Activating transcription factor 4 (ATF4) is a transcription factor induced under severe hypoxia and a component of the PERK pathway involved in the unfolded protein response (UPR), a process that protects cells from the negative consequences of endoplasmic reticulum (ER) stress. In this study, we have used small interfering RNA (siRNA) and microarray analysis to provide the first whole-genome analysis of genes regulated by ATF4 in cancer cells in response to severe and prolonged hypoxic stress. We show that ATF4 is required for ER stress and hypoxia-induced expansion of autophagy. MAP1LC3B (LC3B) is a key component of the autophagosomal membrane, and in this study we demonstrate that ATF4 facilitates autophagy through direct binding to a cyclic AMP response element binding site in the LC3B promoter, resulting in LC3B upregulation. Previously, we have shown that Bortezomib-induced ATF4 stabilization, which then upregulated LC3B expression and had a critical role in activating autophagy, protecting cells from Bortezomib-induced cell death. We also showed that severe hypoxia stabilizes ATF4. In this study, we demonstrate that severe hypoxia leads to ER stress and induces ATF4-dependent autophagy through LC3 as a survival mechanism. In summary, we show that ATF4 has a key role in the regulation of autophagy in response to ER stress and provide a direct mechanistic link between the UPR and the autophagic machinery.
Tumour hypoxia drives malignant progression, confers chemoresistance and radioresistance and is associated with a poor prognosis (Graeber et al., 1996). In vivo tumour O2 levels can be extremely low, and up to 35% of tumour cells have hypoxia below 0.1% O2 (Olive et al., 2002). Furthermore, the fraction of cells existing below 0.5 mm Hg (0.06% O2, commonly referred to as the radiobiologically hypoxic fraction) exhibit almost complete radioresistance, whereas cells exposed to more than 0.1% and less than 2% O2 (in radiology defined as an intermediate oxygen tensions) show only partial radioresistance (Wouters and Brown, 1997). The hypoxic response mounted by eukaryotic cells is predominantly orchestrated by hypoxia inducible factor 1 (HIF1), a transcription factor that upregulates the expression of a range of proteins involved in hypoxic adaptation (Harris, 2002; Semenza, 2003). Hypoxia inducible factor 1α is induced by a broad range of oxygen concentrations ranging from mild 5% to moderate 0.1% hypoxia. However, severe and prolonged hypoxia <0.01% is considered a functionally different state (Wenger and Gassmann, 1996). The endoplasmic reticulum (ER) is highly sensitive to severe hypoxic stress (<0.01% O2) as it perturbs and reduces protein folding capacity, which is oxygen dependent, resulting in the accumulation and aggregation of misfolded proteins in the ER lumen (Tu and Weissman, 2004). Protein aggregation is sensed by the eukaryotic translation initiation factor 2 alpha kinase 3 (PERK; Bi et al., 2005), which initiate cytoprotective measures collectively termed the unfolded protein response (UPR; Harding et al., 1999; Lu et al., 2004). The phosphorylation of eukaryotic initiation factor 2α by PERK reduces global protein synthesis, but results in preferential translation of selected messenger RNAs (mRNAs), including ATF4 (activating transcription factor 4; Karpinski et al., 1992; Harding et al., 1999; Koumenis et al., 2002; Ameri et al., 2004). Activating transcription factor 4 regulates the integrated stress response, a gene expression program involved in oxidative stress, amino acid synthesis, differentiation, metastasis, angiogenesis (Harding et al., 2003) and drug resistance (Rzymski et al., 2009). Acute HIF1α and chronic ATF4-dependent gene expression programs are temporarily separated in cancer cells during exposure to hypoxia (Rzymski et al., 2008).
Autophagy is an evolutionarily conserved process involving the formation of double-membraned vesicles called autophagosomes. The autophagosome vesicles encapsulate cytoplasmic contents, including organelles, and deliver this cargo to the lysosome for degradation. Although in yeast autophagy is mainly involved in adaptation to starvation, in multicellular organisms this route has emerged as a multifunctional pathway involved in a variety of additional processes, such as programmed cell death (Kroemer and Jaattela, 2005), and the response to microenvironmental or cellular stress (Kuma et al., 2004; Shimizu et al., 2004; Shintani and Klionsky, 2004; Lum et al., 2005). It has been shown that prolonged hypoxia can induce autophagic cell death in apoptosis-competent cells, through a mechanism involving HIF1α and BCL2/adenovirus E1B 19-kDa protein-interacting protein 3 (Azad et al., 2008). However, additional HIF1α-independent molecular events are necessary for cell death observed under severe hypoxia (Papandreou et al., 2005). Recently mitochondrial autophagy also has been shown to represent an adaptive metabolic response maintaining redox homeostasis and cell survival under hypoxia (Semenza, 2008).
One of the key steps in autophagosome generation is the association of a lipidated form of MAP1LC3B (light chain 3; LC3B) with the pre-autophagosomal structure. Previously, we showed that the inhibition of the 26S proteasome by Bortezomib leads to the accumulation of misfolded proteins, ER stress and induction of autophagy (Milani et al., 2009). Induction of autophagy by Bortezomib was crucially dependent on the proteasomal stabilization of ATF4 and upregulation of LC3B and protected Bortezomib-treated cells from cell death. Recently, we also identified that lysosomal-associated membrane protein 3 was induced by the PERK/eIF2alpha/ATF4 arm of the UPR and hypoxia (Mujcic et al., 2009).
