April 9, 2002
In the first late breaker session, Danny Douek (from the new Vaccine Research Center at NIH) presented new data illuminating the intimate interaction between HIV replication and the HIV-specific T cell response (Douek, 2002). Douek compared the frequency of proviral HIV DNA in T cells specific for HIV compared to T cells specific for CMV. CD4 T cells were sorted based on whether they produced the cytokine interferon-gamma in response to HIV- or CMV-derived peptide pools (peptides are small slices of viral proteins). The amount of HIV gag DNA in each cell population was then measured using real time quantitative PCR. The PCR test was shown to be sensitive down to 1 copy of HIV's gag gene in a single T cell.
Douek analyzed peripheral blood samples from 12 chronically infected individuals with detectable viral loads. The results revealed that 1-10% of all infected CD4 T cells were HIV-specific when analyzed based on interferon-gamma production (although not all HIV-specific T cells make this cytokine, suggesting that this could be an underestimate). HIV-specific CD4 T cells comprised by far the largest proportion of infected cells relative to those of other specificities, including CMV. Despite this preferential infection, however, Douek noted that the majority of HIV-specific T cells appeared uninfected at any given time. Douek went on to theorize that during acute infection, HIV might preferentially infect naïve CD4 T cells as they become activated by their first encounter with HIV antigens and attempt to mature into HIV-specific "effector" and "memory" T cells. To test this hypothesis, he exposed activated naïve CD4 T cells to R5-using HIV isolates in vitro and measured the percentage of cells staining positive for p24 antigen over time. The results demonstrated that viral replication increased as a function of each cell division. After one division, 3.1% of the T cells were producing p24, but over the next two divisions the proportion increased to 12% and 13.6%, respectively (the number of proviral HIV DNA copies detected followed the same pattern). Thereafter, the percentage of virus-producing cells began to decline, coincident with an increasing ability of the activated T cells to produce interferon-gamma.
The current model in immunology posits that the majority of activated naïve T cells die within a day or two, a phenomenon called "activation-induced cell death." Some, however, mature over the course of multiple divisions (between 7 and 14) into "memory" T cells. Memory T cells acquire enhanced infection-fighting skills due to permanent cellular changes that occur during the division process (for example: cytokines can be produced immediately upon activation, without the delay Douek saw in activated naïve T cells). Some memory T cells can rapidly reinitiate cell division upon re-encounter with their specific antigen. Douek next tested whether activated memory CD4 T cells were as susceptible to HIV infection as activated naïve T cells. Again, he found that replication appeared to increase as a function of division, but at considerably lower levels: the proportion of infected memory CD4 T cells at each of the first four divisions was 1.3%, 4.4%, 2.4% and 1%, respectively. The apparently reduced susceptibility of memory T cells to R5 HIV infection echoes recent results from researchers in the UK (Vyakarnam, 2001), and may have important implications for vaccines: the goal of many new preventive and therapeutic vaccines is to induce the differentiation of naïve T cells into HIV-specific memory T cells, either before a person is exposed to HIV or while viral replication is controlled by antiretroviral therapy. One possible implication of Douek's work is that protecting activated, dividing naïve CD4 T cells -- which he describes as "exquisitely sensitive" to HIV infection -- may allow the maturation of an HIV-specific memory T cell population that is less susceptible to invasion by the virus.
In concluding his presentation, Douek sounded a note of caution regarding the impact of structured treatment interruptions (STI) on HIV specific CD4 cells in chronic infection. In an analysis of four NIH study participants undergoing STI, the number of virus-infected HIV-specific CD4 T cells increased dramatically, representing over 50% of the proviral DNA-containing cells in one individual. These results suggest that STI alone may be something of a Battle of the Somme approach to immunotherapy, because the attempt to trigger new memory T cell responses by allowing HIV viral load rebound could be offset by the enhanced susceptibility of activated naïve CD4 T cells to infection. However, Douek did not speculate as to whether changes in the relative contribution of memory vs. naïve T cell responses during repeated STI might influence the eventual outcome, as has been suggested by renowned cellular immunologist Rafi Ahmed.
