PI3K/Akt is required for heat shock proteins to protect hypoxia-inducible factor 1alpha from pVHL-independent degradation.

Hypoxia inducible factor 1 (HIF-1), a heterodimeric transcription factor composed of HIF-1alpha and HIF-1beta subunits, serves as a key regulator of metabolic adaptation to hypoxia. The amount of HIF-1alpha protein is regulated either by attenuating von Hippel-Lindau protein (pVHL)-dependent ubiquitination and subsequent 26 S proteasomal degradation or by enhancing cap-dependent mRNA translation, presumably involving a phosphatidyinositol 3-kinase (PI3K)/Akt-regulated pathway. In addition, it became apparent that Hsp90 protects HIF-1alpha from oxygen-independent degradation. Here we present evidence that PI3K/Akt is required for heat shock proteins to stabilize HIF-1alpha. In pVHL-deficient renal cell carcinoma cells, PI3K inhibition by LY294002 and wortmannin or transfection of either a dominant negative PI3K or a kinase-dead Akt mutant substantially lowered constitutively expressed HIF-1alpha without altering HIF-1alpha mRNA. Inhibitors of mitogen-activated protein kinase kinase (MAPKK) such as PD98059 or the p38 MAPK inhibitor SB203580 showed no interference. Considering that PI3K inhibitors down-regulated heat shock protein 90 (Hsp90) as well as Hsp70 expression and moreover attenuated heat- or hypoxia-induced Hsp70 as well as hypoxia-induced Hsp90 up-regulation we conclude that PI3K inhibition promoted degradation of HIF-1alpha indirectly by reducing steady state concentrations of Hsp90 and/or Hsp70. HIF-1alpha co-immunoprecipitated with Hsp90/Hsp70 and direct binding of Hsp70 to the oxygen-dependent degradation domain (ODD) of HIF-1alpha was proven by a pull-down assay and a peptide array. PI3K-mediated degradation of HIF-1alpha was confirmed in HEK 293 cells under hypoxia, suggesting that heat shock proteins constitute an integral component for HIF-1alpha accumulation. We conclude that PI3K/Akt contributes to HIF-1alpha stabilization by provoking expression of heat shock proteins.

Hypoxia inducible factor 1 (HIF-1), a heterodimeric transcription factor composed of HIF-1␣ and HIF-1␤ subunits, serves as a key regulator of metabolic adaptation to hypoxia. The amount of HIF-1␣ protein is regulated either by attenuating von Hippel-Lindau protein (pVHL)-dependent ubiquitination and subsequent 26 S proteasomal degradation or by enhancing cap-dependent mRNA translation, presumably involving a phosphatidyinositol 3-kinase (PI3K)/Akt-regulated pathway. In addition, it became apparent that Hsp90 protects HIF-1␣ from oxygen-independent degradation. Here we present evidence that PI3K/Akt is required for heat shock proteins to stabilize HIF-1␣. In pVHL-deficient renal cell carcinoma cells, PI3K inhibition by LY294002 and wortmannin or transfection of either a dominant negative PI3K or a kinase-dead Akt mutant substantially lowered constitutively expressed HIF-1␣ without altering HIF-1␣ mRNA. Inhibitors of mitogen-activated protein kinase kinase (MAPKK) such as PD98059 or the p38 MAPK inhibitor SB203580 showed no interference. Considering that PI3K inhibitors down-regulated heat shock protein 90 (Hsp90) as well as Hsp70 expression and moreover attenuated heat-or hypoxia-induced Hsp70 as well as hypoxia-induced Hsp90 up-regulation we conclude that PI3K inhibition promoted degradation of HIF-1␣ indirectly by reducing steady state concentrations of Hsp90 and/or Hsp70. HIF-1␣ co-immunoprecipitated with Hsp90/Hsp70 and direct binding of Hsp70 to the oxygen-dependent degradation domain (ODD) of HIF-1␣ was proven by a pull-down assay and a peptide array. PI3K-mediated degradation of HIF-1␣ was confirmed in HEK 293 cells under hypoxia, suggesting that heat shock proteins constitute an integral component for HIF-1␣ accumulation. We conclude that PI3K/Akt contributes to HIF-1␣ stabilization by provoking expression of heat shock proteins.
