|Home | About | Journals | Submit | Contact Us | Français|
Cells encounter oxygen deprivation (hypoxia) in various physiological and pathological contexts. Adaptation to hypoxic stress occurs in part by suppressing MYC, a key regulator of cellular metabolism, proliferation, and survival. Hypoxia has been reported to inhibit MYC through multiple means, including disruption of MYC transcriptional complexes and decreased MYC protein abundance. Here we identify enhanced proteasomal degradation and cathepsin-mediated proteolysis as important mechanisms for hypoxic MYC inhibition in human colon carcinoma cells. MYC protein levels were similarly reduced in hypoxic primary keratinocytes. Increased MYC turnover at low O2 tension was dependent on the E3 ubiquitin ligases FBXW7 and DDB1, as well as hypoxic induction of cathepsins D and S. Reduced MYC protein levels coincided with hypoxic inhibition of RNA polymerase III-dependent MYC target genes, which MYC regulates independently of its binding partner MAX. Finally, MYC overexpression in hypoxic cells promoted cell cycle progression but also enhanced cell death via increased expression of the proapoptotic genes NOXA and PUMA. Collectively, these results indicate that hypoxic cells promote MYC degradation as an adaptive strategy to reduce proliferation, suppress biosynthetic processes, and promote cell survival under low O2 tension.
The proto-oncogene MYC is broadly expressed in proliferating tissues. Decades of study have revealed crucial roles for MYC in the promotion of cell division, ribosomal assembly, and anabolic metabolism in both normal and cancer cells (1). MYC family deregulation occurs in more than 40% of all cancers, including Burkitt's lymphoma, neuroblastoma, and multiple myeloma, and high levels of MYC activity are frequently a poor prognostic indicator (2, 3). Multiple mechanisms contribute to MYC overexpression in tumors, such as chromosomal translocation, amplification, or stabilizing mutations. MYC activity is also regulated by growth factor signaling pathways, which are in turn influenced by microenvironmental factors, such as nutrient or O2 availability (4).
One of the principal functions of MYC is to coordinate the expression of multiple proteins responsible for cell cycle progression. MYC activates the transcription of its targets—e.g., the cyclin D2 (CCND2) and cyclin-dependent kinase 4 (CDK4) genes—by binding to CACGTG (E-box) DNA sequences in association with its heterodimeric partner MAX (4). MYC and MAX can also bind to and inactivate the transcription factors MIZ1 and SP1 at initiator elements, thus repressing transcription of CDK inhibitors CDKN1A and CDKN2B (5). MYC activity is negatively regulated by the MAD family of proteins, including MXD1 and MXI1, which competitively titrate MAX away from MYC (4). However, MYC target gene transcription by RNA polymerase III (Pol III) does not require MAX, MXD1, or MXI1. MYC binds transcription factor IIIB (TFIIIB) subunits TBP and BRF1 directly to enhance Pol III-dependent transcription of 5S rRNA (RN5S) and tRNA genes (6, 7). These effects on RNA synthesis and cell cycle progression comprise a key mechanism whereby MYC coordinates cell growth and proliferation.
Elevated MYC activity can commit cells to bioenergetic and synthetic demands that exceed available nutrient supplies (1). In this context, cell death resulting from hyperactive MYC can be a tumor-suppressive response to prevent unrestrained tumor growth. MYC-induced apoptosis involves a number of effector molecules, including the tumor suppressors ARF and p53. For example, MYC-dependent induction of ARF stabilizes p53 by inhibiting its negative regulator MDM2 (8, 9). Stabilized p53 in turn stimulates the expression of proapoptotic proteins NOXA and PUMA, resulting in activation of the downstream effector BAX (10, 11). MYC can also induce cell death independently of p53, for example, by directly regulating the expression of NOXA and other apoptotic genes (12, 13). Importantly, elevated MYC activity sensitizes cells to numerous apoptotic stimuli, including tumor necrosis factor alpha (TNF-α) death receptor signaling, DNA damage, and O2 and nutrient deprivation (14–17).
To circumvent MYC-induced cell death under conditions of decreased nutrient and growth factor availability, some cells reduce their metabolic and proliferative requirements by downregulating MYC activity. In particular, MYC protein expression and activity can be modulated by nutrient- and growth factor-responsive signal transduction pathways. For example, inhibition of RAS signaling reduces MYC stability via changes in MYC phosphorylation and subsequent FBXW7-dependent ubiquitylation and proteolysis (18). Similarly, activation of SIRT1, a sensor of cellular metabolic state, leads to MYC deacetylation and degradation (19). Furthermore, cytoplasmic proteases, such as calpains, regulate MYC activity and cell differentiation via proteolytic cleavage (20, 21). The control of MYC abundance and activity is therefore an important response to fluctuations in nutrient and growth conditions, including changes in O2 tension.
