HIF-1 (hypoxia-inducible factor 1) is a master regulator of cellular adaptive responses to hypoxia. The expression and transcriptional activity of the HIF-1α subunit is stringently controlled by intracellular oxygen tension through the action of prolyl and asparaginyl hydroxylases. In the present study we demonstrate that PG (n-propyl gallate) activates HIF-1 and expression of its downstream target genes under normoxic conditions in cultured cells and in mice. The stability and transcriptional activity of HIF-1α are increased by PG. PG treatment inhibits the interaction between HIF-1α and VHL (von Hippel–Lindau protein) and promotes the interaction between HIF-1α and p300, indicating that PG inhibits the activity of both prolyl and asparaginyl HIF-1α hydroxylases. We conclude that PG activates HIF-1 and enhances the resultant gene expression by directly affecting the intracellular oxygen sensing system in vitro and in vivo and that PG represents a lead compound for the development of a non-toxic activator of HIF-1.
Understanding molecular mechanisms regulating angiogenesis and systemic adaptive responses to reduced oxygen availability may lead to novel therapies for ischaemic disorders. Tissues use many strategies, including induction of angiogenesis and alterations in metabolism, to survive under hypoxic conditions. HIF (hypoxia-inducible factor)-1 activates the transcription of genes whose protein products mediate adaptive responses to hypoxia/ischaemia, including EPO (erythropoietin), GLUT1 (glucose transporter 1), NOS2 (nitric oxide synthase 2) and VEGF (vascular endothelial growth factor) [1–4]. Ischaemic heart disease is a major cause of mortality that is treated by pharmacologic agents, balloon angioplasty and coronary artery-bypass graft surgery. Novel therapeutic strategies aiming to stimulate neovascularization and to induce expression of cytoprotective factors are now under development.
HIF-1 is a heterodimer consisting of a constitutively expressed HIF-1β subunit and an O2-regulated HIF-1α subunit . Under normoxic conditions, members of the PHD (prolyl hydroxylase domain-containing protein)/EGLN (egg laying nine) family hydroxylate the HIF-1α subunit on two conserved prolyl residues in an O2-, Fe2+- and 2-oxoglutarate-dependent manner [6–8]. Hydroxylated HIF-1α molecules are polyubiquitinated and hence marked for proteasomal destruction by an ubiquitin ligase that contains the VHL tumour suppressor protein. Under conditions of low O2 or low Fe2+, or in the absence of VHL, HIF-1α accumulates in its active form and activates the transcription of genes involved in adaptation to hypoxia. Concurrently with prolyl hydroxylation, hydroxylation of a conserved asparaginyl residue by FIH-1 (factor inhibiting HIF-1) regulates HIF-1 transcriptional activation by modulating recruitment of the co-activators CBP and p300 also in an O2-, Fe2+-, and 2-oxoglutarate-dependent manner [9–11]. Thus hydroxylation of prolyl and asparaginyl residues in HIF-1α regulates its protein stability and transactivation function in an oxygen-dependent manner.
Physiological stimuli other than hypoxia can also induce HIF-1 activation and the transcription of hypoxia-inducible genes. Signalling via the HER2 (human epidermal growth factor receptor 2)/neu or IGF-1 (insulin-like growth factor 1) receptor tyrosine kinase induces HIF-1 expression by an oxygen-independent mechanism [12,13]. In addition to growth factors, nitric oxide, prostaglandin E2, thrombin, angiotensin II, 5-hydroxytryptamine and acetylcholine induce HIF-1 activation under non-hypoxic conditions [14,15]. Taking into account evidence that activation of HIF-1 is therapeutically beneficial in diseases characterized by acute or chronic ischaemia [16,17], low-molecular mass compounds that induce HIF-1 may be candidates for the treatment of ischaemic diseases.
