journal of alzheimer’s disease 23 (2011) 257–269 ios press

Journal of Alzheimer’s Disease 23 (2011) 257–269DOI 10.3233/JAD-2010-101083IOS Press

257

Redox Proteomics Analysis of Brains fromSubjects with Amnestic Mild CognitiveImpairment Compared to Brains fromSubjects with Preclinical Alzheimer’sDisease: Insights into Memory Loss in MCI

Christopher D. Aluisea,b,c, Renã A.S. Robinsona,b,c, Jian Caid, William M. Pierced,William R. Markesberyc,† and D. Allan Butterfielda,b,c,∗aDepartment of Chemistry, University of Kentucky, Lexington, KY, USAbCenter of Membrane Sciences, University of Kentucky, Lexington, KY, USAcSanders-Brown Center on Aging, University of Kentucky, Lexington, KY, USAdDepartment of Pharmacology, University of Louisville, Louisville, KY, USA

Handling Associate Editor: Patrizia Mecocci

Accepted 20 September 2010

Abstract. Alzheimer’s disease (AD) is a central nervous system disorder pathologically characterized by senile plaques, neu-rofibrillary tangles, and synapse loss. A small percentage of individuals with normal antemortem psychometric scores, afteradjustments for age and education, meet the neuropathological criteria for amnestic mild cognitive impairment (MCI) or AD;these individuals have been termed ‘preclinical’ or ‘asymptomatic’ AD (PCAD). In this study, we employed the immunochemicalslot-blot method and two-dimensional gel-based redox proteomics to observe differences in protein levels and oxidative modi-fications between groups with equal levels of AD pathology who differ in regards to clinical symptoms of memory impairment.Results of global oxidative stress measurements revealed significantly higher levels of protein carbonyls in the MCI inferiorparietal lobule (IPL) relative to PCAD (and controls), despite equal levels of neuropathology. Proteomics analysis of the IPLrevealed differences in protein levels and specific carbonylation that are consistent with preservation of memory in PCAD andapparent memory decline in MCI. Our data suggest that marked changes occur at the protein level in MCI that may cause orreflect memory loss and other AD symptoms.

Keywords: Alzheimer’s disease, brain, mild cognitive impairment, oxidative stress, preclinical Alzheimer’s, proteomics, redoxproteomics, two dimensional gel electrophoresis

† Deceased Jan. 30, 2010.∗Correspondence to: Prof. D. Allan Butterfield, Department of

Chemistry, Center of Membrane Sciences, and Sanders-BrownCenter on Aging, University of Kentucky, Lexington, KY 40506,USA. Tel.: +1 859 257 3184; Fax: +1 859 257 5876; E-mail:[email protected].

INTRODUCTION

Alzheimer’s disease (AD) affects more than5 million Americans. This disease progresses pheno-typically by slight memory complaints initially, andculminates in the development of dementia. An inter-mediate stage between cognitively normal individuals

ISSN 1387-2877/11/$27.50 © 2011 – IOS Press and the authors. All rights reserved

mailto:[email protected]

258 C.D. Aluise et al. / Comparative Redox Proteomics in PCAD and MCI

and persons with AD is mild cognitive impairment(MCI). Persons with amnestic MCI have memoryloss but normal activities of daily living and demon-strate AD-like pathology [1]. AD is characterizedpathologically by senile plaques (SPs), neurofibrillarytangles (NFTs), and synapse loss. SPs are composedof amyloid-� peptide (A�), surrounded by numer-ous protein and dystrophic neurites, while NFTs arecomposed of hyperphosphorylated tau deposits. Bothplaques and tangles are a requirement for the defi-nite diagnosis of AD in concert with apparent signsof memory impairment and/or dementia. A� is pro-duced via the amyloidogenic pathway involving betaand gamma secretase-mediated enzymatic cleavage ofthe amyloid-� protein precursor (A�PP) [2–4], result-ing in the formation of the 42-amino acid fragment.A�1-42 is thought to be the most toxic form of A�,particularly upon oligomerization [5–9].

As mentioned previously, SPs, NFTs, and dementiaare all requirements for complete antemortem diag-nosis of AD. Interestingly, brains from a number ofcognitively intact individuals at autopsy reveal anextensive SP and NFT load. Therefore, the possibilityof a preclinical or presymptomatic disease state exists[10–12]. Following accepted criteria [10] and similarto our previous report on preclinical AD (PCAD) [13],we have defined PCAD as those individuals with suf-ficient AD pathological alterations to meet NationalInstitute on Aging-Reagan Institute (NIA-RI) interme-diate or, rarely, high likelihood criteria (Braak stageIII or higher and moderate or frequent neuritic plaquescores) for diagnosis of AD, but who had normal cog-nitive function as shown by antemortem psychometrictest scores within the normal range after adjustmentfor age and education. Literature on these individu-als is limited, most likely due to the rarity of sampleavailability; however, in addition to significant levelsof pathological hallmarks and no observable behav-ioral/memory problems, other anatomical/biochemicalfeatures of PCAD include neuronal hypertrophy [11],no hippocampal cell loss [14], increased synapticplasticity [15], and alterations in zinc transporters[16].

Proteomics has been used extensively in our labora-tory to identify differentially regulated and oxidativelymodified proteins in AD brain [17]. Proteomics-determined differences in proteins in AD brain aswell as mouse models thereof have provided valuableinsight into regulation or disruption of biological path-ways contributing to neurodegeneration. In the currentstudy, we used redox proteomics to compare the infe-rior parietal lobule (IPL) proteome from groups ofsubjects with equivalent levels of AD-like pathologywho differed in their respective clinical symptoms. IPLsamples from MCI (AD pathology, behavior/memoryimpairment) subjects were compared to IPL fromPCAD (AD pathology, no behavior/memory impair-ment) and healthy controls (no AD pathology, nobehavior/memory impairment) to examine biochem-ical differences at the protein level which may shedlight into mechanisms underlying memory dysfunc-tion in AD.

METHODS

Participants

MCI, PCAD, and control brainsIPL samples were obtained from 6 patients with

MCI, 12 individuals with PCAD, and 12 age-matchedindividuals who were cognitively intact without post-mortem neuropathologic changes of AD (controls).The Rapid Autopsy Program at the University ofKentucky Alzheimer’s Disease Research Center (UKADRC) obtained samples with extremely short post-mortem intervals (PMI) (Table 1). A short PMI isadvantageous in proteomics studies using human tissuesince PMI related artifacts are minimized and resultsmore likely reflect the intrinsic situation in PCAD andMCI brain.

