Regulation of the balance between survival, proliferation, and apoptosis on carcinogenic polycyclic aromatic hydrocarbon (PAH) exposure is still poorly understood and more particularly the role of physiologic variables, including intracellular pH (pHi). Although the involvement of the ubiquitous pHi regulator Na+/H+ exchanger isoform 1 (NHE1) in tumorigenesis is well documented, less is known about its role and regulation during apoptosis. Our previous works have shown the primordial role of NHE1 in carcinogenic PAH-induced apoptosis. This alkalinizing transporter was activated by an early CYP1-dependent H2O2 production, subsequently promoting mitochondrial dysfunction leading to apoptosis. The aim of this study was to further elucidate how NHE1 was activated by benzo(a)pyrene (BaP) and what the downstream events were in the context of apoptosis. Our results indicate that the mitogen-activated protein kinase kinase 4/c-Jun NH2-terminal kinase (MKK4/JNK) pathway was a link between BaP-induced H2O2 production and NHE1 activation. This activation, in combination with BaP-induced phosphorylated p53, promoted mitochondrial superoxide anion production, supporting the existence of a common target for NHE1 and p53. Furthermore, we showed that the mitochondrial expression of glycolytic enzyme hexokinase II (HKII) was decreased following a combined action of NHE1 and p53 pathways, thereby enhancing the BaP-induced apoptosis. Taken together, our findings suggest that, on BaP exposure, MKK4/JNK targets NHE1 with consequences on HKII protein, which might thus be a key protein during carcinogenic PAH apoptosis. [Cancer Res 2007;67(4):1696–705]
Polycyclic aromatic hydrocarbons (PAH) are widely spread pollutants to which humans are commonly exposed. Many of these compounds exhibit carcinogenic effects due to their metabolism by cytochromes P450, thereby forming highly reactive metabolites that bind to DNA (1). This genotoxicity is also the major trigger of apoptotic induction in several cellular models (2). However, recent works showed that PAH can lead to deleterious effects apparently following DNA damage–independent processes (3, 4) and that proapoptotic and antiapoptotic signals often occur simultaneously (5). In this context, it seems important to further elucidate which physiologic variables influence the balance between survival and death.
In our previous works, we focused on intracellular H+ homeostasis following PAH treatment. Indeed, intracellular pH (pHi) is a highly regulated cellular variable, whose perturbations by activation or inhibition of transporters or by metabolic changes induce proliferative, differentiation, or apoptotic signals (6). About the balance between survival and death, acidification has been considered as a proapoptotic signal, whereas alkalinization has commonly been associated with proliferative situations (7). Whereas intracellular acidification is well documented, apoptosis-related alkalinization has thus far been the subject of only a few investigations, although its emerging involvement as an early event in the apoptotic cascade could make it a crucial target for cancer therapy or for prevention against chemically induced toxicity (8, 9). The origin of intracellular alkalinization seems to be mainly attributed to the activation of Na+/H+ exchanger isoform 1 (NHE1), as observed on cytokine deprivation or potassium antimonyl tartrate exposure (see ref. 10 for review), and often found to be transient followed or not by an acidification. This biphasic pHi variation is involved in PAH-induced apoptosis in F258 rat liver epithelial cells (3).
We have previously shown that toxic PAHs activate the ubiquitous NHE1 following CYP metabolism and H2O2 production. The induced H+ efflux leads to an early but transient alkalinization, which is followed by a late acidification depending on a mitochondrial F0F1-ATPase reversal (11). The use of cariporide, a specific NHE1 inhibitor, blocks both alkalinization and acidification and allows partial prevention of PAH-related apoptosis. These studies underlined the importance of NHE1 activation for the control of mitochondrial homeostasis during PAH-induced cell death. Moreover, we showed that F0F1-ATPase reversal, leading to acidification, promotes recruitment of multiple apoptotic effectors. Thus, these results strengthen the idea that disturbance of energetic metabolism can modulate pHi and consequently apoptosis.
