Human Immunodeficiency Virus Type 1 Induces Apoptosis in CD4⁺ but Not in CD8⁺ T Cells in Ex Vivo-Infected Human Lymphoid Tissue

Abstract

Progression of human immunodeficiency virus (HIV) disease is associated with massive death of CD4⁺ T cells along with death and/or dysfunction of CD8⁺ T cells. In vivo, both HIV infection per se and host factors may contribute to the death and/or dysfunction of CD4⁺ and CD8⁺ T cells. Progression of HIV disease is often characterized by a switch from R5 to X4 HIV type 1 (HIV-1) variants. In human lymphoid tissues ex vivo, it was shown that HIV infection is sufficient for CD4⁺ T-cell depletion. Here we address the question of whether infection of human lymphoid tissue ex vivo with prototypic R5 or X4 HIV variants also depletes or impairs CD8⁺ T cells. We report that whereas productive infection of lymphoid tissue ex vivo with R5 and X4 HIV-1 isolates induced apoptosis in CD4⁺ T cells, neither viral isolate induced apoptosis in CD8⁺ T cells. Moreover, in both infected and control tissues we found similar numbers of CD8⁺ T cells and similar production of cytokines by these cells in response to phorbol myristate acetate or anti-CD3–anti-CD28 stimulation. Thus, whereas HIV-1 infection per se in human lymphoid tissue is sufficient to trigger apoptosis in CD4⁺ T cells, the death of CD8⁺ T cells apparently requires additional factors.

AIDS is a multifactorial disease which starts with human immunodeficiency virus (HIV) infection of CD4⁺ cells and eventually leads to deterioration of lymphoid tissue (12). This is accompanied by massive death of CD4⁺ T cells along with death and/or dysfunction of various CD4⁻ cells thought not to be infectible by the virus (7, 16, 25, 56). Among CD4⁻ cells, CD8⁺ T cells are thought to play a major role in HIV disease control as a source of soluble factors (8, 38) and as cytotoxic T lymphocytes (CTLs) (63). Failure of these cells to contain HIV infection is thought to be a major cause of HIV disease progression (63). Early stages of HIV disease, when CCR5-tropic (R5) HIV variants dominate (10, 54, 55), are characterized by expansion (14) and accelerated turnover (53, 65) of CD8⁺ T cells. At later stages of HIV disease, when variants that use either CXCR4 alone (X4) or both CXCR4 and CCR5 (R5/X4) often become dominant (10, 54), the absolute number of CD8⁺ T cells gradually decreases (14, 39). The emergence of X4 or R5X4 variants often coincides with a rapid decline in T-cell counts and the development of AIDS (55, 60).

In vivo, both HIV infection per se and host factors may contribute to the death and/or dysfunction of CD4⁺ and CD8⁺ T cells in lymphoid tissue (11, 47). From our experiments with ex vivo human lymphoid tissue, we know that productive HIV infection alone is sufficient to cause CD4⁺ T-cell depletion (20, 49, 59). Whereas productive R5 infection of lymphoid tissue ex vivo depletes the small fraction of CD4⁺ T cells (20, 49) that express CCR5 (28), X4 infection severely depletes the entire CD4⁺ T-cell population (20, 28, 49). Here, we address the question of whether infection of human lymphoid tissue ex vivo with R5 or X4 HIV variants also depletes or impairs CD8⁺ T cells. We compared the frequency of apoptosis and the absolute number of CD4⁺ and CD8⁺ T lymphocytes in human lymphoid tissue infected ex vivo with prototypic X4 and R5 isolates. We found that whereas HIV type 1 (HIV-1) infection per se is sufficient to induce apoptosis of tissue CD4⁺ T cells, it neither triggers apoptosis of CD8⁺ T cells nor decreases the responsiveness of these cells to activators.

MATERIALS AND METHODS

Viruses.

Prototypic R5 HIV-1 isolate SF162 and X4 isolate LAV.04 were obtained through the AIDS Research and Reference Reagent Program.

HIV infection of human lymphoid tissue ex vivo.