O2 is often in limited supply in solid tumors because of defective and inadequate vascularization in the context of rapid cell division (22). In its absence, cells are unable to generate ATP via oxidative phosphorylation and must undergo metabolic adaptations in order to survive. Many of these adaptations are mediated by the stabilization of hypoxia-inducible factors HIF1α and HIF2α, which activate transcription of genes encoding angiogenic, hematopoietic, and metabolic effectors (23). HIF induction in hypoxic cells suppresses oxidative phosphorylation and promotes nonoxidative forms of ATP production, such as glycolysis (24). HIF also promotes autophagosomal and lysosomal activity to relieve cellular energy demand and recycle cellular nutrient sources (25). Concurrently, HIF-dependent angiogenesis improves O2 delivery. Because these adaptive changes require time, hypoxia decreases energy consumption by reducing cell proliferation, mitochondrial metabolism, and DNA replication and repair, often by inhibiting MYC activity (26–29).
Hypoxic inhibition of MYC largely occurs via HIF-dependent effects on MYC-interacting proteins. For example, HIF1α directly induces MXI1 expression to inhibit MYC-dependent mitochondrial biogenesis and O2 consumption (29, 30). At the protein level, HIF1α competes with MYC for binding to SP1 at the promoters of MYC target genes, such as MSH2, MSH6, and NBS1, which encode DNA repair proteins. MYC displacement from these promoters represses gene expression and causes genomic instability in tumor cells (28, 31). This effect is exclusive to HIF1α, as HIF2α fails to bind SP1 in the same manner (32). HIF1α has also been reported to bind to MAX and disrupt MYC-MAX complexes, leading to reduced CCND2 expression, induction of CDKN1A, and G1-phase arrest (26). However, it is unclear whether hypoxia affects important MYC activities that operate independently of MYC-interacting proteins (e.g., MAX, MXI1, MIZ1, or SP1), such as transcription of MYC target genes by Pol III. Furthermore, the diversity of MYC-dependent cellular processes suggests that the biological consequences of MYC suppression under low O2 tension remain to be fully elucidated.
Here we investigate the mechanisms and consequences of MYC suppression in hypoxic adaptation. We show that hypoxic stress decreases MYC protein stability and transcriptional activity in normal and cancer cells. Hypoxia represses not only MYC target genes transcribed by RNA polymerase II (Pol II), such as CCND2 and MCM5, but also Pol III-dependent MYC target genes encoding ribosomal and transfer RNAs. MYC suppression is independent of changes in MYC transcription and mRNA translation rates but results from its increased proteasomal and nonproteasomal degradation under hypoxic conditions. Hypoxic MYC degradation requires hypoxia-induced cathepsin expression, as well as expression of the E3 ubiquitin ligases DDB1 and FBXW7. Forced expression of a stabilized MYC variant in hypoxic cells increases NOXA expression and enhances hypoxia-mediated cell death. Therefore, we propose that MYC suppression under hypoxia is an adaptive response that promotes cell survival under low-O2 conditions.
The HCT116, DLD1, and REF52 cell lines were maintained in Dulbecco's modified Eagle's medium (DMEM) with 10% fetal bovine serum at 37°C in 5% CO2. Primary keratinocytes were isolated from neonatal mice as described in reference 33 and cultured in complete MCDB 153 medium (Sigma) with 45 μM calcium chloride at 34°C in 8% CO2. To generate cell lines expressing hemagglutinin (HA)-tagged MYC and the T58→A MYC mutant (MYCT58A), HCT116 or REF52 cells were transduced with retroviral pMSCV-HA-MYC or pMSCV-HA-MYCT58A vector (Addgene) and selected with puromycin (Sigma). For hypoxia treatment, cells were cultured in an InVivo2 400 hypoxia workstation (Ruskinn) at 0.5% O2. For proliferation studies, 3 × 104 cells were seeded in duplicates in 60-mm tissue culture dishes and grown under normoxic or hypoxic conditions for 1 to 6 days. Cell number was counted using a hemacytometer.
Proteasome and protease inhibitors were purchased from Calbiochem and used at the following concentrations: MG132, 25 μM; lactacystin, 25 μM; cathepsin inhibitor I (CAT I), 20 μM; cathepsin inhibitor III (CAT III), 10 μM; and chloroquine, 20 μM.
Total RNA from HCT116 cells was reverse transcribed as described below. Specific primers were used to amplify full-length MXD1, MXI1A, and MXI1C cDNA with 5′ and 3′ restriction sites for HindIII and XbaI, respectively (primer sequences available on request). PCR products were purified using QIAquick gel extraction kit (Qiagen) and cloned into pcDNA3.1 vector (Invitrogen) for sequence analysis and transient expression.
MXD1, MXI1A, MXI1C, HIF1A-FLAG, and HIF2A-FLAG were transiently expressed in HEK293T cells using Lipofectamine 2000 (Invitrogen), and cell lysates were prepared after 48 h. Six-histidine (6×His)-tagged and FLAG-tagged ubiquitin were overexpressed in HCT116 cells using Lipofectamine 2000. For knockdown studies, small interfering RNAs (siRNAs) targeting CUL4A, CUL4B, DDB1, FBXW7, HIF1A, HIF2A, MYC, MXD1, MXI1, NOXA, PUMA, TRPC4AP, UBR5, and the scrambled control (Qiagen) or siRNAs against MIR34B, MIR34C, and the scrambled control (Dharmacon) were transfected into cells at a 50 nM concentration using HiPerFect (Qiagen) or DharmaFECT 2 (Dharmacon), respectively. Hypoxic treatment was initiated 24 h after transfection and maintained for a further 24 h unless otherwise specified.
