阿尔茨海默病,诊断标记物,脑脊液7

Journal of Alzheimer’s Disease31(2012)865–878DOI10.3233/JAD-2012-120211IOS Press865 The Load of Amyloid-␤Oligomersis Decreased in the Cerebrospinal Fluidof Alzheimer’s Disease PatientsGiulia M.Sancesario a,Maria T.Cencioni b,Zaira Esposito a,Giovanna Borsellino b,Marzia Nuccetelli a, Alessandro Martorana a,b,Luca Battistini b,Roberto Sorge a,Gianfranco Spalletta b,Davide Ferrazzoli a, Giorgio Bernardi a,b,Sergio Bernardini a and Giuseppe Sancesario a,b,∗a Tor Vergata General Hospital,Faculty of Medicine and Surgery,The University of Rome Tor Vergata,Rome,Italyb Santa Lucia Foundation,Rome,ItalyAccepted19May2012Abstract.Amyloid-␤(A␤)oligomers are heterogeneous and instable compounds of variable molecular weight.Flow cytometry andfluorescence resonance energy transfer(FRET)-based methods allow the simultaneous detection of A␤oligomers with low and high molecular weight in their native form.We evaluated whether an estimate of different species of A␤oligomers in the cerebrospinalfluid(CSF)with or without dilution with RIPA buffer could be more useful in the diagnosis of Alzheimer’s disease (AD)than the measurement of A␤42monomers,total tau(t-tau),and phosphorylated tau(p-tau).Increased t-tau(p<0.01)and p-tau(p<0.01),and decreased A␤42(p<0.01),were detected in the CSF of patients with AD(n=46),compared to patients with other dementia(OD)(n=35)or with other neurological disorders(OND)(n=56).In native CSF(n=137),the levels of A␤oligomers were lower(p<0.05)in AD than in OD and OND patients;in addition,the ratio A␤oligomers/p-tau was lower in AD than in OD(p<0.01)and OND(p<0.05)patients,yielding a sensitivity of75%and a specificity of64%.However, in CSF diluted with RIPA(n=30),A␤oligomers appeared higher(p<0.05)in AD than in OND patients,suggesting they become partially disaggregated and more easily detectable after RIPA.In conclusion,FRET analysis in native CSF is essential to correctly determine the composition of A␤oligomers.In this experimental setting,the simultaneous estimate of low and high molecular weight A␤oligomers is as useful as the other biomarkers in the diagnosis of AD.The low amount of A␤oligomers detected in native CSF of AD may be inversely related to their levels in the brain,as occurs for A␤monomers,representing a biomarker for the amyloid pathogenic cascade.Keywords:A␤42,Alzheimer’s disease,amyloid-␤,biomarkers,cerebrospinalfluid,flow cytometry,fluorescence resonance energy transfer,FRET,oligomersINTRODUCTIONThe amyloid-␤peptides1-40and1-42(A␤40,A␤42) are physiological derivatives of the amyloid-␤protein precursor,a normal constituent of neuronal and glial membranes.A␤oligomers are soluble aggregates of a ∗Correspondence to:Prof.Giuseppe Sancesario,Department of Systems Medicine,1Montpellier Street–00133Rome, Italy.Tel.:+39620903013;Fax:+39620903118;E-mail: sancesario@med.uniroma2.it.variable number of A␤monomers and constitute the precursors of insoluble largefibrils that deposit in amy-loid plaques in the brain,representing,together with neurofibrillary tangles,the histological hallmarks of Alzheimer’s disease(AD).A␤42,and to a minor extent A␤40,oligomers have been regarded in recent years not just as the intermediate to A␤fibrils,but as a pri-mary pathological factor,affecting synapses’function and neuronal viability,and secondarily rising an amy-loidogenic cascade to the large insoluble A␤fibrils,and the formation of amyloid plaques and neurofibrillaryISSN1387-2877/12/$27.50©2012–IOS Press and the authors.All rights reserved866G.M.Sancesario et al./AβOligomers in CSFtangles[1–3].A soluble pool of A␤40and A␤42aggre-gates together with the large insoluble A␤fibrils can be extracted by analytical ultracentrifugation of tissue homogenates from the cerebral cortex of AD affected brains,demonstrating that soluble oligomeric A␤species are intrinsic to the brain AD pathology[4–6]. The proposed role of A␤oligomers in the brain pathology has stimulated an extensive search for their presence in the cerebrospinalfluid(CSF),since,unlike the insolublefibrils deposited in brain tissue,solu-ble