Alzheimer’s Disease Cybrids Replicateb-Amyloid Abnormalities through

Cell Death PathwaysShaharyar M. Khan, BA, David S. Cassarino, PhD, Nicole N. Abramova, BS, Paula M. Keeney, BS,

M. Kate Borland, BS, Patricia A. Trimmer, PhD, Clara T. Krebs, BS, Jason C. Bennett, Janice K. Parks, MS,Russell H. Swerdlow, MD, W. Davis Parker, Jr, MD, and James P. Bennett, Jr, MD, PhD

Alzheimer’s disease (AD) is characterized by the deposition in brain of b-amyloid (Ab) peptides, elevated braincaspase-3, and systemic deficiency of cytochrome c oxidase. Although increased Ab deposition can result from mutationsin amyloid precursor protein or presenilin genes, the cause of increased Ab deposition in sporadic AD is unknown.Cytoplasmic hybrid (“cybrid”) cells made from mitochondrial DNA of nonfamilial AD subjects show antioxidant-reversible lowering of mitochondrial membrane potential (D(gYm), secrete twice as much Ab(1-40) and Ab(1-42), haveincreased intracellular Ab(1-40) (1.7-fold), and develop Congo red–positive Ab deposits. Also elevated are cytoplasmiccytochrome c (threefold) and caspase-3 activity (twofold). Increased AD cybrid Ab(1-40) secretion was normalized byinhibition of caspase-3 or secretase and reduced by treatment with the antioxidant S(2)pramipexole. Expression of ADmitochondrial genes in cybrid cells depresses cytochrome c oxidase activity and increases oxidative stress, which, in turn,lowers DCm. Under stress, cells with AD mitochondrial genes are more likely to activate cell death pathways, which drivecaspase 3–mediated Ab peptide secretion and may account for increased Ab deposition in the AD brain. Therapeuticstrategies for reducing neurodegeneration in sporadic AD can address restoration of DCm and reduction of elevated Absecretion.

Khan SM, Cassarino DS, Abramova NN, Keeney PM, Borland MK, Trimmer PA, Krebs CT, Bennett JC,Parks JK, Swerdlow RH, Parker WD, Bennett JP. Alzheimer’s disease cybrids replicate b-amyloid

abnormalities through cell death pathways. Ann Neurol 2000;48:148–155

Alzheimer’s disease (AD) is the most prevalent neuro-degenerative disease in human beings. Over 90% ofAD occurs with a sporadic (nonfamilial) pattern, al-though a minority of AD cases have autosomal domi-nant inheritance (familial AD [FAD]). FAD has beenlinked to mutations in either the amyloid precursorprotein (APP)1 or one of the presenilin (PS) genes2

(PS13,4 and PS25).A neuropathological hallmark of AD is extensive and

regionally specific deposition of b-amyloid (Ab) pep-tide fragments through increased processing of APP byseveral putative “secretase” enzymes, including the re-cently described b-secretases.6–9 FAD due to APP orPS mutations is generally associated with increased se-cretion10 and extensive deposition in brain of theAb(1-42) fragment, particularly in senile plaques.11 Inthe sporadic AD brain, insoluble Ab(1-40) levelsare substantially higher than Ab(1-42) levels,12 andAb(1-40) is also deposited in senile plaques to adegree greater than in FAD.11

Transgenic mice with APP or PS mutations showextensive cerebral amyloid deposition but limited neu-ropathology beyond amyloid deposition and variabledegrees of cognitive impairment.13–17 This has ledsome to question as to whether amyloid deposition is aprimary pathological event in either sporadic or FADcompared with its being a secondary marker for theunderlying abnormality responsible for neurodegenera-tion.18 For sporadic AD in particular, the mechanismsof increased Ab deposition remain unclear.

Caspase enzymes are capable of processing APP intosmaller fragments, which are deposited in AD brainalong with Ab and, in turn, can stimulate secretase-mediated production of Ab peptides.19 Because ADbrains demonstrate elevated caspase-3 protein19,20 andincreased caspase-3 immunoreactivity in neurons dis-playing DNA fragmentation,21 and because caspase-3is a primary “executioner” caspase in apoptotic cas-cades, an etiological relation between Ab deposition inAD and activation of cell death pathways is possible.

From the Center for the Study of Neurodegenerative Diseases, Uni-versity of Virginia Health System, Charlottesville, VA.

