The up-regulation of BACE1 mediated by hypoxia and ischemic injury: role of oxidative stress and HIF1α

Authors


Address correspondence and reprint requests to Tamagno Elena, Department Experimental Medicine and Oncology, University of Turin, Corso Raffaello 30, 10125, Turin, Italy. E-mail: elena.tamagno@unito.it; Tabaton Massimo Department of Neurosciences, Via de Toni 2, 16132, Genova, Italy. E-mail: mtabaton@neurologia.unige.it

Abstract

While it is well established that stroke and cerebral hypoperfusion are both significant risk factors for Alzheimer’s disease, the molecular link between ischemia and amyloid precursor protein processing has only been recently established. Specifically, hypoxia significantly increases β-site APP cleaving enzyme (BACE1) gene transcription through the over-expression of hypoxia inducible factor 1α, resulting in increased BACE1 secretase activity and amyloid-β production. In this study, we significantly extend these findings both in vitro, in differentiated SK-N-BE neuroblastoma cells, and in vivo, in rats subjected to cerebral ischemia, showing that hypoxia up-regulates BACE1 expression through a biphasic mechanism. The early post-hypoxic up-regulation of BACE1 depends on the production of reactive oxygen species mediated by the sudden interruption of the mitochondrial electron transport chain, while the later expression of BACE1 is caused by hypoxia inducible factor 1α activation. The involvement of reactive oxygen species released by mitochondria in the BACE1 up-regulation was confirmed by the complete protection exerted by complex I inhibitors such as rotenone and diphenyl-phenylen iodonium. Moreover, the oxidative stress-mediated up-regulation of BACE1 is mediated by c-jun N terminal kinase pathway as demonstrated by the protection exerted by the silencing of c-jun N-terminal kinase isoforms 1 and 2. Our study strengthens the hypothesis that oxidative stress is a basic common mechanism of amyloid-β accumulation.

Abbreviations used
AD

Alzheimer’s disease

APP

amyloid precursor protein

amyloid-β

BACE1

β-site APP cleaving enzyme

CoCl2

cobalt chloride

DPI

diphenyl-phenylen iodonium

ERK

extracellular signal mitogen activated protein kinase

HIF1α

hypoxia inducible factor 1α

JNK

c-jun N-terminal kinase

LDH

lactate dehydrogenase

MTT

3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrasodium bromide

ROS

reactive oxygen species

Rot

Rotenone

Alzheimer’s disease (AD) is a progressive neurodegenerative disorder affecting the elderly population (Lage 2006). The disease is pathologically characterized by the formation in the brain of intracellular neurofibrillary tangles and extracellular amyloid deposits (Braak et al. 1993; Jellinger 1998). Mechanistically, a major role in the disease is attributed to the altered proteolytic processing of the amyloid-β (Aβ) protein precursor (APP) that results in the accumulation and aggregation of neurotoxic forms of Aβ (Selkoe 2001; Hardy 2006).

Amyloid-β derives from the sequential proteolytic cleavage of the β- and the γ-secretases on type I integral membrane APP. β-site APP cleaving enzyme (BACE1), identified as the β-secretase that cleaves APP within the ectodomain (Vassar et al. 1999), is an integral type I transmembrane protein detected in the trans-Golgi network and endosomes (Benjannet et al. 2001; Pasternak et al. 2004). The role of BACE1 in Aβ peptide generation has been convincingly supported by the lack of Aβ peptides in mice carrying a homozygous deletion of BACE1 (Roberds et al. 2001).

While the etiology of AD is incompletely understood, it is clear that stroke and hypoperfusion are important risk factors for the disease (de la Torre 2008). Notably, hypoxia significantly increases BACE1 gene transcription through the over-expression of hypoxia inducible factor 1α (HIF1α), resulting in increased β-secretase activity and Aβ production, (Sun et al. 2006; Zhang et al. 2007). As the mitochondrial electron chain acts as an oxygen sensor, releasing reactive oxygen species (ROS) in response to hypoxia, ROS released during hypoxia act as signaling agents that trigger functional responses, ranging from transcription factor activation to the control of the vascular tone in the pulmonary circulation (Chandel et al. 1998).

The author and others have shown that the expression and activity of BACE1 is increased by oxidant agents and by the lipid peroxidation product 4-hydroxynonenal (Tamagno et al. 2002, 2005; Kao et al. 2004), and that there is a significant correlation between BACE1 activity and oxidative markers in sporadic AD brain tissue (Borghi et al. 2007).

Here, we provide evidence that hypoxia acts through a biphasic mechanism involving an early up-regulation of BACE1 dependent on release of mitochondrial ROS and a late up-regulation of BACE1 mediated by HIF1α activation.

