Scientist Stories: Herbert "Skip" Virgin, Mammalian Virome in Health and Disease Pathogenesis

Good afternoon, everyone. I'd like to welcome everyone who's here and is watching from abroad or wherever you are watching the videocast to the Rola E. Dyer Lecture, established in 1950 in honor of the former NIH Director, Dr. Rola E. Dyer, who was a noted authority in infectious diseases. The lectureship features internationally renowned researchers who have contributed substantially to medical as well as biological knowledge of infectious diseases. My name is Brian Kelsall, and I've had the honor of introducing this year's speaker, Herbert W. Virgin IV, who we like to call Skip, and everyone else knows as Skip. Skip is the Edward Mallin Crawley Professor and Chairman of the Department of Pathology and Immunology, Professor of Molecular Microbiology, and the Professor of Medicine, as well as the Director and a Principal Investigator of the Midwest Regional Center of Excellence in Biomedical and Emerging Infectious Disease Research at the Washington University School of Medicine. Skip grew up in Florida, where he actually sailed competitively and where his father and he were in multiple races together and where he became the real Skip. We found out last night. So Skip was an undergraduate at Harvard. He graduated magna cum laude in biology in 1977, went to Harvard Medical School where he got MD-PhD with his advisor Emil Yonanaway, ended up doing residency in internal medicine at the Brigham and Women's Hospital, a postdoctoral fellow with Bernie Fields, who many of you know is one of the giants of virology. And then went on to do a fellowship in infectious disease at Washington University in 1990 and has been there ever since. He's really had multiple academic appointments and has recently become the professor of pathology and immunology in 19— in 2002 and became the Malenkoff professor and chair of pathology and immunology in 2006 and the professor of medicine in 2008. Dr. Virgin is a world expert in the field of immunology, immunology, virology, and infectious diseases. His laboratory studies the interaction of the host, pathogen, and symbiotic microbes in the pathogenesis of acute and chronic diseases using several experimental models of infection with DNA and RNA viruses, including herpes simplex virus, a lot of— a number of studies with murine gamma herpes virus 68. And murine norovirus, together with genetic, structural, computational sequencing methods, his laboratory has really defined the mechanisms of many viral and host genes in viral pathogenesis, latency, oncogenesis, vasculitis, and recently inflammatory bowel disease. In addition, they discovered the first murine norovirus, developed the first culture system for a norovirus, And they also showed that persistent norovirus infection can trigger intestinal inflammation in genetically susceptible mice which have a hypomorphic mutation in an autophagy gene, ATG16L1. And this really established the concept that cooperative roles of viruses— viral infection and genetic susceptibility can result in the development of inflammatory diseases. More recently, they have developed next-generation sequencing technologies to define the intestinal virome and have really been pioneers identifying potential roles for virome in diseases such as inflammatory bowel disease and HIV enteropathy. In recent work, they've also focused on the role of interferon and autophagy proteins in innate and adaptive immunity, pathogen persistence, and protein secretion. In fact, they were the first laboratories to identify non-canonical roles of autophagy proteins, in this case in innate immunity, B-cell development, panacell development, and in cellular immunity to intracellular pathogens. Finally, they have recently provided fascinating evidence for transkingdom metagenomic interactions in regulating immunity. He's recently published 5 or 6 really brilliant manuscripts that really have described multiple interactions between organisms in regulating immune system. And this has included the demonstration that chronic herpes virus infection can actually be beneficial to its host in stimulating innate immunity and to provide innate immunity to other infectious agents such as bacteria. They've also described the— that helminth infection can reactivate chronic viral infections. Through a mechanism involving cytokine action on viral replication. So given the fact that we all harbor a vast virome in the form of latent infections and commensal viruses and bacteriophage populations, I think Dr. Virgin's studies have broad implications for understanding the pathogenesis of susceptibility to chronic inflammatory and infectious diseases. Dr. Virgin has published over 200 papers and chapters. He's won numerous awards and honors, including a Burroughs-Wellcome Young Investigator Award in virology. He was a Mallinckroft Scholar. He was a fellow in the American Association for Advancement of Science. He's a member of the American Association of Physicians, a member of the American Academy of Microbiology, and was recently inducted into the Henry Kunkel Society. Not only that, he actually performs a number of roles both at WashU and abroad and in other institutions in the United States. He has been the— on steering committees for the Pasteur Institute, for the Rogen Institute at MGH, Harvard, and MIT, and he's also been on the steering committee of the NIH Genome Centers for Infectious Diseases. He's been an advisory board member for Stanford for the Human Systems Immunology Program, as well as others. He's really a very well-recognized expert in virology and its interactions with the immune system. So finally, I should mention that in addition to science, he's very committed to education and performs a number of roles at Washington University, administrative roles as well as teaching., and mentoring. And this is evidenced by his awards for mentorship and teaching at WashU, as well as his contributions to the MSTP and other training programs. As you'll find out, he's a very engaging, enthusiastic person and colleague and speaker. And I really— it's a joy to have him visit the NIH and to present the Dyer Lecture today.

