p53 cannot be induced by hypoxia alone but responds to the hypoxic microenvironment


Solid tumors frequently contain hypoxic subregions due to insufficient blood supply. In these domains, cells can undergo p53-dependent apoptosis. Therefore, hypoxia has been implicated as a physiological stimulus for p53 accumulation and activation. In such an environment, p53 mutant cells exhibit a selective growth advantage. Hypoxic regulation of p53 has been proposed to be hypoxia inducible factor (HIF) dependent; however, controversy remains over whether and to what extent low oxygen (O2) tension by itself enhances p53 protein stability. Here, we examined the p53 response to hypoxia and hypoxia mimetics in several cell lines expressing different HIF-α proteins. Most cells exhibited elevated levels of p53 in response to hypoxia mimetics such as deferoxamine mesylate and CoCl2, regardless of their HIF-α protein expression profile. However, over a range of O2 levels, from 1.5% to less than 0.02%, we failed to observe p53 accumulation or p53 nuclear translocation in any cell lines tested. Only after treatment with a combination of hypoxia and acidosis/nutrient deprivation did some cells exhibit p53 induction. Our results suggest that, although hypoxia induces p53 accumulation in vivo, secondary effects such as acidosis caused by a hypoxic Pasteur effect (instead of low O2 by itself) are necessary for p53 accumulation. Therefore, the expression of HIF-1α and p53 proteins is not coupled during the cellular hypoxia response.


Cells adapt to low oxygen (O2) levels in several ways: metabolism shifts from oxidative phosphorylation to glycolysis with increased glucose and iron uptake, factors stimulating angiogenesis (VEGF) and red blood cell production (erythropoietin) are expressed at increased levels, and growth arrest and/or apoptosis may be induced (Graeber et al., 1996; Schmaltz et al., 1998; Dang and Semenza, 1999; Semenza, 2000; Gardner et al., 2001). The exact molecular pathways leading to cell cycle arrest and apoptosis during hypoxia are not very clear (An et al., 1998; Carmeliet et al., 1998; Giaccia and Kastan, 1998; Schmaltz et al., 1998; Stempien-Otero et al., 1999; Bruick, 2000; Gardner et al., 2001; Seagroves et al., 2001), but an attractive candidate transducing this signal is the tumor suppressor p53 (Graeber et al., 1994, 1996; Kim et al., 1997; Giaccia and Kastan, 1998; Alarcon et al., 1999; Denko et al., 2000; Hammond et al., 2002). In the absence of stress, p53 is expressed at low levels due to its association with the ubiquitin ligase Mdm2, causing rapid degradation via the proteasome (Haupt et al., 1997; Kubbutat et al., 1997). However, numerous stimuli have been reported to cause p53 accumulation (Pluquet and Hainaut, 2001). p53 is stabilized and activated in response to various signals, such as DNA damage, abnormal proliferation signals (e.g. E2F or Myc overexpression), hypoxia, and osmotic stress (Prives, 1998; Pluquet and Hainaut, 2001). Activated p53 functions as a transcription factor to stimulate a variety of genes involved in DNA damage repair, cell cycle arrest, and apoptosis (Sharpless and DePinho, 2002).

Several lines of evidence suggest that hypoxia directly causes p53 protein accumulation. Histological studies show that cells expressing detectable p53 protein frequently reside within hypoxic tumor subdomains (Zhong et al., 1999). Furthermore, in tumors expressing wild-type p53, cells undergoing apoptosis strongly correlate with such hypoxic regions (Graeber et al., 1994, 1996). In direct contrast, tumors expressing mutant p53 exhibit significantly lower levels of apoptosis in hypoxic domains (Graeber et al., 1996; Stempien-Otero et al., 1999). Using in vitro cell culture systems, hypoxia-induced apoptosis has been correlated with p53 stimulation in human umbilical vein endothelial cells and neurons (Stempien-Otero et al., 1999; Zhu et al., 2002). Furthermore, p53−/− mouse embryo fibroblasts (MEFs) appear to be more resistant to hypoxia-induced apoptosis when O2 levels drop below 0.2% and demonstrate a selective growth advantage over wild-type cells under hypoxic conditons both in vitro and in vivo (Graeber et al., 1996). Human HCT116 cells also exhibit p53-dependent apoptosis after hypoxia treatment, probably through its downstream target PUMA (Yu et al., 2003).

Additional evidence suggests that hypoxic p53 accumulation is linked to the central hypoxia transcriptional response pathway – hypoxia inducible factor (HIF). The HIF transcription complex is a dimer composed of α and β subunits (Wang and Semenza, 1995; Salceda et al., 1996). Three α subunits, namely HIF-1α, HIF-2α, and HIF-3α, and two β subunits, ARNT and ARNT2, have been identified (Semenza and Wang, 1992; Wang and Semenza, 1995; Gu et al., 1998; Wiesener et al., 1998; Semenza, 2001). All the three α subunits exhibit different expression patterns and probably carry out distinct physiological functions (Gu et al., 1998; Jain et al., 1998; Tian et al., 1998; Makino et al., 2002; Hu et al., 2003). Levels of ARNT, the primary HIF-β subunit (Salceda et al., 1996; Kallio et al., 1997; Maltepe et al., 2000; Keith et al., 2001), do not change as O2 levels decrease (Salceda and Caro, 1997). In contrast, HIF-α subunits are expressed at very low levels under normoxia and dramatically increase as O2 concentration decreases. This results in nuclear translocation, pairing with the β subunit, and activation of a wide variety of genes involved in metabolism, angiogenesis, and red blood cell production (Semenza, 2002). p53 and HIF-1α have been shown to associate when the expression levels of both proteins are high (An et al., 1998; Ravi et al., 2000). Based on these observations, it has been suggested that interaction with HIF-1α can cause p53 accumulation during hypoxia. DFX and CoCl2, which mimic hypoxia in HIF-α subunit regulation via inhibition of the HIF-prolyl hydroxylases (Epstein et al., 2001), also induced p53 accumulation in one study (An et al., 1998). However, hypoxia did not induce HIF-1α protein in ARNT-deficient Hepa1c4 cells and the accumulation of p53 in response to DFX and CoCl2 was diminished (An et al., 1998). A separate report showed p53 accumulation in wild-type embryonic stem (ES) cells challenged with anoxia/hypoglycemia, but not in HIF-1α-deficient cells (Carmeliet et al., 1998). Collectively, these studies suggest that HIF-1α and p53 can affect each other's expression and/or activity.

