Longitudinal Assessment of Changes in HIV-Specific Effector Activity in HIV-Infected Patients Starting Highly Active Antiretroviral Therapy in Primary Infection

Study population

We studied 22 subjects enrolled in the Quebec HIV PI cohort and six subjects in the chronic phase of infection. The institutional review boards of all participating study sites approved the study and all participants signed informed consent. All subjects were antiretroviral drug-naive at study entry. Subjects in the Quebec PI cohort were followed clinically at study entry, at wk 2, 4, 6, and 8, and every 3 mo until mo 24. Subjects in chronic infection were followed monthly for 24 mo. All study participants, with the exception of MQPI018 and MQPI020, started HAART consisting of at least two reverse transcriptase inhibitors and at least one protease inhibitor or non-nucleoside transcriptase inhibitor at the first clinic visit. MQPI018 started HAART at visit 5 (2 mo after study entry), and MQPI020 started HAART at visit 8 (9 mo after study entry). The date HAART was started was defined as baseline for the results reported here.

Entry criteria for the Quebec HIV PI cohort have been described previously (19). The presumed date of infection was estimated for each individual using clinical and laboratory data as well as patient history information. The following guidelines proposed by the Acute HIV Infection and Early Disease Research Program (Bethesda, MD) sponsored by the National Institutes of Allergy and Infectious Disease Division of AIDS were used to estimate the date of infection: the date of the first indeterminate Western blot minus 35 days; and the date of a positive HIV RNA test or p24 Ag assay available on the same day as a negative HIV enzyme immunoassay (EIA) test minus 14 days. The date of onset of symptoms of an acute retroviral syndrome minus 14 days was also used to estimate the date of infection (18). Information obtained from questionnaires addressing the timing of high-risk behavior for HIV transmission was used when available to confirm the presumed date of exposure.

Participants were classified into study groups based on the time elapsed between infection and start of HAART. Subjects in group 1 (n = 6) started HAART before seroconversion as defined by a negative or indeterminate standard HIV-1 Ab EIA and confirmatory Western blot. They were also negative in a less sensitive HIV-1 Ab EIA (LS-EIA). This assay can distinguish recent HIV infection from chronic infection (21). The cutoff for the LS-EIA used to classify participants into groups for this study was 1.0, which identifies individuals as infected for <170 days (95% confidence intervals 162–183 days). Subjects were classified as belonging to group 2 (n = 11) if they began HAART at a time when they had positive results using the standard EIA but were negative using the LS-EIA. Subjects classified as group 3 (n = 5) were positive in both the standard and LS-EIA. For comparison, group 4 included six asymptomatic chronically infected subjects who were seropositive for a median 4.4 years (range 2–14 years).

Laboratory testing

HIV-1 Ab testing was performed at three university hospital sites. The LS-EIA was conducted at the University of California (San Francisco, CA) using the Vironostica HIV-1 EIA (Organon-Tecnika, Boxtel, The Netherlands). Western blot analysis for HIV-1 Abs and p24 Ag capture assays were performed at the Laboratoire de Santé Publique du Québec (St. Anne-de-Bellevue, Quebec, Canada). Plasma viremia was measured using the Roche Amplicor Assay (Roche Diagnostics, Mississauga, Ontario, Canada) with a detection limit of 500 HIV-1 RNA copies/ml of plasma. Plasma samples falling below the detection limit of this assay were retested using the ultrasensitive method (Ultradirect; Roche Diagnostics) with a detection limit of 50 copies/ml of plasma. T cell subset distribution was measured by flow cytometric analysis.

Cells and peptides

PBMCs were isolated by density gradient centrifugation (Ficoll-Paque; Pharmacia, Uppsala, Sweden) and frozen in 10% DMSO (Sigma-Aldrich, St. Louis, MO) in 90% FCS (Montreal Biotech, Montreal, Quebec, Canada). Subjects were typed for MHC class I Ag expression by the amplification refractory mutation system-PCR using 95 primer sets amplifying defined MHC class I alleles (ABC SSP Unitray; Pel-Freez Clinical Systems, Brown Deer, WI) (22). Genomic DNA for molecular HLA typing was prepared from either fresh blood or EBV-transformed B cell lines using the QIAamp DNA blood kit (Qiagen, Mississauga, Ontario, Canada).