Skip. Am I— you can hear me? Not 100% sure who Brian just introduced. But it is indeed a great honor to be here and to speak in the name of Dr. Dyer. As you'll find, I love infectious diseases and this is one of those things that you get to do. It's a particular honor to be at the NIH. To give this lectureship. So it's much appreciated. Plus, in each of the meetings today, I have learned something new that I did not know before. And that's why we do this, right? So I'm going to today make what's really a genetic argument. And I'm going to say some things that may be controversial to those who think that all the genes that matter to us are on our own chromosomes. And I'm going to make an argument that you've heard before in terms of the bacterial microbiome, but I'm going to try to add the viruses that live on and in us as genetic elements that create the composite organism that we are, and make the argument that some of the dark matter of inheritance and the fact that we don't always understand everything when we analyze a chromosomal gene may actually just be because you're ignoring some of the genes that are in you that don't happen to be on your own chromosomes. So that's the argument that I'm going to make today. And I'll start by telling you how we got there, and I'll summarize two studies which gave us the idea that we should then pursue an avenue of research towards understanding how viruses that are present in or on us might influence our fundamental biology and immunity, and then I'll tell you a series of more recent stories which are news that are published, and then I'll end with a set of unpublished data which I think is interesting in regards to how we think about species differences between immune responses. And so that's the outline of what I'll try to accomplish today. So everything is about the people who work in your lab. And we have groups working on viral persistence, the virome, which I'll define, and inflammatory bowel disease, a number of different genes. And I'm going to refer to the work done by these people. But I want to start by telling you that I have 3 people who are on the job market right now. Donna McDuff, who published this eLife paper showing herpes viruses can complement immunodeficiency. Jason Norman, who published this paper in Cell showing disease-specific alterations in the enteric virome, particularly bacteriophages. Tim Nice, who found that there is such a thing as sterilizing innate immunity. I used to think that B cells and T cells were required to get rid of viruses. Turns out, at least in mice, that's not true. And this very smart young man discovered that. Megan will come out on the job market next year. She has found that bacteria control viral persistence in the gut, and also that inherited changes in the bacterial microbiome control IgA and the immunologic phenotype of the host. So these are the kind of young people who, if you can, you should hire. So the viral. So this is from a review in 2009. This is the number of people on Earth. At that time, about 7.75 billion. This is a series of virus infections and this is the proportion of the total people on Earth predicted to be infected with these viruses. And you immediately see that there are a number of viruses, amongst them the herpes viruses, that are in fact present in almost everyone. And in fact, in this room right now, each of you has around 10 of these viruses. They are inside your body. You will never get rid of them. There are things like EBV. If I were to swab your throats, most of you, I'd be able to detect EBV DNA in the swab from your tonsils, for example. And it's occurred to us based on the two stories that I'll start with that we haven't thought enough about this. Because we know that when you immunocompromised a person, for instance with AIDS, you can get CMV retinitis. So the CMV is in equilibrium with the immune system. In fact, all of these viral infections are in equilibrium with the immune system. But if that's true, then this is a dynamic interaction. These elements are permanent to us and they're in constant motion stimulating the immune system. And since being recognized by the immune system and since cytokines made by B cells, T cells, other innate immune cells don't actually just focus on infected cells, it's possible that there's substantial effects of this dynamic equilibrium on the host itself. And that's the concept that we have come to and that I'll be talking to you about. And then I'll talk to you about what might be regulating some of these things. Is it only our genes or is it also genes that would be present in other organisms that inhabit us? So this is the concept, that there is a virome And that the virome is individual to different people and it's constantly in exchange with other mammals, birds, bats, et cetera, and that it evolves very quickly. That there are eukaryotic viruses, endogenous viral elements such as retroviruses or retrovirus-derived elements, prokaryotic and archaeal organisms, and that they interact with each other in transkingdom interactions.— as a virologist, forgive me, I'm going to call viruses alive for the sake of this argument. And that this in turn, these interactions interact with genetic variations to define our genotype phenotype relationship in terms of systemic inflammation, the immunophenotype of the host, which is the basal immune system we think of, how the basal activation of the immune system works when a new antigen or a new agent comes in, that would be the immunophenotype of the individual at the time of challenge. Disease susceptibility, I think some of these things are clearly good for you, so they're mutualistic symbiosis, and perhaps substantial interindividual variation would come from the interaction of these viruses and these genes, and of course, their traditional viral pathogens which actually cause disease in the majority of people who are infected with them. So that's the concept that we've come to test. So how did we get there? So we thought about the virome of changing the immunophenotype and changing the relationship between the phenotype of the individual and the genotype based on two studies that were published about 7 or 8 years ago. And one of them, a virus we discovered, a chronic norovirus, interacted with an autophagy gene to change the transcriptional pattern of epithelial cells and generated phenotypes in mice that actually look just like a subset of patients with Crohn's disease. And I'll show you a few slides from the study to introduce it. And in another, we found that chronic herpes virus infection could change gene expression in the host and actually protect against bacterial infection and tumors. And so in each of these cases, we had a chronic virus altering transcription leading to very substantial new phenotypes that did not exist just with the mutation or just with the organism. And that's what's made us think about this idea that there's this interaction with the host and the virome that changes biology. So some preliminary data, some preliminary slides of those data. So this is the study here. It's a very simple experiment. We took mice and we latently infected them with a gamma herpes virus shown in red here. And then we challenged them with a very high dose of Listeria monocytogenes, an intracellular bacteria. And this bacteria killed mice except when they were latently infected. And this lasted for many months, out to 6 months it's been observed. And in fact, we being viral geneticists had engineered versions of this virus by deleting a gene so that we could get acute infection but not latent or chronic infection. And this effect disappeared. So this was chronic infection protecting the host. This is a class of viruses which most of you have 4 or maybe 5 of. We think it's reasonable to consider the possibility that that chronic infection might alter your susceptibility to bacterial disease, for example. The mechanism was published, but I'll just outline it. It turns out that this chronic viral infection sets the innate immune system by stimulating interferon gamma secretion. Activates macrophages. We also showed for NK cells in other studies. And those activated macrophages protect against infection. So this is the virus stimulating the immune system through interferon gamma to protect the host. If you do a very simple experiment and just take an animal and latently infect it and just harvest the organs and gene chip them, this is liver, spleen, and brain. The first 3 are from the virus that came can establish chronic infection, the next 3, the mutant that's only acute, chronic, acute, chronic, acute. And just look at the genes. You can immediately see that the latently infected animals have substantial changes in just basal gene expression. And this is different in the liver, spleen, and brain. And in fact, if you look at these genes, there are several genes which have been linked in GWAS and other studies to susceptibility to human diseases, celiac disease, Crohn's, multiple sclerosis, that are regulated by these chronic infections. So what this says is that we as transcriptional entities are not the same when we are chronically infected, and yet we are all chronically infected by multiple viruses. So this leads to the idea that maybe these changes in gene expression have very important phenotypic effects on us. So that's an example of the virome regulating the immunophenotype of the host. The second story is this chronic norovirus. Noroviruses are the cause of about 95% of the epidemic non-bacterial gastroenteritis in the world. They haven't been cultured. We were fortunate enough to discover a murine virus, develop a culture system, reverse genetics and such. So we have a nice animal model and we were studying autophagy genes and we found out that we got this interesting Crohn's disease-like phenotype when we combined this virus with this mutation. So this is a virus plus gene interaction. This is a combinatorial concept. The idea here is that you don't get the phenotype with just the gene or just the virus. It's only the two together. This is the way it looks. Barrier-raised mice which are mutated for the autophagy gene have normal Paneth cells. These are the Paneth cell granules. They get colitis when you give them DSS. And they have normal small intestine. This is the ileum. But if you now take this mutant animal and infect with the virus, but not if you infect the control animal with the virus, you get abnormal Paneth cells, much changed response to DSS, including some phenotypes that look more like Crohn's disease. And now we have a completely new phenotype, which is ileal atrophy. So if you were looking for a genetic