In this study, we show that ATF4 protects against cell death in breast cancer cells under severe hypoxia. Expansion of autophagy is critically dependent on transcriptional upregulation of the key autophagy regulator LC3B in response to direct binding of ATF4 to a cyclic AMP response element binding site in the LC3B promoter and protects cells from the hypoxic cell death. Thus, the activation of ATF4 provides a key survival pathway in response to hypoxia by linking the UPR to autophagy. This is potentially important in the survival of cancer cells in vivo and normal tissue after vascular occlusion. While this work was in submission, similar results were observed by Wouters’ group (Rouschop et al., 2009), but there were differences, which are also discussed.
Identification of hypoxia (<0.01% O2) and ATF4-dependent genes by a microarray analysis
In wild-type MCF7 cells exposure to 0.1% O2 was not sufficient for the significant induction of ATF4, however, hypoxic stress (<0.01% O2) led to a rapid and sustained accumulation of ATF4 protein within 24 h, which is consistent with previous reports (Scheuner et al., 2001; Figure 1a). The MCF7 cells transiently transfected with small interfering RNA (siRNA) duplexes, specific to ATF4, showed a significant 93.7±1% reduction in ATF4 mRNA levels under normoxia and a 93±3% reduction after 24 h of hypoxia, as measured by quantitative RT–PCR (qPCR) and reduction in the hypoxic ATF4 protein levels (Figure 1b). Consistent with this, ATF4 knockdown resulted in a greatly reduced gene expression level of the ATF4 target gene CHOP, when MCF7 cells were exposed to hypoxia for 24 h (Figure 1b). Illumina Human-6 BeadChip v1 gene expression microarrays were used to identify genes that may be transcriptionally regulated by ATF4 and essential for the adaptive response to hypoxic stress. The MCF7 cells were transiently transfected with siRNA specific for ATF4 or scrambled control siRNA (SCR) followed by exposure to either normoxia or hypoxia. We focused first on genes differentially expressed between normoxia and hypoxia. Of the approximately 47 000 complimentary DNA transcripts represented on the microarray, expression levels of 403 increased significantly after 24 h of hypoxia (⩾twofold with ⩽5% false discovery rate using Limma analysis with Benjamin and Hochberg correction for multiple testing; Supplementary Information). Interestingly, many of the upregulated genes had roles in protein metabolism, protein modification and homeostasis. In contrast, of the 274 genes that were significantly repressed by 24 h hypoxia (0.5-fold, 5% false discovery rate), many had roles in apoptosis and cell cycle control. Genes upregulated after 48 h hypoxia (Supplementary Information) and a subset of genes that showed a significant difference in response between 24 h and 48 h hypoxia (Supplementary Information) were also identified.
To directly determine the role of ATF4 in hypoxia-induced gene expression, we compared gene expression in ATF4 siRNA and SCR siRNA knockdown cells under normoxia and 24 h hypoxia. Candidate genes were classified as those exhibiting a change in expression after 24 h hypoxia but were unaffected by ATF4 knockdown in normoxia. All genes that met these requirements are displayed (Supplementary Information).
We observed 412 genes that were regulated under hypoxia in an ATF4-dependent manner. A total of 224 genes were repressed in ATF4 knockdown cells after 24 h of hypoxia, therefore representing genes induced by ATF4. Of these, 80% (179) genes were induced at 24 h hypoxia in the controls. The other 188 of 412 genes were specifically induced in cells lacking ATF4 at 24 h hypoxia, and therefore represented genes specifically repressed by ATF4. Of this number, 63% (118) genes were downregulated by 24 h of hypoxia in the controls.
Validation of the microarray results
We next used qPCR analysis to validate 25 of the candidate genes regulated by ATF4 under severe hypoxia (Table 1). With the exception of BAG1 and PDK3, 23 of the 25 genes demonstrated significant (P<0.05) upregulation under hypoxia. The expression levels of 19 of these genes were repressed in ATF4 knockdown cells exposed to 24 h hypoxia, whereas expression of five genes was specifically increased in ATF4 knockdown cells under hypoxia. The degree of correlation between expression microarrays and qPCR fold changes indicated a high level of reproducibility and reliability of our microarray data sets (Spearman's ρ=0.82, P<0.0001 for hypoxia regulated genes and ρ=0.72, P<0.0001 for ATF4-regulated genes).
ATF4 upregulates LC3B under severe hypoxia
Our initial microarray analysis identified LC3B, a major component of autophagic vesicles as a gene robustly upregulated by ATF4 under hypoxia. By using qPCR (Figure 2a) and western blot analysis (Figure 2b), we determined that LC3B levels increased markedly (22-fold) after 48 h of severe hypoxia in MCF7 cells. The induction of phosphor-eIF2alpha preceded ATF4 expression, which was then followed by LC3B expression (Figure 2b) Similar induction levels of LC3B were obtained for HCT116, HeLa and HEF551 (Detroit 551) cells exposed to severe hypoxia (Figure 2c).
