Perspective: HIV Controllers: Can the Human Genome Project Advance AIDS Vaccine Development?
By Toshiyuki Miura, Kholiswa Ngumbela, Danni Ramduth, Florencia Pereyra, and Bruce Walker*
At the pivotal United Nations General Assembly Special Session on HIV (UNGASS) meeting in New York in June 2001, there was an urgent call to action regarding the expanding HIV pandemic, including increased treatment access for HIV-infected persons worldwide. Since then over two million individuals in low- and middle-income countries have gained access to antiretroviral (ARV) therapy through expanded global treatment programs, such as the Global Fund to Fight AIDS, Tuberculosis and Malaria, the US President's Emergency Plan for AIDS Relief, as well as increasing national treatment programs.
On an individual level, treatment results have been spectacular, showing that ARV therapy can be effective even in the most resource-constrained environments. Even so, the overall effort to contain the epidemic is actually losing ground. In some areas more persons are in need of therapy now than before these programs were started and issues about the long-term sustainability of these programs in the face of daunting medical needs remain to be addressed. Prevention efforts, including behavioral and vaccine efforts, have not kept abreast with need and the pandemic continues to expand in developed and developing countries.
Clearly the only global solution to the AIDS epidemic is an effective vaccine, yet six years after the UNGASS meeting and almost a quarter-century since the viral etiology of AIDS was defined, an effective preventive vaccine still eludes our grasp. Indeed, in the interim, additional insights into the challenges posed in generating broadly-neutralizing antibodies have resulted in the rather stark realization that a fully preventive vaccine—one that prevents infection from occurring through sterilizing immunity—is not a realistic immediate goal. Instead vaccine efforts are currently focused on a hypothesis that has shown promise in animal models—namely a vaccine that will keep the virus in check to such a degree that disease progression will not occur if a vaccinated individual does become infected, and that maintains viral load at such low levels that the likelihood of transmission will be markedly reduced.
Cohort studies have shown that viral load is directly linked to both disease progression and transmission and the likelihood of either of these outcomes markedly diminishes at viral loads below 2000 HIV RNA copies/ml plasma(1-5). Although on an individual level persons with viral loads this low can still transmit virus and/or undergo disease progression, on a population level these events would diminish to the point that the epidemic would be considerably curtailed if this level of control could be achieved with a vaccine. There is reason to be optimistic, given that infected individuals who represent precisely this desired phenotype have been identified. These persons, who have been termed HIV controllers, are able to maintain suppression of viremia and stable CD4+ T-cell counts without ARV medication(6-8), some now for close to 30 years (S. Deeks and B. Walker, manuscript submitted). The most dramatic examples are those who maintain viral loads below the level of detection by the most sensitive commercial assays presently in use.
For the purposes of this article we will focus on two groups of HIV controllers: elite controllers, who maintain viral loads of <50 HIV RNA copies/ml plasma in the absence of therapy, and viremic controllers, who maintain viral loads between 50 and 2000 RNA copies/ml. This article seeks to draw attention to recent international efforts to focus research efforts on these individuals, and to use advances from the Human Genome Project to dissect the mechanisms behind these remarkable outcomes, and thereby guide efforts to recreate this phenotype through an AIDS vaccine.
Long-term nonprogressors and HIV controllers
Some of the earliest data showing dramatic differences in long-term outcome of HIV infection came from the San Francisco City Clinic, where blood samples stored during a hepatitis B vaccine study conducted in the late 1970s enabled precise determination of the timing of HIV infection as the epidemic spread through that population(9). As the extent of the HIV epidemic became clearer in the 1990s, there was a remarkable realization that some individuals in this cohort were remaining healthy (as defined by stable CD4+ T-cell counts) despite never being treated with the ARVs that were becoming increasingly available. These persons were detected in this and other cohort studies with sufficient frequency that the US National Institutes of Health sponsored a meeting on what were to be termed long-term nonprogressors (LTNPs) in the early 1990s. A subset of these LTNPs with even more remarkable characteristics was then defined as sensitive viral load testing became available: those who had plasma viral loads below the limits of detection by the most sensitive assays (initially <400 RNA copies/ml and later <50 RNA copies/ml)(7,10). Here was a group of persons who appeared to be able to contain HIV as is typically seen for many other chronic virus infections—such as Epstein-Barr virus and varicella zoster virus—that are effectively held in check despite the continued presence of infectious virus.