HIF-1 activity is also affected by cellular redox conditions, which are maintained by the thioredoxin-redox factor 1 cascade [18,19]. We screened redox-acting low-molecular mass compounds to find an inducer of HIF-1 activity. In this report we demonstrate that PG (n-propyl gallate) strongly induces HIF-1 activity by inhibiting the activity of both prolyl and asparaginyl hydroxylases in various cell types, including epithelial, endothelial and smooth muscle cells. Moreover, we demonstrate that PG is active in the kidney after systemic administration, as determined by transcription profiling and production of EPO, indicating that HIF-1 activity can be manipulated in vivo with an orally active low-molecular mass compound.
MATERIALS AND METHODS
Cell culture and reagents
HEK-293 cells (human embryonic kidney cells) were maintained in DMEM (Dulbecco's modified Eagle's medium) supplemented with 10% FBS (fetal bovine serum), 100 units/ml penicillin and 0.1 mg/ml streptomycin. Hep3B (human hepatoma) and Caco-2 (human colonic adenocarcinoma) cells were maintained in MEM (modified Eagle's medium) with Earl's salts supplemented with 10% FBS, 1 mM nonessential amino acids and 1 mM sodium pyruvate, 100 units/ml penicillin, and 0.1 mg/ml streptomycin. The human colon adenocarcinoma cell lines HT-29 and SW48 were grown in RPMI 1640 medium supplemented with 10% FBS, and T-84 cells were grown in RPMI 1640 medium containing 10% FBS and Ham's F12:DMEM (1:1) containing 2 mM glutamine. HUVECs (human umbilical-vein endothelial cells) and HASMCs (human pulmonary arterial smooth muscle cells) were obtained from Kurabo (Osaka, Japan). GA (gallate, 3,4,5-trihydroxybenzoic acid), PG (3,4,5-trihydroxybenzoic acid propyl ester), methyl gallate (3,4,5-trihydroxybenzoic acid methyl ester), octyl gallate (3,4,5-trihydroxybenzoic acid octyl ester), EGCG [(−)-epigallocatechin-3-gallate, (−)-cis-2-(3,4,5-trihydroxyphenyl)-3,4-dihydro-1(2H)-benzopyran-3,5,7-triol 3-gallate], the iron chelator DFX (desferrioxamine), ascorbate and α-tocopherol were obtained from Sigma. CHX (cycloheximide), wortmannin, LY294002, genistein, PD98059, rapamycin, NAC (N-acetylcysteine) and DTT (dithiothreitol) were obtained from Calbiochem. Methanol (1 μl), used as vehicle, was the highest grade available.
The reporter plasmid p2.1, which contains a 68-bp HRE (hypoxia response element) from the human enolase 1 (ENO1) gene inserted upstream of an SV40 (simian virus 40) promoter and Photinus pyralis (firefly) luciferase coding sequences, and the HRE mutant plasmid p2.4 were described previously . The reporter pVEGF-Kpn I contains nucleotides −2274 to +379 of the VEGF gene inserted into the luciferase reporter pGL2-Basic (Promega) . The expression vector pGAL4/HIF-1α(531–826) and the reporter vector pG5E1bLuc, which contains 5 copies of a GAL4 binding site upstream of a TATA sequence and firefly luciferase coding sequences, were described previously . The pCS2-Venus expression vector which encodes Venus, the F64L/M153T/V163A/S175G mutant of green fluorescent protein, was provided by Dr Atsushi Miyawaki (RIKEN, Waco, Japan) . The expression vector encoding a Venus–HIF-1α fusion protein was made from pCS2-Venus and HIF-1α cDNA by inserting HIF-1α sequences 3′ to Venus coding sequences and is designated pVenus-HIF-1α. The expression plasmid pCH-NLS-HIF1α(548–603)-LacZ was described previously . The FLAG-tagged HIF-1α expression vector pcDNA3-FLAG-HIF-1α was described previously .
