|Home | About | Journals | Submit | Contact Us | Français|
Hypoxia-inducible factor-1 alpha (HIF-1α) and purine nucleosides adenosine and inosine are critical mediators of physiological responses to acute and chronic hypoxia. The specific aim of this paper was to evaluate the potential role of HIF-1α in purine-mediated neuroprotection. We show that adenosine and inosine efficiently rescued PC12 cells (up to 43.6%) as well as primary cerebellar granule neurons (up to 25.1%) from hypoxic insult, and furthermore, that HIF-1α is critical for purine-mediated neuroprotection. Next, we studied hypoxia- or purine nucleoside-increased nuclear accumulation of HIF-1α in PC12 cells. As a possible result of increased protein stabilization or synthesis an up to 2.5 fold induction of HIF-1α accumulation was detected. In cerebellar granule neurons, purine nucleosides induced an up to 3.1 fold HIF-1α accumulation in cell lysates. Concomitant with these results, siRNA-mediated reduction of HIF-1α completely abolished adenosine- and inosine-mediated protection in PC12 cells and severely hampered purine nucleoside-mediated protection in primary neurons (up to 94.2%). Data presented in this paper thus clearly demonstrate that HIF-1α is a key regulator of purine nucleoside-mediated rescue of hypoxic neuronal cells.
The reduction of apoptosis in the brain is central to functional recovery after a stroke. Adenosine and its receptors were proposed as targets for therapeutic approaches to treat stroke and related disorders (Rudolphi et al. 1992, review; Sweeney 1997; von Lubitz 1999). Adenosine is the final metabolite in the stepwise dephosphorylation of ATP and it is produced and released in response to ischemia and hypoxia in the central nervous system (Winn et al. 1981; Zetterstrom et al. 1982; Phillis et al. 1994; Melani et al. 1999; Latini and Pedata 2001, review). It acts as a powerful endogenous neuroprotectant during ischemia-induced energy failure (Higgins et al. 1994; Brundege and Dunwiddie 1997), by decreasing neuronal metabolism, increasing cerebral blood flow, and by playing a variety of different roles as an intercellular messenger. These effects are mediated through the interaction of adenosine with specific receptors (Rudolphi et al. 1992; Sweeney 1997; von Lubitz 1999; Kobayashi et al. 1998), and this stimulation was hypothesized to result in an effective treatment for stroke (Dunwiddie and Masino 2001, review). Likewise, the adenosine derivative inosine was shown to have protective effects against insults related to ischemia and reperfusion (Shen et al. 2005), to induce neurite outgrowth (Benowitz et al. 1998), and to stimulate the extension of new neuronal projections into denervated areas in adult rats with unilateral cortical infarcts (Chen et al. 2002). Our own data showed that purine nucleosides protected neuronal cells from rotenone-induced cell death (Bocklinger et al. 2004; Heftberger et al. 2005). The discovery that purine nucleosides play a role in endogenous neuroprotection has given rise to extensive efforts towards developing future ischemia/reperfusion drug therapies. To achieve such a goal, however, a complete understanding of the intracellular signaling mechanisms involved in purine-mediated neuroprotection is required.
HIF-1α is a transcription factor that plays an essential role in cellular and systemic homeostatic responses to hypoxia (Semenza 2000). Under hypoxia, the HIF-1α protein is stabilized and translocation into the nucleus is increased. In the nucleus, HIF-1α associates with HIF-1β to form the active transcription factor complex. The target genes of the HIF-1 complex are involved in energy metabolism and cell viability. The HIF-dependent hypoxic response pathway plays a prominent role in mediating the consequences of many disease states, including cerebral ischemia (for review, see Semenza 2000). Along these lines, it was recently demonstrated (Baranova et al. 2007) that HIF-1-mediated responses have an overall beneficial role in the ischemic brain.