Immunoprecipitation-Cells were treated, scraped off the dishes and collected. Cell pellets were reconstituted with 300 l of buffer B (50 mM Tris, 150 mM NaCl, 5 mM EDTA, 0.5% Nonidet P-40, 5% glycerol, 1 mM phenylmethylsulfonyl fluoride, protease inhibitor mixture, pH 7.5), followed by immediate vortexing (3 ϫ 15 s). Following centrifugation (15,000 ϫ g for 30 min) supernatants were transferred to fresh tubes. Supernatants, 1 mg of protein each, were supplied with 1 g of anti-HIF-1␣-, anti-Hsp90-, or anti-Hsp70-antibody and incubated at 4°C for 1 h. Thereafter, 50-l protein G microbeads were added and incubations continued at 4°C overnight. Beads were magnetically collected following the manufacturer's manual, washed 3 times with 100 l of buffer B each. The co-precipitated proteins were finally eluted by 95°C preheated SDS-PAGE sample buffer according to the manufacturer's manual. Eluted samples were loaded on 7.5% SDS-PAGE. Western analysis was performed using anti-HIF-1␣, anti-Hsp90, or anti-Hsp70 antibodies.
GST-HIF-1␣ODD/Hsp70 Pull-down Assay-4 ϫ 10 6 RCC4 cells were seeded in 10-cm dishes 1 day prior to transfection. At a rate of 60% confluence, cells were transfected using SuperFect TM transfection reagent following the instructions provided by the manufacturer. The plasmid allows expression of the HIF-1␣ODD domain from residue 401 to 603, fused to GST. Following treatments with inhibitors, cells were scraped off the dishes and collected. 300 l of buffer B was added to each pellet, followed by immediate vortexing (3ϫ 15 s). Following centrifugation (15,000 ϫ g for 30 min) supernatants were transferred to fresh tubes. Each supernatant (1 mg of protein) was supplied with 50 l of GSH-agarose and incubated at 4°C for 16 h. Thereafter, beads were collected, washed 3 times with 100 l of buffer B, finally supplemented with 50 l of 2ϫ SDS-PAGE sample buffer, and boiled at 95°C for 10 min. Beads were removed by centrifugation and supernatants were loaded on 7.5% SDS-polyacrylamide gels. Western analysis was performed using anti-Hsp70 or anti-GST antibodies.
Hsp70 Protein Expression-Human Hsp70 was cloned into pPROEX. Protein was expressed in Escherichia coli as a fusion protein with a hexahistidine tag using BL21(DE3)pLysS cells and LB medium. Expression was induced by 0.5 M isopropyl-1-thio-␤-D-galactopyranoside for 8 h at 37°C. Purification of the Hsp70 was performed by Ni-NTAagarose chromatography. Briefly, cells were lysed in buffer C (50 mM Tris-HCl, 150 mM NaCl, 1 mM phenylmethylsulfonyl fluoride, protease inhibitor mixture, pH 7.5) by sonification. Following centrifugation, the lysate was loaded on Ni-NTA columns equilibrated with buffer C. The columns were washed with a 5-fold volume of buffer C and a 5-fold volume of buffer C containing 50 mM imidazol. Finally, Hsp70 protein was eluted by buffer C containing 500 M imidazol.
Peptide Array Assay-The peptide array containing HIF-1␣ peptide spots was generated following SPOT synthesis. Briefly, 272 overlapping peptide fragments of 15 amino acids in length with an offset of 3 amino acid residues were generated such that the complete HIF-1␣ protein sequence was covered. These HIF-1␣ peptides were chemically synthesized as an array of spots on an amino-polyethyleneglycol modified cellulose membrane (AIMS Scientific Products GmbH, Braunschweig, Germany) as described previously (35). All peptides are NH 2 -terminal acetylated and remain covalently attached to the membrane via their carboxyl termini. Binding studies were performed according to an adapted protocol from Frank and Overwin (36). In brief, unspecific binding sites were blocked with buffer D (50 mM Tris-HCl, 150 mM NaCl, 5 mM EDTA, 0.5% Nonidet P-40, 20% glycerol, 1 mM NaF, 1 mM Na 3 VO 4 , 1% bovine serum albumin, pH 7.0). Afterward the membrane was incubated with 6 g/ml purified His-tagged Hsp70 protein in buffer D overnight at 4°C. For detection of binding either an anti-Hsp70 antibody (1:1000 in buffer D) or anti-pentahistidine tag antibody (1:1000 in buffer D) was added and incubated for 2 h at room temperature. Thereafter, the membrane was washed 3 times for 5 min each with phosphate-buffered saline (pH 7.0). For visualization, the membrane was incubated with a horseradish peroxidase-labeled goat antimouse secondary antibody (1:2000 in buffer D) for 1 h at room temperature, washed two times for 5 min each with TTBS and 5 min with phosphate-buffered saline, followed by ECL detection. 35 S-Radioisotopic Labeling-RCC4 cells were starved for 1 h in serum-and methionine-free medium, followed by replacement with methionine-free medium containing 10% fetal calf serum and 100 Ci/ml [ 35 S]methionine for the time indicated. Cells were then washed with phosphate-buffered saline and lysed with buffer B. Hsp90 or Hsp70 were immunoprecipitated from lysates containing 1 mg of total protein, using 1 g of anti-Hsp90 or anti-Hsp70 antibody as described above. After a 10% SDS-PAGE, the gel were dried and exposed to x-ray films.
