Scientist Stories: Britt Glaunsinger, How Viruses Reshape the Gene Expression Mammalian Cells

[Speaker] Thank you. So what my lab is interested in is understanding pathways that control RNA destruction, in particular those that can have global effects on gene expression. And so we're trying to figure out basically what pathways regulate RNA stability in cells and then also How do broad changes in RNA levels influence viral replication and pathogenesis? And so to approach these questions, we study how viruses interface with these pathways, in particular those viruses that can promote a widespread messenger RNA turnover phenotype in cells. And our primary model systems are gamma herpesviruses like Kaposi's sarcoma-associated herpesvirus, shown here. This is a virus that is the etiologic agent of Kaposi's sarcoma, one of the main AIDS-defining illnesses of the 1980s. Also causes several B-cell lymphoproliferative disorders. And one of the reasons that we study it is that during lytic infection with KSHV or other gamma herpesviruses like Epstein-Barr virus or murine relatives of these viruses, their lytic infection promotes a very prominent destruction of messenger RNAs in cells. So in fact, the vast majority of host messenger RNAs are downregulated during infection. And what we've shown is that this is primarily carried out by a single viral protein called SOX. So SOX stands for shut-off and exonuclease, and that's because this protein is bifunctional. It was actually characterized originally a number of years ago as a DNase that's important for resolution of branch structures that arise during rolling circle DNA replication, viral DNA replication in the nucleus. So because of this function, it's actually essential for viral replication. Delete the gene, or you will kill the virus, basically. And it's also highly conserved in all herpesviruses. However, as it turns out, this ability to restrict host gene expression and promote messenger RNA destruction is only found in the gamma subfamily of herpesviruses, like KSHV and EBV. And so we can actually generate mutants that can genetically separate these two functions just shown here, this is a crystal structure of the SOX protein that was solved in the last year and a half or so, and in blue are residues that retain only the DNase activity, but they lack this RNA turnover activity, and those tend to map to the surface of the protein. So this is just to emphasize that, as this audience well knows, viruses are extremely good at getting the host cell to do things for the most part for them. So they've evolved very closely with host cells and can use them to their maximal advantage. And we think that this is no exception for the case of this host shut-off phenotype. We think it's an interconnected series of events that's actually set off by activities from this viral SOX protein, but then largely carried out by host pathways and host proteins. So this is just to summarize what what we've been studying about the mechanism of how SOX induces this phenotype over the past few years. What we think happens is that SOX comes in and specifically recognizes translationally competent messages, or RNAs undergoing translation. And this is how it is able to get rid of messenger RNAs instead of other types of RNAs in the cell. And what it does is that it induces a site-specific endonucleolytic cleavage, one or a small number of these, on these translating messages, and And perhaps that's all it needs to do. After that's done, it may not play a large role in actually getting rid of most of these RNAs. Instead, for that, it relies on host RNases. In particular, we've shown that XRN1 plays a major role in then subsequently clearing these messages from the cell. So you get widespread messenger RNA degradation in the cytoplasm, and that basically what SOX is allowing these host enzymes to do is access these RNAs in a way that bypasses sort of the rate-limiting steps of deadenylation and decapping that they normally have to sort of wait for before they can access messages. So this, though, turns out to set off an additional sort of cascade of events that further downregulates gene expression. One of these events that we've spent a lot of time studying is that cytoplasmic poly-A binding protein, which is normally bound to the poly-A tails of messenger RNAs— it plays an important role in translation and messenger RNA RNA stability, gets liberated from the 3' ends of these messages while they're undergoing sort of this rampant degradation. And what we've found is that this liberation of PABCI reveals one or more non-canonical nuclear localization signals in the protein, causing it to translocate to the nucleus and accumulate there at higher than physiologic levels. Now, what we found is that if PABCI accumulates in the nucleus, this actually drives an RNA processing abnormality in which in which newly transcribed messenger RNAs acquire longer than usual poly-A tails, so they get hyperadenylated. And this doesn't, as it turns out, destabilize the messages, but what it does is we think it renders these messages sort of targets of nuclear quality control machinery. They're recognized as aberrant and they cannot be