22R-Hydroxycholesterol protects neuronal cells from

b-amyloid-induced cytotoxicity by binding to b-amyloid peptide

Zhi-Xing Yao,*,� Rachel C. Brown,* Gary Teper,*,� Janet Greeson� and Vassilios Papadopoulos*,�

*Division of Hormone Research, Departments of Cell Biology, Pharmacology and Neurosciences and �Samaritan Research

Laboratories, Georgetown University School of Medicine, Washington DC, USA

�Samaritan Pharmaceuticals, Las Vegas, Nevada, USA

Abstract

22R-hydroxycholesterol, a steroid intermediate in the pathway

of pregnenolone formation from cholesterol, was found at

lower levels in Alzheimer’s disease (AD) hippocampus and

frontal cortex tissue specimens compared to age-matched

controls. b-Amyloid (Ab) peptide has been shown to be neu-

rotoxic and its presence in brain has been linked to AD

pathology. 22R-hydroxycholesterol was found to protect, in a

dose-dependent manner, against Ab-induced rat sympathetic

nerve pheochromocytoma (PC12) and differentiated human

Ntera2/D1 teratocarcinoma (NT2N) neuron cell death. Other

steroids tested were either inactive or acted on rodent neu-

rons only. The effect of 22R-hydroxycholesterol was found to

be stereospecific because its enantiomer 22S-hydroxycho-

lesterol failed to protect the neurons from Ab-induced cell

death. Moreover, the effect of 22R-hydroxycholesterol was

specific for Ab-induced cell death because it did not protect

against glutamate-induced neurotoxicity. The neuroprotective

effect of 22R-hydroxycholesterol was seen when using Ab1)42

but not the Ab25)35 peptide. To investigate the mechanism of

action of 22R-hydroxycholesterol we examined the direct

binding of this steroid to Ab using a novel cholesterol-protein

binding blot assay. Using this method the direct specific

binding, under native conditions, of 22R-hydroxycholesterol to

Ab1)42 and Ab17)40, but not Ab25)35, was observed. These

data suggest that 22R-hydroxycholesterol binds to Ab and the

formed 22R-hydroxycholesterol/Ab complex is not toxic to

rodent and human neurons. We propose that 22R-hydroxy-

cholesterol offers a new means of neuroprotection against Ab

toxicity by inactivating the peptide.

Keywords: Alzheimer’s disease, neuroprotection, neuro-

steroids.

J. Neurochem. (2002) 83, 1110–1119.

Amyloid b-peptide (Ab), which is produced by proteolyticcleavage of b-amyloid precursor protein, is a major compo-nent of senile plaques and cerebrovascular angiopathy.

Genetic, biochemical as well as histological studies strongly

implicated Ab in the pathogenesis of Alzheimer’s disease

(AD), which is clinically characterized by progressive

cognitive impairment and memory deficit (Selkoe 1989,

1999; Kalaria 1996; Yankner 1996; Christen 2000).

Recent studies have provided evidence that estrogens

protect neuronal cells from the cytotoxicity induced by Ab(Behl et al. 1995; Bonnefont et al. 1998; Liu and Schubert

1998; Pike 1999). Estrogen replacement therapy can reduce

the risk of women developing AD (Tang et al. 1996) and

decreased the symptoms of the disease for older women with

AD (Paganini-Hill and Henderson 1996). On the other hand,

there are reports that neurosteroids, such as pregnenolone and

dehydroepiandrosterone (DHEA), play a physiological role

in preserving and/or enhancing cognitive function (Vallee

et al. 1997) and induce memory improvement in aged

animals (Roberts 1990). These links between steroids, Aband AD led us to investigate the relationship between

neurosteroid biosynthesis and AD. We specifically investi-

gated the levels of various steroids in brain tissues from

Received July 2, 2002; revised manuscript received August 26, 2002;

accepted September 4, 2002.

Address correspondence and reprint requests to V. Papadopoulos,

Division of Hormone Research, Department of Cell Biology, George-

town University School of Medicine, 3900 Reservoir Road, Washington,

DC 20057, USA. E-mail: papadopv@georgetown.edu

Abbreviations used: Ab, b-amyloid peptide; AD, Alzheimer’s disease;CPBBA, cholesterol-protein binding blot assay; 5-cholesten-3b, 22S-diol; cholesterol, 5-cholesten-3b-ol; DHEA, 5-androsten-3b-ol-17-one ordehydroepiandrosterone; 22R-hydroxycholesterol, 5-cholesten-3b,22R-diol; 22S-hydroxycholesterol, 17a-hydroxypregnenolone,5-pregnen-3b,17a-diol-20-one; NT2, Ntera2/D1 teratocarcinoma cells; NT2N,

differentiated human NT2 neurons; pregnenolone, 5-pregnen-3b-ol-20-one.

Journal of Neurochemistry, 2002, 83, 1110–1119

1110 � 2002 International Society for Neurochemistry, Journal of Neurochemistry, 83, 1110–1119

AD and age-matched controls and examined the ability of

steroids whose levels change in AD to protect neuronal cells

from Ab-induced cytotoxicity.We report herein that 22R-hydroxycholesterol, an inter-

mediate in the cytochrome P450 C27 side chain cleavage

enzymatic pathway of pregnenolone formation from choles-

terol (Dixon et al. 1970), was lower in AD patient’s brain

tissue compared to age-matched controls. 22R-hydroxycho-

lesterol was found to have neuroprotective properties against

Ab-induced cell death examined on both human Ntera2/D1teratocarcinoma cells (NT2) and differentiated NT2N neurons

as well as rat sympathetic nerve pheochromocytoma (PC12)

cells. In search of the mechanism of action of 22R-hydroxy-

cholesterol, we found that its neuroprotective property is due

to its ability to bind to Ab peptide, thus rendering it inactive.