Peptide selection

The HIV epitopes used for stimulation were chosen from the National Institutes of Health HIV molecular immunology database (23). Peptides of 9, 15, or 20 aa in length containing these sequences were obtained from the Medical Research Council AIDS Reagent Project (Hertz, U.K.) and the National Institutes of Health AIDS Research and Reference Reagent Program (Rockville, MD). Lyophilized peptides were diluted to 1 mg/ml in HBSS (Life Technologies, Grand Island, NY) containing 10% DMSO and stored at −70°C. They were used at a final concentration of 10 μM (20–40 μg/ml, depending on the peptide size).

ELISPOT assay for single-cell IFN-γ release

IFN-γ secretion by virus-specific cells was quantified by ELISPOT assay as described (19). Stimulating peptides were chosen on the basis of their having amino acid sequence binding motifs for the MHC class I alleles expressed by the person being tested. Stimulatory panels contained peptides restricted to multiple (two to five) MHC class I alleles per subject. Medium, containing the equivalent amount of DMSO as present in peptide stimulation conditions, was used as a negative control. Anti-CD3 Ab (Research Diagnostics, Flanders. NJ) was used as a positive control stimulus. Cells were plated at two concentrations (2 × 10⁵ cells/well and 5 × 10⁴ cells/well) for each peptide condition. For a subset of experiments, a pool of immunodominant HLA-A2- or HLA-B7-restricted EBV- and CMV-derived peptides was also used for positive control stimuli. The frequency of reactivity to anti-CD3 and the EBV/CMV peptide pool stimuli occurring in longitudinally collected samples was used to control for between-time point variability in cell responsiveness. Results are expressed as spot-forming cells (SFCs)/10⁶ PBMCs after subtraction of negative controls. Negative control stimulation produced less than five spots per well in >90% of experiments. The average number of spots in the negative control wells was 3.13 ± 3.12. In experimental wells, the signal was considered positive if at least 10 spots per well were present (>2 SDs above the mean), the number of spots obtained was proportional to the number of cells plated, and the number of spots per well was at least twofold greater than the negative control wells. The identity of IFN-γ secreting cells as CD8⁺ was confirmed by the reduction of SFC numbers after depletion of CD8⁺ cells with magnetic beads (Dynal, Lake Success, NY).

Intracellular cytokine staining

The frequency of Gag p55-specific CD4⁺ and CD8⁺ T cells was quantitated by intracellular cytokine staining. Frozen and thawed PBMCs were resuspended at 10⁶ cells/ml in RPMI 1640 containing 10% FCS, 2 mM l-glutamine (ICN Biomedical Canada, Costa Mesa, CA), 50 IU/ml penicillin (ICN Biomedical Canada), 50 mg/ml streptomycin (ICN Biomedical Canada), and 50 μM 2-ME (Sigma-Aldrich) and were stimulated with a pool of 122 HIV 15-mer peptides with 11 aa overlaps corresponding to the HIV-1 HXB2 Gag p55 sequence (National Institutes of Health AIDS Research and Reference Reagent Program). Staphylococcal enterotoxin B (Sigma-Aldrich) was used as the positive control; medium alone served as the negative control. Cells were incubated for 2 h at 37°C in 5% CO₂ with stimulatory peptides and 1 μg/ml each of anti-CD28 and anti-CD49d mAbs (BD Biosciences, Mississauga, Ontario, Canada). Brefeldin A (Sigma-Aldrich) was added at a final concentration of 10 μg/ml and incubated for an additional 4 h at 37°C in 5% CO₂. Cells were permeabilized with FACSPerm (BD Biosciences) according to the manufacturer’s direction. IFN-γ secretion by recently stimulated CD4⁺ and CD8⁺ T cells was detected by staining samples with FITC-conjugated anti-CD4, PE-conjugated anti-CD69, and PerCP-conjugated anti-IFN-γ mAbs (BD Biosciences) and with allophycocyanin-conjugated anti-CD8 (BD Biosciences) for 30 min in the dark. In parallel, control samples were left unstained and stained with Ig isotype control Abs (BD Biosciences). After washing, cells were resuspended in 1% paraformaldehyde (Polysciences, Warrington, PA) until four-color flow cytometric analysis was performed on a FACSCalibur instrument (BD Biosciences). Fifty thousand to 200,000 events were acquired and analyzed using CellQuest software (BD Biosciences). A response was considered positive if the frequency of CD69⁺IFN -γ⁺CD4⁺ or CD69⁺IFN-γ⁺CD8⁺ cells present after an antigenic stimulation was >0.1% over that seen with cells stimulated with no Ag.

