Elicitation from virus-naive individuals of cytotoxic T lymphocytes directed against conserved HIV-1 epitopes

Cytotoxic T lymphocytes (CTL) protect against viruses including HIV-1. To avoid viral escape mutants that thwart immunity, we chose 25 CTL epitopes defined in the context of natural infection with functional and/or structural constraints that maintain sequence conservation. By combining HLA binding predictions with knowledge concerning HLA allele frequencies, a metric estimating population protection coverage (PPC) was computed and epitope pools assembled. Strikingly, only a minority of immunocompetent HIV-1 infected individuals responds to pools with PPC >95%. In contrast, virus-naive individuals uniformly expand IFNγ producing cells and mount anti-HIV-1 cytolytic activity. This disparity suggests a vaccine design paradigm shift from infected to normal subjects.


Background
Although it has been more than 20 years since the discovery that HIV-1 is the cause of acquired immune deficiency syndrome (AIDS), we are not yet close to realizing a vaccine to halt the devastation created by the AIDS pandemic [1]. HIV-1 clearance by the human host immune system and development of effective natural immunity have never been observed in AIDS patients. Consequently, correlates of immune protection are unknown (reviewed in [2]). Several features of HIV-1 biology have resulted in unprecedented challenges to vaccine development [3,4]. First, the virus has extraordinary genetic diversity on a global population scale as well as at the level of the single infected person [5,6]. With up to a billion new viral particles produced per day in an infected individual, HIV-1 genetic variability can be greater within one host than the worldwide variability of influenza A virus in any year. This sequence/epitope variability thwarts effective cellular and humoral immune responses. Second, HIV-1 infection targets CD4 + helper T cells, thereby blunting mechanisms that normally amplify adaptive immunity. Third, HIV-1 is difficult to neutralize because the viral envelope glycoproteins are protected by a glycan shield, and exist in several distinct conformations, rendering conserved epitopes largely inaccessible to antibody-mediated neutralization [4].
CD8 + cytotoxic T lymphocytes have been shown to contribute to HIV-1 containment through direct killing of infected cells and production of soluble anti-viral media-tors [7][8][9]. In one study, AIDS patients with high anti-HIV-1 specific CD8 + CTL levels in blood generally exhibit better clinical status [10]. Furthermore, in vivo containment of HIV-1 replication appears coincident with the emergence of specific CTL, well before neutralizing antibody [8]. Conversely, viral mutations resulting in loss of HIV-1 or simian immunodeficiency virus (SIV) recognition by CTL are linked to increased viral replication in vivo and diminished immune function [11,12]. Experimental depletion of CD8 + T cells prior to SIV infection in monkeys leads to uncontrolled viral replication, accelerated disease and death [13]. Collectively, these data suggest that the induction of CTL responses to an appropriate selection of viral CTL epitopes prior to HIV-1 exposure might comprise one important element in HIV-1 vaccine development. This current study addresses elicitation of CTL against conserved HIV-1 epitopes using blood from healthy donors and the implications of our findings on future preventive HIV-1 vaccine designs.

Generation of a consensus HIV-1 proteome
A consensus HIV-1 proteome bearing a consensus sequence of the HIV-1 gene products (GAG, POL, ENV, VIF, TAT, REV, VPU/VPX, VPR, and NEF) with the variable residues masked was obtained as indicated below. First, multiple sequence alignments for the relevant HIV-1 gene products including representative sequences from all HIV-1 clades were retrieved from the HIV databases in Los Alamos [14] (Table 1). Then, sequence variability in the alignments was computed using a variability metric (V) formally identical to the Shannon entropy equation [15]. Briefly, V per site is given by: where Pi is the fraction of residues of amino acid type i, and M is equal to 20, the number of amino acid types. V ranges from 0 (total conservation, only one amino acid type is present at that position) to 4.322 (all 20 amino acids are equally represented in that position). Note that in order to achieve the maximum value V = 4.3, at least 20 sequences are required. Gap symbols (-) were considered for deriving the consensus sequence but are not computed for the variability calculations. Finally, consensus sequences with the variable positions masked were obtained as follows. A variability threshold (Vt) was set 1.0, and at each position in the alignment V was compared to the Vt, if V b Vt the most frequent residue is selected as the consensus, otherwise if V > Vt a "." symbol is used to mask the sequence at that position (the HIV-1 consensus proteome generated in such a way is shown in Additional file 1).
ranking peptides. Cycling through the above steps results in a table consisting of a predicted HLA I binding profile for each CTL epitope.

