Biochimica et Biophysica Acta 1783 (2008) 994–1002

Contents lists available at ScienceDirect

Biochimica et Biophysica Acta

j ourna l homepage: www.e lsev ie r.com/ locate /bbamcr

brought to you by COREView metadata, citation and similar papers at core.ac.uk

provided by Elsevier - Publisher Connector

Insulin neuroprotection against oxidative stress is mediated by Akt and GSK-3βsignaling pathways and changes in protein expression

Ana I. Duarte a,b, Paulo Santos c, Catarina R. Oliveira a,d, Maria S. Santos a,b, A. Cristina Rego a,d,⁎a Center for Neuroscience and Cell Biology, University of Coimbra, Coimbra, Portugalb Department of Zoology, Faculty of Sciences and Technology, University of Coimbra, Coimbra, Portugalc Center for Histocompatibility of the Centre, Coimbra, Portugald Institute of Biochemistry, Faculty of Medicine, University of Coimbra, Coimbra, Portugal

a r t i c l e i n f o

⁎ Corresponding author. Institute of Biochemistry, FNeuroscience and Cell Biology, University of Coimbra, 30+351 239 820190; fax: +351 239 822776.

E-mail address: acrego@cnc.cj.uc.pt (A.C. Rego).

0167-4889/$ – see front matter © 2008 Elsevier B.V. Aldoi:10.1016/j.bbamcr.2008.02.016

a b s t r a c t

Article history:Received 12 September 2007Received in revised form 28 January 2008Accepted 11 February 2008Available online 29 February 2008

Previously we demonstrated that insulin protects against neuronal oxidative stress by restoring antioxidantsand energy metabolism. In this study, we analysed how insulin influences insulin- (IR) and insulin growthfactor-1 receptor (IGF-1R) intracellular signaling pathways after oxidative stress caused by ascorbate/Fe2+ inrat cortical neurons. Insulin prevented oxidative stress-induced decrease in tyrosine phosphorylation of IRand IGF-1R and Akt inactivation. Insulin also decreased the active form of glycogen synthase kinase-3β (GSK-3β) upon oxidation. Since phosphatidylinositol 3-kinase (PI-3K)/Akt-mediated inhibition of GSK-3β maystimulate protein synthesis and decrease apoptosis, we analysed mRNA and protein expression of “candidate”proteins involved in antioxidant defense, glucose metabolism and apoptosis. Insulin prevented oxidativestress-induced increase in glutathione peroxidase-1 and decrease in hexokinase-II expression, supportingprevious findings of changes in glutathione redox cycle and glycolysis. Moreover, insulin precluded Bcl-2decrease and caspase-3 increased expression. Concordantly, insulin abolished caspase-3 activity and DNAfragmentation caused by oxidative stress. Thus, insulin-mediated activation of IR/IGF-1R stimulates PI-3K/Aktand inhibits GSK-3β signaling pathways, modifying neuronal antioxidant defense-, glucose metabolism- andanti-apoptotic-associated protein synthesis. These and previous data implicate insulin as a promisingneuroprotective agent against oxidative stress associated with neurodegenerative diseases.

© 2008 Elsevier B.V. All rights reserved.

Keywords:Cortical neuronsERK1/2GSK-3βInsulinOxidative stressPI-3K/Akt

1. Introduction

In the central nervous system (CNS), insulin binding and activationof IR and/or IGF-1R [1] initiate a cascade of intracellular signalingpathways that regulate brain activity [2]. Peripherally synthesizedinsulin can be transported into the brain either via the blood–brainbarrier, or directly through the area postrema [3]. In addition, insulincan be synthesized de novo in the brain, as deduced from the ob-servation of preproinsulin I and II mRNA and insulin immunoreactivityin neurons [4,5].

IR and IGF-1Rexpressionhas beenalsodetected in theCNS [5]. Insulinbinds to the IR with higher affinity (b1 nM) than IGF-1 (100–500-foldlower affinity), whilst IGF-1 (b1 nM) preferentially binds IGF-1Rcompared to insulin (100–500-fold lower affinity) [6]. Binding of insulinto the α subunits of IR or IGF-1R results in Tyr autophosphorylation ofβ subunits, stimulating its Tyr kinase activity [7]. This then activatesphosphatidylinositol 3-kinase (PI-3K) and the Ser/Thr kinase Akt [8,9].Phosphorylated (active) Akt has a prosurvival role against neuronal

aculty of Medicine, Center for04-504 Coimbra, Portugal. Tel.:

l rights reserved.

apoptosis due to the phosphorylation and subsequent inactivation ofproapoptotic proteins, such as Bad, caspase-9 and FoxO1 [9]. Akt alsopromotes glycogen synthase kinase-3β (GSK-3β) phosphorylation atSer9, inhibiting its activity [10]. cAMP response-element binding (CREB)-induced expression of the anti-apoptotic protein Bcl-2 can also beactivated [11]. IR/IGF-1R-induced signaling may also activate Ras, Raf,MEK (MAPK, mitogen-activated protein kinase) and ERK1/2 (extracel-lular signal-regulated kinase), which control gene expression [9].

High content of polyunsaturated fatty acids, transition metals andoxygen and poor antioxidant defenses render neuronal cells highlyvulnerable to reactive oxygen species (ROS) [12]. This culminates inneuronal death [13], which may underlie the pathophysiology of neu-rodegenerative diseases, stroke, and diabetes complications affectingthe CNS [12,14]. Insulin was previously reported to protect againsthydrogen peroxide (H2O2)- and serum deprivation-induced neuronalapoptosis [15]. Thus, effects of oxidative stress on neuronal IR/IGF-1Rsignaling influence brain function, impairing glucose metabolism andleading to neuronal apoptosis [4]. Interestingly, brain insulin levelsappear to correlate with oxidative stress, type 2 diabetes mellitus andAlzheimer's disease (AD) [16].

We have previously demonstrated that insulin protects againstoxidative stress-mediated neuronal damage [17]. Furthermore, insu-lin stimulated glucose uptake and metabolism upon ascorbate/Fe2+,

Fig. 1. Analysis of immunocytofluorescence and tyrosine phosphorylation activity of IR (A,C) and IGF-1R (B,D) in cultured cortical neurons. The cells were incubated in the absence orpresence of 0.1 and 10 µM insulin, incubated, respectively, for 48 h or immediately before, and also during and after the induction of oxidative stress,with ascorbate plus Fe2+ (Asc/Fe2+).Immunocytofluorescence was performed using antibodies against the β-subunits of IR (A) or IGF-1R (B) and anti-MAP2 to label differentiated neurons. IR and IGF-1R activities weremeasured after immunoprecipitation of cytosolic extracts with rabbit anti-IR, β-subunit (C) or rabbit polyclonal anti-IGF-1Rβ (D) antibodies, followed by SDS/PAGE (10%) with mousemonoclonal anti-P-Tyr. Results are themean±SEM of three independent experiments performed in duplicate (considering 139523.3 and 125190.6 INT/mm2 as 100% for IR and IGF-1R,respectively). Statistical significance: ⁎Pb0.05, ⁎⁎Pb0.01 vs Control; £Pb0.05 vs Asc/Fe2+ without insulin.