Materials and methods

Cell culture differentiation and treatments

SK-N-BE neuroblastoma cells were maintained in RPMI (Roswell Park Memorial Institute) 1640 medium containing 2 mmol/L glutamine and supplemented with 100 mL/L fetal bovine serum, 10 mL/L non-essential amino acids, and 10 mL/L antibiotic mixture (penicillin–streptomycin–amphotericin), in a humidified atmosphere at 37°C with 5% CO2. For differentiation, 2 x 106 cells were plated in 75 cm2 culture flasks (Costar, Lowell, MA, USA) and exposed to 10 μmol/L retinoic acid for 10 days. Serum-deprived SK-N-BE neuroblastoma cells were incubated in strictly controlled hypoxic conditions (3% O2) for up to 72 h. Rotenone (Rot) was used at 2.5 μM 1 h before hypoxia and diphenyl-phenylen iodonium (DPI) at 1 μM at the time of hypoxia. Cobalt chloride (CoCl2), used as positive control for HIF1α induction, was used at 100 μM for 20 h (Yao et al. 2008).

Animals and treatments

Male Wistar rats (Harlan-Italy, Udine, Italy) weighing 80–100 g were housed in a controlled environment at 25 ± 2°C with alternating 12 h light and dark cycles. They were provided with Piccioni pellet diet (No. 48, Gessate Milanese; Italy) and water ad libitum. All rats were acclimatized in our animal facility for at least 1 week before experiments. Stressful stimuli were avoided. Animal care was in compliance with Italian regulations on the protection of animals used for experimental and other scientific purposes (D.M. 116/92) as well as with the Guide for the Care and Use of Laboratory Animals, as adopted and promulgated by the U.S. National Institutes of Health. The experimental protocol, approved by the Turin University Ethics Committee, was performed as described elsewhere (Collino et al. 2006). Briefly, rats were anesthetized through intraperitoneal injection (30 mg/kg) of Zoletil 100 (mixture of Tiletamine and Zolazepam; Laboratoires Virbac, France), supplemented as needed. Both common carotid arteries were exposed over a midline incision and a dissection was made between the sternocleidomastoid and the sternohyoid muscles, parallel to the trachea. Each carotid artery was freed from its adventitial sheath and vagus nerve, which were carefully separated and maintained. Ischemia was achieved by clamping the bilateral common carotid arteries for 1 h using non-traumatic artery clamps (Micro Bulldog Clamps; Harvard Apparatus Ltd, Kent, UK). During ischemia the animals were monitored for body temperature, respiration pattern, loss of righting reflex and unresponsiveness, corneal reflexes, and fixed and dilated pupils. At the end of the ischemia, anesthetized rats were killed by decapitation. Sham-operated rats underwent identical surgical procedures except that no artery clamps were applied. After decapitation, the forebrain was rapidly dissected at 0°C and the brain and cortices from both hemispheres were rapidly removed and transferred to an appropriate ice-chilled homogenizing medium for biochemical assays.

PEG-SOD (Sigma Chemical Company, Beverly, MA, USA) and PEG-CAT (Sigma Chemical Company) as well as α-tocopherol (Sigma Chemical Company) were used. PEG-SOD/CAT (1000 U/kg/10 000 U/kg, respectively) was administered for 4 days ‘per os’, and α-tocopherol was administered intraperitoneally at the dose of 50 mg/kg for 5 days.

Tissue and cell extracts

Cytosolic and nuclear extracts of rat brain and cortex tissues were prepared as described previously (Meldrum et al. 1997). Briefly, cerebral tissues were homogenized at 10% (w/v) in a Potter Elvehjem homogenizer (Wheaton, Millville, NJ, USA) using a homogenization buffer containing 20 mmol/L HEPES, pH 7.9, 1 mmol/L MgCl2, 0.5 mmol/L EDTA, 1% Nonylphenyl-polyethylene glycol, 1 mmol/L EGTA, 1 mmol/L dithiothreitol, 0.5 mmol/L phenylmethylsulfonyl fluoride, 5 mg/mL aprotinin, and 2.5 mg/mL leupeptin. Homogenates were centrifuged at 1000 g for 5 min at 4°C.

Supernatants were removed and centrifuged at 15 000 g at 4°C for 40 min to obtain the cytosolic fraction, The pelleted nuclei were resuspended in extraction buffer containing 20 mmol/L HEPES, pH 7.9, 1.5 mmol/L MgCl2, 300 mmol/L NaCl, 0.2 mmol/L EDTA, 20% glycerol, 1 mmol/L EGTA, 1 mmol/L dithiothreitol, 0.5 mmol/L phenylmethylsulfonyl fluoride, 5 mg/mL aprotinin, and 2.5 mg/mL leupeptin. The suspensions were incubated on ice for 30 min for high-salt extraction followed by centrifugation at 15 000 g for 20 min at 4°C. The resulting supernatants containing nuclear proteins were carefully removed and protein content was measured using a commercially available assay (Bio-Rad, Segrate, Italy). Preparation of cell lysates and nuclear extracts were performed as described previously (Tamagno et al. 2002, 2005).