The present study focuses on the activation of NHE1 by considering two aspects: how NHE1 is activated in PAH-induced apoptosis and what its implications are in this process. Indeed, this transporter has been associated with tumor growth and chemotherapy resistance (7, 12). Moreover, it has been shown that a highly carcinogenic PAH, benzo(a)pyrene (BaP), induced NHE1 expression in HepG2 tumoral cells (13). At this stage, it was important to identify the stimuli involved in NHE1 activation during the apoptotic process elicited by PAH. Finally, activation of NHE1 might seem as a new pathway for modulating cellular response to an apoptotic stimulus. Few targets for this increase of pHi have been identified thus far, among which Bax whose conformation changes on alkalinization, thus allowing its mitochondrial translocation (14, 15). However, in our experimental system, Bax did not seem to be involved (11). The sensitivity of metabolic enzymes for pHi changes has also been described previously (16); in this context, although it was neglected in the past, the energy metabolism may play a primordial role in various transduction pathways, including those leading to cell death (17). About that point, it is worth noting that recent works have suggested that hexokinase II (HKII), a key glycolytic enzyme bound to mitochondria, would be an important regulator of apoptotic processes (18, 19).
Our data provide evidence that c-Jun NH2-terminal kinase (JNK), activated by mitogen-activated protein kinase (MAPK) kinase 4 (MKK4) following its stimulation by BaP-induced H2O2 production, triggers NHE1 activation. The consecutive alkalinization may modify the activity of the mitochondrial complex III, leading to superoxide anion production. About p53, we showed that its activation and translocation on one hand, and NHE1 activation on the other hand, occur independently to each other. However, the fact that inhibiting p53 prevents mitochondrial reactive oxygen species (ROS) production points out the existence of a common mitochondrial target for NHE1 activation and p53 pathway. By using glucose derivative and small interfering RNA (siRNA), we found that HKII might be the “pivot protein” between NHE1 and p53 pathways during PAH-induced apoptosis.
Materials and Methods
Chemicals. BaP, α-naphthoflavone (α-NF), antimycin A, RNase A, propidium iodide, carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone (FCCP), DEVD-AMC (Asp-Glu-Asp-7-amino-4-methylcoumarin), tetrarhodamine isothiocyanate (TRITC)-conjugated anti-rabbit IgG, thiourea, and 5-thioglucose (5ThioG) were purchased from Sigma Chemical Co. (St. Louis, MO). Cyclic pifithrin α (PFT; ref. 20), a transactivation inhibitor of p53, and rabbit polyclonal JNK were purchased from Calbiochem (France Biochem, Meudon, France). Cariporide was a kind gift from Aventis (Frankfurt, Germany). All these products were used as a stock solution in DMSO; final concentration of this vehicle in culture medium was <0.05% (v/v), and control cultures received the same concentration of vehicle as treated cultures. Pyruvate, Hoechst 33342, MitoTracker Red CMXROS M-7512, mouse monoclonal COX-IV antibody, and dihydroethidine (DHE) were purchased from Invitrogen (Carlsbad, CA). Z-Asp-2,6-dichlorobenzoyloxymethyl-ketone and L-JNK inhibitor (JNKi) were from Alexis Co. (Lausanne, Switzerland). Rabbit polyclonal anti–phosphorylated p53 (Ser15), anti-HKII, anti-MKP1, mouse monoclonal anti-β-actin, and anti-HSC70 antibodies were from Santa Cruz Biotechnology (Tebu-bio SA, Le Perray en Yvelynes, France). Rabbit polyclonal SEK1/MKK4, phosphorylated c-Jun (Ser73), c-Jun, phosphorylated SEK1/MKK4 (Thr261), and mouse monoclonal phosphorylated stress-activated protein kinase/JNK (Thr183/Tyr185; G9) antibodies were purchased from Cell Signaling (Ozyme, Saint Quentin en Yvelynes, France). Secondary antibody conjugated to horseradish peroxidase was from DAKO A/S (Glostrup, Denmark).
Cell culture and apoptosis measurement. The F258 rat liver epithelial cell line was cultured as described previously (11). Detection of apoptosis was done using Hoechst 33342 labeling, and caspase activity assay was done as described previously (3). Both experiments were done after 72 h of treatment with 50 nmol/L BaP and after 48 h with higher concentrations.
Transfection and siRNAs. See details in Supplementary Data.