Human tonsils removed during routine tonsillectomy were received within several hours of excision. The tonsils were washed thoroughly with medium containing antibiotics and then sectioned into 2 to 3-mm blocks with an average weight of 3 mg. To compare the absolute number of lymphocytes in infected and control tissues, blocks were matched by size and weight. Tissue blocks were placed on top of collagen sponge gels in the culture medium at the air-liquid interface and infected the next day as described earlier (20, 21). The spontaneous loss of cells from the tissue blocks occurs in the first 36 h after dissection. Therefore, in several experiments infections were delayed for 36 h. In a typical experiment, 3 to 5 μl of clarified virus-containing medium (approximately 300 50% T-cell infective doses/block) was applied to the top of each tissue block. At this concentration both viruses replicate to approximately the same level (20). Productive HIV infection was assessed by measuring p24 in the culture medium using an HIV-1 p24 antigen enzyme-linked immunosorbent (ELISA) (AIDS Vaccine Program, National Cancer Institute, Frederick, Md.): specifically, the concentration of p24 accumulated in 4 ml of culture medium bathing nine tissue blocks during the 3 days between the successive medium changes was used as a measure of virus replication.

Flow cytometry.

Flow cytometry was performed on cells mechanically isolated from control and infected tissue blocks (21). Lymphocytes were first identified according to their light-scattering properties and then analyzed for expression of lymphocyte markers. More than 95% of cells in the lymphocytic gate express lymphocytic markers (21) and exclude propidium iodide. Depletion of CD4⁺ T cells was assessed as described earlier (21). For determination of the CD4⁺ and CD8⁺ T cells, cells were stained for surface markers using anti-CD3-TriColor, anti-CD4-phycoerythrin (PE), and anti-CD8-fluorescein isothiocyanate (FITC) antibodies (Caltag, Burlingame, Calif.). To enumerate cells in tissue blocks, we used Trucount tubes (Becton Dickinson, San Jose, Calif.) containing a known number of fluorescent beads. The absolute number of cells in the sample can be determined by normalizing to the number of acquired beads. To evaluate proliferation, cells were first stained with anti-CD3-TriColor, anti-CD8-FITC, and anti-CD4 allophycocyanin (APC) antibodies, washed, fixed and permeabilized with Cytofix-Cytoperm (PharMingen, San Diego, Calif.) according to the manufacturer's protocol, and then stained with anti-Ki67-PE (PharMingen) or with control isotype antibodies. To evaluate activation, cells were stained with anti-CD8-FITC, anti-CD27-PE, and anti-CD28-TriColor antibodies and with APC-labeled antibodies against either HLA-DR or CD38 and then fixed. Cells were acquired on a FACSCalibur using CellQuest software for both acquisition and analysis.

Apoptosis.

Cells were stained for one of the following cell surface antigen combinations: CD3-TriColor, CD8-FITC, and CD4-APC (Caltag); or CD4-TriColor (Caltag), CCR5-APC, and CXCR4-APC (PharMingen). After washing, cells were fixed and permeabilized using Cytofix-Cytoperm (PharMingen) according to the manufacturer's instructions and stained for the mitochondrial antigen 7A6 using monoclonal antibody Apo 2.7-PE (Immunotech, Marseille, France). In another apoptosis assay, the activity of caspase 3 was measured by the fluorogenic substrate PhiPhiLux (OncoImmunin, Inc., College Park, Md.). Cells were isolated from tissue blocks, thoroughly washed, incubated with the substrate solution, and then stained for cell surface antigens. Cells were analyzed by flow cytometry. For a positive control for apoptosis, peripheral blood mononuclear cells (PBMCs) were incubated overnight with 3 μM staurosporine (Sigma, St. Louis, Mo.). The number of 7A6⁺ cells increased from 5% to more than 60%.

Cytokine production in response to stimulation.

Lymphocytes were mechanically isolated from control and infected tissue blocks 8 to 10 days postinfection. Cells were counted and stimulated either with phorbol myristate acetate (PMA) plus ionomycin or with anti-CD3 and anti-CD28 antibodies. For direct protein kinase C (PKC) stimulation, 10⁶ cells/ml were incubated with PMA (50 ng/ml) and ionomycin (500 ng/ml Sigma) in the presence of brefeldin A (GolgiPlug; Pharmingen) for 4 h. Cells were washed, stained for cell surface antigens CD8-FITC and CD3-TriColor (Caltag), and then fixed and permeabilized with Cytofix-Cytoperm according to the manufacturer's instructions. Permeabilized cells were stained with either of three cytokine-specific monoclonal antibodies (interleukin-2 [IL-2]-PE, gamma interferon [IFN-γ]-PE, and tumor necrosis factor alpha [TNF-α]-PE [Caltag]) and relevant isotype controls. To stimulate CD8⁺ T cells via the T-cell receptor complex, 4 × 10⁶ cells/ml were cultured in anti-CD3 antibody-coated wells (UCHT1; 10 μg/ml; PharMingen) in the presence of soluble anti-CD28 antibody (CD28.2; 2 μg/ml; PharMingen). Cells were washed, permeabilized, and stained for cytokine production as described above. Preliminary experiments with noninfected tissues showed that optimal stimulation was achieved at 40 h. Therefore, we chose this time point to compare stimulation of CD8⁺ T cells in control and HIV-infected tissues.