Computing the cumulative phenotype frequency
Given a set of HLA I alleles (N), their cumulative phenotypic frequency (CPF) in a given population can be computed from the gene and haplotype frequencies of the individual alleles using the equation 2.
where the EPF i term account for the Expected Phenotypic Frequency of each of the alleles (i), and the ECPFij term accounts for the Expected Combined Phenotypic Frequency of each two distinct pair of alleles (ij) in the N set. EPFi is calculated through equation 3: where GF i is the gene frequency of the allele i. On the other hand, given a pair of alleles (ij), ECPF ij is computed upon considering whether they belong to the same HLA I gene (HLA-A, -B, -C) or not. If they do belong to the same gene, we assume a Hardy-Weinberg equilibrium and calculate ECPF ij as follows: However, if the two alleles belong to different genes and their haplotype frequency is available (HF ij ) we calculate ECPF ij using the following formula: If no haplotype frequency is available, ECPF ij is computed as if they belonged to the same gene using equation 4. In this 'study, HLA I allelic and haplotype frequencies used for the above calculations were those published by Cao et al [17], which corresponded to 5 major American ethnic groups (Black, Caucasian, Hispanic, Native American, and Asian). Note that for computing CPF, we have used a method similar to that of Dawson et al. [18] which, unlike the methods of Schipper [19] and Gulukota [20], considers haplotype frequencies and hence linkage disequilibrium. However, in contrast to Dawson et al., we minimized the number of peptides to be included in the vaccine rather than the alleles that are required to be the binding targets for those peptides.

Selection of minimal a set of epitopes providing a determined population protection coverage (PPC)
For a set of CTL epitopes, the PPC defines the population proportion that could potentially mount an immune response to any of these epitopes. Given the required HLA I restriction of CTL responses, the PPC for a set of CTL epitopes matches the proportion of the population that exhibits at least one of the HLA I alleles targeted for binding any of those epitopes, and is in turn given by the CPF of the targeted HLA I alleles. Provided a set of CTL epitopes (N CTL ) and their binding profiles (HLA I alleles they bind to), an algorithm for the selection of a minimal subset of epitopes (K) providing a PPC ≥ 95% was implemented as follows: 1. We begin by considering all independent subsets (X) of epitopes in N CTL of size K (number of epitopes), begining with 2.
2. For each independent subset of epitopes, their HLA I binding profiles are pooled into a single non-redundant array of HLA I alleles, and the CPF is computed for 5 major American ethnicities using the HLA I gene and haplotype frequencies reported by Cao et al [17]. Only epitope subsets reaching a CPF ≥ 95% for the five ethnicities are saved and returned by the algorithm.
3. If none of the independent epitope subsets X of size K is found to reach a CPF ≥ 95% the algorithm increases the subset size by 1 and loops over 1 through 3.
4. The algorithm stops cycling and returns the results when at least 4 independent subsets of epitopes of size K reach a CPF ≥ 95%. If only 3 independent sets of size K are found to reach a CPF ≥ 95%, the program will cycle once more and find the subsets of size K + 1 that reach a CPF ≥ 95%, and then exit.