Viral protein R (vpr) is an HIV accessory protein that has been associated with a number of effects on virus-infected T cells in vitro, including apoptosis (cell suicide), blocked activation and cell cycle arrest. The latter phenomenon represents an inhibition of cell division: as cells divide, they go though a number of phases known as G1, S, G2 and M (for mitosis or division). In vitro, vpr has been shown to arrest cells at the G2/M phase of the division process, but it has been unclear whether the protein can cause this problem in vivo. At the Retrovirus conference, Michael Sherman from the Gladstone Institute in San Francisco presented some of the first evidence that vpr-induced cell cycle arrest does in fact occur in HIV-infected humans (Sherman, 2002).
Sherman's study used an antibody that stains intracellular p24 to identify HIV-infected CD4 T cells from the peripheral blood of recent seroconverters (individuals who had become HIV-positive within the past than six months). Real time PCR was also employed to assess proviral DNA content. Sherman reported that infected cells exhibited a four-fold increase in the ratio of G1 compared to G2 DNA content, an indicator of cell cycle arrest. The even more dramatic finding was that, at least in Sherman and coworkers hands, essentially all p24-positive CD4 T cells were arrested in vivo. In stark contrast, activated but uninfected T cells from the same individuals showed normal cell cycle distributions.
The Gladstone team conducted a separate analysis to look for evidence that HIV infection of CD4 T cells increases levels of apoptosis, another in vitro effect often attributed to vpr. CD4 T cells from ten individuals were isolated based on staining for both p24 and annexin V, a known marker for cells undergoing apoptosis. This experiment could not demonstrate significant differences in annexin V staining between infected and uninfected cells, suggesting that some of the in vitro effects of vpr may not be borne out in vivo.
The significance of vpr-induced cell cycle arrest in the pathogenesis of HIV infection has yet to be fully elucidated, but a number of research groups have reported that arresting infected, cycling CD4 T cells is likely to enhance virus production in vivo. The fact that progression through the cell cycle is critical for certain CD4 T cell functions also raises the possibility that vpr-induced arrest represents yet another immune evasion tactic employed by HIV.
Renowned Parisian immunologist Brigitte Autran helped kick off the first full day of the Retrovirus conference with a half-hour plenary on immune reconstitution and correlates of long-term non-progression (Autran, 2002). Autran began by reviewing seminal work from her group at the Hopital Pitié-Salpêtrière, originally published in the journal Science in 1997. These studies described the rapid redistribution of mainly memory CD4 T cells from the lymph nodes after HAART initiation, accompanied by a precipitous decline in markers of immune activation and followed by a slow but steady increase of naïve T cells over subsequent years. Concomitant improvements in T cell responses to common opportunistic pathogens were also reported. All of Autran's initial observations regarding HAART-induced immune reconstitution have since been confirmed and extended by other research groups. The one type of immune response that is typically absent or weak in HAART-treated individuals is HIV-specific; CD4 T cell responses to the virus can be hard to detect and, if detectable, may be functionally compromised. CD8 T-cell responses are typically detectable but also show evidence of functional defects and appear unable to keep pace with viral evolution.
The new research priority for Autran's group is to ascertain whether this apparent deficit in HIV-specific immunity can be addressed with therapy. As a first step, a cohort of long-term non-progressors (LTNP) was analyzed in order to try and define clear correlates of immunologic control of HIV infection. No factors related to the infecting virus were identified, but a number of associations relating to the immune response emerged. As reported by others, certain class I and II HLA genotypes are more common in LTNP, suggesting that the ability of both CD8 and CD4 T cells to recognize and respond to HIV is a key variable influencing the outcome of infection. Using an ELISpot assay to identify HIV-specific CD8 T cells based on their ability to produce interferon-gamma, Autran reported that stronger and more broadly targeted responses were detected in LTNP compared to progressors, although this finding has not been duplicated by some groups (Betts, 2001). Autran also looked at the expression of the cell-killing enzyme perforin in HIV-specific CD8 T cells and found that fewer cells were perforin-positive in LTNP than people with progressing disease, indicating that this is not a useful correlate (as previous study results had suggested -- see Appay, 2000).