All participants underwent annual mental status test-ing and annual physical and neurological exams, asa part of the UK ADC normal volunteer longitudinalaging study, and did not have a history of dementiaor other neurologic disorders. All control and PCADparticipants had test scores in the normal range. Neu-

Table 1Patient characteristics

Group Age Gender Brain weight Postmortem interval Braak MMSE(M/F) (g) (h)

Control 85.67 (1.88) 4M/8F 1218 (44) 2.99 (0.22) I 29PCAD 84.33 (1.38) 3M/9F 1201 (36) 2.76 (0.18) IV* 29MCI 89.00 (2.32) 2M/4F 1101 (40) 3.00 (0.46) IV* 23**

Average values ± SEM for age, brain weight, and post mortem interval; Median values for Braak andMMSE; *p < 0.001 versus control; **p < 0.005 versus PCAD and control.

C.D. Aluise et al. / Comparative Redox Proteomics in PCAD and MCI 259

ropathological evaluation of the brain samples usedas controls revealed only age associated gross andhistopathological alterations. Hematoxylin-eosin andmodified-Bielschowsky staining, A�antibody (10D5),and alpha-synuclein immuno-histochemistry wereused on multiple neocortical, hippocampal, entorhi-nal, amygdala, brainstem, and cerebellum sections fordiagnosis. All procedures to obtain human brain sam-ples were approved by the University of KentuckyInstitutional Review Board and experiments wereapproved by the Executive Committee of the Univer-sity of Kentucky Alzheimer’s Disease Clinical Center.

Treatment of samples

Brain samples for oxidative stress analyses andproteomics were homogenized and suspended inMedia I buffer containing protease inhibitors: leu-peptin (0.5 mg/mL), pepstatin (0.7 �g/mL), and apro-tinin (0.5 �g/mL). Homogenates were centrifuged at2000× g for 5 min to remove debris. Protein concentra-tion was determined by the bicinchoninic acid (BCA)protein assay (Pierce, Rockford, IL, USA).

Protein carbonyl assay

Samples (5 �l), 12% SDS (5 �l), and 10 �l of10 times-diluted 2,4-dinitrophenylhydrazine (DNPH)from a 200 mM stock solution were incubated atroom temperature for 20 min, followed by neutral-ization with 7.5 �l neutralization solution (2 M Trisin 30% glycerol). This neutralized solution (250 ngprotein) was loaded in each well on a nitrocellulosemembrane under vacuum using a slot–blot apparatus.The membrane was blocked in blocking buffer (3%bovine serum albumin) in PBS 0.01% (w/v) sodiumazide and 0.2% (v/v) Tween 20 for 3 h and incubatedwith a 1 : 100 dilution of anti-DNP polyclonal anti-body in PBS containing 0.01% (w/v) sodium azideand 0.2% (v/v) Tween 20 for 2 h. The membranewas washed three times in PBS following primaryantibody incubation at intervals of 5 min each. Themembrane was then incubated following washing withan anti-rabbit IgG alkaline phosphatase-linked sec-ondary antibody diluted in PBS in a 1 : 8000 ratio for1 h. The membrane was washed three times in PBS for5 min and developed in Sigmafast tablets [5-bromo-4-chloro-3-indolyl phosphate/Nitro blue tetrazoliumsubstrate (BCIP/NBT substrate)]. Blots were dried,scanned with Adobe Photoshop, and quantified withScion Image (PC version of Macintosh compatible

NIH image). Controls included samples with NaBH4prior to adding DNPH, in which case no immunochem-ical signal was observed (data not shown).

Two-dimensional gel electrophoresis

Protein samples (250 �g) were derivatized with 4times volume of 10 mM DNPH for western blots or 2NHCl for 2D gels and precipitated by adding ice-cold100% trichloroacetic acid (TCA) to a final concen-tration of 15% for 10 min on ice. Precipitates werecentrifuged for 2 min at 14,000× g at 4◦C. The pelletwas retained and washed three times with 1 ml of 1 : 1(v/v) ethyl acetate/ethanol. The final pellet was dis-solved in rehydration buffer (8 M urea, 2 M thiourea,2% CHAPS, 0.2% (v/v) biolytes, 50 mM dithiothreitol(DTT), and bromophenol blue). Samples were soni-cated in 15 s intervals three times each.

Two-dimensional polyacrylamide gel electrophore-sis was performed with a Bio-Rad IEF Cell systemusing 110-mm pH 3–10 immobilized pH gradients(IPG) strips and Criterion 8–16% resolving gels. IPGstrips were actively rehydrated at 50 V 20◦C followedby isoelectric focusing: 300 V for 2 h linear gradient,1200 V for 4 h slow gradient, 8000 V for 8 h lineargradient, and 8000 for 10 h rapid gradient. Gel stripswere equilibrated for 10 min prior to second-dimensionseparation in solution A [0.375 M Tris–HCl (pH 8.8),6 M urea (Bio-Rad, Hercules, CA), 2% (w/v) sodiumdodecyl sulfate (SDS), 20% (v/v) glycerol, and 0.5%dithiothreitol (DTT, Bio-Rad, Hercules, CA)], and thenre-equilibrated for 10 min in solution A containing4.5% iodoacetamide (IA) instead of DTT. ControlPCAD, and MCI strips were placed on the 8–16%Criterion gels, unstained molecular standards wereapplied, and electrophoresis was performed at 200 Vfor 65 min.

SYPRO Ruby stainingGels were fixed in a solution containing 10% (v/v)

methanol, 7% (v/v) acetic acid for 40 min and stainedovernight at room temperature with agitation in 50 mlof SYPRO Ruby gel stain (Bio-Rad, Hercules, CA,USA). Gels were then destained with 50 ml deionizedwater overnight.

Western blottingFor 2D western blotting, gels of derivatized sam-

ples were transferred to a nitrocellulose membrane90 mA/gel for 2 h and blocked for 2 h with 3%bovine serum albumin (BSA) in wash blot. Blotswere incubated with primary rabbit anti-DNP anti-


body (1 : 200) for 2 h. Membranes were washed threetimes in wash blot and incubated for 1 h with ananti-rabbit IgG alkaline phosphatase secondary anti-body diluted in wash blot in a 1 : 8000 ratio. Themembrane was washed three times in wash blot for5 min and developed using SigmaFast tablets (5-bromo-4-chloro-3-indoylphosphate/nitroblue tetrazo-lium, BCIP/NBT substrate). Membranes were driedovernight and scanned for PD-Quest analysis.

