HIV-1 MN Env 15-mer peptides better detect HIV-1 specific CD8 T cell responses compared with consensus subtypes B and M group 15-mer peptides

Objective: To compare the ability of three Env (15-mer) peptide sets derived from the HIV-1 MN, the subtype B consensus, and the group M consensus to detect HIV-1 specific interferon (IFN)-γ responses in HIV-1 subtype B infected subjects.

Methods: Peripheral blood mononuclear cells were obtained from 17 HIV-1 subtype B seropositive and 5 HIV-1 seronegative subjects. Peptide matrices comprising each peptide set were used in IFN-γ Elispot assays to screen for T cell epitopes. Following matrix deconvolution, individual peptides were analyzed by IFN-γ intracellular cytokine-staining to confirm and characterize the responding cells.

Results: HIV specific IFN-γ responses were detected in 17 of 17 HIV-1 seropositive and none of 5 HIV-1 seronegative subjects by Elispot. Within the 17 HIV-1 seropositives, 16, 14, and 11 subjects responded to MN, B consensus, and group M env peptides, respectively. Responses were confirmed by intracellular cytokine analysis in 14 subjects and were in the CD3CD8 compartment. Cross-recognition of 'equivalent' peptides (i.e., peptides mapping to the same sequence region from the three peptide sets) was observed in 9 of 17 subjects. Peptide set specific responses to individual peptides were also observed; 11, 1, and 1 subjects demonstrated peptide set specific responses to MN, B consensus, and consensus group M, respectively.

Conclusion: MN derived Env peptides were better able to detect HIV-1 specific CD8 T cell responses, many of which were not detectable by the equivalent clade or group consensus peptides. No single peptide set detected all the IFN-γ responses within an individual. These results demonstrate the importance of reagent selection for monitoring of HIV responses in HIV-1 infected individuals and subsequently vaccine recipients.

Introduction

Viral strains from the same HIV-1 clades can differ by 5-25% of amino acids, depending on the HIV-1 protein ^[1-3] (Los Alamos National Laboratory, http://www.lanl.gov). This diversity poses a major obstacle to the rational design of HIV vaccines and to selection of appropriate reagents for detecting immune responses in HIV infected individuals and vaccine recipients ^[4-6].

A critical component of controlling HIV infection is mediated by CD8 T lymphocytes through MHC directed cytolysis and cytokine production. Accurate evaluation of vaccine strategies against HIV and other viruses has promoted the development of various new methods to detect and quantify HIV specific CD8 T cells. Several studies have evaluated and compared different methodologies, such as the Elispot assay and intracellular cytokine analysis, and peptide matrices have been frequently used for the comprehensive assessment of HIV-1 specific T cell responses ^[5,6]. Matrix sets consisting of peptides of different lengths and overlap have been used to compare breadth and magnitude of T cell responses ^[6-8]. They are highly sensitive, specific, and specimen sparing, making them an attractive tool for large-scale clinical or vaccine trials. Matrix sets spanning the entire HIV-1 genome or specific proteins have been used to characterize HIV-1 specific T cell responses in HIV-1 infected individuals ^[7,9-11]. HIV specific T cell responses detected by autologous versus consensus viral sequences have also been explored ^[9,12,13].

Intra- and interclade sequence variability of HIV-1 may pose a major challenge for the accurate characterization of HIV-1 specific T cell responses. In an attempt to overcome this problem, centralized HIV-1 sequences have been proposed for HIV-1 immunogen design. Using computational methods to minimize sequence diversity between virus isolates, consensus, most recent common ancestor and center of the tree sequences have been generated ^[14-16]. Consensus sequences are based on the current sequence database for circulating strains and represent the most frequent amino acid residue across an alignment of homologous sequences. Single subtype consensus and M-group consensus (the consensus of consensus sequences for subtypes A, B, C, D, F, G, H) have also been constructed ^[17]. In another approach, the most recent common ancestor, an ancestral sequence reconstructed on the basis of an evolutionary model, and center of the tree in an unrooted phylogeny have been derived, both try to minimize the distance from the 'center' to each tip on the tree.