The aim of this study was to determine (1) whether adenosine and its derivative inosine might regulate the cellular response to hypoxia in neuronal cells, and (2) whether HIF-1α might contribute to adenosine-mediated neuroprotection. Two cell models were used: first we studied the O2-sensitive (Zhu et al. 1996; Seta et al. 2002) clonal rat pheochromocytoma (PC12) cell line, a widely used model for sympathetic ganglion-like neurons (Greene and Tischler 1976). PC12 cells express abundant A2A adenosine receptors (A2AR) (Arslan et al. 1999; Hide et al. 1992; van der Ploeg et al. 1996; Tomaselli et al. 2005b), which are involved in cellular responses to hypoxia (Kobayashi et al. 1998; Kobayashi and Millhorn 1999). Second, we studied a primary cell model, cerebellar granule neurons, which express predominantly the A1R (A1 receptor), intermediate levels of A3R, and negligible amounts of A2R (Wojcik and Neff 1983; Heftberger et al. 2005; for reviews see: Fredholm et al. 2001b; Ribeiro et al. 2002). The latter cells were also used previously to study the effects of oxygen and glucose deprivation followed by reoxygenation (Scorziello et al. 2001).
1) PC12 cells (LGC Promochem, Germany) were cultured on collagen-S type I (BD Biosciences, Austria) coated culture dishes in RPMI 1640 medium (PAA Laboratories, Austria) supplemented with 10% horse serum (Gibco, Austria), 5% fetal calf serum (PAA Laboratories, Austria), 1% Pen/Strep (PAA Laboratories, Austria), and 1% L-glutamine (PAA Laboratories, Austria) and maintained at 37°C under 5% CO2. Cells were split once per week and one or two days before beginning an experiment. For all experiments hypoxia was induced in an incubator designed for maintaining hypoxia (HERAcell 240, Kendro, Austria) that was set to continuous conditions of 1% O2 balanced with N2, and 5% CO2, maintained at 37°C, and controlled with O2- and CO2-sensors. Small culture volumes guaranteed the fast equilibration of medium to the 1% oxygen atmosphere.
Prior to incubating the PC12 cells under reduced oxygen, the RPMI medium was changed to serum reduced medium (RPMI 1640: 1.25% horse serum, 0.63% fetal calf serum, 1% Pen/Strep, and 1% L-glutamine). In order to account for the eventual loss of cells due to the reduction in serum, we matched all experimental assays with control cells treated under the same conditions. The purine nucleosides adenosine and inosine (Sigma, Germany) were dissolved in RPMI 1640 medium and maintained in a 10mM stock solution at −20°C. Prior to experiments, adenosine and inosine were added directly to the cells to obtain final concentrations of 500μM. Cells were exposed to normoxia or hypoxia for 24 hours. For control cells, the medium was also changed, but no further drugs were added. For CoCl2 controls (cobalt stabilizes HIF-1α by inhibiting its hydroxylation), cells were incubated for 4 h with with 100μM CoCl2 (stock solution 10mM in serum reduced medium).
Equal concentrations of nucleosides were used based on the observation in vivo that interstitial levels of adenosine and inosine are roughly similar under both basal conditions and after hypoxia or ischemia (reviewed by Fredholm et al. 2001a). The initial concentrations of 500 μM for each nucleoside was based on observations that basal adenosine levels in interstitial fluid are between 30 and 300nM and increase some 10-fold during hypoxia and 100- to 1000- fold in ischemia (Zetterstrom et al. 1982; Hagberg et al. 1987; Dux et al. 1990), and (Ballarin et al. 1991; Rudolphi et al. 1992; Latini et al. 1999; Pearson et al. 2003). Thus, based on (1) data in the literature (Litsky et al. 1999; Gysbers and Rathbone 1996; Lynch et al. 1998), (2) our own experience with chemically-induced hypoxia (Tomaselli et al. 2005b; Tomaselli et al. 2005a), and (3) the fact that purine nucleosides are only added once during the experiment, we used adenosine and inosine at a concentration of 500μM throughout our studies.
We used nucleosides purified by HPLC chromatography throughout the studies. In our experience the purine nucleosides we used have shown very low toxicity (measured by FACS analysis of propidium iodide stained- as well as by microscopic analysis of Hoechst/propidium iodide stained-PC12 cells) even after longer incubation periods (B.T. and V.P. unpublished data).