Analysis of 2D imagesImages from SYPRO Ruby stained gels, used to

measure protein content, were obtained using a phos-phorimager (excitation = 470 nm, emission = 618 nm,Molecular Dynamics, Sunnyvale, CA, USA). PDQuest2-D Analysis Software (Bio-Rad) was used to matchand analyze visualized protein spots among differentialgels and blots. Spot intensities were normalized rela-tive to total density of spots in corresponding gel or blotto correct for slight differences in staining and/or load-ing. Specific carbonylation of proteins was assessedby dividing the normalized blot intensity of spot bythe normalized gel intensity of the corresponding spotin each sample. After completion of spot matching,the densitometric and specific carbonylation valuesfrom individual 2D images were compared betweengroups using statistical analysis. Method reproducibil-ity was assessed by performing 4 gels from a sampleand completing spot matching to estimate analyticalvariance. After completion of spot matching from thissame-sample set, an average %CV value of 12% wasobtained. All reported proteins in this paper had foldchanges much greater than this value to eliminate thepossibility that methodological variance is responsiblefor observed protein changes.

In-gel digestionSamples were prepared according to the method

described by Thongboonkerd et al. [18]. Briefly, theprotein spots were cut and removed from the gelwith a clean razor blade. The gel pieces were placedinto individual, clean 1.5 ml microcentrifuge tubesand kept overnight at −20◦C. The gel pieces werethawed and washed with 0.1 M ammonium bicarbonate(NH4HCO3) (Sigma) for 15 min at room temperature.Acetonitrile (Sigma) was added to the gel pieces andincubated for an additional 15 min. The liquid wasremoved and the gel pieces were allowed to dry. Thegel pieces were rehydrated with 20 mM DTT (Bio-Rad, Hercules, CA, USA) in 0.1 M NH4HCO3 (Sigma)and incubated for 45 min at 56◦C. The DTT wasremoved and replaced with 55 mM IA (Bio-Rad) in0.1 M NH4HCO3 for 30 min in the dark at room tem-

perature. The liquid was drawn off and the gel pieceswere incubated with 50 mM NH4HCO3 at room tem-perature for 15 min. Acetonitrile was added to the gelpieces for 15 min at room temperature. All solventswere removed and the gel pieces were allowed to dryfor 30 min. The gel pieces were rehydrated with addi-tion of a minimal volume of 20 ng/�l modified trypsin(Promega, Madison, WI, USA) in 50 mM NH4HCO3.The gel pieces were chopped and incubated with shak-ing overnight (∼18 h) at 37◦C.

Mass spectrometryTryptic peptides were analyzed with an automated

nanospray Nanomate (Advion Biosystems) OrbitrapXL MS (ThermoFinnigan) platform. The Orbitrap MSwas operated in a data-dependent mode whereby the 8most intense parent ions measured in the FT at 60,000resolution were selected for ion trap fragmentationwith the following conditions: injection time 50 ms,35% collision energy, MS/MS spectra were measuredin the FT at 7500 resolution, and dynamic exclusionwas set for 120 seconds. Each sample was acquiredfor a total of ∼2.5 min. MS/MS spectra were searchedagainst the International Protein Index (IPI). HumanDatabase using SEQUEST and search results were fil-tered with the following criteria: Xcorr > 1.5, 2.0, 2.5,3.0 for +1, +2, +3, and +4 charge states, respectively,�Cn > 0.1, and p-value (protein and peptide) <0.01. IPIaccession numbers were cross-correlated with Swis-sProt accession numbers for final protein identification.

Statistical analysisFor patient data and slot blot data, one-way ANOVA

followed by Newman-Keuls post hoc test was usedto assess statistical significance. For proteomics data,Mann-Whitney U test was used for significance. pvalues less than 0.05 were considered statisticallysignificant.

RESULTS

Global oxidative stress in MCI IPL

Oxidative modifications to proteins have been impli-cated in the pathology and progression of AD [17,19–27]. Several reports from our laboratory have alsoshown elevations in oxidative damage in several brainregions in MCI, including (but not limited to) theIPL. In the current study, we observed consistent find-ings with previous reports showing increased oxidativedamage in MCI IPL as assessed by protein carbonyls(p < 0.01, Fig. 1). Furthermore, a significant increasein protein carbonyls was observed in MCI relative


Fig. 1. Oxidative stress marker protein carbonyls in IPL in control,PCAD, and MCI patients. No significant difference was observedbetween PCAD and control. A significant increase in the level ofprotein carbonyls was observed in MCI relative to both groups(*p < 0.01).

to PCAD (p < 0.01), despite equal levels of AD neu-ropathology. Consistent with our previous report andothers examining levels of protein carbonyls in PCADrelative to controls [13, 28] we observed no significantdifferences between these two groups (Fig. 1).

Proteomics determined differences in proteinlevels distinguishing PCAD from MCI

Two dimensional gel-based proteomics has beenused extensively to examine differences in pro-tein levels in AD brain, MCI brain, and transgenicmouse models of AD [17]. Oftentimes, proteomics-determined alterations in levels of proteins areindicative of up-regulation or down-regulation of path-ways involved in disease pathogenesis or responseto a given treatment. Therefore, we performed gel-based proteomics (Fig. 2) to investigate differences

Fig. 2. 2D gel map of proteins displaying differential levels amongthe three groups. Protein spots were identified by MS/MS and labeledwith their corresponding identifications. Data on extent of foldchange and MS identification data is available in Tables 2a and b.

in protein expression in conditions with equal patho-logical levels (PCAD and MCI) with differing clinicalsymptoms. Interestingly, 6 proteins were observed to

Table 2aSummary of differentially expressed proteins. Identification data of proteins exhibiting differ-

ences in levels between control, PCAD, and MCI IPL as determined by MS/MS

Protein name MW/pI@ SwissProt accession Peptide hits$ p of id#

EfTu 50/7.36 P49411 5 (5) 4.00E-07UCHL1 25/5.22 P09936 4 (4) 8.00E-06PK M1 58/7.53 P14618-2 10 (10) 7.00E-08PEBP1 21/7.38 P30086 7 (6) 8.00E-10Dihydro DEH 54/7.50 P09622 2 (2) 4.00E-08ATP synthase 60/9.46 P25705 7 (7) 3.00E-08

Albumin 72/6.40 ∧Q56G89; Q5D0D7 4 (4) 8.00E-08GAPDH 36/8.68 P04406 4 (3) 3.00E-05

Cu, Zn SOD 16/5.68 P09772 3 (1) 1.00E-04

$Peptide hits represents total number of peptides matches, while number in parentheses repre-sents number of unique sequences identified; @MW = molecular weight, in kDa; pI = isoelectricpoint; ∧TRMBL no; #refers to probability associated with SEQUEST score.


