Abstract
Granzyme B (GzmB) plays a major role in CTLs and NK cell–mediated elimination of virus-infected cells and tumors. Human GzmB preferentially induces target cell apoptosis by cleaving the proapoptotic Bcl-2 family member Bid, which, together with Bax, induces mitochondrial outer membrane permeabilization. We previously showed that GzmB also induces a rapid accumulation of the tumor-suppressor protein p53 within target cells, which seems to be involved in GzmB-induced apoptosis. In this article, we show that GzmB-activated p53 accumulates on target cell mitochondria and interacts with Bcl-2. This interaction prevents Bcl-2 inhibitory effect on both Bax and GzmB-truncated Bid, and promotes GzmB-induced mitochondrial outer membrane permeabilization. Consequently, blocking p53–Bcl-2 interaction decreases GzmB-induced Bax activation, cytochrome c release from mitochondria, and subsequent effector caspases activation leading to a decreased sensitivity of target cells to both GzmB and CTL/NK-mediated cell death. Together, our results define p53 as a new important player in the GzmB apoptotic signaling pathway and in CTL/NK-induced apoptosis.
Introduction
Cytotoxic T lymphocytes and NK cells eliminate virus-infected or malignantly transformed cells principally by releasing the contents of cytotoxic granules into the immune synapse formed with their target cell (1). The granule mediators of target cell lysis are serine proteases, known as granzymes (Gzms), which induce programmed cell death (2–4) after they are delivered into the target cell cytoplasm by the pore-forming protein perforin (PFN) (5–7).
Granzyme B (GzmB), one of the main mediators of cell death by the cytotoxic granules pathway, is the most extensively studied Gzm. Human GzmB (hGzmB) preferentially induces target cell apoptosis in a mitochondria-dependent manner, which is a highly regulated process involving the Bcl-2 family proteins. HGzmB cleaves the proapoptotic Bcl-2 family member, BH3-only protein, Bid (8, 9). Truncated Bid (tBid) disrupts the mitochondrial outer membrane (MOM) to cause the release of the proapoptotic factors cytochrome c and Smac leading to caspase 3 activation and target cell apoptosis (10). Moreover, hGzmB induces cell death in a Bax or Bak-dependent manner (8, 11), while this pathway is blocked by overexpression of important prosurvival Bcl-2 family proteins, including Bcl-2 or Bcl-XL (12). It has also been suggested that GzmB may induce target cell death in absence of Bid, Bax, and Bak (13), by directly activating caspase 3 (8, 14), cleaving the inhibitor of caspase-activated DNase (CAD) to free CAD (15, 16), and/or by cleaving several caspases (17) or noncaspase substrates (18, 19).
Our previous studies also suggested the existence of a relationship between the tumor suppressor p53 and hGzmB-induced apoptosis. Indeed, we demonstrated that hGzmB induces a rapid activation of the endonuclease CAD in a Bid- and caspase 3–independent manner that triggers early DNA damages. An ATM-related genotoxic stress-activated kinase, hSMG-1, is then rapidly activated in response to hGzmB-induced early DNA damage (20) and would be, at least in part, responsible for a rapid wild type p53 (wtp53) stabilization and activation in a melanoma target model after interaction with autologous CTL clones or after hGzmB treatment (21). Furthermore, RNA interference-mediated inhibition of p53 significantly decreased target cell killing induced by CTLs or hGzmB (21). Taken together, these results indicate that wtp53 is involved, by an unknown mechanism, in the modulation of the PFN/hGzmB-dependent apoptotic pathways during the killing of target cells by immune effector cells.
It is well established that wtp53 acts as a stress-activated transcription factor that activates transcription of genes involved in apoptosis induction or regulation (22, 23), although some effects of p53 are independent of transcription (24). In this regard, compelling evidences show that the transcription-independent function of p53 relies on its interaction with several members of the Bcl-2 family (25, 26). Indeed, it has been reported that wtp53 protein can promote the permeabilization of the MOM by forming complexes with the protective Bcl-XL or Bcl-2 proteins, hence inhibiting their antiapoptotic effects (27). It was also demonstrated that p53 can accumulate in the cytoplasm, where it directly activates the cytosolic proapoptotic protein Bax (28) or mitochondrial Bak (29) to induce the mitochondrial release of apoptogenic factors.
Thus, the aim of this study was to investigate the mechanisms of p53-dependent regulation of the hGzmB signaling pathway leading to apoptotic cell death. We demonstrate that after PFN/hGzmB treatment, p53 quickly accumulates on target cell mitochondria and interacts with Bcl-2. This p53–Bcl-2 interaction prevents the inhibitory interaction of hGzmB-activated tBid with Bcl-2 and also favors the release of Bax from Bax–Bcl-2 complexes. Consequently, p53–Bcl-2 interaction is of major importance in hGzmB apoptotic signaling pathway as shown by the pharmacological inhibition of this interaction leading to hGzmB cell death signaling blockage at the mitochondria level and to a decrease of both hGzmB- and CTL/NK-induced apoptosis.
Materials and Methods
Cell lines
MCF7/Caspase3 (30), T1, and A549 cells were grown in RPMI 1640/GlutaMAX supplemented with 10% FCS, 100 U/ml penicillin, 100 mg/ml streptomycin, and 1% sodium pyruvate (Life Technologies). The T1 tumor cell line was established from the primary lesion of a patient suffering from a melanoma (31). The LT12 CTL clone was isolated from T1 autologous tumor-infiltrating lymphocytes, recognized a peptide derived from the MelanA/MART-1 Ag, and was maintained in culture as described previously (20, 31). The NK92 cell line and NK cells isolated from healthy donors with human CD56+ selection kit (Stemcell) were cultured in RPMI 1640/GlutaMAX supplemented with 10% of FCS and 300 U/ml recombinant human IL-2. The H33 CTL clone was maintained in culture as described previously and recognizes the allogeneic MCF7 cell line in an HLA-A2–restricted manner (32).