effect on ileal atrophy, you would only see it in the presence of this chronic virus infection. And that pathology, it turns out, if you now look in humans with Crohn's disease who are wild-type for ATG16, they have these nice panacells, but the humans with the mutation in ATG16 have the same paneth cell abnormality that we found in the mouse. So think about what we did. We took mouse with a gene mutation, normal paneth cells, we added a chronic infection, we got a phenotype which is duplicated in humans. And here is shown just the staining for those paneth cell granules. You can see the abnormalities inside the cell here. So those are the two concepts. An interesting similar message is that if you look at transcription in the presence of the virus in wild-type versus the presence of the virus in these mutant animals, you get substantial changes in transcription, and it isn't just quantitative. This is for sets of genes, for instance, protein localization. You get complete inversion. Things go up instead of down. So what that means is that our transcriptional identity can be influenced in a combinatorial fashion between the virus and a gene. So that's an example of the virome interacting with a host genetic variation to cause a pathology. And before I go to the newer data, I want to respond to a question that I was asked in a meeting in Argentina, which is, Dr. Virgin, isn't that terribly complicated? And the answer is yes, but it's not my fault. Okay? This is the way biology is. And if we try to look in silos at one picture, we're not going to get the whole picture. So I know it's complicated, but it's not my fault. So with regard to the immunophenotype, if a virus could make a wild-type animal more resistant to a bacteria, occurred to us that a chronic virus infection might actually be able to complement an immunodeficiency. And since people with immunodeficiencies have broadly variant phenotypes, we thought that maybe the presence of a virus could alter such a phenotype. And so we've collaborated with Jean-Laurent Casanova's group. They identified this gene called HOIL in two families which had three affected children who died from a disease which was both immunodeficient and hyperinflammatory. We obtained the mice for this, and I'm just going to summarize some recently published data. These mice are very immunodeficient. They die when infected with toxoplasma. They die when infected with listeria. But they're the only animals we've ever studied which have a rather unique phenomenon, which is that they are resistant to Mycobacterium tuberculosis. Resistant to chronic herpes virus infection, limiting the reactivation of this virus. So this gene seems to control the relationship between inflammation and immunodeficiency. But it allowed us an experimental tool to test a hypothesis. So this animal doesn't die from chronic herpes virus infection, so we could look at the effect of this on the genetic susceptibility. And so the mechanism of this gene is shown here. It's responsible in this case after bacterial infection for generating cytokines such as TNF, IL-12, IL-6. Knockouts are in red here. It's at the transcriptional level. Each of these genes is essential for resistance to Listeria monocytogenes in a mouse. So basically in the absence of HOIL, you don't make the cytokines that protect you from Listeria infection. Now the reason we did this experiment came here. The first 3 patients that were identified with Hoyl deficiency had this disease that I referred you to, plus a myopathy. But then a number of other patients were published who had mutations, many of which looked like null mutations, but they didn't have the same phenotype. They all, including the mice, have the myopathy, but they didn't have the immunodeficiency and inflammation. So there are a lot of explanations for this. Maybe these data are not correct. Maybe there's another gene that modifies this, but the other possibility is that maybe these patients and these patients have a different environmental exposure and these patients are lacking something that these patients have that complement the immunodeficiency. And so we developed the hypothesis that chronic herpes virus infection, to which these mice are resistant, will complement that severe immunodeficiency and generate autoinflammation, the second phenotype. So we did a very simple experiment. We took the knockout animals infected with the herpes virus and we looked for cytokines. Sure enough, these animals, when chronically infected with herpes virus, are hyper-inflammatory. Okay? But they are not hyper-inflammatory in the absence of the virus. And then we did the key experiment from our point of view. We took a number of different knockout animals and we asked, can the virus complement in trans the genetic immunodeficiency? And here's data for the Hoyl knockout. Let me take you through it. These are animals that died. The Hoyle knockout animals die. The wild types die. If you chronically infect the Hoyle knockout, you can complement. If you do the same thing for IL-6, the virus complements. And in fact also for caspase deficiency not shown here. So we can make an animal resistant to a virus by many orders of magnitude that to a bacterial infection by chronic viral infection. It makes us wonder whether some of the variation in clinical presentation of genetic immunodeficiency might be due to elements of the virome. So that's an example of phenotypic complementation of a genetic immunodeficiency by chronic herpes virus infection influencing the genotype phenotype relationship in that animal model. Now, I introduced the idea, and Brian mentioned, what controls the virome? Is it just our genes and the immune system, which we all know and love, or are there other elements? And so we developed the hypothesis that transkingdom interactions, and by metagenomic I mean including all of the sequences in or on the host, will alter viral pathogenesis and immunity. And we really did that because that virus plus gene phenomena that I showed you that was the ATG16 plus the norovirus, we could actually cure that with antibiotics. So bacteria were required for a virus phenotype that only occurred in the presence of a host gene mutation that really suggested to us that bacteria were interacting with viruses in a manner controlled by the host immune system. And so in each of these stories, it turns out that exploring this concept has led us to some interesting discoveries. So with regard to the interaction between a virus and an autophagy gene, it turns out that the bacterial microbiome actually controls viral persistence. And that's a mechanism by which antibiotics can change chronic virus infection. I'll show you those data. And we found that helminth infection substantially altered chronic herpes virus infection, including— I'll show you some human viral data. And it's through a cytokine. And so since these helminths are amongst the most common infections of humans on Earth, as are these, it's potentially important that there would be an evolutionarily conserved genetic interaction between the inducer of IL-4 and IL-13 and a viral promoter. And that's what I'm going to tell you. So the concept here is that the virus is controlled by the worm through something we call cytokine competition at a viral promoter. And I'll explain that. I want to presage the argument with the fact that the ancestral herpes virus is shared between birds, reptiles, and mammals. And that means that the herpes viruses have been studying you far longer than you have been studying the herpesviruses. There has been coevolution of herpesviruses with each species from the time that the species separated. So when mammals speciate, their herpesviruses come along with them and coevolve. So I'm going to show you that that gives the chance for the virus to do some pretty remarkable stuff. So we did a very simple experiment. We took a virus which was engineered when it was latent, quiescent in the host. If you reactivated it, it would express luciferase, and we could just measure the amount of reactivation of the virus by imaging the amount of light that came out of the mouse, and then we infected with the helminth and we looked to see what happened. And the helminth infection reactivated the virus. You can see this here. This is 7 days after the helminth infection. There is significantly more light coming from the the animals that have this virus. So this is reactivation of the virus. We showed it by other methods as well. So the worm is reactivating the virus from latency. So our hypothesis was that there was something about the immune response from the worm that actually triggered the virus and that therefore IL-4, which is induced by helminth infection as part of a TH2 response, might increase MHV replication. I won't show you all the data, but testing this hypothesis led to the discovery that both IL-4 and IL-13 actually make the virus grow better. That this requires the transcription factor STAT6, which is downstream from the receptor for these. And that from a virologist point of view, very interestingly, this is because the viral gene 50, which is essential for growth and reactivation, it's a switch gene basically. When the virus is latent, this gene is off. When the gene turns on, the virus emerges from latency and makes infectious virus. Turns out that these cytokines turn on that gene. Okay? And that's the mechanism we believe by which the cytokines reactivate the virus. And that's shown here. This is just expression of gene 50 from the virus. And if you treat— this is the expression of the gene without cytokines. If you put interferon gamma, which blocks reactivation, you can inhibit the expression of the gene, but IL-4 added here counters. IL-4 counters. So this gene somehow is sensing both host cytokine that turns it off and a host cytokine that turns it on. So how does that work? Well, it turns out that there are 5 different promoters in the virus for this for this gene, gene 50, and one of them is directly responsive to IL-4. So this is not the immune system controlling the virus. This is the virus acting as a sensor by having its own promoters actually respond to host cytokines, and we show that this is through STAT6 binding using ChIP assays. If this is a true general phenomenon, we should be able to reactivate a human herpes virus with IL-4. And we should be able to use in mice both cytokines and get a greater effect. And what I mean by that is we should be able to— if we inhibit reactivation with interferon gamma, anti-interferon gamma, we should see limited reactivation. If we treat with the animals with IL-4, limited. But if you do both, you get huge