Maintenance or Emergence of Chronic Phase Secondary Cytotoxic T Lymphocyte Responses after Loss of Acute Phase Immunodominant Responses Does Not Protect SIV-Infected Rhesus Macaques from Disease Progression

The simian immunodeficiency virus- (SIV-) infected rhesus macaque is the preferred animal model for vaccine development, but the correlates of protection in this model are not completely understood. In this paper, we document the cytotoxic T lymphocyte (CTL) response to SIV and its effects on viral evolution in an effort to identify events associated with disease progression regardless of MHC allele expression. We observed the evolution of epitopes targeted by CTLs in a group of macaques that included long-term nonprogressing (LTNP), slowly progressing (SP), normally progressing (NP), and rapidly progressing (RP) animals. Collectively, our data (1) identify novel CTL epitopes from an SP animal that are not restricted by known protective alleles, (2) illustrate that, in this small study, RP and NP animals accrue more mutations in CTL epitopes than in SP or LTNP macaques, and (3) demonstrate that the loss of CTL responses to immunodominant epitopes is associated with viral replication increases, which are not controlled by secondary CTL responses. These findings provide further evidence for the critical role of the primary cell-mediated immune responses in the control of retroviral infections.


Background
There is compelling evidence that the virus-specific cytotoxic T lymphocyte (CTL) response is an important factor in the control of human and simian immunodeficiency viruses (HIV and SIV, resp.) in infected individuals. Failures in antibody-based vaccines emphasize the importance of the CD8+ T lymphocyte response, but the disappointing results of the STEP HIV vaccine trial indicates a need for a more complete understanding of CTL responses in lentiviral infections [1]. The CTL response is restricted by the repertoire of major histocompatibility complex (MHC) Class I molecules that present viral epitopes, and a greater knowledge of these allele: epitope combinations is vital for identifying antigenspecific CTLs and measuring effector functions of antiviral cellular immunity. In addition, more information is needed about the selective pressure applied by CTLs to viral epitopes and how that influences viral evolution during SIV infection [2].
The expression of the rhesus macaque major histocompatibility complex (MHC) Class I molecule Mamu A * 01 has been determined to be associated with slower disease progression [3,4], which is similar to the protective effects of HLA B * 57 expression in humans [5]. One important technological development from this research was the creation of Mamu A * 01 tetramers [6]. Unfortunately, animals with 2 Journal of Biomedicine and Biotechnology Mamu A * 01 are in great demand, and the limited availability of these animals has slowed research in this area [7]. Other protective alleles include Mamu B * 17, which was initially reported as a protective allele that was enriched in longterm nonprogressor (LTNP) animals [8], but recent work indicates that animals with different Mamu B * 17 haplotypes have divergent disease courses [9]. A third Mamu allele-B * 08-was also reported to be associated with slower disease progression and is enriched in macaque populations that exhibit LTNP phenotypes [10].
CTL responses during HIV and SIV infection generally consist of several CTL clones recognizing many different epitopes; the size of those CTL clones can differ substantially resulting in dominant and subdominant responses [11,12]. Immunodominance in CTL response is influenced by several factors [13], including MHC expression [14], T cell avidity [15,16], proteosome processing [17], frequency of naïve precursors specific for a viral epitope [18], and viral sequence variability [19]. It has been shown that in SIV infection CTL selective pressure is a cause of viral escape mutations [20][21][22][23], and this phenomenon is one, but not the only, cause of vaccine failure [24,25]. Escape variants have been identified and characterized in Mamu A * 01- [26,27], B * 08- [23], and B * 17-restricted [25] CTL epitopes, and it has been shown that these viral escape variants are poorly recognized by subsequent MHC-matched hosts but can revert to wild type if CTL selective pressure is removed by transmitting the mutated viruses to MHC mismatched animals [28][29][30]; interestingly, SIV can also adapt to the fitness cost of certain escape mutations by generating compensatory mutations [29]. This paper presents detailed results from a group of eight rhesus macaques followed for over 4 years (or until progression to disease) to determine the interactions between the CTL response and viral evolution in chronic SIV infection. Our comprehensive analysis of CTL response monitoring and viral sequencing in animals of varying disease progression phenotypes enabled us to identify novel epitopes recognized by antiviral CTL responses. We also show that, in this study, slow progressors (SPs) and LTNP animals accrue fewer mutations in CTL epitopes than either normal or rapid progressor animals (NP and RP, resp.), and that increased mutations are associated with a lower CD4+ T cell count. Most importantly, we demonstrate that a loss of CTL response to immunodominant epitopes can be associated with disease progression despite the presence of other subdominant CTL responses. This was observed in an animal that expressed the protective Mamu A * 01 allele as well as an animal that expressed no known protective allele.

