What I am going to talk about today is our use of the SIV nonhuman primate model for AIDS to investigate the feasibility of using a DNA vaccine for immunotherapy of HIV.
(Click on image for a larger version)
It is clear that highly active antiretroviral therapy (HAART) has had a significant impact in HIV infection by restoring immune function and reducing viral load. However, long-term benefits are limited. Drug-resistant strains emerge and drug toxicity and treatment fatigue force many people to go off the drug. HAART doesn’t eliminate the viral reservoir, and virus can sequester in certain tissues. And, importantly, HAART does not restore HIV-specific immunity.
(Click on image for a larger version)
The next frontier for treatment of HIV infection is in HIV immunotherapy, and specifically by vaccination in combination with HAART. The goals of HIV immunotherapy are to reduce viral load, target the viral reservoir, induce host-mediated control of the infection in the absence of antiretroviral drugs and restore HIV-specific immunity. Vaccines will likely need to restore CD4+ and CD8+ T cell responses, which are critical for clearance of HIV infection. Inducing these responses should contain viral rebound after discontinuing the drug.
(Click on image for a larger version)
The strategy that we are mostly focused on is in the use of DNA vaccines, and particularly particle-mediated epidermal delivery (PMED), a strategy that is more commonly known as the gene gun. DNA immunisation results in direct delivery of DNA into cells. PMED delivers gold beads coated with DNA directly into cells of the skin. Once the DNA gets inside the cell then the antigen is expressed in the cell resulting in antigen presentation via both class I and class II MHC.
The gene gun delivers DNA vaccines into the epidermis. We and others have shown that the gene gun directly transfects epidermal Langerhans cells as well as resident dendritic cells (DCs). These cells then traffic to local draining lymph nodes as well as to mucosal compartments.
Delivering DNA vaccines into the skin results in the induction of immune responses in the blood and the mucosal tissues, including in the lung and the gut. A number of studies have demonstrated that there is significant cross-talk between the skin and the mucosal immune system. So it is possible to induce a mucosal immune response by peripheral immunisation in the skin.
This particular aspect of gene gun DNA immunisation is especially important for the development of a therapeutic vaccine against HIV. The primary reservoirs for HIV are in the mucosa and a vaccine that induces mucosal responses may be able to eliminate the virus at its source.
(Click on image for a larger version)
The first type of DNA vaccine that we worked with was one that was specifically focused on the induction of a CTL response by encoding virus-specific CTL epitopes. CTL epitopes are short sequences that are often poorly immunogenic. To increase the immunogenicity of epitopes delivered by a PMED DNA vaccine, we constructed a fusion plasmid that fuses the epitopes to a gene encoding hepatitis B core antigen. The concept of fusing epitopes to hepatitis B core is not new in the context of a protein vaccine. However, delivery in the context of a DNA vaccine offers a unique advantage: the plasmid is initially expressed inside the cell, resulting in endogenous class I presentation and expression of a virus-like particle carrying the epitopes. The particles can be picked up and again processed by antigen-presenting cells for exogenous class I presentation. The combined mechanisms significantly increase presentation and immunogenicity of the encoded epitope.
We have shown that this particular DNA vaccine strategy also increases the ability of B cells to present antigen, providing another mechanism to increase epitope immunogenicity.
Importantly, we have shown that with this particular type of DNA vaccine can be used to induce responses against epitopes that are generally subdominant or even silent in the context of natural infection. Thus, it is possible to expand the repertoire of the CD8+ T cell response via this strategy.
To test this concept for therapy of HIV, we constructed a hepatitis B core plasmid containing 19 CTL epitopes that were Mamu-A*01 restricted for the SIV virus. These epitopes can be inserted either into the internal region of the core antigen gene or at its carboxy terminus. Three epitopes can be inserted into both of these regions without disrupting expression of a fully assembled core particle.
There are two other important aspects of this vaccine. One is that the selection of the epitopes for the vaccine was based on the high level of conservation of these epitopes between the strain that they were derived from, which is SIVmac239, and the challenge strain that we used, which is SIV/DeltaB670. SIV/DeltaB670 is a primary isolate that mimics exposure to diverse HIV isolates in the human population. In this vaccine, 14 of the 19 epitopes we selected are 100 per cent conserved between these two viruses.
