|Home | About | Journals | Submit | Contact Us | Français|
Animal models for HIV research have been indispensible in fulfilling Koch’s postulate and in exploring issues of viral infectivity and pathogenesis, sequence divergence, route(s) of acquisition, tissue distribution and tropism, immunogenicity and protection capacity of vaccine candidates, escape from adaptive immunity, and more. Did they fail to predict the efficacy of T cell vaccines in humans? This article summarizes progress and status of models to inform and complement clinical work.
The past two years of HIV research have been sobering years, during which a T cell based vaccine was tested in the clinic, and the trial was halted when it failed to achieve its objectives of limiting or controlling infection. More disturbingly, some investigators have suggested that the vaccines may have enhanced HIV acquisition in a subset of subjects. How could this have happened, given the many years of research to understand animal models, at least one of which predicted potential success of T cell based vaccines (1) (2)? And where do we stand, given the failures of vaccines designed to elicit potent neutralizing antibodies? This article addresses key aspects of the animal models and the outcomes that they predicted. In my view, lessons learned from the trial, combined with the preclinical data provided by the models, place us in much better position to predict future vaccine success than before the STEP trial. Some of the questions that are most hotly debated today are:
The lack of productive replication of HIV-1 in all animals, with the exception of the chimpanzee, led to a search for a mouse lentivirus that was never discovered. Rabbits and mice (non-transgenic) are used for immunogenicity studies, and currently the rabbit is a favored host for testing constructs aimed at eliciting neutralizing antibodies. There is no challenge model for the rabbit, so this species is used to compare vaccines without the option of measuring protective efficacy. Other species such as guinea pigs or goats are used by some investigators. For infection models, mice or NHP have been productively explored. In general, rodents transgenically altered (e.g. non-obese diabetic/SCID mice) and infused with human components (e.g. autologous human fetal thymic and liver CD34+ stem cells) results in the establishment and maintenance of human T and B cells, monocytes, and macrophages. They are thus susceptible to infection and can serve to test concepts (3), although the models developed to date have limited duration or longevity and retain the murine MHC antigens, thus potentially complicating immune response interpretations. Even without a murine pathogenesis model, this research has provided key insights into understanding blocks to replication and productive infection. These models have also provided protection data for human monoclonal antibodies using HIV-1.
NHP models have been developed and refined since the discovery of an AIDS-like illness and similar pathology in captive Asian macaques in 1985 (4). Cloning of the virus, simian immunodeficiency virus (SIV) and subsequent isolation of many related SIV from African monkeys has been instrumental in understanding the origins and evolution of the primate lentiviruses. The lack of pathogenic consequences despite viremia in their “natural” African hosts may provides clues for eventual control, despite infection (5). At the same time, the fulminant infection in Asian macaques has led to better understanding of important disease sequelae (6). Some examples of lessons learned from SIV research include demonstrations of escape from T cell epitopes and neutralizing antibodies, as well as direct destruction of germinal centers and gut-associated lymphocytes that predicted such findings in humans (7). Chimpanzees support productive infection, but disease does not occur for at least 10 years. Baboons can support replication of strains of HIV-2, but not HIV-1.
In addition to the obvious host differences between humans and NHP, the biggest issue is that SIV is not HIV-1, rather a relative of HIV-2. While these two viruses utilize the same cellular receptors (CD4) and many of the same coreceptors (principally CCR5), they are genetically and antigenically distinct. Isolates of HIV-1 that utilize CXCR4 as a coreceptor can arise later in infection in some subjects, but it appears that most transmitted viruses uniformly share CCR5. Antibodies that neutralize HIV do not neutralize SIV, and vice versa, due to differences in the Envelope protein. Furthermore, T cell epitopes are not shared between the viruses. Vaccine experiments using SIV virus challenge must utilize SIV-based immunogens, preventing the direct testing and validation of specific HIV vaccine antigens for protection prior to their testing in the clinic. Therefore the field has turned to the use of genetic chimeras termed “SHIV” that utilize the backbone of one virus (typically SIV) and have one or more genes swapped with HIV to gain the function needed. These SHIV are typically not pathogenic unless passaged in vivo in NHP to increase fitness and viral persistence. Enzymes (e.g. protease) differ as well and require the use of SHIV with the appropriate swapped gene to perform drug testing in vivo in macaques. The use of Envelope replacement SHIV has the target of neutralizing antibodies - the HIV Envelope - in the challenge virus, thereby allowing testing of concepts designed to stimulate neutralizing Ab. Finally there is the issue of the well-known worldwide genetic diversity of HIV-1. Few groups have ventured beyond testing vaccines exactly matched to the challenge virus, a hurdle that must be overcome for any vaccine to be successful.
