|Home | About | Journals | Submit | Contact Us | Français|
We previously elucidated an important role for gangliosides in RCC-mediated T-lymphocyte apoptosis, though the mechanism by which they mediated lymphocyte death remained unclear. Here we demonstrate that when added in purified form, GD3 is internalized by activated T-cells, initiating a series of proapoptotic events including the induction of reactive oxygen species, an enhancement of p53 and Bax accumulation, an increase in mitochondrial permeability, cytochrome-c release, and the activation of caspase-9. GD3-induced apoptosis of activated T-cells was dose dependent, and inhibitable by pre-treating the lymphocytes with N-acetyl cysteine, cyclosporin A or bongkrekic acid, emphasizing the essential role of reactive oxygen species and mitochondrial permeability to the process. Ganglioside-induced T-cell killing was associated with the caspase-dependent degradation of NFκB-inducible, anti-apoptotic proteins including RelA; this suggests that their loss is initiated only after the cascade is activated, and that their disappearance amplifies but not trigger GD3 susceptibility. Resting T-cells did not internalize appreciable levels of GD3, and did not undergo any of the proapoptotic changes that characterize activated T-lymphocytes exposed to the ganglioside. RelA overexpression endows Jurkat cells with resistance to GD3-mediated apoptosis, verifying the intact transcription factor’s role in mediating protection from the ganglioside.
Evidence accumulated by our laboratory over the last several years has increasingly implicated gangliosides as being soluble tumor-derived factors that contribute significantly to T-cell apoptosis(1–5). Gangliosides are structurally diverse acidic glycosphingolipids that are present in the outer leaflet of plasma membranes(6), and consist of a sialic acid-containing carbohydrate component attached to ceramide. Many tumors exhibit enhanced synthesis of select gangliosides, some of which are shed into the tumor microenvironment(7, 8). Malignant melanomas and neuroblastomas over-express GD3, GD2 and GM2, whereas renal cell carcinomas were reported to display increased levels of GD1a, GM1 and GM2, as compared to cells of the normal kidney(9). Some tumor-derived gangliosides inhibit specific aspects of the immune response, and thereby contribute to tumor formation and progression(10–15). Because neuraminidase abrogated the apoptotic effects that RCC supernatants have on T-cells, we hypothesized that gangliosides were involved in RCC-mediated immune dysfunction(16), a finding later confirmed both by the capacity of the ganglioside synthesis inhibitor 1-phenyl-2-hexadecanoylamino-3-pyrrolidino-1-propanol (PPPP) to abrogate RCC-mediated killing of co-cultured T-lymphocytes(2), and the ability of TNFα treatment of RCC lines to coordinately enhance their ganglioside accumulation and apoptogenicity for co-incubated T-cells(3).
Our present work focuses on the mechanism by which gangliosides induce T-cell apoptosis. Current evidence implies that gangliosides mediate their pro-apoptotic effects by directly activating the intrinsic apoptotic pathway. In separate reports, Garcia-Ruiz et al.(17) and Rippo et al.(18) demonstrated that the ganglioside GD3 could stimulate a burst of ROS and mitochondrial permeability in purified mitochondria, leading to the release of apoptogenic factors such as cytochrome-c and apoptosis inducing factor (AIF). Later studies showed that mitochondria are targeted even when intact cells are exposed to gangliosides, as GD3-treated hepatocytes underwent apoptosis in association with ROS production, MPT, cytochrome-c release and activation of caspase-9(19, 20).