exported from the cell. And so you have a prominent RNA export block in these cells where you have nuclear PAB-C accumulated. And the interesting thing for us is that it's— we and others have now shown that this doesn't just happen during KSHV infection, but infection with a wide variety of unrelated viruses like rotavirus and bunyavirus and SARS coronavirus also causes PAB-C nuclear relocalization, and which we think then results in an export block. And so this failure to repopulate the cytoplasm with newly transcribed messages then serves to increase the magnitude of the overall gene expression block in cells. And so I'm not going to— well, a lot of the lab sort of works on the mechanics of these processes. Instead, what I'm going to talk about today are efforts to understand why does the virus do this in the first place. So when you think about viruses inducing host shut-off phenotype, there are lots of different viruses that do it via a variety of mechanisms. And generally, people would suggest, well, it's because you have a generalized immune evasion type phenotype that's achieved by this, or it helps the virus not have to compete with the cell for resources. But it's been very difficult to sort of tease out exactly— those make sense, but is that really the case of what's going on here? And so to get at this question, instead of using the human herpesviruses, which are incredibly difficult to work with genetically, and there's no good small animal model system for them, they're very species-specific, we instead use a related murine herpesvirus called mouse gamma herpesvirus 68, or MHV-68. And so this has served as a very useful model to understand the pathogenesis of the human gamma herpesviruses. And so it's very genetically tractable and infects mice, and we've shown that it has a functional homolog of the SOX protein, which we call muSOX for murine SOX. So what a graduate student, Justin Richner, in the lab did was he set about— basically did a random mutagenesis screen and was able to identify a single-function, one-amino acid change mutant in the MUSOX gene which allowed it to selectively lose this ability to trigger RNA turnover but retain the wild-type DNase activity. And remember, this is critical because if you get rid of both activities, you're just going to kill viral replication because DNase activity is essential for the mechanics of viral replication. But instead this mutant, while it has wild-type DNase activity, is defective for all of the hallmarks of host shutoff, including messenger RNA turnover, PAB-C relocalization, and hyperadenylation. And so we could build this mutation back into the virus, and in fact a postdoc in the lab, Karen Clyde, has recently shown using microarray analysis that if you compare the overall levels of of cellular endogenous genes in cells infected with the wild-type virus relative to this mutant virus, which we call the delta host shut-off virus. In fact, in the wild-type virus, the vast majority of genes are downregulated, indicating that this virus is defective globally in messenger RNA turnover. So one of the first things that Justin did was just ask, is this phenotype necessary for viral replication in cultured cells? And so he's using murine fibroblasts here in a multi-step growth curve, and basically sees that in this cell type at least, there is no replication defect over a series of replication events. And so he's comparing wild-type virus with the Delta Ho-Shut-Off virus. This MR is a mutant rescue virus in which we replace the mutation that we made with a wild-type codon to make sure there are no other mutations in the genome that could be causing these effects. So this isn't particularly surprising. You'd think tissue culture cells are bathed in a lot of growth factors, you might not have a resource allocation problem, and you're certainly probably not having as big of an immune evasion problem there. And so what we really wanted to look at was in an animal model. So how does MHV-68 replicate in a mouse? Well, it's thought the natural route of infection is intranasal, so we can do an intranasal infection on these mice, and about 3 days to a week or so later, The first thing that happens is you get a burst of lytic replication within the mouse lung. And this is sort of the major phase of lytic replication within these mice and probably serves to seed the rest of the events that happen within the mouse. From the lung, the virus then traffics to the lymph nodes and then subsequently to the spleen where it establishes latency, probably in germinal center B cells, and then eventually will basically persist for the lifetime of the host, similar to how anybody gets infected with a herpes virus, you're basically never going to clear it. And so we have this hypothesis that— so host shutoff is an exclusively lytic event, right? It's going to be toxic for cells, and latently infected cells are not destined to die. So we know that it doesn't happen during latency. MUSOX is a lytic gene. It's expressed with delayed early kinetics. So if you're looking at these stages of the viral life cycle in a mouse, the obvious one is this burst of lytic replication in the lung. These three all