Organized by the same people who put together the annual Retrovirus conference, the 2003 AIDS Vaccine conference drew several thousand researchers and a smattering of community activists to the Sheraton hotel in New York City this past September. The take home message from the meeting was that a number of vaccine constructs now in human testing appear able to induce HIV-specific cell-mediated immunity (comprising CD4 and CD8 T cell responses), and at least some of these candidates are likely to wend their way toward efficacy trials over the next several years. It remains unclear, however, as to whether such candidates can offer anything resembling complete protection from HIV infection—the more prudent hope is that they will reduce post-infection viral load and confer something akin to long-term non-progressor status on vaccinated individuals. The majority opinion is that effective neutralizing antibodies will be essential for complete protection, but, for the most part, efforts to induce such responses remain in the pre-clinical stages of testing. This report summarizes a few of the interesting data presentations from the conference (webcasts and abstracts can be found online at http://www.aidsvaccine2003.org/2003/).

Occult HIV Infection and Late Seroconversions in Exposed Seronegatives
Tuofu Zhu from Juliana McElrath's group at the University of Washington presented an update on a cohort of individuals who, despite repeated exposure to HIV from an infected partner, have remained seronegative over several years of follow-up. Zhu and colleagues have recently published data showing that some of these individuals harbor extremely low levels of HIV DNA, suggesting that they may in fact be infected but their immune systems are exerting extremely strict control of the virus despite the absence of antibody responses (or, alternatively, that the infecting virus is defective in some way—see TAG's May/June 2003 Basic Science Review (http://www.aidsinfonyc.org/tag/science/0603review.html#occult).

At the conference, Zhu reported that out of 93 exposed seronegative individuals identified to date, 12 have shown at least transient evidence of low-level infection, with two repeatedly testing positive for HIV DNA. Ten of these study participants have now fully seroconverted after a prolonged period of seronegativity ("late seroconversions"). According to Zhu, they appear to break down into two groups: five have maintained low viral loads, typically between 2,000-3,000 copies, while the other five have high viral loads of 10,000 copies or greater. Sequence analysis of the breakthrough virus suggests it is not closely related to that of the HIV-infected partner. As Zhu put it, this finding suggests that these individuals may have possessed "selective immunity" that protected them from their partner's virus.

This hypothesis was explored further by comparing env sequences (the genetic code of responsible for making the virus's outer envelope proteins) from the late seroconverters (LSC) to that of their partners and then conducting a similar analysis comparing env sequences from LSC to two consensus sequences—the first derived from 53 clade B samples from the Los Alamos database and the second from 19 clade B samples specifically taken from the Seattle area (where Zhu's cohort is based). Consensus sequences are essentially a composite version of the genetic codes of all the viruses analyzed, based on the most common features of each. Comparing LSC and their partners, sequences were on average 19.3% different, while LSC and consensus clade B sequences were 14.5% different and LSC and the consensus for Seattle were 13.7% different. The degree of difference between the env sequences of LSC and their partners versus LSC and the two consensus sequences was statistically significant (p≤0.001). In other words, the genetic code of the viral envelope in the LSC was more similar to the composite code of circulating viruses than it was to their partner's HIV. Although these analyses are preliminary, Zhu suggested they provide evidence that the LSC may have possessed immune responses that protected them from their partner's virus. Further work will look at the impact of pre-seroconversion HIV-specific T cell responses on the infecting viruses (and, hopefully, post-infection viral load).