The means by which tumor-derived gangliosides stimulate the apoptosis of T-cells remains undefined, however. Garcia-Ruiz et al. showed that in response to TNFα, endogenous GD3 redistributes from the outer leaflet of hepatocyte membranes to mitochondria via Rab5-and Rab7-positive endosomes, where it induces the same series of pro-apoptotic events observed when isolated mitochondria are treated with the same ganglioside(20). Studies pertaining to ganglioside transport in Niemann-Pick disease indicate that even exogenous gangliosides can be internalized and targeted to the Golgi complex within Rab-expressing vesicles(21), potentially localizing to the mitochondria where they can induce toxic levels of ROS in glutathione-depleted cells, as described previously for the transported, endogenous gangliosides(20). The notion that exogenous gangliosides may also stimulate T-cell apoptosis in a mitochondrial-dependent manner is suggested by the ability of the Bcl-2 transgene to protect CEM lymphoma cells from GD3-induced caspase-9 activation and death(18). The experiments described below suggest that such a scenario may in fact be the mechanism by which GD3 kills T-lymphocytes: we find that GD3 specifically induces the apoptosis of activated but not resting T-cells. Our results indicate that activated T-cells internalize abundant levels of exogenously administered GD3 within 90 minutes, causing ROS production, the mitochondrial permeability transition and cytochrome-c release by 24h and apoptosis by 48h. GD3-mediated apoptosis additionally involves p53 stabilization and induction of Bax, and is amplified by the caspase-dependent degradation of NFκB-inducible anti-apoptotic proteins. Resting T-cells, which fail to internalize appreciable levels of the ganglioside, do not undergo any of these pro-apoptotic changes within the time frame examined, thus explaining the comparative resistance of that lymphocyte population to GD3-mediated destruction.
Anti-Bcl-xL, anti-Bcl-2, anti-CIAP-2, anti-p53, anti-Bax, anti-RelA and horseradish peroxidase-conjugated donkey anti-goat IgG were from Santa Cruz Biotechnology (Santa Cruz, CA). Anti-XIAP was from BD Transduction Laboratories (San Diego, CA), and anti-Actin (AC-15) was from Novus Biologicals (Littleton, CO). Anti-pro-caspase-9 and procaspase-8 antibodies were from Oncogene Research Products (Boston, MA), and anti-cytochrome-c and anti-GD3 were from BD PharMingen (San Diego, CA). HRP-conjugated sheep anti-mouse and donkey anti-rabbit secondary antibodies were from Amersham (Arlington Heights, IL), and Caspase inhibitor III , Caspase-9 Inhibitor II, and Caspase-8 inhibitor I (IETD-CHO) were all obtained from Calbiochem (La Jolla, CA). Anti-CD3 antibody (OKT3, Ortho Biotech, Raritan, NJ) and anti-CD28 antibody (Becton Dickenson Immunocytometry Systems, San Jose, CA) were used for the stimulation of T-lymphocytes. Human recombinant interleukin-2 (IL-2) (CHIRON Corporation, Emeryville, CA) was used at 20U/ml to maintain the viability of activated T-cells.
Wild type Jurkat cells, caspase-8-negative Jurkat cells(22) (gift from Dr. John Blenis (Harvard University), HA-RelA over-expressing Jurkat cells(1), primary RCC lines(23), the long-term renal cell carcinoma lines SK-RC-45 and SK-RC-26b(3) (gift of Dr.Neil Bander, The New York Hospital, Cornell University Medical College), and the normal kidney epithelial (NKE) cell line(24) were derived and maintained as described previously. The caspase-9 D/N Jurkat cell line was generated using a cDNA expression vector encoding the active site mutation cysteine-287 to alanine (25). A well characterized, long-term glioblastoma cell line U87 was obtained courtesy of Dr. Michael Vogelbaum (Cleveland Clinic) and CCF52 was a short-term GBM line sub-cultured from a fresh glioma by our Brain Tumor Registry.
Tumor cells were immunostained to assess GD3 expression levels. RCC tumor cells, NKE control cells and glioblastoma cells were fixed, blocked and stained with 2µg/ml anti-GD3 antibodies as described previously(5). After washing, the cells were incubated with an Alexa 594-labeled secondary antibody (Molecular Probes Eugene, OR), washed and counterstained with DAPI to visualize the nuclei.
Resting and activated T-cells were also assessed for their capacity to internalize GD3. The lymphocytes were adhered to lysine-coated glass slides for 2h at 4°C in complete medium, at which time the medium was replaced with a solution of 25µg/ml GD3 in plain RPMI. Ganglioside was allowed to adhere to the cells for 1h at 4°C, at which point the slides were moved to 37°C for the designated times prior to being fixed, permeabilized and immunostained for GD3 expression as described above.
