Margalit: Can you explain more about the use of tetramers in this context?

De Groot: We are working on West Nile virus which is a big problem in Israel and now the USA. As a clinician, I know that when I have a patient come into the emergency room with aseptic meningitis, I am now committed to a 5—6 day hospitalization because I have to rule out West Nile. This requires acute serology and follow-up serology. There is no method ofdetecting whether this person might have recently been exposed to West Nile when they walk into the emergency room. Our idea is that you might be able to identify T cells in their circulation that are specific for West Nile virus using the tetramer technology. This is what we proposed. It wouldn't be a screen for blood banks because there they just do PCR. In our procedure you could mix the whole blood with the tetramer, run it through the FACS machine in the clinical lab and then detect whether the person had recently been exposed to the West Nile virus. T cell responses are much quicker than the antibody response. We are also very interested in looking at smallpox and vaccinia. You could differentiate people who had been immunized with vaccinia versus people who had been exposed to smallpox in an exposure situation. How else would you know how to quarantine people?

Petrovsky: The problem with that strategy is that it is MHC specific and therefore you can't really develop a generic reagent. It is also epitope specific and you'd need an enormous cocktail to cover most of the epitopes.

De Groot: It is a new idea and there are many potential obstacles. If you include the main five MHC types, you could say for these people with these HLAs, it is not a 100% reliable test in all cases, but if it is positive then you have your answer. A negative result doesn't help.

Rammensee: The critical issue is to verify the predictions. In the case of HIV, this is done with T cells from patients. As I understand it, you test them for recognition of the synthetic peptides, but you don't know whether these peptides are actually processed.

De Groot: We do know that the epitopes were actually processed, because if they are naturally infected and they respond to the peptides, then presumably they have been exposed to a peptide like that which was naturally processed and presented.

Rammensee: There still could be some kind of cross-priming by other cells, and this might not be the same peptide as presented by the infected cell.

De Groot: It could be slightly different, but I don't think it would be very different.

Rammensee: The key question here is whether you could test a cell that is not infected but which is transfected with different genes of interest to see whether they are recognized. This would ensure that these peptides are really relevant.

De Groot: It is hard to know what is in a human. We are going to be doing studies in transgenic mice which are immunized with a construct which contains the whole gene that may contain some of our epitopes. Then we will be coming back (in vitro) with the epitopes. This may answer your concerns.

Rammensee: Will you have some cells expressing the virus or part of the virus?

De Groot: No, they will be getting a DNA vaccine containing a gene from the virus. We will know if the epitope is naturally processed or not.

Rammensee: It is critical to show that the peptides are naturally processed.

De Groot: This is hard to do in humans who are naturally infected, because we don't know the sequence of the original strain of the virus. If there is a negative response, is it because the epitope wasn't processed, or because the subject have a different strain of HIV and they have never seen that epitope? Presumably, we are finding conserved epitopes, but maybe not. The mouse experiment will be able to answer some of these questions.

Borras-Cuesta: I have a question and a comment with respect to the strategy you have used. You immunize with DNA with multiple epitopes. If I understood you correctly, you said that at the end you had a response when you stimulated with the fuller peptides. What happens to individual peptides? I know from my own experience that sometimes you just get a response to one peptide.

De Groot: It is a very frustrating collaboration. We will be doing individual peptides, but this hasn't been done yet. Currently we only have results showing response to all of the epitopes, in a pool of peptides.

Borras-Cuesta: We have done this in our lab and other people have reported the same thing. Ifyou immunize with these types ofplasmids, you can end up having a response to only one of them. It is a crucial issue.

De Groot: This is obviously of great interest to me, because I want to know how many peptides are coming in.

Borras-Cuesta: You mentioned that a peptide was recognized by about 60% of people, and you thought it was a good candidate. In general terms I think you are right, but then you have to ask yourself why these people are still infected. The fact that this peptide is recognised by 60% of the people doesn't make it a good peptide

De Groot: It is big problem. I don't think anyone who is working in HIV knows how to sort this out. There are some data from Bruce Walker looking at the types of T cell responses in patients who are acutely infected. They actually recognize different epitopes in the early infection phase than they recognize during chronic infection. We can't, however, dissect whether this is due to the evolution of the virus or whether it is due to differential processing. The best you can do is base your hypothesis on what is known about effective containment of HIV infection. It appears to be due to the recognition of multiple epitopes. Livingston et al (2001) have shown that if epitopes are presented as single units in a string, this will cause better or broader recognition of a greater number of epitopes than is done if just the gene is presented. You can pack more information in a pseudogene like this than you can using a single gene as an immunogen. It is a mix of hypothesis-driven research and practical experience. We are trying to make something that will work on the basis of what we know. The problem is we don't know whether an immune response to an epitope we identify in this manner is going to work to protect against infection, since the people we are testing already have HIV infection. I should also add that for the Th epitopes we use long-term non-progressors. Perhaps these are better, because people who do not progress to AIDS do recognize the epitopes.

