p53 Accumulation at the ER/MAM Compartments After Anticancer Treatments Enhances Cell Death.

To establish whether another nonmitochondrial p53 aspect could be involved in a nontranscriptional proapoptotic pathway, we verified the intracellular localization of p53 using biochemical and immunofluorescence techniques. Using a previously described subcellular fractionation protocol (19), we purified ER and MAM fractions from primary mouse embryonic fibroblasts (MEFs) and the human colon cancer HCT-116 p53^+/+ cell line. Similar to PML, p53 was localized to the ER and MAMs as well as the cytosolic fraction (Fig. 1A and Fig. S1A) under untreated conditions.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

p53 localizes at the ER and MAMs. (A–C) Detection of p53 by immunoblotting in HCT-116 p53^+/+ fractions. (A) p53 localization in untreated condition (UNT). Accumulation of p53 at the ER and MAMs in HCT-116 p53^+/+ cells after adriamycin (ADRIA) induction (1 μM, 6 h) (B) or after H₂O₂ treatment (500 μM, 6 h) (C). (D–F) Colocalization of p53 (red) and Sec61-GFP (used as ER marker; green) in p53^+/+ MEFs under untreated conditions (UNT) (D) and after ADRIA (E) or H₂O₂ (F). (Insets) A higher magnification of the images is presented. (G) p53 activation increased its ER colocalization. Colocalization of p53 and ER in p53^+/+ MEFs quantified as the proportion of total ER marker overlapping the p53 signal (by Mander’s coefficient colocalization method). To allow for a better appreciation of colocalization of p53 with the ER, a cytoplasmic portion was selected and the contrast was increased. Bars, SEM; *P < 0.05.

Therefore, we investigated whether adriamycin (ADRIA), a chemotherapeutic agent, or H₂O₂, an oxidative stress mediator, would change the subcellular localization of p53. An enrichment at the ER/MAMs was detected after p53 induction by either treatment (Fig. 1 B and C and Fig. S1 B and C). We confirmed p53 accumulation at the ER/MAMs by immunofluorescence using digital imaging 3D deconvolution (Fig. 1 D–F). The colocalization of p53 and the ER was analyzed in p53^+/+ MEFs as an overlap between the p53 and ER-marker (Sec61b-GFP) signals. The suitability of a colocalization analysis was verified by sampling the cytoplasmic portion of each cell and performing the randomized Costes et al. method (20). In all of the analyzed samples, the colocalization probability was higher than 99%. As expected, the overlap of p53 signal and ER-marker signal was increased in response to stress (Fig. 1G).

MEFs have previously been shown to be resistant to apoptosis induced by thapsigargin (TG) in the absence of p53 (21). In addition, p53 localizes to the ER/MAM compartments. Therefore, we investigated whether p53 could be a fundamental component of the ER stress-induced apoptotic pathway. Using different ER-stress inducers, we showed a marked reduction in the number of apoptotic cells in p53^−/− MEFs compared with wild-type (WT) cells using flow cytometry analysis (Fig. 2 A and B), cytochrome c release (Fig. 2 C and D), and automated cell analysis based on morphological parameters and propidium iodide staining (Fig. 2E). Apoptotic cell death, evoked by H₂O₂, was blocked in cells pretreated with a caspase inhibitor (Fig. 2F) but enhanced after ADRIA-induced p53 accumulation at the ER/MAMs (Fig. 2 G–I).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Activation and accumulation of p53 at the ER/MAMs render cells more prone to death. (A and B) Percentage of apoptosis induced by (A) H₂O₂ (500 μM, 12 h) in p53^+/+ or p53^−/− MEFs or (B) ceramide (C2; 60 μM, 12 h), thapsigargin (TG; 2 μM, 12 h), tunicamycin (TUN; 6 μM, 12 h), brefeldin A (BFA; 5 mg/mL, 12 h), or menadione (MEN; 15 μM, 12 h) in p53^−/− MEFs. The data show the percentage of cell death in the whole cell population negative for annexin-V-FITC and propidium iodide (PI) staining, as analyzed by flow cytometry. (C) Detection of cytosolic cytochrome c release and supernatant HMGB1 release (a necrotic marker) by immunoblotting in p53^+/+ or p53^−/− MEFs treated with H₂O₂ (500 μM, 12 h) compared with the untreated condition. Actin was used as a loading control for the cytosolic fraction. (D) Cytosolic cytochrome c release in p53^+/+ and p53^−/− MEFs treated with C2 (60 μM, 12 h), TUN (6 μM, 12 h), or TG (2 μM, 12 h). (E) Percentage of apoptosis versus necrosis analyzed by automated imaging and cell scoring based on morphological parameters and PI staining in p53^+/+ and p53^−/− MEFs treated with H₂O₂ (500 μM, 12 h), MEN (15 μM, 12 h), TUN (6 μM, 12 h), TG (2 μM, 12 h), or C2 (60 μM, 12 h). (F) Z-VAD-FMK treatment inhibits cell death in p53^+/+ MEFs (H₂O₂, 500 μM, 6 h). (G) Quantification of cell survival induced by H₂O₂ (500 μM, 12 h) through automated nucleus count analysis. Bars, SEM. (H) Representative microscopic fields of p53^+/+ and p53^−/− MEFs under untreated conditions, pretreated with ADRIA (1 μM, 6 h) and then H₂O₂ (1 mM, 12 h). (I) Detection of apoptosis by immunoblotting in p53^+/+ and p53^−/− MEFs and p53^+/+ pretreated with ADRIA (1 μM, 6 h) under untreated conditions and with H₂O₂ (500 μM, 6 h).