T-cells were isolated from the peripheral blood of healthy volunteers with informed consent, as described before(3). The T-cell isolation procedure yielded cells that were more than 95% positive for CD3 as defined by immunocytometry. T-cells were stimulated with cross-linked anti-CD3 (OKT3) and anti-CD28 antibodies for 3d(26), and were then transferred to fresh flasks for further expansion in the presence of 20U/ml IL-2 prior to use.
Resting and/or activated T-cells (1 × 106/mL) were cultured in the presence or absence of GD3 for specified times prior to being analyzed for ganglioside-induced apoptosis by staining the lymphocytes with AnnexinV and 7-aminoactinomycin-D (7-AAD; BD Pharmingen, San Diego, CA). Events were acquired on a multivariable FACScan, and analysis was done using quadrant software (CellQuest v3.3, Becton Dickenson, San Jose, CA). Lymphocytes were additionally stained with 4’6’-diamidoino-2-phenylindole (DAPI) and analyzed by fluorescent microscopy to assess the number of T-cells with apoptotic nuclei in the GD3-treated or untreated cultures. In some experiments, cells were pretreated with an antioxidant (N-acetyl cysteine (Sigma-Aldrich, San Diego, CA), 10mM, NAC) or with inhibitors of mitochondrial permeability (Cyclosporin-A, Bongkrekic Acid, Calbiochem, San Diego, CA) prior to their exposure to GD3, to assess the role of mitochondrial involvement in ganglioside-mediated T-cell apoptosis. In other experiments, activated T-cells were pretreated with 150µM of caspase-8 inhibitor I, caspase-9 inhibitor II, or the pan-caspase inhibitor III prior to ganglioside administration, to assess the role of specific caspases in mediating GD3-induced lymphocyte killing.
Resting and activated T-cells were pretreated or not with 100µg/ml GD3 for 48h prior to stimulating the lymphocytes for 0, 15 and 30 min with PMA (10ng/ml) and ionomycin (0.750µg/ml), at which point nuclear extracts were isolated and prepared in binding reaction buffer, as previously described(13). Oligonucleotide corresponding to the IL2Rα (5’-CAACGGCAGGGGAATCTCCCTCTCCTT-3’) was obtained from Qiagen Operon (Alamed, CA) and used as a probe. Radiolabeled double-stranded oligonucleotides were prepared and labeled with [α-32P]dCTP, and were used in binding reactions with 10µg of nuclear protein, as described previously . Samples were resolved on 6% polyacrylamide gels, and were dried and analyzed as before(1).
Cell lysates were prepared and run on SDS PAGE gels, transferred to nitrocellulose membranes and probed with primary antibodies as described previously(1;27). The immunoreactive proteins were visualized using horseradish peroxidase-linked secondary antibodies and enhanced chemiluminiscence (ECL western blotting kit, Amersham, Piscataway, NJ). Cytochrome-c assays were performed as before(4).
Measurement of intracellular ROS was as described previously(28). Briefly, 3 × 106 resting or activated peripheral blood T-cells were co-cultured with GD3 for 18h, prior to isolating the lymphocytes and incubating them for an additional hour with 10µM 5, 6-carboxy-2’,7’-dichlorofluorescene-diacetate (CM-H2DCFDA, Molecular Probes, Carlsbad, CA). Cells were analyzed flow cytometrically for DCFDA fluorescence using an excitation wavelength of 488nm and an emission wavelength of 535nm. For measuring mitochondrial membrane potentials (ΔΨm), T-cells were treated with GD3 for 18h as described above, prior to incubating the lymphocytes for an additional 15 minutes at 37°C in 500µl of 40nmol/l DiOC6. The cells were then analyzed immediately by FACS analysis using an excitation wavelength of 488 nm and an emission wavelength of 530nm (29).
The Student T test (paired two samples for mean or two samples using equal variances) was used to determine p using Microsoft Excel Software (version 2003) SEM was calculated from SD using Microsoft Excel software.
