The heterodimeric HIF (hypoxia-inducible factor)-1 is a transcriptional master regulator of several genes involved in mammalian oxygen homoeostasis. Besides the well described regulation of the HIF-1α subunit via hydroxylation-mediated protein stability in hypoxia, there are several indications of an additional translational control of the HIF-1α mRNA, especially after growth factor stimulation. We identified an interaction of CPEB (cytoplasmic polyadenylation-element-binding protein) 1 and CPEB2 with the 3′-UTR (untranslated region) of HIF-1α mRNA. Overexpression of CPEB1 and CPEB2 affected HIF-1α protein levels mediated by the 3′-UTR of HIF-1α mRNA. Stimulation of neuroblastoma SK-N-MC cells with insulin and thus activation of endogenous CPEBs increased the expression of a luciferase reporter gene fused to the 3′-UTR of HIF-1α as well as endogenous HIF-1α protein levels. This could be abrogated by treating the cells with CPEB1 or CPEB2 siRNAs (short interfering RNAs). Injection of HIF-1α cRNA into Xenopus oocytes verified the elongation of the poly(A)+ (polyadenylated) tail by cytoplasmic polyadenylation. Thus CPEB1 and CPEB2 are involved in the regulation of HIF-1α following insulin stimulation.
Adaptation of cells towards a decreased oxyen supply involves hypoxia-mediated gene expression which is mainly initiated by the HIF (hypoxia-inducible factor)-1. The hypoxic stabilization and activation of the HIF-1α transcription factor is tightly regulated by oxygen [1,2]. Under normoxic conditions, the HIF-1α protein is hydroxylated by the consumption of molecular oxygen by three prolyl-4-hydroxylase-domain-containing enzymes as described previously [3,4]. In its hydroxylated form, HIF-1α is ubiquitinated by the pVHL (von Hippel–Lindau protein) and then degraded in the proteasome [5,6]. Under hypoxic conditions, oxygen-dependent hydroxylation is alleviated and HIF-1α is stabilized, heterodimerizes with HIF-1β and activates HIF-1-dependent target genes, including genes required for oxygen transport, oxygen supply and metabolic adaptation, as well as for pH regulation .
Besides these well described hypoxia-induced features at the level of protein stability and transcriptional activation, additional modulations in gene expression based on translational changes under severe hypoxic/anoxic conditions have been described previously [7–10]. The inhibition of overall mRNA translation under hypoxia and anoxia paradoxically leads to the privileged translation of certain mRNAs [11,12]. Blais et al.  identified several genes with a translational control during severe hypoxia/anoxia, including VEGF (vascular endothelial growth factor), ATF-3 (where ATF is activating transcription factor) and ATF-4. Interestingly, translational regulation has also been described for HIF-1α. The most prominent effect demonstrated so far on HIF-1α translation has been alterations in the pathways which control mTOR (mammalian target of rapamycin) activity or the AMP protein kinase [8,14,15]. Furthermore, a recent study has identified HuR and PTB (polypyrimidine tract binding) as two HIF-1α mRNA-binding proteins that function jointly to stimulate HIF-1α translation in response to cobalt chloride treatment . A functional IRES (internal ribosome entry site) element, which facilitates translation under stress conditions when protein synthesis via cap-dependent translation is reduced, has also been identified . In subsequent studies, however, it has been demonstrated that IRES activity is minimal and cannot account for changes in translation observed after exposure to hypoxic conditions .
In the 3′-UTR (untranslated region) of HIF-1α mRNA, AU (adenosine and uridine)-rich elements can be found. Those elements have been shown to control translation by promoting the assembly of RNA-binding proteins. In the present study, we provide evidence that the 3′-UTR of HIF-1α mRNA has regulatory functions by binding CPEB (cytoplasmic polyadenylation-element-binding protein) 1 and CPEB2.