Turning to HIV-specific CD4 T cells, the correlation between the proliferative response to p24 and control of viral load in Autran's study was highly significant (p=0.0001). A similar link was seen when, instead of proliferation, the frequency of CD4 T cells making interferon-gamma in response to p24 was measured by ELISpot. Since other studies have found that the frequency of HIV-specific CD4 T cells (as measured by ELISpot) is not always correlated with control of viral load (see also Betts, 2001), Autran's group identified another marker that may address the functionality of the CD4 T cell response. Interferon-gamma production is typically associated with a type of CD4 T cell response known as T-helper type 1 or Th1, considered to be important in the defense against many viral infections, including HIV. It is known from basic immunology research that Th1 responses drive B-cells to make a particular class of antibody called IgG2. Although many studies have measured HIV-specific antibodies in infected individuals, Autran's colleague Nicole Ngo-Giang-Houng looked specifically for those belonging to the IgG2 subclass, reasoning that they might be a marker for robust, Th1-type HIV-specific CD4 T cell activity. As reported last year (Ngo-Giang-Houng, 2001), she found that IgG2 antibodies targeting HIV's gp41 protein were strongly associated with persistent LTNP status. Perhaps surprisingly, the presence of these antibodies was not linked to an ability to neutralize two primary isolates or the lab strain HIVLAI (although all viruses tested were syncytium-inducing, leaving open the possibility that non-syncytium-inducing isolates would be affected differently).
Autran updated these findings at the conference by assessing whether maintenance of LTNP status (over five years of follow-up subsequent to study entry) could be predicted by combining the IgG2 and ELISpot results. The analysis revealed that a strong HIV-specific CD4 T cell response (over 170 spot-forming units or SFC in the ELISpot test) and the presence of IgG2 antibodies directed against gp41 correlated strictly with LTNP. One hundred percent of individuals with responses maintained their non-progressor status over five years of follow-up. In contrast, more than half the individuals with the same HIV-specific CD4 T cell response measured by ELISpot but no gp41-specific IgG2 antibodies experienced disease progression during this period. Over 80% of the individuals in the remaining two categories (those with gp41-specific IgG2 antibodies but lacking a strong HIV-specific CD4 T cell response and those with neither IgG2 nor a strong CD4 T cell response) developed progressive disease.
Having outlined the immune responses that may protect against progression, Autran described the strategies her research team is employing to try and create (or boost) the same type of HIV-specific immunity in people with acute and chronic HIV infection. The current goals are straightforward: to use candidate vaccines in conjunction with HAART in order to induce strong, broad and durable HIV-specific CD4 and CD8 T cell responses, in the hope of reducing viral load set point when HAART is stopped and thus extending the time that drug therapy can safely be withheld. A number of studies are underway (see table below) using just about every vaccine that's been shown to be safe and to induce at least a meager level of HIV-specific T cell activity: ALVAC, lipopeptides and Remune. Autran coyly referred to "encouraging intermediate results" from these studies, which indicate that new HIV-specific CD4 T cell responses can be induced in both primary and chronic infection. Autran reported that the frequencies of these responses are comparable to LTNP in vaccinated individuals with primary infection, but are slightly lower in study participants with chronic infection. CD8 T cell data are still pending. New trials are now planned with newer and potentially more immunogenic vaccines such as Merck's DNA/adenovirus combination.
|Study||Vaccine||Stage of Infection|
|ANRS 094||ALVAC vCP1433 (gag, env, epitopes from nef and RT)||Chronic|
|ANRS 093||ALVAC vCP1433 + lipopeptides + IL-2||Chronic|
|ANRS 095||ALVAC vCP1433 + lipopeptides + IL-2||Acute|
|QUEST||ALVAC vCP1452 (gag, env, pro, epitopes from nef and RT) +/- Remune||Acute|
In a rapid-fire half-hour plenary session, Merck's head of HIV vaccine research Emilio Emini reviewed the company's development program and unveiled the first human immunogenicity data from phase I trials of their DNA and adenovirus type 5 (Ad5) constructs (Emini, 2002). Currently, both vaccines contain only the gag gene from HIV-1 (clade B) but Merck plans to add nef and pol as the program moves ahead.
Emini outlined the studies that are in progress in both HIV negative and HIV positive volunteers (Merck is also pursuing these vaccines as potential therapeutics).