GD3 was used as the prototypic ganglioside in these studies for multiple reasons: it is over-expressed in numerous tumors of neuroectodermal and epithelial origin, including RCC(30) (Figure 1); it is associated with both tumor progression(31) and immunosuppression(32, 33), and unlike some species of gangliosides, it has already been observed to induce the apoptosis of multiple cell types(19, 20, 34–36). Indeed, we now show that resting and activated T-cells demonstrate the same relative susceptibilities to purified GD3 as they do to ganglioside-producing tumor cells(3, 5) (Figures 2A and 2B). When resting and 5d activated T-cells were treated in vitro with 100µg/ml GD3, approximately 52% of the activated lymphocytes became apoptotic by 48h, as assessed by FACS analysis of Annexin V/7AAD stained cells (Figure 2A, 2B). This was in stark contrast to the ganglioside-treated resting T-cells, which remained mostly viable throughout the 72h period monitored. Other criteria were also used to gauge ganglioside-mediated T-cell damage. Viable cells can also be differentiated by DAPI staining, which reveals the nuclear fragmentation and nuclear membrane invagination that uniquely typify apoptotic cells. Although the percentage of activated, GD3-treated T-cells with apoptotic nuclei per visual field significantly increased following 48h of culture, ganglioside-treated resting lymphocytes retained morphologically normal nuclei throughout this time frame (Figure 2C). Minimal killing of resting T-cells was observed even at 200µg GD3/ml, again contrasting with activated cells, which showed progressively escalating levels of susceptibility to GD3 as the ganglioside concentration was increased from 25µg/ml to 100 µg/ml (Figure 2D).
To assess the possibility that T-cell apoptosis induced by exogenous GD3 might require its internalization and intracellular trafficking, resting and activated T-cells were compared for their relative abilities to internalize exogenous GD3 over the course of 2h. Cells were adhered to lysine-coated glass slides prior to being treated or not with 25µg/ml GD3 for 30, 60, 90 or 120 min, at which points the cells were fixed, permeabilized, and stained with anti-GD3 antibodies. When intracellular GD3 staining was compared at each time point by fluorescent microscopy, it was apparent that, unlike the resting cells, which remained mostly GD3-negative through 60 minutes of treatment and showed only hints of positivity at 90 and 120 minutes, the activated T-cells showed definite signs of GD3 internalization by 30min, stained highly positive by 60 and 90min, but had diminishing levels at 2h (Figure 3A). Interestingly, even when washed and incubated in medium following a 30 minute exposure to GD3, approximately 40% of the activated T cell population still became apoptotic by 48h (Figure 3B), suggesting that lymphocytes need not be continuously exposed to gangliosides for the molecules to mediate their toxic effects. To address the argument that activated T-cells have highly ruffled membranes which bind more ganglioside and hence merely give the appearance of increased ganglioside internalization, FACS analysis and confocal microscopy were performed on GD3-treated resting and activated T-cells that were permeabilized or not, in order to discriminate between adherent and internalized ganglioside. There was comparable binding of anti-GD3 to non-permeabilized resting and activated T cells, reflecting the fact that the ganglioside adhered equivalently to both. FACS analysis of permeabilized cells, however, revealed the extent to which resting and activated T-cells differed in their capacities to internalize GD3. GD3-treated permeabilized and non-permeabilized resting T-cells stained equivalently with the anti-GD3 antibody, indicating that resting T cells did not internalize much ganglioside. This contrasted with the permeabilized, GD3-treated activated T cells, which stained GD3 positive with far greater intensity than either non-permeabilized activated T cells, or permeabilized resting T cells (Figure 3C). Similar results were obtained when the same cells were subjected to examination by confocal microscopy (Figure S1).
