DNA vaccines against tuberculosis - Nature

18 downloads 0 Views 3MB Size Report
culosis in mice and raising hope for a new kind of vaccine to replace bacille Calmette-Guerin (BCG), DNA encoding only one or a few protein antigens was ...
Immunology and Cell Biology (!997) 75. 591-594

DNA vaccines against tuberculosis DOUGLAS B LOWRIE,' CELIO L SILVA^ and RICARDO E TASCON" ^National Institute for Medical Research, Mill Hill, London. United Kingdom and ^University of Sao Paulo. Brazil Summary This edited transcript of a presentation at the 'Vaccines Beyond 2000' conference describes a series of investigations by the authors throwing light on the mechanisms of protective immunity against tuberculosis in mice and raising hope for a new kind of vaccine to replace bacille Calmette-Guerin (BCG), DNA encoding only one or a few protein antigens was found capable of conferring persistent protection equal to the effect of BCG. The essential features seem to be an endogenous origin of the antigen within transfected mouse cells which favours the development of CD8VCD44^/IFN-y-producing T cells with antigen-specific cytotoxicity. Such cells were the most efficient in adoptive transfer of protection from infected or DNA-vaccinated mice to naive mice. Key words: antigen presentation, cytotoxic immunity, passive transfer of immunity, recombinant cell-mediated immunity. T lymphocytes, tuberculosis vaccines.

It would be a mistake to think that there has not been much interest in tuberculosis in recent years. In fact there has. A quick scan through the Science Citation Index database. pulling out publications on tuberculosis immunology over the last 16 years indicates that, expressed as a percentage of the whole of immunology publications, there has been an approximate four-fold increase in interest over that period. I think this is partly due to the fact that tuberculosis is still seen as a wonderful immunological paradigm of totally cell-mediated immunity being responsible for protection. But even more so. it is due to the sudden perception within mainstream medical and public awareness at the beginning of this decade that tuberculosis had not gone away but was increasingly a serious killer in the civilized world. The significance of 3 million people dying of tuberculosis per annum in the rest of the world somehow had not previously registered. Having said that, we have not made a lot of progress over this period, in the sense that we have known that we need new vaccines. we have known that we need new drugs yet at least in the case of the drug requirement, they are still some distance away and in any case, there will always be drug resistance against new drugs and it is not the final answer. For vaccines, we know that BCG is not all that it should be; it fails in .some populations, protection lasts perhaps 15 years in the UK, we do not know about the rest of the world. But we have known for a long time that the sub-unit approach to replacing BCG does not look terribly promising. With dead BCG and purified components you can get .some protection in animal models but it does not last very long; this is the main problem. We have heard from Peter Andersen of some very elegant work and we now know that there are individual protective antigens. Some of them have been identified and shown to work using some of the newer adjuvants. However, I want to talk about a different approach; DNA vaccination, which I think is at least as promising, but will it be better? Correspondence: Dr Douglas B Lowrie, National In.-ilitute for Medical Research. The Ridgeway. Mill Hill, London NW7 lAA. United Kingdom. Email: Received 15 Augu,st 1997; accepted 15 August 1997.

We do not know at this stage. Tn any case, how will new vaccmcs bo tested? This is a whole separate issue that 1 cannot properly deal with here although it is the main outstanding question. The take-home message is that DNA vaccination works quite well. Even a single protein antigen given as DNA can protect as effectively as BCG and DNA might work where BCG fails. The work started when it became possible to clone mycobacterial genes into eukaryotic vectors and express them within APC. This meant that we could introduce the antigens in new ways into the immune system and perhaps we would leam something about the basic immunology. When 1 started this work I was not expecting to come up with a line into new vaccine.s. but it looks as if that may be the outcome. The work I want to present really comes in two parts. The first part concerns expression of the mycobacterial DNA from a retroviral vector that integrates into the cell nucleus of J774 macrophage-like tumour cells, making a transgenic cell line that produces the protein. We used this to vaccinate mice and look at the immune response. We have leamt a lot about the immune response using this model. We then learned that we could use naked DNA, so the second part of this work concerns DNA vaccination.

