CD8+ T cell responses to human immunodeficiency viruses type 2 (HIV-2) and type 1 (HIV-1) gag proteins are distinguishable by magnitude and breadth but not cellular phenotype



The mechanisms underlying the relatively slow progression of human immunodeficiency virus type 2 (HIV-2) compared with HIV-1 infection are undefined and could be a result of more effective immune responses. We used HIV-2 and HIV-1 IFN-γ enzyme-linked immunospot assays to evaluate CD8+ T cell responses in antiretroviral-naive HIV-2- (‘HIV-2+’) and HIV-1-infected (‘HIV-1+’) individuals. Gag-specific responses were detected in the majority of HIV-2+ and HIV-1+ subjects. Overlapping gag peptide analysis indicated a significantly greater magnitude and breadth of responses in the HIV-1+ cohort, and this difference was attributable to low responses in HIV-2+ subjects with undetectable viral load (medians 2107 and 512 spot-forming units per 106 PBMC, respectively, p=0.007). We investigated the phenotype of viral epitope-specific CD8+ T cells identified with HLA-B53- and HLA-B58-peptide tetramers (8 HIV-2+, 11 HIV-1+ subjects). HIV-2-specific CD8+ T cells were predominantly CD27+ CD45RA, and only a minority expressed perforin. The limited breadth and low frequency of CD8+ T cell responses to HIV-2 gag in aviremic HIV-2+ subjects suggests that these responses reflect antigen load in plasma, as is the case in HIV-1 infection. Immune control of HIV-2 does not appear to be related to the frequency of perforin-expressing virus-specific CD8+ T cells.


Enzyme-linked immunospot


Highly active antiretroviral therapy


Long-term nonprogressor


Spot-forming unit


Despite extensive genetic sequence homology, infection with the retrovirus HIV-2 leads to a distinct clinical outcome from that observed in HIV-1-infected (HIV-1+) individuals. Most HIV-2+ individuals are long-term nonprogressors (LTNP) with low or undetectable viral loads 1, 2. Therefore, HIV-2 infection has the features of an ‘attenuated’ HIV-1 infection, but evidence for a true attenuated phenotype is lacking 3. It has been proposed that HIV-2 replication in activated CD4+ T cells is relatively inefficient 4. An alternative but not mutually exclusive explanation for slow progression is that host immunity is more effective in HIV-2 control. Preservation of IL-2 and IFN-γ secretion by nonspecifically activated T cells in HIV-2 relative to HIV-1 infection has been reported 5. Strong virus-specific humoral and CTL responses have been observed in HIV-2-infected subjects 6, 7, but the role of virus-specific CD8+ T cells in controlling HIV-2 viral replication is less well established than in HIV-1 infection 8. Comparative studies of CD8+ T cell responses to vaccinia virus-expressed antigens or overlapping peptides in HIV-2+ and HIV-1+ antiretroviral-naive Gambian and Senegalese subjects showed broad similarities in the response to the two infections, despite significant differences in plasma viral loads in the two infections 9, 10. Therefore, superior control of virus replication in HIV-2 infection does not appear to be readily explained by frequencies of IFN-γ-producing CD8+ T cells. Recent studies in HIV-1-infected individuals have failed to find a correlation between the frequency of virus-specific IFN-γ-secreting CD8+ T cells and viral load 11, 12. This, together with observations that virus-specific T cells decline after initiation of highly active antiretroviral therapy (HAART) 13, 14, suggests that the IFN-γ response is antigen driven and that other qualitative aspects of the CD8+ T cell response determine virological control.

The characterization of CD8+ T cell subsets and delineation of CD8 T cell differentiation on the basis of cell surface markers including CD45RA, CD27, CD28 and CCR7 has been described in several chronic viral infections, including HIV-1. CD8+ T lymphocyte compartments have been categorized as follows: naive, CD45RA+ CD27+ CD28+ CCR7+; central memory, CD45RA CD27+ CD28 or CD45RA CD27 CD28+; and effector memory, CD45RA+ CD27 CD28 CCR715, 16. It has been proposed that T cells differentiate in a linear fashion along this pathway (reviewed in 17). HIV-1-specific CD8+ T cells, identified by tetramer staining, were found to display a CD28 CD27+ perforinlow phenotype and to have reduced cytolytic capacity relative to CD8+ T cells specific for other viruses such as human cytomegalovirus 18; however, the significance of these differences in diverse viral infections remains controversial. Failure to control HIV-1 replication has been attributed to a block in maturation of virus-specific CD8+ T cells to a CD45RA+ CCR7 phenotype 19. Expansion of terminally differentiated (CD27 CD28) total CD8+ T cells in HIV-2-infected individuals has been reported, but it is not known whether HIV-2-specific CD8+ T cells share this phenotype 4.