For cell cycle analysis, HCT116 cells were pulsed with 10 μM bromodeoxyuridine (BrdU) for 20 min after 24 h of culture under the appropriate conditions. Cells were stained with Alexa Fluor 488–anti-BrdU (Invitrogen) and propidium iodide (PI) for analysis in a FACScalibur flow cytometer (Becton, Dickinson). Cell viability was measured using an annexin V-fluorescein isothiocyanate (FITC) apoptosis detection kit (BD Pharmingen) and analyzed by flow cytometry.
HCT116 cells were grown on collagen-coated glass coverslips and exposed to 21% or 0.5% O2 for 24 h. Cells were incubated in the dark in growth medium containing 50 nM LysoTracker Red solution (Invitrogen) for 1 h. Subsequently, cells were fixed in 3% paraformaldehyde and incubated in cathepsin D (NB600-1295; Novus) or cathepsin S (Ab92780; Abcam) antibody at 4°C overnight. Cells were probed with Alexa Fluor 488-conjugated secondary antibody (Invitrogen), mounted in Prolong Gold antifade reagent (Invitrogen), and imaged on a Zeiss LSM 710 inverted confocal microscope.
For quantitative reverse transcription-PCR (qRT-PCR), total RNA was extracted with TRIzol reagent (Invitrogen) and cDNA was produced using a high-capacity RNA-to-cDNA kit (Applied Biosystems). Analysis of gene expression was performed in a 7900HT sequencer (Applied Biosystems) using specific primers (sequences available on request). Expression levels were normalized to levels of HPRT1. MicroRNA was extracted using the miRNeasy kit (Qiagen), and expression analysis was performed using the TaqMan microRNA assay (Applied Biosystems) in a 7900HT sequencer.
Chromatin immunoprecipitation (ChIP) was performed as described in reference 34. Cells were cross-linked in 1% formaldehyde for 10 min. Precleared chromatin was immunoprecipitated with anti-MYC N262 antibody (Santa Cruz) and protein A-agarose beads (Roche) for 16 h at 4°C. Immunoprecipitated DNA and total DNA were quantified in a 7900HT Sequencer using specific primers (sequences available on request).
For immunoprecipitation (IP), cells were lysed in 25 mM Tris (pH 8.0), 100 mM NaCl, 1 mM dithiothreitol (DTT), 1 mM EDTA, and 1% NP-40 containing Complete mini-protease inhibitor cocktail (Roche), 100 mM N-ethylmaleimide (NEM; Sigma), and 200 μM deferoxamine (DFO) (Sigma). Precleared lysates were immunoprecipitated in lysis buffer with anti-MYC N262, anti-MAX H2 (Santa Cruz), or anti-FLAG M2 (Sigma) antibody.
Twenty to 50 μg of total protein extract was electrophoresed on 10% sodium dodecyl sulfate-polyacrylamide gels. Antibodies against DDB1 (Cell Signaling), MYC C33, MYC N262, MYC 9E10, phospho-T58/S62-MYC E203 (Epitomics), MAX C17, MAX H2, HIF1α C terminal (Cayman), HIF2α NB100-122 (Novus), HA (Roche), FLAG-M2 (Sigma), cyclin E M20, MXD1 C19, MXI1 G16, β-actin (Sigma), and ARNT H172 were used for Western blotting. (All antibodies were from Santa Cruz Biotechnology unless otherwise noted.) Optical densities were quantified using ImageJ software (NIH).
Cells were starved of methionine and cysteine for 30 min. Labeling was carried out in HCT116 or REF52 cells for 15 min in DMEM without methionine or cysteine, supplemented with 10% or 2% dialyzed fetal calf serum, respectively, and 100 μCi/ml of [35S]methionine-cysteine. As a negative control, cells were treated with 100 μg/ml cycloheximide (CHX) (Calbiochem). Total protein extract (0.5 to 1 mg) was used for immunoprecipitation with anti-MYC C33 antibody, anti-HA 3F10 antibody, or rabbit preimmune serum. Bound proteins were eluted with 2× sample buffer, separated by SDS-PAGE, and visualized by autoradiography. Phosphorescence was quantified using Storm 820 molecular imager (Molecular Devices) and ImageQuant software (GE).
Data are presented as means + standard errors of the mean (SEM). Differences between groups were analyzed for significance using Student's t test, with P values <0.05 considered significant.