HIV-1 specific CTL epitopes and assembly of peptide pools
The 25 peptides considered in this study consisting of HIV-1 specific CTL epitopes were synthesized by Invitrogen Co. (Carlsbad, CA). HPLC analysis showed that the purity of the synthesized crude peptides was 86-92%. All peptides had expected masses as confirmed by mass spectrometry. Crude peptides were used for initial peptide binding assays and selected crude peptides were purified to >96% by reverse phase HPLC for further experiments. Several peptide pools were obtained by combining an excess number of CTL epitopes providing a PPC ≥ 95%, with peptide pool #1 containing all selected peptides and smaller pools #2-#5 as defined in Results. All peptides pools were prepared in DMSO at 200 μM each.

IFNγ-ELISPOT assay on HIV-1 infected donor cells
CD8 + T-cell responses to pools of HIV-1 epitopes were quantified by gamma interferon (IFN-γ) ELISpot assay as follows. Peripheral blood mononuclear cells (PBMC) isolated from HIV-1 infected patients were plated at 100,000 per well with peptide pools at a final concentration of 10 μM in anti-interferon gamma mAb 1-D1K (Mabtech, Stockholm, Sweden) coated polyvinylidene 96-well plates (Millipore.MA) and processed as previously described [21]. For each individual peptide, the assay was run in triplicate. Negative and positive controls were obtained by incubating individual PBMC with medium alone (negative control) and phytohemagglutinin (PHA) as a positive control for naive patients. For the assessment of general immunocompetence of HIV-1 infected patients, a CEF peptide pool was used as an internal positive control. The CEF pool consists of optimal T cell epitopes for CMV/EBV and Influenza viruses provided by the NIH reagent program. Only HIV-1 patients whose blood yielded positive CEF and/or PHA responses were chosen as subjects for the present study.
The number of specific IFN-γ secreting T cells was determined with an automated ELISPOT reader (AID, Strassberg, Germany), calculated by subtracting the average negative control value and expressed as the number of spot-forming cells (SFC) per 10 6 input cells. Negative controls were approximately 50 SFC per 10 6 input cells for HIV-1 infected patients and around 40 for naïve subjects.
A response was considered positive if there were 50 SFC per 10 6 input cells and the activity was at least three times as great as the mean background activity.

IFNγ -production on HIV-1 specific T cell lines generated from naive donors
HIV-1 peptide pool-specific T cell lines were generated as described above. Two weeks after the second stimulation, 25,000 per well HIV-1 specific T cells were plated into anti-interferon gamma mAB 1-D1K (Mabtech) coated polyvinylidene 96-well plates (Millipore) with 25,000 per well irradiated donor PBMCs and stimulated with 10 μM of each HIV-1 peptide pool. Plates were processed and spot forming cells were calculated as described above.

DimerX staining
A2:Ig fusion protein DimerX (BD Biosciences) was passive loaded with individual A0201 restricted HIV-1B peptides or positive control EBV peptide at 640 molar excess in PBS pH 7.2 overnight. HIV-1B specific CD8 + cells were stained with peptide loaded A2:Ig fusion protein DimerX reagent and detected by PE-conjugated A85-1 mAb (anti-mouse IgG1, BD biosciences). DimerX staining was determined by flow cytometry (FACSAria, BD Biosciences). The negative control straining was obtained using the HIV-1 specific T cell line and a non-relevant HIV-1 peptide loaded DimerX. The positive control was obtained using donor PBMC's stimulated with 1uM BMLF-1 EBV peptide for one week and then stained with EBV peptide loaded Dimer X.

