|Home | About | Journals | Submit | Contact Us | Français|
The eukaryotic translation initiation factor 4GI (eIF4GI) serves as a central adapter in cap-binding complex assembly. Although eIF4GI has been shown to be sensitive to proteasomal degradation, how the eIF4GI steady-state level is controlled remains unknown. Here, we show that eIF4GI exists in a complex with NAD(P)H quinone-oxydoreductase 1 (NQO1) in cell extracts. Treatment of cells with dicumarol (dicoumarol), a pharmacological inhibitor of NQO1 known to preclude NQO1 binding to its protein partners, provokes eIF4GI degradation by the proteasome. Consistently, the eIF4GI steady-state level also diminishes upon the silencing of NQO1 (by transfection with small interfering RNA), while eIF4GI accumulates upon the overexpression of NQO1 (by transfection with cDNA). We further reveal that treatment of cells with dicumarol frees eIF4GI from mRNA translation initiation complexes due to strong activation of its natural competitor, the translational repressor 4E-BP1. As a consequence of cap-binding complex dissociation and eIF4GI degradation, protein synthesis is dramatically inhibited. Finally, we show that the regulation of eIF4GI stability by the proteasome may be prominent under oxidative stress. Our findings assign NQO1 an original role in the regulation of mRNA translation via the control of eIF4GI stability by the proteasome.
In eukaryotes, eukaryotic translation initiation factor 4G (eIF4G) plays a central role in the recruitment of ribosomes to the mRNA 5′ end and is therefore critical for the regulation of protein synthesis (14). Two homologues of eIF4G, eIF4GI and eIF4GII, have been cloned (15). Although they differ in various respects, both homologues clearly function in translation initiation. The most thoroughly studied of these is eIF4GI, which serves as a scaffolding protein for the assembly of eIF4F, a protein complex composed of eIF4E (the mRNA cap-binding factor) and eIF4A (an ATP-dependent RNA helicase). Thus, via its association with the mRNA cap-binding protein eIF4E and with another translation initiation factor (eIF3) which is bound to the 40S ribosomal subunit, eIF4GI creates a physical link between the mRNA cap structure and the ribosome, thus facilitating cap-dependent translation initiation (25). eIF4GI functions also in cap-independent, internal ribosome entry site (IRES)-mediated translation initiation. For instance, upon picornavirus infection, eIF4G is rapidly attacked by viral proteases. The resulting eIF4GI cleavage products serve to reprogram the cell's translational machinery, as the N-terminal cleavage product inhibits cap-dependent translation of host cell mRNAs by sequestering eIF4E while the C-terminal cleavage product stimulates IRES-mediated translation of viral mRNAs (23). Similarly, apoptotic caspases cleave eIF4G into an N-terminal fragment that blocks cap-dependent translation and a C-terminal fragment that is utilized for IRES-mediated translation of mRNAs encoding proapoptotic proteins (22).
The regulation of eIF4GI cleavage by viral proteases or apoptotic caspases has been extensively studied. Little is known, however, about the regulation of eIF4GI steady-state levels. Yet the eIF4GI amount that exists at a given moment results from the sum of the effects of de novo synthesis and ongoing degradation. Many cellular proteins are physiologically degraded by the proteasome. This has been shown to be true for eIF4GI, as the factor can be degraded by the proteasome in vitro (5) and in living cells (6). However, how eIF4GI targeting for or protection from destruction by the proteasome is regulated remains unknown.
There are two major routes to degradation by the proteasome. In the more conventional route, polyubiquitinated proteins are targeted to the 26S proteasome. Alternatively, a few proteins can be degraded by the 20S proteasome (and sometimes by the 26S proteasome) in a ubiquitin-independent manner (16). Interestingly, it has been shown recently that a few of these proteins (1, 2, 13) can be protected from degradation by the 20S proteasome by binding to the NAD(P)H quinone-oxydoreductase 1 (NQO1). It has been proposed that NQO1 may interact with the 20S proteasome and may consequently block access of target proteins to the 20S degradation core. Because eIF4GI can be degraded in vitro by the 20S proteasome (5) and since it appears that proteasomes can degrade eIF4GI in living cells independently of ubiquitination (6), we asked whether NQO1 could protect eIF4GI from degradation by the proteasome.
Three cell lines were used: human embryonic kidney (HEK-293) cells, simian virus 40 large T antigen-transformed monkey kidney (Cos-7) cells, and immortalized mouse embryo fibroblast (NIH 3T3) cells. Cells were grown as described previously (4).