To assess the extent of hypoxic effects on MYC-dependent gene transcription, we cultured HCT116 colon carcinoma cells under hypoxia (0.5% O2) for 24 h. Quantitative RT-PCR (qRT-PCR) analysis revealed that hypoxia-inducible genes PGK1 (encoding phosphoglycerate kinase 1), MXD1 (MAX dimerization protein 1), and MXI1 (MAX interactor 1) were robustly induced (Fig. 1A). Hypoxia also inhibited the expression of MYC target genes involved in cell cycle progression (CCND2) and DNA replication (MCM5 [minichromosome maintenance-deficient 5] and MCM7 [minichromosome maintenance-deficient 7]) (Fig. 1A). Conversely, expression of the MYC-repressed gene CDKN1A (p21 cyclin-dependent kinase inhibitor 1A) was increased, consistent with prior reports (26, 28) (Fig. 1A).
To determine whether MYC-induced genes transcribed by Pol III were also repressed by hypoxic treatment, we examined the expression of MYC target genes involved in protein translation. Transcript levels of TRNAR (tRNA arginine gene), TRNAL (tRNA leucine gene), and TRNAY (tRNA tyrosine gene) were also decreased under low O2 tension (Fig. 1A). In contrast, expression of RPLP0 (ribosomal protein large P0 gene), which is not a MYC target gene, was unchanged by hypoxia (Fig. 1A). The chromatin immunoprecipitation assay demonstrated that the reduction in MYC-dependent gene transcription was accompanied by decreased MYC occupancy at target gene promoters (Fig. 1B). These results demonstrate that O2 limitation not only antagonizes the expression of MYC target genes transcribed by Pol II but also represses Pol III-dependent MYC target gene expression. In this way, hypoxic inhibition of growth, ribosome assembly, and protein synthesis are coordinated via MYC suppression.
Suppression of MYC activity in hypoxic cells could be mediated by disruption of MYC-MAX complexes, induction of the negative regulator MXI1, or degradation of MYC protein (26, 30, 35). Consistent with previous reports (26, 36), MYC's association with its positive binding partner MAX decreased after 24 h of hypoxia (Fig. 2A). While the loss of MYC-MAX heterodimers was expected to inhibit Pol II-dependent transcription of MYC target genes, it could not account for the observed reduction in Pol III-dependent MYC target gene transcription. Western blot analysis showed that whereas MAX protein expression remained constant, MYC protein levels were decreased in hypoxic HCT116 cells and also in primary murine and human keratinocytes (Fig. 2B; see Fig. S1A in the supplemental material) (data not shown). Taken together, these results suggest that hypoxia reduces MYC protein abundance and thus represses MYC target gene transcription by both Pol II and Pol III. Furthermore, decreased MYC protein abundance may be a general response to hypoxia in both malignant and untransformed cells.
To determine which regulatory steps in MYC expression are O2 responsive, we used qRT-PCR to measure MYC mRNA levels. MYC mRNA abundance was maintained after the same duration of hypoxic exposure that reduced MYC target gene expression (Fig. 2C). Because MYC mRNA translation is repressed by microRNA 34B/C (MIR34B/C) induction in response to DNA damage (37), we investigated the possibility that microRNAs regulate MYC expression during O2 deprivation. Interestingly, hypoxia induced MIR34B and MIR34C expression (see Fig. S1B in the supplemental material). However, RNA interference (RNAi)-mediated inhibition of MIR34B or MIR34C failed to prevent MYC suppression under hypoxia (see Fig. S1C and D). MYC protein was also suppressed by hypoxia in both control and DICER1 mutant HCT116 cells, which have reduced expression of most mature microRNAs (38) (see Fig. S1E).
To examine MYC mRNA translation under hypoxia directly, HCT116 cells were pulse-labeled with [35S]cysteine-methionine to evaluate new protein synthesis. The amount of radiolabeled protein in anti-MYC immunoprecipitates was unchanged in normoxic (21% O2) versus hypoxic (0.5% O2) cells, demonstrating that hypoxia does not inhibit MYC protein synthesis (Fig. 2D). These results indicate that hypoxia decreases MYC protein abundance in a manner that is independent of transcriptional and translational effects, including regulation by microRNAs.
Although O2 limitation is known to induce the expression of MYC antagonists (30, 39), MXD1 or MXI1 overexpression was insufficient to reduce MYC protein levels in normoxic cells (see Fig. S1F in the supplemental material). In the converse experiment, in which MXD1 or MXI1 was silenced using RNAi (see Fig. S1G), MYC protein levels were unaffected (data not shown). However, MXI1 knockdown resulted in derepression of MYC target gene expression (see Fig. S1H). These observations are consistent with a role for MXD1 and MXI1 in regulating MYC activity but not MYC protein abundance. Intriguingly, MXD1 and MXI1 knockdown suppressed hypoxic induction of CDKN1A (see Fig. S1H). The mechanism by which CDKN1A suppression occurs is not immediately apparent, although it has been reported that enhanced MYC-MAX dimerization can antagonize MYC-mediated repression (40). Inhibition of MXI1 may therefore lead to increased MYC-MAX dimerization and, consequently, reduced MYC interaction with MIZ1.