Experimental analysis.

Data obtained with tissue from one donor constitute the results of one experiment. Both viral replication and the proportion of cells in various leukocyte subsets varied from tissue to tissue (21). To compare results obtained in different experiments, we normalized the data for such variation: for each experiment we compared infected and control tissues in replicates of at least three histocultures of nine tissue blocks each, obtained from an individual donor. To average the results of different experiments and analyze them statistically, we normalized the data as percentage of the control.

RESULTS

We compared apoptosis in CD4⁺ and CD8⁺ T lymphocytes in human lymphoid tissue infected ex vivo with prototypic X4 isolate LAV.04 or prototypic R5 isolate SF162. Apoptosis was evaluated by flow cytometry using antibodies against the mitochondrial membrane antigen 7A6, which is exposed at early stages of apoptosis (36, 41), before cells lose their normal optical characteristics, surface markers, and become permeable to propidium iodide. Therefore, we were able to confine our analysis of apoptotic cells to the lymphocytic gate as defined by a light scatter plot in which cells are readily identifiable.

Both isolates productively infected tissue blocks with the typical kinetics described earlier (20, 49) and presented in Fig. 1. Viral replication, as evaluated by p24 concentration in culture medium, was detectable typically between days 6 and 9 postinfection and continued to increase until days 12 to 13 postinfection. At that time point in LAV.04-infected cultures, CD4⁺ T lymphocytes were depleted by about 80%. Among the remaining CD4⁺ T cells, apoptosis, as defined by expression of 7A6 mitochondrial antigen, was fourfold higher than in matched uninfected controls (Fig. 2a), which on average was 6 ± 2% (mean ± standard error of the mean [SEM]; n = 6). (When data for all uninfected control tissues were pooled, the average proportion of apoptotic cells on days 11 to 13 in culture was 5% ± 1% [n = 11].) An increase in CD4⁺ T-cell apoptosis in LAV.04-infected tissues was also revealed in experiments with PhiPhiLux, which measures caspase 3 protease activity (data not shown). In contrast, the frequency of apoptosis among CD8⁺ T cells from the same LAV.04-infected tissues was similar to that from uninfected controls (Fig. 2a). Apoptosis in both CD4⁺ and CD8⁺ T cells from SF162-infected tissue blocks was also similar to that from uninfected controls (Fig. 2a).

FIG. 1.

Replication of HIV-1 in tissues infected with X4 and R5 isolates. Human tonsils removed during routine tonsillectomy were dissected into 2 to 3-mm blocks inoculated with virus-containing medium and cultured as described in Materials and Methods. Productive HIV infection was assessed by measuring p24 released into the culture medium.

FIG. 2.

Apoptosis of T lymphocytes in tissues infected with X4 and R5 HIV-1 isolates. Lymphocytes mechanically isolated from control and ex vivo-infected tissues were stained for CD3, CD4, CD8, and the apoptosis-reporting mitochondrial antigen 7A6 and then analyzed by flow cytometry. (a) Apoptosis in CD4⁺ and CD8⁺ T cells from control and HIV-infected tissues (mean + SEM [n = 6]). Note the significant increase in the number of apoptotic CD4⁺ T cells remaining in the LAV.04 (X4)-infected but not SF162 (R5)-infected tissue, while there is no increase in apoptosis of CD8⁺ T cells. (b) Apoptosis in CD4⁺ and CD8⁺ T cells isolated at different time point postinfection (mean + SEM [n = 3]). In control uninfected tissues, the proportion of apoptotic CD4⁺ and CD8⁺ T cells was between 5 and 7% and 3 and 5%, respectively. Note the increase in apoptosis in CD4⁺ but not CD8⁺ T-cell subset. (c) Apoptosis in CXCR4⁺ and CCR5⁺ subsets of CD4⁺ T cells isolated from LAV.04- and SF162-infected tissues on day 12 postinfection (mean + SEM [n = 3]). Note the selective increase in the number of apoptotic CCR5⁺ CD4⁺ T cells in tissues infected with the R5 HIV isolate.