The objective of this study was to compare the ability of different peptide sets to detect HIV-1 specific T cell immune responses in HIV-1 infected individuals. Peptide sets based on a specific strain, on a subtype consensus sequence, and on the group M consensus were used. Complete sets of 15-mer peptides overlapping by 11 amino acids, derived from the Env sequences of the HIV-1 subtype B (MN) isolate, the HIV-1 consensus subtype B (Con B), and the HIV-1 consensus M group (Con M) were used to monitor HIV-1 specific interferon (IFN)-γ responses in a cohort of HIV-1 subtype B infected individuals.

Materials and methods

Subjects

Seventeen consenting adults were recruited under an institutional review board approved protocol (RV149) conducted within the United States Military HIV Research Program (USMHRP), to obtain leukapheresis samples from healthy HIV-1 positive asymptomatic anonymous blood donors. Leukapheresis samples were also obtained from five anonymous HIV-1 seronegative donors kindly supplied through the NIH Blood Donor Center (Bethesda, Maryland, USA) and BRT laboratories (Baltimore, Maryland, USA). Peripheral blood mononuclear cells (PBMC) were isolated by Ficol-Hypaque (Pharmacia Biotech, New Jersey, USA) gradient centrifugation and cryopreserved in liquid nitrogen.

HIV-1 subtype confirmation, plasma viral load, CD4 cell count and HLA typing

HIV Viral RNA was purified from plasma using the QIAamp HIV Viral RNA Extraction kit (Qiagen Inc., Valencia, California, USA). The HIV sequence for the protease and the reverse transcriptase gene regions was obtained using the TRUGENE HIV-1 Genotype Assay (Bayer Health Care Diagnostics, Norwood, Massachusetts, USA). The HIV subtype was determined by phylogenetic analysis of the HIV sequence using neighbor joining, maximum likelihood and distance analyses (Wisconsin Package Version 10, Genetics Computer Group, Madison, Wisconsin, USA). HIV viral RNA load was determined on RNA extracted from plasma using the Amplicor HIV-1 Monitor test, Version 1.5 (Roche Molecular Systems, Branchburg, New Jersey, USA). CD4 cell counts were determined using the multiTEST kit (Becton Dickinson, New Jersey, USA) according to the manufacturers protocol. HLA typing was performed on all subjects using DNA extracted from either PBMC or autologous B lymphoblastoid cell lines at the University of Alabama, Birmingham. Briefly, high molecular weight genomic DNA was extracted from immortalized B cells or PBMC using the QIAamp Blood Kit and protocols recommended by the manufacturer (Qiagen Inc.). The four-digit alleles at the HLA-A and -B loci were resolved by automated reference-strand conformation analyses of PCR amplicons corresponding to exon 2, intron 2, and exon 3 sequences (Pel-Freez Clinical Systems, Brown Deer, Wisconsin, USA). The two-digit HLA-C specificities were defined separately by PCR with sequence specific primers (Pel-Freez Clinical Systems).

Peptides

Complete overlapping synthetic peptide sets were obtained from the NIH AIDS Reagent and Reference Reagent Program. Peptide sets consisted of: 212 peptides 15 amino acids in length overlapping by 11 amino acids spanning the HIV-1 subtype B MN isolate Env sequence (catalogue number 6451); 211 peptides 15 amino acids in length overlapping by 11 amino acids spanning the HIV-1 consensus subtype B env sequence (catalogue number 9480); and 211 peptides 15 amino acids in length overlapping by 11 amino acids, spanning the HIV-1 consensus M group env sequence (catalogue number 9487). Peptide purity was greater than 80%, and all were reconstituted in dimethyl sulfoxide prior to storage. Matrices for each peptide set were prepared by creating 15 linear pools of 15 peptides each and 14 pools of every 15th peptide in the series. The genetic distances between the three HIV-1 envelope genes were calculated by DNADIST using the Kimura two-parameter algorithm. The average genetic distance for all pair-wise comparisons was 10.8% between MN and consensus subtype B Env sequences; 12.1% between consensus B and consensus M Env sequences and 19.4% between MN isolate and consensus M Env sequences, respectively.