Pifithrin treatment
Target cells were pretreated 15 h with 5–10 μM pifithrin (PFT)-μ, 20 μM PFT-α (Sigma-Aldrich), or DMSO before incubation with cytotoxic cells, buffer, sublytic native rat PFN, and/or hGzmB in presence of PFT or DMSO.
Treatment with PFN and GzmB
Native rat PFN was purified from RNK16 cells, and native hGzmB was purified from YT-Indy cells as described previously (33). Target cells were diluted in HBSS with 10 mM Hepes pH 7.5, 4 mM CaCl2, 0.4% BSA before adding sublytic PFN ± 100 nM hGzmB, diluted in PFN buffer (HBSS, 10 mM Hepes pH 7.5). The PFN sublytic concentration that delivers Gzms to induce apoptosis without excessive necrosis was determined independently for each experiment by flow cytometry (BD Accuri C6) as described previously (33).
Inhibition of GzmB activity
The GzmB inhibitor Z-AAD-CH2Cl (a synthetic peptide that irreversibly inhibits GzmB activity; Biovision) (34) was used at 100 μM to preincubate effector cells for 2 h before incubation with target cells (concentration of inhibitor during coincubation: 50 μM).
Western blot
Total cellular extracts were prepared by lysing cells in RIPA buffer (Pierce) containing a mixture of protease inhibitors (Roche) and 2 mM sodium orthovanadate before denaturation by boiling in Laemmli buffer and SDS-PAGE separation. Blots were probed with the following Abs: anti-p53 mouse mAb (clone DO-1), anti–Bcl-2 mouse mAb (clone 100), and anti-Bax mouse mAb (clone N20) from Santa Cruz Biotechnology; anti–phopho-p53(Ser37) rabbit polyclonal Ab (pAb), anti–caspase 3 mAb (clone 3G2), anti-cleaved caspase 3 (p19/p17) rabbit pAb (clone 5A1E), anti–caspase 9 rabbit pAb, anti-PARP rabbit pAb, anti–Bcl-XL mouse mAb (clone 54H6), and anti-Bid rabbit pAb from Cell Signaling; anti-tBid (p15) rabbit pAb from Invitrogen; and HRP-conjugated anti-actin mouse mAb from Sigma-Aldrich. Western blot quantification was performed using the ImageJ densitometry software.
Formation of E:T cell conjugates
Target cells were incubated with NK92 or H33 effector cells in Ca2+ free medium during 20 min at 37°C to allow conjugate formation. Synchronized exocytosis of the cytotoxic granules contents was triggered by adding 4 mM CaCl2 before analysis by microscopy, flow cytometry, or cell fractionation at the indicated time.
Microscopy
Target cells were grown on collagen-coated glass coverslips (Sigma-Aldrich), incubated with Cy5-CytoPainter mitochondrial staining solution (Abcam), and treated with sublytic PFN ± 100 nM hGzmB, NK92, or H33 effector cells (E:T ratio 3:1). After 30 min, cells were fixed and stained with an anti-p53 mouse mAb (clone DO-1) followed by incubation with a goat Alexa Fluor 488–conjugated secondary Ab (Life Technologies) as described previously (5). Coverslips were then washed three times in PBS and mounted in Vectashield mounting medium containing DAPI (Vector Laboratories) before imaging (IX83 microscope; Olympus) and analysis (CellSense Dimension software; Olympus).
Cytosol-mitochondria-nucleus fractionation
After pretreatment with DMSO or 10 μM PFT-μ and 30 min treatment with PFN ± 100 nM hGzmB or NK92 cells (E:T ratio 3:1), 5 × 106 target cells per condition were fractionated into cytosolic and high-purity mitochondrial fraction using a Qproteome mitochondria isolation kit (Qiagen) or into cytosolic, mitochondrial, and nuclear fractions using a standard cell fractionation kit (Abcam) following manufacturer’s protocol. Cytosolic, mitochondrial, and/or nuclear extracts from the same number of cells were denatured in Laemmli buffer before immunoblotting with anti-p53 (clone DO-1) or anti-cytochrome c (clone 7H8.2C12; BD Pharmingen) mouse mAbs. Anti-vinculin (clone Vin-11-5; Sigma-Aldrich), anti-mitochondrial Hsp70 (MtHsp70/Grp75; clone JG1; Abcam), and anti-PCNA (clone PC10; Cell Signaling) or anti-topoisomerase II α (clone Ki-S1; Millipore) mouse mAbs were used to verify the purity of cytosolic and mitochondrial fractions and the absence of nuclear contamination.
Bcl-2 immunoprecipitation
After pretreatment with PFT-μ or DMSO and PFN ± hGzmB loading, Bcl-2 immunoprecipitation was performed using a classical immunoprecipitation kit (Pierce) and an anti–Bcl-2 mouse mAb (clone 100) overnight at 4°C. Immune complexes were eluted and washed before boiling in Laemmli buffer. Immunoprecipitates and 1% of the cell lysates (input) were run in 4–20% precast SDS-PAGE gel and immunoblotted with HRP-conjugated anti–Bcl-2 mouse mAb (clone 100), anti-p53 rabbit pAb (clone FL-393; Santa Cruz Biotechnology), HRP-conjugated anti-Bax rabbit pAb (clone N20), or anti-tBid (p15) rabbit pAb (Invitrogen).
Bid and p53 small interfering RNA transfection
Silencer select validated Bid, p53, and control small interfering RNAs (siRNAs) were purchased from Life Technologies. Subconfluent cells were transfected with siRNA using Lipofectamine RNAiMAX (Life Technologies) in Opti-MEM according to the manufacturer’s instructions.