Attack of the Replicons
An important new vaccine technology that has emerged in recent years involves using "naked DNA" to deliver genes encoding vaccine antigens into the body. However, although DNA vaccines can induce potent immune responses in mice and, to a lesser degree, non-human primates, the results from human trials have been disappointing. During a session on novel vaccine vectors, Peter Liljestrom from the Karolinska Institute in Sweden outlined a new approach to improving the immunogenicity (ability to induce immune responses) of DNA constructs. Liljestrom has created DNA "replicons" by inserting the replicase gene from Semliki Forest Virus (SFV) into a DNA plasmid (he has christened this construct a "self-replicative suicidal DNA vaccine"). In a typical DNA vaccine, a promoter gene is inserted from CMV and this drives the production of whatever protein antigens are encoded by the DNA (HIV Gag, for example). Liljestrom's construct also contains a CMV promoter, but in this case it triggers the expression of the SFV replicase, which in turn drives the production of the encoded vaccine antigens.

In order to compare the immunogenicity of this novel construct with a standard DNA vaccine, Liljestrom has collaborated with Andrew McMichael and Tom Hanke from Oxford University. Hanke & McMichael have developed a DNA vaccine encoding the majority of the HIV Gag protein along with a number of CD8 T cell epitopes from the virus, and Liljestrom has inserted the same genes into his DNA replicon. In laboratory experiments, both constructs appear to produce roughly the same amount of protein antigens when inserted into cells. But in immunogenicity studies in mice, it required three injections of 10 micrograms of the standard DNA vaccine to induce detectable HIV-specific T cell responses. In comparison, it only required one shot of 1 microgram of the DNA replicon to induce an equivalent HIV-specific T cell response.

The search for an explanation for this difference in immunogenicity led to an interesting finding. By conducting a series of experiments in the lab, Liljestrom showed that the DNA replicon was not able to directly infect dendritic cells, which are primarily responsible for presenting antigens to T cells and thus trigger the development of an antigen-specific T cell response. Instead, the inclusion of the SFV replicase leads to the apoptosis (cell suicide) of non-dendritic cells that take up the DNA replicon vaccine. It is these dying cells—which contain the vaccine antigens that the DNA replicon has produced within them—that are then taken up and processed by dendritic cells, which then go on to present the vaccine antigens to T cells. As it turns out, dying cells appear to be particularly efficient at triggering dendritic cells to perform their antigen-presenting function (the technical term is "cross-priming"), and Liljestrom's hypothesis is that it is this phenomenon that explains the significantly enhanced immunogenicity of his DNA replicon vaccine.

With support from the International AIDS Vaccine Initiative (IAVI), The Karolinska Institute will soon be moving a DNA replicon HIV vaccine into phase I human trials. It is hoped that the replicon will be able to offer enhanced immunogenicity compared to standard DNA vaccines, at relatively low doses (although researchers have attempted to improve the response to standard DNA vaccines by simply increasing the dose, the potential of this approach is limited by the fact that large amounts of DNA vaccine turn into an unwieldy goo, which is extremely difficult to inject).

Liljestrom's findings may also help explain the disappointing immunogenicity of the supposedly "optimized" version of Aventis Pasteur's ALVAC HIV vaccine vector. This construct, dubbed ALVAC vCP1452, was modified to included an anti-apoptosis gene in the hopes that it would lead to prolonged expression of vaccine antigens by cells that took up the vector. But results from human trials suggest that vCP1452 induced HIV-specific CD8 T cell responses less effectively than the original vCP205 construct. It appears possible that the inclusion of the anti-apoptosis gene reduced the potential for cross-priming and thus negated, rather than enhanced, the immunogenicity of the vCP1452 vector.

Modeling Breastfeeding Transmission in Baby Macaques: ALVAC to the Rescue?
The most commonly used animal model for evaluating HIV vaccine candidates involves rhesus macaques, which can be infected with the simian immunodeficiency virus (SIV). One of the criticisms of this model is that animals must be challenged with a high dose of SIV in order to ensure that all unvaccinated controls become infected. This does not accurately mirror the human situation, where infection typically occurs after repeated low-dose exposures and the risk of infection after a single exposure is thought to be very low (in the range of 0.001-0.1%). Martha Marthas from the University of California at Davis has developed a novel vaccine challenge experiment which involves exposing neonatal macaques to low doses of SIV contained in milk, which is fed orally to the animals thrice-daily for a period of five days.