MG-132, lactacystin, dicumarol (dicoumarol), doxorubicin, H2O2, cycloheximide, and puromycin were from Sigma and were dissolved as recommended by the manufacturer.
A plasmid expressing wild-type NQO1 (pEFIRES-NQO1) was a kind gift of Gad Asher. pcDNA3-HA-eIF4GI (encoding wild-type eIF4GI) and pcDNA3-HA-eIF4GI-N, pcDNA3-HA-eIF4GI-M, and pcDNA3-HA-eIF4GI-C (encoding various eIF4GI deletion mutants) were kind gifts of Nahum Sonenberg (for depictions of the corresponding proteins, see Fig. Fig.1B1B).
Transient (co)transfections of HEK-293, Cos-7, or NIH 3T3 cells with plasmids were carried out with FuGene 6 reagent (Roche) as described previously (4). In all experiments, proteins were expressed for 36 h and cells were either processed immediately (for immunoblotting and immunoprecipitation) or processed following treatment with various compounds (where indicated).
siRNA duplexes (On Target Plus grade) of 21 bp were synthesized by Dharmacon. The targeted sequence for NQO1 siRNA duplexes was 5′-GAACCUCAACUGACAUAUA-3′, while control siRNA was the On Target Plus nontargeting siRNA furnished by Dharmacon. Transfections with siRNA duplexes (concentrations are indicated in the figures) were carried out using DharmaFECT as specified by the manufacturer (Dharmacon) and as described previously (17). Following transfections, expression of the different proteins was visualized by Western blotting (see below).
The rate of protein synthesis was monitored using the recently described nonradioactive method termed surface sensing of translation (SUnSET) (28). Briefly, cells were incubated with a sublethal dose (10 μg ml−1) of puromycin for 20 min prior to being harvested. Cell extracts were then simply processed for SDS-PAGE and Western blotting with antipuromycin antibodies as described below.
Preparation of cell extracts in lysis buffer and coimmunoprecipitation (26) and immunoblotting (8) were carried out as described previously. The antibodies used were as follows: mouse monoclonal antipuromycin (28), rabbit anti-eIF4GI (a kind gift of Nahum Sonenberg), goat anti-NQO1 antibodies C-19 and R-20 (Santa Cruz), mouse monoclonal anti-NQO1 A180 (Santa Cruz), mouse antihemagglutinin (anti-HA) antibody HA-7 (Sigma), rabbit anti-p73 (Cell Signaling), rabbit antibody to poly(ADP-ribose) polymerase (PARP) cleaved at Asp214 (Cell Signaling), rabbit anti-caspase 3 (Cell Signaling), rabbit antibody to caspase 3 cleaved at Asp175 (Cell Signaling), rabbit anti-cyclin D1 (Cell Signaling), mouse monoclonal anti-β-actin (Sigma), mouse monoclonal anti-β-tubulin (Sigma), mouse monoclonal anti-poly(A)-binding protein (anti-PABP) antibody 9E10 (Sigma), rabbit anti-eIF4A (Cell Signaling), rabbit anti-eIF4E (Cell Signaling), and rabbit anti-4E-BP1 (Cell Signaling). Anti-HA immunoprecipitations were carried out using agarose-conjugated mouse monoclonal anti-HA antibodies (Sigma), and eIF4E immunoprecipitations were carried out using agarose-conjugated mouse monoclonal anti-eIF4E antibody (Santa Cruz). Anti-eIF4GI, anti-NQO1 R-20, and anti-NQO1 C-19 antibodies were conjugated to protein A-Sepharose beads prior to incubation with cell extracts. For RNase treatment, beads were washed five times with 0.5 ml of lysis buffer following immunoprecipitation, incubated with 5 μl of an RNase cocktail (a mixture of RNase A and RNase T1; Ambion) for 30 min at 37°C, washed again five times with 0.5 ml of lysis buffer, and processed for Western blotting.
Ornithine decarboxylase (ODC) activity was measured as described previously (12). Cells were rinsed twice in ice-cold phosphate-buffered saline, scraped into lysis buffer (250 mM Tris-HCl, pH 7.4, 1 mM EDTA, 1 mM dithiothreitol), and subjected to two freeze-thaw cycles. The lysate was centrifuged at 14,000 rpm for 10 min to remove cellular debris. The protein content was quantified, and equal amounts (100 μg) of total protein were incubated with a mixture of 2.5 μCi of [14C]ornithine (Amersham) and 50 μM pyridoxal 5-phosphate for 1 h at 37°C. Incubations were performed in 96-well microtiter plates. Liberated 14CO2 was trapped in a covering 3MM paper saturated with a solution of barium hydroxide. The 3MM paper was rinsed with acetone, dried, and exposed to X-ray film.
