Could I take this opportunity of thanking the organising committee for inviting me to present our work on the polyepitope vaccine.
This sort of technology was developed by our group almost a decade back, and we have done a lot of work on the technology to exploit it, particularly for herpesviruses. The main types of viruses we have been working on are Epstein-Barr virus (EBV) and cytomegalovirus (CMV). Today I will show you some results on that and show you how we have progressed from early preclinical studies to a clinical trial and a series of trials which are also planned in the next few months.
(Click on image for a larger version)
I think one of the critical steps in the initiation of adaptive cellular response is the presentation of the antigen. I must say that a lot of vaccinology seems to overlook this very important step in the initiation of immune responses. We have spent too much time selecting antigens and formulating the antigens; we rarely look at what role the antigen presentation itself plays in the whole initiation of the immune responses. Eugene Maraskovsky very nicely highlighted that point. When the adjuvant and the antigen are presented in the correct way, it will be properly presented to the immune system and that is when you get an effective response at the other end.
The initiation of response can take place when the virus or a pathogen gets into the antigen-presenting cell, or you can induce it simply by vaccinating, thus transferring the antigen.
One very important point I would like to make here is that contrary to the mouse models, where the issue of cross-presentation has become such a big concept in vaccine development, our experience with the human studies, particularly with Epstein-Barr virus, has shown that, if anything, in human systems cross-presentation may not be the major component of the initiation of immune responses. I think that is where we have to be very careful. When you are developing any vaccine strategies for humans, studies done in the mouse model using the cross-presentation methodologies may not be applicable to the human system.
In a lot of examples we have seen here, the vaccines work in the mouse model or animal models, for HIV, for cancer vaccines, but when we take them into the human setting the same formulation doesn’t work any more. So we have to be very careful about translating those findings.
I think the major drawback of the mouse model has been that we have relied too much on the cross-presentation concept, and taken it into the human setting but found that actually it fails miserably in that setting. We need to think more carefully about what relevance cross-presentation has in the human setting.
(Click on image for a larger version)
Once these antigens get access to the antigen-presenting cell, either through direct infection or through other pathways, the antigens are then broken down and processed and presented through the class I or the class II pathway, and they activate the helper T cells and cytotoxic T lymphocytes (CTLs). These epitopes are crucial in activating the responses of these T cells.
(Click on image for a larger version)
Once these T cells are activated, the cytotoxic T cells in particular can directly lyse the virus-infected cells; and occasionally the helper T cells, besides helping the CD8 cytotoxic T cells, can also directly lyse virus-infected cells. So we have a double arm of the immune system working, both arms trying to clear these virus-infected cells.
We have done a lot of work with EBV, as other groups have done with CMV. In persistent viruses, in particular, the T cells play a crucial role in controlling latent infection. If you remove these T cells, you get an uncontrolled proliferation of virus-infected cells. The classic example of this is in transplant patients, where the immunosuppressive therapy which impairs this T cell function, particularly the CD8 T cell immunity, can lead to uncontrolled proliferation of B cells, for example in B cell lymphomas in EBV. And if you reconstitute the T cell immunity by adoptive immunotherapy you regress the lymphomas. (I will show you a few examples of that, one of which being an example we have done a trial with.)
(Click on image for a larger version)
As we have heard in many talks in the last day and a half, the T cell vaccine has been a major challenge. We have been struggling to make an effective T cell vaccine for the human setting. I think the reason is that in most of the strategies we have adopted we have been trying to take full-length proteins, and most of the time these are poorly immunogenic in terms of generating the T cell response.
The other thing is that, at least in the viral setting that we work with, the use of these full-length proteins is going to be almost impossible, especially in the setting of EBV, where many of these proteins have potentially oncogenic functions as well associated with them. Regulatory authorities are not going to allow us to use them.
Taking into account those limitations and constraints on the use of full-length proteins, we started thinking that another approach is to use polyepitope technology.
(Click on image for a larger version)
The other important issues you need to consider while designing the T cell based vaccine are two major components of this approach: the population and the viral diversity.
When you try to cover different ethnic groups around the world, you have to make sure that your T cell vaccine will cover as many HLAs (human leukocyte antigens) as possible, in different ethnic groups. And taking multiple epitopes restricted through multiple HLA alleles would avoid any problems and allow you to give multiple coverage.
The other problem with the full-length antigen vaccine arises because of the underlying viral diversity, and not only in terms of the different antigens included in the vaccine. You might also have another problem, such as we have seen in HIV or HCV or other viruses, where there is a genetic variation within different sequences in itself. So you end up with a vaccine which actually is effective against one isolate of one strain of the virus.