Marthas has now used this system to evaluate the potential of a number of different vaccine approaches to protect neonatal macaques from SIV infection, and she presented her latest data at the conference. The following table demonstrates the results obtained to date:

So far, the search for correlates of protection has not produced any candidates. SIV-specific CD8 T cell responses in the blood (as measured by an ELISpot assay that captures cells based on their ability to produce the cytokine interferon-gamma) have only rarely been detected, although Marthas noted that more work on CD4 T cell responses might be worthwhile. The dramatically poorer outcome after CD8 T cell depletion also suggests to Marthas that these cells are playing a key role in the observed protection even though they don't seem to be being captured using the interferon-gamma ELISpot assay. Marthas reported that a trial designed to evaluate the ability of ALVAC to protect human newborns against breastfeeding transmission—HIVNET 027—has been in the works for some time and is still slated to take place in Uganda sometime in the future.

En Route to Remembrance: IL-7 Receptor Identifies Developing Memory T cells
On the basic immunology front, Rafi Ahmed presented new data from his lab at Emory University that may shed light on one of the key outstanding questions facing T cell immunologists (the data have since been published in Nature Immunology 4;12:1191-8). Upon initial exposure to a viral infection, naïve CD4 and CD8 T cells that specifically recognize viral antigens typically undergo a burst of activation and proliferation that, under most circumstances, eventually leads to clearance or control of the virus. The majority of these newly-produced T cells then die in a process of cellular suicide called activation-induced cell death (AICD). However, a subset of virus-specific T cells do not die but persist, often for the lifetime of the animal. These are known as memory T cells, and they play a key role in preventing reinfection with (or for persisting infections, reactivation of), the virus that triggered their development. For a long time, immunologists have searched for markers that might identify which of the T cells produced during the initial proliferative burst are likely to become long-lived memory cells, but without success.

Ahmed's team has now identified the expression of a particular T cell surface molecule, the receptor for the cytokine interleukin-7 (IL-7R), as a marker for developing memory CD8 T cells (whether this also applies to memory CD4 T cells is not yet known). Using a murine model involving infection with lymphocytic choriomeningitis virus (LCMV), Ahmed and colleagues noticed that a subset of CD8 T cells that proliferated during acute infection (about 8%) were expressing high levels of IL-7R and that, numerically, this approximated the proportion that would be expected to survive as memory T cells. As the mice were followed over time, and the majority of the CD8 T cells that initially responded underwent AICD, the IL-7R-expressing cells made up an increasing proportion of the LCMV-specific CD8 T cell response (51% at day 15, 67% at day 22 and >95% at day 40 and beyond).

To further prove that these were indeed developing memory CD8 T cells, additional experiments were conducted. Equal numbers of LCMV-specific CD8 T cells either expressing IL-7R (IL-7R+) or not (IL-7R-) were isolated from infected mice at various timepoints post-infection (days 8, 11 or 15) and transferred into uninfected animals. Three weeks later, these animals were infected with a bacteria (listeria monocytogenes) that was modified to encode the part of LCMV to which these CD8 T cells responded (an epitope called gp33). As predicted, the IL-7R+ cells responded as memory CD8 T cells would be expected to, undergoing a robust proliferation so that by day 5 post-infection, they comprised 20-30% of circulating CD8 T cells. In contrast, transferred IL-7R- cells were incapable of such an expansion. At the same day 5 post-infection timepoint, these cells only comprised only 0.2-0.5% of circulating CD8 T cells. Using mice genetically deficient in IL-7, Ahmed was also able to show that the presence of the cytokine and the survival signals it delivers through the IL-7R were essential for the development of memory CD8 T cells. While it will be important to first ensure that this finding is relevant to the human setting, there is reason to hope that the data will provide a means to better monitor the development of memory T cell responses to vaccination and infection (particularly HIV infection, where the development of a functional memory T cell response is believed to go awry).