Statistical Analysis-Each experiment was performed at least three times and representative data are shown.

PI3K Is Required to Maintain HIF-1␣ Expression in RCC4
Cells-Considering some controversy on the role of PI3K in affecting HIF-1␣ expression we wanted to better understand the involvement of PI3K in stability regulation of the HIF-1␣ protein. We started experiments in RCC4 cells because they should show constitutively expressed HIF-1␣ because of a pVHL deficiency. RCC4, but as expected not RCC4 cells with pVHL being reintroduced (RCC4/pVHL), revealed basal HIF-1␣ expression. Cells were treated with PI3K inhibitors such as 3 to 30 M LY294002 or 10 to 100 nM wortmannin for 8 h (Fig. 1, A and B) and HIF-1␣ disappeared following dosedependent inhibition of PI3K.
To confirm HIF-1␣ down-regulation by interfering with PI3K/Akt signaling and to exclude potential side effects of inhibitory drugs we transfected RCC4 cells with dominantnegative kinase mutants of either PI3K or Akt (Fig. 1, C and D). RCC4 cells were transfected with 1 g of plasmids allowing expression of a dominant-negative PI3K (pSR␣-⌬p85) or a kinase-dead mutant of Akt (pCMV5.-m/p-PKB␣K179A). Compared with transfection with wild-type PI3K (pSR␣-WTp85) or an empty vector plasmid (pCMV5) dominant-negative kinases decreased HIF-1␣ expression substantially. In contrast to PI3K inhibitors, the MAPKK inhibitor PD58059 or the p38 MAPK inhibitor SB203580 left HIF-1␣ expression in RCC4 cells unaltered (Fig. 1E). These experiments imply that pharmacological intervention or a molecular approach to attenuate PI3K/Akt signaling decreased HIF-1␣ expression in RCC4 cells. It remains to be clarified how an inactive PI3K/Akt pathway contributes to HIF-1␣ down-regulation.
Inhibition of PI3K Promoted Degradation of Constitutively Expressed HIF-1␣-HIF-1␣ accumulation either results from enhanced protein synthesis or attenuated protein degradation. As a first potential target of PI3K inhibitors we examined HIF-1␣ mRNA levels in RCC4 cells by semi-quantitative RT-PCR ( Fig. 2A) and quantitative real-time PCR (Fig. 2B). A total overlap of traces from controls, LY294002-, and wortmannintreated samples showing the dependence on the cycle number relative to the fluorescence increase during quantitative realtime PCR excluded variations in HIF-1␣ mRNA to account for alterations noticed at the HIF-1␣ protein level. Quantification showed that neither LY294002 (mRNA amount is 1 Ϯ 0.03 FIG. 1. HIF-1␣ destabilization by PI3K/Akt inhibition. RCC4 cells were exposed to: A, the PI3K inhibitor LY294002; B,wortmannin; C, transfected with 1 g of plasmids encoding PI3K-p85 wild-type (pSR␣-WTp85) or a dominant-negative PI3K-p85 subunit (pSR␣-⌬p85); D, 1 g of plasmid encoding a dominant-negative Akt-kinase (pCMV5.m/p-PKBK197A) or 1 g of empty vector plasmid (pCMV5); or E, the MAPKK inhibitor PD58059 or the p38 MAPK inhibitor SB203580 for 8 h at the concentrations indicated. Controls or RCC4/pVHL cells remained without treatment. HIF-1␣ was determined by Western analysis as described under "Experimental Procedures." Blots are representative of three independent experiments. normalized to the control) nor wortmannin (mRNA amount is 1 Ϯ 0.08 normalized to the control) caused mRNA alterations.
