Identification of cryptic MHC I-restricted epitopes encoded by HIV-1 alternative reading frames

Abstract

Human immunodeficiency virus (HIV) 1 major histocompatibility complex (MHC) I-restricted epitopes are widely believed to be derived from viral proteins encoded by primary open reading frames. However, the HIV-1 genome contains alternative reading frames (ARFs) potentially encoding small polypeptides. We have identified a panel of epitopes encoded by ARFs within the gag, pol, and env genes. The corresponding epitopic peptides were immunogenic in mice humanized for MHC-I molecules. In addition, cytotoxic T lymphocytes recognizing these epitopes were found in HIV-infected patients. These results reveal the existence of atypical mechanisms of HIV-1 epitope generation. They indicate that the repertoire of epitopes recognized by the cellular anti-HIV-1 immune response is broader than initially thought. This should be taken into account when designing vaccine strategies aimed at activating these responses.

Figure 1.

Identification of an antigenic epitope (Q9VF) derived from a nontraditional HIV-1 reading frame. (A) Nucleotide and amino acid sequences of a potential alternative reading frame (ARF) protein encoded within the gag gene. Start and stop codons are in bold, and the Kozak sequence is underlined. Nucleotide numbering is according to HIV_LAI sequence. The predicted amino acid sequence of the corresponding polypeptide (Gag-ARFP) is shown, and the HLA-B7 potential epitope Q9VF is underlined (bold). (B) Cytotoxic activity of splenocytes from six HLA-B7 humanized mice immunized with Q9VF synthetic peptide. Splenocytes were tested using RMA-B7 target cells pulsed with the Q9VF peptide or with a nonrelevant CMV peptide (R10TV) as a control. Data are mean ± SD of six mice and are representative of three independent experiments. Cytolytic responses were considered positive when specific lysis was >10% at an effector/target (E/T) ratio of 30:1.

Figure 2.

The Q9VF epitope from Gag-ARFP is presented by HIV-infected cells. (A) IFN-γ ELISPOT assay was performed using T1-B7 cells infected with HIV_LAI or HIV_LAI/stop or incubated with the indicated synthetic peptide as stimulators. Effector CTLs were splenocytes from HLA-B7 mice immunized with synthetic peptides corresponding to Gag-ARFP (Q9VF), HIV-1 Nef (F10LR), and CMV (R10TV) epitopes. Data are mean ± SD of triplicates for IFN-γ–producing cells per 1,000 effectors and are representative of three independent experiments. (B) Gag-ARFP is not required for HIV replication in T cell lines. The replication kinetics of HIV_LAI and HIV_LAI/stop were measured in two human cell lines (T1-B7 and Jurkat) by measuring p24 production in the supernatants at the indicated days after infection. HIV_LAI/Stop was derived from HIV_LAI by inserting a stop codon three nucleotides after the start codon of the sequence encoding for Gag-ARFP. T1-B7 cells were derived from T1 cells and express the HLA-B7^mα3 protein. NI, noninfected cells.

Figure 3.

CTL responses to Q9VF in HIV-infected individuals. (A) IFN-γ ELISPOT assay was performed with PBMCs from eight HIV-infected HLA-B7⁺ patients (nos. 1–8). As controls, PBMCs from seven HIV⁻ HLA-B7⁺ individuals (nos. 9a–9g) and one HIV⁺ HLA-B7⁻ (no. 10) individual were similarly analyzed. Responses against subject no. 9a are represented. The test was performed with the indicated concentrations of PBMCs and incubated with either CMV (T10AM), Gag-ARFP (Q9VF), Gag (Y10LF), or HCV (G9AT) peptides (at 5 μg/ml). Data are means of duplicates. Responses were considered positive when IFN-γ production was >50 spots/10⁶ PBMCs and exceeded at least twofold background levels (measured with the HCV peptide). Similar results were obtained with Q9VF for individual nos. 9a–9g (not depicted). (B) IFN-γ intracellular labeling of PBMCs from HIV⁺ HLA-B7⁺ patients (nos. 1 and 2) and from an HIV⁻ HLA-B7⁺ individual (no. 9a). Cells were incubated with the indicated peptides, stained for CD3, CD4, CD8, and intracellular IFN-γ expression and analyzed by flow cytometry. The indicated percent corresponds to the proportion of IFN-γ⁺ cells within CD3⁺CD8⁺ or CD3⁺CD4⁺ populations. Data are representative of three independent experiments. (C) Cytotoxic activity of PBMCs from two HIV-infected HLA-B7⁺ patients (nos. 1 and 2) and one HIV⁻ HLA-B7⁺ individual (no. 9a). ⁵¹Cr release assay was performed after two rounds of stimulation of PBMCs with the indicated peptides. Specific lysis was measured by using as targets T2-B7 cells pulsed with the corresponding peptides (closed circles) or with HCV G9AT peptide (open circles) as a control. Data are mean of duplicates at the indicated effector/target (E/T) ratio, and are representative of two independent experiments. Cytolytic responses were considered positive when specific lysis was >10% at an effector/target ratio of 30:1.