Antibodies and immunoblot analysis

The following antibodies were used: polyclonal anti-BACE1 antibody (Chemicon, Temecula, CA, USA); monoclonal β actin antibody (Sigma Chemical Company); polyclonal HIF1α, monoclonal HIF1β, polyclonal lamin A (Santa Cruz Biotechnology, Santa Cruz, CA, USA), polyclonal pJNK and polyclonal c-jun N-terminal kinase (JNK) antibodies (Cell Signalling Technology, Beverly, MA, USA). Lysates, nuclear and membrane fraction extracts were subjected to sodium dodecyl sulfate–polyacrylamide gel electrophoresis on 9.3% (BACE1, β-actin, JNK and lamin A) or 7.5% (HIF1α) acrylamide gels using the mini-PROTEAN II electrophoresis cell (Bio-Rad). Proteins were transferred onto nitrocellulose membranes (Hybond-C extra Amersham Life Science, Arlington Heights, IL, USA). Non-specific binding was blocked with 50 g/L non-fat dry milk in 50 mmol/L Tris–HCl, pH 7.4, containing 200 mmol/L NaCl and 0.5 mmol/L Tween-20 (Tris-buffered saline Tween). The blots were incubated with different primary antibodies, followed by incubation with peroxidase-conjugated anti-mouse or anti-rabbit immunoglobulins in Tris-buffered saline Tween containing 20 g/L non-fat dry milk. Reactions were developed with an enhanced chemiluminescence system according to the manufacturer’s protocol (Amersham-Pharmacia Biotech Italia, Cologno Monzese, Italy).

Oxidative stress determination

Differentiated neuroblastoma cells were exposed to hypoxia for 15, 30, 60, or 180 min. Intracellular generation of ROS has been detected by using the conversion of 2′,7′-dichlorodihydrofluorescein diacetate (used at 5 μM concentration), once taken up by cells and de-acetylated by esterase, into the corresponding fluorescent derivative (Rezvani et al. 2007). Cells were observed and photographed under a Zeiss fluorescence microscope. The amount of ROS was expressed as percentage of fluorescent cells. For in vivo oxidative stress determinations, ROS were detected in cytosolic fraction derived by total brain and cortex.

Analysis of gene expression

For the quantitative SYBR Green (2x iQ YBR Green PCR Super Mix; Bio-Rad Laboratories) real-time PCR, 40 ng of cDNA was used per reaction. Primer sequences, designed with PRIMER 3 software (Applied Biosystems, Monza, Italy) were:

Human BACE1: 5′-CATTGGAGGTATCGACCACTCGCT-3′ and 5′-CCACAGTCTTCCATGTCCAAGGTG-3′.

Human β actin: 5′-GGCACTCTTCCAGCCTTCCTTC-3′ and 5′-GCGGATGTCCACGTCACACTTCA-3′.

Murine BACE1: 5′-GCATGATCATTGGTGGTATC-3′ and 5′-CCATCTTGAGATCTTGACCA-3′.

Murine βactin: 5′-AGCTATGAGCTGCCTGACGGC-3′ and 5′-CATGGATGCCACAGGATTCCA-3′.

Quantitative PCR was performed on a real-time iCycler sequence detector instrument (Bio-Rad Laboratories). After 3 min of initial denaturation, the amplification profile included 30 s denaturation at 95°C and extension at 72°C. Primer annealing was carried out for 30 s at 60°C. The results were obtained with the comparative Ct method using the arithmetic formula 2−ΔΔCt. Samples obtained from a least three independent experiments were used to calculate the mean and SD.

For the semi-quantitative reverse transcription PCR primer sequences, designed with PRIMER 3 software were:

Human HIF1α: 5′-CGGCGCGAACGACAAGAAAAAGAT-3′ and 5′-TCGTTGGGTGAGGGGAGCATTACA-3′ (Chu et al. 2008).

Human G3PDH: 5′-GTCGGAGTCAACGGATTTGG-3′ and 5′-GGGTGGAATCATATTGGAACATG-3′.