The Indian-origin rhesus macaque (Macaca mulatta) has become the favored species for AIDS vaccine and pathogenesis research, but it is not the only species in use. Cynomolgous (Macaca fascicularis) and pigtailed macaques (Macaca nemestrina) are also utilized for pathogenesis, vaccine and microbicide research (8) (9). Differences in SIV and SHIV replication in these species have been observed and has led to preferential use, depending upon the question being asked (10). Vaccine testing in the different species of macaques with similar vaccine modalities provides an opportunity to compare and to rank outcomes and to gain confidence that these outcomes may be more broadly reproducible (11). Due to a high and increasing demand for the Indian-origin rhesus macaque, it is also a practical matter to maintain other species and to continue to utilize them, so that the most highly demanded species can be reserved and used judiciously.
As with financial investments, diversification in challenge models is important for HIV research. Reliance upon a single clone of SIV - SIVmac239- and a single clone of SHIV- SHIV89.6P - as challenge viruses was championed as an ideal strategy for comparing vaccines in the Indian-origin rhesus macaque. This strategy resulted in opportunities for direct and useful comparisons of different vaccine modalities, with a focus on vaccines identically matched genetically to the challenge virus. In hindsight, interpretation of these “matched” challenge experiments may have been optimistic and potentially risky in the absence of a validated model (12). Guided and reassured by the reproducibility of the high virus loads and rapid CD4 loss of SHIV-89.6P and several other CXCR4-utilizing SHIV, vaccines that could reduce or prevent this loss were touted as “protective.” Work by Martin and colleagues (13) showed that major differences in the pathogenesis of X4 SHIV versus R5 SIV were likely to be very important in choosing models that more closely resemble the pathogenesis of HIV in human. In addition to this point, the issue of matching the challenge virus to the vaccine looms large over us. Challenges with heterologous viruses from within the subtype are rare, and the ability to protect from viruses in another clade remains a future goal. SHIV constructed with Envelopes that utilize CCR5 for entry from both B and non-B subtypes are in development and show promise (14, 15). These will allow more realistic challenges using viruses that are not exact matches to the vaccine. Similarly, more SIV isolates are needed to use as ‘non-self’ or heterologous challenge virus, and this work is proceeding. The need for additional SIV and SHIV challenge stocks and the shared resources to make these widely available are two requirements that will remain for the foreseeable future.
The failure of the STEP trial—a ‘proof-of-concept trial designed to test replication-defective adenovirus to stimulate T cells—was met with disappointment and initial resignation. This trial took forward the non-replicating adenovirus 5 (Ad5) vectors expressing HIV Gag, Pol, and Nef. As discussed in several excellent articles, most likely we were overly optimistic about the effectiveness of vaccines that “protected”—i.e. brought virus loads down several orders of magnitude-- in the SHIV-89.6P model, when the SIV model provided a more sobering view with minimal to no reduction in virus loads in adenovirus-vaccinated macaques (16). Yet, how could we have known in advance which of these two alternative models was more predictive? The answer is that we could not have known, because animal models are not validated until there is success in the clinic to tie down specific correlates of protection. What we have done is to examine the outcome and say with some certainty that protection from SHIV-89.6P (and possibly any CXCR4-utilizing virus) is likely to give misleading results for T cell vaccines. We still do not know whether the SIV model is valid, but it is a more stringent model and has gained in favor in the last year for testing vaccines designed to elicit T cells (12, 17, 18). SHIV models that utilize CCR5 to gain entry are gaining acceptance for testing Envelope-based vaccines and neutralizing monoclonal antibodies.
Where do we go from here? We certainly need to understand the role of anti-vector (adenovirus) immunity, as is being done in the samples from the STEP trial; another is to test ideas directly using one of the NHP models. In both cases the ultimate success of modeling disease susceptibility and in understanding correlates will come from much more sophisticated understanding of the host genetics as it relates to innate and adaptive responses, as eloquently noted by Moore et al. (19). The way forward will by necessity include tighter integration and coordination of preclinical work and clinical trial designs, as now is being strongly advocated by both the animal and clinical groups (12). Animal models can provide informative data to rank orthogonal approaches and to rule out poorly immunogenic vaccines, for example. They belong squarely on the critical path, not as gatekeepers but as informants.
Should viral challenges be as close to human exposures as possible? Repeated low dose mucosal viral challenge has been proposed in lieu of high dose mucosal or intravenous exposure. Knowing that the overwhelming majority of infections worldwide take place via the mucosal route, it makes sense to work toward mucosal challenge models. Early work had shown that a much higher dose - 100 to 1,000-fold - of HIV or SIV was required to assure establishment of infection via mucosal (vaginal or rectal) exposure compared with intravenous inoculation. NHP experiments typically have utilized a dose high enough to infect 100% of the control or sham vaccinated animals, raising questions about the relevance of this dose of virus compared with the average human sexual exposure. The use of repeated low dose challenges has significant consequences on resources and experimental design. Not only are challenges much more labor intensive and requiring higher volumes of valuable virus stocks, but they also require greater numbers of animals per group to obtain statistically significant differences. Within the last few years, much more effort has gone into the testing of low dose repeated mucosal exposures, and it appears that they are feasible and are gaining acceptance (20) (21). Hand in hand with these challenge models comes the need to better understand how to elicit effective responses at the mucosal site of exposure and entry (22). Both are critically important areas for development and should be supported.
