p53 can eliminate damaged cells through the induction of mitochondria-mediated apoptosis. Recent observations have provided strong evidence that a fraction of total p53 translocates to mitochondria specifically in response to a death stimulus. Unexpectedly, mutant p53, which is expressed at much higher levels than wild type in unstressed cells, is apparently always present at the mitochondria, independent of apoptotic signal. This prompted us to ask whether cell lines with intact p53-dependent apoptosis and cell cycle arrest pathways exist in which the mitochondrial localization of wild-type p53, like that of mutant, is independent of a death stimulus and instead, correlates with the total p53 levels. Here, we document that human HCT116 colorectal carcinoma cells treated with adriamycin or 5-fluorouracil (5FU) can accumulate total p53 to equally high levels, and mitochondrial p53 to proportionate levels, although only 5FU treatment provoked p53-dependent apoptosis. Along the same line, HCT116 derivatives with increased basal p53 levels, and glioblastoma cells with a doxycycline-inducible p53, also revealed proportionate mitochondrial p53 levels, and even unstressed HCT116 cells had some p53 located at the mitochondria. Finally, mitochondrial and total p53 showed distinct post-translational modifications. Thus, cell lines exist in which the mitochondrial p53 levels parallel total levels independent of apoptosis.
p53 is a predominantly nuclear, homotetrameric protein that is usually dormant in normal, young, and unstressed cells (for a recent review, see Vousden and Lu, 2002). Various stresses, including DNA damage, hypoxia, and derailed oncogene expression, all important for the development of cancer, can activate p53. Active p53 is multifunctional, owing to its ability to bind to specific DNA and RNA sequences as well as to numerous cellular proteins. The best understood consequences of p53 activation are the arrest of cell cycle in the G1 and G2 phases, mostly through the transcriptional activation by p53 of the p21Waf1/Cip1 (CDKN1A) inhibitor of cyclin-dependent kinases, and the induction of programmed cell death (apoptosis). Together, these two effects appear to constitute the core armament of cells against neoplastic transformation, with the remarkable general consequence that cancers apparently cannot develop with all p53 and connected pathways fully intact.
With more than 4000 putative p53-binding sites identified in the genome (Wang et al., 2001), with at least 70 p53-responsive genes already known, and with the irreversibility and definitiveness of apoptosis demanding delicate regulation, it is no wonder that cell survival control by p53 is complex and cell type dependent. Among the 20 or so known p53-responsive, cell survival-regulating genes are the sequences for the BH (Bcl-2 homology domain) 123 protein Bax (Miyashita and Reed, 1995), the BH3-only proteins Noxa and PUMA (Nakano and Vousden, 2001; Oda et al., 2000a; Yu et al., 2003), the apoptosome-constituent Apaf-1 (Fortin et al., 2001; Moroni et al., 2001; Robles et al., 2001), mitochondria-localized p53AIP (Oda et al., 2000b), plasma membrane-localized PERP (Attardi et al., 2000), the death domain-containing PIDD (Lin et al., 2000), and the death receptors Fas and KILLER/DR5 (Owen-Schaub et al., 1995; Wu et al., 1997). Besides transactivating proapoptotic genes, p53 can transcriptionally repress survival factor genes like the ones for Bcl-2 and the IAPs (reviewed in Schuler and Green, 2001; Johnstone et al., 2002). Notably, none of these genes appears to satisfy the requirements for a master switch, although the recent discovery that irradiation-induced thymocyte apoptosis is abrogated in knockout mice lacking PUMA points to this mitochondria-localized, Bcl-2- and Bcl-XL-associated protein as an important candidate.
Apoptosis seems to constitute a threshold phenomenon requiring the integration of various signals transmitted through different routes. Cell survival regulation by p53 is further complicated by the crosstalk with other pathways. For example, signals stimulating the overproduction of antiapoptotic proteins such as Bcl-2 may over-rule the full mitochondria-mediated death program by p53 (Schmitt et al., 2002). Not least, stimulation of p21Waf1/Cip1 expression by p53 can attenuate apoptosis induction in some contexts (Bunz et al., 1999; Mahyar-Roemer and Roemer, 2001), perhaps by preventing the resolution of the inhibitory complex between the retinoblastoma protein and the proapoptotic E2F1 transcription factor.
p53 can provoke death in the absence of any transcription or translation in certain cell types (Caelles et al., 1994; Wagner et al., 1994; Yan et al., 1997; Gao and Tsuchida, 1999). However, it is still a subject of debate as to how much, and through which routes, the active p53 protein per se, independent of transcription, contributes to the phenotype of p53-mediated apoptosis. Studies using cell-free cytosolic extracts have documented for the first time that p53 protein may directly mediate the activation of the major apoptosis-executing proteinase caspase-3 through the activation of the precursor caspase-8 (Ding et al., 2000). Furthermore and most strikingly, Moll and co-workers have recently demonstrated that a fraction of cellular wild-type p53 localizes to the mitochondria under conditions that provoke apoptosis, and that mitochondria-localized p53 is sufficient to launch cell death via the release of cytochrome c (Marchenko et al., 2000; Sansome et al., 2001). The same researchers went on to show that p53 achieves this through the direct physical interaction via its central DNA-binding domain with the antiapoptotic Bcl-2 and Bcl-XL proteins (Mihara et al., 2003). They suggest that p53 may act by sequestering the antiapoptotic proteins away from the proapoptotic Bax/Bak proteins, in analogy to the effects of the proapoptotic BH3-only proteins and the BH3 domain-lacking p53AIP1, which also binds Bcl-XL. As a result, Bax/Bak can multimerize and permeabilize the outer mitochondrial membrane, and thereby release cytochrome c from the intermembrane space (Scorrano and Korsmeyer, 2003). One further intriguing result of these studies, and a primer of the work presented here, was the observation that whereas wild-type p53 is apparently translocated to the mitochondria in a highly regulated way, only during apoptosis but not cell cycle arrest, mutant forms of the protein (including DNA-contact mutants with wild-type conformation) are localized at the mitochondria all the time, even in the absence of a death stimulus and proportional to the total p53 levels (Mihara et al., 2003). We therefore asked whether wild-type, like mutant, p53 can be present at the mitochondria even in the absence of apoptosis, at least in certain cell types, and if so, how the mitochondrial p53 levels relate to the total cellular levels of the tumor suppressor.
