(Click on image for a larger version)
I am going to talk about Epstein-Barr virus (EBV) vaccines, and in the short time available I am going to pick three of what I think to be the key developments, and even some old approaches which I think are coming back into vogue, and I will review some of the issues that generally apply to the business of making antiviral cancer vaccines – many of which were very clearly presented by Ian Frazer earlier.
This slide shows an EB virus. It is a herpesvirus: a very large piece of double-stranded DNA, 170 kilobases. It has 11 latent genes and 70 or 80 lytic cycle products. It is a very, very complicated beast indeed.
The main point I want to make at the beginning is that you should regard EBV really not so much as an ‘infectious agent’, like influenza or HIV. It is a virus that peacefully coexists with us, and has done for millions and millions of years since we were swinging around in trees. It has co-speciated with us. You find homologues in Old World monkeys and even in New World monkeys. So it is part of us. One has to have that perspective in thinking, ‘How are we going to design vaccines to combat the ill effects that take place in a tiny minority of people?’ – but that tiny minority is a very large number of people, as it happens, because almost everybody on the planet, above the age of 3 or 4, is infected with this virus. So I think you need to have a slightly different perspective when asking how we are going to develop a vaccine.
You might even ask, ‘Why do you want to develop a vaccine?’ Well, I hope to make that clear in the next few slides.
(Click on image for a larger version)
Let’s stick with normal EBV biology. I have to say that since 1964, when Sir Anthony Epstein discovered this virus in a Burkitt’s lymphoma – which actually was massive news at the time, that here was a virus in a human cancer, probably the first virus associated with a human cancer – our understanding of the behaviour of the virus in vitro in the target cell, the B-cell, is very extensive, and probably very misleading. Our understanding of the real biology of this virus in us is pretty thin on the ground after all these years.
Our understanding at the moment is that it is an oral infection. Fair enough, one person transmits it to another by the oral route. What happens at the oropharyngeal epithelium is still basically a complete mystery. For many years it was thought that the virus could replicate in epithelial cells, but if you look at normal healthy epithelium in an infected individual you can’t find any virus there. Nobody has ever found virus in these cells, and that is a bit of a paradox.
One way or another, the virus gets across that layer and infects its real target cell, the circulating B-lymphocyte. And what happens then, in vitro at least, is that the B-cell is transformed. This is an oncogenic virus; it transforms or immortalises the B-cell. One assumes that in vivo similar things are happening, and that in the normally healthy individual these now transformed and infected B-cells, which are expressing quite a range of the latent antigens, are perfectly good targets for a healthy immune system and these infected cells are eliminated very quickly.
But a very small number of those infected B-cells escape the attention of the immune system, because they are not expressing EBV products at all, as far as we can tell – the so-called Latency 0 infected B-cell. These are B-memory cells, and they hang around, essentially, forever, and are quiescent and not doing anything and represent the reservoir of our EBV infection.
From time to time, by mechanisms that are not understood, when that circulating latently infected B-cell is in the region of the oropharynx, it may be stimulated to enter lytic production of infectious virus, which then is excreted in the saliva and infects another unsuspecting person.
(Click on image for a larger version)
Our understanding of EBV biology is, I think, fairly limited. This becomes apparent when you start asking how we are going to vaccinate to stop the diseases associated with the virus. I have listed here the principal conditions that are associated with EBV. The ones that are of most importance, really, are the three in the middle – nasopharyngeal carcinoma, half of Hodgkin’s lymphomas, and what looks like at least 10 per cent of all gastric carcinomas have an EBV association.
Obviously, the others are not unimportant, but I am not going to talk about them. I have never included multiple sclerosis in my talks until recently, but I think the evidence linking EBV infection with this condition has substantially increased in the past couple of years. I won’t be talking about it today but I think it is now worthy to be included in the list of potential conditions that might be dealt with, with an EBV vaccine.
(Click on image for a larger version)
These are the headings I am going to address, and I am referring to experiments that were actually started more than 20 years ago by Sir Anthony Epstein himself. In fact, I was doing my first postdoc in his lab all those years ago, and I think they have become important again. They were always important but they are becoming more important now, especially when we bear in mind the huge boost to morale that we have received from the Ian Frazer-HPV story. I am not saying that that strategy is applicable to EBV; it is just that if you have worked in viruses and cancer for all these years it was good to hear what Ian Frazer had to say this morning.