2) Cerebellar granule cells were isolated from Sprague-Dawley rats. Briefly, the whole cerebellum was removed at postnatal day 7 (p7) and meninges were carefully stripped off. The tissue was mechanically dissociated and then digested with 1x trypsin (Invitrogen, Austria) for 15 min at 37° C. The digestion was stopped with trypsin inhibitor (520μg/ml) (Gibco, Austria) combined with DNAse (100μg/ml) (Roche, Germany). To remove cell aggregates, the cell suspension was filtered through a cell strainer (40μm mesh size, BD Falcon, Bedford (MA), USA). Cerebellar granule neurons were further enriched by preplating on an untreated culture dish for 1 hour in full medium (FM, ‘Neurobasal medium’ (Invitrogen, Austria), supplemented with 2% B-27 (Invitrogen, Austria), 1% glutamine, and 1% Pen/Strep (PAA Laboratories, Austria). This cerebellar granule cell preparation yielded over 90% neurons (B.T., V.P. unpublished data), and therefore cells are called granule neurons throughout the paper. For transfection experiments a two-step gradient of Percoll (35/65% in PBS-EDTA, GE Healthcare Biosciences, Sweden) was used for further purification. Neurons were then seeded on culture plates coated with poly-L-ornithine (Sigma, Germany) in FM. After 3 days in culture, the medium was changed, drugs were added, and cells were maintained for 6-8 hours under normoxic /hypoxic conditions in the presence or absence of adenosine/inosine (500μM).
Compared to PC12 cells, primary cerebellar granule neurons (1) showed a higher spontaneous loss of viability and (2) died more rapidly under oxygen deprivation for 24 hours. Based on this experience, we decided to incubate primary neurons for a shorter time (6-8 h) in low oxygen.
Cell death was quantified by staining with Hoechst 33342 (Molecular Probes, Oregon, USA) and propidium iodide (Sigma, Germany). Hoechst penetrates both living and dead cells and was used to estimate total cell number, and propidium iodide is membrane impermeable and therefore only stains dead cells. Cells were incubated with 10 μg/ml Hoechst 33342 dissolved in 1x PBS for 10 minutes followed by 5μg/ml propidium iodide for 5 minutes at 37°C. Pictures were taken under an inverted fluorescence microscope (Zeiss Axioplan2, Austria) equipped with a spot camera (RT-slider 2.3.1 Visitron Systems, Germany) using fluorescence filters for Hoechst and propidium iodide. Pictures of both fluorescence excitations were superimposed in ‘Adobe Photoshop 7.0’ and the percentage of cell death was calculated as follows: percent of dead cells = propidium iodide positive cells x 100/total cell number. Protection of cell viability was calculated as follows: percent protected cells = 100-(%cell death of treated cells x100)/ % cell death of control cells).
PC12 cells were cultured in RPMI 1640 medium with supplements in 4 well chambers (Nunc, Rochester (NY), USA) for 1-2 days. Cells were stimulated with adenosine/inosine under normoxic/hypoxic conditions for 24 hours. Pictures were taken from 4-5 independent microscopic fields and neurite-bearing cells (cells with neurites longer than one cell diameter) were counted.
For western blot analysis of total cell extracts, PC12 cells and cerebellar granule neurons were cultured in 60mm plates. The medium was aspirated and 1x lysis buffer was added directly to the cells. Lysis buffer contained 100mM Tris pH 8.5 (Merck, Germany), 2% NP-40 (Applichem, Germany), 10mM EDTA (Applichem, Germany), 10mM NaP-P (Sigma, Germany), 10mM NaF (Sigma, Austria), 100mM NaCl (NeoLab Migge, Germany), 10mM Na3Vo4 (Sigma, Germany), 60μg/ml aprotinin (Sigma, Germany), and 60μg/ml leupeptin (Sigma, Germany). Cells were removed from the plates with a cell scraper, transferred to a precooled tube, and kept on ice for 20 minutes with frequent vortexing. After centrifugation (14000 x g, 4°C, 15 minutes) supernatants were collected and mixed with 4x Laemmli sample buffer composed of 40% glycerin (Merck, Germany) and 240mM TRIS pH 6.8, 4% SDS, 0.008% bromophenol-blue sodiumsalt, and 0.2M mercaptoethanol (all from Applichem, Germany). Samples were boiled at 95°C for 5 minutes and stored at −20°C for western blot analysis.