We next measured the effect of O2 deprivation on MYC protein turnover. MYC protein is highly unstable, with a reported half-life of 20 to 30 min (41, 42). Pulse-chase analysis indicated that hypoxic treatment reduced the half-life of MYC protein from 40 min to 18 min (Fig. 2E). To study the mechanism for increased MYC degradation under hypoxia, we used chemical compounds to inhibit different components of the protein degradation machinery. Treatment of hypoxic cells with either or both of the proteasomal inhibitors MG132 and lactacystin partly restored MYC protein levels (Fig. 3A). Nonproteasomal contributions to hypoxic MYC turnover were evaluated using cathepsin inhibitors I and III (inhibiting cathepsins B, L, and S and papain) and chloroquine. While chloroquine treatment had no consistent effect on MYC expression under hypoxic conditions, cathepsin inhibition significantly restored hypoxic MYC levels (Fig. 3A).
Nonproteasomal mediators of MYC turnover were further evaluated using pulse-chase experiments. Treatment with cathepsin inhibitor III (CAT III) extended MYC half-lives at both 21% and 0.5% O2 (Fig. 3B). Importantly, CAT III treatment significantly and almost completely abrogated the effect of hypoxia on MYC stability (Fig. 3B). CAT III primarily inhibits cathepsin activity, but it also displays some activity toward other cysteine proteases, including papains and calpains. While calpains have been associated with MYC proteolysis (20, 21), a role for cathepsins in MYC stability has not been reported previously. The effect of cathepsin and calpain inhibition on extending MYC half-life can also be contrasted with its effect on MYC protein levels under hypoxia (Fig. 3A and andB).B). MYC half-life was restored to essentially normoxic values after CAT III treatment in cells that had been exposed to 24 h of hypoxia (Fig. 3B). However, after 44 h of hypoxic treatment, CAT I and CAT III had only partial effects on MYC protein levels (Fig. 3A). In these cells, combined inhibition of cathepsin and proteasome activity using CAT III, a cathepsin inhibitor, and lactacystin, a proteasome inhibitor, significantly and almost completely restored MYC protein abundance to normoxic levels (Fig. 3A). These differences suggest that cytoplasmic proteases and the proteasome may be activated by hypoxia with different response latencies. Taken together, these experiments indicate that low O2 tension promotes MYC degradation via both proteasome-dependent and protease-dependent mechanisms.
The requirement of cathepsin activity for hypoxic MYC degradation correlated with hypoxic gene induction of cathepsins S, K, B, and D (Fig. 4A). Cathepsins S, K, and B are cysteine proteases, while cathepsin D is an aspartyl protease. Although most of these cathepsins are acidic proteases that operate within the low-pH environment of the lysosome, cathepsin S is stable at neutral pH and plays physiological roles both inside and outside the lysosome. To further evaluate the role of hypoxia in regulating lysosomal activity, we measured the expression of cathepsins D and S in hypoxic HCT116 cells using confocal immunofluorescence microscopy. Cathepsins D and S were both induced by hypoxic treatment (Fig. 4B). Significant colocalization was observed between cathepsin D or S immunofluorescence and acidic lysosomes, detected by LysoTracker Red dye (Fig. 4B). As expected, cathepsin S induction was also observed in the cytoplasm. LysoTracker Red staining also colocalized with that of the lysosomal membrane protein LAMP1, confirming dye specificity (data not shown). Interestingly, lysosome formation, as measured by LysoTracker Red and LAMP1 staining, was also increased under hypoxic treatment (Fig. 4B) (data not shown). Therefore, hypoxic induction of cathepsin expression and lysosome formation comprise a mechanism by which MYC levels are regulated by low O2 tension.
Multiple E3 ubiquitin ligases for MYC have been identified, including SKP2, FBXW7, HUWE1, BTRC, TRPC4AP, and DDB1 (43–48). Hypoxic reduction of MYC protein levels was not accompanied by enhanced expression of any MYC E3 ubiquitin ligase (data not shown). However, RNAi-mediated silencing of FBXW7 and DDB1 significantly restored MYC protein levels under hypoxic conditions (Fig. 5A). To assess the role of FBXW7 in hypoxic MYC degradation, we made use of colorectal cancer cell lines in which FBXW7 had been deleted (49). Similar to proteasomal inhibition by MG132, FBXW7 deletion partially (but significantly) restored MYC levels in hypoxic colorectal cancer cells (see Fig. S2A in the supplemental material). Pulse-chase analysis in FBXW7-deleted cells indicated that MYC protein was stabilized under both normoxic and hypoxic conditions in the absence of FBXW7: the MYC half-life at 21% O2 was 87 min, while that at 0.5% O2 decreased to 69 min (see Fig. S2B). Furthermore, the extent of MYC destabilization under hypoxic settings was abrogated in cells lacking FBXW7 (31% reduction compared to normoxic levels in wild-type cells and 21% reduction in FBXW7 cells) (compare Fig. S2B in the supplemental material with Fig. 2E). These results confirmed that FBXW7 partially contributes to MYC inhibition under O2 deprivation. However, the reduction in MYC half-life in hypoxic FBXW7−/− cells indicates that other (proteasomal or nonproteasomal) processes also play important roles in MYC regulation under low O2 tension.