Measurement of steady-state pHi and following an acid load. The pHi of F258 cells cultured on glass coverslips was monitored using the pH-sensitive fluorescent probe carboxy-seminaphtorhodafluor-1 (carboxy-SNARF-1; Invitrogen) in HEPES-buffered solution (see ref. 3 for detailed protocol). Addition and subsequent removal of NH4Cl were used to induce an acid load to activate the pHi regulatory mechanisms.
Western blotting immunoassays. The detailed protocol was described previously (11).
Flow cytometry analysis of apoptosis (sub-G1 population). After a 72-h treatment, quantification of sub-G1 hypoploid apoptotic cells was done as described previously (11).
Immunofluorescence assays. Immunostaining with rabbit anti–phosphorylated p53, anti–phosphorylated JNK, or anti-HKII and MitoTracker Red CMXROS in F258 cells was done as described previously (11).
ROS detection. See details in Supplementary Data.
Total RNA isolation and reverse transcription-PCR assays. See details in Supplementary Data.
Statistical analysis. All data are quoted as mean ± SE of mean along with number of observations, n, corresponding, if not otherwise stated, to the number of separate cultures used. ANOVA followed by Newman-Keuls' test was used to test the effects of BaP. Differences were considered significant at the level of P < 0.05.
BaP-induced NHE1 activation relies on a H2O2-dependent MKK4/JNK pathway. Experiments were first done to identify the intermediate signals between H2O2 and NHE1. Several reports suggested that MAPK could be involved in NHE1 activation elicited by exogenous H2O2 in cardiac cells (21). One of the stress-related MAPKs, JNK, is activated by BaP (22). In Fig. 1A, immunoblotting and immunolocalization of phosphorylated JNK (Thr183 and Tyr185) revealed an increase of JNK phosphorylation, with a nuclear accumulation of this protein following a 24-h treatment of F258 cells with 50 nmol/L BaP. This state of phosphorylation was maintained after 48 h but decreased after 72 h of BaP exposure. JNK activation was associated with phosphorylation of its major substrate, c-Jun, as detected by Western blot (Fig. 1A). Moreover, c-Jun activation was also detected by gene reporter transfection and luciferase activity measurements as soon as 24 h of treatment with 50 nmol/L BaP (Supplementary Fig. S1). The use of the JNKi (1 μmol/L), as well as thiourea [10 mmol/L; to block H2O2 production (3)], prevented both JNK phosphorylation and translocation as well as c-Jun phosphorylation (Fig. 1A).
MKK4 is known to directly activate JNK in a context of ROS-dependent apoptosis (23). We therefore tested its possible involvement upstream JNK activation on BaP exposure. Western blot analysis in Fig. 1A revealed the same phosphorylation pattern as for JNK, with a peak at 24 h of treatment. We further found that MKK4 phosphorylation was sensitive to thiourea pretreatment and, as expected, insensitive to JNKi. To firmly confirm the role of MKK4, we next did experiments using siRNA against MKK4 (Si MKK4) or RNA-negative control (Si Neg). As shown in Fig. 1A, MKK4 expression was markedly reduced in Si MKK4–transfected cells compared with Si Neg–transfected cells; most interestingly, this down-regulation consequently abrogated JNK phosphorylation, pointing to MKK4 as the major activator of JNK following BaP-induced H2O2 production.
In an attempt to elucidate the role of JNK in BaP-induced apoptosis, we also examined the putative protective effect of JNKi and Si MKK4 on BaP toxicity in F258 cells. By estimating the number of apoptotic nuclei using Hoechst 33342 staining and the sub-G1 population (Fig. 1B and C), apoptosis was found to be decreased by ∼40% on JNKi cotreatment; likewise, Si MKK4 decreased the number of apoptotic cells (Fig. 1B). Moreover, the caspase-3 activity measured on BaP treatment was found to be similarly reduced by the presence of JNKi or in Si MKK4–transfected cells compared with vehicle or Si Neg–transfected cells, respectively.