HCT116 is a poorly differentiated, growth factor-insensitive human colorectal adenocarcinoma cell line exhibiting the microsatellite instability (MIN) phenotype as a result of deficiency for hMLH1, the human homologue of the bacterial Mut L protein and essential factor of DNA mismatch repair. HCT116 cells are responsive to various stresses, including DNA damage and spindle disruption, primarily because of the functional integrity of the p53/p21Waf1/pRb tumor suppressor pathway. This and the availability of several derivative cell lines with genes knocked out by targeted homologous recombination (Waldman et al., 1995; Bunz et al., 1998) have made these cells a widely used tool for the study of the p53 pathway. We (Mahyar-Roemer and Roemer, 2001; Mahyar-Roemer et al., 2001) and others (Bunz et al., 1999) have recently reported that the presence or absence of p53 or p21Waf/Cip1 can profoundly affect proliferation and survival of HCT116 cells treated with anticancer or chemopreventive agents, and furthermore that the actual cellular responses can be dependent on the type of drug.
p53 protein levels in drug-treated HCT116 cells
When exponentially growing HCT116 cultures were either mock-treated or incubated with the DNA-damaging drug adriamycin (ADR) or the DNA- and RNA-damaging inhibitor of thymidylate synthase, 5-fluorouracil (5FU), p53 levels began to rise within 3 h and plateaued at 12–24 h. Immunoblotting furthermore showed that this rise in total cellular p53 levels was accompanied by a simultaneous increase in the levels of p53 phosphorylated at the serine at amino-acid position 15, the target of the DNA damage induced ATM/ATR kinases (Iliakis et al., 2003) (Figure 1a). A careful inspection of the steady-state levels of p53 with two different p53 antibodies at 18 h after exposure to doses of ADR (0.34 μ M) and 5FU (375 μ M) as they have been employed in the recent literature revealed that, under the culture conditions used by us, p53 accumulated to higher levels in the presence of 5FU. When the ADR dose was increased to 1.36 μ M, similar steady-state levels of total p53 were reached in the ADR- and 5FU-treated cultures within 18 h (Figure 1b). Figure 1c documents that the accumulation of total p53 under ADR and 5FU to comparable levels was paralleled by a greatly and proportionally increased phosphorylation of p53 at S15. In contrast, phosphorylation of S46, which has been associated with apoptosis in some cellular contexts (Oda et al., 2000b), was very weak and detectable only after a greatly prolongated exposure of the immunoblot (120 vs 5 min). Several recent studies have shown that the p300/CBP histone acetyltransferase can acetylate p53 on various lysine residues within the C-terminus, among them K382 (Gu and Roeder, 1997; Prives and Manley, 2001). Approximately equal quantities of p53 from ADR- and 5FU-treated cells were acetylated at position K382. In accord with the presence of active p53, both ADR- and 5FU-treated cultures produced high levels of p21Waf/Cip1. Thus, 1.36 μ M ADR and 375 μ M 5FU induce, in proliferating HCT116 cultures, the accumulation of similar levels of total p53 and of p53 protein carrying post-translational modifications known to be associated with the active form of the tumor suppressor.
Effects of ADR and 5FU on HCT116 cells
When exponentially growing HCT116 cultures were incubated with ADR or 5FU for 3–18 h, p53 protein accumulated predominantly in the cell nucleus (Figure 2a). In agreement with recent reports by Bunz et al. (1999), flow cytometric analysis showed that ADR causes the arrest of the cell cycle in the G2/M phase within 48 h, but not apoptosis, whereas 5FU induces a substantial peak of cells with a sub-2n DNA content indicative of apoptotic degradation (Figure 2b). In contrast to the findings by Bunz et al. (1999), however, and in accord with the increased expression of p21Waf/Cip1 (see Figure 1c), we consistently observed the arrest of a proportion of the cells in G1. The inclusion of isogenic p53-negative HCT116 cells in the study (Bunz et al., 1998) revealed that the (minor) G1 arrest following ADR treatment as well as the apoptosis following 5FU treatment were dependent on p53, confirming previous observations (Bunz et al., 1999). Apoptosis under 5FU was detectable as early as 24 h after the beginning of treatment and continued to increase for further 24 h (Figure 2c). To substantiate that the observed cell death in 5FU-treated cultures was apoptotic, cell fractionations and immunoblot analyses were carried out. Figure 2d shows that in the presence of 5FU, but not ADR, the apoptosis-mediator cytochrome c is gradually released from the mitochondrial intermembrane space and is accumulating in the cytoplasm. Furthermore, it shows that the cytochrome c release is accompanied by the appearance of cleaved (active) caspase-3, indicative of the stimulation of the mitochondrial death pathway by 5FU. Accordingly, apoptosis by 5FU could be prevented by the incubation of the cultures with increasing doses of the pancaspase-inhibitor z-VAD-fmk (Figure 2e). Thus, ADR and 5FU at doses chosen to induce primarily nuclear localization of similar levels of active p53 provoked very different, yet p53-dependent responses in HCT116 cells: mitochondria-mediated apoptosis in the case of 5FU and a minor G1 cell cycle arrest (superimposed by a major, p53-independent G2/M arrest) without significant apoptosis in the case of ADR.