Whole cell lysates were prepared using ice-cold lysis buffer [0.1% SDS, 1% NP40 (Nonidet P40), 5 mM EDTA, 150 mM NaCl, 50 mM Tris/HCl (pH 8.0), 1 mM sodium orthovanadate and complete protease inhibitor (Roche Diagnostics)] following a protocol described previously . Aliquots (100 μg protein) were fractionated by SDS/PAGE (7.5% gel) and subjected to immunoblot assay using a mouse monoclonal antibody against HIF-1α (BD Biosciences) or HIF-1β (H1β234; Novus Biologicals) at 1:1000 dilution and HRP (horseradish peroxidase)-conjugated sheep antibodies against mouse IgG (GE Healthcare) at 1:1000 dilution. The signal was developed using ECL® (enhanced chemiluminescence) reagents (GE Healthcare).
RT–PCR (reverse transcription–PCR)
The RT–PCR protocol was performed as described previously . Cells were harvested and RNA was isolated with TRIzol (Invitrogen). RNA (1 μg) was subjected to first strand cDNA synthesis using random hexamers (SuperScript II reverse transcriptase kit, Invitrogen). cDNAs were amplified with TaqGold polymerase (Roche) in a thermal cycler with the specific primers (sequences provided on request). For each primer pair, PCR was optimized for cycle number to obtain linearity between the amount of input RT product and output PCR product. Thermocycling conditions were 30 s at 94 °C, 60 s at 57 °C, and 30 s at 72 °C for 25 [VEGF and ITF (intestinal trefoil factor)], 27 (EPO), 25 (HIF1A), or 20 (18S rRNA) cycles preceded by 10 min at 94 °C. PCR products were fractionated by 3% Nusieve agarose gel electrophoresis, stained with ethidium bromide, and visualized with UV.
Reporter gene assay
Reporter gene assays were performed in HEK-293 cells [15,26,27]. Cells (5×104) were plated per well in 24-well plates on the day before transfection. In each transfection, 200 ng of reporter gene, and 50 ng of the control plasmid pRL-SV40 (Promega), containing an SV40 promoter upstream of Renilla reniformis (sea pansy) luciferase coding sequences, were pre-mixed with FuGENE™ transfection reagent (Roche). Cells were incubated with the reagents for 6 h, and then exposed to 20% or 1% O2 for a further 18 h. The cells were harvested and the ratio of firefly to sea pansy luciferase activity was determined using the dual-luciferase reporter assay system (Promega). For each experiment, at least two independent transfections were performed in triplicate and representative results are shown. Results shown represent means±S.D. of three independent transfections. β-Gal (β-galactosidase) activity was determined using a commercial assay system (Boehringer Ingelheim). To normalize β-gal activity of each sample, the pGL-Control plasmid was co-transfected with a β-gal-coding plasmid. The net β-gal count of each sample was divided by its luciferase count, and normalized mean count±S.D. of three independent transfections was shown as relative activity.
Confocal microscopic analysis
Chamber slides were mounted in 90% glycerol with 1 mg/ml p-phenylenediamine and examined with an MRC-1024 confocal microscope (Bio-Rad Laboratories).
In vitro HIF-1α-VHL and HIF-1α-p300 interaction assays
The plasmids used in assays were described previously [10,15]. GST (glutathione transferase)–HIF-1α(429–608) and GST– HIF-1α(531–826) fusion proteins were expressed in Escherichia coli as described previously [10,15]. Biotinylated lysine-labelled proteins were generated in reticulocyte lysates with the TNT T7-coupled transcription/translation system using Transcend™ biotinylated tRNA (Promega), based on the T7-promoter driven plasmids vectors coding VHL and the CH (cysteine/histidine-rich) domain of p300 . Aliquots (25 μg) of HEK-293 cell lysate were pre-incubated with PG or DFX for 30 min at 30 °C, then 2.5 μg of GST–HIF-1α(429–608) was added and incubated for 30 min at 30 °C. A 5 μl aliquot of in vitro-translated biotinylated VHL was mixed with 4 μg of GST fusion protein in a final volume of 200 μl in binding buffer [Dulbecco's PBS (pH 7.4) plus 0.1% Tween-20 ] and incubated for 2 h at 4 °C with rotation followed by addition of 10 μl of glutathione-Sepharose 4B beads (GE Healthcare) and incubation at 4 °C for 1 h. The beads were pelleted, washed 3 times in binding buffer, pelleted again, resuspended in Laemmli sample buffer and analysed by SDS/PAGE. Proteins were transferred to PVDF membrane and visualized using streptavidin-labelled horseradish peroxidase and ECL® reagent.
