|Home | About | Journals | Submit | Contact Us | Français|
Control of HIV-1 replication following nonsterilizing HIV-1 vaccination could be achieved by vaccine-elicited CD8+ T-cell-mediated antiviral activity. To date, neither the functional nor the phenotypic profiles of CD8+ T cells capable of this activity are clearly understood; consequently, little is known regarding the ability of vaccine strategies to elicit them. We used multiparameter flow cytometry and viable cell sorts from phenotypically defined CD8+ T-cell subsets in combination with a highly standardized virus inhibition assay to evaluate CD8+ T-cell-mediated inhibition of viral replication. Here we show that vaccination against HIV-1 Env and Gag-Pol by DNA priming followed by recombinant adenovirus type 5 (rAd5) boosting elicited CD8+ T-cell-mediated antiviral activity against several viruses with either lab-adapted or transmitted virus envelopes. As it did for chronically infected virus controllers, this activity correlated with HIV-1-specific CD107a or macrophage inflammatory protein 1β (MIP-1β) expression from HIV-1-specific T cells. Moreover, for vaccinees or virus controllers, purified memory CD8+ T cells from a wide range of differentiation stages were capable of significantly inhibiting virus replication. Our data define attributes of an antiviral CD8+ T-cell response that may be optimized in the search for an efficacious HIV-1 vaccine.
Recent human immunodeficiency virus type 1 (HIV-1) vaccine clinical trial results have put an increasing emphasis on the need for definition of the qualities of HIV-1-specific CD8+ T-cell responses required to lower virus set point and maintain long-term virus control. While some aspects of CD8+ immune responses have been reproducibly associated with virus control, analysis of a recent HIV-1 vaccine efficacy trial in which enhanced acquisition of HIV-1 occurred (5), underscored the limitations in our understanding of how surrogate T-cell functional markers relate to CD8+ T-cell response rate and vaccine efficacy. Several lines of evidence show that CD8+ T-cell responses are key determinants of disease progression early after infection. During human acute HIV-1 infection and acute simian immunodeficiency virus (SIV) infection in nonhuman primates, CD8+ T-cell responses have been associated with the initial control of viremia (4, 11, 21). Moreover, a recent study demonstrated rapid appearance of cytotoxic T-lymphocyte (CTL) escape mutations in the transmitted/founder virus during acute infection (10). Thus, rapid and durable elicitation of potent HIV-1-specific CD8+ T-cell responses following vaccination from which the transmitted/founder virus cannot escape will likely be critical for extinguishing virus replication in vivo and for control of viral load in the absence of sterilizing immunity.
The precise attributes of CD8+ T cells that suppress viral production in vivo are not defined. A detailed understanding of the properties of CD8+ T cells that correlate with virologic control will inform vaccine development by focusing on strategies that elicit functional cellular responses. In this regard, it is critical to determine the phenotypic and functional properties of CD8+ T cells that can mediate viral suppression. In HIV-1-infected subjects, measurements of T-cell polyfunctionality (1, 3), Gag specificity (6, 20), and the ability to suppress virus replication (19) have all been associated with virus control and slower disease course in HIV-1+ individuals. However, the direct relationships between CD8-mediated antiviral activity and specific effector functions and CD8+ T-cell differentiation stage have not been defined.
Polyfunctional CD8+ T cells are defined as those cells that have multiple effector functions, including degranulation (measured by CD107a) and production of cytokines/chemokines such as gamma interferon (IFN-γ), tumor necrosis factor alpha (TNF-α), interleukin-2 (IL-2), and macrophage inflammatory protein 1β (MIP-1β) (3). Several studies have characterized the polyfunctional T-cell responses in HIV-1-infected subjects as a surrogate of T-cell efficacy (22). It remains unknown which subsets of T-cell functions are associated with virus inhibition and, importantly, whether or not polyfunctional T cells elicited by vaccines have increased capacities to inhibit viral replication. Gag-specific or total CD8+ T-cell responses have also been associated with the ability of CD8+ T cells to control plasma viremia in elite controllers (20). Thus, it is critical to determine whether vaccine-elicited Gag- or Env-specific CD8+ T cells have the functional capacity to inhibit HIV-1 replication.
While the presence of more differentiated cells may allow for a more rapid response to pathogen, durable protection in the setting of vaccination may also require generation of a substantial memory pool. Evaluation of vaccine strategies eliciting antiviral CD8+ T-cell function must include determination of the phenotypic profiles of these cells as well as an understanding of whether their differentiation stage defines their capacity to inhibit virus.
