Control of HIV-1 Pathogenesis in Viremic Nonprogressors Is Independent of Gag-Specific Cytotoxic T Lymphocyte Responses

ABSTRACT Viremic nonprogressors (VNPs) constitute a very scarce group of untreated human immunodeficiency virus type 1 (HIV-1)-infected individuals who maintain stable CD4+ T cell counts despite high levels of HIV-1 replication. The specific factors associated with this atypical control of the HIV infection have been poorly described. Since specific T cell responses seem to be one of the main causes of HIV-1 control in elite controllers, we studied whether HIV-1 Gag-specific cytotoxic T lymphocyte (CTL) responses could also modulate disease control in VNPs. We characterized the immune responses from four VNPs compared to those of five standard progressors (SPs) during the first years of HIV-1 infection. We observed no differences in the breadth and frequency of Gag-specific cellular responses. Furthermore, we obtained 217 HIV-1Gag clonal sequences in which the viral variability of Gag increased over 3 years of infection for synonymous and nonsynonymous mutations in both VNPs and SPs. VNPs evolution rates in gag were comparable to SPs. This observation is in line with a similar accumulation of CTL putative escape mutations in Gag epitopes targeted by CTL responses. Altogether, the absence of viral pathogenesis in VNP individuals seems to be independent of HIV-Gag-specific CTL responses. This novel information guides to the study of alternative mechanism of HIV-1 pathogenesis control. IMPORTANCE Control of HIV infection has been widely studied in elite controllers or long-term nonprogressor models. However, there is a less-known group of individuals, termed viremic nonprogressors (VNPs), who maintain stable CD4+ T cell counts despite high plasma viremia. The mechanisms involved in this remarkable control of HIV-1 pathogenesis clearly have implications for the development of new drugs and vaccines. We show here for the first time that VNPs have immune responses and HIV-gag evolution similar to those of standard progressors. Remarkably, we demonstrate that the mechanism of pathogenesis control in these individuals differs from some elite controllers that are reported to have improved immune control. This is noteworthy since it opens the door to new, as-yet-unknown mechanisms for HIV control. Our novel results advance the understanding of mechanisms involved in viremic nonprogression and suggest that there are alternative mechanisms to the adaptive immune responses for an effective control of viral pathogenesis.

U ntreated human immunodeficiency virus (HIV) infection leads to a progressive reduction on CD4 ϩ T lymphocytes, which finally results in the development of AIDS-defining symptoms. This severe immunosuppression is typically associated with high levels of viral replication (1, 2). However, the termed HIV-1 viremic nonprogressors (VNPs) maintain stable CD4 ϩ T cell counts despite high plasma viremia (3). This infrequent phenotype resembles the natural infection of sooty mangabeys and African green monkeys by the simian immunodeficiency virus (SIV) (4). SIV-infected natural host nonhuman primates rarely progress to AIDS and have normal life spans despite displaying high levels of viral replication. Both human VNPs and sooty mangabeys have the common characteristic of presenting low immune activation despite the high viremia (3,5).
The specific viral and host factors associated with this atypical control of the HIV-1 infection are poorly described. Recent data suggest that VNPs might be infected with virus with impaired fitness (6), although previous studies showed no significant alterations in env-and nef-associated functions (7,8). On the other side, limited infection of Tcm and Tscm CD4 ϩ T cells in VNPs has been reported (9), suggesting that host factors rather than specific viral properties may allow VNPs to dodge HIV-1 pathogenesis and chronic immune activation. Research performed in nonhuman primates have suggested that SIV-specific cytotoxic T lymphocyte (CTL)-mediated immune responses are not responsible for the lack of disease progression in the SIV natural host (10,11). However, no further immunological studies have been performed in humans.
An effective HIV-specific CTL response is postulated as one of the main causes of HIV-1 control in elite controllers (12), but there are no studies addressing whether VNPs could avoid the HIV-1 disease progression through alternative CTL-related mechanisms. It is therefore essential to understand whether cellular HIV-specific immune responses play a role in VNPs. This study evaluates whether HIV-specific CTL responses are effective in controlling the viral pathogenesis in VNPs compared to HIV-infected standard progressors (SPs) during 3 years in the absence of combination antiretroviral therapy. Adaptive immune pressure in a highly replicating viral environment might contribute to select viral mutants with impaired viral fitness and consequently reduced viral pathogenesis.