Because NHE1, involved in BaP-induced apoptosis, was also activated by H2O2, we examined the potential role of JNK as an intermediate between the observed NHE1 activation and early ROS production. We thus assessed the effect of JNKi on NHE1-dependent alkalinization by measuring resting pHi in F258 cells treated with 50 nmol/L BaP and 1 μmol/L JNKi. Figure 2A shows that BaP-induced alkalinization was prevented by JNKi, thus suggesting a role for JNK in NHE1 up-regulation following BaP treatment. To confirm this hypothesis, we next measured the rate of pHi recovery following an acid load induced by an NH4+ prepulse in F258 cells treated with 50 nmol/L BaP in the presence or absence of JNKi. As expected from our previous work, the rate of pHi recovery was markedly increased after a 48-h exposure to 50 nmol/L BaP due to NHE1 activation, and JNKi seemed to prevent this phenomenon (Fig. 2A). Estimations of the mean H+ equivalent efflux, JeH (dpHi / dt × βi) were as follows: in the absence of JNKi: control, 0.76 ± 0.06 versus BaP, 1.48 ± 0.36 mequiv/L/min; in the presence of JNKi: control, 0.99 ± 0.20 versus BaP, 1.06 ± 0.08 mequiv/L/min (n = 5). These results clearly show an inhibition of the increase in NHE1-related acid extrusion by JNKi at 48 h. Because JNK phosphorylation was dependent on MKK4 activation, we also tested Si MKK4 on resting pHi. As shown in Fig. 2B, the alkalinization elicited by BaP in Si Neg–transfected cells did not occur in Si MKK4–transfected cells. Altogether, these data indicated that JNK, following MKK4 activation by ROS production, was necessary for NHE1 activation following BaP treatment.
BaP-induced p53 phosphorylation is not involved in NHE1 activation. BaP is a powerful carcinogen, and DNA damage and oxidative stress are often related to p53 activation. We have previously shown that p53 is phosphorylated on Ser15 following a 24-h treatment with 50 nmol/L BaP and then translocated into the nucleus at 48 h (11). To examine whether p53 activation could play a role in NHE1 activation, we used PFT that was previously shown to prevent phosphorylated p53 translocation (11, 20). Resting pHi was then measured in F258 cells treated with both 50 nmol/L BaP and 5 μmol/L PFT for 48 h. As shown in Fig. 2A, alkalinization still occurred under such conditions. Moreover, the use of siRNA directed against p53 (Si p53) confirmed this result because alkalinization was still detected in Si p53–transfected cells compared with cells transfected with the Si Neg (Fig. 2C), whereas p53 expression and phosphorylation were blocked under these conditions (Fig. 2C). These data therefore showed that p53 was not involved in the signaling pathway between H2O2 production and NHE1 activation observed on BaP.
NHE1 activation does not modify BaP-induced p53 phosphorylation and nuclear translocation. To determine the targets of NHE1 activation in BaP-induced apoptosis, we next tested RNA interference against NHE1 (Si NHE1) or the NHE1-specific inhibitor cariporide on p53 activation. In Fig. 3A, Si NHE1 efficiency was controlled both at its mRNA expression level [by reverse transcription-PCR (RT-PCR)] and at its functional level (by measuring pHi by microspectrofluorimetry after an acid load). Our data clearly showed a marked decrease of NHE1 mRNA expression, in the presence or absence of BaP, in Si NHE1–transfected cells compared with Si Neg–transfected counterparts; moreover, NHE1-dependent acid extrusion was completely prevented, in these cells, similarly to what was already observed in the presence of cariporide (3). Besides, after a 48-h treatment with 5 μmol/L BaP, estimation of apoptosis by Hoechst staining and caspase activity measurements revealed a decrease of cell death in Si NHE1–transfected cells; for comparison, data obtained from Si p53–transfected cells have also been plotted. In Fig. 3B, Western blot analysis showed that cariporide (30 μmol/L) did not prevent Ser15-p53 phosphorylation following a 48-h treatment whatever the concentration of BaP tested (50 nmol/L and 5 μmol/L). Furthermore, immunolocalizations revealed that phosphorylated p53 translocation was still maintained in the presence of cariporide (Fig. 3B). Finally, p53 phosphorylation pattern was not modified by Si NHE1 transfection compared with Si Neg, thus confirming results obtained with cariporide. Therefore, NHE1 activation and/or alkalinization do not seem to play any role in either activation or nuclear translocation of p53.