We know that lentivirus disease outcome depends upon the virus-host combination. How can the study of nonpathogenic outcomes in “natural” hosts help to inform the disease consequences in the new host? A related, equally key question in HIV infection that is also observed with many of the SIV and SHIV models is the variability in pathogenic outcomes in the outbred population. A minority of susceptible Asian macaques (10–20%) infected with SIV become elite controllers—are these similar to the adapted hosts in Africa—or are they genetically disposed? A recent study shows that disease has been observed in SIV infected African monkeys, and thus these adapted hosts may be persistently nonprogressive (5). Continued studies of these populations are clearly warranted on the basis of this new finding. We also need to understand cases of transient or occult infection (23), which may give further clues as to how to control HIV.
Despite the challenges inherent in the animal models, there has been significant progress in the past year or two. I list a few representative and key findings in Table 1. This is not meant as an exhaustive list, rather a taste of recent progress. In the first section of this table are some advances in pathogenesis research, while the second section describes new developments in vaccines, microbicides, and therapies. In Table 2, I list a number of recent and excellent reviews that address key areas of research relevant to this discussion. The sections below highlight some of the most exciting findings.
There has been progress in both murine and NHP models for HIV-1. In the past year, the female reproductive tract of humanized bone marrow-liver-thymus (BLT) mice was reconstituted with human CD4+ T and other relevant human cells, rendering these humanized mice susceptible to intravaginal infection by HIV-1 (24). There was also CD4+ T cell depletion in gut-associated lymphoid tissue (GALT) that closely mimics what is observed in HIV-1-infected humans. The authors demonstrated drug prophylaxis, suggesting that this might be a good model for microbicide studies. After much parsing of the HIV genome, a solution has been found to engineer an HIV-1 that can replicate well in pigtailed macaques and retain fairly good persistence (25). The importance of TRIM5-alpha as a restriction element was discovered several years ago, and this new, exciting model is based on the removal of the vif gene from HIV-1, thereby allowing the productive infection in M. nemestrina.
A few years ago (26, 27) it was observed that PD-1 is expressed on HIV-specific T cells that are associated with T-cell exhaustion and disease progression, suggesting a potential novel therapeutic pathway for enhancing immunity (27). This year two groups (28) (29) demonstrated that PD-1 blockade using an antibody to PD-1 is well tolerated and results in improved CD8+ T cell functional quality, proliferation of memory B cells, and increases in SIV envelope-specific antibody. The improved immune responses were concomitant with significant reductions in plasma viral load and also prolonged the survival of SIV-infected macaques. Blockade was effective during the early (week 10) as well as late (approximately week 90) phases of chronic infection even under conditions of severe lymphopenia.
Several novel approaches were published in the last year:
One important clinical goal is to determine if HIV can ever be eliminated, since even very long-term ART results in residual integrated virus—the viral reservoir (37). One of the most important roles that animal models can play is to test risky concepts prior to the clinic. Efforts to eliminate the virus can be modeled in the macaque without risk to HIV-infected subjects (38), and thus should be explored prior to clinical testing.
Animal models for AIDS research will continue to be developed and adapted for specific uses. If the exciting new “nearly HIV” model can be expanded and further utilized for drug and vaccine testing, it has the potential for saving many millions of dollars in preclinical stages of testing, since SIV-specific versions of the vaccine would no longer be necessary for proof-of-concept testing. The utility of this virus may still be limited because of differences in host factors, including MHC, between humans and NHP. The judicious use of NHP models in concert with clinical data offers a means to accelerate vaccine and microbicide development, as well as potential approaches to reduce the viral reservoir. As Dr. Anthony Fauci and colleagues noted in an opinion piece in Science in 2008 in summarizing the outcomes of an Think Tank sponsored by Fauci’s National Institute of Allergy and Infectious Disease of the NIH (17), the way forward stresses “attracting and retaining young researchers, developing better NHP models, and more closely linking NHP and clinical research.”
The author has funding from the U.S. National Institutes of Health (National Institute of Allergy and Infectious Disease (AI074379 and AI078064), the National Center for Research Resources (RR00163), and the National Institute for Child Health and Human Development (HD38653), and from the Bill & Melinda Gates Foundation (Global Health Innovation Grant).
Conflict of interest: The author declares no financial or commercial conflict of interest.