Platelet-activating factor: a candidate human immunodeficiency virus type 1-induced neurotoxin.

(1)

0022-538X/94/$04.00+0

Copyright X 1994, American Society forMicrobiology

Platelet-Activating

Factor:

a

Candidate Human

Immunodeficiency

Virus

Type

1-Induced Neurotoxin

HARRIS A. GELBARD,l.2* HANS S. L. M. NOTTET,3SUSANSWINDELLS,4 MARTI

JETT,5

KIRK A. DZENKO,1 PETERGENIS,3 ROBERT WHITE,1 LEO WANG,1 YUN-BEOMCHOI,6 DONGXIAN ZHANG,6STUART A. LIPTON,6WALLACE W.TOURTELLOTTE,7

LEON G. EPSTEIN,1'2'8 AND HOWARD E.GENDELMAN3'4'9

ChildNeurology Division, Department of

Neurology,'

and Departments ofPediatrics,2 and Microbiology and Immunology,8 UniversityofRochesterMedicalCenter, Rochester,New York14642; Laboratory ofViralPathogenesis,

Departmentsof Medicine4 and PathologyandMicrobiology3 andEppleyInstituteforResearch in Cancer and Allied

Diseases,9

UniversityofNebraska MedicalCenter, Omaha, Nebraska 68198;DivisionofMolecularPathology, Walter Reed Army InstituteforResearch, Washington, D.C.s;Departmentsof Neurology, Children's BethIsrael, Brighamand

Women's, and Massachusetts GeneralHospitals, Harvard MedicalSchool, Boston, Massachusetts021156; and NationalNeurological Research Specimen

Bank,

Wadsworth and BrentwoodDivisions, VA MedicalCenter,

LosAngeles, Califomia 900737 Received 8February1994/Accepted 12April 1994

Thepathogenesis of centralnervoussystem diseaseduringhumanimmunodeficiencyvirus type1 (HIV-1) infection revolves aroundproductiveviralinfection ofbrain macrophages and microglia. Neuronallosses in thecortexandsubcortical gray matteraccompanymacrophageinfection. Thequestionof how viral infection of brainmacrophages

ultimately

leadstocentral nervous system (CNS) pathologyremains unanswered. Our previousworkdemonstratedhigh-level production oftumornecrosis factoralpha, interleukin 1 ,arachidonic acid metabolites, and platelet-activating factor (PAF) from HIV-infected monocytes and astroglia (H. E. Gendelman,P. Genis,M.Jett,and H. S.L. M.Nottet,in E.Major, ed., Technical Advances in AIDS Research in the Nervous System, in press; P. Genis, M. Jett, E. W. Bernton, H.A. Gelbard, K. Dzenko, R. Keane, L. Resnick, D. J.Volsky,L.G. Epstein, and H. E.Gendelman, J. Exp. Med. 176:1703-1718, 1992).Thesefactors, together, were neurotoxic. The relative role(s) of eachof these candidate neurotoxins in HIV-1-related CNS dysfunction was not unraveled by these initial experiments. We now report that PAF is produced during HIV-1-infected monocyte-astroglia interactions. PAFwas detected at high levels in CSF of HIV-1-infected patients withimmunosuppression and signs of CNS dysfunction.Thebiologic significanceof the results for neurological disease wasdetermined by addition of PAF to cultures ofprimary human fetal cortical or rat postnatalretinalganglion neurons. Here, PAFatconcentrations of -300pg/ml producedneuronal death. The N-methyl-D-aspartate receptor antagonist MK-801 or memantine partially blocked theneurotoxic effects of PAF.Theidentification of PAFas anHIV-1-induced neurotoxin providesnewinsights into how HIV-1causes neurological impairmentand howit mayultimately be ameliorated.

Human immunodeficiency virus type 1(HIV-1) infection of the central nervous system (CNS) results in cognitive and motor abnormalities in a majority of infected individuals. HIV-1-associated dementia or AIDS dementia complex (ADC), adevastating complication of direct viral infection of the brain, arguably represents one of the most severe and significant clinical manifestations of HIV infection (6, 13, 16, 26,43). The pathological hallmarks coincident with cognitive and motor dysfunctions include HIV-1 infection of brain macrophages, microglia, and multinucleated giant cells; astro-cyte proliferation; and neuronal loss in discrete areas of the retina,neocortex,and subcortical brain (14, 18, 29, 40, 49, 50, 53, 59,60).Interestingly, this disease complex is most prevalent invirus-infected children (6).

The molecular mechanisms governing neurologic dysfunc-tionduringHIV-1infection of the brain remain an enigma. No laboratory has convincingly demonstrated HIV-1 mRNA or p24 antigens in neurons or oligodendrocytes within virus-infected brain tissue. How does HIV-1 cause neuronal loss

*Corresponding author. Mailing address: Division of Pediatric Neurology, Box 631, University of Rochester Medical Center, 601

Elmwood Ave., Rochester, NY 14642. Phone: (716) 275-4784. Fax: (716)275-3683.

when neurons are not infected and perhaps only a small number ofmacrophages support productive viral replication (13, 17,40)? Indeed, secretory productsfrom HIV-1-infected macrophages are the likely source of neurotoxic activities. Neuronal deathcanoccur soonafter exposuretoculturefluids from HIV-1-infected monocytes. Monocyte production of neu-rotoxic factors appears to require HIV-1 infection (21, 47), monocyte activation, and astroglia and/or neuronal monocyte interactions(2,12, 17, 20,52).HIV-1 gene productsalso affect neuronalviability. Picomolar amountsofHIV-1 gpl20added tomixedmicroglial-neuronal cells in thepresence of sublethal concentrations ofglutamate are toxic to rat retinal ganglion cells (10, 11,33-35). Thistoxicity is associated with increased neuronal

Ca2'

and is reversed by calcium channel or N-methyl-D-aspartate (NMDA) antagonists (33-35).

Ourrecent workhas demonstrated that coculture of HIV-1-infected monocytes and human astrocytoma cell lines in-duces cytokines, arachidonic acid metabolites, and platelet-activating factor (PAF) from virus-infected monocytes (20). Astroglia cells serve to activate macrophages for their produc-tion of these neurotoxic factors (17). Our initial studies, however, didnotprovide evidence that anyorall of the factors areneurotoxinsordetermine whetherthey playabiologically important role in neurological disease. The purpose of the 4628

on November 9, 2019 by guest

http://jvi.asm.org/

(2)

HIV-MEDIATED NEUROTOXICITY 4629

present study was to identify PAF as a candidate HIV-1

neurotoxin and to provide a conceptual framework for how

"putative"HIV-1 neurotoxinsare identified.PAFis produced

within cocultures of HIV-1-infected monocytes and astroglia and ispresentinthe cerebrospinal fluid (CSF)of HIV-infected patients during neurologic dysfunction. PAFaddedtoprimary human orrat neurons at levelsapproximating those found in the CSF of HIV-1-infected patientsproduces dose-dependent neurotoxicity. NMDA receptor uncompetitive antagonists MK-801 and memantine (4, 22, 35, 39) block PAF-mediated neurotoxicity. These findings, taken together, provide insights into the pathobiology of ADC and newideas for therapeutic

interventions for neurological impairments associated with HIV-1 infection.