In this study, we characterized the CD8+ T cell response to HIV-2 in antiretroviral-naive viremic and aviremic HIV-2+ Gambian subjects, first in terms of breadth and magnitude using a peptide matrix-based IFN-γ enzyme-linked immunospot (Elispot) assay 11. We focused on the gag protein, as preferential targeting of gag has been shown to correlate with viral control in HIV-1 infection 12 and because several CD8+ T cell epitopes have been defined in HIV-2 gag. We next used pre-defined gag epitope peptides to generate HLA class I tetramers which enabled both quantification and phenotypic characterization of HIV-2-specific CD8+ T cells.


Magnitude and breadth of virus-specific CD8+ T cells in HIV-2+ and HIV-1+ subjects determined using a peptide matrix-based IFN-γ Elispot assay

Fresh PBMC were isolated from cohorts of 17 HIV-2+ and 11 HIV-1+ individuals whose CD4 T cell counts were >20% of the total peripheral lymphocyte count (Group A; Table 1). CD8+ T cells were positively selected from PBMC and tested for recognition of pools of peptides spanning the HIV-2 and HIV-1 gag proteins, presented in a matrix array. We observed responses to all matrix peptide pools in both HIV-1+ and HIV-2+ cohorts. A peptide-specific response greater than 100 spot-forming units (SFU)/106 PBMC in at least one of the peptide pools 1–6 and of pools 7–12 was detected in 13/17 (76.5%) HIV-2+ subjects and 11/11 HIV-1+ subjects (Fig. 1). However, the magnitude of the total gag response was significantly higher in HIV-1+ subjects (median 2107 SFU/106 cells) than in HIV-2+ subjects (median 872 SFU/106 cells) (p=0.015), with three of the HIV-2+ subjects giving a total gag response ⩽100 SFU/106 CD8+ T cells. Analysis of the breadth of the response, indicated by the number of peptide pools stimulating an HIV-specific response >100 SFU/106 cells, revealed significantly broader recognition in HIV-1+ subjects (median 3/6 HIV-2 pools targeted, range 1–6 vs. median 1.5/6 HIV-1 pools, range 0–3; p=0.012).

Table 1. Demographic characteristics of HIV-2+ and HIV-1+ subjectsa)
  1. a) NA: Not available

original image
Figure 1.

Recognition of gag peptide matrix pools by CD8+ T cells from HIV-2- and HIV-1-infected subjects (Group A) as determined by IFN-γ Elispot assay. Results are expressed as SFU/106 CD8+ PBMC. Negative control well values have been subtracted. Each circle represents one subject; where identical values for different subjects were obtained, circles overlap.

Since plasma viral load was below the limit of detection (100 copies/ml) in 10/17 HIV-2+ subjects, we performed a separate analysis comparing gag-specific responses in the viremic and aviremic subgroups with the HIV-1+ subjects. CD8+ T cell frequencies were lowest in the HIV-2+ aviremic subjects (median 512 SFU/106 cells), and this difference was highly significant when compared with the HIV-1+ subjects (p=0.007) (Fig. 2). Comparisons of the viremic HIV-2+ subjects’ responses (median 1090 SFU/106 cells) with those of the other two groups did not reveal statistically significant differences.

Figure 2.

Group A: total gag-specific IFN-γ CD8+ T cell response in HIV-2+ subjects stratified by viral load <100 or >100 copies/ml, indicated as ‘HIV-2 <100′ and ‘HIV-2 >100′, respectively, and in HIV-1+ subjects. Horizontal bars indicate median values.

