Mechanical Weathering - CurrTech Integration

An analysis of HIV-1-associated in¯ammatoryproducts in brain tissue of humans and SCID micewith HIV-1 encephalitis

Yuri Persidsky,1,3 Manuel Buttini4, Jenae Limoges1,3, Paul Bock1,3 and Howard E Gendelman1,2,3

Departments of 1Pathology and Microbiology, 2Medicine, Eppley Institute for Research in Cancer andAllied Diseases, 3Center for Neurovirology and Neurodegenerative Disorders, University of NebraskaMedical Center, Omaha, Nebraska 68198-5215; and the 4Gladstone Molecular Neurobiology Program andthe Department of Neurology, University of California, San Francisco, California 94141-9100, USA

The human immunode®ciency virus type 1 (HIV)-associated dementia complex(ADC) is a neuroimmunological disorder fueled by viral replication inmononuclear phagocytes (MP) (brain macrophages and microglia). Theelucidation of MP in¯ammatory factors involved in neurological dysfunctionis pivotal for unraveling pathogenic mechanisms and in developing newtherapies for this disease. Recent advances in animal model systems for ADCand its associated encephalitis have provided important insights into howvirus-infected macrophages cause brain injury. Indeed, the stereotacticinoculation of HIV infected monocytes into the basal ganglia/cortex of micewith severe combined immunode®ciency disease (SCID) results in pathologicalfeatures similar to those of human HIV-1 encephalitis (HIVE). We used thisSCID model to study the roles of macrophage secretory factors in HIVE. Theexpression of interleukin-1 (IL-1b, IL-6, IL-10), tumor necrosis factor-alpha(TNFa), vascular endothelial growth factor (VEGF), and adhesion molecules (E-selectin, intracellular cell adhesion molecule (ICAM-1), and vascular celladhesion molecule-1 (VCAM-1)) in encephalitic brains of mice and humans wasevaluated by semi-quantitative polymerase chain reaction (PCR). In SCID micewith HIVE, human and mouse TNFa, and mouse IL-6, VEGF, VCAM-1 and E-selectin were expressed at high levels. These results paralleled, to a greatextent, those in HIVE brain tissues. Laser scanning confocal microscopyperformed to assess the associated neuronal damage showed that microtubuleassociated protein-2 (MAP-2) immunoreactive dendrites were signi®cantlyreduced in both the ipsilateral and contralateral hemispheres of encephaliticmice. These results demonstrate the importance of macrophage in¯ammatoryproducts in the pathogenesis of HIVE and further validates this model of viralencephalitis in SCID mice.

Keywords: animal model; in¯ammatory products; HIV encephalitis; neurotox-ins; neuronal damage

Introduction

Primary infection of the brain with human im-munode®ciency virus type 1 (HIV) is the mostfrequent cause of central nervous system (CNS)morbidity in people with the acquired immunode-®ciency syndrome (AIDS) (Navia et al, 1986a,b).Up to 90% of AIDS patients have pathologicalchanges in the brain at death (Navia et al., 1986a,b).Clinical neurological manifestations of HIV infec-

tion collectively referred to as the HIV-associateddementia complex (ADC) occur in 15 ± 20% ofpatients (Anders et al, 1986; Navia et al, 1986a;Price et al, 1988a,b) usually during advancedimmunosuppression. The early stage of ADC ischaracterized by cognitive, motor, and behaviouralimpairments (Navia et al, 1986a). In later stages, themajority of patients develop global cognitivedysfunction which typically progresses to confu-sion, delusions, visual hallucinations, seizures,coma, and death (Navia et al, 1986a; Price et al,1988a; Kieburtz and Schiffer, 1989; Parisi et al,1991; Brew et al, 1995).

HIV infection of the CNS is manifested patholo-gically by the in®ltration of virus-infected macro-

Correspondence: Howard E Gendelman, Director, Center forNeurovirology and Neurodegenerative Disorders, University ofNebraska Medical Center, 600 S 42nd Street, PO Box 985215;Omaha, NE 68198-5215, USAReceived 22 January 1997; revised 9 October 1997; accepted 14October 1997

Journal of NeuroVirology (1997) 3, 401 ± 416

ãhttp://www.jneurovirol.com

1997 Journal of NeuroVirology, Inc.

phages and the presence of multinucleated giantcells in the brain called HIV-1 encephalitis (HIVE)(Sharer et al, 1985; Budka, 1986; Koenig et al, 1986).Pathological abnormalities, including immunoreac-tive HIV infected brain macrophages/microglia,gliosis, myelin pallor and microglial nodules, aremost prominent in the white and deep gray matter(especially the caudate and putamen) (Navia et al,1986b; Petito et al, 1986; Wiley et al, 1986). In AIDSbrains evaluated at autopsy, there is commonlydiffuse myelin pallor and reactive astrocytosis(Navia et al, 1986b). Although most patients withHIV dementia have HIVE (Wiley and Achim, 1994;Glass et al, 1993), the two are not always associated(Brew et al, 1995). The underlying CNS abnormalitythat accounts for clinical neurological de®cits isneuronal injury (including loss of synaptic contactsand dendrite alterations). Loss of neuronal sub-populations is most signi®cant in the neocortex anddeep grey matter (Masliah et al, 1992a,b).

Despite a clear clinical de®nition of ADC (Brew etal, 1995) and despite the relationship of ADC tobrain macrophage in®ltration, the pathogenesis ofthe disease remains uncertain. Although a numberof factors have been identi®ed during the lastdecade, including the role of neurotoxic productssecreted by virus-infected and immune-activatedmononuclear phagocytes (MP) (brain macrophagesand microglia) (Genis et al, 1992; Tyor et al, 1992;Wesselingh et al, 1993; Grif®n et al, 1994; Glass et al,1993; Lipton and Gendelman, 1995; Nottet et al,1995; Gendelman et al, 1997), the importance ofneurotropic strains of HIV (Power et al, 1994; Korberet al, 1994), and the damaging effects of viralproteins on neurons (Lipton, 1992; Toggas et al,1994; Weeks et al, 1995; Koka et al, 1995; Wyss-Coray et al, 1996), no consensus theory has emerged.

Relevant in vivo models were designed to testpathogenic mechanisms of ADC and to developtherapeutic interventions. To these ends, ourlaboratory recently developed a small animal modelfor HIV encephalitis (Persidsky et al, 1996). Severecombined immunode®ciency mice (SCID), micestereotactically inoculated with HIV-infected

monocytes into basal ganglia/cortex produced thesalient features of HIV encephalitis astrogliosis,microglial nodules, MP activation, and neuronaldeath in response to relatively small numbers ofvirus-infected monocytes.

To analyze the utility of this mouse model forADC in this study, we investigated the role of viralproteins, cytokines and other immune potentiallyneurotoxic products in HIVE. Analysis of humanand mouse in¯ammatory factors in affected braintissue showed similarities between the in¯amma-tory responses of humans and SCID mice withHIVE. Importantly, the presence of HIV infectedmonocytes correlated with neuronal injury. These®ndings support the idea that HIVE is a metabolicencephalopathy caused by viral and host immuneneurotoxins from macrophages and fueled by viralreplication.

Results

Neuropathological features of HIVE in humansTo analyze the morphologic and molecular simila-rities between HIVE in humans and in SCID mice we®rst evaluated postmortem brain tissue from fourpatients with HIVE, two who died of opportunisticinfectious complications but without neurologicalimpairments, and three who died of causes notrelated to HIV infection (Table 1). In humans withHIVE the pathological features included the pre-sence of multinucleated giant cells (MGC), perivas-cular macrophages, microglial nodules andastrogliosis (Figure 1a,c,e,g). MGC and macrophagebrain in®ltration were identi®ed in the deep whitematter of the cerebral hemispheres and in the basalganglia. These cells expressed HLA-DR and HIV-1p24 antigens (Figure 1c,g). In some instances theMGC had long cell processes, which suggested thatthe fused macrophages had arisen from, or hadincorporated microglia (Figure 1a,c). Focal degen-eration of white matter was associated with pallor,astrogliosis, and vacuolation. Macrophages contain-ing myelin debris were readily observed. Rare foci

Table 1 Clinical and morphological ®ndings in nine patients from postmortem brain tissue

Case Age (years) AIDS* HIVE Clinical and neuropathological diagnosis

123456789

373935124264617151

++++++777

++++77777

HIVE/progressive dementiaHIVE/progressive dementia

HIVE/progressive dementia/Pneumocystis carinii pneumoniaHIVE/progressive dementia

Disseminated intravascular coagulation/multiorgan failureGeneralized cachexia/malignant B cell lymphoma

Multiple sclerosisSquamous cell carcinoma of lung

Hepatocellular carcinoma/altered mental status

*AIDS (Center for Disease Control and Prevention case de®nition of HIV-seropositive subject with CD4+ T lymphocytes5200 mm3).HIVE, HIV-1 encephalitis.

