Quantitation of HIV-1-Specific Cytotoxic T Lymphocytes and Plasma Load of Viral RNA

See allHide authors and affiliations

Science  27 Mar 1998:
Vol. 279, Issue 5359, pp. 2103-2106
DOI: 10.1126/science.279.5359.2103


Although cytotoxic T lymphocytes (CTLs) are thought to be involved in the control of human immunodeficiency virus–type 1 (HIV-1) infection, it has not been possible to demonstrate a direct relation between CTL activity and plasma RNA viral load. Human leukocyte antigen–peptide tetrameric complexes offer a specific means to directly quantitate circulating CTLs ex vivo. With the use of the tetrameric complexes, a significant inverse correlation was observed between HIV-specific CTL frequency and plasma RNA viral load. In contrast, no significant association was detected between the clearance rate of productively infected cells and frequency of HIV-specific CTLs. These data are consistent with a significant role for HIV-specific CTLs in the control of HIV infection and suggest a considerable cytopathic effect of the virus in vivo.

At all stages of disease, plasma RNA viral load remains the most potent predictor of outcome in HIV-1–infected individuals (1). Cross-sectional stratification of RNA loads provides a highly significant indicator of the likelihood of progression to acquired immunodeficiency syndrome (AIDS) and mortality (1). Even shortly after seroconversion the virological setpoint is significantly associated with prognosis, suggesting that in most individuals the determinants of progression are present early in the course of infection (1). The dominant host immunological factors involved in the control of progression remain unresolved, but most infected individuals show a vigorous HIV-specific immune response comprising both cellular and humoral mechanisms.

Cytotoxic T lymphocytes (CTLs) are believed to be important in the control of HIV infection (2-6) because the emerging HIV-specific CTL response observed during primary infection follows a close temporal association with acute viral load reduction (2, 3). Evidence also comes from the characterization of virus mutants that escape recognition by CTLs that have been identified at seroconversion, during asymptomatic HIV infection, during AIDS, and after CTL immunotherapy (4, 5), suggesting that CTLs exert a considerable selective pressure at many stages of disease. In addition to killing infected cells, CD8+ T cells effectively inhibit HIV replication in vitro through the production of chemokines (6).

Mathematical modeling of viral dynamics has predicted that if CTLs are important in the control of HIV infection, there should be an inverse association between HIV-specific CTL activity and plasma RNA viral load (7, 8). However, previous studies have demonstrated no significant correlation between circulating HIV-specific CTL activity and plasma RNA viral load during the chronic phase (9-11). High CTL activity in the presence of low levels of plasma viral RNA has been reported, but these findings were based on qualitative associations and indirect assays of CTL function that require culture and expansion in vitro (12). In a study of early HIV infection only, the levels of Env-specific cultured memory CTLs were inversely correlated with RNA viral load, but this finding did not extend to Gag- or Pol-specific memory CTLs (13). Direct measurement of lytic activity by uncultured, circulating HIV-specific CTLs (effector CTLs, CTLe) has only been possible in a subset of patients because the assay is relatively insensitive, requiring antigen-specific CTLe frequencies above 1 in 100 peripheral blood mononuclear cells (PBMCs) for lysis to be detectable (14). Here we have used highly sensitive human leukocyte antigen (HLA)–tetrameric complexes for a cross-sectional analysis of HIV Gag- and Pol-specific CTLe frequencies from 14 untreated HLA A*0201–positive individuals at different stages of infection.