All cloning work was carried out using Gateway Technology (Invitrogen). The CPEB1 entry vector was generated by cloning a PCR fragment into the NcoI/EcoRV-digested pENTR4 vector. The coding region of the CPEB1 zinc finger (amino acids 514–544) was deleted in the entry vector by site-directed mutagenesis using the QuikChange® site-directed mutagenesis kit (Stratagene) following the manufacturer's instructions. pDONR221 (Invitrogen) containing full-length CPEB2 was used to perform a BP Clonase reaction (Invitrogen) to obtain a CPEB2-containing entry vector. The inserts of both entry vectors were verified by DNA sequencing. To generate plasmids expressing fusion proteins, DNA inserts were transferred from entry clones to destination vectors using the LR Clonase recombination enzyme mixture (Invitrogen) following the manufacturer's instructions. The destination vector pcDNA3.1/nV5-Dest was used to express N-terminal V5-tagged proteins in mammalian cells. pMal-c2x (New England Biolabs) was converted into a destination vector by ligation of the Gateway vector conversion cassette B (Invitrogen) into the EcoRI site (blunt-ended by Klenow) of pMal-c2x, allowing the expression of MBP (maltose-binding protein)-tagged fusion proteins in Escherichia coli after recombination. EGFP (enhanced green fluorescent protein) was amplified by PCR and cloned into the Gateway compatible HindIII/EcoRV-digested pcDNA3.1nV5DEST vector (Invitrogen). This vector was used to perform LR Clonase recombinations to obtain fluorescence-coupled expression of CPEB1 or CPEB2. HIF-1α 3′- and 5′-UTR plasmids (gifts from Professor R. H. Wenger, Institute of Physiology, University of Zürich, Zürich, Switzerland) have been described previously . Site-directed mutagenesis and deletion variants of the pGlmHIF3′-UTR were performed using the QuikChange® site-directed mutagenesis kit (Stratagene) following the manufacturer's instructions. The NC-CPE [non-consensus-CPE (cytoplasmic polyadenylation element] (TTTTCAT) was mutated to the non-CPE-related sequence TTGGTAC, whereas the classical CPE (TTTTAT) was deleted in these constructs. The mouse cyclin B1 3′-UTR that contains two CPEs was amplified from mouse testis cDNA, subcloned into pENTR4 and transferred by the LR Clonase reaction into the luciferase reporter gene plasmid pcDNA3.1Luc. The pCRIIrat HIF-1α 3′-UTR, which was used for in vitro transcription of HIF-1α 3′-UTR mRNA, was provided by Professor S. W. Spaulding (Buffalo Veterans Affairs Medical Center, Buffalo, NY, U.S.A.) . The CPE and the two NC-CPEs were deleted by site-directed mutagenesis using the QuikChange® site-directed mutagenesis kit.
HeLa cells were obtained from the A.T.C.C. The neuroblastoma cell line SK-N-MC and the neuroglioma cell line H4 were obtained from Professor R. H. Wenger. Cells were cultured in high glucose Dulbecco's modified Eagle's medium containing 10% (v/v) fetal calf serum, 50 i.u./ml penicillin G and 50 μg/ml streptomycin (Invitrogen) in a humidified 5% CO2/95% air atmosphere at 37 °C. For hypoxic conditions, O2 levels were decreased to 1% O2 with N2 in an oxygen-controlled incubator (Invivo; Ruskinn, Bridgend, U.K.).
For immunofluorescence analyses, cells were fixed with freshly prepared 3.7% (w/v) formaldehyde in PBS (pH 7.4) for 10 min, washed with PBS, permeabilized with 0.5% Triton X-100 for 5 min and rinsed again with PBS. After blocking non-specific binding sites with 3% (w/v) BSA in PBS for 30 min, the cells were incubated for 1 h at room temperature (20 °C) with an anti-HIF-1α antibody (BD Transduction Laboratories) diluted 1:100 in 3% (w/v) BSA in PBS, followed by incubation for 1 h at room temperature with a tetramethyl rhodamine iso-thiocyanate-coupled anti-mouse antibody (Dako) diluted 1:100 with 3% (w/v) BSA in PBS. After extensive washing with PBS, the slides were mounted and analysed by fluorescence microscopy (Axioplan 2000; Zeiss).