amounts of reactivation. This is now a log scale. So it means that the virus is actually a two-signal model. It's using a promoter to sense one thing to turn it off, one thing to turn it on. And so it's actually learned how to sense the immunologic milieu of the host. And with regard to the human virus, if we add IL-4 to cultured cells with Kaposi sarcoma herpes virus, we induce the expression of similar genes and the reactivation of the virus. That was done in Rolf Rennie's lab at the University of Florida. So this is an evolutionarily conserved mechanism by which part of our viroma herpes virus actually senses the immunologic milieu of the host. So helminth co-infection reactivates through a transkingdom interaction between a worm and the virus, and that same virus controls the immunophenotype of the host and resistance to bacterial infection. We also at the same time working with David Artis found that this helminth infection, same infection, actually inhibited the CD8 T-cell response to another chronic virus. So these interactions are very complicated. The same worm reactivated a herpes virus and inhibited the CD8 response to our norovirus. All right. Second transkingdom interaction story. The bacterial microbiome controls chronic norovirus infection. There are two effects here. One is that the bacteria are actually required for this virus to establish chronic infection, and the other is that a specific cytokine and cytokine receptor are involved in that, and we can leverage that to cure chronic virus infection. So this story starts with our discovery of the murine norovirus. Turns out that human noroviruses, a certain proportion of people chronically shed these viruses. For long periods of time. This explains how the virus can cause epidemics even though there's no animal reservoir. Some proportion of you in the audience or out wherever else you're watching this lecture are probably shedding chronic norovirus infection, and occasionally someone who's shedding the virus makes the food for a picnic and you generate an epidemic by contaminating the food, or a cruise ship, for example, or a military base. All of them are susceptible to epidemics of this virus. And it's probably from the chronic carriage amongst a certain number of us of these viruses. When we discovered the mirroring viruses, we discovered some viruses in orange here which can persistently be shed in the stool and other viruses which could not. This virus, the blue virus here, it's not an attenuated virus, it actually spreads systemically goes to the spleen and lymph nodes. This virus doesn't. So this virus is adapted to the intestine. And that gives us a model for understanding this persistent carriage of a norovirus. And that's where the story begins. We did a very simple experiment. And that is we just treated with antibiotics for 14 days and then infected with the virus. During the reception afterwards, you can ask me why we would ever do this experiment. There was a good reason. It wasn't just luck, I don't think. But what happened was when we treated with antibiotics, the virus which would normally become persistent was actually cleared. And here is shown across many experiments, many of the animals are cured. They don't get this persistent infection. Even though they clearly have the virus early, they don't get the infection. Something about the antibiotics is preventing the virus from establishing a chronic infection. That's a transkingdom interaction, bacteria to virus. Is it the bacteria? Well, we just did a simple experiment again. We stopped the antibiotics and we fed back fecal pellets. We did a fecal transplantation and we restored— this is untreated, antibiotics, fecal feedback. Restores the virus in colon, lymph node, ileum, and shedding in the stool. So it's really something in the fecal material that the antibiotics are removing. And to understand the mechanism, we decided we would assess the requirement for host genes in this effect. We took a broad range of knockout mice and did the same experiment. And these are all genes which are not required for the antibiotics to work. This is not projecting. That's the interferon alpha beta receptor, gamma receptor, both the interferon alpha beta and gamma receptor, all adaptive immunity in Rag mice is missing, Mighty 88, toll-like receptors, sensors for the virus, autophagy proteins, transcription factors. The antibiotics are fine. But there were 3 genes, the interferon lambda receptor, STAT1 and IRF3 where the antibiotics did not work. So what this means is that the immune system is somehow involved in this transkingdom interaction. That led us to do the following experiment. Interferon lambda is what is also called type 3 interferon. It's less well studied. It's been used to treat hepatitis C in humans. It's less well studied than type 1 and type 2 interferons, interferon gamma, interferon alpha, beta, and the other type 1 interferons. It signals through a similar molecule, STAT1, another one of the genes required in our experiments. And so we just took type 3 interferon supplied by Bristol-Myers Squibb and treated before infection and the animals do not become persistently infected. So you can pretreat with interferon lambda and the persistent infection does not occur. And then we did what I think is the key experiment. We waited for the persistence to be established. So this is now day 21. And the virus is shedding. This is a wild-type animal. And we treated with interferon lambda and we could cure the persistent infection. In an animal lacking the type 1 interferon receptor, it still works. But in an animal lacking the interferon lambda receptor, there's no effect. So interferon lambda can cure a persistent virus infection with this norovirus. Now, I was always taught that the innate immune system controls infection until T cells and B cells can come into the fray and clear an infection out, and that sterilizing immunity is due to the effect of T and B cells. But one of the mice that we'd studied in the antibiotic treatment was a Rag knockout. Which suggests that the innate immune system actually can do whatever it is that we're studying here. And so we did the following experiment. This is Tim Nice's key last experiment in his paper. We took persistently infected Rag knockouts and we treated them with interferon lambda and cured them. This is by PCR, but we also took the fecal material here and put it back into STAT knockout and Rag knockout mice. So we think this is an actual cure of a virus infection in the absence of T and B cells. And that suggests to us that there is sterilizing innate immunity, which suggests in turn that we might be able to induce this kind of immunity to eliminate virus infection even when T and B cells are compromised. It's an interesting concept to us. So these are examples of transkingdom interactions where commensal microbes and a host molecule interfere on lambda. Are responsible for controlling chronic norovirus infection and where we can then discover something new about the immune system by leveraging the initial studies of the transkingdom interactions to identify what parts of the immune system might then in turn be involved in controlling virus infection in the intestine. Now, if we're right about all this, we should be able to demonstrate changes in the virome in disease states. And so we began to look at that in human studies. And we performed a study together with Miles Parks in the United Kingdom, Ramnik Xavier and Dirk Gevers at MGH, and Ali Keshvarzian and E.C. Mutlu at Rush, where we looked at the enteric virome in patients with inflammatory bowel disease. And this cohort is particularly notable. I won't go into it in detail, but because they have a remarkable clinical scenario there, setting there, they could get us household controls, and the data is much more powerful when you do shotgun sequencing or metagenomic analysis of patients compared to household controls, because a number of studies have shown that our microbiomes and viromes are actually related to the people that we live with. This was a particularly powerful cohort. And it has been long published that patients with inflammatory bowel disease have abnormalities in their bacterial microbiome. In fact, these patients have decreased species diversity and species richness compared to controls. So there's something not complicated, decreased complexity of their bacteria. And so the first thing we did in our Cambridge, Chicago, and Boston cohorts is look at the number of species and the diversity, and we showed that both for Crohn's disease and ulcerative colitis, purple and red here, we got exactly what everyone else had published. Okay? So this is bacterial 16S data, and these patients have an abnormality of the bacterial microbiome. Their bacteria are less complex. So then we purified virus particles from the fecal material from these patients. We deep sequenced them. I'll be happy to talk about the bioinformatics of this. But what we found was the following, that these are household controls, these are ulcerative colitis patients, and these are Crohn's disease patients, and this is the number of different species of bacteria. So the bacteriophages are becoming more diverse and richer in— there are more viruses even when there are less bacteria. We believe that it is a stretch that this decrease in bacterial diversity could explain the presence of greater numbers of different bacteriophages. In fact, this signature is disease-specific in that the viruses in ulcerative colitis are different than the viruses in Crohn's. And this repeated to a greater or lesser degree in all of our cohorts. So across the ocean and across 3 different cohorts. So there are bacteria which are changing in a downward direction and viruses which are now greater in number. Is there a relationship? So we developed an approach to take the discriminant bacteria, these are the bacteria in the UK cohort which correlate with disease. And put them statistically against the top 30 or so viruses, and a red dot means there's a negative correlation. That means when there's more of this phage, there's less of this bacteria, and so on. A blue is a positive relationship. These are the household controls. And then we did Crohn's disease, and what's really striking is that the pattern of relationships here and here are completely different. So there is, in fact, a signature of the relationship between these bacteriophages and the bacteria, which is in Crohn's disease compared to controls, and in turn, ulcerative colitis is different than either. So these data demonstrate that there is a correlation between the enteric virome and a human disease. The