oligomers can diffuse into the CSF.If detected in sensitive assays,they would constitute,in essence,a useful biomarker for the amyloid pathogenic cascade. An increase of A␤oligomers in the CSF of AD patients would be expected as a corollary of their increased occurrence and neurotoxic role in the brain tissue. The detection of single A␤aggregates in the CSF wasfirst observed usingfluorescence correla-tion spectroscopy,a very sensitive biophysical method, demonstrating the occurrence of a faster A␤self assembly in the CSF of AD affected patients[7]. Although several methods have been recently devel-oped attempting to detect diffusible A␤oligomers in CSF,their detection in the CSF has not been an easy task to be pursued for diagnostic purposes in the setting of common biomedical research institutions[8].San-tos and co-workers[9]published a very sensitive and specific method for the detection of different species of A␤oligomers in the CSF of non-demented neurologi-cal patients,based on afluorescence resonance energy transfer(FRET)setup,and subsequent detection by flow cytometry.We used FRET andflow cytometric analysis and unexpectedly found that levels of A␤oligomers are lower in the CSF of probable AD than in control patients in contrast to the results recently reported by Santos et al.[10].Thesefindings raise the possibility that the detectability of A␤oligomers with the FRET cytometric analysis could be dependent on different pre-analytical and analytical factors.We sought to val-idate the quantitative reproducibility of storage and detection procedures of oligomers by FRET cytometric analysis,andfinally we evaluated whether an estimate of oligomer A␤species in their native form in the CSF could be more useful in the diagnosis of AD than A␤monomers and tau.PATIENTS AND METHODSOne hundred and thirty seven patients were included in this study after admission to the Neurological Centre of the Tor Vergata General Hospital between2009 and2011because of cognitive or motor impairments. Exclusion criteria were acute trauma of the head and of the spinal cord,chronic subdural hematoma, hydrocephalus,brain tumors and paraneoplastic neu-rological syndrome,acute and chronic infections of the CNS,vitamin deficiency,hypothyroidism,liver and kidney insufficiency,or rmed written consent was obtained from each patient or relatives for routine clinical and instrumental investigations. The study was conducted under the guidelines of Hip-pocrates deontology and following approval of the local Ethical Committee.Clinical evaluationMedical history,thorough clinical and neurolog-ical investigation,Mini-Mental State Examination (MMSE),complete blood screening,electrocardio-gram,electroencephalogram,and magnetic resonance imaging of the brain were performed on all patients.A more comprehensive neuropsychological and psy-chiatric examination was performed on patients with cognitive deficit,using the standardized Mental Dete-rioration Battery[11]and the Hamilton Rating scale [12].Moreover,CSF was obtained as part of routine clinical evaluation;A␤42,total tau(t-tau),and phos-phorylated tau(p-tau)were measured for differential diagnosis of dementia[13,14].Patients werefirst divided in two groups:patients with different degrees of cognitive impairment (demented,MMSE=18.2±6.1)and patients without cognitive impairment(non demented, MMSE=27.8±1.2).The non demented group (n=56;40women,16men)included patients with heterogeneous neurological diseases:arterial hypertension and chronic ischemic encephalopathy (n=9),Parkinson’s disease without cognitive impair-ment(n=13),cerebellar atrophy(n=3),familial tremor(n=3),late onset epilepsy(n=5),spondilo-genetic mielopathy(n=4),optic papillitis(n=1), Sjogren syndrome(n=1),peripheral neuropathies (n=7),pseudotumor cerebri(n=2),narcolepsy(1), depressive pseudo-dementia(n=6),and psycho-somatic syndrome(n=1).Demented patients were further classified,according to clinical criteria,as affected by probable AD(n=46;27women,19 men;MMSE=20.1±6.5)and patients affected by other dementias(n=35;14women,21men; MMSE=15.5±3.8).The