MATERIALS ANDMETHODS

HIV-1 infection ofmonocytes. Monocytes were recovered

fromperipheral blood mononuclear cells ofHIV-and hepatitis B-seronegative donors after leukapheresis and purified by

countercurrent centrifugal elutriation. Cell suspensions were

>98% monocytes by criteria of cell morphology on

Wright-stainedcytosmears,by granular peroxidase,andby nonspecific

esterase (15, 19). Cellswere cultured asadherentmonolayers

(106/ml in 24-mm-diameterplastic culture wells) in Dulbecco modified Eagle medium (Sigma, St. Louis, Mo.) with

recom-binant humanmacrophage colony-stimulatory factor (Genetics Institute, Cambridge, Mass.). The human brain astroglia

tu-mor-derived cell line U251 MG, a generous gift from D.

Bigner,wasutilized in cocultivationassayswith HIV-1-infected

monocytes as described previously (20). These cells were

grown as adherent monolayers in Dulbecco modified Eagle medium (Sigma)-10% heat-inactivated fetal calf serum

(Sig-ma)-50 ,ugofgentamicin perml.

Monocytes were exposed to HIV-1ADA (accession number M60472) (42) at a multiplicity ofinfection of 0.01 infectious

virions pertarget cell (19, 57). The viral inoculawere free of

mycoplasma contamination (Mycoplasma Detection Kit III;

Geneprobe, San Diego, Calif.). Human macrophage colony-stimulating factor-treated monocyteswere cultured as

adher-entmonolayers 7 to 10 days prior to use asviral targetcells. Under these conditions, 10 to 20% of the monocytes are

productivelyinfected 7days after HIV-1 inoculation (27).All cultures were refed with fresh medium every 2 to 3 days. Reversetranscriptase activitywasdetermined in culture fluids

added to a reaction mixture of Nonidet P-40 (Sigma), poly(rA)-oligo(dT) (Pharmacia, Piscataway, N.J.), dithiothrei-tol(Pharmacia), MgCl2,and [oa-32P]dTTP(400 Ci/mmol; Am-ersham Corp., Arlington Heights, Ill.) for 24 h at 37°C. The mixture wasapplied to chromatographypaper, airdried, and washed five times in 0.3 M NaCl-0.03 M sodium citrate (pH 7.4) and twice in 95% ethanol. The paperwasdried and cut,

and the radioactivity wascountedby liquid scintillation

spec-troscopy (27). Five to 7 days after infection and during the peak ofreverse transcriptase activity (107 cpm/ml)in

HIV-1-infected monocytes, equal numbers of astroglial cells, U251 MG, were added and cell lysates were recovered for PAF

determinations.

PAFassay.Cultured cellsorCSFsampleswerefrozenindry ice-ethanol mixtures. CSF samples were frozen immediately after procurement frompatients. Cellsampleswere extracted

within 1 week offreezingtopreservetheintegrityof thelipids, whichare generallyunstableinaqueoussolutions.The

extrac-tionprocedurewasperformedwithaC18disposable cartridge (Waters, Milford, Mass.) as previously described (18). PAF

was eluted as a separate fraction, with approximately 70%

TABLE 1. PAFlevels in CSF of HIV-1-infectedpatients with or

without neurological impairment

Patient ADC No. of CD4+ Mean PAF concn no. Age stage' T cells (pg/ml)± SEM

1 7mo 1 480 22

2 10 yr lb 182 32

3 7yr ob 20 61

4 47yr 1 530 124 17

S Syr 3b 256 152 6

6 61 yr 0 355 157 13

7 37yr 0 116 192 27

8 35yr 0 446 195 11

9 48 yr 2 4 197 73

10 41yr 2 12 198 20

11 34yr 1 374 232 62

12 42yr 4 4 241 22

13 31yr 2 258 245+32

14 23yr 0 320 260 2

15 31yr 1 157 287 35

16 17mo 3b 85 299 24

17 25 mo 3b 354 322 51

18 37 yr 3 6 335 20

19 34 yr 3 20 343 56

20 45yr 2 58 353±151

21 42yr 3 12 398 17

22 35yr 3 45 624 45

23 34yr 1 50 853 248

aNeurologic dysfunction (ADCstage) (46a) ratingwasdonein accordance

with theguidelinesofthe AmericanAcademy of Neurology Task Force on AIDS

(26).

bPediatric patientswereassigned"best-approximation" ADC stages to con-form with the adult data. All of the subjects were male.

recovery,based on inclusion of internal standards of

[3H]PAF

during extraction. The material was dissolved in 1.2 ml of methanol and aliquoted into 0.1-, 0.4-, and 0.6-ml fractions. These were dried by vacuum centrifugation (Savant Instru-ments, Inc.,Farmingdale, N.Y.), and PAF was quantitated by radioimmunoassay (E.I.Du Pontde Nemours & Co., Boston, Mass.).The detectable levels ofPAFin these assayswere30to 1,200pg/ml (20).

Patient material. Forty-four patients were clinically evalu-atedattheUniversityofNebraska, theUniversity of Medicine and Dentistry of New Jersey, the Wadsworth VA Medical Center (UniversityofCalifornia Los Angeles), and the Mas-sachusetts General Hospital. Patients were retrospectively evaluated. Nine HIV-1-seropositive patients had complete postmortem examinations. The HIV-positive patients ranged in age from7monthsto61years(Table1).Twenty-one control patients were between 13 and 78 years old (Table 2). The HIV-seropositive patients were classified as to severity of neurologic dysfunction in accordance with ADC guidelines (46a) and those established by the American Academy of Neurology TaskForce onAIDS (26).

Humanneuronal cellcultures. Humanfetalbrain tissuewas obtained from elective therapeutic abortions under thestrict ethicalguidelinesof theNational Institutes of Health and the University of Rochester Medical Center. Neurons were ob-tainedfrom thetelencephalonwith both cortical and ventric-ularsurfaces ofsecond-trimester

(13

to 16weeksof

gestation)

humanfetal brain tissuebyamodification of the

procedure

of Banker andCowan (1).Briefly,brain tissuewascollected and washedin30 ml of coldHanks balanced salt solution

(contain-ing

Ca2+, Mg2+,

HEPES

[N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic

acid],

and 50 ,ug of

gentamicin

per

ml).

Brain tissue was separated from adherent

meninges

and blood and

VOL. 68, 1994

on November 9, 2019 by guest

http://jvi.asm.org/

(3)

[image:3.612.67.305.100.335.2]

TABLE 2. PAFlevelsin CSF of non-HIV-1 disease and controlsubjects

Patient Age . Mean PAFconcn

no. (yr)/sex' Systemic disease (pg/ml) ±SEM

1 38/F Noneb 30 ± 4

2 14/F None 32+7

3 78/F None 37 ± 12

4 61/F MSC(minimal deficits) 64 ± 13

5 49/F Localized melanoma 68 ± 7

6 56/M None 97+ 4

7 71/M None 97 + 13

8 72/M Myocardial infarction 111 ± 10

9 30/M None 121 ± 13

10 13/M Leukemia(CNS 125 ± 8

involvement)

11 36/M TIAd 144 ± 21

12 54/M TIA 160

13 32/F MS(minimal deficits) 168 ± 26

14 79/M Pneumonia 172 ± 24

15 93/M ProstaticCae 190 ± 16

16 61/M Prostatic Ca 213±40

17 44/M Hepatic failure 235±79

18 4/M Leukemia(CNS) 366±55

19 57/M Nasopharyngeal Ca 417 ±53

20 22/M Diffusehistiocytic 709± 201

lymphoma

21 58/M Metastatic melanoma 825 + 446

a F, female; M,male.

bForsubjectswithoutsystemicdisease, lumbarpunctures wereperformedto

excludemeningitis. Patientspresented withheadache, fever, and/orweakness.