Both ADR and 5FU induce p53's presence in mitochondrial fractions
Previous work by Moll and co-workers (Marchenko et al., 2000; Sansome et al., 2001; Mihara et al., 2003) has documented that a proportion of cellular wild-type p53 localizes to the mitochondria of a variety of primary cells and wild-type p53-positive tumor cell lines, including the human colorectal carcinoma line RKO, under conditions that provoke apoptosis but not cell cycle arrest. To study mitochondrial localization, Moll and co-workers employed the well-established classic discontinuous sucrose gradient procedure to generate highly enriched mitochondrial fractions from whole cells and then analysed these by immunoblotting. Contamination of the mitochondrial fraction by nuclei or soluble nuclear proteins leaked into the cytoplasm was monitored by detecting nuclear or cytosolic marker proteins in the immunoblots. An antibody directed against the strictly mitochondrial cytochrome oxidase subunit IV (OX IV) was used to control for the enrichment of the mitochondrial fraction and to serve as a reference when comparing mitochondrial p53 levels in differently treated cultures. We have employed the same techniques (Bogenhagen and Clayton, 1974; Marchenko et al., 2000) in the following studies. Exponentially proliferating HCT116 cultures (108 cells) were mock treated or incubated with 1.36 μ M ADR or 375 μ M 5FU for 18 h. After cell fractionation, 5 μg each of total cell protein (t), mitochondrial (HM) fraction and free protein (f) fraction containing cytosolic plus leaked nuclear proteins were analysed by immunoblotting (Figure 3a). Both drugs induced a robust, readily detectable expression of nuclear p21Waf/Cip1. Absolutely no p21Waf/Cip1 was detectable in the HM fraction. In contrast, a strong p53 signal was present in the HM fraction, in accord with the results by Moll and co-workers (Marchenko et al., 2000), indicating that a proportion of p53 can localize to the mitochondria. However, ADR and 5FU produced equally strong signals, although the former cannot provoke apoptosis in HCT116 cells, indicating that in this cell type and under the described conditions, p53 can associate with mitochondria at 18 h after drug exposure even in the absence of apoptosis.
To further examine the localization of p53 to the mitochondrial fraction, we next made use of the isogenic HCT116 p21−/− cells deficient for p21Waf/Cip1 due to targeted homologous recombination (Waldman et al., 1995). HCT116 p21−/− cells have a higher basal level of p53 compared to the parental HCT116 cells; this can be further increased in response to DNA-damaging drugs such as ADR (Figure 3b). The advantage of the comparison of HCT116 cells with the p21-negative derivatives for the present study is twofold: first, flow cytometry has revealed that exponentially growing HCT116 p21−/− cultures have a higher basal rate of apoptosis compared to HCT116 cultures (7 vs 1.9%; see Figure 3c), which seems to result at least partially from the elevated p53 levels (Mahyar-Roemer and Roemer, 2001). Second, ADR, producing only cell cycle arrest in HCT116 cells, induces programmed cell death in the p21-negative derivatives (29.1 vs 3.2% apoptosis; Figure 3c) (Bunz et al., 1999; Mahyar-Roemer and Roemer, 2001). We therefore treated HCT116 cultures with ADR or 5FU, provoking either cell cycle arrest or cell death, as before, or incubated HCT116 p21−/− cultures with ADR and compared them with mock-treated cultures. After 18 h, the cells were carefully fractionated and the loaded protein quantities were adjusted to approximately equal levels of total p53 prior to final immunoblot analysis (Figure 3d; note that slightly lower quantities of the ADR-treated HCT116 p21−/− extracts were loaded to compensate for the higher p53 levels under ADR compared to mock; see also Figure 3b). Consistent with the results shown in Figures 1b and 3a, 1.36 μ M ADR and 375 μ M 5FU produced equal levels of total p53, and gave rise to equal levels of mitochondrial p53 (compare also with the levels of mitochondrial OX IV in the differently treated cultures). Thus, the levels of p53 in the mitochondrial fractions paralleled the levels of total cellular p53 regardless of whether 5FU provoked apoptosis in HCT116 cells, whether ADR induced cell cycle arrest in HCT116 cells or apoptosis in the p21−/− derivatives, or whether p53 levels were elevated in the absence of drug, as was the case with the mock-treated HCT116 p21−/− cells. Combined, the data therefore suggest that the ratios of mitochondrial p53 to total p53 are constant in HCT116 cells after 18 h of treatment and not influenced by stress response.
