Expression of the CD8αβ-Heterodimer on CD8⁺ T Lymphocytes in Peripheral Blood Lymphocytes of Human Immunodeficiency Virus⁻ and Human Immunodeficiency Virus⁺Individuals

EDTA-anticoagulated peripheral blood specimens were obtained from 21 healthy volunteers and 103 HIV-1–infected individuals. HIV-1 seropositive individuals provided written informed consent before donating blood or blood was evaluated from specimens available in the course of routine clinical testing. HIV-1 seronegative blood donors provided verbal consent. Blood samples from eight HIV-1–infected individuals participating in the prospective open-label AIDS Clinical Trials Group (ACTG) study 343 were obtained at baseline and at weeks 4, 8, and 16 after starting treatment. These patients received highly active antiretroviral therapy (HAART), including zidovudine (ZDV; Glaxo Wellcome, Inc, Research Triangle Park, NC) 300 mg twice daily, lamivudine (3TC; Glaxo Wellcome, Inc) 150 mg twice daily, and indinavir (IDV; Merck & Co, Inc, Whitehouse Station, NJ) 800 mg three times a day.

The MoAbs used for this study (all from Coulter Corp, Miami, FL) were directly conjugated to fluorescein isothiocyanate (FITC), phycoerythrin (PE), phycoerythrin-Texas red (ECD), or allophycocyanin (APC) (Table 1). The following four-color combinations were used: (1) CD8α-FITC, CD4-PE, CD8αβ-ECD, and CD3-APC; (2) CD8α-FITC, TCRγδ-PE, CD8αβ-ECD, and CD3-APC; (3) CD8α-FITC, CD56-PE, CD8αβ-ECD, and CD3-APC; (4) CD8α-FITC, C1.7-PE, CD8αβ-ECD, and CD3-APC; (5) CD8α-FITC, HLA-DR-PE, CD8αβ-ECD, and CD3-APC; (6) CD8α-FITC, CD11a-PE, CD8αβ-ECD, and CD3-APC; (7) CD8α-FITC, CD28-PE, CD8αβ-ECD, and CD3-APC; (8) CD8α-FITC, CD38-PE, CD8αβ-ECD, and CD3-APC; (9) CD8α-FITC, CD45RA-PE, CD8αβ-ECD, and CD3-APC; (10) CD8α-FITC, CD45RO-PE, CD8αβ-ECD, and CD3-APC; (11) CD8α-FITC, CD49d-PE, CD8αβ-ECD, and CD3-APC; (12) CD8α-FITC, CD57-PE, CD8αβ-ECD, and CD3-APC; (13) CD8α-FITC, CD58-PE, CD8αβ-ECD, and CD3-APC; (14) CD8α-FITC, CD62L-PE, CD8αβ-ECD, and CD3. In addition, a two-color reagent (CD45-FITC and CD14-PE) was used for verifying the recovery and purity of the lymphocytes within the light scatter gates according to the Centers for Disease Control (CDC) guidelines for performing CD4⁺ T-cell determinations in persons infected with HIV-1.25 Aliquots of 100 μL of EDTA-anticoagulated blood from each donor were incubated with each of these reagent combinations for 15 minutes at room temperature before lysis and fixation using Coulter Immunoprep Reagent System and Q-prep Workstation (Coulter, Miami, FL). To reduce the background level of staining, the Q-Prep procedure was modified, and lysed samples were washed with 1.0 mL phosphate-buffered saline (PBS) and centrifuged for 3 minutes at 300g. The supernatants were decanted, cells were resuspended in 0.5 mL PBS containing 1% paraformaldehyde and maintained for 24 hours at 4°C before flow cytometric analysis.