Statistical analysis

Two-tailed unpaired t tests were used to assess between-group differences in age, CD4 count, CD8 count, and viral load at study entry as well asbetween-group differences in the number of HIV peptides used to screen for HIV-specific responses. Wilcoxon matched pair tests were used to assess in-group differences at study entry vs during therapy. Nonparametric Wald-Wolfowitz tests with adjusted z-score and corresponding adjusted p value for small sample size were used to evaluate the significance of between-group differences in the magnitude of HIV-specific responses. Fisher P-exact tests were used to test the significance of within-group and between-group differences in the breadth of HIV responses. The Fisher P-exact test with Bonferonni correction was used to assess between-group differences in the expression of common HLA alleles. Only p values <0.05 were considered significant.

Study population

A description of the study population is provided in Table I⇓. Subjects in PI were separated into three groups based on the interval between the presumed date of infection and start of HAART, as described in Materials and Methods. There were no differences between groups with respect to age, ratio of males to females, or their distribution in risk behavior categories. All study groups had similar CD4⁺ and CD8⁺ T cell counts at the time HAART was started and were followed over a similar period of time (Table I⇓). Viral load at entry was higher for the individuals in group 1 compared with those in groups 2 and 4. This is consistent with their being closer to acute infection at a time when viremia peaks before the establishment of a viral load set point (1, 2, 3, 4).

Table II⇓ provides information for each individual on the composition of HAART prescribed, the classification of each individual into the study groups described in Materials and Methods, estimated date of infection, clinical information and laboratory results used to derive an estimated date of infection, and expression of MHC class I alleles. The composition of the HIV peptide panels used to test individual subjects for HIV-specific responses was dependent on the MHC class I alleles expressed by the subject being tested. For this reason, the distribution of common MHC class I alleles, such as HLA-A1, A2, A3, A24, B7, B8, B35, and B44, between groups was compared. No significant skewing in the distribution of these alleles was evident between the groups in PI (p value was NS, Fisher P-exact test with Bonferronni correction), nor were between-group differences detected with respect to the distribution of alleles likely to be homozygous. Although all individuals in group 4 expressed HLA-A2, the inclusion of HIV peptides restricted to between two and five alleles per study subjects in each peptide screening panel makes the possibility that the results reported are due to skewed distribution of MHC class I alleles between groups unlikely.

For many of the within-group and between-group comparisons, an on-therapy time point was examined. The time from infection to which this on-therapy sample corresponds was the available time point closest to 12 mo from the start of treatment. This time point was a median of 348 days (range 184–429 days) from the start of HAART for group 1, 349 days (range 84–385 days) for group 2, 189 days (range 100–357 days) for group 3, and 350 days (range 266–362 days) for group 4. The time used as the on-therapy time point was not significantly different between groups (p = NS, unpaired t tests). Furthermore, decline in the breadth and magnitude of HIV-specific responses in study subjects in which this occurred after starting HAART was usually evident within 2 mo and corresponded to control of viremia.