For the HIF-1α transcription factor assay we used a nuclear extraction kit (TransAM HIF-1α Transcription Factor Assay kit; Active motif, Belgium). With its patented TransAMTM method*, Active Motif introduced the first ELISA-based kits to detect and quantify transcription factor activation. The TransAM HIF-1 Kit includes a 96-well plate that features immobilized oligonucleotides containing the hypoxia response element (HRE). HIF dimers present in nuclear extracts bind with high specificity to this response element and are subsequently detected with an antibody directed against HIF-1α. Addition of a secondary antibody conjugated to horseradish peroxidase (HRP) provides a sensitive colorimetric readout that is easily quantified by spectrophotometry. Briefly, 4×107 cells (two 100mm plates) were harvested and centrifuged at 50 x g, 4°C for 5 minutes. The pellet was resuspended in 500μl 1x hypotonic buffer (Active Motif, Belgium) and the suspension was incubated for 15 minutes on ice. Next, 20μl of detergent (Active motif, Belgium) was added, the suspension was vortexed, and then it was centrifuged at 14000 x g, 4°C, for 30 seconds. The nuclear pellet was then resuspended in 50μl complete lysis buffer (Active Motif, Belgium), vortexed, and incubated for 30 minutes on ice on a rocking platform. After centrifugation (14000 x g, 4°C, 10 minutes), the supernatants (nuclear fraction) were stored at −80°C. They were subsequently used for the HIF-1α transcription factor assay analysis and for western blot analysis.
Equivalent cell concentrations (~6×106cells/total cell extract and ~1×107 cells/nuclear extract) were loaded on a 7.5% SDS-polyacrylamide gel and run for 30 minutes at 80 V (stacking gel) and 1.5h at 120 V (separating gel). Proteins were transferred from the gel onto a nitrocellulose membrane (GE Healthcare Biosciences, Sweden) by a wet-blotting system (Bio-Rad, Austria) for 1h 20min at permanent 100 V. Membranes were blocked for 1h at room temperature in 5% skim milk (dissolved in 1xTBS) and subsequently immunoprobed overnight on a rocking platform at 4°C with an HIF-1α antibody (New England Biolabs, Germany, 1:1000 dilution or Santa Cruz Biotechnology, Santa Cruz (CA), USA, 1:400 dilution; the diluting solution was 5% BSA dissolved in 1xTBS-T). On the next day membranes were washed with 1xTBS-T (3 times for 10 minutes) and incubated with a secondary antibody coupled to horseradish-peroxidase (goat-anti-rabbit-HRP-linked, Pierce, Rockford (IL), USA, 1:2500 dilution with 5% skim milk dissolved in 1xTBS) for 1 hour at room temperature. After further washing in 1x TBS-T, immunoreactive bands were visualized by using the ECL-kit (Pierce, Rockford (IL), USA) and exposing a photographic film (GE Healthcare Biosciences, Sweden) to the membrane. The films were removed for developing and membranes were stripped in 0.1M glycine (NeoLab Migge, Germany) for 10 minutes and reprobed with DNA Pol δ cat (Santa Cruz Biotechnology, Santa Cruz (CA), USA, 1:500 dilution, in 5% TBS-T). The immunoprobing, washing, and exposure procedures were repeated as described above. For quantification, films were scanned with a Molecular Dynamics Personal Densitometer SI scanner. To normalize the HIF-1α signal of each assay to the protein amount loaded, the ratio of HIF-1α and DNA polymerase was calculated as follows: ratio = HIF-1α signal/DNA polymerase signal.