NHE1 activation seems to be involved in mitochondrial ROS production. We have previously shown that cariporide inhibited not only the early BaP-induced alkalinization but also the subsequent mitochondria-dependent acidification (3). Thus, NHE1 activation seems to play a role in mitochondrial homeostasis. Moreover, BaP and other PAHs have been reported to induce mitochondria-related ROS production during carcinogenesis and apoptosis (24, 25). We therefore decided to examine ROS production following treatment of F258 cells with BaP (50 nmol/L) for 48 h by using the superoxide anion–specific probe DHE and by flow cytometry analysis. In Fig. 3C, the shift of the peak observed on BaP treatment revealed an O2− production following 48 h of treatment, which was dependent on CYP metabolism since inhibited by a CYP inhibitor, α-NF (data not shown). The addition of JNKi and cariporide fully blocked this O2− production. Thus, regardless of whether NHE1 was inhibited by cariporide or JNKi, ROS production was prevented. Moreover, to test the involvement of the p53 pathway in this ROS production, we finally tested the effect of PFT. Like cariporide, this compound prevented the O2− generation (Fig. 3C). All these results obtained with chemical inhibitors were confirmed by using Si NHE1–transfected or Si p53–transfected cells. Indeed, the shift of the DHE fluorescent peak observed when exposing Si Neg–transfected cells to 500 nmol/L BaP for 48 h was markedly inhibited when Si NHE1–transfected or Si p53–transfected cells were tested. Therefore, whereas activation of NHE1 and p53 seems to occur independently following BaP treatment, both pathways would modulate ROS production.
To determine the involvement of ROS in the apoptotic cascade, we first assessed the effect of the antioxidant N-acetylcysteine (NAC) on BaP-induced apoptosis. Experiments of Hoechst 33342 staining, caspase activity measurement, and sub-G1 population analysis (Fig. 4A–C) revealed that NAC was capable of reducing BaP toxicity by ∼50%. This protection was related to the inhibition of O2− production (Fig. 4D). The mitochondrial uncoupler FCCP also blocked BaP-induced O2− production (Fig. 4D), indicating that such an event was dependent on the mitochondria. Moreover, as antimycin A (25 μmol/L), a specific inhibitor of mitochondrial complex III, prevented BaP-induced ROS production (Fig. 4D), this complex was supposed to be involved in this effect. It is worth noting that antimycin A was previously shown to reduce BaP-induced cell death (11). On the whole, our data indicated that NHE1 and p53 activation promoted apoptosis by leading to a mitochondrial dysfunction and O2− generation.
Our previous works showed an important mitochondrial dysfunction, with a hyperpolarization (26) and the reversal of the F0F1-ATPase (11). The presently detected mitochondrial ROS production further underlines the fact that BaP exposure disrupts mitochondrial homeostasis. However, the absence of depolarization and cytochrome c release in our cell model led us to consider the possibility that mitochondrial alterations might stem from changes in the energy metabolism (27). Accordingly, we tested a glucose derivative, the 5ThioG, shown to inhibit HKII (27). This enzyme has been shown to interact with the voltage-dependent anion channel (VDAC) and to regulate mitochondrial ROS production (28) as well as apoptosis (17). To maintain ATP supply, pyruvate (110 mg/L) was added, without any effect on BaP-induced apoptosis rate (Fig. 4A and B). As shown in Fig. 4A–C, cotreatment with BaP and 5ThioG (5 mmol/L) resulted in the reduction of both nuclear fragmentation and caspase activity (by about a factor of 3.5). Moreover, cotreatment with 5ThioG strongly inhibited ROS production (Fig. 4D). Altogether, these results suggested that glycolysis and, more particularly, the glycolytic enzyme HKII might be a potential target for BaP.