PCR was performed in a Gene Amp PCR System 9600 (Perkin-Elmer, Wellesley, MA, USA), with 1 μL of cDNA reaction mixture in a volume of 50 μL containing 200 μM dATP, dGTP, dCTP, and dTTP; 1 μM 5′-and 3′-primer; and 1.25 units of Taq DNA polymerase (Polymed, Florence, Italy). The amplification program consisted of a denaturing step at 94°C for 40 s, an annealing step at 58°C for 40 s, and an extension step at 72°C for 40 s, for 35 cycles for HIF1α or for 25 cycles for G3PDH. A 10 μL sample of the PCR mixture was separated on a 1% agarose gel and amplification products were stained with GelStar nucleic acid gel stain (FMC BioProducts, Rockland, ME, USA). Densitometric analysis was performed using a software program (Multi-analyst, version 1.1; Bio-Rad).

RNA interference

To knockdown HIF1α expression in differentiated SK-N-BE neuroblastoma cells we used siRNA duplex (Qiagen Italia, Milano, Italy). The following target sequence was used: 5′-AGGAAGAACTATGAACATAAA-3′. The siRNA and related non-silencing control (negative control) were transfected in SK-N-BE cells with lipofectamine transfection reagent (Invitrogen srl, Milano, Italy) according to the manufacturer’s instructions for 48 h.

RNA interference experiments to knockdown the JNK1/2 in SK-N-BE neuroblastoma cells were performed using a RNAi human/mouse control kit, including the small RNA duplex (Qiagen Italia). The siRNAs (2.5 μg) were transfected into cells using the properRNAiFect reagent ratio (1 : 6) in a RPMI (Roswell Park Memorial Institute) medium for 48 h. Transfected cells in fresh medium were then exposed for 1, 6, 24, or 48 h to hypoxia and then harvested for sample preparation.

Analysis of cell death

In preliminary experiments, designed to evaluate the rate of cell death, differentiated neuroblastoma cells were exposed to hypoxic conditions up to 72 h and cell death evaluated by analyzing release of lactate dehydrogenase (LDH) in culture medium as well as cleavage of 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrasodium bromide (MTT). Enzymatic analysis of LDH activity released by cells in culture medium was performed as previously described (Tamagno et al. 2003). Values for control and treated cells were expressed as a percentage value of the total LDH activity released by untreated cells after exposure to Triton X-100.

3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrasodium bromide is a pale yellow substrate that is cleaved by living cells to yield a dark-blue formazan product. The MTT test was determined using a commercially available kit from Chemicon International according to the manufacturer’s protocol.

Statistical analysis

Data in bar graphs represent mean ± SEM, and were obtained from average data of at least three independent experiments. Luminograms and morphological images are representative of at least three experiments with similar results. Statistical analysis was performed by Student’s t-test or anova when appropriate (p < 0.05 was considered significant).

Results

Hypoxia up-regulates BACE1 expression and determines intracellular production of ROS

To test the effect of hypoxia on BACE1, we determined mRNA expression and protein levels of BACE1 in differentiated SK-N-BE neuroblastoma cells exposed to hypoxia up to 72 h. Hypoxia led to a marked biphasic increase in BACE1 mRNA with peaks after 1 h (2.3-fold increase) and 24 h (2.6-fold increase) of hypoxia (Fig. 1a). The same biphasic result was observed in analyzing BACE1 protein levels with peaks at 6 h (2.5-fold increase) and 48 h (threefold increase) of hypoxia (Fig. 1b). We predicted that the early induction of BACE1 could be mediated by production of ROS, as we have previously observed that treatment with hydrogen peroxide, and the consequent overproduction of ROS, induced an increase of BACE1 expression after 1 h of treatment and of BACE1 protein levels after 6 h (Tamagno et al. 2008a,b). On the other hand, the late augmentation of BACE1 is consistent with the activation of HIF1α, a transcription factor that responds to cellular oxygen reduction (Sharp and Bernaudin 2004) that has also been shown able to increase BACE1 transcription (Zhang et al. 2007).

Figure 1.

 Hypoxia increases BACE1 expression and protein levels in SK-N-BE cells. (a) Hypoxia leads to a marked biphasic increase in BACE1 mRNA, as revealed by real time PCR, with peaks after 1 and 24 h of hypoxia; (b) BACE1 protein levels, as revealed by the immunoblot and by the densitometric analysis, are significantly increased, with two peaks after 6 and 48 h of hypoxia. Experiments were performed in triplicate; *p < 0.05 versus control; **p < 0.02 versus control.

To confirm the intracellular generation of ROS under hypoxic conditions we used a morphological semi-quantitative technique based on the use of 2′,7′-dichlorodihydrofluorescein. The levels of intracellular fluorescence, because of the interaction of intracellularly generated ROS with the deacetylated dye, were significantly increased in differentiated SK-N-BE neuroblastoma cells when exposed to hypoxic conditions, reaching the highest level of intracellular fluorescence after 1 h of hypoxia (Fig. 2a and b). Of note, we tested the cell viability in SK-N-BE differentiated cells exposed to hypoxia up to six 3 h in terms of LDH release and MTT cleavage. Cells did not show any change in viability respect to control cells not exposed to hypoxia (data not shown).