In¯ammatory responses in HIV-1 encephalitisY Persidsky et al

402

Figure 1 Morphological features of HIV encephalitis in human brain tissue (A, C, E, G) and in SCID mouse brain (B, D, F, H). Multi-and single nucleated CD68-positive cells were present in basal ganglia (A, B). Most expressed HIV-1 p24 antigen (C, D). Pronouncedand widely spread astrogliosis was detected by anti-GFAP immunostaining in areas containing HIV-1 infected cells (E, F). Activatedhuman (G) and mouse microglia (H) formed nodules. A, C, E, G present serial sections of human brain, and panels B, D, F, H presentserial coronal sections of SCID mouse brain immunostained with antibodies against CD68 (A, B), HIV-1 p24 antigen (C, D), GFAP (E,F), HLA-DR (G) or F4/80 (mouse macrophage antigen) (H). Primary antibodies were detected by Vectastain Elite Kit usingdiaminobenzidine (DAB) as a substrate. F4/80 was visualized by immuno¯uorescence assay. Tissue sections were counterstained withMayer's hematoxylin. Original magni®cation: A ± D, G and H, 6400; E and F, 6200.


403

of neuronal degeneration were seen in the middlelayers of the cortex. Opportunistic infections of thebrain were not found.

Neuropathological features of HIVE in SCID miceThe neuropathological features of HIVE describedabove were reproduced by stereotactic inoculationof human HIV-infected monocytes into the brains ofSCID mice. Animals (three per group) were injectedwith HIV infected or control uninfected monocytes.For both the infected and control animals humanmonocytes were observed in mouse brain up to 5weeks after injection, as demonstrated by immu-nostaining for CD68 antigen (Figure 1b) or vimentin(data not shown). Monocytes were found in thebasal ganglia (putamen and globus pallidus) andcortex along the needle track. Some of the mono-cytes were localized around microvessels as inhuman HIVE. The number of human monocytes inmouse CNS decreased from an average of 60 persection 3 ± 7 days after inoculation to an average ofeight cells 5 weeks after manipulation. Infectedmonocytes were seen in equal numbers anddistribution as the uninfected cells, but readilyformed MGCs (Figure 1b,d) and expressed viralproteins. About 30% of the human monocytesexpressed HIV-1 p24 antigen at 1 week afterinjection, and 80% were antigen positive at 5weeks. HLA-DR antigen expression was detectedin 70% of the inoculated human cells. A reactiveastrogliosis which was initiated at 3 days, peaked at7 days and persisted for 5 weeks afer monocyteplacement. The intensity of the astroglial reactioncorrelated with the number of cells injected.Astrocytes around the injected human cells dis-played thick processes, enhanced GFAP immunor-eactivity and big `empty' nuclei with chromatincondensation (Figure 1f). Few monocytes weredemonstrated in the contralateral ventricle andmeninges, and their presence provoked a wide-spread astrogliosis. Microglial nodules were ®rstdemonstrated 3 days after human monocyte inocu-lation and at 7 days had readily identi®able multi-ple branching processes. The nodules were F4/80-positive and clustered around human monocytes(Figure 1h). The size of microglia paralleled thedegree of astrocytosis. Endothelial cells expressingvascular cell adhesion molecule (VCAM)-1 werelargely in and around microvessels in mouse brainscontaining HIV infected monocytes. Intercellularadhesion molecule (ICAM)-1 antigen was mini-mally increased. The astrocyte and microglialreactions resulting from the needle trauma werelocalized and distinct from those in brains contain-ing human monocytes (data not shown).

Neuronal damage identi®ed in brains of SCID micewith HIVENeuronal damage is a central neuropathologicalfeature of HIVE of humans (Masliah et al, 1992a,b).

To determine if similar ®ndings occur in SCID micewith HIVE we studied neuronal damage induced bymonocytes inoculated into brain after 5 weeks bysemi-quantitative microtubule associated protein(MAP)-2 immunohistochemistry and laser scanningconfocal microscopy in combination with compu-ter-aided image analysis (Table 2).

Neuronal damage, was shown by loss ofdendritic processes on MAP-2-immunolabeledbrain sections in the caudate/putamen and neo-cortex of animals inoculated with human mono-cytes (Figure 2). This neuronal damage waspatchy. In some areas, the damage was moderate,in other areas, severe damage with substantialreduction in the number of MAP-2-immunoreac-tive dendrites was seen. Moderate neuronaldamage was found most frequently in the ipsilat-eral hemibrains of mice inoculated with unin-fected monocytes (Figure 2b) and in thecontralateral hemibrains of mice inoculated withHIV infected monocytes (Figure 2d). Importantly,severe neuronal damage was found most fre-quently in the ipsilateral hemibrains of miceinoculated with HIV-1 infected monocytes (Figure2c). After inoculation with uninfected monocytes,little or no neuronal damage was detected in thecontralateral hemibrains (data not shown). Thedistribution of neuronal damage in the neocortexwas similar to that in the caudate/putamen, someareas show extensive loss of dendritic processes(Figure 2f) and others show no apparent changes(data not shown).

Since the neuronal damage was observed moreconsistently in the caudate/putamen than in theneocortex, the caudate/putamen was chosen for

Table 2 Quantition of MAP-2 immunostaining in brains of SCIDmice with HIVE

Mean neuropil area occupiedby MAP-2 positive dendrites

Treatment Animal Ipsilateral Contralateral

None 123

22.1+4.827.4+0.628.1+2.0

28.9+2.624.7+3.027.7+3.3

Sham 45

27.5+3.626.0+3.7

30.0+3.128.0+2.9

Inoculation of uninfectedmonocytes

6789

20.7+3.322.9+2.514.2+3.617.0+3.0

22.3+2.622.9+3.724.6+5.720.9+6.1

Inoculation of HIV-1infected monocytes

101112131415

12.1+2.315.3+4.714.1+4.313.9+4.89.4+3.68.0+1.1

19.9+3.518.2+4.319.7+6.520.3+5.014.5+6.012.6+2.7

Brain sections were analyzed immunohistochemically by laserconfocal microscopy as described in Materials and methods.Values represent mean (+s.d.) of percent of neuropil occupiedby MAP-2 immunoreactive dendrites for each animal assayed.


404

semi-quantitative analysis of dendritic loss asdescribed in Materials and methods (Table 2). BothHIV-infected and uninfected monocytes inducedneuronal damage in the ipsilateral hemibrains(Figure 3, left panel), but only the infectedmonocytes induced signi®cant neuronal damagein the contralateral hemibrains (Figure 3, rightpanel). Notably, dendritic loss was signi®cantlymore severe in brains of mice inoculated with HIV-1-infected monocytes than in brains of mice in-ocuated with uninfected monocytes (Figure 3).Since the mice were inoculated with similarnumbers of infected or uninfected cells, the moresevere dendritic loss seen in the virus-infectedmonocyte animals is probably due either directlyor indirectly to HIV infection.

In¯ammatory products in human brainsTo further substantiate the notion that theneuropathological features of HIVE in humanscan be produced in SCID mice we undertook an

extensive evaluation of the in¯ammatory productsproduced in human brains during HIV encepha-litis. Initial experiments used semi-quantitativeRNA polymerase chain reaction (PCR) assays todetect proin¯ammatory cytokines in differentregions (cortex, basal ganglia, white matter,cerebellum) of encephalitic and control brains(Table 1). The primers used for these andsubsequent experiments are shown in Tables 3and 4. The results are summarized in Table 5 andFigure 4.

Because the number of macrophages in braincan re¯ect the severity of ADC (Glass et al., 1993)we ®rst assessed the level of CD14 expression (amarker for MP cells) in encephalitic and controlbrain tissues (Table 5 and Figure 4). IncreasedCD14 mRNA levels were uniformly higher inpatients with HIVE than in controls. These resultsre¯ected the distribution of macrophages demon-strated by CD68 or HAM 56 immunostaining (datanot shown). Importantly, CD14 mRNA levelscorrelated signi®cantly with tumor necrosis factoralpha (TNFa) expression, but not interleukin-1(IL-1b), HIV-1pol or other cytokines, adhesionmolecules or metalloproteinases. HIV-1pol mRNAwas observed in all brain regions with morpho-logical signs of encephalitis (Table 5, Figure 4). Incontrol brains without encephalitis HIV-1pol wasabsent.