995A.I. Duarte et al. / Biochimica et Biophysica Acta 1783 (2008) 994–1002

996 A.I. Duarte et al. / Biochimica et Biophysica Acta 1783 (2008) 994–1002

increasing intracellular energy and adenosine levels [18], which maybe metabolized to uric acid [17]. Increased recycling of reducedglutathione (GSH) (due to the stimulation of glutathione reductase(GRed) and inhibition of glutathione peroxidase (GPx) activities) in-duced by insulin also contributed to the replenishment of antioxidantdefenses [17]. Additionally, insulin neuroprotecion against oxidativestress involved a decrease in extracellular adenosine [18]. Based on theeffect of selective inhibitors, protective effects of insulin under oxi-dative stress appeared to be mediated by PI-3K and ERK signalingpathways [17,18]. Thus, the aim of the current work was to investigatethe effect of insulin on IR- and IGF-1R-mediated intracellular signalingpathways and the changes in the expression of “candidate” proteinsinvolved in the regulation of antioxidant defenses, energy metabolismand apoptosis following oxidative stress in rat primary cortical neu-rons. We demonstrate that activation of IR/IGF-1R mediate insulinneuroprotection against oxidative stress through stimulation of PI-3K/Akt and inhibition of GSK-3β, involving the regulation of proteinexpression and neuronal survival.

2. Materials and methods

2.1. Materials

Insulin from porcine pancreas, protease inhibitors, anti-IR (β-subunit) and-microtubule-associated protein 2 (MAP2), and 3-[(3-cholamidopropyl) dimethylam-monio]-1-propane-sulfonate (CHAPS) were from Sigma-Chemical Co. (St. Louis, MO,USA). Neurobasal Medium and B-27 supplement were from Gibco (Paisley, Scotland,UK). Hoechst 33342, Alexa Fluor 594 anti-rabbit IgG, Alexa Fluor 488 anti-mouse IgGand SuperScript™ III reverse transcription reagents were from Invitrogen (Eugene, OR,USA). Protein A sepharose CL-4B, polyvinylidene difluoride (PVDF) Hybond-P mem-

Fig. 2. Insulin-mediated activation of Akt (A), ERK1/2 (B,C) and GSK-3β (D,E) signaling paanalysed by western blotting with mouse anti-P-Akt (Ser473) monoclonal antibody, rabbit m(Ser9) and mouse anti-P-GSK-3β (Tyr216). Then, the membranes were reprobed with rabbitantibody, respectively. Results are the mean±SEM of three-seven independent experiments pAkt, P-p44 ERK/total p44 ERK, P-p42 ERK/total p42 ERK, P-GSK-3β (Ser9)/total GSK-3β and P-£Pb0.05, ££Pb0.01 vs Asc/Fe2+ in the absence of insulin.

branes, anti-mouse and -rabbit IgG antibodies conjugated to alkaline phosphatase andthe Enhanced ChemiFluorescence reagent (ECF) were from Amersham Biosciences(Little Chalfont, UK). Anti-α-tubulinwas from Zymed (San Francisco, CA, USA). Anti-IGF-1Rβ (H-60) and -Bcl-2 (C-2) were from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA,USA). Anti-P-Akt (Ser473), -P-p44/42 ERK (Thr202/Tyr204), -P-GSK-3β (Ser9), -p44/42ERK and -caspase-3 were from Cell Signaling (Beverly, MA, USA). Anti-Akt, -P-GSK-3β(Y216) and -total GSK-3β were from BD Biosciences (San Diego, CA, USA). Anti-type IIhexokinase (Hxk) and -glyceraldehyde-3-phosphate dehydrogenase (GAPDH) werefrom Chemicon International (Temecula, CA, USA). Anti-GPx-1 was from Abcam(Cambridge, UK). RNeasy Mini Kit, QIAshredder spin columns, RNase-Free DNase Set,and primers for Bcl-2 (Rn_Bcl2_1_SG QuantiTect Primer Assay), caspase-3(Rn_Casp3_1_SG QuantiTect Primer Assay), GPx-1 (Rn_Gpx1_1_SG QuantiTect PrimerAssay), Hxk-II (Rn_Hk2_1_SG QuantiTect Primer Assay), β-actin (Rn_Actb_1_SGQuantiTect Primer Assay) and β-2-microglobulin (Rn_B2m_1_SG QuantiTect PrimerAssay), and QuantiTect® SYBR Green PCR Master Mix were from Qiagen (Hilden,Germany). N-acetyl-Asp-Glu-Val-Asp-p-nitroanilide (Ac-DEVD-pNA) was from Calbio-chem (Darmstadt, Germany). RNA NanoLabChip (RNA 6000 Nano Assay) and DNA 1000LabChip Kit were from Agilent Technologies (Santa Clara, CA, USA). All other reagentswere of the highest grade of purity commercially available.

2.2. Primary culture of cortical neurons

Pregnant female Wistar rats were obtained from our local colony (Animal Facilities,Faculty of Medicine, University of Coimbra). Animals were kept under controlled lightand humidity conditions, being anesthesized under halothane atmosphere beforesacrifice by cervical displacement and decapitation (EU guideline 86/609/EEC). Corticalneurons were isolated from brains of 15–16 day-old rat embryos, using a previouslydescribed procedure [17]. Cells were maintained in Neurobasal Medium supplementedwith 2% B-27, 0.2 mM glutamine, 100U/ml penicillin, and 100mg/ml streptomycin andcultured for 6 days in 95% air and 5% CO2. Cortical neurons were plated on poly-L-lysine(0.1mg/ml) plates, at 3.1 × 105–2.5 × 106 cells/cm2, according to the experimentalprocedures. Cortical cultures contained few glial cells (less than 10%), as assessed byimmunofluorescence using anti-MAP2 and anti-GFAP (glial fibrillary acidic protein)[Almeida and Rego, unpublished data].

