|Home | About | Journals | Submit | Contact Us | Français|
Human immunodeficiency virus type 1 (HIV-1)-specific immune responses over the course of rapidly progressive infection are not well defined. Detailed longitudinal analyses of neutralizing antibodies, lymphocyte proliferation, in vivo-activated and memory cytotoxic T-lymphocyte (CTL) responses, and viral sequence variation were performed on a patient who presented with acute HIV-1 infection, developed an AIDS-defining illness 13 months later, and died 45 months after presentation. Neutralizing-antibody responses remained weak throughout, and no HIV-1-specific lymphocyte proliferative responses were seen even early in the disease course. Strong in vivo-activated CTL directed against Env and Pol epitopes were present at the time of the initial drop in viremia but were quickly lost. Memory CTL against Env and Pol epitopes were detected throughout the course of infection; however, these CTL were not activated in vivo. Despite an initially narrow CTL response, new epitopes were not targeted as the disease progressed. Viral sequencing showed the emergence of variants within the two targeted CTL epitopes; however, viral variants within the immunodominant Env epitope were well recognized by CTL, and there was no evidence of viral escape from immune system detection within this epitope. These data demonstrate a narrowly directed, static CTL response in a patient with rapidly progressive disease. We also show that disease progression can occur in the presence of persistent memory CTL recognition of autologous epitopes and in the absence of detectable escape from CTL responses, consistent with an in vivo defect in activation of CTL.
In the ongoing search for effective preventative vaccines and immunotherapy for human immunodeficiency virus type 1 (HIV-1) infection, much attention has been focused on the mechanisms by which long-term nonprogressors control infection (6, 7, 18, 36). Equally important is an understanding of why certain individuals infected with HIV-1 progress more rapidly than others. Viral (10, 22, 23, 27), host (11, 29, 37), and immunologic (16, 34, 39, 41) factors have been postulated to play important roles in determining the rapidity of progression to AIDS after infection with HIV-1.
Virus-specific immune responses are likely to play an important role in modulating disease progression. HIV-1-specific cytotoxic T lymphocytes (CTL) are considered to be important in reducing viral load and containing infection (2, 3, 13, 26). A broadly reactive, adaptable HIV-1-specific CTL response has been demonstrated in several groups of long-term nonprogressors (15, 16, 20, 24, 50), and it has been postulated that a narrowly directed, fixed CTL response may play a role in progressive infections (35); however, there are few detailed longitudinal studies of CTL epitope specificity. Clonal exhaustion due to continued high level of antigen has also been postulated to explain immune system failure in persons with chronic viral infections (31); however, this concept has yet to be convincingly demonstrated in human populations. Viral escape from immune system pressure has also been suggested to play a role in progressive infection. Generation of specific CTL responses has been associated with immune system pressure and viral escape during both acute (3) and chronic (13) HIV-1 infection, and it has been postulated that a narrowly directed and unadaptable initial CTL response may lead to rapid production of viral escape variants in persons with rapidly progressive infection. Lack of adequate CD4+-T-lymphocyte help may also contribute to rapidly progressive infection (41), but few studies have addressed the relationship between CD4 helper responses and CTL.
To characterize the immune system responses during rapidly progressive infection, we performed a longitudinal analysis of HIV-1-specific neutralizing antibodies, T-helper-cell function, CTL function, CTL epitope specificity, and the emergence of viral variants over the entire course of disease in an individual with rapidly progressive infection. The patient, who provided whole-blood samples approximately every 6 months, presented with an acute HIV-1 infection syndrome in November 1992, developed an AIDS-defining illness 13 months later, and died 45 months after the initial presentation. The patient’s CTL response was compared to that of a second individual with rapid disease progression. The immunodominant CTL response in both individuals was directed against a B7-restricted CTL epitope within gp41, and the CTL response did not broaden over time.