Vaccination with transgenic tumour cells Protective

efficacy

Having made the transgenic J774 cell line expressing hsp65 antigen of Mycobacterium leprae from a relroviral vector (J774-hsp65 cells) we introduced the cells into mice and tested what effect they had on protective immunity. They received large doses of the tumour cells, the tumour cells disappeared and the mice were left with substantial immunity. When challenge infected with BCG the bacteria multiplied in controls and there was very dramatic protection with an exponential decrease during 5 weeks in mice that had received cells expressing hsp65.' It was totally unexpected that with a

592

DB Lowrie et al.

single antigen we could get such a major protective effect. Even more remarkable was that this was also manifest against vii-ulent Mycohacterium luberculosis challenge in an exactly pajallel experiment: the bacteria multiplied in the controls and were killed exponentially in the transgenic tumour cellvaccinated animals. Hitherto this degree of protection had only been seen with whole live BCG. So we naturally wanted to know what had happened to the immune system. It could have been a totally non-specific effect. Cutting a long story short, it was not non-specific, it was antigen-specific acquired cell-mediated immunity that we could adoptively transfer into naive mice,- So naive BALB/c mice were gamma-irradiated, and then reconstituted with T cell clones that had been derived from spleens of mice that had been vaccinated in this way. The numbers of live bacteria in the spleens were measured 3 weeks after reconstitution and challenge infection with virulent M. tuhercutosis. The CD4'^ clone gave a small amount of protection, the CD8* clone gave much bigger protection and the y/S elone was similar to the CD4'^ clone. When we combined the clones, for example, taking half the number of CD4"^ and half the number of CDS"*^ cells and mixing them together we got an additive effect. Adding all three clones together we saw synergy, a greater effect than would be expected with the individual clones by just adding them together and really a very substantial protective effect. The most protective clone individually was the CDS* clone, which was consisteni with the hypothesis that the most important thing happening in this novel vaccine system is that the antigen is being generated endogenously which would favour presentation on MHC class I and generation of the CD8"*" T cell response. CDS"^ was the most protective phenotype, but we had only looked at one clone of each and. of course, we needed to look at some more. A lot of work has been done in Brazil generating and testing more clones.

Protective T cell clones Twelve CD4+ and 12 CD8+ antigen-specific clones were characterized extensively and were shown to have antimycobacterial activity in vitro.-^ If supematants from the clones were used to activate macrophages they would activate to inhibit multiplication of intracellular tubercle bacilli and if the clones were put into cellxell contact with infected macrophages in the pre,sence of anti-IFN-y those cells that had antigen-specific cytotoxicity would inhibit the growth of the bacteria too. So they had two properties that were associated with in vilro antimycobacterial activity: IFN-y production and cytotoxicity. We used representative clones in adoptive transfer of protection."* These clones represented the spectrum of activities among CD4"^ and CDS"^ clones in teniis of IFN-y and IL-4 production and cytotoxicity. Some of the clones resulted in bacterial killing; there was a quite substantial reduction in numbers of bacteria in the internal organs. The most effective clones were CDS"^ cytotoxic clones. If the clone produced IFN-7 as well, that was good and we could neutralize the effect attributable to IFN-y by giving the animal monoclonal antibody against IFN-y. From this study we concluded that CDS"^ clones probably are more effective than CD4+ clones and IFN-y and cytotoxicity both contribute to protection, perhaps with cytotoxicity being more important.