TNFa mRNA levels were higher in all HIVinfected brain tissues than in controls (Table 5).TNFa levels were highest in HIVE brains (Figure 4).TNFa expression correlated signi®cantly with thatof IL-1b, IL-6, E-selectin, and VCAM-1. No correla-tions between TNFa, HIV-1pol or glial ®brillaryacidic protein (GFAP) expression were found. IL-1blevels were statistically higher in HIV-1 infectedpatients. Interestingly, IL-1b expression correlatedwith ICAM-1 levels. IL-6 mRNA levels were nothigher in HIV-1 infected patients. Interestingly, IL-1b expression correlated with ICAM-1 levels. IL-6mRNA levels were not higher in HIV-1 infectedpatients.

Comparison of expression of two adhesionmolecules (E-selectin and VCAM-1) in the braintissue of patients with HIV-1 encephalitis andcontrols failed to reach statistical signi®cance.VCAM-1 and E-selectin RNAs correlated wellwith each other and with TNFa. CD14 levels didnot correlate with adhesion molecule expressionsuggesting that the upregulation of adhesionmolecules does not necessarily correlate withthe numbers of brain macrophages. ICAM-1 wasupregulated in nonencephalitic tissue but lowerin brains with HIVE. Vascular endothelial growthfactor (VEGF) expression was similar, in allbrains examined, regardless of whether HIVEwas present. There was no relationship betweenVEGF expression and encephalitis. However,VEGF did correlate with GFAP.

Figure 2 Loss of dendritic processes in the brains of SCIDmice inoculated with uninfected (B) or HIV-1-infected (C, D, F)human monocytes revealed by confocal microscopy of brainsections immunolabeled with MAP-2. MAP-2 immunoreactivitywas associated with neuronal cell bodies and dendrites in thecaudate/putamen of a sham-inoculated mouse (A) and theneocortex of an unmanipulated mouse (E). Areas with moderateloss of dendritic processes are shown in B (ipsilateral caudate/putamen, uninfected monocytes) and D (contralateral caudate/putamen, HIV-infected monocytes). Areas with severe loss ofdendritic processes are shown in C (ipsilateral caudate/puta-men, HIV-infected monocytes) and F (ipsilateral neocortex, HIV-infected monocytes). Original magni®cation, 6400 for allpanels.


405

Human in¯ammatory products in SCID mousebrainsHIV-infected or uninfected monocytes were stereo-tactically inoculated (ten SCID mice each) into theleft hemisphere of the brain. After 1 week theanimals were sacri®ced and RNA extracted from thewhole hemisphere of the inoculated and nonin-jected area. RNA ± PCR assays were performed toassay differences between viral, cell and in¯amma-tory factors in HIV-1 infected and control monocytecontaining brains. Differences between inoculatedand noninjected hemispheres were also assessed(see below). The results of these experiments areshown in Figure 5 and Table 6.

To evaluate secretory products of the humanmonocytes placed into the SCID brains, we ®rstcon®rmed the monocytes presence by analyzingthe levels of human CD14 (Figure 5). CD14 is amarker for human monocytes and its mRNA wasfound in the injected hemispheres of animals

inoculated with control (uninfected) human cells(Table 6). CD14 mRNAs were also observed in thecontralateral (noninjected) hemispheres demon-strating that the uninfected human monocytescan migrate (by whatever mechanism(s)) into theopposite hemisphere. CD14 mRNA levels were,surprisingly, not uniformly detected in brainsinoculated with virus-infected monocytes. Thiscould be related to a downregulation of CD14 as aresult of HIV infection or to lower numbers of HIVinfected cells that migrated through the subarach-noid space or brain parenchyma. The ®ndings mayalso re¯ect, in measure, virus-induced cell death,or to MGCs in the HIV-1 cell preparationsrendering the cells less able to migrate from theinoculated site from the syringe placement. Insupport of the former, a 430% downregulation ofCD14 expression was found in cultured HIV-1infected monocytes than in uninfected controls(data not shown).

Figure 3 Neurodegeneration induced by HIV-infected and uninfected human monocytes in the ipsilateral and contralateral caudate/putamen of SCID mice. Laser scanning confocal images of MAP-2-immunolabeled brain sections from SCID mice inoculated with HIV-infected (six mice) or uninfected (four mice) human monocytes were analyzed as described in Materials and methods. Unmanipulated(three mice) and sham-inoculated (two mice, not shown), SCID mice served as controls. Values are the mean+s.e.m. of the percentarea of the neuropil occupied by MAP-2-immunoreactive dendrites. The values obtained for the brains of the two sham-inoculatedanimals (27.5% and 26% for the ipsilateral hemispheres, 30% and 28% for the contralateral hemispheres) were similar to the valuesobtained for unmanipulated mice (average of 25.9% for the ipsilateral hemisphere and of 26.9% for the contralateral hemisphere).MAP-2 immunoreactive dendrites were signi®cantly reduced in both hemispheres of mice inoculated with HIV-infected monocytesand in the ipsilateral hemisphere of mice inoculated with uninfected monocytes. In both hemispheres, the damage induced by HIV-infected monocytes was signi®cantly more severe than the damage induced by uninfected monocytes. *P50.05; **P50.02;***P50.01; ns, not signi®cant.


406

To con®rm the presence of HIV infectedmonocytes in mouse brains by the PCR assayswe designed nested primers for CD14. Thenested RNA ± PCR showed CD14 mRNA in allanimals inoculated with HIV-1 infected mono-cytes. All mouse brains inoculated with virus-infected monocytes had detectable HIV-1pol.HIV-1pol did not correlate with in¯ammatorymolecule expression. This suggests that the levelof viral infection does not necessarily correlatewith cytokine levels or that the infected cellspreferentially produce more TNFa per cell, butthe same amount or less IL-10 and IL-6.

Human TNFa was signi®cantly higher (morethan twofold) in mouse brains inoculated withHIV infected monocytes as compared to unin-fected controls (Table 6, Figure 5). Both theinfected and the contralateral hemisphere hadhigh levels of TNFa. This suggested that TNFacould be readily detected despite the reducednumber of human cells. CD14 and TNFa RNAexpression correlated.

Modest decreases in human IL-10 and IL-6were detected in mouse brains with virus-infectedmonocytes as compared to controls. Such differ-ences were likely related to their decreasedproduction in infected monocytes. CD14 and IL-

6 levels correlated with each other but not withIL-10. IL-1b, IL-8, and TGFb were not detected inmouse brains inoculated with HIV infected orcontrol monocytes.

Mouse in¯ammatory products in SCID mousebrainsIn these experiments, a variety of in¯ammatoryproducts including TNFa, IL-1b, IL-6, VEGF, theinducible form of nitric oxide synthase (iNOS), E-selectin, ICAM-1 and VCAM-1 were analyzed inSCID mice inoculated with HIV infected mono-cytes and in control mice inoculated withuninfected monocytes or with media alone. The®ndings are shown in Table 7 and Figure 5.Mouse TNFa, IL-6, VEGF, and VCAM-1 levelswere all higher in mouse brains containing HIVinfected monocytes than in controls. AlthoughTNFa levels were 1.7 times higher in brains withHIV infected than uninfected monocyte brains thedifferences were not signi®cant (Table 7) (Figure5). Differences between TNFa mRNA betweeninjected monocytes in both hemispheres weresigni®cant for encephalitic mice only. IL-6 mRNAlevels were fourfold higher in mice with HIVinfected monocytes than in animals inoculatedwith control cells. Mouse VEGF was 2.3-fold

Table 3 Oligonucleotide primers used in RNA ± PCR assays for human products

Ampli®cation product (size) Nucleotide position Primer Sequence

Nested CD14 (527 bp)Nested CD14 (527 bp)CD14 (268 bp)

TNFa (237 bp)

IL-1b (179 bp)

IL-6 (159 bp)

IL-10 (265 bp)

VCAM-1 (281 bp)

E-Selectin (257 bp)

ICAM-1 (287 bp)

VEGF (153 bp)

