeric and dimeric forms. In fact, retroviruses always package dimeric RNA, and I believe the packaging signal really includes the dimer structure, the structure of the dimer of the RNA. Okay, so setting out to do fluorescence correlation spectroscopy. Here's what was done. The different RNAs I just showed you were put in solution at 15 nanomolar, very dilute. So that's how we got around virus assembly, is the RNAs were so dilute that when gag bound and RNA, those gag RNA complexes could not find other gag RNA complexes and assemble. The buffer includes 0.2 molar salt that is even somewhat above physiological ionic strength. That is a crucial variable in this whole story. And so then gag was titrated into these four solutions and fluorescence correlation spectroscopy was performed. And what you see is that as gag binds the RNA, RNAs, the diffusion constants go up, that is the gags are becoming smaller. Gag is condensing the RNA. This was not what we were expecting, but maybe it should have been. This has been seen from Peter Stockley's lab with capsid proteins of other RNA viruses. So however, Mauricio noticed that in addition to condensing the RNA, changing its diffusion constant, gag was also quenching the CY5 fluorophore that he had put on the three prime end of the RNA. And so he was able to extract binding data from the quenching data. And that is how all the data I'm gonna show you now was obtained. And so here are the binding curves for those four RNAs obtained by titrating gag into 15 nanomolar solution of those different RNAs in the buffer with 0.2 molar salt. And as you see here, it's pretty hard to tell those four curves apart. These differences in affinities, I mean, so you can get KDs from these curves and they differ by like a factor of two. This difference in affinities is certainly not enough to explain selective packaging. Further analysis of these curves shows that they are, that there is cooperativity in the binding. This is a consistent feature of all of the binding curves, that there is essentially a hill coefficient of two or three somewhere around that. I think it would, we would probably expect that gag of a virus building block would bind cooperatively to RNA, but this has not been documented before. However, I am fascinated to tell you that binding to the different RNAs are not as similar as they appear. So Mauricio started attacking the binding in different ways and every time he attacked it, he found that there really is, gag really can discriminate between the packaging signal and other RNAs. So one way he modified, oops, one way he modified the experiment was to add a 50 fold excess by mass of tRNA in addition to the 15 nanomolar labeled RNAs. And what he saw was that adding this cold, irrelevant competitor RNA had a small effect on the binding to the psi RNAs, but a much larger effect on binding to the control RNAs as you see here. So adding tRNA changed the KD for the right RNA by about a factor of two, but changed the KD for the control RNA by a factor of eight. This is one way of showing that the binding to the wrong RNA is non-specific binding, which can be competed by an irrelevant RNA as contrasted with binding to the right RNA, which gag will still bind to despite the excess of tRNA. Another way he, so it's been very informative to use mutants of gag protein. One mutant he used has been called eight N in the literature, so eight positive residues in the matrix domain, matrix domain at the N-terminus of gag is quite basic as well as the nuclear capsid domain. So in this mutant, the basic character has been diminished by changing eight basic residues to asparagine. And this mutant gag also really discriminates quite strikingly between the right RNA and the wrong RNA. That is, it has lost a lot of its affinity for the wrong RNA. The affinity for the wrong RNA seems to represent arise from basic charges in the matrix domain unlike the affinity for the right RNA. In fact, if you now add tRNAs into this experiment, so now this is binding curves for the mutant gag to the right RNA or the wrong RNA, whether without tRNA, addition of tRNA makes essentially no difference now. Thus, the binding of tRNA apparently all depends on the matrix domain and depends on those basic residues in the matrix domain. And then another way of attacking the experiment in the way it's the simplest way of attacking it, and I think it's very straightforward and informative, is just to raise the salt concentration. And in this experiment, we are really following the path blazed by Yulia and Kareem Yuzier-Forsyth a couple of years earlier. And the result is binding to the right RNA is far more salt resistant than binding to the wrong RNA. So what you see here is binding curves. So everything I've showed you now has been in 0.2 molar salt. If we just raise the salt to 0.3 or 0.4 molar, there is very little effect on binding to the right RNA. However, binding to the wrong RNA is practically eliminated by 0.4 molar salt. So therefore, I think that is a really basic result property of the system, showing that the binding to the wrong RNA has a higher electrostatic component or a smaller non-electrostatic component than binding to the right RNA. Now, 10 minutes, okay? All right. There's an interesting aspect to this which I don't quite understand, but I'm gonna mention it. Now, the literature includes classic work from Rekerd and Lohman, who showed in their experiments that if you plot the log of the KD versus the log of the sodium concentration, you get a straight line. And the slope of the line represents the number of sodium ions displaced by binding of one protein molecule to the nucleic acid. This is from one of their papers 40 years ago, but we don't get that. Our curves for binding to either the right RNA or the wrong RNA are bent. And in fact, they're bent upward. In other words, as we raise the salt, the binding becomes more salt sensitive. It doesn't make too much sense, but that's what we get. It suggests that changing the salt, I mean, I think the only way to explain these results is to say that the gag protein itself is changing as we change the salt concentration. And in fact, we have seen quite a bit changes in gag. For example, gag in solution is bent over or compact, but in the virus particle, it is an extended rod. So we know gag is capable of significant changes in its confirmation. I had no particular reason to expect changing the salt concentration would do that, but I think that's what this result implies. Mauricio