Received Dec 27, 1999, and in revised form Mar 22, 2000. Ac-cepted for publication Mar 22, 2000.

Address correspondence to Dr Bennett, Box 394 HSC, Universityof Virginia, Charlottesville, VA 22908.

148 Copyright © 2000 by the American Neurological Association

Because each process might stimulate the other, it re-mains unclear what their order of appearance is in thepathogenesis of AD.

We have explored this question with AD cybrids.The cybrid technique is based on the introduction ofmitochondria and mitochondrial DNA (mtDNA) intoreplicating clonal host cells that are deprived of theirown mtDNA (r0 cells) by long-term incubation withlow levels of ethidium bromide. Removal of specificnutritional support from r0 cells after cybrid formationselects for cybrids with adequate mitochondrial re-population. Six weeks of cell (and mitochondrial) rep-lications before assay eliminate any potential effects ofenvironmental factors on mitochondrial function. Thenuclear genetic component of cybrids is equivalent,and the mitochondrial genotype is left as the only vari-able. Thus, differences among various cybrid lines mustbe attributable to differences in sources of mtDNA.

We have shown that AD cybrids recapitulate manypotentially pathogenic features of AD, including de-creased cytochrome c oxidase (COX) activity,22 in-creased oxidative stress,22 abnormal intracellular cal-cium signaling,23 reduced mitochondrial membranepotential (DCm),24 and abnormal mitochondrial mor-phology and movement velocities.25 We performed thepresent study to test the hypothesis that AD mitochon-drial genes are also responsible for a critical neuro-pathological feature of AD, cellular oversecretion of Abpeptides.

Materials and MethodsCell CultureCybrid cell lines were produced in human neuroblastomaSH-SY5Y cells using methodology previously described22 andwere maintained in Dulbecco’s Modified Eagle’s Medium(DMEM) supplemented with 10% fetal calf serum, penicil-lin/streptomycin, and pyruvate/uridine. Cells were grown in75-cm2 flasks to 90 to 100% confluence at 37°C and 5%CO2. In some experiments conducted 24 hours before mea-suring amyloid secretion, the culture media were changed tomedium without serum and supplemented with glutamineand pyruvate/uridine. In selected experiments, 100 nM ofDEVD.fmk (caspase-3 inhibitor) or 100 nM of AEBSF(secretase inhibitor) was added 2 hours before beginning thesecretion experiments. In experiments measuring DCm, cellswere grown in medium with 5 to 10% serum. All experi-ments used mixed population (nonclonal) cell lines fromeach cybrid preparation.

Sandwich Enzyme-Linked Immunosorbent Assay forAb(1-40) and Ab(1-42)Ab secretion was assayed in two independent experiments. Inthe first experiment, which was unmasked, five AD and fiveage-matched control cybrid lines were incubated for 24hours in DMEM (5 ml in t75 flasks) supplemented withpyruvate/uridine. Media from these cells were assayed forAb(1-40) levels as described below. In the second masked

experiment performed 2 months later, independent culturesof the same AD and control cybrid lines were incubated for24 hours in DMEM (10 ml in t75 flasks) supplemented withpyruvate/uridine. Media from these cells were assayed forAb(1-40) and Ab(1-42) levels. The media were removed,centrifuged, and concentrated 310 to 320 with Centricon-3tubes to a final volume of 0.5 ml. Cells were lysed in Oltavibuffer and assayed for intracellular Ab peptides by enzyme-linked immunosorbent assay (ELISA) or Congo red bind-ing.26 ELISA assay (QCB/Biosource International, Cama-rillo, CA) was performed according to the manufacturer’sprotocol. In brief, a 96-well clear-bottom plate suited for flu-orescence was coated with a goat anti-mouse polyclonal an-tibody. A second “capture” antibody that was N-terminal–specific for Ab was added to the plate. Approximately 100ml of sample or Ab peptide standard was added to each well.A tertiary rabbit polyclonal antibody specific to the carboxy-terminus of the 1-40 or 1-42 species of amyloid was addedto the plate. For detection of the antibody sample complex,a final alkaline-phosphatase linked anti-rabbit goat polyclonalantibody was allowed to incubate in the wells. After the ad-dition of an alkaline-phosphatase substrate supplied with thekit, wells were read at 460 nm excitation and 560 nm emis-sion wavelengths. Results are expressed as picograms per mil-liliter of Ab(1-40) or Ab(1-42) normalized with respect tomicrograms per milliliter of protein from cells secreting amy-loid.