Suppression of HIV-1 replication by specific CD8+ T cells can be mediated either through major histocompatibility complex (MHC)-restricted cytolytic activity (16, 24, 26) or through nonlytic mechanisms of virus inhibition (e.g., MIP-1β and IFNs) (7, 23, 25). The measurement of virus inhibition ex vivo, using infected primary autologous CD4+ T lymphocytes as targets, encompasses the totality of the potential inhibitory mechanisms (direct target cell lysis as well as cytokine or β-chemokine release) by CD8+ T cells. Thus, the virus inhibition assay gauges the overall potency of the CD8+ cell-mediated antiviral response elicited by infection or vaccination.
In this study, we used polychromatic flow cytometry and a CD8 virus inhibition assay to examine the relationship between phenotypic and functional subsets of HIV-1-specific CD8+ T cells and the ability to mediate direct virus inhibition in vitro. We found that a multiclade DNA prime/recombinant adenovirus type 5 (rAd5) boost vaccine (14) elicited CD8+ T cells with antiviral activity that was best associated with expression of CD107a and MIP-1β. Elicitation of this antiviral activity did not require maturation to the late stages of differentiation, suggesting that vaccination in the absence of chronic antigen stimulation is sufficient for eliciting CD8+ T-cell-mediated virus inhibition.
Vaccinees were enrolled as healthy volunteers under Institutional Review Board (IRB) approval through the Vaccine Research Center (VRC) (NIH, NIAID, Bethesda, MD) protocols VRC008 and VRC011. Thirty-one donors from VRC008 (3 DNA vaccinations with clades A and B Gag, Pol, and Nef and rAd5 boost with clade B Gag/Pol and clades A, B, and C Env) were studied at week 38 (14 weeks post-rAd5 boost). Nine donors from VRC011 were studied following randomized administration of the either the DNA-rAd5 immunizations described above (n = 3) or two doses of rAd5 (n = 6). Cellular activity was analyzed in samples from weeks 4 to 17 (median 7) following rAd5 boost collected by apheresis by the NIH Clinical Center Department of Transfusion Medicine.
HIV-1-positive donors were recruited under Duke University IRB approval through the adult and pediatric infectious diseases clinics at Duke University Medical Center. Virus controllers (VCs) were identified as subjects who had been diagnosed as HIV-1 positive for greater than 1 year, are antiretroviral therapy (ART) naïve, have CD4 counts >350 cells/μl blood, and have been controlling virus replication to less than 5,000 viral RNA copies/ml blood. Chronic HIV-1 donors were recruited by the Immunology Quality Assessment Center (IQAC) of the Duke Human Vaccine Institute (DHVI) or the Duke Infectious Diseases clinics and were defined as HIV-1 positive for greater than 1 year, with a viral load (VL) of >5,000 or on antiretroviral therapy. HIV-1-negative control CD8+ T cells were obtained as leukopaks, either from the American Red Cross or from healthy uninfected donors through the IQAC, DHVI.
NL4-3 and transmitted/founder Env (13)-containing Renilla luciferase (LucR) reporter viruses were generated by Fugene HD (Roche) DNA transfection of 293T cells. Working stocks were prepared by passaging virus in peripheral blood mononuclear cells (PBMCs). Briefly, PBMCs from a pool of 10 healthy HIV-1 negative donors were activated using OKT3 (R&D Systems) and anti-CD28 (BD Biosciences) for 3 days. PBMC targets were depleted of CD8+ T cells by magnetic bead separation (Dynal) and spinoculated (17) for 2 h at 1,200 × g. Virus supernatants were collected every 2 to 3 days and filtered through a 0.45-mm syringe filter, and titers were determined on TZM-bl cells. The virus nomenclature is as follows: NL4.3-LucR.T2A, NL-LucR.T2A-CH040.ecto, NL-LucR.T2A-CH058.ecto, NL-LucR.T2A-CH077.ecto, NL-LucR.T2A-WITO.ecto, NL-LucR.T2A-WEAU.ecto. Throughout the article, the viruses will be identified by envelope.