HeLa or SK-N-MC cells were transiently transfected with luciferase reporter gene constructs, where the 3′-UTR or 5′-UTR of HIF-1α or cyclin B1 had been fused to the firefly luciferase cDNA, or with the luciferase HIF reporter gene plasmid pH3SVL (gift from Professor R. H. Wenger). In this plasmid, the luciferase reporter gene is driven by a SV40 (simian virus 40) promoter and contains a total of six HIF DNA-binding sites derived from the transferrin gene. Transfections were performed using either the calcium phosphate co-precipitation method [20a] or by lipofection using Rotifect (Carl Roth) according to the manufacturer's instructions. Cells were seeded in 24-well plates at 5×104 cells/well. One day after seeding, cells were co-transfected with 1 μg of the reporter gene construct together with 0.5 μg of pcDNA3.1, 0.5 μg of pcDNA3.1/nV5-Dest containing full-length CPEB1 or CPEB2 and 0.002 μg of Renilla luciferase control plasmid pRL-SV40 (Promega). Cells were subsequently exposed to normoxic (21% O2) or hypoxic (1% O2) conditions for 24 h. Luciferase activity was determined using the Dual-Luciferase assay kit (Promega). Results were normalized against the pcDNA3.1-transfected control values, which were arbitrarily defined as 1. In the case of the UTR-hybrid vectors, normalization was based on the firefly luciferase counts (which were generally higher) obtained with pGLmHIFI.2 3′-UTR plasmid containing HIF-1α 5′-UTR and HIF-1α 3′-UTR [231.092±11.114 firefly luciferase counts (results are means±S.E.M., n=3) without co-transfection of CPEBs] compared with the plasmids containing only the HIF-1α 3′-UTR [65.462±11.457 firefly luciferase counts (results are means±S.E.M., n=3) without co-transfection of CPEBs] or the HIF1α 5′-UTR [45.095±3.250 firefly luciferase counts (results are means±S.E.M., n=3) without co-transfection of CPEBs].
Exposure of mice to inspiratory 0.1% carbon monoxide has been described previously [21,22]. Total RNA from various tissues was purified as described previously . RNA concentrations were determined spectrophotometrically, and RNA integrity was monitored by denaturing formaldehyde/agarose-gel electrophoresis (1% gels).
siRNA (short interfering RNA) transfection
HeLa cells were transfected with 20 nM siRNAs (including two siRNA sequences for each gene) targeting CPEB1, CPEB2 or a non-target control using Lipofectamine™ (Invitrogen) according to the manufacturer's instructions. The following siRNA oligonucleotides were used: CPEB1 siRNA, 5′-ACCCACUGGGAAAUGUCCUAGGAA-3′ and 5′-GGAUAUUACAGAAGCUGGAUUAGUU-3′; CPEB2 siRNA, 5′-GAAAUAACUGCUAGCUUCAGAAGAU-3′ and 5′-GAGCUCAGUUCAGGCACUCAUUGAU-3′. All Star control siRNA (Qiagen) was used as a non-targeting control.
Protein samples were analysed for expression of HIF-1α, V5 and β-actin. Protein concentrations were determined by the Bradford method [23a] using BSA as a standard. For immunoblot analysis, cellular protein (50 μg) was resolved by SDS/PAGE (7.5% gels) and electrotransferred on to nitrocellulose membranes (Amersham Biosciences) by semi-dry blotting (PeqLab). HIF-1α, V5 and β-actin were detected using mouse anti-HIF-1α (1:1000 dilution; BD Transduction Laboratories), mouse anti-V5 (1:1000 dilution; Invitrogen) and mouse anti-β-actin (1:30000 dilution; Sigma) antibodies respectively, followed by a goat HRP (horseradish peroxidase)-labelled anti-mouse antibody (1:1000 dilution; Santa Cruz Biotechnology). Chemiluminescence detection of HRP was performed by incubation of the membranes with 100 mM Tris/HCl (pH 8.5), 2.65 mM H2O2, 0.45 mM luminol and 0.625 mM coumaric acid for 1 min at room temperature and analysed by imaging with a chemiluminescence camera (LAS3000; Fujifilm).
Protein expression and purification
MBP fusion proteins were expressed in the E. coli TB1 strain (New England Biolabs) transfected with pMal-c2x-DestCPEB1 or pMal-c2x-DestCPEB2 and purified using the pMal purification system (New England Biolabs) according to the manufacturer's instructions.