Caspase-3 ActivityCells (2 3 107 per flask) were collected with phosphate-buffered saline (PBS) and harvested by centrifugation at 450gfor 6 minutes at 4°C. Pellets were resuspended in a hypo-tonic lysis buffer (25 mM of HEPES, 5mM of MgCl2, 5mMof EDTA, 1 M of dithiothreitol, and Sigma protease inhib-itor cocktail) at 2 3107 cells per 100 ml of lysis buffer. Ly-sates were subjected to four cycles of freezing and thawingand then centrifuged at 16,000g for 30 minutes at 4°C. Thesupernatant fractions were collected, and 100 mg of eachprotein was incubated with 2.5 mM of Ac-DEVD-AMC andPromega incubation buffer. Fluorescence of the caspase-3product was measured at 0, 15, 30, and 40 minutes. Accu-mulation of product was always linear with time. Results aregiven as fluorescence units after 30 minutes of incubation.

Western Blotting for Cytochrome cA total of 150 mg of cytoplasmic protein per well was run outon 4 to 20% Tricine polyacrylmide gels (Bio-Rad). Proteinswere transferred to polyvinyldifluoride membrane overnight.Blots were blocked with nonfat milk and incubated with pri-mary antibody to cytochrome c (Pharmingen, San Diego,CA). Blots were washed in tripton 3100–tris-buffered salineand incubated with secondary antibody. After washing off ex-cess secondary antibody, the blots were developed with thePierce Ultra-Chemiluminescence kit (Rockford, IL) and de-tected using the Fluor-S (Bio-Rad).

Immunohistochemical Staining for AbGrowth media were removed from the flasks, cultures werewashed three times with PBS (pH 7.4), fixed for 10 minutesat room temperature with 4% phosphate-buffered parafor-

Khan et al: Cybrids Replicate AD Amyloid Abnormalities 149

maldehyde (pH 7.4), and washed three times with PBS.Fixed cells were incubated 10 minutes in 4°C methanol,washed with PBS, incubated 3 minutes in 50% formic acid,and washed with PBS until pH 7.4 was confirmed with pHstrips. Pretreated cultures were then incubated in 300 nM ofMitoTracker Green (Molecular Probes, Eugene, OR) for 25minutes at room temperature, washed in PBS, incubated 1hour at room temperature in a 1:50 monoclonal Ab anti-body (Novocastra Laboratories, Newcastle upon Tyne, UK),washed in PBS, incubated in 15 mg/ml of horse anti-mouseTexas Red (Vector Laboratories, Burlingame, CA) for 1 hourat room temperature, washed in PBS, and coverslipped usingVectashield mounting medium (Vector Laboratories). Singleconfocal images of cell aggregates were taken using an Olym-pus Confocal Laser Scanning System mounted on an Olym-pus IX70 microscope (Olympus, Melville, NY).

CONGO RED STAINING Cells were fixed as described forAb staining, washed with water, and incubated for 20 min-utes at room temperature in NaCl working solution (2.5%NaCl in 80% EtOH with 0.0025% NaOH added just be-fore filtration and use), followed by a 20-minute room tem-perature incubation in Congo red working solution (0.5%Congo red in 2.5% NaCl/80% EtOH with 0.0025% NaOHadded just before filtration and use). Cells were then washedrapidly with 100%, 95%, and 50% EtOH and PBS, incu-bated in 300 nM of MitoTracker Green in PBS for 25 min-utes at room temperature, washed in PBS, and coverslippedusing Vectashield mounting medium.

JC-1 Staining for DCm

Cells were grown in medium with 10% fetal bovine serum,penicillin (100 IU/ml), streptomycin (100 mg/ml), ampho-tericin (2.5 mg/ml), uridine (50 mg/ml), and pyruvate (100mg/ml). Forty-eight hours before JC-1 uptake, cells were re-plated in number 0 glass-bottom, poly-D-lysine–coated,35-mm dishes (MatTek Corp, Ashland, MA) at a concentra-tion low enough to visualize single cells after uptake. At thetime of uptake, frozen aliquots (1 mg/ml in dimethylsulfox-ide) of JC-1 (Molecular Probes) were diluted to a final con-centration of 4 mg/ml in maintenance media without phenol

red. A 35-mm dish was incubated in 4 mg/ml of JC-1 for 10minutes at 37°C with 5% CO2, washed twice with Hank’sbalanced salt solution, and held in maintenance media with-out phenol red during imaging. Confocal images were col-lected from a stained dish for 30 minutes after completion ofstaining and then the plate was discarded.