HLA-tetrameric complexes can be used to directly visualize antigen-specific T cells by flow cytometry (15). HLA heavy chain is expressed in Escherichia coli with an engineered COOH-terminal signal sequence containing a biotinylation site for the enzyme BirA (16). After refolding of heavy chain, β2-microglobulin (β2M), and peptide, the complex is biotinylated and tetramer formation induced by the addition of streptavidin. By means of fluorescently labeled streptavidin, the tetramer can be used to stain and sort antigen-specific cells. The staining is highly specific such that CTL clones and lines directed to different epitope peptides bound to the same HLA molecule do not stain (15). Figure 1, A to C, shows examples of HLA-tetrameric staining with (A) HLA B*3501 tetramer refolded around the envelope peptide DPNPQEVVL [Env(77–85)] (17), (B) HLA A*0201 tetramer refolded around the Gag peptide SLYNTVATL [Gag(77–85)] (18), and (C) HLA A*0201 tetramer refolded around the Pol peptide ILKEPVHGV [Pol(476–484)] (19, 20). Using freshly isolated PBMCs, we characterized epitope-specific cytolytic activity with 86 51Cr-release cytotoxic assays, each performed in triplicate, and compared the results to HLA-tetrameric complex staining (Fig.1D). Extending previous data (15), we found a highly significant positive correlation (P < 0.001) between fresh cytolysis and tetramer staining. Low levels of cytolysis (<5%) previously discounted as below an empirical limit in other studies may in fact reflect significant numbers of circulating epitope-specific CTLe. By means of two-color staining, we were able to accurately detect tetramer-positive cells at levels as low as 0.02% of CD8+T lymphocytes. Direct sorting of tetramer-stained PBMCs into Elispot plates (21) is a very rapid and reliable protocol that can be applied to establish the specificity of low-frequency CTL responses, such as tumor-specific CTL responses in patients with melanoma (15).

Figure 1

Analysis of PBMCs for the expression of cell surface markers with a FACS Calibur (Becton Dickinson) and CellQuest software (Becton Dickinson). Antibody to CD38 (anti-CD38)–fluorescein isothiocyanate (Dako) and anti-CD8–Tricolour (Caltag) were used according to standard protocols. Briefly, 106 PBMCs were centrifuged at 300g for 5 min and resuspended in 50 μl of cold phosphate-buffered saline. Tri-color analysis was performed with tetramer-phycoerythrin, anti-CD8–Tricolor, and anti-CD38. The cells were incubated with tetramer and antibodies on ice for 30 to 60 min and then washed twice before formaldehyde fixation. Gates were applied to contain >99.98% of control samples. Controls for the tetramers included both A*0201-negative individuals and A*0201-positive HIV-1–uninfected donors. (A toC) CD8+ T cells from three HIV-1–infected individuals with staining for CD38 along the x axes and for B*3501-Env tetrameric complex, A*0201-Gag, and A*0201-Pol along they axes, respectively. The tetramer-positive cells make up 0.2 to 2.7% of all CD8+ cells (values are indicated in each plot) and are all CD38+. (D) Comparison of the percentage of CD8+ cells staining with tetramer and the uncultured peptide-specific cytolytic activity of PBMCs. Each experiment was performed in triplicate. Donors were all HLA A*0201–positive and were either untreated or treated with combination antiretroviral therapy. Responses to three epitopes were measured [A2Gag, A2Pol, A2EBV BMLF1 280-8 GLCTLVAML (27)] and were all included in the data. Subgroup analyses gave identical results, namely, a significant positive correlation (Pearson correlation coefficient) between percentage peptide-specific lysis and percentage of CD8+ T cells staining with each tetramer. By multiple stainings on single samples, we have found tetramer binding to be highly reproducible with variation of less than 5% between stains.

To address the sensitivity of tetramer binding, we performed a parallel experiment using an HLA A*0201 influenza A matrix peptide GILGFVFTL (residues 58 to 66) tetramer to sort stained cells from a source population of 0.04% of CD8+ T lymphocytes derived from an HLA A*0201–positive donor. When duplicate sets of tetramer-positive cells were placed directly into Elispot wells, every cell could be accounted for after influenza matrix–specific interferon-γ (IFN-γ) production was determined (21). In contrast, CD8+ T cells that did not stain with the influenza A matrix tetramer were unable to produce IFN-γ in response to peptide-specific stimulation. Tetramer-positive cells were also plated directly into cloning wells (22), and after 3 weeks in culture all expanded clones were able to secrete IFN-γ when exposed to HLA A*0201–matched target cells coated with the influenza A matrix peptide (22). Therefore, staining of PBMCs with HLA-tetrameric complexes provides a highly sensitive and specific method for detecting antigen-specific CTLs.