Let’s start with gp350.
(Click on image for a larger version)
This is the major envelope glycoprotein in the virus. In the old days, when our immunological appreciation was relatively simple, our modest ambition was to induce neutralising antibodies in vaccines. We thought, ‘Well, that will take care of everything.’ Of course, that didn’t turn out to be true, but the target of virus neutralising antibodies in healthy individuals is this molecule, this envelope glycoprotein gp350. So the very first vaccines were designed around this viral product.
(Click on image for a larger version)
We isolated the natural product, we made all sorts of different recombinants and replication-defective adenoviruses, vaccinia viruses, recombinant subunit products, a whole range of adjuvants, including ISCOMs (immunostimulating complexes), and engaged in a series of immunisation studies in this primate, the cottontop tamarin, which is susceptible to EBV-induced B-cell lymphomas, rather like the B-lymphomas that appear in iatrogenically immunosuppressed individuals.
It is a model that has limitations, but it is the only animal model of EBV that is available – or was available then. The upshot was that we could protect these animals against EBV-induced B-cell lymphoma by all of the various gp350 products that we generated. One of the puzzling bits of information was that antibodies did not seem to be necessary. Never mind neutralising antibodies; this was a cell-mediated protective response in these animals. Furthermore, of course, the tumour cells only have a latent profile of infection. They don’t produce virus or structural components of the virus.
So you are wondering what the mechanism might be here. I think one can only conclude that the protective mechanism is that the effective dose of the virus is reduced by the existence of an immune response, so that the reduction in the dose of virus given to induce tumours goes below that threshold which in fact can induce tumours. I believe that is the mechanism of protection in this particular circumstance.
(Click on image for a larger version)
Some years have elapsed. I must say that we had a product as early as 1985 that could probably have been put in human trials, but we had to wait a lot longer for anything to happen. GlaxoSmithKline took an interest in this whole area, and relatively recently – in 2004 – they told us about the effects of a phase II gp350 subunit vaccine trial in seronegative adolescents in Western Europe. And a most interesting result this is, indeed.
Although EBV infection was not prevented, these people did not develop glandular fever. This is really the ideal outcome. I don’t think we want to stop people being infected with EBV. It is likely to open a can of worms. We are normally infected with EBV; it cohabits with us. It is probably doing some good. Don’t ask me what good it is doing, but I suspect that in evolutionary terms it is there and we are happily coexisting with it for perfectly good reasons. So this would be an ideal result, I think.
What will happen in terms of EBV-associated tumours is a completely open question. We have no idea what might happen if people were immunised in this way. You will see, as the talk develops, why this is quite an issue.
Most recently we have completed a phase I trial of this type of vaccine in paediatric organ transplant recipients, who have a high incidence of B-cell lymphomas, and at this stage all I can say about that is that appropriate immune responses developed in these children and we hope to proceed to a phase II in the not too distant future.
(Click on image for a larger version)
Change in EBV immune status?]
To take an overview of the reasoning behind doing gp350 vaccinations: I think there are one or two pretty insoluble problems here.
Most people are infected very early on. In the Third World it is really in the first year of life. And then there is a time period of anything between 40 and 60 years before any tumour or nasopharyngeal carcinoma (NPC) is going to develop. All kinds of things are likely to be happening during that long period of time, but one point that is worth making is that what is actually diagnostic of the onset of NPC is a rise of serum antibodies and mucosal antibodies against structural proteins in the virus. This either reflects an increase in virus replication and a corresponding immune response, or some change in the immune status of the individual which may allow the emergence of the tumour; who knows? We don’t know what the answer to that is.
So if someone said to you, ‘I’m going to let you make a gp350 vaccine and give you many millions of pounds to do a human trial on 250,000 subjects in southern China, and you’ll have to wait 50 years for the result,’ you might not be too interested in taking that approach and I would agree. Although this goes against certain immunological dogmas, I think that there is scope for considering a post-infection, post-exposure vaccination in the age groups that are susceptible to nasopharyngeal carcinoma. Although, obviously, the results in papillomaviruses indicate that that doesn’t work in that situation, EBV and HPV are from different planets. They have completely different life cycles and strategies, and this comparison may not be useful.