latter group consisted of patients affected by vascular dementia(n=15), frontotemporal dementia(n=1),Creutzfeldt-JacobG.M.Sancesario et al./AβOligomers in CSF867disease(n=1),Parkinson’s disease with dementia (n=14),progressive supranuclear palsy(n=1), amyotrophic lateral sclerosis with dementia(n=1), and chronic depressive state(n=2).The diagnosis of probable AD was made according to the DSM IV and NINCDS-ADRA guidelines[15].Diagnosis of frontotemporal dementia was suggested accord-ing to neuropsychological deficits and behavioral symptoms[16].Diagnosis of probable vascular dementia included patients according to NINDS-AIREN criteria[17].Clinical and neuropsychological examination was repeated six months later in each patient to check the clinical course and accuracy of the diagnosis.CSF withdrawalLumbar punctures were performed following stan-dard procedures.Briefly,the skin in the lumbar area was cleansed with povidone-iodine solution10%, and local anesthesia was induced with ethyl chlo-ride.A20-gauge needle was inserted at the L3/L4 or L4/L5interspaces while the patients were at bed rest in the lateral decubitus or in the sitting position. 6–8ml were routinely drained in three consecutive polypropylene tubes without preservative,and imme-diately carried to the chemistry department.Blood samples were taken at the same time to evaluate blood-brain albumin ratio and blood-brain barrier integrity. Polypropylene tubes(Eppendorf Safe-Lock tubes) and polypropylene pipette tips(Gilson-DIAMOND Tips)were used in the collection and storage of CSF [18].One CSF sample from each patient was used for routine chemical analysis and microscopic observa-tion.Patients with blood contaminated CSF samples containing>500erythrocytes/␮l were excluded from this study.CSF samples were centrifuged(2000rpm) at4◦C for10min,divided in aliquots of300␮l,and stored at–80◦C within2h after withdrawal.Within1 to3months after freezing,the samples were thawed just once and immediately handled for the detection of A␤oligomers and for the measurements of A␤42, t-tau,and p-tau.Aβ42,t-tau and p-tau measurementsThe levels of A␤42,t-tau,and p-tau,classical biomarkers in CSF,were determined according to pre-viously published standard procedures[19,20],using commercially available sandwich enzyme-linked immunosorbent assays(ELISA)(Innotest␤-Amyloid 1-42,Innotest h-T-tau,Innotest PhosphoT-tau–Inno-genetics).Preparation of AβaggregatesWe obtained aggregates from A␤monomers using Dahlgren’s modification of Lambert’s protocol[21, 22].Recombinant A␤42and synthetic A␤40were purchased respectively from rPeptide(Athens,GA) and American Peptide(Sunnyvale,CA).Both A␤40 and A␤42monomers(1mg)were dissolved in anhy-drous dimethyl sulfoxide to5mM,and sonicated for 10min.Thereafter,the monomeric A␤solutions were diluted to100␮M either in cold HAM’s-F12cul-ture medium(Euroclone)and incubated over night at 4◦C for oligomer formation,or in HCl10mM and incubated at37◦C overnight forfibril formation.All samples were then stored at−20◦C.Analysis of synthetic AβaggregatesFrozen oligomer andfibril preparations were thawed just once,and used to confirm the formation of A␤aggregates by SDS-PAGE analysis without heat denat-uration followed by silver staining.Monomers and aggregated forms of A␤40and A␤42,corresponding to an original content of1␮g of monomers for each sample,were separated in16%SDS-PAGE tris-tricine buffer(5ml Acrylamide Bis/tri33%,0.88ml Glyc-erol,3.3ml Tris-HCl3M pH8.45,1ml SDS1%, 100␮l APS10%,10␮l TEMED).Polypeptide SDS PAGE molecular weight standards(Bio-Rad)were included as reference.After run,gels werefixed in 40%methanol and10%acetic acid over night,and then were silver stained(Bio-Rad Silver Stain kit).Densit-ometric analysis of scanned blots was performed using the NIH ImageJ version l.29program(NIH,Bethesda, MD,USA).Fluorescence labeling of antibodiesThe monoclonal A␤antibodies4G8(Chemicon) and6E10(Covance),recognizing different epitopes of the A␤peptide(17-24and1-16,respectively), were labeled respectively with thefluorescent chro-mophores