Gene silencing using siRNA (short interfering RNA)
Caco-2 cells were transfected with 100 nM siRNA using HiPerFect Transfection Reagent (Qiagen) following a protocol provided by the manufacturer.
Transepithelial electrical resistance measurements
Measurement of transepithelial electrical resistance in the Caco-2 cell monolayer was performing using a Millicell ERS (electric resistance system) ohmmeter (Millipore Corporation) as a barrier function assay. This device can measure electric resistance of epithelial cells in culture using a separate pair of Ag-AgCl electrodes and a resistance meter. Fluid resistance was subtracted and net resistance was calculated as ohm/cm2. Results are presented as means±S.D. of three independent wells.
Oral administration of PG and EPO ELISA
PG (100 mg/kg) was administered to male 57BL/6 mice daily for a week by oral gavage. Mice were anesthetized using sevoflurane 24 h after the last administration of PG, and blood was collected via the heart. All precautions were taken to ensure that the animals did not suffer unduly during and after the experimental procedure. The protocol was approved by the institutional animal care committees (The Tazuke Kofukai Medical Institute and Kyoto University). All samples were allowed to clot for 2 h at room temperature (25 °C) before centrifuging for 20 min at 1000 g at 25 °C, and the serum was transferred to a 1.5 ml tube for EPO analysis using a commercial ELISA kit specific for mouse EPO (R&D Systems) according to the manufacturer's instructions. Results are presented as means±S.D. of five mice (ELISA study). Statistical analysis was performed using the Student's t test and a P value <0.05 was considered significant.
Results are presented as means±S.D. Statistics were analysed with Student's t test or, when appropriate, one-way ANOVA followed by Dunnett's test.
PG induces the accumulation of HIF-1α protein under non-hypoxic conditions
To examine the effect of PG on HIF-1 activity, HEK-293 and Hep3B cells were exposed to 100 μM PG under non-hypoxic conditions (20% O2) for 4 h. In both cell lines, 100 μM PG promoted the accumulation of HIF-1α (Figure 1A), similar to the effect of DFX, a known inducer of HIF-1 . Expression of HIF-1β was not affected by PG treatment (Figure 1A). PG also induced HIF-1α accumulation in primary cultures of HUVECs and HASMCs (Figure 1B) without affecting the expression of HIF-1β.
Effect of PG on HIF-1α protein levels in cells
Next, we investigated the dose-dependency of the PG effect on HIF-1α protein expression in HEK-293 cells. PG promoted the accumulation of HIF-1α at doses from 25–400 μM in a dose-dependent manner (Figure 1C). The effect of 100 μM PG peaked at 1 h, which was sustained through to 4 h and then gradually returned to baseline levels by 24 h of treatment (Figure 1C). The time course of HIF-1α accumulation induced by PG is quite different from that induced by DFX (Figure 1C).
PG activates HIF-1-dependent gene expression
PG (100 μM) induced expression of VEGF mRNA and EPO mRNA in HEK-293 cells without affecting the expression of HIF-1α mRNA (Figure 2A), suggesting that PG directly affects HIF-1α protein accumulation, which leads to increased HIF-1 transcriptional activity. To test this hypothesis, HEK-293 cells were transfected with the reporter p2.1, which contains a HIF-1-dependent HRE, or p2.4, which contains a mutation in the HIF-1 binding site . PG induced HRE-dependent gene expression in a dose-dependent manner that was comparable with DFX (Figure 2B). Expression of a dominant negative form of HIF-1α reduced p2.1 reporter gene expression (Figure 2C). Transcription of the mutated reporter p2.4 was not significantly activated by PG (Figure 2D). These results demonstrate that reporter gene activation in PG-treated cells was HRE- and HIF-1-dependent. PG also induced dose-dependent transcription of a luciferase reporter gene containing the VEGF promoter encompassing nucleotides −2274 to +379 relative to the transcription start site (Figure 2E). Thus PG treatment induces expression from both a native promoter and an isolated HRE.
