HIV-1 Directly Kills CD4⁺ T Cells by a Fas-independent Mechanism

CEM is a human CD4⁺ T lymphoblastoid cell line originally isolated from a child with acute leukemia (20, 21). SupT1 is a CD4⁺ T cell line isolated from a pleural effusion of a non-Hodgkin's lymphoma patient (22). Jurkat clone E6 is a CD4⁺CD3⁺ acute T cell leukemia line (23). CEM (from Peter L. Nara), SupT1 (from James Hoxie), and Jurkat clone E6 (from Arthur Weiss) were obtained from the NIH AIDS Research and Reference Reagent Program. The hFasL/L5178 cell line, which stably expresses human FasL (hFasL; reference 24), was provided by Hideo Yagita of the Juntendo University School of Medicine, Tokyo, Japan. The CEM, SupT1, Jurkat, and hFasL/L5178 cell lines were grown in RPMI 1640 medium supplemented with 10% fetal bovine serum, 100 U/ml penicillin, 100 μg/ml streptomycin, and 2 mM glutamine. PBMCs from HIV-seronegative blood donors were purified by centrifugation of buffy coat samples over a Ficoll-Hypaque (Pharmacia Biotech AB, Uppsala, Sweden) density gradient. CD4⁺ T cells were purified by negative selection of PBMCs using a CD4⁺ T cell enrichment column (R&D Systems, Minneapolis, MN) according to the manufacturer's recommendations. Patients with autoimmune lymphoproliferative syndrome (ALPS) arising as a result of mutations in Fas and their families were evaluated and followed at the Warren Magnuson Clinical Center at the National Institutes of Health as part of an Institutional Review Board–approved clinical research protocol. PBMCs were isolated as above and T cells and EBV-transformed B cells were tested for sensitivity to anti-Fas antibody (CH11; Kamiya Biomedical, Seattle, WA) and plate-immobilized anti-CD3ε monoclonal antibody (64.1) as previously described (25, 26). PBMCs from these families were cryopreserved after donation. Before infection with HIV-1, PBMCs and purified CD4⁺ T cells were activated for 2 d in culture medium containing 5 μg/ml PHA (Sigma Chemical Co., St. Louis, MO), and then maintained in culture medium with 100 U/ml recombinant IL-2 (Genzyme Corp., Cambridge, MA).

The HIV molecular clone NL43 (27) was obtained from Dr. Malcolm Martin through the AIDS Research and Reference Reagent Program. The HIV-1 molecular clone HXB2 was provided by Dr. Mark Feinberg (Emory University, Atlanta, GA; R7 clone) and contains a repaired nef open reading frame (28). The construction of PLAP-expressing NL43 (NL-PI) and PLAP-expressing HXB2 (HXBnPLAP) have been previously described (29, 30). Virus was produced by calcium phosphate transfection of HIV DNA constructs into 293 cells and viral supernatants were harvested at 48 h after transfection as previously described (30). Virus was quantitated by p24 ELISA to normalize all infections to equivalent antigenic input (31).

Target cells at 1–2 × 10⁶ cells/ml were incubated with HIV-1 (NL43 or HXB2) or HIV-PLAP at different multiplicities of infection in the presence of 4 μg/ml polybrene for 8–12 h at 37°C in a humidified incubator. Infection of PBMCs with HIV-PLAP was carried out by spin infection as follows: 1–2 × 10⁶ cells in 1 ml of culture medium containing HIV-PLAP at different multiplicities of infection and 4 μg/ml polybrene were centrifuged in a sealed biosafety container for 90 min at 2,000 rpm in a Jouan CR4.12 centrifuge (Jouan Inc., Winchester, VA), and then returned to a humidified incubator for 8–12 h. After infection, cells were washed with 5 volumes of PBS and then resuspended in culture medium. Cultures were split every 2–3 d to maintain a concentration of 5–20 × 10⁵ cells/ml. For experiments involving the caspase inhibitor z-VAD-fmk (Enzyme Systems Products, Livermore, CA), after infection CEM and Jurkat T cells were resuspended in culture medium in the presence or absence of 50 μM z-VAD-fmk. Cells were maintained at 5–20 × 10⁵ cells/ml and fresh z-VAD-fmk was added every 2–3 d.