To further substantiate that the p53 protein present in the mitochondrial fraction is not just contamination from the nuclear p53 pool, and to begin simultaneously to probe for differences in post-translational modifications between the two p53 fractions, we next analysed total and mitochondrial protein from HCT116 cultures exposed to either ADR or 5FU for the presence of modified p53. As before, there was no p21Waf/Cip1 signal detectable in the mitochondrial fractions, despite a robust signal being present in the total protein fractions, documenting the purity of the HM preparation (Figure 3e). In contrast, p53 was present in the mitochondrial fractions prepared from the ADR- and 5FU-treated cultures in approximately equal quantities. Moreover, a similar portion of the mitochondrial p53 in the ADR and 5FU lanes was phosphorylated at S15 (compare also with the total p53 and p53-S15 levels in Figure 1c). However, and remarkably, we were unable to find any mitochondrial p53 acetylated at position K382, although acetylated p53-K382 was readily detectable in the total p53 pools (Figure 3e). We conclude that the post-translational modification of the p53 in the mitochondrial fraction is distinct from that of the remaining p53 protein.
Mitochondrial p53 in unstressed HCT116 cells
HCT116 cells have a functional wild-type p53-dependent stress-response pathway whose activity can be sensitively monitored by the expression level of the p21Waf/Cip1 inhibitor of cyclin-dependent kinases (Waldman et al., 1995; Mahyar-Roemer and Roemer, 2001). In addition, these cells contain appreciable quantities of p53 even in the absence of stress, that is, under conditions of exponential proliferation and background p21Waf/Cip1 levels. In light of our observation that subcellular fractionation and immunoblot analysis were able to detect small quantities of p53 even in the mitochondrial fraction of unstressed cells, we asked next whether it would be possible to observe p53 associated with mitochondria in a blinded study by immunogold electron microscopy. For this purpose, parental HCT116 cells and HCT116 p53−/− cells were cultured and monitored for exponential growth (background p21Waf/Cip levels, approximately 28 h population doubling time). The unstressed cultures were rapidly cooled and mitochondria were immediately prepared from the p53-proficient and -deficient cells. As a further control, mitochondria from the p53-negative cells were incubated for 10 min on ice with cytosol from the parental HCT116 cells. All three mitochondrial preparations (HCT116, HCT116 p53−/−, and p53−/− plus HCT116 cytosol) were prepared for electron microscopy, incubated either with mouse IgG or anti-p53 antibody DO-1 and stained with colloidal gold. Intact mitochondria in randomly selected, photographed fields were subjected to analysis. Figure 4 shows that gold grains indicative of p53 can be observed at and within the mitochondria of unstressed HCT116 cells. In contrast, mitochondria from p53-negative cells as well as p53-negative mitochondria incubated with p53-positive cytosol produced lower labelling levels. Incubations with IgG control antibody produced background labelling in all three mitochondrial preparations. Altogether, up to four times as many grains were counted in the HCT116-derived mitochondria incubated with DO-1 as compared to the controls, suggesting that p53 can be associated with the mitochondria of unstressed HCT116 cells.
Mitochondrial p53 levels parallel total p53 levels in glioblastoma cells
LN-Z308 is a p53-null human glioblastoma cell line known to respond with growth arrest and morphological changes to the expression of ectopic wild-type p53, and LN-Z308 wt53 cells conditionally produce wild-type p53 under the control of a doxycycline-sensitive promoter (Van Meir et al., 1994). When proliferating LN-Z308 wt53 cultures were exposed to very small and growing doses of doxycycline for 24 h, increasing steady-state levels of p53 could be induced (Figure 5a). Quantitation of the p53 levels revealed that approximately four times less p53 was present in the cultures at a drug dose of 30 ng/ml when compared to a dose of 80 ng/ml. However, even at high doxycycline doses and correspondingly high p53 levels, there was only weak expression of the p53 target gene p21Waf/Cip1. In accord with the apparently weak activity of the p53 expressed in these cells, flow cytometry analyses at 48 h after drug treatment showed that there was no difference in the rate of sub-G1 (apoptotic) cells between the mock-treated and drug-treated cultures and, in agreement with the lack of p21Waf/Cip1 expression, no FACS profile changes indicative of a cell cycle block (Figure 5b). Since the steady-state p53 levels were regulatable in these cells and the cell responses were similar under low (30 ng/ml) and high (80 ng/ml) doxycycline doses, we again looked for the presence of p53 in the mitochondrial cell fractions, using the same preparation procedure as for the HCT116 cells. For immunoblot analysis, four times as much protein from the 30 ng/ml samples than from the 80 ng/ml samples were loaded to assure the presence of equal total p53 levels, as 80 ng/ml doxycycline induced four times more p53 (Figure 5c; note the proportionally higher levels of OX IV in the 30 ng/ml lanes). Again, equal levels of total p53 led to equal levels of p53 in the mitochondrial fractions, while Rb was not detectable in this fraction. Together, these results indicate that mitochondrial fractions prepared from HCT116 and LN-Z308 wt53 cells can contain p53 at levels proportional to the total p53 levels and independent of apoptosis. Furthermore, mitochondrial and total p53 from HCT116 cells differ in their post-translational modifications. Finally, even unstressed HCT116 cells may have small amounts of p53 associated with their mitochondria.