Cells were fixed with 4% paraformaldehyde (Sigma, Germany) for 20 minutes. After washing with 1x PBS cells were permeabilized with 0.3% Triton X-100 (Applichem, Germany) diluted in 1x PBS for 10 minutes. Then a blocking solution (3% BSA (Calbiochem, Germany) with 0.1% Triton X-100 diluted in 1xPBS) was added to the cells for 30 minutes at room temperature. Cells were incubated with the HIF-1α antibody (Santa Cruz Biotechnology, Santa Cruz (CA), USA, diluted 1:1000 in 1% BSA with 0.1% Triton X-100 in 1xPBS) over night at 4°C. After another wash the goat-anti-rabbit-FITC-labeled secondary antibody (Molecular Probes, Oregon, USA, 1:1000 in 1% BSA) and 10μg/ml Hoechst 33342 were added for one hour. Cells were covered in fluorescence mounting medium (Dako, Austria) and pictures were taken under an inverted fluorescence microscope (Zeiss Axioplan2, Austria) equipped with a spot camera (RT-slider 2.3.1 Visitron Systems, Germany).
‘Nucleofector technology’ (Amaxa Biosystems, Germany) was used for transient transfection. Synthetic siRNA that targeted HIF-1α (siGENOME SMARTpool reagent, Dharmacon, Chicago (IL), USA) was used as a probe and a pool of scrambled non-targeting siRNA duplexes (control siRNA, Dharmacon, Chicago (IL), USA) were used as negative controls. The transfection of PC12 cells was performed according to a general protocol for nucleofection of adherent cell lines described in the instructions for the ‘Neuronal Cell Line NucleofectorTM kit V’ (Amaxa biosystems, Germany). Approximately 1.7×107 cells were transfected with 100 nM HIF-1α siRNA duplexes and control cells were transfected in parallel with non-targeting siRNA. Briefly, the cell suspension was transferred to an Amaxa cuvette and cells were nucleofected using the U-29 program. Immediately after transfection the cells were transferred with the recommended plastic pipettes into provided culture dishes. Two days after transfection cells were stimulated with purine nucleosides for 24 hours in serum-reduced medium under normoxic and hypoxic conditions.
The transfection of cerebellar granule neurons was performed according to the general protocol for nucleofection of primary neurons (Gartner et al. 2006) described in the instructions for the rat neuron Nucleofector kit (Amaxa biosystems, Germany). Approximately 1×107 cells were transfected with 1μM HIF-1α siRNA duplexes and control cells were transfected in parallel with 1μM non-targeting siRNA. Briefly, the suspension was transferred to an Amaxa cuvette and nucleofected using the O-03 program. Afterwards, DMEM medium (PAA Laboratories, Austria) supplemented with 10% horse serum, 1% glutamine, and 1% Pen/Strep was added and cells were cultured on poly-L-ornithine coated glass cover slips (Assistant, Germany). After three hours the medium was changed to ‘Neurobasal medium’. Three days after transfection cells were stimulated with purine nucleosides for 6-8 hours under normoxic and hypoxic conditions. Hoechst 33342 and propidium iodide staining were performed and cell death estimated. The knockdown of HIF-1α was assessed by western blot analysis.
All values in the figures were expressed as the mean ± S.E.M. The SPSS 15.0 statistic program was applied for analysis of samples. The unpaired one-tailed t-test was used to compare independent groups (normoxia vs. hypoxia, purine nucleosides vs. control and siRNA knockdown of HIF-1α vs. siRNA scrambled control values). p-Values of <0.05 were considered statistically significant (*p<0.05, **p<0.01, ***p<0.001) .
After incubating PC12 cells in low oxygen (1%) for 24 hours cell death was quantified by fluorescence microscopy analysis (Fig. 1Aa-d). Hypoxia induced a 3.3 fold increase in PC12 cell death (Fig. 1B). However, cells were significantly rescued in the presence of adenosine (43.6%), and inosine (31.0%) (Fig. 1C). Neurite outgrowth was investigated (Fig. 1D-F) in parallel. Neurite formation was increased under hypoxic conditions by 2.3 fold (Fig. 1D,E), and was further enhanced by the presence of adenosine (1.5 fold) and inosine (1.1 fold) (Fig. 1D,F).
After incubating cerebellar granule neurons under reduced oxygen (1%) for 6-8 hours dead cells were quantified by fluorescence microscopy (Fig. 2Aa-d). Hypoxia induced a 1.5 fold increase in neuronal cell death (Fig. 2B). Primary neurons were partially rescued in the presence of adenosine (25.1%), and inosine (23.0%) (Fig. 2C). In addition, neurite outgrowth in granule neurons (Fig. 2Ae-h) reflected the cell fate after hypoxia as well as the protection by adenosine and inosine.