In a second step we approached the question of HIF-1␣ degradation. For these experiments LY294002 or wortmannin were used during an 8-h incubation period to decrease HIF-1␣ expression in RCC4 cells (Fig. 3A). Thereafter we supplied the proteasome inhibitor MG132 at a concentration of 5 M for 4 h, which allowed full recovery of HIF-1␣ expression that otherwise was compromised by PI3K inhibition. As expected, decreasing HIF-1␣ expression by blocking protein translation with CHX did not allow regaining of the protein upon MG132 treatment, implying that LY294002 did not block basal translation of HIF-1␣ mRNA compared with CHX. The proteasome inhibitor MG132 blocked HIF-1␣ protein degradation but did not disturb HIF-1␣ protein synthesis. Choosing a low exposure time of the x-ray film allowed detection of increased HIF-1␣ protein after 2 h of MG132 (5 M) treatment with a maximal HIF-1␣ response seen around 4 h with no further increase noticed at 8 h (Fig. 3B).
In contrast, the protein translation inhibitor CHX blocked HIF-1␣ protein synthesis. As seen in Fig. 3C, a 4-h CHX treatment lowered HIF-1␣ expression substantially compared with controls and HIF-1␣ disappeared when CHX was supplied for 8 h. Once HIF-1␣ had been depleted with CHX, the protein could not be recovered by subsequent addition of 5 M MG132 for 4 h.
We conclude that a pVHL-independent but proteasome-dependent degradation process explains the HIF-1␣ disappearance by PI3K inhibitors. Considering that besides the HIFprolyl hydroxylases/pVHL/proteasome degradation system Hsp90 may play a role in pVHL-independent HIF-1␣ degradation, we investigated whether heat shock proteins are involved in decreasing HIF-1␣ upon PI3K inhibition.
Expression of Hsp90 and Hsp70 Is Decreased by PI3K Inhibition-Hsp90 is known to participate in HIF-1␣ stability regulation (10,11). Therefore, it was our intention to search for a connection between PI3K signaling and heat shock proteins. For these experiments RCC4 cells were treated with 1 M geldanamycin, 30 M LY294002, or 100 nM wortmannin for 8 h (Fig. 4A). As anticipated the Hsp90 inhibitor geldanamycin decreased HIF-1␣ expression without altering protein amounts of Hsp90 or Hsp70. In contrast, the PI3K inhibitors LY294002 and wortmannin not only decreased the amount of HIF-1␣ but also abrogated expression of Hsp90 as well as Hsp70. A timedependent study revealed that LY294002 effects became apparent after 8 h and persisted at least up to 24 h (Fig. 4B). Interestingly, Western analysis showed that protein disappearance of HIF-1␣, Hsp90, and Hsp70 followed a similar time Interactions of HIF-1␣ with Hsp90 and Hsp70 -To prove that HIF-1␣ interacts with Hsp90 and Hsp70 in vivo, immunoprecipitation studies were performed using the anti-HIF-1␣ antibody (Fig. 5). RCC4 cells were treated with 30 M LY294002 or left untreated. Immunoprecipitation results confirmed a direct association between HIF-1␣, Hsp90, and Hsp70 in control cells.
However, under conditions of LY294002 treatment we precipitated less HIF-1␣ because the HIF-1␣ protein amount decreased as seen in Fig. 4. Therefore, it was without surprise that less Hsp90 or Hsp70 co-immunoprecipitated with HIF-1␣ under those conditions. Decreased co-immunoprecipitation of HIF-1␣ with either Hsp90 or Hsp70 under LY294002 treatment reflects less input, i.e. HIF-1␣ loading rather than allowing the assumption that the protein-protein interaction is attenuated.