Detection of virus-specific CD8+ T cells in HIV-2+ and HIV-1+ subjects using pre-defined optimal epitope peptides

In view of the low-magnitude IFN-γ response in HIV-2+ aviremic subjects, together with previous reports that control of HIV-1 correlated poorly with virus-specific CD8+ IFN-γ+ T cell numbers 11, we synthesized HLA class I tetramers to enable further analysis of HIV-2 gag-specific CD8+ T cells. Few CD8+ T cell epitopes have been defined in HIV-2. However, the MHC class I allotypes HLA-B*5801 and -B53, which are highly represented in the Gambian population 20, have been identified as the restricting element for at least one HIV-2 gag p24 epitope each and for the equivalent epitopes in HIV-1. These epitopes are HIV-2 gag182–190 TPYDINQML (TPY) and HIV-2 gag241–250 TSTVEEQIQW (TSTV) 6, 7. The equivalent sequences of HIV-1 are HIV-1 gag180–188 TPQDLNMML (TPQ) and HIV-1 gag240–249 TSTLQEQIGW (TSTL) 21, 22. We performed an initial screen for recognition in IFN-γ Elispot assays of gag p24 epitope peptides with fresh PBMC from HLA-B*5801- or HLA-B53-positive HIV-2+ (n=8) and HIV-1+ (n=11) subjects (Group B). We detected responses in 7/8 HIV-2+ and 10/11 HIV-1+ subjects (Table 2). The target of these responses differed in the two infections in that 4/5 HIV-2-infected HLA-B*5801+ subjects recognized the TSTV epitope, but no HLA-B*5801+ HIV-1+ subjects (n=4) responded to the equivalent HIV-1 TSTL epitope. Instead, another epitope, HIV-1 gag308–316 QASQEVKNW (QAS), which is cross-presented by HLA-B58 and HLA-B53 21, 22, was preferentially targeted. Of the HLA-B53+ HIV-2+ subjects, 3/3 made a response to the TPY epitope, and 5/7 HLA-B53+ HIV-1+ subjects recognized both its equivalent epitope, TPQ, and the QAS epitope.

Table 2. Responses to pre-defined HLA-B*5801- and HLA-B53-restricted gag epitopes by HIV-2+ and HIV-1+ subjects (Group B)
original image

Next, we employed HLA-B*5801 and HLA-B*5301 tetramers refolded with these peptides to determine the frequency of gag-specific CD8+ T cells by flow cytometry. Tetramer+ CD8+ T cells were detected in 6/8 HIV-2+ and 10/11 HIV-1+ subjects (Table 2). Comparison of the frequencies of CD8+ T cells specific for HIV-2 and HIV-1 epitopes obtained by tetramer staining and Elispot assay showed a trend towards higher numbers of IFN-γ-secreting cells relative to tetramer+ CD8+ T cells in HIV-2+ subjects, which was not statistically significant. Responses to mitogen (PHA) between subjects with high and low virus-specific IFN-γ responses were similar (data not shown).

Phenotypic profile of HIV-2 gag-specific CD8+ T cells

HIV-1-specific CD8+ T cells are reported to display a CD28 CD27+ perforinlow phenotype 16. We investigated the expression of CD27, CD28, CD45RA, Bcl-2 and perforin in CD8+ HIV-2 gag tetramer+ T cells and CD8+ T cells as a whole using three-color flow cytometry. Phenotype analysis was based on data from 6/8 HIV-2+ individuals due to the low frequencies of tetramer-staining cells in two subjects (Fig. 3A). Representative staining profiles from an HLA-B53+ and an HLA-B*5801+ HIV-2+ subject are shown in Fig. 3B. The majority of HIV-2-specific CD8+ tetramer+ cells expressed CD27, but there was variation between subjects in the expression of CD28 by these cells. A preponderance of CD8+ tetramer+ cells that co-expressed CD27 and CD28 were detected in two subjects. CD45RA and perforin were each expressed on <40% of HIV-2 tetramer+ cells, and there was considerable heterogeneity within the six HIV-2+ subjects with respect to the proportion of tetramer+ cells expressing Bcl-2. Taken together, these observations suggest that HIV-2-specific CD8+ T cells have a similar phenotype to that described in chronic HIV-1 infection 16. The phenotype of total CD8+ T cell populations from HIV-2+ individuals differed from that of virus-specific T cells in that a greater proportion of CD8+ T cells were CD45RA+ and CD28.