Table 2bDifferences in levels of proteins between control, PCAD, and MCI IPL as determined by 2-D gel electrophoresis

Protein name Control PCAD MCI p (PCAD p (MCI p (MCIvs control) vs control) vs PCAD)

EfTu 839.48 ± 66.29 818.73 ± 44.12 570.45 ± 39.80 0.665 0.0218* 0.0043*UCHL1 1553.84 ± 90.88 1689.03 ± 121.32 1016.73 ± 74.87 0.453 0.0057* 0.0043*PK M1 1037.04 ± 72.95 852.66 ± 67.69 570.7 ± 33.88 0.0606 0.0076* 0.01*PEBP1 3046.79 ± 145.02 3041.19 ± 127.02 1958.27 ± 303.31 0.885 0.0131* 0.0131*Dihydro DEH 1574.97 ± 77.69 1519.02 ± 83.70 1026.85 ± 189.46 0.7508 0.0169* 0.0218*ATP synthase 3477.62 ± 156.68 3416.02 ± 296.66 2336.55 ± 292.24 0.977 0.01* 0.0351*

Albumin 2936.45 ± 559.09 2404.59 ± 287.41 6336.52 ± 741.69 0.624 0.0076* 0.0012*GAPDH 1276.56 ± 65.74 1309.14 ± 49.56 1694.47 ± 112.13 0.977 0.0169* 0.0169*

Cu, Zn SOD 1192.7 ± 213.67 1863.53 ± 71.24 1246 ± 274.24 0.0036* 0.743 0.0218*

Values for Control, PCAD, and MCI represent group average of normalized densitometric spot volume ± SEM; *p < 0.05.

Table 2cGeneral trends of proteins observed as having differential levels control, PCAD, and MCI IPL

have decreased levels in MCI relative to both PCADand control IPL, while 2 proteins showed signifi-cant increases in MCI IPL relative to both PCADand controls. One protein (Cu, Zn superoxide dismu-tase (Cu, Zn SOD)), displayed a significant increasein PCAD relative to both control and MCI IPL.

Proteins whose levels were unchanged in PCAD ver-sus control, but show a significant increase in MCIrelative to both of these groups were identified asalbumin and glyceraldehyde-3-phosphate dehydroge-nase (GAPDH) (Tables 2a–c, Fig. 3). Conversely,proteins that showed no significant differences in


Fig. 3. Enhanced 2D images clearly showing differences in protein levels reported in Table 2.

levels between PCAD and control, but show a sig-nificant decrease in MCI IPL relative to both groupswere identified as ubiquitin carboxy-terminal hydro-lase 1 (UCHL1), dihydrolipoyl dehydrogenase (DHD),ATP synthase, elongation factor Tu (EfTu), phos-phatidylethanolamine binding protein 1 (PEBP1), andpyruvate kinase (PK) (Tables 2a–c, Fig. 3). Severalof these proteins have been implicated in AD-relatedcell death and memory impairment in previous inde-pendent studies. The alterations observed here mayprovide insight into the connection between pathol-ogy and biochemical events/pathways as they relate tomemory.

Redox proteomics analysis of MCI IPL comparedto PCAD for carbonylated proteins

Reactive oxygen species (ROS) or reactive nitro-gen species (RNS) are capable of causing oxidativemodifications to proteins, lipids, nucleic acids, andcarbohydrates. Identification of specific targets of oxi-dation can shed light on dysfunctional biochemicalprocesses/pathways in a disease state, as increasedoxidation or nitration is typically met with compro-mised function. Therefore, we employed 2-D redoxproteomics in order to identify which proteins in MCIbrain undergo specific carbonylation relative to PCAD.

As with any redox proteomics experiment investigatingmodified proteins, there exists the possibility that someexcessively oxidized proteins can become cross-linkedand, therefore, unable to be solubilized under SDS-PAGE conditions [29]. Therefore, observed changesin relative carbonyl content of the reported proteinsare restricted to those proteins whose extent of car-bonylation is measurable given the current solubilityconditions.

We observed significant increases in the specific car-bonyl levels of alpha enolase and heat shock protein90 (HSP90) in MCI IPL relative to PCAD (p < 0.05)(Tables 3a–c, Fig. 4). Furthermore, HSP90 was alsoshown to be increasingly oxidized in MCI relative tocontrol (p < 0.05) (Fig. 4). No significant differences inlocal carbonylation were observed between PCAD andcontrol groups, further supporting our finding here (andelsewhere) that oxidative stress is absent from PCADbrain, possibly corresponding to cellular preservationand protection from memory impairment.

DISCUSSION

Pathologically, AD is indexed by SPs and NFTs.SPs are composed primarily of A�, a peptide thoughtto be a contributing factor to neurodegeneration; thisis controversial, as a small percentage of cognitively


Table 3aSummary of oxidized proteins. Identification data of proteins showing differences in specific carbonylation between

control, PCAD, and MCI IPL as determined by MS/MS

Protein name MW/pI@ SwissProt accession Peptide hits$ p of id#

Alpha enolase 47.10/7.16 P06733-1 4 (4) 6.00E-09HSP90 98.10/4.93 P07900-2 13 (3) 1.00E-06

$Peptide hits represents total number of peptides matches, while number in parentheses represents number of uniquesequences identified; @MW=molecular weight, in kDa; pI = isoelectric point; #refers to probability associated withSEQUEST score.

Table 3bDifferences in specific carbonylation of proteins between control, PCAD, and MCI as determined by 2-dimensional

redox proteomics

Protein name C ave P ave M ave p (PCAD p (MCI p (MCIvs control) vs control) vs PCAD)

Alpha enolase 1.80 ± 0.36 1.11 ± 0.15 1.80 ± 0.23 0.19 0.98 0.031*HSP90 3.29 ± 1.00 2.63 ± 0.79 6.63 ± 1.07 0.834 0.0277* 0.01*

$Densitometric spot ratio average (specific carbonyl, blot/gel); *p < 0.05.

