The Role of the SIV Model in Aids Vaccine Research

By Thomas C. Friedrich and David I. Watkins*

Despite over twenty years of research, there is still no effective vaccine against HIV.

Only one vaccine candidate has been through a Phase III (large-scale efficacy) trial: AIDSVAX (1, 2), a gp120-based candidate that was intended to induce antibody responses against the virus that would prevent individuals from becoming infected with HIV. Although this vaccine candidate elicited antibodies in most volunteers, it failed to prevent infection.


The failure of AIDSVAX was likely due to its inability to induce potent neutralizing antibodies (NAbs). A vaccine that elicits both a broadly-specific NAb response and a robust cellular response is the Holy Grail of AIDS vaccine research since it might impart sterilizing immunity, preventing the virus from binding to and entering target cells and thereby completely protecting vaccinated individuals. Unfortunately, very few antibodies have been identified that bind the viral envelope are capable of neutralization, so while the handful of characterized NAbs are the subject of intensive research, a vaccine based on NAbs remains a distant hope for the clinic.

As a result, researchers have focused on developing vaccines that elicit potent cellular immune responses. These vaccines are comprised of recombinant virus vectors to deliver HIV genes or gene segments. Unfortunately, recent large-scale trials of recombinant virus vector-based vaccines have also proven disappointing as the regimens tested have not induced potent T cell responses in many vaccinees. Recent evidence also indicates that HIV infections have occurred even in vaccinees who did make detectable responses (3, 4).

The disappointing results to date of antibody- and T cell-based vaccine strategies may seem to add up to a bleak picture for AIDS vaccines. But in fact we are still ignorant of the correlates of protection, so it remains unclear exactly what a successful vaccine needs to do. Why do a few rare individuals seem to make broadly neutralizing antibodies, while most antibodies do not neutralize? Do antibodies need to recognize intact envelope multimers? What titers of antibodies would be protective? Similarly, what defines effective T cell responses? Must they be of particularly high avidity? Should they traffic to the right anatomical compartments? Should they recognize a broad array of viral epitopes? Or epitopes that are under functional constraints, and thus poorly tolerant of mutation? The answers to these and other questions about effective immunity to immunodeficiency viruses can be informed by experiments in the macaque model.

Cellular immunity to HIV and SIV

As T cell biologists, we will be somewhat parochial here and focus on T-cell responses to HIV and SIV. In fact there are practical reasons for this T cell-centrism. The enormous difficulty in generating NAbs encourages us to focus on T-cell responses. Unlike NAbs, these responses are detected in most HIV-infected individuals. This may on first glance seem to argue against the wisdom of inducing T-cell responses through vaccination. After all, vigorous T-cell responses can be detected even in progressive infection, so what use could T cells be in controlling virus replication, let alone in preventing infection? We believe that these responses could provide the “raw material” of immune responses that could be augmented and focused by cleverly designed AIDS vaccines to ameliorate disease and help prevent HIV transmission. Several lines of evidence suggest that CD8+ T cell responses are important in controlling acute infection and maintaining the chronic phase viral set point. CD8+ T cells were first implicated in suppressing HIV replication in 1994 in two studies demonstrating that the reduction in viremia in acute infection was temporally associated with the appearance of HIV-specific CD8+ T cells (5, 6). A vigorous antibody response occurs subsequent to this initial CD8+ T cell response, after viremia has been controlled. Moreover, vigorous CD8+ T cell responses have been reported in some subjects with long term non-progressive infection (7, 8). Certain HLA alleles, such as B27 and B57, are also associated with long term non-progression (9-12), and some CD8+ T cell responses exert selective pressure on replicating viruses, resulting in the outgrowth of escape mutant viruses.

The relative contributions of cellular immune responses to control of virus replication and evolution have been difficult to discern from human studies alone. The existence of viral escape from CD8+ T cell responses remained controversial until evidence from the macaque model showed that these responses did indeed select for de novo escape variants (13). This is a particularly apt illustration of the utility of the animal model: Infection of macaques with a molecularly-cloned virus, SIVmac239, allows for the unambiguous identification of mutations that were not present during initial infection. Subsequently, cloned viruses derived from SIVmac239 have been used to show that escape from CD8+ T cell responses can exact a cost to viral fitness, an idea that may turn out to be important in the design of successful vaccines (14).