Vaccination with DNA General background At that point we saw a publication saying that DNA injected stTaight into animals could give specific protection against infection* and this was very interesting because here we were getting protection using a single mycobacterial gene, expressing that //( vivo, albeit using a tumour cell that makes it a messy system. Obviously it is not a practical vaccine and it is a bit complicated to sort out what is happening. But if one could u,se the DNA directly, this was going to be a very useful thing to do in order to investigate the immunology behind this protection. So we began to do this against a growing background of publications showing that DNA vaccination could give protection against a wide range of infectious diseases in a wide range of animal species. In fact, DNA vaccination has seized the imagination and there are very many laboratories currently working on it. The essence of this approach is that a very small amount of the DNA that is injected into tissue is taken up into cells and expressed. This has been clearly demonstrated with DNA encoding marker proteins that can be readily detected, such as P-galactosidase that can be stained histochemically. Another feature of the system is that expression of the protein can continue for a very long time. There is an initial peak of protein expression that declines asymptotically and probably never comes down to zero, so that there can be long-term expression of foreign antigen in the tissues of the animal. This system has gone into clinical trials very quickly since its initial description. There are at least half a dozen ongoing phase I clinical trials around the world, mainly with HIV but also with herpes and influenza. The trials are now including normal volunteers, both in HIV and influenza vaccine trials, a reflection of the fact that it is a very benign vaccination procedure: reaction is minimal to non-existent. We are all waiting, of course, to see how effective it is going to be. The initial rumours are that the big problem DNA vaccination is going to face is one of efficacy — getting it to be sufficiently effective. As I have mentioned it is very inefficient; a veiy small amount of DNA is taken up and expressed.

Efficacy of hsp65 DNA However, applying the procedure to our mouse system, we made the DNA using a cytomegalovirus immediate early gene promoter and the hsp65 gene, and grew up lots of plasmid, injected 50 pg into the right hind leg and 50 jig into the left hind leg four times at 3 week intervals, waited 1 month then challenge infected with virulent tuberculosis and counted the numbers of bacteria in the spleen, liver and lungs 6 weeks after infection. In outbred Parkes strain mice BCG vaccine gave a nice protection, substantial reduction in the infection in each of the tissues, and so did hsp65 DNA, although not quite as well as BCG,"^ Vaccination with DNA encoding a completely different antigen, the 36 kDii prolinerich antigen of M. leprae had essentially no effect. It was good to get substantial protection with hsp65 DNA in outbred mice because it meant that there was not too much of a problem with genetic restriction ofthe response, but it does make a difference when we use different mouse strains. In an

DNA vaccines against tuberculosis

experiment with CBA/BIO cross-bred mice the protection with BCG was again substantial with, for example, about 100-fold reduction of counts in the spleen; hsp65 DNA equalled BCG, there is no difference between BCG and DNA vaccination and, in fact, there was also substantial and significant protection with DNA encoding the proline-rich antigen; hence hsp65 was not a special antigen,^

593

antigen presentation. There was protection, but it declined just as it did with the J774 vaccine vehicle. So it looks as though the important mechanism here, with DNA and live BCG, is likely to be the persistence of expression with the protein consequently being presented to the immune system for a longer period of time.

Memory Additiotiat antigens We have now made a range of constructs with different promoters and different antigens; for example, ESAT-6. the 6 kDa early secretory antigen of M. tuberculosis, and we have heard a lot of evidence from Peter Andersen that this is an important protective antigen, MPT83 and hsp70. All of these have given protection, sometimes as good as BCG but we do get experiment-to-experiment variation. An altemative promoter that we have used is from the mouse 'house-keeping' gene hydroxymethylglutaryl-CoA-reductase and this also works.'^ We have not done enough to say that one works better than the other. In an experiment in BALB/c mice comparing and mixing plasmids expressing different antigens we found that hsp70 and ESAT-6 were better than hsp65, but I would hesitate to say that one antigen was better than another at this stage and we need to do much more comparative work,^ When we mix plasmids together and inject them simultaneously we can get greater protection than when they are injected individually, but sometimes there appears to be an interaction and protection is less. The significance of this phenomenon needs to be resolved in addition to deciding which are the best antigens.

Duration of protection I want to tum now to what is, I think, the most important outstanding question in the development of replacements for BCG and that is the duration of protection. Right from the very early attempts to make altemative vaccines, many years ago now, it was established that you could, in fact, get quite good transient protection using proteins but this did not last very long. So does protection last with DNA? We have recently compared a number of different ways of vaccinating in BALB/c mice.** In this experiment the level of protection obtained with live BCG was quite modest. We do not know why. However, the point can still be made that when we come in with the challenge infection with M. luberculosis 2 weeks after vaccination or at intervals up to 8 months after vaccination, the protection we got is maintained — it did not decline over an 8 month period. When we vaccinated with hsp65 DNA we saw the same phenomenon: challenging 1 week after completion of DNA vaccination was the same as challenging after 8 months, and that, I think, is potentially a very important observation. If we used different vaccination procedures any protection declined during 8 months of delay before challenge. For example, with J774-hsp65 we did get strong protection if we challenged immediately but it declined over 8 months. When we entrapped the protein in positively charged liposomes and injected them intravenously there was protection. The protein is presumably discharged into the cytosol, so this is another way of getting endogenous