Patient 012-053i (designated RP1) was 29 years old when he presented to medical attention in November 1992 with a severe febrile illness associated with pharyngitis, lymphadenopathy, and aseptic meningitis. He reported having engaged in receptive anal intercourse approximately 1 month prior to presentation. Antibodies to HIV-1 and HIV-2 were undetectable by both enzyme-linked immunosorbent assay (ELISA) and Western blotting. The HIV-1 viral load, performed on banked serum, was 800,000 copies/ml, and the CD4 count was 310/mm3 (Fig. (Fig.1).1). On follow-up, 3 months after presentation, the subject was clinically well and declined antiretroviral therapy. The ELISA result was equivocal, and the Western blotting result was indeterminate, showing antibodies to HIV-1 p24 and p55. The viral load had fallen to 60,000 copies/ml, and the CD4 cell count had risen to 500/mm3. Eight months after presentation, the subject continued to be clinically well, with a stable CD4 cell count and a positive Western blot result with reactive bands at p24, p55, and gp160; however, the viral load had risen to 200,000 copies/ml. Thirteen months after presentation, with a CD4 cell count of 318 and a rising viral load, the subject developed Pneumocystis carinii pneumonia. In June 1994, therapy consisting of a combination of zidovudine and delaviridine was initiated. (Didanosine was added in June 1995.) Despite treatment, the CD4 cell count continued to fall and the viral load increased. Twenty-five months after presentation, he developed severe gastrointestinal cryptosporidiosis infection and had a CD4 cell count of 114. Thirty months after presentation, he developed renal insufficiency consistent with HIV nephropathy and hemolytic anemia thought to be due to increasing dapsone levels. The antiretroviral therapy was changed to stavudine, lamivudine, and indinavir; however, clinical deterioration continued with a falling CD4 cell count and rising viral load. He developed cutaneous Kaposi’s sarcoma, hypogonadism, and peripheral neuropathy. Forty-four months after presentation, he developed recurrent P. carinii pneumonia in association with severe adult respiratory distress syndrome and possible pulmonary Kaposi’s sarcoma. Despite aggressive treatment, his condition failed to improve. He died in August 1996, 45 months after presentation.
Comparison was made to a second patient with rapidly progressive HIV-1 infection. This patient, designated RP2, seronconverted to HIV-1 in 1986, developed clinical AIDS within 4 years, and also demonstrated persistent lack of control of viremia.
Autologous virus was isolated and expanded in culture by using the qualitative peripheral blood mononuclear cell (PBMC) macroculture assay technique. A total of 107 patient PBMC were cocultured at 37°C with 107 phytohemagglutinin (PHA)-stimulated donor PBMC in RPMI 1640 medium supplemented with 50 u of recombinant interleukin-2 per ml (rIL-2) and 20% fetal calf serum. On days 4, 7, 10, and 14, half the medium was removed without disturbing the cells and replaced with fresh medium. On day 7, an additional 107 PHA-stimulated donor PBMC were added to the culture. On day 14, the culture was terminated, the supernatant was removed and frozen, and p24 ELISA was performed. Positive supernatants were subjected to titer determination by the virus stock infectivity titer determination technique and the 50% tissue culture infective dose was calculated by the Spearman-Karber method. The viral phenotype was determined by the standard ACTG HIV syncytium-inducing (MT-2) assay (12a).
Neutralizing antibodies against heterologous laboratory strains of HIV-1 (IIIB, MN, and SF2) were measured in MT-2 cells in 96-well plates by monitoring a reduction in virus-induced cell killing as described previously (30). Briefly, cell-free virus (50 μl containing 1,000 50% tissue culture infective doses) was added to multiple dilutions of test plasma in 100 μl of growth medium in triplicate wells of 96-well microtiter plates and incubated at 37°C for 1 h. MT-2 cells (5 × 104 cells in 100 μl) were added to each well, and the plates were incubated until syncytium formation and virus-induced cell killing were observed in wells that did not contain antibodies (virus control wells). Cell viability was quantified with Finter’s neutral red in poly-l-lysine-coated plates. The percentage of viable cells was calculated by finding the difference in absorption at 540 nm between test wells (cells plus serum sample plus virus) and virus control wells (cells plus virus), dividing this by the difference in absorption between cell control wells (cells only) and virus control wells, and multiplying the result by 100. Neutralization titers were expressed as the reciprocal of the plasma dilution required to protect at least 50% of cells from virus-induced killing. Neutralization was measured at a time when virus-induced cell killing in virus control wells was greater than 70% but less than 100%.
Antibody-mediated neutralization of primary isolates was measured in PHA-stimulated PBMC by use of a p24 immunoassay to quantify virus production (38). Briefly, 50 μl of cell-free virus was incubated with 50 μl of diluted plasma sample in triplicate for 1 h at 37°C (the final plasma dilutions were 1:6, 1:18, 1:54, and 1:162). Next, 25 μl was transferred to corresponding wells of 96-well U-bottom plates containing 4 × 105 PHA-stimulated PBMC in 175 μl of IL-2 growth medium. An additional six wells of cells received an equivalent amount of virus that had not been incubated with a serum sample (virus control). The cells were incubated with virus or virus-plasma mixtures for 1 day at 37°C and then washed three times with 250 μl of IL-2 growth medium to remove the virus inoculum and antibodies. Washed cells were suspended in 250 μl of IL-2 growth medium and incubated at 37°C for the duration. Culture supernatants (25 μl) were collected daily, mixed with 225 μl of 0.5% Triton X-100, and stored at 4°C for later p24 measurements. This volume was replaced with 25 μl of fresh IL-2 growth medium at each collection. Viral p24 was quantified for harvest on a day when p24 production in virus control wells was >5 ng/ml and was in a linear phase of increase. Serum samples were considered positive for primary isolate neutralization if they caused >80% reduction in p24 relative to a negative control serum.