You need to make sure that you design a vaccine which would cover different antigens but also different genotypes, subtypes and variants. And you should include, potentially, T cell and CTL epitopes as well as helper epitopes.
So on both sides of these issues the polyepitope vaccine, we feel, covers that and provides an opportunity to address those issues.
As I said, our laboratory has been working on using this polyepitope technology on human cytomegalovirus and Epstein-Barr virus. I will present to you some of the data on these two viruses.
(Click on image for a larger version)
CMV is actually a human β herpesvirus, and just as with EBV most of us carry this virus as a latent infection. And once you get this infection, the infection persists for life.
There are two major risk groups for CMV which have been recognised by the Institute of Medicine for the development of a prophylactic vaccine. One infection risk group is transplant patients, particularly stem cell transplant and whole organ transplant patients. The second biggest risk is intrauterine infection in pregnant women and transmission of that infection to the unborn baby.
The Institute of Medicine has estimated that the total cost to the US health system from these two infectious presentations would be almost $4 billion to $5 billion each year, and has rated the CMV vaccine as the top priority amongst six other pathogens.
(Click on image for a larger version)
When we first started thinking of working on a CMV vaccine, we were a bit taken aback by the complexity of the genome of the CMV, like that of the EBV. Its genome is one of the largest in the herpesviruses, around 230kb, and it encodes for 145 genes. Of those, 70 proteins had been characterised. So there was no way we were going to take these 70 proteins and try to link them to one another, and try to map what immunogenic determinants were there.
The first approach we used was the immunomics approach, where we took the whole genome analysis approach. We used computer-based algorithms and tried to split the genome into four different categories: early-late antigens, immediate-early antigens, glycoproteins, and even the immune evasion proteins. We grouped them separately and then conducted extensive immunomic analysis, trying to find potential epitopes in these protein sequences.
(Click on image for a larger version)
What we did then, after we had the predictions for something like close to 10,000 normal sequences, was to run through a large panel of healthy virus carriers and transplant patients to do ex vivo T cell assays, to map down the potential critical sequences which could go into the polyepitope vaccine.
And after spending almost four years of work on that, we were able to map out individual sequences within these proteins, and that gave us a pool of 200 novel sequences.
(Click on image for a larger version)
When we took those and then analysed the immune responses to various proteins in CMV, we noticed that most of the dominant responses in CMV are directed towards the structural or the immediate-early proteins. What surprised us was that, although there is some response there, we could also detect some response to the immune evasion genes. That tells you that although these proteins inherently inhibit immune responses, they can then be subtargets for T cell control as well.
(Click on image for a larger version)
Based on these data, we then selected a short list of 46 epitopes.
(Click on image for a larger version)
What we did then was simply to take these individual sequences and link them as a string of beads. So, simply, you have the minimum possible sequence you require to induce a response. You have now eliminated any potential flanking sequences which may have any damaging effect on the immune system or may have some oncogenic potential or a similar modulatory property. You are now targeting your immune response specifically to these epitopes.
(Click on image for a larger version)
Once we had that polyepitope in hand, a second question occurred to us, that in the case of CMV we needed a neutralising domain as well. So we covalently linked this polyepitope sequence to the extracellular domain of glycoprotein B. Glycoprotein B is a critical molecule in CMV for the binding of the virus to the cell surface of the host cell. If you block that binding, you block the virus infection.
So in this formulation we now had an ability to induce CD4 and CD8 responses through the polyepitope sequence, as well as an antibody response to the gB (glycoprotein), but that sequence would also provide opportunity to induce the CD8 and CD4 responses.
 |  |
| (Click on images for larger versions) | |
The next step was to introduce this into a viral vector. We have chosen an endoviral vector which is a chimeric vector. (I won’t go into more detail here about the advantages of this viral vector because time is short.)
(Click on image for a larger version)
We have taken this viral vector and expressed it into the cell system, and we can see that the polyepitope with the glycoprotein B complex is expressed quite nicely. As I said before, it has 46 epitopes, which gives greater than 98 per cent HLA coverage, and it includes the extracellular domain of gB.
(Click on image for a larger version)
We have taken this polyepitope construct and tried to do a series of experiments with HLA-A2 transgenic mice, to look at the T cell responses, the antibody responses, and a quasi-protection experiment as well.
(Click on image for a larger version)
I can’t present all the data, but this is one of the experiments we have done. In the graph at the left we are looking the ELISPOT responses in these mice, and in the right-hand graph we are looking at the neutralising responses against the virus.
Within 10 days you get massive ELISPOT responses in these mice when you vaccinate with single intramuscular immunisation, and these responses are virtually maintained right up to day 75. In terms of virus-neutralising responses, you see a small neutralising response but it keeps improving as you progress to the later time points. It showed us quite nicely that covalent linking of gB to the polyepitope was not impairing either of these two components in their ability to induce either the antibody, or the T cell immunity in this case.