HKII is an apoptotic target for BaP-induced NHE1 activation. To gain further insight into the possible role of HKII during BaP-induced apoptosis, siRNA directed against HKII (Si HKII B66 and Si HKII B67) was tested; its efficiency was first confirmed in Fig. 5A. After 72 h of treatment with BaP, we observed a potentialization of BaP-induced apoptosis (Fig. 5A) by ∼93.7% for Si HKII B67. Moreover, similar results were obtained by using a HKII VDAC binding domain peptide (HKII-VDAC; 20 μmol/L) after Hoechst staining. This cell-permeant peptide acts as a competitor with HKII for its binding to VDAC. After a 72-h cotreatment with BaP and HKII-VDAC, the percentage of apoptotic cells was increased by 2.0 ± 0.6–fold (n = 3) compared with BaP alone (data not shown). In conclusion, changing HKII expression or displacing it from the VDAC amplified the toxic effect of BaP. These results further emphasized the observation that HKII might be important for toxicity of BaP in our cell model.
The next set of experiments was therefore done to test the effect of BaP on HKII expression. Western blotting on total lysates prepared from BaP-treated (50 nmol/L) cells for 24, 48, and 72 h showed that BaP markedly decreased total HKII expression at 48 and 72 h (Supplementary Fig. S2). Because the energy control by HKII is directly dependent on its binding to mitochondria, we next focused on mitochondrial HKII expression. In Fig. 5B, after a 48-h treatment with 50 nmol/L BaP, immunolocalization experiments indicated that HKII (green) was translocated from the mitochondria to the cytosol, as visualized by the loss of costaining of HKII with MitoTracker (red), a specific mitochondrial fluoroprobe. Such an effect was further confirmed by the decrease in the mitochondrial expression of HKII as evidenced by Western blotting on mitochondrial fractions (Fig. 5C). This decrease was prevented by the P-450 inhibitor α-NF, as well as by PFT or Si p53, to prevent p53 activation; this therefore suggested that p53 could be involved in HKII localization regulation (Fig. 5C). p53 involvement was further confirmed by HKII immunostaining in Si p53–transfected cells; indeed, as shown in Fig. 5B, HKII was maintained at the mitochondrial level in the presence of BaP, whereas it was translocated in Si Neg–transfected cells on BaP. To evaluate any possible role for NHE1 in the process, the effects of cariporide and Si NHE1 were next tested; our data indicated that cariporide as well as Si NHE1 blocked the decrease in the expression level of HKII at the mitochondria following BaP exposure (Fig. 5C) and that Si NHE1 prevented HKII translocation (Fig. 5B). 5ThioG, which inhibited BaP-induced apoptosis, also prevented the decrease of mitochondrial HKII expression (Fig. 5C). pHi measurement after a 48-h treatment proved that 5ThioG inhibited ROS production and HKII translocation without any effect on NHE1 activation because the alkalinizing effect of BaP was maintained [(BaP treated − control cells) in pyruvate-treated cells: ΔpHi = +0.11 ± 0.04; in pyruvate + 5ThioG: ΔpHi = +0.12 ± 0.07; n = 9].
Whereas p53 consensus sequences have been found in HKII promoter (29), how NHE1 could act on HKII expression has not yet been examined. By doing RT-PCR experiments (Fig. 5D), we found that BaP (50 nmol/L, 48 h) induced a down-regulation of HKII mRNA, which could be prevented by cariporide. Moreover, in Si p53–transfected and Si NHE1–transfected cells, HKII expression was not decreased on BaP exposure (5 μmol/L, 48 h), whereas such a decrease occurred in Si Neg–transfected cells. Taken together, these results suggested that BaP reduced total HKII protein levels by inhibiting HKII mRNA expression through a process involving both p53 and NHE1 activation and/or alkalinization.
Our previous works showed the prominent role of NHE1 and pHi during genotoxic PAH-induced apoptosis (3). The aim of the present study was to elucidate (a) how H2O2 production related to PAH metabolism could activate NHE1 exchange and (b) which apoptotic events were sensitive to NHE1 activation. Our results show for the first time that the MKK4/JNK pathway is capable of activating NHE1 during BaP-related apoptosis. Moreover, NHE1 activation plays an essential role in mitochondrial superoxide generation, which originates from the complex III. More precisely, its activation triggers a decrease of the mitochondrial HKII expression, leading to the uncoupling of the mitochondrial energy metabolism and ROS production. Finally, whereas p53 and NHE1 activations seem as two independent phenomena in early PAH-elicited biological effects, our data suggest possible cross-talks at the mitochondrial level, which might explain the similar effects of NHE1 and p53 inhibitions on mitochondria-dependent events elicited on BaP application [i.e., cytosolic acidification and ROS production (Fig. 6)].