While FBXW7 has been suggested to regulate MYC levels under hypoxic conditions, DDB1 involvement had not been previously identified. RNAi-mediated silencing of DDB1 in HCT116 cells lacking FBXW7 restored MYC protein abundance under hypoxia to higher levels than in cells deficient in either DDB1 or FBXW7 alone (Fig. 5B to toD;D; see Fig. S2C in the supplemental material). Interestingly, knockdown of DDB1 binding partners CUL4A and CUL4B also restored hypoxic MYC protein levels, albeit insignificantly (Fig. 5A). Taken together, these results suggested that the E3 ligases FBXW7 and DDB1 are both responsible for hypoxic MYC degradation. DDB1 involvement in MYC protein regulation is not a consequence of indirect effects on FBXW7, as DDB1 silencing in HCT116 cells did not reduce FBXW7 expression (see Fig. S2D).
The activities of MYC E3 ligases are regulated by distinct upstream signaling events. For example, MYC phosphorylation on Thr58 and Ser62 is a prerequisite for FBXW7-dependent ubiquitylation; however, MYC mutants lacking phosphorylation sites at amino acid residues 58 and 62 can still be ubiquitylated and degraded (45). We did not observe any significant increase in MYC phosphorylation on Ser62 and Thr58 in hypoxic cells (see Fig. S2E in the supplemental material). Antibody specificity against phospho-T58/S62-MYC was verified using RNAi-mediated MYC knockdown and treatment of cell lysates with λ protein phosphatase (see Fig. S2E and F). Despite the absence of changes in MYC phosphorylation, MYC ubiquitylation was modestly increased in hypoxic HCT116 cells. FLAG-tagged ubiquitin was transiently expressed in HCT116 cells, and MYC protein was immunoprecipitated from cell lysates. MYC was more highly ubiquitylated in hypoxic cells than in the normoxic control (see Fig. S2G). These results indicate that hypoxia promotes MYC polyubiquitylation and subsequently its proteasomal degradation via FBXW7 and DDB1.
Previous reports have demonstrated that HIF proteins can regulate hypoxic MYC activity (50). To characterize the requirement for HIF in hypoxic MYC degradation, we employed RNAi-mediated knockdown to inhibit HIF1α or HIF2α alone or in combination (Fig. 6A). The absence of HIF1α or HIF2α alone had no effect on MYC degradation under hypoxia. When both HIF isoforms were inhibited simultaneously, MYC protein abundance was partially restored (Fig. 6A). We next examined whether HIF activity contributed toward proteasome- or protease-dependent MYC regulation. HCT116 cells in which HIF1α and HIF2α had been inhibited by RNAi were treated with cathepsin inhibitor (CAT III) or proteasome inhibitor (lactacystin). Compared to control cells treated with CAT III, HIF knockdown and cathepsin inhibition produced an additive effect toward restoring hypoxic MYC protein levels (Fig. 6B). In contrast, combined inhibition of HIF and proteasomal activity did not further enhance MYC protein levels under hypoxic conditions (Fig. 6B). These results suggest that HIF acts largely via the proteasome to enhance MYC degradation under low O2 tension.
Although HIF1α and HIF2α are required for hypoxic MYC degradation, we were unable to detect direct interactions between MYC, MAX, HIF1α, and HIF2α using in vitro translated proteins (see Fig. S3A in the supplemental material). Coimmunoprecipitation studies in endogenous and overexpression settings confirmed MYC-MAX and HIF-ARNT binding but not interactions between HIF1α/HIF2α and MYC/MAX (see Fig. S3B to D). This result was somewhat surprising, because HIF1α has been suggested to bind to MYC/MAX and thereby inhibit MYC transcriptional activity (26–28, 31). Importantly, MYC-MAX dimerization in HCT116 cells was unaffected by 6 h of hypoxic exposure compared to the normoxic control, but it decreased after 24 h of hypoxia, coincident with a reduction in MYC protein levels (compare Fig. S3B with Fig. 2A and andB).B). Therefore, we propose that hypoxic inhibition of MYC activity largely occurs via disruption of MYC-MAX dimerization due to MYC protein degradation.
To study the biological consequences of MYC inhibition under hypoxia, we compared the behavior of cells transduced with retroviral vectors encoding HA-tagged versions of wild-type HA-MYC or the MYCT58A mutant (containing the Thr58→Ala58 mutation). The absence of Thr58 phosphorylation renders MYCT58A relatively resistant to degradation via the ubiquitin-proteasome pathway (51). Indeed, our experiments indicate that MYCT58A undergoes hypoxia-mediated degradation, albeit at a lower rate. Pulse-chase analyses in REF52 fibroblasts confirmed that MYCT58A stability was enhanced compared to that of wild-type HA-MYC in normoxic settings (half-life of 63 min versus 27 min) as well as under hypoxic conditions (51 min versus 20 min) (Fig. 7B). Western blotting in HCT116 cells verified that HA-MYC and MYCT58A were expressed at similar levels under normoxic conditions (data not shown). Furthermore, MYCT58A degradation proceeded at a lower rate than that of endogenous wild-type MYC, such that under hypoxic conditions, HCT116-MYCT58A cells express MYCT58A predominantly (Fig. 7C).