X-ray films were scanned and eIF4GI, β-tubulin, or β-actin bands were quantified using the ImageJ software developed by the National Institutes of Health. Histograms in Fig. 2A and C, 4B and C, 5B and C, and 7 represent eIF4GI amounts normalized to either β-tubulin or β-actin (as indicated in each figure) and expressed as mean percentages of amounts in untreated cells ± standard deviations (SD) (where applicable).
NQO1 has been shown previously to protect four different proteins, p53 and p73 (3), ODC (1), and p33 (13), from degradation by the proteasome. One important feature in the molecular mechanism responsible for stabilization by NQO1 is that NQO1 and the protected protein are found in the same complex. Therefore, to test for the possibility that NQO1 can also control eIF4GI stability, we first evaluated the capability of NQO1 to coimmunoprecipitate with eIF4GI and vice versa. NQO1 was revealed by the use of a monoclonal antibody following immunoprecipitation with polyclonal antibodies that specifically recognize eIF4GI (Fig. (Fig.1A,1A, lanes 1 and 4). This interaction was specific, as NQO1 was not detected when a cell extract was incubated with agarose beads alone (Fig. (Fig.1A,1A, lane 2), and the signal did not emanate from cross-reaction with immunoglobulins, as no band was detected when agarose-conjugated immunoglobulins were loaded onto the SDS-PAGE gel (Fig. (Fig.1A,1A, lane 3). In addition, in a reverse experiment, eIF4GI was detected following coimmunoprecipitation with two distinct polyclonal NQO1 immunoglobulins (Fig. (Fig.1A,1A, lanes 5 and 6). We then wanted to decipher further the region of eIF4GI that interacted with NQO1. Cells were transfected with three HA-tagged eIF4GI deletion fragments (26) (Fig. (Fig.1B):1B): HA-N (the N-terminal third of eIF4GI, which contains the PABP- and eIF4E-binding sites), HA-M (the middle portion, which contains one eIF4A-binding site and the eIF3-binding sites), and HA-C (the C-terminal third, which contains the Mnk1-binding site and the other eIF4A-binding sites). All full-length proteins or deletion fragments were efficiently expressed (Fig. (Fig.1C,1C, left). However, in contrast to endogenous eIF4GI protein (see above) or to full-length HA-tagged eIF4GI introduced by transfection, which was used as a positive control (Fig. (Fig.1C,1C, right, lane 1), none of the HA-tagged eIF4GI fragments were capable of interacting with NQO1 under the conditions tested (Fig. (Fig.1C,1C, right, lanes 2 to 4). These data may indicate that none of the eIF4GI deletion mutants are localized properly following transfection and that the mutants are therefore unable to encounter NQO1. An alternative explanation may be (i) that NQO1 creates multiple contacts with different parts of eIF4GI and that a combination of multiple contacts is necessary for stable interaction or (ii) that the conformation of full-length eIF4GI protein (which can be influenced by its binding partners) is important for interaction with NQO1.
Because eIF4GI is an RNA-binding protein (24) which interacts with different proteins, we asked whether eIF4GI interaction with NQO1 could be dependent on RNA and whether the eIF4GI-NQO1 complex could contain other proteins. Incubation of eIF4GI immunoprecipitates with an RNase A-RNase T1 mixture did not abolish eIF4GI-NQO1 interaction (Fig. (Fig.1D,1D, compare lane 1 to lane 2). Also, incubation of NQO1 immunoprecipitates with the same RNase mixture did not abolish coimmunoprecipitation with eIF4GI or its partner eIF4E or eIF4A, while in contrast, coimmunoprecipitation of PABP with NQO1 was prevented by RNase treatment (Fig. (Fig.1D,1D, lanes 3 and 4). All these data indicate that the interaction between eIF4GI and NQO1 is independent of mRNA and suggest that NQO1 exists in a complex that contains eIF4GI and at least two of its protein partners, eIF4E and eIF4A.