thways. Intracellular activation of PI-3K/Akt/GSK-3β or ERK signaling pathways wereonoclonal anti-P-p44/42 ERK (Thr202/Tyr204) antibody, rabbit polyclonal anti-P-GSK3βpolyclonal anti-Akt antibody, rabbit polyclonal p44/42 ERK antibody and mouse GSK-3βerformed in duplicate (considering 0.52, 0.92, 0.62, 0.69 and 0.92 as 100% for P-Akt/totalGSK-3β (Tyr216)/total GSK-3β, respectively). Statistical significance: ⁎Pb0.05 vs Control;

997A.I. Duarte et al. / Biochimica et Biophysica Acta 1783 (2008) 994–1002

2.3. Incubation with insulin and induction of oxidative stress

As previously reported [17,18], 0.1 or 10 μM insulin was added to Neurobasalmedium, 48 h (at 4 days in vitro, renewed every 24 h) or immediately (at 6 days in vitro)before induction of oxidative stress with 1.5 mM ascorbic acid and 7.5 μM FeSO4, for15 min, at 37 °C, in sodium saline solution, containing (in mM): 132.0 NaCl, 4.0 KCl,1.2 Na2HPO4, 1.4 MgCl2, 1.0 CaCl2, 6.0 glucose and 10.0 HEPES, pH 7.4. Cells were furtherincubated for 5h in Neurobasal medium in the absence of the oxidizing agents. Insulinwas present throughout the experiments. Ascorbate and Fe2+ solutions were preparedimmediately before use and maintained at 0–4 °C, protected from light.

2.4. IR and IGF-1R immunocytofluorescence

Neurons were washed twice with saline buffer (PBS, in mM: 137 NaCl, 2.7 KCl,1.4 K2HPO4, 4.3 KH2PO4, pH 7.4), and fixed in 4% paraformaldehyde containing 4%sucrose for 15 min. Cells were then permeabilized with 0.2% Triton X-100 for 2 min andblocked in 3% BSA for 90 min. Permeabilized cells were then co-incubated with rabbitanti-IR (β-subunit) or anti-IGF-1R (β-subunit) (10 μg/ml or 1:200, in 3% BSA, re-spectively), andmouse anti-MAP2 (1:500, in 3% BSA), for 60min at room temperature. IRor IGF-1R and MAP2 were detected by using the secondary antibodies Alexa Fluor 594anti-rabbit IgG (1:100in 3% BSA) and Alexa Fluor 488 anti-mouse IgG (1:200 in 3% BSA),respectively, incubated for 60 min. Nuclei were stained with Hoechst 33342 (2 μg/ml, inPBS) for 5 min, protected from light. Fixed cells were treated with DakoCytomationFluorescentmounting solution on amicroscope slide, and neuronswere visualized in anepifluorescence Axiovert 200 fluorescence microscope (Zeiss Axioscope, Zeiss,Germany).

2.5. Immunoprecipitation of IR and IGF-1R

After incubation, neurons were scraped with immunoprecipitation buffer (IPB)containing (in mM): 20 Tris–HCl, 100 NaCl, 2 EDTA and 2 EGTA, pH 7.0, supplementedwith 1% SDS, 50 mM NaF, 1 mM sodium o-vanadate, 0.1 μM okadaic acid, 1 mM DTT,100 μM PMSF and 1:1000 of protease inhibitors cocktail (containing chymostatin,pepstatin A, leupeptin, and antipain,1mg/ml). Themixturewas homogenized in 1:5 IPB,containing 1% Triton X-100, at 0–4 °C, and centrifuged at 2500 rpm for 12 min in anEppendorf 5415C centrifuge, to remove the nuclei. The resulting supernatant was col-lected and the pellet resuspended in IPB, and centrifuged again at 2500 rpm for 12 min.

Fig. 3. Relative gene expression of GPx-1 (A), Hxk-II (B), Bcl-2 (C) and caspase-3 (D) upon treQiagen RNeasy Mini Kit. Single-stranded cDNAwas synthesized by incubating 50 ng/µl of RNtime RT-PCR. Amplification reaction mixture contained the cDNA sample, 2× QuantiTect® SYlevels of the target genes were normalized to the housekeeping genes β-actin and β-2-micduplicate. Statistical significance: ⁎Pb0.05 vs Control; £Pb0.05, ££Pb0.01 vs Asc/Fe2+ in the

The supernatant was added to the previous one and protein content was determinedusing the BioRad reagent.

Immunoprecipitation of neuronal cytosolic extracts (100 μg) was performed aspreviously described [17]. Briefly, cytosolic extracts were mixed with a slurry of protein Asepharose beads (1:1, in IPB), for 1 h at 4 °C. Simultaneously, a slurry of protein A sepharosebeads were mixed with 5 μl of rabbit anti-IR (β-subunit) or anti-IGF-1Rβ in IPB plus 1%Triton X-100, for 2 h, at 4 °C. Mixtures were centrifuged at 14,000 rpm, 0–4 °C, for 10 min.The supernatant of the first mixture was recovered and incubated with IR or IGF-1Rantibodies covalently cross-linked to protein sepharose A (pellet) for 4 h, at 4 °C. Immunecomplexes were then centrifuged at 14,000 rpm, 10 min, and the pellet was washed withIPB. Finally, the immune complexes were denatured with SDS sample buffer, containing0.5M Tris–HCl, 0.4% SDS, pH 6.8, supplemented with 30% glycerol, 10% SDS, 0.6M DTT and0.012% bromophenol blue, at 100 °C, for 5min, and then centrifuged at 14,000 rpm,10min,using Spin-X® centrifuge tube filter to separate the protein A sepharose beads.

2.6. Western blotting

After incubation, neurons were scraped with buffered solution containing (in mM):250 sucrose, 20 HEPES, 10 KCl, 1.5 MgCl2, 1 EDTA, and 1 EGTA, pH 7.4, supplemented asdescribed for the preparation of immunoprecipitation samples.