Samples were analyzed on a Coulter EPICS Elite ESP equipped with argon and helium neon lasers, a gated amplifier, and a 120 μm flow cell tip. The instrument was run at high bandwidth and alignment was controlled on a daily basis using DNA-CHECK EPICS Alignment Fluorospheres (Coulter Corp) to maintain the same sensitivity levels during the entire study. Linear performance in each channel was controlled using the EPICS Immuno-Brite Standards Kit (Coulter Corp). The sensitivity of the photo multiplier tube for detection of the ECD-characteristic fluorescence was controlled during the entire study with a single lot of anti-CD8αβ–ECD stained Coulter CYTO-TROL control cells (Coulter Corp). Voltage and compensation levels were established using both unstained cells for adjusting the negative/background levels of fluorescence to the first log step and single-color stained cells for adjusting spectral overlap. A total of 5,000 lymphocytes were analyzed in a manually set acquisition gate and positive cutoffs for fluorescence were set to the first log step to include less than 1% of nonstaining cells. Data analysis was performed using the EPICS Elite software version 4.02 (Coulter Corp). Absolute numbers of lymphocyte subsets were calculated using routine diagnostic lymphocyte counts obtained from the same blood specimens analyzed on a Coulter Hematology Analyzer T-540 (Coulter Corp).

All results showing percent or absolute count data in Tables 2-4 are expressed as median (25th percentile, 75th percentile). Comparisons of the three data values per row in Tables 2and 3 were first done using the Kruskal-Wallis test.28 If the results of the Kruskal-Wallis test were statistically significant (P < .05), the three data pairs per row were compared by the Dunn test.29 The crude significance level used for the Dunn tests within each row of Tables 2 and 3 was P < .01 to account for the three tests being done in each row. This method was also used to obtain theP values in Fig 1C (P < .01).

The comparison of the mean fluorescence (MF) of anti-CD8αβ staining of PBL from all three blood donor groups is shown in Fig 1. Linear regression analyses were performed relating CD8αβ MF to absolute counts of CD4⁺ T lymphocytes/μL blood (Fig 1A and B). The F test was used to test whether the slope was significantly different from 0.30 The Wilcoxon matched pairs31 test was used for comparison of the CD8αβ MF of PBL from the eight HIV-1 seropositive study subjects receiving HAART.

Within each of the three blood donor groups in Table 4 the anti-CD8αβ staining of the two different CD8αβ⁺ T lymphocyte subsets was investigated using the Wilcoxon matched pairs test (paired within each patient). For example, the test at the top of the leftmost column in Table 4 compared the MF of anti-CD8αβ staining for HLA-DR⁺ versus HLA-DR⁻ CD8⁺ T lymphocytes from HIV-1 seronegative individuals. Statistical tests in Table 4 were considered significant if the P values were < .05. These P values were not adjusted for multiple comparisons.

All statistical analyses were performed using Microsoft Excel software version 5.0 (Microsoft Corp, Redmond, WA) and GraphPad PRISM software version 2.01 (GraphPad Software, San Diego, CA). Data presentation was performed using WinMDI software version 2.3 (Joseph Trotter, La Jolla, CA) and Microsoft PowerPoint software version 4.0c (Microsoft Corp).

The absolute CD4⁺ T lymphocyte counts of the HIV-1 seropositive blood donors ranged from 1 to 1,359 cells/μL. Of the 103 HIV-1 seropositive donors analyzed, the 58 with > 200 CD4⁺ T cells/μL were designated as early stage of disease and the 45 donors with < 200 CD4⁺ T cells/μL were designated as late stage of disease. The CD4 counts of the 21 healthy, HIV-1 seronegative donors all fell within the expected normal range for our laboratory. For the purpose of those studies, lymphocyte subsets were defined as follows: CD8⁺ T lymphocytes (CD8α⁺CD3⁺), NK cells (CD56⁺CD3⁻), TCRγδ⁺ T lymphocytes (TCRγδ⁺CD3⁺), and T-helper cells (CD4⁺CD3⁺).

We found that CD8αβ-heterodimer positive lymphocytes were almost all CD3⁺ (99%, Fig 2). We sought to clarify which subset of CD8⁺ lymphocytes express the CD8αβ-heterodimer in HIV-1 seronegative and seropositive individuals. A median of 91% of CD8α⁺CD3⁺ T lymphocytes of HIV-1 seronegative blood donors and 97% (<200 CD4⁺ T cells/μL) or 98% (<200 CD4⁺ T cells/μL) of HIV-1 seropositive blood donors expressed the CD8αβ-heterodimer (Table 2). It has previously been suggested that the subset of CD8^bright T lymphocytes predominantly contains true MHC class I–restricted CD8⁺ T lymphocytes with potential cytotoxic function.32 33 We found that almost all CD8α^bright cells were CD8αβ⁺ (median, 97% of the HIV-1 seronegative and 99% of the HIV-1 seropositive blood donors) (Table 2).