Recent work by Moll and co-workers (Marchenko et al., 2000; Sansome et al., 2001; Mihara et al., 2003) has provided strong evidence that p53 can contribute to apoptosis induction not only through target gene stimulation in the cell nucleus but also through the initiation of proapoptotic mechanisms that involve the physical interaction of the core DNA-binding domain of p53 with apoptosis regulators such as Bcl-2 and Bcl-XL at the mitochondria. One remarkable finding by these authors was the selectivity of the novel p53 function: a fraction of total cellular p53 translocated to the mitochondria only when undergoing p53-dependent apoptosis but not when experiencing p53-mediated cell cycle arrest, although p53 protein rose to high total levels in both instances (Marchenko et al., 2000). Obviously, these observations imply a stress signal-specific, regulated translocation mechanism. The more surprising it was to learn that, in contrast to wild-type p53, mutant p53 including DNA contact mutants with wild-type p53 conformation, were associated with mitochondria all the time, independent of apoptotic stimulus (Mihara et al., 2003). Although the impairment of the core DNA-binding domain of these mutants precluded proapoptotic function, it is not obvious why wild-type p53 would be strictly nonmitochondrial in the absence of a death signal, whereas mutant p53 of the same conformation would not. Given the usually low levels of wild-type p53 and the high levels of mutant p53 in unstressed cells, we hypothesized that the mitochondrial wild-type p53 levels parallel the total p53 levels and that the direct proapoptotic function of p53 at the mitochondria constitutes a threshold phenomenon, at least in some cell types. In accord with this hypothesis, we showed here that the ratios of mitochondrial and total p53 are indeed approximately constant in stressed HCT116 cells and independent of cellular response.
Moll and co-workers have furthermore shown that mitochondrial translocation of p53 is an early event, observable as early as 1–6 h after a death stimulus (Marchenko et al., 2000; Sansome et al., 2001; Mihara et al., 2003). However, what precisely is the status of p53 at later time points, when its levels have plateaued (12–18 h)? Does it still localize to the mitochondria only in response to a death stimulus but not other stresses? Or does p53 at these later times associate with the organelle regardless of the type of stress? If the latter is the case (as our data on HCT116 cells indicate), then why does it not always induce apoptosis under these conditions? We have initially also studied the subcellular localization of p53 at early time points, but have found at these times the accumulating p53 levels to be highly variable and inhomogeneous within a culture and between experiments (data not shown). For instance, at 3 h after ADR or 5FU exposure, only a few individual cells with relatively high intracellular p53 levels were usually present, whereas the large majority of cells showed no or almost no signal in immunofluorescence. At later times (12–18 h after treatment), it seemed that primarily the numbers of cells staining positive for p53 increased dramatically, and hence the cultures appeared more homogeneous. With these observations in mind, we decided to examine p53 localization at the earliest time points at which p53 expression had reached stable steady-state levels.
Marchenko et al. (2000) have used classical cell fractionation and immunoelectron microscopy to demonstrate for the first time that p53 protein can be present at the mitochondria in cells provoked to undergo p53-mediated apoptosis. In their studies, they have routinely used nuclear markers including PCNA that produced strong signals in the total cell fraction and in the free cytosolic protein fraction but no signal in the mitochondrial fraction, to exclude contamination of the mitochondrial fraction by nuclei or leaked nuclear proteins. P21Waf/Cip1, employed as another readily detectable nuclear protein by us, was – like PCNA in the previous studies – present in large amounts in both the total cell and free cytosolic protein fractions, but produced absolutely no signal in the mitochondrial (HM) fraction. Similarly, other markers such as Rb, c-Myc or cyclin E were absent from the HM fraction. In contrast, p53 signals were readily detectable in the HM fractions as well as by immunoelectron microscopy, in accord with the previous findings that a fraction of p53 can associate with the mitochondria (Marchenko et al., 2000; Mihara et al., 2003). At variance with these reports, however, we found similar levels of mitochondrial p53 regardless of whether p53-dependent apoptosis or cell cycle arrest was induced in human HCT116 colon adenocarcinoma cells, a cell line widely used for the study of the p53 pathways (for example Bunz et al., 1998; Hwang et al., 2001; Mahyar-Roemer and Roemer, 2001; Mahyar-Roemer et al., 2001; Bunz et al., 2002; Jallepalli et al., 2003; Lohr et al., 2003). We (Mahyar-Roemer et al., 2002) and others (Bunz et al., 1998) have previously shown that HCT116 cells express no detectable Bcl-2 but high levels of mitochondrial Bcl-XL, a target of mitochondrial p53 interaction (Mihara et al., 2003). It is thus conceivable that mitochondrial p53 lowers the threshold for apoptosis in HCT116 cells to levels that require additional (nuclear?) signals. These may then be provided only by apoptotic but not cell cycle arrest stimuli. In other words, we interpret our observations to suggest that mitochondrial p53 may be necessary but is insufficient for HCT116 cell death to occur. Alternatively, the cell cycle arrest induced by ADR but not the death stimulus by 5FU may activate a survival signal capable of over-ruling mitochondrial p53. In any event, our data show that cell types exist in which mitochondrial localization of p53 is not strictly associated with the p53-dependent apoptotic stress response, and the electron microscopy studies suggest that even in unstressed (p21Waf/Cip1-negative, exponentially growing) HCT116 cultures, low levels of p53 can be present at the mitochondria.