Samples containing immunoprecipitated denatured proteins or the subcellularsuspension were immediately subjected to SDS/PAGE (10%) analysis and transferredonto PVDF Hybond-P membranes. Membranes containing immunoprecipitated IR andIGF-1R were incubated with mouse anti-P-Tyr (1:1000), overnight at 4 °C. Membranescontaining subcellular proteins were incubated with mouse P-Akt (Ser473) (1:1000),rabbit P-p44/42 ERK (Thr202/Tyr204) (1:1000), mouse P-GSK-3β (Tyr216) (1:1000),mouse Bcl-2 (1:200), rabbit caspase-3 (1:1000), rabbit GPx-1 (1:200) and rabbit Hxk-II(1:5000) antibodies. Membranes were reprobed with rabbit Akt (0.40 μg/ml), rabbitp44/42 ERK (1:1000), rabbit P-GSK3β (Ser9) (1:1000) and mouse GSK-3β antibo-dies (1:2500), respectively. Membranes were also labelled withmouse GAPDH (1:2000)or α-tubulin (1:2500) antibodies (in the case of caspase-3 labelling due to the proximityof molecular weights). Membranes were then incubated with anti-rabbit and anti-mouse IgG secondary antibodies (1:25,000), for 2 h at room temperature, and developedusing ECF. Immunoreactive bands were visualized by the VersaDoc Imaging System(BioRad, Hercules, CA, USA). Fluorescence signal was analysed using the QuantityOnesoftware and the results were given as INT/mm2. Phosphorylated protein/total pro-tein and protein expression levels, corresponding to the ratio of each protein vs GAPDHor α-tubulin, were expressed as % of control.

atment with insulin. Total RNA was extracted from cortical neuronal cultures using theA with SuperScript™III reverse transcription reagents. The mRNAwas analysed by real-BR Green PCR Master Mix, QuantiTect® Primer Assay and RNase-free water. The mRNAroglobulin. Results are the mean±SEM of five independent experiments performed inabsence of insulin.

998 A.I. Duarte et al. / Biochimica et Biophysica Acta 1783 (2008) 994–1002

2.7. Total RNA extraction and reverse transcription

Total RNAwas extracted using the RNeasyMini kit, according to the manufacturer'sprotocol. After incubation, cortical neurons were rapidly scrapped with diluted RLTbuffer, containing β-mercaptoethanol (10μl/ml), collected into RNase- and DNase-freeeppendorf tubes and stored immediately at −80 °C. Furthermore, RNA samples weretreated with RNase-Free DNase set, according to manufacturer's instructions, to removeany residual genomic DNA. RNA sample was diluted in RNase-free water and the sam-ples were stored at −80 °C.

RNA integrity of randomly chosen samples was analyzed in an Agilent 2100 Bio-analyzer using a RNA NanoLabChip (RNA 6000 Nano Assay) and RNA integrity numbers(RIN) were greater than 7 in all analyzed samples. Single-stranded cDNA was syn-thesized using SuperScript™ III reverse transcription reagents, by incubating total RNA(50 ng/μl of DNase-treated RNA) with 2.5 ng/μl random hexamers, 0.5 mM dNTP mixand RNase-freewater, for 5min at 65 °C. Then, RNA/primermixwas incubated in a GeneAmp PCR System 9600 thermocycler (Perkin-Elmer, Wellesley, MA, USA) for 10 min at25 °C with cDNA synthesis mix containing 10 mM DTT, 2 U/μl RNase OUT, 10 U/μlSuperScript™ III reverse transcriptase and FS buffer. cDNA synthesis occurred at 50 °Cfor 50 min and reactionwas stopped by inactivation of reverse transcriptase at 85 °C for5 min. Samples were incubated for 20 min at 37 °C with RNase H.

2.8. Real-time RT-PCR

Real-time RT-PCR was performed to monitor the expression of Bcl-2, Caspase-3(Casp3), GPx1 and Hxk-II, and of the endogenous control genes β-actin (Actb) and β-2-microglobulin (B2m). The HotStarTaq DNA Polymerase technology was used and theresults were analyzed in a 7900 HT sequence detector system (Applied Biosystems,Foster City, CA, USA).

Amplification reaction mixture contained cDNA sample, QuantiTect® SYBR GreenPCR Master Mix, QuantiTect® Primer Assay and RNase-free water. For real-time PCR,each mixture was added to 96-well optical reaction plate. For control purposes, non-template samples were subjected to PCR amplification. Thermal cycling conditionsincluded 10min at 95 °C, proceeding with 40 cycles of 95 °C for 15 s and 60 °C for 1 min.The size of the PCR product was determined in an Agilent 2100 bioanalyzer using a DNA1000 LabChip Kit. mRNA levels of the target genes were normalized to that of thegeometric mean of endogenous control genes Actb and B2m, and expressed relative tocontrol in the absence of insulin using the ΔΔCt method.

Fig. 4. GPx-1 (A), Hxk-II (B), Bcl-2 (C) and procaspase-3 (D) protein expression levels in coblotting with rabbit polyclonal GPx-1, rabbit type II Hxk, mouse Bcl-2 (C-2) monoclonal, rabantibodies. Results are the mean±SEM of three-seven independent experiments performeGAPDH, Bcl-2/GAPDH and procaspase-3/α-tubulin, respectively). Statistical significance: ⁎P

2.9. Colorimetric evaluation of caspase-3-like activity

Caspase-3-like activitywasdetermined by following a previously describedprocedure[19]. Briefly, 25 μg protein were added to a reaction buffer containing 25 mM HEPES, 10%sucrose, 0.1%CHAPS, pH7.5, supplementedwith 10mMDTTand colorimetric substrateAc-DEVD-pNA (100 μM). After incubation at 37 °C for 2 h, formation of pNAwas measured at405 nm in a Synergy™-HT Multidetection Microplate Reader (Bio-Tek®Instruments, Inc,Winooski, VT, USA). Caspase-3-like activity was calculated as the increase above thecontrol for equal protein loads, and the results expressed in % of control.

2.10. Assessment of apoptotic neurons

Apoptotic cells were characterized using the fluorescent probe Hoechst 33342.Briefly, 24 h after oxidative stress, cells were loaded with 15 μg/ml Hoechst 33342 inPBS, for 15 min at 37 °C. Cells (viable and apoptotic) were visualized using anepifluorescence Axiovert 200 fluorescence microscope. Condensed/fragmented (apop-totic) nuclei were markedly bright and small or divided into several chromatin clumpsafter staining with Hoechst 33342. Viable cells showed diffuse Hoechst labeling.

2.11. Data analysis and statistics

Results are expressed as mean ± SEM of the indicated number of independentexperiments, run in duplicates or triplicates. Statistical significance was analyzed usingone-way-ANOVA for multiple comparisons, with Bonferroni post-test. Pb0.05 wasconsidered significant.