Figure 4.

Identification of five potential epitopes derived from Pol⁻ and Env⁻ ARFPs. (A) Nucleotide and amino acid sequences of ARFPs encoded within the pol and env genes. Start and stop codons are in bold, and the Kozak sequence is underlined. Nucleotide numbering is according to HIV_LAI sequence. The predicted amino acid sequences of the corresponding polypeptides (Pol-ARFP and Env-ARFP) are shown, and HLA-B7 potential epitopes (A9RL, S10PV, Q10QM, G10QT, and M9PT) are underlined. (B) Peptide relative affinity for HLA-B7 and immunogenicity in HLA-B7 mice. Peptide relative affinity was calculated as described in Materials and Methods. A low relative affinity (<5) denotes a strong binding. To measure immunogenicity, six HLA-B7 mice were immunized with each peptide. Splenocytes were tested in a ⁵¹Cr release cytotoxic assay using autologous target cells pulsed with the corresponding peptide or with a nonrelevant CMV peptide as a control. The number of responding mice is indicated. Data are mean ± SD of responder mice and are representative of two independent experiments. Cytolytic responses were considered positive when specific lysis was >10% at an effector/target (E/T) ratio of 30:1.

Figure 5.

CTL responses to the six ARFP-derived epitopes in HIV-infected individuals. (A) IFN-γ ELISPOT assay was performed with the indicated peptides. PBMCs from eight HIV-infected HLA-B7⁺ patients (nos. 1–8) and, as controls, from seven HIV⁻ HLA-B7⁺ (nos. 9a–9g) and one HIV⁺ HLA-B7⁻ (no. 10) individuals were analyzed. For controls, responses against subjects 9d and 10 are shown (similar results were obtained for individuals 9a–9g for all HIV peptides tested; not depicted). Results are presented as IFN-γ⁺ cells/10⁶ PBMCs. Responses were considered positive when IFN-γ production was >50 spots/10⁶ PBMCs and exceeded at least twofold background levels (measured with the HCV peptide). (white bars) Negative responses (below background thresholds, which are represented by the black line). (gray bars) Weak responses (twofold background levels). (black bars) Strong responses (at least threefold background levels). (+) Responses ≥1,500 IFN-γ⁺ cells. ND, not determined. (B) Cytotoxic activity of PBMCs from two HIV-infected HLA-B7⁺ patients (nos. 1 and 2). ⁵¹Cr release assay was performed after two rounds of stimulation of PBMCs with the indicated peptides. Specific lysis was measured by using T2-B7 cells pulsed with the corresponding peptides (closed circles) or with a HCV peptide (open circles, G9AT) as a control. Data are means of duplicates at the indicated effector/target (E/T) ratio, and are representative of two independent experiments. Cytolytic responses were considered positive when specific lysis was >10% at an effector/target (E/T) ratio of 30:1.

Figure 6.

Conservation of ARFP-derived epitope sequences among circulating HIV-1 strains. (A) For each ARFP-derived peptide, the percentage of HIV isolates carrying the corresponding sequence is shown. Analysis was performed among clade B and all clade HIV-1 strains present in the Los Alamos database. Strains without ARF translation potential (absence of AUG start codon or presence of stop codons within the ARFP) are also indicated. The Gag-ARFP Q9VF (QPRSDTHVF) has a major variant, Q9VF/5N (QPRSNTHVF). WT, wild type, corresponding to the HIV_LAI sequence. (B) Cross-reactivity between Q9VF and Q9VF/5N peptides. Six HLA-B7 mice were immunized with Q9VF or Q9VF/5N. Splenocytes were tested using autologous target cells pulsed with either Q9VF or Q9VF/5N peptide or with a nonrelevant CMV peptide as a control (R10TV). Data are mean of six mice and are representative of three independent experiments. Cytolytic responses were considered positive when specific lysis was >10% at an effector/target (E/T) ratio of 20:1.