Table 3cGeneral trends of proteins observed as exhibiting differences in specific oxidation between control, PCAD, and MCI

intact individuals at autopsy have significant levels ofplaque and tangle pathology that would otherwise war-rant diagnosis of AD, or at least, MCI. A� has beenshown to cause oxidative stress, both in AD and MCIbrain, and in mouse models of AD [17, 19, 20]. How-

ever, in our previous report on PCAD, we observedno significant differences in global oxidative stressrelative to controls, possibly due to a lack of oligomer-ization of A� [13]. Others also observed no differencesin protein carbonyls in PCAD hippocampus, superior


Fig. 4. 2D gels and oxyblots of oxidatively modified proteins in MCI relative to control and PCAD.

and medial temporal gyrus, and cerebellum [28]. Fur-thermore, in the current study, we confirm this resultin the IPL, and also, observed no significant differ-ences in local carbonylation of proteins in PCADcompared to control (as assessed by 2D redox pro-teomics). Because oxidative imbalance in a cell canlead to adverse protein and lipid modifications culmi-nating in apoptosis, an absence of significant oxidativestress in PCAD may be central to the lack of cell loss[14], as well as the neuronal hypertrophy [11] andincreased synaptic plasticity [15] observed in previousreports of PCAD.

In this study, we chose to compare PCAD to MCIinstead of AD due to the fact that PCAD patients typ-ically displayed MCI-like levels of pathology, with noobservable memory impairment (Table 1). Therefore,the proteomics-determined differences in protein lev-els and oxidative modifications to proteins betweentwo groups that display equal levels of AD pathol-ogy, but differ in clinical manifestations of memoryimpairment, provide valuable insight into the connec-tion between pathology and symptoms. Proteomics hasbeen used extensively by our lab to identify differencesin protein levels and oxidative modifications to proteinsin AD and MCI brain as well as mouse models of AD[17]. Because oxidative modification to proteins typi-cally results in lowered functionality, identification ofmodified proteins may be indicative of dysfunctionalcellular processes in AD. Also, differential levels of

proteins in a diseased versus control comparison oftenshed insight into pathways or mechanisms that are up-or downregulated, possibly contributing to or actingagainst an inherent insult.

Initial studies on brains of AD subjects observedincreased protein carbonylation in AD subjects relativeto controls [30]. Our group showed that brain proteinoxidation in AD occurred in regions rich in A�, butnot in A�-poor cerebellum [31]. Other reports havealso observed increased carbonyls in AD brain [31],as well as increases in the specific carbonyl contentof several proteins [24, 32, 33]. This modification hasbeen suggested to underlie the reduced activity of sev-eral proteins whose function is diminished in AD [34].The appearance of increased protein carbonylation inAD is evident in brain areas such as hippocampus[24, 30, 35, 36], and IPL [32, 33], both of whichare memory-associated regions. Furthermore, in brainregions not associated with memory/cognition, such ascerebellum, protein carbonylation is largely absent [31,35]. Therefore, increased levels of protein carbonyla-tion leading to decreased enzymatic activity, loss offunctional processes and, eventually, neuronal death,have been proposed as a possible driving force of ADpathogenesis [30].

Between PCAD and MCI, we observe a signifi-cant increase in the level of protein carbonyls in MCI,indicative of significantly elevated oxidative stress inMCI despite equal levels of pathology. We also found


significantly higher protein carbonyl levels in MCIrelative to control, confirming previous reports onincreased protein oxidation in MCI [20, 21, 37]. Con-sistent with this observation, we have shown alteredlevels and activities of antioxidant enzymes in brainsof AD/MCI patients relative to healthy controls [26].In the present study, we observed differences in thelevels of Cu, Zn SOD and DHD between the threegroups using 2D proteomics. Cu, Zn SOD showed asignificant increase in PCAD relative to control andMCI, while DHD was significantly lower in MCI rel-ative to PCAD and control. Cu, Zn SOD is a cytosolicantioxidant that catalyzes the dismutation of super-oxide to hydrogen peroxide; this protein has beenpreviously observed to highly correlate with mem-ory and increased black/white reversal learning in acanine study of aging [38]. Other groups have observeddecreased DHD in AD brain relative to healthy con-trols [39]; here we replicate this finding in MCI IPL,but also observe that the level of this protein is signif-icantly suppressed in MCI relative to PCAD as well,while no difference was observed between PCAD andcontrol. DHD is the E3 subunit of the enzyme pyruvatedehydrogenase, and, more specifically, is involved inthe oxidation-reduction of lipoic acid for the (eventual)production of NADH. A deficiency in DHD, a criticalsubunit of pyruvate dehydrogenase, may be responsi-ble for the decreased activity of this complex observedin brains of AD patients in another study [40].

Glucose metabolizing enzymes have been reportedto be differentially expressed in many previous studiesof AD. The observation of aberrant levels of proteinsinvolved in glycolysis/TCA cycle/electron transport isthought to cause or contribute to the observed decreasein glucose metabolism as evidenced by PET scanningin brains of AD patients [41]. Because depletion of cel-lular ATP production via mitochondria can lead to celldeath, a key event in AD pathogenesis, it is thought thatthese events are interrelated. Here, we observe a signif-icant increase in GAPDH in MCI relative to PCAD andcontrols, and a significant decrease in pyruvate kinaseand ATP synthase in MCI relative to PCAD and con-trols. ATP synthase has been observed to be decreasedin AD hippocampus [42], and beta subunit mRNA forthis protein has also been reported to be decreased[43]. PK converts phosphoenolpyruvate to pyruvateand ATP in the final step of glycolysis; lower levelsof this enzyme may reduce cellular ATP levels crit-ical for maintaining neuronal survival. Additionally,we observed a significant increase in the carbonylationstatus of ENO1 in MCI compared to PCAD, despitean equal level of AD pathology. Oxidative damage to

glycolytic enzymes has been previously observed tolower function, as observed previously [44]. In additionto adversely affecting its glycolytic activity, oxidativemodification to enolase and the subsequent functionalalteration may also affect its multitude of other func-tions, as enolase is also a neurotrophic factor, 14-3-2[45–47], a hypoxic stress protein [48], c-Myc bind-ing protein and transcription factor [49, 50], and astrong plasminogen binding protein [51, 52]. Further-more, many of these non-glycolytic functions havebeen implicated in AD [53, 54].

Cells utilize proteasomes to break down aber-rantly folded or damaged proteins. Proteins that areincorrectly folded can trigger a heat shock response,leading to the induction of intracellular heat shock pro-teins (HSP). Therefore, dysfunction in the heat shockresponse/ubiquitin/proteasomal pathways can lead tointracellular buildup of damaged biomolecules that canlead to the death of the cell. We observed a signif-icant increase in the specific carbonylation of HSP90 in MCI relative to PCAD and control. Further-more, previous reports have shown a downregulationof UCHL1 in AD brain compared to controls [55];here, we observe lower levels of this protein in MCIrelative to both control and PCAD. Based on this data,we suggest an imbalance in the proteasomal networkin MCI (that is absent in PCAD), leading to a buildupof aberrantly folded proteins, possibly contributing tomemory loss and neuropathological changes in MCI.Further, HSP90 is critical to suppressing inflamma-tion through degradation of hypoxic-inducing factor 1alpha [56]; oxidation of this protein may contribute tothe rampant brain-resident inflammation in AD/MCI.