Effect of PG on HIF-1-mediated transcriptional activity
PG induces nuclear localization and prolongs the half-life of HIF-1α protein
We next investigated the subcellular localization of HIF-1α. The fluorescent protein Venus was present in both the nucleus and cytoplasm (Figure 3A, panel a). The Venus–HIF-1α fusion protein was localized mainly in the cytoplasm of untreated, non-hypoxic cells (Figure 3A, panel b). Within 4 h, PG treatment induced translocation of Venus–HIF-1α from the cytoplasm into the nucleus (Figure 3A, panel c), similar to the effect of DFX (Figure 3A, panel d). Next, we examined the intracellular localization of HIF-1α using FLAG-tagged HIF-1α overexpressed in HEK-293 cells. When FLAG–HIF-1α is overexpressed, neither PG nor DFX induces further HIF-1α protein accumulation in HEK-293 cells. However, PG or DFX treatment facilitated nuclear translocation of overexpressed HIF-1α (Figure 3B). Although these results indicated that PG as well as DFX induced at least green fluorescent protein- or FLAG-tagged HIF-1α overexpression by a promoter derived from virus in HEK-293 cells, further analysis is still to be performed.
Effect of PG or DFX on HIF-1α protein stability
To determine whether PG treatment affected the half-life of HIF-1α, HEK-293 cells were treated with 100 μM PG or 100 μM DFX for 4 h to induce HIF-1α expression, and then CHX (100 μM final concentration) was added to block ongoing protein synthesis (Figure 3C). In the presence of CHX, the half-life of HIF-1α was > 60 min in PG-treated cells, which was longer than the half-life of HIF-1α in cells treated with DFX, which is known to block O2-dependent degradation of HIF-1α. Taken together, our results indicate that PG induces the accumulation of HIF-1α protein by causing it to stabilize and by facilitating its nuclear translocation.
Effect of other gallates on HIF-1α accumulation
To explore the molecular mechanism by which PG induces HIF-1α, we tested GA and EGCG, which are compounds that are structurally related to PG. As shown in Figure 4(A), PG induced the accumulation of HIF-1α (lanes 2 and 3) but neither GA (lanes 4 and 5) nor EGCG (lanes 6 and 7) promoted accumulation at any dose tested in HEK-293 cells. Consistent with these results, the reporter gene assay demonstrated that only PG induced HRE-dependent gene expression (Figure 4B). To explore the mechanisms behind PG-induced HIF-1 activation, we examined trihydroxy compounds with similar structures to PG. The effects of methyl gallate, ethyl gallate, and octyl gallate were examined in addition to GA and PG (Figure 4C). Octyl gallate (100 μM, lane 6) induced greater HIF-1α accumulation than 100 μM PG (lane 5). In contrast, 100 μM methyl gallate (lane 3) and ethyl gallate (lane 4) had a weaker effect compared with 100 μM PG.
Effect of gallates, antioxidants, and kinase inhibitors on HIF-1α protein levels and HRE-dependent gene expression in HEK-293 cells
Next, we performed experiments focused on HIF-1α hydroxylases. Both classes of HIF-1α hydroxylases, PHDs and FIH, are members of the family of 2-oxoglutarate dioxygenases. The enzymes require Fe2+ and ascorbate as cofactors, and 2-oxoglutarate and dioxygenase serve as prime substrates for the reaction. We therefore investigated the effect of 2-oxoglutarate, ascorbate, and Fe2+ on PG-induced HIF-1α accumulation. As shown in Figure 4(D), 15 mM 2-oxoglutarate inhibited PG-induced HIF-1α accumulation. Fe2+ (100 μM) also suppressed the accumulation, but the effect was weaker than that of 15 mM 2-oxoglutarate. In contrast, 100 μM ascorbate did not affect PG-induced accumulation.