p53 Induction at the ER/MAMs Regulates Ca²⁺ Homeostasis, Allowing for Mitochondrial Fragmentation and Apoptosis.

The key process connecting apoptosis to the ER–mitochondrial interaction is an alteration in Ca²⁺ homeostatic mechanisms (13, 22, 23) that results in massive and/or prolonged mitochondrial Ca²⁺ overload (13, 24, 25). We thus examined the effect of p53 down-regulation and induction on Ca²⁺ homeostasis. Using recombinant aequorin probes (26), [Ca²⁺] was measured selectively in the cytosol and in organelles acting as a source (ER) or target (mitochondria) of cellular Ca²⁺ signals. A striking difference was evident in [Ca²⁺]_ER (Ca²⁺ concentration within ER lumen) steady-state levels (Fig. 3A). After p53 induction by ADRIA the [Ca²⁺]_ER was higher, whereas the loss of p53 caused a reduction in the [Ca²⁺]_ER compared with WT. Representative traces are shown in the figures whereas the full dataset is included in Table S1.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Deregulation of Ca²⁺ homeostasis after p53 induction is a stress signal for mitochondrial structure and a trigger for apoptosis. (A–C) Measurements of [Ca²⁺] using recombinant aequorin upon agonist stimulation (100 μM ATP) in the ER (A), cytosol (B), and mitochondria (C). (D) ER Ca²⁺ release induced by H₂O₂ measured using a FRET-based Ca²⁺-sensitive D1ER-YC4.3 probe; the normalized FRET ratio of D1ER-YC4.3 was assumed as the intraluminal [Ca²⁺]. (Insets) A magnified portion of the first 2 min of the recording as basal. (E) Cytosolic Ca²⁺ response induced by H₂O₂ (2 mM) in MEFs loaded with the Ca²⁺-sensitive fluorescent dye Fura-2. The kinetic behavior of the [Ca²⁺]_c (Ca²⁺ concentration within cytoplasm) response is presented as the ratio of fluorescence at 340 nm/380 nm. (F) Analysis of [Ca²⁺]_m (Ca²⁺ concentration within mitochondrial matrix) during oxidative stress upon H₂O₂ stimulation (2 mM). Isosurface rendering of representative p53^+/+ (G and H) and p53^−/− (I and J) MEFs expressing mitochondrial GFP in basal conditions (UNT), after adriamycin (1 μM, 6 h), and/or H₂O₂ exposure (500 μM, 3 h). (H and J) High-resolution imaging of mitochondrial fragmentation during p53 activation and oxidative stress induction in p53^+/+ and p53^−/− MEFs.

In agreement with the [Ca²⁺]_ER data, the [Ca²⁺] increases evoked by agonist stimulation (ATP) in the cytosol and in the mitochondria were significantly higher after ADRIA treatment and lower in p53^−/− MEFs than in WT MEFs (Fig. 3 B and C). Similarly, increased Ca²⁺ traffic from the ER to the mitochondria was observed in both HeLa cells overexpressing a WT p53 construct (4) (Fig. S2 A–C) and p53^+/+ HCT-116 cells upon ADRIA treatment (Fig. S2D), as well as in p53^−/− MEFs after the reintroduction of WT p53 (Fig. S2E). In contrast, ADRIA treatment in p53^−/− MEFs had no effect on Ca²⁺ homeostasis (Fig. 3 A–C, blue traces). The effect of p53 at the ER was also confirmed by analyzing mitochondrial Ca²⁺ uptake in permeabilized cells exposed to the same [Ca²⁺]. Under these conditions, no differences were observed with regard to the p53 levels (Fig. S2F), indicating that the p53-dependent Ca²⁺ responses previously described were due to alterations of the source of the Ca²⁺ signals, the ER.