way we are thinking about this is shown in this picture. This is a picture of a virus. Okay. Here's another virus. And this is a bacteria. And another bacteria. We think that there could be a predator/prey relationship between the bacteria— between the viruses and the bacteria such that maybe inflammatory bowel disease is actually a viral disease in some senses. That bacteriophages either emerge or are introduced, that that changes the bacterial microbiome into a new equilibrium state and that that may in fact be how the viruses might trigger inflammatory bowel disease. This is purely hypothetical, and we're in the process of setting up animal models to try to understand this. Now, that's an example of a transkingdom interaction between a virus and a bacteria. So we've seen worms interacting with viruses. We've seen bacteria interacting with viruses in a positive sense where bacteria are required for persistent virus, and this is a negative relationship we think between a virus and a particular group of bacteria. And so that we think is the first example of a relationship between the virome, in this case bacteriophages, and a human disease. What do we think are the implications? We wonder whether this is relevant to other diseases where the bacterial microbiome has been identified to be a risk factor. Type 1 diabetes, cardiovascular disease, nutritional deficiencies, have all— there's been extensive data— Type 1 and Type 2 diabetes— there's extensive data relating the bacterial microbiome to these diseases, but no one's looked at the viruses yet. And so we wonder whether that's a good idea. We also think that probiotic therapies added into the intestine might fall prey to the same bacteria which— bacteriophages which we identified. And speculative, but I wonder whether one could make provirotics. That would be viruses which you could put into a host to manipulate the bacteria in a way that was useful for the health of the host. And maybe that could correct some of the abnormalities in bacteria that we see in different diseases. So again, we're trying to develop animal models to look at these ideas. So coming back to the immunophenotype, so I made an argument that chronic infections can alter the fundamental nature of the immune system. Now I'm going to show you some unpublished data. This is one of these experiments where I've always wanted to do this experiment and it was merely a matter of me finding someone I could convince to do it. Because I tried to convince a number of people to do this experiment, and no one was brave enough except for Tiffany Reese, who published that worm paper and who is now a new faculty member at UT Southwestern on the tenure track, and she deserves credit for doing this experiment. Now, there is a bias that mice are not humans, and I completely understand that mice look very different than humans. But there's also a bias that the mouse immune system is not reflective of the human immune system, and therefore human immunologists at times look down on the mouse model because they say the mouse model is not reflective of human immune responses. I cannot get a prediction from a mouse model that is relevant for my human vaccine study. And so I understand that concern. But what if there was a way to make mice look more like humans? Instead of transplanting human tissues into them, altering their environment. So now think about what we've done with experimental mice. We put them in ever cleaner environments, and then we have made the data from them very reproducible. The same thing happens every time, more or less. But at the same time, we've removed them from all of their environmental triggers. But all of us are sequentially infected by herpes viruses and worms and other things. So is it possible that the reason that the immune system in a mouse is different than the immune system in humans is in part not just the chromosomal abnormalities but something about the environment of the mouse? So that's the hypothesis. The idea is that serial coinfections that mimic the early life exposure of humans might alter the murine immune response to vaccination. That's the first hypothesis. We'll come to the second hypothesis after this initial data. So we did a truly massive experiment with a lot of mice and 4 different replicates, and we left some mice uninfected and then we just sequentially infected mice with different pathogens, a gamma herpes virus, a beta herpes virus, influenza, and a worm. And then we waited various periods of time and then we gave yellow fever virus vaccine. 17D is the strain of virus in the yellow fever vaccine. We bled the mice at day 0 and we bled them 3 days, 7 days, 21 and 34 days after vaccination. We picked this vaccination for a reason that will become clear in a moment. And we asked firstly the question is, is the immune system different when the mouse has had a history of being infected with some of these chronic viruses and other pathogens? Firstly, if you take the animals, this is mock or co means co-infected, these are the animals that were sequentially infected. They all make a nice immune response early, but the co-infected animals lose their antibody responses over time. And so we just looked at what genes are expressed in response to vaccination. And this is the mock. And so if you were to do this with mice from a clean mouse facility, give them 17D yellow fever, on day 3, 318 genes would be up, 142 down. Etc. This is what a non— a control mouse would look like. If you take the co-infected mice, genes are changed, but there is negligible overlap. So what this figure shows is that the immune response of the mouse is completely different if it had those environmental infections sequentially when it was growing up. Okay. Now, is this more like a human immune response? It could be that those serial infections would alter the mouse immune response so that it looked more human. So that's the second hypothesis. Coinfection-induced changes will include the expression of genes related to human vaccine responses. So if we looked at the genes in the co-infected mice versus the non-co-infected mice, we would find a resemblance between between humans and the co-infected mice that was not seen with the mock-infected mice. So let me introduce you to this kind of a figure. This is a figure looking at the genes which are— let me, sorry, come back to the cohort. I'm so excited about this I jumped ahead. This is a collaboration between Bali Pulindran, Rafiq Sikaly, and Ali Falali. Rafiq has two cohorts of humans vaccinated with yellow fever virus. One is in Entebbe, Uganda, and one is in Lausanne, Switzerland. And in each, they did exactly the same experiment that I showed you in that they bled the patients at day 0 and then day 3, day 7, and then sequentially afterwards and did array analysis to look at the gene expression patterns. So we have the mouse data and now we have human data from 2 different cohorts. And we're going to ask, is there a relationship? So this is Entebi day 3, subtracting out baseline on day 0, and we're looking at the genes that are co-infection regulated on day 3 in the mouse, and the graphs are all going to be the same. These are genes that are up in both cohorts, down in both cohorts, and are inverted, and you can see a very significant overlap between the co-infected regulated genes and the genes expressed in patients in Uganda after vaccination. If you look at the day 0 data, this is corrected for day 0, the genes expressed in Lausanne on day 0 are highly related to the co-infected genes Lausanne on day 3. And if you now look at this set of genes in a network analysis, you discover that the genes which are here are things like STAT1, TRAFs, chemokines. So this is actually the mouse immune response being made to overlap with the human immune response. Similarly, if you look at in Tebian day 7 and Tebian day 3 and you look at the network, again you find STAT1, STAT2, IFITs, HLA molecules. What we're seeing here is that the immune response of the mouse was made to be more similar in response to a vaccine to the immune response in a human by serial coinfection. So that is an example we think that the genotype phenotype relationship in this case by measured by a vaccine response may be altered by coinfections. So today what I've shown you or I've tried to argue to you is a genetic argument. That when we look at our own genotype-phenotype relationship, that we need to account for the virome. That in fact, there are additional interactions in the metagenome between bacteria and viruses, between worms and viruses, between viruses and bacteria, which can influence or be related to inflammation or human diseases. That host genetic variation plays a role. That when we're mutated in a gene, it could be that the expression of that mutation is not solely defined by its presence on our chromosomes in homozygous or heterozygous state, that it may take a trigger or an element from the virome in order to have that host genetic variant actually generate a phenotype. I've given you multiple examples of how the genotype phenotype relationship can be changed. We can complement immunodeficiency, we can create new pathologies that mimic human pathologies. I've argued you for each of these. Systemic inflammation changes the immunophenotype, a relation to disease susceptibility. Symbiosis of viruses, it's not all bad news. Interindividual phenotypic variation in profound changes in the transcriptional milieu when you have a chronic virus between or you don't in multiple organs, the brain, the spleen, the liver. And I've shown you that multiple different elements of the virome can play a role. And I think we're in the very infancy of this field. The number of sequences we can actually annotate with current databases is very limited, and so I think there's a huge opportunity to define the virome and its relationship to these important pieces of biology in the future. I'm going to end by telling you that Megan Baldridge was involved in the studies of bacteria controlling chronic virus infection. Sunmin Park was involved in the complementation studies along with Donna McDuff. Tim Nice found sterilizing innate immunity. Jason Norman and Scott Handley with able assistance from a number of others were the the authors and the discoverers of the relationship between the enteric virome and inflammatory bowel disease. This is why I love this job. These are smart people. They challenge you every day. They don't believe what you say and they try to disprove you. And in so doing, you discover sometimes exciting things. So I'm very indebted to them. And with that, I'll stop and be happy to take any questions.

