Molecular Snapshots from CROI 2007

Highlights in virology and immunology, from HIV variation and endogenous retroviruses T-cell memory and exhaustion

By Richard Jefferys*

Historically, the annual CROI meeting has concentrated heavily on virology, while immunology and vaccines commanded the attention that you might expect for, say, the Oscar for best animated short film. But the first day of this year's CROI offered two parallel morning sessions on virology and immunology and, surprisingly, it was the latter session that was standing room only. In a similar shift, vaccines commanded a special symposium that lasted the entirety of Tuesday afternoon. The increased balance of the program is perhaps a welcome sign that the relative success of antiretroviral therapies has sharpened the focus on the complex immunological questions that confront researchers in the disparate fields of AIDS pathogenesis, preventive vaccine research, and immune-based therapy development.

HIV variation

During the CROI symposium on AIDS vaccines, Francine McCutchan from the US Military HIV Research Program (USMHRP) delivered one of her signature talks on global HIV diversity and inter-subtype recombination. McCutchan reviewed some of the ways HIV variation impacts vaccine research, including how it affects choices regarding vaccine antigens, clinical trial site selection, and also evaluation of breakthrough infections in vaccinees to assess whether they resulted from a mismatch between vaccine antigens and the infecting virus.

In terms of the current global pandemic, McCutchan listed the nine known HIV subtypes (A through K, except E which had its subtype status rescinded when it was found to be a recombinant) and reported that there are now 35 circulating recombinant forms (CRFs) in the Los Alamos HIV sequence database, as well as many unique recombinant forms (URFs) that have been reported from single individuals that are a potential reservoir of new CRFs.

Despite this extensive variation, McCutchan estimates that—at least currently—six globally prevalent subtypes (A, B, C, D, CRF01 and CRF02) account for more than 95% of HIV infections worldwide. Vaccine candidates in advanced stages of testing are based on one or more of these subtypes but many trial sites have a complex mix of subtypes and recombinant forms co-circulating. In preparation for vaccine trials in East Africa, McCutchan's team sequenced 120 full-length genomes from Kenya, Uganda, and Tanzania. While the majority of sequences were subtypes A, D, and C respectively, all three subtypes were present in each country and, unexpectedly, a third or more of the sequences represented URFs.

McCutchan explained how such recombinant forms arise: HIV is a diploid virus—each virion contains two RNA templates—and during each replication cycle the viral reverse transcriptase switches back and forth between the templates to generate its proviral DNA. Usually the RNA templates are virtually identical, but in individuals infected dually (or multiply) with different subtypes the RNA of two different viruses can become co-packaged into the same viral particle. Subsequently, when replication occurs, a provirus is generated that is a chimera of the two subtypes; the RNA template switching causes the new virus genome to contain alternating sections from each subtype.

McCutchan's logical inference is that circulating virus/vaccine antigen mismatch are more likely to occur in high-risk populations where dual infections and recombination are more common, and most vaccine trials are, by necessity, carried out in these populations. She has assessed the extent to which there is an epidemiological link between high-risk groups, dual infections, and recombinant forms, using a novel assay that can distinguish between these various viral forms using probes specific to the subtypes of interest. The technique was employed to study viruses from 1420 HIV-infected individuals in East Africa and 806 individuals in Thailand whose risk for HIV infection varies from low to high.

McCutchan showed that, as predicted, the proportion of individuals infected either dually or with inter-subtype recombinant forms increased based on their degree of risk; more than half the individuals from an urban high-risk cohort in East Africa were infected dually or with recombinant forms, and in Thailand these forms were more often seen in injection drug users. She also found that, over time, dually-infected individuals typically generate 5-10 different URFs that replicate to high enough levels to become detectable.

To look at how recombinant forms spread among high risk social networks, McCutchan and her colleagues developed a novel technique that analyzes the breakpoints in the recombinant HIV genome where the different subtypes have been joined together, and these breakpoints can potentially track inter-subtype recombinants through different social networks and populations. They conducted a painstaking analysis of breakpoints in 125 complete HIV sequences from Uganda, Kenya, Tanzania, Thailand, Myanmar, and China, and after protracted analysis created a database of each breakpoint. A grand total of 896 breakpoints were eventually identified, 521 in viruses from East Africa and 375 in those from Asia.