Hypoxia is one of the most common metabolic stresses encountered by tumors and such conditions are often associated with apoptosis. As p53 is a tumor suppressor that induces apoptosis in response to multiple stresses, we dissected the signaling pathway leading to p53 accumulation and activation during hypoxia. We were especially interested in possible cross-talk between p53 accumulation and HIF, and whether a correlation between p53 accumulation and specific HIF-α subunits (i.e. HIF-1α or HIF-2α) exists. We evaluated p53 levels during treatment with hypoxia mimetics (DFX and CoCl2), true hypoxia, and anoxia in several cell lines expressing different HIF-α proteins. We found that chemicals which induce HIF-α proteins also induce p53 accumulation, regardless of the presence or level of HIF-1α or HIF-2α. This suggests that elevated p53 expression is independent of HIF-α protein stimulation during chemical treatment. O2 deprivation, even to the extent of anoxia (<0.02%), failed to induce p53 accumulation in these cells when pH and nutrient levels were maintained. Moreover, we observed decreased levels of p53 during severe hypoxia (0.05%) or anoxia (<0.02%) in multiple cell lines. The only time we observed p53 accumulation was when hypoxia treatment was accompanied by acidosis and nutrient deprivation. We conclude, therefore, that although hypoxia can induce p53, low O2 is unlikely to be a direct stimulus of p53 protein accumulation and p53 expression during hypoxia is not directly linked to the accumulation of either HIF-1α protein or HIF-2α protein.


In order to determine whether HIF-1α or HIF-2α proteins exhibit differential effects on p53 stimulation, we treated mouse ES cells with mutations in either the Hif- or Hif- genes and examined their p53 expression after treatment with hypoxia mimetics. We used etoposide, a known DNA damage reagent, as a positive control for a functional cellular p53 pathway. When treated with etoposide, all the three cell lines (wild-type, Hif-−/−, and Hif-−/−) exhibited robust p53 induction (Figure 1a). Both 100 μ M DFX and 100 μ M CoCl2 caused a visible and reproducible p53 induction, albeit to lower levels than found in etoposide-treated cells (Figure 1a). However, there were no obvious differences in p53 accumulation among wild type, Hif-−/−, and Hif-−/− ES cells. Moreover, the obvious increase in p53 levels did not occur until 8–16 h of treatment, while levels of HIF-α proteins increase dramatically within an hour under such conditions (Hu et al., 2003). We used both the phosphorylation-sensitive antibody PAb421 and the phosphorylation-insensitive antibody PAb240 (Maya and Oren, 2000) and obtained identical results. We also examined p53 steady-state levels in ES cells harboring mutations in either Hif- or Hif- gene, and found no correlation between p53 protein expression and the presence or absence of either protein (data not shown). Therefore, neither HIF-α protein appears to affect the p53-kinase pathway. However, based on these results, it was unclear whether HIF-1α or HIF-2α exhibit redundant function for p53 accumulation or if p53 accumulation occurs independent of either HIF-α protein. In vivo, pVHL functions as the HIF ubiquitin ligase (Maxwell et al., 1999). Human RCC4 renal carcinoma cells lack the VHL gene; therefore, HIF-1α and HIF-2α are constitutively stabilized and expression levels are not altered by hypoxia or hypoxia mimetics (Maxwell et al., 1999; Hu et al., 2003). Consequently, we expect no change in p53 levels as a result of hypoxia mimetics if p53 accumulation is dependent on HIF-α protein stabilization. However, untreated RCC4 cells express low p53 levels when HIF-α protein levels are high. Moreover, when treated with DFX, RCC4 cells exhibited p53 accumulation to the same extent as detected in those treated with the DNA-damaging agent etoposide. In contrast, CoCl2 did not affect p53 levels in RCC4 cells (Figure 1b). These results suggest that p53 accumulates in response to hypoxia mimetics such as DFX and CoCl2. However, this accumulation is not coupled to HIF-α protein levels and accumulation of HIF-α protein is not the causal factor leading to p53 stabilization when cells are treated with chemicals mimicking hypoxia.

Figure 1

Hypoxia mimetic chemicals can induce p53 in cultured murine ES cells and RCC4 cells. (a) Gelatin-adapted ES cells were grown to about 50% confluence and treated with 100 μ M CoCl2 and 100 μ M DFX to mimic hypoxia and 50 μ M etoposide to induce DNA damage for the indicated period of time. Cells were then harvested for Western analysis to examine their p53 expression. * indicates nonspecific bands picked up by the p53 antibody PAb421 and serve as loading control. (b) Log-phase RCC4 cells were treated with 100 μ M DFX and 100 μ M CoCl2 to mimic hypoxia treatment and 50 μ M etoposide to induce DNA damage, respectively, for the indicated period of time. Whole-cell lysates were used for Western analysis of p53 expression level. * indicates a nonspecific band serving as a loading control

Hypoxia mimetics DFX and CoCl2 recapitulate hypoxia in terms of their ability to induce HIF-α protein. Beyond this, there are many distinct independent intracellular effects of these two disparate stimuli. For example, it has been reported that hypoxia mimetics can affect p53 phosphorylation differently from hypoxia itself (Achison and Hupp, 2003). Therefore, we considered the possibility that p53 accumulation is dependent on HIF-α protein only during true hypoxia and not upon treatment by chemicals mimicking hypoxia. We challenged Hif-−/− and Hif-−/− ES cells with O2 deprivation to verify that hypoxic stimulation of p53 is coupled to upregulation of either HIF-1α or HIF-2α protein. To our surprise, we failed to observe any upregulation of p53 in cells cultured at 1.5% O2 (Figure 2). We examined HIF-2α protein expression in these experiments as a control for O2 deprivation and observed elevated levels of HIF-2α in wild type as well as Hif-−/− cells as expected (Figure 2). These results further suggest that HIF-α protein expression is not correlated with p53 accumulation. Moreover, our data are apparently inconsistent with the scenario that low O2 levels promote p53 stabilization, and raise the possibility that hypoxia mimetics and hypoxia have different effects on p53 induction.