(Click on image for a larger version)
So what next with this particular approach? Obviously, we hope that much bigger trials are going to take place on infectious mononucleosis. I don’t know whether that is happening. You would have to ask GlaxoSmithKline whether they are doing that or not. I suspect they are considering it, at least.
We are looking at phase II trials in paediatric organ transplant patients.
Of the rest of the things you see on the list here, the appeal of a Burkitt’s lymphoma trial in West Africa is that you would only have to wait about 12 years before you got an answer. So I think that strategy may be re-emerging. It is an idea that Epstein had, all those years ago. I think it is re-emerging as a real possibility, and I think efforts should be made in that direction.
Again, NPC trials in southern China: I think post-exposure vaccination is a possibility. I think it unlikely that a prophylactic vaccination is on the cards, for the reasons that I have already outlined.
As far as Hodgkin’s disease, gastric carcinomas and multiple sclerosis are concerned, I think they are pretty much in the same situation. It is too early to say whether one could justify resources on the scale required to take these forward.
There are other developments taking place in this type of vaccine. There are transformation-defective viruses being made and also live virus vectors expressing gp350, notably a varicella-zoster recombinant – another herpesvirus that was developed years ago, actually, by myself, Elliot Kieff and Ron Ellis – which may, given the events of recent years, come back into vogue. We will have to see.
(Click on image for a larger version)
Now let’s completely change the subject and talk about vaccines that you could actually develop and that would, hopefully, work within one’s own lifetime! That is to say, if a patient presents with nasopharyngeal carcinoma, could we have a therapeutic vaccine that would bump up that person’s immune response to at least offer some scope for reducing the tumour or even getting rid of it?
The next few slides I am going to show are the work of Alan Rickinson and Graham Taylor, and colleagues in Birmingham and Hong Kong, and have absolutely nothing to do with me at all. I was asked if I would show these slides, and I am quite happy to do so, but I must say it is worth reminding ourselves that, as anybody who has ever given a talk on somebody else’s work would know, it is much trickier than you think, because when somebody asks that killer question you are suddenly faced with issues that you had never really given any thought to. So I’m ready for pretty much anything.
(Click on image for a larger version)
Anyway, to give it a bit more background: listed here are the target infected cells that you might encounter in an EBV-infected person.
At the top we have the reservoir cell, the B-memory cell – no protein expression, not a target.
The next one is a bit of a freak, really. It is the Burkitt’s lymphoma phenotype – only the EBNA1 latent antigens expressed.
Latency II I will talk about in a second.
And the Latency III infected B-cell is the transformed B-cell that we think a healthy immune system gets rid of fairly readily.
Then there is the cell which is actually engaged in lytic production. These are in a tiny minority.
(Click on image for a larger version)
We know an awful lot about CD8 cytotoxic T cell reactivity against some of these targets, from in vitro studies, and amazingly little about what might actually be happening in vivo. The significance of particular antigens or particular types of CD8 T cell and so on is still pretty foggy. But this is basically the picture.
We know that there are strong CD8 T cell responses against the Latency III phenotype and lytically infected cells. We are very confident about those, but much less confident about CD8 T cell activities against the other latency phenotypes – more about that in a second.
(Click on image for a larger version)
Much more recently CD4 T cells have come back into the picture, because people have finally realised that they provide essential help, not only for antibody production but for longevity of CD8 T cells responses, and also that they are actually killer cells in their own right, in certain circumstances. We have certainly shown that with EBV lytically infected B-cells. But again with the actual targets, the other latency phenotypes, it is much less clear what is happening in real life.
These latent antigens EBNA1, LMP1 and LMP2 have, in my view, intrinsic properties which interfere with their own processing. That is absolutely crystal clear with EBNA1, and less clear with LMP1 and 2 but pretty likely. It is very difficult indeed to actually isolate T cells which are specific for LMP1. We know that EBNA1 has a structural feature which prevents its proteasomal presentation through MHC class I, although EBNA1 actually contains many CD4 T cell epitopes, and we certainly believe that LMP2 has intrinsic properties which interfere with its own presentation.