Several possible explanations exist for the discrepancies between the previous results and the data presented here. If mutant p53, including the (conformationally wild-type) p53-273H DNA-binding mutant, can be located at the mitochondria even in the absence of a death stimulus (Mihara et al., 2003), yet wild-type p53 is translocated to the mitochondria only in response to an apoptosis signal in some cell types (Marchenko et al., 2000; Sansome et al., 2001), a regulatable cellular factor specifically associating with wild-type but not mutant p53 and inhibiting translocation might be inferred. How might such a factor distinguish between the two p53s? If the interaction with wild-type p53 required an intact core DNA-binding domain, interactions with the mutant were precluded. Alternatively, the expression of the hypothetical factor might depend on the transactivation of its gene by p53. Obviously, the absence or malfunction of this protein factor would impose upon a cell mitochondrial p53 levels paralleling total p53 levels, as has been observed with HCT116 and LN-Z308 cells. Finally, p53's interaction with such a protein might be regulated through defined post-translational modifications.
Both ADR and 5FU treatment of HCT116 cells caused a robust phosphorylation of p53 on S15. This modification was inhibitable by wortmannin or caffeine (data not shown), indicating ATM/ATR in the cell nucleus as the relevant kinases. Mitochondrial p53 induced by both stimuli was also S15 phosphorylated, strongly suggesting that this p53 had been translocated from the nucleus to the mitochondria. In contrast, only nonmitochondrial p53 was acetylated at K382 by the p300/CBP histone acetylase at K382. Acetylation of p53 is not important for its DNA-binding activity, but may serve a role in the regulation of gene transcription/repression and in the direction of p53 to nuclear compartments (reviewed by Prives and Manley, 2001). We conclude that mitochondrial and nuclear p53 harbor distinct post-translational modifications, which may be involved in the regulation of the subcellular localization as well as the functions of p53 in the different organelles. Furthermore, mitochondrial p53 levels are proportional to total p53 levels independent of the type of stress response in some cell types, and a fraction of the total p53 is localized at the mitochondria even in the absence of stress.
Materials and methods
Cell lines and chemicals
Human HCT116 adenocarcinoma cells and the p53- and p21-negative derivatives (Waldman et al., 1995; Bunz et al., 1998) were cultured in McCoy's 5A medium supplemented with 10% FCS; tet-inducible LN-Z308-p53wt cells were grown and maintained in DMEM with 4.5 g/l glucose supplemented with 10% of a tet-negative FCS purchased from Clontech (Palo Alto, USA). Both cell lines were cultured as monolayers at 37°C in a humidified 7% CO2 atmosphere. ADR, 5FU, propidium iodide (PI), DAPI, doxycycline and the anti-β-actin monoclonal antibody were from Sigma (St Louis, USA). All chemicals were dissolved in water. The p53 monoclonal antibody DO-1 (Ab-6), the rabbit polyclonal antibody directed against p53 acetylated at K382, the monoclonal p21Waf/Cip1 antibody, the monoclonals against cytochrome c (Ab-2) and caspase-3, and the pancaspase inhibitor z-VAD-fmk were purchased from Calbiochem (San Diego, USA). Monoclonal G59-12 against p53 and the anti-Rb monoclonal were from Transduction Laboratories/PharMingen (San Diego). Rabbit polyclonals against p53 phosphorylated at S15 or S46 were from Cell Signaling Technology (Beverly, MA, USA). The monoclonal anti-cytochrome OX IV antibody was from Molecular Probes (Eugene, USA), and the anti-c-Myc antibody 9E10 from Santa Cruz Biotechnologies (Santa Cruz, CA, USA). All antibodies were diluted as outlined in the figure legends.
Flow cytometry and immunofluorescence
Cells were seeded in six-well dishes to approximately 30% confluency 24 h prior to drug treatment. At the indicated times, the cultures were harvested by trypsinization, combined with cells floating in the medium, washed in PBS, resuspended in 200 μl of 0.9% NaCl, squeezed through a 23.5-gauge needle into 1.8 ml of methanol, and fixed for 30 min at −20°C. Cells were resuspended in PBS supplemented with RNase A (25 μg/ml) at approximately 106 cells per ml, and were stained with PI (25 μg/ml) for 1 h at 4°C. DNA fluorescence was determined in a Beckton Dickinson FACScan (Bedford, USA); data were analysed with the Cell Quest software from Beckton Dickinson. For immunofluorescence, cells were washed in PBS and fixed for 15 min at 37°C in PBS plus 3.7% formaldehyde. The cells were then washed with PBS and permeabilized with 0.2% Triton X-100 for 5 min at RT. Following repeated washes with PBS, cells were incubated with the primary anti-p53 antibody DO-1 (1 : 1000) or appropriate controls for 1 h, washed twice in PBS, incubated with secondary FITC-labeled anti-mouse antibody (Sigma), and finally examined with a Leitz fluorescence microscope.