3H-Tetraphenylphosphonium Ion IncubationAD and control cybrid cell lines were incubated for 3 days inmedium with 10% serum with or without antioxidants,washed, and incubated with 3H-tetraphenylphosphoniumion (3H-TPP1) as described.20 3H-TPP1 accumulation wasnormalized to monoamine oxidase activity as described.20

ResultsWe used cybrids made from platelet mitochondria of 5sporadic AD subjects and 5 age-matched neurologicallynormal subjects to examine amyloid metabolism andmitochondrial function in this disease model system.Using sensitive and specific sandwich ELISAs to detectAb(1-40) and Ab(1-42) peptides, we found that themedia of AD cybrids contained about twice as manyAb(1-40) and Ab(1-42) peptides as did the media ofcontrol cybrids (Fig 1A). Intracellular levels of Ab pep-tides in AD cybrids were also elevated, whether assayedby specific sandwich ELISAs or with Congo red spec-trophotometry (see Fig 1B).26

We then examined the effect of caspase-3 inhibitionwith the specific inhibitor DEVD.fmk on Ab(1-40) ac-cumulation in cybrid cell media. We found thatcaspase-3 inhibition lowered the elevated Ab(1-40) se-cretion in AD cybrids to control basal levels (see Fig1C). Elevated Ab(1-40) secretion in AD cybrids couldalso be reduced to control levels by nonselective inhi-bition of secretase activities (see Fig 1C). Incubation ofAD cybrids with S(2)pramipexole, an agent we haveshown to be an effective oxygen radical scavenger invitro and in vivo and an inhibitor of opening of the

Fig 1. (A) Ab(1-40) and Ab(1-42) secretion by five Alzheimer’s disease (AD) and five control (CTL) cybrid lines in serum-freemedia. Results are expressed as percentage of mean control values 6 SEM. For Ab(1-40), data were combined from the unmaskedand masked studies, which gave equivalent results. Ab(1-42) secretion was measured in the masked study only. *p , 0.05 by Stu-dent’s t test. For comparison, the mean values from the second (masked) study for Ab(1-40) secretion (picograms of peptide per mil-ligram of cell protein) were 12.0 for AD cybrids and 6.3 for control cybrids; for Ab(1-42) secretion, the mean values were 18.0 forAD cybrids and 9.2 for control cybrids. All data represent averages of duplicate measurements from each cell line. (B) IntracellularAb peptides in five AD and five control cybrid lines. Cell lysates from (A) were assayed by sandwich enzyme-linked immunosorbentassays for Ab(1-40) and Ab(1-42) or with Congo red spectrophotometry. All data are from the masked study. *p , 0.05 by Stu-dent’s t test. Mean levels for intracellular Ab(1-40) (picograms of peptide per milligram of cell protein) were 12.6 for AD cybridsand 7.4 for control cybrids. For intracellular Ab(1-42), levels were 19.5 in AD cybrids and 0.97 in control cybrids (p 5 0.09).All data represent averages of duplicate measurements from each cell line (C) Suppression of elevated AD cybrid Ab(1-40) secretionby inhibition of caspase-3 (DEVD.fmk) or secretases (AEBSF)31 or after incubation with 1 mM of S(2)pramipexole. *p , 0.02by Student’s t test compared with AD basal Ab(1-40) secretion. #p , 0.04 by Student’s t test compared with control basalAb(1-40) secretion. For S(2)pramipexole effect on AD Ab(1-40) secretion, p 5 0.06 by Student’s t test. All data represent aver-ages of duplicate measurements from each cell line. PPX 5 S(2)pramipexole. (D) AD and control cybrid basal caspase-3 activity.*p , 0.05. All data represent averages of triplicate measurements from each cell line. (E) Western blot analysis of intracellular pro-tein from AD and control cybrids probed with antibody to Ab(1–40). Molecular weight markers are located to the right. The twolanes on the left are from cells treated with the caspase-3 inhibitor DEVD.fmk.

‹

150 Annals of Neurology Vol 48 No 2 August 2000

mitochondrial transition pore,27 also reduced elevatedAb(1-40) secretion (see Fig 1C). We then examinedbasal caspase-3 activities and found a twofold elevationin AD compared with control cybrids (see Fig 1D).