Apoptosis assays
To assess apoptosis, we analyzed cells incubated for 2 h at 37°C with buffer or sublytic PFN ± 100 nM hGzmB for caspase activation by flow cytometry using M30-FITC mAb staining according to the manufacturer’s protocol (M30 CytoDEATH; Roche) to detect an effector caspase-cleavage product of cytokeratin 18 (35).
Mitochondrial outer membrane permeabilization assay
MOM permeabilization (MOMP) was evaluated by 3,3′-dihexyloxacarbocyanine iodide DioC6 (3) labeling (Life Technologies). After sublytic PFN ± 100 nM hGzmB treatment or incubation with NK92 (E:T ratio 5:1), cells were incubated 10–15 min at 37°C with 40 nM DioC6 (3) in prewarmed HBSS. The stained cells were then separated from the staining solution by centrifugation, resuspended in PBS, and immediately analyzed by flow cytometry (BD Accuri C6). For experiments with NK cell treatment, MCF7-Casp3 cells were prestained with 1 nM Vybrant DiD (Life Technologies) to exclude NK cells and to allow analysis of the target cells.
Chromium release assay
Cytotoxicity was measured by a 4 h chromium release assay as previously described (21). Inhibition of PFN/Gzm-mediated cytotoxic pathway was performed by using effector cell preincubation for 2 h with 100 nM concanamycin A (CMA; Sigma-Aldrich) (concentration of CMA during lysis: 50 nM) or with the GzmB inhibitor Z-AAD-CH2Cl before incubation with target cells. Experiments were performed in triplicate. Data were expressed as the percentage of specific 51Cr release from target cells, calculated as (experimental release − spontaneous release)/(maximum release − spontaneous release) × 100.
Measurement of Bax activation and detection of p53 intensity by flow cytometry
After 30 min treatment with PFN ± 100 nM hGzmB, cells were washed with ice-cold PBS and fixed for 5 min in PBS/0.25% PFA at room temperature (RT). Cells were then washed with PBS and incubated 30 min at RT with an anti-Bax mAb (clone 6A7; Santa Cruz Biotechnology) diluted in PBS/200 μg/ml digitonin (Sigma-Aldrich). Cells were then washed with PBS, incubated 30 min at RT with a goat anti-mouse FITC-conjugated secondary Ab (Beckman Coulter) in PBS, and washed twice with PBS before flow cytometry analysis (BD Accuri C6). For the detection of p53 intensity, cells were fixed and permeabilized with Cytofix/Cytoperm (BD Biosciences) 30 min after PFN/GzmB treatment, then incubated with an anti-p53 mouse mAb (clone DO-1) and an Alexa 488–conjugated secondary Ab before analysis of the mean fluorescence intensity by flow cytometry (BD Accuri C6).
Statistical analysis
Data are expressed as mean ± SD. The p values were determined by unpaired two-tailed Student t tests.
Results
hGzmB induces wtp53 accumulation on target cell mitochondria
To study the role of p53 in the regulation of hGzmB apoptotic signaling pathway and in hGzmB-mediated cell death, we used the wtp53 breast carcinoma cell line MCF7. Because MCF7 are deficient for procaspase 3, which is important for hGzmB-induced cell death, we used cells stably expressing physiological level of procaspase 3 (MCF7-Casp3; Supplemental Fig. 1A). p53 can be activated by gamma irradiation in these cells, and hGzmB activates the mitochondrial pathway of apoptosis in this cell line, with Bid cleavage, Bax activation, procaspase 3, and PARP cleavages leading to apoptosis (Supplemental Fig. 1B–F). As previously reported in melanoma cells (20, 21), the treatment of MCF7-Casp3 cells with 100 nM hGzmB in the presence of sublytic concentration of PFN results in a significant wtp53 accumulation within ∼30 min (Fig. 1A, 1B) and a wtp53 phosphorylation (Ser37; one of the key events in DNA damage-induced stabilization and activation) within ∼20 min after PFN/GzmB treatment (Fig. 1C). Moreover, fluorescence microscopy analysis of p53 subcellular localization after 30 min target cell loading with PFN/hGzmB or incubation with the NK cell line NK92 or the allogeneic CTL clone H33 showed a mainly mitochondrial p53 accumulation within target cells, which can be abrogated by preincubating effector cells with the irreversible GzmB inhibitor Z-AAD-CH2Cl (Fig. 1D, 1E). Similarly, after a classical cytosol/mitochondria/nucleus or a high-purity fractionation of mitochondria, p53 accumulates in the MCF7-Casp3 target cell mitochondrial fraction after 30 min treatment with PFN/hGzmB (Fig. 2A–C), but this accumulation can also be detected earlier (Fig. 2D). Of note, similar results were obtained using the Casp3− parental MCF7 cell line (data not shown). Moreover, we also detected p53 in the mitochondrial fraction after target cell incubation with the NK92 cell line in the absence of the irreversible GzmB inhibitor (Fig. 2E). Together, these observations demonstrate that hGzmB induces a rapid p53 accumulation in the mitochondrial compartment of treated cells. We also concluded that p53 might control hGzmB-induced cell death through its nontranscriptional activity at the mitochondria level. We thus hypothesized that p53 regulates hGzmB-induced cell death by interacting with Bcl-2 to facilitate tBid-dependent Bax activation, MOMP, apoptogenic factors release, and effector caspases activation.