Preparation of subcellular fractions and immunoblot analysis
Mitochondrial and free cytosolic protein fractions were prepared as reported before (Bogenhagen and Clayton, 1974; Marchenko et al., 2000). In brief, 108 cells were scraped into 5 ml ice-cold TD buffer (135 mM NaCl, 5 mM KCl, 25 mM Tris-HCl pH 7.5), and 0.5 ml of the cell suspension was saved and dissolved in Laemmli buffer (50 mM Tris-HCl pH 6.8, 100 mM DTT, 2% SDS, 20% glycerol) as total cell protein. After the cells were spun at 4°C at 1500 g for 5 min, they were resuspended in 0.8 ml 4°C CaRSB buffer (10 mM NaCl, 1.5 mM CaCl2, 10 mM Tris-HCl pH 7.5, protease inhibitors) and allowed to swell on ice for 10 min. The cell suspension was then Dounce homogenized with 30 strokes, and 0.75 ml of 4°C 2.5 × MS buffer (0.525 M mannitol, 175 mM sucrose, 12.5 mM EDTA, 12.5 mM Tris-HCl pH 7.5) was added to stabilize mitochondria. Nuclei were spun out twice at 4°C, 3000 r.p.m. for 15 min in an Eppendorf fuge. The remaining, nuclei-free supernatant was layered over a sucrose gradient consisting of a layer of 1.5 M sucrose solution (1.5 M sucrose in 10 mM Tris-HCl pH 7.5, 5 mM EDTA, 2 mM DTT, protease inhibitors) and an upper layer of 1 M sucrose solution (1 M sucrose dissolved in same buffer). The gradient was spun at 4°C for 45 min at 80 000 g in a Beckman SW41 rotor. Mitochondria were collected at the gradient interphase, diluted and washed with 4 × volume of 1 × MS buffer (210 mM mannitol, 70 mM sucrose, 5 mM EDTA, 5 mM Tris-HCl pH 7.5), spun down at maximum speed for 15 min in an Eppendorf fuge, and dissolved in Laemmli buffer. The top layer was collected as free protein fraction containing cytosolic, organellar, and leaked nuclear protein. For immunoblot analysis, equal protein quantities or the quantities indicated in the figure legends were subjected to 8–14% SDS–PAGE and transferred to a nitrocellulose membrane (Immobilon-P, Millipore, USA). Signals were detected after overnight exposure of the membranes with the primary antibodies at dilutions indicated in the figure legends, followed by incubation with a peroxidase-conjugated anti-mouse or anti-rabbit antibody (Sigma) diluted at 1 : 3000 and 1 : 1000, respectively, and exposure of the membranes to film for 5 min, unless indicated otherwise. The complexes were visualized with the Renaissance Enhanced Luminol Reagents, as outlined by the manufacturer (NEN, Boston, USA).
Mitochondrial preparations were fixed in 4% formaldehyde/0.05% glutaraldehyde, dissolved in 0.1 M cacodylate buffer (pH 7.4) at RT, and stored overnight at 4°C. Pellets were resuspended in 2% low-melting point agarose at 40°C and solidified on ice. Mitochondria were fixed to the agarose gel with the formaldehyde/glutaraldehyde fixative (see above). After washing the gel with 0.1 M phosphate buffer pH 7.2, small blocks (maximum 2 × 2 × 2 mm3) were cut out and dehydrated by the processive lowering of temperature method using the following ethanol series and temperatures: 30%, 0°C; 50%, −20°C, 70, 90, 100% at −35°C, for 1 h each. Dehydrated gel blocks were infiltrated and embedded with the acrylate resin Lowicryl K4M (Polyscience, Eppelheim, Germany) at −35°C. The resin was UV polymerized for 1 day at −35°C, 1 day at 0°C, and 1 day at RT. Ultrathin sections (70–80 nm) were placed on droplets (30 μl) of the following: glycine (50 mM in PBS); blocking solution; anti-p53 antibody DO-1 or IgG control diluted in blocking solution; blocking solution; sheep anti-mouse antibody coupled to 10 nm colloidal gold (British Biocell, Cardiff, UK); blocking solution; PBS; 2.5% glutaraldehyde in 0.1 M phosphate buffer; PBS; and aqua dest. The blocking solution contained (0.5% cold water fish gelatine, 0.5% BSA, 0.01% Tween-20, all from Sigma) dissolved in PBS. The incubations with the antibodies were carried out overnight at 4°C in a wet chamber. Finally, the sections were dried and stained with uranyl acetate and methylcellulose. All intact mitochondria (n=20 or 30 per cell type) detected at × 68 000 magnification in randomly chosen fields were analysed using morphometric software (Analysis, SIS, Münster, Germany).
Attardi LD, Reczek EE, Cosmas C, Demicco EG, McCurrach ME, Lowe SW and Jacks T . (2000). Genes Dev., 14, 704–718.
Bogenhagen D and Clayton DA . (1974). J. Biol. Chem., 249, 7991–7995.
Bunz F, Dutriaux A, Lengauer C, Waldman T, Zhou S, Brown JP, Sedivy JM, Kinzler KW and Vogelstein B . (1998). Science, 282, 1497–1501.
Bunz F, Fauth C, Speicher MR, Dutriaux A, Sedivy JM, Kinzler KW, Vogelstein B and Lengauer C . (2002). Cancer Res., 62, 1129–1133.
Bunz F, Hwang PM, Torrance C, Waldman T, Zhang Y, Dillehay L, Williams J, Lengauer C, Kinzler KW and Vogelstein B . (1999). J. Clin. Invest., 104, 263–269.
Caelles C, Helmberg A and Karin M . (1994). Nature, 370, 220–223.
Ding HF, Lin YL, McGill G, Juo P, Zhu H, Blenis J, Yuan J and Fisher DE . (2000). J. Biol. Chem., 275, 38905–38911.
Fortin A, Cregan SP, MacLaurin JG, Kushwaha N, Hickman ES, Thompson CS, Hakim A, Albert PR, Cecconi F, Helin K, Park DS and Slack RS . (2001). J. Cell Biol., 155, 207–216.
Gao C and Tsuchida N . (1999). Jpn. J. Cancer Res., 90, 180–187.
Gu W and Roeder RG . (1997). Cell, 90, 595–606.
Hwang PM, Bunz F, Yu J, Rago C, Chan TA, Murphy MP, Kelso GF, Smith RA, Kinzler KW and Vogelstein B . (2001). Nat. Med., 7, 1111–1117.