Hsp70 Interacts with the ODD Domain of HIF-1␣-Previous studies have shown that Hsp90 interacts with the bHLH-PAS domain of HIF-1␣ (7). Having established that Hsp70 co-im-munoprecipitates with HIF-1␣, it was our intention to locate the potential binding site. Therefore, we transfected the RCC4 cell with pGST or pGST-HIF1␣ODD plasmids, and used GSHagarose to pull-down HIF-1␣401-603 and its potentially associated protein, i.e. Hsp70. Western analysis proved that Hsp70 directly associated with the ODD domain of HIF-1␣ (Fig. 6A). Similar to observations in Fig. 5, the apparent decreased interaction between Hsp70 and pGST-HIF1␣ODD under LY294002 treatment was based on a lower input of Hsp70. Without surprise, PD58059 had no effect on the interaction between Hsp70 and pGST-HIF1␣ODD. To narrow the exact binding site, we used a HIF-1␣ peptide spot array and incubated the membrane with bacterial expressed, His-tagged Hsp70, followed by anti-Hsp70 or anti-pentahistidine tag antibody detection (Fig. 6B).
This approach allowed us to identify two individual spots, located next to each other, highlighted on the peptide array. Deduced from the spot location we mapped the binding site of Hsp70 to amino acids 406 -420 with the sequence 406 DTIISLD-FGSNDTET 420 . This sequence is located within the ODD domain of HIF-1␣.
PI3K Inhibitor Attenuated HIF-1␣ Accumulation in HEK 293 Cells under Hypoxia-To confirm inhibition of PI3K in close association with HIF-1␣ degradation in cells other than RCC4 we employed human embryonic kidney (HEK 293) cells that contain a functional pVHL/ubiquitination/degradation system. HEK 293 cells, exposed to hypoxia (0.5% oxygen) for 8 h, responded with a robust HIF-1␣ accumulation signal. At this time we added 30 M LY294002 or 100 nM wortmannin and continued incubations under hypoxic conditions for an additional 16 h (Fig. 7). Western analysis showed that inhibition of PI3K down-regulated Hsp90 and Hsp70 expression and caused disappearance of HIF-1␣, whereas in controls prolonged hypoxia for a total of 24 h allowed high expression of HIF-1␣. It was noticed that hypoxia up-regulated Hsp90 as well as Hsp70 expression, an effect completely antagonized by inhibition of PI3K.
The Impact of PI3K Inhibition on Hsp90/Hsp70 Expression-Having established that PI3K inhibitors decreased basal expression of Hsp90 as well as Hsp70 in RCC4 cells, we went on to explore the impact of PI3K inhibition during heat treatment (Fig. 8A). RCC4 cells were heat shocked at 42°C for 8 h, followed by Western analysis of Hsp90 and Hsp70 expression. As expected, Hsp70 responded with increased protein synthesis toward heat treatment. Interestingly, pretreatments for 30 min with either LY294002 or wortmannin largely attenuated this response. Although Hsp90 was not induced by heat treatment, inhibition of PI3K decreased basal Hsp90 expression as seen before. In addition, we investigated whether hypoxia (0.5% oxygen, 8 h) might affect heat shock protein expression (Fig.  8B). Hypoxia induced both Hsp90 and Hsp70. Preincubations with LY294002 for 30 min largely attenuated hypoxia-evoked  5. HIF-1␣ co-immunoprecipitates (IP) with Hsp90 and Hsp70. RCC4 cells were treated with 30 M LY294002 (LY) for 8 h or remained as controls (C). Co-immunoprecipitation was performed with an anti-HIF-1␣ monoclonal antibody. Immunoprecipitation of Hsp90 as well as Hsp70 along with HIF-1␣ was determined by Western blot (WB) analysis as described under "Experimental Procedures." Experiments were performed at least three times and representative data are shown.
Hsp90 and Hsp70 induction, which pointed to a demanding role of PI3K in heat shock protein synthesis.
Consistent with results obtained by employing chemical inhibitors such as LY294002 and wortmannin, transient transfection of a dominant negative p85-PI3K subunit revealed inhibition of Hsp90 as well as Hsp70 expression (Fig. 8C). In contrast, expression of wild-type p85 left Hsp protein appearance unchanged. The role of Akt as a downstream target of PI3K signaling in promoting Hsp90 and Hsp70 expression was confirmed by transfecting a kinase-dead mutant of Akt (Fig.  8D). Whereas the kinase-dead mutant of Akt (pCMV5.-m/p-PKB␣K179A) lowered Hsp90 and Hsp70 expression, an empty vector plasmid (pCMV5) showed no interference. These results allow the prediction of an essential role of the PI3K/Akt pathways in expression regulation of Hsp90 as well as Hsp70.