The median differences in the absolute cell counts between all CD8α⁺CD3⁺ T lymphocytes and the CD8αβ⁺ T lymphocytes were less than 57 cells/μL (HIV-1 seronegative donors, 56; HIV-1 seropositive donors > 200 CD4⁺ T cells/μL, 38; HIV-1 seropositive donors < 200 CD4⁺ T cells/μL, 21). The median differences between the absolute cell counts of bright CD8α⁺ CD3⁺ T lymphocytes and the CD8αβ⁺ T lymphocytes within this subset were less than 10 cells/μL (HIV-1 seronegative donors, 9; HIV-1 seropositive donors donors > 200 CD4⁺ T cells/μL, 7; HIV-1 seropositive donors < 200 CD4⁺ T cells/μL, 5).

The CD8αβ molecule was only rarely expressed on NK cells (Tables 2and 3). Less than 1.7% of NK cells in PBL of HIV-1 seronegative and seropositive donors expressed the CD8αβ-heterodimer (Table 2). In the small subset of peripheral blood T lymphocytes expressing TCRγδ fewer cells expressed the CD8αβ-heterodimer than the CD8αα-homodimer (Tables 2 and 3). We found no significant difference in the absolute counts of TCRγδ⁺ T cells between these groups of subjects. However, more (percent and absolute counts) CD8αβ⁺TCRγδ⁺ T lymphocytes were observed in HIV-1–infected than in normal donors; higher numbers of this cell subset were also seen in patients with > 200 CD4⁺ T cells/μL than in individuals with < 200 CD4⁺ T cells/μL (Table 3). The CD8αβ-heterodimer was only expressed on a subpopulation of the small subset of CD4⁺ T-helper cells, which expressed the CD8α-chain. We observed a biologically negligible increase in the percentage of CD8αβ⁺CD4⁺ T lymphocytes in the blood of HIV-1–infected individuals and a decrease in absolute counts of CD8αβ⁺CD4⁺ T lymphocytes in donors with < 200 CD4⁺ T cells/μL (Tables 2 and 3).

These observations indicate that the MoAb 2ST8-5H7, directed against the CD8αβ-heterodimer, identifies most CD8⁺ T lymphocytes in HIV-1 seronegative donors, and virtually all CD8⁺ T lymphocytes in HIV-1 seropositive individuals. Therefore, we used this anti-CD8αβ MoAb in place of the conventional antibody combination anti-CD3 and anti-CD8α to detect CD8⁺ T lymphocytes in the four-color analyses in these studies. By using this MoAb, an additional fluorescence channel was available in multiparameter flow cytometric analysis for characterization of discrete CD8⁺ T lymphocyte subsets.

Interestingly, we observed a decrease in the MF of the binding of this CD8αβ-heterodimer–specific antibody on CD8⁺ T lymphocytes of 101 HIV-1 seropositive individuals as compared with 21 seronegative individuals. When analyzed for expression of the CD8α-chain using an anti-CD8α–specific MoAb, CD8⁺ T lymphocytes of HIV-1 seronegative and seropositive blood donors showed a similar MF. Data from representative blood donors are shown in Fig 3.

We sought to determine whether the extent of decrease in CD8αβ expression by PBL correlated with clinical disease status in the study subjects. Linear regression analysis of CD8αβ MF of PBL and absolute CD4 cell counts from the HIV-1 seronegative subjects showed no significant deviation from a slope of 0 (Fig 1A). However, we detected a significant positive correlation between CD8αβ MF and absolute CD4 cell counts (slope, 0.007) when we investigated the data values of the HIV-1 seropositive subjects (Fig 1B). In fact, we found significant differences (HIV-1 seronegative v either HIV-1 seropositive patient group: P < .0001; between HIV-1 seropositive patient groups: P < .01) in the CD8αβ MF between PBL of each pair of the three donor groups (median CD8αβ MF, 41 channel [HIV-1 seronegative]; 27 channel [HIV-1 seropositive with > 200 CD4⁺ T cells/μL]; and 21 channel [HIV-1 seropositive with < 200 CD4⁺ T cells/μL]) (Fig 1C).