CSFexaminations showed normalproteinandglucoselevels andcellcounts.

c MS,multiple sclerosis.

dTIA,transient ischemicattack.

'Ca, carcinoma.

cut into 2-mm3 pieces. Dissociated tissue was transferred to andthenforcedthrougha230-jimNitexbag.Thedissociated tissuewasgentlytriturated and thencentrifugedat100xgfor 5 min at 4°C. The pellet was resuspended in 5 to 10 ml of MEM-hipp (2 mM D-glucose, 10 mM HEPES, 1 mM Na pyruvate,20 mMKCI) containingNicomponents(insulinat5 mg/liter,transferrinatS mg/liter,seleniteat5 ,ug/liter, proges-teroneat 20nM, andputrescine at 100 ,uM),aswell as 10% fetal calf serum, PSN antibiotic mix (penicillinat 50mg/liter, streptomycinat50mg/liter,andneomycinat100mg/liter),and amphotericin B(Fungizone;2.5 mg/liter). The cellcountand viability were determined by diluting Hanks balanced salt solutionwith0.4%trypanblue (1:1, vol/vol)andcountingwith ahemocytometer.Cellsweregentlytriturated fivetimes witha 10-ml pipet, plated at a densityof 105/12-mm-diameter glass coverslipprecoated withpoly-L-lysine (70K-1SOKMW;Sigma), andplacedin 24-wellculture dishes. Cells were cultured for 10 to28daysat37°Cin a humidified atmosphere of 5%C02-95% air,andthe medium was changed every 3 days (7).

Under these conditions, neuronal cultures were >70% homogeneousas determined by anti-human PGP 9.5 staining (Ultraclone, Ltd., Wellow, Isle of Wight, England) (28) and anti-MAP-2 staining. Glial fibrillary astrocyte protein staining-positiveastrocytes made up25% of the total cell population. Microglia-macrophages made up less than 5% of the popula-tionas determined by RCA-1 lectin and CD68 staining.

Neuronal cultures maintained for >28 days under the conditions described above expressed NMDA receptors. Neu-rons exposed to 300

jiM

NMDA for 10min were reduced in number to20% of that of untreated neurons. This neurotox-icitywasblocked by coincubation with 10

jiM

MK-801 during exposure to NMDA (data not shown). Evaluation of

PAF-induced humanneurotoxicitywasperformedwith

1-O-hexade-cyl-2-O-acetyl-sn-glycero-3-phosphorylcholine

PAF

(Biomol

Research

Laboratories,

Inc.,

Plymouth Meeting, Pa.).

Coverslips

with human neurons were fixed with 4% paraformaldehydeat48 h after exposuretoculture fluids from HIV-iADA-infected monocytes and/or astroglial cells. Cells were incubated with Neurotag Red

(Boehringer Mannheim,

Indianapolis, Ind.)

tostainsomasand neuritic processes. Cell morphologyofneurons wasexamined

by

fluorescence micros-copy. In

replicate

experiments,

neurons were evaluated as described aboveat48 h after exposuretoPAFwithorwithout MK-801. Cellswere

immunocytochemically

stained with PGP 9.5 antiserum. To assay

neurotoxicity,

examiners

(K.A.D.,

H.A.G.,and

L.G.E.)

wereblindedtothe neuronaltreatments. All assayswere

performed

in

triplicate

determinations.

Digi-tized

images

of PGP 9.5-stained neurons in .15

microscopic

fieldswere

analyzed

for numbers of intact neuronalsomasper 5OX field with a densitometer

(Imaging

Research

Inc.,

St.

Catharines,

Ontario,

Canada).

Datawere

expressed

as mean

neuronal cell counts ± the standard errors of the means

(SEM).

Tests of statistical

significance

of the differences between control and

experimental

treatments were deter-mined

by paired

ttests.

Ratneuronal cell cultures. Ratretinal

ganglion

cells from 7-to

10-day-old

postnatal Long-Evans

rats werelabeled insitu

by

retrograde

transport of the fluorescent

dye granular

blue. Since

ganglion

cellsarethe

only

cellsto

project

from the retina to deeperbrain structures, this

labeling

was

accomplished by

injectionof

granular

blue

(as

a1% solution in

saline)

into the

superior colliculus;

maximum

labeling

of retinal

ganglion

cells occurred within 2

days

of the

injections,

so theretinas were harvestedatthis time after therats weresacrificed

by

cervical dislocation

(36).

With the

retrograde

labeling technique,

the

ganglion

cell neurons could be

specifically

identified

by

the presence of the bluefluorescent

label,

even

following

dissoci-ation from the retina and

placement

in

culture,

as

previously

described

(31, 36).

The

resulting

retinal cultures contained mixed neuronal and

glial

elements. These cells were

plated

onto

glass

coverslips

coated with

poly-L-lysine

and culturedin Eagle's minimumessential medium

supplemented

with 0.7%

(wt/vol)

methylcellulose, 2 mM

glutamine,

1 mgof

gentamicin

per

ml,

16 mM

glucose,

and 5%

(vol/vol)

rat serum. This culturemediumcontained 1.8mM

CaC12

butno

MgCl2,

since at

negative

membrane

potentials magnesium

is knowntoblock NMDA

receptor-operated

channels andtherefore

might

have partially masked NMDA

receptor-operated

channels and NMDA

receptor-mediated neurotoxicity

(4,

22,

39,

44).

In

parallel experiments, however, qualitatively

physiologic

concentration of

Mg2+

(0.8

mM). Replicate

cultures received PAF with or without the NMDA

antagonist

MK-801

(dizocilpine)

ormemantine. Ret-inalcultures treated in thismanner were incubatedovernight at

37°C

in ahumidifiedatmosphereof5%

C02-95%

air,and then

viability

was assessed as described below. For these

experiments,

carbamyl-PAF

(c-PAF;

Biomol Research

Labo-ratories, Inc.)

wasusedto avoidrapidmetabolism ofPAFby tissue

hydrolases (45).

The existence of PAF degradation products might have obfuscated the results of the neuronal assays.Inthis respect,wewouldnothave known ifPAFitself or ametabolitewas aneurotoxic factor in this system. c-PAF hasapharmacologic profilesimilartothat ofPAFbut is 5-to 10-fold less potent in binding and physiological studies of membrane surface receptors for PAF (45). Previous studies

using high-pressure liquid chromatographic analysis

showed that these retinal cultures contain

approximately

25

jiM

glu-tamate

(an

NMDA receptor

agonist)

and 15

jiM glycine (an

on November 9, 2019 by guest

http://jvi.asm.org/

(4)

NMDA receptor coagonist), but by themselves these levels of the amino acids are not neurotoxic in this system (35).