At least 66% of HLA A*0201–positive individuals have circulating CTLs that recognize Gag(77–85), an A*0201-restricted Gag epitope. Most (70%) of those individuals that do not recognize the A2Gag epitope will recognize Pol(476–484), an A*0201-restricted Pol epitope, so that the addition of both the A2Gag and A2Pol responses gives a representation of total A*0201-directed CTL activity (23). Using HLA-tetrameric complexes, we compared the percentage of uncultured CD8+ T cells specific for the A2Gag epitope with plasma RNA viral load in 14 HLA A*0201–positive untreated individuals at different stages of disease (Fig. 2A) and found a highly significant inverse correlation (Pearson correlation coefficientr = −0.80, P = 4.05 × 10 6). The addition of A2Gag and A2Pol responses improved the P value to 3.2 × 10 7 (Fig. 2B). The CD4 counts ranged from 235 to 711 cells/μl (median, 457 cells/μl), and viral load from <500 to 359,500 copies per milliliter (median, 10,418 copies per milliliter). No correlation was detected between CD4 count and viral load or between CD4 count and the circulating HIV-specific A*0201-restricted CTLe activity. Such a lack of association between total CD4 count and circulating CTLe activity has been reported previously (7, 10, 11, 24).

Figure 2

Association between plasma RNA viral load and the percentage of CD8+ cells staining with (A) A2Gag tetramer alone and (B) A2Gag and A2Pol tetramers.

On the basis of previous calculations, many of those individuals with sufficient circulating HIV-specific CTLs to have fresh HIV-specific cytolytic activity will lyse target cells with half-lives of less than 1 day (7, 8). For a cytopathic virus (short half-life of infected cells), CTL-mediated lysis would be unlikely to significantly further shorten the half-life of infected cells while still causing the death of most infected cells. A small change in the life-span of productively infected cells may have a large effect on viral burst size. In contrast, for a noncytopathic virus, CTL-mediated lysis would significantly shorten the otherwise relatively long half-life of infected cells. Thus, if the virus is cytopathic, CTL-mediated lysis should reduce RNA viral load while having minimal effect on the clearance rate of infected cells. We examined this hypothesis in 9 of the 14 individuals who received triple-combination therapy shortly after the CTL measurements were made. The subsequent decline in plasma viral RNA was used to derive the exponential clearance rate of productively infected cells (25) (Fig.3A). The clearance rate did not correlate with the percentage of A2Gag or A2Pol staining or with CD4 count, confirming that the decline in productively infected cells was independent of disease stage (CD4 count) or immune status of the patients.

Figure 3

Effect of antiretroviral therapy. (A) The exponential decay in plasma viremia from nine individuals treated with combination antiretroviral therapy. (B) Changes in A2Gag- (▪) and A2Pol-specific (□) CTLs after 6 months of combination antiretroviral therapy during which suppression of viremia was maintained. (C toE) Changes in B35Env (C), A2Gag (D), and A2Pol (E) staining of PBMCs from three individuals before and after 6 months of therapy. The x axes show staining with anti-CD38, and the y axes staining with tetramer. The values in each plot are the percentage of CD8+ T cells staining with the relevant tetramer. Before treatment, the plasma RNA viral loads were 46500, 1420, and 27720 copies per milliliter, respectively. After 6 months of therapy all three individuals had plasma RNA viral loads below the levels of detection.

To examine whether the inverse association of HIV-specific CTLe and plasma RNA viral load was due to suppression of the virus by CTLs and not vice versa, we artificially reduced viral load using triple-combination antiretroviral therapy. The nine treated individuals were studied again at 6 months after starting treatment when all had viral loads below the detectable limit (<500 copies per milliliter). At 6 months after starting therapy, a new steady state will have been reached, thus eliminating the rapid early fluctuations in the CD8 population observed in previous studies (26). In all cases, we observed a decrease in the HIV-specific CTLe, which demonstrates that the virus did not have a significant inhibitory effect on CTLs and consistent with the dependence of HIV-specific CTLs on continued viral replication (Fig. 3, C to E).

Using HLA-tetrameric complexes, we have demonstrated a significant inverse association between A*0201-restricted HIV-specific CTLe and plasma RNA viral load. No significant correlation was detected between the A*0201-restricted HIV-specific CTLe and CD4 count or clearance rate of productively infected cells. These findings strongly support the involvement of CTLs in the control of HIV infection. Furthermore, these data add weight to the search for a vaccine or therapeutic strategies designed to boost the HIV-specific CTL response.


View Abstract

Stay Connected to Science

Navigate This Article