In a recent study, we demonstrated that the kinetics of the hypoxic activation of the UPR might be influenced by other factors (for example, the surface of tissue culture dishes (plastic, which retains oxygen vs glass; Mujcic et al., 2009), preconditioning of media and type of incubation (palladium catalyst vs no catalyst). In our experience, 0.1% oxygen levels induced HIF1 response, but did not result in the upregulation of ATF4. 0.1% O2 did not upregulate LC3B mRNA levels, although PHD3 and BCL2/adenovirus E1B 19-kDa protein-interacting protein 3 (HIF1α-dependent genes) were induced (Figure 2d). We also did not observe induction of the LC3B protein in cells exposed to 0.1% hypoxia (data not shown).
The MCF7 cells were further transiently transfected with siRNA specific to ATF4 and exposed to 24 h severe hypoxia, resulting in a significant 65% reduction in LC3B mRNA levels after 24 h hypoxia, as measured by qPCR (Figure 3a) and confirmed by western blotting (Figure 3b). Hypoxic induction of LC3B was independent of two other branches of the UPR, the IRE1/XBP1 and ATF6 pathways, as it could not be repressed by transient transfection with siRNA specific to transcription factors: XBP1 (Figure 3c) and ATF6 (Figure 3d).
Induction of LC3B by other inducers of UPR
We next investigated whether induction of LC3B was a unique response to hypoxia or if it occurred in response to other UPR inducers. The LC3B mRNA levels were increased in MDA MB 231, LS174T and HCT116 cells treated with the ER Ca2+ ATPase inhibitor, Thapsigargin, or the N-linked glycosylation inhibitor, Tunicamycin, as measured by qPCR (Figure 4a). Notably, two cell lines (LS174T and HCT116) did not induce GADD34, which inhibits the ATF4 pathway, and these exhibited the greatest fold induction of LC3B. These results were further corroborated by the increased protein level and processing of LC3B in MCF7 cells treated by Thapsigargin, preceded by PERK phosphorylation and ATF4 induction (Figure 4b). The LC3B protein levels were reduced in ATF4 siRNA transfected MCF7 cells as compared with control Thapsigargin-treated cells (Figure 4c).
Specific protein–DNA interactions within the promoter of LC3B under hypoxia
Promoter Inspector software (Genomatix Software GmbH, Munich, Germany) revealed that the 5′ untranslated region of the LC3B promoter region contains putative DNA-regulatory elements similar to the consensus composite ATF4/XBP1-binding site (Figure 5a). To characterize the cis-regulatory elements and trans-acting factors that have a role in the transcriptional regulation of the LC3B gene under hypoxia, we performed a series of electro-mobility shift assays (EMSAs). Synthetic biotin-labelled oligonucleotides containing putative unfolded protein response elements (UPREs/ATF4) in the LC3B promoter were tested for protein binding with nuclear lysates from normoxic and hypoxic cells (Figure 5b). The UPRE/ATF4 DNA fragment bound strongly to proteins from the hypoxic (24 h) cell lysate (Figure 5b). The binding of proteins to this fragment was specific since competition with 100-fold excess of the same unlabelled oligonucleotide completely prevented binding of the labelled fragment. Mutated variants of the UPRE/ATF4 were also tested for protein binding. Although the first mutated oligonucleotide (MUT1) efficiently bound proteins from the hypoxic lysate, single nucleotide substitution in MUT2 and double substitution in MUT3 completely disrupted binding of the hypoxic extract to the putative stress elements in EMSA. We investigated further whether specific binding to the putative UPRE/ATF4 elements involved ATF4. Preincubation of the hypoxic 24 h extract with ATF4 antibodies disrupted binding to the putative UPRE/ATF4 elements in EMSA (Supplementary Figure 1). Furthermore, we also analysed the binding of ATF4 to the promoter of LC3B in vivo by using chromatin immunoprecipitations (ChIP; Figures 5c and d). Chromatin samples from cells grown under normoxia and 24 h hypoxia were immunoprecipitated with antibodies against ATF4 and control rabbit antisera. Recovered DNA was amplified by quantitative PCR using primers covering different regions of the LC3B promoter, LC3B non-promoter and intergenetic regions and the occupancy was calculated as percentages of IP samples relative to inputs. The ATF4 occupancy was increased in the LC3B promoter +541/+656 relative to the transcription start site (Figure 5c). There was also higher ATF4 recruitment to the LC3B promoter activated by 24 h hypoxia (Figure 5d).
Severe hypoxia induces accumulation of autophagosomes
As severe hypoxia induces LC3B through ATF4, we next investigated the ability of hypoxia to induce autophagosomes. A specific 16-kDa band was detected under hypoxia by western blotting, which corresponded to LC3B-II, the membrane-bound form of LC3B. The 18-kDa band, corresponding to the cytosolic form LC3B-I was barely detectable in most hypoxia experiments, indicating a high level of autophagy in these cells (compare with Figures 2b and c). The high levels of the membrane-bound form compared with precursor indicate a high level of autophagy. Owing to the degradation of autophagosomes, to assess the amount of LC3B-II produced cells were treated with Bafilomycin A1—a specific inhibitor of vacuolar-type H+-ATPase that prevents maturation of autophagic vacuoles (Yamamoto et al., 1998; Figure 6a). This showed the greatly increased amount of LC3B produced in severe hypoxia and its ATF4 dependence and indicated increased turnover of LC3B.