Iliakis G, Wang Y, Guan J and Wang H . (2003). Oncogene, 22, 5834–5847.
Jallepalli PV, Lengauer C, Vogelstein B and Bunz F . (2003). J. Biol. Chem., 278, 20475–20479.
Johnstone RW, Ruefli AA and Lowe SW . (2002). Cell, 108, 153–164.
Lin Y, Ma W and Benchimol S . (2000). Nat. Genet., 26, 122–127.
Lohr K, Moritz C, Contente A and Dobbelstein M . (2003). J. Biol. Chem., 278, 32507–32516.
Mahyar-Roemer M, Katsen A, Mestres P and Roemer K . (2001). Int. J. Cancer, 94, 615–622.
Mahyar-Roemer M, Kohler H and Roemer K . (2002). BMC Cancer, 2, 27.
Mahyar-Roemer M and Roemer K . (2001). Oncogene, 20, 3387–3398.
Marchenko ND, Zaika A and Moll UM . (2000). J. Biol. Chem., 275, 16202–16212.
Mihara M, Erster S, Zaika A, Petrenko O, Chittenden T, Pancoska P and Moll UM . (2003). Mol. Cell, 11, 577–590.
Miyashita T and Reed JC . (1995). Cell, 80, 293–299.
Moroni MC, Hickman ES, Denchi EL, Caprara G, Colli E, Cecconi F, Muller H and Helin K . (2001). Nat. Cell Biol., 3, 552–558.
Nakano K and Vousden KH . (2001). Mol. Cell, 7, 683–694.
Oda E, Ohki R, Murasawa H, Nemoto J, Shibue T, Yamashita T, Tokino T, Taniguchi T and Tanaka N . (2000a). Science, 288, 1053–1058.
Oda K, Arakawa H, Tanaka T, Matsuda K, Tanikawa C, Mori T, Nishimori H, Tamai K, Tokino T, Nakamura Y and Taya Y . (2000b). Cell, 102, 849–862.
Owen-Schaub LB, Zhang W, Cusack JC, Angelo LS, Santee SM, Fujiwara T, Roth JA, Deisseroth AB, Zhang WW, Kruzel E and Radinsky R . (1995). Mol. Cell. Biol., 15, 3032–3040.
Prives C and Manley JL . (2001). Cell, 107, 815–818.
Robles AI, Bemmels NA, Foraker AB and Harris CC . (2001). Cancer Res., 61, 6660–6664.
Sansome C, Zaika A, Marchenko ND and Moll UM . (2001). FEBS Lett., 488, 110–115.
Schmitt CA, Fridman JS, Yang M, Lee S, Baranov E, Hoffman RM and Lowe SW . (2002). Cell, 109, 335–346.
Schuler M and Green DR . (2001). Biochem. Soc. Trans., 29, 684–688.
Scorrano L and Korsmeyer SJ . (2003). Biochem. Biophys. Res. Commun., 304, 437–444.
Van Meir EG, Polverini PJ, Chazin VR, Su Huang HJ, de Tribolet N and Cavenee WK . (1994). Nat. Genet., 8, 171–176.
Vousden KH and Lu X . (2002). Nat. Rev. Cancer, 2, 594–604.
Wagner AJ, Kokontis JM and Hay N . (1994). Genes Dev., 8, 2817–2830.
Waldman T, Kinzler KW and Vogelstein B . (1995). Cancer Res., 55, 5187–5190.
Wang L, Wu Q, Qiu P, Mirza A, McGuirk M, Kirschmeier P, Greene, JR, Wang Y, Pickett CB and Liu S . (2001). J. Biol. Chem., 276, 43604–43610.
Wu GS, Burns TF, McDonald III ER, Jiang W, Meng R, Krantz ID, Kao G, Gan DD, Zhou JY, Muschel R, Hamilton SR, Spinner NB, Markowitz S, Wu G and el-Deiry WS . (1997). Nat. Genet., 17, 141–143.
Yan Y, Shay JW, Wright WE and Mumby MC . (1997). J. Biol. Chem., 272, 15220–15226.
Yu J, Wang Z, Kinzler KW, Vogelstein B and Zhang L . (2003). Proc. Natl. Acad. Sci. USA, 100, 1931–1936.
We thank Bert Vogelstein, Johns Hopkins University, Baltimore, USA for the cell lines HCT116, HCT116 p53−/− and HCT116 p21−/−; we are also grateful to Erwin Van Meir, Emory University, Atlanta, USA for the cell line LN-Z308 p53wt. Finally, we thank Gabi Kiefer, Institute of Anatomy, for excellent technical assistance with electron microscopy. This work was supported by the German Research Foundation (DFG) Grant RO 1205/5-1 to KR.
About this article
Cite this article
Mahyar-Roemer, M., Fritzsche, C., Wagner, S. et al. Mitochondrial p53 levels parallel total p53 levels independent of stress response in human colorectal carcinoma and glioblastoma cells. Oncogene 23, 6226–6236 (2004). https://doi.org/10.1038/sj.onc.1207637
- tumor suppressor
Mutant p53 exhibits trivial effects on mitochondrial functions which can be reactivated by ellipticine in lymphoma cells
Cell Death & Differentiation (2008)
The post-translational phosphorylation and acetylation modification profile is not the determining factor in targeting endogenous stress-induced p53 to mitochondria
Cell Death & Differentiation (2005)
The EMBO Journal (2005)