Strategy for selection of HIV-1 specific CTL cell epitopes and generation of peptide pools
In this study, we approached the formulation of CTL targets from a collection of 199 unique HIV-1 CTL epitopes retrieved from the Los Alamos HIV database [14]. The peptides were all 9mers, which is the optimal peptide length for MHCI binding, and are known to elicit CTL responses in one or more HIV-1-infected individuals, indicating that they are naturally processed during the course of infection. Identification of processed epitopes eliminates one hurdle in CTL epitope-based vaccine design. Nonetheless, a successful CTL-based strategy must confront the sequence variability of HIV-1 strains and polymorphism of the human MHCI molecules (HLA I). HIV-1 sequence variation largely contributes to immune evasion. In fact, it has been suggested that at a population level, HIV-1 variants have evolved to adapt to CTL responses [23,24]. Certainly, CTL escape variants can arise from mutations within and/or flanking HIV-1 CTL epitopes [12].
To select target epitopes less likely to escape from CTL, we first subjected representative viruses from all HIV-1 clades (Table 1) to a Shannon Entropy (H) variability analysis [25], thereby removing or "masking out" any residue with H >1. See Additional file 1 for the resulting HIV-1 consensus proteome. All residues in the selected CTL epitopes are therefore quasi-invariant. The most proximal residue flanking the C-terminus of the CTL epitope is a determinant for cleavage by the proteasome, and mutations in that residue can abrogate T cell recognition [26]. Accordingly, we also excluded those CTL epitopes with a C-terminal flanking residue with H >1. As a result, only 37 of the 199 HIV-1 CTL peptides were chosen.
HLA I polymorphisms complicate development of a broadly protective CTL epitope vaccine by limiting population coverage. For a given selection of CTL epitopes, the population protection coverage (PPC) is given by the proportion of cumulative phenotypic frequency (CPF) of the HLA alleles restricting these epitopes. CPF can be computed from the HLA gene and haplotype frequency in the population. Using the HLA I gene and haplotype frequency reported by Cao et al. [17] for five major American ethnicities (Black, Caucasian, Hispanic, Native American, and Asian), we developed an algorithm that first computed the binding of each epitope to 55 HLA I alleles and then selected epitope combinations providing a PPC ≥ 95% for all ethnic groups considered (details in supporting methods). From this analysis, we predicted that as few as five epitopes from the 37 conserved HIV-1 specific CTL epitopes should be recognized by = 95% of the population, regardless of ethnicity. Furthermore, we identified 5epitope combinations using only 25 of the 37 conserved HIV-1 specific CTL epitopes ( Table 2). These 25 CTL epitopes were distributed as follows: POL (reverse transcriptase, integrase and protease), 14 epitopes; GAG, 5 epitopes; ENV, 3 epitopes; and NEF, 3 epitopes. None of the epitopes were found in the five remaining protein open reading frames (ORFs) encoded by the HIV-1 genome (VIF, TAT, REV, VPU/VPX, and VPR). This distribution reflects, in large part, the overall size as well as degree of conservation of the different HIV-1 ORFs. Furthermore, visual inspection of the relevant 3D structures reveals that these epitopes encompass residues that are important either for structural integrity and/or catalytic activity. For example, epitopes TLVDVGDAY and VIYQYMDDL from HIV-1 reverse transcriptase (POL) bear 6 and 4 residues, respectively, involved in substrate binding and catalysis [27]. The penalty for mutations at any of these sites vis-a-vis viral fitness presumably precludes ready development of escape variants.