3. Results

3.1. Insulin prevents the inhibition of neuronal IR and IGF-1R activityupon exposure to ascorbate/Fe2+

Immunocytofluorescence revealed a co-localization of both IR andIGF-1Rwithanti-MAP2 labeling, suggesting that both receptors arewidelyspread throughout cultured cortical neurons (Fig. 1A,B). Exposure toascorbate/Fe2+ did not affect the distribution of both receptors (Fig. 1A,B).

rtical neurons subjected to insulin treatment. Cell extracts were analysed by westernbit polyclonal caspase-3, mouse monoclonal GAPDH and mouse monoclonal α-tubulind in duplicate (considering 0.72, 0.91, 1.29 and 1.82 as 100% for GPx-1/GAPDH, Hxk-II/b0.05 vs Control; £Pb0.05, ££Pb0.01, £££Pb0.001 vs Asc/Fe2+ without insulin.

999A.I. Duarte et al. / Biochimica et Biophysica Acta 1783 (2008) 994–1002

Insulin may activate both IR and IGF-1R, stimulating their Tyrkinase activity and initiating intracellular signaling pathways [7].Although insulin (0.1 and 10 μM) per se rapidly activated IR and IGF-1Rafter incubation for 10 or 30 min (particularly for 10 μM) (data notshown), we did not observe a significant activation of these receptorsupon longer incubation periods used throughout this study (Fig. 1C,D).Nevertheless, insulin significantly prevented oxidative stress-induceddecrease in Tyr phosphorylation of both receptors (Fig. 1C,D),suggesting the restoration of IR and IGF-1R activities.

3.2. Insulin modulates intracellular PI-3K/Akt and GSK-3β signalingpathways upon oxidative stress

Following our previous results, we further analysed the role ofinsulin on the phosphorylation of Akt, ERK and GSK-3β (Fig. 2A–E).Although insulin per se had no significant effect on P-Akt, con-cordantly with no changes in receptor activity (Fig. 1C,D), it preventedthe decrease in P-Akt associated with oxidative stress (Fig. 2A),suggesting that PI-3K/Akt signaling plays an important role on insulin-mediated effects on cortical neurons. It should be noted that, undercontrol conditions, exposure to 0.1 or 10 μM insulin for 10 or 30 minincreased P-Akt (data not shown).

Fig. 5. Apoptotic neuronal death in cortical neurons treated with insulin. Apoptosis wcondensation/fragmentation (B). Caspase-3-like activity was determined by following thefluorescent probe Hoechst 33342. Results are the mean±SEM of five-ten independent exp⁎⁎Pb0.01 vs Control; ££Pb0.01, £££Pb0.001 vs Asc/Fe2+ in the absence of insulin. In (B) are10 µM insulin (Ins0.1 and Ins10, respectively), visualized in an inverted Nikon Diaphot fluo

Ras activates Raf-1, which sequentially phosphorylates and acti-vates MEK and ERK1/2 [9]. However, neither P-ERK p44 nor P-ERK p42was significantly affected by insulin (0.1 and 10 μM) or ascorbate/Fe2+

(Fig. 2B,C).Phosphorylation of GSK-3β on Ser9 inhibits its activity, whereas at

Tyr216 activates the enzyme [20]. We also evaluated the involvementof GSK-3β on insulin neuroprotection (Fig. 2D,E). Neither insulin norascorbate/Fe2+ changed significantly the levels of GSK-3β phosphory-lated on Ser9 (Fig. 2D). No significant changes were observed in thephosphorylation of Tyr216 upon oxidative stress either (Fig. 2E).Interestingly, insulin (0.1 and 10 μM) decreased GSK-3βTyr216phosphorylation, the active form of GSK-3β, by about 26%, particularlyin the presence of ascorbate/Fe2+ (Fig. 2E).

3.3. Insulin regulates mRNA and protein expression levels of “candidate”proteins upon oxidative stress

Following our previous data [17,18], we analysed the role of insulinon the expression of “candidate” proteins involved in these pathways,namely GPx-1, Hxk-II, Bcl-2 and caspase-3 (Figs. 3 and 4).

Upon ascorbate/Fe2+ exposure, both 0.1 and 10 μM insulin pre-vented the increase in GPx-1 and caspase-3 mRNA levels (Fig. 3A,D).

as analysed by the activation of the effector caspase-3 (A), and by assessing DNAextent of cleavage of Ac-DEVD-pNa, at 405 nm, whereas DNA was analysed with theeriments performed in duplicate (considering 0.015 as 100%). Statistical significance:representative images of viable and apoptotic neurons upon treatment with 0.1 and

rescence microscope.

1000 A.I. Duarte et al. / Biochimica et Biophysica Acta 1783 (2008) 994–1002

Additionally, insulin rescued oxidative stress-induced decrease inHxk-II and Bcl-2 mRNA expression levels (Fig. 3B,C). These resultssuggested that insulin-mediated activation of intracellular signalingpathways under oxidative stress conditions culminated in theregulation of gene expression, thus affecting subsequent protein ex-pression. To further test this hypothesis, we measured the expressionlevels of the “candidate” proteins GPx-1, Hxk-II, Bcl-2 and procaspase-3 by western blotting (Fig. 4A–D). Similarly to the results shown inFig. 3, insulin (0.1 and 10 μM) completely prevented the increase inGPx-1 (Fig. 4A) and procaspase-3 (Fig. 4D) and the decrease in Hxk-II(Fig. 4B) and Bcl-2 (Fig. 4C) protein expression induced by ascorbate/Fe2+. Insulin per se did not significantly affect the mRNA or proteinexpression levels of these “candidate” proteins.

Hence, insulin neuroprotection against oxidative stress mayinvolve the regulation of the expression of genes and proteins in-volved in glutathione redox cycle, glycolytic energy metabolism andanti-apoptotic processes.

3.4. Insulin protects against apoptosis under oxidative stress

Since insulin stimulates Bcl-2 expression (Figs. 3C and 4C) andinhibits caspase-3 expression (Figs. 3D and 4D), we further investi-gated its effects on caspase-mediated apoptosis, by determiningcaspase-3 activity and DNA fragmentation (Fig. 5A,B). We observedthat ascorbate/Fe2+-induced increase in caspase-3 activity (by about52%) was completely prevented by 0.1 and 10 μM insulin. No changeswere induced by insulin per se (Fig. 5A). These results were furthercorroborated by the observation that oxidative stress-mediated DNAfragmentation was prevented by insulin (Fig. 5B).

Data suggest that, by activating IR/IGF-1R signaling pathways,insulin rescues cortical neurons from caspase-3-mediated apoptoticcell death.