(Click on image for a larger version)
The other thing we looked at in the polyepitope vaccine was the qualitative concept of these T cells. We not only looked at the gamma-interferon but also looked at the ability to produce TNF, or gamma-interferon and TNF together. You can see from the right-hand graph that 30 to 40 per cent of these T cells coexpressed TNF and interferon-gamma, and a majority of the T cells also expressed CD107, which means they can degranulate and that has a cytotoxic T cell function associated with it.
(Click on image for a larger version)
One problem with the human CMV was that for a protection experiment it cannot go into the mouse. So one of the model systems which we have used, and which other groups have also used, is a quasi-protection experiment where you can immunise the mice and then challenge with recombinant poxvirus which encodes one of the CMV antigens which is included in your vaccine. In this case the data I am showing are for the gB which is part of the vaccine composition.
You can see here, looking at the virus titre along the left side, that animals either immunised by CMV vaccine or naïve, when challenged by gB showed 2–3 log lower virus load when we challenged with vaccinia virus, clearly showing that the protection is very much antigen-specific, because you don’t get protection when you give these mice gB or TK-negative.
(Click on image for a larger version)
What was interesting is that if you take these mice and then analyse their T cell responses, particularly in the ones which have been challenged with gB, you see a massive increase in the CD4 T cell responses to gB in these mice as compared with the TK-negative, which have also been immunised. In the ones which have been challenged with gB, within four days of challenge you see almost a three- to fourfold increase in the CD4 responses.
As shown in the right-hand graph, we have also done a protection experiment with another vaccinia construct, which includes the immediate-early antigen. And similarly you get protection and also an increase, in this case, of CD8 responses in the same vaccine formulation.
So what we have shown to date is that this sort of formulation can actually induce both CTL helper immunity and antibody responses. It has always been a challenge for the CMV vaccines that in most of the previous attempts which have been made with CMV vaccines they have been able to induce either cellular immunity or humoral immunity, but the two things have never come together. I think the approach we have used here will overcome that limitation.
(Click on image for a larger version)
The other application of polyepitope technology that we have been exploring in our group is to use it for adoptive immunotherapy. This was one of the interests that came into our group when we were originally doing some work on transferring autologous T cells into transplant patients.
You see here a CT scan of a patient we treated many years back who developed post-transplant lymphomas following lung transplant. Virtually within two weeks we were able to transfer the T cells and we saw a 50 per cent reduction in the lymphomas, and by 20 weeks the patient was completely free of these tumours.
In this case the tumours were of a phenotype for which we could use polyclonal T cells which are directed against multiple EBV antigens, but as Andy Morgan presented to you yesterday, when you go to the other malignancies which are not like polyclonal lymphomas, which are not like normal EBV-infected B cells, the situation is quite different. The gene expression pattern is quite different.
So what we have done is to address those model systems, or the tumour systems.
(Click on image for a larger version)
We have developed an EBV polyepitope which selectively expresses the epitopes from those antigens which are expressed in malignancies such as nasopharyngeal carcinoma (NPC) or Hodgkin’s lymphoma. Using that strategy, we have now been able to develop a very, very simple protocol where we can get autologous T cell lymphocytes which can grow from day 0 to day 14 by a 50–150 fold expansion.
The procedure it requires is very simple. You take your recombinant adenovirus which expresses the EBV polyepitope, you infect the autologous PBMC (peripheral blood mononuclear cells) at a very, very low MOI (multiplicity of infection), you mix them together, you simply add IL-2 on days 3, 7 and 10, and by day 12 or day 14 you harvest these cells.
(Click on image for a larger version)
Here is an example of what we have done, from one of the Hodgkin’s lymphoma patients. At day 0 you are looking at different epitopes in the polyepitope; the responses are 0.24, 0.02, 0.04 per cent of the CD8s. And within 12–14 days you see massive expansion with this adenoviral vector, up to 150-fold. These T cells not only express gamma-interferon but also show cytolytic function against the tumour cells expressing the LMP protein and also the vaccine construct expressing LMP2 protein, which are the critical genes expressed in Hodgkin’s lymphoma that you want to target your immune response to.
(Click on image for a larger version)
The other thing which we have been trying to do – we have just received some funding from NCRIS (National Collaborative Research Infrastructure Strategy) – is to develop a concept we are calling a T cell bank. We are collaborating with the Australian Red Cross. What we will be able to do is to feed the EBV polyepitope adenovirus into the peripheral blood mononuclear cells from the Red Cross donors, and then grow the T cell cultures by exactly the protocol I described earlier. These T cells will be characterised and stored in liquid nitrogen.