Using the data to look at the connections between different risk groups and countries, they found that in Thailand the recombinant viruses among IDUs shared more breakpoints and yet also contained more novel breakpoints compared to those found in lower risk heterosexuals. However, the analysis also showed that the heterosexual and IDU networks were interconnected and had been since the beginning of the Thai epidemic. Compared to the East African countries studied, there were fewer connections across borders and the breakpoints also showed that HIV strains found in Myanmar bridged the epidemics in Thailand and China. In East Africa, recombinants with shared breakpoints were also commonly identified in different countries, suggesting long chains of transmission of viruses from specific recombinant lineages.

McCutchan's groundbreaking work promises to add another dimension to the study of the molecular epidemiology of HIV infection. Among her conclusions from these initial studies is that dual infection—the proximal source of recombinant forms—is occurring at a higher rate than had previously been surmised. She also believes that CRFs may represent recombinant viruses that have entered lower-risk social networks and therefore have fewer opportunities to recombine due to the lower incidence of dual infections, a phenomenon evolutionary biologists call reproductive isolation. In terms of vaccine trials, McCutchan suggested the optimal approach will be to test candidates in many different types of social networks with varying degrees of risk, so that efficacy in multiple settings can be evaluated.

Electrifying DNA vaccines

Initial excitement regarding the immunogenicity and efficacy of DNA vaccines in small animal models largely dissipated when results in larger animals and humans proved far less impressive. At CROI, Michael Egan from Wyeth Vaccines explained why DNA vaccines may be about to undergo a renaissance. He first cited recently published data from the NIH's Vaccine Research Center showing that an optimized DNA vaccine consisting of multiple plasmids could induce detectable, albeit low-level and limited, T-cell responses in the majority of recipients in a Phase I trial, a first for any DNA vaccine (J. Infect. Dis. 194, 12, 1650). Egan then outlined the strategies now being pursued to try to build on this important milestone. Wyeth has long been interested in 'molecular adjuvants,' such as cytokine genes that are delivered in tandem with DNA-encoding vaccine antigens, but Egan showed data indicating that T-cell responses to an HIV DNA vaccine construct were only marginally enhanced by the inclusion of an interleukin (IL-) 12 gene. As a result of this and similar data from other groups, Egan explained that the focus has now shifted to new delivery methods. He listed the variety of ways DNA vaccines can be administered, including particle and jet delivery devices (intramuscularly or intradermally), topical applications, and—the focus of Wyeth Vaccines and the remainder of his talk—electroporation.

Electroporation involves delivering brief electrical pulses to the muscle after the injection of the DNA vaccine using a special 'wand.' The pulses induce transient pore formation in cell membranes, facilitating entry of the DNA and then the production of vaccine-encoded antigens. Electroporation also attracts inflammatory cells, including antigen-presenting cells, to the site of immunization. Egan showed data from an experiment in macaques demonstrating a massive enhancement of DNA vaccine immunogenicity using this approach. Two weeks post-immunization, animals that received a multi-gene HIV DNA vaccine plus IL-12 showed HIV-specific T-cell responses that averaged 816 spot forming cells (SFCs). The same vaccine, delivered with electroporation, led to responses that averaged 8141 SFC. Over longer term follow-up the differences between the two groups of animals increased. At 22 weeks post-immunization, T-cell responses averaged 3394 SFCs in electroporation recipients versus 74 SFCs, only marginally above background, in animals that received injection only. Egan pointed out that this represented a roughly 225-fold increase in DNA vaccine immunogenicity. However, Egan stressed that data from non-human primates has often failed to be duplicated in humans so confirmation of the electroporation buzz awaits results from Wyeth's ongoing Phase I clinical trials.