These studies used the animal model to complement investigations of HIV-infected humans but the macaque model also allows experiments that would be impossible in humans. For example, crucial evidence for the role of CD8+ cells in controlling immunodeficiency virus replication came from studies of transient depletion of CD8+ cells in SIV-infected macaques (15-17). In these studies, anti-CD8 mononclonal antibodies (MAbs) were used to transiently deplete circulating CD8+ cells, which resulted in an increase in viremia lasting until these cells repopulated the periphery. In some cases administration of these MAbs during acute infection seems to prevent control of virus replication, leading to rapid disease progression.

Antibodies and T cells

Although we have chosen to focus on cellular immune responses here we will admit to some degree of antibody envy. In many ways the antibody field is years ahead of the T cell field. A few broadly-neutralizing antibodies have been identified and investigators are now beginning to understand the mechanisms that confer their neutralizing activity. There is also a straightforward conceptual model guiding our thinking: NAbs must be broadly reactive and bind the viral envelope protein in such a way as to prevent infection of target cells. Although such antibodies seem to be extremely rare, at least we think we know what they should do (18).

In contrast we have not yet been able to define what an effective CD8+ T cell response might be. Many investigators have speculated that not all CD8+ T cells will be equally effective in eliminating virus-infected cells and it has recently been shown in an in vitro system that HIV-specific CD8+ T cell clones have differential abilities to select for escape variant viruses (19). There are several viral and host parameters that could affect CTL efficacy, including (i) avidity of the T cell receptor (TCR) for the MHC/peptide complex; (ii) CTL frequency and anatomical localization; (iii) the kinetics of expression of the viral protein from which the epitope is derived; (iv) fitness constraints on the epitope region; (v) immunodominance relationships among CTL specificities. We cannot yet evaluate the contributions of each of these factors to T cell efficacy but a rational approach to AIDS vaccine design will depend on our ability to understand them.

The role of the SIV model in vaccine research

In twenty years only one AIDS vaccine candidate has progressed all the way through clinical trials and it failed to protect people from infection. In the immediate aftermath of those results the feeling of disappointment in the field was palpable. But we believe that the AIDSVAX trial should be a call back to the drawing board. In order to find a safe and effective vaccine, first we need to know what we’re looking for. Sterilizing immunity is an ideal but, perhaps for now, unrealistic goal. We find reason for hope in studies of discordant couples, in which one partner is HIV-infected but the other is not. Longitudinal studies of such couples suggest that there is a threshold of viremia, about 1,700 vRNA copies/ml plasma, below which transmission of HIV is extremely unlikely (20). Furthermore, a study of over 550 mother-child pairs found no cases of mother-to-child transmission when the mother’s viral load was less than 1,000 vRNA copies/ml (21).

Figure 1. The goal of a CTL-based vaccine is to reduce chronic phase transmission of HIV.Recent evidence suggests that transmission of HIV is unlikely when the infected individual’s virus load is less than 1,700 vRNA copies/ml plasma. This represents a reduction of about 1.5 logs below the median virus load of 30,000 vRNA copies/ml in untreated subjects. We therefore suggest that the goal of a CTL-based vaccine candidate in macaques should be to reduce virus load by 1.5 logs.

 [View Larger Image].

We therefore propose that the immediate goal of an effective AIDS vaccine should be to reduce transmission of the virus (Fig. 1). This may be a more realistic goal for a CTL-based vaccine, since cellular responses would need to allow infection of at least some host cells, and could not provide absolutely sterilizing immunity. Such a vaccine would be a truly positive development for public health, since a reduction in transmission rates would roll back the epidemic. Furthermore, lower chronic phase virus loads are associated with slower progression to AIDS, so a vaccine that reduces transmission should also prolong survival in vaccinated individuals who become infected. A CTL-based AIDS vaccine should therefore reduce chronic phase viremia from its present level of ~30,000 vRNA copies/ml in untreated subjects to ≤1,700 vRNA copies/ml, about a 1.5-log reduction. When designing and evaluating experiments in the macaque model, we must keep this goal in mind. To be labelled promising, then, an experimental vaccine should reduce chronic phase replication of a pathogenic SIV by at least 1.5 logs below controls. Given this goal, how can we evaluate vaccine candidates in the monkey model?