GADPH

719 ± 7401226 ± 1244

881 ± 9071135 ± 1153

963 ± 982503 ± 527740 ± 716588 ± 608480 ± 500659 ± 638549 ± 567317 ± 357476 ± 455399 ± 420128 ± 148393 ± 373244 ± 263234 ± 253515 ± 495372 ± 391612 ± 640878 ± 857669 ± 690334 ± 353621 ± 601367 ± 390152 ± 171283 ± 302273 ± 292199 ± 217394 ± 374280 ± 299

senseantisense

senseantisense

probesense

antisenseprobesense

antisenseprobesense

antisenseprobesense

antisenseprobesense

antisenseprobesense

antisenseprobesense

antisenseprobesense

antisenseprobesense

antisenseprobe

TCTCTGTCCCCACAAGTTCCCCGCCAAGGCAGTTTGAGTCC

ATGCATGTGGTCCAGCGCCCCCACCGACAGGGTCGAACG

AGCCAAGCTCAGAGTGCTCGGAGCTGAGAGATAACCAGCTGGTGCAGATAGATGGGCTCATACCAGGG

CCCTCCACCCATGTGCTCCTCAAAAGCTTGGTGATGTCTGGTTTCAACACGCAGGACAGGATGGAGCAACAAGTGGTG

GTGTGAAAGCAGCAAAGAGGCCTGGAGGTACTCTAGGTATACGGATTCAATGAGGAGACTTGCGATCTCCGAGATGCCTTCAGCGGAAGAAATCGATGACAGCGCAGCCTTGTCTGAGATGATCCTCTCATTGACTTGCAGCACC

ACTTGACTGTGATCGGCTTCCACGAACACTCTTACCTGTGCTACACTTGCAAGTGTGACCC

TGTCACAGCATCACACTCAACCATTGTGAACTGTACAGCCCTGG

TATTCAAACTGCCCTGATGGTAGTGCGGCACGAGAAATTGG

AAAACCTTCCTCACCGTGTACTGGAGGAGGGCAGAATCATCACGTCGCATCAGGGGCACACAGGGGCACACAGGATGACTTGAA

CCATGGAGAAGGCTGGGGCAAAGTTGTCATGGATGACC

CTGCACCACCAACTGCTTAGC


407

higher in mouse brains inoculated with HIV-infected monocytes than in those inoculated withcontrol monocytes (Table 7). Signi®cant differ-ences were shown between the injected andcontralateral hemispheres. Similar results wereobserved for mouse VCAM-1. VCAM-1 wasconstitutively expressed in mouse brain tissue(Figure 5). Nonetheless, VCAM-1 levels weretwofold higher in brain tissues inoculated withHIV-1 infected monocytes than in the contral-ateral hemisphere or in tissues inoculated withcontrol uninfected monocytes (Table 7). Similarly,E-selectin mRNA was nearly twofold higher inHIV-infected mouse brains (for both inoculatedand contralateral hemispheres) than in brainsinoculated with uninfected monocytes. MouseICAM-1 was increased at statistically signi®cantlevels only in brains inoculated with controluninfected monocytes. VCAM-1 correlated withE-selectin and ICAM-1 but only in mice inocu-lated with HIV infected monocytes. iNOS RNAwas increased in all monocyte inoculated brainsbut did not reach statistical signi®cance. IL-1bmRNA levels in brains inoculated with HIVinfected monocytes as compared to controls failedto reach statistical signi®cance. IL-10 levelsshowed no signi®cant differences between HIV-1

monocyte or controls. Mouse TNFa levels corre-lated with mouse IL-6, IL-1b, and GFAP in braintissues containing infected or uninfected mono-cytes. Interestingly, VCAM-1 and iNOS correlatedwith mouse TNFa only in the mice inoculatedwith HIV infected monocytes. Human CD14correlated with mouse TNFa, GFAP, ICAM-1,and IL-1b expression. HIV-1pol correlated withmouse TNFa and VCAM-1 in SCID mice injectedwith virus-infected cells. Human TNFa correlatedwith mouse TNFa in all monocytes inoculatedbrains.

GFAP expression was higher in hemispheresinoculated with HIV-infected monocytes or unin-fected monocytes as compared to the contral-ateral noninjected hemispheres. GFAP levelswere higher in brains inoculated with infectedmonocytes. These differences did not, however,reach signi®cance (Table 7). Human TNFacorrelated with GFAP. GFAP correlated withVCAM-1 and iNOS in brain tissue with HIVinfected monocytes.

These results, taken together with the previousneuropathological analyses strongly support therelevance of the SCID mouse model of HIVencephalitis to human disease. Most importantly,the results demonstrate that such in¯ammatory

Table 4 Oligonucleotide primers used in RNA ± PCR assays for mouse products

Ampli®cation product (size) Nucleotide position Primer Sequence

TNFa (200 bp)

IL-1b (242 bp)

IL-6 (157 bp)

IL-10 (180 bp)

VCAM-1 (255 bp)

E-selectin (223 bp)

ICAM-1b (346 bp)

VEGF (250 bp)

GFAP (212 bp)

iNOS (191 bp)

124 ± 143302 ± 321154 ± 174157 ± 176377 ± 396203 ± 222127 ± 146262 ± 251160 ± 179881 ± 900

1039 ± 1058931 ± 950734 ± 753501 ± 521599 ± 619821 ± 840620 ± 640714 ± 733204 ± 224528 ± 547420 ± 439176 ± 195404 ± 423254 ± 273

1458 ± 14771648 ± 16671563 ± 1582

384 ± 404554 ± 573476 ± 495

senseantisense

probesense

antisenseprobesense

antisenseprobesense

antisenseprobesense

antisenseprobesense

antisenseprobesense

antisenseprobesense

antisenseprobesense

antisenseprobesense

antisenseprobe

CTCCCTCCAGAAAAGACACCTTTGGGGACCGATCACCCCG

AAAGCATGATCCGCGACGTGGGACCCCAAAAGATGAAGGGCATGAGTCACAGAGGATGGGCTCCAGATGAGAGCATCCAGCACTTCACAAGTCCGGAGAGGTTCATACAATCAGAATTGCCGATACCACTCCCAACAGACCACTTTATAGTATTTAAAGGGTTCTTCACAACTCTCTTAGGCCTTGGGAAGCAATTGAAGCGGGATGTACAGAGATCGTCG

GGAGATTGATCTGTTCAAGGGCCTTTACTCCCGTCATTGAGGGCAGGCGCACTCCACTCTCC

ACCTGCAAGTGCCACCCTGGCGCTCCCACCCGTTCGGCCCC

CTTCCTCATGCAAGGAGGACCCCCACCCACTGGCTGGCGGCCAGCCTGGCAGCAAGTAGGCAGAAGGAGAGCAGAAGTCCCTTTGGTGAGGTTTGATCCGCCCTGGTGGACATCTTCCAGGCCCCAGCTGGTTAGAATTGGTGGCATTCGTGATGCATAGGTGGAAGTTTCCACAGGCTGCCCTAAGAGTCACCAAAATGGGGACTTGCAAGTGAAATCCGCCCGTCCACAGTATGTGAGG


408

effects are most likely mediated by virus-infectedmacrophages.

Discussion

The present study demonstrates that neuronalinjury and in¯ammatory factors in human HIV

encephalitis can be produced in a SCID mouse.Inoculation with HIV-infected monocytes upregu-lated a number of in¯ammatory factors includingTNFa, IL-6, VEGF, VCAM-1, and E-selectin, in thebrains of SCID mice. These products also showedassociations one with another. Interestingly, IL-1b, acytokine not associated with human HIV encepha-litis, was found at higher levels in mouse brainsinoculated with control uninfected monocytes.GFAP expression was only minimally upregulatedin brains with HIV-1 encephalitis than in humanand mouse suggesting that the changes found at theprotein level likely occur at a posttranscriptionallevel. Similar patterns of cytokine regulation wereobserved in human and mouse brains with HIV-1encephalitis including the upregulation of bothhuman TNFa and mouse TNFa, VCAM-1, and Eselectin; the absence of statistically signi®cantcorrelation between HIV-1pol and the expressionof in¯ammatory factors, and the capacity of virus-infected macrophages to induce a signi®cant in-¯ammatory reaction in brain tissue. Taken togetherthese ®ndings support the notion that the SCIDmouse model of HIV encephalitis is relevant formirroring aspects of the human disease.

