Broad, Intense Anti-Human Immunodeficiency Virus (HIV) Ex Vivo CD8⁺ Responses in HIV Type 1-Infected Patients: Comparison with Anti-Epstein-Barr Virus Responses and Changes during Antiretroviral Therapy

Subjects.

We studied retrospectively anti-HIV CD8⁺ T lymphocytes obtained from 34 HIV-1-infected patients and from 10 healthy HIV-seronegative volunteers who served as control subjects. Cohorts were established with the approval of the local ethics committee (Comité Consultatif de Protection des Personnes dans la Recherche Biomédicale de l’Hôpital Cochin), and all participants gave their written informed consent for the constitution of cell banks.

Cells.

PBMC isolated from freshly drawn heparinized venous blood were isolated by density gradient centrifugation (separation medium; Flow, Irvine, United Kingdom) and used after freezing and thawing. Subjects were serologically HLA typed by complement-mediated lymphocytotoxicity.

Peptides.

The HIV-1 and EBV epitopes used are listed in Table 1. The sequences of these epitopes are available online (19a, 32a). Peptides were synthesized by Neosystem (Strasbourg, France) and supplied by the Agence Nationale de la Recherche sur le SIDA. Lyophilized peptides were diluted to 1 mg/ml in water plus 10% dimethyl sulfoxide aliquoted, and stored at −20°C. They were used at a final concentration of 1 μg/ml (0.01% dimethyl sulfoxide in cell culture medium).

ELISPOT assay.

The IFN-γ ELISPOT assay was adapted from that of Scheibenbogen et al. (40). Ninety-six-well nitrocellulose plates (Millipore, Bedford, Mass.) were coated with capture mouse anti-human IFN-γ monoclonal antibody (2 μg/ml; code 1598-00; Genzyme, Rüsselheim, Germany). PBMC, either freshly isolated or thawed and cultured overnight in complete medium, were plated in triplicate at serial dilutions (3 × 10⁵ to 10⁴ cells per well). Appropriate stimuli were then added, and the plates were incubated for 20 h at 37°C and 5% CO₂. After the plates were washed, the cells were incubated with 100 μl of rabbit polyclonal anti-human IFN-γ antibody (code IP500; 1:250 dilution; Genzyme), then with an anti-rabbit immunoglobulin G-biotin conjugate (1:500 dilution; Boehringer, Mannheim, Germany), and finally with alkaline phosphatase-labeled extravidin (Sigma, St. Louis, Mo.). Spots were developed by adding chromogenic alkaline phosphatase conjugate substrate (BioRad, Hercules, Calif.). Colored spots were counted in a stereomicroscope. The signal was considered positive if the number of spots was significantly greater than that obtained with the negative controls at least at one of the cell concentrations used and if the number of spots obtained was proportional to the number of plated cells. Frequencies of IFN-γ spot-forming cells (SFC) were then calculated.

Positive controls consisted of six wells containing 300 to 1,000 cells with phorbol myristate acetate (50 ng/ml) and ionomycin (500 ng/ml). This strong mitogenic stimulus permitted verification that freezing and thawing various cell samples did not introduce artifactual differences in T-cell reactivities, thus constituting an indirect check of overall T-lymphocyte viability.

Negative controls consisted of HLA-matched and mismatched epitopic peptides derived from HIV or other viruses (including the HLA-A2-restricted peptide Tax 11-19, derived from the irrelevant human T-cell lymphotropic virus type 1). Negative controls never elicited a specific response compared with PBMC incubated in medium alone. The PBMC from the HIV-seronegative individuals secreted no IFN-γ in response to HIV peptides, but they did in response to some EBV or influenza virus peptides, depending on their HLA haplotype (data not shown). For some HIV-1-infected patients, we checked that all epitopes recognized in the IFN-γ ELISPOT assay were able to expand antipeptide cell lines. ELISPOT assays performed on purified T-cell subsets showed that only CD8⁺ T cells secreted IFN-γ in response to viral HLA class I epitopes (data not shown). These points show the high specificity of the ELISPOT assay.

Intracellular staining of IFN-γ.

