Screening for HIV-specific T-cell responses using overlapping 15-mer peptide pools or optimized epitopes

The IFN-γ enzyme-linked immunospot (ELISpot) assay is often used to map HIV-specific CD8 T-cell responses. We compared overlapping 15-mer pools with optimized CD8 epitopes to screen ELISpot responses in HIV-infected individuals. The 15-mer pools detected responses to previously undefined epitopes, but often missed low-level responses to predefined epitopes, particularly when the epitope was central in the 15-mer, rather than at the N-terminus or C-terminus. These factors should be considered in the monitoring of HIV vaccine trials. HIV-specific CD8 T lymphocytes are important in the host control of HIV replication. There is a clear temporal relationship between the development of HIV-specific CD8 T cells and viral control during acute HIV infection [1], and between experimental CD8 T-cell depletion and increases in viral load [2]. In addition, epitope-specific CD8 T cells place considerable evolutionary pressure on the virus in both primates and humans [3–8]. Various HIV-specific immune responses have been described in highly exposed, persistently seronegative (HEPS) populations [8,9], most commonly HIV-specific CD8 cells [10]. Moreover, the loss of peripheral HIV-specific CD8 effector responses in the absence of HIV exposure has been associated with the seroconversion of HEPS individuals [11]. These findings, among others, have led to the development of HIV vaccines that are designed to induce a virus-specific cellular immune response [12,13]. As HIV vaccines move into human trials, one issue faced is the choice of assay to monitor CD8 T-cell responses [14]. CD8 response frequencies in both vaccinated and HEPS individuals are approximately 10-fold lower than in HIV-infected individuals [15,16], and may target novel virus epitopes [17,18]. The IFN-γ enzyme-linked immunospot (ELISpot) assay has been used successfully to screen responses in HIV-infected [19], HEPS [20] and vaccinated [16] cohorts. The HIV antigen(s) used to elicit responses in this system have most commonly been HIV-vaccinia constructs [21,22], overlapping peptide pools spanning an entire HIV-1 gene or the entire HIV-1 genome [23], or optimized CD8 epitope libraries [20]. Optimized epitopes can be pooled to screen responses without losing sensitivity [24], and the sensitivity of overlapping 15-mers and 20mers is similar [25]. However, as the detection of low frequency CD8 responses is likely to be important in vaccine studies, we compared the use of individual optimized HIV CD8 epitopes with overlapping 15-mer peptides. HIV-1 Gag-specific responses were screened in 49 HIV-1-infected individuals at various clinical stages, and in 11 low-risk control subjects. Molecular HLA class I genotyping was performed for all subjects using polymerase chain reaction sequence-specific primers [26], and IFN-γ ELISpot assays were set up using peripheral blood mononuclear cells (PBMC) at 1 × 105 per well, with a final peptide concentration of 10 μM, as previously described [15,27]. A total of 122 15-mer peptides overlapping by 11 amino acids and spanning HIV-1 Gag (obtained through the AIDS Research and Reference Reagent Program, Division of AIDS, National Institute of Allergy and Infectious Diseases, National Institutes of Health) were set up in a matrix of 23 pools, so that each peptide appeared uniquely in two separate pools, and all responses detected using the 15-mer matrix were confirmed using individual 15-mer peptides. In addition, a panel of all 58 HIV-1 Gag cytotoxic T lymphocyte epitopes listed in the Los Alamos Immunology Database [28] was synthesized, and screening was performed against all appropriate class I HLA-matched epitopes. Plates were counted using an automated ELISpot counter (Autoimmun Diagnostika GmbH, Strassberg, Germany). The criteria for a positive assay were: (i) HIV-specific response 50 or more spot-forming units (SFU)/million PBMC; (ii) peptide response two or more times background (media alone); and (iii) a response to the positive control (phytohemagglutinin; Murex Biotech Ltd., Dartford, Kent, UK). CD8 T-cell response frequencies were further classified as weak (< 200 SFU/106 PBMC), intermediate (200–1000 SFU/106 PBMC) or strong (> 1000 SFU/106 PBMC). No positive responses, either to the overlapping peptides or to the optimized epitopes, were seen in 11 low-risk, HIV-uninfected control subjects. The 49 HIV-1-infected subjects made 102 CD8 T-cell responses to HIV-1 Gag, with at least one response seen in 48 out of 49 (98%) individuals. Of the 102 IFN-γ CD8 T-cell responses detected with the HIV-1 Gag matrix, 15 (14.7%) were to 15-mer peptides that did not contain a CD8 epitope previously known to be restricted by any of the subjects’ class I HLA alleles. These novel responses spanned 12 immunogenic regions of Gag, and fine mapping of these novel epitopes is currently underway. The sole use of a predefined epitope panel would therefore have missed a substantial number of CD8 T-cell responses. The predefined epitope panel detected 87 responses in 49 study