DNA HIV gag vaccine. The ongoing program involves two doses (1mg or 5mg) of gag-expressing DNA in saline or with one of two adjuvants: aluminum phosphate or the experimental nonionic block co-polymer CRL-1005.
Adenovirus 5 (Ad5) HIV gag vaccine. Separate trials are evaluating four escalating doses (108, 109, 1010 and 1011 virus particles or vp) of the gag-expressing Ad5 vector and the final phase of the program (which began in January) is offering recipients of the DNA vaccine an Ad5 booster immunization.
Emini presented data from the separate phase one studies of the DNA and Ad5 candidate vaccines, both conducted in HIV-negative volunteers.
The DNA vaccine trial is evaluating doses of 1 or 5mg compared to placebo, with injections given at 0, 4, 8 and 26 weeks. Side effects have thus far been mild: headaches, muscle aches and injection site tenderness but no consistent laboratory test abnormalities. Some transient detections of anti-DNA antibodies (an area of concern with DNA vaccines) were reported, but levels did not increase after repeat immunizations. The findings could not be confirmed with the standard Farr assay, leading Emini to believe that these results were likely to be false positives.
Presenting the immunogenicity data, Emini emphasized the role of DNA as a prime (in other words, a method of establishing a low-level memory T cell response that can be boosted by a subsequent immunization) and characterized the T cell responses overall as "moderate." At week 30, four weeks after the fourth immunization, 16 out of 38 individuals (42%) in the 5mg dose group are showing a positive response to the gag antigen on Merck's ELISpot assay (a positive response is defined as >55 spot-forming cells or <4 fold above the background). The geometric mean for these 16 individuals was 109 spot-forming cells (SFC). In the 1mg dose group, only 7/34 (21%) individuals showed a positive response. (T cells were not separated into CD4 and CD8 subsets for these assays, but Emini reported that both cell types were detected and the majority of the response appeared to be CD8 T cell mediated.) While these results obtained with DNA in saline are not particularly impressive, data on individuals receiving DNA with adjuvant is not yet available, and Emini believes that even these low-level responses might be sufficient to prime the HIV-specific T cell response prior to Ad5 boosting.
The second phase I trial from which preliminary data are available is the dose-escalation of the Ad5 vector. Twelve individuals (including three placebo recipients) are in each dose group and inoculations were performed at weeks 0, 4 and 26. Emini showed data after all three immunizations in the two lower dose groups (108 and 109 vp) and limited data from individuals who have received two immunizations of the 1010 vp dose. Looking first at adverse events, the vector has caused some moderate and sporadic injection site reactions along with malaise and body aches. There have been no consistent lab abnormalities reported. The immunogenicity data is, as anticipated, considerably more encouraging than that obtained with DNA. At the lowest dose, 6/9 individuals have mounted a significant T cell response to gag (as with the DNA study, both CD4 and CD8 T cell responses were included in the assay). At week 8, after the first two shots, the average reactivity (again measured by ELISpot) was 228 SFC. Four weeks after the final immunization (at week 30), the average (among the same 6 responders) was 439 SFC. Data at the next dosing group (109 vp) is only complete at the week 8, at which time 4/9 participants have mounted a positive T cell response averaging 149 SFC. After the third immunization, 4 of 7 (two remain to be analyzed) are responding with an average of 337 SFC, and one of these individuals lacked a response at week 8 (leading Emini to anticipate the total number of responders in this dose group will be 5/9). The 1010 vp dose group has only been evaluated at week 8, and 5/9 individuals show a positive T cell response to Gag, averaging 257 SFC. To put these numbers in some context, compared with other vaccine vectors studied to date these ELISpot results represent some of the best T cell responses yet seen in humans. Although study follow-up is still in the early stages, Emini also provided a brief look at some data relating to the longevity of the T cell responses induced by the Ad5 vector. Four responding participants in the lowest Ad5 dose group have been followed out to 16 weeks post-boost and their gag-specific T cell responses are essentially unchanged.