(Click on image for a larger version)
Focusing on the important EBV tumours, this is the phenotype that is presented: EBNA1 LMP1 LMP2. And we know these antigens have problems with being processed and presented in the class I and class II pathways.
(Click on image for a larger version)
You can take two approaches, really, two sides of the equation, for thinking about dealing with these tumours. You can say, ‘Well, let’s make much more, much better CD8+ and CD4 T cells by vaccinating a person.’ There is good reason for doing this, because we already know that if you grow ex vivo EBV-specific T cells and put them into transplant lymphoma patients, you can achieve a great deal in removing or preventing the tumours in question. But you have to set that against the other side of the argument: that the tumour antigens themselves may not be such great targets. So you could have large numbers of very, very good T cells swimming around but unable to set their eyes on the tumour cell targets because they are still, to all intents and purposes, invisible. There are two extremes of the argument.
(Click on image for a larger version)
The first bit I am going to deal with is the generation of CD4 and CD8 T cells. This is the work entirely of Alan Rickinson and Graham Taylor, and colleagues in Hong Kong, so I would like you to bear that in mind as I try and explain my way through it.
Essentially what they have made is an MVA vaccinia vector containing a truncated EBNA1 and a full-length LMP2 molecule. They have made a fusion protein.
There are a couple of features of these proteins that are worth mentioning. First of all, all the transactivation functions and the structural feature that prevents proteasome processing have been removed in the EBNA1. The CD4 T cell epitope region remains. The LMP2 is exactly the same as the normal one, except there are a couple of residues in the amino terminus which normally interact with Syk and Lyn kinases and transmit one kind of signal or another. These have been taken out. So, as far as the known signalling activation functions are concerned, these have been eliminated in this fusion protein.
This is expressed in Modified Vaccinia Ankara, which by all accounts would be a good vector to use to vaccinate individuals who have nasopharyngeal carcinoma, gastric carcinoma.
(Click on image for a larger version)
One of the most interesting things that came out of this was that the processing of EBNA1 now has been changed by its association with LMP2. If you look at the top series of images on this slide, you see that EBNA1 is a nuclear protein, restricted entirely to the nucleus. You see a marker for early endosomes, EEA1, which is dotted all over the cytoplasm. And the two essentially, EBNA1 and that marker, do not co-localise at any point. This is what you would normally expect. EBNA1 is not expressed in the cytoplasm and is not, for one reason or another, processed through the proteasome.
But what you see using the EBNA1 LMP2 fusion protein is a different distribution. You find that not all the EBNA1 is located in the nucleus, and there are some concentrated dots in the cytoplasm. And when you look across at the overlay you find that quite a lot of these dots are actually co-localising with the early endosome. That doesn’t tell you that functionally things are different, and I am not going to show all the slides from Alan, but they have got operational read-outs in CD4 T cell activities which indicate that EBNA1 is now being processed through the MHC class II presentation pathway, although it is an endogenously produced protein. The original dogma – class I, class II, external, internal – is looking weaker and weaker, in my view, as time goes by. And here is yet another example.
(Click on image for a larger version)
This is an ELISPOT assay from one of the patients. There are two phase I trials going on with this MVA product, one in Hong Kong and one in Birmingham. This is one example (I have seen the others and they are very similar) from an NPC patient in Hong Kong.
On the left you see an ELISPOT against a variety of CD8 and CD4 EBV epitopes, and there is a background level of spot formation. At the bottom is a positive control with PHA, which is obviously a completely black plate.
If you look in a vaccinated individual, you will see that there is a very marked increase in CD8 T cell interferon responses to an EBNA1 CD8 epitope. That is pretty clear. Perhaps less clear, but marked, are responses against LMP2 epitope. But also against the CD4 T cell epitope PQC, again the responses are greatly enhanced. Again my impression is that these are backed up by operational T cell assays and it is not just an interferon ELISPOT phenomenon – because I do gather (although I am no expert on this) that ELISPOT is not 100 per cent reliable in indicating a functional cytotoxic T cell, by any means.
(Click on image for a larger version)
So let’s go to this version of a diagram I showed earlier. That is one side of it: let’s vaccinate somebody, boost their immune system, generate the right kind of effector T cells. But I did say that there might be an issue there, in that all the effector T cells in the world, if they have got nothing to look at, may not be a big help.