We next looked for morphological consequences ofelevated Ab(1-40) secretion by AD cybrids by growingcybrid lines to confluence and maintaining them in aminimum volume of media for 10 to 12 days, followedby washing, fixation, and immunohistochemical stain-ing for Ab(1-40). When imaged with confocal mi-croscopy, AD cybrid lines showed multiple large hypo-

cellular areas with dense Ab deposition (Fig 2).Intracellular amyloid staining was also increased in ADcybrids (see Fig 2), consistent with the biochemicaldata (see Fig 1B) and Western blot analysis (see Fig1E). Control cybrids showed less prevalent and muchless intense areas of positive Ab(1-40) immunoreactiv-ity. We also used the Ab(1-40) antibody to stain amy-loid deposits in paraffin sections of AD brain tissueand observed typical plaques, which were smaller thanthe amyloid deposits seen in cell culture (see Fig 2).Congo red staining of amyloid deposits in AD cybrids

Khan et al: Cybrids Replicate AD Amyloid Abnormalities 151

revealed dense homogeneous patterns, which con-trasted with reticular Congo red staining patterns seenin AD brain (see Fig 2).

We then sought potential mechanisms for the ele-vated caspase-3 activity in AD cybrids, which waslinked to the increased Ab(1-40) secretion. If mito-chondrial depolarization leads to cytochrome c release,our previously described lowering of averaged DCm inAD cybrids24 may explain elevations in caspase-3 activ-ities. To explore this possibility further, we character-ized DCm abnormalities in AD cybrids in two ways.First, we studied the population distribution of DCm

among AD and control cybrids compared with r0 cells,using calibrated confocal microscopy with JC-1 as anindicator of DCm. JC-1 distributes across mitochon-dria as a function of DCm. At lower levels of mito-chondrial energization, JC-1 fluorescence is detectedmainly as the green monomer, whereas higher DCm

and resulting elevations of mitochondrial JC-1 levelslead to greater degrees of red fluorescing JC-1 aggregate.We examined the utility of using ratios of aggregate andmonomer fluorescence intensities in our confocal systemby calibrating JC-1 fluorescence in isolated rat liver mi-tochondria against potassium diffusion potentials,28

and we obtained a linear semilogarithmic relation (Fig3). We then examined the distribution of DCm in ADand control cybrids and r0 cells and found that AD

cybrids are highly overrepresented in the lower energymitochondria group and highly underrepresented inthe higher energy mitochondria group (see Fig 3). Thedistribution of DCm in AD cybrids resembled that ofr0 cells.

Because our AD cybrid lines have selective and rel-atively small losses of COX, it seemed unlikely thatelectron transport chain (ETC) deficits per se could ex-plain the observed loss of DCm. We hypothesized thatoxygen radicals resulting from ETC dysfunction causedsecondary oxidative damage to critical mitochondrialmembrane components involved in DC regulation. Weincubated AD and control cybrids with the free radicalscavengers S(2)pramipexole, trolox (a water-soluble vi-tamin E analogue), or both agents for 3 days and esti-mated DCm with accumulation of 3H-TPP1.24 Wefound that anti-oxidant treatment partially restored thereduced DCm of AD cybrids (Fig 4). The basal 3H-TPP1 accumulation by AD cybrids was 73% of that ofcontrol cybrids.24 Thus, the maximum resoration ofDCm that we observed would be equivalent to 88% ofbasal control values. This supports the concept that theloss of DCm in AD cybrids results from oxidative dam-age to mitochondrial membrane components responsiblefor DCm regulation and is not due to an irreversiblyreduced capacity of AD mitochondria to maintain anormal DCm.

Fig 2. Confocal microscopic images of immunohistochemical staining for b-amyloid (Ab) (1-40) in Alzheimer’s disease (AD) andcontrol cybrids grown to confluence and maintained at high density for 10 to 12 days. The Ab primary antibody is localized withTexas Red. Sections are costained with MitoTracker Green. Intracellular Ab appears as yellow (red 1 green). Paraffin sections ofAD hippocampus were immunostained for Ab(1-40) and localized with fluorescein isothiocyanate. Positive Ab immunostainingappears as green. Congo red staining was performed in both AD cybrid and paraffin sections of AD hippocampus. All cybrid imagesare at 3150, and all AD brain paraffin section images are at 3400.