The ability of retinal ganglion cells to take up and cleave fluorescein diacetate to fluorescein was used as an index of their viability and lack of injury, as described previously (17, 31). Under a combination of phase-contrast microscopy and UV epifluorescence, retinal ganglion cells are the only cells that appear blue with a standard 4',6-diamidino-2-phenylin-dole filter set because of the presence of retrogradely trans-porteddye; viablecells, visualized with a fluorescein filter set, appearyellow-green. Thus, viable retinal ganglion cell neurons appear blue with one fluorescence filter set and yellow-green withasecond, facilitating their rapid scoring. This test appears tobe a more sensitive indicatorof subtle forms of cell injury than dye exclusion studies, which have been used to monitor cell death inmanystudies.Results werereplicated in triplicate dishes on at least four separate occasions. Viable rat retinal ganglion cellswerescored ineach culture dish over an area of 75 um

2.

RESULTS

PAFdetected in HIV-infected monocyte-astroglial mixtures. Our previous work demonstrated that tumor necrosis factor alpha (TNF-oz), interleukin 1B, arachidonic acid metabolites, and PAF are produced during interactions between HIV-infected monocytes and astroglia tumor cell lines. These factors predict neurotoxicity (20). The contribution that each of these factors made toneurotoxicitywas not demonstrated. Therefore,weperformedaseries of experimentsto determine whetherPAF isaneurotoxin associated with HIV-1 infection andneurological disease.

PAF levelswere examined by radioimmunoassay in HIV-1-infected cell cultures. Monocytes were infected with HIV-lADAat amultiplicity of infection of0.01 for7days then added in a 1:1 cell-to-astroglia ratio (U251 MG cells). Methanol extracts of cells were prepared during monocyte-astroglial cocultureat 1,3, 6, 12, 20, and 30min.Asshown inFig. 1, high levels of PAF were selectively observed (3 to 20 min) in cocultures of HIV-infected monocytes and astroglia. PAF concentrations were significantly lower (i) in HIV-1-infected monocytes or (ii) in uninfected monocytesin the presence or absence ofastroglia. PAFcould notbe detected reproducibly atlonger time intervals(hourstodays) (datanotshown).This mayhave beenduetothe temporal regulationofPAFandits instability in culture solutions for longer time intervals(3, 8,9, 20). Alternatively, PAF may be synthesized only after the initial cell-to-cell contact between HIV-1-infected monocytes and astroglia.

Measurement of PAF in CSF of HIV-1-infected control subjects. CSF samples from 23 patients with documented HIV-1 infection and 21 controlswere collected for

measure-mentofPAF levels (Tables 1 and2). Neurologic dysfunction, defined by the clinicalparameters oftheADC,wasscoredon

a scale of 0 to 4. Immunosuppression was measured by determining numbers ofCD4+ Tcells inblood. Six of the 23 patientswerechildren,with agesrangingfrom7monthsto10 years. Inthepediatricgroup, threeofsixchildren had elevated PAF levels (152 to 322 pg/ml),

correlating

with

neurologic

dysfunction and immunosuppression

(Fig.

2 and Table

1).

In adultpatients, however,elevatedPAFlevels

(124

to853

pg/ml)

correlatedin 17of23 patientswith theonsetbutnot progres-sion ofneurologic

dysfunction (ADC

stages 1to

4).

PAF levels inCSF from this groupofpatientswere

significantly

different fromthose inCSFfrom control patients (P<

0.0002).

Five of 23patientswithno

neurologic dysfunction

(ADC

stage

0)

had

260

195

I..

1-1

, 1 30

0c

65

-0

t'

I

,~~~~~%

.';

,~~~~~%

v~~~~~%

v~~~~~~~~%

8 16 24 32

Time (minutes)

FIG. 1. Monocytes cultured for 7 days as adherent monolayers wereexposed to

HIV-lAI1A

at a multiplicity ofinfection of 0.1. At 7 days after infection, HIV-infected or control uninfected monocytes werecocultured 1:1 (monocytes-astroglial cells). At various times (in minutes) after coculture, HIV-infected monocytes, uninfected mono-cytes, ormonocyte cocultures werefrozen on dry ice in ethanol and then celllysateswerecombined with methanol washes of broken cells and extracted. PAF was eluted withapproximately 70% recovery and

quantitated by radioimmunoassay. The detectable levels ofPAF in these assayswere30to1,200pg/mI.Levels ofPAFareshownasclosed diamonds (HIV-1-infected monocytes and astroglial cells), closed circles(uninfectedmonocytes andastroglial cells),open boxes

(HIV-infected monocytes), and open circles (uninfected monocytes). The

errorbarsrepresent standard deviations.

low levels of PAFinCSF (61 to 195 pg/ml)(Fig. 3 and Table 1). Elevation of PAF in CSF was not specific for HIV-1-infected individuals. PAF was detected in patients with a variety of other medical conditions (multiple sclerosis, cere-brovascular disease, leukemia, disseminated cancer, and he-patic failure) in whom CNS symptoms (changes in cognitive and/ormotorfunctions) werepresent. Patientswith paraneo-plastic syndromes that included CNS dysfunction (Table 2, patients 18 to 21), interestingly, had the highest PAF levels (366 to 825 pg/ml) ofthe non-HIV patients. This includes a

single leukemic patientwith CNS involvement. PAF levels in the CSF of this group of patients were also significantly differentfrom those in the CSF of controlpatients (P<0.006; Fig. 3).

Neurotoxicactivityof PAF.Couldtheobservedelevation of PAF in CSFmean that it plays an important role(s) in CNS dysfunction?Totestthistheory,we

exposed primary

neuronal fetal cultures to PAF at concentrations equal to andgreater thanthose present in coculture fluids. PAF(50to6,000 pg/ml) appliedtoprimaryfetalcultureswastoxictohumanneuronsin

a dose-dependent manner within 48 h

(Fig. 4).

At PAF concentrations of300 and

1,250 pg/ml,

neuronal cell counts

were reduced to 54 and

28%,

respectively,

of the control (untreated neurons) value. The PAF-induced

neurotoxicity

VOL.68, 1994

0t

on November 9, 2019 by guest

http://jvi.asm.org/

[image:4.612.324.559.81.338.2]

(5)

300-1

0

I T

0

normal ADC stage 0 ADC stage 1-4

FIG. 2. Box plots of PAF levels in CSF obtained from control

(HIV-1-seronegative) patients without systemic disease (n = 6),

HIV-1-seropositive patientswithout evidence ofneurologic dysfunc-tion (ADC stage 0; n = 5), and HIV-1-seropositive patients with

neurologic dysfunction (ADCstages 1 to4;n = 18). Comparisons by

t test: controlversusADC stage0group,P< 0.025; control versus

ADCstage1to4group,P<0.0002;ADC stage 0groupversusADC

stage1to4group,P<0.05. Because threecomparisonsweremade, the level ofsignificancewasset ata=0.05/3 =0.017.Therefore, only

the difference between control patients and the ADC stage 1 to 4

group wassignificant.

4Ji

03

0 0

z

250

-200

150

100

50

o0

*

_~~~

0 50 300

PAF (pg/mi)

1250

FIG. 4. Dose response of human fetal neuronal cell cultures to

PAF. Culturesweregrownoncoverslipsasdescribedin Materialsand

Methods.Human cortical neuronalcell cultureswereincubated for48

hwith thevehicle aloneorwithincreasingdosesof PAF(50, 300,and

1,250pg/ml [97 pM,557pM,and 2.3nM, respectively]).The cultures

were washed, fixed in paraformaldehyde, and subjected to PGP 9.5

immunocytochemical labeling.Neuronal cellcountingwasperformed asdescribedinMaterials and Methods. Observerswereblinded tothe

treatment groups. Each neuronal cell count shown represents the

meanof .15microscopicfields selected from fourreplicates± SEM.