Filter-binding assays were carried out essentially as described previously . For determination of the binding constants, 10 fmoles of radioactively labelled RNA were incubated with increasing amounts of MBP–CPEB1 or MBP–CPEB2 in 40 μl of RNA-binding buffer [50 mM Tris/HCl (pH 8.0), 10% (v/v) glycerol, 0.2 mg/ml methylated BSA, 0.01% Nonidet P50, 1 mM EDTA, 1 mM dithiothreitol and 100 mM KCl]. After incubation for 30 min at room temperature, 35 μl of each reaction mixture was applied to nitrocellulose filters (Schleicher and Schuell) pre-treated with 1 ml of wash buffer [10 mM Tris/HCl (pH 8.0) and 100 mM NaCl] containing 5 μg/ml rRNA. After rinsing with 5 ml of ice-cold wash buffer, the filter-bound radiactivity was measured by scintillation counting. Apparent Kd values were determined both from direct and double-reciprocal plots.
Xenopus oocyte polyadenylation assays
HindIII-linearized pCRIIrat HIF-1α 3′-UTR or the mutant construct, in which the CPE and the two NC-CPEs are deleted, were amplified in the presence of [α-32P]UTP using the T7 mMessage Machine Kit (Ambion) following the manufacturer's instructions. Radioactively labelled RNA was injected into stage VI Xenopus oocytes (1 fmole/oocyte). Subsequently, oocytes were incubated at 16 °C in 0.5× L15 medium overnight. Oocytes which had undergone germinal vesicle breakdown and therefore presented a white spot at the animal pole were separated. RNA was extracted from the matured oocytes by phenol/chloroform precipitation. One portion of the isolated RNA was digested with 2.5 units RNAse H (New England Biolabs) in the presence of a polyT18 primer to remove the poly(A)+ (polyadenylated) tail. Non-digested and RNAse H-digested RNA was separated by denaturating 6 M urea/PAGE (4% gels). Radioactive bands were visualized using a phosphoimager system (Bio-Rad).
RNA extraction and real-time RT-PCR (reverse transcription-PCR)
Total RNA from mouse tissue or cultured cells was extracted as described previously . First-strand cDNA synthesis was performed with 1 μg of RNA using the First Strand Synthesis kit (Fermentas). Subsequently, mRNA expression levels for mCPEB (mouse CPEB)1, mCPEB2, hCPEB (human CPEB)1, hCPEB2, hCAIX [human CAIX (carbonic anhydrase IX)] and hHIF-1α (human HIF-1α) were quantified using 1 μl of the appropriate cDNA for real-time RT-qPCR (reverse transcription-quantitative PCR) using a SybrGreen Q-PCR reagent kit (Invitrogen) in combination with the MX3000P light cycler (Stratagene). Initial template concentrations of each sample were calculated by comparison with serial dilutions of a calibrated standard. To verify mRNA integrity and equal input levels, ribosomal human L28 or mouse S12 mRNA was also analysed and results are expressed as ratios relative to L28 or S12 levels. The oligonucleotide primers used were as follows: mCPEB1 5′-AGAAGTCCGTCCGGGCCT-3′ and 5′-GGTTGTGAGCTGCAGATAAG-3′; hCPEB1, 5′-GTGATCCCCTGGGTATTAGC-3′ and 5′-CAGATATGACACAGAGAATCTT-3′; m/hCPEB2, 5′-GCACTCTCTGGAAAATTCCC-3′ and 5′-AGAGCTTTCCATCTTCTTCAAT-3′; mS12, 5′-GAAGCTGCCAAGGCCTTAGA-3′ and 5′-AACTGCAACCAACCACCTTC-3′; hHIF-1α, 5′-TGCATCTCCATCTCCTACCC-3′ and 5′-GTAGCTGCATGATCGTCTGG-3′; hCAIX, 5′-GGGTGTCATCTGGACTGTGTT-3′ and 5′-CTTCTGTGCTGCCTTCTCATC-3′; and hL28, 5′-GCAATTCCTTCCGCTACAAC-3′, and 5′-TGTTCTTGCGGATCATGTGT-3′.