A minority of HIV-1 patients responds to HIV-1 specific CTL peptide pools
That as few as 5 peptides may provide the requisite PPC largely stems from their ability to bind to many different HLA I molecules. The peptide SPRTLNAWV itself is anticipated to bind to eight distinct HLA I allelic variants, offering a PPC ranging from 35% in Asians to 52% in North American Indians. Such promiscuous MHCI-peptide binding often resides in previously defined HLA I supertypes (i.e. alleles with similar peptide binding specificity [28]), e.g., the peptide SPRTLNAWV binds to the B7 supertype and the peptide AVFIHNFKR to the A3 supertype. Nevertheless, binding promiscuity is not exclusively confined to HLA I supertypes. For instance, the peptide EKEGKISKI is predicted to bind to the alleles B2701, B3801 B39011, B3909, B4402, B5101 and B0801, which do not conform to any known supertype, and peptide LVGPTPVNI is predicted to bind to several alleles within the A2 supertype (A0201 A0202 A0205 A0209) but additionally to other non-related alleles of the HLA-B locus (B2701 and B3801).
In order to assess the accuracy of the predicted PPC, five peptide pools were created, each solubilized in DMSO and combining CTL epitopes (200 μM each) to provide a PPC ≥ 95 % (peptides included in the pools are shown in Table 3). Peptide 1 pool contained all 25 selected peptides whereas peptide pools #2 (15 peptides), #3 (13 peptides), #4 (11 peptides), #5 (7 peptides) comprising 55, 13, 11, 5, 2, respectively, of distinct 5-peptide combinations producing ≥ 95% PPC. Peptides pools 1-5 were designed so that peptides in the smaller pools are contained in the larger pools. CD8+ T cell responses to these pools were first checked by IFNγ ELISPOT assay using PBMC samples from a cohort of 47 HIV-1 infected patients. Patients were largely heterogeneous with regard to their HLA I background and included those diagnosed during acute or chronic HIV-1 with or without HAART therapy and some long-term non-progressors (Table 4). Furthermore, these patients were immunocompetent as judged by their ability to elicit responses to a pool of CMV, EBV and Influenza virus peptides and/or PHA [29]. A summary of the patient antigen-specific CD8 responses is given in Figure 1. Surprisingly, the percentage of patients that exhibit responses to the tested pools was far lower than expected, and ranged from 31% to 45%. There were no significant differences to HIV-1 peptide pools between HIV-1 infected patient groups (p < 0.0001) (see Figure 1A and 1).

T cell responses obtained from all HIV-1 naive donors primed with HIV-1 specific CTL peptide pools
Errors in the prediction of HLA I binding profiles of epitopes may have contributed to the above discrepancies between expected and experimental responses. Nonetheless, prediction of peptide-MHCI binding per se is quite accurate, and in general experimental HLA I binding can be confirmed for at least 75% of HLA I predicted binders [30]. Thus, while we do not discard the possibility that errors in predicting epitopes by HLA I binding profiles might exist, they cannot account for the great divergence between the expected CTL responses and those observed experimentally. More likely, functional impairment of CTL in chronically HIV-1-infected patients may render CTLs unable to produce IFNγ [31,32]. In addition, the specificities of CTL responses change during disease progression [33]. Those targets of CTL responses observed in individuals during the course of any viral infection are Antigen-specific responses to conserved HIV-1 peptide pools in HIV-1-infected and naive individuals Figure 1 Antigen-specific responses to conserved HIV-1 peptide pools in HIV-1-infected and naive individuals. Conserved HIV-1 peptides were collected in five pools [1][2][3][4][5]. CTL responses of 47 immunocompetent HIV-1 infected patients to each peptide pool were analyzed by IFNγ ELISPOT (Panel A). The same HIV-1 peptide pools were used to prime PBMCs from naïve individuals and after two stimulations, similarly tested by IFNγ ELISPOT (Panel B).  Table 1  #2  15  13  FPVRPQVPL, KIQNFRVYY, AVFIHNFKR, GPKVKQWPL, LVGPTPVNI, TLNAWVKVI, GLNKIVRMY,  TLFCASDAK, VIYQYMDDL, PLVKLWYQL, SPRTLNAWV, YVDRFFKTL, YFPDWQNYT,  RAIEAQQHL, FVNTPPLVK  #3  13  11  FPVRPQVPL, KIQNFRVYY, AVFIHNFKR, GPKVKQWPL, LVGPTPVNI, TLNAWVKVI, GLNKIVRMY,  TLFCASDAK, VIYQYMDDL, PLVKLWYQL, SPRTLNAWV, YVDRFFKTL, YFPDWQNYT  #4  11  5  FPVRPQVPL, KIQNFRVYY, AVFIHNFKR, GPKVKQWPL, LVGPTPVNI, TLNAWVKVI, GLNKIVRMY,  TLFCASDAK, VIYQYMDDL, PLVKLWYQL, SPRTLNAWV  #5  7  2 FPVRPQVPL, KIQNFRVYY, AVFIHNFKR, GPKVKQWPL, LVGPTPVNI, TLNAWVKVI, GLNKIVRMY *5-PC: distinct 5-peptide combinations in the peptide pool each providing ≥ 95% PPC determined by subtle immunodominance patterns influenced by many factors including other HLA molecules and previous pathogenic encounters [34]. To address these possibilities, we analyzed responses in HIV-1 naive individuals by priming PBMC of ten donors at high cell density with HIV-1 peptide-pool pulsed autologous Langerhans-like cells followed by restimulation with peptide pulsed fresh PBMCs. We tested the resulting HIV-1 peptide pool specific T cell lines by ELISPOT analysis. ( Figure  1B) All donors generated IFNγ with a magnitude significantly greater (p < 0.0001) than ELISPOT responses generated by HIV-1-infected subjects.