The virus inhibition assay (VIA) was performed as follows. PBMCs from each donor were thawed in RPMI 1640 supplemented with 20% fetal bovine serum (FBS), penicillin/streptomycin, and 20 U/ml recombinant human interleukin-2 (rhIL-2) (PeproTech) and activated for 3 days as described above. PBMC targets were depleted of CD8+ T cells by magnetic bead separation (Dynal; Miltenyi) as described above. CD8 effectors were isolated by negative selection using anti-CD4 beads (Dynal) or the CD8+ T-cell isolation kit (Miltenyi). Target PBMCs were infected at 2.25 × 105 cells/ml and a multiplicity of infection (MOI) of 0.004 to 0.06 by spinoculation at 1,200 × g for 2 h. Following spinoculation, cells were resuspended and plated at 2.25 × 104 cells per well in a flat-bottom 96-well tissue culture plate. CD8+ effectors were serially diluted 2-fold from 4.5 × 105 to 5.6 × 104 cells/ml and added to autologous infected targets at corresponding effector/target (E:T) ratios of 2:1, 1:1, 1:2, and 1:4. Infectivity controls consisted of infected targets without added CD8+ effectors. Cultures were incubated at 37°C and 5% CO2 for 3 days. Viral replication was measured as Renilla luciferase production (in relative light units [RLU]) from infected target cells using the Renilla luciferase assay kit (Promega) and read on a Berthold E & G luminometer using MicroWin software. Viral inhibition was calculated as a percentage of the infectivity control for each E:T ratio and log reduction of RLU from infectivity controls at an E:T of 2:1.
A dead-cell discriminator, ViViD (violet viability dye; Invitrogen), together with CD14- and CD19-PacificBlue (conjugated according to standard protocols; http://drmr.com/abcon/index.html), all read on the same fluorescent channel, were used to identify dead cells, B cells, and monocytes, and exclude them from the analysis. Surface markers used to determine T-cell maturation included CD45-ROQD705, CD57-QD585, CCR7-Ax680 (in-house conjugation), and CD28-Cy5-PE (BD Biosciences). Additional antibodies to identify T cells included CD8-QD800 (in-house conjugation) and CD4-Cy5.5-PE (Caltag), both surface stained, and CD3-Cy7-APC (BD), stained intracellularly (IC). Functional T cells were identified based on their staining with IFN-γ fluorescein isothiocyanate (FITC), IL-2 allophycocyanin (APC), TNF-α Cy7-phycoerythrin (PE), MIP-1β-PE (all from BD), and CD107a Ax680 (conjugated in house).
Fifteen-mer peptides overlapping by 11 amino acids were synthesized to more than 85% purity as confirmed by high-performance liquid chromatography (HPLC). Individual peptides of HIV-1 Gag, Env, Pol, and Nef were diluted to 100 mg/ml in dimethyl sulfoxide (DMSO) or deionized water and then pooled to a final concentration of 500 μg/ml. The TRVV peptide pools, which include B clade Tat, Rev, Vif, Vpr, and Vpu were pooled to a concentration of 100 μg/ml. Peptide mixes in each experiment were at a final concentration of 2.5 μg/ml, except for the TRVV peptides, which were used at a final concentration of 0.5 μg/ml.
Peripheral blood, with subsequent PBMC isolation using Ficoll-Paque Plus (GE Healthcare) density-gradient centrifugation, was collected from healthy donors. PBMCs were frozen (90% fetal bovine serum-10% DMSO) and stored at −180°C until analyzed. On the day of analysis, cells were thawed in a 37°C water bath, transferred to prewarmed R10 (RPMI 1640 [HyClone] supplemented with 10% heat-inactivated FBS [Invitrogen], 100 U/ml penicillin/streptomycin, and 2 mM l-glutamine [Gibco]), and examined for recovery and viability. PBMCs were cultured overnight at 1 × 106 to 2 × 106 cells/ml in a 37°C, 5% CO2 incubator. After overnight rest, cells were washed and resuspended in R10 at 10 × 106 cells/ml, and mixes of peptide pools composed of HIV-1 overlapping protein sequences were prepared. In addition, brefeldin A (BFA; final concentration of 10 μg/ml), monensin (final concentration of 0.7 μg/ml), and anti-CD107a (pretitered volume) were added to each peptide mix. For each assay day, an unstimulated control and a positive control (Staphyloccoccus enterotoxin B) were included in the experiment. To each plate well, 100 μl of cells and 100 μl of peptide mix were added, mixed, and incubated for 6 h at 37°C and 5% CO2. The plates were then placed in the refrigerator at 2 to 8°C overnight.
PBMCs were washed once in phosphate-buffered saline (PBS) and surface stained with a violet amine dye for dead-cell discrimination. Cells were then washed with wash buffer (PBS with 1% FBS) and surface stained with directly conjugated antibodies. After incubation, cells were washed, fixed/permeabilized with a Cytofix/Cytoperm kit (Becton Dickinson-PharMingen), washed twice in permeabilization/wash buffer, and stained intracellularly for markers of interest. The cells were then washed three times in permeabilization/wash buffer and resuspended in 1% paraformaldehyde (Electron Microscopy Systems) in PBS. Samples were acquired on a LSRII (Becton Dickinson), and data analysis was performed using FlowJo version 8.8.6 (TreeStar).