The few differences in in¯ammatory factorexpression between human and mouse brain tissues(for example IL-6, VEGF) could be related to thetrauma of cell inoculation or altered responses ofdifferent mammalian species to virus and/orcytokine production. These could affect the com-plex regulation of in¯ammatory products by humanor mouse cells. Moreover, the elevated levels ofcytokines and other proin¯ammatory products inthe peripheral blood of HIV-infected patients mayalso affect the expression of in¯ammatory factors inbrain. No experiments that could generate suchsimilarities were attempted in this study. Never-theless, similarities between the expression of avariety of in¯ammatory factors in human andmouse brains were shown clearly and reproducibly.Importantly, these data show that in¯ammatoryneurotoxins produced by human monocytes can

1 2 3 4 5 6 7 8 9

CD14

HIV-1pol

TNF-α

lL-1β

GFAP

VCAM-1

GAPDH

Figure 4 Expression of mRNA for CD14, HIV-1pol, TNFa,IL-1b,GFAP, VCAM-1 in human brains affected by HIV-1 encephalitis/HIV-1 associated dementia (cases 1, 2, 3, and 4), HIV positivepatients without signs of encephalitis (cases 5 and 6), andseronegative controls (cases 7, 8, and 9). RNA was reversetranscribed and then ampli®ed by PCR for 28 cycles. Coupledreverse transcription/PCR ampli®cation products were visua-lized by Southern blot hybridization. The mRNA for the cellularenzyme glyceraldehyde 3-phosphate dehydrogenase (GAPDH)served as an internal control to allow analysis and comparison ofRNA species between different samples using a MolecularDynamic's PhosphorImager 425B.

Table 5 RNA ± PCR analysis of viral and cellular monocytes secretions in human brains with HIVE

Viral and cellular Non-encephalitic brains P valuesproducts HIVE brains HIV seropositive HIV seronegative A B C

CD14HIV-1 polTNFaIL-1bIL-6VCAM-1ICAM-1

32+5.72.82+1.14272+82

27+4.476+11.4

101+10.8130+18.2

15+60

140+5217+0.3

136+4.5085+4.5

405+83.7

10+60

17+6.76+0.6

42+5.433+3.347+8.8

50.05NA*

50.05NSNSNS

50.01

NA0.0099NSNSNS

50.03NS

NANA50.0250.004

0.05NS

50.03

NS, not signi®cant: NA, not applicable. Levels of RNA isolated from 5 different brain regions were quantitated after cDNAamplication for 28 cycles. Comparisons between species of RNA were made utilizing a ratio to GAPDH (mean+s.e.m.6103) as aninternal control and quantitation of the ampli®ed products with a phosphorimager. A=HIV-1 encephalitic versus non-encephaliticbrains; B=HIV-1 encephalitic versus HIV seronegative brains; C=non-encephalitic versus HIV nonencephalitic brains.


409

affect neuroimmune activities in mouse braintissue. The results further provide an explanationfor the neuropathological changes (including micro-glial nodule formation, astrogliosis, and neuronalinjury) demonstrated in the SCID mouse brainsinoculated with HIV-infected monocytes (Persidskyet al, 1996). These changes are likely mediated byimmunological factors secreted in response to viralinfection and immune activation of brain macro-phages and microglia. These secretory activitiescould affect both the autocrine and paracrineproduction of cytokines in the affected brains ofhumans and mice.

Levels of CD14 RNA served as an importantindicator for monocyte brain in®ltration in bothanimals inoculated with HIV-1 infected and controluninfected cells. The inability to detect CD14 geneexpression in brains of animals inoculated withvirus-infected monocytes could re¯ect a down-regulation of CD14 within the infected monocytesand/or decreased ability of the infected monocytesto migrate from the areas of inoculation. Indeed, theMGCs may have less ability to move within theaffected hemisphere or to cross hemispheres. Thiscould be related to cell death or defects in migratoryfunction as related to cell size or viral infection.This explains, in measure, the apparent discre-pancy between the histopathological data, whichshowed equal numbers of virus infected and controlmonocytes in immunostained tissues, and theRNA ± PCR results which showed decreased levelsof CD14 RNA in the brains inoculated with infectedas compared to uninfected monocytes.

The central feature of HIV encephalitis isneuronal damage. Neuronal injury resembling thatin humans with ADC was elicited in the brains ofSCID mice by human monocytes. The area occupiedby MAP-2-immunoreactive dendrites was signi®-cantly reduced in both the ipsilateral and contral-ateral hemibrain of mice inoculated with HIV-infected monocytes and in the ipsilateral hemi-

brains of mice inoculated with uninfected mono-cytes. Virus-infected monocytes inducedsigni®cantly more pronounced neuronal injury thanuninfected monocytes. Such dendritic damage andreduction were previously found distinct in hu-mans with HIV-1 encephalitis (Masliah et al,1992a).

This study also demonstrates that HIV-infectedmonocytes can induce in¯ammatory reactions inbrain that affect neurotoxic activities. Therefore,these cells may be key effectors of neuronal damagein the brains of HIV-1-infected patients. Theneurotoxic properties of human monocytes couldbe explained by several, possibly synergistic,mechanisms. Implantation of puri®ed humanmonocytes into mouse brain activates the cellsand induces the production of neurotoxins (Per-sidsky et al, 1996), which may be increased by HIVinfection. HIV infection primes monocytes foractivation (Gendelman et al, 1997) and, as aconsequence, stimulated virus-infected monocytesmay produce a variety of potential neurotoxins ingreater abundance than stimulated uninfected cells.In the brains of SCID mice inoculated with humanmonocytes, autocrine stimulation of the implantedmonocytes by monocyte-derived factors, as well asparacrine stimulation by factors of endogenousmouse brain cells, such as proin¯ammatory cyto-kines expressed by microglia (Persidsky et al, 1996),may all contribute to the persistent activation of theimplanted monocytes. This hypothesis is supportedby the upregulation of several distinct in¯ammatoryfactors, produced in large measure by activatedmacrophages and proposed to play central roles inneurotoxic responses in affected brains with HIV-1encephalitis.

Potentially noxious molecules produced byactivated monocytes include in¯ammatory cyto-kines, such as TNFa and IL-1b (Genis et al, 1992;Merrill and Martinez-Maza, 1993; Nottet et al,1995), eicosanoids (Genis et al, 1992; Nottet et al,

Table 6 RNA ± PCR analysis of human monocyte secretory products in mouse brain tissue

Human Unifected monocytes HIV-1-infected monocytes P valuescytokines Ipsilateral Contralateral Ipsilateral Contralateral A B C

CD14HIV-polTNF-aIL-6IL-10MMP-1MMP-9

26+40

25+466+913+120+111+3

6+0.60

14+221+79+2

20+31+0.3

0*172+57

52+10#

28+118+2

24+33+0.4

0*5+1

33+622+14

6+214+2

2+0.3

50.0001NA50.0350.002

NSNS

50.02

NA0.0099NSNSNS

50.03NS

NANA50.0250.0040.05NS

50.03

*Presence of monocytes was veri®ed by nested PCR for CD14. NA, not applicable; NS, not signi®cant. Levels of RNA werequantitated after cDNA ampli®cation for 28 cycles. Comparisons between species of RNA were made utilizing a ratio to GAPDH(mean+s.e.m.6103) as an internal control and quantitation of the ampli®ed products with a Phosphorimager. P values are shown(right) between brains where control uninfected monocytes were inoculated (A) and compared to the non-inoculated contralateralhemisphere; (B) the HIV-1 infected monocytes inoculated hemisphere compared to the non-inoculated contralateral hemisphere and(C) HIV-1 inoculated as compared to control monocyte-injected brain hemisphere.


410

1995), free radicals (Piani et al, 1992; Dawson etal, 1993; Bukrinsky et al, 1995), platelet-activatingfactor (Gelbard et al, 1994), and excitotoxic aminoacids (Piani et al, 1992; Lipton, 1992; Lipton andRosenberg, 1994). In addition, secretion of HIV-1proteins such as gp120, gp41 and tat, which havebeen shown to induce neuronal injury in vivo(Mucke et al, 1995; Wyss-Coray et al, 1996; Weekset al, 1995; Adamson et al, 1996), may contributeto the neurotoxicity caused by HIV-infectedmonocytes (Wahl et al, 1989; Lipton, 1992;

Guilian et al, 1993). The combined action ofendogenous monocyte products, HIV factors, andsecretions derived from mouse cells likely ac-counts for the dramatic neuronal damage observedin the brains of mice inoculated with infectedmonocytes.