Materials and methods

Materials

Ab1)42 and Ab peptide fragments were purchased from Ameri-

can Peptide Co. (Sunnyvale, CA, USA). Polyclonal rabbit anti-

b-amyloid peptide (cat. no. 71–5800) was obtained from Zymed

Laboratories (San Francisco, CA, USA). [22–3H]R-hydroxy-

cholesterol (sp. act. 20 Ci/mmol) was synthesized by American

Radiolabeled Chemical (St Louis, MO, USA). Cholesterol,

22R-hydroxycholesterol, 22S-hydroxycholesterol, pregnenolone,

17a-hydroxypregnenolone and DHEA were purchased from

Sigma-Aldrich (St Louis, MO, USA). Fetal bovine serum and horse

serum were supplied from Biofluids Inc (Rockville, MD, USA). Cell

culture supplies were purchased from Gibco (Grand Island, NY,

USA), and cell culture plasticware was from Corning (Corning, NY,

USA). Electrophoresis reagents and materials were supplied from

Bio-Rad (Richmond, CA, USA). All other chemicals used were of

analytical grade and were obtained from various commercial sources.

Tissue samples

All human tissue samples were obtained from the Harvard Brain

Tissue Resource Center (Belmont, MA, USA). Samples for steroid

measurements were either snap frozen or passively frozen in liquid

nitrogen. Brain hippocampus and frontal cortex samples were

obtained from 19 patients, 12 AD (six men and six women) and

seven age-matched control patients (four men and three women).

AD patients were classified by the Harvard Tissue Resource Center

as having �severe AD�. Mean age for all patients was

74.6 ± 7.2 years for AD patients and 73.4 ± 10.5 years for control.

Mean post-mortem interval was 10.2 h for AD patients and 14.7 h

for control. Protocols for the use of human tissue were approved by

the Georgetown University Internal Review Board.

Purification and measurement of 22R-hydroxycholesterol

Samples were extracted and purified by reverse-phase high-perform-

ance liquid chromatography (HPLC) as we previously described

(Brown et al. 2000). Fractions containing 22R-hydroxycholesterol

were collected (retention time of 22R-hydroxycholesterol ¼ 55 min)

and levels of 22R-hydroxycholesterol were determined using the

cholesterol oxidase assay (Gamble et al. 1978).

Cell culture, cellular toxicity and viability assays

Rat PC12 cells were cultured (50 000 cells/well in 96-well plates) in

RPMI-1640 supplemented with 10% fetal bovine serum and 5%

horse serum as we previously described (Yao et al. 2001). Cells

were cultured for 12 h before addition of the various indicated

treatments. Human NT2 precursor (Ntera2/D1 teratocarcinoma)

cells were obtained from Stratagene (La Jolla, CA, USA) and

cultured (10 000 cells/well in 96-well plates) following the

instructions of the supplier. Differentiated human NT2 neurons

(NT2N) were obtained after treatment of the NT2 precursor cells

with retinoic acid (Andrews 1984). Ab was dissolved in media andused either in the aggregated (left overnight at 4�C) or soluble(containing oligomers such as dimers and tetramers) forms

examined by electrophoresis as we previously described (Yao et al.

2001). Cellular toxicity for Ab and Ab fragments was assayed usingthe 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide

(MTT) assay (Trevigen, Gaithersburg, MD, USA) as we previously

described (Yao et al. 2001). Although this assay has been widely

used to assess cytotoxicity in neuronal cells treated with Ab, itreflects the Ab-dependent vesicle recycling leading to increased

MTT formazan exocytosis and loss (Liu and Schubert 1998). Cell

viability was measured using the trypan blue exclusion method as

we previously described (Yao et al. 2001). In brief, cells were

treated for 72 h with steroids in presence or absence of increasing

concentrations of Ab. At the end of the incubation, the cells werewashed three times with phosphate-buffered saline (PBS) and

incubated for 15 min with 0.1% trypan blue stain solution at room

temperature. After washing three times with PBS, 0.1 N NaOH was

added to the cells and trypan blue staining was quantified using

the Victor quantitative detection spectrophotometer (EGG-Wallac,

Gaithersburg, MD, USA) at 450 nm. Cell protein levels were

determined in the same samples by the method of Bradford

(Bradford 1976), where Coomassie blue staining is detected at

590 nm.

Cholesterol-protein binding blot assay

Purified Ab1)42 protein (50 lM) or various Ab fragments (50 lM)and 3H-22R-hydroxycholesterol were incubated either alone or in

the presence of increasing concentrations of unlabeled 22R-

hydroxycholesterol in 20 lL volume for 8 or 24 h at 37�C. At theend of the incubation time, samples were separated by 1.5% agarose

(Type I-B) gel electrophoresis under native conditions and trans-

ferred to nitrocellulose membrane (Schleicher & Schuell, Keene,

NH, USA) in 10 · saline–sodium citrate (SSC) buffer. The

membrane was exposed to tritium-sensitive screen and analyzed

by phosphoimaging using the Cyclone Storage phosphor system

(Packard BioScience, Meridien, CT, USA). Image-densitometric

analysis was performed using the OptiQuant software (Packard).

This method allows for the separation, visualization and identifica-

tion of Ab complexes, which have incorporated radiolabeled

cholesterol (Yao and Papadopoulos 2002) and 22R-hydroxycholes-

terol (this report) under native conditions. Low molecular weight

unincorporated 22R-hydroxycholesterol is separated and eliminated

during electrophoresis.