From an AIDS vaccine standpoint, understanding this subset of infected individuals, defined by the ability to control viral load rather than by CD4+ T-cell count or duration of infection, may be most relevant in terms of the rationale for a vaccine to prevent disease progression. Cohort studies suggest that elite controllers occur at a frequency of about one in 300 HIV-infected persons(11,12). A larger subgroup of HIV-infected persons called viremic controllers consists of those who maintain viral loads between 50 and 2000 RNA copies/ml plasma without treatment. At a viral load of less than 2000 copies, two important needs for a vaccine would likely be met: there would be less likelihood of progression in persons infected(2), and, since these viral loads are associated with delayed progression and reduced transmissibility(3,5), such individuals are also likely to provide key insights into the factors involved in successful viral containment.
Reasons for HIV control
Understanding the ability to maintain viral loads at levels that would diminish the likelihood of both disease progression and transmission is key to current vaccine development efforts. Multiple factors have been implicated in contributing to this outcome, including viral, host genetic, and immunologic factors, but the bottom line is that we still lack a fundamental understanding of the pathways leading to this remarkable and particularly relevant outcome of natural infection.
Viral factors: At the time LTNPs were first being defined, the identification of a group of transfusion recipients in Australia who had all received blood products from the same donor and all had slowly progressing diseases suggested that some persons were simply infected with less pathogenic viruses(13). These persons ultimately progressed, and though additional studies have shown that gross defects in HIV genes can attenuate viral pathogenicity, such deletions are not a common feature of improved outcome(14,15). Some studies demonstrated that virus isolated from LTNPs are attenuated(11,16,17), but replication competent viruses have clearly been isolated from persons maintaining viral loads below 50 RNA copies/ml1(5,18,19). Even minor differences in viral fitness may have a long-term impact over the typical 10-year course from HIV infection to AIDS, and enhanced methods for discriminating fitness differences are needed to address this issue(20). Insights into viral factors contributing to this phenotype are also likely to come from detailed characterization of viruses from persons who establish elite control and then subsequently lose this control.
Host genetics: Although large host genetic studies focused solely on elite controllers are yet to be done, associations between host genetic variation and susceptibility to HIV infection and disease progression have been studied extensively across the entire spectrum of viral loads. Some examples are chemokine co-receptor polymorphisms such as CCR5Δ32 and CCR2V64I, CCR5 promoter polymorphisms, RANTES-28G, -403A, SDF-1 3'A, IL-4 589T, and others(21-26). Variation in host factors that restrict viral replication such as APOBEC3G and Trim5a may also influence disease, but are yet to be studied in detail in HIV controllers(27,28). Except for resistance to HIV-1 infection of persons homozygous for CCR5Δ32, it is unclear how many of these factors would remain statistically significant after multivariate analysis taking into account all reported gene variation. A recent study of over 4000 HIV-infected and uninfected individuals demonstrated that combination of CCL3L1 gene (that encodes MIP-1a) copy number and CCR5 haplotype is associated with HIV/AIDS susceptibility and disease progression(29). As more host factors are discovered and included in these studies, greater numbers of patients will be required to perform multivariate analyses and to define true associations. Without question the strongest genetic association thus far with elite control is the expression of certain HLA class I alleles(8,30,31). Since these surface molecules present viral peptides to the immune system for recognition, they suggest a link between host genetics and the adaptive arm of the immune responses, particularly suggesting a key role of virus-specific CD8+ T cells.
Immunologic: HIV controllers are particularly relevant to AIDS vaccine design if it actually is the immune system in these persons that controls HIV replication. Although by no means conclusive, current data suggest that this is indeed the case. Perhaps the strongest indicator that the immune system influences outcome is the high frequency of certain HLA class I alleles among elite controllers, such as HLA B27 and HLA B57(8,30). These cell-surface molecules complexed with a viral protein indicate to the immune system that the cell is infected and should be eliminated, so the enrichment for certain HLA alleles implicates the host immune response rather than the virus itself. But the issues are far from resolved—quantitatively, using the interferon (IFN)g ELISPOT assay to detect responses, there is no difference in magnitude or breadth of HIV-specific CD8+ T-cell responses and viral load(32,33). Instead it may be functional qualities of these cells, such as the ability to proliferate or to avoid exhaustion, or to directly inhibit virus replication in vitro, that play a central role in outcome(34-36). Whether these are simply associations or causative will remain very difficult to determine. Recent data indicate that even in elite controllers there is evidence of ongoing immune selection pressure on the virus, as indicated by mutations arising within targeted epitopes, suggesting ongoing functional significance of these responses(37). Moreover, depletion of CD8+ T cells in an animal model of elite control led to increased viremia, again indicating active containment of the virus by these cells(38).