Thank you very much. We have a— thanks for the great talk. We have time for a couple of questions, and then I want to make sure everyone knows there is a reception in the library afterwards if you'd like to talk more with Skip.

Julie? So Skip, really nice talk. And so you talk about it like there's two two worlds, there's the viruses that infect bacteria and there's the viruses that infect the mammalian cells, and I wonder if you're thinking that there's going to be some community, some assembly in some way that they sense each other, or is it really through who they interact with rather than their recognition of each other? Yeah, so I don't have any data with which to answer that, so that gives me permission to explain my bias. I think that these things are all going to interact with one another. It's impossible for me to believe that viruses that change the inflammatory milieu of the host do not then in turn change bacteria and that that doesn't in turn interact with other. So I think that there will be a very complex and integrated matrix. And I believe that this is an evolved process. These viruses have evolved with us, as have the bacteria. So it would be very surprising if certain interactions hadn't developed genetically which allow those community interactions that you're talking about. A simple answer is that we haven't published this, but there are a few papers in the literature that show that host cells, our eukaryotic cells, can actually sense bacteriophages. So you could also argue that the bacteriophage can interact with the virus, the bacteria, but it could also be interacting with the host. And it was actually shown, I believe, here in the 1960s that bacteriophages can be immunogenic, studies of phi X. So I think that there's a lot yet to be discovered, but my bias is that there's going to be the network that you described. Yes. With respect to asthma and the hygiene hypothesis, I've seen that there are contrasting points of view in terms of exposure, and some people say that leads to it, some say it's the other.