Figure 2

Low O2 tension does not induce p53 accumulation in cultured ES cells. In all, 50% confluent ES cells were cultured under 21% O2 (N) and 1.5% O2 (H) for 16 h and then harvested for p53 Western analysis. * indicates a nonspecific band serving as a loading control

To rule out that these distinct stimuli confer different effects on p53 expression that is restricted only to murine ES cells, we analysed a variety of common human and murine tumor cell lines frequently used for hypoxia and p53 studies. We tested their response to chemicals that mimic hypoxia or hypoxia itself. In agreement with previous results (An et al., 1998), CoCl2 and DFX increased p53 expression in human HepG2 hepatoma cells, albeit to a lower level than that obtained with UV irradiation (Figure 3a). We then tested p53 expression in cells exposed to a decrease in O2 partial pressure. HepG2 cells were subjected to 21% O2, 1.5% O2, and 0.1% O2 for different lengths of time. Although HIF-1α and HIF-2α proteins were substantially induced by hypoxia, there was no increase in p53 after either short or long periods of hypoxia (Figure 3b and data not shown). On the contrary, there was actually some reduction in p53 expression in cells cultured under hypoxic conditions. This reduction was more obvious in cells cultured at 0.1% O2 and in cells exposed to hypoxia for a prolonged period of time, that is, 24 h (Figure 3b). We also assessed the level of Mdm2, which is a transcriptional target of p53 (Wu et al., 1993; Zauberman et al., 1995). The expression of Mdm2 was reduced to an even greater extent than p53, suggesting that p53 was not transcriptionally active in these hypoxic cells.

Figure 3

Hypoxia-mimetic chemicals but not true hypoxia induce p53 in cultured HepG2 cells. (a) Log-phase HepG2 cells were treated with 100 μ M DFX and 100 μ M CoCl2 for the indicated period of time to mimic hypoxia treatment. Another aliquot of cells prepared at the same time was subjected to 100 J/m2 UV irradiation to induce DNA damage, and harvested after the indicated time after irradiation. Whole-cell lysates were used for Western analysis of p53 expression level. * indicates a nonspecific band picked up by the secondary antibody and serves as a loading control. (b) In all, 50% confluent HepG2 cells were cultured under 21% oxygen, 1.5% oxygen (generated by hypoxic workstation), and 0.1% oxygen (generated by flushing a sealed box with a premixed tank composed of 95% nitrogen, 5% carbon dioxide, and 0.1% oxygen) for the indicated period of time. Cells were then harvested for Western analysis to examine the p53 expression level. * indicates a portion of the Ponsau S staining of the membrane used for Western analysis and used here as a loading control

Severe hypoxia, or anoxia, has been shown to cause p53 accumulation and apoptosis (Graeber et al., 1994; Wenger and Gassmann, 1996; Koumenis et al., 2001; Hammond et al., 2002). Thus, we considered the possibility that O2 levels used in our previous experiments (0.1–1.5%) were not low enough. Therefore, we further decreased O2 tension to determine if p53 accumulation can be induced under such conditions. We employed chemical O2 sensing strips and a quantitative EF5 staining method to ensure that we obtained severe hypoxia. EF5 staining of human HCT116 colon carcinoma cells was used to monitor the average hypoxic levels in each experiment. EF5 staining results suggested that we can achieve an O2 level as low as 0.02–0.05% in our standard experimental setup (Table 1). By using an O2 scavenger (100 mM Na2S2O3 and 200 mM boric acid, pH 8.5), we can achieve even lower O2 levels (Materials and methods, Table 1). Therefore, we quantitatively confirmed that we have achieved the range of O2 levels intended.

Table 1 EF5 staining of HCT116 cells under different O2 levels

Human MCF7 mammary carcinoma cells and HCT116 colon carcinoma cells, previously shown to exhibit p53 accumulation upon hypoxia treatment (Hammond et al., 2002; Yu et al., 2003), were treated with 0.02% O2, 100 μ M DFX, and 50 μ M etoposide, respectively. Instead of observing an increase in p53 levels, we detected a significant decrease in p53 levels in cells treated with extreme O2 deprivation (Figure 4). DNA-damaging agents such as etoposide significantly induced p53 in both cell lines, indicating that the p53-signaling pathway was functional in each (Figure 4). When assaying the levels of two downstream targets of p53, p21 (Wenger and Gassmann, 1996) and Mdm2, we saw a reduction of both proteins only in cells treated with severe hypoxia (Figure 4, lane 2). In direct contrast, both proteins were strongly induced in etoposide-treated MCF7 cells, while p21 levels increased in etoposide-treated HCT116 cells. These results strongly suggest that p53 can be activated by DNA damage but not severe hypoxia (Figure 4). The fact that both 0.02% O2-treated and DFX-treated cells exhibit a strong increase in HIF-1α protein expression without p53 protein accumulation also suggests that the expression of p53 and HIF-1α proteins is uncoupled in these two additional cell lines. These results support the conclusion that low O2 tension is not a direct acute stimulus for p53 accumulation and activation.

