PML Regulates Apoptosis at Endoplasmic Reticulum by Modulating Calcium Release

The promyelocytic leukemia gene (PML)was originally identified at the breakpoint of the t(15;17) translocation of acute promyelocytic leukemia (APL), and function of the PML protein is frequently lost or aberrant in human solid tumors and hematopoietic malignancies (1, 2). PML is a nuclear protein and an essential component of subnuclear structures termed nuclear bodies (NBs) (3). However, many, if not all, PML isoforms have shown both cytoplasmic and nuclear localization (4, 5). Pml^−/− mice and primary cells are protected from apoptosis triggered by a number of diverse stimuli (6).

To determine how PML could regulate such broadly diverse apoptotic responses, we analyzed its intracellular localization by cell fractionation (7, 8). We fractionated homogenates of primary mouse embryonic fibroblasts (MEFs) by ultracentrifugation, focusing on the mitochondria, endoplasmic reticulum (ER), and mitochondria-associated membranes (MAMs), the structures that contain sites where the ER contacts mitochondria. Pml localized both to the nucleus and the cytosol and appeared to localize also to the ER, MAM, and crude mitochondrial fractions but not to “pure” mitochondrial fraction free of ER and nuclear markers (Fig. 1A). These results were confirmed by immunogold labeling of ultrathin cryosections showing that Pml associates with the surface of the ER (Fig. 1B, a and b) and in the proximity of the mitochondrial membrane at contact sites between the ER and mitochondria (Fig. 1B, d to g).

In view of the localization of Pml at the ER and MAM, we investigated its requirement in apoptosis induced by ER stress (9). Matched wild-type (Pml^+/+) and Pml^−/− MEFs were treated with ER stress inducers: H₂O₂ and menadione (MEN), two oxidizing agents that induce ER Ca²⁺ release; tunicamycin (TN) or an inhibitor of protein N-glycosylation; and thapsigargin (TG), an inhibitor of the sarcoplasmic/ER Ca²⁺-ATPase (adenosine triphosphatase). After 12 hours of treatment, the percentage of apoptotic cells in Pml^−/− MEFs was much lower than that observed in Pml^+/+ MEFs under all treatment conditions (Fig. 1C and fig. S1).

MAMs are specialized domains selectively enriched in critical Ca²⁺ signaling elements, which mediate Ca²⁺ transfer between ER and mitochondria (10, 11), such as the inositol 1,4,5-trisphosphate receptor (IP₃R) (12). Ca²⁺ signaling has a major role in the regulation of cell death (13, 14). Release of the ER Ca²⁺ pool through the type 3 IP₃R (IP₃R3) appears to induce a sensitization of cells to apoptotic stimuli (15, 16).

To investigate the role of Pml in Ca²⁺ homeostasis, we used recombinant Ca²⁺-sensitive bioluminescent protein aequorin (17). In Pml^+/+ MEFs, the Ca²⁺ concentration ([Ca²⁺]) in the lumen of the ER ([Ca²⁺]_ER) at steady state was ~450 µM, whereas in Pml^−/− MEFs it was lower. When the cells were stimulated with adenosine 5′-triphosphate (ATP), the P2Y receptor agonist that causes release of Ca²⁺ from the ER, the decreases in the [Ca²⁺]_ER observed in Pml^+/+ MEFs in quantitative and kinetic terms were larger and faster than in Pml^−/− MEFs, reflecting a more rapid flow of Ca²⁺ through the IP₃R (Fig. 2A). In turn, the [Ca²⁺] increases evoked by stimulation with ATP in the cytosol ([Ca²⁺]_c) and mitochondria ([Ca²⁺]_m) were smaller in Pml^−/− than in Pml^+/+ MEFs (Fig. 2, B and C, and fig. S2). A mitochondrial Ca²⁺ deregulation was observed also in human-derived cells in which PML was depleted (fig. S3) and in different cellular models of APL (fig. S4).

We then investigated whether the absence of Pml could alter the increases in [Ca²⁺]_c and [Ca²⁺]_m induced by apoptotic stimuli. In Pml^−/− MEFs, the increases in [Ca²⁺]_c and [Ca²⁺]_m, evoked by the oxidative apoptotic stimuli, such as MEN and H₂O₂ that trigger both a progressive release of Ca²⁺ from the ER and an activation of the capacitative Ca²⁺ influx (18), were smaller as mentioned above (Fig. 2, D and E).