Figure 2.

 Hypoxic conditions induces intracellular generation of ROS in SK-N-BE cells. (a) and (b) detection of intracellular generation of ROS was performed using the 2′,7′-dichlorodihydrofluorescein (DCFH-DA) probe in cells exposed to hypoxia for 15 and 30 min; 1 and 3 h. The peak of stained cells was observed after 30 min and 1 h of hypoxia. Cells was observed and photographed under a Zeiss fluorescence microscope equipped with phase contrast objectives. Immunofluorescence and phase contrast images of the same fields were collected and electronically merged (see images in the right column) for all conditions. The amount of ROS was expressed as percentage of fluorescent cells. Experiments were performed in triplicate; *p < 0.05 versus control; **p < 0.02 versus control.

Given that the hypoxia-dependent ROS production is coincident with BACE1 over-expression, we investigated the direct relationship of the two events. SK-N-BE differentiated neuroblastoma cells were pre-treated with Rot or DPI, compounds that inhibit complex I - dependent mitochondrial generation of superoxide anion. As shown in Fig. 3(a), hypoxia-dependent generation of ROS was completely blocked by pre-treatment of differentiated neuroblastoma cells with Rot or DPI. As expected, pre-treatment with Rot or DPI also blocked the up-regulation of BACE mRNA. Specifically, 1 h hypoxia was followed by a twofold increase of BACE1 mRNA that is fully prevented by pre-treatment with Rot or DPI (Fig. 3b). The course of BACE1 protein levels paralleled the results obtained by BACE1 mRNA expression (Fig. 3c). Next, we examined the behavior of HIF1α during the post-hypoxia phase. As shown in Fig. 4(a), in differentiated SK-N-BE neuroblastoma cells exposed to hypoxia up to 3 h we observed only a slight, not significant increase in nuclear translocation of HIF1α. Treatment of differentiated SK-N-BE cells with CoCl2, used as positive control, significantly increased nuclear traslocation of HIF1α (Fig. 4a). To further confirm that HIF1α is independent of the early up-regulation of BACE1 we silenced HIF1α with RNA interference. Differentiated neuroblastoma cells were silenced for HIF1α for 48 h and then exposed to 1 or 6 h hypoxia with or without pre-treatment with Rot. The silencing of the gene was evaluated with semi-quantitative PCR and, as shown in Fig. 4(b), HIF1α transcription has been decreased by about 60%. As shown in Fig. 4(c) 1 h of hypoxia caused a 2.4-fold increase of BACE1 expression both in HIF1α-silenced and not silenced cells whereas the BACE1 over-expression was almost completely prevented by pre-treatment with Rot. BACE1 protein levels paralleled results obtained by real-time PCR (Fig. 4d).

Figure 3.

 Inhibitors of complex I-dependent mitochondrial generation of ROS blocks BACE1 up-regulation in SK-N-BE. (a) Detection of intracellular generation of ROS was performed by using the 2′,7′-dichlorodihydrofluorescein (DCFH-DA) probe in cells pre-treated for 30 min with Rotenone or DPI and exposed to hypoxia for 1 h. Pre-treatment with Rot or DPI completely blocks the intracellular ROS production mediated by hypoxic condition; (b) pre-treatment with Rot and DPI completely inhibits the over-expression of BACE1 induced by 1 h of hypoxia, as revealed by real-time PCR as well as (c) the increase of BACE1 protein levels observed after 6 h of hypoxia. Experiments were performed in triplicate; **p < 0.02 versus control.

Figure 4.

 Hypoxia inducible factor 1α (HIF1α) is not involved in the early up-regulation of BACE1. (a) Detection of nuclear HIF1α levels by western blot analysis in differentiated SK-N-BE cells exposed up to 6 h of hypoxia. Change in nuclear traslocation of HIF1α was not detected. Treatment with cobalt chloride (CoCl2), used as positive control, significantly increase nuclear traslocation of HIF1α; (b) 48 h silencing of HIF1α, evaluated with semi-quantitative PCR in SK-N-BE exposed to hypoxia for 1 h. HIF1α transcription was decreased of about 60%; (c) BACE1 expression evaluated by real-time PCR in SK-N-BE cells silenced or not for HIF1α for 48 h. Cells were then pre-treated or not with Rot and exposed to hypoxia for 1 h. Hypoxia caused a significant increase in BACE1 mRNA both in silenced and not silenced cells whereas Rot pre-treatment almost completely prevent this increase; (d) western blot analysis of BACE1 protein levels in SK-N-BE cells silenced or not for HIF1α for 48 h. Cells were then pre-treated or not with Rot and exposed to hypoxia for 6 h. Results completely paralleled those obtained by real-time PCR. Experiments were performed in triplicate; **p < 0.02 versus control.