To follow its kinetics, we evaluated apoptosis in CD4⁺ and CD8⁺ T cells from LAV.04-infected tissues at different time points after infection. Apoptosis in uninfected tissues remained low and varied between 5 and 7% for CD4⁺ T cells and between 4 and 5% for CD8⁺ T cells (n = 3). LAV.04 infection increased apoptosis in CD4⁺ T cells (Fig. 2b). The increase of apoptosis became significant on day 6 postinfection (P = 0.016, n = 4), and the difference between control and infected tissue continued to increase thereafter. At none of the time points was apoptosis significantly elevated in CD8⁺ T-cell subset (P = 0.84 on day 6 and P = 0.53 on day 12) (Fig. 2b).

Since analysis of general CD4⁺ T-cell population in SF162-infected tissues revealed no apoptosis, we analyzed whether R5 and X4 HIV variants induced apoptosis differently in CD4⁺ T cells expressing viral coreceptors CCR5 or CXCR4 on their surface. More than 80% of CD4⁺ T lymphocytes in cultured human lymphoid tissue express CXCR4 but not CCR5, and less than 10% express CCR5 but not CXCR4 (28). LAV.04 increased apoptosis almost sixfold in both CXCR4⁺ CCR5⁻ and CXCR4⁻ CCR5⁺ subsets of CD4⁺ T cells (Fig. 2c). In SF162-infected tissues, apoptosis was increased more than fourfold in the CXCR4⁻ CCR5⁺ subset selectively but did not change at all in CXCR4⁺ CCR5⁻ subset of CD4⁺ T cells (Fig. 2c).

Apoptosis measurements represent a snapshot, which does not account for cells that had already undergone apoptosis and had disintegrated. Therefore, we measured the cumulative effect of apoptosis by enumerating CD4⁺ and CD8⁺ T cells over the course of infection. On day 0, six control uninfected tissues from different donors contained on average 41,900 ± 7,200 CD4⁺ T cells and 7,500 ± 1,400 CD8⁺ T cells per mg. The number of T cells in these control blocks decreased over the next 24 h and then stabilized (Fig. 3a and b). On average, the numbers of CD4⁺ T cells on days 1 and 13 were similar (P = 0.08). The same was true for CD8⁺ T cells (P = 0.47). As was shown earlier and confirmed here, CD4⁺ T cells in the same tissue blocks were depleted by almost 80% by days 11 to 13 of LAV.04 infection compared with matched uninfected control (Fig. 3c). In contrast to CD4⁺ T cells, there was no significant difference (P = 0.07, n = 13) in CD8⁺ T-cell counts between LAV.04-infected and matched uninfected controls (Fig. 3c). In the case of SF162-infected tissues, there was mild but significant (P = 0.02) depletion of CD4⁺ T cells, whereas depletion of CD8⁺ T cells was not significant (P = 0.75) (Fig. 3c). To determine whether the number of CD8⁺ T cells was conserved over the entire 13-day period of experimental infection, we compared the absolute numbers of CD4⁺ and CD8⁺ T cells in infected and control tissues at different time points postinfection (Fig. 3d). Whereas the number of CD4⁺ T cells in LAV.04-infected tissue blocks declined starting from about day 6 postinfection, the numbers of CD8⁺ T cells remained similar to those in uninfected controls at all time points (P > 0.06).

FIG. 3.