GzmB induces p53 accumulation in target cell. (A and B) Within 30 min of treatment with sublytic PFN and hGzmB, p53 accumulates in MCF7-Casp3 target cells. A representative Western blot analysis of three independent experiments is shown in (A). Actin was used as loading control. p53 mean fluorescence intensity (MFI) fold increase analyzed by flow cytometry analysis from three independent experiments is displayed in (B). The p values were determined by unpaired two-tailed Student t test. (C) Within 20 min of treatment with sublytic PFN and hGzmB, p53 is phosphorylated (Ser37). An equal amount of total p53 protein was loaded in each well and served as loading control. (D and E) p53, mitochondria, and nucleus staining of MCF7-Casp3 target cells treated for 30 min with PFN ± hGzmB (D) or with the NK92 cell line or the H33 CTL clone preincubated or not with an irreversible GzmB inhibitor (E:T ratio 3:1) (E). Scale bars, 10 μM (D), 5 μM (E). Dashed lines represent plasma membrane.
GzmB–induced p53 accumulates on target cell mitochondria. (A–C) After classical cytosol/mitochondria/nucleus (A) or high-purity cytosol/mitochondria (B) fractionation experiments of target cells after 30 min treatment with PFN/hGzmB, p53 is detectable in the mitochondrial fraction of treated cells. Representative immunoblots are shown. Immunoblotting with anti-vinculin, anti-mitochondrial Hsp70, and anti-PCNA or Topoisomerase II mAb was used to verify the purity of cytosolic (C), mitochondrial (M) fraction, and nuclear (N), and the absence of nuclear contamination in mitochondrial extract, respectively. Blot exposure was titrated down to the point where p53 is almost undetectable in control conditions (buffer, PFN alone, and GzmB alone) to show increased mitochondrial accumulation in PFN/GzmB-treated cells. The p53/mtHsp70 ratio from (B) was calculated by densitometry and display in (C) (mean ± SD from three independent experiments). The p values were determined by unpaired two-tailed Student t test. (D) Mitochondrial and cytosolic fractions were isolated as in (B) after 25 to 30 min treatment with hGzmB or PFN/hGzmB treatment. (E) MCF7-Casp3 target cell incubation for 30 min with the NK92 cell line (E:T ratio 3:1) also induces p53 mitochondrial translocation, which can be strongly decreased after NK cell preincubation with an irreversible GzmB inhibitor.
hGzmB-activated p53 interacts with Bcl-2
To demonstrate the possible p53–Bcl-2 interaction in target cells after PFN/GzmB treatment, we used the cell-permeable sulfonamide PFT-μ that blocks p53 interaction with Bcl-XL and Bcl-2 proteins, and selectively inhibits p53 translocation to mitochondria without affecting the transactivation function of p53 (36). We first verified that PFT-μ did not affect p53 stabilization in response to PFN/hGzmB treatment (Supplemental Fig. 2A, 2B). We then performed a cytosol/mitochondria fractionation of DMSO or 10 μM PFT-μ–pretreated MCF7-Casp3 cells, loaded with PFN ± hGzmB. DMSO-pretreated cells showed an accumulation of p53 in the mitochondrial fraction after 30 min treatment with PFN/hGzmB, which can be blocked by PFT-μ (Fig. 3A, 3B). Similarly, PFT-μ pretreatment decreased p53 mitochondrial accumulation in target cells after incubation with the NK92 cell line, whereas PFT-α, which mainly blocks p53 transcriptional activity (37), has no effect (Fig. 3C). Furthermore, Bcl-2–p53 complexes were detected by immunoprecipitating Bcl-2 30 min after PFN/hGzmB loading (Fig. 3D). Moreover, Bcl-2–p53 complexes were detected in control cells pretreated with DMSO after PFN/hGzmB loading, but not in PFT-μ pretreated cells (Fig. 3E). Together, these results indicate that hGzmB-activated p53 interacts with Bcl-2, which constitutively resides at MOM.
p53 interacts with Bcl-2 in response to GzmB. (A and B) PFT-μ, a cell-permeable sulfonamide that blocks p53 interaction with Bcl-2, inhibits p53 accumulation on mitochondria of PFN/hGzmB-loaded cells. A representative immunoblot is shown in (A). The p53/mitochondrial Hsp70 ratio was calculated by densitometry and display in (B) (mean ± SD from three independent experiments). The p values were determined by unpaired two-tailed Student t test. (C) PFT-μ also blocks p53 mitochondrial accumulation in target cells after MCF7-Casp3 interaction with NK92 cells (E:T ratio 3:1). NK92 preincubated with an irreversible GzmB inhibitor or MCF7-Casp3 target cells pretreated with PFT-α, which blocks p53 transcriptional activity, were used as controls. Blot exposure (A and C) was titrated down to the point where p53 is almost undetectable in control conditions (buffer or without NK treatment) to show the increased or decreased mitochondrial accumulation in PFN/hGzmB or NK-treated cells. (D) p53 interacts with Bcl-2 after 30 min PFN/hGzmB treatment. Bcl-2 was immunoprecipitated before immunoblotting for Bcl-2, p53, and actin as a negative control. Whole-cell lysate (input) used for the immunoprecipitation was also immunoblotted. The experiment was performed twice. (E) PFTμ pretreatment inhibits p53–Bcl-2 interaction after PFN/hGzmB loading. Bcl-2 was immunoprecipitated before immunoblotting for Bcl-2, p53, and actin as in (C). The experiment was performed twice.