To determine which GD3-induced pro-apoptotic events are differentially induced in activated and resting T-cells, thereby explaining the differential susceptibility of those two populations to ganglioside-mediated death, lymphocytes were treated with 100µg/ml GD3 prior to being analyzed for various parameters of apoptosis over the course of 72h. DCFDA fluorescence revealed that the GD3-treated, activated T-cell population generated elevated levels of ROS that were readily detectable by FACS within 24h (Figure 4A). This contrasted with our findings for resting T-cells, which, consistent with their decreased sensitivity to GD3-induced apoptosis, and their only minimal capacity to internalize ganglioside, did not produce measurable levels of ROS within the same time frame (Figure 4A). When DiOC6 fluorescence was used as a measure of increased mitochondrial permeability in response to GD3, the activated T-cells were found to have progressed beyond ROS production to this advanced state by 48h post-ganglioside treatment (Figures 4B). Western analysis revealed that the GD3 treatment had also induced the release of cytochrome-c into the cytoplasm of the CD3/CD28-stimulated T-cells, a requisite step for the direct activation of caspase-9 (Figure 4C). Resting T-cells, on the other hand, weren’t stimulated to augmented ROS production by GD3 (Figure 4A), and didn’t undergo MPT or cytochrome-c release in response to the ganglioside(Figures 4B and 4C), consistent with the comparative resistance of the naïve lymphocytes to GD3-mediated killing.
Because GD3-induced reactive oxygen intermediates appear to act as mediators of T-cell apoptosis, we explored the capacity of N-acetylcysteine (NAC) to protect the lymphocytes from ganglioside-induced death. When activated T-cells were treated with GD3 in the presence of 10mM NAC, lymphocyte killing was reduced by approximately 50% (Figure 4D). The effects of bongkrekic acid (BA) and cyclosporine-A (CsA) on GD3-induced lymphocyte apoptosis were next determined, as each specifically inhibits a unique aspect of mitochondrial permeability. Only 12–15% of activated T-cells underwent apoptosis when treated with GD3 in the presence of either CsA or bongkrekic acid, a significant reduction from the approximately 52% of cells that were killed when GD3 was administered in media alone (Figure 4D). The abilities of NAC, CsA and BA to each inhibit GD3-mediated apoptosis of activated T-cells collectively point to the integral role of ROS and mitochondria in mediating the ganglioside-induced effects.
Activated T-cells were next treated with 100µg/ml GD3 for 72h in the presence or absence of either a pan-caspase inhibitor or specific inhibitors of caspases-8 or 9, prior to being assessed for AnnexinV/7AAD positivity. The pan-caspase inhibitor provided the greatest protection, decreasing GD3-mediated apoptosis of the treated cells five fold to about 10% of the total population. The caspase-9 inhibitor was also extremely effective, as it reduced the killing from 52% to approximately 22% (Figure 5A). The fact that the caspase-8 inhibitor also significantly reduced GD3-mediated killing of activated T-cells (Figure 5A) was somewhat surprising since these and previous studies suggest the initial insult occurs to the mitochondrion(17,18, 20). This result was supported, however, by experiments gauging the relative susceptibilities of wildtype, caspase-8-negative- and caspase-9-negative Jurkat cells to the ganglioside (Figure 5B). As compared to the 55% of wildtype Jurkat cells that underwent apoptosis following a three day exposure to GD3, both mutant lines were significantly protected, with killing of the caspase-8- and caspase-9-negative cells dropping to 25% and 17% of the ganglioside-exposed populations, respectively (Figure 5B). The dependence on both caspases-8 and 9 for optimal GD3-mediated killing of activated T-cells and Jurkat cells contrasts with results obtained using TPEN, a prototypic and potent agonist of the mitochondrial pathway, which induced close to 75% of the activated T-cells to undergo apoptosis (Figure 5A). Here, caspase-9 was observed to be of foremost importance to the response, with only a minimal requirement for caspase 8: while both the pan caspase inhibitor and the caspase-9 inhibitor reduced TPEN-induced T-cell apoptosis from 72% of the cells to 20% and 30%, respectively, the caspase-8 inhibitor had only a negligible protective effect on TPEN-induced killing(Figure 5A).