Cells (1.8 × 10⁶) were cultured for 6 h in 300 μl of complete medium with 10 μg of brefeldin A per ml in 12- by 75-mm test tubes, in the presence of various stimuli. The positive control consisted of mitogenic activation with phorbol myristate acetate (25 ng/ml) and ionomycin (1 μg/ml). Stimulator cells were autologous lymphoblastoid cell lines obtained by transforming PBMC with EBV (EBV-LCL). They were tested either untreated or after infection with vaccinia virus recombinants for various HIV-1_LAI genes as described elsewhere (12). Reactivity against autologous EBV-LCL alone and against the same cell line infected with vaccinia virus recombinants expressing the env, gag, pol, and nef genes of HIV-1_LAI were tested for each patient. Negative controls were medium alone and EBV-LCL infected with wild-type vaccinia viruses. The intracellular staining procedure of IFN-γ was adapted from the Becton Dickinson protocol. Cells were stained directly in the culture tubes with 15 μl of PerCP (peridinin chlorophyll a protein)-conjugated anti-CD8 and 15 μl of phycoerythrin CPE)-conjugated anti-CD28 or with isotype-matched negative control reagents. Cells were then permeabilized and incubated with 30 μl of anti-human IFN-γ-fluorescein isothiocyanate (FITC) or with isotype-matched negative control reagent. Finally, cells were washed, resuspended in 100 μl of cell wash containing 1% paraformaldehyde, and stored at 4°C in the dark for flow cytometry analysis.

Five-parameter analysis (forward and side scatter, FITC, PE, and PerCP) was performed on a FACScan flow cytometer using the Cellquest software (Becton Dickinson). For each sample, 125,000 to 800,000 events were acquired and gated on CD8 expression; a scatter gate was designed to include only viable lymphocytes.

Statistical analyses.

Data analyses were performed with the StatView 4.5 software (Abacus Concepts, Berkeley, Calif.). Comparisons between the anti-HIV and the anti-EBV responses were performed by using the Mann-Whitney test. Correlations between variables were identified by using simple linear regression analysis. Changes over time of CD4⁺ T-cell counts and intensity of antiviral CD8⁺ responses were summarized by the least-squares estimate of the slope of the linear regression plots of each measurement against time. The Wilcoxon rank test was used to compare the antiviral responses before and after antiretroviral therapy. P values of 0.05 or less were considered significant.

Comparison between the anti-HIV and the anti-EBV ex vivo CD8⁺ responses in unstimulated PBMC of HIV-seropositive individuals.

We compared the breadth and intensity of anti-HIV and anti-EBV ex vivo CD8⁺ T-cell responses in the natural history of HIV infection in a large population, 34 HIV-infected patients harboring at least 1 of 17 HLA alleles for which several HIV or EBV epitope peptides have been described (Table 1). At the time of the test, these patients had been diagnosed as HIV positive for an average of 4.8 years (range, 6 months to 12 years; median, 5 years), none had received HAART, and they harbored a wide range of CD4⁺ T-cell counts (27 to 952; mean, 395; median, 325) and viral loads (undetectable to 2,503,000; mean, 215,054; median, 36,000) (Table 2).

The number of peptides tested for each patient (range, 7 to 20; median, 13) depended on each combination of HLA-A and HLA-B alleles. The positive responses of each individual are shown in Fig. 1. Three parameters were defined to describe the anti-HIV and anti-EBV CD8⁺ T-cell repertoires of each patient: (i) polyclonality (number of viral epitope peptides recognized), (ii) total intensity (sum of the numbers of antiviral CD8⁺ T cells per 10⁶ PBMC for all viral peptides recognized), and (iii) mean intensity (total intensity divided by polyclonality) (Table 2).

All patients responded to HIV. Most recognized at least one HIV epitope for each HLA molecule tested. Anti-HIV responses showed great polyclonality (mean, 5; median, 5; range, 1 to 13), high total intensity (mean, 4,084; median, 3,576; range, 50 to 16,413), and high mean intensity (mean, 762; median, 720; range, 50 to 2,052).

Most (24 of 30) patients also responded to EBV with moderate total intensity (mean, 355; median, 160; range, 0 to 1,856) and moderate mean intensity (mean, 303; median, 136; range, 27 to 1,344). As fewer CD8⁺ epitopes have been described for EBV, the polyclonality (each patient recognized one EBV peptide on average [mean, 1; range, 0 to 3]), and total intensity of anti-EBV responses may be greatly underestimated. Therefore, with this experimental approach, it is not possible to compare the total intensity and polyclonality of the anti-EBV and anti-HIV repertoires. However, the mean intensity of anti-EBV responses was significantly lower than that of anti-HIV responses (Fig. 1; a mean of 303 SFC per 10⁶ PBMC was obtained for each EBV peptide recognized, versus 762 for HIV [P < 0.0001]).