p53–Bcl-2 interaction regulates GzmB-induced Bax activation and MOMP
We next investigated whether p53–Bcl-2 interaction plays a role in tBid and Bax-dependent GzmB-induced MOMP. As previously described, MCF7-Casp3 target cell treatment with PFN/GzmB generated a significant amount of tBid (p15), which is mainly localized in the mitochondrial fraction. We also detected a significant amount of Bax in both cytosolic and mitochondrial compartments in PFN or hGzmB alone-treated cells, but Bax became enriched at the mitochondria after PFN/GzmB treatment (Fig. 4A). Moreover, Bid inhibition by siRNAs strongly inhibits PFN/GzmB-induced caspase 3 activation (Fig. 4B, 4C), confirming that Bid is crucial for hGzmB-induced cell death in our model. These results demonstrate that GzmB-induced mitochondrial p53 cannot lead to cell death by itself but more likely regulates tBid-dependent Bax activation, a key step in hGzmB-induced MOMP and apoptosis (10). It is known that Bax changes conformation before oligomerization in the MOM. The conformational change of Bax exposes its N terminus and can be specifically detected by an anti-Bax Ab (clone 6A7) that binds to this region (38). Flow cytometry experiments using the 6A7 Bax mAb show an ∼50% decrease of Bax activation after PFN/GzmB loading after PFT-μ pretreatment (Fig. 4D, 4E). Moreover, this decrease of Bax activation correlates with a decrease of PFN/hGzmB-induced MOMP after PFT-μ preincubation, as measured by DioC6 (3) staining (Fig. 4F, 4G). Similar results were also obtained after NK92 effector cell incubation with PFT-μ–pretreated or p53 siRNA-transfected target cells (Supplemental Fig. 2C, 2D). Finally, PFN/hGzmB-induced cytochrome c release from the mitochondria to the cytosol is also significantly decreased by a PFT-μ pretreatment (Fig. 4H, 4I). These results suggest that hGzmB-induced p53–Bcl-2 interaction promotes tBid-dependent Bax activation, MOMP, and cytochrome c release.
Inhibition of p53–Bcl-2 interaction decreases GzmB-induced Bax activation and MOMP. (A) Bax and tBid accumulate in mitochondria of PFN/hGzmB-loaded cells. A representative immunoblot of two independent experiments is shown. (B and C) Bid is essential for GzmB-induced caspase 3 activation. MCF7-Casp3 target cells were transfected with control (Ctrl) or Bid siRNAs (B) before loading with PFN ± hGzmB for 45 min (C). Bid siRNAs inhibit caspase 3 and PARP cleavage. (D and E) PFT-μ pretreatment decreases Bax activation after 30 min PFN/hGzmB loading, assessed by flow cytometry using anti-Bax 6A7 mAb. Representative flow cytometry histograms are shown in (D) and mean ± SD of three independent experiments in (E). (F and G) PFT-μ pretreatment decreases MOMP after 30 min PFN/hGzmB loading, assessed by flow cytometry using DioC6 (3) staining. Representative flow cytometry histograms are shown in (F) and mean ± SD of three independent experiments in (G). (H and I) PFT-μ pretreatment decreases cytochrome c release from the mitochondrial (m) to the cytosolic fraction (c) of PFN/hGzmB-loaded cells. A representative immunoblot is shown in (H). The ratio of cytosolic/mitochondrial cytochrome c was calculated by densitometry and display in (I) (mean ± SD from three independent experiments). The p values were determined by unpaired two-tailed Student t test.
GzmB-induced p53–Bcl-2 complexes regulate Bax and tBid interaction with Bcl-2
We next investigated the mechanisms involved in p53-dependent regulation of Bax activation in response to hGzmB after its interaction with Bcl-2. It has been shown that the mitochondrial pathway is blocked by prosurvival Bcl-2, which binds and blocks the active form of Bid or Bax by different mechanisms, depending of the nature of the apoptotic stimulus (39–41). The “activator” model (or mode 1) postulates that some BH3-only proteins (e.g., tBid) transiently bind to Bax and induce conformational changes required for stable membrane insertion. In this model, Bcl-2 proteins sequester the activator BH3-only proteins and neutralize them (42). In contrast, the “inactivator” model (or mode 2) proposes that the primary function of antiapoptotic protein Bcl-2 is to neutralize the proapoptotic effector proteins Bax by forming heterodimeric complexes (28, 43), limiting their capacity to form pores in the MOM. Based on these models, the earliest step in the process of hGzmB-induced cell death leading to MOMP that could be modulated by p53–Bcl-2 interaction is binding of tBid or Bax to Bcl-2. To answer this question, we immunoprecipitated Bcl-2 after PFN/hGzmB loading in DMSO- or PFT-μ–pretreated cells before immunoblotting with Bcl-2, p53, Bax, and tBid mAbs (Fig. 5). Bcl-2–p53 complexes were only detected in DMSO-pretreated cells after PFN/hGzmB loading and inhibited by PFT-μ. Bax did coprecipitate with Bcl-2 in control DMSO-pretreated cells and was released from Bcl-2 after PFN/hGzmB treatment. Importantly, the inhibition of p53–Bcl-2 interaction by PFT-μ inhibits the release of Bax from Bcl-2 after PFN/hGzmB treatment. With respect to tBid, we did not detect Bcl-2–tBid interaction in DMSO-pretreated cells after PFN/hGzmB loading. However, Bcl-2–tBid complexes were detected after PFN/GzmB treatment when Bcl-2–p53 interaction was inhibited by PFT-μ. Together, these data suggest that p53–Bcl-2 interaction in response to PFN/hGzmB treatment favors the release of Bax from its inhibitor Bcl-2 and also prevents tBid interaction with Bcl-2.
p53–Bcl-2 interaction favors the release of Bax from Bcl-2 and prevents tBid interaction with Bcl-2 in response to GzmB. p53–Bcl-2 interaction in response to hGzmB promotes the release of Bax from Bcl-2 and inhibits tBid interaction with Bcl-2. DMSO or 10 μM PFT-μ–pretreated MCF7-Casp3 cells were loaded with hGzmB alone (negative control) or PFN/GzmB for 30 min before Bcl-2 immunoprecipitation and immunoblotting with anti–Bcl-2, anti-p53, and anti-Bax and anti-tBid (p15) mAb. Representative immunoblots from two independent experiments are shown. p53/Bcl2, Bax/Bcl2, and tBid/Bcl2 ratios were calculated by densitometry. A.U., arbitrary unit.