Other proteins whose levels were found to differin MCI relative to PCAD and control were albumin,PEBP1, and EF-Tu. Albumin in brain/cerebrospinalfluid is indicative of blood brain barrier (BBB) dys-function [57], and an increase in the level of this proteinin MCI relative to PCAD and control may indicateBBB breakdown as the AD phenotype develops. Ef-tu has been previously shown to be HNE-bound inMCI brain by our laboratory [23]; here we observeddecreased levels of this protein in MCI, possibly indi-cating a decrease in tRNA transport from the cytosol tothe ribosome for cytosolic protein production. Lastly,PEBP, or neuropolypeptide h3, has been implicated inAD [58, 59], due to its involvement in the production ofcholine acetyltransferase, a critical enzyme in acetyl-choline synthesis; decreased levels of this protein inMCI IPL relative to PCAD and controls potentiallysignal decreased acetylcholine production as has beenpreviously reported in AD. In fact, one of the most


commonly prescribed classes of drugs to treat AD onthe market today is the acetylcholinesterase inhibitors,implicating decreased acetylcholine levels as a majorpathogenic event in AD. A lack of significant alter-ation of PEBP1 in PCAD versus control IPL may inferthat levels of this enzyme are sufficient to maintainacetylcholine levels to preserve memory.

In closing, the data presented here are supportive ofthe notion that in PCAD there is a relatively protectiveperiod against AD-related neurodegeneration, whichas oxidative stress increases in MCI, is lost. The factthat some adverse changes in protein modifications andprotective protein levels observed here correlate withclinical manifestations of AD supports previous stud-ies that posit that these changes are critical to memoryimpairment. The observed changes in oxidative stressand protein levels in MCI relative to PCAD in this studymay help explain previous studies that observed signif-icant neuron loss in very mild dementia of the AD typethat was absent in PCAD [14, 60, 61]. A large caveatto working with PCAD is whether or not these indi-viduals would have developed the AD phenotype hadthey lived long enough; nevertheless, our data furthersupport the concept of PCAD, and highlight consid-erable differences at the protein level between PCADand MCI, despite equal levels of pathology.

ACKNOWLEDGMENTS

This work was supported in part by NIH grant toDAB [AG-05119]. We are grateful to the Clinical andNeuropathology Cores of the University of KentuckyAlzheimer’s Disease Clinical Center for providingwell-characterized tissue samples from volunteers inthe longitudinal normal aging study.

Authors’ disclosures available online (http://www.j-alz.com/disclosures/view.php?id=626).

REFERENCES

[1] Petersen RC (2004) Mild cognitive impairment as a diagnosticentity. J Intern Med 256, 183-194.

[2] Haass C, Schlossmacher MG, Hung AY, Vigo-Pelfrey C,Mellon A, Ostaszewski BL, Lieberburg I, Koo EH, SchenkD, Teplow DB, et al. (1992) Amyloid beta-peptide is pro-duced by cultured cells during normal metabolism. Nature359, 322-325.

[3] Seubert P, Vigo-Pelfrey C, Esch F, Lee M, Dovey H, Davis D,Sinha S, Schlossmacher M, Whaley J, Swindlehurst C, et al.(1992) Isolation and quantification of soluble Alzheimer’sbeta-peptide from biological fluids. Nature 359, 325-327.

[4] Shoji M, Golde TE, Ghiso J, Cheung TT, Estus S, Shaffer LM,Cai XD, McKay DM, Tintner R, Frangione B, et al. (1992)

Production of the Alzheimer amyloid beta protein by normalproteolytic processing. Science 258, 126-129.

[5] Geula C, Wu CK, Saroff D, Lorenzo A, Yuan M, YanknerBA (1998) Aging renders the brain vulnerable to amyloidbeta-protein neurotoxicity. Nat Med 4, 827-831.

[6] Li S, Hong S, Shepardson NE, Walsh DM, Shankar GM,Selkoe D (2009) Soluble oligomers of amyloid beta proteinfacilitate hippocampal long-term depression by disruptingneuronal glutamate uptake. Neuron 62, 788-801.

[7] Pike CJ, Walencewicz AJ, Glabe CG, Cotman CW (1991)In vitro aging of beta-amyloid protein causes peptide aggre-gation and neurotoxicity. Brain Res 563, 311-314.

[8] Shankar GM, Bloodgood BL, Townsend M, Walsh DM,Selkoe DJ, Sabatini BL (2007) Natural oligomers ofthe Alzheimer amyloid-beta protein induce reversiblesynapse loss by modulating an NMDA-type glutamatereceptor-dependent signaling pathway. J Neurosci 27, 2866-2875.

[9] Shankar GM, Li S, Mehta TH, Garcia-Munoz A, ShepardsonNE, Smith I, Brett FM, Farrell MA, Rowan MJ, LemereCA, Regan CM, Walsh DM, Sabatini BL, Selkoe DJ(2008) Amyloid-beta protein dimers isolated directly fromAlzheimer’s brains impair synaptic plasticity and memory.Nat Med 14, 837-842.

[10] Price JL, McKeel DW Jr, Buckles VD, Roe CM, Xiong C,Grundman M, Hansen LA, Petersen RC, Parisi JE, DicksonDW, Smith CD, Davis DG, Schmitt FA, Markesbery WR,Kaye J, Kurlan R, Hulette C, Kurland BF, Higdon R, KukullW, Morris JC (2009) Neuropathology of nondemented aging:presumptive evidence for preclinical Alzheimer disease. Neu-robiol Aging 30, 1026-1036.

[11] Iacono D, Markesbery WR, Gross M, Pletnikova O, RudowG, Zandi P, Troncoso JC (2009) The Nun study: clinicallysilent AD, neuronal hypertrophy, and linguistic skills in earlylife. Neurology 73, 665-673.

[12] Goldman WP, Price JL, Storandt M, Grant EA, McKeel DW Jr,Rubin EH, Morris JC (2001) Absence of cognitive impairmentor decline in preclinical Alzheimer’s disease. Neurology 56,361-367.

[13] Aluise CD, Robinson RA, Beckett TL, Murphy MP, Cai J,Pierce WM, Markesbery WR, Butterfield DA (2010) Preclin-ical Alzheimer disease: brain oxidative stress, Abeta peptideand proteomics. Neurobiol Dis 39, 221-228.