4. Discussion

Recent hypothesis suggest that insulin and IGF-1 can be themissing link between diabetes and AD, and possibly other neurode-generative diseases, for which oxidative stress has been extensively

Fig. 6. Stimulation of IR/IGF-1R-induced PI-3K/Akt signaling pathway mediates insulin neconditions induces the phosphorylation of IR and/or IGF-1R, whichmaintains the activation oproteins, namely GPx-1, Hxk-II, Bcl-2 and caspase-3, thus helping to counterbalance oxidatwith GSK-3β signaling, decreasing its activated form and inhibiting apoptotic neuronal deathhexokinase-II; GSH, reduced glutathione; GSSG, oxidized glutathione.

described [16]. Thus, deleterious effects induced by oxidative stress onneuronal IR/IGF-1R signaling are related with altered brain function.

In the present study, we analysed the involvement of IR/IGF-1Rintracellular signaling pathways on insulin neuroprotection againstoxidative stress [17,18]. Both IR and IGF-1R were present in corticalneurons, in agreement with other reports [5,21]. Although rat brain IRmRNA is most abundant at embryonic day 20 and at birth, parallelingthe changes observed in insulin binding and IR density [22], theduration of IR activation remains uncertain. Insulin-mediated neuro-nal IR activation phosphorylates IRS-2 within 2–5 min, with a peak at20–30min [23,24], further phosphorylating PI-3K. Neuronal Akt, GSK-3β, ERK1/2 and FoxO1 were also maximally phosphorylated in re-sponse to IR activation 10 min after peripheral insulin injection [24].This was not observed in NIRKO mice, a neuronal/brain-specific IRknockout [24] or upon wortmannin exposure [8]. Therefore, it is notsurprising that insulin per se did not show IR or IGF-1R Tyrphosphorylation under our experimental conditions, since the short-est time-interval for insulin exposure of cortical neurons exceeds 5h.Nevertheless, Huang et al. [25] showed that insulin can induce a long-lasting (30 h, 1 nM) PI-3K-dependent phosphorylation of Akt, incultured adult sensory neurons. Unlike peripheral IR, brain IR is notdownregulated upon exposure to high insulin levels [26], supportinginsulin-mediated effects upon ascorbate/Fe2+.

In this study, ascorbate/Fe2+-induced caspase-3 activation was ac-companied by a decrease in Bcl-2 expression, suggesting that caspase-3 mediates apoptotic neuronal death under oxidative stress, support-ing our previous observations [17]. This agrees with the observationthat ROS-induced mitochondrial dysfunction may result in mitochon-drial cytochrome c release, apoptosome activation, caspase-3 cleavageand apoptosis [27].

PI-3K/Akt regulates the survival response against oxidative stress-associated neuronal apoptosis [28]. In the present study, ascorbate/Fe2+

neurotoxicity occurred via inhibition of both IR and IGF-1R, leading toimpairment of Akt phosphorylation. Insulin neuroprotection againstoxidative stress involved the stimulation of IR and IGF-1R activity andthe restoration of PI-3K/Akt signaling pathway. Neuroprotection can befurther afforded by phosphorylation of GSK-3β at Ser9, which becomesinactive, as occurs in primary cultured neurons and in the hippocampal

uroprotection against oxidative stress. Treatment with insulin under oxidative stressf the PI-3K/Akt signaling pathway. This appears to regulate the expression of “candidate”ive stress, impaired glucose metabolism and neuronal apoptosis. Insulin also interferesunder oxidized conditions. Casp-3, caspase-3; GPx-1, glutathione peroxidase-1; Hxk-II,

1001A.I. Duarte et al. / Biochimica et Biophysica Acta 1783 (2008) 994–1002

HT22 neurons [29]. As a consequence, GSK-3β-induced inhibition ofheat shock factor-1 [30] and CREB activity is abolished, stimulating Bcl-2expression [11] and protecting against caspase-3-dependent and-independent neuronal apoptosis [31]. Conversely, stimulation of GSK-3β activity due to chronic inhibition of insulin signaling caused apop-tosis involving oxidative and endoplasmic reticulum stress [32–34].Despite no changes in phosphorylation of GSK-3βSer9 (inactive) in thepresent study, insulin decreased GSK-3βTyr216 (active) phosphoryla-tion. Oxidative stress did not interfere with GSK-3β phosphorylation,suggesting that its deleterious effects may involve other mechanisms,namely decreased phosphorylation of Bad [14]. Although less investi-gated, inhibition of GSK-3βTyr216 phosphorylation in neurons can beinvolved in insulin stimulation of Bcl-2 expressionuponoxidative stress.Indeed, under oxidative stress, insulin prevented changes in expressionof genes linked to cell survival, namely Bcl-2 and caspase-3.

We previously showed that ascorbate/Fe2+ impaired neuronalglucose uptake and metabolism [18], which was accounted for by anincreased oxidation of GLUT3 transporter by 4-hydroxynonenal [17].Decreased Hxk-II levels induced by oxidative stress, as observed in thepresent study, may further contribute to glycolysis inhibition andlower intracellular ATP levels [18]. Hxk-II represents the rate-limitingstep in neuronal glycolysis — the formation of glucose-6-phosphate[35], being also a downstream effector of Akt-mediated cell survival inrat fibroblasts [36]. Thus, stimulation of Hxk-II expression by insulincorrelates with the increase in glucose metabolism and intracellularATP and adenosine levels upon oxidation [18].

Here we show that exposure to ascorbate/Fe2+ increased GPx-1mRNA and protein expression, which correlates with increased GPxactivity [17], thus contributing to glutathione redox cycle impairment.Since GPx-1 is the main neuronal antioxidant enzyme in the cytosoland mitochondria [37], insulin-induced decrease in expression andactivity of GPx-1 may contribute to the replenishment of antioxidantdefenses under oxidative stress [17].

Neuronal IR and IGF-1R can also activate ERK1/2 signaling, thusmodulating gene expression [38]. H2O2 may activate MEK-ERK1/2signaling, stimulating neuronal gene expression [28] or protectingagainst acute neuronal apoptosis [39]. Although ERK1/2 signalingcascade was unmodified in our experimental conditions, we cannotexclude its involvement in insulin effects against oxidation-mediatedinjury. Indeed, we previously observed that insulin-mediated protec-tive effects under oxidative stress were precluded in the presence ofthe MEK inhibitor PD98059. This inhibitor altered the levels of in-tracellular uric acid, ATP and adenosine, highly suggesting the in-volvement of the ERK1/2 pathway [17,18]. The apparent discrepancymay be explained by the fact that ERK1/2 signaling was determined atthe end of the experiments, whereas PD98059 was incubated prior toinsulin exposure.