Gallates are reported to have potent antioxidative effects in vitro and in vivo [29,30]. To test whether PG induces the accumulation of HIF-1α via its antioxidant activity, we treated HEK-293 cells with other reagents that have antioxidant activity. DTT, NAC, ascorbate or α-tocopherol did not induce HIF-1α protein accumulation (results not shown). Expression of the potent intracellular redox regulator thioredoxin or Cu/Zn superoxide dismutase did not induce HIF-1α accumulation in HEK-293 cells (results not shown). Moreover, the PI3K (phosphoinositide 3-kinase) inhibitor LY294002, the tyrosine kinase inhibitor genistein, the MEK [MAPK (mitogen-activated protein kinase)/ERK (extracellular-signal-regulated kinase) kinase] inhibitor PD98059, the p38 MAPK inhibitor SB203580 and the mTOR (mammalian target of rapamycin) inhibitor rapamycin did not inhibit PG- or DFX-induced HIF-1α accumulation in HEK-293 cells (results not shown).
PG inhibits HIF-1α hydroxylases
We examined whether prolyl or asparaginyl hydroxylase activity is affected by PG using in vitro pulldown assays. Incubation of a GST–HIF-1α(429–608) fusion protein with lysate from untreated cells demonstrated prolyl hydroxylation of HIF-1α as determined by its interaction with VHL (Figure 5A, lane 2), which is hydroxylation-dependent. Lysate from cells treated with PG or DFX did not promote the interaction of GST–HIF-1α(429–608) with VHL (Figure 5A, lanes 3–5). We also examined the effect of GA using two different protocols. In one protocol, cell lysates were prepared from untreated and GA-treated cells (Figure 5B, left panel). In the other protocol, lysate was directly incubated with GA or vehicle (Figure 5B, right panel). Lysate from GA-treated cells promoted the interaction of GST–HIF-1α(429–608) with VHL (Figure 5B, left panel), similar to the effect of lysate from untreated cells.
Effect of PG on the activity of HIF-1α hydroxylases
Next, we examined asparaginyl hydroxylase activity in PG-treated HEK-293 cell lysate. A GST–HIF-1α(531–826) fusion protein encompassing Asn803 was incubated with lysate and its interaction with the CH1 domain of p300 was tested. As shown in Figure 5(C), PG treatment promoted the interaction between HIF-1α and p300, which was similar to the effect of DFX treatment.
Next, we investigated the impact of PG on HIF-1α transcriptional activity. There are two independent TADs (transactivation domains) present in HIF-1α, which are designated as the N-terminal (amino acids 531–575) and C-terminal (amino acids 786–826) TADs (TAD-N and TAD-C respectively) . A fusion protein consisting of the GAL4 DNA-binding domain fused to HIF-1α amino acid residues 531–826, which contains both of the TADs, is expressed at similar levels under hypoxic and non-hypoxic conditions and thus can be used to examine the transcriptional activity of HIF-1α independent of its protein expression [10,15,22]. PG treatment increased transactivation mediated by GAL4–HIF-1α(531–826) in a dose-dependent manner, similar to the effect of DFX (Figure 5D). These results demonstrate that PG promotes HIF-1α accumulation and HIF-1α transcriptional activity by inhibiting prolyl and asparaginyl hydroxylase activity respectively.