Figure 3.

(A) Dot plots showing percentages of gated CD8+ tetramer+ (‘tetr+’) cells and total CD8+ T (‘tCD8+’) cells expressing CD28, CD27, CD45RA, perforin and Bcl-2 in HIV-2-infected subjects. CD8+ lymphocytes were defined by forward and side scatter and CD8bright expression as shown in (B). The percentages of cells expressing phenotypic markers were obtained from quadrant statistics. (B) Representative staining profiles of CD8+ tetramer+ T cells (FL-2 fluorochrome) from an HLA-B53+ and an HLA-B*5801+ HIV-2-infected subject obtained with various phenotypic markers (FL-1 fluorochrome).


HIV-2 infection is largely confined to West Africa and, consequently, has not been studied as intensively as HIV-1. While a small number of studies have investigated HIV-2-specific CD8+ T cell responses 5, 10, 23, only a limited number of T cell epitopes have been defined 6, 7. We have extended the findings of these studies by showing that: (i) the frequency of gag-specific CD8+ IFN-γ+ T cells is significantly lower in HIV-2+ subjects, particularly in those with undetectable viral load in plasma, than in HIV-1+ subjects; (ii) CD8+ T cells from HIV-2-infected individuals are less broadly directed at gag epitopes than CD8+ T cells from HIV-1-infected individuals; and (iii) CD8+ T cells identified by HIV-2 gag tetramer staining express low levels of perforin, similar to that reported in HIV-1 infection 16, 18.

The use of a peptide matrix comprising overlapping 15–18-mers in the IFN-γ Elispot assays has been shown to be highly sensitive for the detection of HIV-1-specific CD8+ T cells, particularly when these are present at a frequency of at least 200 SFU/106 PBMC 11. This may account for the higher frequencies of HIV-2 gag-specific CD8+ T cells and the greater proportion of responders we identified than those reported in studies where recombinant antigens were used 9, 23. It could also explain the significant difference we observed in the magnitude of the total CD8+ gag-specific responses in HIV-2+ and HIV-1+ subjects. Our findings could be influenced by the smaller number of subjects in our cohort, although we used the same selection criteria and observed similar CD4 T cell counts and percentages in the HIV-2+ subjects to those reported in the study by Jaye et al., who reported no difference in gag-specific responses in HIV-2+ and HIV-1+ Gambian subjects 9.

In common with another larger study, over half of the HIV-2+ subjects in our cohort had undetectable plasma viral load 10, and these individuals were found to have the lowest virus-specific CD8+ T cell response in the Elispot assay. In HIV-1 infection, high frequencies of virus-specific IFN-γ+ CD8+ T cells can be found at all stages of disease and in LTNP, but they are highest in those with chronic untreated infection and lowest in patients on HAART with fully suppressed virus 11, 24. The observation that these T cells reappear in the circulation upon virological failure or HAART interruption suggests that they are maintained by continuous antigenic stimulation in untreated individuals 25. It is therefore possible that the relatively low-frequency HIV-2-specific CD8+ T cell response we observed in aviremic HIV-2+ subjects is the result of low levels of virus replication as detected by plasma viral load. Jin et al. proposed a model to describe the relationship between viral load and the CD8+ T cell response in HIV-1 infection, in which these two parameters are positively correlated when antigen concentration is limiting (e.g. after HAART initiation), but under conditions of antigen excess (e.g. untreated progressive disease) this relationship is lost 26. This model could explain our findings, although confirmation in a larger cohort is needed. It is noteworthy that CD8+ T cell responses of the HIV-2+ and HIV-1+ subjects in our study could be distinguished by breadth also. We do not know whether this is a consequence of differing plasma viral loads, and hence increased viral diversification in HIV-1 relative to HIV-2 infection, or whether this is due to characteristics of the virus-specific cellular response early in infection which determines the viral set-point. A more effective T helper response (M. Sayeid-Al-Jamee, unpublished observations) 10 or broader T cell receptor usage in HIV-2 infection 27 could potentially limit escape of HIV-2 from CD8+ T cell recognition, which would be manifested in narrowly focused CD8+ T cell responses. Alternatively, the presence of fewer CD8+ T cell epitopes in the HIV-2 gag protein could explain this observation. Further studies that include all structural and accessory proteins are required to determine whether a low-magnitude, narrow CD8+ T cell response is a universal feature of HIV-2 infection.