As mentioned above, there is a strong agreement in the literature linking Ca²⁺ transfer from the ER to the mitochondria and the effects of apoptotic stimuli (13, 22, 23, 25). Therefore, we investigated whether the absence or the induction of p53 could alter [Ca²⁺]_m after apoptotic stimuli. We observed that the agonist-dependent mitochondrial Ca²⁺ response, after the oxidative apoptotic inducer H₂O₂, was reduced proportional to p53 expression (Fig. S3A). Using the ER-targeted, FRET-based Ca²⁺-sensitive D1ER-YC4.3 probe, we measured the effect of p53 on the progressive release of Ca²⁺ from the ER caused by H₂O₂. The normalized FRET ratio (proportional to [Ca²⁺]_ER) was observed to correlate with p53 levels (Fig. 3D and Fig. S3B). Administration of H₂O₂ caused a progressive depletion of Ca²⁺ from the ER (as revealed by a reduction in the normalized FRET ratio) with consequent increases in [Ca²⁺] in the cytosol (Fig. 3E and Fig. S3C) and mitochondria (Fig. 3F and Fig. S3D). Furthermore, this event appeared to be proportional to p53 levels.

To assess whether the observed ER Ca²⁺ overload is a proapoptotic condition, we analyzed the mitochondrial morphology after apoptotic stress induction. The mitochondria of WT p53 and p53^−/− MEFs were labeled with targeted GFP, and the mitochondrial structure was evaluated by confocal microscopy. Treatment with H₂O₂ for 3 h caused a strong reduction in the average mitochondrial volume in WT cells, as expected, upon network breakage (Fig. 3 G, i and ii, and H, i and ii, and Fig. S4 A and B). The induction of p53 by ADRIA alone did not significantly affect mitochondrial morphology (Fig. 3 G, iii, and H, iii, and Fig. S4 A and B), whereas ADRIA treatment followed by H₂O₂ exposure induced a stronger increase in the fragmentation index value (Fig. 3 G, iv, and H, iv, and Fig. S4 A and B) compared with H₂O₂ alone. p53^−/− cells treated with H₂O₂ did not show significant changes in the mitochondrial network (Fig. 3 I and J and Fig. S4 C and D).

Naturally Occurring p53 Mutants Lose Their Ability to Modulate Ca²⁺ Responses.

To exclude the possibility that a transcription-dependent pathway of p53 accounts for its effect on Ca²⁺ homeostasis (and, in turn, on the sensitivity of ER-stress apoptotic stimuli), we used different strategies: specific drugs blocking the transcriptional arm of p53 (Fig. 4A) and p53-targeted chimeras, p53-ΔNLS (a nuclear import-deficient p53 mutant; Fig. 4 B and C) and ER-p53 (a chimera containing the human p53-ΔNLS protein targeted to the outer surface of the ER; Fig. 4 B and C) (27).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

p53 controls mitochondrial Ca²⁺ homeostasis and, in turn, apoptotic sensitivity from ER/MAM compartments. (A) Agonist-dependent [Ca²⁺]_m response in p53^+/+ MEFs after pharmacological block of the transcriptional arm of p53. (B) Schematic representation of p53-ΔNLS and ER-p53 chimeras. (C) Immunofluorescence images of p53^−/− MEF cells expressing the p53-ΔNLS or ER-p53 constructs stained with anti-p53 antibody (green) and Hoechst (nuclear marker). (D) Mitochondrial Ca²⁺ response in p53^−/− MEFs after the reintroduction of an ER-targeted chimera, ER-p53 or p53-ΔNLS. (E) Representative microscopic fields, from three independent experiments, of p53^−/− MEFs expressing p53-ΔNLS and ER-p53 before and after H₂O₂ treatment (1 mM, 12 h). (F) Evaluation of cell-death induction by H₂O₂ (500 μM, 12 h) through automated nucleus count analysis in p53^−/− MEFs, p53^−/− MEFs expressing ER-p53, and p53^−/− MEFs expressing p53-ΔNLS. Bars, SEM. (G) Analysis of apoptotic markers by immunoblot in p53^−/− MEFs and p53^−/− MEFs expressing p53-ΔNLS and ER-p53 under untreated conditions and after H₂O₂ treatment (500 μM, 6 h).