Alexa Fluor488and Alexa Fluor594using the Monoclonal Antibody Labeling Kit(Invitrogen), following manufacturer’s instructions.868G.M.Sancesario et al./AβOligomers in CSFFRET andflow cytometric analysisExtensive FRET analysis was carried out in frozen samples of native CSF(n=137)and of synthetic oligomer andfibril preparations;frozen samples were thawed just once and directly incubated with 4G8-Alexa Fluor488(2nM)and6E10-Alexa Fluor 594.Moreover,some CSF samples(n=15of patients with AD,and n=15with other neurological disor-ders(OND))were studied in duplicate to evaluate the interference of dilution and detergents on detec-tion of oligomers.To this purpose,two CSF aliquots of each patient were studied:the undiluted one (defined as native)was directly incubated withfluo-rescent antibodies,according to our protocol,whereas the other was preliminary diluted(1:1)with RIPA buffer which contains both non ionic(0.5%NP-40and0.25%deoxycholate)and ionic detergents (SDS0.05%)according to Santos et al.[9,10]. Furthermore,in a separate and adjunctive group of patients(n=5)FRET analysis was carried out in duplicate in fresh native CSF samples incubated withfluorescent antibodies within2h after lum-bar puncture,as well as after one freezing-thawing cycle.Equal volumes(300␮l)of native or diluted CSF, of synthetic A␤40and A␤42monomers,and of A␤aggregate preparations(corresponding to an original content of0.1␮g of monomers for each sample)were incubated for90min at4◦C with4G8-Alexa Fluor 488(2nM)and6E10-Alexa Fluor594(8nM),accord-ing to Santos et al.[9].CSF samples and standard in vitro A␤aggregates were analyzed for monomer and oligomer content with a FACS Calibur cytome-ter(BD biosciences).FRET analysis was acquired from samples incubated either with thefluorescent chromophores Alexa Fluor488and Alexa Fluor594 (blank),or with thefluorescent antibody4G8-Alexa Fluor488only or with6E10-Alexa Fluor594,and finally with bothfluorescent antibodies.FRET involves the energy transfer between twofluorescent chro-mophores,from the excited donor molecule,Alexa Fluor488,to the acceptor molecule,Alexa Fluor594. The FRET can occur only if the distance between the two chromophore-carrying antibodies is less than 10nm,that is the distance between the donor and acceptor probe at which the energy transfer is(on average)50%efficient(F¨o rster distance).According to Santos et al.[9],oligomer particles were gated in logarithmic forward/side scatter dot plot(FSC ver-sus SSC).The green or redfluorescence respectively of the dyes Alexa Fluor488and Alexa Fluor594were detected in logarithmic scale by the correspond-ing FL1and FL3photomultipliers through530/30 or670LP bandpassfilter,respectively.To avoid dif-ference in analysis,known number of Count Bright absolute counting beads(Invitrogen)were added to all sample and gated in logarithmic FL4/SSC dot. Incubation of synthetic Aβ42oligomers in human CSFTo evaluate the stability of A␤42oligomers in human CSF,known amounts of synthetic A␤42oligomers were incubated in CSF of control(n=5)and AD patients(n=5).Briefly,synthetic A␤42oligomer preparation,corresponding to an original content of 0.5␮g monomers,was diluted to25␮M in native CSF 300␮l of AD and control patients,and incubated at 37◦for5h.SDS PAGE was used to determine the frequency distribution of synthetic A␤42monomers and oligomers:samples of control and probable AD patients were run in duplicate in16%tris-tricine SDS-PAGE at room temperature.After the run,the gels were alternativelyfixed in40%methanol and10% acetic acid for silver staining(Bio-Rad),or were trans-ferred onto nitrocellulose membrane in Tris glycine 20%methanol for western blot.Thereafter,nitrocellu-lose membranes were boiled3min in distilled H2O, blocked in3%BSA,and probed with the mono-clonal anti A␤antibody6E10(Covance)at a1:2,000 dilution in PBS at4◦C overnight.Bound antibodies were detected by incubation for2h at room tem-perature with an HRP-conjugated secondary antibody (DAKO)1:10,000in PBS.HRP reaction products were detected using the ECL system(Perkin Elmer), followed by