Two additional plasmids expressing the whole SIV gag and tat genes and also derived from SIVmac239 were included in the vaccine to stimulate virus-specific T cell help in combination with the CTL responses.
(Click on image for a larger version)
Rhesus macaques were infected with SIV/DeltaB670 and then treated with the antiretroviral drug PMPA two weeks after the infection. Six months later, the drug was removed to measure viral rebound.
During drug therapy a set of eight monkeys received a series of 6 DNA immunisations spaced one month apart. A second set of monkeys that were initially used to test the vaccines for immunogenicity were included in this study and immunised both before and post-infection. A control group was mock-immunised with empty vectors only both before and after infection.
(Click on image for a larger version)
The shaded box indicates the period of time the animals received uninterrupted treatment with the antiretroviral drug, PMPA. The dotted line shows the level of virus load that was our criterion for vaccine efficacy. We selected this level because it is comparable to viral loads in long-term non-progressors. So efficacy is defined as induction of a long-term non-progression status in these animals. In the mock-vaccinated control group all of the animals except for one rebounded after the drug was withdrawn.
In the vaccinated groups, one group was immunised before and after infection, and the other group immunised only after infection. In the animals immunised both before and after infection, seven of eight completely controlled the virus for a minimum period of six months, and of these, six continued to maintain control for a period of one year after the drug was removed. In the group that was immunised only post-infection, four of the eight animals were able to control the virus for six months, and three of these four continued to maintain viral control for a full year.
Looking more closely at the data, some animals responded very well to the drug and some animals failed to respond, and only animals that responded well showed a significant response to vaccination
Therefore, vaccine efficacy was highly dependent on the response to the drug. Among the 11 animals that responded to the drug, the vaccine was able to induce durable control of the virus in essentially 100 per cent of all of the animals.
With respect to disease progression, all eight of the animals in the mock-vaccinated group showed CD4 decline, including the one animal that had maintained control of virus load. This particular animal likely had most of its viral replication occurring somewhere other than in the blood.
In animals immunised before and after infection, one animal lost control of the virus after about six months. This animal showed some degree of CD4 decline. In addition, the one animal that did not respond to the drug also showed CD4 decline. In animals immunised only post-infection, the four animals that were drug non-responders showed CD4 decline, but all four animals that were vaccinated with the DNA vaccine were protected against CD4 decline for a full year after the drug was removed.
Immune responses were compared in animals that did not respond to the drug to those that did respond well to the drug. CD4+ T cell responses measured by proliferation assay showed no effect of the vaccines in animals that didn’t respond to the drug. However, in animals that responded well to the drug, both vaccines and immunising before and after infection or only after infection, induced a significant enhancement of CD4 responses.
CD8+ T cell responses were also measured by ELISPOT by depleting CD4 T cells. The vaccines had no effect on the CD8 response in animals that failed to respond to the drug but both vaccines elevated CD8 responses in animals that responded well to the drug when compared to the unvaccinated controls.
We also examined the breadth of the CD8 response by measuring the immune responses against eight of the 19 CTL epitopes that were included in our vaccine. The breadth of the response was measured in different phases – before immunisation, after immunisation was completed, and then after the drug was removed in controls and vaccinated animals. Overall, vaccinated animals developed a greater breadth of CD8 response when compared to the controls, demonstrating that the vaccine was able to broaden the CD8+ T cell response.
Animals that were immunised before infection had a broad CD8 response both early and late in the study whereas those that were immunised only after infection initially developed a more limited response, but over time the CD8 response broadened to ranges that were comparable to the animals immunised both before and after infection.
(Click on image for a larger version)
To summarise this particular study: we used a regimen where the DNA vaccine was combined with antiretroviral therapy and administered for a period of six months starting two weeks post-infection. Eleven of the 16 vaccines were protected against viral rebound and disease for over one year after the drug was stopped. In contrast, all eight controls demonstrated viral rebound and/or CD4 decline. Immunising before and post-infection enhanced post-infection therapy, suggesting that prophylactic vaccines that fail to protect against infection may still improve prognosis of post-infection immunotherapy.