Emini next addressed one of the key questions relating to the Ad5 vector: the effect of pre-existing adenovirus neutralizing antibodies on immunogenicity. Many people have been exposed to adenoviruses and consequently possess antibodies against the virus. The prevalence of these antibodies in the population remains uncertain, however, with estimates for North America starting at around 40%. Because it was clear from the ELISpot data that some study participants were not responding at all to Ad5, Emini drew a plot to assess whether this lack of response was associated with an anti-adenovirus neutralizing antibody (NAb) titer above 1:200. The results showed that only 1/3 non-responders in the lowest Ad5 dosing group had a NAb titer above this cut-off. However, 4/5 non-responders at the 109 vp dose were above the 1:200 level and the fifth had a NAb titer greater than 1:100. Notably, all the responders in these dosing groups had NAb titers less than 1:200. The one encouraging observation that Emini could point to is that although three non-responders at the 1010 vp dose had titers over 1:200, so did 4/5 responders -- a strong hint that increasing the Ad5 dose might allow the blocking effect of pre-existing NAbs to be circumvented. This fits with Emini's theory that if antibodies neutralize 99% of the 1010 dose, there will still be 108 virus particles remaining and the data from the lowest dose group shows that this can be enough to induce an HIV-specific T cell response. Additionally, Emini noted that the cut-off of 1:200 was not entirely arbitrary: surveys estimate that 70% of North Americans that have been exposed to adenovirus have NAb titers below this level, and preliminary data from other regions is so far looking similar. Emini also expressed his hope that the DNA prime, Ad5 boost strategy would show greater efficacy in the face of high NAb titers. Ultimately, this issue can only be resolved by Merck's ongoing studies.
Another facet of Merck's vaccine development program involves analyzing the potential of HIV-specific T cells to recognize viruses from multiple clades. Emini reported that the Merck team has looked at T cell responses in HIV-infected individuals from the US, Brazil, Thailand and Malawi (infected with viruses from clades, A, B and C) and assessed their ability to react against a panel of clade B-derived consensus peptides (comprising the viral proteins gag, pol, nef, rev and tat) in an ELISpot assay. Both the frequency and magnitude of the response to gag, pol and nef was indistinguishable by region, indicating substantial cross-clade reactivity. The proteins rev and tat induced a lower level of T cell response from all study participants, and individuals from Malawi did not appear to respond to clade B rev and tat at all. Emini then expressed the correlation between T cell responses to the different clades using a logarithmic scale. The magnitude of T cell reactivity to gag and nef consensus peptides from clades A and C compared to the equivalent peptides from clade B was shown to be close to 1:1 in this analysis, although Emini did point out a few outlying individuals whose cross-clade T cell reactivity was poor. The ongoing phase I vaccine studies have provided Merck with another opportunity to assess cross-clade recognition. Testing of vaccine responders demonstrated that gag-specific T cells from 10/13 (77%) could cross-react with consensus gag peptides derived from clade A or clade C viruses.
Emini closed his presentation by outlining the results yet to come, which include immunogenicity data from the highest (1011 vp) Ad5 dose and the DNA/Ad5 prime-boost approach in HIV negative volunteers, as well as the safety and immunogenicity information from ongoing trials of these vaccines in HIV-infected individuals. Merck anticipates that these results will become available sometime in the summer of 2002.
In one of the noteworthy vaccine presentations at Retrovirus, scientists at the Wistar Institute in Philadelphia reported that they may have come up with a novel solution to Merck's anti-adenovirus antibody problem (Fitzgerald, 2002). The Wistar team, headed by vaccinologist Hildegund Ertl, reasoned that adenoviruses from a closely-related animal species might work as vaccine vectors while avoiding neutralization by antibodies targeted to human adenovirus strains. They tested their hypothesis by isolating a chimpanzee adenovirus of the serotype C68 (AdC68) from the lymph node of an infected animal. As predicted, antibodies against human adenoviruses were unable to cross-neutralize AdC68. To produce a vaccine vector, Ertl and colleagues followed the same procedure commonly used with human adenovirus type 5 (Adhu5), deleting the E1 region to make the virus replication-defective and allow carriage of DNA encoding the desired vaccine antigen.