ChemiDoc acquisition(Bio-Rad).Densit-ometric analysis of scanned blots was performed using the NIH ImageJ version l.29program(NIH,Bethesda, MD,USA).Statistical analysisAll data analysis was performed using the Statistical Package for the Social Sciences Windows,version17.0 (SPSS,Chicago,Illinois,USA).Descriptive statistics consisted in the mean±SD for variables with Gaus-sian distributions after confirmation with histograms and the Kolgomorov-Smirnov test,or median(min-max)for parameter categorical(non–parameters).The homogeneity of the variance was evaluated with Lev-ene’s test.One-way analysis of variance(ANOV A) and test for multiple comparison(Bonferroni test) was used to evaluate significant differences amongG.M.Sancesario et al./A βOligomers in CSF 869groups.Receiver operating characteristic (ROC)curveanalysis determined cut-off of sensibility and speci-ficity of the data.Statistical significance was set atp value <0.05.RESULTSSynthetic A βoligomersSilver stained SDS polyacrylamide gels were usedto distinguish aggregated species of synthetic A ␤40and A ␤42after applying the oligomer-or fibril-forming conditions.Analysis of the A ␤40gel showeda large band of monomers (∼3.5kDa)and an unspe-cific (∼9kDa)band equally detectable under differentexperimental conditions (Fig.1A).Analysis of theA ␤42gel showed an heterogeneous complex ofmonomers and oligomers of different sizes depend-ing on the experimental conditions (Fig.1B):oligomerpreparations contain monomer,tetramer,and pentamerbands of ∼14–17kDa,corresponding to low molecularweight (LMW)oligomers;fibril preparations containmonomer and tetramer bands,and large aggregatesof 27and ∼50–100kDa,which correspond to highmolecular weight (HMW)oligomers.The presenceof fibrils in the fibril preparation was not,however,detected by SDS-PAGE and immunoblotting:fibrilsinstead could be visualized by electron microscopyand atomic force microscopy according to previousworks [21,22].Although it is not possible to equatestructural A ␤42oligomer species observed in vitro andin CSF,it is worth noting that A ␤42fibril prepara-tion contains A ␤42monomers,and LMW as well asHMW oligomers,corresponding at least in part to theheterogeneous molecular size of the oligomer mixturedetected in CSF[23].Fig.1.SDS-PAGE and silver staining of synthetic A ␤40(A),andA ␤42(B),corresponding to an original content of 1␮g of monomers,in monomeric (lane 1),fibril (lane 2),or oligomeric preparations(lane 3).M =molecular weight markers.FRET and Flow cytometric analysis of in vitro formed A βaggregates The particles positive only for 4G8-Alexa Fluor 488can be detected by the FL1photomultiplier through a 530/30bandpass filter,while the particles posi-tive for both 4G8-Alexa 488and 6E10-Alexa 594can be detected by FL3photomultiplier through the 670LP bandpass filter,and can be identified as cor-responding to FRET events indicative of oligomer particles (Fig.2).FRET analysis readily distinguishes A ␤monomer and A ␤aggregate preparations.Sam-ples containing synthetic A ␤42monomers did not show detectable FRET signal after incubation with both monoclonal 4G8and 6E10anti-A ␤antibodies,labeled respectively with Alexa Fluor 488or Alexa Fluor 594(data not shown).Moreover,no FRET signals were detected in samples containing A ␤42oligomer preparation after incubation either with both chromophores Alexa Fluor 488and Alexa Fluor 594not bound to anti-A ␤antibodies (Fig.2A),or after incubation with just one labeled antibody,i.e.,4G8-Alexa Fluor 488or 6E10-Alexa Fluor 594separately (Fig.2B,C,E).On the contrary,after incubation with both anti-A ␤antibodies 4G8-Alexa Fluor 488and 6E10-Alexa Fluor 594,the samples of oligomer and fibril preparations displayed highly specific FRET signals in the FL3photomultiplier (670LP),indica-tive of single A ␤aggregates and of their numbers in the sample (Fig.2D,F).FRET signal intensity was stronger for fibril preparation,containing both LMW and HMW oligomers (Fig.2F),compared to oligomer preparation containing LMW oligomers only (Fig.2D).However,it is not presently clear whether fibrils in fibril preparation remain in solution and are efficiently detected by FRET signals.Anyway,fibril preparations were more stable than oligomer prepa-rations and were used to assess the sensitivity of the assay.As shown by Santos et al.