Ab aggregation assay

Purified Ab (1–42) protein (50 lM) in cell culture media was

incubated either alone or in the presence of increasing concentrations

22R-Hydroxycholesterol and Ab-induced cytotoxicity 1111

� 2002 International Society for Neurochemistry, Journal of Neurochemistry, 83, 1110–1119

of 22R-hydroxycholesterol for 24 h at 37�C. At the end of the

incubation proteins were separated by sodium dodecyl sulfate

polyacrylamide gel electrophoresis (SDS–PAGE) on 4–20% gradient

acrylamide-bis-acrylamide gel at 125 V for 2 h. Proteins were

visualized by Coomassie blue staining. Ab species were identified byimmunoblot analysis (Yao et al. 2001).

Immunoblot analysis

The membrane with the 22R-hydroxycholesterol-Ab peptide com-

plexes was then used to examine Ab levels. Membranes were

blocked by incubating the nitrocellulose in 5% milk and treated for

immunodetection of Ab using ECL reagents (Amersham-Pharmacia,Piscataway, NJ, USA; Li et al. 2001). Anti-Ab antibody and

secondary antibodies were used at 0.2 lg/mL and 1 : 5000 dilution,respectively.

Peptide modeling and 22R-hydroxycholesterol docking

Computer docking of 22R-hydroxycholesterol with Ab1)42, Ab17)40and Ab25)35 were accomplished using structures generated from the

solution structure of Ab1)40Met(O) (MMDB Id: 7993 PDB Id:

1BA) resulting from data generated by CD and NMR spectroscopy

(Watson et al. 1998). The Met(O) SME 35 residue was replaced by

Met retaining the adjacent backbone dihedral angles and the

co-ordinates for residues 17–40 extracted or the structure was

completed with the addition of the last two amino acids, Ile41 and

Ala42, of Ab1)42. The 22R-hydroxycholesterol structure was

generated using the Alchemy 2000 program (Tripos, St Louis,

MO, USA). The docking was accomplished using Monte Carlo-

simulated annealing as previously described (Li et al. 2001) and

implemented in modified versions of Autogrid/Autodock (Morris

et al. 1998). The conformation of minimum energy of approxi-

mately 109 conformations was evaluated. Seven sessions consisting

of 100 runs, each starting at a random initial relative location and

orientation of the ligand with the target, were executed. Each run

was comprised of 100 annealing cycles using about 2 · 104

improvement steps. The total computation time using the modified

program was about 15 min using a 1.7 GHz, 1GB RAM PC.

Statistics

Statistical analysis was performed by one-way analysis of variance

(ANOVA) and unpaired Student’s t-test using the INSTAT 3.00

package (GraphPad, San Diego, CA, USA).

Results

Figure 1 shows that levels of 22R-hydroxycholesterol in

hippocampus of AD patients’ brain specimens were

decreased by 60% (p ¼ 0.04) compared to age-matched

controls. 22R-hydroxycholesterol levels were also decreased

by 50% in frontal cortex of AD patient’s brain specimens

compared to age-matched controls, although in a non-

significant manner.

The ability of 22R-hydroxycholesterol to rescue rat PC12

neuronal cells from Ab-induced cytotoxicity was examined.PC12 cells were treated for 24 h with increasing concentra-

tions of 22R-hydroxycholesterol in the presence or absence

of increasing concentration of Ab1)42. Cell viability was

measured using the mitochondrial diaphorase assay MTT.

Ab1)42 induced a dose-dependent loss of MTT formazan thatreached 26% (p < 0.001) and 40% (p < 0.001) in the

presence of 25 and 50 lM Ab, respectively (Fig. 2a).

Increasing concentrations of 22R-hydroxycholesterol alone

did not affect PC12 cell viability, although a non-significant

improvement was seen in the presence of 10 and 100 lM of

22R-hydroxycholesterol (Fig. 2a). 22R-hydroxycholesterol

was able to protect all the cells from 25 lM Ab-inducedMTT loss (p < 0.001) and to protect 50% (p < 0.01) of the

cells from losing MTT formazan in the presence of 50 lMAb (Fig. 2a). Interestingly, 22R-hydroxycholesterol was

effective only when present at the same time with Ab. Pre-treatment of PC12 cells with 22R-hydroxycholesterol fol-

lowed by treatment with Ab failed to offer any protection tothe cells (data not shown). The neuroprotective effect of 22R-

hydroxycholesterol could not be replicated using either its

precursor cholesterol (Fig. 2b) or its metabolite pregneno-

lone (Fig. 2c). In contrast, both cholesterol and pregnenolone

alone were toxic to the cells. Moreover, the presence of

cholesterol accentuated the MTT loss seen with low

concentrations of Ab. 17a-hydroxypregnenolone alone wasalso toxic to the cells (Fig. 2d). 100 lM DHEA had a positiveeffect on cell viability. The same concentration of DHEA

protected against the 5 lM (p < 0.001), but not 50 lM,Ab-induced MTT formazan loss (Fig. 2e). The effect of

22R-hydroxycholesterol was stereospecific because

22S-hydroxycholesterol not only failed to protect against

the Ab-induced MTT loss, but at a 100-lM concentration wasneurotoxic (Fig. 2f).

It should be noted that, in the presented studies, aggregated

Ab (left overnight at 4�C) was used. In separate experiments,soluble Ab (containing oligomers) was directly added to

PC12 cells and found to be induce MTT formazan loss (data

not shown). 22R-hydroxycholesterol also protected against

the Ab oligomer-induced MTT formazan loss (not shown).

Fig. 1 22R-hydroxycholesterol levels in AD and control brain speci-

mens. Endogenous 22R-hydroxycholesterol levels in human brain

were measured by the cholesterol oxidase assay after HPLC purifi-

cation. Data presented is means ± SEM for duplicate measurements

from 12 AD and seven age-matched control samples.