Experimental Animals and Tether
System. Animal care and treatment were in accordance with SFBR Institutional Animal Care and Use Committee (IACUC) approved protocols. Four Indian origin rhesus macaques (16037, 16040, 16041, and 16044), which had been vaccinated 24 months prior to this experiment with a Nef -deleted SIVmac239 live attenuated vaccine [31], were determined to be virus negative by RT-PCR and virus isolation at the time of challenge. Four additional Indian-origin, retrovirus-negative naïve rhesus macaques (11896, 12416, 12900, and 14757) were used as a control group. These eight animals were placed on a tether system during the first 4 weeks of infection in order to demonstrate the usefulness of this technology in the study of viral immunology during the acute phase of infection. Details of the placement of the tether system, which allowed for blood sampling without sedation, are provided elsewhere [32].

Viral
Challenge and Viral Loads. The animals were challenged intravenously with 1 mL of the pathogenic isolate SIVmac251 at a dose of 100 TCID 50 . PBMCs were isolated by Ficoll-Hypaque separation for use in IFN-γ ELISPOT assays, CTL functional assays, flow cytometry, and viral isolation. Lymphocyte phenotyping was performed by multicolor flow cytometry as described elsewhere [31]. Plasma was utilized to determine viral loads by (nucleic acid sequence based analysis) NASBA assay (from Advanced Bioscience Labs, Inc.) with a lower detection limit of 500 RNA copies per 100 μL plasma. NASBA technology was utilized instead of real-time RT-PCR because all the plasma samples collected while animals were on tether contained sodium heparin. For relevant comparison, the chronic viral loads were also performed with this assay.

MHC Allele
Genotyping. RNA was isolated from macaque PBMCs with the RNAeasy kit (Qiagen). RNA was converted to cDNA by amplification with reverse transcriptase and random primers. This cDNA was then subjected to PCR with a forward primer common for the human A, B, and C alleles and reverse primers specific for either alleles A or B [33]. The amplified fragments were cloned into the pCR2.1 (Invitrogen) plasmid and sequenced. In addition, animals were screened for Mamu A * 01, A * 02, B * 08, and B * 17 by sequence-specific PCR as previously described in [34].

CTL Response Measurement and Epitope Identification by ELISPOT Assay.
Peptide-specific release of IFN-γ from lymphocytes was measured by ELISPOT in accordance with the manufacturer's instructions (U-Cytech, Utrecht, The Netherlands). SIV-specific responses were determined with 15-mer overlapping SIV peptides (NIH AIDS Research and Reference Reagent Program) at a concentration of 2 μg/mL. All conditions were performed in duplicate with a positive control consisting of cells stimulated with Staphylococcus enterotoxin A and B (SEA/B), and a negative control of RPMI containing 0.5% dimethyl sulfoxide (DMSO) (Sigma) alone. To minimize bias, all plates were counted by one individual in a blinded fashion.
Responses were normalized for CD8+ T cells determined by flow cytometry and adjusted for DMSO background per animal. For those epitopes of known restriction, tetramer studies were performed to ensure that it was the CD8+ T lymphocytes that recognized the epitope (data not shown). For epitopes of unknown restriction, CD107 and Cytoxilux (OncoImmunin, Inc.) assays were utilized to determine that apoptosis of target autologous lymphoblastoid B cells was peptidespecific and that the CD8+ T lymphocyte population was positive for degranulation in a peptide-specific manner [32]. Epitopes identified by IFN-γ ELISPOT are putative only, because they were determined by single 15-mer peptides or overlapping regions of two or more 15-mer peptides that contain the 8-to 10-amino acid epitope. To identify these putative epitopes, PBMCs from each animal were utilized in ELISPOT assays with peptide pools representing the full SIVmac239 amino acid sequence. Positive pools were split into smaller pools, and assays were repeated. This deconvolution continued until either one single 15-mer peptide or two overlapping 15-mer peptides were identified as peptides that elicited production of IFNγ. No additional mapping of epitopes was performed in this study.