152 Annals of Neurology Vol 48 No 2 August 2000

Because mitochondrial depolarization can release cy-tochrome c within the cytoplasm, activating the caspasecell death–associated protease cascade, we isolated cy-toplasm fractions of AD and control cybrids and esti-

mated relative levels of cytoplasmic cytochrome c byWestern blot analysis. We found extremely low levelsof cytochrome c in cytoplasmic preparations of controlcybrids, but we observed approximately threefold ele-vated levels of cytoplasmic cytochrome c in AD cybrids(Fig 5). This elevated cytoplasmic cytochrome c mayaccount for increased caspase-3 activities and caspase-3–dependent Ab secretion.

DiscussionBy introducing AD mitochondrial genes into r0 hostcells, we have produced cybrid cells that oversecreteAb(1-40) and Ab(1-42) peptides and have elevated in-tracellular levels of Ab peptide. AD cybrids in denseculture also lay down Ab deposits as demonstrated byboth immunohistochemical and Congo red staining.Our findings implicate mitochondrial genetic abnor-malities as contributing to pathogenic amyloid metab-olism in sporadic AD.

Although serum deprivation represents a form of cel-lular stress, our observed elevations of Ab peptide se-cretion in AD cybrids most likely do not arise from theeffects of serum deprivation alone. Twenty-four hoursof serum deprivation did not cause any increase inAb(1-40) secretion by human neurotrophin-2 terato-carcinoma cells.19 In addition, culture conditions wereidentical for all AD and control cybrid lines, whichwere studied simultaneously in two independent exper-iments. Rather, Ab oversecretion in AD cybrids ap-pears to derive from oxidative stress–mediated damageto components that regulate DCm, with resulting in-creased release of cytochrome c and activation ofcaspase-3. Although increased caspase 3–mediated pro-

Fig 3. (Top) Ratios of JC-1 aggregate and monomer fluores-cence intensity in isolated rat liver mitochondria as a functionof potassium diffusion potential. (Middle) Confocal laser mi-croscopic images of JC-1 monomer (green) and aggregate (red)fluorescence in control (CTL) and Alzheimer’s disease (AD)cybrids. (Bottom) Histogram of distribution of JC-1 aggregateand monomer fluorescence intensity in AD and control cybridsand SH-SY5Y r0 cells. X2 analysis showed that AD and con-trol proportions were significantly different (p 5 0.002).

Fig 4. Effects of 3 days of anti-oxidant treatment on mito-chondrial membrane potential (DCm ) in Alzheimer’sdisease and control cybrids. DCm was estimated with 3H-tetraphenylphosphonium ion accumulation. PPX 5 1 mM ofS(2)pramipexole. Trol 5 100 mM of trolox.

Khan et al: Cybrids Replicate AD Amyloid Abnormalities 153

cessing of APP19 could account for our observed eleva-tions in Ab peptide secretion, we have not yet identi-fied specific caspase-3–produced APP derivatives.19 Atthe moment, our data support only the caspase-3 de-pendency of the elevated Ab(1-40) secretion.

Although much attention has recently been paid tothe pathophysiology of the longer Ab(1-42) fragment,which is oversecreted in several FAD variants, theAb(1-40) fragment (both soluble and insoluble) ismore prevalent than the Ab(1-42) fragment in sporadicAD brains12 and is regularly observed in mature senileplaques and vascular deposits in the AD brain. We ob-served increased secretion of both Ab fragments in ourAD cybrids. Both Ab species are pathologically depos-ited in sporadic AD brain and exhibit neurotoxicity incell models, but the argument continues as to whetherdeposition of either Ab species is a primary event ofsporadic AD and causes the neurodegeneration orwhether it is a secondary marker for the underlyingpathological process responsible for cell death.18 Basedon our findings, we propose that abnormal mitochon-drial genes in sporadic AD drive an oxidative stress–and caspase-3–mediated Ab(1-40) and Ab(1-42) over-secretion and deposition and that resulting increases inintracellular and extracellular Ab peptide further accel-erate cell death cascades.

Our findings support the hypothesis that a patho-

genic site of action of mitochondrial gene–driven oxi-dative stress in sporadic AD is damage to mitochon-drial mechanisms of DCm regulation. Our observationsthat relief of oxidative stress with antioxidants both im-proves DCm and reduces Ab peptide oversecretion sug-gest but do not prove a causal link between the twoprocesses. The greater abundance of more depolarizedAD mitochondria may account for the increased cyto-plasmic cytochrome c and doubling of basal caspase-3activity that we observed, which could then stimulateAPP processing into toxic Ab fragments. Increased in-tracellular Ab peptide can then worsen the underlyinggenetic complex IV lesion29,30 and increase the like-lihood of mitochondrial transition pore opening,30

worsening the deficit in maintaining DCm. SecretedAb peptide can potentially increase caspase-3 activationfurther, setting up a spiraling cycle toward cell death.