We previously showed that RCC-mediated killing of co-cultured activated T-lymphocytes was associated with the depletion of anti-apoptotic T-cell proteins by a mechanism that could be abrogated by pre-treating the tumor cells with the ganglioside synthesis inhibitor, PPPP(1, 23). Here, we asked whether GD3 treatment alone would decrease anti-apoptotic protein expression, and if so, whether the effect was limited to the GD3-susceptible, activated T-cell population. We found that in addition to activating caspases-8 and 9 in the activated T-cells, as evidenced by the depletion of their zymogens (Figure 5, Panel C), GD3 also mediated a dramatic and progressive decline in Bcl-2 and Bcl-xL expression levels (Figure 5, Panel C), findings deemed significant given the important contributions those proteins make to maintaining the integrity of mitochondrial membranes(37). Ciap-2 and XIAP are also NFκB-dependent anti-apoptotic proteins which function by blocking the activity of effector caspases,(38)and GD3 decreased these too from the elevated levels that characterize activated T-cells (Figure 5, Panel C). Importantly, none of these protein modifications occurred in GD3-treated resting T-cells, consistent with their resistance to ganglioside-induced apoptosis(Figure 5, Panel C)
Because we observed that GD3-mediated apoptosis of activated T-cells (Figure 2) was associated with GD3-induced ROS (Figure 4A), we next assessed the involvement of p53 in ganglioside-stimulated lymphocyte killing. Cytoplasmic lysates made from activated T-cells treated or not with 100µg/ml GD3 for 48h were subjected to western analysis, using antibodies to p53 and Bax. We found that the ganglioside did in fact up-regulate both p53 and Bax (Figure 5, Panel D), results consistent with the capacity of the ganglioside to sequentially induce downstream ROS accumulation, increased mitochondrial permeability cytochrome-c release, and initiation of the caspase cascade. When GD3-treatment of activated T-cells took place in the presence of a pan-caspase inhibitor, the ganglioside-mediated loss of Bcl-xL, Ciap-2 and XIAP expression levels was abrogated, suggesting that the activation of caspases either directly or indirectly diminished the expression of those proteins (Figure 5, Panel D). The GD3-mediated inductions of p53 and BAX, on the other hand, were not prevented by the pan-caspase inhibitor, results expected of molecular events upstream of ganglioside-induced caspase activation (Figure 5, Panel D).
Previous studies from our laboratory demonstrated that SK-RC-45 induced RelA degradation in co-cultured, activated T-cells by mechanism that was both tumor ganglioside and caspase dependent(1). To assess the relative susceptibilities of resting and activated T-cells to GD3-mediated proteolysis of RelA, western analysis was performed on whole cell lysates made from each population following a 48h exposure to the ganglioside. RelA levels in resting T-cells were not altered by the ganglioside, but dropped precipitously in GD3-treated activated T-cells by a mechanism that was caspase-dependent (Figure 6, Panels A and B), results paralleling those observed for the anti-apoptotic proteins. Consistent findings were obtained when RelA activity in the GD3-treated cells was monitored by EMSA. As compared to unstimulated cells, treatment of resting and activated T-cells with PMA/ionomycin resulted in the translocation of RelA to the nucleus, as evidenced by its increased binding to an oligonucleotide encoding the kB element of the IL-2Rα gene (Figure 6, Panel C). A 48h pre-treatment of the activated T-cells with GD3 completely inhibited the PMA/ionomycin-induced nuclear translocation of RelA, however, though the ganglioside had no such effect on the resting cells (Figure 6, Panel C).
The possibility that ganglioside-induced NFκB degradation contributes to GD3-mediated T-cell apoptosis led us to ask whether over-expressing RelA would confer protection to GD3-treated Jurkat cells. A Jurkat cell clone permanently transfected with the PcDNA3/HA-RelA construct(1) was therefore compared to wildtype Jurkat cells for their susceptibility to GD3. Neither wildtype nor RelA-transfected Jurkat cells demonstrated significant susceptibility to the ganglioside after 24h of treatment, though by 48h the RelA over-expressing cells had a distinct survival advantage: 33% of wildtype Jurkat cells had succumbed to GD3, but only 12% of the RelA-transfected cells stained positive for Annexin V/7AAD (Figure 6, Panel D). The ability of the RelA transgene to inhibit GD3-mediated killing of Jurkat cells by 66% points to the importance of continuous RelA-induced anti-apoptotic gene transcription in protecting T cells from ganglioside-induced apoptosis.