Virus Isolation and Epitope Sequencing.
PBMCs from infected animals were cocultured with CEMx174 cells with or without CD8+ lymphocyte depletion with magnetic beads coated with anti-CD8+ antibodies (Dynal). Supernatant from these cultures was assayed by ELISA for SIV gag protein p27 [31]. Infected CEMx174 cells from ELISApositive samples were collected and DNA extracted with the PureGene DNA Isolation Kit (Gentra Systems) in accordance with the manufacturer's protocols. The DNA was then used in PCR reactions to amplify fragments of SIV provirus containing relevant epitopes for Gag, Rev, Vif, Tat, Pol, and Nef. The primers employed were as follows: Gag F , 5 GCCTGGTCAACTCGGTACTC3 ; Gag R , 5 GTGGAC-CTAACTCTATTCCTGTTACAA3 ; Vif F , 5 GAAGGGACC-CGGTGAGCTATTG3 ; Vif R , 5 AGGAGGAGGTCCTGG-TCTCCATC3 ; Tat F , 5 GCTGCAGGTTCCCGAGAGCT3 ; Tat R , 5 ACAAAACTGGCAATGGTAGCAACACT3 ; Pol F , 5 CAGGTCCCAAAATTCCACTTACCAG3 , Pol R , 5 AAT-GCCATGAGAAATGCTTCCAATT3 ; Nef F , 5 CTCTCT-GCGACCCTACAGAGG3 ; Nef R , 5 GCATTTCGCTCT-GTATTCAGTC3 . These fragments were then cloned into pCR2.1 (Invitrogen) and sequenced. A minimum of five clones of each gene were sequenced for each time point for each animal to determine the consensus sequence. Any differences in sequence were resolved by repeated cloning and sequencing. In animals in which virus could not be isolated by coculture, but viral loads were detected by NASBA, the viral epitopes were sequenced by utilizing the viral RNA extracted from plasma for the NASBA assay. In these samples, a calibrator RNA was added for the NASBA assay; the calibrator consisted of a mismatched Gag sequence and that precluded our cloning of Gag epitopes for these time points.
2.6. Statistical Analysis. All statistical analyses were performed with GraphPad Prism (Windows version 4.03, GraphPad Software, San Diego, CA). Data was first tested for normality using the Kolmogorov-Smirnov (KS) test, with the P-value determined by the Dallal and Wilkinson test. The acute phase data sets of both the vaccinated and unvaccinated groups failed the normality test (P = .0253 and P < .0001, resp.). Data were then transformed to log scale to account for the variability of animal modelling and to improve further statistical analyses by normalizing the data. Subsequent analyses are detailed in the results and included, Mann-Whitney comparison of mean responses, as well as Pearson's correlation to determine significant correlations of the CTL response over time.

Outcome of SIV Challenge and Expression of Protective
Alleles. Using the tether system to study the acute phase after viral challenge, we recently demonstrated that all the macaques inoculated with the live-attenuated SIV vaccine were protected from pathogenic SIV infection [32]. As Table  1 shows however, while three (16037, 16041, and 16044) of four animals controlled SIV infection very efficiently and remained SP or LTNP, one vaccinated animal (16040, Mamu A * 01 + ) progressed to simian AIDS (SAIDS) at 118 weeks postchallenge. On the other hand, and as expected, unvaccinated animals became infected with SIV and progressed to disease at 12 (11896), 16 (12900), and 52 (12416) weeks postchallenge (WPC); one naïve animal, however, spontaneously controlled infection and remained as LTNP (14757). Three of the four animals that controlled infection had the protective Mamu A * 01 allele (16037, 16041, and 14757); the naïve LNTP animal (14757) also had the Mamu B * 08 allele. Table 1 also includes a summary of the epitopes that were recognized by CTL responses (as measured by IFN-γ ELISPOT responses) that significantly changed over the course of infection. The epitopes that triggered IFN-γ ELISPOT responses were also shown to induce degranulation (surface expression of CD107a) of CD8 + T cells and to induce killing of autologous lymphoblastoid B cell lines (data not shown). The SP/LTNP group included two vaccinated animals with the Mamu A * 01 allele, 16037 and 16041, and one naive animal with the Mamu A * 01 and B * 08 alleles, 14757. All animals in the Mamu A * 01 positive subgroup showed a statistically significant positive correlation in the CTL response to the previously described Mamu A * 01restricted immunodominant Gag CTPYDINQM epitope [35]. None of the viral loads in these animals exceeded 500 RNA genome copies per 100 μl of plasma in the 4 years of this study. In addition, unvaccinated animal 14757 had significant increases in CTL responses to the recently reported Mamu B * 08 restricted Nef RRLTARGLL epitope and to an unreported putative Gag TAPSSGRGGNY epitope late in infection.