We also investigated the CD8+ T cell response to pre-defined HLA-B*5801- and HLA-B53-restricted gag epitopes using IFN-γ Elispot and tetramer assays, but found no correlation between the numbers of virus-specific CD8+ T cells obtained by these two methods. This could be due to the finding that tetramer+ cells were detected in only six HIV-2+ subjects and to the limited number of epitopes available for study. Migueles et al. found no difference in the frequency of CD8+ IFN-γ+ T cells in HIV-1+ LTNP and untreated progressors, but LTNP were distinguishable by enhanced CD8+ T cell proliferative responses and perforin expression 24; therefore, further assessment of these parameters in HIV-2 infection is warranted.

While simultaneous analysis of multiple phenotypic markers was limited in this study, through assessment of individual differentiation marker expression we observed that the majority of HIV-2-specific T cells were CD27+ perforinlow, while CD28 expression was more variable. This is consistent with the incomplete or ‘intermediate’ differentiation phenotype described in chronic HIV-1 infection. Two HIV-2+ subjects were found to have expansions of tetramer+ cells that were CD27+ CD28+, a ‘naive’ or ‘early’ differentiation phenotype not commonly observed in chronically infected HIV-1+ patients 16. Therefore, our findings do not show a distinct phenotype for HIV-2-specific CD8+ T cells, but may have been influenced by inclusion of HIV-2 ‘progressors’ in this analysis, as patients were selected for tetramer studies on the basis of HLA class I haplotype. Nevertheless, it is noteworthy that perforin was expressed in a minority of virus-specific and total CD8+ T cells (medians 11% and 18%, respectively) from HIV-2+ subjects. Higher proportions of mature, perforin+ T cells were detected in HIV-1+ slow progressors relative to those present in progressors during chronic disease 19 and in nonprogressors after in vitro stimulation 24. Our data suggest that HIV-2+ subjects cannot be differentiated from HIV-1+ subjects on the basis of ex vivo perforin expression in HIV-2-specific CD8+ T cells. We observed greater inter-subject variation in the proportion of HIV-2 tetramer+ cells expressing the anti-apoptotic factor Bcl-2 than that which has been reported in HIV-1 infection 16. The total CD8+ T cell pool in HIV-2+ subjects was also heterogeneous in terms of CD28, CD45RA and Bcl-2 expression. The phenotypic heterogeneity of CD8+ T cells in HIV-2-infected individuals might be a reflection of the polyclonality of expanded T cell populations 27.

In summary, we have used a highly sensitive peptide matrix assay, together with HLA class I-peptide tetramers, to dissect the HIV-2 gag-specific CD8+ T cell response. We speculate that the relatively narrow specificity and low frequency of HIV-2 gag-specific CD8+ T cells, which is most evident in aviremic subjects, are a consequence of low antigenic drive. The lack of a distinct phenotype for HIV-2 gag-specific CD8+ T cells suggests that other factors determine virological control in HIV-2 infection. The efficacy of T cell help may be key, possibly because viral replication is reported to be less efficient in activated CD4+ T cells 4 or because HIV-2 is less likely to infect virus-specific CD4+ T cells than HIV-1 28. Studies designed to address these questions will be critical to our understanding of the pathogenesis of both HIV-2 and HIV-1 infections.

Materials and methods

Patient cohort

Antiretroviral-naive HIV-2+ and HIV-1+ adults (n=47) were recruited from the HIV clinic at MRC Laboratories, Fajara, The Gambia. Approval for venesection was obtained from the Gambian Government/MRC Ethical Committee, and all subjects gave verbal consent to donate 20 ml blood up to a maximum four times per year. HIV antibody testing was performed by a combined HIV-1 and HIV-2 enzyme immunoassay (ICE* HIV.1.0.2; Murex Diagnostics, Kent, UK) and positive results were confirmed by monospecific ELISA (Wellcozyme HIV Recombinant and Murex HIV-2). Percentages of CD4 T lymphocytes were determined by flow cytometry using BD Tritest reagents and Multiset software (Becton Dickinson Immunocytometry System) (Table 1).