Symbols: *,P < 0.0001, control versusPAF at 300 pg/mI; +,P <

0.0001,controlversusPAFat1,250 pg/ml.

was partially blocked by coincubation with 10 ,uM MK-801 (Fig.5A).The numbers ofneuronstreated with both PAF and MK-801 were 80% of the numbers of untreated control neurons.

Similar results were obtained with rat postnatal retinal cultures. Retinalganglioncellswerespecifically depletedfrom cultures containing c-PAF. Assay ofrat retinal ganglion cell viabilitywasbasedupontheability of healthyneuronstotake

up and cleave fluorescein diacetate to fluorescein. Retinal ganglioncellswerespecificallylabeledby retrogradetransport

ofafluorescentdye injectedinto thesuperiorcolliculus. Viable

retinalganglion cellsweredual labeledwithgranular blueand

fluoresceinyellow.A 35% reduction ofneuronalviabilitywas

1000

8001

.

600-

U-<

400-200

u

0

[image:5.612.321.552.77.219.2]

systemic diseases normal

FIG. 3. Boxplot of PAF levels in CSF ofcontrol (HIV-1-seroneg-ative) patients without systemic disease (n = 6)and patients (HIV-1

seronegative) witha variety of systemic diseases (metabolic, cancer, andothers;n = 15). Comparisonbyttestof control patientsversus

patientswithsystemic disease, P<0.006.

observed following addition of c-PAF and was largely

pre-ventedby the specificNMDAantagonist MK-801 or

meman-tine (each at6 ,uM) (Fig. 5B). Thus, PAF-induced neurotox-icity was seen in both human and rat neuronal cell culture

systems.

DISCUSSION

Productive HIV infection of the brain occurs in brain

macrophagesandmicroglia. Cognitive andmotordysfunction

occurs, as well as neuronal loss in infected brain tissue.

HIV-1-infectedbrain macrophagesandmicrogliamayleadto HIV-1-associated dementia, but neurologic deficits may not directly correlate with productive viral infection. A discor-dance exists between the level ofhistopathologic changes and the small numbers of detectable HIV-infected cells. This

supports the notion that diffusible virus-induced neurotoxins

areresponsible, inpart,for theneurological manifestations of HIV-1infection.Moreover,thishypothesis remains likelyeven

if farmoremicroglialcells andastrocytes(48)areinfected with

HIV-1 and low-level or restricted infection results in local

neurotoxic activities. Herewedemonstrate thatPAFisoneof

possibly many HIV-induced neurotoxins. This was shown by

severalassaysystems.First,wedetected PAF inan

experimen-tal model systemforHIVCNS disease. The levels ofPAFin

ourlaboratorysystemwerehigh butnotsustained, supporting thenotion of itscyclical production. Second,weconfirmed the

biological relevance ofour in vitro observations by assay of

PAF in the CSF of HIV-infected patients. Although PAF elevations were not specific for HIV infection and CNS disease, they did correlate with the onset of neurological impairment. Finally, we inoculated PAF, in concentrations

similartothose found inCSF,intoprimaryneurons.Here PAF

was a potent neurotoxin. These findings, taken together, stronglysuggestthat PAFplaysarole inADC. Moreover,the

dataprovidetheidentityofasolubleneurotoxin, distinct from

aviralgeneproduct, produced by HIV-infected cells.

PAF is alipid mediatorthat initiates a large repertoire of

1000

-800

-E

600--

400-0.

200

-

0-..:.."

71

L-1 .. t...'iiiiLl

on November 9, 2019 by guest

http://jvi.asm.org/

[image:5.612.60.295.77.238.2] [image:5.612.65.295.520.681.2]

(6)

A

200-c

0

-a

c

0

z

150

-100

-50

0

+

Control

II

c

0

coCD=

0

c 4-. -C

0

4-0 .0

B

120- 100- 80- 60- 40- 20-

o0-*

PAF PAF/MK-801

77 --F~1

[image:6.612.57.299.65.354.2]

PAF PAF/MK-801 PAF/memantine

FIG. 5. Neurotoxic effects of PAF onhuman fetal neuronal cell

cultures(A)andratpostnatalretinalganglioncells inculture(B)with

partialblockade ofneurotoxicity byNMDAantagonists.Cultureswere grown on coverslips as described in Materials and Methods. (A)

Humancorticalneuronal cultureswereincubated with PAF for48 h.

The cultures were then washed, fixed in paraformaldehyde, and

subjected to PGP 9.5 immunocytochemical labeling. Neuronal cell

counts were determined as described in Materials and Methods.

Observerswere blinded tothe treatmentgroups. Each neuronal cell

count shown representsthemeanof .15 microscopicfields selected

from fourreplicates±SEM. PAFwasusedat6,000pg/ml (10.9 nM);

MK-801 was usedat 10,uM. Symbols: *,P< 0.0001, controlversus

PAF; +,P< 0.0001, PAFversusPAF-MK-801. (B)Rat retinal cell

cultures were incubated in c-PAF overnight. Retinal ganglion cell

neurons werespecificallyidentified with thefluorescentdye granular

blue andthen scored forviabilityinamaskedfashiononthe basis of

theabilitytotakeupand cleavefluorescein diacetatetofluorescein.

Thus,viable retinalganglioncellsweredual labeled withgranularblue

andfluoresceinyellow.c-PAF(54 ng/mlor100nM) injured32% of the

retinalganglioncells(68%remainedviable). MK-801 ormemantine

(eachat6 p,M) largely preventedthec-PAF-inducedneuronalinjury.

In the experiments whose results are shown, all of the cultures contained 25p,M glutamate.Control dishes containedapproximately

150 viable retinalganglioncellneuronsandwereusedtonormalize the data obtained from each of fourseparate experiments.

pathophysiologicresponses,including bronchoconstriction, hy-potension, neutropenia, thrombocytopenia,and increased

vas-cularpermeability leadingtocardiac and renalimpairment,as

well aspulmonary edema, anaphylaxis,and death(8, 9). PAF has diverseeffects at the cellularlevel, including phagocytosis, exocytosis, aggregation, chemotaxis, calcium mobilization, ei-cosanoid production, and superoxide production (24, 25). Theseresponsesaremediatedby G-protein-coupled receptors (24, 25). PAFhasbeen shown to induce transient increases in

intracellular calcium in cultured neuronal cells (3, 30). PAF

can facilitate excitatory neurotransmission andglutamate

re-lease in mammalian brains(3, 5, 37, 51, 58). Thus,

experimen-tal evidence suggests that PAFplays important roles in neu-ronal function. In contrast to its beneficial effects, it remains possiblethat PAF contributes to brain injury. This could occur by different mechanisms, including superoxide production, increased calcium mobilization, or by glutamate elevation. Theseeventswouldcollectively injure susceptible neurons.

Thepathophysiologic relevance of PAF in HIV-1 infection of the CNS was first demonstrated by analyses of HIV-1-infectedmonocyte-astroglia mixtures. Cytokines, most notably TNF,are detected in braintissues of patients with advanced ADC(54). Reports demonstrate that TNF-a and interleukin 113 are regulated by PAF and arachidonic acid metabolites (23). Indeed, inhibitors of arachidonic acid metabolites, such as dexamethasone and nordihydroguaiaretic acid, decrease TNF-aproduction in HIV-1-infected cultures (20) and dimin-ish neurotoxic activities (14a). Furthermore, PAF and arachi-donic acid metabolitesare potentstimuli for TNF-a produc-tion. The rise in cytokines, PAF, and arachidonic acid metabolites in these HIV-1-infected cell systems correlates with neurotoxicity, suggesting that all of these factors play importantroles inprogressiveHIV-1 CNS disease.