The ®ndings of this study may also support invivo evaluations of therapeutic compounds aimedat preventing neuronal injury or reducing macro-phage in¯ammatory responses that affect neuronalimpairments in HIV-infected patients. Substancessuch as glutamate receptor antagonists, anti-in¯am-matory compounds, or free radical scavengers mayall help limit the neuronal damage caused byactivated MPs. For example, the N-methyl-D-aspartate receptor antagonist memantine preventneuronal injury in the brains of gp120 transgenicmice (Toggas et al, 1996), and the antiin¯ammatorycompound and phosphodiesterase inhibitor pen-toxifylline suppresses TNFa transcription (Chao etal., 1993). TNFa may affect neuronal viability aswell as stimulate HIV replication in monocytes (Poliet al, 1990) and microglia (Peterson et al, 1992).Since TNFa is strongly suspected to contribute toneuronal damage in the brains of HIV-infectedpatients (Merrill and Martinez-Maza, 1993), sup-pression of TNFa secretion may limit monocyte-mediated neuronal injury.

In summary, we have demonstrated that HIV-infected monocytes in the brains of SCID mice elicitpronounced in¯ammatory responses resulting inneuronal damage. These results suggest that HIVinfection primes macrophages for activation andlead to increased secretion of neurotoxins andincreased neuronal injury associated with HIVinfection. Such effects could be limited by therapiesthat target HIV replication, prevent the productionof in¯ammatory factors from activated MP, inhibitmonocyte transendothelial migration into the brain,or protect neurons from neurotoxic activities. Theseresults further substantiate the relevance of theSCID mouse model to the neuropathogenesis ofHIV-1 infection and the importance of macrophagesin disease.

Materials and methods

Isolation and culture of primary human monocytesMonocytes were recovered from PBMC's of HIV andhepatitis B seronegative donors after leukopheresis,and were puri®ed by counter current centrifugalelutriation. Cell suspensions were 498% mono-cytes by criteria of cell morphology and Wright-stained cytosmears, by granular peroxidase, bynonspeci®c esterase, and CD68 labeling (mono-cyte-macrophage marker). Monocytes were cul-tured in Te¯on ¯asks (26106 cells/ml, 1506106

cells/¯ask) in Dulbecco's modi®ed Eagle's media(DMEM) (Sigma, St. Louis, MO) with 10% heat-inactivated pooled human serum, 50 mg/ml genta-

+ –

21 3 4

CD14

HIV-1pol

TNF-α

TNF-α

lL-1β

lL-1β

MMP-1

MMP-9

GFAP

VCAM-1

GAPDH

Human Cytokines

Mouse Cytokines

Inoculated Hemisphere:

Animal number:

+ – + – + –

Figure 5 Detection of mRNA for human and mouse proin-¯ammatory products in the SCID mouse brains with HIVencephalitis. SCID mice were stereotactically inoculated with16105 HIV infected (ten animals) or uninfected monocytes (tenanimals). Mice were sacri®ced 1 week after inoculation, andRNA was extracted from injected and contralateral (unmanipu-lated) hemispheres. The results are shown for four miceinoculated with HIV infected monocytes (inoculated hemi-spheres labeled with `+') and are represented for all animalsinvestigated. Oligonucleotide primers for speci®c human (CD14,HIV-1pol, TNFa, IL-1b, MMP-1 and -9) and mouse products(TNFa, IL-1b, GFAP, VCAM-1) were designed from sequencesobtained from Genbank. mRNA for mouse cytokines did notcross-react with analogous human cytokine mRNAs. RNA wasreverse transcribed and then ampli®ed by PCR for 28 cycles.Coupled reverse transcription/PCR ampli®cation products werevisualized by Southern blot hybridization. The mRNA for thecellular enzyme glyceraldehyde 3-phosphate dehydrogenase(GAPDH) served as an internal control to allow analysis andcomparison of RNA species between different samples using aMolecular Dynamic's PhosphorImager 425B.


411

micin (Sigma), 10 mg/ml cipro¯oxacin (Sigma Che-mical Co.), and 1000 U/ml highly puri®ed recombi-nant humanmacrophage CSF (a generous gift fromGenetics Institute Inc., Cambridge, MA) (Kalter etal, 1991). All tissue culture reagents were screenedbefore use and found negative for endotoxin(510 pg/ml; Associates of Cape Cod, Woods Hole,MA) and mycoplasma contamination (Gen-probe II;Gen-probe, San Diego, CA).

HIV infection of monocytesAfter 7 days in suspension, cultured monocyteswere infected with HIV-1ADA at multiplicity ofinfection (MOI) of one infectious viral particle/target cell (Gendelman et al, 1988). The culturemedium was exchanged every 3 days. Reversetranscriptase (RT) activity was determined intriplicate samples of culture ¯uids added to areaction mixture of 0.05% Nonidet P-40 (Sigma),10 mg/ml poly(A), 0.25 mg/ml oligo(dt) (PharmaciaFine Chemicals, Piscataway, NJ), 5 mM DTT(Pharmacia), 150 mM KCl, 15 mM MgCl2, and[3H]dTTP (2 Ci/mmol; Amersham, ArlingtonHeights, IL) in pH 7.9 Tris-HCl buffer for 24 h at378C. Radio labeled nucleotides were precipitatedwith cold 10% TCA and 95% ethanol in automaticcell harvester (Skatron, Sterling, VA) on paper®lters. Radioactivity was estimated by liquidscintillation spectroscopy. The percentage of HIV-1 p24 monoclonal antibody (Dako, Carpenteria,CA). Cytospin preparations of HIV-1 infected ornoninfected cells were ®xed in cold absoluteacetone/methanol (1 : 1) for 10 min at 7208C, andthen anti-p24 monoclonal antibody (Dako, 1 : 10)was applied for 45 min at room temperature,followed by treatment with FITC-conjugated anti-mouse IgG, F(ab')2 fragment (Boehringer MannheimCorp., Indianapolis, IN; 1 : 100). Viral antigen-positive cells were counted with FITC-®lter in aNikon Microphot-FXA microscope (Nikon, Tokyo,Japan) using a620 objective in ten random ®elds.All experiments were performed in triplicate.

SCID mouse modelThe SCID mice (male C.B-17/IcrCrl-SCID-bgBR 3 ±4 weeks old) were purchased from Charles RiverLaboratories (Wilmington, MA). Animals weremaintained in sterile microisolator cages underpathogen-free conditions in the Laboratory ofAnimal Medicine at the University of NebraskaMedical Center (UNMC) in accordance withethical guidelines for care of laboratory animalsset forth by the National Institutes of Health. Allanimal manipulations (including intracranial in-oculations) were made in a laminar ¯ow hood.Intracerebral injections of HIV-1ADA-infected orcontrol uninfected monocytes were performedfollowing intraperitoneal (i.p.) ketamine/xylazineanesthesia (100 mg/kg ketamine and 16 mg/kgxylazine). On the day of brain inoculation (day 7post infection) cells were pelleted. A monocytesuspension of 26107 cells/ml was prepared forinjections. SCID mice were anesthetized andplaced in a stereotaxic apparatus (Stoetling Co.,Wood Dale, IL) designed speci®cally for mice.The animal's head, following anesthesia, wassecured with earbars and mouthpiece. Eachanimal was inoculated with 15 ml of suspensioncontaining 26105 cells. The coordinates for theinoculation (putamen) were: 3.5 mm behind thebregma, 3.5 mm lateral from the sagittal midlineat a depth and angle of 4.0 mm and 358 fromvertical line respectively with 100 ml Hamiltonsyringe and a 26 guage needle. This technique ofintracerebral inoculation was reproducible inmultiple separate experiments. Animals weresacri®ced 1 week after intracerebral inoculation.Ten mice were inoculated in each of threeconditions: human monocytes, HIV-1-infected hu-man monocytes, and DMEM culture media (sham)inoculum. At 7 days post-inoculation all micewere sacri®ced, and the whole brain was col-lected. The brain stem was removed, the hemi-spheres separated, and RNA was extracted fromeach hemisphere individually.