PI3K Regulates Hsp90 and Hsp70 Protein Synthesis-To further substantiate an essential role of PI3K signaling in affecting Hsp90 as well as Hsp70 protein synthesis we performed [ 35 S]methionine pulse labeling experiments (Fig. 9). Within 6 -12 h substantial protein synthesis of Hsp90 and Hsp70 occurred, as detected by incorporation of radioactivity.
In the presence of 30 M LY294002, 35 S labeling of either Hsp90 or Hsp70 was drastically attenuated compared with controls. This suggests protein synthesis of Hsp90 and Hsp70 to be under the control of the PI3K pathway.

DISCUSSION
Our study corroborates earlier observations on the involvement of PI3K in affecting HIF-1␣ protein appearance. In pVHL-negative RCC4 cells pharmacological inhibition of PI3K or a molecular approach to eliminate PI3K as well as Akt activity largely attenuated HIF-1␣ expression although HIF-1␣ mRNA expression and basal protein translation remained unchanged. PI3K inhibition not only decreased HIF-1␣ expression but also lowered protein amounts of Hsp70 and Hsp90 and effectively antagonized heat-elicited up-regulation of Hsp70 or hypoxia-induced Hsp90 and Hsp70 expression. Our study shows direct Hsp70-HIF-1␣ binding and suggests that an active PI3K/Akt pathway is required for heat shock protein, i.e. Hsp70 and Hsp90 synthesis. Long lasting PI3K inhibition decreased Hsp expression and thus favored pVHL-independent HIF-1␣ proteasomal degradation rather than blocking protein translation.
PI3K/Akt Signaling via Hsp90 and Hsp70 Contributes to HIF-1␣ Protein Stability-More recently it became apparent that HIF-1␣ and Hsp90 physically interact (6,7). In addition, it was noticed that geldanamycin, a Hsp90 inhibitor, promoted efficient proteasomal, however, pVHL-independent degradation of HIF-1␣ (10,11). These results are confirmed in our study by using geldanamycin to decrease expression of HIF-1␣. Unexpectedly, inhibition of PI3K/Akt not only repressed HIF-1␣ but also constitutive expression of both, Hsp70 as well as Hsp90, whereas geldanamycin left heat shock protein expression unaltered. The impact of PI3K inhibition and Hsp expression became further evident when LY294002 or wortmannin attenuated heat-induced Hsp70 expression or hypoxiaevoked Hsp90 and Hsp70 up-regulation. Along that line dominant negative p85 or a kinase-dead mutant of Akt lowered Hsp90 and Hsp70 expression and LY294002 blocked Hsp90 as well as Hsp70 protein synthesis as shown by a [ 35 S]methionine pulse labeling experiment. It must be concluded that basal as well as stimulated expression of heat shock proteins demands a functional PI3K/Akt pathway. Stress proteins, i.e. heat shock proteins, regulate fundamental cellular processes, such as folding, sorting, degradation, resolubilization of proteins, and assembly of proteins into larger aggregates. At least expression of Hsp70 is transcriptionally regulated by heat shock transcription factor 1, which seems to require phosphorylation for full activity (20). There are indications that PI3K either upstream of PKC␦ or upstream of Rac/PKA contributes to heat shock transcription factor 1 activation and Hsp70 expression (20). This may explain why inhibition of PI3K down-regulated Hsp expression.