PC12 cells were incubated for 24 h in 1% O2. In the subsequent western blot analysis, we compared ratios of HIF-1 α/DNA polymerase in total cell extracts (see Fig.3Aa and b) to ratios in purified nuclei (see Fig.3Ba and b). The ratios in total cell extracts were 0.24 ± 0.18% (normoxia) and 0.57 ± 0.22% (hypoxia), and in nuclear extracts 0.32 ± 0.09% (normoxia) and 0.62 ± 0.13% (hypoxia). Interestingly, this effect was augmented in total cell extracts by adenosine (1.5 fold) as well as inosine (1.4 fold) (Fig. 3Ab), in nuclear extracts the augmented effect was enhanced with adenosine (2.5 fold) and similar with inosine (1.4 fold) (Fig. 3Bb). CoCl2 treated samples are shown for comparison (Fig. 3Ba and Fig. 3C, panels a, f, and k). Concomitant with western blot analysis, fluorescence analysis revealed that incubation of PC12 cells with inosine and particularly with adenosine enriched the nuclear HIF-1α signal (Fig. 3C, panels a-j). Next, the transactivation potency of HIF-1α was analyzed by an electrophoretic mobility shift assay (EMSA). Results showed that adenosine and inosine induced a 1.1 fold activation of HIF-1α binding capacity compared to controls (Fig. 3D).
In cerebellar granule neurons hypoxia induced a 2.5 fold increase of HIF-1α protein measured by western blot analysis of total cell extracts (Fig. 4Aa). This effect was significantly augmented with adenosine (3.1 fold) and inosine (2.7 fold) (Fig. 4Ab). In the nuclei of granule neurons hypoxia increased the amount of HIF-1α detected by fluorescence analysis. Again, this effect was significantly augmented with adenosine and inosine (Fig. 4Ba-h).
To further study the importance of HIF-1α for the protection of hypoxic PC12 cells a siRNA-approach was employed to specifically knockdown HIF-1α protein expression. Partial knockdown of HIF-1α sensitively hampered the HIF-1α protein levels in PC12 cells (42.0% downregulation, Fig. 5A). Analysis of Hoechst 33342 and propidium iodide stained cells (Fig. 5Aa-f) revealed that siRNA-mediated knockdown of HIF-1α led to a 34.7% increase in cell death following incubation in reduced oxygen. Purine nucleoside-mediated rescue of cell viability was completely blocked upon knockdown of HIF-1α (Fig. 5Ag), though adenosine and inosine efficiently rescued cells transfected with control siRNA. In parallel assays, we observed a 46.1% inhibition of neurite formation after siRNA-mediated knockdown of HIF-1α, and 14.0% and 15.4% inhibitions of the slight increases in neurite outgrowth mediated by adenosine and inosine, respectively (Fig. 5Bg).
Western blot analysis of total cell extracts showed that knockdown of HIF-1α sensitively hampered the HIF-1α protein levels in cerebellar granule neurons (39.2% downregulation, Fig. 6A). Analysis of Hoechst 33342 and propidium iodide stained cells (Fig. 6Aa-l) revealed that siRNA-mediated knockdown led to an 11.0% increase in cell death during hypoxia compared to cells transfected with control siRNA. Adenosine- and inosine- mediated rescue of cell viability was significantly reduced by 94.2% and 61.8%, respectively, upon knockdown of HIF-1α (Fig. 6B), though adenosine and inosine efficiently rescued viability in hypoxic cells transfected with control siRNA.