CPEB1 and CPEB2 bind to the HIF-1α 3′-UTR
CPEBs are key factors in regulating mRNA translation via defined changes in the poly(A)+ tail length. Four different CPEBs have been described in mammalian cells [25,26]. For CPEB1 and CPEB2, a polyadenylation-dependent function has been determined, whereas CPEB3 and CPEB4 seem to regulate translation of specific mRNAs independently of polyadenylation . To determine the tissue-expression profile of CPEB1 and CPEB2, RT-PCR was performed with samples derived from various mouse tissues. Whereas high levels of CPEB2 mRNA were detected in the testis only, high expression levels of CPEB1 mRNA were additionally found in the brain and, to a lower extent, in the kidney, heart, skeletal muscle and ovary (Figures 1A and 1B). Exposure of mice to inspiratory 0.1% carbon monoxide for 4 h did not change the expression of CPEB1 and CPEB2 (Figures 1C and 1D). Testing different tumour cell lines (including H4, SK-N-MC, PC12, HepG2 and HeLa cells), we found high endogenous CPEB1 and CPEB2 mRNA levels in the neuroblastoma SK-N-MC and the neuroglioma H4 cell lines. Hypoxic incubation (1% O2 for 24 h) of the SK-N-MC and H4 cells did not affect CPEB1 or CPEB2 mRNA levels (Figure 2). Additionally, we tested a shorter incubation time, i.e. 1% O2 for 4 h, which likewise did not affect CPEB1 or CPEB2 mRNA levels (results not shown). The RNA expression of the well described HIF target gene CAIX served as a positive control for hypoxia-induced gene expression in these experiments.
Organ expression of CPEB1 and CPEB2
CPEB1 and CPEB2 mRNA levels are not induced by hypoxia
Target mRNAs of CPEBs in mammalian cells include Cdk1 (cyclin-dependent kinase 1), cyclin B1, CaMKII (Ca2+/calmodulin-dependent protein kinase II)α, CaMKIIδ, tPA (tissue plasminogen activator), ABP (α-amino-3-hydroxy-5-methyliso-xazole-4-propionic acid-binding protein), RCM3 (calmodulin), Map2 (microtubule-associated protein 2) and β-casein [28–33]. Comparable with these mRNAs, the 3′-UTRs of mouse, human and rat HIF-1α mRNA contain one CPE (TTTTAT; nucleotides 952–958, 969–975 and 964–970 after the stop codons of human, mouse and rat HIF-1α mRNAs respectively) near to the hexanucleotide polyadenylation signal (Figure 3A). In addition, one or two NC-CPEs (TTTTCAT) can be found in mouse, human and rat HIF-1α 3′-UTRs respectively. We tested whether CPEB1 and CPEB2 are able to bind the CPE and NC-CPE in mouse HIF-1α 3′-UTR. Therefore we first characterized the binding affinity of CPEB1 and CPEB2 to a fragment containing the mouse 3′-UTR which contains the CPE (Figure 3B). CPEB1 and CPEB2 did bind the mRNA fragment with Kd values of 7.8±1.5 nM and 4±0.6 nM respectively. Specific binding of CPEB2 to the CPE was verified by a lack of binding to an mRNA fragment without CPE or NC-CPE characteristics (nucleotides 777–950 after the stop codon) as well as by competition of CPEB2 binding to the CPE and NC-CPE-containing fragment (nucleotides 950–1098 after the stop codon) by poly(U), but not by poly(A)+ or poly(C) (Figure 3C).
CPEB1 and CPEB2 bind to HIF-1α 3′-UTR
CPEB1 and CPEB2 affect the translational efficiency of HIF-1α via its 3′-UTR
Recently Pique et al.  described a combinatorial code for CPE-mediated translational control. In this previously published report, it has been demonstrated that the relative position of CPEs, NC-CPEs and the hexanucleotide polyadenylation signal AATAAA determines whether an mRNA will be translationally repressed by CPEBs, as well as determining the extent and time of cytoplasmic polyadenylation-dependent translational activation. Performing a software-assisted analysis, which has been developed based on the recently published combinatorial code (http://genome.imim.es/CPE/server.html), CPEB-mediated translational repression was predicted when CPEB is not stimulated.