Figure 4

0.05% O2 cannot induce p53 accumulation in cultured MCF7 and HCT116 cells. In all, 50% confluent MCF7 and HCT116 cells were placed in HEPES balanced CO2-independent medium and immediately subjected to 16 h of 21% O2 treatment (N) or 0.05% O2 treatment (A). As a control, cells cultured in the same medium were treated with 100 μ M DFX (D) and 50 μ M etoposide (E) for 8 h. Cells were then harvested for Western analysis

To obtain the lowest O2 level possible, we employed a sealed aluminum chamber to culture our cells. Oxygen was purged out of the container by flushing with gas composed of 95% N2 and 5% CO2 several times. Oxygen scavenger was added in a separate dish in the chamber to eliminate any residual O2. Differences between the presence and absence of the O2 scavenger in the aluminum chamber were confirmed by a plutonium electrode. We estimate that the O2 level in this case was well below 0.02%, which is near the threshold of detection by plutonium electrodes. When we tested MCF 7 cells in such low pO2 levels, there was no perceptible increase in p53 accumulation (data not shown).

Under normal conditions, p53 is expressed at very low levels and shuttled out of the nucleus by Mdm2 (Freedman and Levine, 1998; Roth et al., 1998). Upon stimulation, p53 accumulates and translocates to the nucleus to induce target genes such as Mdm2 and p21. Therefore, even if a small amount of p53 translocated from the cytoplasm to the nucleus, it may affect the transcription of its downstream targets. To examine the possibility that increased p53 nuclear translocation occurs in hypoxic cells although total protein levels do not increase, we performed immunofluorescence (IF) experiments to monitor p53 subcellular localization upon hypoxia treatment. We treated HCT116 and MCF7 cells with severe hypoxia (0.05% O2) for 16 h and then examined p53 levels and cellular localization. We observed no p53 nuclear translocation in hypoxia-treated cells. In contrast, such translocation was obvious in etoposide-treated cells (Figure 5). We further tested the p53 localization after severe hypoxia treatment using murine B16F10 melanoma cells which have a very robust p53 response to etoposide treatment. We treated these cells with severe hypoxia (0.05% O2) for 16 h and then examined p53 levels and cellular localization. By Western blot analysis, we observed a slight reduction in total p53 levels upon hypoxia treatment of B16F10 cells and a strong induction of p53 upon etoposide treatment (Figure 6a). Based on IF analysis, we noticed an obvious nuclear translocation of p53 in etoposide-treated cells, but no translocation was observed in 0.05% O2-treated cells (Figure 6b). There was even a slight reduction in p53 levels in hypoxic cells, which is consistent with the Western data. Furthermore, the p53 downstream target p21 was significantly increased in etoposide-treated cells, but not in 0.05% O2-treated cells (Figure 6a). Therefore, based on expression profile, subcellular localization, and activity, there is no evidence for p53 activation in B16F10 cells upon treatment with severe hypoxia. In addition, similar treatment of RCC4 cells failed to result in p53 accumulation (Figure 7a) or nuclear translocation (Figure 7b). Since we observed no p53 nuclear translocation, it is unlikely that p53 is stimulated to act as either a transcriptional activator or a transcriptional repressor after 0.05% O2 treatment in these cells.

Figure 5

0.05% O2 does not induce nuclear translocation of p53 in cultured HCT116 and MCF7 cells. HCT116 and MCF7 cells were plated on a gelatin-treated glass cover slip and grown to 50% confluence. Immediately after changing the culture medium to HEPES balanced CO2-independent medium, cells were treated for 16 h under 21% O2 (N) and 0.05% O2 (A). As a control, cells were treated for 8 h with 50 μ M etoposide (E). At the end of each treatment, cells were used for IF staining

Figure 6

0.05% O2 does not induce nuclear translocation of p53 in cultured B16F10 cells. B16F10 cells were plated in a tissue culture dish containing a gelatin-treated glass cover slip and grown to 50% confluence. Immediately after changing the culture medium to Hepes balanced CO2-independent medium, cells were treated for 16 h under 21% O2 (N) and 0.05% O2 (A). As a control, cells were treated for 8 h with 50 μ M etoposide (E). At the end of each treatment, cells grown in the dish but not on the glass cover slip were harvested for Western analysis for p53 and p21 expression (panel A). * indicates a nonspecific band serving as loading control. Cells grown on glass cover slips were used for IF staining (B)

Figure 7

0.05% O2 does not induce nuclear translocation of p53 in cultured RCC4 cells. Experiments were performed the same way as in Figure 5, except that RCC4 cells were used (N, 16 h 21% O2 treatment; A, 16 h 0.05% O2 treatment; E, 8 h 50 μ M etoposide treatment; * indicates a nonspecific band serving as loading control). Note that the IF signals in 0.05% O2-treated cells are nonspecific and do not overlap with the nucleus

Previous studies suggested that acidosis rather than low O2 itself is the direct cause of cellular apoptosis under hypoxic conditions (Graeber et al., 1996; Stempien-Otero et al., 1999; Koumenis et al., 2001). Therefore, we compared the p53 response to hypoxia using different cell densities to generate acidosis during hypoxia in samples plated at high cellular density but not in those plated at low density. HCT116 cells were plated at high density (3 × 105 cells per 60 mm dish 48 h before treatment) and low density (5 × 104 cells per 60 mm dish 48 h before treatment) and cultured at 0.2% O2 overnight. In order to monitor any potential p53 nuclear translocation, we fractionated the cell lysates into cytoplasmic and nuclear fractions (see Materials and methods section) and performed Western analysis. Of note, dense cultures exhibited increased levels of p53 in the nucleus (Figure 8, compare lane 2 to lane 6) in those samples plated at a density high enough to lower the pH of the medium below 6.9. In fact, this sample exhibited p53 accumulation to levels comparable to the γ-irradiated control samples (Figure 8, compare lane 2 to lane 8). Consistent with our previous results, cells plated at low density with no significant decrease in pH of the medium failed to demonstrate an increase in p53 protein (Figure 8, compare lane 4 to lane 6 and lane 2). We quantitated total p53 levels (nuclear plus cytoplasmic) in each sample and found a 30% increase in p53 in γ-irradiated cells compared to untreated cells, a 15% increase in hypoxia-treated cells at high density, and a 20% decrease in hypoxia-treated cells cultured sparsely. We assayed the nuclear p53 levels in these cells and found that γ-irradiated cells exhibit a 60% increase compare to control cells, densely cultured hypoxic cells exhibit an 32% increase, and sparsely cultured cells exhibit a 14% decrease. We also observed p53 accumulation to similar levels in densely plated cells subjected to 4% O2 treatment, in which the pH of the medium also dropped to below 6.9 (data not shown). These results further support a scenario where hypoxia can trigger p53 accumulation, but not through low O2 levels alone. Since these samples underwent a strong Pasteur effect, not only was there a decrease in pH but also deprivation of glucose and other nutrients. As many biochemical pathways are influenced by these stimuli, the direct cause of p53 accumulation is not clear from our results. It is very possible that multiple factors in various combinations are responsible.