RESULTS
Subjects. Clinical characteristics from the studied individuals are depicted in Table  1 CTL responses between VNPs and SPs. HLA class I was genotyped for all the individuals (Table 1). None of the participants carried protective HLA allele B*5701 or B*2701. Rapid progression alleles (B*3503) were also absent in both groups.
In order to characterize HIV-specific CTL responses in the VNP group, we assessed the production of gamma interferon (IFN-␥) by peripheral blood mononuclear cells (PBMCs) stimulated with Gag peptides. Longitudinal analysis was only performed in the VNPs, since PBMC samples were not available for SPs at t 0 . Nonetheless, comparisons between VNPs and SPs were performed at t 3 .
CTL responses per subject are depicted in Fig. 1a and b. Specific epitopes in p24 Gag and p6 Gag -32 (MREPRGSDIAGTTSTL), 41 (YVDRFYKTLRAEQASQEV), and 66 (KELYPLAS LRSLFGNDPSSQ)-were overrepresented in both VNPs and SPs. Those regions seemed to be immunodominant in the early phase of the HIV-1 infection. The magnitudes of the responses slightly increased over time in VNPs, but not their breadths ( Fig. 1c and  d). When both groups were compared at t 3 , similar responses were found, but the frequency was slightly higher in SPs ( Fig. 1c and d). Together, these data suggest that VNPs and SPs presented similar T cell HIV-specific responses, although the breadth tends to be wider in SPs. Viral evolution in VNPs and SPs. In order to identify the impact of HIV-specific CTL responses in viral evolution, we analyzed 217 clonal Gag sequences obtained from plasma by limiting dilution. For each individual, we obtained a median of 12 independent Gag sequences (range, 10 to 14) in two different time points separated by 2.5 to 3 years. In all cases, sequences grouped independently per subject (Fig. 2). Since VNP-4 carried non-B subtype virus, these sequences clustered distantly from the rest. Gag sequences diverged over time in all subjects, with t 0 and t 3 clustering separately in most cases. Using a p-distance algorithm, we calculated the diversity of the sequences represented in the tree. This parameter measures the proportion of changes in each sequence calculated as synonymous (pS) or nonsynonymous (pNS) (Fig. 3). Consistently, we observed a longitudinal increase in genetic diversity over time for both VNPs and SPs ( Fig. 3a and b), although statistically significant differences were not reached. This supports that genetic diversity increases over time, as suggested by the phylogenetic tree.
To compare whether the degree of diversity is different between VNPs and SPs, we measured the global evolution per individual (Fig. 3c). We observed no changes between the two groups in either synonymous or nonsynonymous mutations. Likewise, there were no differences when we analyzed Gag regions p7, p24, or p2 separately (data not shown). Finally, we checked whether there was positive evolution by examining the pNS/pS ratio. No significant differences were found, and both ratios were Ͻ1, indicating that there was not a predominant positive selection in any of the groups (Fig. 3d).
Altogether, these data reflect continuous viral evolution between VNP and SP during the study period despite differences in viral pathogenesis. Therefore, no changes in the rate of evolution were observed between groups.
CTL-associated epitope evolution. We further evaluated the appearance of Gag immune mutants as an indirect measure of CTL pressure. For each individual, we checked intraepitope viral evolution during follow-up (Fig. 4). Both groups developed putative escape mutations in gag epitopes during follow-up. The mutations were mainly concentrated in the regions shared with a specific HLA-restricted epitope, but also potential compensatory adjacent mutations were observed. Putative escape mutation appeared in epitopes that induced low-, medium-, and high-magnitude CTL responses, as shown in Fig. 4.