Here we show that GD3 can mediate the apoptosis of activated T-cells by a mechanism involving enhanced production of reactive oxygen species, p53 and Bax accumulation, the induction of mitochondrial permeability, the stimulation of cytochrome-c release and the activation of caspase-9, events all initiated and occurring sequentially following GD3 internalization by the lymphocytes. Resting T-cells did not readily internalize GD3, and were not observed to undergo any of the aforementioned proapoptotic changes in response to the ganglioside. The GD3-induced apoptosis of activated T-cells was first detectable 48h post-ganglioside treatment and was dose-dependent, becoming evident at 25µg/ml, appreciable at 50µg/ml and plateauing at 100µg/ml. Interestingly, unfractionated, tumor-derived gangliosides collectively induce T cell apoptosis at even 7 µg/ml, suggesting that the glycosphingolipids likely synergize to mediate appreciable lymphocyte death at much lower, more physiologically relevant concentrations in vivo(23). Bcl-xL, Ciap-2 and Xiap were degraded in the GD3-treated activated T-cells by a mechanism that was caspase-dependent, suggesting that while a deficiency of anti-apoptotic proteins might have amplified GD3-induced T-cell apoptosis, their loss did not initiate the process. Previous studies performed in other cell types and on the purified organelles indicated that GD3 exerts its apoptogenic effects by acting directly on mitochondria. Our observation that activated T-cells pretreated with NAC, CsA or BA were able to resist GD3-induced apoptosis attested to a similar role for GD3-stimulated ROS production and mitochondrial permeability in mediating the death of intact lymphocytes as well.
Based on our results, it would appear that T-cell internalization of GD3 is requisite for the ganglioside to mediate its apoptotic effects: when assessed by both confocal microscopy and FACS analysis, it was only the activated T-cells that internalized appreciable amounts of the ganglioside, which correlated with the unique susceptibility of that population to GD3-mediated killing. Though we are presently examining the itinerary of internalized gangliosides in T-cells, there is precedence in other cell types for endogenous GD3 to be transported from the plasma membrane to mitochondria in response to apoptotic stimuli such as TNFα(20), a molecule synthesized by activated but not resting T cells(5). Consistent with their common structural features, it is hypothesized that GD3, like ceramide, initiates the accumulation of toxic ROS by disrupting electron flow at complex III of the respiratory chain(17, 28). Resulting oxidative stresses trigger conformational changes in inner mitochondrial membrane proteins, leading to the MPT, cytochrome-c release, activation of caspase-9 and the apoptosis observed in these ganglioside-treated cells(17). A role for ROS in ganglioside-mediated apoptosis of activated T-cells was demonstrated by the ability of NAC to inhibit the killing by more than 50%. The capacity of both CsA and BA to reduce GD3-mediated apoptosis of activated T cells 3–5 fold from the levels observed with GD3 alone further underscored the important contribution of GD3-induced mitochondrial permeability to T-cell killing.
Associated with the GD3-mediated apoptosis of activated T-cells was the augmented expression of p53 and Bax, the mitochondrial release of cytochrome-c, and the activation of caspase-9. P53 is a transcription factor activated in response to cellular stresses, and mediates its effects by inducing the de novo expression of growth-inhibitory or pro-apoptotic molecules that prevent the proliferation of damaged or infected cells(39). The protein is ordinarily short-lived, being maintained at low levels by an inhibitor that restricts its transcription and promotes its degradation. When the cell perceives DNA damage or toxic ROS, however, p53 is stabilized, leading to the transcriptional activation of Bax, a proapoptotic Bcl-2 family member that forms the mitochondrial pores through which cytochrome-c enters the cytoplasm to initiate the intrinsic pathway(39). Consistent with our findings, Chandra et al. previously detected elevated p53 and Bax in murine bone marrow cells en route to an apoptotic cell death following a 24h treatment with purified T-cell lymphoma gangliosides(40). Though in our studies the pan-caspase inhibitor prevented GD3-mediated apoptosis, it could not abrogate the ganglioside-induced elevations of p53 or Bax expression, indicating that GD3-stimulated ROS and the sequential elevations in p53 and Bax are caspase-independent steps in the pathway to lymphocyte death.