Viral control correlated with higher and broader T cell responses, including both the CD4 and the CD8 response. Importantly the vaccine efficacy was dependent on the virological response to PMPA.
However, in animals that did respond to the drug, we found that vaccination was unable to prevent viral rebound in all 11 drug responders.
(Click on image for a larger version)
We have demonstrated the feasibility of using a DNA vaccine for therapy of HIV and have shown that a DNA vaccine effectively increased T cell responses in SIV-infected macaques and prevented viral rebound and AIDS after the drug was withdrawn.
In this study, antiretroviral drugs were started early, during acute infection, whereas in HIV-infected populations drug therapy is usually started after establishment of a chronic infection. So this was an idealised model. An important issue is that the vaccine was ineffective in the drug low responders.
So in our next study, we started HAART after establishment of a chronic infection to more closely mimic treatment of HIV infection in the clinic. We investigated a novel adjuvant strategy to improve the effects of vaccine immunotherapy in the drug poor responders.
(Click on image for a larger version)
Adjuvants that can improve DNA vaccine immunogenicity may improve their therapeutic efficacy. We also wanted to determine if an adjuvanted DNA vaccine was effective if started after chronic infection was established. We hypothesised that adjuvants that increase systemic and mucosal responses would improve therapeutic efficacy, and may be able to improve the outcome in the drug low responders.
The adjuvant that we used is a plasmid encoding the A and B subunits of the E. coli heat-labile enterotoxin (LT) or cholera toxin (CT). The plasmids encode the whole toxin, but there is no systemic or local toxicity even when much larger quantities of these toxins are delivered as proteins to the skin. The DNA plasmid expresses nanogram quantities of these toxins in the skin and plasmids expressing the whole toxin were much more immunogenic than mutant forms. These adjuvants activate DCs. In mice, we previously showed that CT and LT increased Th1 CD4 and CD8 T cell responses and protection against HSV and influenza challenges. These studies have been published.
One of the important reasons for selecting these adjuvants is that these are mucosal adjuvants, and delivering them with the DNA into the skin may be able to enhance the induction of mucosal immunity and perhaps improve therapeutic efficacy by targeting the viral reservoir in the GALT.
(Click on image for a larger version)
For this study, we used a DNA vaccine encoding RT, nef and gag. This vaccine was selected because as part of a collaboration with GlaxoSmithKline, and in parallel, GSL was testing a similar HIV DNA vaccine in humans.
We included an additional plasmid DNA vaccine expressing envelope because we had evidence from our first study that animals with envelope-specific responses better controlled the virus. The plasma expressed in the LT adjuvant had both the A and B subunits expressed on the same plasmid.
(Click on image for a larger version)
We again used the SIV/DeltaB670 which is a primary isolate and heterologous to the vaccine sequences. The drug was started in this study six weeks after infection, or after establishment of chronic infection using this virus. This is a large study, so we had two cohorts of animals with staggered starting times. The data I am going to show you as cohort A is at eight months after the drug was stopped and cohort B is at five months after the drug was stopped.
During the course of drug therapy we administered six-monthly DNA doses, starting eight weeks after the drug was started. At that point we were able to stratify the animals into groups based on virus load because we knew from our previous study that the drug was not going to be effective in half of the animals. The animals were stratified so that we had an equal number of drug non-responders in each of the three groups. As shown here, we had a total of 16 animals per group: one group that was immunised only with the DNA vaccine, a second group was immunised with the DNA vaccine co-delivered with the LT adjuvant and a third group served as ART treated controls.
We increased the number of animals per group from eight in our previous study to 16 in this study because we expected that half of the animals would not respond to the drug.
Prior to the therapy study we first tested the LT and CT adjuvants for immunogenicity in rhesus macaques, to determine whether CT and LT was the more effective adjuvant. The results showed that CT and LT adjuvants both augmented CD8 responses, but the LT adjuvant was better than the CT adjuvant. Both CT and LT augmented CD4 responses, with about comparable enhancement.