Increased accumulation of autophagosomes under hypoxia is dependent on upregulation of LC3B by ATF4
To investigate whether the maintenance of autophagosomes under severe hypoxia was dependent on ATF4, we used synthetic siRNA duplexes to specifically inhibit ATF4 expression and thus upregulation of LC3B. Control MCF7 cells showed a significant intracellular increase in the LC3B-positive foci after 48-h severe hypoxia, the hallmark of increased autophagosome formation, whereas ATF4 knockdown cells failed to induce positive foci (Figure 6b). The percentage of autophagy-positive cells was counted and a significantly (P=0.0007) higher percentage of positive cells was observed among the SCR control cells (62.3±10.2%) compared with the ATF4 knockdown (6.6±1.8%) population. Moreover, we observed a marked inhibition of induction of fluorescence measured quantitatively, with ATF4 knockdown (Supplementary Figure 2A). Hypoxic cells also showed a significantly increased lysosomal mass as determined by LysoTracker (Invitrogen, Paisley, UK) staining (Supplementary Figure 2B), consistent with elevated autophagic degradation. In contrast, the lysosomal content in ATF4 and LC3B knockdown cells under hypoxia was significantly reduced. Thus, these results suggest that LC3B upregulation by ATF4 is necessary for the accumulation and expansion of autophagosomes that occurs under hypoxia.
Autophagy increases survival of cells under severe hypoxia
To determine the role of increased autophagy in cells exposed to hypoxic stress, we specifically reduced LC3B levels using siRNA (Figure 7; Supplementary Figure 2C). The LC3B knockdown cells showed a decreased survival compared with SCR control when exposed to severe hypoxia as shown using a clonogenic assay (Figure 7a), whereas normoxic cells remained unaffected by LC3B knockdown over a 48-h period.
To assess the role of increased death in these effects, MCF7 cells transiently transfected with siRNA duplexes against LC3B and compared with control cells, showed a marked increase in apoptosis as measured by protease release from cells that have lost membrane integrity (Figure 7b) and by decrease in number of viable cells as shown by fluorescence-activated cell sorting (Supplementary Figure 3A). We also found that the loss of LC3B leads to a significant increase in the percentage of annexin V-positive cells and in the caspase activity, suggesting that the increase in cell death was at least in part due to apoptosis (Figures 7c and d). Consistently, we observed an increased cleavage of poly (ADP-ribose) polymerase by immunoblotting in MDA231 knocked down for LC3B (Supplementary Figure 3B) under hypoxia. Moreover, inhibition of the initiation of the formation of the autophagasome by downregulation of Beclin 1 using siRNA transfection in MCF7 cells resulted in increased poly (ADP-ribose) polymerase cleavage and reduced viability under severe hypoxia compared with SCR control cells (Supplementary Figure 4).
Co-localization of autophagy and the UPR in hypoxic tumours
To show the expression of UPR and autophagy markers in tumour tissues, U87 xenografts were treated with Bevacizumab to increase hypoxia and stained with antibody specific to CHOP and LC3B (Supplementary Figure 4). Specific nuclear staining for CHOP was observed in the tumour cells, which localized near the area of tumour undergoing necrosis and only scattered nuclei were observed in the viable tumour far from the necrotic area (Supplementary Figure 5A). Serial sections of the same specimen were stained for LC3B showing the cytoplasmic signal, specific for LC3B, in the tumour cells; moreover the granular pattern, consistent with the presence of autophagosomes, was observed and was stronger in the perinecrotic area (Supplementary Figure 5B).
Despite increasing interest in PERK/eIF2α/ATF4 pathway as a new target for anti-cancer modalities, a role of ATF4 in cancer cells is still poorly characterized. Previous reports have established that the transcription factor ATF4 has a regulatory role in the cellular response to ER stress and amino-acid starvation (Averous et al., 2004; Ohoka et al., 2005). Activating transcription factor 4 transcriptionally upregulates the expression of ER chaperones and also the proapoptotic genes: CHOP and SKIP3 (Harding et al., 2000; Blais et al., 2006). Thus, ATF4 regulates both prosurvival and proapoptotic signalling during ER stress, however, the pathways that can shift this balance either towards survival or apoptosis still need to be characterized in detail.
Earlier studies showed the importance of the PERK pathway for tumour progression and the response to UPR (Blais et al., 2006). The adaptation of cells to chronic hypoxic stress requires translational attenuation by phosphorylation of eukaryotic initiation factor 2α by PERK (Mizushima et al., 2004; Wu et al., 2006). Severe hypoxia upregulates ATF4 in a PERK-dependent manner and ATF4-deficient mouse embryonic fibroblasts are sensitive to hypoxia (Koumenis et al., 2002). However, the mechanisms through which ATF4 promotes survival under hypoxia have remained unknown. To investigate this, we used synthetic siRNA duplexes to inhibit ATF4 expression in breast cancer cells. To the best of our knowledge, our study provides the first whole-genome analysis of ATF4-dependent genes in cancer cells regulated under severe hypoxia.
We demonstrated that the upregulation of LC3B by ATF4 under severe hypoxia was required to sustain an increased level of autophagosomes and both LC3B and ATF4 contributed to the increased lysosomal mass under hypoxia.
