An understanding of the mechanism(s) by which some individuals spontaneously control HIV/SIV replication may aid in rational vaccine design. Here, we studied elite control of SIVmac239 in Mamu-B*08+ rhesus macaques. The association with MHC-I alleles implicates CD8+ T cells and/or natural killer cells in the control of viral replication. We attempted to break the ability of Mamu-B*08+ animals to control viremia by introducing point mutations into eight Mamu-B*08-restricted CD8+ T-cell epitopes, impairing the generation of epitope-specific CD8+ T-cell responses. Ten Mamu-B*08+ macaques were infected with this mutant virus, 8X. We compared the immune responses and viral loads of these animals to those of WT-infected Mamu-B*08+ macaques. The five typically strongest Mamu-B*08-restricted CD8+ T-cell responses were barely detectable in 8X-infected animals. Interestingly, these 8X-infected animals made unusually large responses to several unaltered, subdominant Mamu-B*08-restricted epitopes. By 48 weeks postinfection, 2 of 10 8X-infected Mamu-B*08+ animals controlled viral replication to <20,000 vRNA copy eq/ml plasma, while 10 of 15 WT-infected Mamu-B*08+ animals had viral loads of <20,000 vRNA copy eq/ml. These results suggest that high-frequency epitope-specific CD8+ T-cell responses restricted by Mamu-B*08 play an important role in establishing the control of viral replication.
Studies of chronically HIV-infected patients have failed to find correlations between the magnitude of T-cell responses directed against epitopes restricted by protective alleles and the control of viral replication (39
). Likewise, in recent work, no correlation between acute-phase T-cell responses and later set-point viral loads was seen in persons with a variety of HLA backgrounds (22
). Similarly, in our study of acutely infected Mamu-B*08+
animals, there was no apparent relationship between the magnitude of responses and control of viremia. For instance, at week 2 postinfection, WT-infected animal r99019 had massive CD8+
T-cell responses against Mamu-B*08-restricted epitopes, in excess of 35% of its CD8+
T cells (Fig. ). This animal's viral set point was >1 million vRNA copy eq/ml plasma. Interestingly, animal r99019's Mamu-B*08-restricted CD8+
T-cell responses contracted dramatically between weeks 2 and 4 postinfection, which was not the case with WT-infected ECs’ responses, although they were not as high in frequency at week 2 (Fig. ). Additionally, there was no clear difference between which Mamu-B*08-restricted epitopes were targeted by WT-infected controllers and progressors. WT-infected animals generally made a consistent repertoire of responses, targeting multiple Mamu-B*08-restricted epitopes, albeit with various magnitudes. Thus, simply making immunodominant responses targeting Mamu-B*08 epitopes during the acute phase of infection is not sufficient to establish control in WT-infected animals.
The sequence of plasma virus did not clarify why some animals became ECs and others did not. Controllers’ and progressors’ viral loads diverged by 5 weeks postinfection (see Fig. S2 in the supplemental material). In 8X-infected animals, six of the eight progressors had variation in the region containing Nef246-254
RL9b and Nef245-253
RL9c, whereas neither of the two ECs had such variation despite making immune responses against this epitope (Fig. ). However, the earliest viral variation detected in this epitope was at week 6 postinfection (data not shown), indicating that circulating virus in all animals should have been susceptible to T-cell responses targeting the Nef246-254
RL9b and Nef245-253
RL9c epitopes during at least the first 6 weeks of infection. Likewise, in the WT-infected animals, controllers and progressors had similar patterns and kinetics of escape (Fig. and data not shown). However, a caveat to these conclusions is that our sequencing identifies only dominant quasispecies circulating in the plasma. It is possible that applying new deep-sequencing methodologies may reveal important dynamics in sequence variation that occur below our current level of detection (5
If simply making high-frequency CD8+
T-cell responses against Mamu-B*08-restricted epitopes is not sufficient for control, what other factors contribute to elite control in WT-infected animals? We have yet to extensively investigate the heterogeneity of T-cell responses against the same epitope in the setting of acute infection. For instance, the use of different T-cell receptor repertoires could render some T-cell responses more effective than others, as was shown previously in the setting of vaccination (48
). Additionally, it is possible that different effector phenotypes of T cells are key. The responsiveness to antigen has been found to differ between cells from chronically infected progressors and controllers, with EC cells being able to secrete a wider array of cytokines and greater amounts of granzyme than cells from progressors (40
). Perhaps most fundamentally, T-cell responses do not occur in isolation but act in concert with the many other components of the immune system. Qualitative differences in CD8+
T-cell responses may be determined by the immunological milieu in which the cells are initially primed. What sort of signals (numbers of T-cell receptors engaged, cytokines present, and levels of costimulatory engagement) are desirable during priming and whether such signals can be manipulated by vaccination could prove to be fruitful areas of future research.