1112 Z.-X. Yao et al.

� 2002 International Society for Neurochemistry, Journal of Neurochemistry, 83, 1110–1119

The neuroprotective effect of 22R-hydroxycholesterol was

not restricted to PC12 cells but was replicated on differen-

tiated human NT2N neurons (Fig. 3). 25 lM Ab inhibit by

50% (p < 0.001) human neuron MTT loss, while 1 and

10 lM 22R-hydroxycholesterol protected by 50% (p < 0.01)

and 100% (p < 0.001), respectively, against Ab-inducedMTT formazan loss (Fig. 3). To assess whether

22R-hydroxycholesterol rescues human NT2 cells against

other toxic insults, we treated NT2 cells for 3 days with

5 mM glutamate in the presence or absence of 1–50 lM 22R-hydroxycholesterol. Glutamate induced a 30% decrease in

cell viability, determined using the MTT assay and the

presence of 22R-hydroxycholesterol failed to protect the cells

(data not shown).

Fig. 2 Effect of increasing concentration of

various steroids and steroid intermediates

on rat PC12 neuronal cell viability in the

absence or presence of increasing con-

centration of Ab1)42. PC12 cells were

treated for 24 h with the indicated concen-

trations of Ab1)42 (j, control; n, 0.5 lM; .,

5 lM; e, 50 lM) in the absence or presence

of increasing concentrations of 22R-hy-

droxycholesterol (a), cholesterol (b), preg-

nenolone (c) or 17a-hydroxypregnenolone

(d), DHEA (e) or 22S-hydroxycholesterol (f).

Levels of cell viability were measured

using the MTT assay as described under

Materials and methods. Results shown are

means ± SD (n ¼ 6–12).

22R-Hydroxycholesterol and Ab-induced cytotoxicity 1113

� 2002 International Society for Neurochemistry, Journal of Neurochemistry, 83, 1110–1119

The results obtained from using MTT assay were further

confirmed with the trypan blue dye exclusion assay. Addition

of 1, 5, and 50 lM Ab1)42 for 72 h on either rat PC12 orhuman NT2 cells resulted in the loss of 5, 10 and 25–35% of

the cells, respectively. Figure 4 shows that 22R-hydroxycho-

lesterol rescued both the rat PC12 (Fig. 4a) and human NT2

(Fig. 4c) cells from Ab1)42-induced cell death. In contrast,DHEA only protected the rat PC12 cells from Ab1)42-induced cell death but not NT2 cells (Figs 4a and c). Neither

22R-hydroxycholesterol nor DHEA could rescue the PC12

and NT2 cells from the Ab25)35-induced cell death (Figs 4band d).

We examined the ability of 22R-hydroxycholesterol to

alter Ab aggregation. Figure 5 shows that 22R-hydroxycho-lesterol did not affect Ab aggregation identified by immu-

noblot analysis (Fig. 5b) of the Coomassie blue stained gels

(Fig. 5a). Ab aggregation can be seen on the top of the geland it is absent in control-media lane. A 100-kDa band

recognized by the Ab polyclonal antiserum used in all

samples, including control-media, probably reflects non-

specific binding of the antiserum.

The mechanism of action of 22R-hydroxycholesterol was

then examined. Considering that 22R-hydroxycholesterol

was neuroprotective only when in presence of Ab, weexplored the direct interaction between 22R-hydroxycholes-

terol and Ab. For that we used a novel method, the

cholesterol-protein binding blot assay (CPBBA) method.

Co-incubation of radiolabeled 22R-hydroxycholesterol

together with Ab1)42 for 24 h at 37�C demonstrated the

presence of a high molecular weight radiolabeled band

(Fig. 6a) recognized by an antibody specific to Ab(Fig. 6b). The specificity of the radiolabeling of Ab1)42by 22R-hydroxycholesterol was demonstrated by competi-

tion studies using unlabeled 22R-hydroxycholesterol

(Fig. 6a). In these studies, 50 and 200 lM 22R-hydroxy-

cholesterol inhibited by 50 and 90%, respectively, the

binding of radiolabeled 22R-hydroxycholesterol to 50 lMAb1)42, as indicated by image analysis of the radiolabeledAb1)42 (Fig. 6a). Equal loading of Ab1)42 in the incubationreactions and in CPBBA was assessed by immunoblot

Fig. 4 Effect of 22R-hydroxycholesterol and DHEA on Ab1)42 or

Ab25)35-induced toxicity on rat PC12 neuronal cell and human NT2

cells. PC12 cells were treated for 72 h with increasing concentrations

of Ab1)42 (a) or Ab25)35 (b) in the presence or absence of 100 lM 22R-

hydroxycholesterol (n) or DHEA (s). NT2 cells were treated for 72 h

with increasing concentrations of Ab1)42 (c) or Ab25)35 (d) in the

presence or absence of 25 lM 22R-hydroxycholesterol or DHEA.

Levels of viability were measured using the trypan blue assay as

described under Materials and methods. Results are expressed as fold

trypan blue stained cells per total cell protein over control untreated

cells. Results shown are means ± SD (n ¼ 6–12). j, Control.

Fig. 3 Effect of 22R-hydroxycholesterol on differentiated human

NT2N neuron viability determined in absence (j) or presence (n) of

Ab1)42. Differentiated human NT2N neurons were treated for 72 h with

25 lM Ab1)42 in the presence or absence of 22R-hydroxycholesteol.

Levels of cell viability were measured using the MTT assay as des-

cribed under Materials and methods. Results shown are means ± SD

(n ¼ 6–12).