PBMC were isolated from heparinized whole-blood samples by Ficoll-Hypaque (Sigma Chemical Co., St. Louis, Mo.) density centrifugation. The PBMC were resuspended in RPMI 1640 medium (Sigma) containing 10% human AB serum (Sigma), 10 mM HEPES buffer, 2 mM l-glutamine, and 50 U of penicillin-streptomycin per ml (R-10 medium). The cells (105 cells/well) were cultured in 4 to 6 replicate wells of 96-well U-bottom plates in the presence of HIV-1 recombinant proteins, control proteins, tetanus toxoid (Connaught), or medium alone. Six days later, the cells were pulsed with [3H]thymidine (New England Nuclear, North Billerica, Mass.) at 1.0 μCi/well, and uptake was measured 6 h later with a TopCount Microplate Scintillation Counter (Packard Instrument Co., Meriden, Conn.). The HIV-1 p24 Gag and gp160 Env proteins (Microgenesys, Meriden, Conn.) are recombinant proteins that were derived from the NY-5 and LAV strains of HIV-1, respectively, and were produced in a baculovirus system with 90 to 95% purity. These proteins were tested over a range of concentrations, with 2 μg/ml used as the standard concentration. A mixture of baculovirus proteins were used as controls. Tetanus toxoid was used at 2 μg/ml. For the recombinant HIV-1 proteins, a stimulation index was defined as the ratio of the mean counts per minute (cpm) of the HIV-1 protein wells to the mean cpm of the control protein wells. For tetanus toxoid, the stimulation index was defined as the ratio of the mean cpm of the stimulated wells to the mean cpm of six control wells containing PBMC and medium alone.
PBMC were prepared from fresh whole-blood samples by Ficoll-Hypaque density gradient centrifugation. Bulk PBMC were used as effector cells in standard chromium release assays as previously described (49) at effector-to-target-cell ratios of 100:1, 50:1, and 25:1. Autologous Epstein-Barr virus-transformed B-lymphoblastoid cell lines (B-LCL) infected with recombinant vaccinia virus (VV; provided by Therion Biologics, Cambridge, Mass.) containing pol, gag, or env coding sequences from the IIIB strain of HIV-1 (VVpol, VVgag, and VVenv) were used as target cells. Autologous B-LCL infected with the New York City Board of Health (NYCBH) VV construct were used as negative controls.
Autologous EBV-transformed B-LCL (107 cells) were incubated overnight with recombinant VV containing the combined pol, gag, and env coding regions (vABT 489) at a multiplicity of infection of 3. Following 16 h of incubation at 37°C in 5% CO2, the cells were washed and resuspended in 5 ml of R-10. Psoralen (HRI Associates, Concord, Calif.) was added at a final concentration of 10 μg/ml, and the cells were added to a single well of a six-well flat-bottom plate, which was placed under a long-wave UV light source for 5 min. The cells were then washed three times and resuspended in R-20 medium (RPMI 1640 medium with 20% fetal calf serum) and placed in a flask with 1 × 106 freshly thawed PBMC and 4 × 107 irradiated allogeneic PBMC, isolated from seronegative donors. One day later, 10 ml of R-10/50 (R-10 with 50 U of rIL-2 per ml, kindly provided by M. Gately, Hoffman-LaRoche, Nutley, N.J.) was added, and the cells were incubated for 14 days with twice-weekly feedings of R-10/50. Expanded cells were used as effector cells in standard chromium release assays at effector-to-target-cell ratios of 100:1, 50:1, and 25:1. Autologous transformed B-LCL infected with recombinant VVgag (vABT 141), VVpol (vABT 204), VVenv (vABT 299), or NYCBH construct (control) were used as target cells (14).
Thawed cryopreserved PBMC were plated at limiting dilution (100/well, 50/well, 25/well, and 5/well) and cultured with allogeneic irradiated PBMC at of 2 × 105 cells/well in a final volume of 200 μl/well of R10/100 (R10 medium with 100 U of rIL-2 per ml). The CD3-specific monoclonal antibody, 12F6, was added at 0.1 μg/ml as a stimulus for T-cell proliferation (47). Plates were incubated for 2 weeks in 5% CO2 at 37°C and fed twice weekly with medium exchanges. Wells showing growth after 14 to 21 days were transferred to 24-well plates and restimulated with 106 irradiated allogeneic PBMC and 12F6. Clones thus obtained were then used as effector cells in standard cytotoxicity assays with autologous B-LCL infected with recombinant VV vectors, VVgag, VVpol, VVenv, and VVnef, as target cells (14).
HLA typing was performed by the Massachusetts General Hospital Tissue Typing Laboratory, using a standard lymphocytotoxicity assay. Molecular typing of HLA-A*0201 was performed as described previously (4).