Importantly, only the LT adjuvant was able to augment mucosal immune responses in the gut-associated lymphoid tissue. Since induction of mucosal responses in the viral reservoir was a goal of our study, we selected the LT adjuvant for testing in combination with drug therapy in SIV-infected macaques.
Analysis of viral loads in the unvaccinated control animals treated with antiretroviral drug therapy (ART) showed that that only 38% (5/13) of animals responded very well to the drug whereas 46% (6/13) of animals did not respond. Two additional animals were moderate responders to ART. Animals regarded as good responders to ART demonstrated viral load reduction during therapy that was below our proposed threshold for vaccine efficacy. In the control group, all but one of the animals that responded well (N=5) or moderately well (N=2) to ART showed viral rebound after ART was removed.
In the DNA vaccine groups, 13 of the 28 animals (46%) didn’t respond to ART whereas the others responded well (N=10, 36%) or moderately well (N=5, 18%) Once viral responsiveness was determined, the animals were stratified into three groups, each containing approximately equal numbers responding to the drug.
In the animals that did not respond to the drug, DNA immunisation had no apparent effect in improving viral control. SIV/Delta B670 is a very aggressive virus and most untreated animals will progress to AIDS by 60 weeks post-infection. However, 57% the ART non-responders in the DNA plus LT group, were still surviving by 84 weeks post-infection compared to only 14% and 17% in the unadjuvanted DNA vaccine and control groups, respectively, suggesting that despite having no apparent effect on viral replication, the adjuvant may still have conferred some benefit in delaying disease.
Data from animals that did not respond to drugs were excluded from analysis with ART responders so we could more effectively look at the effects of the DNA vaccine in ART responders versus ART non-responders.
Looking only at the animals that responded well to the drug, all animals but one showed viral rebound after the drug was stopped. In the unadjuvanted DNA vaccine group four of the eight animals were able to control the virus below 1000 viral RNA copies, whereas the other four animals had virus rebound to high levels. In the group that was immunised with the LT adjuvant, all of the animals except one (6 of 7) were able to control the virus for a period of either five or eight months after the drug was removed. The difference between the LT group and the controls, with respect to the number of animals in each group that contained viral rebound, was statistically significant (P < 0.05, Fisher’s exact test).
Immunisation with either the unadjuvanted or LT-adjuvanted DNA vaccine induced insignificant elevation of T cell responses during ART as measured by ELISPOT.. We also looked at the drug non-responders, and you may remember that in our previous study we didn’t see any effect of the vaccine on the T cell responses. In this study, the responses were wildly variable in the drug non-responders group, but ART nonresponder animals immunised with the LT adjuvanted DNA vaccine showed consistently higher T cell responses throughout the study..
I am going to focus the rest of my talk on the effects that we saw in the drug responders.
T helper cell responses were analysed in drug responders prior to the initiation of DNA vaccination, and then after the DNA vaccination was completed but before the drug was removed. We observed elevated responses only in the LT adjuvanted group. Then after the drug was removed these responses were further increased.
We also looked at the effects of vaccination on cytokine responses, using a cytometric bead array analysis. We saw augmented gamma-interferon, TNF-alpha, interleukin-2 and, interestingly, an augmented interleukin-6 response as well, in the CD4 response.
We looked at correlates of viral control in the CD8+ T cell responses by comparing animals that controlled the virus in the vaccinated groups with those that failed to control the virus. Controller animals maintained the virus below 100 copies for a period of at least 3 months after the drug was removed compared to noncontroller animals that have a virus load over 5000. ICS was used to analyse responses besides the gamma-interferon response, including CD107a (which is a marker for a cytolytic effector response), TNF-alpha and interleukin-2.
Animals that were controllers of the virus showed enhanced cytolytic effector responses (CD107a) and TNF-alpha when compared to animals that did not control the virus.
Of course, a significant question in this particular study was whether we were able to have an impact on the viral reservoirs in the lymph node and gut. Was the vaccine controlling the viral replication in the peripheral blood, or did it actually target the gut and reduce the viral reservoir?