Number of CD4⁺ and CD8⁺ T cells in tissues infected with X4 and R5 HIV-1 isolates. Blocks of human tonsils of equal size and weight were infected ex vivo with HIV-1 and maintained in culture for 11 to 13 days. Lymphocytes were isolated and enumerated by flow cytometry at different time points postinfection. (a and b) Absolute numbers of CD4⁺ (a) and CD8⁺ (b) T cells in uninfected control tissues (mean ± SEM [n = 6]). (c) Numbers of CD4⁺ and CD8⁺ T cells in SF162-infected (n = 3) and LAV.04-infected (n = 13) tissues on days 11 to 13 postinfection (mean ± SEM). (d) Change in the number of T cells in the course of LAV.04 infection. Each time point represents mean ± SEM numbers of CD4⁺ and CD8⁺ T cells (n = 3 to 8). Note conservation of CD4⁺ and CD8⁺ T-cell numbers in uninfected controls (a and b) versus severe depletion of CD4⁺ T cells and the lack of depletion of CD8⁺ T cells in X4 HIV-1-infected tissues (c and d). R5 HIV-1 infection results in a mild depletion of CD4⁺ T cells and no depletion of CD8⁺ T cells (c).

To ascertain that spontaneous loss of T cells on the first day of culture does not obscure depletion of T cells in HIV-infected tissues, we delayed HIV infection for 36 h, or until the tissue cellularity stabilizes (Fig. 3a and b). This procedure did not change the severe depletion of CD4⁺ T cells and conservation of CD8⁺ T cells. In a typical experiment, there were 590 CD4⁺ T cells/mg in LAV.04-infected tissue and 9,500 CD4⁺ T cells/mg in the uninfected control, whereas for CD8⁺ T cells these numbers were 1,600 and 1,400 cells/mg in LAV.04-infected and control tissues, respectively.

To assess cell turnover in ex vivo human lymphoid tissue, we evaluated T-cell proliferation using Ki67. Ki67 is expressed in cycling cells (19) and is considered to be an adequate marker of cell proliferation (5, 19). To verify that Ki67 was evaluated properly, we used a B-cell lymphoma as a positive control. Ki67 was expressed in approximately 75% of these cells, a level of expression similar to that described for lymphoma cell lines (6). Although approximately 10% of CD4⁺ and 25% of CD8⁺ T cells expressed Ki67 in tonsillar tissues a few hours after tonsillectomy, proliferation dropped fivefold in the first 24 h of culture, and at the time of infection (36 h posttonsillectomy), only 0.7% ± 0.1% of CD8⁺ and 0.6% ± 0.1% of CD4⁺ T cells expressed Ki67 (n = 6). Cell proliferation remained low throughout the experiment. In tissues from three donors, on average 0.8% ± 0.8% and 0.8% ± 0.4% of CD8⁺ T cells and CD4⁺ T cells, respectively, were Ki67⁺ at day 2 postinfection; these numbers were even lower at day 4 (less than 0.25%), and no Ki67⁺ T lymphocytes were detected on day 7 of culture irrespective of LAV.04 infection.

Also, the level of activation was low in human tonsillar tissue: four-color flow cytometry revealed that at day 0, 1 to 2% of CD8⁺ T cells were of HLA-DR⁺ CD27⁻ CD28⁻ or CD38⁺ CD27⁻ CD28⁻ phenotype. On day 5 or 8 postinfection, these cells constitute less than 1% of CD8⁺ T cells irrespective of LAV.04 infection. Thus, in the absence of significant activation, replenishment, and apoptosis, the numbers of CD8⁺ T cells in both HIV-infected and control tissues are conserved.

The question arising from these results is whether HIV tissue infection ex vivo renders CD8⁺ T cells dysfunctional. We evaluated the ability of these cells to produce cytokines in response to the PKC activator PMA. When T cells were stimulated with PMA-ionomycin, there was no significant difference (P > 0.5, n = 3) in cytokine production by CD8⁺ T cells isolated on day 8 postinfection from LAV.04-infected and matched uninfected control tissue (Fig. 4a). To test whether HIV infection impaired the early T-cell receptor signaling events, we stimulated T cells with anti-CD3 and anti-CD28 antibodies. As shown in Fig. 4b, there was no significant difference (P > 0.5, n = 3) between the numbers of IL-2-, TNF-α- and IFN-γ-producing CD8⁺ T cells isolated on day 8 from LAV.04-infected and matched control tissues. Similar results were obtained when CD8⁺ T cells were isolated on day 10 postinfection.

FIG. 4.