ROS production regulates the expression of BACE1 in ischemic rats

To confirm the results obtained in vitro we used a well-established animal model of ischemia, obtained by clamping common carotid arteries for 1 h (Collino et al. 2006). Mice that had undergone ischemic injury, exhibited a 30% and 40% increase in ROS in the total brain and cortex, the main brain area affected by AD, respectively, as shown in Fig. 5(a and b). The increased production of ROS was accompanied by 2.4 and 1.8 increase of BACE1 mRNA (Fig. 5c and d). The causative role of oxidative stress in the up-regulation of BACE1 was confirmed by the pre-treatments with the antioxidant compounds SOD/Cat and α-tocopherol, that prevented BACE1 over-expression (Fig. 5a–d). As observed in neuroblastoma cells, after 1 h of ischemia, an activation of HIF1α was not shown, ruling out its involvement in the early up-regulation of BACE1 (Fig. 5e and f).

Figure 5.

 Reactive oxygen species (ROS) production regulates the expression of BACE1 in ischemic rats. (a) and (b) detection of ROS in brain (a) and cortex (b) lysates of rats subjected to cerebral ischemia for 1 h pre-treated or not with SOD/Cat or α-tocopherol. Rats that had undergone ischemic injury exhibited a significant increase in ROS production whereas antioxidants completely prevented ROS production; (c) and (d) BACE1 expression evaluated by real time PCR in brain (c) and cortex (d) of ischemic mice pre-treated or not with SOD/Cat or α-tocopherol. Pre-treatment with antioxidants prevented BACE1 over-expression mediated by ischemia; (e) and (f) detection of nuclear HIF1α levels by western blot analysis in brain (e) and cortex (f) lysates of rats subjected to ischemic injury for 1 h pre-treated or not with antioxidant compounds such as SOD/Cat or α-tocopherol. We did not observed any change in nuclear traslocation of HIF1α. Experiments were performed in triplicate; *p < 0.05 versus control; **p < 0.02 versus control.

HIF1α is responsible for the late induction of BACE1

We investigated the mechanism of the late up-regulation of BACE1 by focusing on HIF1α (Zhang et al. 2007). First, we tested the cell viability in SK-N-BE differentiated cells exposed to hypoxia for 12–72 h in terms of LDH release (Fig. 6a) and MTT cleavage (Fig. 6b). Cells showed only a slight decrease of viability after 48 h of hypoxia (15% LDH release and 40% MTT) (Fig. 6a and b).

Figure 6.

 Cell death mediated by hypoxic conditions. (a) LDH release evaluated in culture medium of differentiated SK-N-BE neuroblastoma cells exposed to hypoxia up to 72 h. Cells showed a slight increase in LDH release only after 48 h of hypoxia; (b) detection of MTT cleavage in neuroblastoma cells exposed to hypoxia up to 72 h. Cells showed only a slight decrease of viability after 48 h of hypoxia. Experiments were performed in triplicate; *p < 0.05 versus control; **p < 0.02 versus control.

We then evaluated the nuclear level of HIF1α in differentiated SK-N-BE neuroblastoma cells exposed to hypoxia up to 48 h. In contrast to the findings with hypoxia up to 6 h, a nuclear translocation of HIF1α became significant after 24 h and persisted up to 48 h (Fig. 7a). As expected, treatment with CoCl2, significantly increased nuclear traslocation of HIF1α (Fig. 7a).

Figure 7.

 Hypoxia inducible factor 1α was responsible of the late induction of BACE1. (a) Detection of nuclear HIF1α levels by western blot analysis in differentiated SK-N-BE cells exposed up to 48 h to hypoxia. The nuclear translocation of HIF1α became significant after 24 h of hypoxia; treatment with cobalt chloride (CoCl2), used as positive control, significantly increase nuclear traslocation of HIF1α. (b) 48 h silencing of HIF1α, evaluated with semi-quantitative PCR in cells exposed to hypoxia for 48 h. Approximately 60% silencing was obtained; (c) BACE1 expression evaluated by real-time PCR in SK-N-BE cells silenced or not for HIF1α for 48 h. Cells were then pre-treated or not with Rot and exposed to hypoxia for 24 h. Silencing of HIF1α completely prevents the up-regulation of BACE1, both in presence or absence of Rot. On the other hand, Rot is not able to prevent the increase of BACE1 mRNA in non-silenced cells; (d) western blot analysis of BACE1 protein levels in cells silenced for HIF1α for 48 h. Cells were then pre-treated or not with Rot and exposed to hypoxia for other 48 h. Results completely paralleled those obtained by real-time PCR technique. Experiments were performed in triplicate; *p < 0.05 versus control; **p < 0.02 versus control.