The epitopes targeted by the CTL response were determined initially by using autologous B-LCL infected with recombinant VV vectors containing truncated HIV-1 gene inserts as target cells (19, 48). After the general region of specificity was determined, peptides spanning the region were synthesized and used to sensitize autologous B-LCL at 10 μg/200 μl. After 1 h of incubation and labeling with 50 to 100 mCi of Na251CrO4 (New England Nuclear, North Billerica, Mass.), the cells were washed and used as targets in standard chromium release assays. Fine mapping was achieved by synthesizing smaller overlapping peptides until the minimum sensitizing peptide was identified.
Peptides were diluted serially 1:10 to determine the lowest peptide concentration required for sensitization of target cells. The log dose-response relationship was modeled with the median-effect equation: percentage of lysis/lysismax = 1/[1 + (SD50/peptide)m], in which lysismax is the expected percentage of lysis at saturating doses of peptide, SD50 is the sensitizing dose of peptide required to achieve one-half of lysismax, and m is the slope of the function (8, 32). All statistical analyses were performed with the Statistica for Windows 5.1 software package (Statsoft, Tulsa, Okla.). The optimal epitope was defined as the peptide which sensitized target cells for lysis at the lowest peptide concentration (44, 45). To determine the stability of peptide-target cell binding, synthetic peptides were incubated with autologous EBV-transformed B-LCL at 10 μg/200 μl for 1 h, washed once with R-10, and placed in 24-well plates in R-20 for periods up to 96 h before chromium labeling and use as target cells.
Synthetic peptides were synthesized as free acids on an automated peptide sequencer (model 432A; Applied Biosystems, Foster City, Calif.). Lyophilized peptides were reconstituted at 2 mg/ml in sterile distilled water plus 10% dimethyl sulfoxide (Sigma) and 1 mM dithiothreitol (Sigma).
Proviral DNA was extracted from cryopreserved PBMC (Puregene DNA isolation kit; Gentra Systems, Inc.), and the region flanking the Env CTL epitope (Env amino acids [aa] 843 to 851) was amplified with primers 5F8514 (5′-GCT ACC ACC GCT TGA GAG ACT T) and 3R8864 (5′-GCT CCC ACC CCA TCT GCT [positions numbered according to the pNL4-3 sequence in reference 1]). The PCR mixture contained 3.3× XL buffer (Perkin-Elmer), 1.5 mM magnesium acetate, 200 mM each deoxynucleoside triphosphate, 25 pmol of each primer, and 2 U of rTth DNA polymerase (XL DNA PCR; Perkin-Elmer), and the following cycling conditions were used: 94°C for 1 min, followed by 25 cycles of 94°C for 10 s, 55°C for 30 s, and 68°C for 45 s, and a final extension at 72°C for 10 min. The PCR products were cloned (TA cloning system; Invitrogen), and the clones were cycle sequenced with a fluoresceinated primer, 5F8683 (5′-TGA GGG GAC AGA TAG GGT TA), and modified Taq polymerase (ThermoSequenase; Amersham) on an automated sequencer (ALF; Pharmacia). Data were analyzed on a Macintosh computer (Sequencher; Gene Codes). Amplification of the region including the Pol epitope (Pol aa 156 to 164) was performed with primers 5F2027 (5′-GTA GGA AAT GTG GAA AGG AAG GAC ACC AAA T) and 3R4150 (5′-GCT AGG TAG ACC TTT TCC TTT TTT ATT A) in a reaction mixture containing 3.3× XL buffer, 0.8 mM magnesium acetate, 200 mM each deoxynucleoside triphosphate, and 2 U of rTth DNA polymerase, and the following cycling conditions were used: 94°C for 1 min followed by 25 cycles of 94°C for 15 s, 62°C for 30 s, and 68°C for 90 s, and a final extension at 72°C for 10 min. In addition, HIV-1 RNA was extracted from plasma samples (QIAamp viral RNA kit; Qiagen), and the pol region was reverse transcribed with primer 3R4226 (5′-TGG GCC TTA TCT ATT CCA TCT AAA AAT AGT) and Superscript II reverse transcriptase (Gibco BRL) and amplified by nested PCR (12). RNA PCR products were separated from primers and nucleotides (QIAquick PCR purification; Qiagen) and directly sequenced.
Rapid disease progression has been associated with early emergence of T-tropic (syncytium-inducing) virus. Autologous viral samples obtained from plasma supernatants were evaluated for the syncytium-inducing (SI) or the non-syncytium-inducing (NSI) phenotype by the standard MT-2 assay. Seven viral samples obtained at approximately 6-month intervals displayed the NSI phenotype. At no time point was the SI phenotype seen. Virus isolates from these same time points readily infected PHA-stimulated PBMC as well as autologous CD4 cells which had previously been cleared of virus; however, the titers obtained were lower than those observed with IIIB or pNL4-3 laboratory isolates (data not shown). A 32-bp deletion within the gene coding for CCR5, the “second receptor” for macrophage-tropic strains of HIV-1, has been associated with decreased susceptibility to infection in individuals who are homozygous for the mutation (11) and a decreased rate of disease progression in patients who are heterozygous. Analysis of the CCR5 gene of the patient in our study showed that he was homozygous for the wild-type receptor.