In contrast to the well-characterized mechanism of autophagy, activation during nutritional stress, which does not require de novo synthesis of the LC3B (He et al., 2003), we demonstrated that hypoxia-induced maintenance of autophagy is attained by the transcriptional upregulation of LC3B by ATF4. A minimal increase in LC3B level observed under acute hypoxia in ATF4 siRNA-transfected cells, compared with that at 12 and 24 h, was within experimental variation and notably, LC3B I levels were not increased.
This indicates that when cells are under hypoxic stress, LC3B levels can be rate limiting with regard to autophagic degradation and increased turnover of LC3B. The expression of LC3B required prolonged hypoxia and previously we established that HIF1-stabilizing agents, such as DMOG, DFO and cobalt chloride were not sufficient for LC3B inductions (Milani et al., 2009).
We studied several human colon and breast cancer cell lines, non-transformed human embryonic cells, and murine cells, all showing the same marked induction of LC3B, implying a general response pathway. The effect of the UPR on LC3B transcription was similar, using other mechanisms, such as thapsigargin, tunicamycin and proteasomal inhibitors (Milani et al., 2009) to induce the UPR. While this study was in submission, similar results were published, indicating that increased expression of LC3B under severe hypoxia was regulated by PERK-dependent activation of ATF4 and CHOP (Rouschop et al., 2009). Rouschop et al. (2009) performed the ATF4 ChIP assays followed by promoter-specific qPCR and demonstrated enrichment of LC3B (−458 to −376), however, the enrichment was observed under both normoxic and hypoxic conditions. In contrast, we used primers covering a larger portion of the proximal promoter. We detected the highest enrichment within the first exon of LC3B (+541 to +656) activated by the hypoxic exposure and our results were further corroborated by the EMSA. We observed some mediocre enrichment within the regions upstream of the LC3B transcription start site and we could not exclude the possibility that ATF4-binding site detected by Rouschop et al. (2009) could be also functional. Functionality of these binding sites may be influenced by relative amounts of other transcription factors that dimerize with ATF4 and by the severity and origin of stress. Rouschop et al. (2009) also demonstrated the binding of CHOP to the LC3B promoter. Interestingly, we observed that although LC3B was induced by the UPR in each cell line tested, there were substantial differences in the extent of induction and relative induction to CHOP, demonstrating the complexity and heterogeneity of the response.
Thus, our results provide a new mechanistic link between the UPR and autophagy. The increased pool of LC3B was readily converted from LC3B-I to LC3B-II in response to severe hypoxia, resulting in the accumulation of LC3B-II in punctuate autophagosomes. We have shown LC3B is critical in maintaining autophagy, however, we did not see induction of other genes involved in the initiation of autophagy at RNA level, such as Beclin1, hAPG7, hAPG12 or hAPG5. Thus, the mechanism of autophagy initiation in severe hypoxia still needs evaluation. Importantly, cells with impaired eukaryotic initiation factor 2α signalling could form autophagosomes during short exposures to hypoxia (Rouschop et al., 2009).
The relationship between autophagy and ER stress in mammalian cancer cells is not well defined (Yorimitsu et al., 2006). In yeast, accumulation of misfolded proteins in the ER induces an autophagic response (Bernales et al., 2006) and sequestration of ER structures into autophagosomes helps to maintain homeostasis during the accumulation of misfolded proteins (Ogata et al., 2006). In agreement with our study, previous reports have identified the eIF2α/PERK pathway as being required for autophagy (Kouroku et al., 2007). Contradictory to our findings, one other report identified the IRE1–c-Jun N-terminal kinase pathway as required for autophagy activation after ER stress in SK-N-SH neuroblastoma cells (Ogata et al., 2006).
Rouschop et al. (2009) convincingly demonstrated that PERK signalling was necessary for the accumulation and increased turnover of LC3B under hypoxia, which could indicate a high autophagic flux. We further extended this hypothesis by showing that activation of ATF4 and increased levels of LC3B directly increased number of autophagosomes and acidic compartments in hypoxic cells.
Although it remains controversial whether autophagy serves a protective or detrimental role, we have shown that inactivation of autophagy by loss of LC3B and Beclin 1 increased susceptibility of cancer cells to metabolic consequences of the hypoxic stress. Rouschop et al. (2009) used pharmacological inhibitors of autophagy and similar effects were observed. Thus, we have shown that both the initiation of autophagy by Beclin 1 and the expansion of autophagy by LC3B were essential for maintaining cell survival after exposure to severe hypoxia.
Our experiments also extend the study of Bi et al. (2005), and show a major role of ATF4 in regulating the gene expression programme after hypoxia, although ATF6 and XBP1 are also induced by this pathway. This suggests that the HIF and BCL2/adenovirus E1B 19-kDa protein-interacting protein 3 pathway may be less important under chronic and severe hypoxia and our results show that ATF4 predominates in these conditions. Such conditions may occur frequently in tumours with inadequate or intermittent blood flow and in areas furthest from vessels. Rouschop et al. (2009) demonstrated that autophagy was strongly associated with hypoxic (pimonidazole positive) tumour regions. We also observed a hypoxic gradient of autophagy towards necrotic regions with a striking co-localization of the granular LC3B pattern and the UPR marker CHOP within perinecrotic regions of the hypoxic tumours.
These results also extended our previous study on the resistance to proteasomal inhibitor,Bortezomib, in which ATF4-dependent expression of LC3B was also a limiting factor for the maximum induction of autophagy (Milani et al., 2009). Our data suggests that ATF4 antagonists would be worth investigating for anti-tumour effects and we are currently screening for such compounds.