Borras-Cuesta: So you are hoping to use a good CTL response in vaccination so you would not reach that situation.

De Groot: Yes.

Perelson: I think with HIV vaccines we have to be somewhat careful in distinguishing them from other vaccines. With most vaccines we try to generate what we call sterilizing immunity, so that one person will not become successfully infected by the disease. In HIV no one has been able to establish that state of protection. Most of the vaccines that have been tried in animal studies are non-sterilizing and act as therapeutic vaccines: where we have seen the most success is in generating enough of an immune response to maintain levels of CD4+ T cells that are higher than in unvaccinated animals, and hence extend their lives.

De Groot: This is a major shift in our thinking about vaccines. This deserves emphasis here. In Barcelona, Larry Corey talked about modifying the goal for HIV vaccines. He said that we should not demand that the vaccine act as prophylaxis. Instead, an acceptable new goal is to contain infection. This is a completely new way of thinking about HIV. Perhaps the reason we have 'lowered our expectations' is because we now realize that it will be very difficult to make a vaccine that works prophylactically, so we will accept something that works after infection, by containing the virus better than a non-immunized host. The concept of containing infection is an interesting one, and reflects a shift in the vaccine community in general: therapeutic vaccines will be better accepted in the future on the basis of this work in HIV.

Rammensee: Regarding HIV vaccination, we would probably think in terms of applying the vaccine between phases of HAART.

De Groot: Brigitte Autran in France has set up a network of collaborators and they are looking at therapeutic vaccination for HIV infected patients. The idea is to treat with HAART and get the virus load low, and then vaccinate during HAART with the intention of educating (priming) the maximum number of T cells to respond so that when you take the HAART away you can look at the slope of viral load increase to see whether or not the vaccine is working.

Brusic: Coming back to immunoinformatics, you can see here that there are a number of analytical steps or models that have been put together, starting from genomic information all the way to constructing candidates for vaccines. We would expect that with so many steps involved, errors would creep in and we might not successfully find vaccine targets. Fortunately, the results are quite encouraging. A significant number of peptidic vaccine studies started with predicted targets which were subsequently shown to be functional in patients. This is an illustration of how immunoinformatics can help move the whole field forward. However these are only preliminary studies and we can improve the synergy between predictions and experimentation. I am pleased to see that you built 74 models for MHC class II peptide binding. This takes a lot of effort. How many people do you have to maintain your prediction system? De Groot: Just one.

Brusic: We have recently developed a prediction system where a single model predicts peptide binding to an array of HLA molecules. This was achieved by modelling interactions between peptides and multiple MHC molecules. Our single predictive model can in parallel predict peptide binding to multiple HLA-DR molecules, and another model predicts peptide binding to multiple HLA-A2 molecules. This is an example where computational immunology can help us do things more efficiently. There are two sides to the problem — how to discover better vaccines and also how to improve the research methodology. We should strive to advance both these aspects of our research work.

De Groot: Don't you think that we are also expanding our horizons in immunology? Regardless of how you find the protein (by microarray or by direct sequencing), we are looking at the immunogenicity ofproteins that people haven't even been able to isolate. What excites me is this 'immunome' problem: how much information (in terms of epitopes) is required in order to get a host to respond effectively to a pathogen? I also don't think that the information required just involves T cell epitopes; it is also B cell epitopes (which I can't model). I think the question 'how much immune information is required to generate a protective response?' is an interesting question to ask. The more we apply these tools, perhaps the closer we will get to answering it.


Livingston BD, Newman M, Crimi C, McKinney D, Chesnut R, Sette A 2001 Optimization of epitope processing enhances immunogenicity of multiepitope DNA vaccines. Vaccine 19:4652-5660

Immunoinformatics: Bioinformatic Strategies for Better Understanding of Immune Function:

Novartis Foundation Symposium 254. Volume 254 Edited by Gregory Bock and Jamie Goode Copyright © Novartis Foundation 2003. ISBN: 0-470-85356-5

Was this article helpful?

0 0

Post a comment