Figure 8

p53 accumulates in HCT116 cells under conditions of hypoxia-induced nutrient depletion and acidity. HCT116 cells were plated at high (H) or low density (L) and cultured at 0.2% O2 or 21% O2 overnight. Cytoplasmic (C) and nuclear (N) fractions were prepared for p53 Western blotting. At the end of the experiment, the medium of the high-density cells treated with 0.2% O2 was below pH 6.9 and the pH of the medium of all other cells was normal. Cells treated with 6 Gy γ-irradiation (G) were harvested after 6 h of treatment. The same blot was probed for β-tubulin (a control for cytoplasmic fractionation) and Ku-80 (a control for nuclear fractionation) to ensure equal loading of each sample

Based on the data presented above, we conclude that low O2 by itself cannot induce cellular p53 accumulation. Furthermore, elevation of either HIF-1α nor HIF-2α protein by themselves does not directly cause p53 accumulation. Therefore, it is very likely that secondary effects of hypoxia such as acidosis and nutrient deprivation are required for p53 induction in tumor cells situated in a hypoxic microenvironment.


Although it has been suggested that p53 accumulates under hypoxia in an HIF-1α-dependent manner, other studies challenge the notion that hypoxia induces p53 or that hypoxic p53 regulation is HIF-dependent. Using ES cells lacking either HIF-1α or ARNT and ARNT-deficient Hepa1c4 cells, Wenger et al. (1998) demonstrated that p53 is not induced or activated by hypoxia treatment, regardless of HIF protein status. Similar to our findings, hypoxia mimetics such as DFX induced p53 in these cells in a HIF-1α protein-independent manner (Wenger et al., 1998). These data argue against the hypothesis that HIF, or a functional cellular hypoxia response pathway, is required for p53 accumulation during low O2. However, anoxia was not tested in these studies. Furthermore, hypoxic growth arrest and apoptotic effects were shown to be carried out via p53-independent mechanisms and p53 accumulation was not consistently observed in primary murine fibroblasts, splenic B cells, and human hepatoma cells (Schmaltz et al., 1998; Wenger et al., 1998; Gardner et al., 2001; Goda et al., 2003), suggesting that p53 stimulation is not a universal consequence of hypoxia treatment. Therefore, it is unclear from the current literature whether HIF-α protein levels affect p53 accumulation and, if yes, which HIF-α protein is responsible. Due to the technical difficulties of measuring the degree of hypoxia when O2 levels are very low and the complex environment generated via hypoxic treatment, current literature also remains controversial about whether low O2 can directly cause p53 accumulation, and whether true anoxia (<0.02% O2) rather than hypoxia is required.

In this study, we hoped to determine if either HIF-1α or HIF-2α plays a role in stimulating p53 stabilization and whether hypoxia and/or anoxia can function as a bona fide signal to induce p53 stabilization. In almost all cell lines assayed, we observed that p53 expression is at least modestly upregulated after treatment with DFX and/or CoCl2. However, the induction of p53 did not correlate with the expression of either HIF-1α or HIF-2α. Very convincingly, ES cells with either the Hif- or Hif- gene ablated, responded to these chemicals with respect to p53 in a manner indistinguishable from wild-type ES cells. In human RCC4 renal carcinoma cells, both HIF-α proteins are constitutively stabilized and DFX and CoCl2 do not further increase their expression (Maxwell et al., 1999; Hu et al., 2003). However, p53 expression can still be induced in this cell line after DFX treatment. This strongly suggests that the upregulation of p53 is not due to an increase in HIF-α protein expression. DFX and CoCl2 cause HIF stabilization via disruption of HIF prolyl hydroxylase iron binding; HIF prolyl hydroxylase function is required for rapid HIF-α protein degradation (Bruick and McKnight, 2001; Epstein et al., 2001; Jaakkola et al., 2001; Hon et al., 2002). Since both DFX and CoCl2 interfere with other iron-containing enzymes as well, it is more likely that DFX and CoCl2 confer a stress signal to the cell causing p53 accumulation due to uncharacterized toxic effects (rather than induced HIF-α protein expression). We subsequently observed that low O2 tension, which dramatically elevated HIF-α protein levels, had no impact on p53 expression in all cell lines used. Our results also demonstrate that ‘chemical hypoxia’ is quite different from true hypoxia.

Previously, HIF-1α and p53 have been shown to physically associate with each other and affect each other's stability and activity (An et al., 1998; Ravi et al., 2000). However, the majority of these experiments were done in cells overexpressing both proteins due to transfection, raising the possibility that the interaction only happens when the concentrations of both proteins are atypically high. We believe that endogenous HIF-α and p53 protein levels are not high enough to affect each other in our experiments.