Now, this is just a hypothesis, of course, but if you take that extreme position you could then say, ‘Well, let’s do something about the tumour cell which makes it visible to the immune system’ – to the existing immune system, that is. There are several ways of doing this. You can induce these cells to go into lytic cycle, that is to say, to produce virus, which causes the cell to self-destruct. That is another subject which I won’t go into here, but it is certainly an approach.
Our approach in Bristol has been to ask: can we enhance the processing and presentation of LMP1 and LMP2 in these tumour cells, and make them much more visible to any T cells that happen to be in the vicinity?
(Click on image for a larger version)
I am going to wind up by explaining this.
(Click on image for a larger version)
The classic EBV target cell – that is to say, the lymphoblastoid cell line which expresses the range of EBV latent products – is commonly used as the target in assessing cytotoxic cells. And what we and others find is that if you load peptides onto these cells they are very, very good targets; you kill them very easily. If you express the latent product with vaccinia in these target cells, again they are very good targets. But if you just take the LCLs themselves, which express relatively low amounts of LMP2, they are actually rather poor targets and are not easily killed.
So here we have a situation which goes against the gut feeling that most people have and that LMP2-specific cytotoxic T cells don’t actually kill LMP2-positive autologous B-cell targets that well.
Of course, it is not an absolute truth. From this population of CTLs you can isolate, clone out, LMP2-specific T cells that do kill these targets. But it is an issue that needs to be faced up to, I think, and we have looked at this in the following way.
(Click on image for a larger version)
We are all aware that LMP1 and 2, in either plasma membrane or internal membranes, coexist in lipid raft-like structures which are rich in the ganglioside GM1.
(Click on image for a larger version)
Just by a complete coincidence, down the corridor from us in Bristol lived Tim Hirst, who was a bacteriologist. (He is actually here now, at ANU, but he was with us in Bristol.) He worked on this wonderful thing, this Escherichia coli heat-labile enterotoxin. It is made of two subunits, as A1B5 toxins are. The A-subunit does the business, and the pentamer shown below it, the B-subunit, delivers the A-subunit into the cell by binding to GM1, with an absolutely massive affinity.
(Click on image for a larger version)
Tim had created some recombinants of the B-subunit which still did all the business but had none of the toxic A-subunit. And it occurred to us that if we put this molecule into those lipid rafts which have LMP1 and LMP2, we could interfere with the turnover or the processing of these EBV molecules and have an effect on their susceptibility to cytotoxic T cells.
(Click on image for a larger version)
I won’t show the various confocal images of the co-localisation of these three components. You will just have to take my word for it.
(Click on image for a larger version)
But what was absolutely clear was that EtxB could now render autologous lymphoblastoid cell lines susceptible to LMP2-specific CTLs where they had not previously been so. And we have now shown this effect to be proteasome-dependent. It doesn’t matter whether epitopes are TAP-dependent or non-TAP-dependent. We know that it is class I dependent; it is a class I phenomenon. (We don’t know anything about class II yet.) It works for both LMP1 and LMP2 epitopes. And we have also recently demonstrated this to occur in epithelial cells – the tumour cells in question are, in fact, epithelial tumour cells.
So we have, I think, demonstrated a potentially therapeutic effect with EtxB. We are now trying to develop this in various murine tumour models where nasopharyngeal carcinoma can be actually grown on mice.
(Click on image for a larger version)
EtxB by itself has quite a large number of immunological effects of one kind or another. Tim has actually come up with a mutant where the so-called signalling functions of this molecule have been eliminated, so it is much more inert with respect to these other side effects but it still delivers, in terms of being able to traffic and enhance the visibility of these tumour antigens to antigen-specific T cells.
(Click on image for a larger version)
A simple view of this, really, would be that here we have an EBV tumour cell, we have LMP1 and LMP2 – actually, LMP2 is the predominant element in nasopharyngeal carcinoma, LMP1 is present only about 30 per cent – and we have EBNA1. These are the two antigens that are expressed in these tumour cells. We know that EBNA1 class I processing is poorly efficient, because it can’t get into the proteasome. We don’t know what is happening in the LMP1 and LMP2 processing pathways. It is not generally known how membrane proteins are processed so that they gain access to the proteasome. We don’t know what is happening there.