As pharmacological treatments, we used α-amanitin, a highly specific and potent inhibitor of RNA polymerase II transcription, or a combination of pifithrin α, which selectively blocks p53-mediated transcription, and ADRIA to activate the remaining p53 pathways (28). As expected, we observed increased mitochondrial Ca²⁺ responses under both conditions (Fig. 4A), reflecting increased ER Ca²⁺ release and indicating a nontranscriptional role for p53 in Ca²⁺ modulation.

Moreover, the expression of p53-ΔNLS or ER-p53 chimeras in p53^−/− MEF (Fig. 4D), HeLa (Fig. S5A), and H1299 (Fig. S5B) cells enhanced mitochondrial Ca²⁺ signaling, similar to the effect of p53 induction by ADRIA (Fig. 3C). This effect was further associated with a reestablished sensitivity to apoptosis induced by ER stress (Fig. 4E), as determined by cell count analysis (Fig. 4F) and PARP and caspase 3 cleavage (Fig. 4G).

In contrast, MDA-MB 468 cells, harboring the p53 273H mutant, were not sensitive to p53-ADRIA induction (Fig. 5A).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

p53 mutants cannot modulate the mitochondrial Ca²⁺ response and thus apoptosis. (A) Mitochondrial Ca²⁺ response after agonist stimulation in HCT-116 p53^−/− cells and HCT-116 p53^−/− cells after reintroduction of the p53-ΔNLS and ER-p53 chimeras or naturally occurring p53 mutants R175H and R273H. (B) Mitochondrial [Ca²⁺] after ATP stimulation measured in MDA-MB 468 cells, harboring p53 273H mutation, under control conditions and after adriamycin treatment (1 μM, 6 h). (C and D) Evaluation of apoptosis induction in HCT-116 p53^−/− cells expressing p53-ΔNLS and ER-p53 chimeras or naturally occurring p53 mutants R175H and R273H after treatment with H₂O₂ using (C) immunoblot detection of cleaved PARP and cleaved caspase 3 (500 μM, 6 h) and (D) automated cell count analysis (500 μM, 12 h). Bars, SEM. (E) Representative images of HCT-116 p53^−/− cells expressing different p53 constructs under untreated conditions and after H₂O₂ treatment (1 mM, 12 h).

Accordingly, naturally occurring p53 mutants expressed in HCT-116 p53^−/− cells (or in HeLa and H1299 cells; Fig. S5 A and B) lost their ability to increase the Ca²⁺ response (Fig. 5B).

Moreover, those mutants were unable to modulate Ca²⁺ homeostasis also failed to rescue the sensitivity to apoptosis after oxidative stress treatment (Fig. 5 C–E), although the apoptotic genes were expressed equally in cells expressing mutant p53 (Fig. S5C). A similar effect was observed when these mutations were introduced into p53-ΔNLS or ER-p53 chimeras (Fig. S6).

These data show that Ca²⁺-mediated apoptosis is a transcription-independent pathway regulated by p53 at ER/MAMs.

p53 Modulates Ca²⁺ Homeostasis and Apoptosis by Interacting with the Sarco/ER Ca²⁺-ATPase Pump at the ER, Changing Its Redox State.

To dissect the mechanism by which WT p53 exerts its impact on Ca²⁺ homeostasis upon activation and accumulation at ER/MAM compartments, we examined whether p53 modulates the activity of sarco/ER Ca²⁺-ATPase (SERCA) pumps, which mediate ER Ca²⁺ reaccumulation. Thus, we tested whether p53 functionally and physically interacts with SERCA. The in vitro interaction between p53 and endogenous SERCA was first detected using an MBP pull-down assay (Fig. 6A and Fig. S7A) in H1299 cells lacking p53, and the interaction was confirmed by the endogenous coprecipitation of the two proteins in WT MEFs (Fig. S7B). Next, we mapped the region of p53 involved in the interaction with SERCA. To this end, we used HA-tagged p53 deletion constructs: HAp53 1–175, HAp53 175–393, HAp53 294–393, and the full-length HA-p53. In coimmunoprecipitation experiments performed in H1299 cells transfected with the HA-p53 constructs, SERCA selectively bound to the C-terminal regulatory domain of p53, a region where posttranslational modifications can modify the interaction of p53 with partner proteins (Fig. 6B and Fig. S7C). However, the interacting fragment alone was not sufficient to modulate Ca²⁺ homeostasis and apoptosis (Fig. S7 D and E), suggesting that the entire p53 protein (or the majority of it) is required for its biological activity.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 6.