Under the conditions described, hypoxia or anoxia alone did not cause p53 accumulation. We have tried multiple cell lines, different degrees of hypoxia, and even different experimental apparatuses to achieve the intended low O2 tension. The degree of hypoxia was confirmed by HIF-α protein stabilization, EF5 staining, and O2-sensitive chemical or electrode probes (Table 1). However, we have not observed p53 accumulation when using well-buffered medium and relatively sparse cell densities to avoid pH change and nutrient deprivation. We also checked the localization of p53 and expression of Mdm2 and p21, two p53 transcriptional targets, upon hypoxia treatment. We observed neither p53 nuclear translocation nor increased levels of Mdm2 and p21 expression. Although temporary suppression of Mdm2 expression after stress may serve as a driving force for p53 accumulation at early stages of the p53 response (Sax et al., 2002), we observed reduction in Mdm2 even after prolonged hypoxia treatment (typically 16–24 h and up to 72 h). Therefore, we believe the reduction in Mdm2 expression we observed after hypoxia treatment is an indication of lowered p53 activity (or lowered cellular transcription and translation rates in general), rather than an indication of p53 accumulation. Previous studies also suggested that hypoxia may attenuate the p53 response instead of promoting it, presumably via suppression of phosphorylation on certain serine residues (Achison and Hupp, 2003). Based on lack of nuclear translocation, we think our experimental conditions are unlikely to yield a transcriptional repression function of p53 as suggested (Koumenis et al., 2001) or a transcriptional activation function. Therefore, we propose that p53 accumulation and activation are unlikely to be a direct result of hypoxia, but more likely the result of one or more secondary effects conferred by hypoxia.

In agreement with this scenario, the only time we observed p53 accumulation after low O2 treatment was when we plated HCT116 cells at densities allowing nutrient deprivation and significant decrease in pH. According to these results, we think it is more likely that the acidosis and/or glucose deprivation caused by hypoxic treatment is the direct cause of p53 accumulation and not low O2 itself. Our results are consistent with previous findings that low pH was able to induce p53 accumulation (Ohtsubo et al., 1997; Schmaltz et al., 1998). Hypoxia induces acidosis and accelerated glucose deprivation because of a metabolic shift to glycolysis (Seagroves et al., 2001). This effect of hypoxia depends on the accumulation of HIF-1α (Seagroves et al., 2001; Hu et al., 2003). Therefore, it is conceivable that p53 accumulates in hypoxic cells if the buffer capacity is not sufficient to counter HIF-1α-induced acidosis and glucose deprivation. Differential apoptotic effects of hypoxia rendered in cells with different Hif- gene status in previous studies (Carmeliet et al., 1998) may also reflect such a connection among HIF-1α, acidosis, and p53 accumulation. Such a connection can also explain results from multiple studies, indicating that the hypoxic environment in tumors is responsible for activating the p53 pathway leading to tumor cell apoptosis (Graeber et al., 1994, 1996; Stempien-Otero et al., 1999) and why tumor cells lacking functional p53 actually survive this environment better. Therefore, in vivo p53 accumulation can correlate with HIF-1α protein expression if the acidic and glucose deprivation effects of hypoxia are pronounced. However, rather than directly associate with p53 and cause its stabilization, HIF-1α protein exerts its effect through its downstream targets.

From our study, we conclude that low O2 level by itself is not a bona fide signal for p53 accumulation. Rather, secondary effects of hypoxia such as acidosis and glucose deprivation are responsible for increasing p53 levels.

Materials and methods

Cell culture

Human HepG2 heptoma cells, HCT116 colon carcinoma cells, MCF7 mammary carcinoma cells, RCC4 renal carcinoma cells, and murine B16F10 melanoma cells were obtained from ATCC and cultured in DMEM supplemented with 10% FBS, 1% penicillin/streptomycin (PSN), and 2 mM L-glutamine. Wild-type, Hif-−/−, and Hif-−/− ES cells (kind gifts from Dr Peter Carmeliet) were cultured in DMEM supplemented with 15% FBS, 1% PSN, 2 mM L-glutamine, 0.14 mM β-ME, and (leukemia inhibitory factor (LIF)). To reduce acidification under hypoxic conditions, 20 mM HEPES was added to the media and cell density was kept below 70% confluency throughout all experiments. For experiments at 0.05% O2 and lower, cells were plated on permanox dishes (Nunc) or glass dishes to minimize O2 released from plastic, and put into a CO2-independent medium balanced with 25 mM HEPES immediately before treatment.

Western blot assays

Whole-cell lysates were prepared for Western blot analysis. For each sample, cells were scraped and pelleted. After one wash with PBS, cells were lysed directly into SDS lysis buffer (10 mM Tris, pH 7.4, 2 mM EDTA, 1% SDS) and boiled immediately at 95°C for 10 min. For subcellular fractionation, cells were first lysed in a hypotonic buffer to extract cytoplasm and nuclei were subsequently lysed in a hypertonic buffer containing NP-40 to extract nuclear protein. Protein concentration was determined using standard BCA methods (Pierce) and equal amounts of total protein were loaded for gel electrophoresis.

For detecting human p53 protein, Clone DO-1 (Ab-6, Oncogene) was used. For detecting murine p53 protein, PAb421 (Ab-1, Oncogene), which recognizes both murine and human p53, was used. For detecting human Mdm2 protein, Clone IF-2 (Ab-1, Oncogene) was used. Anti-human HIF-1α antibody was purchased from Transduction Laboratories (H72320). Anti-HIF-2α antibody was purchased from Novus. Anti-p21 antibody was obtained from Santa-Cruz. Quantitation was performed by digital means using β-tubulin to normalize sample signals.

IF assays

Cells were grown on gelatin-coated glass cover slips. After fixation in 1% paraformaldehyde at room temperature for 10–15 min, cells were treated with 0.1% Triton X-100/PBS on ice for 10 min and then blocked and stained with appropriate antibody and mounted on Dapi-containing mounting medium (Vector Laboratories). Murine p53 was detected using CM-1/5 (NovoCastra Laboratories) and FITC-conjugated anti-rabbit IgG secondary antibody (Vector Laboratory). Human p53 was detected using DO-1 (Oncogene) for HCT116 cells and MCF7 cells and PAb421 for RCC4 cells (Oncogene) and FITC-conjugated anti-mouse IgG secondary antibody (Vector Laboratories). Different antibody and cell lines used could account for the different fluorescence signal intensity achieved.