Surprisingly, enhanced MYC expression did not result in increased cell numbers as measured by serial cell counts: HCT116-MYCT58A cell counts were lower than HCT116-HA-MYC cell counts at both 21% and 0.5% O2 (Fig. 7D). BrdU incorporation experiments revealed that, paradoxically, MYCT58A cells displayed a trend toward increased G1/S cell cycle progression, whereas control cells underwent S/G2 arrest under hypoxia (Fig. 7E). Therefore, cell cycle changes in response to low O2 tension cannot account for the growth disadvantage of HCT116-MYCT58A cells.
Because both hypoxia and MYC are known to induce apoptosis under certain conditions, we asked whether MYC overexpression in HCT116 cells contributed to apoptotic cell death. Cell viability was measured by annexin V and propidium iodide (PI) staining. The proportion of annexin V- and PI-negative (i.e., live) cells was reduced by the combination of O2 limitation and MYCT58A expression (Fig. 8A and andB).B). Correspondingly, the proportion of dying (annexin V-positive, PI-negative) and dead (annexin V- and PI-positive) cells increased in hypoxic HCT116-MYCT58A cells (Fig. 8B). To identify the mediators of cell death in response to hypoxia and MYC activity, we evaluated the expression of various BCL2 family proteins. Hypoxia induced PUMA expression in both control and MYCT58A cells (Fig. 8C). In contrast, expression of NOXA, a direct MYC target gene (13), was induced only in hypoxic MYCT58A cells (Fig. 8D). These experiments suggest that high MYC activity under hypoxia results in an additional apoptotic stimulus, which correlates with increased NOXA induction.
Finally, we evaluated the contribution of NOXA and PUMA expression to hypoxic cell death in MYCT58A cells. NOXA silencing alone was insufficient to improve HCT116-MYCT58A cell viability. However, combined inhibition of NOXA and PUMA restored the viability of hypoxic HCT116-MYCT58A cells to normoxic levels (Fig. 8E; see Fig S4 in the supplemental material). In contrast, HCT116-HA-MYC cell viability was unaffected by either hypoxic treatment or NOXA and PUMA knockdown (Fig. 8E). This finding suggests that the coinduction of NOXA and PUMA in hypoxic cells with high MYC activity contributes to hypoxia-induced cell death. HCT116-HA-MYC cells, which are able to downregulate MYC under hypoxic conditions, did not induce NOXA expression (Fig. 8D); PUMA induction in these cells was insufficient to promote hypoxia-induced cell death. Collectively, the data presented here indicate that hypoxic MYC downregulation is likely a physiologically important response to circumvent NOXA induction and avoid NOXA- and PUMA-dependent cell death, thereby preserving cell viability under hypoxic stress (Fig. 9).
MYC is an important regulator of cell division, metabolism, and apoptosis. These cellular processes are coordinated with nutrient and energy availability by regulatory mechanisms that modulate MYC in response to different types of stress. In this study, we demonstrated that MYC protein levels are suppressed under hypoxic conditions in both transformed and untransformed cells. We have extended our analysis of MYC activity during O2 deprivation to show that Pol III-dependent transcription of MYC target genes is reduced in low O2. MYC effects on Pol III-dependent transcription are known to be mediated via its direct interaction with the transcription factor TFIIIB and do not require binding to MAX, MIZ1, SP1, or other related proteins (6, 7). Therefore, the effect of O2 on MYC-dependent transcription cannot be fully explained by previous models of hypoxic MYC inhibition, in which MYC-containing complexes are disrupted either by direct HIF binding or by induction of negative regulators MXD1 and MXI1 (50). We have identified the E3 ubiquitin ligase DDB1 and cathepsin proteases as novel regulators of hypoxic MYC degradation; we also independently confirmed that the E3 ligase FBXW7 contributes to increased MYC degradation under hypoxia (35). When MYC degradation under hypoxic conditions was prevented by forced expression of MYCT58A, cell death occurred via NOXA and PUMA induction, indicating that hypoxic MYC suppression is important for maintaining cell survival under low O2 tension.
MYC expression and activity are known to be controlled at virtually all levels of regulation (4). However, MYC suppression in response to various stresses occurs via distinct pathways. For example, the absence of microRNA effects on hypoxic MYC abundance illustrates the complexity of MYC regulatory mechanisms in response to O2 limitation and DNA damage (37). Interestingly, MIR34B or MIR34C inhibition enhanced normoxic MYC expression (see Fig. S1C in the supplemental material), suggesting that these microRNAs may be involved in homeostatic but not hypoxic MYC regulation. While our work indicated that hypoxic MYC inhibition occurred posttranslationally, it is possible that prolonged hypoxic exposure (i.e., greater than 48 h) and the accumulation of additional stresses could in time inhibit MYC expression. For example, hypoxic suppression of Wnt signaling via LEF1/TCF1 could eventually decrease MYC transcription (52). Hypoxia also inhibits mTORC1 activity, potentially suppressing MYC expression due to global inhibition of cap-dependent translation (53–55). Although we did not observe these effects under our experimental conditions and time frame, mechanisms of MYC regulation under hypoxia may vary depending on the severity and duration of hypoxic stress.