Materials and methods
Thapsigargin was used at 300 nM and Bafilomicin was used at 100 nM concentration (Calbiochem, Nottingham, UK).
The cell lines MCF7, MDA231, HCT116, HeLa and HEF (Detroit 551) were provided by Cancer Research UK (London, UK). Cells were maintained in Dulbecco's modified Eagle medium (with 4.5 mg/ml glucose) supplemented with 10% (vol/vol) fetal calf serum, penicillin (100 U/ml) and streptomycin (100 μg/ml), and 4 mM L-glutamine (Gibco, Paisley, UK).
A humidified gas-sorted anoxic incubator-gloved box (InVivo2 400; Ruskinn, Bridgend, UK) was used for hypoxic experiments resulting in 5% H2, 5% CO2 and 90% N2. Humidity level was greater than 90%. Severe hypoxic conditions (<0.01% O2) were controlled using an anaerobic indicator (Oxoid, Cambridge, UK). Palladium catalysts were used to scavenge traces of oxygen. Moderate hypoxia (0.1% O2) was obtained using gas mixer (Ruskinn).
siRNA treatment of cells and transfection procedures
Transfections of siRNA duplexes diluted to give a final concentration of 20 nM in Opti-Mem I (Invitrogen, Paisley, UK) were performed with cells at 30–40% confluency using Oligofectamine transfection reagent (Invitrogen).
Gene expression microarrays processing and analysis
Microarray analysis of gene expression was performed in triplicate using the Illumina BeadChip system (Illumina, Little Chesterford, UK). In brief, 200 ng of RNA was used to perform in vitro transcription amplification using the Illumina RNA amplification kit (Ambion, Austin, TX, USA). Amplified RNA (1.5 μg) was hybridized to the Sentrix Human-6 Expression BeadChips (Illumina) containing probes for 48 000 transcripts. All labelling, hybridization and scanning steps were performed according to the manufacturers’ instructions. Quality assessment on the array was done using the Bioconductor package BeadExplorer (http://www.bioconductor.org/) (which facilitates visualization of control data and comparison of samples across and between chips.
Whole cell lysates were resolved by sodium dodecyl suphate–polyacrylamide gel electrophoresis, electroblotted onto polyvinylidene fluoride membrane (Millipore, Watford, UK) and probed by indicated antibodies. Horseradish peroxidase-conjugated anti-rabbit or anti-mouse, goat or mouse secondary antibodies (Dako, Ely, UK) were used with ECL Plus system (Amersham Biosciences, Little Chalfont, UK) to visualize immunoreactive bands. The rabbit polyclonal antibody against ATF4 was obtained from Santa Cruz Biotechnology (Heidelberg, Germany); the mouse monoclonal antibody against MAP1LC3 was purchased from NanoTools (Teningen, Germany); the rabbit monoclonal antibody against Phospho-eIF2a and the rabbit monoclonal antibody against poly (ADP-ribose) polymerase were obtained from Cell Signaling Technologies (Danvers, MA, USA). The β-tubulin antibody was purchased from Sigma and anti-PERK-phosphorylated (Thr980) was obtained from Biolegend (San Diego, CA, USA).
Real-time quantitative PCR
In real-time quantitative PCR experiments, the methods used by us for extraction, quantification and evaluation of the quality of RNA were as previously described (Rzymski et al., 2008), the data of real-time quantitative PCR were normalized to tubulin alpha 6 and β-actin (for the list of primers, see Supplementary Materials).
Cells were fixed in −20 °C cold methanol for 15 min, washed with phosphate-buffered saline and blocked with 1% bovine serum in phosphate-buffered saline (pH 7.5) for 30 min followed by overnight incubation with primary antibody at 4 °C. Secondary fluorescent Alexa Fluor 488 anti mouse antibody (Invitrogen) were used at dilution 1:500 in blocking buffer for 2 h. The nuclear staining was performed using Vectashield mounting medium with 4′,6-diamidino-2-phenylindole (Vector Laboratories, Burlingame, CA, USA). Cells were visualized by using the confocal microscopy (Carl Zeiss LSM510 META with supplied software, Cambridge, UK). Images from five random fields were taken for each condition. Positive cells were counted for each field. Positive cells were considered those that showed more than seven fluorescent dots (Sarkar et al., 2007).
LysotrackerRed fluorescent dye (Invitrogen) was used on cells grown in monolayer for 30 min at 50 nM in the dark. Cells were briefly washed thrice with fresh medium and visualized using a microscope.
Cell death assay
This was performed by using MultiTox-Fluor Multiplex Cytotoxicity Assay (Promega, Madison, WI, USA), according to manufacturer's instructions.
To determine the caspase activity was used the luminescent ELISA-like assay from Promega (Caspase 3/7 Glo) according to manufacturer's instructions (Promega).
Electrophoresis mobility shift assays (EMSA)
These assays were performed using LightShift Chemiluminescence EMSA kit (Pierce, Rockford, IL, USA) following the manufacturer's instructions. The 5′-biotin-labelled oligonucleotides are described in Supplementary Materials.