Virus loads in the gut and lymph nodes prior to removal of the drug show that the DNA vaccinated group without the adjuvant had no apparent effect on virus in the gut when compared to the control group. But the group that received the LT adjuvant showed a significant reduction in viral load in axillary lymph nodes as well as the lamina propria or the gut-associated lymphoid tissue.
We analysed the mucosal reservoir three weeks after the drug was removed, at a critical timepoint prior to the point when an animal is going to rebound or not. Consistent with the virus load data in the gut measured before the drug was removed, we saw that only the LT adjuvanted group was able to show significant reduction in virus loads in the mucosal viral reservoir.
We then looked at the effects of the therapeutic vaccine on mucosal immune responses in the gut. In this experiment, we take a small section of jejunum from the macaque and isolate the mononuclear cells from the laminar propria. The LT adjuvanted vaccine provided some elevation of the mucosal responses in the gut relative to either the control or the DNA vaccinated only, but this was not statistically different and it doesn’t necessarily explain the significant impact that we had with this vaccine on the mucosal viral reservoir. So we investigated other possible immune correlates.
In this experiment, we also looked at whether the vaccine influenced the breadth of the T cell response in 5 representative animals from each group.
In the control group, we observed a limited dominant response against only 1 or 2 peptide pools in the nef and/or pol genes.. In contrast, in the unadjuvanted and LT-adjuvanted DNA vaccine groups T cell responses were directed against 4 – 8 peptide pools that included the gag, envelope, nef, and pol genes.
One animal in the control group showed a broader response against 4 peptide pools However, this animal was undergoing a significant viral rebound at the time we took this sample and typically at the end stage of the disease, we see a transient broadening of the T cell response just as the virus begins to rebound.
The DNA-vaccinated group showed a trend toward a dominant gag response. Responses against one particular gag peptide pool were consistently detected in three animals that maintained durable control of the virus. In contrast, two animals in this group that did not control viral rebound lacked this response.
The LT-adjuvant induced an even broader immune response that included responses against gag and envelope as well as polymerase and nef. The only animal that showed viral rebound in this group also had a more limited response against nef and pol that resembled responses in the control group.. In contrast, the one animal in this group that maintained control of its virus load below the limits of detection throughout the study showed the broadest response against 8 peptide pools that included gag, envelop, nef, and pol. We have not yet been able to isolate virus from the blood of this particular animal. Two additional animals that showed a similar broad T cell response profile also maintained a virus-negative status after ART was stopped
(Click on image for a larger version)
In summary, what we found was that DNA vaccine immunotherapy, when combined with an effective drug, prevented viral rebound after the drug was stopped.
We found that the LT adjuvant increased both DNA vaccine immunogenicity and therapeutic efficacy. LT increased responses in both the drug responders and non-responders, but had no effect on the viral load in the low responders. Significantly, the LT adjuvant significantly increased viral control in the gut reservoir and lymph nodes, indicating that there was significant clearance of the virus in that viral source.
Viral control post-drug was also associated with increased T cell function. We saw increased CD4+ T cell responses as well as an increase in CD8+ T cell effector responses of multiple functional capacity.
We have a number of experiments still in progress. This is a study that is still ongoing, in which all the animals will be brought to a full year after the drug is removed. We are continuing to look at immune correlates of viral control to try to understand what immune responses are associated with viral control in a relatively large subset of animals in which we have induced viral control that resembles long-term non-progressors. We are also looking at B cell responses, B cell ELISPOTs, particularly in the mucosal compartment.
(Click on image for a larger version)
I would like to acknowledge and credit my laboratory at Albany Medical College and in particular Mickey Corb at the University of Pittsburgh, who is a collaborator on this study and does all of the virology and pathology for these studies; David Watkins and Todd Allen, who contributed significantly to the design of the epitope-based DNA vaccine; PowderJect, where this study actually got started when I was employed there; GlaxoSmithKline for contributing the DNA vaccines in the second trial; and of course funding from the National Institutes of Health.
Discussion
Question: I am curious: are you able in this model to look at, especially with ART therapy, the evolution of drug-resistant viruses, and can you say anything about whether the immunotherapy either delays it or changes the distribution of mutants?