PAFproduction in HIV-1-infectedmonocytes was ephem-eral.Thiswasreported previously for arachidonic acid metab-olites induced in gp120-stimulated monocytes (56). Indeed, detection of PAF was difficult at later time points (hours and/or days). There are several explanations for this result, including the following: PAF biosynthesis may occur only during initial cell-to-cell contact between HIV-1-infected monocytes and astroglial cells; uninfectedor infected mono-cytes maysecrete afactor that inhibitsastroglialcell-mediated biosynthesisofPAF;and enzymesimportant inthe generation of PAF may be expressed at low levels in HIV-1-infected monocytesandastrogliabut mayundergocell-to-cell transfer sothat PAFbiosynthesisis increasedtransiently(38).

Despitetheevanescent natureof PAFmetabolisminthese experimental conditions,themarked increasein PAF biosyn-thesis inthe CSF of HIV-1-infectedpatientsand the ability of PAFtostimulate TNFproductionsuggestthatPAFis biolog-ically relevant for ADC. One of the advantages of the CSF assaysis thatspecimenswereobtainedwith minimalrisk to the patientsandlikely reflected ongoingbrainpathology. The level ofPAFintheCSF represents the average ofdifferentratesof

production

and degradationofPAF in all brain areas.Thus, levels of PAF intheCSFmaybe different from those observed in

specific

neocorticalareasinvolved inADC.Notwithstanding this

limitation,

severalpotentially importantobservationswere made.First, in pediatric patients, elevatedPAFlevels in CSF correlated with severe encephalopathy. Second, in adult

pa-tients,

elevated PAF levels inCSF correlatedwith both

immu-nosuppression

and onset of

neurological impairment

(ADC stages 1 to

4).

PAFwas not

specific

for

neurological

manifes-tations of HIV infection. PAF in CSFwaselevated inavariety of non-HIVdiseases with CNS

manifestations,

demonstrating that it is not a

specific

marker for HIV-1-associated CNS

dysfunction.

PAF

produced dose-dependent neurotoxicity

(Fig. 4).

Neu-rotoxicity

associated with PAFwas

blocked,

in part,

by

coin-cubation with the NMDA

antagonist

MK-801 ormemantine,

demonstrating

that PAFcan

directly

activate NMDA receptors with

subsequent

excitotoxic neuronal

damage.

PAF canalso increase neuronal

Ca2"

and lead to enhanced

excitatory

neurotransmission

through

increased

glutamate

release

(3,

5, 37,

51,

58).

In

addition,

recentreportshave demonstrated that arachidonic acidcaninhibit

high-affinity glutamate uptake

in synaptosomes and astrocytes

(55)

and

potentiate

NMDA receptorcurrents

by increasing open-channel probability

(41). VOL.68, 1994

T-r

I I~~~~~~~~~~~~~~~~~~~~~~

on November 9, 2019 by guest

http://jvi.asm.org/

(7)

These mechanisms might be responsible for an increase in glutamateinthesynapticcleft and subsequentNMDA

recep-tor-mediated excitotoxic damage to neurons in HIV-infected

brains. These findings, in toto, support a role of PAF in HIV-1-mediated neurotoxicity.This doesnotmeanthat PAF is

anexclusive HIV neurotoxin. It isgenerallybelievedthatmany

cellular and viral neurotoxins contribute to CNS disease. Indeed, arachidonic acid metabolitesmayplay equally impor-tant roles in HIV-1-induced brain disorders.

The abilityof NMDAantagoniststoblockneurotoxicityhas been a central feature in other models of HIV-1-mediated

neuronal damage (21, 35). The consequences of productive HIV-1infection, includingviralgeneproducts,e.g.,gpl20,may

stimulate arachidonic acid metabolite andcytokine production

in macrophages (56). These alterations in macrophage

secre-toryfunctionmay,inturn,result inneurotoxicityandsupport, for the first time, commonalityamong mechanismsfor HIV-1 neuropathogenesis.

ACKNOWLEDGMENTS

We thankKarenSpiegelforoutstandingadministrative support. We thank Michael P. McDermott for advice on statistical analysis and Bradford Navia and the National Neurological Research Specimen

Bankfor contributionofCSFsamples.

This workwas funded in partby PAF/AmFARgrants

500258-12-PG, 500278-14-PGR (H.A.G.), and 001433-11-R&R (S.A.L.), the

Strong Children's Research Fund (H.A.G.), Universityof Nebraska

Biotechnology start-up funds, and NIH grants PO1 NS31492-01 (H.A.G., L.G.E., and H.E.G.), P01 HD29587-01, ROI EY09024-03

(S.A.L.),and P01 HL43628-05(H.E.G.). Howard E. Gendelman isa

Carter Wallace Fellow of theDepartmentofPathologyand

Microbi-ologyof theUniversityof Nebraska Medical Center. Hans Nottet isa

Nicholas B. Badami Scholar of the Laboratoryof ViralPathogenesis, Department of Pathology and Microbiology, Universityof Nebraska

Medical Center.

REFERENCES

1. Banker, G. A., andW.M.Cowan.1977. Rathippocampalneurons

indispersed cell culture.Brain Res. 126:397-425.

2. Bernton, E., H. Bryant, M.Decoster, J.M.Orenstein, J. Ribas,

M.S.Meltzer, andH. E. Gendelman. 1992. No direct

neurono-toxicity byHIV-1virionsorculture fluids from HIV-1 infected T cellsormonocytes.AIDS Res. Hum. Retroviruses 8:495-503.

3. Bito, H., M. Nakamura, Z. Honda, T. Isumi, T. Iwatsubo, T.

Seyama, A. Segura, Y. Kido, and T. Shimizu. 1992.

Platelet-activating factor (PAF) receptor in rat brain: PAF mobilizes intracellularCa2+inhippocampal neurons.Neuron 9:285-294.

4. Chen, H.-S. V., J. W. Pellegrini, S. K. Aggarwal, S. Z. Lei, S.

Warach,F. E.Jensen, and S. A.Lipton. 1992. Open-channel block

of NMDAresponsesbymemantine:therapeutic advantage against

NMDA receptor-mediated neurotoxicity. J. Neurosci.

12:4427-4436.

5. Clark,G.D.,L. T.Happel, G. F. Zorumski, and N. G. Bazan. 1992.

Enhancement ofhippocampal excitatory synaptic transmission by platelet-activating factor. Neuron 9:1211-1216.

6. Cohen,S.E.,T.Mundy, B. Karassik, L. Lieb, D. D. Ludwig,and

J.Ward. 1991.Neuropsychological functioning in human immu-nodeficiency virus type 1 seropositive children infected through neonatalbloodtransfusion. Pediatrics 88:58-68.

7. Cole, R., and J. de Vellis. 1989. A dissection and tissue culture manual of the nervous system, p. 133. Alan R. Liss, Inc., New York.

8. Criscuoli, M.,andA.Subissi. 1988.Catecholamine released from

e adrenal medulla exert a compensatory, protective effect at

b2-adrenoreceptors against PAF-induced death in mice. Br. J.