Flow cytometric analyses for PLAP and CD4 were performed as previously described (30). In brief, cells were stained with a 1:100 dilution of rabbit anti–human PLAP (Zymed Laboratories, So. San Francisco, CA) and a 1:100 dilution of anti–human CD4-biotin (Caltag Labs., Burlingame, CA). Goat anti–rabbit FITC that was human and mouse serum adsorbed (BioSource International, Camarillo, CA) at 1:250 and Ultra-avidin PE (Leinco Technologies, Ballwin, MO) at 1:100 were used as secondary stains. The control for the anti-PLAP staining was incubation of cells with secondary antibody only. Triple staining for CD3, CD4, and CD8 was done using anti-CD3-FITC (PharMingen, San Diego, CA) at a 1:25 dilution, anti-CD4-PE (Caltag Labs.) at a 1:100 dilution, and anti-CD8-CyChrome (PharMingen) at a 1:25 dilution. Two-color staining for surface PLAP and Fas was performed using rabbit anti–human PLAP as above and mouse anti–human Fas IgM (Upstate Biotechnology, Inc., Lake Placid, NY) at a 1:200 dilution; secondary staining was done with goat anti–rabbit PE (Southern Biotechnology Associates, Birmingham, AL) at 1:200 and goat anti-mouse IgM FITC (Caltag) at 1:250. FasL staining was performed using the NOK-1 antibody (PharMingen) at 20 μl per reaction followed by biotin-conjugated goat anti–mouse immunoglobulin (PharMingen) at 1:100 and then Ultra-avidin PE at 1:100. Cells that were stained for FasL were grown and stained in the presence of 10 μM of the metalloproteinase inhibitor KB8301 (24), which was provided by Hideo Yagita. After staining, cells were fixed in 1% paraformaldehyde overnight. Flow cytometry was performed on a FACScan^® (Becton Dickinson, San Jose, CA) as previously described.

The number of viable cells in each culture was determined by using trypan blue exclusion. To perform two-color staining for PLAP and TUNEL, cells were first stained for PLAP, and goat anti–rabbit PE was used as a secondary stain as above. The cells were then fixed in 1% paraformaldehyde for 1 h, washed in PBS, resuspended in 70% EtOH, and stored overnight at 4°C. The cells were then washed twice in PBS and a TUNEL assay was carried out using the In Situ Cell Death Detection Kit, Fluorescein (Boehringer Mannheim, Indianapolis, IN) as per the manufacturer's recommendations, except that the cells were permeabilized with 70% EtOH (see above) rather than with Triton X-100. After the TUNEL reaction, cells were washed in 20 volumes of PBS twice. Cells were then analyzed by flow cytometry for both PLAP and TUNEL. Apoptosis was also assessed by staining cells with propidium iodide for DNA content and by determining the percentage of cells that had hypodiploid DNA content. Cells were permeabilized in 70% EtOH for 1 h, washed in PBS, and then resuspended in PBS containing 0.1 mg/ml propidium iodide and 1 mg/ml RNAse. After 90 min in the dark at room temperature the cells were analyzed by flow cytometry.

To study how HIV-1 kills CD4⁺ T cells, we infected two human CD4⁺ T lymphoblastoid cell lines, CEM and SupT1, with either the HIV-1 molecular clone NL43 or NL-PI (Fig. 1,A), and then determined the number of viable cells every 2–4 d. We observed consistent depletion of both cell types in the HIV-infected cultures by 7–10 d after infection (Fig. 2 A). Although the reporter virus NL-PI grew more slowly than did the wild-type virus (data not shown), it also induced significant T cell depletion in these cells, thereby allowing us to study the mechanism of HIV-induced T cell killing.

We also examined the effect of HIV-1 infection of PHA-activated PBMCs from uninfected donors on the percentage and absolute number of CD4⁺ T cells. After infection with NL43 or NL-PI, a portion of each culture was stained daily for surface CD3, CD4, and CD8 expression and analyzed by flow cytometry. This method allowed us to distinguish loss of CD4 cells from the CD4 downmodulation induced by multiple HIV-1 gene products (30). CD4 downmodulation was detected by the appearance of CD3⁺ CD4⁻CD8⁻ T cells in the culture, and these cells were included in the determination of CD4⁺ cells present in the sample. We found that infection with NL43 led to a rapid depletion of CD4⁺ T cells from the culture (Fig. 2 B). Similar results were obtained with NL-PI (data not shown). At early time points (2–3 d after infection), the extent of depletion was proportional to the amount of virus added to the culture (as measured by nanogram per milliliter of HIV p24). When PBMCs were infected with 50 ng/ml p24 of NL43 or 200 ng/ml p24 of NL-PI, the majority of CD4⁺ cells were lost within 2 to 4 d after infection.