1. Define the parameters of effective CD8+ T cells. The characteristics of “effective” CD8+ T cell responses remain a mystery. Since the cellular field lags behind the antibody field in defining the characteristics of “good” T cell responses, the SIV model will be instrumental in testing hypotheses about what defines a desirable T cell response. Let us consider two examples of the ways in which such hypotheses could be addressed in the animal model. First, CD8+ T cell-based vaccines must deal effectively with the ability of the virus to escape. As we mentioned above, it may even be possible to use the plasticity of the viral genome to our advantage by targeting epitopes that are under functional constraints and poorly tolerant of variation. These epitopes should be less likely to accumulate escape mutations in individual patients or in the population. If escape variants are selected, they may reduce viral fitness, resulting in a mutant virus that is easier to control. Indeed it seems likely that vaccine-induced CD8+ T cells specific for an epitope in SIV Gag have done just that in a recent trial (22). In that study, 5 of 8 vaccinated Burmese rhesus macaques rapidly controlled an intravenous SIV challenge. Control in each case was associated with sequence variation in gag. One mutation appeared in each of three animals that shared an MHC haplotype, and appeared to have a detrimental effect on viral fitness. Unfortunately, though, CD8+ T cells that target constrained epitopes may not be immunodominant in natural infection. Indeed the eventual progression of most infected individuals to AIDS indicates that the natural hierarchy of T-cell responses, that is, their relative frequencies in natural infection, is usually not sufficient to control virus replication. A vaccine, therefore, should not simply aim to induce as many CD8+ lymphocyte responses as possible. In many cases, it will be necessary to alter the natural immunodominance of the HIV- or SIV-specific CD8+ T cell response since immunodominant responses may limit the development of potentially more effective sub-dominant responses (23, 24). For a vaccine, it may be important to force escape at a price to the virus.

2. The challenge SIV strain must be appropriate to the experiment. The identification of escape mutations associated with reduced fitness was facilitated by the use of a cloned challenge virus, SIVmac239. This illustrates another important consideration in the monkey model: which challenge virus is the most appropriate? Few CD8+ T cell-based vaccine regimens have significantly lowered viral load or affected disease course in macaques using the most stringent SIV challenge models available. Recently several vaccines have claimed success in control of the chimeric virus SHIV89.6P (25-28). However, these results have the potential to be misleading for two reasons. First, all but one of these recent vaccine regimens includes a similar Env in the vaccine. The only regimen that did not use Env, a DNA prime/recombinant adenovirus (rAd) boost encoding only Gag, was ineffective against a SIVmac239 challenge despite its success against SHIV89.6P (Casimiro et al., J. Virol. in press). Second, SHIV viruses were developed to test vaccines designed to elicit Env-specific antibodies in the macaque model. While they have had utility in this regard, the biology of these chimeric viruses has raised some doubts about their relevance for use in evaluation of CTL-based vaccines (29). SHIV89.6P, for example, causes a dramatic, acute loss of virtually all circulating CD4+ T cells, which is thought to be a main reason for its pathogenicity. Such a dramatic loss of CD4+ T cells form the periphery is rare in HIV infection. In contrast, pathogenic SIV strains such as the molecular clone SIVmac239, or the biological isolates SIVmac251 or SIVsmE660, replicate to high titer in the chronic phase but do not cause as dramatic a depletion of CD4+ T cells (30, 31). Uncloned SIV isolates that are difficult to neutralize may be the most relevant challenge stocks since they mimic the diversity of quasispecies that are likely to be present in transmission of HIV.

3. Repeated low-dose challenge may help the SIV model more closely mimic natural HIV infection. It is possible that high-dose SIV challenge is too stringent a test for our vaccine regimens. In most SIV experiments animals are challenged with a single very high dose of virus, equivalent to thousands of infectious particles. This approach has a sound rationale: in order to detect a vaccine effect when animal numbers are low, it is essential that all animals in the control group become infected. But it is possible that an extremely high dose of infectious virus will overwhelm any vaccine-induced immune responses, causing us to reject what may actually be promising approaches. Several laboratories have recently begun developing lower-dose challenge models in which animals are exposed to 10- or 100-fold less infectious virus than in conventional high-dose experiments (32). Challenges must be repeated multiple times in order for animals to become infected, complicating experiment schedules. Low-dose challenge approaches must therefore be optimized to deliver as small a dose of virus as possible while maintaining experimental tractability.