Table 7 RNA PCR analysis of mouse monocyte secretory products in the brains of SCID mice with HIV encephalitis

Uninfected monocytes HIV-1-infected monocytes P valuesMouse Cytokines Ipsilateral Contralateral Ipsilateral Contralateral A B C

TNF-aIL-1bIL-6VEGFVCAM-1E SelectinGFAP

42+535+85+1

58+12129+14

11+1143+16

31+416+36+2

46+8126+17

14+286+6

74+1523+621+5

136+29225+33

28+9180+30

41+817+318+7

111+15115+26

22+6103+15

50.0450.04

NSNSNSNS

50.004

NSNSNSNS

50.02NS

50.05

NSNS

50.0350.0550.02

NSNS

NS, not signi®cant. Levels of RNA were quantitated after cDNA ampli®cation for 28 cycles. Comparisons between species of RNAwere made utilizing a ratio to GAPDH (mean+s.e.m.6103) as an internal control and quantitation of the ampli®ed products with aPhosphorimager. P values are shown (right) between brains where control uninfected monocytes were inoculated (A) and comparedto the non-inoculated contralateral hemisphere; (B) the HIV-1 infected monocytes inoculated hemispere compared to the non-inoculated contralateral hemisphere and (C) HIV-1 inoculated as compared to control monocyte-injected brain hemisphere.


412

ImmunohistochemistryBrain tissue was collected at necropsy and frozen inOCT compound in cassettes for cryostat sectioning.Six coronal blocks (about 1 mm in thickness) wereprepared from each brain, and ten sections (6 mm)were cut from each block and double immuno-stained for GFAP (astrocytosis) and CD68 orvimentin (monocyte distribution). Additional serialsections were obtained from the block containingthe inoculation site (usually block 3), 30 and fromthe immediately adjacent blocks (u = ten sectionsper block). Monocytes were identi®ed in mousebrain tissue by morphological and immunohisto-chemical assays in each study.

Immunohistochemistry was performed on 6 mmthick cryostat sections ®xed in absolute acetone/methanol (1 : 1) for 10 min at 7208C. Humanmonocyte-derived macrophages were identi®edwith anti-CD68 KP-1 (Dako, 1 : 100) or anti-vimentin(Boehringer Mannheim, Indianapolis, IN; 1 : 25)monoclonal antibodies. Mouse astrocytes wererecognized with polyclonal antibodies against glial®brillary acidic protein (GFAP) (Dako 1 : 100).Mouse microglia/macrophages were identi®ed withrat monoclonal antibody F4/80 (Serotech, Kidling-ton, UK). Mouse VCAM-1 was identi®ed with ratmAb (PharMingen, San Diego, CA) at a 1 : 25dilution). Mouse ICAM-1 was detected with ham-ster mAb (PharMingen and used at 1 : 50 dilution).Human HLA-DR (Boehringer Mannheim 1 : 25) andHIV p24 antigens (Dako 1 : 10) mAbs were used todetect macrophage activation and HIV-1 proteinexpression respectively.

Indirect immuno¯uorescence detection (second-ary Abs) was performed with FITC-coupled anti-mouse IgG, F(ab')2 fragment, or anti-rabbit Abs IgG,F(ab')2 fragment (Boehringer Mannheim 1 : 100)labeled with rhodamine. Goat anti-rat Ab, F(ab')2

fragment (Biosource) coupled with FITC was usedas the detection step for the primary rat mAbs.Immunostained tissues were analyzed with a NikonMicrophot-FXA microscope. Double labeling (withanti-CD68 and anti GFAP) was done on replicatetissue sections to characterize astrocyte reactions tohuman monocytes. Absence of the primary anti-body or use of mouse IgG or rabbit IgG (Dako) servedas controls. To detect antigens (CD68, HLA-DR,HIV-1, p24, GFAP, vimentin, NSE, MAP-2, neuro-®laments and PGP 9.5) on paraf®n sections, avidin-biotin immunoperoxidase staining Vectastain EliteABC kit (Vector Laboratories, Burlingame, CA) wasused with 3,3'-diaminobenzidine as the chromogen.Then sections were counterstained with Mayer'shematoxylin.

Tissue preparation for laser confocal microscopyFor laser confocal microscopic analysis of neuronaldamage, mice inoculated with uninfected mono-cytes (n=4) HIV-infected monocytes (n=6) ormedium only unmanipulated controls (n=4) were

anesthetized and perfused transcardially with 0.9%saline ®ve weeks after implantation of the mono-cytes. The brains were removed and post®xed for48 h in 4% paraformaldehyde in phosphate-buf-fered saline (PBS: 2.6 mM KCl, 1.4 mM KH2PO4,136 mM NaCl, 8 mM Na2HPO4). Vibratome sections(40 mm) were prepared and stored in cryoprotectantmedium (0.1 M phosphate buffer containing 30%glycerin and 30% ethylene glycol) at 7208C untiluse. Brains of additional animals were removed 1week (n=6) and 5 weeks (n=2) after monocyteimplantation to prepare frozen or paraf®n-em-bedded coronal sections covering a distance ofabout 600 mm (rostrally and caudally) from theinjection site. These sections were used to con®rmthe presence of human monocytes and of HIV byimmunostaining with anti-CD68 and anti-p24 anti-bodies, respectively. Primary antibody binding wasdetected by the avidin-biotin immunoperoxidasetechnique (Vectastain Elite ABC kit; Vector Labora-tories, Burlingame, CA), and sections were counter-stained with Mayer's hematoxylin.

Laser confocal microscopyTwo vibratome brain sections at the level of thecaudate/putamen (near the inoculation site of hu-man monocytes) per animal were immunolabeledwith an antibody directed against MAP-2, a markerfor neuronal dendrites, and then incubated with aFITC-labeled secondary antibody. For each mouse,(six to nine different laser scanning confocal imagesfor the caudate/putamen of each hemibrain) cover-ing a surface of 2106140 mm each, were obtainedand analyzed, essentially as described (Masliah etal, 1992a), with a Bio-Rad MRC-1024 mounted on aNikon Optiphot-2 microscope. Digitized imageswere transferred to a PowerMacintosh 8500 runninga public domain program written by Wayne Ras-band (NIH Image). The area of the neuropiloccupied by MAP-2-immunolabeled dendrites wasquanti®ed and expressed as a percentage of the totalimage area, as described (Masliah et al, 1992a).

In view of the patchy distribution of neuronaldamage after implantation of human monocytes andto obtain a representative value for loss of dendriticprocesses in each group of animals, four to six of themost severely affected areas and two to three lessseverely affected areas were imaged for each hemi-sphere in which neuronal damage was identi®ed.For each control animal (unmanipulated and sham-inoculated mice), two or three images for thecaudate/putamen of each hemibrain were analyzed.Values for individual hemibrains represent themean of all the readings obtained from the differentareas imaged.

Oligonucleotides for RNA ± PCR assaysTo analyze murine cytokine response to humanmacrophages, oligonucleotide primers for speci®cmouse cytokine mRNA were designed from mRNA


413

sequences obtained from Genbank. The computerprogram Amplify 1.2 was utilized to assessef®ciency and reactivity of primer sequences.Only primers that produced one product ofsuitable size from mouse mRNA, and no productfrom analogous human mRNA, were used. Tofurther test the reactivity of the primers, PCRagainst HIV-infected and uninfected human mono-cytes RNA, and total cellular RNA from mice braintissue were analyzed. For controls, total cellularRNA from liver, spleen, and brain from controlanimals, and animals challenged with 50 mg and100 mg of LPS, were also examined. Primers forhuman and mouse products used in this work arelisted in Tables 2 and 3.

Isolation of total cellular RNATotal cellular RNA was isolated from mouse orhuman brain tissue (see below) using TRIzol reagent(Life Technologies). Brie¯y, whole tissue washomogenized in 1 ml of TRIzol per 50 ± 100 mgtissue, or cells were lysed by repetitive pipetting in1 ml of TRIzol reagent per 50 ± 100 million cells.RNA was extracted using chloroform, and precipi-tated with isopropanol. The samples were treatedtwo times with 10 ml RNAse free RQ1 DNAse(1000 u/ml; from Promega). The DNA-free RNAwas then precipitated using 3M sodium acetate andethanol. To verify that DNA contamination wasabsent, a DNA ± PCR for glyceraldehyde 3-phos-phate dehydrogenase (GAPDH) was utilized.

RNA ± PCR assayLevels of cytokine RNA's were examined afterreverse transcription with antisense primers andPCR ampli®cation of cDNA transcripts. RNA fromglyceraldehyde 3-phosphate dehydrogenase(GAPDH) served as an internal standard forquantitation utilizing a Molecular Dynamic's Phos-phorImager 425B equalizing results to GAPDH cellexpression for densitometric analysis.