p53–Bcl-2 regulation of GzmB-induced MOMP promotes subsequent effector caspase activation
To further demonstrate that p53–Bcl-2 interaction regulates hGzmB apoptotic signaling pathway at the mitochondrial level, we then performed immunoblotting analysis of key proteins involved in this pathway upstream and downstream of mitochondria. Bid (p22) cleavage into tBid (p15) and cleavage of procaspases 9 and 3 were thus detected 15–30 min after PFN/hGzmB loading of DMSO-pretreated target cells, as expected. In contrast, PFT-μ pretreatment did not interfere with hGzmB-mediated Bid cleavage but inhibits the cleavage of procaspases 9 and 3 (Fig. 6A). Nevertheless, a first cleavage of procaspase 3 (p19) is still observed in PFT-μ–pretreated cells after 20 min treatment with PFN/hGzmB, which might reflect a direct but incomplete cleavage of procaspase 3 by hGzmB. In this regard, it has been proposed that GzmB can initiate effector caspases activation but cannot fully process procaspase 3 without disruption of the mitochondrial membrane and the release of second mitochondrial activator of caspases/Diablo and HtrA2/Omi that facilitate caspase 3 full activation by blocking the inhibitory action of inhibitor of apoptosis proteins (10). Similarly, the reduction of p53 expression by siRNAs strongly decreased PFN/hGzmB-induced procaspases 9 and 3 activation but did not interfere with Bid cleavage (Fig. 6B, 6C). Finally, neither DMSO nor PFT-μ pretreatment interfered with Bcl-2, Bcl-XL or Bax expression (Fig. 6D). Overall, these data demonstrate that inhibiting p53–Bcl-2 interaction after PFN/GzmB treatment blocks MOMP, leading to the inhibition of effector caspases activation.
Inhibition of p53–Bcl-2 interaction blocks GzmB signaling pathway downstream of mitochondria. (A) Analysis of hGzmB signaling pathway in MCF7-Casp3 cells pretreated with DMSO or 10 μM PFT-μ and loaded with PFN ± hGzmB for 20–45 min. Bid (p22), tbid (p15), procaspase 9 (p47), procaspase-3 (p35), and cleaved caspases-3 (p19/p17) were analyzed by immunoblot. PFT-μ does not interfere with GzmB-mediated Bid cleavage but inhibits the cleavage of procaspases 9 and 3. (B and C) p53 knockdown decreases GzmB-induced caspase 9 and 3 activation but does not affect Bid cleavage. MCF7-Casp3 target cells were transfected with control (Ctrl) or p53 siRNAs (B) before loading with PFN ± hGzmB for 45 min (C). The p53/actin ratio from (B) was calculated by densitometry and normalized to “1” in untreated (mock) cells. (D) PFT-μ pretreatment does not interfere with Bcl-2, Bcl-XL, or Bax expression. Actin was a loading control. Data (A–D) are representative of three independent experiments.
Blocking p53–Bcl-2 interaction inhibits GzmB-induced apoptosis
Our results suggest that blocking p53–Bcl-2 interaction after PFN/hGzmB treatment inhibits hGzmB-induced MOMP and hGzmB apoptotic signaling pathway downstream of mitochondria, leading to a resistance to GzmB-induced cell death. To confirm this finding, we compared PFN/hGzmB-induced apoptosis by measuring effector caspases activation in either DMSO or 5–10 μM PFT-μ–pretreated cells. For effector caspase activation, we used a flow cytometry–based assay using the M30-FITC mAb to detect a caspase 3 cleavage product of cytokeratin 18 (35). Pretreatment of MCF7-Casp3 cells with 7.5 or 10 μM PFT-μ before PFN/hGzmB loading results in a significant dose-dependent inhibition of hGzmB-mediated apoptosis (Fig. 7A, 7B). Similar results were also obtained in MCF7-Casp3 cells transfected with p53 siRNA (Fig. 7C) or using the wtp53 melanoma and lung carcinoma cell lines T1 and A549 pretreated with PFT-μ (Fig. 7D). Moreover, the pretreatment of target cells with PFT-α, which mainly blocks p53 transcriptional activity (37), only slightly decreases PFN/hGzmB-induced apoptosis compared with PFT-μ (Fig. 7E), suggesting that p53 function on hGzmB-induced cell death is mainly mediated by its nontranscriptional activity. Together, these data demonstrate that hGzmB-induced wtp53 accumulation to mitochondria and interaction with Bcl-2 is an important modulator of hGzmB-induced apoptosis.
Inhibition of p53–Bcl-2 interaction alters GzmB–induced apoptosis. (A and B) PFT-μ decreases PFN/GzmB-induced apoptosis in a dose-dependent manner. MCF7-Casp3 cells were pretreated with 5–10 μM PFTμ before PFN ± hGzmB loading. Apoptosis was measured by flow cytometry using M30 mAb staining (which recognizes a cytokeratin 18 epitope, revealed after effector caspase cleavage). Representative flow cytometry histograms (A) and mean ± SD of percentage M30+ cells from four independent experiments (B) are shown. (C) MCF7-Casp3 transfection with control (ctrl) or p53 siRNAs (leading to ∼65% p53 inhibition compared with p53 basal level, measured by Western blot) also decreases PFN/GzmB-induced apoptosis. (D) PFT-μ (10 μM) also decreases PFN/hGzmB-induced apoptosis of wtp53 T1 melanoma and A549 lung carcinoma cell lines. (E) PFT-μ is more efficient than PFT-α to decrease PFN/hGzmB-induced apoptosis. MCF7-Casp3 cells were pretreated with 10 μM PFT-μ and/or 20 μM PFT-α before PFN/hGzmB loading. Apoptosis in (C)–(E) was measured as in (A) and shown as means ± SD of percentage M30+ cells from three independent experiments. The p values were determined by unpaired two-tailed Student t test.