was able to model the data by assuming that the non-electrostatic component changes gradually as the, diminishes gradually as the salt concentration is raised and using this slope actually, that the real affinity is the sum of electrostatic and non-electrostatic affinity of binding and this slope represents the electrostatic component and this slope actually comes from data. Data I'll show you in a minute with another gag mutant. So it's possible to model this behavior. Couple other interesting results about the binding. So we showed years ago, gag is in monomer dimer equilibrium solution, the dimer interface is within the capsid domain and we generated a point mutant WM, we called it at that interface, which is defective in dimerization. Remarkably, this mutant has mostly lost its ability to bind the wrong RNA. So this is what you've seen before, binding of wild type gag to the right RNA or the wrong RNA in increasing salt concentrations and as you see the green line here is 400 millimolar salt and binding to the right RNA survives, binding to the wrong RNA is largely lost. However, if we now do the same experiment with the mutant gag, we see binding to the right RNA well is significantly more salt sensitive, but it still binds and but binding to the wrong RNA is so weak that we actually, Mauricio could not detect any binding when he raised the salt above 0.2 molar. So this is very weak in its binding to the wrong RNA, although it's chain, it's this mutant, this protein is altered in the capsid domain, which does not interact with nucleic acids at all. Its function is in protein-protein interaction. Now I mentioned the zinc fingers in nucleic capsid and so this is a mutant, this is now testing a gag mutant, which we call SSH or SSHC in which the zinc chelating cysteines in nucleic capsid have been replaced with serines. In other words, this no longer has the zinc fingers and as you see, this mutant has, it's binding to the right RNA is much more salt sensitive than the wild type, which I've shown you a couple of times before and again, it's binding to the wrong RNA is, it was unmeasurable above 0.2 molar salt. We think that the zinc fingers, although we do see a difference between this and this, we still think that the zinc fingers are primarily responsible for the non electrostatic specific component of binding to the right RNA. By the way, this mutant protein in vivo in cells assembles into viruses okay, but fails to package viral RNA. What is Cy? And so I'm going to tell you a little about this and I'm sure Mauricio will fill in the blanks and pick up where I left off. So Cy is located in the five prime on translated region and beginning of the gag coding region. It has a number of regulatory functions for the virus. So this is from paper by Kevin Weeks. This is the secondary structure of the five prime end of viral RNA, about 700 bases shown here. This is obtained by chemical probing of the RNA inside virus particles. And so, well, there's various secondary structures, which I won't discuss, but I need to point out what Kevin called the nucleocapsid interaction domain. So another thing Kevin did was to take that virus, treat it with an oxidized, a mild membrane permeable oxidizing agent. This chemical penetrates into the virus and takes away the electrons with which the cysteines are holding on to the zinc in the zinc fingers, thus destroys the zinc fingers. And then he did the chemical probing and he found that this pretreatment caused the, I mean, everything was the same except there were these seven little turquoise clusters of bases that were exposed by the treatment and he inferred that these were bases which are protected by nucleocapsid inside the virus particle. So if they're bound to nucleocapsid, maybe they're important in the packaging signal. And so Mauricio has mutated those bases and tested binding of GAG to that mutant and also some other mutants which I am gonna skip over for reasons of time. And again, at 0.2 molar salt, wild type GAG binds all RNAs pretty much the same. However, if you challenge the binding by raising the salt to 0.4 molar, then you see that this mutant which is called NBSM first generation here has really lost quite a bit of its affinity as shown in this curve here for GAG. So we think that these results and more that I imagine Mauricio will tell you about are give some insight into what is Cy, what is it that GAG discriminates from random RNA, both in our binding experiments and in packaging by our RNA. So in summary, GAG is a nucleocapsid chaperone. It uses cooperative binding to RNA to bring multiple GAG molecules close together. This triggers assembly. Packages Cy containing RNA with high selectivity. If it's present in the cell, otherwise it packages mRNAs with very little selectivity. It binds with very similar high affinity to all RNAs tested in 0.2 molar salt. Therefore high affinity cannot explain selective packaging. The binding is clearly the sum of specific and non-specific interactions. Specificity was revealed when mutant GAGs were used or a non-specific competing RNA was present or when the salt concentration was raised. Properties of the mutant GAGs showed that the non-specific binding is largely attributable to the MA domain and apparently GAG-GAG interaction also makes a major contribution to non-specific binding. The salt resistant binding apparently requires unpaired GAGs in what Kevin called the nucleocapsid interaction domain. Speculative remarks, just two slides. Michael Summers years ago determined by NMR the structure of complexes between nucleocapsid and specific stem loops within the five gram untranslated region. He found that NC binds well to unpaired GAGs and that hydrophobic residues with the zinc fingers stack within the Gs in these complexes. So you see here a G base in an RNA in a cleft between phenylalanine and an isoleucine in one of the zinc fingers. We propose that the key to selective packaging is the efficiency of nucleational particle assembly as the immature particle is a hexameric lattice of GAG perhaps when the NC domains of six GAGs each bind to one of those little stretches of unpaired Gs assembly is initiated. This can also occur on other RNAs. What we propose that it happens faster or more efficiently on these sequences. What I showed you today is largely the work of Mauricio. He was other contributions to the worker from CID, Laura, Rajat and Reddy. And I also showed you work going back over the decades from a few other lab members. Thank you.