Elispot assays

All assays were performed using complete medium (RPMI 1640 medium; Quality Biological, Maryland, USA) containing 10% fetal bovine serum (Gemini, California, USA), penicillin (100 IU/ml) and streptomycin (100 μg/ml; Quality Biological, USA). PBMC were recovered from liquid nitrogen, thawed and washed twice using complete media containing 50 U/ml Benzonase nuclease (Novagen, Madison, Wisconsin, USA), resuspended in complete medium and rested overnight at 37°C and 5% CO₂. The PBMC were recounted prior to addition to the Elispot plates at 2 × 10⁵ cells per well. The Elispot assay was performed as previously described ^[10]. Peptides were added to wells at a final concentration of 2 μg/ml peptide in complete medium. Negative controls consisted of six wells of PBMC in complete medium only. To test for functional integrity of the cells, staphylococcal enterotoxin B (SEB) (Sigma, St. Louis, Missouri, USA) was added to triplicate wells at 5 μg/ml as a positive control.

Elispot assay analysis

Elispot plates were examined under a stereomicroscope, and spots were evaluated with an Automated Elispot Reader System with KS 4.3 software (Carl Zeiss, Thornwood, New York, USA). Background was determined as the mean response in the six PBMC-only wells. A positive response was defined as greater than 50 spot forming units per 1 × 10⁶ PBMC and at least twice background. Peptides within pools that induced positive responses were then analyzed individually by IFN-γ intracellular cytokine-staining to confirm and characterize responses.

Intracellular cytokine analysis

PBMC (1 × 10⁶, thawed as above) were stimulated with individual peptides (2 μg/ml) for 6 h at 37°C/5%CO₂ in a 96-well polypropylene plate. Cells incubated in complete media served as negative controls. As a positive control PBMC were stimulated with 5 μg/ml SEB. PBMC were co-stimulated with 1 μg/ml anti-CD28/CD49d (BD Biosciences, USA). Brefeldin A (Sigma Aldrich, St Louis, Missouri, USA) was added at a concentration of 10 μg/ml to block the release of cytokine from the Golgi. Following incubation, PBMC were washed with 0.1% NaN₃ and 0.5% bovine serum albumin in phosphate-buffered saline (flow buffer), fixed with 2% formaldehyde and permeabilized with 0.5% saponin (Sigma) in flow buffer. Cells were then stained for cytometry with fluorescein isothiocyanate-conjugated anti-IFN-γ, phycoerythrin-conjugated anti-CD8, peridinin chlorophyll protein Cy5.5-conjugated anti-CD4 and allophycocyanin-conjugated anti-CD3 (Becton Dickinson) for 45-60 min and washed three times with flow buffer prior to acquisition. A minimum of 1 × 10⁵ events were collected in the lymphocyte gate on the Becton Dickinson FACSCalibur. From this population, CD3CD4 and CD3CD8 populations were identified and analyzed individually for cytokine production. Analysis was done using the Flowjo software (TreeStar, San Carlos, California, USA).

Results

Participant characteristics

Clinical and demographic data of the 17 HIV-1 infected study participants are shown in Table 1. Subjects were between 29 and 50 years of age (mean, 39 years), 16 males and 1 female. Duration of infection ranged from 1 to 17 years. All participants were infected with an HIV-1 subtype B virus strain by partial pol sequencing (data not shown). The CD4 cell count ranged from 238 × 10⁶ to 800 × 10⁶/l (mean, 524 × 10⁶ cells/l) and HIV-1 RNA viral load ranged from 783 to 86 174 copies/ml (mean, 25 438 copies/ml). Three of the 17 volunteers had received highly active antiretroviral therapy. A broad range of HLA alleles were present within this study group; 16 different HLA-A, 21 HLA-B and 15 HLA-C alleles.

Table 1

T cell immune responses

Elispot assays

All 22 individuals (17 HIV seropositive and 5 HIV seronegative subjects) had a positive response to SEB of > 200 spot forming units/1 × 10⁶ PBMC. All 17 of the HIV seropositive subjects had positive IFN-γ responses to at least one of the three peptide sets. None of the HIV seronegative donors responded to any HIV-1 peptide (data not shown). Of the HIV positive subjects, 16 of 17 responded to the MN pools, 14 of 17 to the HIV-1 consensus B pools and 11 of 17 to consensus M group pools. There was no statistically significant difference between the number of responders to each of the three Env peptide sets (P = 0.08; Fisher's Exact Test). A minimum of 1 peptide, and up to 17 individual peptides, were selected per subject for further analysis by flow cytometry. Overall 146, 77, and 60 peptides from MN, Con B, and Con M group peptide sets respectively, were analyzed by flow cytometry.