HIV-specific effector responses

Panels of MHC class I-restricted HIV peptides were used to stimulate PBMC samples collected longitudinally from the time HAART was initiated. Table III⇓ lists the sequence and the MHC class I restriction specificity of peptides used in this analysis together with their sequence, location, and MHC class I restriction specificity. For each peptide stimulus, the frequency of IFN-γ-secreting cells generated per 10⁶ PBMCs was assessed in an ELISPOT assay. The magnitude of the response to the HIV peptide panel tested was measured by adding the number of SFCs/10⁶ PBMCs generated to each stimulus that induced above-background responses. The breadth of the responses was measured by calculating the number of positive responses in each peptide panel divided by the total number of peptides tested. Longitudinal changes in these parameters were assessed.

Fig. 1⇓A shows the changes in the magnitude and breadth of responses to an HIV peptide panel used to screen one subject from group 1. Subject MQPI005 began HAART 36 days from infection, at which time the subject was positive in a standard EIA but indeterminate in a confirmatory Western blot test and negative in an LS-EIA. This representative group 1 individual maintained the magnitude and the breadth of his response to the HIV peptide panel tested for up to 684 days on therapy, despite suppression of viremia soon after starting HAART. The five other members of group 1 similarly maintained or increased the magnitude and breadth of their HIV-1 response to the peptides tested throughout the follow-up period of 184, 358, 518, 535, and 630 days from the start of HAART, which successfully controlled viral load. A within-group comparison of the magnitude and breadth of responses at baseline vs those at the on-therapy time point closest to 12 mo revealed a similar breadth and magnitude on therapy vs that at baseline (Table IV⇓).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

Longitudinal assessment of HIV-specific effector responses to MHC class I-restricted panels of HIV peptides. Stacked bar graphs show changes with time in the number of SPCs per million PBMCs (SFCs/10⁶ PBMCs) generated after stimulation of lymphocytes from HIV-infected patients starting HAART at various times after infection with a panel of MHC class I-restricted HIV peptides. The x-axis indicates, for each bar, the number of days on HAART at the time tested. The legends list the HIV peptides tested for each study subject. Peptides are identified by the HIV (Figure legend continues) gene product from which they were derived. Also shown are changes in log₁₀ viral load keyed to the right-hand y-axis. Results are presented for one subject in group 1 (starting HAART before seroconversion) (A), 11 subjects in group 2 (starting HAART postseroconversion but in early HIV infection) (B–L), one subject in group 3 (starting HAART postseroconversion and in the second half of the first year of infection) (M), and one subject in group 4 (starting HAART in the chronic phase of infection) (N). The shaded areas in I, M, and N correspond with changes in reactivity to EBV-derived control peptides with time.

Fig. 1⇑, B–L, displays the longitudinal assessments of reactivity to panels of MHC class I-restricted peptides for all 11 individuals belonging to group 2 who began HAART early during infection but after seroconversion, as defined by their being negative in the LS-EIA. Therapy was associated with reduced viremia in this group as well. Two distinct patterns of change in HIV-specific responses were observed in this group after the start of HAART. Fig. 1⇑, B–F, shows results for five subjects who maintained responses over a follow-up of 230–815 days. Fig. 1⇑, G–L, shows results from all six individuals belonging to group 2 whose HIV-specific responses fell in intensity and narrowed in specificity after HAART was initiated and viral load was controlled. This decline occurred despite the maintenance of a stable frequency of cells specific for the EBV-derived HLA-B7-restricted peptide RPPIFIRRL (Fig. 1⇑I) and the EBV-derived HLA-B8-restricted peptide FLRGRAYGL (Fig. 1⇑J) (24). Comparison of the intensity and breadth of HIV peptide panel-specific responses at baseline to those at the time point closest to 12 mo on-therapy for subjects in group 2 showed a nonstatistically significant trend toward reduction in the magnitude of the response (p = 0.7, Wilcoxon matched pairs test), whereas the breadth of the response fell significantly from baseline to the on-therapy time point (p = 0.048, Fisher P-exact test) (Table IV⇑).

Group 2 subjects who maintained their HIV-specific responses started HAART a median of 97 days (range 69–138 days) from infection, whereas those who lost these responses began treatment a median 134 days (range 72–173 days) from infection. Although there was a tendency to maintain persistent effector responses after starting therapy in subjects treated at earlier vs later times after infection and seroconversion, this difference was not statistically significant (p = 0.27, unpaired t test). No differences were detected in the viral load at therapy initiation between the subgroup that maintained vs the one that lost HIV-specific effector activity.