HERVing into view

Two novel presentations at CROI described interactions between HIV and human endogenous retroviruses (HERVs), which may have intriguing implications for vaccine research. HERVs are essentially fossil remnants of retroviruses that humans encountered many millennia ago. These retroviruses can no longer replicate but are incorporated into our chromosomes and now make up a surprising 8% of the human genome. The envelope protein of one HERV has been co-opted by humans and plays an indispensable role during pregnancy, illustrating the ancient and ongoing relationship between humans and retroviruses. Certain rare HERV sequences, for example HERV-K, exist as full length proviruses, but fragments known as human endogenous retrotransposable elements (HERE) are more common. Human cells have restriction factors like the recently discovered APOBEC proteins which suppress retroviral activity (see Guardian of the genome, IAVI Report 9, 2, 2005) and these factors are believed to be involved in keeping endogenous retroviral sequences dormant. Only a few instances of HERE transcription have so far been described in breast and testicular cancer tissue. However, HIV has mechanisms, such as the viral protein Vif, which disable host restriction factors. This led researcher Brad Jones from the University of Toronto to ask whether HIV infection of CD4+ T cells would awaken endogenous retroviral sequences and transcriptionally activate them.

When Jones looked at HIV-infected primary CD4+ T-cell lines in vitro, this is exactly what he found. These cells accumulated increasing numbers of genomic copies of the HEREs that Jones tested for (AluSX, LINE-1, and segments of HERV-K), but uninfected CD4+ T cells did not. To take a preliminary look at whether such events occur in vivo, Jones searched databases of HIV sequences and found a primary HIV isolate into which a HERE (LINE-1) had become inserted, strongly suggesting that HIV and LINE-1 can be active in the same cell.

Keith Garrison from UCSF, one of Jones's collaborators, also presented a poster at CROI (Abstract 457). Garrison studied 16 people with HIV and found they had higher levels of HERV-K genetic material in their blood compared to four uninfected controls. Garrison also looked for evidence of T-cell responses to HERV-K derived epitopes and found that they could be detected in individuals with primary HIV infection but not in uninfected controls. This finding held true even for HERV-K epitopes with little similarity- less than three amino acids in common-to HIV epitopes. There was even a weak but statistically significant inverse correlation between HERV-K-specific T-cell responses and HIV viral load.

The authors of these studies note that induction of CD8+ T cell responses to HERV epitopes may offer a novel way of selectively targeting HIV-infected cells, with the advantage that HERV epitopes cannot mutate to escape immune surveillance. Further work will be needed to demonstrate whether this interesting idea can be translated into a safe vaccine strategy.

Memory center

Louis Picker From the Vaccine & Gene Therapy Institute at Oregon & Health Science University delivered a plenary talk on pathogenesis that focused on lessons learned from the SIV/macaque model of HIV infection. Picker is focused on understanding the host factors that distinguish progressive from non-progressive disease using macaques infected with either the highly pathogenic clone SIVmac239 or its attenuated derivative SIVmac239Δnef. Picker stressed a basic concept that underpins his work: T cells (both CD4+ and CD8+) can be grossly subdivided into long-lived resting T cells (either inexperienced naïve T cells or antigen-experienced central memory T cells) and short-lived CCR5-expressing effector memory T cells. Resting T cells primarily circulate through the lymphoid system while effector memory T cells migrate to the tissues. Importantly, central memory T cells represent a reservoir of effector memory T cells because they generate these cells upon activation. Picker has previously shown that loss of effector memory CD4+ T cells from tissue (e.g. the bronchioalveolar lavage and lamina propria) is a critical determinant of disease progression in SIV-infected macaques (J. Exp. Med. 200, 1299, 2004). At CROI, he described his research team's efforts to understand how this loss occurs.

Picker homed in on SIV's effects on central memory CD4+ T cells as they represented the likely source of the vanishing effector memory cells. Around one-fifth of all the central memory CD4+ T cells detectable in lymph nodes were lost during acute SIV infection, and this was followed by a more gradual attrition and abrogation of proliferative capacity among the remaining central memory population. SIV DNA was always detectable in a small proportion of the central memory CD4+ T cell pool, even though effector memory CD4+ T cells were more extensively infected. In addition to direct infection, Picker also found that immune activation drove the accelerated conversion of central memory CD4+ T cells into effector memory cells and he suggested that this may lead to replicative senescence or other functional disturbances.