We then asked whether depletion of primary CD4⁺ T cells by HIV-1 depends upon the presence of other cells because it has been suggested that macrophages may upregulate FasL after HIV-1 infection and then kill CD4⁺ T cells expressing Fas (32). We purified CD4⁺ T cells to >95% purity using a column that removes CD8⁺ T cells, macrophages and B cells. Infection of PHA-activated purified CD4⁺ T cells with HXB2 led to rapid depletion of all the cells in culture (Fig. 2 C), suggesting that HIV-1 can cause the death of CD4⁺ T cells in the absence of other cell types.

To determine whether HIV-1 infection was killing T cells by apoptosis, we performed a TUNEL assay on infected and uninfected cell cultures. In both T cell lines and primary CD4⁺ T cells, the cultures infected with HIV-1 had a significantly higher percentage of TUNEL-positive cells as compared with an uninfected culture (Fig. 3, A and B). In purified CD4⁺ T cells, there was a sixfold increase in apoptosis in the infected culture 5 d after infection.

We then determined whether infected or uninfected cells within a culture are preferentially undergoing apoptosis by infecting cultures with HIV-PLAP and performing two-color flow cytometric analysis for detection of PLAP and TUNEL. After infection of both CEM and primary CD4⁺ T cells with HIV-PLAP, a higher fraction of PLAP-positive cells was TUNEL-positive as compared with PLAP-negative cells (Fig. 4). In CEM cells, we infected 75.6% of cells by day 4 after infection (Fig. 4,B, upper right plus lower right quadrants). The percentage of TUNEL-positive and PLAP-positive cells was 16.1% (Fig. 4,B, upper right quadrant). Therefore, 21.3% (16.1 out of 75.6%) of infected (PLAP-positive) cells were undergoing apoptosis (TUNEL-positive). The percentage of TUNEL-positive and PLAP-negative cells undergoing apoptosis was 1.7% (Fig. 4,B, upper left quadrant), which is comparable to the background apoptosis in CEM cells. In purified CD4⁺ cells, we infected 7.75% of cells by day 5 after infection and 1.94% of cells were TUNEL- and PLAP-positive (Fig. 4 E). Thus, 25% (1.94 out of 7.75%) of infected (PLAP-positive) CD4⁺ T cells were undergoing apoptosis (TUNEL-positive). The fraction of PLAP-negative cells that were TUNEL-positive in the infected culture was nearly the same as that in the uninfected culture (8%), which is the background apoptosis after activation of primary cells. A similar result was obtained when primary cells were examined 3 d after infection (data not shown). Therefore, in both CEM and primary CD4⁺ T cells, virtually all of the HIV-induced apoptosis occurs in the infected population. We find in vitro that HIV-1 infection has little influence on levels of apoptosis occurring in uninfected bystander cells.

The Fas pathway plays a key role in regulating the survival of T lymphocytes, and induction of this pathway has been proposed as a mechanism by which HIV-1 causes T cell apoptosis. In PBMCs from HIV-infected patients, a higher percentage of T cells express Fas, and the T cells are more sensitive to killing by anti-Fas antibody (15, 17). However, it is not known whether these effects are specific to HIV-infected cells. Therefore, we examined whether Fas expression is upregulated in HIV-infected cells by analyzing cultures infected with HXBnPLAP (Fig. 1,B) for expression of Fas and PLAP. PLAP-positive PBMCs did not have an increase in surface Fas expression as compared with PLAP-negative cells (Fig. 5). Infection with NL-PI also did not affect surface expression of Fas on PBMCs when assessed 4–9 d after infection (data not shown). Since NL-PI expresses all of the regulatory genes of HIV-1, the lack of Fas modulation was not due to the absence of a specific HIV-1 gene product. Similar results were obtained when CEM cells were infected with NL-PI (data not shown). This finding suggests that HIV-1 does not specifically upregulate Fas expression on the surface of infected cells.

We also stained uninfected and infected CEM cells and activated PBMCs for FasL. In uninfected cells, levels of surface FasL were extremely low even in the presence of a metalloproteinase inhibitor, which previously has been shown to stabilize FasL on the cell surface (24). We observed no difference in FasL levels in infected cultures as compared with uninfected cultures (data not shown). As a control, surface FasL was easily detected on hFasL/L5178Y cells stably expressing hFasL (data not shown). The low level of FasL detected on T cells makes it difficult to rule out that HIV-1 infection subtly affects its surface expression.