Neutralizing-antibody responses have been postulated to contribute to nonprogressive HIV infection. In this subject, we examined neutralizing-antibody titers to three heterologous laboratory strains of HIV-1 as well as to sequential autologous viral isolates by using plasma samples obtained at eight different time points (Table (Table1).1). The patient showed an initial antibody response that was capable of neutralizing the heterologous laboratory strains, MN, SF2, and IIIB; however, this response waned over time. Only a single positive neutralization titer was seen against autologous virus and occurred with the latest plasma sample assayed against the earliest viral isolate. The overall results are consistent with poor seroconversion.
Previous studies have demonstrated a negative correlation between HIV-1 viral load and p24-specific T-helper-cell responses (41). In our patient with a rapidly progressing infection, no HIV-1-specific lymphocyte proliferation was detected at any time point tested. Three months after presentation, proliferation to the recall antigen, tetanus toxoid, was detected, and a vigorous response to mitogen (PHA) was noted; however, there was no detectable proliferation following stimulation with either p24 or gp160 antigen (Fig. (Fig.2).2). Nineteen months following presentation, the response to tetanus toxoid was lost, proliferation in response to mitogen had decreased, and HIV-1-specific proliferative responses remained undetectable. Proliferation in response to mitogen continued to decrease throughout the patient’s disease course and eventually was lost completely. These data indicate that the patient either failed to make a significant HIV-1-specific CD4+-T-helper-cell response or had lost it within 3 months of presentation with the acute HIV-1 infection syndrome.
The presence of in vivo-activated HIV-1-specific CTL was determined by using fresh PBMC preparations as effector cells in standard chromium release assays. Activity was tested against the Gag, Pol, and Env regions of HIV-1 by using autologous B-LCL infected with recombinant VV as target cells. Three months after presentation, at a time when the viral load was at its nadir (60,000 copies/ml), vigorous CTL activity was detected against Pol and Env regions (Fig. (Fig.3a).3a). At 19 months after presentation, weak Env-specific activity was still detectable but only at high effector-to-target-cell ratios; and by 21 months after presentation, no in vivo-activated HIV-1-specific CTL were detected. These data indicate that the subject generated an initial strong CTL response in the absence of detectable virus-specific helper cell responses but that the initial CTL response was narrowly directed and declined coincident with a rise in viral load and fall in CD4 cell count. Furthermore, they show that the dominant in vivo-activated response never included recognition of Gag proteins, despite expression of the HLA class I alleles (HLA A*0201,19 and HLA B7,18) for which responses to Gag proteins have been reported (5).
Cryopreserved samples of PBMC from multiple time points were stimulated in vitro with inactivated autologous B-LCL infected with a VV construct expressing HIV-1 Gag, Pol, and Env proteins. Stimulated cells were utilized as effectors in standard chromium release assays. Three months after presentation, high levels of Env- and Pol-specific CTL activity were detected (Fig. (Fig.3b).3b). This response attenuated over time, and by 39 months after presentation, no significant HIV-1-specific memory CTL activity could be detected by this approach; however, HIV-1-specific CTL clones could be generated by in vitro stimulation of PBMC until the time of the patient’s death, indicating that these cells were still capable of expansion and effector function. Clones obtained from PBMC at 45 months continued to be propagated in culture for up to 18 months, indicating an ability of these CTL to expand in response to appropriate stimuli. As noted in the evaluation of the in vivo-activated CTL response, the detectable memory CTL response was directed at only the Pol and Env regions of HIV-1 and did not broaden over time. These data indicate that despite ongoing viremia, the patient maintained a narrowly focused CTL response. Moreover, they suggest a defect in in vivo activation and/or expansion as memory CTL were clearly present but could be overcome by in vitro propagation and stimulation. Similar results were found in a second patient with rapidly progressive infection (RP2), in whom longitudinal analysis of bulk CTL activity revealed CTL to Env alone without evidence of a broadening response over time (data not shown).
Having identified a CTL response to Pol and Env regions of HIV-1, we next determined the precise epitopes targeted within these proteins. For these studies, we tested samples obtained from the time of presentation with the acute HIV-1 infection syndrome until the time of death. Established clones were screened for recognition of target cells expressing Gag, Pol, Env, and Nef proteins. The percentage of CTL clones obtained by standard cloning techniques which were confirmed to be HIV-1 specific peaked 3 months after presentation and declined steadily thereafter. Although no formal qualitative analysis was performed, the percentage of wells with HIV-1-specific CTL peaked at 20% at the time of seroconversion and fell to 2% at the time of death. This peak was coincident with the nadir of the viral load and initial clinical improvement. As disease progressed and the viral load rose, the percentage of CTL clones which could be detected declined, but the specificity of these responses remained stably targeted at the same two epitopes.