They talk about Th1 and Th2.

Does your work in any way relate to this?

I know you've talked about inflammatory bowel disease.

People talk about that and the microbiome, et cetera. So I guess I'm trying to say, could you shed some light on how this relates to asthma and the hygiene hypothesis? So I think that this is the hygiene hypothesis, except what I'm saying is that in addition to the things that one would normally perhaps have thought about when the hygiene hypothesis was originated, that the viruses which live in us permanently and on us are a part of that interaction between organisms that sets up our immune system, which is the heart of the hygiene hypothesis. If we change that, we could change our immune responses. That could lead to inflammatory disease. We have no direct data to support that hypothesis. It is a very— for viruses. It is a very attractive hypothesis. There are clear changes in the epidemiology of these chronic viruses that occur if you compare what we think of as westernized countries to non-westernized countries. When you compare environments in which one is more exposed to the natural world versus more sterile from the natural world. All of those things do in fact change virus infection. So I think it's the same general idea. And I'm saying that you should think about viruses, but I don't have any data which would specifically say that a certain viral infection occurred early versus late would predispose or not predispose to asthma. Thank you. Well, looks like you covered all the genomes. So— and also you showed the beneficial effect and also adverse effects of this combination. So when do you think you will make some cocktails for people who want to take a cruise? So the question is, if I may paraphrase, if it's that complicated, how is it possible that you're going to be able to get a specific change in the host with one particular approach to altering the viruses? I think that's a real challenge. I think that probably the bacterial microbiome is the best early target. For doing this. But I would point out that even in primate studies, there are magnificent data showing that certain viral vectors, CMV, can induce very valuable immune responses when other viral vectors do not. And that some of those viral vectors, CMV, the work of Lewis Picker and others, are actually chronic viruses. And it's probably their chronicity that allows you to get that particular response. So I'm not— I don't think that we could today design a cocktail of viruses which would get rid of asthma or change your susceptibility to IBD, but neither do I think that there is no hope that specific results could occur if one could just understand the fundamental relationship between the virus, its host cell, and then the tissue for which it has tropism. So when do you think we will have fecal transplantation for Inflammatory bowel disease and Crohn's disease? So this is not an area that I work in. I can tell you that I've just come from a Keystone Microbiome meeting, and there was a lot of interest in data in fecal transplantation. I think you're probably aware the data on human fecal transplantation in C. diff colitis is nothing short of stunning. So far, there have been some trials of fecal transplantation in IBD which have been less than impressive, but it's also possible that that is evolving, and some people believe that they have positive data. So it's interesting. I don't think there's anything published that says that we should drop everything, stop immunomodulatory drug therapies, and fecally transplant people, but it might actually have a role.

Congratulations.

Good luck. I was wondering about the demographics of the bacteriophages. And have you looked at cohorts of family members and whether or not they have similar profiles, or do they differ? Or taking it further, if I go to Africa, I imagine I would have one profile, and if I come to the United States, I would have another, if I go to India, another. I'm just curious to know. This is a really great question. And the answer is that there are profound differences in the data if you compare Chicago, Boston, and the United Kingdom. Now, I'm going to— the caveat to that statement that I just made is that at the current level of capacity to annotate viruses, given the weakness of current databases, okay, it could be that some of that apparent variation is due to failure to fully annotate the sequences, and there is no question that the bacterial metagenomic world is leagues ahead of the viral metagenomic world, and in my view, given data like this and data from others, that difference needs to be fixed. We need to be able to annotate these things as well as we can annotate bacteria today. Now, the most interesting difference is between Boston and the other two sites, and there are more bacteriophages in people in Boston. Now, I have a number of colleagues in Boston, and as a virologist, we occasionally get the permission to name a virus, and I'm strongly considering naming the extra viruses in Boston chowderviruses for the fact that these are people who eat clam chowder. And maybe they have chowder viruses. But it's actually kind of striking. If you look at the controls there, I don't have an explanation for why that might be, but there are definitely going to be geographic variations.

How about family members?

Like if you have a patient— So that's been done in some recently studied work, and it is clear, and we have some other data that's unpublished, it is clear that we share not only our bacteria, with our family members, but also our viruses with our family members. And you know that because, you know, children get CMV or chickenpox from patient, you know, grandparents with zoster, and so you know that there is this kind of transfer. But that clearly happens with bacteriophages as well, and with the environment. If you go into a household and you sequence the environment, door jambs, you will find that the family microbiome and very likely virome are significantly controlled by the immediate family environment.