Epitope Emergence in Slow-and Long-Term Nonprogressors.
The fourth animal in the SP/LTNP group was the vaccinated SP animal 16044, who lacked Mamu A * 01, Mamu B * 08, and Mamu B * 17 alleles as confirmed by cloning and sequence-specific PCR screening, and showed a significant negative trend in a previously unreported putative Vif epitope YFPCFTAGEVR. The establishment of an early strong CTL response to the Vif YFPCFTAGEVR epitope,   LAV: live-attenuated vaccine, LTNP: long-term non progressor, SP: slow prog., NP: normal prog., RP: rapid prog., (time of euthanasia in weeks post-challenge). [1] Bolded alleles have been reported as protective. [2] Variant (less than 5 amino acid changes in α1 and α2 domains from published sequence of allele listed). [3] Novel (more than 5 amino acid differences in α1 and α2 domains from published sequence of allele listed). P Putative Epitope, * Immunodominant Epitope.
coupled with initial control of SIV replication, implied that this might be an epitope with protective characteristics. This epitope is located in a functionally important area of the Vif protein and is highly conserved across SIV and HIV strains. The Vif YFPCFTAGEVR epitope showed a threonine (T) to alanine (A) mutation within the putative epitope first noted at 48 weeks post-challenge with no apparent increase in viral replication prior to that time. To determine what effect the CTL response was having in protecting this animal from progressing to disease, we performed additional ELISPOT pool studies on fresh and frozen PBMC samples. As expected, CTL responses to the wild-type peptide were significantly lower at timepoints at which the circulating virus had mutated (P = .0271, one-tailed Mann Whitney test, Figure 1(a)), which is consistent with previous studies on escape mutations and implies that the CTL population specific for the wild-type epitope was declining, presumably because the virus in the animal no longer had the wild-type epitope. Pool studies on archival PBMC samples identified additional epitopes in the Gag protein that were being targeted by CTLs. The strongest ELISPOT response was seen with the emergence of the putative Gag MYNPTNILDVK epitope (<45 WPC versus >45 WPC, P < .0022, onetailed Mann Whitney test) that occurred after 45 weeks post-challenge, which was concurrent with a significant loss of response to the Vif YFPCFTAGEVR epitope (<45 WPC versus >45 WPC, P < .0001, one-tailed Mann Whitney test). Coincidentally, viral loads steadily increased from this time point onwards, despite strong CTL responses to Gag MYNPTNILDVK (Figure 1(b)). Whether the loss of control over viral replication by CTLs targeting these epitopes was due to viral escape, an intrinsic deficit of effector functions, or some other silencing mechanism such as increased PD-1 expression, was not determined. to isoleucine (I) mutation was first observed in position 2 of the Gag CTPYDINQM epitope at 70 WPC. This is a well characterized escape mutation known to abrogate MHC binding [26] and was associated with a significant decrease in CTL response (Figure 2(a)). Other known escape mutations were first detected at 48 WPC in position 1 and 8 of the Tat STPESANL epitope (serine to proline, and leucine to proline) and these have also been reported to abrogate MHC binding [26]. This is consistent with declining numbers of CTLs specific for the wild-type epitope, which would be expected in animals no longer carrying virus with wildtype sequence. Interestingly, a threonine (T) to serine (S) mutation was observed in position 2 of the Mamu A * 01 Pol STPPLVRLV epitope prior to 48WPC, but was not associated with a decrease in CTL response (Figure 2(a)). Analysis of viral loads (Figure 2(b)) illustrated that the Pol STPPLVRLV epitope was not protective as the viral load increased despite a CTL response to this epitope. The unvaccinated normal progressor macaque 12416 (lacking known protective alleles) showed an initial ELISPOT response to a Gag peptide pool that declined before individual epitopes could be deconvoluted. This animal had significant ELISPOT response increase over time to a putative Nef epitope RTMSYKLAIDM. Despite the increased response to this epitope, the viral load at 54 WPC was 420,270 RNA genome copies per 100 μL of plasma (data not shown). Neither of the rapid progressors (12900, 11896) showed any detectable ELISPOT response.

Mutations in Relation to Disease Progression.
We counted the number of mutations that we identified for each animal throughout the course of this study and its relation to disease progression. We found significantly higher percentages of CTL epitope mutations in the normal or rapid progressing animals (Figure 3(a)) as compared to SP/LTNP animals. Animals that carried viruses with wild-type CTL epitopes maintained significantly higher CD4+ T cells counts than animals with virus that has mutated at CTL epitopes (Figure 3(b)). To determine patterns in the emergence of CTL epitope mutations, we also compared the percentage of mutated CTL epitopes found on each viral protein for all time points and observed that accessory proteins appeared to mutate at a higher frequency than the structural proteins in this sample (Figure 3(c)).