PG treatment enhances cellular barrier function
Epithelial cells provide a barrier against external antigens and bacteria. Mucosal organs, including stomach and intestine, are dependent on an extensive underlying vasculature, and therefore are susceptible to hypoxic-ischaemic tissue damage. We exposed intestinal epithelial cell lines to PG under non-hypoxic conditions and examined HIF-1 activity and epithelial barrier function. Caco-2, HT-29, SW48, and T89 cells, which were derived from colorectal carcinomas, were exposed to PG at concentrations ranging from 50 to 400 μM. Compared with vehicle-treated cells, PG induced HIF-1α protein expression in a dose-dependent manner in all of the cell lines tested (Figure 6A). We examined whether PG affects the expression of two genes that are regulated by HIF-1 by using RT–PCR to quantify mRNA expression in Caco-2 cells. PG induced the expression of ITF and VEGF mRNA to levels that were similar to those induced by hypoxia (Figure 6B). In contrast, HIF-1α mRNA expression was not affected by exposure of the cells to PG or hypoxia. Pre-treatment of cells with a siRNA that targets HIF-1α mRNA for degradation blocked the induction of ITF mRNA expression in response to PG (Figure 6C).
Effect of PG on cells derived from intestinal epithelium and serum EPO expression
To determine whether PG altered the function of the mucosal barrier, cultured intestinal epithelial cells were treated with vehicle, PG or GA, and the electrical resistance of the cells, which is a measure of barrier function, was examined using the Millicell-ERS ohmmeter system. Transepithelial resistance was increased 24 to 36 h after PG treatment, whereas treatment with vehicle or GA had no significant effect (Figure 6D).
PG induces EPO expression in vivo
Preclinical results suggest that a HIF-1 activator might be useful for the treatment of anaemia and ischaemic diseases [16,31]. HIF prolyl and asparaginyl hydroxylation can be inhibited with small organic molecules [32,33], leading to increased HIF activity. To explore whether PG can induce HIF-1 and its downstream gene expression in vivo, we administrated PG to mice and examined serum levels of the erythropoietic cytokine EPO, which is the product of a HIF-1 target gene. Daily administration of PG to mice by oral gavage increased circulating EPO levels (Figure 6E).
The results reported above demonstrate that treatment of several different cell types with PG induces HIF-1α protein accumulation and HIF-1 transcriptional activation, resulting in HIF-1-regulated gene expression. PG treatment inhibited the interaction between HIF-1α and VHL, increased the half-life of HIF-1α protein, induced nuclear translocation of overexpressed HIF-1α, promoted the interaction between HIF-1α and p300, and stimulated HIF-1α transactivation function. We demonstrated that PI3K, tyrosine kinase, MAPK and mTOR inhibitors do not have any significant effects on PG-induced HIF-1α accumulation. In contrast, MAPK, PI3K and certain tyrosine kinases play significant roles in growth factor- or PGE2 (prostaglandin E2)-induced HIF-1α accumulation, in which increased HIF-1α translation is induced [12,13,34,35]. The evidence indicates that PG is targeting a general mechanism in the HIF-1 signalling pathway, similar to hypoxia. In fact, PG suppresses prolyl HIF-1α hydroxylase activity, which determines the stability of HIF-1α protein . Because O2 is a required substrate for prolyl hydroxylase activity, these enzymes provide a direct link between reduced O2 availability and adaptive responses to hypoxia that are mediated by HIF-1. In addition, the asparaginyl hydroxylase FIH-1 governs the interaction between HIF-1 and the transcriptional co-activator p300 [10,11,36]. PG also enhances the transcriptional activity of HIF-1α by suppressing the asparaginyl hydroxylase activity. Thus PG inhibits the intracellular O2 sensing system by blocking both prolyl and asparaginyl hydroxylase activity under non-hypoxic conditions.
Previous studies reported that green tea catechins such as EGCG induced HIF-1 activation in T47D human breast carcinoma and PC3 human prostate cancer cells [37,38]. In contrast, we did not observe an effect of EGCG on HIF-1 activity in HEK-293 cells, Hep3B cells, HUVECs, or HASMCs. The varying effectiveness of different gallates as inducers of HIF-1α activity may reflect differences in their ability to enter different cell types. The cell membrane permeability of PG is approx. 10-fold higher than that of GA . The results shown in Figure 5(B) are consistent with this hypothesis, as GA inhibited prolyl hydroxylase activity when added directly to lysates but not when added to intact cells.