To test if non-stimulated CPEBs indeed down-regulate HIF-1α protein levels, we overexpressed GFP (green fluorescent protein) alone, GFP–CPEB1 or GFP–CPEB2 in HeLa cells. Subsequently, cells were exposed to 1% O2 for 4 h. After incubation under hypoxic conditions, cells were fixed and HIF-1α protein expression was analysed by immunofluorescence. Cells overexpressing CPEB1 or CPEB2 demonstrated a significantly diminished hypoxic accumulation of HIF-1α compared with non-transfected cells or GFP-transfected control cells (Figures 4A and 4B). Similar results were obtained by testing the influence of increasing amounts of V5–CPEB2 on the expression of HIF-1α as determined by immunoblotting (Figure 4C). Additionally, CPEB1 and CPEB2 significantly diminished HIF-1 activity, which was analysed by a reporter gene assay (Figure 4E). Moreover, a significant decrease in hypoxia-induced CAIX RNA expression was detected as a result of CPEB1 or CPEB2 overexpression (Figure 4F). Diminished HIF-1α protein levels and HIF-1 activity were not due to changes in gene expression or mRNA stability, since overexpression of CPEB1 or CPEB2 did not change HIF-1α mRNA levels (Figure 4D).
CPEB1 and CPEB2 affect HIF-1α protein levels and HIF activity
To determine whether CPEBs have an effect on HIF-1α expression levels via the HIF-1α 3′-UTR, we performed chimaeric reporter gene assays (Figure 5A). Cells were transfected with constructs harbouring the 5′-UTR (pGLmHIFI.2), the 5′- and 3′-UTRs (pGLmHIFI.2 3′UTR) or the 3′-UTR (pGLmHIF3′UTR) of HIF-1α in frame with the firefly luciferase-coding region. Co-expression of CPEB1 or CPEB2 with the HIF-1α 5′-UTR (pGLmHIFI.2) did not decrease luciferase activity, whereas the presence of the 3′-UTR alone (pGLmHIF3′UTR) or in combination with the 5′-UTR (pGLmHIFI.2 3′UTR) significantly diminished firefly luciferase activity under the influence of CPEB1 and CPEB2 (Figures 5B and 5C). As a control for the reliability of the assay, we transfected cells with the 3′-UTR of the known CPEB target mRNA cyclin B1 fused to the firefly luciferase coding region. For cyclin B1, a CPEB1-mediated repression has been shown previously . In line with the fact that cyclin B1 mRNA is a target for CPEB1, co-expression of CPEB1 attenuated luciferase activity (Figure 5E).
CPEB1 and 2 change the translational efficiency of HIF-1α via its 3′-UTR
CPEBs contain two RNA-recognition motifs and one CH (cysteine and histidine) region, which is reminiscent of a zinc finger. For CPEB1, the CH region has been shown to be critical for RNA binding . Deletion of amino acids 514–544, which code for the CH region, abolished CPEB1-induced repression of luciferase activity after co-transfection with pGLmHIF3′UTR (Figure 5D).
HIF-1α mRNA is polyadenylated after injection into Xenopus oocytes
To determine whether the HIF-1α mRNA poly(A)+ tail is elongated by cytoplasmic polyadenylation after CPEB stimulation, we performed a commonly used assay to test for cytoplasmic polyadenylation . To this end, stage VI Xenopus oocytes were injected with radiolabelled rat HIF-1α cRNA 3′-UTR containing the CPE and NC-CPEs (Figure 6A). Oocytes were subsequently incubated with progesterone until a white spot was oberserved on the animal pole, which is an indication of germinal vesicle breakdown and thus oocyte maturation and CPEB activation. RNA from the matured oocytes was then extracted and analysed by agarose-gel electrophoresis and autoradiography. As demonstrated in Figure 6(B), the poly(A)+ tail length of the 3′-UTR was increased after injection into oocytes. This increase in length was due to polyadenylation, since treatment of the isolated RNA with an oligo(dT) primer and RNase H shortened the fragment. In addition, deletion of the CPE and NC-CPEs in the HIF-1α 3′-UTR abrogated polyadenylation.