p53–Bcl-2 interaction is an important modulator of CTL/NK-mediated cell death through the PFN/GzmB pathway
To confirm the physiological relevance of the GzmB-induced p53–Bcl-2 interaction, we then tested whether blocking hGzmB-induced p53–Bcl-2 interaction with PFT-μ can interfere with CTL and NK cell–mediated killing of wtp53 target cells. The lysis of MCF7-Casp3 target cells by the allogeneic H33 CTL clone and by the NK92 cell line was thus strongly decreased by PFT-μ pretreatment (Fig. 8A, 8B) in a dose-dependent manner (Supplemental Fig. 3A). Similarly, the lysis of wtp53 T1 target cells by the autologous LT12 CTL clone and by the NK92 cell line was strongly decreased after PFT-μ pretreatment (Fig. 8C, 8D). Similar results were also obtained after p53 inhibition using siRNA or with NK cells obtained from two healthy donors (Supplemental Fig. 3B–F). Moreover, the inhibition of exocytosis-mediated pathway by CMA resulted in the abrogation of H33, LT12, NK92, and NK cells isolated from healthy donors’ cytotoxicity, indicating that their observed killing is mostly mediated by the PFN/Gzms pathway (Fig. 8A–D and Supplemental Fig. 3B–E) and particularly by GzmB, as confirmed using the GzmB inhibitor Z-AAD-CH2Cl (34) (Supplemental Fig. 4). Together, these data demonstrate that hGzmB-induced p53 accumulation to mitochondria and interaction with Bcl-2 is an important regulator of the CTL/NK-mediated killing of wtp53 target cells (Fig. 9).
Inhibition of p53–Bcl-2 interaction decreases CTL/NK-mediated cell death. PFT-μ–pretreated cells are less susceptible to the CTL/NK-mediated cell death. MCF7-Casp3 lysis by the H33 CTL clone (A) or by the NK92 cell line (B) and T1 lysis by the autologous CTL clone LT12 (C) or by the NK92 cell line (D) after preincubation with 10 μM PFT-μ or DMSO and cocultured at different E:T ratios are shown (mean ± SD of at least three independent experiments). CMA, which inhibits calcium-dependent exocytosis of cytotoxic granules, was used as an inhibitor of PFN/Gzm-dependent lysis (positive control). The p values were determined by unpaired two-tailed Student t test.
Proposed model for the role of p53 in GzmB–induced apoptotic signaling pathway. hGzmB-induced p53 accumulates on target cell mitochondria where it interacts with the prosurvival protein Bcl-2. This interaction allows the release of the proapoptotic protein Bax from its inhibitory interaction with Bcl-2 and prevents hGzmB-activated tBid sequestration by Bcl-2. In this case, free tBid activates free Bax to induce MOMP, cytochrome c release, and apoptosis. Blocking p53–Bcl-2 interaction with PFT-μ inhibits Bax release from Bcl-2, favors tBid–Bcl-2 binding, and consequently inhibits MOMP.
Discussion
Based on our previous results showing that p53 is involved in PFN/GzmB-dependent cell death during the killing of target cells by immune effector cells, this study was intended to provide insight into the functional relationship between hGzmB and p53 during CTL and NK-mediated killing. In this study, we demonstrated that the death signal induced by hGzmB and mitochondrial pathway are interconnected through p53 and tBid during the CTL/NK-induced killing of target cells. We showed that hGzmB-activated p53 translocates to target cell mitochondria. Because PFT-μ, which targets only the mitochondrial branch of the p53 pathway without affecting the important transcriptional functions of p53, decreases both hGzmB- and CTL/NK-mediated cell death, we concluded that p53 nontranscriptional activity is an important determinant in the hGzmB apoptotic signaling pathway. Of note, PFT-α also has, to some extent, an effect on hGzmB-induced cell death, even if PFT-μ inhibition is much more potent. PFT-α was shown to block p53 transcriptional activity by disrupting the nuclear transport of p53 leading to inhibition of p53-dependent transactivation of its target genes (37). Thus, we cannot exclude that p53-controlled transactivation might be, at least in part, an important feature for hGzmB-mediated apoptosis. However, recent evidence demonstrated that PFT-α can also block the apoptosome-mediated processing and activation of caspases 9 and 3, independently of p53 (44). Moreover, cDNA microarrays analysis of p53 direct target genes expression (including Bax, Bcl-2, Bid, Puma, Noxa) after PFN/hGzmB treatment did not emphasize significant upregulation or downregulation (data not shown and Ref. 20), suggesting that p53 transcriptional activity is not essential for the control of hGzmB-induced cell death.