have a lot to do with latency. This is sort of the major lytic replication time, and so we thought, well, we're likely to see a defect there if something's going to happen. And of course, we were totally wrong in that if Justin looks at 3, 5, and 7 days post-infection, this is when you have most of the lytic replication happening within the lung. There's basically almost no defect whatsoever at all. So this is comparing the mutant rescue virus, which is like the wild-type virus, and our delta host shut-off virus here. And so no difference at 3 days and at 5 days. There's a very slight half-log defect at 7 days, but really nothing to write home about. And so we scratched our heads over this for a little while and then decided to look at later times. And so he then looks at 2 weeks what happens within the spleen when you have peak establishment of viral latency. And the first clue that something interesting was going on is when he looks at the spleens of these mice, the— really the only major pathology you see in an MHV-68-infected mouse is they have this sort of prominent splenomegaly, or enlarged spleen phenotype. And that was completely absent from the mice that were infected with the mutant virus. So these spleens are normal size, whereas mice infected with the wild-type virus or the mutant rescue virus have spleens that are 2 to 4 times the normal size of a mouse. So this suggested, okay, well, there's basically— you're missing the main pathological phenotype here. And so then he could go on to actually calculate the frequency with which of MHV-68 infection in splenocytes at this time, and he does this using a limiting dilution PCR-type assay where he can actually map down almost single-copy sensitivity which splenocytes, or the number of splenocytes, that harbor the viral genome. And what he found is that similar to what other people have found in a wild-type infection or a mutant rescue infection, approximately 1 in 200 or so splenocytes will harbor the viral genome, probably in the latent state. In contrast with the host shut-off defective virus, this number is dramatically reduced to, in fact, calculated to be less than 1 in 10,000 cells are latently infected with the virus. So there's this major latency establishment defect with virus that can't promote RNA turnover. So is it just that you have a latency establishment phenotype, which is sort of completely surprising to us in the first place, given that this is a lytic-exclusive phenotype, or could there be more upstream events in between lytic replication in the lung and establishment of latency in the spleen that could be driving this effect. And so Justin looked at an intermediate time, 10 days post-infection, when the virus is mostly gone from the lung but not yet really establishing latency in the spleen, it's trafficked to the lymph nodes. And even at this time, he sees that the virus which cannot promote RNA turnover is getting to the spleen at lower levels than the mutant rescue virus. So this suggests that there's an earlier trafficking defect in the ability of the virus to get from the lung to the lymph nodes. And so then you can envision, well, our failure to see it accumulating in splenocytes is just because you're not getting trafficking from the lung, then even less efficiently from the lung to the spleen. But we don't think that's the case, and that's because he can bypass the need for trafficking altogether by doing an intraperitoneal injection, which basically the virus can get directly to the spleen, and looks and, and still sees that there is a significant defect in the ability of this mutant virus to accumulate in these splenocytes. And so turns out that RNA degradation during lytic gamma herpes virus infection is involved both in in trafficking of the virus from the lung to the lymph and in establishing latency within these spleen cells, probably B cells within the spleen. So how could this be happening? We started to think about— well, we knew that this MusOx protein was obviously involved in downregulating the vast majority of host messenger RNAs. What about viral RNAs? We get asked this all the time. How do they escape this phenotype? And so, as I mentioned, lots of viruses promote host shut-off by a variety of mechanisms. And in almost all the cases that have been studied, viral RNAs have a specific way which they can avoid this, because why would they be subject to turnover at the same rate that host messages are subject to turnover? So in gamma herpesviruses, you have sort of this cascade of gene expression beginning with immediate early genes right after the virus infects the cell, followed by delayed early genes, which are pre-viral DNA replication, and late genes, which are post-DNA replication. In these viruses, these genes come on with different kinetics but then basically stay on throughout the life cycle of the virus. And so we predicted that probably genes in any of these three classes would need to escape. And so now that we have this host shut-off defective virus, we could actually directly measure the half-lives of viral RNAs in this mutant virus that couldn't promote RNA turnover versus the wild-type virus. And so this work was done