Characterization of A0201 restricted responses in HIV-1 naive individuals
To further evaluate specific T cell responses, we selected the most common HLA subtype, A0201 for detailed analysis. Eight HIV-1 peptides with A0201 restriction ( was used to normalize positive peptide binding (value = 1). A predicted A0301 restricted HIV-1 peptide was used as a negative control (value = 0). Each HIV-1 peptide was found to bind A0201 as shown by increased A0201 surface expression on peptide-pulsed T2 cells (Figure 2A). Subsequently, A0201 restricted HIV-1 peptides were combined in a pool and were used to prime an A0201 donor, resulting in selective proliferation to the HIV-1 peptide pool compared to control ( Figure 2B). These cells produced IFNγ upon stimulation with HIV-1 peptide-pulsed antigen-presenting cells (PBMCs) in ELISPOT assays (Figure 2C). After three rounds of stimulation, 25% of the cells in culture were CD8 + and 62% were CD4 + ( Figure  2D). Given that ~20-30% of MHCI binding peptides are predicted to bind to MHCII as well, this result is not unexpected (data not shown). We also generated two control cell lines: the first, a tetanus toxoid specific T cell line with 78% CD4 + and no CD8 + T cells elicited by whole protein stimulation; and the second, an EBV specific T cell line with 62% CD8 + and 29% CD4 + cells elicited by stimulation with an immunodominant EBV epitope BMLF1 (GLCTLVAML). CD8 + T cells were sorted from the HIV-1 Generation and characterization of HLA-A0201 restricted HIV-1 peptide-specific T cell lines from uninfected individuals In Panel E, the negative control straining was obtained using the HIV-1 specific T cell line and a non-relevant peptide loaded DimerX. The EBV positive control was obtained using donor PBMC's stimulated with 1uM BMLF-1 EBV peptide for one week and then stained with EBV peptide loaded Dimer X. One to nine percent of CD8 T cells were specific for each one of the selected HIV-1B peptides (Panel F). (Fig.  2E, first plot) and their peptide specificities determined by individual peptide-loaded DimerX staining ( Figure 2E). One to nine percent of CD8 + T cells were specific for each one of the selected HIV-1B peptides ( Figure 2F).