HIF-1α mRNA is polyadenylated after injection into Xenopus oocytes
CPEB mediates insulin-stimulated HIF-1α expression
CPEB is post-translationally phosphorylated and thereby activated after stimulation of Xenopus oocytes with progesterone. In mammalian cells, the phosphorylation of CPEB can be triggered by treatment with insulin [33,38]. CPEB activation determines the switch from CPEB-mediated repression to CPEB-induced polyadenylation, resulting in translational activation of target mRNAs. Thus we determined the influence of insulin on the CPEB-mediated effects on the HIF-1α 3′-UTR in SK-N-MC cells (Figure 7A). To this end, we treated the neuroblastoma SK-N-MC cells with insulin after transfection with the chimaeric 3′-UTR reporter gene plasmid pGLmHIF3′UTR. Stimulation of the cells with insulin increased reporter gene activity to a minor extent in normoxia. However, exposing cells to hypoxic conditions and simultaneous treatment with insulin resulted in a significant increase in reporter gene activity. Regarding the reporter gene assay, a similar effect was not detectable after transfecting cells with pGLmHIFI.2, which contains the HIF-1α 5′-UTR in frame with the luciferase-coding region, excluding a non-specific insulin effect (Figure 7A). In addition, the insulin-stimulated increase in luciferase activity could be abrogated by deleting the CPE and the NC-CPE (Figure 7B). The insulin-dependent effect on the HIF-1α 3′-UTR was in accordance with increased endogenous HIF-1α protein levels after treating SK-N-MC cells with insulin in hypoxic conditions (Figure 7C). To test the effects of CPEBs on the hypoxic stabilization of HIF-1α, as well as the insulin-dependent increase in HIF-1α in hypoxia, we transiently transfected SK-N-MC cells with siRNAs targeting CPEB1 or CPEB2. Since there are no suitable antibodies available for detecting endogenous CPEB1 or CPEB2 protein expression in human cells, we determined the efficacy of the siRNA effects at the mRNA level. As a result of the transient transfection, a significant down-regulation of CPEB1 and CPEB2 was observed, which was not accompanied by a change in HIF-1α mRNA levels or hypoxia-induced mRNA expression of the HIF target gene CAIX (Figure 8). In line with the results obtained with the 3′-UTR reporter gene assays, down-regulation of CPEB1 or CPEB2 also did not affect hypoxic stabilization of the HIF-1α protein, and prevented the insulin-dependent increase of HIF-1α protein levels in hypoxia (Figure 9A). Mirroring this effect, CPEB1 and CPEB2-down-regulation inhibited an insulin-stimulated increase of CAIX mRNA expression in hypoxia (Figure 9B). This was not accompanied by changes in the expression of the HIF-1α mRNA levels, excluding changes in HIF-1α mRNA expression or stability.
Insulin stimulates the CPEB-dependent translational efficiency of HIF-1α
CPEB1 and CPEB2 knockdown does not affect HIF-1α RNA and CAIX mRNA levels after hypoxic incubation
CPEB1 and CPEB2 mediate the stimulatory effect of insulin on HIF-1α protein levels in hypoxia
The hexanucleotide sequence AAUAAA is required for the nuclear polyadenylation of mRNAs. The same sequence is necessary for cytoplasmic polyadenylation, which is facilitated by CPEB1 and CPEB2 (for a review, see ). CPEBs are the key factors for regulating mRNA translation via defined changes in the poly(A)+ tail length. They bind CPEs in the 3′-UTR of specific target mRNAs. After stimulation, CPEB promotes polyadenylation. The newly elongated tail is then bound by the poly(A)+-binding protein, which stimulates general translation by augmenting the assembly of the eIF4F (eukaryotic initiation factor 4F) initiation complex and thus the recruitment of the 40S ribosomal subunit to the 5′-end of the mRNA . This complex formation is inhibited under dormant conditions by the association of the translation inhibiting protein Maskin . Results obtained with CPEB1 knockout mice demonstrate that CPEB1 plays a role in germ-cell development and neuronal synaptic plasticity, as well as in cellular senescence [42–45].
The hypoxic response of HIF-1α is mainly regulated by hydroxylation-dependent protein stability. However, after stimulation of cells with hormones or growth factors or under strong hypoxic/anoxic conditions, an additional translational regulation of HIF-1α expression has been described [16,46–48]. Whereas general cap-dependent translation is inhibited during hypoxia/anoxia, HIF-1α mRNA is refractory to translational repression. Consistently, whereas most mRNAs shift to less actively translated monosomes, HIF-1α mRNA remains in actively translated polysomes under hypoxic/anoxic conditions .