subjects. A matching response was seen to the corresponding 15-mer in 70 out of 87 (81.5%), with no corresponding response in 17 out of 87 (19.5%). Weak responses (< 200 SFU/106) to optimized epitopes were more likely to be missed by the 15-mer matrix than stronger (≥ 200 SFU/106) responses (12/21 versus 58/66; likelihood ratio 8.5; P = 0.002). There was a stepwise association between the strength of the CD8 T-cell response to an optimized epitope and the probability of detecting this response with the 15-mer peptide matrix: 28 out of 30 (93.3%) strong responses were detected using the 15-mer pools, 30 out of 36 (83.3%) intermediate responses, and 12 out of 21 (57.1%) weak responses (likelihood ratio 10.1; P = 0.006; Fig. 1a). Predefined epitope responses were retested for 43 epitopes in which cryopreserved PBMC were available, and were confirmed in 42 out of 43 cases (98%). In one case (2%), the epitope response could not be confirmed; this response had been weak, and was not detected using the 15-mer matrix.Fig. 1. Screening for CD8 T-cell responses with overlapping 15-mer pools misses low-level responses detected by optimized CD8 epitopes.: (a) Solid black bars on the X-axis represent IFN-γ responses against optimized HIV epitope peptides, divided into ‘weak’ [< 200 spot-forming units (SFU)/million peripheral blood mononuclear cells (PBMC)], ‘intermediate’ (200–1000 SFU/million PBMC) and ‘strong’ (> 1000 SFU/million PBMC). The Y-axis shows the percentage of these responses that were also detected using pools of overlapping 15-mer HIV peptides. (b) Representative peptide titration curve for HIV-specific IFN-γ responses in an HIV-infected participant. The antigen used was either an optimized HIV epitope peptide (solid line), or the 15-mer peptide containing this epitope (dotted line). —•— TPQDLNTML; - - -♦- - - EGATPQDLNTMLNTV.Serial fivefold peptide titrations were performed for five optimized epitope 15-mer pairs, using cryopreserved PBMC samples (Fig. 1b; representative example). When responses to both the optimized epitope and the 15-mer peptide were detected at a given peptide concentration, responses to the former were stronger in all cases. More importantly, the threshold concentration for response detection was consistently lower for the optimized epitope than for the 15-mer (mean 1.4 fivefold dilutions lower; range 0–3 dilutions). In one subject, the response to an optimized epitope (QASQEVKNW) could be detected down to a concentration of 0.2 μg/ml, whereas no responses to the 15-mer (RAEQASQEVKNWMTE) could be detected below 25 μg/ml. It was hypothesized that the position of an optimized epitope within the 15-mer itself might affect the assay sensitivity, because issues of steric hindrance and amino acid charge might be more important for centrally located epitopes than for those located at the C or N-terminus. To test this hypothesis, we analysed responses in which the same optimized epitope was contained within two different 15-mers, centrally in one and terminally in another (N = 27 responses). Responses to the 15-mer with the epitope in a central position tended to be weaker than those with the same epitope in a terminal position (mean response 598 versus 740 SFU/million PBMC; P = 0.1, paired-samples t-test), and were more likely to miss an optimized epitope response altogether (11/27 versus 5/27 responses missed; P = 0.001). In summary, screening for HIV-1-specific CD8 T-cell responses in an HIV-1-infected cohort using a matrix of overlapping 15-mer peptide pools detected several novel responses that would have been missed using a panel of predefined CD8 epitopes, confirming that this will be a useful technique for mapping CD8 responses in HIV-1-infected individuals and HIV-1 vaccine recipients. However, overlapping 15-mers were less sensitive than optimized CD8 epitopes in detecting low frequency CD8 responses (< 200 SFU/million PBMC), especially when the epitope was located centrally in the 15-mer peptide. As CD8 responses in exposed seronegative and vaccinated cohorts are up to 10-fold lower in frequency than those in HIV-infected individuals [15,16], the merits of each approach should be considered carefully when planning and monitoring HIV-1 vaccines trials. Acknowledgements The authors would like to thank Kati di Gleria for peptide synthesis. Overlapping HIV-1 peptide pools were obtained through the AIDS Research and Reference Reagent Program, Division of AIDS, National Institute of Allergy and Infectious Diseases, National Institutes of Health. Sponsorship: This work was supported by grants from the UK Medical Research Council (S.R.J., T.D.); the Elizabeth Glaser Pediatric AIDS Foundation (S.R.J.); the Canadian Institutes of Health Research (F.P.); and the Connaught Foundation of the University of Toronto (R.K.). S.R.J. is an Elizabeth Glaser Scientist of the Pediatric AIDS Foundation; R.K. and F.P. are supported by the Canadian Research Chair Programme. These data were presented in part at the International Meeting of the Institute of Human Virology, Baltimore, USA, 2003.

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.