We have used CD44(j| antigen expression as a T cell surface marker for memory and activation and investigated what happens following these vaccination procedures.^ Takmg spleens from the mice at intervals up to 8 months after vaccination we found that there was a high frequency of CDSVCD44^| ceils showing hsp65-specific cytotoxicity. The cells from the spleens were amplitied by culture on y-irradiated J774-hsp65 (iJ774-hsp65) and were then characterized. With hsp65 DNA vaccination we got a significant rise in these cells and the frequency continued to increase during 8 months, so that, if you like, the memory ofthe activated population \^as increasing, and presumably this reflected what was happening in VIVO. Remember that although we were seeing ver>' high frequencies here, the cells had been amplified in vitro. After vaccinating with J774-hsp65 the frequency was initially very high but this declined and was the same with the liposomes, all of which is. I ihink, consistent with the idea that continued production of endogenous antigen is important to sustain these cells.

ThI versus Th2 responses We looked to see what functional phenotype might be associated with protection, comparing spleen cell populations taken either 1 week after completing hsp65 DNA vaccination or 2 weeks after giving live BCG, After an initial amplification on iJ774-hsp65 we separated the cell types using a combination of lytic monoclonal antibodies and adherence and then titrated the cells out in wells containing iJ774-hsp65 and coated with appropriate antibodies in ELISPOT assays to compare the frequency of cells producing lFN-y and those producing iL-4, We found thai the IFN-y-producing ceiLs were more frequent than those producing IL-4 in all of those vaccination procedures where an endogenous source of antigen was provided, except for BCG, where it was the other way round — the lL-4-producing cells were more frequent.

Comparisons between responses to infection and responses to DNA vaccination We wanted to take this observation a bit further and see whether the whole live mycobacterium affects the immune system a bit differently to an endogenous single antigen. We have compared the T cell populations in spleens of mice infected with virulent M. tuberculosis, 30 days into the infection, with T cells from spleens of hsp65 DNA-immunized mice. 1 week after completion of immunization. The T cells were separated into CD4VCD8- and CD8*/CD4 . amplified by culture for 1 week with iJ774-hsp65 ceils then stained with monoclonal antibody against CD44. and FACS

594

DB Lowrie et al.

analysed. Among the cells from infected mice the ^ populations, at 22-28% of stained cells, were fewer than the CD44jy populations, whereas from the DNA-immunized mice it was the other way around, with the majority, 56-78% of stained cells, being

Characteristics of protective splenocytes When we took these cells and used them in adoptive transfer of protection against challenge with M. tuberculosis, as described earlier, the cells from infected mice had only a small protective effect, whereas the protection with cells from the DNA-immunized mice was really quite significant. The most protective cells were the CD8'' cells from either source. When we used the separated CD44^- and CD44[^ cells it was clear that protection was particularly associated with the CD44^, cells. We did get better protection with the CD44|^| cells from infected mice and the most effective cells were clearly the CDS"^ cells. The cells from DNA-vaccinated mice gave much better protection and the effect was almost all associated with the CD8VCD44^i population. When we characterized the protective cells in terms of IFTM-y or IL-4 production, and obviously we are using these properties as markers for cytokine type 1 and cytokine type 2 T cell responses, we saw a difference between the infected and DNA-immunized mice. A higher proportion of the cells from the infected mice were producing IL-4 than were producing IFN-y and the reverse was true for the cells from DNAimmunized mice, where a higher proportion were producing IFN-y than were producing IL-4, It did not matter whether we were looking at CD4'^ or CD8* cells. When the cells were separated into CD44^| and CD44|^^, as was done for the adoptive transfer of protection experiment, it was evident that IFN-y production was associated with CD44jj- whereas IL-4 production was associated with CD44|^ cells, irrespective of whether the cells came from infected or DNA-immunized mice. You may remember that most of the cells from spleens of infected mice were CD44|^, whereas most of the cells from DNA'immunized mice were CD44^j, so it appears that CD44|^L-4-producing cells predominated in cells from the infected mice and CD44^|/IFN-y-producing cells predominated in cells from DNA-immunized mice. On the basis of this evidence, it appears that protection is mediated by CD8VCD44^./IFN-y-producing T cells.