Hypoxia treatment

1.5% O2 was achieved using an In Vivo hypoxia workstation (Ruskinn Technologies, Leeds, UK). 0.1% oxygen was achieved by flushing a sealed box containing the cultured cells with premixed gas composed of 95% nitrogen, 5% CO2, and 0.1% O2. 0.05% O2 was obtained by flushing a sealed box containing the cultured cells with premixed gas composed of 95% N2, 5% CO2, and 0% O2 or using 100% N2 combined with culturing cells in CO2-independent medium. 0.02% O2 was achieved by putting culture dishes in a sealed aluminum container and multiple gas exchanges using O2-free N2 to achieve the desired O2 level. For all experiments, one or more of the following methods was used to monitor the actual O2 level: via EF5 staining and plotting a reference curve, or O2-sensing BBL GasPack strips (Becton Dickinson) and/or a polarographic O2 sensor. For experiments to induce different levels of acidosis via hypoxia treatment, HCT116 cells were plated either at 3 × 105 cells per 60 mm dish (high density) or 5 × 104 cells per 60 mm dish (low density) 48 h before each experiment. Cells were then changed into CO2 independent medium and buffered with 25 mM HEPES. Immediately after removing cells from the chamber, the pH was measured by carefully decanting the medium into a tube which was just larger than the Corning pH electrode.

EF5 staining

This method provided a measure of the ‘average’ intracellular pO2 (Koch, 2002). We used HCT116 as a standard cell line. HCT116 cultured in an aluminum chamber was stained with EF5. The level of EF5 staining in these HCT116 cells was correlated with absolute chamber pO2 measured by a plutonium electrode to generate a standard curve. O2 levels in other experimental apparatuses were determined by culturing HCT116 in these systems in the presence of EF5, and the EF5 staining of HCT116 in these conditions was plotted on a standard curve to obtain the actual pO2. For each EF5 staining, 0.03 mM EF5 was added to the culture medium immediately before the experiment and was present for the entire treatment. At the end of each treatment, cells were trypsinized and fixed in 4% paraformaldehyde and then stained with Cy5-conjugated EF5 antibody (Elk3–51). Flow cytometry analysis was then performed to measure the degree of hypoxia achieved quantitatively.


  1. Achison M and Hupp TR . (2003). Oncogene, 22, 3431–3440.

  2. Alarcon R, Koumenis C, Geyer RK, Maki CG and Giaccia AJ . (1999). Cancer Res., 59, 6046–6051.

  3. An WG, Kanekal M, Simon MC, Maltepe E, Blagosklonny MV and Neckers LM . (1998). Nature, 392, 405–408.

  4. Bruick RK . (2000). Proc. Natl. Acad. Sci. USA, 97, 9082–9087.

  5. Bruick RK and McKnight SL . (2001). Science, 294, 1337–1340.

  6. Carmeliet P, Dor Y, Herbert JM, Fukumura D, Brusselmans K, Dewerchin M, Neeman M, Bono F, Abramovitch R, Maxwell P, Koch CJ, Ratcliffe P, Moons L, Jain RK, Collen D, Keshert E and Keshet E . (1998). Nature, 394, 485–490.

  7. Dang CV and Semenza GL . (1999). Trends Biochem. Sci., 24, 68–72.

  8. Denko NC, Green SL, Edwards D and Giaccia AJ . (2000). Exp. Cell Res., 258, 82–91.

  9. Epstein AC, Gleadle JM, McNeill LA, Hewitson KS, O'Rourke J, Mole DR, Mukherji M, Metzen E, Wilson MI, Dhanda A, Tian YM, Masson N, Hamilton DL, Jaakkola P, Barstead R, Hodgkin J, Maxwell PH, Pugh CW, Schofield CJ and Ratcliffe PJ . (2001). Cell, 107, 43–54.

  10. Freedman DA and Levine AJ . (1998). Mol. Cell. Biol., 18, 7288–7293.

  11. Gardner LB, Li Q, Park MS, Flanagan WM, Semenza GL and Dang CV . (2001). J. Biol. Chem., 276, 7919–7926.

  12. Giaccia AJ and Kastan MB . (1998). Genes Dev., 12, 2973–2983.

  13. Goda N, Ryan HE, Khadivi B, McNulty W, Rickert RC and Johnson RS . (2003). Mol. Cell. Biol., 23, 359–369.

  14. Graeber TG, Osmanian C, Jacks T, Housman DE, Koch CJ, Lowe SW and Giaccia AJ . (1996). Nature, 379, 88–91.

  15. Graeber TG, Peterson JF, Tsai M, Monica K, Fornace Jr AJ and Giaccia AJ . (1994). Mol. Cell. Biol., 14, 6264–6277.

  16. Gu YZ, Moran SM, Hogenesch JB, Wartman L and Bradfield CA . (1998). Gene Expr., 7, 205–213.

  17. Hammond EM, Denko NC, Dorie MJ, Abraham RT and Giaccia AJ . (2002). Mol. Cell. Biol., 22, 1834–1843.

  18. Haupt Y, Maya R, Kazaz A and Oren M . (1997). Nature, 387, 296–299.

  19. Hon WC, Wilson MI, Harlos K, Claridge TD, Schofield CJ, Pugh CW, Maxwell PH, Ratcliffe PJ, Stuart DI and Jones EY . (2002). Nature, 417, 975–978.

  20. Hu CJ, Wang LY, Chodosh LA, Keith B and Simon MC . (2003). Mol. Cell. Biol., 23, 9361–9374.

  21. Jaakkola P, Mole DR, Tian YM, Wilson MI, Gielbert J, Gaskell SJ, Kriegsheim A, Hebestreit HF, Mukherji M, Schofield CJ, Maxwell PH, Pugh CW and Ratcliffe PJ . (2001). Science, 292, 468–472.

  22. Jain S, Maltepe E, Lu MM, Simon C and Bradfield CA . (1998). Mech. Dev., 73, 117–123.

  23. Kallio PJ, Pongratz I, Gradin K, McGuire J and Poellinger L . (1997). Proc. Natl. Acad. Sci. USA, 94, 5667–5672.

  24. Keith B, Adelman DM and Simon MC . (2001). Proc. Natl. Acad. Sci. USA, 98, 6692–6697.

  25. Kim CY, Tsai MH, Osmanian C, Graeber TG, Lee JE, Giffard RG, DiPaolo JA, Peehl DM and Giaccia AJ . (1997). Cancer Res., 57, 4200–4204.