In this paradigm, Ab oversecretion would serve acritical secondary role in the pathogenesis of sporadicAD even though its origin is from mitochondrial ge-netic defects in COX activity and oxidative stress–induced damage to regulation of DCm. In this scheme,mitochondrial genetic defects and Ab oversecretion areetiologically intertwined, and both would be necessaryfor the cell death and cognitive deficits that define thisdisease.

Our results are important in light of the potentialfor feed-forward neurotoxicity of Ab peptides. Accord-ing to this concept, a given neuron oversecreting Abpeptides may induce toxicity in neighboring cellsthrough Ab effects on those cells. Our results haveshown that this Ab oversecretion can derive solelyfrom the effects of AD mitochondrial genes oncaspase-3 activities. Conceivably, cells may survive to acertain level of Ab buildup but then succumb as Abfurther inhibits COX and increases the probability ofmitochondrial transition pore opening.

Therapeutic interventions in this cascade of celldeath could involve caspase inhibition, anti-oxidativeor other therapies that can restore reduced DCm, orcombinations thereof. Because of the necessity of apo-ptosis in many peripheral cell systems and the depen-dence of many such forms of apoptosis on caspases,anti-oxidant agents may be a more realistic approach tochronic neuroprotective therapy. Therefore, futureneuroprotective clinical trials in AD patients shouldemploy agents that both restore DCm in AD mito-chondria and reduce Ab oversecretion so as to preventneuronal cell loss and the disabling effects of this dev-astating neurodegenerative disease.

This study was supported by NIH grants AG14373 and NS35325.Pharmacia and Upjohn donated the S(2)pramipexole used in thisstudy.

Fig 5. (A) Cytoplasmic cytochrome c levels in Alzheimer’s dis-ease (AD) and control (CTL) cybrids by Western blot analysis.(B) Quantitative densitometry of cytoplasmic cytochrome c inAD and control cybrids.

154 Annals of Neurology Vol 48 No 2 August 2000

References1. Goate A, Chartier-Harlin MC, Mullan M, et al. Segregation of

a missense mutation in the amyloid precursor protein gene withfamilial Alzheimer’s disease. Nature 1991;349:704–706

2. Hardy J. Amyloid, the presenilins and Alzheimer’s disease.Trends Neurosci 1997;20:154–159

3. Sherrington R, Rogaev EI, Liang Y, et al. Cloning of a genebearing missense mutations in early-onset familial Alzheimer’sdisease. Nature 1995;29:754–760

4. Alzheimer’s Disease Collaborative Group. The structure of thepresenilin 1 (S182) gene and identification of six novel muta-tions in early onset AD families. Nat Genet 1995;11:219–222

5. Levy-Lahad E, Wijsman EM, Nemens E, et al. A familial Alz-heimer’s disease locus on chromosome 1. Science 1995;269:970–973

6. Vassar R, Bennett BD, Babu-Khan S, et al. Beta-secretase cleav-age of Alzheimer’s amyloid precursor protein by the transmem-brane aspartic protease BACE. Science 1999;286:735–741

7. Sinha S, Anderson JP, Barbour R, et al. Purification and clon-ing of amyloid precursor protein beta-secretase from humanbrain. Nature 1999;402:537–540

8. Yan R, Bienkowski MJ, Shuck ME, et al. Membrane-anchoredaspartyl protease with Alzheimer’s disease beta-secretase activity.Nature 1999;402:533–537

9. Hussain I, Powell D, Howlett DR, et al. Identification of anovel aspartic protease (Asp 2) as beta-secretase. Mol Cell Neu-rosci 1999;6:419–427

10. Scheuner D, Eckman C, Jensen M, et al. Secreted amyloidbeta-protein similar to that in the senile plaques of Alzheimer’sdisease is increased in vivo by the presenilin 1 and 2 and APPmutations linked to familial Alzheimer’s disease. Nat Med1996;2:864–870

11. Iwatsubo T, Odaka A, Suzuki N, et al. Visualization ofAb42(43) and Ab40 in senile plaques with end-specific Abmonoclonals: evidence that an initially deposited species isAb42(43). Neuron 1994;13:45–53