To investigate the role of the translocation of HIF1α on BACE1 up-regulation in the late effect of hypoxia, we silenced HIF1α gene in neuroblastoma cells exposed to 24 or 48 h to hypoxia with or without pre-treatment with Rot. As previous, approximately 60% silencing was obtained (Fig. 6b).

Figure 7(c) reports BACE1 mRNA levels after 24 h of hypoxia in cells silenced or not for HIF1α. As shown, 24 h of hypoxia was followed by a 2.4-fold increase of BACE1 expression while pre-treatment with Rot was not able to prevent this increase in non-silenced cells. However, on the other hand, silencing of HIF1α completely prevented the up-regulation of BACE1, both in presence or absence of Rot. BACE1 protein levels paralleled results obtained by real-time PCR. After 48 h of hypoxia, a significant (threefold) increase in BACE1 protein levels was observed (Fig. 7d); pre-treatment with Rot was not able to protect this induction in non-silenced cells, and HIF1α silencing completely prevents the hypoxia mediated up-regulation of BACE1 protein.

JNK activation mediates the up-regulation of BACE1 in the early post-hypoxia phase

The signalling pathways involved in the early hypoxia dependent up-regulation of BACE1 were next investigated. The JNK signalling pathway is a potential candidate, as it was shown to transmit the over-expression of BACE1 induced by oxidative stress (Tamagno et al. 2005, 2008a,b). First, we tested if the JNK pathway was activated in the early phase of hypoxic conditions. As shown in Fig. 8(a) a robust increase of nuclear JNK phosphorylation was observed up to 6 h of hypoxia in SK-N-BE differentiated cells while in the late phase (after 12 h) this activation disappeared. The role of JNK pathway in the up-regulation of BACE was confirmed by the complete protection exerted by the silencing of JNK 1 and 2 (Fig. 8b). This finding confirm a role of JNK pathway in the early up-regulation of BACE1 mediated by ROS but not in the late, HIF1α-mediated, induction. As expected a strong increase in the JNK phosphorylation was also observed after 1 h of ischemia both in nuclear fractions derived from total brain (Fig. 8c) as well as by cortex (Fig. 8d).

Figure 8.

 c-jun N-terminal kinase activation is responsible of BACE1 induction in the early hypoxia phase. (a) Detection of nuclear JNK phosphorylation levels by western blot analysis in differentiated SK-N-BE cells exposed up to 12 h of hypoxia. Change in JNK phosphorylation was observed up to 6 h while in the late phase (after 12 h) this activation disappeared. (b) Western blot analysis of BACE1 protein levels in cells silenced or not for JNK1/2 for 48 h. Cells were then exposed to hypoxia for 6 h. Silencing of JNK1/2 completely prevents the up-regulation of BACE1. (c) and (d) Detection of nuclear JNK phosphorylation levels by western blot analysis in nuclear fractions derived by total brain (c) and cortex (d) of rats subjected to 1 h of ischemic injury. A strong increase in the JNK phosphorylation was observed.

Discussion

A link between stroke, brain ischemia, and AD is hypoxia, the direct consequence of cerebral hypoperfusion. Indeed, hypoxia increase BACE1 expression and activity, resulting in Aβ overproduction, as has been shown in vitro as well as in AD transgenic mice (Sun et al. 2006). One mechanism of the effect of hypoxia on BACE1 up-regulation that has been proposed is the activation of HIF-1, a transcription factor that regulates oxygen homeostasis, and has been shown to bind to BACE1 promoter and regulate its gene expression (Zhang et al. 2007). Our study extends these findings, showing that hypoxia, both in vitro and in vivo, up-regulates BACE1 mRNA expression in a biphasic manner, through two distinct mechanisms: (i) an early release of ROS from mitochondria, and (ii) a late activation of HIF-1.