Production of IL-2, IFN-γ, and TNF-α in stimulated CD8⁺ T cells in tissues infected with X4 HIV-1 isolate LAV.04. Cells were isolated from control and HIV-infected tissue blocks on days 8 to 10 postinfection, activated, stained for CD3, CD8, and either IL-2, TNF-α, or IFN-γ, and analyzed by flow cytometry. (a) Cells were stimulated with PMA-ionomycin for 4 h in the presence of the secretion inhibitor monensin (mean ± SEM [n = 3]). (b) Cells were stimulated with immobilized anti-CD3 and soluble anti-CD28 antibodies for 40 h (mean + SEM [n = 3]). Note that there is no difference between control and HIV-infected tissues in the number of CD8⁺ T cells producing cytokines.

DISCUSSION

HIV-1 infection causes profound changes in the cells constituting the human immune system. In particular, CD4⁺ T lymphocytes are progressively depleted during the course of infection (12). In contrast, the number of CD8⁺ T cells in HIV-infected patients is initially increased but often declines at late stages of HIV disease (14, 35, 39). The mechanism responsible for the increased death of CD4⁺ and CD8⁺ T cells in HIV-infected human lymphoid tissue is not known. In vivo, the death of T cells is normally balanced by their replenishment (reviewed in reference 18). HIV or simian immunodeficiency virus (SIV) infection disturbs this balance (44, 53). Although the data on both CD4⁺ and CD8⁺ T-cell turnover are controversial (15, 33, 46, 51, 65, 66), it is believed that at late stages of HIV/SIV disease the immune system fails to keep pace with cell death (31). Many host factors (e.g., chronic immune activation [9, 12]) modulate HIV-triggered T-cell death and replenishment in lymphoid tissue, where the critical events in HIV disease occur in vivo. On the other hand, it was reported that in vitro CD8⁺ T-cell death can be caused in PBMCs by HIV infection per se (32).

Apoptosis was implicated as a major mechanism for the death of both CD4⁺ and CD8⁺ T cells in HIV-infected patients (42, 43, 52; reviewed in reference 27). Here, we tested whether this mechanism also applies to isolated tissue blocks infected with HIV-1 ex vivo. We found that (i) R5 HIV-1 productive infection induces apoptosis selectively in the CCR5⁺ subset of CD4⁺ T cells, whereas X4 HIV-1 infection induces apoptosis in both CXCR4⁺ CD4⁺ and CXCR4⁻ CD4⁺ T cells; (ii) neither of these isolates induces apoptosis in tissue CD8⁺ T cells, and their number remained similar to those in uninfected controls; (iii) cytokine production in response to direct PKC or to T-cell receptor stimulation is similar in CD8⁺ T cells from HIV-infected and control tissues. Thus, in lymphoid tissue, HIV productive infection per se is sufficient for the death of CD4⁺ but not CD8⁺ T cells.

The system of human tonsillar tissue ex vivo has been used to study various aspects of HIV pathogenesis (20, 22, 24, 28, 40). This system is different from thymus organ cultures, where 90% of cells undergo gradual apoptosis in 12 days of culture (2). Tonsillar histocultures lose approximately 60% of cells in the first day (probably as a combined result of tissue cutting and background apoptosis), but starting from the next day tissue cellularity stabilizes and the subsequent cell loss is negligible unless the tissue is infected with HIV-1. There is no statistical difference between the number of either CD4⁺ or CD8⁺ T lymphocytes in uninfected tissue at any day of culture starting from day 1.

Human lymphoid tissue ex vivo supports replication of different HIV variants and does not require exogenous stimulation (20, 21). In this regard this system is different from isolated PBMCs. One can speculate that tissue blocks do not require exogenous stimulation because lymphocytes in tonsils are already in an activated state since tonsils have been removed from the patients due to their size and probably were infected by bacteria. However, noninflamed lymph nodes (21), spleen (49), and thymus (2, 50) support ex vivo HIV-1 replication also without exogenous stimulation. Moreover, activation seems not to be an absolute condition for a CD4⁺ T cell to be productively infected in lymphoid tissue either in vivo (69) or ex vivo (23). These results do not necessarily contradict the common notion that it is difficult to infect resting T cells, since the above-mentioned cells may have been in activated state at the time of infection. We think that the endogenous cytokine network together with a preserved system of cell-cell interactions rather than inflammation-related activation make human tonsillar tissue sufficient to support productive HIV-1 infection both in vivo and ex vivo.