Of over 400 expanded clones tested, more than 50 HIV-specific CTL clones were obtained over the course of the patient’s infection, 41 of which were directed against the gp41 region of Env. The other 9 clones were directed against Pol and were detected in different amounts throughout the course of infection. No Gag- or Nef-specific clones were identified at any time. HIV-1-specific CTL clones could be detected in declining numbers until the time of the subject’s death, and these clones were active against peptide-sensitized as well as VV-infected autologous B-LCL. Similar analysis of the second rapid progressor (RP2) revealed that HIV-1-specific CTL clones in this individual targeted only the Env protein.
We used these established clones to determine the precise epitopes targeted by the CTL response. Use of overlapping peptides to map CTL clone specificities demonstrated that the immunodominant Env response in patient RP1 was targeted against an optimal 9-aa peptide within the gp41 region of Env, IPRRIRQGL (aa 843 to 851), which was HLA-B7 restricted (data not shown). Remarkably, CTL clones from RP2 targeted the identical epitope and only this epitope, which was also HLA-B7 restricted. Analysis of the activity of Env-restricted CTL clones from both patients showed similar, efficient lysis of B-LCL even at low peptide concentrations (Fig. (Fig.33c).
Epitope mapping of the subdominant Pol response in subject RP1 showed that all Pol-specific clones were directed against the 9-aa peptide SPAIFQSSM (aa 156 to 164) and were also B7 restricted. Both this and the Env epitope fit the predicted motif for B7 with proline at position 2 and an aromatic or hydrophobic residue at position 9 (43).
The CTL response remained remarkably narrow over time in both rapid progressors. No CTL responses restricted by any of the other class I alleles expressed by either individual were detected. This included the lack of HLA-A*0201 restricted CTL responses in RP1, including responses to the Gag protein, SLYNTVATL (p17 aa 77 to 85), which have been found in the majority of A*0201-positive persons (4). For RP1 and RP2, CTL specificities were unchanged from the time of presentation until the development of AIDS. These data support findings from bulk CTL assays showing that both rapid progressors generated a narrowly directed initial CTL response to HIV-1 which did not broaden over time.
Recent studies have shown a correlation between A*0201-restricted CTL to an epitope within Gag-p17 and control of viremia. Approximately, 70% of A*0201-positive HIV-infected individuals make a CTL response against this epitope. Because of the surprising lack of Gag-specific CTL in RP1, who expressed the HLA-A*0201 allele, we sought to determine if such activity was indeed present but masked by the vigorous immunodominant Env-specific CTL response. Bulk PBMC obtained at two time points, 3 and 45 months after presentation, were stimulated with autologous B-LCL expressing the peptide SLYNTVATL in the presence of IL-2 and feeder cells. The expanded cells were then used as effectors in standard chromium release assays. Despite specific stimulation with this peptide, no Gag-specific or SLYNTVATL-specific CTL activity was seen at either time point (data not shown), making it unlikely that small amounts were present but “overgrown” by the dominant Env-specific CTL. Sequencing of the patient’s PBMC for mutations within the Gag epitope showed a stable mutation, SLYNTVATL→SLFNTVAVL; however, this mutation is recognized well by SLYNTVATL-specific CTL (4). In addition, the patient’s own cells were able to present this epitope for CTL recognition by SLYTNVATL-specific CTL clones from other donors (data not shown).
To determine whether the antiviral CTL response was associated with immune system selection pressure, we examined autologous virus for evidence of CTL escape mutations. Proviral DNA obtained from PBMC from RP1 at 3, 25, and 45 months after presentation was sequenced in the regions of the Env and Pol epitopes recognized by the subject’s own CTL. In addition, plasma samples from identical time points were sequenced in the area of the Pol epitope. Three months after presentation, a single amino acid substitution in Env (IPRRIRQGL→IPRRTRQGL) was noted in 8 of 10 sequences obtained (Fig. (Fig.4a).4a). Twenty-five months after presentation, this variant and the index peptide were undetectable. The majority of sequences (8 of 10) at this time had a single terminal amino acid substitution (IPRRIRQGL→IPRRIRQGF). Of 10 sequences, 1 showed a second amino acid substitution (IPRRILQGF) and another 1 had both C- and N-terminal substitutions (IPRRIRQGL→VPRRIRQGF). Terminally (45 months after presentation), the variant showing both C- and N-terminal mutations (VPRRIRQGF) was codominant with the variant expressing a single terminal phenylalanine substitution (IPRRIRQGF). Mutations did not affect the B7 binding motif, which includes a stringent requirement for proline in position 2 and allows for a variety of hydrophobic and aromatic residues in position 9 (43). Sequencing was carried out to 10 amino acids flanking the established 9-amino-acid epitope, and no substitutions were noted in the flanking sequences (data not shown).