The frequency and phenotype of antigen-specific CD8+ T cells were determined in preliminary experiments, where cells were stimulated overnight with pooled peptides in the presence of brefeldin A (BD, San Jose, CA). Cells were subsequently washed in PBS and stained with the following commercial reagents: LIVE/DEAD fixable violet stain (Invitrogen, Eugene, OR) and antibodies to CD3 APC-Cy7 (BD), CD4 PE-cyanin (Cy) 5.5 (Caltag), CD27 PE-cyanin (Cy5) (Coulter), and CD127 PE (Coulter; in some experiments). In addition, the following reagents were manufactured in our laboratory and used: antibodies to CCR7 Alexa 680, CD45RO Qdot (QD545), CD57 QD585, and CD8 QD800. Cells were then treated with Cytofix/CytoPerm (BD), washed, and stained with the following commercial cytokine antibodies (all from BD): IFN-γ FITC, IL-2 APC, and TNF-α PE-Cy7. Antigen-specific T cells were identified based on their expression of at least one of the cytokines measured, and distribution of these cells among various memory cell populations was determined for each subject studied. The most frequent memory cell subsets in each individual were identified, stained (in a separate aliquot of cells, using the cell surface reagents listed above), and isolated by sorting samples on a 20-parameter BD FACS Aria.
Population comparisons of virus inhibition and cellular phenotypes were performed using Student's t test, assuming unequal variances (MedCalc v11.01.1 [MedCalc Software, Mariakerke, Belgium] and SAS v9.1 [SAS, Cary, NC]). Correlations between CD8-mediated virus inhibition and markers of effector function or phenotype were determined using Pearson's Correlation (MedCalc v11.01.1). There were no adjustments for multiple comparisons, and P values should be interpreted with this in mind. Polyfunctionality comparisons of multicategorical distributions were performed by a permutation test (SPICE v4.2). Comparisons of individual distributions were performed using the Wilcoxon rank sum test or Student's t test (SPICE v4.2; NIAID, NIH).
We characterized the ability of peripheral blood CD8+ T cells to mediate virus inhibition in vitro using cells from three donor cohorts: HIV-1+ virus controllers, HIV-1+ chronically infected subjects, and HIV-1 DNA prime/rAd5 boost vaccinees (see Table S1 in the supplemental material). The 16 virus controllers were defined as antiretroviral (ART)-naïve HIV-1+ subjects with CD4 counts of ≥350 cells/μl and viral loads of <5,000 copies/ml for at least 1 year. Two of these subjects were elite controllers with either undetectable viral load or undetectable with viral blips of <2,000 for greater than 2 years. The second cohort comprised 21 subjects with chronic HIV-1 infection (HIV-1 seropositive greater than 4 years, VL ≥ 5,000, or on ART). Sixteen of 21 subjects in this cohort were on ART and were matched to the virus controller cohort for the prevalence of controlling alleles (43% in the chronic infection group compared to 40% in the VCs) (Table S1). In the third cohort, we assessed the ability of a DNA prime/rAd5 boost vaccine regimen to elicit functional antiviral activity in CD8+ T cells. We examined 40 vaccinees: 31 enrolled in VRC008 and 9 enrolled in VRC011. Both protocol regimens included a multiclade DNA prime (VRC HIVDNA016-00-VP, clade B Gag, Pol, Nef and clades A, B, and C Env) given three times at months 0, 1, and 2, followed by an rAd5 boost (VRC-HIVADV014-00, clade B HIV-1 Gag/Pol and clades A, B, and C Env) given at month 6. For our study, cellular responses were examined 4 to 17 weeks postboost (VRC011) or at 14 weeks postboost (VRC008). We found no significant differences between samples from the two studies (or arms within either study) for any responses reported here; hence, all vaccinees were grouped together for the purposes of our analysis.