[9],FRET signals are dose-dependent and linear for the concentrations of monomers used for A ␤42aggregates,ranging between 10pM to 2.0nM (Fig.3).Although estimation of the A ␤oligomer size is possible due to FRET signal intensity of each single particle [24],characteriza-tion of the molecular size of the detected particles cannot be easily inferred among a multiplicity of FRET signals.Moreover,a reliable internal standard for the detection of A ␤oligomer is presently not avail-able,since both natural and synthetic A ␤oligomer mixtures are heterogeneous,composed by variable species of different molecular sizes and conforma-tions.Therefore,our FRET analysis is not indicative870G.M.Sancesario et al./AβOligomers in CSFFig.2.Flow cytometric detection of A␤42oligomers prepared in vitro(original content of0.1␮g of monomers).Fluorescent signals are represented as black spots.The FRET signals were detected by the FL3photomultiplier(670LP bandpassfilter)and represented toward the middle of each graph if the acceptor Alexa Fluor594was excited by a very close donor Alexa Fluor488,whose signals were instead only detected by the FL1photomultiplier(530/30bandpassfilter)and represented on the left of each graph.A–D)Samples of A␤42oligomer preparations after:A)fluorescent chromophores Alexa Fluor488and Alexa Fluor594,or B)4G8-Alexa Fluor488,or C)6E10-Alexa Fluor594labeled antibodies were added separately,or D)4G8-Alexa Fluor488and6E10-Alexa Fluor594labeled antibodies were added in combination.E, F)Samples of A␤42fibril preparations after the6E10-Alexa Fluor594acceptor antibody was added alone(E),or in combination with the 4G8-Alexa Fluor488labeled antibody(F).The numbers reported on the corners of eachfigure represent the percentage offluorescent signals detected in each channel respective of the total of thefluorescent signals detected on the whole sample.G.M.Sancesario et al./AβOligomers in CSF871Fig.3.Sensitivity of the detection of A␤42oligomers infibril prepa-ration byflow cytometric analysis.A␤42oligomers were prepared in vitr o usingfibril preparation as described in methods,and progres-sively diluted from2nmol to10pmol(original content of0.1␮g of monomers).All data points represent mean values(±standard error) of measurements performed in triplicate.of the molecular size of the detected oligomers in the samples.FRET andflow cytometric analysis of native CSF Human CSF incubated with both chromophores Alexa Fluor488and Alexa Fluor594did not emitfluo-rescent signals(Fig.4A).Incubation of CSF with one labeled antibody plus a cromophore,i.e.,4G8-Alexa Fluor488plus Alexa Fluor594or with6E10-Alexa Fluor594plus Alexa Fluor488,displayedfluorescence signals in the FL1photomultiplier only,as expected (Fig.4B,C).Incubation of CSF with both labeled antibodies4G8-Alexa Fluor488and with6E10-Alexa Fluor594showedfluorescence signals in the FL3pho-tomultiplier,detecting the energy transfer of FRET events between contiguous donor and acceptor probes, and demonstrating the presence of A␤oligomers in human CSF(Fig.4D).The quantitative evaluation of FRET signals in CSF samples from control and AD patients’groups pre-sented a large variation in the amount of detected oligomers(Fig.5).In order to avoid differences in the measurements due to uncontrolled variations in vol-ume of the CSF samples and of other undetermined variables,FRET events in each sample were normal-ized to the amounts of Count Bright absolute counting beads detected in the same sample.In restricting the analysis to the relationship between the A␤FRET events relative to the counting beads,we unexpectedly found that the levels of A␤oligomers were lower in the CSF of probable AD patients than in OND and other dementia(OD)patients(p<0.05)(Fig.6B). Relationship between Aβoligomers and other biomarkers of ADLow levels of CSF A␤oligomers were