Future directions in vaccine research

Reviewing the recent state of AIDS vaccine research could understandably be a depressing exercise. Like the stage at the end of a Shakespearean tragedy, the field is littered with approaches that, noble attempts though they were, were tragically brought low. We believe, however, that recent results from the SIV model may offer a glimmer of hope, but we need to view these results as critically and honestly as possible to avoid future disappointment and the prolongation of an epidemic whose human toll is already disastrous. Antibodies and CTL alike need to deal with the prodigious ability of the virus to spawn variants that evade immune responses. But there may be chinks in this armor of variability. Constrained epitopes may persist in the virus population and represent attractive targets for CD8+ T cell responses. New generations of vaccine vectors may be able both to induce more potent mucosal responses and to refocus CTL on epitopes in which variation results in a loss of viral fitness. Indeed, we may already be on the trail of successful vaccine approaches whose tracks are currently obscured by too-high doses of challenge virus.

We hope that we are experiencing a new era of innovation in AIDS vaccine research in which many exciting new ideas are being put forward from many investigators. Not all of these ideas can be tested in human trials. The SIV model can provide some road signs to the clinic by allowing us to evaluate hypotheses in ways that are impossible in human subjects. Immunogenicity is not enough. We must challenge our new ideas, figuratively by refining our hypotheses and experimental approaches, and literally by infecting vaccinated animals with appropriate pathogenic viruses.


1. T. Ready, Nat Med. 9, 376 (2003).

2. WHO. HIV Vaccines, http://www.who.int/hiv/topics/vaccines/Vaccines/en/ (2005).

3. D. Lee, B. S. Graham, et al. J. Infect. Dis. 190, 903 (2004).

4. M. R. Betts, B. Exley, et al. Proc. Natl. Acad. Sci. USA 102, 4512 (2005).

5. P. Borrow, H. Lewicki, et al. J. Virol. 68, 6103 (1994).

6. R. A. Koup, J. T. Safrit, et al. J. Virol. 68, 4650 (1994).

7. T. Harrer, E. Harrer, et al. AIDS Res. Hum. Retroviruses 12, 585 (1996).

8. T. Harrer, E. Harrer, et al. J. Immunol. 156, 2616 (1996).

9. S. J. O'Brien, X. Gao, and M. Carrington, Trends Mol. Med. 7, 379 (2001).

10. A. D. Kelleher, C. Long, et al. J. Exp. Med. 193, 375 (2001).

11. M. Altfeld, M. M. Addo, et al. AIDS 17, 2581 (2003).

12. S. A. Migueles, M. S. Sabbaghian, et al. Proc. Natl. Acad. Sci. USA 97, 2709 (2000).

13. D. T. Evans, D. H. O'Connor, et al. Nat. Med. 5, 1270 (1999).

14. T. C. Friedrich, E. J. Dodds, et al. Nat. Med. 10, 275 (2004).

15. X. Jin, D. E. Bauer, et al. J. Exp. Med. 189, 991 (1999).

16. T. Matano, R. Shibata, et al. J. Virol. 72, 164 (1998).

17. J. E. Schmitz, M. J. Kuroda, et al. Science 283, 857 (1999).

18. D. R. Burton, R. C. Desrosiers, et al. Nat. Immunol. 5, 233 (2004).

19. O. O. Yang, P. T. Sarkis, et al. J. Exp. Med. 197, 1365 (2003).

20. R. H. Gray, M. J. Wawer, et al. Lancet 357, 1149 (2001).

21. P. M. Garcia, L. A. Kalish, et al. N. Engl. J. Med. 341, 394 (1999).

22. T. Matano, M. Kobayashi, et al. J. Exp. Med. 199, 1709 (2004).

23. W. Chen, L. C. Anton, et al. Immunity 12, 83 (2000).

24. M. J. Palmowski, E. M. Choi, et al. J. Immunol. 168, 4391 (2002).

25. J. W. Shiver, T. M. Fu, et al. Nature 415, 331 (2002).

26. N. F. Rose, P. A. Marx, et al. Cell 106, 539 (2001).

27. D. H. Barouch, S. Santra, et al. Science 290, 486 (2000).

28. R. R. Amara, , F. Villinger, et al. Science 292, 69 (2001).

29. M. B. Feinberg and J. P. Moore, Nat. Med. 8, 207 (2002).

30. Y. Nishimura, T. Igarashi, et al. Proc. Natl. Acad. Sci. USA 101, 12324 (2004).

31. Y. Nishimura, C. R. Brown, et al. Proc. Natl. Acad. Sci. USA 102, 8000 (2005).

32. A. B. McDermott, J. Mitchen, et al. J. Virol. 78, 3140 (2004).