by a postdoc in the lab, Karen Clyde, and much to our surprise, when she started picking some genes from individual classes, she found that in all the cases she looked at, the viral RNAs all had significant dramatically reduced half-lives in the wild-type virus compared within the virus that could not promote messenger RNA turnover. So it turns out there isn't a broad escape mechanism, at least, and viral RNAs are also subject to downregulation by this phenotype. And so she wondered, is this just an effect of some of the RNAs and just the ones we happened to randomly pick? Or is this— how global is this phenotype across the Mhv68 transcriptome? And so to answer this question, we established a collaboration with Laurie Krug at Stony Brook who had generated a tiled microarray across the Mhv68 genome, allowing us to look at all the RNAs on both strands. And you don't have to be able to read any of these just to get the basic idea that, in fact, when we compare the levels of viral RNAs in a wild-type infection versus an infection with a Delta host shut-off virus, you can see that the vast majority of viral RNAs are in fact downregulated in the wild-type infection. So MusOx broadly influences viral transcription, transcript accumulation, as well as host transcript accumulation. There are some things that escape. Turns out a lot of these look to be non-coding RNAs, or RNAs that are transcribed from intergenic regions. This would make sense if we think that RNAs have to be translationally competent to be targeted. And we think that there are some other coding genes that can escape as well, but this is highly selective. So just to give you one example, it turns out the major lytic transactivator, called RTA, within the virus, which is essential for driving virolytic replication, is one such escapee. So Karen always finds that the half-life of the RTA message is identical in viruses that are defective for host shut-off versus wild-type. And she can actually take the RTA coding region out of the context of a virus and just co-transfect it into cells with MUSOX versus the MUSOX mutant or GFP as a control, and it's not downregulated at all. In fact, sometimes it's even upregulated a little bit, suggesting that this gene is directly refractory to turnover. So there are examples, viral examples, and we know of cellular examples as well where genes are refractory. But for the most part, viral gene expression is heavily controlled by this phenotype as well. So how might this tie in then to our observation that you have a defect in trafficking and latency establishment? Driven by this lytic process? Well, we have some clues as to how this might be happening, although mostly still at the hypothesis stage with a little bit of data. One of the things that we were thinking about is, well, we and anybody else who works with MHV-68 knows that this virus is a pain to plaque. And the reason that it's a pain to plaque is that you don't get nice homogeneous large round plaques with it, you get plaques that are completely variable in size. Some of them are large, a lot of them are really teeny, some of them are medium-sized. So counting plaques with these viruses is sort of an art form. What Justin noticed though is that with his delta host shut-off virus— actually, I have these swapped, sorry— with his delta host shut-off virus, he found that plaques are in fact much more uniform in size. And very, very much easier to count. They tend to be larger. He can actually plot the frequency distribution of this and sees that with his Hoshitoff mutant, he generally gets larger plaques. They tend to be more of the same size. And so we thought, well, is this providing us a clue about something? And the idea that we came up with is, well, maybe you have such heterogeneity in plaque size with Mhv68 because you're not having a synchronous reactivation, and so the cell— infected cells are not all engaging in lytic replication at the same time, that some of them there's some sort of a delay, they're temporarily perhaps engaging in a latent-type infection before they're reactivating, whereas potentially because you have dysregulated viral gene expression and cellular gene expression, and much higher levels of lytic viral antigens in cells where you have when you're lacking host shut-off, you get a more synchronous drive into the lytic replication cycle. And so to test this idea, what Justin did is he took a— he acquired MHV-68, which is marked with YFP, so it's constitutively expressing yellow fluorescent protein. What this allows us to do is to— when you infect the cells now, you can see any cell that is expressing YFP, you know, is infected. Either latently infected or lytically infected. And then he can co-stain for lytic antigens like Museox to identify the percentage of infected cells that are detectably entering lytic replication and see is the ratio of lytic infection to total infection different when you have a host shut-off defective virus compared with wild-type. And in fact, this is what he consistently sees, is that he can do this by IFA or flow and for a number of MOIs, but the trend is always the same, that there is a higher percentage of cells that are detectably