In summary, we demonstrate that insulin neuroprotection againstoxidative stress is mediated by IR- and/or IGF-1R-induced activation ofPI-3K/Akt and inhibition of GSK-3β signaling pathways, which appearto modulate the expression of proteins involved in glucose metabo-lism (Hxk-II), antioxidant defenses (GPx-1) and anti-apoptotic mech-anisms (Bcl-2 and caspase-3) under ascorbate/Fe2+ treatment, thusprotecting cortical neurons against caspase-3-dependent apoptosis(Fig. 6). These results are in agreement with insulin stimulation ofneuronal glucose uptake and metabolism, the increase in intracellularATP, adenosine, uric acid and glutathione redox cycle and the decreasein lipid and protein oxidation upon oxidative stress [17,18] (Fig. 6).

These data are important, since oxidative stressmay be themissinglink between the loss of β-pancreatic cells and neurons that underliestype 2 diabetes and AD [2,40]. AD has been considered “brain-typediabetes” [16] and insulin may constitute a therapeutic agent againstAD via activation of neuronal IR/IGF-1R signaling pathways. Never-theless, systemic insulin administration in AD should avoid hypergly-cemia-associated cognitive and memory deficits [41]. According toReger et al. [42], this limitation can be circumvented by intranasal

insulin administration, allowing its direct entry into the brainwithoutaffecting peripheral glucose or insulin levels.

Acknowledgements

This workwas supported by Fundação para a Ciência e a Tecnologia(FCT) (POCI/SAU-NEU/57310/2004 and PTDC/SAL-FCF/66421/2006).Ana Duarte is supported by Science and Technology Foundation (POCI2010) and European Social Fund (SFRH/BD/5338/2001). We thankTeresa Oliveira (Center for Neuroscience and Cell Biology) for kindlyhelping with fluorescent microscopy and immunocytochemistry ana-lysis. We also thank Dr. Cláudia Pereira (Center for Neuroscience andCell Biology) for kindly providing the antibodies against GSK-3β andcaspase-3, and to Dr. Paula Agostinho (Center for Neuroscience andCell Biology) for kindly supplying the antibody against total ERK1/2.

References

[1] C.R. Kahn, M.F. White, S.E. Shoelson, J.M. Backer, E. Araki, B. Cheatham, P. Csermely,F. Folli, B.J. Goldstein, P. Huertas, P.L. Rothenberg, M.J.A. Saad, K. Siddle, X.J. Sun, P.A.Wilden, K. Yamada, S.A. Kahn, The insulin receptor and its substrate: moleculardeterminants of early events in insulin action, Recent Prog. Horm. Res. 48 (1993)291–339.

[2] S. Craft, G.S. Watson, Insulin and neurodegenerative disease: shared and specificmechanisms, Lancet Neurol. 3 (2004) 169–178.

[3] R. Sankar, S. Thamotharan, D. Shin, K.H. Moley, S.U. Devaskar, Insulin-responsiveglucose transporters — GLUT8 and GLUT4 are expressed in the developingmammalian brain, Mol. Brain Res. 107 (2002) 157–165.

[4] S. Hoyer, Memory function and brain glucose metabolism, Pharmacopsychiatry 36(Suppl 1) (2003) S62–S67.

[5] W. Zhao, H. Chen, H. Xu, E. Moore, N. Meiri, M.J. Quon, D. Alkon, Brain insulinreceptors and spatial memory. Interrelated changes in gene expression, tyrosinephosphorylation, and signaling molecules in the hippocampus of water mazetrained rats, J. Biol. Chem. 274 (1999) 34893–34902.

[6] R. Conejo, M. Lorenzo, Insulin signaling leading to proliferation, survival, andmembrane ruffling in C2C12 myoblasts, J. Cell. Physiol. 187 (2001) 96–108.

[7] G.M. Cole, S.A. Frautschy, The role of insulin and neurotrophic factor signaling inbrain aging and Alzheimer's Disease, Exp. Gerontol. 42 (2007) 10–21.

[8] M.D. Bruss, E.B. Arias, G.E. Lienhard, G.D. Cartee, Increased phosphorylation of Aktsubstrate of 160 kDa (AS160) in rat skeletal muscle in response to insulin orcontractile activity, Diabetes 54 (2005) 41–50.

[9] L.P. van der Heide, G.M.J. Ramakers, P.S. Marten, Insulin signaling in the centralnervous system: learning to survive, Prog. Neurobiol. 79 (2006) 205–221.

[10] A. Wada, H. Yokoo, T. Yanagita, H. Kobayashi, New twist on neuronal insulin receptorsignaling in health, disease and therapeutics, J. Pharmacol. Sci. 99 (2005) 128–143.

[11] S. Pugazhenthi, A. Nesterova, C. Sable, K.A. Heidenreich, L.M. Boxer, L.E. Heasley, J.E.Reusch, Akt/protein kinase B up-regulates Bcl-2 expression through cAMP-response element-binding protein, J. Biol. Chem. 275 (2000) 10761–10766.

[12] M. Valko, D. Leibfritz, J. Moncola, M.T.D. Cronin, M. Mazura, J. Telser, Free radicalsand antioxidants in normal physiological functions and human disease, Int. J.Biochem. Cell Biol. 39 (2007) 44–84.

[13] M.R. Duchen, Roles of mitochondria in health and disease, Diabetes 53 (Suppl. 1)(2004) S96–S102.

[14] Z.Z. Chong, F. Li, K. Maiese, Oxidative stress in the brain: novel cellular targets thatgovern survival during neurodegenerative disease, Prog. Neurobiol. 75 (2005)207–246.

[15] B.R. Ryu, H.W. Ko, I. Jou, J.S. Noh, B.J. Gwag, Phosphatidylinositol 3-kinase-mediated regulation of neuronal apoptosis and necrosis by insulin and IGF-1,J. Neurobiol. 39 (1999) 536–546.

[16] S.M. de la Monte, J.R. Wands, Molecular indices of oxidative stress and mito-chondrial dysfunction occur early and often progress with severity of Alzheimer'sdisease, J. Alzheimer's Dis. 9 (2006) 167–181.

[17] A.I. Duarte, M.S. Santos, C.R. Oliveira, A.C. Rego, Insulin neuroprotection againstoxidative stress in cortical neurons: involvement of uric acid and glutathioneantioxidant defenses, Free Radic. Biol. Med. 39 (2005) 876–889.

[18] A.I. Duarte, T. Proença, C.R. Oliveira, M.S. Santos, A.C. Rego, Insulin restoresmetabolic function in cultured cortical neurons subjected to oxidative stress,Diabetes 55 (2006) 2863–2870.