The capacity of CD8+ T cells to inhibit HIV-1 replication was assessed using an autologous CD8+ T-cell virus inhibition assay with replication-competent reporter viruses. HIV-1 inhibition was determined for six different replication-competent Renilla luciferase reporter viruses, isogenic except for the env ectodomain: four CCR5-tropic transmitted/founder envelopes (CH040, CHO77, CH058, and WITO), a CCR5/CXCR4-dualtropic transmitted/founder envelope (WEAU), (13), or the standard CXCR4 reference virus (NL4-3) (Fig. (Fig.1).1). CD8+ T cells from 23 HIV-1-seronegative donors were assessed for anti-HIV activity and were used to define a cutoff for positive inhibition as 0.8 log10 (3 standard deviations [SD] above the mean). In comparison to the seronegative group, we found that the DNA/rAd5 vaccine cohort, the virus controller cohort, and the chronic HIV-1 subjects had statistically significant virus inhibition (P < 10−4). Of note, the chronic subjects in this study were predominately on antiretroviral therapy and thus had low virus loads (see Table S1 in the supplemental material), which likely contributed to the preservation of the CD8+ T-cell-mediated virus inhibition in this study. CD8+ T cells from 65% of vaccinees inhibited replication of either transmitted/founder or NL4-3 enveloped viruses; the vaccine group had 0.7 log10 less-potent suppressive activity (P < 0.004) than the virus controller group. Although all six viruses showed similar magnitudes of suppression for the controller and chronic groups, three of the viruses (NL4-3, CH040, and CH058) were more sensitive to virus inhibition (P < 0.05) in the vaccinees than the less-sensitive virus (WEAU).
Differences in inhibitory activity might be due to sequence variation between the target viruses and the immunogen. Thus, we compared the sequences of the HIV-1 envelopes in the vaccine regimen to those envelopes used in the virus inhibition assay. As expected, the sequence of the VRC B envelope was most similar to the NL4-3 enveloped virus; this is consistent with the relatively greater sensitivity of this virus to inhibition (see Table S2 in the supplemental material). However, the envelope sequences for CH040 and CH058 are no more similar to any of the 3 vaccine strain envelopes than the other viruses. Thus, sequence identity alone does not account for differences in the magnitude of virus inhibition that were observed.
Finally, there was no correlation between inhibition and time since vaccination. Thus, the HIV-1 DNA/rAD5 prime-boost vaccine regimen elicited T-cell-mediated HIV-1 inhibition against a range of different viruses, including transmitted/founder enveloped viruses, for at least 4 months following vaccination.
Within these cohorts, a number of subjects (six controllers, seven chronic infection patients, and seven vaccinees) have known protective MHC class I alleles (HLA B*27, HLA B*57, or HLA B*5801). Of the vaccinees that had suppressive activity, 20% were HLA B*27, B*57, or B*58; thus, 80% of vaccinees showed CD8+ T-cell virus inhibition in the absence of these alleles. Figure Figure22 shows the magnitude of virus suppression for each of six reporter viruses for subjects with and without protective alleles. Although overall there is a trend toward greater virus suppression in subjects with controlling alleles, this reached significance only for vaccinees tested against the viruses NL4-3 and WEAU. Of all six viruses, the NL4-3 and WEAU viruses have the most conserved B*27 and B*57 CTL epitopes (7/10 and 4/10, respectively) that have been defined (LANL database) (see Fig. S1 in the supplemental material); this may account for the increased magnitude of suppression in allele-positive individuals. In contrast, virus inhibition of the other four viruses was not significantly different between those with and without known controlling alleles. Thus, HIV-1 vaccines can elicit functional CD8+ T-cell-mediated virus inhibitory activity in subjects that are not known to be genetically predisposed to have better virus control.
We sought to define differences in CD8+ T-cell responses in vaccinees with antiviral activity against the standard reference virus (NL4-3) compared to those without antiviral activity. We assessed multiple CD8+ T-cell effector functions using 13-color flow cytometry (see Fig. S2 in the supplemental material) after in vitro stimulation with overlapping peptides covering the individual HIV-1 proteins (Fig. (Fig.3).3). Overall, CD8+ responses were greatest for Env- and Gag-stimulated cells, with Env responses accounting for nearly 60% of the total response to HIV-1. We found no significant differences in percentage of IL-2- or TNF-α expressing cells in vaccinees with inhibitory activity over those without inhibitory activity. However, antiviral activity was strongly associated with increased numbers of CD8+ T cells expressing CD107a or MIP-1β following either Env (Fig. (Fig.3A)3A) or Gag (Fig. (Fig.3B)3B) stimulation, and additionally with IFN-γ for Env-stimulated cells. There was not a direct correlation between the presence of known protective alleles in this cohort and with the total magnitude of any T-cell response nor with any individual functions.
Similarly, virus controllers and chronically HIV-1-infected donors with CD8+ T-cell-mediated antiviral activity also had CD8 T-cell responses dominated by production of CD107a, MIP-1β, and IFN-γ (Fig. (Fig.4).4). CD8+ T-cell responses following Env or Gag peptide stimulation showed that, for HIV-1-infected subjects, Gag-specific responses were predominant. Specifically, Gag responses accounted for 80% (virus controllers) or 65% (chronically infected) of the total HIV-1-specific response, and thus represent a significant portion of the CD8 T cells with antiviral capacity.