lytically infected when he's using the mutant virus, which cannot promote RNA turnover, compared with cells infected with a wild-type virus, which can promote RNA turnover. And so we think we have this disruption perhaps in the lytic-latent ratio that's going on here, and so maybe you're disfavoring latency in the host as well, and that's why you're not getting efficient establishment of latency and potentially trafficking, which we think may need to occur in latently infected cells so that they're not cleared. So in conclusion, we see that the SOX protein broadly downregulates both cellular and viral gene expression, that the RNA turnover function appears largely dispensable for productive infection, at least in the lung, but important for subsequent trafficking to the lymph and latency establishment, and that we think there may be a possible involvement in maintaining the appropriate lytic-to-latent balance, although we certainly need to do a lot more work to confirm that. And so we think there are still— there's several possible roles that this viral protein could be having during a viral infection, one of which could be immune evasion. It can be that through downregulation of immune modulators. We certainly know that RNAs coding for MHC proteins and interferon and lots of other immune modulators are downregulated, as well as decreasing the overall levels of viral antigens may make these cells less able to be detected and cleared by the immune system. So it could be a general immune evasion mechanism. None of these are mutually exclusive. Exclusive, by the way. As I showed you, the host shut-off could also be important for maintaining an appropriate lytic-latent ratio. Any herpes virus, when it infects a cell, has a decision to make. Does it engage in a latent program, or does it engage in lytic replication? And so there's probably tons of factors that go into that decision, cell-type-specific things, et cetera. But one of those factors could be appropriate balance of viral and cellular antigens. That play a role in latency establishment or driving the lytic cycle. And then finally, it could still have a role in viral replication, but in a cell-type-specific way. So we don't see an obvious role in 3T3 cells in culture, and we don't see an obvious role in lung epithelial cells. But it could very well be defective for lytic replication in certain cell types like B cells or dendritic cells or macrophages, and we would like to to begin to look at that to deconstruct specific types of cells that it might be involved in. And so with that, I'll end. And I think I mentioned that most of the virus work was done by graduate student Justin Richner. And the measuring viral RNA levels was done by a postdoc in the lab, Karen Clyde. And these are two undergraduates, Tina Wang and Benson Cheng, who participated in this. And all of the mouse work was done in collaboration with Laurent Koskoe's lab because we are not immunologists, and he's teaching us a lot about that together with his graduate student Andrea Pesda. And with that, I'd be happy to take any questions.

Hi. I wondered whether you'd consider the possible role of some of the innate signaling molecules, like the DNA sensors or things like that, or RNA sensors, obviously, that could be activated or inappropriately activated under these conditions. Right.

We do think about that, and that's sort of, I think, What we're envisioning is the next step as we try and deconstruct how is it that this virus is not getting to where it's supposed to be or doing what it's supposed to be doing in the host. And one of the obvious things is, are these things getting cleared? If you have too high levels of viral RNAs in cells, are they getting sensed more in a more robust way, and that's leading to clearance by the innate immune system as well? And that's sort of the next thing that we're looking to check out.

Britt?

Leonard?

Yeah, thanks. I just wanted to ask, do you now discount the possibility that the virus is trying to upregulate its messages relative to the host because the virus downregulates its own messages, or is it still possible— if the viral transcription is relatively higher?

Yeah, so for the resource reallocation type model, it's still possible, right? I mean, you can envision that while the polysomes are getting cleared of host messages and the virus can tolerate some loss and still be translated perfectly efficiently. And so I think that's a possibility. But again, it seems like an inefficient way of going about doing it. So alpha herpesviruses, for example, that model is thrown out there because they have also host shut-off through RNA degradation but by a different protein, VHS, and it's thought that that can happen there too. But in this case, the viral protein is itself then subsequently downregulated and turned off, and we have no evidence that SOX's activity is tempered throughout the course of infection. We do think that, you know, it's of course possible. These RNAs are certainly getting translated. It's just that they're also getting turned over. So it could be a combination of things. I don't think we can formally exclude that. All right, we'll move on.

Thank you.