Retrospective statistical analysis of group 2, categorized by whether they maintained (group 2a) or lost (group 2b) HIV-specific effector responses, revealed that responses were as intense and broad in subgroup 2a after 1 year of therapy as they were before starting HAART. In contrast, for subjects belonging to group 2b, the breadth of this activity at the closest available time tested to 12 mo on-therapy was lower than at baseline (p < 0.02, Fisher P-exact test). The magnitude of HIV-specific reactivity showed a nonsignificant trend toward a reduced response (p = 0.07, Wilcoxon matched pairs test) (Table IV⇑).

Fig. 1⇑, M and N, displays longitudinal analyses of HIV-specific reactivity for a group 3 subject (Fig. 1⇑M) and for a group 4 subject (Fig. 1⇑N). As has been described by others, persons starting HAART in the chronic phase of infection displayed reduced magnitude and breadth of HIV-specific responses after start of HAART in association with viral load suppression (25, 26, 27, 28). Possibly due to small group size, the reduction in the magnitude and breadth of the HIV responses from baseline to the closest available time point to 12 mo on-therapy was only statistically significant for breadth for group 3 (p < 0.001, χ² test) (Table IV⇑). No differences were detected for comparisons of the breadth and magnitude of HIV peptide panel-specific responses at baseline and 12 mo between groups 3 and 4. Therefore, we combined the study subjects in groups 3 and 4 into a single data set and assessed the significance of changes in breadth and magnitude of HIV-specific responses at baseline vs the on-therapy time points. For both measures, responses were reduced on-therapy compared with baseline (p = 0.02, Wilcoxon matched pairs test; p < 0.001, Fisher P-exact test).

All of the subjects in the PI study groups were recategorized according to whether they maintained or lost HIV-specific responses and were compared for within-group and between-group differences in HIV-specific effector responses. To determine whether viral load could account for whether an individual maintained or lost HIV-specific immune responses in this population, we compared viral load at baseline. No differences were detected between subjects who maintained (group 1 + 2a) vs those who lost (group 2b + 3) HIV-specific responses after starting HAART (Fig. 2⇓A). Comparisons of the breadth of responses to HIV peptide panels showed maintenance of the reactivity from baseline to the on-therapy time point for group 1 + 2a (p = NS, Fisher P-exact test) and a decline in responsiveness for group 2b + 3 (p < 0.001, Fisher P-exact test) (Fig. 2⇓B). HIV-specific responses in these two groups were similar at baseline but were significantly narrower at the time point closest to 12 mo on-therapy for groups 2b and 3 vs groups 1 and 2a (p < 0.001, Fisher P-exact test) (Fig. 2⇓B). Comparison of the magnitude also showed maintenance of the reactivity from baseline to the on-therapy time point for group 1 + 2a (p = NS, Wilcoxon matched pairs test) and a decline in responsiveness for group 2b + 3 (p < 0.03, Wilcoxon matched pairs test) (Fig. 2⇓C). HIV-specific reactivity appeared to be greater before starting HAART for subjects in group 1 + 2a compared with those for group 2b + 3, but this difference was not statistically significant (Fig. 2⇓C). At the on-therapy time point, between-group comparisons revealed that subjects in group 1 + 2a had a nonstatistically significant trend toward having a more intense response to the peptide panel tested than did subjects in group 2b + 3 (p = 0.08, Wald-Wolfowitz with adjusted z-score) (Fig. 2⇓C).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

Comparison of viral load, breadth, and magnitude of responses between subjects in PI who maintain or lose HIV-specific effector activity after initiation of HAART. Subjects in PI were separated based on whether they maintained or lost effector responses specific for a panel of major histocompatibility complex-restricted HIV peptides. Group 1 + 2a included six subjects starting HAART preseroconversion (group1) and five recently infected persons who started HAART postseroconversion (group 2a). Group 2b + 3 included six recently infected individuals who started HAART postseroconversion (group 2b) and five persons who began HAART in the second half of the first year of infection (group 3). Shown are scatter plots of viral load at baseline (A), breadth (percent of reactive peptides) of HIV-specific responses at baseline and on-therapy time point (B), and magnitude (number of SFCs per million PBMCs [SFCs/10⁶ PBMCs]) of HIV-specific responses at baseline and on-therapy time point (C). The on-therapy data were generated from a lymphocyte sample taken at the closest available time point to 12 mo on therapy, as described in Results.