What will happen is that if a patient comes down with any EBV malignancies we will do the HLA matching and we will be able to provide these T cells to those patients for adoptive immunotherapy.
So, rather than waiting for the patient-specific therapy to be prepared, after the patient’s blood comes to us you will have this T cell bank which is stored at our institute and you can match the HLA, just as in a bone marrow registry, and provide these T cells for adoptive immunotherapy.
(Click on image for a larger version)
We have just initiated a phase I/II clinical trial on NPC patients in collaboration with the University of Hong Kong, using the autologous T cell approach, and we plan to initiate a similar trial in collaboration with the PeterMac Cancer Institute for Hodgkin’s lymphoma patients later this year.
(Click on image for a larger version)
This slide is to show you examples of the polyepitope technology other than being used for EBV and CMV. A number of other groups have moved on from the preclinical to phase I or phase II clinical trials. As you can see here, it has been applied to melanoma, breast cancer, HIV, malaria, hepatitis B and hepatitis C, and in a cervical cancer model system as well.
(Click on image for a larger version)
I will just summarise what I have presented to you.
We feel that with polyepitope technology you can select your epitopes from the highly conserved regions, which avoids any potential viral proteins which are likely to have escape mutants.
You can combine both CTL and helper epitopes in a single formulation, and also virus-neutralising components as well, without compromising the immunogenicity of each of these components.
The other important part is that polyepitope technology is much safer than delivering the full-length protein vaccines, and provides the maximum HLA coverage you need for different ethnic groups.
The inclusion of multiple epitopes from multiple antigens also diversifies your response. In fact, we have got very early preliminary results indicating that you can actually combine EBV and CMV vaccine in a single formulation and you will have a single vaccine for multiple pathogens from polyepitope technology.
(Click on image for a larger version)
I wish to acknowledge some of my lab members. Jie Zhong has been with Ian Frazer and has moved into our lab. He has been actively working with the CMV vaccine with us. Corey Smith has worked on the EBV vaccine, and Leone Beagley is doing the NPC patients. And you see here the names some of the other members of the group who have contributed to this work.
Discussion
Question: This question relates to the polyepitope constructs. Once, say, an antigen-presenting cell takes them up, there is always going to be peptide competition in terms of loading to the various class I molecules, for example. And there is immunodominance in any given individual, whether there is pre-existing immunity or not. In the studies that you have done, say in the patients, do you see or can you detect equal epitope representation in the T cell response, or is there a dominance towards certain epitopes and not others? In other words, are you losing opportunity because some of those epitopes are just never able to load, because of the competition?
Rajiv Khanna: Good question. It depends on what time you see that, when you do the analysis. If you do an early-stage, as it happens with any other stimulation system, you will find the specificity is much narrower, but if you allow the cultures to progress, surprisingly rather than getting more focused it gets more diversified.
We have done multiple donors, and we find, actually, as the culture progresses the specificity of the culture tries to include those epitopes which we hadn’t seen previously responded to when we were simply stimulating with single peptides.
We don’t know the answer to why that is happening, but we have seen it repeatedly in NPC patients, in healthy individuals and in Hodgkin’s patients. If you look at the culture at day 10, it will have three or four of the 10 peptides positive, and then suddenly it will become six and 10. It diversifies. I don’t know why it does it, but that is how we have seen the responses.
Question: I am sure you have been asked this question before, but it concerns an issue when you move peptides around like this. The criticism that has been around a long time is that you could make new epitopes and those new epitopes could be self antigens inducing autoimmunity. What is your standard answer to that question?
Rajiv Khanna: I haven’t got any standard answer! We have tried to look at that issue, and we haven’t found any responses like that. But it can never be ruled out.
To extend that point: what surprised us is that we have, if anything, found the responses to the exact epitopes, but not through the HLAs we originally had mapped. We are now finding that it can actually cross the HLA restrictions more than we expected. For example, one of the epitopes we had was HLA-A24 restricted. We found that it also can be presented by HLA-Cw6, which we had originally not found through using a peptide mapping system. In fact, there is a paper on that concept published in the European Journal of Immunology, I think by someone from Harvard, that you have a so-called promiscuity which we previously had been seeing in the class II system but not for class I. This is not the supertype, which is a separate concept – this is beyond the supertype concept, where the HLA-A restricted allele can be presented by the B or C allele as well.
We had originally thought the coverage would be 95 per cent or 96 per cent, but we probably are underestimating the coverage. That was originally based on the restriction elements we mapped from our epitope mapping process, but when we do the polyepitope technology and stimulate the immune responses we are finding additional HLA restrictions.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