CD8+ T cells are capable of expressing multiple effector molecules at once. The association between viral inhibition and CD107a, MIP-1β, or IFN-γ may reflect inhibitory activity of all of these functions. Alternatively, only a subset of the functions may be inhibitory, with a broader correlation for all functions arising from the frequent coexpression of these functions by individual cells. To address this question, we measured all functions on a cell-by-cell basis and examined whether CD107a, MIP-1β, or IFN-γ expression was independently associated with virus suppression. Among the vaccinees, for both Env (Fig. (Fig.5A)5A) - and Gag (Fig. (Fig.5B)-stimulated5B)-stimulated CD8+ T cells, virus inhibition was associated with CD107a and MIP-1β expression independently, but not with IFN-γ. Both polyfunctional and monofunctional cells expressing MIP-1β and CD107a were associated with the presence of antiviral activity (see Fig. S3 in the supplemental material) following Env (Fig. S3A) or Gag (Fig. S3B) stimulation. The strongest correlation between the potency of antiviral activity and number of HIV-1-specific cells was found for monofunctional cells expressing MIP-1β alone. Furthermore, among the functional subsets showing significant correlations with antiviral activity against at least three viruses, all expressed either CD107a or MIP-1β; notably, there was no correlation for cells expressing IFN-γ in the absence either of these two functions. These results indicate that the association of IFN-γ with viral suppression is a consequence of the frequent coexpression of this cytokine by those cells with true suppressive activity.
Similar results were observed for HIV-1+ subjects. For virus controllers, significant associations with antiviral potency were found for functional subsets expressing CD107a, IFN-γ, or MIP-1β (see Fig. S4 in the supplemental material) for Env- or Gag-stimulated cells. Again, by examining individual responses on a cell-by-cell basis, it is apparent that the suppressive activity is associated with MIP-1β or CD107a expression. Suppressive activity correlated with Env- and Gag-specific MIP-1β monofunctional cells and polyfunctional cells expressing CD107a and/or MIP-1β cells, but not with cells that produced IFN-γ, IL-2, or TNF-α in the absence of MIP-1β or CD107a (see Table S3 in the supplemental material).
CD107 and MIP-1β have been shown to be dominant CD8 responses to HIV antigens in chronically infected adults (3, 19); thus, correlations between these functions and antiviral activity could be driven by this dominance. Indeed, in our cohort, the fractions of the total HIV response that included CD107a and MIP-1β were 72% and 88%, respectively (noncontrollers), and 89% and 88% (controllers). However, for vaccinees, these fractions were 28% and 60%: i.e., these functional responses do not dominate the CD8 T-cell response like they do in natural infection, but still correlate with viral suppression activity. Taken together, these data indicate that virus suppression can be mediated by CD8+ T cells specific for multiple HIV-1 antigens (Env and Gag) and producing MIP-1β and/or CD107a, but not by CD8+ T cells producing only IFN-γ, IL-2, or TNF-α.
By including markers of T-cell maturation in our antigen-specific assays, we determined whether T cells at various stages of the differentiation pathway could mediate virus inhibition. Cells that have reached the terminal effector stage may be more likely to quickly mediate cytolytic activity than earlier memory phenotypes; however, such effector cells may not be as long-lived. To define the phenotype of CD8+ cells that mediate virus inhibition, we characterized the phenotype of HIV-1-specific CD8+ T cells from subjects with demonstrable CD8-mediated antiviral activity. We grouped CD8+ T cells that respond to HIV-1 antigens (and are therefore memory cells) into five categories: naive-like (CD45RO− CCR7+ CD28+), central memory (CD45RO+ CCR7+ CD28+), transitional memory (CD45RO+ CCR7− CD28+), effector memory (CD45RO+ CCR7− CD28−), or terminal effector (CD45RO− CCR7− CD28−). Figure Figure6A6A illustrates the distribution of the HIV-1 antigen-specific cells among these five different maturation stages for the three cohorts. We observed a significant fraction of antigen-specific T cells with a phenotype that is typically associated with naïve T cells (predominantly in the vaccinees). The significance of these cells is unknown; we have observed a similar phenotype for antigen-specific cells after influenza virus vaccination in healthy individuals (M. Roederer, unpublished data).