detected in AD patients in the early as well as in the late stage of the cognitive impairment assessed with MMSE scores(data not shown),even though the number of AD patients in the different severity stages in our study is insufficient to clarify whether CSF oligomers could be related to the disease progression.To vali-date our results with other investigated biomarkers for AD,we also measured the levels of monomeric A␤42, t-tau,and p-tau in this set of samples as detected by ELISA.Consistent with previous reports,increased t-tau(p<0.01)and p-tau(p<0.01),and decreased A␤42(p<0.01and p<0.05)levels were detected in the CSF of patients with probable AD,compared to OND and to OD patients,respectively(Fig.6A,C,E). The values of A␤42,t-tau,and p-tau detected in our lab were within the confidence interval of the average±2 SD in the external quality control program for CSF biomarkers of the Alzheimer’s Association[25]. Thereafter,we studied whether any relationship exists between A␤oligomers detected by FRET and the other biomarkers of AD detected by ELISA. According to Santos et al.[9]and Fukumoto et al.[23],the ratio between A␤oligomers and monomeric A␤42in CSF was not significant(data not shown). Conversely,the ratio of A␤oligomers/t-tau was significantly lower in the CSF of patients with probable AD compared to the OND control group (p<0.01)but not to the OD group(Fig.6D);the ratio of A␤oligomers/p-tau was significantly lower in AD patients compared to OND control(p<0.05) as well as to OD patients(p<0.01)(Fig.6F).We next performed ROC curve analysis to assess the ability of the ratio A␤oligomers/t-tau or A␤oligomers/p-tau to discriminate the patients with probable AD from the other two groups.The cutoff level of0.40for the ratio A␤oligomers/t-tau yielded a sensitivity of78% and a specificity of71%,while a cutoff level of2.50 for the ratio A␤oligomers/p-tau yielded a sensitivity of75%but a specificity of64%.Such values were comparable to sensitivity and specificity yielded by the other biomarkers alone or in combination in our study:A␤monomer sensitivity78%,specificity65%; t-tau sensitivity80%,specificity62%;p-tau sensitivity 80%,specificity63%;A␤monomers/t-tau sensitivity872G.M.Sancesario et al./A βOligomers inCSFFig.4.Representative flow cytometric detection of A ␤42native oligomers in CSF samples.A)CSF with fluorescent chromophores Alexa Fluor 488and Alexa Fluor 594.B–D)CSF after 4G8-Alexa Fluor 488(B),or 6E10-Alexa Fluor 594(C)labeled antibodies were added separately,or(D)in combination.Significant FRET signals were detected by the FL3photomultiplier (670LP bandpass filter)only in D.78%,specificity 71%;A ␤monomers/p-tau sensitivity80%,specificity 62%.Influence of CSF sample preparation on A βoligomersTo examine whether the RIPA buffer,which wasused in the original protocol of Santos et al.[9,10],could affect the detection of A ␤oligomers,we alsoconducted FRET analysis in duplicate in native CSFand in CSF diluted with RIPA buffer,using CSF sam-ples obtained from 30individuals (15patients withAD,and 15with OND).Taking into account the dilu-tion of CSF with RIPA (1:1)compared to native CSF,there was no correspondence between the two ana-lytical procedures since most of the data set of CSF diluted with RIPA lie above the identity line in a scatter diagram of pooled data from AD and control patients (Fig.7A)(p <0.01);indeed,the aliquots of CSF diluted with RIPA showed a concentration of A ␤oligomers significantly higher than the aliquots of native CSF from the same individuals examined at the same time (Fig.7B).Moreover,the influence of CSF prepara-tions was comparatively evaluated in the AD and OND group.In native CSF,the levels of A ␤oligomers were not significantly different between the two groups due to the large standard deviation and to the small sample size of this subset of patients (Fig.7C).Conversely,the levels of A ␤oligomers in the CSF aliquots diluted with RIPA were significantly higher for AD compared to OND patients (p <0.05):RIPA buffer increases the overall amount of A ␤oligomers and reduces the large。