In pre-clinical studies in mice, a dose of 5x105 particle forming units (pfu) of the AdC68 vector (expressing a truncated form of the HIV gag gene) induced a strong CD8 T cell response. An average of 9.7% of circulating CD8 T cells were gag-specific post-immunization, and this increased to 18.6% when a dose of 107 pfu was employed. In Ertl's experiments, a 10-fold higher dose of Adhu5 was needed to achieve an equivalent CD8 T cell response. In earlier murine studies employing a rabies virus antigen, a strong T-helper type 1 (Th1) bias was seen with the AdC68 vector compared to Adhu5 (these studies have just been published -- see Xiang, 2002). Th1-type immune responses are thought to be important in controlling many viral infections, including HIV. Seeking to explain the impressive immunogenicity of AdC68, Ertl and the Wistar team have conducted a number of in vitro studies, and this work is ongoing. Adenoviruses are known to target specialized antigen-presenting cells called dendritic cells (DC), and Ertl showed that AdC68 induces expression of an array of surface molecules on DC that are known to potently stimulate T cell responses. Venturing into virtually uncharted territory, Ertl also compared changes in gene expression in DC infected with AdC68 vs. Adhu5. Out of 5,000 genes studied, Adhu5 induced over-expression of five while AdC68 appeared to upregulate 150. While intriguing, more work is required to fully understand the functional implications of these changes. Ertl reported that the Wistar team is moving ahead with plans to study AdC68 vectors in non-human primates and eventually hopes to conduct clinical trials in humans.
At the single oral abstract session on vaccines, Genoveffa Franchini from the National Institutes of Health presented data from a macaque challenge experiment utilizing DNA and NYVAC (NYVAC is a highly attenuated version of the vaccinia virus) vaccine vectors (Hel, 2002). Animals were assigned to three groups of eight:
Franchini assessed post-immunization SIV-specific T-helper responses by lymphoproliferation (LPR) to the p27 gag protein and gp120 and found the most robust responses in the group C. A similar story unfolded for CD8 T cell responses, as assessed by tetramers containing the gag CM9 epitope (the full pre-challenge immunogenicity data from this study was published last year in the Journal of Immunology -- see Hel, 2001). Six months post-boost, all macaques were challenged intrarectally with the highly pathogenic primary SIV isolate SIVmac251.
Franchini reported that all but one control animal sustained high-level SIV viremia post-challenge, while 5/8 group B NYVAC-immunized macaques showed some evidence of restriction of SIV replication (to less than 100,000 copies at set point after a typical acute phase peak). In contrast, 5/8 group C animals that received the prime-boost vaccine regimen showed significantly lower acute viral load peaks and controlled SIVmac251 to undetectable levels at set point. Although complete data were only available out to six months post-challenge when her presentation was prepared, Franchini stated that these five group C animals have maintained control of SIV replication out to a year now with no cases of viral breakthrough (CD4 T cell counts have also been completely preserved). Several analyses have been conducted to look for both pre- and post-challenge correlates of this salutary outcome. Franchini discovered statistically significant correlations between pre-challenge peak T-helper responses to p27 gag and CD8 T cell responses to CM9 and post-challenge containment of SIV replication. Blunting of acute viremia was associated with CD8 T cell responses to both the gag CM9 and tat SL8 epitope, while the strongest post-challenge correlate of immune control of SIV was the strength of the T-helper response to p27 gag (as measured both by LPR and intracellular cytokine staining for SIV-specific interferon-gamma production).
While these results are impressive given the virulence of SIVmac251 compared to other challenge isolates, there is one potential caveat that needs to be borne in mind. Franchini's own research has recently shown that macaques with the class I MHC type known as Mamu *A01 tend to control SIVmac251 somewhat better than Mamu *A01-negative animals (Pal, 2002). Of the five macaques in group C that are controlling viral load to undetectable levels, four are Mamu *A01-positive. But similarly, 4/5 animals in group B that brought their SIV viral load below 100,000 copies are Mamu *A01, so Franchini feels the study still demonstrates a significant effect of prime-boost compared to NYVAC alone regardless of the potential influence of the class I MHC type. Although Franchini did not discuss plans for human trials with these vaccines, the European Vaccine Consortium (EuroVac, http://www.eurovac.net) is working on phase I human trials of a NYVAC vector containing gag, pol, env and nef.