Chromatin immunoprecipitation (ChIP)
The ChIP assays were performed using EZ ChIP Chromatin Immunoprecipitation Kit (Upstate, Watford, UK) according to manufacturers’ instructions. DNA was cross-linked with formaldehyde (1% final concentration for 10 min). Cells were lysed in the presence of protease inhibitors and lysates were sonicated using Bioruptor from Diagenode (Cambridge, UK) to shear DNA (average size 200–600 bp). Immunoprecipitations were performed with 5 μg of anti-ATF4 (Santa Cruz Biotechnology, Heidelberg, Germany) and 5 μg of normal rabbit IgG (Upstate). Complexes were collected with protein A/G agarose (Upstate).
All experiments were run in triplicate or quadruplicate, and each experiment was repeated two or three times. Data from real-time quantitative PCR were compared by Student's t-test of treated vs control samples. The criterion for significance was a P-value <0.05 for all comparisons.
Ameri K, Lewis CE, Raida M, Sowter H, Hai T, Harris AL . (2004). Anoxic induction of ATF-4 through HIF-1-independent pathways of protein stabilization in human cancer cells. Blood 103: 1876–1882.
Averous J, Bruhat A, Jousse C, Carraro V, Thiel G, Fafournoux P . (2004). Induction of CHOP expression by amino acid limitation requires both ATF4 expression and ATF2 phosphorylation. J Biol Chem 279: 5288–5297.
Azad MB, Chen Y, Henson ES, Cizeau J, McMillan-Ward E, Israels SJ et al. (2008). Hypoxia induces autophagic cell death in apoptosis-competent cells through a mechanism involving BNIP3. Autophagy 4: 195–204.
Bernales S, McDonald KL, Walter P . (2006). Autophagy counterbalances endoplasmic reticulum expansion during the unfolded protein response. PLoS Biol 4: e423.
Bi M, Naczki C, Koritzinsky M, Fels D, Blais J, Hu N et al. (2005). ER stress-regulated translation increases tolerance to extreme hypoxia and promotes tumor growth. EMBO J 24: 3470–3481.
Blais JD, Addison CL, Edge R, Falls T, Zhao H, Wary K et al. (2006). Perk-dependent translational regulation promotes tumor cell adaptation and angiogenesis in response to hypoxic stress. Mol Cell Biol 26: 9517–9532.
Graeber TG, Osmanian C, Jacks T, Housman DE, Koch CJ, Lowe SW et al. (1996). Hypoxia-mediated selection of cells with diminished apoptotic potential in solid tumours. Nature 379: 88–91.
Harding HP, Zhang Y, Ron D . (1999). Protein translation and folding are coupled by an endoplasmic-reticulum-resident kinase. Nature 397: 271–274.
Harding HP, Zhang Y, Bertolotti A, Zeng H, Ron D . (2000). Perk is essential for translational regulation and cell survival during the unfolded protein response. Mol Cell 5: 897–904.
Harding HP, Zhang Y, Zeng H, Novoa I, Lu PD, Calfon M et al. (2003). An integrated stress response regulates amino acid metabolism and resistance to oxidative stress. Mol Cell 11: 619–633.
Harris AL . (2002). Hypoxia—a key regulatory factor in tumour growth. Nat Rev Cancer 2: 38–47.
He H, Dang Y, Dai F, Guo Z, Wu J, She X et al. (2003). Post-translational modifications of three members of the human MAP1LC3 family and detection of a novel type of modification for MAP1LC3B. J Biol Chem 278: 29278–29287.
Karpinski BA, Morle GD, Huggenvik J, Uhler MD, Leiden JM . (1992). Molecular cloning of human CREB-2: an ATF/CREB transcription factor that can negatively regulate transcription from the cAMP response element. Proc Natl Acad Sci USA 89: 4820–4824.
Koumenis C, Naczki C, Koritzinsky M, Rastani S, Diehl A, Sonenberg N et al. (2002). Regulation of protein synthesis by hypoxia via activation of the endoplasmic reticulum kinase PERK and phosphorylation of the translation initiation factor eIF2alpha. Mol Cell Biol 22: 7405–7416.
Kouroku Y, Fujita E, Tanida I, Ueno T, Isoai A, Kumagai H et al. (2007). ER stress (PERK/eIF2alpha phosphorylation) mediates the polyglutamine-induced LC3 conversion, an essential step for autophagy formation. Cell Death Differ 14: 230–239.
Kroemer G, Jaattela M . (2005). Lysosomes and autophagy in cell death control. Nat Rev Cancer 5: 886–897.
Kuma A, Hatano M, Matsui M, Yamamoto A, Nakaya H, Yoshimori T et al. (2004). The role of autophagy during the early neonatal starvation period. Nature 432: 1032–1036.
Lu PD, Jousse C, Marciniak SJ, Zhang Y, Novoa I, Scheuner D et al. (2004). Cytoprotection by pre-emptive conditional phosphorylation of translation initiation factor 2. EMBO J 23: 169–179.
Lum JJ, Bauer DE, Kong M, Harris MH, Li C, Lindsten T et al. (2005). Growth factor regulation of autophagy and cell survival in the absence of apoptosis. Cell 120: 237–248.
Milani M, Rzymski T, Mellor HR, Pike L, Bottini A, Generali D et al. (2009). The role of ATF4 stabilization and autophagy in resistance of breast cancer cells treated with Bortezomib. Cancer Res 69: 4415–4423.