In the present study, we demonstrate that the HIF-1α 3′-UTR contains a functional CPE and NC-CPEs. The CPE is able to bind CPEB1 and CPEB2, resulting in HIF-1α 3′-UTR cytoplasmic polyadenylation. In terms of the functional consequences in vivo, simple overexpression of CEPB1 or CPEB2 decreased HIF-1α protein levels, in line with the assumption that CPEBs need to be activated for polyadenylation activity. CPEB-mediated repression under resting conditions is in accordance with the specific arrangements of the CPE and the NC-CPE in the HIF-1α 3′-UTR. Activation of CPEBs has been described after treatment with insulin or progesterone [33,38]. Different phosphorylation pathways, including the phosphoinositide 3-kinase GSK3 (glycogen synthase kinase 3), the aurora A kinase and the CDC2 (cell division cycle 2 kinase) pathways, are associated with CPEB activation [32,38]. Accordingly, stimulation of SK-N-MC cells with insulin resulted in a CPEB-dependent increase in the chimaeric reporter gene activity mediated by the HIF-1α 3′-UTR. This result resembles the described effects of progesterone and insulin on β-casein regulation. Both progesterone and insulin stimulate CPEB-dependent polyadenylation synergistically, resulting in an increased rate of β-casein mRNA translation .
Clear evidence has emerged during previous years that HIF-1α is affected by many stimuli, including insulin, under normoxic conditions. We have found that insulin has an effect on HIF-1α expression through a 3′-UTR-mediated mechanism most prominently under hypoxia. These results indicate that the combination of hypoxia and stimulation of insulin results in the integration of various signals affecting CPEB activation. One candidate pathway besides the above-mentioned phosphorylation-dependent CPEB activation could be the mTOR pathway, which is affected by hypoxia and insulin. In addition, mTOR has been described as being involved in CPEB activation . Further studies are needed to analyse this hypothesis.
The present study contributes to the knowledge of the complex regulation of the HIF system. Whereas hypoxia-regulated protein stability of HIF-1α is probably the general control mechanism of the HIF system, additional regulatory pathways may account for the precise control of the HIF system after stimulation. In this regard, the specific organ-expression patterns of CPEB1 and CPEB2 should be noted, which indicates that the function of CPEB may be tissue specific. Most interestingly, high expression levels of CEPB1 and CEPB2 were found in the brain and testis. The high expression level of CPEB2 in the testis is in accordance with a previous report . The spatial pattern of HIF-1 protein expression in vivo in response to oxygen tension as a single specific physiological stimulus in a healthy organism has been examined in detail by Stroka et al. . Most interestingly, although in all organs HIF-1α expression responded to a decrease in the inspiratory oxygen concentration, basal HIF-1α levels differed greatly between various organs, with high levels in the brain and testis. These results indicate that organ-specific and mitogen-inducible factors, such as CPEB, may contribute to oxygen-dependent HIF-1α expression.
We thank Mrs Markmann and Professor B. C. Burckhardt for providing Xenopus oocytes. We thank Professor R. H. Wenger (Institute of Physiology, University of Zürich, Zürich, Switzerland) for the HIF 3′- and 5′-UTR plasmids, the pH3SVL plasmid and the SK-N-MC and H4 cell lines, and Professor S. W. Spaulding (Buffalo Veterans Affairs Medical Center, Buffalo, NY, U.S.A) for the pCRIIrat HIF-1α 3′-UTR.
activating transcription factor
carbonic anhydrase IX
Ca2+/calmodulin-dependent protein kinase II
cysteine and histidine
cytoplasmic polyadenylation element
cytoplasmic polyadenylation-element-binding protein
green fluorescent protein
internal ribosome entry site
mammalian target of rapamycin
reverse transcription-quantitative PCR
short interfering RNA
simian virus 40
The work was supported by the Deutsche Forschungsgemeinschaft [grant number DFG Ka 1269/5-1]; and by the Deutsche Krebshilfe [grant number 107757].