The protection from proteasomal degradation exerted by NQO1 can be relieved by pharmacological inhibition of the enzyme. It has been shown that upon treatment of cells with dicumarol, the proteins p53 (and p73), ODC, and p33 (see above) are targeted for destruction by the proteasome. The molecular mechanism that underlies the dicumarol effect of targeting proteins for degradation is not completely understood. It is believed that dicumarol precludes NQO1 binding to its partners because the compound competes with NADH (or NADPH) for a common binding site on NQO1, while NADH (or NADPH) association with NQO1 appears to be critical for NQO1 binding to its partners and consequent protection from destruction by the proteasome. As expected from the results of the binding experiments described above, kinetic analyses with 300 μM dicumarol showed that a decrease in the eIF4GI steady-state level became visible at 6 h and was maximal after exposure for 10 h and that the carrier (NaOH) had no effect (Fig. (Fig.2A).2A). In this experiment, the amount of β-tubulin also diminished and could therefore not be used as a loading control. This finding is consistent with earlier data showing that dicumarol exerts NQO1-independent side effects which impinge directly upon β-tubulin function (20). A Western blot analysis using antibodies specific to β-actin, however, revealed that β-actin could be used as a loading control when cells were treated with 300 μM dicumarol (Fig. (Fig.2A).2A). As a positive control in this experiment, we monitored ODC activity (Fig. (Fig.2B).2B). Treatment of cells with 300 μM dicumarol significantly inhibited ODC activity, an effect which was shown to be correlated to a decrease in the ODC amount due to the relief of NQO1-mediated protection of ODC from degradation by the proteasome. An experiment with increasing concentrations of dicumarol then showed that induction of eIF4GI degradation by dicumarol was dose dependent, with a maximal effect at 400 μM (Fig. (Fig.2C2C).
One probable downstream effect of dicumarol treatment and the consequent degradation of eIF4GI (and 4E-BP1 activation [see below]) may be the inhibition of protein synthesis. To address this point, we monitored the rate of peptide chain elongation by a new, elegant, nonradioactive method termed SUnSET (28). The data revealed that treatment of cells with dicumarol dramatically and rapidly inhibited protein synthesis (Fig. (Fig.2D).2D). One could therefore argue that the decrease in the eIF4GI steady-state level we observed in dicumarol-treated cells may result from a combination of degradation of the existing protein and inhibition of neosynthesis. To appreciate the involvement of the inhibition of protein synthesis in eIF4GI decrease, we evaluated the eIF4GI half-life upon cycloheximide treatment. The data we obtained clearly showed that, compared to a short-lived protein (cyclin D1) which was used as a positive control for cycloheximide treatment, eIF4GI had a long half-life (much longer than 10 h) (Fig. (Fig.2E),2E), thus indicating that eIF4GI decrease upon dicumarol treatment could not be explained solely by the inhibition of protein synthesis. Instead, protein degradation appears to play a major role in the diminution of the eIF4GI amount upon dicumarol treatment.
One could also argue that upon dicumarol treatment, eIF4GI is cleaved by caspase activation instead of the proteasome. This argument is supported by the fact that dicumarol is known to provoke intracellular accumulation of reactive oxygen species, which in turn may trigger apoptosis (18). Furthermore, treatment with dicumarol generated the apparition of eIF4GI fragments that could be interpreted to be fragments obtained following caspase activation (Fig. (Fig.3A).3A). However, we obtained two sets of indirect evidence which tended to indicate that eIF4GI degradation provoked by dicumarol treatment was not due to caspase activation. First, eIF4GI products generated by the incubation of cells with dicumarol were quite different from those generated by treatment with doxorubicin used at a dose which induces apoptosis (Fig. (Fig.3B).3B). Second, under the conditions tested, doxorubicin but not dicumarol induced apoptosis, as attested by the apparition of caspase 3 cleavage fragments solely in cells incubated with doxorubicin (Fig. (Fig.3B).3B). To argue more directly in favor of a role of the proteasome in dicumarol-mediated degradation of eIF4GI, we then used MG-132 and lactacystin, two pharmacological compounds that specifically inhibit proteasome activity. It was, however, difficult to anticipate the data that could be obtained with such inhibitors, as a priori contradictory, or at least cell type-dependent, results were described in the literature. eIF4GI accumulates upon the treatment of LoVo colon cancer cells with proteasome inhibitors (6), while proteasome inhibitors have no effect on eIF4GI in C2C12 myoblasts (9) or, in sharp contrast, provoke eIF4GI cleavage in Jurkat T cells (21). Our data obtained using NIH 3T3 fibroblasts revealed that MG-132 and lactacystin exhibit apparently opposing effects on the steady-state amount of full-length eIF4GI. eIF4GI was cleaved in MG-132-treated cells (Fig. (Fig.4A,4A, compare lane 2 to lane 1), while the protein accumulated in lactacystin-treated cells (compare lane 4 to lane 3). This apparent discrepancy could be explained by the fact that MG-132, but not lactacystin, induced NIH 3T3 cell apoptosis, as attested by the appearance of cleaved PARP and cleaved caspase 3 solely in cells exposed to MG-132 (Fig. (Fig.4A).4A). To avoid complex data interpretation due to the dual effects of MG-132 (inhibition of the proteasome but induction of apoptotic proteases), we decided to use lactacystin in the rest of the study. To test for the involvement of the proteasome in dicumarol-mediated eIF4GI degradation, kinetic studies similar to those described in the legends to Fig. Fig.2A2A and and3B3B were performed. eIF4GI (and p73, which was used as a positive control) gradually accumulated upon treatment of cells with lactacystin alone for 2, 6, and 10 h (Fig. (Fig.4B),4B), and pretreatment with lactacystin prevented dicumarol-mediated degradation of eIF4GI (Fig. (Fig.4C4C).