While previous reports have suggested that hypoxia increases MYC phosphorylation at amino acid residues Thr58 and Ser62 (30, 35), our experiments indicated that phospho-T58/S62-MYC remained unchanged relative to total MYC protein levels. The reason for this discrepancy is unclear, although we have found that different commercially available antibodies exhibit various degrees of specificity toward phospho-T58/S62-MYC (data not shown). Nonetheless, consistent with published data (35), we found that hypoxia increased MYC ubiquitylation and proteasomal degradation. Proteasomal MYC degradation under low O2 tension was dependent on two E3 ubiquitin ligases, FBXW7 and DDB1. Other modes of regulation included nonproteasomal degradation mediated by cathepsin proteases.
Hypoxic treatment simultaneously induced cathepsin gene expression and lysosome formation. Importantly, we found that cathepsin expression colocalized with lysosomal structures upon hypoxic induction, suggesting that hypoxic MYC degradation may take place in lysosomes. We demonstrated for the first time that increased cathepsin expression under hypoxic conditions is correlated with enhanced lysosome formation and MYC degradation. While the role of low O2 tension in regulating lysosomal proteolysis remains poorly understood, hypoxia and HIF proteins are known activators of autophagy. During the last stage of autophagy, autophagosomes fuse with lysosomes to form autolysosomes. One possible extension of this work would be to assess whether hypoxic stimulation of autophagy is coordinated with hypoxic induction of lysosomal formation and cathepsin expression. However, the lack of clear and consistent effects of chloroquine treatment on MYC protein levels suggests it is unlikely that hypoxia-induced autophagy contributes to MYC degradation under low-O2 conditions (Fig. 3A).
Our finding that hypoxic MYC inhibition requires the concerted activity of HIF1α and HIF2α, while consistent with some previous reports (29, 35), is in contrast to the unique and antagonistic contributions of HIF1α and HIF2α to MYC activity in renal clear cell carcinoma (ccRCC) (50). Although HIF2α expression in HCT116 cells contributed to MYC inhibition, the exclusive expression of HIF2α in ccRCC cells clearly enhances MYC transcriptional activity (26). One possible explanation is that the relative expression levels of HIF1α and HIF2α determine whether O2 deprivation leads to MYC inhibition or activation. Another contentious issue lies in the significance of physical interactions between HIF and MYC family proteins in hypoxic MYC regulation. While we have previously reported the existence of HIF-MAX binding in HCT116 cells (26), we have been unable to replicate this finding. Instead, the data presented here indicate that MYC-MAX heterodimerization is necessarily disrupted—regardless of physical interactions between HIF and MAX—because of the loss of MYC protein under low O2 tension.
Elevated MYC expression in HCT116-MYCT58A cells exacerbated cell death under low O2 levels. Hypoxia promotes apoptosis by modulating p53 activity and expression of downstream effectors, including the BH3-only proteins BNIP3, PUMA, and NOXA (56–58). We have confirmed that hypoxia activates expression of the proapoptotic gene PUMA. Moreover, high MYC activity during O2 limitation resulted in additional apoptotic drive via increased NOXA induction. This observation is consistent with previous reports that NOXA is a direct MYC target gene (13). Our experiments indicate that the combination of high MYC activity and hypoxic stress promotes cell death via NOXA- and PUMA-dependent mechanisms.
The role of MYC in promoting cell death during hypoxia suggests a survival benefit in MYC suppression under low O2 tension. In support of this argument, inhibition of MYC-induced apoptosis in pancreatic β cells promotes tumor development in vivo (59). Alternatively, tumors with high MYC activity frequently lose expression of p53, ARF, or downstream apoptotic mediators, such as BAX and PUMA, in order to escape cell death (60–63). Intriguingly, induction of the proapoptotic protein BIM is attenuated in MYCT58A-overexpressing mouse embryonic fibroblasts (MEFs) compared to MEFs overexpressing wild-type MYC (64), suggesting that at least in some contexts, the stabilization of MYC mutants can inhibit MYC-dependent cell death. We did not observe consistent effects on BIM induction in HCT116-MYCT58A cells (data not shown). Instead, we found that NOXA and PUMA induction in MYCT58A-overexpressing cells enhanced apoptosis under low-O2 conditions (Fig. 8). We propose, therefore, that hypoxia-induced MYC inhibition is an adaptive mechanism to ensure cell survival at the expense of unrestricted growth.
We thank members of our laboratory for helpful discussions and critical reading of the manuscript and Bert Vogelstein, Tatyana Svitkina, and Andrei Thomas-Tikhonenko for reagents.
This work was supported by grant P01-CA104838 (M.C.S.) from the National Cancer Institute and by the Howard Hughes Medical Institute (M.C.S.).
Published ahead of print 1 July 2013
Supplemental material for this article may be found at http://dx.doi.org/10.1128/MCB.00853-12.