Intracellular cytokine analysis

We first established a cut-off for the intracellular cytokine assay. The background responses in the negative controls for the CD8 IFNγ-positive population ranged from 0.006 to 0.060% (mean, 0.021%), The 99% confidence interval (CI) for CD8 IFNγ-positive background responses was 0.051%. The background responses for the CD4 IFNγ-positive population ranged from 0.006 to 0.042% (mean, 0.016%), The 99% CI for CD4 IFNγ-positive background responses was 0.047%. The 99% CI were used as the cut-offs to determine positive responses. Responses to SEB ranged from 1.4 to 36% for the CD8 IFNγ-positive populations and from 2 to 44% for the CD4 IFNγ-positive populations. HIV specific CD8 IFNγ-positive responses were confirmed in 14 of 17 subjects. For three subjects, epitopes identified in the Elispot assay were below the limit of detection by the intracellular cytokine assay. Using the defined cut-off, HIV specific CD4 IFNγ-positive responses were not detected in any of the subjects studied with the panel of reagents that were used.

Responses towards individual HIV-1 15-mer Env peptides

A total of 32 distinct responses were detected using the three-peptide sets (Fig. 1). The majority of responses were towards the MN peptide set (a total of 25 CD3CD8 IFNγ responses were detected in 13 subjects), of which 17 were MN specific. Fourteen responses were detected with peptides from Con B, seven of which were also detected with peptides from Con M. There were only two responses that were detected by all three peptide sets; these were to the Con B peptides (837-851) and (581-595) and to the corresponding peptides from MN and con M (Table 2). Five responses were detected using both Con B and Con M peptide sets. One response was unique to the Con B peptide set and another was unique to the Con M group peptide set; these seven responses were not detected using the MN peptide set. The magnitude of the positive responses ranged from 0.097 to 1.18% (median, 0.28%) for the MN set, 0.065-1.64% (median, 0.29%) for the Con B set, and 0.066-1.39% (median, 0.71%) for the Con M set (Fig. 2). No one peptide set detected all 32 responses; however, the MN set detected a larger number of responses than either of the two Env consensus sets.

Fig. 1

Table 2

Fig. 2

Of 17 subjects, 9 demonstrated a cross-recognition of at least two of the peptide sets (Table 2). Five subjects had responses to equivalent peptides derived from Con B and Con M group sequences but not to the MN pool (Table 2). In one of five of the reactive peptides, the amino acid peptide sequences matched exactly. In four of five peptides the sequence of the equivalent peptides did not match exactly. For the equivalent MN peptides that did not induce IFNγ T cell responses, residue substitutions ranged from one to six amino acids. Five subjects had responses to equivalent peptides from Con B and MN (Table 2). In three of the reactive peptides, the sequences matched exactly, while in the remaining three the sequence of the equivalent peptides differed by at least one amino acid. The equivalent Con M group peptides that did not induce IFNγ T cell responses differed by at least one and often more amino acid residues. For the two subjects who had responses to equivalent peptides derived from all three peptide sets (Table 2) the equivalent peptide sequences did not match exactly for the three peptide sets.

Of 17 subjects, 11 demonstrated peptide set specific responses (Table 3). All of these individuals had a peptide set specific IFNγ CD8 T cell response to at least one MN 15-mer peptide (Table 3). The number of responses ranged from one to four individual peptides per subject. Equivalent 15-mer Env peptides from Con B and Con M group did not induce detectable IFNγ responses. In all cases, these 'equivalent' peptides did not match the MN isolate exactly. Two individuals (1000 and 1001), in addition to their peptide specific responses towards the MN set, also demonstrated peptide set specific responses to Con B (Table 3) and Con M (Table 3), respectively. Here again, the equivalent peptides that did not induce IFNγ responses did not match exactly. Overall, the majority of responses observed, whether peptide set specific or cross-reactive, were detected when using 15-mer peptides derived from the MN isolate sequence.

Table 3

When equivalent peptides from Con B and Con M were not recognized, they did not match the MN isolate exactly, and in fact the use of con M alone would have missed about 75% (24/32) of responses. Con B alone would have missed 56% (18/32) of responses, whereas the MN peptide set alone would miss 22% (7/32) of these responses (Fig. 1). Using a combination of the MN isolate peptide set with either one of the consensus peptide sets, 97% (31/32) of responses could be detected.