The association of the intensity of the CD8αβ staining and cell expression of activation, adhesion, and maturation-associated molecules was then explored. Eleven blood specimens from each of the three groups of subjects were studied: HIV-1 seronegative donors, HIV-1 seropositive donors with > 200 CD4⁺ T lymphocytes/μL, and HIV-1 seropositive donors with < 200 CD4⁺ T lymphocytes/μL. Four-color analyses were performed on these samples using 14 different antibody combinations. We observed a heterogeneous expression of CD8αβ in PBL from subjects in all three donor populations. Based on this heterogeneity, CD8⁺ T lymphocytes could be divided into two groups: cells with a relative high MF of anti-CD8αβ staining and cells that showed only a low to intermediate MF of anti-CD8αβ staining.

In PBL of HIV-1 seronegative individuals, we could demonstrate a pattern of expression of activation-, adhesion-, and maturation-antigens that was associated with the intensity of CD8αβ-specific staining. CD8αβ⁺ lymphocytes that expressed HLA-DR, C1.7, CD11a, CD45RO, CD49d, CD56, CD57, or CD58 showed a significantly lower CD8αβ MF than did lymphocytes not expressing these molecules (Table 4, Fig4). Furthermore, CD8αβ⁺ lymphocytes that expressed CD28, CD45RA, or CD62L showed a significantly higher CD8αβ MF than lymphocytes not expressing one of these molecules. These observations suggest that CD8⁺ T lymphocytes from normal donors can be divided into a fraction containing nonactivated cells with higher levels of CD8αβ expression and a fraction of activated cells with a lower level of CD8αβ expression.

In PBL of HIV-1–infected individuals, we similarly detected significant differences in the CD8αβ MF of specific T-cell subsets, although the differences were not as large as in HIV-1 seronegative individuals (Table 4, Fig 4). Lower levels of CD8αβ MF were seen on cells expressing C1.7, CD11a, CD49d, CD56, or CD57. Higher levels of anti-CD8αβ MF were seen on cells expressing CD28 or CD62L. Interestingly, in PBL of most HIV-1–infected blood donors, we did not find significant differences in CD8αβ MF on CD8⁺lymphocyte subsets which expressed either CD45RA or CD45RO (Table 4).

Blood specimens from eight HIV-1–infected study subjects receiving HAART were investigated for expression of CD8αβ MF. We found a significant increase in the MF of CD8αβ staining of PBL during the 16 weeks after initiation of therapy (median increase, 11 channel;P < .008 [two-sided Wilcoxon test]). One patient had an increase of only 5 channels, the other seven patients had an increase of at least 10 channels. The median absolute count of CD4⁺T lymphocytes increased from 500 to 647 cells/μL, and the median absolute count of CD8⁺ T lymphocytes from 910 to 960 cells/μL. Interestingly, the subject with the highest increase in CD4⁺ T lymphocytes (before treatment: 400 cells/μL and after treatment: 780 cells/μL) showed the greatest increase of CD8αβ MF (Fig 5B). In three individuals, the CD8αβ MF after 16 weeks of HAART was higher than the median CD8αβ MF in the HIV-1 seronegative individuals.

In most investigative studies and in routine clinical immunophenotyping, the expression of the CD8 molecule on human lymphocytes is determined with antibodies that recognize the α-chain of the CD8αα-homodimer and the CD8αβ-heterodimer. Some heterogeneity in the intensity of staining of cells using anti-CD8α antibodies has been observed in those studies.23,26,27,32,34 The difference in the intensity of anti-CD8α staining has been used to distinguish dim CD8α⁺ NK cells from bright CD8α⁺ T cells.35,36 However, these two CD8α⁺ cell fractions overlap. Therefore, the use of a MoAb specific for CD3 has been recommended in addition to an anti-CD8α MoAb to differentiate CD8⁺ T cells from non-T cells in routine immunophenotyping.25-27 

In the present study, we have shown that a MoAb directed against the CD8αβ-heterodimer, 2ST8-5H7, binds almost exclusively to CD3⁺ T cells and recognizes nearly all CD8α⁺T cells from HIV-1 seronegative and seropositive individuals. A small subset of CD8⁺ T lymphocytes expresses only the CD8αα-homodimer and, presumably, matures through extrathymic pathways.37 Interestingly, this cell subset represents a significantly lower percentage of CD8⁺ T lymphocytes after HIV-1 infection. The MoAb, 2ST8-5H7, binds to virtually no NK cells, to substantially fewer TCRγδ⁺ T cells and CD4/CD8 coexpressing T cells than does a CD8α-chain–specific MoAb. These observations suggest that this CD8αβ-specific MoAb is an acceptable substitute for the anti-CD3 and anti-CD8α MoAb combination to detect CD8⁺ T lymphocytes.