Deborah Fuller: Well, that goes to why we had some animals that responded to the drug and others that didn’t. Drug non-responsiveness was not associated necessarily with an emergence of drug-resistant variants, although they did eventually emerge in those ones. Animals that tended to respond better to the drug and the vaccine had a more limited viral repertoire in the beginning of the infection. Those animals that failed to respond to the drug had much broader viral diversity by the time they went on the drug.
It would suggest that there is a predisposition, in a sense, to be able to respond better to the combination of a vaccine and drug therapy, although in humans, drugs tend to be much more effective. So in future studies, we will try a better drug that can control virus completely in the majority of the animals and address the question of whether the effects of the vaccine are due to synergy with the drug responsiveness or if there is just some predisposition for these animals to respond better.
Question: I find it so interesting that breadth of response is important, that you basically have T cell epitopes that are conserved. I just wondered if you had thoughts about the contrast between your studies and the studies performed by Merck. I know those were prophylactic vaccinations, but they decided not to use DNA vaccines. Clearly you are showing a great effect, but they chose to go with adenovirus. Do you have some thoughts about that?
Deborah Fuller: My understanding of Merck was that the eventual direction was to incorporate a DNA vaccine, after they started with the adenovirus vector. But DNA vaccines alone, particularly in the prophylactic immunisations, especially administered by intramuscular injection, are notoriously poorly immunogenic, and so there is not a lot of promising data initially going into a prophylactic study.
What was different here was that we were immunising animals that have already been primed for an immune response and we are boosting those responses. In fact, the gene gun has been brought into the clinic in several influenza DNA vaccine studies which show that it is hugely immunogenic when there is already a primed response there, as you come back and boost that. The gene gun has worked very effectively alone, even without a viral vector boost, and in naïve subjects as well for hepatitis B and the like.
I think that DNA vaccines got a bad rap initially because the first ones that were brought into the clinic were kind of duds and did not work very well. But there are a lot of new technologies coming out – delivery, electroporation and gene gun alike – in which DNA vaccines may gain some more credibility.
Question: Would you like to comment on how you think the LT is contributing? In particular, there has been some suggestion that when you use it percutaneously it can induce autoimmune disease. Is this a problem when you are actually giving it with the low dose as a DNA construct?
Deborah Fuller: We think that with the small quantities – they are nanogram quantities – you don’t see the kinds of effects that you see when you administer it as a protein or any sort of visible inflammatory reaction. We think we’ll get a microscopic inflammation. As to the purpose of the LT adjuvant, our hypothesis was that it was going to drive a stronger mucosal immune response. We didn’t in the end really see a significant difference, but we did see some effect in the breadth of the response, that maybe it is inducing responses against additional epitopes that would not normally have been induced in the context of an unadjuvanted DNA vaccine.
There are two types of controls I would love to have done if I had more monkeys. (I had 48 monkeys already in the study.) One would be to immunise with the LT alone. Could that alone have had an effect? Were there some innate factors in there? The second was to immunise in the absence of drug therapy, without the drug. Could vaccination have an effect even without the drug? Hopefully, in the future maybe we will be able to do those controls as well.
Question: You started the first studies with multiple epitopes and the core one. Why did you switch to the other one when you were trying to answer a different question? Is that because GSK forced you to do that?
Deborah Fuller: No, it is somewhat political. First of all the funding mechanism that we went after required an industry partner, and that was their interest and their vaccine and we wanted to run these in parallel with the clinical trial. The other issue is that it was very hard to find enough MHC-defined animals, because we would have needed A*01-positive monkeys and they just weren’t out there.
We haven’t abandoned that approach. I have another trial ongoing right now with a much more limited number of monkeys. But what we did was to take eight of the epitopes that we consistently saw in the animals that controlled the virus in the first study and reformulate them into a single hepatitis core vaccine. We are actually testing that right now. It is just not developed.
Question (cont.): I see this as a very worrying trend in the HIV vaccine. When you had such a promising result you could not go any further, and I feel there was probably peer pressure from the company or something. This is the problem that the HIV vaccine has. The approaches which have been used have no scientific validity.
Deborah Fuller: Well, there is politics, but as I mentioned we are still pursuing that particular method. It is just in a much more reduced capacity at this point.
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