[19] J. Gil, S. Almeida, C.R. Oliveira, A.C. Rego, Cytosolic and mitochondrial ROS instaurosporine-induced retinal cell apoptosis, Free Radic. Biol. Med. 35 (2003)1500–1514.

[20] S. Frame, P. Cohen, GSK3 takes centre stage more than 20 years after its discovery,Biochem. J. 359 (2001) 1–16.

[21] K.A. Heidenreich, Insulin and IGF-I receptor signaling in cultured neurons, Ann. N.Y.Acad. Sci. 692 (1993) 72–88.

[22] T.C. Schulingkamp, D. Pagano, R.C. Hung, R.J. Raffa, Insulin receptors and insulinaction in the brain: review and clinical implications, Neurosci. Biobehav. Rev. 24(2000) 855–872.

[23] K.D. Niswender, M.W. Schwartz, Insulin and leptin revisited: adiposity signals withoverlapping physiological and intracellular signaling capabilities, Front. Neuroen-docrinol. 24 (2003) 1–10.

1002 A.I. Duarte et al. / Biochimica et Biophysica Acta 1783 (2008) 994–1002

[24] S. Freude, L. Plum, J. Schnitker, U. Leeser, M. Udelhoven, W. Krone, J.C. Bruning, M.Schubert, Peripheral hyperinsulinemia promotes tau phosphorylation in vivo,Diabetes 54 (2005) 3343–3348.

[25] T.-J. Huang, A. Verkhratsky, P. Fernyhough, Insulin enhances mitochondrial innermembrane potential and increases ATP levels through phosphoinositide 3-kinasein adult sensory neurons, Mol. Cell. Neurosci. 28 (2005) 42–54.

[26] N.R. Zahniser, M.B. Goens, P.J. Hanaway, J.V. Vinych, Characterization andregulation of insulin receptors in rat brain, J. Neurochem. 42 (1984) 1354–1362.

[27] J.W. Russell, D. Golovoy, A.M. Vincent, P. Mahendru, J.A. Olzmann, A. Mentzer, E.L.Feldman, High glucose induced oxidative stress and mitochondrial dysfunction inneurons, FASEB J. 16 (2002) 1738–1748.

[28] H. Hallak, B. Ramadan, R. Rubin, Tyrosine phosphorylation of insulin receptorsubstrate-1 (IRS-1) by oxidant stress in cerebellar granule neurons: modulationbyN-methyl-D-aspartate through calcineurin activity, J. Neurochem. 77 (2001) 63–70.

[29] M. Schäfer, S. Goodenough, B. Moosmann, C. Behl, Inhibition of glycogen synthasekinase 3β is involved in the resistance to oxidative stress in neuronal HT22 cells,Brain Res. 1005 (2004) 84–89.

[30] G.N. Bijur, P. De Sarno, R.S. Jope, Glycogen synthase kinase-3beta facilitatesstaurosporine- and heat shock-induced apoptosis. Protection by lithium, J. Biol.Chem. 275 (2000) 7583–7590.

[31] D. Martin, M. Salinas, N. Fujita, T. Tsuruo, A. Cuadrado, Ceramide and reactiveoxygen species generated by H2O2 induce caspase-3-independent degradation ofAkt/protein kinase B, J. Biol. Chem. 277 (2002) 42943–42952.

[32] T. Hayashi, A. Saito, S. Okuno, M. Ferrand-Drake, R.L. Dodd, P.H. Chan, Damage tothe endoplasmic reticulum and activation of apoptotic machinery by oxidativestress in ischemic neurons, J. Cereb. Blood Flow Metab. 25 (2005) 41–53.

[33] L. Song, P. De Sarno, R.S. Jope, Central role of glycogen synthase kinase-3beta inendoplasmic reticulum stress-induced caspase-3 activation, J. Biol. Chem. 277(2002) 44701–44708.

[34] S. Srinivasan, M. Ohsugi, Z. Liu, S. Fatrai, E. Bernal-Mizrachi, M.A. Permutt,Endoplasmic reticulum stress-induced apoptosis is partly mediated by reducedinsulin signaling through phosphatidylinositol 3-kinase/Akt and increased glyco-gen synthase kinase-3β in mouse insulinoma cells, Diabetes 54 (2005) 968–975.

[35] J.E.Wilson, Isozymes of mammalian hexokinase: structure, subcellular localizationand metabolic function, J. Exp. Biol. 206 (2003) 2049–2057.

[36] K. Gottlob, N. Majewski, S. Kennedy, E. Kandel, R.B. Robey, N. Hay, Inhibition ofearly apoptotic events by Akt/PKB is dependent on the first committed step ofglycolysis and mitochondrial hexokinase, Genes Dev. 15 (2001) 1406–1418.

[37] J. Cai, P. Sternberg Jr., D.P. Jones, Oxidative stress and age-related maculardegeneration, in: R.G. Cutler, H. Rodriguez (Eds.), Critical Reviews of OxidativeStress and Aging. Advances in Basic Science, Diagnostics and Intervention, vol. 2,World Scientific, Singapore, 2003, pp. 987–998.

[38] C.-H. Lin, S.-H. Yeh, C.-H. Lin, K.-T. Lu, T.-H. Leu, W.-C. Chang, P.-W. Gean, A role forthe PI-3 kinase signaling pathway in fear conditioning and synaptic plasticity inthe amygdale, Neuron 31 (2001) 841–851.

[39] H.J. Lin, X.Wang, K.M. Shaffer, C.Y. Sasaki, W.Ma, Characterization of H2O2-inducedacute apoptosis in cultured neural stem/progenitor cells, FEBS Lett. 570 (2004)102–106.

[40] J. Janson, T. Laedtke, J.E. Parisi, P. O'Brien, R.C. Petersen, P.C. Butler, Increased risk oftype 2 diabetes in Alzheimer disease, Diabetes 53 (2004) 474–481.

[41] L. Zhao, B. Teter, T. Morihara, G.P. Lim, S.S. Ambegaokar, O.J. Ubeda, S.A. Frautschy,G.M. Cole, Insulin-degrading enzyme as a downstream target of insulin receptorsignaling cascade: implications for Alzheimer's disease intervention, J. Neurosci.24 (2004) 11120–11126.

[42] M.A. Reger, G.S. Watson, W.H. Frey 2nd, L.D. Baker, B. Cholerton, M.L. Keeling, D.A.Belongia, M.A. Fishel, S.R. Plymate, G.D. Schellenberg, M.M. Cherrier, S. Craft,Effects of intranasal insulin on cognition in memory-impaired older adults:modulation by APOE genotype, Neurobiol. Aging 27 (2006) 451–458.