As a consequence of acute reduction in oxygen tension we observed increased cell death of PC12 cells and primary cerebellar granule neurons. To determine whether purine nucleosides may attenuate hypoxic cell death, cells were exposed to hypoxia in the presence of adenosine and inosine. In agreement with results by others (Huffaker et al. 1984; Braumann et al. 1986; Gysbers and Rathbone 1996; Kobayashi and Millhorn 1999; Muroi et al. 2004; Ribeiro 2005; Tomaselli et al. 2005b), the data presented here demonstrate that adenosine and inosine efficiently rescue the viability of hypoxic PC12 cells and primary cerebellar granule neurons. Earlier data suggested that the effects of adenosine were apparently the result of its conversion to inosine by adenosine deaminase (Haun et al. 1996). In contrast, a more recent study (Benowitz et al. 1998) showed that the addition of adenosine induced goldfish retinal ganglion cells to extend lengthy neurites, and that these effects were highly specific and did not reflect conversion of the nucleosides to their derivatives. The same authors also showed that the action of inosine was not due to its hydrolysis to hypoxanthine, since hypoxanthine was inactive in retinal ganglion cells. Along these lines, our previous results in primary neurons showed that the activity of adenosine is not blocked by EHNA, an inhibitor of adenosine deaminase, suggesting an independent effect of adenosine (Heftberger et al. 2005). Although adenosine is a full agonist of all four human adenosine receptors, inosine may activate A1 but is apparently ineffective on the two A2R (Fredholm et al. 2001a). This may explain the observation that inosine was less effective than adenosine in this study at least in PC12 cells. On the other hand, Jin et al. (1997) showed that inosine is able to bind and activate adenosine A3R, and Hasko et al. (2000) found that A1R- and A2R- antagonists partially blocked the suppressive effect of inosine on proinflammatory cytokine production.
In addition to adenosine receptors, neurons have nucleoside transport systems that play an important role in regulating the concentrations and effects of purine nucleosides (Benowitz et al. 1998; Heftberger et al. 2005; and for a review, see: Rathbone et al. 1999). Along these lines other authors reported that exogenously applied inosine acted directly on an intracellular target, which may coincide with a serine-threonine kinase, protein kinase N (Greene and Tischler 1976; Volonté et al. 1989; Batistatou et al. 1992), and is linked to the response elements of genes associated with axon growth, including GAP-43, L1, and alpha-1 tubulin (Benowitz et al. 1998; Petrausch et al. 2000).
There is a strong link between adenosine and hypoxia-related signaling. The expression levels of adenosine and adenosine receptors are regulated in conditions of cellular stress, and signal transduction increases via one or more of the adenosine receptors. Hypoxia apparently induces a program that shifts the tissue phenotype toward an increase in extracellular adenosine. In turn, adenosine receptor activation tends to limit the potential damage incurred by hypoxia (for review, see Fredholm 2007).
Data regarding the role of the A2AR are controversial. Kobayashi and Millhorn (1999) reported that the increased expression of A2AR during hypoxia might have protected cells against hypoxia. However, results from other studies indicated that there were protective effects of the A2AR antagonist SCH 58261 during ischemia; these effects involved the inhibition of phospho-p38 MAPK and suggested that treatment with the A2AR-antagonist during the first hour to several hours after ischemia may be a useful therapeutic approach in cerebral ischemia (Melani et al. 2006). Along these lines others (Trincavelli et al. 2007) suggested that an increase of A2ARs on neurons and microglial cells may account for the efficacy of the A2AR-antagonists in protecting against the cognitive deficits and neurodegenerative processes that occur after ischemia.