Pharmacol. 93:132.

9. Darius, H., D. J. Lefer, J. B. Smith,and A. M. Lefer.1986.Roleof platelet-activating factor-acether in mediating guinea pig

anaphy-laxis.Science232:58.

10. Dawson, V. L., T. M. Dawson, G.R.Uhl,and S. H. Snyder.1992.

Humanimmunodeficiencyvirus-1coatprotein neurotoxicity

me-diated by nitric oxide in primary cortical cultures. Proc. Natl. Acad. Sci. USA90:3256-3259.

11. Dreyer, E. B., P. K.Kaiser,J.T. Offermann, and S. A. Lipton.

1990. HIV-1 coat protein neurotoxicity prevented by calcium channelantagonists.Science248:364-367.

12. Dzenko,K.A.,R.J. White,P.Genis,H.Nottet,L. G.Epstein,H. E.

Gendelman, and H. A. Gelbard. 1993. Increased secretion of

cytokines and neurotoxinsbycocultures of HIV-1 infected

mac-rophages andhumanfetal astrocytes. Soc.Neurosci. 19(3):1893. 13. Epstein,L.G.,andH. E.Gendelman.1993. Human

immunodefi-ciency virustypeIinfection of the nervoussystem:pathogenetic mechanisms.Ann.Neurol. 33:429-436.

14. Everall,I.P.,P.J. Luthbert,and P. L. Lantos. 1991. Neuronal loss in thefrontalcortexin HIV infection. Lancet337:1119-1121. 14a.Gendelman,H. E.Unpublisheddata.

15. Gendelman, H.E.,L.Baca,H. Husayni, J.M. Orenstein,D. C.

Kalter, J. A. Turpin, D. Skillman, D. L. Hoover, and M. S. Meltzer. 1990.Macrophage-humanimmunodeficiencyvirus inter-action: viral isolationand targetcelltropism.AIDS4:221-228. 16. Gendelman,H.E.,and S.Gendelman. 1992.Neurologicalaspects

of human immunodeficiency virus infection, p. 229-254. In S.

Specter, M. Bendinelli, and H. Friedman (ed.), Neurotropic viruses andimmunity.PlenumPress,NewYork.

17. Gendelman, H. E., P. Genis, M.Jett, and H. S. L. M. Nottet.

Experimental modelsystemsfor studiesofHIV-encephalitis.In E. Major (ed.),Technicaladvancesin AIDS researchin thenervous

system,in press. PlenumPress,NewYork.

18. Gendelman,H.E.,and P. M. Morahan. 1992.Themacrophagein viralinfections,p.156-232.In C. E.Lewisand J.0.McGee(ed.), The natural immune system series: the macrophage. Oxford UniversityPress,London.

19. Gendelman,H.E., J.Orenstein,M. A.Martin,C.Ferrua,G. H.

Pezeshkpour, T. Phipps, L. Wahl, A. S. Fauci, D. Burke, D.

Skillman,and M. S. Meltzer.1988. Efficient isolation and

propa-gation of human immunodeficiency virus on CSF-1 stimulated

macrophages.J.Exp. Med. 167:1428-1441.

20. Genis, P.,M.Jett,E. W.Bernton,H. A.Gelbard,K.Dzenko,R.

Keane, L. Resnick, D. J. Voisky, L. G. Epstein, and H. E. Gendelman. 1992. Cytokines and arachidonic acid metabolites

produced during HIV-infected macrophage-astroglial interac-tions: implicationsforthe neuropathogenesis ofHIV disease.J.

Exp. Med. 176:1703-1718.

21. Giulian, D., K. Vaca, and C. A. Noonan. 1991. Secretion of neurotoxins by mononuclear phagocytes infected with HIV-1.

Science250:1593-1596.

22. Hahn, J. S., E. Aizenman, and S. A. Lipton. 1988. Central mammalian neurons resistant to glutamate toxicity are made

sensitivebyelevatedextracellular calcium:toxicityblockedbythe

N-methyl-D-aspartate antagonist MK-801. Proc. Natl. Acad. Sci.

USA 85:6556-6560.

23. Hogan, M. M., and S. N. Vogen. 1988. Production of TNF by rIFN-g-primed CC3HH/HeJ

(Lps")

macrophagesrequiresthe

pres-enceoflipid-A-associated proteins.J.Immunol.141:4196-4199. 24. Honda, Z., M.Nakamura,I. Miki,M.Minami,T.Watanabe,Y.

Seyama,H.Okado, 0. Toh,K.Ito,T.Miyamoto,and T.Shimizu. 1991.Cloning byfunctionalexpression of platelet-activatingfactor receptorfromguinea pig lung.Nature(London)349:342. 25. Hwang,S. B. 1990.Specificreceptorsofplatelet-activating factor,

receptor heterogeneity, and signal transduction mechanisms. J.

Lipid Mediators2:123.

26. Janssen,R. S.,D.R.Cornblath,L. G.Epstein,R. P. Foa, J. C.

McArthur,R. W. Price,A. K.Asbury,A.Beckett, D. F. Benson, T.P.Bridge,C. M.Leventhal,P.Satz,A.J.Saykin,J. J. Sidtis, and S. Tross. 1991. Nomenclature and researchcasedefinitions

forneurologicalmanifestations ofhumanimmunodeficiencyvirus

type-1 (HIV-1) infection. Report of a Working Group of the American AcademyofNeurology AIDS Task Force.Neurology 41:778-785.

27. Kalter, D. C., M. Nakamura, J. A. Turpin, L. M. Baca, C.

Dieffenbach,P.Ralph,H. E.Gendelman,and M.S. Meltzer. 1991. Enhanced HIVreplicationin MCSF-treated monocytes. J. Immu-nol. 146:298-306.

on November 9, 2019 by guest

http://jvi.asm.org/

(8)

HIV-MEDIATED NEUROTOXICITY 4635 28. Kent, C., and P. J. Clarke. 1991. The immunolocalisationof the

neuroendocrine specific protein PGP9.5 during neurogenesis in therat. Dev. Brain Res. 58:147-150.

29. Ketzler, S., S. Weis, H. Haug, and H. Budka. 1990. Loss ofneurons

inthe frontalcortexinAIDS brains. Acta Neuropathol. 80:92-94. 30. Kornecki, M., V. Hogan, and Y. H. Ehrlich. 1993. Long-term potentiation in the hippocampus induced by platelet-activating factor. Neuron 10:553-557.

31. Leifer, D., S. A. Lipton, C. J. Barnstable, and R. H. Masland. 1984. Monoclonal antibodytoThy-1 enhances regeneration ofprocesses byrat retinal ganglion cells in culture. Science 224:303-306. 32. Lindgren, J. A., T. Hokfelt, S.-E. Dahlen, C. Patrono, and B.

Samuelsson. 1984. Leukotrienes in theratcentralnervoussystem.

Proc.Natl. Acad. Sci. USA81:6212-6216.

33. Lipton, S. A. 1993. Humanimmunodeficiencyvirus-infected

mac-rophages, gp120, and N-methyl-D-aspartate receptor-mediated

neurotoxicity. Ann. Neurol. 33:227-228.

34. Lipton, S. A., and F. E. Jensen. 1992. Memantine, a clinically-tolerated NMDAopen-channel blocker,prevents HIVcoat pro-tein-induced neuronal injury in vitro and invivo. Soc. Neurosci. Abstr.18:757.