Discussion

A successful HIV-1 vaccine will induce broad, long lasting protective immune responses. Many studies characterized HIV-1 immune responses in seropositive individuals to determine possible correlates of protection ^[4]. Overcoming HIV-1 diversity, particularly in envelope, has been a key problem and has influenced the development of reagents. We compared three Env 15-mer peptide sets as reagents to monitor HIV-1 specific T cell responses in HIV-1 subtype B infected individuals. We observed that peptides from one selected subtype B isolate were better able to detect HIV-1 specific responses than were peptides derived from con B and con M groups. Consensus sets underestimated the breadth of responses within this study group. The MN isolate peptide set, or by analogy, any strain that is representative of known circulating viruses, may be a more sensitive reagent than either of the two consensus peptide sets. These results are in line with previous studies where consensus sequences had limited potential to detect HIV-1 specific responses, particularly within the more variable regions of the genome such as envelope, compared to peptides based on an autologous sequences ^[9].

Nevertheless, none of the three peptide sets used was, by itself, able to detect all of the IFNγ responses within any given subject. Single or multiple amino acid changes within the peptides, or the position of the epitopes within the peptides, may have impacted the presentation and/or recognition of potential epitopes. A number of groups recently suggested that the position of the optimal epitope within a peptide is crucial for the success with which T cell responses are detected ^[18,19]. The results we present here, where the alignment at the C terminus did not always match exactly between equivalent peptides from different peptide sets, may support this observation.

Viral sequence variability influences the cellular immune responses generated in an infected individual ^[20]. Single amino acid changes within an epitope have the ability to influence the intracellular processing of the epitope ^[21]. Studies investigating HIV-1 specific T cell responses have used several approaches, which include the use of single peptides ^[22,23] and selections of HIV proteins ^[24]. Several studies have demonstrated that even single amino acid substitutions within T cell epitopes can reduce or abrogate binding to HLA and/or recognition by T cells ^[25-28]. Our results corroborate these findings, particularly when the peptide set specific responses are taken into account. For all of the MN specific responses that were observed, the equivalent con B and con M group peptides did not match the MN peptide exactly and the response was abrogated. Indeed, it was often the variability introduced by derivation of a consensus sequence that interfered with recognition, providing inferior results to a single non-autologous strain. These results have implications for designing panels of overlapping peptides for monitoring HIV-1 specific responses in HIV infected individuals and vaccinees. The HIV-1 subtype B MN isolate may be nearer to the circulating strains than the artificially designed consensus sequences and this may explain its ability to detect a larger proportion of the responses. A selected circulating isolate may be the optimal method for measuring T cell responses, at least in individuals infected within a given subtype. We also note that a combination of selected strain and consensus peptide sets could detect more than 95% of responses in these subjects, which could provide a rationale for similar strategies in future studies. It is possible that the autologous viruses would have detected even more responses, though the use of autologous viruses as reagents is impracticable for cohort studies.

Measurement of T cell responses to HIV-1 continues to present a challenge, but the relationships between peptide reagent sets described here must be taken in context. Less variable proteins of HIV-1, such as Gag and Pol, also incorporated into many candidate vaccines, may permit effective use of consensus sequences, a strategy that seems problematic with the highly variable env gene. Derivation of regional, rather than global, consensus sequences within subtypes could potentially provide benefit as well. Finally, for vaccine trials, it should be possible to derive peptide sets from the actual vaccine strains, abrogating at least one source of variability. Mapping of responses in seropositive individuals may be a particularly challenging element in the overall process of HIV-1 vaccine development.

Acknowledgements

The authors thank the volunteer blood donors without whom this study would not be possible and the clinical coordinator Hannah Flaks, RN for clinical coordination and recruitment of volunteers. We also wish thank the US Military HIV Research Program Dendritic Cell laboratory and Cellular Immunology laboratory for technical assistance.

Sponsorship: Supported by the Department of Defense Collaborative agreement DAMD17-98-8007.

Disclaimer: The views and opinions expressed herein do not necessarily reflect those of the US Army or the Department of Defence.