Synthetic peptides based upon these in vivo viral variants were prepared and used in decreasing amounts to sensitize autologous B-LCL. Env-specific CTL clones obtained from the time of presentation, seroconversion (3 months after presentation), and terminally (45 months after seroconversion) were used as effectors in chromium release assays. Clones from all time points showed identical recognition of the variants (Fig. (Fig.5).5). These data indicate that the Env-specific CTL response was stable over time and indicate that the ability of the detectable CTL response to recognize variant viruses did not change.
Peptide titration assays, as described above, demonstrated that the most stable in vivo variants, IPRRIRQGF and VPRRIRQGF, were as well recognized by Env-specific CTL as were the optimal index peptide (Fig. (Fig.5).5). The dominant variant present 3 months after presentation was not recognized by these CTL; however, it did not come to dominate the in vivo population and was undetectable 25 months after presentation. A second, minor variant present at the 25-month timepoint, IPRRILQGF, was slightly less well recognized by Env-specific CTL than was the index peptide; however, this variant also did not come to dominate the in vivo viral population.
Sequencing of the region of Pol recognized by the subject’s own CTL showed a single, short-lived substitution 3 months after presentation only in plasma samples (SPAIFQSSM→SPSIFQSSM). This variant was as well recognized as the original epitope by Pol-specific CTL obtained from samples 45 months after presentation. A stable substitution was noted 25 months after presentation and terminally in both PBMC and plasma samples (SPAIFQSSM→SPAIFQCSM). This variant was slightly less well recognized than was the index Pol peptide based on SD50 calculations of 0.2 (variant) versus 0.02 (index) (data not shown). Together, these data suggest that immune system escape by viral epitope variation did not play a role in the rapid disease progression in the patient under study.
Previous studies have shown that the optimal viral peptides recognized by CD8+ T lymphocytes bind very tightly to their restricting major histocompatibility complex (MHC) allele and form exceedingly stable complexes with t1/2 of 200 to 600 h (45), whereas mutations associated with immune system escape may fail to bind for more than 2 h (13). To determine if the detectable in vivo variants bound less durably to the target cell, the index Env peptide and the two dominant variants (IPRRIRQGF and VPRRIRQGF) were incubated with autologous B-LCL for 1 h, washed, and allowed to incubate in culture plates for up to 4 days, at which time they were washed again, labeled with 51Cr, and used as targets in standard chromium release assays. All three peptides remained bound to B cells for up to 16 h after incubation (Fig. (Fig.6),6), as demonstrated by maximal lysis with CTL clones. At 48 h after incubation, both the index peptide and terminal codominant variant, IPRRIRQGF, sensitized target cells for lysis equally well while target cells incubated with the other terminal codominant variant, VPRRIRQGF, were less well recognized. By 96 h after incubation, none of the target cells were efficiently lysed. Thus, one of the codominant in vivo viral variants appears to have formed as stable an MHC-peptide combination as the index peptide. While the other codominant Env variant appeared to have dissociated somewhat more rapidly, it was able to sensitize B cells for up to 16 h after incubation. Together, these data indicate that the dominant virus variants present in vivo were still recognized by the endogenous CTL response, arguing against viral escape as a primary mechanism of immune system evasion in this individual. Moreover, since recognized virus was present at high titer throughout the course of this patient’s illness, the data argue against clonal exhaustion contributing to immune evasion.
This longitudinal dissection of the cellular immune response in a person who died within 45 months of presenting with the acute HIV-1 infection syndrome provides important insights into virus-specific cellular immune responses during rapidly progressive infection. Despite strong in vivo activated CTL responses at the time of seroconversion, viremia was not contained and progressive CD4 cell decline ensued. Although strong CTL responses were observed at the time of lowest viremia, increasing viremia was not associated with the development of detectable CTL escape mutations in the immunodominant Env epitope, and the dominant CTL responses in the earliest stages of infection remained present in vivo until the time of death. CTL responses declined progressively in peripheral blood; however, these CTL could be readily expanded in vitro to large numbers, arguing against senescence and exhaustion as contributors to the CTL decline. In addition, no new epitopes were targeted as the disease progressed. Together, these data suggest a defect in in vivo activation of CTL as a contributor to disease progression. The lack of detectable HIV-1-specific CD4 proliferative responses and the lack of broadening of the CTL response over time provide further support for this conclusion.