35. Lipton, S. A., N. J. Sucher, P. K. Kaiser, and E. B. Dreyer. 1991. Synergistic effects of HIV coat protein and NMDA

receptor-mediated neurotoxicity. Neuron 7:111-118.

36. Lipton, S. A., and D. L.Tauck. 1987. Voltage-dependent

conduc-tancesofsolitaryganglion cells dissociated from therat retina. J. Physiol. (London) 385:361-391.

37. Marcheselli, V. L., and N. G. Bazan. 1993. Platelet activating factor(PAF) enhancesglutamic acid release in the retina through apresynapticreceptor. Invest. Ophthalmol. Vis. Sci. 34:1048. 38. Marcus, A. J., L. B. Safier, H. L. Ullman, N. Islam, M. J.

Broekman, J. R. Falck, S. Fischer, and C. von Schacky. 1988. Platelet-neutrophil interactions:

(12S)-hydroxyeicosatetraen-1,20-dioic acid:a neweicosanoid synthesized by unstimulated

neutro-phils from(12S)-20-dihydroxyeicosatetraenoicacid.J. Biol.Chem. 263:2223-2229.

39. Mayer, M. L., G. L. Westbrook, and P. B. Guthrie. 1984. Voltage-dependent block by Mg2+ ofNMDA responses in spinal cord neurons.Nature (London) 309:261-263.

40. Michaels, J., L. R. Sharer, and L. G. Epstein. 1988. Human immunodeficiency virus type 1 (HIV-1) infection of the nervous system:a review. Immunodef. Rev. 1:71-104.

41. Miller, B., M. Sarantis, S. F. Traynelis, and D. Attwell. 1992. Potentiation of NMDA receptor currents by arachidonic acid. Nature (London) 355:722-725.

42. Myers, G., J. A. Berzofsky, B. Korber, andR.F.Smith(ed.).1991.

Human retroviruses and AIDS, p. I A 95. Los Alamos National

Laboratory, Los Alamos, N.Mex.

43. Navia, B. A., B. D. Jordan, and R. W. Price. 1991. The AIDS

dementiacomplex: I.Clinical features. Ann. Neurol. 19:517-524. 44. Nowak, L.,P.Bregestovski,P.Ascher,A.Herbet,and A. Prochi-antz. 1984. Magnesium gates glutamate-activated channels in

mousecentralneurons. Nature(London) 307:462-465.

45. O'Flaherty, J. T., J.R.Redman,Jr., J.D.Schmitt, J.M.Ellis, J.R.

Surles, M. H. Marx, C. Piantadosi, and R. L. Wykle. 1987.

1-O-Alkyl-2-N-methylcarbamyl-glycerophosphocholine: a biologi-cally potent, non-metabolizable analog ofplatelet-activation fac-tor.Biochem. Biophys. Res. Commun. 147:18-24.

46. Poubelle,P.E.,D.Gingras,C.Demers,C.Dubois,D.Harbour, J.

Grassi,and J. Rola-Pleszczynski. 1991. Platelet-activatingfactor

(PAF-acether) enhances the concomitant production of tumor

necrosis factor-alpha and interleukin-1 by subsets of human monocytes. Immunology 72:181-187.

46a.Price,R. W., and B. J. Brew. 1988. TheAIDSdementiacomplex.

J. Infect.Dis. 158:1079-1083.

47. Pulliam, L., B. G. Herndier, N. M. Tang, and M. S. McGrath.

1991. Humanimmunodeficiency virus-infectedmacrophages

pro-duce soluble factors that cause histological and neurochemical alterations in cultured human brains. J. Clin. Invest. 87:503-512.

48. Saito,Y., L.R. Sharer, L. G. Epstein, J. Michaels, M. Mintz, M. Louder, K.Golding,T. A.Cvetkovich,and B. M.Blumberg. 1994.

Overexpression of nefas amarkerfor restrictedHIV-1infection of

astrocytes inpostmortem pediatric central nervoustissues.

Neu-rology44:474-480.

49. Sharer, L. R. 1992. Pathology of HIV-1 infection of the central

nervoussystem(review).J. Neuropathol. Exp.Neurol. 51:3-1 1.

50. Sharer, L. R., L. G. Epstein, E.-S. Cho, V. V. Joshi, M. F. Meyenhofer, L. F. Rankin, and C. K. Petito. 1986. Pathologic features of AIDSencephalopathy in children: evidence for LAV/

HTLV-III infectionof brain. Hum. Pathol. 17:271-284.

51. Shukla, S. D. 1992. Platelet-activatingfactor receptor and signal

transduction mechanisms. FASEBJ. 6:2296-2301.

52. Tardieu,M.,C.Hery, S. Peudenier,0.Boespflug, and L. Montag-nier.1992.Humanimmunodeficiency virustype1-infected

mono-cytic cellscandestroy human neural cells after cell-to-cell adhe-sion.Ann.Neurol. 32:11-17.

53. Tenhula, W. N., S. Z. Xu, M. C. Madigan, K. Heller, W. F.

Freeman,and A. A. Sadun. 1992. Morphometriccomparisonsof optic nerve axon loss in acquired immunodeficiency syndrome. Am.J.Ophthalmol. 15:14-20.

54. Tyor, W. R., J. D. Glass, J. W. Griffin, P. S. Becker, J. C. McArthur, L. Bezman, and D. E. Griffin. 1992. Cytokine

expres-sion in the brainduring the acquired immunodeficiency syndrome.

Ann. Neurol.31:349-355.

55. Volterra, A.,D.Trotti, P.Cassutti, C. Tromba, A.Salvaggio,R.C. Melcangi, and R. Racagni. 1992. High sensitivity of glutamate uptaketoextracellular free arachidonicacid levels inrat cortical

synaptosomes andastrocytes.J.Neurochem. 59:600-606. 56. Wahl, L. M., M. L. Corcoran, S. W. Pyle, L. 0. Arthur, A.

Harel-Bellan,andW. L. Farrar.1989. Human immunodeficiency

virusglycoprotein(gp120)inductionofmonocytearachidonic acid metabolites and interleukin 1.Proc.Natl. Acad. Sci. USA 86:621-625.

57. Westervelt,P., H. E.Gendelman,and L. Ratner. 1991. Identifica-tionofadeterminant within the humanimmunodeficiencyvirus 1 surfaceenvelope glycoproteincritical for productiveinfection of primary monocytes. Proc. Natl. Acad. Sci. USA 88:3097-3101. 58. Wieraszko, A.,G.Li,E.Kornecki,M.V.Hogan,and Y. H. Ehrlich.

1993. Long-term potentiation in the hippocampus induced by platelet-activatingfactor. Neuron 10:553-557.

59. Wiley, C. A., E. Masliah,M.Morey,C. Lemere,R.DeTeresa, M.

Grafe,L.Hansen,and R.Terry. 1991. Neocorticaldamageduring

HIVinfection.Ann.Neurol. 29:651-657.

60. Wiley, C. A., R. D. Schrier, J. A. Nelson, P. W. Lampert, and M. B. A.Oldstone. 1986.Cellular localization of human

immuno-deficiency virus infection within the brains ofacquired immune deficiencysyndromepatients.Proc.Natl. Acad. Sci. USA 83:7089-7093.

VOL. 68? 1994