Vaccinees with antiviral activity had a greater representation of cells from the earlier stages of differentiation, and the HIV-1-positive subjects had a greater representation of cells with the latest stage of differentiation. However, there were no significant associations between the differentiation phenotype of the antigen-specific CD8+ T cells and antiviral activity (i.e., the distribution of memory phenotypes of HIV-specific cells in vaccinees with high inhibitory activity was not different from that of those with low inhibitory activity). These data indicate that CD8+ T cells of various stages of differentiation were capable of mediating virus suppression, as long as they express MIP-1β or CD107a.
To further investigate the relationship between memory differentiation stage and antiviral activity, we purified phenotypically defined CD8+ T-cell subsets by flow cytometry. Viably sorted cells (see Fig. S5 in the supplemental material) from HIV-1+ virus controllers and vaccinees with suppressive activity were tested for antiviral activity (Fig. (Fig.7A).7A). HIV-1 inhibition was essentially absent from sorted naïve CD8+ T cells, but present in all other CD8+ T-cell subsets. All memory subsets inhibited virus replication significantly better than naïve cells (P < 0.03, Student's t test). The degree of inhibition by these purified subsets could be affected by the intrinsic functional capacity of the cells as well as the relative proportion of antigen-specific T cells present. To account for this, we normalized the virus inhibitory activity for each culture by the fraction of HIV-1-specific CD8+ T cells present in that subset (as determined by stimulation of a parallel sample, stained with the same reagents used for sorting plus those identifying intracellular cytokine expression). As shown in Fig. Fig.7B,7B, the central memory subset tends to have the greatest inhibitory activity on a per-cell basis, but this did not achieve statistical significance (P = 0.1, Student's t test). Thus, CD8+ T cells from all memory stages of differentiation exhibit antiviral activity.
Identification of the inducible CD8+ T-cell functions responsible for virus control in HIV-1-infected humans will profoundly impact vaccine design. To this end, it is important to delineate the functional antiviral capacity of HIV-1-specific CD8+ T cells elicited by vaccine strategies and to compare those capacities with CD8+ T-cell inhibitory activities found in HIV-1-infected subjects who control virus production. T cells are remarkably heterogeneous and can be characterized by their capacity to express numerous functions including cytolysis, cytokine secretion, and proliferation. Some of these functions, as well cellular homing properties and differentiation stage, can be inferred from cellular phenotypes. In this study, we demonstrated that vaccination against HIV-1 Env and Gag-Pol by DNA priming followed by rAd5 boosting elicited CD8+ T-cell-mediated antiviral activity. In both vaccinees and in chronically infected HIV-1+ virus controllers, this activity correlated with HIV-1-specific CD107a or MIP-1β expression from HIV-1-specific T cells. Moreover, for vaccinees or virus controllers, purified memory CD8+ T cells from a wide range of differentiation stages were capable of significantly inhibiting virus replication.
The quality of CD8+ T cells can be assessed by examining multiple parameters simultaneously (22). Various functions may have antiviral activity; for example, CD107a expression is associated with cytolytic activity and may reflect the ability of antigen-specific CD8+ T cells to eliminate infected cells. Similarly, MIP-1β secretion can directly inhibit infection of CCR5-using viruses by blocking the viral coreceptor (7, 8). Finally, effector cytokines such as IFN-γ or TNF-α may have antiviral activity, although TNF-α can also activate HIV-1 replication through NF-κB activation. Identifying which of these factors is responsible for antiviral activity by individual CD8+ T cells can be ascertained with multiparameter flow cytometry. Here, we report on the role of multiple CD8+ T-cell functions in inhibiting viral replication in vitro.
In general, the total expression of IFN-γ, CD107a, or MIP-1β correlated with viral inhibition. However, it is important to note that these cytokines are not expressed independently but are coordinately regulated in many T cells. By dissecting these responses down to individual combinations of these cytokines, we found that the viral inhibition was most closely defined by MIP-1β or CD107a expression. Indeed, the secretion of IFN-γ in the absence of these two functions was not correlated with inhibition. These data suggested that IFN-γ is a surrogate for CD107a and/or MIP-1β because of high coexpression but is not likely to contribute to virus inhibition in this assay. Likewise, TNF-α or IL-2 secretion was not observed to be characteristic of antiviral cells.
We also examined the relationship of the CD8+ T-cell phenotype to antiviral activity. In general, vaccinees have HIV-1-specific cells of an earlier differentiation stage than do controllers, and the chronically infected have the most-differentiated cells. Nonetheless, we did not identify strong correlations between the phenotype of HIV-1-specific CD8+ T cells and the magnitude of antiviral activity in either vaccinees or infected subjects. To further delineate the relationship of differentiation stage to inhibition, we isolated pure populations of CD8+ T cells based on their phenotypically defined differentiation stage and quantified the ability of these cells to effect viral inhibition. Overall, we found that cells of any memory phenotype could inhibit HIV-1, whether of the central memory or effector memory categories.