Thank you. I guess I have a related question to IBD because of the diversity of bacteria you see with IBD. Have you looked at family members? Is that a source of the diversity that precedes the disease, or is this something that develops as you Brian asked a very good question.

So in the cohort where we have more viruses and we have household controls, what is the overlap between the viruses and the two? And our data is— it's a really great question. Our data is not robust enough to answer that question. Because what we annotate is not the full genomes of viruses. What we annotate are sections of the virus. And so in order to really know whether I share a virus with my daughter or my wife, we would actually have to sequence the virus and show that unique sequence variations within the overall sequence of that virus are shared between the two of us, and our data is just not robust enough to do that at the current time. As we get bigger and bigger datasets, this is a key question.

Because if your hypothesis is really that the diversity or your predator/prey relationship, there's something that relates to that diversity, which then relates to the prey, right? So what is it that creates that initial diversity if it's not for some other diversity within bacteria and you're getting less diversity?

Well, one, first of all, there's clearly a genetic element to the risk for Crohn's disease, for example. And so one could argue that all the people in the household were exposed to the same viral milieu, but the genetic susceptibility selected out certain patterns within the patient who was predisposed to the disease. But it's also possible that one person ate at a salad bar that the other person didn't eat at and got different viruses introduced into them. Our data can't address those two variant hypotheses at the current time.

Thank you, Skip, for the excellent presentation. It's amazing. Each time I attend your presentation, I learn something new, and it's great. Thank you. The question is about— you know, we have heard or read bacteria in the gut can translocate into the blood, and which may be associated with inflammation, maybe associated with aging and other things. I'm just wondering if you could comment on like we have got so many like viruses in the gut, could this same thing happening with viruses as well or not?

So first of all, the NIH is the source of this hypothesis. Danny Dweck's group was the first to really publish that. That's a paper that we paid a lot of attention to. We fundamentally confirmed those same findings in primates with SIV infection advancing to AIDS. What we found with regard to the virome, there are many hypotheses by which you would get translocation into the systemic circulation and that would drive inflammation. You could change the nature of the immune response to bacteria. That's one good hypothesis. What we found when we deep sequenced the primates, and this was published now a couple years ago, is 21 novel viruses in the primates that were advancing to AIDS. And they are all— they were never discovered before. You wouldn't find them by PCR necessarily. And they were from groups of viruses which are known pathogens. Some of the viruses that we discovered could actually be seen in lesions in the gut where there was an abnormality and a break in the epithelium and we could stain and we could show that a novel adenovirus was actually present in the lesion. So our hypothesis would be that some translocation is due to the presence of as yet undiagnosed viruses that are present which actually may be acting in a more classical fashion. They have a tropism for cells in the intestine. They cause inflammation. They cause breakdown of the barrier. And some of the translocation that you get could be due to the presence of viruses. Viral diagnostics in humans are wonderful things, but they're very targeted to sequences that are in described viruses. If you do metagenomics in those settings, you frequently find the presence of viruses that would be missed by standard PCR-based diagnostics. So you really can't, in my view, fully diagnose, if you will, the enteric virome with current diagnostic tools, and so I think metagenomics, basically shotgun sequencing and careful bioinformatics, is a good way to know whether there's actually a virus there. Our standard diagnostics, you do the diagnostics, you don't find it, you say there's no virus there. I don't think that's accurate. Thank you for your talk. Do you have any data linking cancer susceptibility and the viromes? For instance, cancer due to oncogenic viruses or is it other type of cancer? It's a really good question. The question is whether there's a linkage between the viruses and cancer. We've not done those studies. And so it's an attractive hypothesis. We'll be happy to send people the methods for how to do this stuff.

We'll take one more question and then I think we'll move to the—

So you've mentioned this difference between the ability to annotate bacterial— bacteria versus viruses. It's got to be partly due to the lack of marker gene in viruses like 16S. So how do you currently do something where you want to, given a data set, identify all the viruses? So there's a very— there are two parts to this question. And it's one we think about a lot. So is the reason that it's difficult to annotate viruses because viruses do not share a common gene that can be targeted by conserved PCR as is done for fungal 18S or bacterial 16S? So the answer to that is in part yes. But it is much more than that. Bacteria, which differ by 5% at the nucleotide level or 10% at the nucleotide level are very different from each other. Viruses which vary by 5% within a quasispecies are the same virus. Okay? So the real problem is that the databases are not— don't have enough viral sequences to fully annotate. You cannot do viral metagenomics by nucleic acid homology. It cannot be done. You can identify known viruses that way. But viruses vary so much from each other, even within a single individual, that you can't use strict conservation of nucleic acid. That means you have to do protein-based searches, and that's a lot of what we do. Okay? So not enough viruses, no common gene as an element, you need to do protein, okay? And the last thing is, with due respect to my bacterial colleagues, The bacterial genomes which go into the databases annotated as bacteria contain prophages. Okay? So if you take a prophage, if I make purified viruses from the gut, I get some eukaryotic viruses, I get some prokaryotic viruses. If I search the prokaryotic viral sequences, the database says they're bacteria. Okay? They're not bacteria. They're phages. They're just matching phages in the bacterial database. So there's a problem with the annotation of the databases itself, because they are sometimes not completely annotated for viruses. We're working hard to solve these problems. The last thing, and I think it's a really important point, and it's a good one to end on, not all life is encoded by DNA, okay? Not all life is encoded by DNA. There are RNA organisms. When you see I studied, I defined the microbiome with DNA sequencing, you need to read I did not define the microbiome by doing only DNA sequencing. All of the major viral pathogens, most of the major viral pathogens in the gut of humans are noroviruses, rotaviruses, astroviruses. And adenoviruses, so there's a DNA virus down there. Okay? So we really have to get away from this idea that sequencing DNA is the way to get at the metagenome because there are RNA phages, RNA viruses, and the methods are there. I think that it's just easier to make DNA than it is DNA and RNA, and I think that's a major flaw in many current studies of the virome and the metagenome itself. So thank you for your attention. That was great. Excellent.