Taken together, these data indicate that treatment of cells with dicumarol provokes eIF4GI degradation by the proteasome.
Dicumarol has been shown to exert intracellular effects that can be independent of its target NQO1, as is the case for its off-target effect on β-tubulin (see above) (20). To ensure that NQO1 per se is involved in the stability of eIF4GI, we therefore decided to act specifically on the endogenous NQO1 protein amount by using a previously described siRNA targeted against NQO1 (2). We first searched for the concentration of NQO1 siRNA that was efficient in NIH 3T3 fibroblasts. Two concentrations (10 and 50 nM) were satisfactory 48 h following transfection, as attested by profound NQO1 silencing and consequent downregulation of p73, which was used as a positive control (Fig. (Fig.5A).5A). A kinetic study with 50 nM siRNA then showed that NQO1 silencing was apparent at 24 h posttransfection, persisted for 48 and 72 h, and resulted in a twofold decrease in the eIF4GI amount (Fig. (Fig.4B).4B). In these experiments, however, the pattern of eIF4GI decrease did not perfectly follow that of NQO1 silencing. One possible explanation for this apparent discrepancy is that several pools of eIF4GI (i.e., in complexes with different partners) exist in the cell and that NQO1 regulation may concern only a subset of these pools. The impact of NQO1 overexpression on the eIF4GI amount was tested. In sharp contrast to NQO1 silencing, NQO1 overexpression resulted in intracellular accumulation of eIF4GI (Fig. (Fig.5C).5C). Finally, NQO1 overexpression diminished eIF4GI sensitivity to dicumarol, as attested by lower-level induction of cleavage fragments in cells overexpressing NQO1 than in control cells (Fig. (Fig.5D).5D). Thus, these data indicate that NQO1 is capable of regulating the eIF4GI intracellular level. When combined with the discoveries that eIF4GI is a direct substrate of the 20S proteasome (5) and that NQO1 protects from degradation by the 20S proteasome, these data indicate a highly probability that NQO1 acts as a regulator of eIF4GI degradation by the 20S proteasome.
In cells treated with dicumarol, protein synthesis was rapidly inhibited (Fig. (Fig.2D).2D). Intriguingly, such inhibition of protein synthesis appeared earlier than eIF4GI degradation did (compare the rate of protein synthesis presented in Fig. Fig.2D2D to that of eIF4GI degradation presented in Fig. Fig.2A,2A, ,3A,3A, or 6A). It was therefore probable that dicumarol induced an additional effect that could explain early effects on protein synthesis. The most relevant partner of eIF4GI in regulating cap-dependent translation is the cap-binding protein eIF4E. We therefore first monitored the eIF4E steady-state level upon dicumarol treatment, but the eIF4E amount remained unchanged (Fig. (Fig.6A).6A). This observation, however, did not rule out the possibility that eIF4GI dissociation from eIF4E could precede eIF4GI targeting to the proteasome. One molecular event that could be responsible for eIF4E-eIF4GI complex dissociation was competition between eIF4GI and 4E-BP1 for binding to eIF4E. Indeed, the hypophosphorylated forms of 4E-BP1 naturally compete with eIF4GI for a common binding site on eIF4E, and 4E-BP1 hypophosphorylation has been observed under many different conditions of stress. We therefore tested whether 4E-BP1 hypophosphorylation and consequent binding to eIF4E could result in eIF4G dissociation from eIF4E upon treatment with dicumarol. Direct Western blotting first revealed dramatic accumulation of the hypophosphorylated forms of 4E-BP1 (Fig. (Fig.6A).6A). Then a coimmunoprecipitation analysis showed that hypophosphorylated 4E-BP1 bound tightly to eIF4E and evicted eIF4GI from cap-dependent translation initiation complexes much earlier than eIF4GI degradation. Thus, we can conclude that the dramatic and rapid inhibition of protein synthesis observed in dicumarol-treated cells is due to eIF4F complex disruption caused by 4E-BP1 activation and eIF4GI degradation by the proteasome.