The results of this study demonstrate a strikingly static CTL response over the course of progressive infection. The initial detectable CTL response was directed against a single epitope and broadened to target a second epitope within 3 months of presentation. However, despite ongoing viral replication, the response never broadened further and these two responses could still be detected at the time of death. Clones obtained at the time of initial presentation with the acute HIV-1 infection syndrome were identical in epitope specificity and in their ability to recognize in vivo virus variants, as demonstrated by experiments with limiting peptide concentrations. Furthermore, both of the epitopes were presented in the context of HLA B7. No other class I alleles were identified to be involved in presenting viral proteins for CTL recognition. Importantly, this included the lack of detectable CTL responses to well-defined epitopes presented by HLA-A*0201, which are targeted by more than 70% of A*0201-positive persons (4). CTL directed against the dominant HLA-A*0201 Gag epitope are potent inhibitors of HIV-1 replication in vitro and have been associated with control of viremia in vivo (33, 52). Epitope competition is not likely to have hindered the presentation of other epitopes (42), since we have observed A2-restricted responses in other persons who coexpress HLA B7 and target the same gp41 epitope as our subject (unpublished data). The functional CTL data presented extend studies based on T-cell receptor TCR analysis suggesting that a narrowly directed initial response is associated with more rapidly progressive infection (21, 34). The finding of a second patient with rapidly progressive infection and a monospecific response to the HLA-B7-restricted epitope within gp41 further suggests that the specificity of the initial response as well as its breadth may play a role in determining the rate of disease progression. The present results also indicate that narrowly directed initial CTL specificities can be maintained throughout the course of progressive infection. The inability of these patients to form CTL with new specificities may be related to the phenomenon of original antigenic sin, as has been described in mouse models (25).
These longitudinal studies also provide evidence that viral escape from CTL recognition is not a requirement for disease progression. Consistent with a study of acute infection (2), we noted sequence variation within CTL epitopes, consistent with selection pressure mediated by CTL. However, within the limitations of the assays performed, these mutations did not confer escape from recognition by established CTL responses. Observed mutations did not alter the binding motif which has been suggested to play a role in effective immune system escape (9), and CTL clones recognized epitopes representing dominant in vivo virus variants as well as wild-type virus. Moreover, some species that were detected as escape variants never became dominant in vivo. Thus, we conclude that mutations within CTL epitopes did not account for the inability of the CTL response to control viremia.
The continued presence of CTL in the peripheral blood from the time of presentation until the patient’s death argues against clonal exhaustion as a mechanism for immune system evasion. CTL clonal exhaustion, resulting from chronic exposure to high antigen levels, has been shown to result in immune system escape in a murine model of lymphocytic choriomeningitis virus infection (31), but its role in HIV-1 infection is not clear. Clearly, CTL were not completely lost in this rapid progressor as his disease progressed, since HIV-specific CTL of the identical epitope specificity and HLA restriction were readily obtainable from PBMC harvested as late as 48 h before his death. Such persistence of CTL clones has been observed in other studies, with individual clones being detected for up to 5 years (20, 21). Sequencing of the T-cell receptors of these clones will be necessary, however, before it can be stated definitively that clonal exhaustion did not occur in this case. Additionally, early data on soluble-factor production by the immunodominant Env-specific CTL clones suggest that they are capable of producing large quantities of soluble inhibitory factors when stimulated, arguing against an intrinsic defect in effector function in these activated clones.
The continued presence of CTL able to recognize autologous virus, along with the lack of broadening of the CTL response, suggests that a central immunologic defect in this subject was lack of in vivo activation of CTL. CD4+ helper T cells are known to contribute to the maturation of virus-specific CTL and have recently been shown to be important in the control of viremia (17, 41). Lack of a detectable CD4 proliferative response to HIV-1 antigens suggests that lack of T-helper-cell function may have played an important role in preventing in vivo activation of HIV-specific CTL and additionally in preventing the formation of CTL targeted against a larger array of epitopes. Findings here are consistent with murine models of LCMV infection in mice lacking CD4+ T lymphocytes. In these models, CTL are vigorously induced early in infection but do not persist in the absence of help, and viremia is not controlled during the chronic phase of infection (28, 46). This is precisely what was observed in our study in that a strong CTL response was observed initially but declined rapidly and virus breakthrough occurred.
These data provide a rationale for efforts to augment immunity in HIV-1-infected individuals by broadening the CTL response to include multiple epitopes and augmenting T-helper-cell responses. Earlier studies suggested that CTL directed against Gag regions of HIV-1 are associated with less rapid progression (24, 40), while CTL directed against Env and Pol epitopes may be less efficient at inhibition of HIV-1 replication in in vitro models (51). Further comparison of rapidly progressive HIV-1 infection to nonprogressive infection may therefore allow a more specific targeting of vaccine strategies to ensure that augmented immune responses are directed against a variety of epitopes which most efficiently inhibit viral replication.
We thank Eric Rosenberg for his help with the lymphocyte proliferation assays, David Wong for his help with the statistical analysis, Christian Brander for molecular typing of HLA-A*0201, and Daryld Strick for performing the MT-2 assays. We also thank Martin Hirsch and Philip Goulder for critical reviews of the manuscript.
This work was supported by NIH grants R37 AI28568, R01 AI40873, AI29193, AI36611, and P01 AI42454 and an individual NRSA, 1F32 AI09822-01, awarded to C.M.H.