We also investigated the effect of immunodomination (the phenomenon in which the presence of higher-frequency responses dampens the frequency of subdominant responses) on the magnitude of CD8+
T-cell responses. We observed that typically subdominant responses expanded to higher frequencies in the absence of immunodominant responses restricted by the same MHC allele. A recent vaccine study of macaques similarly found that the deletion of an immunodominant epitope from a simian-human immunodeficiency virus challenge virus resulted in a significantly more robust anamnestic expansion of a typically subdominant epitope (35
). In our study, these subdominant responses did not, however, reach the levels of the high-frequency responses, which were “missing” (Fig. ). Thus, immunodomination has a role in determining the magnitude of CD8+
T-cell responses, but the size of the responses is apparently also limited by other factors. This result, using outbred animals expressing a variety of other MHC-I alleles, agrees with data from previous work conducted largely using inbred mice as models (24
). Several groups have shown that the frequency of naïve precursor CD8+
T cells affects the magnitude of effector responses generated (23
). Differences in precursor frequency could account for the failure of T cells directed against subdominant Mamu-B*08-restricted responses to expand to a greater extent in the absence of higher-frequency Mamu-B*08 responses. If TCR precursor frequency has a substantial role in dictating the magnitude of T-cell responses, altering immunodominance hierarchies by vaccination may prove to be difficult. However, it is also possible that the failure of these subdominant responses to expand to a greater extent could be due to immunodomination by T-cell responses restricted by MHC alleles other than Mamu-B*08—if this is the case, it is possible that dominance hierarchies are more malleable than our data indicate. The regulation of T-cell expansion and contraction during vaccination or chronic infection is currently poorly understood, and detailed studies of this topic would be very useful for vaccine design.
We had hypothesized that in the absence of the normal repertoire of Mamu-B*08-restricted responses, Mamu-B*08+ macaques would not become ECs. While the proportion of animals that controlled viremia was lower than would be expected for WT infection, 2 of 10 Mamu-B*08+ animals infected with the mutated virus still became ECs.
There are several hypotheses to explain why 2 of 10 Mamu-B*08+ macaques were still able to control 8X replication to very low levels. First, the mutant virus may be more easily controlled by an animal's immune system regardless of whether the animal expresses Mamu-B*08 or not. One of the two Mamu-B*08-negative animals infected with 8X became an EC, whereas the other had high viral loads (Fig. ). This may indicate a replicative defect in the 8X virus in vivo. In the Mamu-B*08+ animals, at week 2 postinfection, the geometric mean viral load of those infected with 8X was as high as the geometric mean viral load of those infected with WT virus (Fig. ). Therefore, no fitness defect in 8X was apparent during the first 2 weeks of infection. However, the reversion of four of the 8X mutations in vivo (Fig. ) suggests that mutations in 8X incur a replicative cost. Despite this, the majority of our 8X-infected Mamu-B*08+ macaques (which maintained these mutations) still had high levels of viral replication (Fig. ).
Another possibility is that the remaining responses against Mamu-B*08-restricted epitopes may in some cases be sufficient to control SIVmac239 replication to low levels. The sum of the Mamu-B*08-restricted responses made by the two 8X ECs was not notably different in magnitude than that of other 8X-infected animals (Fig. ). These two animals did have a somewhat skewed repertoire of Mamu-B*08-restricted responses compared to other 8X-infected animals, however, in that their CD8+ T cells targeted predominantly Nef (Fig. ). The strongest Mamu-B*08-restricted response of EC r01088 at week 4 was to Nef245-253RL9c, and the other EC, animal r04135, made an unusual response to the overlapping Nef246-254RL9b epitope, which had a primary anchor residue mutation introduced into 8X. The 8X-infected animals with higher levels of viremia tended to have their strongest Mamu-B*08 responses targeting Env or Vpr, although they generally made, and had escape in, Nef-directed responses also. The hypothesis that subdominant Mamu-B*08 responses mediated control could be tested by introducing point mutations into the additional Mamu-B*08-restricted epitopes that we found to accrue variation and by repeating the experiment reported here.
Third, it is possible that the surprising viral control by two 8X-infected macaques could be due to innate immune responses in Mamu-B*08+
animals tipping the balance to elite control. For instance, we know nothing about the different KIR alleles expressed by animals in this study. Based on human studies (36
), we could reasonably expect KIR to modulate the effect of Mamu-B*08
on viral replication. Information on KIR in rhesus macaques is limited, but a recent paper has defined a variety of KIR3DL polymorphisms, one of which was associated with higher viremia (8
). An interesting future avenue of research with Mamu-B*08+
animals will be the characterization of the KIR genes (or other host factors) in animals that become ECs versus progressors.
In conclusion, we found that high-frequency CD8+ T-cell responses against Mamu-B*08-restricted epitopes may play an important role in establishing control of viral replication. In the case of 8X-infected animals, these responses were not strictly necessary to achieve control of viral replication. Two animals controlled the replication of the 8X virus in the absence of high-frequency Mamu-B*08-restricted responses. There are several possible explanations for this unexpected result, as discussed above. Additionally, the presence of high-magnitude responses was not always sufficient to establish elite control in Mamu-B*08+ macaques infected with WT virus. Thus, our results do not support a model in which the simple presence of a few particular T-cell responses can always determine the outcome of an infection. Our study suggests several lines of future investigation that may help further illuminate the components of a successful immune response against immunodeficiency viruses.