Starting HAART in the acute phase of HIV infection has been shown to maintain HIV-specific CD4⁺ T cells, which are preferred targets for HIV infection (9, 11, 17). We reasoned, based on the observation that HIV-specific effector responses could be preserved in some individuals starting HAART after seroconversion, that this reflected the preservation of functional HIV-specific CD4⁺ T cell levels. Detection of intracellular IFN-γ production in the CD4⁺ and CD8⁺ compartments after stimulation with a pool of HIV-1 Gag p55 peptides was used to screen longitudinal samples from a subset of subjects from each group. All subjects tested from group 1 (24, 31, 54, and 58 days from presumed date of infection) maintained HIV-specific activity in both compartments up to 12 mo from start of HAART (Fig. 3⇓, B and C). Seven subjects from group 2 were tested in this manner. Three individuals classified as belonging to group 2a (MQPI001, MQPI009, and MQPI017; 69, 76, and 138 days from infection, respectively) maintained HIV-specific activity as measured by ELISPOT (Fig. 1⇑, B, D, and F). Four subjects from group 2 (MQPI012, MQPI013, MQPI014, and MQPI016; 83, 135, 146, and 173 days from infection, respectively) lost HIV-specific responses in association with viral control (Fig. 1⇑, H–J and L). By intracellular cytokine staining for IFN-γ HIV Gag p55-specific CD4⁺ and CD8⁺ cells persisted in all group 2a subjects (Fig. 3⇓, B and C). The decline in HIV-specific effector responses in group 2b individuals observed by ELISPOT assay was also seen within 3 mo of initiating HAART for HIV-1 Gag p55 CD4⁺ and CD8⁺ T cells (Fig. 3⇓, B and C). MQPI018 and MQPI021 from group 3 were estimated to be 272 and 279 days from infection when they began therapy. Both lost HIV-specific cell numbers in both the CD4⁺ and CD8⁺ T cell compartments as early as 2 mo from start of HAART (Fig. 3⇓D). The two subjects tested from group 4 (CH5 and CH7) also showed a decline in the frequency of their HIV-1 Gag p55-specific CD4⁺ and CD8⁺ cells to below background values within 1 mo of initiation of HAART (Fig. 3⇓E).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

Longitudinal assessment of changes in the frequency of recently stimulated IFN-γ-producing HIV-1 Gag p55-specific CD4⁺ and CD8⁺ T cells. A, Control and p55 pool-stimulated responses for subject MQPI006 at baseline. Bar graphs show the frequency (by percentage) of CD69⁺ PBMCs producing IFN-γ in response to a pool of 122 15-mer peptides with 11 aa overlaps corresponding to HIV-1 Gag p55. (Figure legend continues) A displays a representative dot plot showing at the top right of each plot the percentage of CD69⁺IFN-γ⁺CD4⁺ cells generated after a 6-h stimulation with no Ag (left) and the Gag p55 peptide pool (middle). Also shown is a dot plot of the CD69⁺IFN-γ⁺CD8⁺ cells generated after a 6-h stimulation with the Gag p55 peptide pool (right). Longitudinal samples were analyzed for HIV-specific activity in the CD4⁺ (B), and CD8⁺ T cell compartments (C) from four subjects from group 1, seven subjects from group 2, two subjects from group 3, and two subjects from group 4 stimulated with no Ag and the Gag p55 peptide pool. The frequency of IFN-γ-producing cells generated after stimulation with no peptide has been subtracted from the results shown. A frequency of 0.1% over background stimulation is considered a positive response.