JNK activation induced by PAH has been previously described during allergic inflammation (30), apoptosis (31), and tumor promotion (32) and is in general related to the occurrence of oxidative stress (30). Our present results revealed that JNK was phosphorylated (Thr183 and Tyr185) and translocated into the nucleus following the previously detected H2O2 production induced by a 24-h treatment with BaP (3). We further found that JNK phosphorylation was dependent on MKK4 activation by H2O2. When testing a role for MKK4/JNK in BaP-induced NHE1 activation (by using JNKi or Si MKK4), we observed that NHE1 recruitment was inhibited by both Si MKK4 and JNKi. Moreover, in accordance with the contribution of NHE1 to BaP-induced apoptosis, JNKi and Si MKK4 prevented BaP toxicity to about the same extent as for the specific NHE1 inhibitor cariporide. The fact that the effects of Si NHE1 on apoptosis were larger than Si MKK4 or JNKi might suggest that reducing NHE1 expression would alter functions of the protein other than only ion translocation (33). Taken together, our results strongly point to MKK4/JNK pathway as the signaling link between H2O2 generation and NHE1 activation during CYP-dependent metabolism of BaP. At this stage, one major concern remains to be resolved: how can phosphorylated JNK act on NHE1 activity? Indeed, no binding or phosphorylation site for JNK has been thus far identified on the COOH-terminal cytoplasmic domain of NHE1 (33). Actually, it might be possible that, under our experimental conditions, JNK would be acting on the serine/threonine kinase p90RSK to phosphorylate and hence activate NHE1 (34) because p90RSK has also been shown to be a substrate of JNK2 on UVA exposure (35), a situation known to produce ROS (36). However, it is important to note that the modalities of NHE1 activation during apoptosis still remain poorly described and it might be possible that this new NHE1 regulation via JNK would be the signature of a proapoptotic signal.
Our previous work suggested that the role of NHE1 activation in BaP-induced apoptosis was linked to mitochondria because cariporide prevented mitochondria-dependent acidification, whose origin seemed to be the reverse activity of F0F1-ATPase (3, 11). Moreover, the mitochondria-related acidification also requires p53 activation. To gain further insight into the signaling pathways elicited by BaP, we first sought whether p53 could also participate to NHE1 activation, thereby acting on mitochondria. This aspect was important to test because p53 has previously been shown to regulate the expression of various transporters and channels (37, 38). However, both the p53 inhibitor PFT and Si p53 were ineffective on BaP-induced alkalinization, indicating a p53-independent NHE1 regulation. Inversely, an effect of NHE1 targeting by BaP on p53 activation could have occurred, as acidosis has been shown to induce p53 accumulation, thereby eliciting apoptosis (39). This was clearly not the case here because no variation in the pattern of p53 phosphorylation on Ser15 and nuclear translocation was detected in the presence of cariporide or in Si NHE1–transfected cells.
Although NHE1 and p53 activations seemed as two independent events upstream mitochondria, they seemed to ultimately converge to these organelles. We then suspected that mitochondrial targeting might be associated with ROS production. Our experiments based on the use of various ROS-sensitive probes showed that a 48-h treatment with 50 nmol/L BaP resulted in a production of superoxide anion O2− related to the metabolism of this PAH. However, unlike H2O2 production detected following 24 h of treatment, BaP-induced O2− production was also inhibited by FCCP, an uncoupler of mitochondrial activity, thus pointing to the mitochondrial origin of the superoxide generation. Most interestingly, when JNK or NHE1 was inhibited, BaP did not induce any ROS production. Therefore, NHE1 activation and/or resulting alkalinization seemed to act on mitochondrion by increasing ROS production. To identify which mitochondrial electron transport chain (ETC) complexes were involved, several specific inhibitors were used. Among them, antimycin A, an inhibitor of complex III (ubiquinol-cytochrome c oxidoreductase; ref. 40), strongly decreased BaP-induced ROS production, supporting the idea of a major role for complex III in this ROS generation. Moreover, the similar pattern of inhibition of mitochondrial ROS production obtained on cotreatment with BaP and cariporide or PFT and confirmed by using Si NHE1–transfected or Si p53–transfected cells, respectively, showed a convergence of both NHE1 and p53 pathways. These results clearly point to the involvement of NHE1 and p53 in the deregulation of the complex III, especially in situations where a dysfunction of the F0F1-ATPase activity occurs, which leads to a backup in ETC (40). Consequently, in a context in which cytochrome c release was not involved (11), one might suppose that all mitochondrial energy metabolism should be altered.
