HIV-1 Vpu represents a minor target for cytotoxic T lymphocytes in HIV-1-infection

We have previously shown that Vpu is rarely targeted by HIV-1-specific cytotoxic T lymphocytes (CTL). The present report extends these findings and describes the characterization of the first CTL epitope within HIV-1 Vpu, identified in an individual with long-term non-progressive HIV-1 infection. The epitope was shown to be highly conserved among HIV clade B sequences and is restricted by HLA-A*3303, an HLA allele commonly seen in Asian and west-African populations. With more than 57 million individuals worldwide infected with HIV-1 since the beginning of the epidemic [1], the development of an effective HIV vaccine either to protect against HIV-1 infection or to attenuate the course of disease is urgently needed. Increasing evidence indicates that HIV-1-specific CD8 cytotoxic T lymphocytes (CTL) and CD4 T helper cells play a critical role in the control of viral replication in HIV-1 infection [2,3]. Recent studies have shown that HIV-1-specific CTL responses are directed against a large variety of different epitopes located within structural as well as regulatory and accessory HIV-1 proteins [4,5]. However, no HIV-1-specific CTL epitopes have been defined to date within the accessory HIV-1 protein Vpu, an 81 amino acid membrane protein that facilitates budding of new virus particles and regulates the degradation of CD4 cells [5,6]. In a previous report investigating CTL responses against all accessory proteins [5] we found detectable CTL responses against HIV-1 Vpu in one out of 60 (1.6%) individuals screened. In an extension of that study, we have now characterized HIV-1 Vpu-specific CTL responses in a total of 85 HIV-1-infected individuals at different stages of HIV-1 infection, including 45 individuals with treated primary infection, two individuals with untreated acute HIV-1 infection, 18 individuals with treated chronic infection, and 20 individuals with untreated chronic infection (14 long-term non-progressors and six progressors). HIV-1 Vpu-specific CTL responses were assessed using 14 overlapping peptides (15–18mer peptides overlapping by 10 amino acids) spanning the entire HIV-1 clade B Vpu sequence in an IFN-γ Elispot assay on fresh and frozen peripheral blood mononuclear cells (PBMC) [4,5]. In addition, all 85 individuals were tested for CTL responses directed against HIV-1 Gag and Nef using the same methods. CD8 T-cell dependence of all responses to synthetic peptides was confirmed by CD4/CD8 T depletion studies using magnetic beads (MACS; Miltenyi Biotech, Germany), as described [4,5]. While all study subjects had detectable HIV-1-specific CTL responses (92% against HIV-1 Gag and 80% against HIV-1 Nef), only two out of the 85 screened subjects (2.4%) had detectable, but low-level CTL responses directed against HIV-1 Vpu. Both individuals, an HIV-1 long-term non-progressor (CMW) [5] and one individual with primary HIV-1 infection (AC-71), had well-controlled viremia in the absence of antiretroviral therapy. These data confirm our previous reports on the remarkably low frequency of HIV-1 Vpu-specific CTL responses in HIV-1 infection compared with other HIV-1 proteins [5]. CD8 T cells in subject CMW targeted the overlapping peptide Vpu-6 (IVFIEYRKILRQRKID) with a frequency of 170 spot-forming cells/106 PBMC. It was found to be a subdominant response in this individual and contributed 5% to the total CTL response against the whole expressed HIV genome (data not shown). In order to confirm and verify the low-level CTL response in CMW, we generated Vpu-6-specific cytotoxic CD8 T cells using the limiting dilution assay [7]. As no HIV-1-specific CTL epitopes have been defined within HIV-1 Vpu to date [6], we set out to characterize the optimal CTL epitope within the Vpu-6 peptide. The HLA class I restriction of the Vpu-6-specific response was determined using an autologous and partially matched antigen-presenting lymphoblastoid B cell line (L-BCL) in a 51chromium release assay, as previously described [4]. The response directed against the Vpu-6 peptide was shown to be restricted by HLA-A*3303, an HLA class I molecule frequently encountered in west Africa and southeast Asia [8,9] (Fig. 1a). The optimal sequence of the CTL epitope contained within Vpu-6 was characterized using PBMC and serial dilutions of peptide truncations in an Elispot assay [10]. As demonstrated in Fig. 1b, the nonamer peptide EYRKILRQR (ER9, Vpu 29-37) was recognized best at the lowest peptide concentration. The epitope lies within one of two α-helical regions in the cytoplasmatic domain of HIV-1 Vpu in proximity to the βTrCP binding motif, which may be implicated in the induction of apoptosis in HIV-infected cells [11]. Despite the high overall sequence variability observed for Vpu (Yusim et al., personal communication), the epitope sequence is 71% conserved among the 60 published HIV-1 clade B Vpu sequences [12] (http://www.hiv.lanl.gov). The EYRKIL RQR epitope represents the first described optimal CTL epitope within the accessory protein Vpu, as well as one of the first HIV-1 CTL epitopes restricted by HLA-A*3303 [6,13]. Its sequence is in concordance with the described peptide-binding motif for HLA-A*3303 [14], containing a tyrosine at the P2 anchor position and an arginine at the C-terminal end of the peptide.Fig. 1.: (a) HLA restriction using autologous and partly HLA-matched lymphoblastoid B cell line (B-LCL) pulsed with the Vpu-6 peptide or no peptide as a negative control in a 51chromium- release assay (E : T ratio 10 : 1). The HLA-A * 3303 allele was identified as the restricting allele for the Vpu-6-specific cytotoxic T lymphocyte (CTL) response. (b) Fine-mapping of the optimal epitope contained within the Vpu-6 peptide using serial dilutions of peptide truncations in an Elispot assay, as previously described [4]. The peptide EYRKILRQR is identified as the optimal CTL epitope.In order to assess the frequency of recognition of the newly defined epitope we also screened five additional HLA-A*3303-expressing HIV-1-infected individuals for responses to the EYRKILRQR epitope. Two of these five individuals and thus a total of three out of six (50%) tested HLA-A*3303-positive individuals had detectable CTL responses against the epitope, suggesting that it may in fact be a fairly common response among individuals expressing the HLA-A*3303 allele. Interestingly, all three HLA-A*3303-positive responders were HIV-1 long-term non-progressors with viral control in the absence of treatment. These data confirm that HIV-1 Vpu is the least targeted protein by HIV-1-specific CTL, as measured using current techniques. This infrequent recognition by CTL is especially remarkable considering that another accessory protein of similar size Vpr (96aa) is highly targeted by CTL in HIV-1-infected individuals [5]. It has been demonstrated that HIV-Vpu is the most variable protein within HIV-1 (Yusim et al., personal communication). Using overlapping peptides based on an HIV-1 clade B consensus sequence in the assays described here could therefore miss responses against this highly variable protein, if the autologous virus sequence is considerably different from the peptide sequence [15,16]. The characterization of HIV-1- specific CTL responses using peptides spanning the autologous virus sequence is needed to address the potential underestimation of immune responses directed against very variable HIV-1 proteins in the future. Further studies will thus be needed to assess whether this infrequent recognition of HIV-1 Vpu reflects low immunogenicity of the protein in vivo, a lack of HLA class I processing of Vpu, or an underestimation of CTL responses directed against this highly variable protein using peptides based on HIV-1 consensus sequences or other factors. Marylyn M. Addo Marcus Altfeld Almas Rathod Michelle Yu Xu G. Yu Philip J. R. Goulder Eric S. Rosenberg Bruce D. Walker