  26. Koch CJ . (2002). Methods Enzymol., 352, 3–31.

  27. Koumenis C, Alarcon R, Hammond E, Sutphin P, Hoffman W, Murphy M, Derr J, Taya Y, Lowe SW, Kastan M and Giaccia A . (2001). Mol. Cell. Biol., 21, 1297–1310.

  28. Kubbutat MH, Jones SN and Vousden KH . (1997). Nature, 387, 299–303.

  29. Makino Y, Kanopka A, Wilson WJ, Tanaka H and Poellinger L . (2002). J. Biol. Chem., 277, 32405–32408.

  30. Maltepe E, Keith B, Arsham AM, Brorson JR and Simon MC . (2000). Biochem. Biophys. Res. Commun., 273, 231–238.

  31. Maxwell PH, Wiesener MS, Chang GW, Clifford SC, Vaux EC, Cockman ME, Wykoff CC, Pugh CW, Maher ER and Ratcliffe PJ . (1999). Nature, 399, 271–275.

  32. Maya R and Oren M . (2000). Oncogene, 19, 3213–3215.

  33. Ohtsubo T, Wang X, Takahashi A, Ohnishi K, Saito H, Song CW and Ohnishi T . (1997). Cancer Res., 57, 3910–3913.

  34. Pluquet O and Hainaut P . (2001). Cancer Lett., 174, 1–15.

  35. Prives C . (1998). Cell, 95, 5–8.

  36. Ravi R, Mookerjee B, Bhujwalla ZM, Sutter CH, Artemov D, Zeng Q, Dillehay LE, Madan A, Semenza GL and Bedi A . (2000). Genes Dev., 14, 34–44.

  37. Roth J, Dobbelstein M, Freedman DA, Shenk T and Levine AJ . (1998). EMBO J., 17, 554–564.

  38. Salceda S, Beck I and Caro J . (1996). Arch. Biochem. Biophys., 334, 389–394.

  39. Salceda S and Caro J . (1997). J. Biol. Chem., 272, 22642–22647.

  40. Sax JK, Dash BC, Hong R, Dicker DT and El-Deiry WS . (2002). Cell Cycle, 1, 90–96.

  41. Schmaltz C, Hardenbergh PH, Wells A and Fisher DE . (1998). Mol. Cell. Biol., 18, 2845–2854.

  42. Seagroves TN, Ryan HE, Lu H, Wouters BG, Knapp M, Thibault P, Laderoute K and Johnson RS . (2001). Mol. Cell. Biol., 21, 3436–3444.

  43. Semenza GL . (2000). J. Appl. Physiol., 88, 1474–1480.

  44. Semenza GL . (2001). Trends Mol. Med., 7, 345–350.

  45. Semenza GL . (2002). Intern. Med., 41, 79–83.

  46. Semenza GL and Wang GL . (1992). Mol. Cell. Biol., 12, 5447–5454.

  47. Sharpless NE and DePinho RA . (2002). Cell, 110, 9–12.

  48. Stempien-Otero A, Karsan A, Cornejo CJ, Xiang H, Eunson T, Morrison RS, Kay M, Winn R and Harlan J . (1999). J. Biol. Chem., 274, 8039–8045.

  49. Tian H, Hammer RE, Matsumoto AM, Russell DW and McKnight SL . (1998). Genes Dev., 12, 3320–3324.

  50. Wang GL and Semenza GL . (1995). J. Biol. Chem., 270, 1230–1237.

  51. Wenger RH, Camenisch G, Desbaillets I, Chilov D and Gassmann M . (1998). Cancer Res., 58, 5678–5680.

  52. Wenger RH and Gassmann M . (1996). Nature, 380, 100.

  53. Wiesener MS, Turley H, Allen WE, Willam C, Eckardt KU, Talks KL, Wood SM, Gatter KC, Harris AL, Pugh CW, Ratcliffe PJ and Maxwell PH . (1998). Blood, 92, 2260–2268.

  54. Wu X, Bayle JH, Olson D and Levine AJ . (1993). Genes Dev., 7, 1126–1132.

  55. Yu J, Wang Z, Kinzler KW, Vogelstein B and Zhang L . (2003). Proc. Natl. Acad. Sci. USA, 100, 1931–1936.

  56. Zauberman A, Flusberg D, Haupt Y, Barak Y and Oren M . (1995). Nucleic Acids Res., 23, 2584–2592.

  57. Zhong H, De Marzo AM, Laughner E, Lim M, Hilton DA, Zagzag D, Buechler P, Isaacs WB, Semenza GL and Simons JW . (1999). Cancer Res., 59, 5830–5835.

  58. Zhu Y, Mao XO, Sun Y, Xia Z and Greenberg DA . (2002). J. Biol. Chem., 277, 22909–22914.

Download references


We thank Dr Peter Carmeliet (Center for Transgene Technology and Gene Therapy, Flanders Interuniversity Institute for Biotechnology, KU Leuven, Belgium) for kindly providing the Hif-−/− and Hif-2α−/− ES cells, and Mark Muelner for technical assistance. We thank all members of the Simon laboratory for helpful discussions. This work was supported by the Howard Hughes Medical Institute (MCS), The National Institutes of Health (MCS, Grant #66310; CK, Grant #P01-CA79862), and the Abramson Family Cancer Research Institute (YP, AMA, MCS).

Author information



Corresponding author

Correspondence to Marie Celeste Simon.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Pan, Y., Oprysko, P., Asham, A. et al. p53 cannot be induced by hypoxia alone but responds to the hypoxic microenvironment. Oncogene 23, 4975–4983 (2004). https://doi.org/10.1038/sj.onc.1207657

Download citation


  • hypoxia
  • anoxia
  • p53
  • HIF
  • tumor cells

Further reading


Quick links