(Click on image for a larger version)
But we do know that the EtxB interferes with whatever that pathway is and allows increased presentation of the appropriate epitopes on the surface of the target cell.
(Click on image for a larger version)
I will finish with a brief summary of the main points I want to make to you. I think the original experiments done by Tony Epstein and myself in the primate model all those years ago should be looked at again, in light of the success of the GlaxoSmithKline glandular fever vaccine and current trials in paediatric transplant patients. But what is a problem with the development of any vaccine is the huge jump from experimental work into human trials, and I don’t need to tell anybody in this audience what a hurdle that is.
(Click on image for a larger version)
I think what is needed in the whole of the EBV vaccine area is a great deal more energy put into actually getting out of the lab into human trial situations, whatever the regulatory difficulties and production difficulties may be. The only information that is going to count is going to come from work in humans, and I would welcome any tips, advice, interactions, from anybody on this issue during the course of this meeting.
(Click on image for a larger version)
(Click on image for a larger version)
Discussion
Question: I am just wondering if you have considered the role of T-regs in EBV, since it is a chronic infection. Is that a barrier to developing a vaccine?
Andy Morgan: It might be a barrier; I don’t think it will be. Just recently Alan Rickinson’s group looked at the notion of T-regs in NPC patients, but it is not clear to me or to anybody whether these are actually having any effect at all. But it is an issue that is bubbling to the surface.
Question: Can you just share with us the evidence on MS and EBV which has suddenly got your interest now?
Andy Morgan: The old evidence that has been around for years is that there is a different serological anti-EBV profile in MS patients as compared to normal. There are higher serum antibodies against the lytic cycle products of the virus as compared to normal. Now, that could be for any reason and is not good evidence.
But, more recently, I think the key paper is a recent one from an Italian group that has actually shown that B-cells present in lymphoid tissue in the brain are infected with EBV. These are EBV-positive B-cells in brain tissue that you don’t find present in other neuro-inflammatory disorders. So I think that is reasonably strong evidence.
But also there is much, much more work coming out about CD4 T cell specificities induced by EBV which may be cross-reactive in terms of the targets of autoimmune response in multiple sclerosis. There is no proof yet but I think there is stronger evidence. I think the interest is being raised there.
Question: It is good to see the progress being made with a prophylactic vaccine for Epstein-Barr virus, Andy. If there were a successful therapeutic vaccine – a really successful therapeutic vaccine – against Epstein-Barr virus, how much of a B-cell immune system would we be left with?
Andy Morgan: It is something that I was going to talk more about, but obviously there is no time. Only a tiny proportion of B-cells are infected and these express no EBV genes at all. The only cells that will be eliminated are those expressing LMP2 so I don’t think that the B-cell system overall would be affected.
Considering the whole business of interfering too much in the balance of power between us and EBV, I think we should proceed with caution. I don’t really want to speak against my own interest, but the notion of having hundreds of thousands of people who are EBV-negative I think is something that we should seriously worry about. I don’t think that is a good idea. If EBV is actually tickling up our immune system in a helpful way, around the clock throughout our lives, we could be really causing some problems. But this is just speculation.
Question: I am showing my ignorance here, but is it known why EBV induces Burkitt’s lymphoma in Africa and so on?
Andy Morgan: And why not anywhere else?
Question (cont.): Yes.
Andy Morgan: The plausible argument is that small children in these parts of the world, in Africa and New Guinea, live in areas of holoendemic malaria and are almost always immunosuppressed for that reason, and as a result are more susceptible to the development of B-cell lymphomas. That is the argument.
Of course, you do have the problem of sporadic B-cell lymphomas. About 15 per cent of all lymphoma around the world is sporadic, and much of that is EBV-negative. Yet both types of cell have similar chromosomal translocations. The only argument I have is that you can reach the same destination by different routes, and that it doesn’t discount the link between EBV and these tumours.
The old work is very strong. The serological prediction that a child will develop Burkitt’s lymphoma is pretty convincing.Visit these websites
The Sir Mark Oliphant Conference Series is supported by the Australian Government under the International Science Linkages Program.
© 2008 Sir Mark Oliphant Conferences.
Website design acidgreen