1114 Z.-X. Yao et al.

� 2002 International Society for Neurochemistry, Journal of Neurochemistry, 83, 1110–1119

analysis of the radiolabeled Ab1)42 (Fig. 6b). It should benoted that, despite the decreased radiolabeling of Ab1)42observed in the presence of 50–200 lM 22R-hydroxycho-

lesterol, there were no differences in the amount of Ab1)42present in each lane. These data demonstrate that, under

native conditions, 22R-hydroxycholesterol binds to Ab.Using CPBBA, and various Ab synthetic peptides, the 22R-hydroxycholesterol-binding site in Abwas mapped to aminoacids 17–40 of Ab (Fig. 6c). Interestingly, peptide Ab25)35,which maintained its neurotoxicity in the presence of 22R-

hydroxycholesterol (Figs 4b and d), did not bind 22R-

hydroxycholesterol (Fig. 6c). In the studies presented

above, radiolabeled 22R-hydroxycholesterol was incubated

with Ab or Ab fragments for 24 h where the binding to theAb polymer is seen. To examine whether 22R-hydroxycho-lesterol could bind to Ab oligomers, shorter time (8 h)

incubations were performed. Figure 6(d) shows that radio-

labeled 22R-hydroxycholesterol binds both to Ab oligomersand polymers.

These data were further confirmed using computational

docking simulations with the entire Ab1)42. The dockingresults show that Ab1)42 forms a pocket in the 19–36 aminoacids area (Figs 7a and d). In agreement with these results,

computational docking simulations using the Ab17)40 aminoacid sequence indicated that this peptide forms a pocket

where 22R-hydroxycholesterol could dock (Figs 7b, c and e).

The orientation R, versus S, is permissive for 22R-hydroxy-

cholesterol docking (data not shown). Similar studies using

Ab25)35 indicated that, despite the presence of some of theamino acids present in the 19–36 area, the docking energy of

Ab25)35 for 22R-hydroxycholesterol (–6.0510 kcal/mol) ishigh relative to Ab17)40 ()8.6939 kcal/mol) and to Ab1)42

()9.6960 kcal/mol), suggesting that this steroid does not bindto Ab25)35 in agreement with the CPBBA data.

Discussion

In brain, neurosteroids (pregnenolone and DHEA) accumu-

late independently of the supply by peripheral endocrine

organs (Baulieu and Robel 1990), act as neuromodulators

(Paul and Purdy 1992) and might serve as pharmacological

tools for various neuropathologies (Costa et al. 1994). Glial

cells can convert cholesterol to pregnenolone. In vitro studies

show that oligodendrocytes, a glioma cell line and Schwann

cells express the cytochrome P450 responsible for the side

chain cleavage of cholesterol and thus pregnenolone forma-

tion (Jung-Testas et al. 1989; Papadopoulos et al. 1992;

Akwa et al. 1993). The P450 side chain cleavage enzyme is

also present in rodent brain (Stromstedt and Waterman 1995)

and in both AD and age-matched control human brain

specimens (Brown et al. 2002). It has been well described

that during this enzymatic reaction one of the three

hydroxylated intermediates formed is 22R-hydroxycholesterol

(Dixon et al. 1970; Hall 1985). 22R-hydroxycholesterol is

more polar than cholesterol and is easily transported through

cell membranes. In the present study, the levels of 22R-

hydroxycholesterol were found to be lower in AD patient’s

brain specimens compared to age-matched controls. Levels

of 22R-hydroxycholesterol were significantly decreased in

hippocampus, a structure in the limbic system of the brain

that is critical to cognitive functions, as learning and

memory, and is affected in AD. We recently proposed that

one of the physiological function of Ab is to control

cholesterol transport (Yao and Papadopoulos 2002). Based

(a) (b)

Fig. 5 Effect of 22R-hydroxycholesterol on Ab aggregation. Purified

Ab1)42 protein (50 lM) in cell culture media was incubated either alone

or in the presence of increasing concentrations of the 22R-hydroxy-

cholesterol for 24 h at 37�C. At the end of the incubation proteins were

separated by SDS–PAGE and visualized by coomassie blue (a). Ab

species formed were identified by immunobloting using an anti-Ab

polyclonal antiserum (b).

22R-Hydroxycholesterol and Ab-induced cytotoxicity 1115

� 2002 International Society for Neurochemistry, Journal of Neurochemistry, 83, 1110–1119

on this finding, it is possible that the decrease of 22R-

hydroxycholesterol might be due to the overproduction of

Ab in AD patient’s brain (Roher et al. 1993; Younkin 1998)

that blocks cholesterol trafficking or decreases cholesterol

uptake by the cells, thus affecting the availability of the

substrate cholesterol for neurosteroid formation resulting in

decreased synthesis of 22R-hydroxycholesterol in AD

patient’s brain. Alternatively, increased de novo synthesis

of pregnenolone and DHEA from cholesterol in AD brain

specimens will also exhaust the available intermediate

(a) (b)

(c) (d)

Fig. 6 Identification of the Ab-22R-hydroxy-

cholesterol binding and binding site by

CPBBA and immunoblot analyses. (a)

Ab1)42 peptide (50 lM) was incubated with3H-22R-hydroxycholesterol (0.1 lCi) in the

presence or absence of increasing con-

centration of unlabeled 22R-hydroxycho-

lesterol in a 20-lL volume for 24 h at 37�C.

CPBBA was performed as described under

Materials and methods. (b) Identification of

Ab1)42 by a polyclonal rabbit antib-amyloid

peptide antiserum on the blot shown in (a).

(c) Identification of the 22R-hydroxycholes-

terol binding site on Ab. Ab peptides (50 lM)

were incubated with 3H-22R-hydroxycho-

lesterol (0.1 lCi) in a 20-lL volume at 37�Cfor 24 h (d) 22R-hydroxycholesterol binding

to Ab oligomers. Ab1)42 peptide (50 lM)

was incubated with 3H-22R-hydroxycholes-

terol (0.1 lCi) in the presence or absence

of increasing concentration of unlabeled

22R-hydroxycholesterol in a 20-lL volume

for 8 h at 37�C. CPBBA was performed as

described under Materials and methods.