In our present work we took advantage of the fact that unlike in vivo, in ex vivo-infected tissues there is no evidence of rapid T-cell turnover, as evaluated by Ki67 expression. Also, in ex vivo-infected human lymphoid tissues the numbers of cells exhibiting phenotypes typical for activated cells, including those commonly associated with CTLs (3, 4, 17, 34), are very low. In this regard, our ex vivo system appears to be different from the situation in vivo (11, 17, 47, 63). Nevertheless, this difference allowed us to evaluate the contribution of HIV infection per se to the death of CD4⁺ and CD8⁺ T cells in lymphoid tissue.

As shown earlier (20, 49) and confirmed here, there is severe depletion of CD4⁺ T cells in ex vivo lymphoid tissues productively infected by X4 isolates, whereas R5 isolates deplete CD4⁺ T cells only mildly. To address the mechanism of this depletion, we evaluated apoptosis in control and infected tissues by measuring expression of the mitochondrial antigen 7A6 (36, 41) and caspase 3 protease activity. The frequency of apoptotic cells in uninfected control tissue remains generally low, with a median (5%) almost equal to that reported for fresh tonsillar tissue (4%) (52). We demonstrated a significant increase in CD4⁺ T-cell apoptosis in lymphoid tissues infected ex vivo by the prototypic X4 isolate LAV.04 but not by the prototypic R5 isolate SF162. This indicates that depletion of CD4⁺ T cells by X4 isolates is the result of cell death rather than their migration out of the tissue. In contrast to the X4 isolate, productive R5 infection increased apoptosis selectively in the CCR5⁺ CXCR4⁻ subset of CD4⁺ T cells. Since CCR5⁺ CXCR4⁻ cells constitute less than 10% of the total CD4⁺ T-cell population (28), the increased apoptosis in this subset does not significantly alter the overall number of apoptotic CD4⁺ T cells. Thus, the R5 variant induces apoptosis in a coreceptor-dependent manner, whereas X4 induces apoptosis irrespective of coreceptor expression. We do not know whether the increased frequency of apoptosis among CXCR4⁻ cells in X4-infected tissues represents a bystander apoptotic death or is a result of direct infection of these cells, which may express CXCR4 below the threshold of detection but at a level that is sufficient for infection. Whatever the mechanism of the differential apoptosis in CCR5⁺ and CXCR4⁺ subsets of tissue CD4⁺ T cells, it accounts for the differential depletion of these cells by R5 and X4 HIV-1 isolates described earlier (28).

The difference in the ability of R5 and X4 isolates to trigger apoptosis is additional evidence that the emergence of X4 viruses may be sufficient for massive CD4⁺ T-cell depletion and progression to AIDS (10, 13, 37, 54, 60). Our results with isolated lymphoid tissue indicate that productive viral infection is sufficient to trigger apoptosis and consequent cell death in cognate CD4⁺ T-cell targets.

In sharp contrast to the effect of HIV infection on CD4⁺ T cells, neither X4 nor R5 infection induced apoptosis or significantly depleted CD8⁺ T cells in ex vivo tissue. Also, the level of proliferation of CD8⁺ T cells was low, as demonstrated by Ki67 measurements. Therefore, we conclude that in ex vivo tissues there is no significant turnover of CD8⁺ T cells. When the absolute numbers of T cells were repeatedly measured in the course of 13 days of infection, CD4⁺ T-cell counts started to drop on day 6 postinfection whereas CD8⁺ T-cell counts even slightly increased at the end of experiment. However, based on our present data this increase in the number of CD8⁺ T cells is not statistically significant. If proven, such an increase may reflect the observed but statistically not significant low level of proliferation unbalanced by apoptosis. Nevertheless, our data demonstrate that the numbers of CD8⁺ T cells is conserved in this system in spite of HIV infection.

In HIV-infected individuals, peripheral CD8⁺ T cells were reported to be dysfunctional (4, 30, 62). Here, we asked whether CD8⁺ T cells in HIV-infected tissue blocks produce IL-2, TNF-α, and IFN-γ upon PMA stimulation differently than do CD8⁺ T cells from matched uninfected controls. In neither case we did find any difference in this regard between CD8⁺ T cells isolated from HIV-infected and matched control tissue blocks. Stimulation of CD8⁺ T cells with PMA bypasses the early signaling events and affects PKC directly. To assess whether CD8⁺ T cells from infected culture responded to physiological stimuli, we treated them with anti-CD3 and anti-CD28 antibodies that stimulate T cells through the T-cell receptors. This assay revealed no difference between CD8⁺ T cells from infected tissue and uninfected control either on day 8 when CD4⁺ T cells become apoptotic and die or at later days when the depletion of CD4⁺ T cells becomes severe. Thus, in human lymphoid tissue, ex vivo HIV infection neither affects CD8⁺ T-cell responsiveness to stimulation nor induces apoptosis.