The interaction of Hsp90 with HIF-1␣ requires the PAS-bHLH structure of HIF-1␣. Making use of a HIF-1␣ peptide array on a spot membrane and confirmatory results from HIF-1␣-Hsp70 pull-down assays, allowed us to assign the binding region of Hsp70 to the ODD domain of HIF-1␣. This interaction may be relevant for stabilizing HIF-1␣. Association of HIF-1␣ with individual heat shock proteins or a combination of heat shock proteins as known for steroid hormone receptors may promote stabilization of HIF-1␣ along the observation that heat stabilized HIF-1␣ (9). In contrast, counteracting binding of Hsp90 to client proteins by geldanamycin, or by down-regulating Hsp expression by blocking the PI3K/Akt pathway will provoke degradation of those clients. The impertinent role of Hsp was further analyzed in pVHL-containing HEK cells. Hypoxia clearly stabilized HIF-1␣, presumably by blocking pVHLevoked hydroxylation and subsequent proteasomal degradation. Taking into account that translational regulation does not participate in HIF-1␣ stabilization under hypoxia it was interesting to see decreased protein amounts with PI3K being blocked in close association with decreased Hsp70 and Hsp90 expression. We may conclude that blocking PI3K for longer time periods interferes only indirectly with HIF-1␣ accumulation by attenuating Hsp70 and/or Hsp90 expression. PI3K/Akt in Translational Regulation of HIF-1␣-The use of chemical inhibitors of PI3K such as wortmannin or LY294002, or kinase-dead mutants of PI3K/Akt were shown to attenuate growth factor-, hormone-, or cytokine-stimulated HIF-1␣ protein accumulation (27,28,30). Reducing the HIF-1␣ protein amount resulted in attenuated DNA binding and a failure to stimulate transcription of reporter constructs or endogenous downstream HIF-1 target genes (27,28). Similarly, inhibition of PTEN (phosphatase and tensin homolog), a negative regulator of the PI3K pathway, increased HIF-1␣ and activated HIF-1 transcriptional activity (27). Mechanistically, PI3K/Akt is linked to an increased rate of HIF-1␣ protein synthesis, rather than inhibition of degradation as known from the action of hypoxia. Heregulin, insulin-like growth factor-1, or insulin stimulate PI3K/Akt to further activate FRAP (FK506 binding protein-rapamycin associated protein; also known as mammalian target for rapamycin), which promotes increased translation of HIF-1␣ mRNA into protein (24,28,37). In detail, FRAP de-represses the translational regulatory protein eIF-4E by phosphorylating and inactivating its binding protein 4E-BP1. FRAP also activates p70 S6K , which stimulates the 40 S ribosomal protein S6. Thus, in a PI3K/Akt and rapamycin-sensitive manner increased translation from the 5Ј-untranslated region of HIF-1␣ mRNA leads to an increase in HIF-1␣ protein (24,38). To explain a functional cellular response, e.g. growth factors, one has to assume that under these conditions the endogenous degradation pathway, i.e. prolyl-hydroxylation, is titrated out and that transcriptional repression because of asparagine hydroxylation is relieved.
The Role of PI3K/Akt in Hsp and HIF-1␣ Degradation Versus HIF-1␣ Translational Control-Interfering with protein translation by attenuating PI3K signaling to block HIF-1␣ protein accumulation is different than the situation seen here. We repressed the PI3K/Akt pathway for at least 8 -16 h and more importantly, followed disappearance of an already stabilized protein by using RCC4 cells. Our observation on LY294002provoked pVHL-independent degradation of HIF-1␣ supports an earlier observation on the role of PI3K and HIF-1␣ appearance (39,40). We went on to show that once HIF-1␣ disappeared its accumulation is regained by blocking proteasomal degradation with MG132. This implies ongoing basal translation of HIF-1␣ mRNA into protein as blocking of protein turnover again causes protein accumulation. As a consequence our experiments suggest a new role of the PI3K/Akt pathway in HIF-1␣ protein degradation besides its established role in enforced HIF-1␣ mRNA translation.
Our observations may help to resolve existing controversy on the role of PI3K in coordinating a HIF-1␣ response. Under conditions of translational regulation of HIF-1␣ as achieved by growth factors, hormones, or cytokines, PI3K inhibitors will suppress assembly of the translational machinery and thus block protein appearance. Basal production of the protein as occurring in RCC4 cells or the classic hypoxic accumulation of HIF-1␣ because of blocked proteasomal degradation may turn out insensitive to PI3K inhibition as long as PI3K inhibition remains short. An exception may be hypoxia or NO under conditions where they stimulate Akt (41,42), a situation when transcriptional and blocked degradation pathways might overlap in accumulating HIF-1␣. With long lasting PI3K inhibition for periods extending 8 h, the mechanism of HIF-1␣ destabilization may involve altered Hsp expression. Blocking Hsp90 function with geldanamycin or blocking Hsp expression by PI3K/Akt inhibition eliminates the ability to stabilize the client HIF-1␣. Heat shock proteins apparently are required to maintain HIF-1␣ stabilization even under conditions of hypoxia. Detailed knowledge on the functional interplay between HIF-1␣ and heat shock proteins will help to understand the full dynamic range of HIF-1␣ stability regulation.