Enhancing the GD3-mediated apoptosis of activated T-cells was the ganglioside-induced depletion of several anti-apoptotic proteins, including Ciap-2, Xiap, Bcl-xL and Bcl-2. Ciap-2 and XIAP both abrogate apoptosis by directly binding and inhibiting caspases, while Bcl-xL and Bcl-2 control survival by inhibiting Bax and BAK, proapoptotic proteins that produce the mitochondrial pores that release cytochrome-c, Smac/Diablo and AIF(41). The ability of the pan-caspase inhibitor to prevent Bcl-xL-, Ciap-2- and XIAP loss during the GD3 treatment of activated T-cells suggests that those proteins are stable until the caspase cascade is activated; so while the depletion of anti-apoptotic proteins during the course of apoptosis might amplify and/or hasten cell death, it is not the loss of those molecules that initially renders the lymphocytes susceptible to the ganglioside. The expression levels of the anti-apoptotic proteins were not altered in GD3-treated resting cells, however, consistent with the resistance of naïve T-lymphocytes to GD3-induced caspase activation.
There are multiple mechanisms by which GD3 might negatively regulate Bcl-xL, Ciap-2 and XIAP expression levels in a caspase-dependent manner. Ganglioside-activated caspases might directly degrade them, or a protease downstream of activated caspases might be mediating the activity. It is also possible that only RelA is cleaved(1), thus inhibiting the transcription of these NFκB-dependent, anti-apoptotic molecules; indeed, caspase inhibitors coincidentally rescued both RelA and the anti-apoptotic proteins in GD3-treated T cells. Arguing for direct caspase-dependent proteolysis of Bcl-2 and Ciap-2 in GD3-treated activated T cells, however, was the long term stability of those proteins in the presence of cyclohexamide, suggesting that merely inhibiting their synthesis would not impact their expression levels within a 24h time frame (data not shown). The degradation of RelA is ultimately critical for GD3-induced apoptosis, however, since RelA over-expression by Jurkat cells is protective.
Our interest in the mechanism by which GD3 induces the apoptosis of T-cells stems in part from the role of tumor-derived gangliosides in mediating immune dysfunction, thereby facilitating the progressive growth of the tumors that produce them. We previously showed that the SK-RC-45 RCC line that over-expresses GD3 was much more effective at inducing apoptosis in activated T-lymphocytes as compared to resting T cells(3, 5), and also demonstrated that one of the most apoptogenic HPLC fractions of two glioma-line-derived ganglioside preparations contain GD1a and GD3(23). This SK-RC-45-mediated killing was inhibitable by PPPP and hence dependent on its ganglioside production, and the fact that T-cell death was accompanied by mitochondrial permeability(4) cytochrome-c release(2) and caspase-9 activation(2) collectively suggested that tumor-induced lymphocyte death occurred at least in part through the mitochondrial pathway. Precisely why the SK-RC-45 tumor line selectively killed activated but not resting T-cells remained enigmatic, however, prompting this comparative study of the GD3-induced effects on both cell types. Though our results with GD3 can’t be generalized to all gangliosides, given the unique molecular characteristics that distinguish them, the effects of GD3 on activated and resting T-cells mirror those induced by SK-RC-45: both selectively kill only activated T-cells through the intrinsic pathway by a mechanism that has ROS accumulation at the apex, and proteolysis of anti-apoptotic proteins as a means of amplifying the response. Because the pro-apoptotic events leading to the GD3-induced death of activated T-cells are initiated by ganglioside internalization, it appears likely that the resistance of resting T-lymphocytes to the ganglioside’s apoptotic effects is based on the only inefficient internalization of the molecule by those cells.
Grant Support :
NIH grants RO1-CA111917(CT), RO1-CA90995(JHF), RO1-CA56937-11(JHF) We also gratefully acknowledge the Department of Science and Technology and Indian Council of Medical Research, Government of India.