Roederer et al38 have shown a small decrease in the density of CD8α expression on CD8⁺ T lymphocytes in PBL of HIV-1–infected individuals. We observed little difference in the CD8α MF between T lymphocytes from HIV-1 seronegative and seropositive individuals. However, the expression of the CD8β-chain was much more heterogeneous in PBL of those donors. The CD8αβ MF was greater in PBL of HIV-1 seronegative individuals than in PBL of HIV-1 seropositive individuals. Furthermore, PBL of HIV-1 seropositive individuals with more than 200 CD4⁺ T lymphocytes/μL blood showed a higher CD8αβ staining than PBL of HIV-1 seropositive individuals with fewer than 200 CD4⁺ T lymphocytes/μL blood. Moreover, using a panel of MoAbs directed against molecules associated with activation, adhesion, maturation, or cytotoxic function, we found that nonactivated CD8⁺ T lymphocytes from HIV-1 seronegative and seropositive individuals had significantly higher levels of anti-CD8αβ staining than did CD8⁺ T lymphocytes with phenotypic evidence of activation. The finding of a decreased cell surface expression of the CD8β-chain on activated CD8⁺ T lymphocytes is consistent with the previous observation that the expression of the CD8β-chain on CD8⁺T lymphocytes decreases following in vitro culture.19 39 

Interestingly, the MF of CD8αβ staining of peripheral blood mononuclear cell (PBMC) from HIV-1 seropositive individuals did not correlate in all instances with the activation or maturation status of CD8α⁺ T cells, as determined by the expression of molecules associated with activation (HLA-DR and CD38) or maturation (CD45RA and CD45RO). This may be due to the fact that the cell surface expression of molecules associated with activation on PBMC of HIV-1–infected individuals does not fully reflect the functional activity of the cells. In addition, it is well known that the CD45RA/CD45RO naive/memory paradigm has not proven particularly useful in analyzing CD8⁺ T cells.40 A substantial fraction of CD45RA⁺ CD8⁺ T cells are probably not true naive cells.

Treatment of HIV-1 infections with HAART results in a substantial impact on viral load and peripheral blood CD4⁺ T lymphocyte counts.41-43 Significant changes in the CD8⁺ T lymphocyte subset in the peripheral blood of treated individuals have also been seen. An early increase and eventual fall in the number of circulating CD8⁺ T lymphocytes has been reported after initiation of HAART.41 Because persistent virus replication is responsible for driving the chronic activation of the immune system and HAART decreases this virus replication, treatment is associated with a decrease in the expression of activation-associated molecules, including HLA-DR and CD38, on CD8⁺ T lymphocytes.41 43 In the present study, we found that the CD8αβ MF in PBL of HIV-1–infected individuals significantly increases during the first 16 weeks of HAART. As we have shown that a relative high MF of staining of the CD8αβ-heterodimer is predominantly associated with nonactivated cells, our observation of treatment-associated increases in CD8αβ MF in PBL of HIV-infected individuals is consistent with a reduction in CD8⁺T-lymphocyte activation.

It is difficult to speculate as to the biologic ramifications of the decreased expression of CD8αβ on CD8α⁺ T cells of HIV-1–infected individuals because it is not clear whether CD8αα and CD8αβ have different cellular functions. It has been suggested that the β-chain of the CD8αβ-heterodimer increases the functional activity of the CD8 molecule.44 Others were unable to substantiate this observation.45 However, the observation made in the present study does indicate that CD8αα and CD8αβ expression are differentially regulated.

For excellent technical support in the antibody-fluorochrome conjugations and optimization of the reagent combinations used, we are indebted to Kirt Toussaint, Ed O'Connell, Lisa Edwards, and MaryLyn Monson. For donation of blood samples, we are grateful to all participating HIV-1 seronegative and seropositive blood donors. For collection of HIV-1 seropositive blood samples, we thank Beryl Chapman, Karen A. McLaughlin, and Dr Jennifer Adelson-Mitty.