DISCUSSION
Due to the rarity of the VNP phenotype among HIV-infected individuals, few cohorts with accurate follow-up have been described. In this study, we monitored VNPs and SPs longitudinally in order to understand whether cellular HIV-specific immune responses play a role in controlling viral pathogenesis. We observed no improved immune response in the VNP group, as happens in other HIV-1 nonprogression phenotypes such as elite controllers or long-term nonprogressors (13,14).
Major distinguishing features of SIV infection in its natural hosts (African green monkeys and sooty mangabeys) include the absence of disease progression (despite displaying high viremia and high viral replication rates in the intestine), the rapid resolution of virus-induced inflammation, the lack of microbial translocation, rapid control of viral replication in secondary lymphoid organs, and a lack of viral trapping by follicular dendritic cells (FDCs) in follicles (17)(18)(19)(20)(21)(22)(23)(24). Recent data demonstrate that, in nonpathogenic SIV infection of African green monkeys, natural killer cells accumulate in an interleukin-15-dependent manner in the follicles of secondary lymphoid organs of the animals and exert efficient control of viral replication within lymph nodes (25). Therefore, the role of NK cells should also be explored in HIV-infected subjects with a VNP phenotype to understand their contribution in limiting viral pathogenesis. However, the extremely scarce number of HIV-infected individuals with this nonprogressive disease profile, along with the current recommendations of rapid antiretroviral therapy initiation in all patients irrespective of the time since HIV-1 transmission (26), may limit the investigation of this hypothesis. In this sense, the conclusions of our study are conditioned to its small sample size. Likewise, the absence of baseline samples for SPs limited our power to reach robust conclusions when global immune responses were measured. However, this limitation was compensated for by performing an extensive study of the immunological marks within the viral genome using a significant number of clonal sequences per individual.
To our knowledge, this is the first report evaluating the immune cytotoxic response in VNP HIV-infected individuals. We here demonstrated that HIV-specific CTL responses are not responsible for the control of HIV-induced pathogenesis in subjects with an absence of CD4 T-cell depletion despite high viral replication. Alternative mechanisms, including humoral responses or differential infection of T cell CD4 ϩ subsets, might be an alternative mechanism of protection against HIV-mediated immune depletion. Further research will unravel whether the VNP phenotype is associated with additional host factors beyond the scope of the present study.

MATERIALS AND METHODS
Subjects. We selected four VNPs and five untreated HIV-infected SPs from the IrsiCaixa VNPs Cohort and the Spanish AIDS Research Network Cohort (CoRIS) (27). VNPs were defined as HIV-infected subjects with viral loads above 10,000 HIV RNA copies/ml that maintained CD4 ϩ T cells counts above 500 cells/l for more than 5 years with decays of less than 100 cells/l/year. SPs are HIV-infected subjects within the same range of plasma viremia but with a decay in CD4 ϩ T cells counts greater than 100 cells/l/year, falling below 500 cells/l after 3 years of monitoring. All of the samples were retrospectively obtained from naive individuals within 4 years after infection. The seroconversion date was estimated using last negative HIV test.
HLA typing and assessment of HIV-specific CTL responses. High-resolution HLA class I typing for alleles A, B, and Cw was performed at the Blood and Tissue Bank of Barcelona. Comprehensive HIV-1 epitope screening of optimal responses was carried out using the IFN-␥ ELISpot assay, as previously described (28). PBMCs were stimulated with a bulk of 66 clade B consensus overlapping peptides covering the entire Gag protein. Wells were considered positive above the background level, as previously reported (29). The breadth of HIV-specific responses was calculated as follows: (number of positive responses/number of peptides tested) ϫ 100. The magnitude of response was the additive response of all the positive wells.
Gag amplification and phylogenetic analyses. Plasma samples were collected at two time points per individual separated by 2 to 3 years (t 0 and t 3 ). Viral RNA was isolated from plasma (Qiagen) and retrotranscribed in cDNA using the 3=-outer primer and SSIII retrotranscriptase (Invitrogen). To prevent resampling, gag was amplified from cDNA using limiting-dilution clonal nested PCR as previously described (30). To ensure single gene amplification, we only sequenced positive PCRs if less of 25% of the reactions were positive. This translated to a minimum of 86% probability of clonality according to Poisson distribution. Additional quality control was done by discarding residual sequences with double peaks in the chromatograms. Only gag clonal sequences were included in the phylogenetic analysis.