Image and densitometric analysis of the

phosphoimage was performed as described

under Materials and methods. Results

shown are representative of four inde-

pendent experiments.

1116 Z.-X. Yao et al.

� 2002 International Society for Neurochemistry, Journal of Neurochemistry, 83, 1110–1119

22R-hydroxycholesterol in AD. We recently reported the

presence of increased levels of pregnenolone and DHEA in

AD hippocampus (Brown et al. 2002), an event induced by

Ab (Brown et al. 2000). It is also possible that both events,Ab-induced decrease in cholesterol trafficking and increasein cholesterol metabolism might occur in AD and lead to

decreased 22R-hydroxycholesterol levels.

For these studies we used the well-established rat PC12

neuronal cell model. However, the neuroprotective effect of

22R-hydroxycholesterol, measured by using the MTT forma-

zan loss and trypan blue staining assays, was not restricted to

rodent neurons but it was also seen in human NT2 and NT2N

neuronal cells. NT2 cells is a clonal line of human

teratocarcinoma cells and NT2N, derived from NT2 cells,

are post-mitotic, terminally differentiated neurons that pos-

sess cell surface markers consistent with neurons of the

central nervous system (Andrews 1984). 22R-hydroxycho-

lesterol was found to protect both rat and human neurons

from Ab-induced toxicity in a dose-dependent manner withIC50 of 10 and 3 lM for PC12 and NT2T cells, respectively.

Treatment of the cells with 22R-hydroxycholesterol offered

full protection against Ab used at 25 lM concentration and

50% neuroprotection against the peptide used at 50 lM.A number of steroids have been tested for their putative

neuroprotective properties against Ab, examined using theamyloid fibril-induced MTT formazan exocytosis assay in

B12 rat neural cells (Liu and Schubert 1998). In these

studies, Liu and Schubert (1998) suggested that compounds

that block amyloid fibril-induced MTT formazan exocytosis,

without affecting that of control cells, should be acting at

upstream targets and thus be neuroprotective. Using the MTT

assay, which measures the formation of blue formazan, we

also examined, in addition to the effect of 22R-hydroxycho-

lesterol, the neuroprotective properties of various steroids

involved in the metabolism of cholesterol. From the steroids

tested on Ab-induced PC12 neurotoxicity, all were toxic

except for 22R-hydroxycholesterol and DHEA. The neuro-

protective effect of DHEA on rodent neurons is in agreement

with previous studies (Kimonides et al. 1998; Cardounel

et al. 2000). However, in contrast to 22R-hydroxycholesterol,

DHEA had no effect on Ab-induced human NT2 cell death,suggesting that the effect of 22R-hydroxycholesterol is not

species-specific, probably because this steroid interacts

directly with Ab. The precursor of 22R-hydroxycholesterol,cholesterol, was found to be neurotoxic. However, the

presence of an hydroxyl group at carbon 22(R) not only

relieves the toxic effect of cholesterol but also protects

against Ab-induced neurotoxicity. The specificity of the

effect of 22R-hydroxycholesterol is further evidenced by the

observation that its enantiomer 22S-hydroxycholesterol is

inactive and at high concentrations neurotoxic.

The findings that 22R-hydroxycholesterol: (i) had no effect

on neuronal cell viability by itself; (ii) failed to protect

against the Ab-induced neurotoxicity when added before Ab;and (iii) failed to protect against the glutamate-induced

neurotoxicity, suggest that its effect is specific for Ab and itcan be exerted only in the presence of Ab. The direct

interaction between 22R-hydroxycholesterol and Ab was

shown using a novel assay, the CPBBA. This assay allows

for the study and visualization of the direct interaction, under

native conditions, between the radiolabeled steroid and Ab,or Ab peptide fragments. Radiolabeled 22R-hydroxycholes-terol binds Ab and the unlabeled 22R-hydroxycholesterol

displaces the bound steroid. CPBBA indicated that 22R-

hydroxycholesterol binds to Ab1)42 and Ab17)40, but barelyinteracts with Ab1)40. Mass spectrometric analysis of

purified amyloid plaques revealed that Ab1)42 is the principalcomponent of amyloid deposits therefore Ab1)42 is believedto be the main culprit in the pathogenesis of AD (Roher et al.

1993; Younkin 1998). The shorter Ab form of 40 amino

Fig. 7 Computational 22R-hydroxycholes-

terol docking simulations to Ab1)42 (a and d)

and Ab17)40 (b, c and e). Simulations were

performed as indicated under Materials and

Methods. Amino acids 19–36 (in blue; f)

form a pocket where 22R-hydroxycholes-

terol could dock.

22R-Hydroxycholesterol and Ab-induced cytotoxicity 1117

� 2002 International Society for Neurochemistry, Journal of Neurochemistry, 83, 1110–1119

acids is believed to have no pathologic effect (Brown et al.

2000) and is less abundant in AD brain (Roher et al. 1993;

Younkin 1998). Computational modeling simulations based

on the reported structure of Ab indicated that amino acids

19–36 capture 22-hydroxycholesterol when the hydroxyl

group of the side chain has the R orientation. Interestingly,

the peptide Ab25)35 that is known for its toxic effects

(Schubert et al. 1995) retained its neurotoxic property even

in presence of 22R-hydroxycholesterol. Computational

modeling simulations and CPBBA failed to show an

interaction between 22R-hydroxycholesterol and peptide

Ab25)35, suggesting that it is the three-dimensional confor-mation of Ab1)42 and Ab17)40 that confers the ability ofamino acids 19–36 to interact with 22R-hydroxycholesterol

rather than the primary amino acid sequence.