Mizushima N, Yamamoto A, Matsui M, Yoshimori T, Ohsumi Y . (2004). in vivo analysis of autophagy in response to nutrient starvation using transgenic mice expressing a fluorescent autophagosome marker. Mol Biol Cell 15: 1101–1111.
Mujcic H, Rzymski T, Rouschop KM, Koritzinsky M, Milani M, Harris AL et al. (2009). Hypoxic activation of the unfolded protein response (UPR) induces expression of the metastasis-associated gene LAMP3. Radiother Oncol 92: 450–459.
Ogata M, Hino S, Saito A, Morikawa K, Kondo S, Kanemoto S et al. (2006). Autophagy is activated for cell survival after endoplasmic reticulum stress. Mol Cell Biol 26: 9220–9231.
Ohoka N, Yoshii S, Hattori T, Onozaki K, Hayashi H . (2005). TRB3, a novel ER stress-inducible gene, is induced via ATF4-CHOP pathway and is involved in cell death. EMBO J 24: 1243–1255.
Olive PL, Banath JP, Durand RE . (2002). The range of oxygenation in SiHa tumor xenografts. Radiat Res 158: 159–166.
Papandreou I, Krishna C, Kaper F, Cai D, Giaccia AJ, Denko NC . (2005). Anoxia is necessary for tumor cell toxicity caused by a low-oxygen environment. Cancer Res 65: 3171–3178.
Rouschop KM, van den Beucken T, Dubois L, Niessen H, Bussink J, Savelkouls K et al. (2009). The unfolded protein response protects human tumor cells during hypoxia through regulation of the autophagy genes MAP1LC3B and ATG5. J Clin Invest 120: 127–141.
Rzymski T, Paantjens A, Bod J, Harris AL . (2008). Multiple pathways are involved in the anoxia response of SKIP3 including HuR-regulated RNA stability, NF-kappaB and ATF4. Oncogene 27: 4532–4543 .
Rzymski T, Milani M, Singleton DC, Harris AL . (2009). Role of ATF4 in regulation of autophagy and resistance to drugs and hypoxia. Cell Cycle 8: 3838–3847.
Sarkar S, Perlstein EO, Imarisio S, Pineau S, Cordenier A, Maglathlin RL et al. (2007). Small molecules enhance autophagy and reduce toxicity in Huntington's disease models. Nat Chem Biol 3: 331–338.
Scheuner D, Song B, McEwen E, Liu C, Laybutt R, Gillespie P et al. (2001). Translational control is required for the unfolded protein response and in vivo glucose homeostasis. Mol Cell 7: 1165–1176.
Semenza GL . (2003). Targeting HIF-1 for cancer therapy. Nat Rev Cancer 3: 721–732.
Semenza GL . (2008). Mitochondrial autophagy: life and breath of the cell. Autophagy 4: 534–536.
Shimizu S, Kanaseki T, Mizushima N, Mizuta T, Arakawa-Kobayashi S, Thompson CB et al. (2004). Role of Bcl-2 family proteins in a non-apoptotic programmed cell death dependent on autophagy genes. Nat Cell Biol 6: 1221–1228.
Shintani T, Klionsky DJ . (2004). Autophagy in health and disease: a double-edged sword. Science 306: 990–995.
Tu BP, Weissman JS . (2004). Oxidative protein folding in eukaryotes: mechanisms and consequences. J Cell Biol 164: 341–346.
Wenger RH, Gassmann M . (1996). Little difference. Nature 380: 100.
Wouters BG, Brown JM . (1997). Cells at intermediate oxygen levels can be more important than the ‘hypoxic fraction’ in determining tumor response to fractionated radiotherapy. Radiat Res 147: 541–550.
Wu J, Dang Y, Su W, Liu C, Ma H, Shan Y et al. (2006). Molecular cloning and characterization of rat LC3A and LC3B—two novel markers of autophagosome. Biochem Biophys Res Commun 339: 437–442.
Yamamoto A, Tagawa Y, Yoshimori T, Moriyama Y, Masaki R, Tashiro Y . (1998). Bafilomycin A1 prevents maturation of autophagic vacuoles by inhibiting fusion between autophagosomes and lysosomes in rat hepatoma cell line, H-4-II-E cells. Cell Struct Funct 23: 33–42.
Yorimitsu T, Nair U, Yang Z, Klionsky DJ . (2006). Endoplasmic reticulum stress triggers autophagy. J Biol Chem 281: 30299–30304.
We thank Dr S Colella for help with initial microarray data analysis. We also thank Dr P Thomas for his help in using the confocal microscope and Dr D Singleton for critical revision of the paper.
The authors declare no conflict of interest.
Supplementary Information accompanies the paper on the Oncogene website
About this article
Cite this article
Rzymski, T., Milani, M., Pike, L. et al. Regulation of autophagy by ATF4 in response to severe hypoxia. Oncogene 29, 4424–4435 (2010). https://doi.org/10.1038/onc.2010.191
Foot-and-mouth disease virus induces PERK-mediated autophagy to suppress the antiviral interferon response
Journal of Cell Science (2021)
Circulation Research (2021)
DIM-C-pPhtBu induces lysosomal dysfunction and unfolded protein response - mediated cell death via excessive mitophagy
Cancer Letters (2021)
A KDM6 inhibitor potently induces ATF4 and its target gene expression through HRI activation and by UTX inhibition
Scientific Reports (2021)