Another strong association between immunologic function and disease outcome derives from studies of CD4+ T-cell function, particularly the ability to secrete multiple cytokines, including TNFa, IFNg, and IL-239. In the largest study to date involving 30 HIV controllers, polyfunctional HIV-specific CD4+ T cells expressing IFNg and IL-2 were the single most consistent correlate of control, but there was still a substantial proportion of subjects who did not demonstrate such responses(39). Another critical host defense that may impact adaptive immune responses in general is innate immunity. Recent data link particular natural killer cell receptors to enhanced control of viremia, supporting a potential role for innate immunity in chronic infection(40). There are almost no data describing the earliest innate immune events following an acute infection that lead to elite control.
In terms of neutralizing antibodies (NAbs), studies of elite controllers indicate generally low-level responses(37). In a comprehensive examination of NAb responses in persons with undetectable viremia (from whom pseudotyped viruses were constructed from autologous Env) there was a lack of strong responses to autologous virus(37), suggesting that there is little ongoing exposure to virus and that NAbs are unlikely to play a major role in durable containment of viremia in these persons.
Unbiased approach: whole genome association scan
It is fair to say that there have been insufficient numbers of elite controllers studied to date to draw firm conclusions, but the data available indicate that none of the current parameters we use to assess virus-host interactions provides high predictability for the observed elite control phenotype. In part we may be constrained by working within existing scientific paradigms of how the host responds to viruses. Now major progress in human genetics allows for a different and largely unbiased approach to defining the molecular basis for disease outcome—high throughput sequencing of the human genome to find the genotype that predicts a particular disease phenotype(41). Until recently such studies could not be contemplated since there was no practical way to sequence the entire three billion nucleotides in the human genome. However, the combination of practical shortcuts as well as reduced sequencing costs now make this a realistic approach and something we believe absolutely needs to be applied to HIV/AIDS.
The goal of whole genome association scans (WGAS) is to define the genetic variability within the host genome that is directly linked to a particular phenotype such as elite control. Fortunately there is limited variability within the human genome, which usually involves single nucleotide polymorphisms (SNPs) that occur with a frequency of about 1 in every 300 base pairs. If one only focuses on the regions of variability, this greatly reduces the amount of sequencing required, and is further diminished because of linkage disequilibrium Figure 1. Closely adjacent SNPs are often transmitted together, so that the presence of one SNP is highly predictive of the presence of a second specific SNP, resulting in what is called a particular haplotype. By selecting one SNP, called a tagSNP, one can often obtain highly reliable information on the surrounding SNPs, greatly reducing the amount of sequencing required. Currently available commercial products allow for the rapid automated sequencing of 650,000 or more SNPs per patient, meaning that large population studies are now possible. With as few as 100 subjects this kind of approach has defined the genetic basis for diseases such as age-related macular degeneration. A recent review summarizes the approach and the power of this type of assay(42).
Figure 1. Whole Genome Association Scan.
Recently a global research consortium, The International HIV Controller Consortium, has been established with the goal of recruiting at least 1000 elite controllers and 1000 viremic controllers, as well as up to 3000 progressor controls, in order to perform a WGAS to define the genetic profile that accounts for elite control. These numbers are the minimum likely to be required to provide appropriate power for such analyses. More is always better given the multiple comparisons being made, but an approximation for this sort of approach is 1000-2000 or more persons with a particular disease phenotype. If the underlying etiology is a single nucleotide change, which for example is the cause of sickle cell disease, then this approach would generate statistical significance with very few subjects compared to controls. For more complex traits or disease states that have multiple different etiologies, the associations can be much more difficult to define in the sea of background noise. There are other potential limitations to this approach, including the possibility that elite control may be defined by some regulatory element which can be more difficult to detect, or by gene duplications, which are also harder to quantitate. Through this international consortium a total of over 600 HIV controllers have already been recruited, of whom 50% are elite and 50% viremic, and sequencing has commenced on these initial subjects. Whether this approach will yield new insights should be known soon.