To compare the total intensities of anti-EBV and anti-HIV responses, we quantitated in six patients the CD8⁺ T lymphocytes capable to produce IFN-γ ex vivo in response to autologous EBV-LCL eventually infected with vaccinia virus recombinants carrying various HIV-1_LAI genes, using intracellular staining and cytofluorometry. A much higher number of CD8⁺ T lymphocytes responded to stimulation with EBV-infected cells expressing also HIV-1 proteins than with cells expressing only EBV antigens (Table 3). We further verified the specificity of the responses observed by verifying that CD8⁺ T cells directed against HIV were mostly CD28⁻ whereas those responding to EBV-LCL alone or infected with wild-type vaccinia virus were mostly CD28⁺, as previously reported (13). Therefore, the total frequency of anti-HIV CD8⁺ T-cell responses appears much more important than that of anti-EBV responses, at least in the six subjects tested.

The data suggested massive ex vivo HIV-specific CD8⁺ T-cell responses, as well as the persistence of weaker anti-EBV responses, at all stages of HIV infection. We therefore examined the influence of HIV-induced immunodepression on anti-HIV and anti-EBV responses and the relationship between anti-HIV responses and virus load.

Influence of HIV-induced immunodepression on antiviral CD8⁺ ex vivo responses.

The CD4⁺ T-cell counts were used as a marker of HIV-induced immunodepression, and their relationship with antiviral responses was examined (Fig. 2). The polyclonality of anti-HIV responses was positively correlated with the CD4⁺ T-cell counts. Neither the mean intensity nor the total intensity of anti-HIV responses was correlated with the CD4⁺ T-cell counts, even in comparisons of patients for the reactivity with a given HLA molecule (not shown). In contrast, the total intensity and mean intensity of anti-EBV CD8⁺ responses were positively correlated with CD4⁺ T-cell counts. Therefore, unlike the anti-HIV responses, the anti-EBV responses tended to decrease with increasing HIV-1-induced immunodepression.

Relationship between virus load and anti-HIV ex vivo CD8⁺ responses.

The polyclonality and mean intensity of anti-HIV CD8⁺ T-cell responses were apparently not correlated with virus load (r = −0.312, P = 0.13; and r = −0.223, P = 0.28, respectively). However, there was a trend toward an inverse relationship between the total intensity of anti-HIV CD8⁺ T-cell responses and virus load, although the correlation was not significant (r = −0.381, P = 0.06). There was also no clear correlation between virus load and anti-HIV ex vivo CD8⁺ T-cell responses in comparisons of patients for reactivity with a given HLA molecule (not shown).

Evolution of IFN-γ secretion in untreated seropositive individuals.

Six of the 34 HIV-seropositive patients studied were followed longitudinally for several years while they received no efficient antiretroviral therapy (Fig. 3). Patients Z038 and Z077 had very stable anti-HIV and anti-EBV responses against all tested epitopes over 8 and 5 years, respectively (Fig. 3A and B). These patients also had remarkably stable CD4⁺ T-cell counts during that period. Patients Z042 and Z108 showed increases in anti-HIV CD8⁺ T-cell responses, while their anti-EBV responses were consistently stable or decreased moderately (Fig. 3C and D). The CD4⁺ T-cell counts of these patients clearly decreased during that period. Finally, patients P04 and P51 showed significant decreases in anti-HIV and anti-EBV CD8⁺ T-cell responses (Fig. 3E to F). Patient P04 had a dramatic decrease in CD4⁺ T-cell counts during follow-up (Fig. 3E). This subject was symptomatic at enrollment and died of AIDS 3 years later. By contrast, patient P51 showed only a moderate decrease of CD4⁺ T-cell counts, even smaller than that of patients Z042 and Z108. However, this subject had a very high virus load, which increased sharply from 120,000 to 600,000 copies of HIV RNA per ml of plasma in the first 1.5 years of study (Fig. 3F).

Impact of HAART on antiviral CD8⁺ T-cell ex vivo responses.

We also studied the impact of HAART on the numbers of anti-HIV and anti-EBV CD8⁺ T cells in five patients treated during the asymptomatic phase of HIV-1 infection. Patients Z108, Z118, Z134, and Z136 were given triple combination therapy, with a protease inhibitor and two nucleoside reverse transcriptase inhibitors, while patient P51 was treated with two nucleoside reverse transcriptase inhibitors. Both treatment regimens rapidly reduced plasma HIV RNA to undetectable levels. There were significant decreases in anti-HIV CD8⁺ T-cell responses in all patients against all HIV epitopes initially recognized (Fig. 4A; P = 0.0003). The anti-EBV CD8⁺ T-cell responses increased in all patients (Fig. 4B).