p53 interacts with SERCA and stimulates Ca²⁺ accumulation in the ER, changing the SERCA oxidative state. (A) In vitro binding of endogenous SERCA to MBP-p53. H1299 cell lysates were incubated with bacterially expressed MBP-p53 protein or MBP as a control. Ponceau staining shows the amount of MBP proteins used in the experiments. (B) Full-length (FL) and HA-tagged p53 deletion mutants transiently expressed in H1299 cells were immunoprecipitated by anti-HA antibody and analyzed by Western blot (WB). IP, immunoprecipitation. (C) H1299 cells were transiently transfected with different p53 constructs (FL, full-length p53 WT; ΔNLS, p53-ΔNLS; R175H, p53 R175H; R273H, p53 R273H) and then harvested for immunoprecipitation and immunoblotting as indicated. WCL, whole cell lysate. (D and E) Rate analysis of Ca²⁺ uptake measured in the ER vesicles isolated from the (D) liver of p53^−/− and p53^+/+ mice and p53^+/+ mice treated with adriamycin (1 μM, 6 h) or (E) ER compartments of p53^−/− and p53^+/+ MEFs at different times after ADRIA (1 μM, 30 min, 3 h, 6 h) treatments. (F) Immunoblot with an antibody reactive to dimedone-conjugated cysteine residues of the protein sample extracts from HCT-116 p53^+/+ and MDA-MB 468 cells after ADRIA induction that were immunopurified using a monoclonal SERCA2 antibody. Cells with an active p53 reveal lower cysteinyl sulfenic acid-modified SERCA. (G) Analysis of ER Ca²⁺ uptake in HCT-116 p53^+/+ and MDA-MB 468 cells after ADRIA induction. Bars, SEM.

Interestingly, the naturally occurring p53 mutants R175H and R273H were unable to be coimmunoprecipitated with the SERCA protein. In contrast, the p53-ΔNLS protein retained this ability (Fig. 6C).

Finally, to establish whether p53 has a direct effect on SERCA activity, we analyzed the kinetics of ER Ca²⁺ accumulation both in vivo and in vitro. The Ca²⁺ accumulation rate was higher in the ER vesicles isolated from the liver of p53^+/+ mice treated intraperitoneally with ADRIA compared with those obtained from p53^+/+ and p53^−/− mice (Fig. 6D). Similarly, the rate of ER Ca²⁺ accumulation, measured in MEFs, increased proportionally in a time-dependent manner with the induction of p53 by ADRIA, indicating a stimulatory role of p53 in SERCA activity (Fig. 6E), without affecting SERCA expression levels (Fig. S8A). These data were confirmed in HeLa cells overexpressing WT p53 and p53-NLS chimera (Fig. S8 B and C). To confirm the importance of SERCA activity in the Ca²⁺-dependent apoptotic pathway, we evaluated the effect of SERCA overexpression on the Ca²⁺ response and cell death in p53^−/− MEFs. We observed an increased mitochondrial Ca²⁺ uptake (Fig. S8D) that correlated with increased levels of cleaved caspase 3 upon H₂O₂ treatment (Fig. S8E), indicating that the activation of SERCA is sufficient to rescue the sensitivity to apoptosis in p53-deficient cells.

Next, we investigated the possible mechanism by which p53 stimulates SERCA activity upon binding. To this end, we analyzed whether p53 activation affects the oxidative state of the SERCA protein, which is known to modulate its activity (29). We thus compared the level of hyperoxidized sulfenylated proteins in HCT-116 p53^+/+ and MDA-MB 468 cells after the induction of WT p53 and mutant p53, respectively. Cells and lysates were exposed to dimedone, a chemical that selectively modifies sulfenylated cysteines, and the dimedone-modified proteins were detected by immunoblotting with an antibody to dimedone (30). In the absence of dimedone, the antibody gave a weak background signal (Fig. S9).

To understand whether the levels of sulfenylated SERCA were different in HCT-116 p53^+/+ and MDA-MB 468 cells, both cell types were exposed to dimedone to quench the sulfenylated cysteines, and the proteins were immunopurified using a monoclonal SERCA antibody. Immunoblotting of the SERCA immunocomplex with an antibody to dimedone revealed the presence of sulfenylated SERCA, which was lower in HCT-116 p53^+/+ cells treated with doxorubicin (Fig. 6F). In contrast, in MDA-MB 468 cells, SERCA oxidation was unchanged after ADRIA treatment. These results were then confirmed by measuring SERCA activity. Fig. 6G shows that ADRIA required WT p53 to be effective. Indeed, in MDA-MB 468 cells, SERCA activity was not sensitive to ADRIA induction of p53.