Peptides and peptide matrix design

HIV-1 (consensus clade A) and HIV-2 (clade A) 20-mer peptides, 36 each, overlapping by ten amino acids spanning the entire region of each gag protein were obtained from the MRC AIDS Reagent Programme. Pre-defined peptide epitopes were synthesized in-house by 9-fluorenylmethoxycarbonyl chemistry on a Zinnser peptide synthesizer (Advanced Chemtech Inc., KY). Peptide purity, as determined by HPLC, was >70%. A matrix was designed for each of the 36 overlapping 20-mer HIV-2 and HIV-1 gag peptide sets, as described 11. Peptide pools 1–6 each comprised six peptides in sequence (i.e. overlapping), while pools 7–12 each comprised six non-sequential peptides; therefore, each individual peptide was represented twice in the matrix. The final concentration of each individual peptide in the Elispot assays was 5 µg/ml.

HLA typing

HLA typing was performed by amplification refractory mutation system PCR using sequence-specific primers 29.

HIV-2 viral load assay

HIV-2 virus loads were assayed by a quantitative PCR method described previously 1, with the following modifications: approximately 100 copies of an internal molecular control were co-extracted and co-amplified to control for inefficiencies of extraction and nonspecific inhibition of amplification. The internal control was a 1-kb synthetic RNA construct spanning the PCR primer and probe binding sites. The modified assay had a cut-off of 100 RNA copies/ml.

IFN-γ Elispot assay

Fresh PBMC or CD8+ cells, positively selected by magnetic bead separation according to the manufacturer's instructions (Miltenyi Biotech Ltd., UK), were stimulated with either the peptide matrix pools (in duplicate) or pre-defined CD8+ T cell epitopes (in triplicate) in an IFN-γ Elispot assay, as described 30. Negative control wells contained cells + medium only and positive control wells contained cells + PHA (1 µg/ml). Wells were counted using an automated Elispot plate reader (AID Elispot Systems, Germany). Responses were reported as SFU per 106 CD8+ T cells (peptide matrix assays) or per 106 PBMC (assays with epitope peptides). Criteria for a positive result were SFU in peptide wells at least three times that observed in negative control wells and >50 SFU/106 cells after subtraction of negative control values. The latter were <100 SFU/106 cells in 87% of the assays. Because each peptide is duplicated in the peptide matrix and the response to a given epitope may be affected by its position within a longer peptide 31, the total gag-specific response was determined by the sum of all peptide responses divided by two.

HLA class I-peptide tetramer synthesis, tetramer staining and phenotypic analysis

Cloning, expression and the generation of HLA-B*5301 and HLA-B*5801 tetramers was performed as described 32. Heparinized fresh blood (150 µl) was stained with 0.5 µg tetramer for 15 min at 37°C, followed by incubation with red cell lysis buffer, and staining with monoclonal antibodies specific for the cell surface antigens CD27, CD28, and CD45RA (Becton Dickinson, Mountain View, CA) at room temperature for 15 min. To assess expression of intracellular antigens, cells were fixed and permeabilized in FACS permeabilization buffer for 10 min and washed prior to staining with anti-Bcl-2 and anti-perforin monoclonal antibodies (Becton Dickinson, Mountain View, CA). Cells were washed and stored at 4°C in PBS/5% formaldehyde and analyzed on a FACSCalibur flow cytometer using CellQuest software.

Statistical analysis

Statistical analyses were performed using SPSS software. Analyses of significance were based on two-tailed tests. Differences in the magnitude and breadth of CD8+ T cell responses to HIV-2 and HIV-1 were analyzed using the Mann-Whitney test.


We are indebted to the patients and staff of MRC Laboratories, Fajara, The Gambia, without whom this study would not have been possible. We wish to thank Kati de Gleria for peptide synthesis and Sowsan Atabani and Victor Appay for technical advice. This work was supported by the Medical Research Council, UK, the Elizabeth Glaser Pediatric AIDS Foundation and the Ministry for Science and Technology, Portugal.


  1. 1


  2. 2


  3. 3