[14] West MJ, Kawas CH, Stewart WF, Rudow GL, TroncosoJC (2004) Hippocampal neurons in pre-clinical Alzheimer’sdisease. Neurobiol Aging 25, 1205-1212.

[15] O’Brien RJ, Resnick SM, Zonderman AB, Ferrucci L, CrainBJ, Pletnikova O, Rudow G, Iacono D, Riudavets MA,Driscoll I, Price DL, Martin LJ, Troncoso JC (2009) Neu-ropathologic studies of the Baltimore Longitudinal Study ofAging (BLSA). J Alzheimers Dis 18, 665-675.

[16] Lyubartseva G, Smith JL, Markesbery WR, Lovell MA(2010) Alterations of zinc transporter proteins ZnT-1, ZnT-4and ZnT-6 in preclinical Alzheimer’s disease brain. BrainPathol 20, 343-350.

[17] Sultana R, Perluigi M, Butterfield DA (2009) Oxidativelymodified proteins in Alzheimer’s disease (AD), mild cogni-tive impairment and animal models of AD: role of Abeta inpathogenesis. Acta Neuropathol 118, 131-150.

[18] Thongboonkerd V, Luengpailin J, Cao J, Pierce WM, Cai J,Klein JB, Doyle RJ (2002) Fluoride exposure attenuatesexpression of Streptococcus pyogenes virulence factors. J BiolChem 277, 16599-16605.

[19] Butterfield DA, Drake J, Pocernich C, Castegna A (2001)Evidence of oxidative damage in Alzheimer’s disease brain:

http://www.j-alz.com/disclosures/view.php?id=626

http://www.j-alz.com/disclosures/view.php?id=626


central role for amyloid beta-peptide. Trends Mol Med 7, 548-554.

[20] Butterfield DA, Poon HF, St Clair D, Keller JN, Pierce WM,Klein JB, Markesbery WR (2006) Redox proteomics iden-tification of oxidatively modified hippocampal proteins inmild cognitive impairment: insights into the development ofAlzheimer’s disease. Neurobiol Dis 22, 223-232.

[21] Lovell MA, Markesbery WR (2007) Oxidative damage inmild cognitive impairment and early Alzheimer’s disease.J Neurosci Res 85, 3036-3040.

[22] Markesbery WR, Kryscio RJ, Lovell MA, Morrow JD (2005)Lipid peroxidation is an early event in the brain in amnesticmild cognitive impairment. Ann Neurol 58, 730-735.

[23] Reed T, Perluigi M, Sultana R, Pierce WM, Klein JB,Turner DM, Coccia R, Markesbery WR, Butterfield DA(2008) Redox proteomic identification of 4-hydroxy-2-nonenal-modified brain proteins in amnestic mild cognitiveimpairment: insight into the role of lipid peroxidation in theprogression and pathogenesis of Alzheimer’s disease. Neuro-biol Dis 30, 107-120.

[24] Sultana R, Boyd-Kimball D, Poon HF, Cai J, Pierce WM,Klein JB, Markesbery WR, Zhou XZ, Lu KP, Butterfield DA(2006) Oxidative modification and down-regulation of Pin1 inAlzheimer’s disease hippocampus: a redox proteomics anal-ysis. Neurobiol Aging 27, 918-925.

[25] Sultana R, Perluigi M, Newman SF, Pierce WM, Cini C,Coccia R, Butterfield A (2010) Redox proteomic analysis ofcarbonylated brain proteins in mild cognitive impairment andearly Alzheimer’s disease. Antioxid Redox Signal 12, 327-336.

[26] Sultana R, Piroddi M, Galli F, Butterfield DA (2008) Proteinlevels and activity of some antioxidant enzymes in hippocam-pus of subjects with amnestic mild cognitive impairment.Neurochem Res 33, 2540-2546.

[27] Smith MA, Perry G (1995) Free radical damage, iron, andAlzheimer’s disease. J Neurol Sci 134(Suppl), 92-94.

[28] Bradley MA, Markesbery WR, Lovell MA (2010) Increasedlevels of 4-hydroxynonenal and acrolein in the brain in pre-clinical Alzheimer’s disease (PCAD). Free Radic Biol Med48, 1570-1576.

[29] Smith MA, Siedlak SL, Richey PL, Nagaraj RH, ElhammerA, Perry G (1996) Quantitative solubilization and analysisof insoluble paired helical filaments from Alzheimer disease.Brain Res 717, 99-108.

[30] Smith MA, Perry G, Richey PL, Sayre LM, Anderson VE,Beal MF, Kowall N (1996) Oxidative damage in Alzheimer’s.Nature 382, 120-121.

[31] Aksenov MY, Aksenova MV, Butterfield DA, Geddes JW,Markesbery WR (2001) Protein oxidation in the brain inAlzheimer’s disease. Neuroscience 103, 373-383.

[32] Castegna A, Aksenov M, Thongboonkerd V, Klein JB, PierceWM, Booze R, Markesbery WR, Butterfield DA (2002)Proteomic identification of oxidatively modified proteinsin Alzheimer’s disease brain. Part II: dihydropyrimidinase-related protein 2, alpha-enolase and heat shock cognate 71.J Neurochem 82, 1524-1532.

[33] Castegna A, Aksenov M, Aksenova M, Thongboonkerd V,Klein JB, Pierce WM, Booze R, Markesbery WR, ButterfieldDA (2002) Proteomic identification of oxidatively modifiedproteins in Alzheimer’s disease brain. Part I: creatine kinaseBB, glutamine synthase, and ubiquitin carboxy-terminalhydrolase L-1. Free Radic Biol Med 33, 562-571.

[34] Smith CD, Carney JM, Starke-Reed PE, Oliver CN, StadtmanER, Floyd RA, Markesbery WR (1991) Excess brain proteinoxidation and enzyme dysfunction in normal aging and in

Alzheimer disease. Proc Natl Acad Sci U S A 88, 10540-10543.

[35] Sultana R, Boyd-Kimball D, Poon HF, Cai J, Pierce WM,Klein JB, Merchant M, Markesbery WR, Butterfield DA(2006) Redox proteomics identification of oxidized proteinsin Alzheimer’s disease hippocampus and cerebellum: anapproach to understand pathological and biochemical alter-ations in AD. Neurobiol Aging 27, 1564-1576.

[36] Smith MA, Sayre LM, Anderson VE, Harris PL, Beal MF,Kowall N, Perry G (1998) Cytochemical demonstrationof oxidative damage in Alzheimer disease by immuno-chemical enhancement of the carbonyl reaction with2,4-dinitrophenylhydrazine. J Histochem Cytochem 46, 731-735.