As shown in this article and by others (8, 9), tBid is crucial for hGzmB-induced Bax activation, MOMP, effector caspases cleavage, and apoptosis induction, but Bcl-2 and Bcl-XL were shown to prevent Bax activation and MOMP in response to different stress via the sequestration of tBid (mode 1) (40, 41). Similarly, we found that p53 participates to hGzmB-induced apoptosis by interacting with the antiapoptotic protein Bcl-2. This interaction prevents hGzmB-generated tBid to interact with Bcl-2 as shown by the Bcl-2–tBid complexes observed in PFN/hGzmB-treated cells after PFT-μ preincubation. Consequently, we suggested that p53 indirectly allows tBid-mediated Bax activation, probably by preventing Bcl-2–tBid inhibitory interaction. Accordingly, we observed that blocking p53–Bcl-2 interaction with PFT-μ strongly decreases Bax activation and MOMP. Of note, we confirmed that PFN/hGzmB-activated tBid mainly localized to mitochondria, and that PFN/hGzmB treatment increased mitochondrial Bax level. These results suggest that hGzmB-activated tBid interacts and activates Bax at the mitochondrial membrane. In this scenario, by blocking Bcl-2–tBid interaction, p53 might also favor tBid-mediated Bax translocation to mitochondria, where it can be activated. In this regard, a critical role of tBid in the regulation of MOMP has been demonstrated (45). In this model, Bax and tBid seem to only interact when a membrane is present: tBid rapidly binds to mitochondrial membranes and its BH3 domain facilitates the insertion of cytosolic Bax into the membrane, leading to its activation/oligomerization, and subsequently to membrane permeabilization (45).
We also observed a perhaps surprising significant amount of Bax in mitochondrial fraction of untreated cells interacting with Bcl-2. However, recent studies indicated that Bax localizes to both the cytosol and to mitochondria in healthy cells, and that transient Bax–Bcl-XL or Bax–Bcl-2 interaction stimulates retrotranslocation of Bax from the mitochondria into the cytoplasm, thereby sustaining Bax in its cytosolic inactive form (46, 47). In stressed cells, transient interaction with direct activators induces extensive conformational changes of Bax, resulting in the formation of more stable, membrane-embedded Bax. As such, antiapoptotic Bcl-2 proteins were shown to prevent Bax activation and MOMP not only via the sequestration of tBid (mode 1), but also by binding Bax (mode 2), thereby preventing Bax activation (40, 41). In particular, Bcl-XL was shown to tie up both tBid and Bax in nonproductive interactions inhibiting Bax binding to membranes (48). Interestingly, we observed that GzmB-activated p53 interaction with Bcl-2 not only prevents tBid interaction with Bcl-2, but also releases Bax from Bax–Bcl-2 complexes. Moreover, in the absence of p53 binding to Bcl-2 after PFT-μ pretreatment, a ternary complex including Bcl-2, Bax, and tBid is observed in response to PFN/hGzmB treatment, but Bax is not activated. These results suggest that tBid binding to Bcl-2 is not enough to free and to activate Bax, and that p53 is important to break this ternary complex. Consistently, recent studies have identified the region of interaction between Bcl-2–Bcl-XL proteins and p53 to its DNA-binding domain (49–51). Moreover, it has been demonstrated that Bcl-XL conformational change upon binding to wtp53 might facilitate the dissociation of Bcl-XL–Bax complexes (49, 52). Thus, we can envision that p53–Bcl-2 interaction in response to PFN/hGzmB treatment also blocks “mode 2” inhibition and favors Bax activation by hGzmB-activated tBid. In particular, p53 might release mitochondrial nonactivated Bax from Bcl-2 interaction, promoting its activation by tBid, but might also facilitate the dissociation of Bcl-2–activated Bax complex, promoting both Bax activation and oligomerization. In this scenario, hGzmB-stabilized p53 would operate as a BH3-only protein of the derepressor type freeing proapoptotic tBid and Bax from Bcl-2–mediated inhibition (Fig. 9). Similarly, ABT-737, a small molecule that acts as a selective inhibitor of Bcl-2, was shown to increase the levels of both tBid and Bax at the MOM that are free to interact, thereby promoting Bax activation (53), and to block hGzmB-activated tBid inhibition by Bcl-2 in Bcl-2–overexpressing cells (54).
In conclusion, this study emphasizes that the coordinated action of hGzmB-activated p53 and GzmB-cleaved Bid is important for GzmB-induced cell death and for CTL/NK-mediated killing of target cells. Nevertheless, it has been demonstrated that wtp53, but not tumor-derived p53 mutants, binds to Bcl-2 via this DNA-binding domain and can trigger MOMP (52). Because our results point out the importance of Bcl-2/wtp53 interaction in hGzmB-induced apoptosis, further studies will be required to determine whether such p53 mutants can be associated with tumor cell resistance to PFN/hGzmB- and CTL/NK-mediated cell death.
Disclosures
The authors have no financial conflicts of interest.
Acknowledgments
We thank J.L. Perfettini for helpful discussion and A.C. Jonckheere for technical assistance.
Footnotes
T.B.S. and L.Z. performed experiments and analyzed data. L.F. and L.L. also performed and helped analyze some experiments. G.G. provided technical assistance with CTL culture. F.M.-C. provided the H33 CTL clone. D.M. purified human GzmB and participated in helpful discussion. J.T. conceived and supervised the project, designed and performed experiments, analyzed data, and wrote the manuscript. S.C. edited the manuscript.
This work was supported by INSERM, by the Association pour la Recherche sur le Cancer (2012-2013; Grant SFI20121205624), and by fellowships from the French Ligue Nationale Contre Le Cancer (to L.Z. and J.T.).
The online version of this article contains supplemental material.
Abbreviations used in this article:
- CAD
- caspase-activated DNase
- CMA
- concanamycin A
- Gzm
- granzyme
- GzmB
- Granzyme B
- hGzmB
- human GzmB
- MOM
- mitochondrial outer membrane
- MOMP
- MOM permeabilization
- pAb
- polyclonal Ab
- PFN
- perforin
- PFT
- pifithrin
- RT
- room temperature
- siRNA
- small interfering RNA
- tBid
- truncated Bid
- wtp53
- wild type p53.
- Received August 4, 2014.
- Accepted October 21, 2014.
- Copyright © 2014 by The American Association of Immunologists, Inc.
References
In this issue