Apart from adenosine receptor-mediated signaling, a number of metabolic pathways have been identified that regulate gene expression during hypoxia (Seta et al. 2002; Bickler and Donohoe 2002). Among these, the activation of HIF-1 transcription factors is the best characterized. HIF-1 is considered one of the master regulators that orchestrate physiological responses to acute and chronic hypoxia (Semenza 2000; Baranova et al. 2007). Therefore, we studied HIF-1α modulation in hypoxia in PC12 cells as well as in cerebellar granule cells. As a consequence of cellular exposure to reduced oxygen, we observed an increased stability of HIF-1α protein in cellular extracts of PC12 cells as well as in primary cerebellar granule neurons. In addition, our results suggested that hypoxia led to an increased nuclear accumulation of HIF-1α as result of possible increased stabilization or synthesis and an increased transcriptional activity. Our data are consistent with results reported by others (Ruscher et al. 1998; Masuda et al. 1994; Krieg et al. 1998) that showed hypoxia stimulated an increase in HIF-1 binding activity in purified neurons and altered the regulation of HIF-1 transcriptional targets in neuronal cell lines and glial cultures. We also tested the potential effect of purine nucleosides. Our results showed that hypoxia-induced HIF-1α was significantly enhanced by purine nucleosides. Our data are consistent with previous findings (De Ponti et al. 2007) that demonstrated that the activation of the A2AR by adenosine treatment induced HIF-1 DNA-binding activity, nuclear accumulation, and transactivation capacity in J774A.1 mouse macrophages. To further study the non-redundant role of HIF-1α in the rescue of hypoxic PC12 cells and cerebellar granule neurons we employed a siRNA approach to specifically knockdown HIF-1α expression. Following partial siRNA-mediated knockdown of the HIF-1α transcription factor, we observed a significant increase in the hypoxia-induced cell death of PC12 cells and cerebellar granule neurons. Compared to cultured PC12 cells, primary cerebellar granule neurons showed a higher spontaneous loss of viability. Consequently, we observed that hypoxic insult led only to a further 1.5 fold increase in neuronal cell death compared to the 3.3 fold increase in PC12 cell death. This fact may be one of the reasons for the observation that the HIF siRNA knockdown had less striking effects on the cell viability of primary neurons compared to PC12 cells. We found even more exciting the observation that purine nucleoside-mediated rescue was completely abrogated upon siRNA-mediated knockdown of HIF-1α.
The future challenge will be to link molecular/genetic events with physiological mechanisms. Various authors have already published work along these lines. One group (De Ponti et al. 2007), reported that HIF-1α activation induced by the A2A receptor-specific agonist CGS21680 required the PI-3K and protein kinase C pathways, but was not mediated by changes in iron levels. Another study (Sodhi et al. 2001) provided evidence that MAPK and PI3K-Akt pathways may interact synergistically in the activation of HIF-1α. A large literature exists on the role of the HIF-1 transcription factor in the expression of genes, including erythropoietin, that enhance oxygen delivery to tissues (Semenza 1999). Hypoxia also increases the expression of genes whose products facilitate alterations in metabolism that optimize cell function during hypoxia (for review, see Bickler and Donohoe 2002). For example, hypoxic preconditioning, a treatment known to protect the newborn rat brain against hypoxic—ischemic injury, markedly increased HIF-1α (Bergeron et al. 1999). The unique feature of HIF-1α is the regulation of its concentration. During normoxia, HIF-1α is rapidly degraded by the ubiquitin proteasome system, and exposure to hypoxic conditions prevents its degradation (Semenza 2000). The amino-terminal half of each subunit contains basic helix-loop- helix (bHLH) and PAS motifs that are required for dimerization and DNA binding. The carboxyl-terminal half of HIF-1α contains domains that mediate hypoxia-inducible nuclear localization, protein stabilization, and transactivation (for review, see Semenza 2000). Among the potential mechanisms that might regulate transactivation, phosphorylation seems to play an important role. Indeed, several studies showed that phosphorylation via the mitogen-activated protein kinases (MAPK) is necessary for activation of HIF-1α transcriptional activity but not for its stabilization in hypoxic conditions (Salceda and Caro 1997; Minet et al. 2000; Hur et al. 2001). Other studies (Conrad et al. 1999), suggested that hypoxia caused specific regulation of the SAPK and p38 MAPK or the p42/44 MAPK signaling pathways, depending on the cell type. Despite these differences, HIF-1 is activated by hypoxia in all cell types (reviewed by Mottet et al. 2003).
In conclusion, the data presented here showed that 1) adenosine and inosine, which have increasingly been recognized as powerful endogenous neuroprotectants, efficiently rescued hypoxic PC12 cells and cerebellar granule neurons, and 2) HIF-1α plays a non-redundant role as a key regulator in the purine nucleoside-mediated rescue of hypoxic neuronal cells.
This work was supported by a grant of the Austrian FWF P19578-B05. We are grateful to Dr. G. Baier, Dr. C. Bandtlow and M.Sc. V. Podhraski for helpful discussion and support.