Because gallates have iron chelating activity, the effect of PG may be attributable to this property. However, GA and EGCG, which have similar iron-chelating activity, did not induce HIF-1α accumulation (Figures 1A and 1B). Cell lysates from cultured HEK-293 cells treated with PG, but not with GA, did not suppress the interaction between HIF-1α and VHL (Figure 5B). Incubation with an equimolar amount of FeSO4 to PG (100 μM) only partially reversed the effect of PG (Figure 4D). In addition, it is reported that PG only has a weak binding constant for Fe2+ . Finally, we indicated that as little as 25 μM PG significantly induced HIF-1α accumulation in HEK-293 cells. We also observed that, in HUVECs, PG as low as 4 μM was enough to induce the accumulation of HIF-1α (results not shown). It is thus unlikely that PG activates HIF-1 by chelating Fe2+. Gallates including PG also have potent antioxidant activity . The finding that treatment with NAC, DTT or α-tocopherol did not affect HIF-1 activity also suggests that the effect of PG cannot be explained by its antioxidant activity. In Figure 4(D), we demonstrated that 15 mM 2-oxoglutarate effectively inhibited the interaction between HIF-1α and VHL. We also previously suggested that GA binds to PHDs by mimicking the structure of 2-oxoglutarate in the PHD active site, and demonstrated that PG added into culture medium passes through the plasma membrane and is hydrolysed into GA . We conclude that PG added to culture medium enters cells and the hydrolysed product, GA, directly inhibits prolyl and asparaginyl hydroxylases of HIF-1α, resulting in HIF-1α protein stabilization and transcriptional activation.
PG is approved as a food preservative that is added to prevent oxidation and is regarded as safe. The implications of a dietary compound affecting the activity of a protein as pivotal as HIF-1 in the process of angiogenesis cannot be overlooked. Increased stabilization of HIF-1α might lead to increased angiogenesis through the co-ordinated induction of HIF-1 target genes encoding the angiogenic growth factors VEGF, platelet-derived growth factor B and placental growth factor [40,41]. Several studies have demonstrated that prolonged expression and activation of HIF-1α alone is sufficient to induce mature, physiologically functional blood vessels and that even transient stabilization of HIF-1α might lead to changes in cellular metabolism, such as increased glucose uptake and glycolysis, that would allow cells to survive acute hypoxic insults . For example, DFX administration has been shown to reduce infarct size in a rat stroke model . In this study, we showed that PG induces ITF, which protects mucous epithelia from a range of insults, contributes to mucosal repair , and maintains barrier function of epithelial cells (Figure 6D). In addition, oral administration of PG increased the levels of EPO mRNA in the kidney and circulating EPO protein in mouse blood (Figure 6E). Taking into account evidence that EPO confers a powerful protective effect to various organs and tissues [44,45], PG may be useful as a lead compound for the development of novel drugs that protect against ischaemic injury. Finally, our findings may have implications for people with diets that are unusually high in PG, such as those consuming ‘health drugs’ that contain this compound. Additional studies are required to investigate further the protective or pathological consequences of PG intake.
We thank Dr Atsushi Miyawaki for providing plasmids. This work was supported in part by a grant-in-aid for scientific research from the Ministry of Education, Culture, Sports, Science and Technology (Tokyo, Japan) to T. A. and K. H.
Dulbecco's modified Eagle's medium
electric resistance system
fetal bovine serum
factor inhibiting hypoxia-inducible factor
human pulmonary arterial smooth muscle cell
- HEK-293 cell
human embryonic kidney cell
hypoxia response element
human umbilical-vein endothelial cell
intestinal trefoil factor
mitogen-activated protein kinase
mammalian target of rapamycin
prolyl hydroxylase domain-containing protein
short interfering RNA
simian virus 40
vascular endothelial growth factor
von Hippel–Lindau protein
These authors contributed equally to this work.