Discussion
Data generated from our studies during the chronic phase of SIV infection adds to the understanding of viral evolution in the context of CTL selection by analyzing epitope evolution over 4 years in slow-and long-term nonprogressors, as well as normal and rapid progressors. The importance of this study lies in the observation, albeit in a small group of animals, that the loss of immunodominant CTL responses established in acute infection may lead to disease progression regardless of the expression of known protective class I MHC alleles. In addition, this increase in viral replication occurred despite late-emerging CTL responses of similar magnitude to the one seen for dominant epitopes. Previous studies using animals with Mamu A * 01 and/or Mamu A * 02 alleles have demonstrated that the evolution of CTL responses to subdominant epitopes can arise after loss of Mamu A * 01 immunodominant responses [36], but a loss of immunodominant responses is followed by increased viral replication, even if normal CTL responses to subdominant epitopes (Pol) are present [20,37]; our data from the Mamu A * 01 + animal 16040 ( Figure 2) agrees with this observation. In the same line of evidence, and outside the context of the Mamu A * 01 allele expression, the loss of the initial immunodominant Vif CTL epitope in animal 16044, resulted in increased viral replication and disease progression despite strong surge of subdominant CTLs (Figure 1). These findings in the SIV system are also in agreement with results from a case report of a closely studied HIV-infected individual [38] and a recent study in a large cohort of acutely HIV-infected individuals that were followed into the chronic phase of the infection, and that showed that the preservation of the initial CD8+ T-cell immunodominance patterns from the acute into the chronic phase of infection was significantly associated with slower CD4+ T-cell decline and subsequent control of viremia [39]. These results appear to be in conflict with a previous study that found that subdominant CTL responses may have been associated with in vivo HIV-1 viral control [40]; however, this latter study utilized few time-points from chronic HIV-infected patients, whereas the study of Streeck et al. [39], like our study, followed the CTL evolution of acutely HIV-1-infected subjects that progressed to chronicity. Similarly to the results of Frahm et al. [40], Streeck et al. did not find an association between immunodominant CTLs and viral control when samples from the chronic phase were evaluated; however, like we report here, the positive correlation was observed when the immunodominant CTLs that aroused in the acute phase of infection were still maintained in the chronic phase [39].
Our studies also identified a novel SIV epitope that elicited strong CTL responses outside of known protective class I MHC allele expression. The putative Vif YFPCF-TAGEVR epitope was the target of a protective CTL response for a slow progressing animal (16044) that did not express any known protective allele. This peptide sequence is highly conserved in various HIV and SIV strains including SIVsmm, SIVmac, HIV2a, and HIV2b; mutagenesis of the Vif protein has identified the same region of the protein as functionally important, with the mutations in the tyrosine (Y), phenylalanine (F), or cysteine (C) residues reducing activity of the protein to 1 to 15% of wild type [41]. This observation is in agreement with recent studies in HIV-1infected individuals that suggest that control of infection may be associated with acute-phase CD8 responses capable of selecting for viral escape mutations in highly conserved regions of the virus [42]. In addition, we demonstrate that slow-and long-term nonprogressors acquire fewer mutations in CTL epitopes than normal or rapid progressors and a higher CD4+ T cell count was associated with CTL responses to wild-type epitopes (Figure 3). This is similar to what has been observed for a large cohort of HIV infected individuals [43] and implies that the protection achieved with certain alleles can be attributed to a failure of the virus to accumulate escape mutations.
All these observations combined suggest that natural containment of AIDS virus replication may be related to a combination of immunodominance and viral escape from CD8 + T-cell responses; that is, those individuals who generate acute dominant CTL that target constrained viral epitopes may have a better chance to control the infection.

Conclusions
This study contributes to the body of knowledge regarding CTL immune responses in SIV-infected rhesus macaques and increases our understanding of viral evolution in the context of SIV infection. Although the sample group is small, our observations are detailed and consistent with previous reports and contribute to the validation of the SIV/rhesus macaque model of AIDS. In addition, our identification of novel epitopes provides preliminary data justifying the continued analysis of CTL populations in these animals to better understand CTL function. Our data also add to the understanding of the associations between CTL epitope mutation and disease progression and CD4+ T cell counts and imply that accessory proteins accrue mutations more frequently. Lastly, we show that a loss of immunodominant immune responses to conserved viral epitopes can lead to disease progression despite maintained or emerging CTL responses outside the context of known protective allele expression, which has implications for the rational selection of antigens for HIV vaccine design.