We found that the ability of HIV-1 to induce cell death in a T cell line was not correlated with the Fas sensitivity of the cell line. As shown in Fig. 2,A, HIV-1 was able to deplete both SupT1 and CEM cells although these cells differ significantly in their susceptibility to Fas-induced killing. This was shown by incubating SupT1 and CEM cells with increasing amounts of anti-Fas antibody (CH11) for 24 h and then staining with propidium iodide to determine their DNA content; the percentage of hypodiploid cells was used as a measure of apoptosis. CEM cells underwent apoptosis with small amounts of anti-Fas antibody, whereas SupT1 cells were resistant to anti-Fas antibody over a wide range of antibody concentration (Fig. 6). This difference was not due to a lack of expression of Fas since both CEM and SupT1 cells expressed Fas on their surface (data not shown).

We also determined whether inhibiting caspases, which are critical downstream effectors of the Fas pathway, would block HIV-induced cell death. Z-VAD-fmk is a cell-permeable irreversible peptide inhibitor of multiple caspase family members (33). When present at a concentration of 50 μM, z-VAD-fmk completely blocked apoptosis of CEM T cells induced by 1 μg/ml of anti-Fas antibody (Fig. 7,A). CEM cells were infected with HXBnPLAP in the presence or absence of 50 μM of z-VAD-fmk. HIV-1 infection depleted CEM T cells as rapidly in the presence of z-VAD-fmk as in its absence (Fig. 7 B). HIV-PLAP also depleted Jurkat T cells in the presence of z-VAD-fmk (data not shown). In fact, we observed a more rapid depletion of Jurkat cells by HXBnPLAP when z-VAD-fmk was present. Staining for PLAP demonstrated that Jurkat cells were more rapidly infected by HIV-1 in the presence of z-VAD-fmk (data not shown), as has been previously reported (34). In CEM cells, the rate of infection was not significantly affected by the caspase inhibitor. In these cells, z-VAD-fmk decreased the percentage of TUNEL-positive cells from 37.7 to 15.5% on day 8 after infection (background apoptosis in uninfected cells was <2%). However, by day 10 virtually all cells were killed by HIV-1 in the z-VAD-fmk–treated and untreated cultures. Therefore, despite blocking Fas-induced apoptosis and decreasing HIV-triggered DNA cleavage, the caspase inhibitor did not prevent virus-associated depletion of CEM cells. There was no effect of z-VAD-fmk alone on growth of uninfected CEM or Jurkat cells.

Finally, we examined whether HIV-1 could kill CD4⁺ T cells from patients who have a genetically defective Fas signaling pathway. A number of patients with ALPS are known to harbor heterozygous dominant-negative mutations in the Fas molecule which renders them relatively insensitive to killing by anti-CD3 stimulation or anti-Fas antibody (25, 35, 36). Many of these patients have missense mutations in the Fas death domain. We studied cells from two ALPS patients and their healthy (Fas-normal) mothers for susceptibility to HIV-induced cell death. These two patients had mutations in exon 9 of the Fas gene that rendered their T cells significantly less susceptible to anti-CD3– and anti-Fas–induced killing (Table 1). We infected PBMCs from these two ALPS patients and their healthy mothers with serial dilutions of NL-PI. Using the PLAP marker, we found that the percentage of infection is proportional to the amount of input HIV over a range of 20– 200 ng/ml p24, and that after spin infection with the highest concentration of NL-PI, >75% of CD4⁺ T cells are infected after 1 d. Importantly, both normal and Fas pathway–defective CD4⁺ T cells were almost completely depleted by day 7 after infection with NL-PI (Fig. 8). Similar results were obtained when these cells were infected with NL43 (data not shown). These results indicate that HIV-1 can kill primary cells that have a defect in their Fas pathway.

The process by which HIV-1 induces CD4⁺ T lymphocyte death was studied using a reporter virus system that distinguishes direct from indirect effects of the virus on its host cell. We find that, in vitro, T-tropic HIV-1 depletes and induces apoptosis in CD4⁺ primary cells and cell lines. After infection of PBMCs with high titers of HIV-1, >50% of CD4⁺ T cells are depleted in 2–3 d. This finding is in agreement with in vivo studies of viral dynamics that indicate that the half-life of the infected cell is ∼2 d (3). Earlier studies found a slower in vitro depletion of primary CD4⁺ T cells by HIV-1 (37); in those experiments HIV-1 was amplified by prolonged cultivation with producer cells, possibly resulting in accumulation of less infectious viral variants. We avoid this problem by producing HIV-1 by transient transfection of 293 cells with viral DNA constructs, generating high titer, homogeneous HIV-1 able to infect at high multiplicity.