To determine whether the effects of Pml on regulation of Ca²⁺ homeostasis depend on its localization to the ER and MAMs, we generated a chimeric protein containing the entire PML protein that was targeted to the outer surface of the ER (19). This chimera, designated erPML, localized to the ER and MAMs in Pml^−/− MEFs, as revealed by immunocytochemical staining, immunogold labeling, and subfractionation (Fig. 3A and fig. S5). The introduction of erPML in Pml^−/− MEFs restored Ca²⁺ signals evoked by either agonist (Fig. 3B and fig. S6) or apoptotic stimuli (MEN or H₂O₂) (Fig. 3C) to values comparable to those in Pml^+/+ MEFs (Fig. 2, C and D). This effect was associated with a reestablished sensitivity to apoptosis induced by ER stress but did not restore the sensitivity to etoposide (ETO) (Fig. 3D and fig. S7A), a DNA-damaging agent that triggers apoptotic death by a Ca²⁺-independent process (fig. S7B). Overall, these experiments indicate that the absence of Pml causes a reduction in the amplitude of Ca²⁺ signals induced by ATP, other agents, or apoptotic stimuli, and that forcing PML to the ER rescues these defects. A PML protein targeted to the nucleus restored the formation of NBs, but did not restore the Ca²⁺ responses and the sensitivity to ER stress–dependent cell death, although it restored response to other apoptotic stimuli such as ETO (fig. S8).

To investigate the mechanism underlying these activities of Pml, we tested whether Pml could functionally and physically interact with the IP₃R3. Immunoprecipitation of IP₃R3 led to the coprecipitation of Pml (Fig. 4A and fig. S9) and vice versa (fig. S10). Amounts of phosphorylated-IP₃R3 (p-IP₃R3) were higher in Pml^−/− than in Pml^+/+ MEFs (Fig. 4A and fig. S11).

Reduced cellular sensitivity to apoptotic stimuli was observed in cells with high activity of the protein kinase Akt, as a result of diminished Ca²⁺ flux from the ER through the IP₃R (20, 21). The amount of phosphorylated Akt (pAkt) (that is, the active form of Akt) coprecipitated with IP₃R3 (Fig. 4A and fig. S9) was higher in Pml^−/− than in Pml^+/+ MEFs (fig. S11). Dephosphorylation of Akt at the MAM might occur through Pml-mediated recruitment of the phosphatase PP2a. Indeed, Pml interacts with PP2a in Pml-NBs (22). Further, the amount of PP2a coprecipitated with IP₃R3 (Fig. 4A and figs. S9 and S10) was diminished in Pml^−/− MEFs (Fig. 4A and fig. S11). Thus, in the absence of Pml, reduced Ca²⁺ release could be caused by increased phosphorylation and activation of Akt at the ER due to an impaired PP2a activity, which in turn impair Ca²⁺ flux through the IP₃R because of its hyperphosphorylated state (figs. S11 and S12).

To determine whether IP₃R3, Pml, Akt, and PP2a interact in a complex, we next performed two-dimensional blue native analysis. We found that Pml, IP₃R3, Akt, and PP2a colocalize in high molecular weight complexes, supporting their possible interaction in the native state (fig. S13). Finally, we demonstrated the localization of all these proteins at the ER and MAM through immunocytochemical staining and subfractionation (Fig. 4, B and C, and fig. S14).

We further investigated the correlation among Pml, Akt, and PP2a at the ER and the regulation of the IP₃R by a selective inhibition of either Akt or PP2a. Pretreatment of cells with okadaic acid (OA, a PP2a inhibitor) caused a reduction in [Ca²⁺]_m responses to ATP stimulation and a reduced H₂O₂- or MEN-induced death in Pml^+/+ MEFs and in Pml^−/− MEFs expressing erPML, but not in Pml^−/− MEFs (Fig. 4, D and E, and figs. S15, S16, and S18), in which PP2a activity is impaired. LY294002 (an inhibitor of Akt) had no effect on the agonist-dependent [Ca²⁺]_m transients and on apoptosis in Pml ^+/+ or Pml ^−/− MEFs expressing erPML, whereas it increased agonist-dependent [Ca²⁺]_m responses and restored sensitivity to H₂O₂ or MEN (Fig. 4, D and E, and figs. S15, S17, and S18) in Pml^−/− MEFs (in which high levels of pAkt are observed; Fig. 4A and fig. S11). These results were confirmed in experiments in which RNA interference was used to deplete cells of Akt or PP2a proteins (fig. S19, A and B) or a constitutively active form of Akt (m/p Akt) was expressed (fig. S19C).

Our data highlight an extranuclear, transcription-independent function of Pml that regulates cell survival through changes in Ca²⁺ signaling in the ER, cytosol, and mitochondria (fig. S20). This effect appears to be specific to Ca²⁺-mediated apoptotic stimuli because alteration in Pml did not influence cell death in cells treated with ETO, which activates the apoptotic pathway in a way largely independent of Ca²⁺.

This mechanism may explain how Pml can so broadly regulate the early (and transcription independent) apoptotic response. Our findings may have implications in tumorigenesis where the function of Pml is frequently lost, or in other patho-physiological conditions where Pml is accumulated such as cell stress, or infection with viral or bacterial pathogens.