Mitochondria have long been known to generate ROS (Chandel et al. 2000; Guzy and Schumacker 2006, Bell et al. 2007; Archer et al. 2008) and play a central role in oxidative stress in AD (Hirai et al. 2001; Reddy and Beal 2005). As extensively described, a quasi-stable ubisemiquinone radical is repeatedly generated during the electron transport process. Molecular oxygen, a lipophilic molecule dissolved in the hydrophobic environment within the membrane, is highly electrophilic and can potentially capture the electron from ubisemiquinone. (Sun and Trumpower 2003). The capture of an electron by O2 yields superoxide, a reactive molecule that can potentially act in either a signaling role or as a non-specific-oxidizing agent capable of contributing to cell damage (Turrens 2003). Moreover, the superoxide generated as described would be likely to contribute to oxidant stress in the mitochondrial matrix compartment. Superoxide dismutase in the matrix or in the intermembrane space yields hydrogen peroxide, which has the capacity to cross membranes (Wallace 2001). Our study confirms this mechanism, as the production of ROS observed in the early phases of hypoxia is completely prevented by pre-treatment of cells with Rot or DPI, compounds that, under hypoxia, are both able to inhibit mitochondrial release of ROS by affecting complex I of mitochondrial electron transport chain (Li and Trush 1998). Our hypothesis is that the generation of ROS by mitochondria have a causative role in mediating early hypoxia-dependent up-regulation of BACE-1 transcription (1 h) and synthesis (6 h), that were indeed prevented by treatment with Rot and DPI. Instead, differently from the observation of Zhang et al. (2007) in our experimental conditions we did not detect an evident nuclear translocation of HIF-1 within 12 h of hypoxia. In fact, silencing of HIF-1 (able to block the late up-regulation of BACE-1) did not prevent neither the early increase in BACE-1 mRNA nor of BACE-1 protein levels.

In addition, the release of ROS, and the consequent up-regulation of BACE1 is paralleled by the activation of the JNK/c-jun pathway, which is quiescent in the late phase of post-hypoxic BACE1 increase, that depends on the activity of HIF-1, as shown unequivocally by data in HIF-1 silenced cells. The early post-hypoxic up-regulation of BACE1 recapitulates the cascade of events induced by oxidant agents and 4-hydroxynonenal in cells and in animal models: an increase of BACE1 mRNA, protein levels, and activity that is mediated by the activation of the JNK/c-jun pathway (Tamagno et al. 2005, 2008a,b). Notable, JNK and c-jun are activated in AD (Zhu et al. 2003; Thakur et al. 2007). The author and others (Jo et al. 2008; Tamagno et al. 2008a) have recently found that BACE1 expression is regulated by the γ-secretase activity, providing evidence of positive feedback loop between the γ- and the β-secretase cleavages on APP. In this connection it is significant that ischemic and hypoxic conditions produces the increase of the γ-secretase activity (Arumugam et al. 2006; Wang et al. 2006; Li et al. 2007). Moreover, the expression of BACE1 is decreased by the activation of extracellular signal regulated mitogen activated protein kinase (ERK)1/2, that inhibits the γ-secretase (Tamagno et al. 2008b).

β-site APP cleaving enzyme expression and enzymatic activity are elevated in the brains of sporadic AD patients and their increase is significantly correlated with oxidative markers (Fukumoto et al. 2002; Holsinger et al. 2002; Tyler et al. 2002; Yang et al. 2003; Li et al. 2004). Oxidative stress is a potential major cause of Aβ overproduction in late-onset, age-related AD. Our results in ischemic rats show that hypoxia because of acute hypoperfusion, as occurs in cerebral ischemia and stroke, produces the increase of BACE1 gene transcription by inducing oxidative stress. Of note, acute ischemia has been also shown to increase BACE1 activity through the impaired degradation of BACE1 mediated by the decrease of GGA3, a trafficking molecule that delivers BACE1 to the endosomal-lysosomal system (Tesco et al. 2007). These findings further highlight the complexity of the regulation of BACE1 activity. BACE 1 expression is controlled by several transcription factors and signaling pathways, that respond to a variety of cellular stress (for review Stockley and O’Neill 2007). Additional non-transcriptional mechanisms have been postulated to account for increased BACE1 protein levels and activity such as the expression of a BACE1 antisense transcript that stabilize BACE1 mRNA (Faghihi et al. 2008) and alternative splicing of BACE1 pre-mRNA (Mowrer and Wolfe 2008). The activity, the interaction with APP, the maturation, and the degradation of BACE1 are modulated by the binding with partners proteins in different cellular compartments (Murphy et al. 1999; Yu et al. 2004, 2005; Pickford et al. 2008).

Chronic hypoxia (lasting at least 24 h, as indicated by the experiments on neuroblastoma cells) triggers the second mechanism of BACE1 up-regulation, the activation of HIF-1. Oxidative stress results from several cellular insults, such as hyperglycemia, glutamate, prion proteins, calcium homeostasis perturbation (Mattson 2002; Ferreiro et al. 2006; Tahirovic et al. 2007; Schaeffer and Gattaz 2008), and our study strengthens the hypothesis that oxidative stress is a basic common pathway of Aβ accumulation, as effected by different AD risk factors (Zhu et al. 2004).

Acknowledgements

The study was supported by the Italian Ministry of Health and Regione Piemonte, Carisa de Mari and Telethon Foundations.

Ancillary