Conclusion
Despite extensive research for over two decades we are still a long way from development of a preventive AIDS vaccine. In contrast, a cellular immunity-based vaccine which controls HIV-1 replication and significantly delays disease progression seems to be a realistic goal for the near future. Extremely skewed distribution of HLA class I alleles amongst elite controllers and increase in viremia after CD8+ T-cell depletion in SIV-infected elite controller monkeys strongly suggest that cellular immunity is playing a major role in containing viral replication in elite controllers. It would be essential for cytotoxic T-lymphocyte vaccine development to dissect the mechanisms behind this strict viremia control. Although conventional approaches have been unsuccessful so far, we are hopeful that novel approaches using a whole genome association scan will provide new key insights to define the correlates of immune protection and guide effective vaccine development. The probability of success of this approach increases with increasing numbers of subjects studied, and readers are encouraged to join in this broad collaborative global effort to recruit the necessary numbers of HIV controllers to allow for the mechanisms of control to be dissected at the HIV Elite Controller Study, with the goal of using the information gained to guide both prophylactic and therapeutic vaccine development.
------------
*Toshiyuki Miura is a research fellow at the Partners AIDS Research Center; Kholiswa Ngumbela and Dhanwanthie Ramduth are members of the HIV Pathogenesis Programme at the Nelson Mandela School of Medicine in the University of KwaZulu Natal; Florencia Pereyra is an Instructor in Medicine at the Partners AIDS Research Center; and Bruce D. Walker is a Professor at Harvard Medical School, a Howard Hughes Medical Institute Investigator, and director of the Partners AIDS Research Center at Massachusetts General Hospital.
References
1. J.W. Mellors et al., Science 272, 1167 (1996)
2. J.V. Giorgi et al., J Acquir. Immune Defic. Syndr. 29, 346 (2002)
3. R.H. Gray et al., Lancet 357, 1149 (2001)
4. D. Finzi et al., Nat. Med. 5, 512 (1999)
5. P.M. Garcia et al., N. Engl. J. Med. 341, 394 (1999)
6. E. Harrer et al., AIDS Res. Hum. Retroviruses 10 Suppl 2, S77 (1994)
7. E.S. Rosenberg et al., Science 278, 1447 (1997)
8. S.A. Migueles et al., Proc. Natl. Acad. Sci. USA 97, 2709 (2000)
9. A.R. Lifson et al., J. Infect. Dis. 163, 959 (1991)
10. T. Harrer et al., J. Immunol. 156, 2616 (1996)
11. J.B. Hubert et al., Aids 14, 123 (2000)
12. O. Lambotte et al., Clin. Infect. Dis. 41, 1053 (2005)
13. N.J. Deacon et al., Science 270, 988 (1995)
14. T.C. Greenough et al., J. Infect. Dis. 180, 1790 (1999)
15. L. Alexander et al., J. Virol. 74, 4361 (2000)
16. H. Blaak et al., J. Infect. Dis. 177, 600 (1998)
17. M.E. Quinones-Mateu et al., J. Virol. 74, 9222 (2000)
18. J.N. Blankson et al., J. Virol. 81, 2508 (2007)
19. J.R. Bailey et al., J. Virol. 80, 6441 (2006)
20. E. Martinez-Hackert et al., J. Biol. Chem. 281, 20205 (2006)
21. M.P. Martin et al., Science 282, 1907 (1998)
22. M.W. Smith et al., Science 277, 959 (1997)
23. H. Liu et al., Proc. Natl. Acad. Sci. USA 96, 4581 (1999)
24. C.M. Hogan, S.M. Hammer, Ann. Intern. Med. 134, 978 (2001)
25. N. Wichukchinda et al., Aids 20, 189 (2006)
26. A. Kawana-Tachikawa et al., J. Virol. 76, 11982 (2002)
27. E.C. Speelmon et al., J. Virol. 80, 2463 (2006)
28. C. Hess et al., Lancet 363, 863 (2004)
29. P.C. Chien, Jr. et al., J. Infect. Dis. 189, 852 (2004)
30. J.H. Bream et al., Science 284, 223 (1999)
31. R.A. Kaslow et al., Nat. Med. 2, 405 (1996)
32. M.M. Addo et al., J. Virol. 77, 2081 (2003)
33. M.R. Betts et al., J. Virol. 75, 11983 (2001)
34. C.L. Day et al., J. Virol. 81, 434 (2007)
35. C.L. Day et al., Nature 443, 350 (2006)
36. A. Saez-Cirion et al., Retrovirology 3, 81 (2006)
37. D.E. Kaufmann et al., J. Virol. 78, 4463 (2004)
38. T.C. Friedrich et al., J. Virol. 81, 3465 (2007)
39. B. Emu et al., J. Virol. 79, 14169 (2005)
40. Martin, M.P. et al., Nat. Genet. 39, 733-740 (2007)
41. International HapMap Consortium, Nature 437, 1299 (2005)
42. K. Christensen, J.C. Murray, N. Engl. J. Med. 356, 1094 (2007)