12. Wang J, Dickson DW, Trojanowski JQ, Lee VM. The levels ofsoluble versus insoluble brain Abeta distinguish Alzheimer’s dis-ease from normal and pathologic aging. Exp Neurol 1999;158:328–337

13. Masliah E, Sisk A, Mallory M, et al. Comparison of neurode-generative pathology in transgenic mice overexpressing V717Fbeta-amyloid precursor protein and Alzheimer’s disease. J Neu-rosci 1996;16:5795–5811

14. Chapman PF, White GL, Jones MW, et al. Impaired synapticplasticity and learning in aged amyloid precursor protein trans-genic mice. Nat Neurosci 1999;2:271–276

15. Moechars D, Dewachter I, Lorent K, et al. Early phenotypicchanges in transgenic mice that overexpress different mutants ofamyloid precursor protein in brain. J Biol Chem 1999;274:6483–6492

16. Chui DH, Tanahashi H, Ozawa K, et al. Transgenic mice withAlzheimer presenilin 1 mutations show accelerated neurodegen-

eration without amyloid plaque formation. Nat Med 1999;5:560–564

17. McGowan E, Sanders S, Iwatsubo T, et al. Amyloid phenotypecharacterization of transgenic mice overexpressing both mutantamyloid precursor protein and mutant presenilin 1 transgenes.Neurobiol Dis 1999;6:231–244

18. St George-Hyslop PH, Westaway DA. Antibody clears senileplaques. Nature 1999;400:116–117

19. Gervais FG, Xu D, Robertson DS, et al. Involvement ofcaspases in proteolytic cleavage of Alzheimer’s amyloid-b pre-cursor protein and amyloidogenic Ab peptide formation. Cell1999;97:395–406

20. Shimohama S, Tanino H, Fujimoto S. Changes in caspase ex-pression in Alzheimer’s disease: comparison with developmentand aging. Biochem Biophys Res Commun 1999;256:381–384

21. Masliah E, Mallory M, Alford M, et al. Caspase dependentDNA fragmentation might be associated with excitotoxicity inAlzheimer disease. J Neuropathol Exp Neurol 1998;57:1041–1052

22. Swerdlow RH, Parks JK, Cassarino DS, et al. Cybrids in Alz-heimer’s disease: a cellular model of the disease? Neurology1997;49:918–925

23. Sheehan JP, Swerdlow RH, Miller SW, et al. Calcium ho-meostasis and reactive oxygen species production in cells trans-formed by mitochondria from individuals with sporadic Alzhei-mer’s disease. J Neurosci 1997;17:4612–4622

24. Cassarino DS, Swerdlow RH, Parks JK, et al. Cyclosporin Aincreases resting mitochondrial membrane potential in SY5Ycells and reverses the depressed mitochondrial membrane po-tential of Alzheimer’s disease cybrids. Biochem Biophys ResCommun 1998;248:168–173

25. Trimmer PA, Swendlow RH, Parks JK, et al. Abnormal mito-chondrial morphology in sporadic Parkinson’s and Alzheimer’sdisease cybrid cell lines. Exp Neurol 2000;162:37–50

26. Klunk WE, Jacob RF, Mason RP. Quantifying amyloid beta-peptide (Abeta) aggregation using the Congo red-Abeta (CR-abeta) spectrophotometric assay. Anal Biochem 1999;266:66–76

27. Cassarino DS, Fall CP, Smith TS, Bennett JP Jr. Pramipexolereduces reactive oxygen species production in vivo and in vitroand inhibits the mitochondrial permeability transition producedby the parkinsonian neurotoxin methylpyridinium ion. J Neu-rochem 1998;71:295–301

28. Reers M, Smith TW, Chen LB. J-aggregate formation of a car-bocyanine as a quantitative fluorescent indicator of membranepotential. Biochemistry 1991;30:4480–4486

29. Canevari L, Clark JB, Bates TE. Beta-amyloid fragment 25-35selectively decreases complex IV activity in isolated mitochon-dria. FEBS Lett 1999;457:131–134

30. Citron M, Diehl TS, Capell A, et al. Inhibition of amyloidbeta-protein production in neural cells by the serine proteaseinhibitor AEBSF. Neuron 1996;17:171–179

Khan et al: Cybrids Replicate AD Amyloid Abnormalities 155