1HKII is a mitochondria-bound glycolytic enzyme that mediates the phosphorylation of glucose and couples intramitochondrial ATP synthesis to glucose metabolism. It favors aerobic glycolysis, proliferation, and maintenance of a low ROS rate (41). Besides, this enzyme is overexpressed in highly malignant tumors (42). Moreover, a link between glucose supply and the control of apoptosis has been established (17, 18). Finally, its role in the antioxidant defense was shown in a recent work (28). The use of 5ThioG as a glycolysis and HKII inhibitor allowed us to show that HKII was involved in BaP-induced O2− production. Moreover, the decrease of HKII expression (by Si HKII) or the disruption of its binding (by using the HKII-VDAC competitor peptide) potentiated BaP-induced apoptosis, whereas maintaining its mitochondrial localization (by 5ThioG) prevented such a toxicity. Similar results have already been identified for apoptosis elicited by growth factor depletion or DNA damage (27, 43), situations also known to be associated with alkalinization (10). The mechanisms underlying disruption of the mitochondrial binding of the HKII on BaP remain to be discovered and might involve a down-regulation of its expression (43), an accumulation of glucose-6-phosphate (44), a protein that enters into competition with VDAC (44, 45), or an involvement of Akt and glycogen synthase kinase 3β (19, 27, 46). About the former aspect, down-regulation of HKII expression would be triggered because reduction of HKII mRNA was observed following BaP treatment and reversed by p53 and NHE1 silencing. The fact that p53 would be capable of regulating HKII promoter (29) is an important point to consider for future experiments in the light of the convergence of p53 and NHE1 pathways. For example, one might suppose the p53 binding on HKII promoter to be modulated by a NHE1-sensitive regulator or signaling pathway, which remains to be determined. Another important concern to resolve in the future is related to the importance of the NHE1 pathway relatively to the DNA damage–triggered p53 pathway, although, based on our results obtained by using Si NHE1 and Si p53, it seems that, in our cell model, both pathways would be as important; however, as described above, decreasing NHE1 expression may alter different transporter-related functions not necessarily involved in BaP effects. In this context, this open question should be tackled by other means, notably by using DNA repair–deficient cells (e.g., cells in which nucleotide excision repair pathway is hampered; see ref. 47 for review).
The present study shows for the first time that BaP can interfere with mitochondrial HKII expression and promote apoptosis. However, very early works by Adachi et al. (48) suggested that PAH would favor tumor growth by increasing hexokinase activities, a phenomenon that would be primordial for the initiation and promotion of cancers (49). Thus, hexokinase might constitute a major protein in controlling the balance of proliferation and apoptosis following PAH exposure and hence related cancer development.
Finally, we have found a new network of pathways in which NHE1, well known for its involvement in tumor progression, has been identified as activated by PAHs via MKK4/JNK in an apoptotic context. It would then affect HKII expression, acting as a proapoptotic actor following genotoxic agent exposure. The present work represents an important step forward in the understanding of the mechanism involved in the regulation of the balance between proliferation and apoptosis following DNA damage, in which both NHE1 and PAH exhibit an ambivalent action on cell survival versus death pathways.
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
Grant support: Institut National de la Santé et de la Recherche Médicale, Ligue Nationale Contre le Cancer (Morbihan, Côte d'Armor, and Ille et Vilaine Comittees), Région Bretagne, and Egide (Aurora programme).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We thank the microscopy platform and Dr. Dutertre (Centre National de la Recherche Scientifique, UMR 6061, Rennes, France) for helpful advice on immunolocalization captures and analyses and David Gilot for helpful technical advice and fruitful discussion on MAPK pathways.