22R-hydroxycholesterol binding to amino acids 17–40 of

Ab1)42 leads to the protection/rescuing of both rodent andhuman neuronal cells from the Ab1)42-induced cytotoxicityand cell death. The exact mechanism by which 22R-

hydroxycholesterol acts to block the neurotoxic effect of

Ab is not known. However, the data presented herein

indicated that it does not affect Ab polymerization, and 22R-hydroxycholesterol binds to both Ab oligomers and poly-

mers. Binding of 22R-hydroxycholesterol to Ab1)42 mighteither change the conformation of the Ab monomer or

polymer, thus rendering it inactive, or prohibit Ab from

interacting with the cell or activating intracellular mechanism

mediating its toxic effect. Thus, the low levels of 22R-

hydroxycholesterol in AD patient’s brain compared to age-

matched controls, in addition to the increased production of

Ab1)42 in AD brains, results in decreased/lost ability of the

brain to fight against the Ab1)42-induced neurotoxicity. Thismight be particularly true for presenilin 1-linked familial

Alzheimer’s disease (FAD) patients, who have the highest

levels of Ab1)42 (Borchelt et al. 1996). We believe that wecan take advantage of this finding and use 22R-hydroxycho-

lesterol as a neuroprotective agent.

Current therapeutic strategies for AD include inhibitors of

Ab production, compounds that prevent its oligomerization

and fibrillization, anti-inflammatory drugs, inhibitors of

cholesterol synthesis, antioxidants and neurorestorative

factors (Selkoe 1999; Emilien et al. 2000). 22R-hydroxy-

cholesterol, a naturally occurring intermediate in neuroster-

oid biosynthesis that is lower in AD brain, might represent a

novel therapeutic strategy that works by binding to the

carboxy-terminal fragment and inactivating Ab1)42, thuspreventing the neurotoxicity of both oligomeric Ab and Abfibrils.

References

Akwa Y., Schumacher M., Jung-Testas I. and Baulieu E. E. (1993)

Neurosteroids in rat sciatic nerves and Schwann cells. C. R. Acad.

Sci. III 316, 410–414.

Andrews P. W. (1984) Retinoic acid induces neuronal differentiation of a

cloned human embryonal carcinoma cell line in vitro. Dev. Biol.

103, 285–293.

Baulieu E. E. and Robel P. (1990) Neurosteroids: a new brain function?

J. Steroid Biochem. Mol. Biol. 37, 395–403.

Behl C., Widman M., Trapp T. and Holsboer F. (1995) 17-beta estradiol

protects neurons from oxidative stress-induced cell death in vitro.

Biochem. Biophys. Res. Commun. 216, 473–478.

Bonnefont A. B., Munoz F. J. and Inestrosa N. C. (1998) Estrogen

protects neuronal cells from the cytotoxicity induced by acetyl-

cholinesterase-amyloid complexes. FEBS Lett. 441, 220–224.

Borchelt D. R., Thinakaran G., Eckman C. B., Lee M. K., Davenport

F., Ratovitsky T., Prada C.-M., Kim G., Seekins S., Yager D.,

Slunt H. H., Wang R., Seeger M., Levey A. I., Gandy S. E.,

Copeland N. G., Jenkins N. A., Price D. L., Younkin S. G. and

Sisodia S. S. (1996) Familial Alzheimer’s disease-linked pres-

enilin 1 variants elevate Ab1)42/1)40 ratio in vitro and in vivo.Neuron 17, 1005–1013.

Bradford M. M. (1976) A rapid and sensitive method for the quantitation

of microgram quantities of protein utilizing the principle of protein-

dye binding. Anal. Biochem. 72, 248–254.

Brown R. C., Cascio C. and Papadopoulos V. (2000) Pathways of

neurosteroid biosynthesis in cell lines from human brain: regula-

tion of dehydroepiandrosterone formation by oxidative stress and

b-amyloid peptide. J. Neurochem. 74, 847–859.Brown R. C., Han J., Cascio C. and Papadopoulos V. (2002) Oxidative

stress-mediated DHEA formation in Alzheimer’s disease patho-

logy. Neurobiol. Aging in press.

Cardounel A., Regelson W. and Kalimi M. (2000) Dehydroepiandro-

sterone protects hippocampal neurons against neurotoxin-induced

cell death: mechanism of action. Proc. Soc. Exp. Biol. Med. 222,

145–149.

Christen Y. (2000) Oxidative stress and Alzheimer disease. Am. J. Clin.

Nutr. 71, 621S–629S.

Costa E., Cheney D. L., Grayson D. R., Korneyev A., Longone P., Pani

L., Romeo E., Zivkovich E. and Guidotti A. (1994) Pharmacology

of neurosteroid biosynthesis. Role of the mitochondrial DBI

receptor (MDR) complex. Ann. NY Acad. Sci. 746, 223–242.

Dixon R., Furutachi T. and Lieberman S. (1970) The isolation of crys-

talline 22R-hydroxycholesterol and 20a, 22R-dihydroxycholesterolfrom bovine adrenals. Biochem. Biophys. Res. Commun. 49, 161–

165.

Emilien G., Beyreuther K., Masters C. L. and Maloteaux J. M. (2000)

Prospects for pharmacological intervention in Alzheimer’s disease.

Arch. Neurol. 57, 454–459.

Gamble W., Vaughan M., Kruth H. S. and Avigan J. (1978) Procedure

for determination of free and total cholesterol in micro- or nano-

gram amounts suitable for studies with cultured cells. J. Lipid Res.

19, 1068–1070.

Hall P. F. (1985) Role of cytochromes P-450 in the biosynthesis of

steroid hormones. Vitamins Hormones 42, 315–368.

Jung-Testas I., Hu Z., Baulieu E. E. and Robel P. (1989) Neurosteroids:

biosynthesis of pregnenolone and progesterone in primary cultures

of rat glial cells. Endocrinology 125, 2083–2091.