We have used the PLAP reporter virus system to demonstrate in vitro that HIV-1 directly induces apoptosis in the cells that it infects. The majority of HIV-induced cell death in vitro occurs in cells that have expressed the PLAP marker. This result is important in light of the continuing debate over whether HIV-1 directly kills infected cells or whether it predominantly induces death of uninfected bystander cells (6). Our finding that in the context of viral infection HIV-1 predominantly causes direct cell death argues against in vitro models in which isolated HIV-1 gene products, such as Tat or Env, induce killing of uninfected bystander cells (38–40). In addition, since HIV-1 is able to directly kill purified primary CD4⁺ T cells in the absence of other cell types, the proposed upregulation of FasL on infected macrophages and resulting parricide of uninfected T cells (32) does not appear to be necessary to explain how HIV-1 induces T cell death. Our demonstration in vitro that HIV-1 directly kills infected cells is consistent with recent indications that HIV-infected patients with different degrees of immune dysfunction have similar rates of infected cell clearance, suggesting that direct cytopathicity of the virus determines the lifespan of the infected cell (1, 3, 41).

The finding that the majority of HIV-specific cell death occurs in cells that express viral gene products appears to contradict observations that in HIV-infected patients apoptosis is increased in CD8⁺ T cells (8, 13) and B lymphocytes (42) as well as in CD4⁺ T cells. We suggest that this increased apoptosis may be due to the generalized immune activation which occurs in response to HIV-1 infection and, thus, may not be responsible for selective CD4⁺ T cell depletion. Other viral infections, such as with EBV, have also been associated with increased apoptosis of uninfected lymphocytes (43). In addition, a study that examined lymph nodes of patients with and without HIV infection concluded that the intensity of apoptosis correlated with the generalized state of activation of the lymph node (i.e., the degree of follicular hyperplasia and germinal center formation) and not with the stage of disease and viral burden (12). The relative role of increased apoptosis secondary to global immune activation in the pathogenesis of the selective CD4⁺ T cell depletion in AIDS is not clear.

Because of its critical role in regulating lymphocyte survival, we also examined the importance of the Fas pathway in HIV-induced cell death. This pathway has been proposed to mediate HIV-induced cell death based in part on observations that patients with HIV-1 infection have a higher percentage of Fas-expressing lymphocytes than uninfected people (17). However, our analysis demonstrates that HIV-1 infection does not directly lead to upregulation of Fas on the surface of the infected cell. Using the HIV-PLAP reporter virus system to distinguish infected from uninfected cells, we find that PLAP-positive primary T cells and CEM cells have the same level of surface Fas as PLAP-negative cells. The elevated percentage of PBMCs expressing Fas in patients with HIV-1 infection may again be a nonspecific manifestation of immune activation. Activation through the antigen receptor is known to induce expression of Fas in T lymphocytes (44–46), and this could explain why both CD8⁺ and CD4⁺ T cells from patients with HIV-1 have been found to have elevated levels of surface Fas (17).

Our findings also raise doubts as to whether the Fas pathway is necessary for HIV-1 to kill CD4⁺ T cells. Although both CEM and SupT1 cells are efficiently killed by HIV-1, only the CEM cells are sensitive to Fas-induced killing. SupT1 cells are resistant to apoptosis induced by anti-Fas antibody even though they express surface Fas. HIV-1 infection also leads to rapid depletion of CD4⁺ T cells from the PBMCs of patients who have a dominant-negative mutation in Fas that impairs Fas-mediated killing. Since the cells from these patients have some residual Fas function (Table 1), this finding does not rule out the possibility that this pathway is involved in HIV-induced cell death. However, we also find that HIV-1 kills CEM and Jurkat T cells in vitro despite the presence of a caspase inhibitor (z-VAD-fmk) that completely abrogates Fas-mediated apoptosis. In fact, we find, in agreement with a previous report (34), that the presence of z-VAD-fmk augments the spread of HIV-1 through Jurkat T cell cultures, which in turn leads to their more rapid death. The caspase inhibitor does decrease the percentage of TUNEL-positive CEM cells after HIV-1 infection, but it does not prevent virus-induced depletion of these cells. This finding is analogous to Bax-induced apoptosis, in which z-VAD-fmk inhibits DNA cleavage but does not block cell death (47). Similarly, HIV-1 may induce cell death both by triggering caspases that cause DNA cleavage as well as by activating other pathways that lead to the death of the cell. These unknown effectors allow HIV-1 to kill CD4⁺ T cells even when the Fas-induced caspases are inhibited. In conclusion, these results suggest that HIV-1 directly kills CD4⁺ T lymphocytes in vitro by a Fas-independent mechanism. The details of this mechanism will be an important topic for future study.

We thank George Cohen for critical comments on the manuscript and members of the Baltimore laboratory for advice and assistance. R.T. Gandhi wishes to thank Bonnie A. Southworth and Marshall A. Wolf.