[37] Butterfield DA, Reed T, Perluigi M, De Marco C, Coccia R,Cini C, Sultana R (2006) Elevated protein-bound levels ofthe lipid peroxidation product, 4-hydroxy-2-nonenal, in brainfrom persons with mild cognitive impairment. Neurosci Lett397, 170-173.

[38] Opii WO, Joshi G, Head E, Milgram NW, Muggenburg BA,Klein JB, Pierce WM, Cotman CW, Butterfield DA (2008)Proteomic identification of brain proteins in the canine modelof human aging following a long-term treatment with antiox-idants and a program of behavioral enrichment: relevance toAlzheimer’s disease. Neurobiol Aging 29, 51-70.

[39] Mastrogiacoma F, Lindsay JG, Bettendorff L, Rice J, Kish SJ(1996) Brain protein and alpha-ketoglutarate dehydrogenasecomplex activity in Alzheimer’s disease. Ann Neurol 39, 592-598.

[40] Sheu KF, Kim YT, Blass JP, Weksler ME (1985) An immuno-chemical study of the pyruvate dehydrogenase deficit inAlzheimer’s disease brain. Ann Neurol 17, 444-449.

[41] Nordberg A, Rinne JO, Kadir A, Langstrom B (2010) The useof PET in Alzheimer disease. Nat Rev Neurol 6, 78-87.

[42] Schagger H, Ohm TG (1995) Human diseases with defectsin oxidative phosphorylation. 2. F1F0 ATP-synthase defectsin Alzheimer disease revealed by blue native polyacrylamidegel electrophoresis. Eur J Biochem 227, 916-921.

[43] Chandrasekaran K, Hatanpaa K, Rapoport SI, Brady DR(1997) Decreased expression of nuclear and mitochondrialDNA-encoded genes of oxidative phosphorylation in associ-ation neocortex in Alzheimer disease. Brain Res Mol BrainRes 44, 99-104.

[44] Reed TT, Pierce WM, Markesbery WR, Butterfield DA (2009)Proteomic identification of HNE-bound proteins in earlyAlzheimer disease: insights into the role of lipid peroxidationin the progression of AD. Brain Res 1274, 66-76.

[45] Takei N, Kondo J, Nagaike K, Ohsawa K, Kato K, Kohsaka S(1991) Neuronal survival factor from bovine brain is identicalto neuron-specific enolase. J Neurochem 57, 1178-1184.

[46] Hattori T, Ohsawa K, Mizuno Y, Kato K, Kohsaka S (1994)Synthetic peptide corresponding to 30 amino acids of theC-terminal of neuron-specific enolase promotes survival ofneocortical neurons in culture. Biochem Biophys Res Commun202, 25-30.

[47] Hattori T, Takei N, Mizuno Y, Kato K, Kohsaka S (1995)Neurotrophic and neuroprotective effects of neuron-specificenolase on cultured neurons from embryonic rat brain. Neu-rosci Res 21, 191-198.

[48] Aaronson RM, Graven KK, Tucci M, McDonald RJ, FarberHW (1995) Non-neuronal enolase is an endothelial hypoxicstress protein. J Biol Chem 270, 27752-27757.

[49] Ray R, Miller DM (1991) Cloning and characterization ofa human c-myc promoter-binding protein. Mol Cell Biol 11,2154-2161.


[50] Subramanian A, Miller DM (2000) Structural analysis ofalpha-enolase. Mapping the functional domains involved indown-regulation of the c-myc protooncogene. J Biol Chem275, 5958-5965.

[51] Nakajima K, Hamanoue M, Takemoto N, Hattori T, Kato K,Kohsaka S (1994) Plasminogen binds specifically to alpha-enolase on rat neuronal plasma membrane. J Neurochem 63,2048-2057.

[52] Pancholi V, Fischetti VA (1998) Alpha-enolase, a novel strongplasmin(ogen) binding protein on the surface of pathogenicstreptococci. J Biol Chem 273, 14503-14515.

[53] Parnetti L, Palumbo B, Cardinali L, Loreti F, Chionne F,Cecchetti R, Senin U (1995) Cerebrospinal fluid neuron-specific enolase in Alzheimer’s disease and vasculardementia. Neurosci Lett 183, 43-45.

[54] Dotti CG, Galvan C, Ledesma MD (2004) Plasmin deficiencyin Alzheimer’s disease brains: causal or casual? NeurodegenerDis 1, 205-212.

[55] Choi J, Levey AI, Weintraub ST, Rees HD, Gearing M, ChinLS, Li L (2004) Oxidative modifications and down-regulationof ubiquitin carboxyl-terminal hydrolase L1 associated withidiopathic Parkinson’s and Alzheimer’s diseases. J Biol Chem279, 13256-13264.

[56] Zhang D, Li J, Costa M, Gao J, Huang C. JNK1 mediatesdegradation HIF-1alpha by a VHL-independent mechanism

that involves the chaperones Hsp90/Hsp70. Cancer Res 70,813-823.

[57] Blennow K, Wallin A, Fredman P, Karlsson I, Gottfries CG,Svennerholm L (1990) Blood-brain barrier disturbance inpatients with Alzheimer’s disease is related to vascular factors.Acta Neurol Scand 81, 323-326.

[58] George AJ, Holsinger RM, McLean CA, Tan SS, ScottHS, Cardamone T, Cappai R, Masters CL, Li QX (2006)Decreased phosphatidylethanolamine binding protein expres-sion correlates with Abeta accumulation in the Tg2576mouse model of Alzheimer’s disease. Neurobiol Aging 27,614-623.

[59] Maki M, Matsukawa N, Yuasa H, Otsuka Y, Yamamoto T,Akatsu H, Okamoto T, Ueda R, Ojika K (2002) Decreasedexpression of hippocampal cholinergic neurostimulatingpeptide precursor protein mRNA in the hippocampus inAlzheimer disease. J Neuropathol Exp Neurol 61, 176-185.

[60] Price JL, Ko AI, Wade MJ, Tsou SK, McKeel DW, Mor-ris JC (2001) Neuron number in the entorhinal cortex andCA1 in preclinical Alzheimer disease. Arch Neurol 58, 1395-1402.

[61] Gomez-Isla T, Price JL, McKeel DW Jr, Morris JC, GrowdonJH, Hyman BT (1996) Profound loss of layer II entorhi-nal cortex neurons occurs in very mild Alzheimer’s disease.J Neurosci 16, 4491-4500.

journal of alzheimer’s disease 23 (2011) 257–269 ios press

Documents