Through the reduction of intracellular quinones, NQO1 plays a role in the cellular response to oxidative stress (27). Furthermore, exposure of cells to oxidative stress has been shown to induce the proteasomal degradation of ODC (1), probably because NQO1 is recruited to exert its enzymatic function and can therefore no longer protect other proteins from degradation by the proteasome. Although oxidative stress has been shown recently to modulate translation initiation rates via components of the eIF4F complex, including eIF4GI (29), the possibility that eIF4GI degradation by the proteasome can be regulated by oxidative stress has never been explored. We therefore examined whether oxidative stress affects the eIF4GI level via proteasomal degradation. Cells were treated with H2O2, and the eIF4GI amount was determined. Time-dependent treatment of NIH 3T3 cells with H2O2 resulted in a significant decrease in the eIF4GI level (Fig. (Fig.7A).7A). In this experiment, ODC activity was used as a positive control. Pretreatment of cells with lactacystin followed by exposure to increasing concentrations of H2O2 precluded eIF4GI diminution, thus showing that a dose-dependent, H2O2-mediated reduction in the eIF4GI amount was the result of proteasomal degradation (Fig. (Fig.7B7B).
Taken together, our data show that the eIF4GI steady-state level is dependent on proteasomal activity and that proteasomal degradation of eIF4GI is controlled by NQO1 but can be induced by oxidative stress.
The intracellular amount of eIF4GI has been shown to be tightly controlled in different biological situations. This is particularly true in virus-infected and in apoptotic cells, where eIF4GI is separated by viral or apoptotic proteases into fragments which are recycled to serve in specific translational functions. Here, we give experimental evidence supporting the idea that eIF4GI cleavage by the proteasome may also generate discrete fragments (Fig. (Fig.3A)3A) which may exert specific translational functions. This view is supported by the data obtained by Baugh and Pilipenko showing that eIF4GI fragments generated by the 20S proteasome (via endoproteolytic cleavage) differently affect the translation of different cellular mRNAs (5). How eIF4GI fragments generated by proteasomal cleavage precisely function in the translation of specific mRNAs remains puzzling, however.
Another important issue that should be tackled in the future is under which physiological circumstances the intracellular amount of eIF4GI is controlled by proteasomal activity. Our data give experimental evidence supporting a model in which eIF4GI is degraded by the proteasome under oxidative stress and in which such degradation may be controlled by NQO1. It would now be of great interest to explore the possibility that such modalities of regulation are altered in tumor situations. This is likely, because NQO1 (10, 19) and eIF4GI are both overexpressed in various cancers. Based on our data, it can be speculated that accumulation of eIF4GI may result from stabilization of the protein through increased interaction with overexpressed NQO1 and consequent protection from proteasome-mediated degradation. Another possibility is that eIF4GI accumulates in cancer cells because increased NQO1 amounts counteract oxidative stress which would have otherwise targeted eIF4GI for degradation by the proteasome.
The possibility that alterations in the control of eIF4GI by the proteasome occur in cancer cells is also supported by the finding that upon induction of p53, eIF4GI degradation is independent of caspase activation (7). Thus, one possibility is that when healthy cells are exposed to stress, p53 induces apoptosis in part via the destruction of eIF4GI and the consequent inhibition of protein synthesis. However, NQO1 overexpression in malignant cells may counteract p53-dependent proteasomal degradation of eIF4GI. Stabilized eIF4GI then maintains high protein synthesis rates and therefore contributes to resistance against apoptosis.
This work was supported by grants from INSERM, from the Association pour la Recherche contre le Cancer (no. ARC-3633 and ARC-3899), from La Ligue contre le Cancer (comités de Haute-Pyrénées et de Lot-et-Garonne), and from Cancéropôle Grand Sud-Ouest. Amandine Alard was the recipient of a 3-year doctoral fellowship from the French Ministry of Research, followed by a 1-year doctoral fellowship from the ARC.
We are grateful to Nahum Sonenberg for eIF4GI antibodies and cDNAs and to Yosef Shaul for NQO1 cDNA.
Published ahead of print on 22 December 2009.