Kalaria R. N. (1996) Cerebral vessels in ageing and Alzheimer’s disease.

Pharmacol. Ther. 72, 193–214.

Kimonides V. G., Khatibi N. H., Svendsen C. N., Sofroniew M. V. and

Hervert J. (1998) Dehydroepiandrosterone (DHEA) and DHEA-

sulfate (DHEAS) protect hippocampal neurons against excitatory

amino acid-induced neurotoxicity. Proc. Natl Acad. Sci. USA 95,

1852–1857.

Li H., Yao Z., Degenhardt B., Teper G. and Papadopoulos V. (2001)

Cholesterol binding at the cholesterol recognition/interaction ami-

no acid consensus (CRAC) of the peripheral-type benzodiazepine

1118 Z.-X. Yao et al.

� 2002 International Society for Neurochemistry, Journal of Neurochemistry, 83, 1110–1119

receptor and inhibition of steroidogenesis by an HIV TAT-CRAC

peptide. Proc. Natl Acad. Sci. USA 98, 1267–1272.

Liu Y. and Schubert D. (1998) Steroid hormones block amyloid fibril-

induced 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bro-

mide (MTT) formazan exocytosis: relationship to neurotoxicity.

J. Neurochem. 71, 2322–2329.

Morris G. M., Goodsell D. S., Halliday R. S., Huey R., Hart W. E., Belew

R. K. and Olson A. J. (1998) Automated docking using a

Lamarckian genetic algorithm and empirical binding free energy

function. J. Comput. Chem. 19, 1639–1662.

Paganini-Hill A. and Henderson V. W. (1996) Estrogen replacement

therapy and risk of Alzheimer’s disease. Arch. Int. Med. 156,

2213–2217.

Papadopoulos V., Guarneri P., Krueger K. E., Guidotti A. and Costa E.

(1992) Pregnenolone biosynthesis in C6–2B glioma cell mito-

chondria: regulation by a mitochondrial diazepam binding inhibitor

receptor. Proc. Natl Acad. Sci. USA 89, 5113–5117.

Paul M. P. and Purdy R. H. (1992) Neuroactive steroids. FASEB J. 6,

2311–2322.

Pike C. J. (1999) Estrogen modulates neuronal Bcl-xL expression and

beta-amyloid-induced apoptosis: relevance to Alzheimer’s disease.

J. Neurochem. 72, 1552–1563.

Roberts E. (1990) Dehydroepiandrosterone and its sulphate (DHEAS) as

neural facilitators: Effects on brain tissue in culture and on memory

in young and old mice: a cyclic GMP hypothesis of action of

DHEA and DHEAS in nervous system and other tissues, in The

Biological Role of Dehydroepiandrosterone (DHEA) (Kalimi M.

and Regelson W., eds), pp. 13–42. de Gruyter, Berlin.

Roher A. E., Lowenson J. D., Clark S., Wolkow C., Wang R., Cotter R. J.,

Reardon I. M., Zurcher-Neely H. A., Heinrikson R. L., Ball M. J.

and Greenberg B. D. (1993) Structural alterations in the peptide

backbone of beta-amyloid core protein may account for its depos-

ition and stability in Alzheimer’s disease. J. Biol. Chem. 286, 3072–

3083.

Schubert D., Behl C., Lesley R., Brack A., Dargusch R., Sagara Y. and

Kimura H. (1995) Amyloid peptides are toxic via a common

oxidative mechanism. Proc. Natl Acad. Sci. USA 92, 1989–1993.

Selkoe D. J. (1989) Amyloid protein precursor and the pathogenesis of

Alzheimer’s disease. Cell 58, 611–612.

Selkoe D. J. (1999) Translating cell biology into therapeutic advances in

Alzheimer’s disease. Nature 399, A23–A31.

Stromstedt M. and Waterman M. R. (1995) Messenger RNAs encoding

steroidogenic enzymes are expressed in rodent brain. Mol. Brain

Res. 34, 75–88.

Tang M. X., Jacobs D., Stern Y., Marder K., Schofield P., Gurland B.,

Andrews H. and Mayeux R. (1996) Effect of oestrogen during

menopause on risk and age at onset of Alzheimer’s disease. Lancet

348, 429–432.

Vallee M., Mayo W., Darnaudery M., Corpechot C., Young J., Koehl M.,

Moal M. L., Baulieu E.-E., Robel P. and Simon H. (1997)

Neurosteroids: deficient cognitive performance in aged rats

depends on low pregnenolone sulfate levels in the hippocampus.

Proc. Natl Acad. Sci. USA 94, 14865–14870.

Watson A. A., Fairlie D. P. and Craik D. J. (1998) Solution structure

of methionine-oxidized amyloid b-peptide (1–40). Does oxida-tion affect conformational switching? Biochemistry 37, 12700–

12706.

Yankner B. A. (1996) Mechanisms of neuronal degeneration in

Alzheimer’s disease. Neuron 16, 921–932.

Yao Z., Drieu K. and Papadopoulos V. (2001) The Ginkgo biloba

extract EGb 761 rescues the PC12 neuronal cells from

b-amyloid-induced cell death by inhibiting the formation of

b-amyloid-derived diffusible neurotoxic ligands. Brain Res. 889,181–190.

Yao Z. and Papadopoulos V. (2002) Function of b-amyloid in cholesteroltransport: a lead to neurotoxicity. FASEB J. in press.

Younkin S. G. (1998) The role of Ab42 in Alzheimer’s disease. J. Physiol.92, 289–292.

22R-Hydroxycholesterol and Ab-induced cytotoxicity 1119

� 2002 International Society for Neurochemistry, Journal of Neurochemistry, 83, 1110–1119