Cytotoxic activity of HIV-1 peptide-specific CD8+ T cells derived from HIV-1 naive donors
Next, to test for potential cytolytic activity of the HIV-1 peptide-specific T cell lines, T2 hybridoma cells were pulsed with A0201 restricted HIV-1B peptide pools and specific lysis was determined by standard 51 Cr release assay. As shown in Figure 3A, the HIV-1 peptide-specific CD8 + T cells killed A0201 restricted HIV-1 peptide poolpulsed T2 cells with ~50% specific lysis at an E:T ratio of 30:1. The same T2 cells loaded with HTLV-TAX peptide as a negative control were not lysed. To determine whether HIV-1 peptide-specific A0201 restricted CTL can recognize and lyse HIV-1 -infected cells, we infected the T1 hybridoma line (A0201+, B5101 + , CD4 + ) with HIV-IIIB virus and then determined whether HIV-1 peptide specific A0201 restricted CTL can recognize HIV-IIIB infected T1 cells and kill target cells in a 51 Cr release assay. HIV-1 specific CD8 T cells from two different A0201 donors recognized and killed 20-28% of HIV-1 infected T1 cells 2 and 3 days after infection at an E:T ratio of 30:1 while uninfected T1 cells were not lysed ( Figures 3B and 3C). We also noted that although some of these peptides (for example, TLNAWYKVI) bind to related alleles, CD8 + T cells stimulated from the A0201* donor fail to specifically lyse peptide-pulsed A0202, A0203 or A0204 targets ( Figure 4). Thus, a given CTL response is less degenerate than peptide binding to MHC.
As shown in Figure 4, A0201 restricted HIV-1-specific CTL demonstrate alloreactivity and kill A0202, A0203 and A0204 transfected 721-221 targets in the absence of HIV-1 peptide loading. Moreover, in the presence of HIV-1 peptides, alloreactivity against A0202, A0203 and A0204 transfected 721-221 targets decrease as shown by diminished lysis. These results suggest that A0201 restricted HIV-1 peptides bind to A0202, A0203 and A0204, replacing possible endogenous peptides that mediate much of the evident alloreactivity. The analysis of the A0204 transfectant is a clear example of this phenomenon.
Allele-restricted fine specificity of CTL directed at HIV-1 peptides Figure 4 Allele-restricted fine specificity of CTL directed at HIV-1 peptides. HIV-1 peptide-specific T cells were generated from an A0201 donor using predicted A0201-restricted HIV-1 peptides (Table 1 and Fig. 1). HIV-1 peptide-specific T cells killed the A0201 transfected 721-221 cell line in a peptide-specific manner by 51 Cr release assay. To test possible cross-reactivity of A0201 restricted HIV-1 peptide-specific CTL to other A02 alleles, we utilized A0202, A0203 and A0204 transfected 721-221 cells as targets. A0201 restricted HIV-1 peptide specific CTLs killed HIV-1 peptide loaded A0201 transfected 721-221 targets at a 10:1 ratio. +pp indicates addition of peptide pool, with solid bar = lysis upon +pp and gray bar = lysis with no peptide pool addition.
Cytotoxic activity of HIV-1 peptide-specific A0201 restricted T cells from normal donors

Conclusion
The present report supports the notion that T cell responses in HIV patients are impaired and argues for optimizing the composition of epitope-based vaccines using data obtained from naive individuals. However, the responses present in HIV-1-infected subjects and the ability to grow peptide-specific CD8 T cells from naive individuals are not directly comparable. Further experimental work shall include analysis of the T cell responses in naive and HIV-1-infected patients using in vitro stimulated CD8 T cells and HIV-specific CTL epitopes. Perhaps this approach will elucidate whether clonal deletion or anergy exists, for example.
Overall, these findings demonstrate that normal, HIV-1 naive individuals possess T cells capable of recognizing and expanding against conserved segments of the HIV-1 proteome. Although only a minority of chronically HIV-1infected patients responds to pools of such epitopes, viral protective responses can be elicited from all normal subjects examined. Thus, the absent response in HIV-1infected patients is not a consequence of intrinsic holes within the human T cell repertoire, but rather the sequelae of chronic exposure to antigens, immunodominance patterns and/or cellular dysfunction. In this regard, the ability of HIV-1 to drive cellular responses away from relevant, invariant segments capable of affording protection toward mutable epitopes that escape CTL-based destruction is reminiscent of the anti-HIV-1 antibody response in chronically infected patients [4]. In the latter, antibodies are directed to the variable loops of the HIV-1-gp160 envelope protein rather than invariant potential neutralization sites. Thus, viral escape mechanisms send both T cells and B cells off in unproductive directions. That human T cell responses can be elicited from virusnaive individuals to conserved segments of the HIV-1 proteome argues that CTL-directed vaccines may target these epitopes, and underscores the importance of assessing immune responses of normal individuals as well as HIV-1-infected patients. Beyond HIV-1, immunoprotection against other chronic infectious diseases such as malaria as well as neoplastic disorders may benefit from a paradigm shift in vaccine design, focusing on potential immune responses in normal rather than in patient populations.