Superiority of DNA vaccination for generating protective T cell clones This conclusion was confirmed and extended when we went back to cloning. We made hsp65-specific T ceil clones from these infected or DNA-immunized mice. Clones were selected on the basis of firstly being strongly growing and then four clones that were CD4VCD44|^, four that were CD4VCD44j,-, four that were CD8VCD44,^ and four that were CD8VCD44^j were selected from clones from infected mice and the same was done for DNA-immunized mice.

After extensive characterization in vitro, so that we knew that they were typical T cells and we knew their IFN-y, IL-4 and antigen-specific cytotoxicity profiles, they were used in an adoptive transfer of protection experiment. Protection was transferred only if the cells were IFN-y producers. The most strongly protective were the four CD8'^/CD44^j/IFN-'yproducing/cytotoxic clones. Whenever the cells were cytotoxic they tended to be the most effective, so the combination of cytotoxicity, IFN-y production, CD44^|, CD8^ seems to be what is needed optimally for protection in this model. We found those cells particularly if we used DNA-immunization, whereas if it was an infection, whether that was BCG or virulent M. tuberculosis, then we found a mixed population of cells in which IL-4 producing CD44j^ cells were predominant; those cells had essentially no capacity to confer protection — they did not produce IFN-y and they were not cytotoxic, I think these conclusions are important because they could be telling us what we should be looking for in developing a vaccine. This is all mouse work and maybe we can find the same sort of thing in man but that remains to be seen.

References 1 Silva CL. Lowrie DB. A single mycobaclerial protein (hsp65) expressed by a transgenic antigen-presenting cell vaccinates mice against tuberculosis. Immunology 1994; 82: 244-8. 2 Siiva CL, Silva MF, Pietro RCLR, Lowrie DB. Protection against tuberculosis by passive transfer with T-cell clones recognizing mycobacterial heat-shock protein 65, Immunology 1994; 83: 341-6. 3 Silva CL. Silva MF, Pietro RCLR, Lowrie DB. Characterization of T cells that confer a high degree of protective immunity against tuberculosis in mice after vaccination with tumor cells expressing mycobacterial hsp65. Infect, fmmun. 1996; 64: 2400-7. 4 Ulmer JB, Donnelly JJ, Parker SE et al. Heteroiogous protection against influenza by injection of DNA encoding a viral protein. Science 1993; 259: 1745-9. 5 Tascon RE, Colston MJ, Ragno S, Stavropoulos E, Gregory D, Lowrie DB. Vaccination against tuberculosis by DNA injection. Nat. Med. 1996; 2: 888-92, 6 Tascon R, Stavropoulos E, Colston MJ, Lowrie DB, Polynucleotide vaccination induces a significant protective immune response against mycobacteria. In: Brown F, Norrby E, Burton D, Mekalanos J (eds). Vaccines 96: Molecular Approaches to the Control of Infectious Diseases. Cold Spring Harbor: Cold Spring Harbor Laboratory Press. 1996; 45-9. 7 Lowrie DB, Silva CL, Colslon MJ, Ragno S. Tascon R. Protection against tuberculosis by a plasmid DNA vaccine. Vaccine 1997; 15: 834-8. 8 Lowrie DB, Colston MJ, Tascon RE, Silva CL. DNA encoding individual mycobacterial antigens protects mice against tuberculosis. In: Brown F. Burton D, Doherty P, Mekalanos J, Norrby E (eds). Vaccines 97: Motecutar Approaches to the Control of Infectious Diseases. Cold Spring Harbor: Cold Spring Harbor Laboratory Press. 1997; 163-6.