In this regard, the fate of CD8⁺ T cells in HIV-infected human lymphoid tissue ex vivo is different from what was reported for CD8⁺ T cells in vivo (16, 45, 52). In particular, increased CD8⁺ T-cell apoptosis was found in tonsils from HIV-infected individuals (52). The discrepancy between the increased level of CD8⁺ T-cell apoptosis in HIV-infected lymphoid tissue in vivo and the lack of such an increase in ex vivo-infected tissues may be related to the possible difference in the level of Tat, Nef, Vpr, or other factors that were implicated in apoptosis (57, 58, 64, 67) in various experimental settings.

The difference in activation status of cells in in vivo and ex vivo tissues may be another reason for the difference in CD8⁺ T-cell apoptosis in these systems. Activation of lymphoid tissue is one of the hallmarks of HIV disease in vivo. Inappropriate overstimulation of CD8⁺ T cells in lymph nodes from HIV-infected patients was documented (1). In contrast, evaluation of cytokine production did not reveal any difference between CD8⁺ T cells isolated from control and ex vivo-infected tissue blocks.

There is a close correlation between apoptosis and the general state of cellular activation (26, 45). Activated CD8⁺ T cells that infiltrate the lymph nodes of HIV-infected patients were shown to be prone to apoptosis (1). Moreover, CD8⁺ T cells that were not apoptotic at the time of isolation from HIV-infected individuals enter apoptosis when activated in vitro or simply left in culture (26, 29, 68). The absence of chronic stimulation of CD8⁺ T cells in ex vivo lymphoid tissue may be the primary reason for the lack of apoptosis in these cells upon HIV infection. This conclusion is in general agreement with the in vivo data that in lymph nodes of HIV-infected individuals apoptosis correlates not with viral burden but rather with the general state of activation (45).

Our results seem to contradict an earlier report that X4 but not R5 HIV-1 isolates induce apoptosis of CD8⁺ T cells in mixed suspensions of phytohemagglutinin-activated PBMCs and macrophages (32). The discrepancy between the two systems may be explained by the fact that PBMCs have to be artificially stimulated by phytohemagglutinin and IL-2 in order to survive in vitro and to support productive HIV infection. In contrast, human lymphoid tissue ex vivo is self-sufficient and requires exogenous stimulation neither for maintenance nor for productive HIV infection (20). In PBMC cultures, HIV-triggered CD8⁺ T-cell apoptosis was dependent on the presence of macrophages, with a significant level of apoptosis reached at a 1:4 to 4:1 macrophage/lymphocyte ratio (32). In both ex vivo and in vivo lymphoid tissues, the ratio of macrophages to lymphocytes is much lower. Also, it is conceivable that within the tissue structure, CD8⁺ T cells are somehow more protected from HIV-induced apoptosis than those in suspension and that the resident lymphocytes are therefore different in this respect from those in the bloodstream. Systematic comparison of the two experimental systems with respect to CD8⁺ T-cell apoptosis will increase our understanding of the mechanisms of CD8⁺ T-cell death during HIV infection in vivo.

In conclusion, in human lymphoid tissue ex vivo, HIV-1 induces significant apoptosis in CD4⁺ T cells but not CD8⁺ T cells. These results suggest that the mechanisms of HIV-triggered death of CD4⁺ and CD8⁺ T cells in vivo may be different. Whereas in lymphoid tissue HIV-1 productive infection is sufficient to trigger apoptosis and consequent depletion of CD4⁺ T cells, the death of CD8⁺ T cells requires additional factors. Chronic activation of the immune system may be one of the factors necessary to trigger CD8⁺ T-cell death, as suggested earlier (1, 15, 26, 34, 45, 48, 61, 62). Human lymphoid tissue ex vivo allows one to study the contribution of productive HIV-1 infection per se to the death and/or dysfunction of CD4⁺ and CD8⁺ T cells.