Picker concluded by proposing a "two tier" schema of HIV infection wherein there is an early assault on effector memory CD4+ T cells, such as those in the gut, that the immune system can withstand due to replenishment from the central memory CD4+ T cell pool. The second tier is the less efficient and more covert infection of central memory CD4+ T cells that occurs during chronic infection, ultimately leading to disease progression and AIDS. Picker's argument appears to dovetail neatly with many recent papers and a presentation at CROI by Mario Roederer from the VRC suggesting that vaccine-mediated protection of the central memory CD4+ T cell pool is a—perhaps the—crucially important correlate of protection against disease in SIV-infected macaques.

Memory cells defend themselves

A presentation that shed some light on factors protecting memory CD4+ T cells from HIV infection was given by Joseph Casazza from Rick Koup's group at the VRC. Casazza compared CD4+ T-cell responses specific for the pp65 protein from CMV, from both HIV-infected and uninfected individuals, to those specific for HIV's Gag protein. He found that Gag-specific CD4+ T-cell responses were less polyfunctional (and therefore unable to produce multiple cytokines and chemokines) than pp65-specific responses. In particular, far fewer Gag-specific CD4+ T cells produced the chemokine MIP-1b. Notably, however, the functional profile of the pp65-specific CD4+ T-cell responses from HIV-infected versus uninfected individuals was similar. To ascertain whether production of MIP-1b was exerting a protective effect, Casazza sorted pp65-specific CD4+ T cells from HIV-infected study participants based on their ability to make the chemokine and then assessed their infection history by looking for the presence of HIV Gag DNA. The results showed that MIP-1b-producing CD4+ T cells consistently had a lower content of Gag DNA than non-producing cells. Casazza suggested that this finding may account for the relative conservation of the CMV-specific CD4+ T cell response in people with HIV. He also noted that MIP-1b production may be a desirable property for vaccine-induced HIV-specific memory CD4+ T cells.

Measuring T-cell exhaustion

Early last year a study from Rafi Ahmed's group at Emory University identified the molecule PD-1 (programmed death 1) as a marker of CD8+ T cell exhaustion (Nature 439, 682, 2006). Since then a plethora of papers have shown that PD-1 is strongly upregulated on the CD4+ and CD8+ T cells of HIV-infected individuals, particularly on their HIV-specific T cells (Nature 443, 350, 2006: Nature Med. 12, 1198, 2006). These papers also reported that PD-1 expression correlates with viral load and inversely with CD4+ T cell counts. At CROI, Brent Palmer from the University of Colorado Health Sciences Center reported on the relationship between viral load and PD-1 expression on HIV-specific CD4+ T cells. Palmer's study cohort included 14 untreated individuals and 17 on ART with viral loads less than 20 RNA copies/ml of blood. PD-1 expression was significantly higher on HIV-specific CD4+ T cells compared to those specific for CMV and was also elevated in untreated individuals versus those on ART. Statistical analyses revealed a strong correlation between PD-1 expression on HIV-specific CD4+ T cells and viral load; however, there was no correlation between viral load and PD-1 expression measured on total CD4+ T cells. Looking at HIV-specific CD4+ T cells based on cytokine production, Palmer found that cells producing only interferon-g expressed the highest levels of PD-1 whereas cells producing IL-2 alone showed low levels of PD-1 expression.

In a subsequent talk on immunopathogenesis, Bruce Walker from Partners AIDS Research Center at Massachusetts General Hospital also touched on the role of PD-1 in HIV infection, adding the information that another molecule associated with T-cell dysfunction, CTLA-4, is also upregulated on HIV-specific CD4+ T cells in untreated HIV infection. Walker pointed out that engagement of PD-1 and/or CTLA-4 in vitro can lead to an apparent restoration of HIV-specific CD4+ T cell function, but it remains uncertain whether these mechanisms can safely be exploited for the purposes of immunotherapy. For the AIDS vaccine field, it remains to be seen whether vaccine-induced HIV-specific CD4+ T cell responses will be less prone to exhaustion than those that differentiate in infected people in the absence of prior vaccination.

*Richard Jefferys coordinates the Michael Palm Basic Science, Vaccines & Prevention Project at the Treatment Action Group, a New York-based community organization advocating for HIV research.