A goal of vaccination is to elicit a strong and durable memory response—a characteristic of central memory cells. However, due to the rapidity in which HIV-1 establishes a latent infection (18), it will be important to arm HIV-1-specific T cells to respond rapidly to infection; thus it is presumed that effector T cells will also be needed. Our results suggest that whether a vaccine elicits central memory- or effector memory-biased CD8+ T-cell responses, a strong antiviral activity can be achieved. A vaccine-elicited CD8+ T-cell-mediated antiviral response could help control early viral replication and reduce the risk of established infection and subsequent transmission. The magnitude and breadth of a vaccine elicited virus inhibitory response needed to impact in vivo virus replication are still unknown. The DNA/rAd5 vaccination induced CD8+ virus inhibitory responses that were 0.7-log less than HIV-1+ virus controllers; it is not clear if the level of CD8+ T-cell responses achieved by this vaccination strategy would be sufficient for controlling virus replication in vivo. Further optimization of vaccine regimens to induce these responses will be important.
The genetic determinants HLA B*57 (2, 9, 16) and HLA B*27 (12) are strongly associated with virus control. An unresolved question is whether T-cell-based vaccines can elicit functional CD8+ T cells in subjects without these MHC alleles. We found that HIV-1-infected individuals bearing these alleles did not have substantially greater inhibitory activity. Importantly, our study also demonstrates that T-cell-based vaccines can effectively target genetically diverse populations and elicit CD8 inhibitory activity even in individuals who are not genetically predisposed to have better control of HIV-1 replication.
Vaccine-elicited CD8+ T cells that can inhibit a diverse panel of HIV-1 isolates are an important goal for vaccine design; thus assay systems need to be established to measure this heterologous virus inhibition, akin to the way neutralizing antibodies are systematically studied (15). In this study, we used a panel of HIV-1 enveloped viruses that include several distinct transmitted/founder viruses and viruses with different coreceptor usage. We found that although all viruses were inhibited, there were some differences in sensitivity, with two of the transmitted viruses (CH040 and CH058) having the most sensitivity to CD8 virus inhibition. Importantly, measurement of inhibitory activity against a wide panel of virus isolates may reveal the potential breadth or coverage of the CD8 virus inhibitory response elicited by a given vaccine approach.
In summary, we show that the expression of MIP-1β and CD107a by HIV-1-specific CD8+ T cells, in either the setting of vaccination or natural infection, most closely corresponds to antiviral activity. In contrast, it appears that the phenotype of the HIV-1-specific CD8+ T cells is less important to this activity. We also demonstrate that, with respect to virus inhibition, HIV-1 vaccines can stimulate anti-HIV CD8+ T-cell subsets similar in function to what is found in HIV-1+ virus controllers. Quantifying these functions will be important for the optimization of vaccine strategies attempting to elicit CD8-mediated antiviral activity.
This work was supported by the National Institutes of Health (NIH/NIAID/DAIDS) through RO1A5779 (G.D.T.) and CHAVI AI067854-05 (G.D.T., J.C.K., and B.F.H.); the Molecular Virology and Clinical Cores of the Duke University Center for AIDS Research P30 AI 64518 (G.D.T., C.K.C., and K.J.W.); the Nucleotide Sequencing and Virology Cores of the UAB Center for AIDS Research P30 AI27767 (J.C.K.); and the Intramural Research Program of the NIAID, NIH.
We thank Bob Bailer, Joe Casazza, Julie Ledgerwood, Richard Nguyen, and the volunteers and staff of the Vaccine Research Center clinical trials teams and Sunita Patil, Jason Stout, Gary Cox, David Holland, Charles Hicks, Susanna Naggie, Mehri McKellar, Deverick Anderson, Nathan Thielman, Vivian Chu, Steven Taylor, and Ross McKinney and the Duke CFAR Clinical Core, as well as the Duke Adult and Pediatric Infectious Diseases patients and staff. We thank Hua-Xin Liao for help with envelope gp140 sequence alignments; Ambrosia Garcia-Louzao, Raul Louzao, Judy Stein, and Erin Klein for clinical and technical assistance with this study; Nathan Vandergrift for statistical support; and George Shaw and Beatrice Hahn for supplying transmitted/founder envelope virus sequences.
Published ahead of print on 3 March 2010.
†Supplemental material for this article may be found at http://jvi.asm.org/.