One mg total cellular RNA in 5 ml was mixed with0.5 mg of antisense primer (primer sequences areattached). The mixture was heated at 708C for10 min then cooled to 48C. RT (Life Technologies,200 U/ml) and 1.5 ml each of the four deoxynucleo-tide triphospates (10 mM, Perkin Elmer) wereadded. RT reactions were at 378C for 15 min, andwere stopped by heating the sample to 958C. ForPCR ampli®cation of cDNA products, one-half ofthe mixture was discarded, then 0.5 mg sense and0.25 mg antisense primers were added, with 1 mleach of the four deoxynucleotide triphosphates, and0.5 ml Amplitaq DNA polymerase (5 U/ml, PerkinElmer). Primer and probe sequences for human andmouse products are listed in Tables 2 and 3,respectively. A semi-quantitative RNA ± PCR analy-sis was made of secretory products from humanmonocytes and mouse brain tissue. Ampli®edproducts (at the linear range of the assay) were

quantitated utilizing ratios of product to GAPDH.The latter was used as an internal standard for allRNA ± PCR products scored with a PhosphorImager425B.

To increase the sensitivity of CD14 nested PCRwas performed with primers were yielding a 527 bpproduct from CD14 RNA. After the initial ampli®ca-tion using the above-mentioned PCR method, 2.5 mlof cDNA was subjected to a second ampli®cationusing primers amplifying an interior portion of the527 bp cDNA, and yielding a 268 bp product. 0.5 mgeach of sense and antisense primers were mixedwith the cDNA in the presence of 1 ml each of thefour deoxynucleotide triphosphates, and 0.5 mlAmplitaq DNA polymerase (5 U/ml, Perkin Elmer).The products of 28 cycles (30 s, 948C; 30 s, 508C;60 s, 728C) were analyzed using Southern Blothybridization, with 32P-labeled oligonucleotides.

Brain autopsy materialsBrain tissue, derived from ®ve separate regions,from 6 HIV-1 infected and three control cases(who were HIV-1 negative) was used for immuno-histological assessment and total cellular RNAisolation. A concise description of clinical historypertaining to each of the brain samples is shownin Table 1. Immunohistochemical evaluation ofmacrophage in®ltration and microglia reaction(anti-CD68 or HAM 56 mAbs), activation (HLA-DR expression), level of infection (HIV-1 p24antigen), astrogliosis (GFAP) and expression ofadhesion molecules (VCAM-1, E-selectin andICAM-1) was performed on frozen tissue sectionsby indirect immuno¯uorescence and avidin-biotinimmunoperoxidase staining Vectastain Elite ABCkit (Vector Laboratories, Burlingame, CA) asdescribed above and before (Persidsky et al,1996). Then, total RNA was isolated from thesame tissue pieces used for immunohistologicalevaluation. This approach enabled us to comparelevels of RNA for secretory products with tissuepathology. Ampli®ed products (at the linear rangeof the assay) were quantitated utilizing ratios ofproduct to GAPDH. The latter was used as aninternal standard for all RNA ± PCR productsscored with a PhosphorImager 425B (as describedabove).

Statistical analysesDifferences between means of the results obtainedfor different products by RT ± PCR and MAP-2staining in the separate groups of mice wereanalyzed with a two-tailed Student's t-test. Differ-ences among means were analyzed by one-wayANOVA with the different treatments as theindependent factor, when signi®cant differenceswere detected, pair-wise comparisons betweenmeans were evaluated by Dunnett's multiplecomparison test. In all analyses, the null hypothesiswas rejected at the 0.05 level.


414

Acknowledgements

We thank Dr E Masliah for helpful advice on laserscanning confocal microscopy and computer-aidedanalysis of confocal images and Dr Lennart Muckefor critical review of the manuscript. Ms KarenSpiegel's efforts are appreciated for excellenteditorial and graphic assistance. This work was

supported by National Institutes of Health grantsRO1 NS36126-01 (HEG), PO1 MH57556-01 (HEG),PO1 NS31492-01 (HEG), RO1 NS 34239-01 (HEG),the University of Nebraska Medical Center start upfunds and the Charles A Dana foundation (HEG).Jenae Limoges is a Nicholas Badami Fellow of theDepartment of Pathology and Microbiology, Uni-versity of Nebraska Medical Center.

References

Adamson CD, Wildemann B, Sasaki M, Glass JD,McArthur JC, Christov VI, Dawson TM, Dawson VL(1996). Immunologic NO synthetase: Elevation insevere AIDS dementia and induction by HIV-1 gp41.Science 274: 1917 ± 1921.

Anders KH, Guerra WF, Tomiyasu U, Verity MA, VintersHV (1986). The neuropathology of AIDS: UCLAexperience and review. Am J Pathol 124: 537 ± 558.

Brew BJ, Rosenblum M, Cronin K, Price RW (1995). AIDSdementia complex and HIV-1 brain infection: Clinical-virological correlations. Ann Neurol 38: 563 ± 570.

Budka H (1986). Multinucleated giant cells in brain: ahallmark of the acquired immune de®ciency syndrome(AIDS). Acta Neuropathol (Berl) 69: 253 ± 258.

BukrinskyM,NottetHSLM,SchmidtmayerovaH,DubrovskyL, Flanagan CR, Mullins ME, Lipton SA, Gendelman HE(1995).Regulationofnitricoxidesynthaseactivityinhumanimmunode®ciency virus type 1 (HIV-1)-infectedmonocytes: implications for HIV-associated neurologicaldisease.JExpMed181:735 ± 745.

Chao CC, Hu S, Close K, Choi CS, Molitor TW, NovickWJ and Peterson PK. (1992). Cytokine release bymicroglia: Differential inhibition by pentoxifylllin anddexamethasone. J Inf Dis. 166: 847 ± 853.

Dawson VL, Dawson TM, Uhl GR, Snyder SH (1993).Human immunode®ciency virus-1 coat proteinneurotoxicity mediated by nitric oxide in primarycortical cultures. Proc Natl Acad Sci USA 90: 3256 ±3259.

Gelbard HA, Nottet HSLM, Swindells S, Jett M, DzenkoKA, Genis P, White R, Wang L, Choi Y-B, Zhang D,Lipton SA, Tourtellotte WW, Epstein LG, GendelmanHE (1994). Platelet-activating factor: a candidatehuman immunode®ciency virus type 1-inducedneurotoxin. J Virol 68: 4628 ± 4635.

Gendelman HE, Orenstein JM, Martin MA, Ferruca C, MitraR, Phipps T, Wahl LA, Lane HC, Fauci SD, Burke DS,Skillman DR, Meltzer MS (1988). Ef®cient isolation andpropagation of human immunode®ciency virus onrecombinant colony-stimulating factor 1-treatedmonocytes. J Exp Med 167: 1498 ± 1506.

Gendelman HE, Persidsky Y, Ghorpade A, Limoges J,Stins M, Fiala M, Morrisett R (1997). Theneuropathogenesis of HIV-1 dementia. AIDS 11suppl. A): S35 ± 45.

Genis P, Jett M, Bernton EW, Boyle T, Gelbard HA,Dzenko K, Keane RW, Resnick L, Mizrachi Y, VolskyDJ, Epstein LG, Gendelman HE (1992). Cytokines andarachidonic metabolites produced during humanimmunode®ciency virus (HIV)-infected macrophage-astroglia interations: implications for theneuropathogenesis of HIV disease. J Exp Med 176:1703 ± 1718.

Giulian D, Wendt E, Vaca K, Noonan CA (1993). Theenvelope glycoprotein of human immunode®ciencyvirus type 1 stimulates release of neurotoxins frommonocytes. Proc Natl Acad Sci USA 90: 2769 ± 2773.

Glass JD, Wesselingh SL, Selnes OA, McArthur JC (1993).Clinical and neuropathologic correlation with HIV-associated dementia. Ann Neurol 33: 2230 ± 2237.

Grif®n DE, Wesselingh SL, McArthur JC (1994). Elevatedcentral nervous system prostaglandins in humanimmunode®ciency virus-associated dementia. AnnNeurol 35: 592 ± 599.

Kalter DC, Nakamura M, Turpin JA, Baca LM,Dieffenbach C, Ralph P, Gendelman HE, Meltzer MS(1991). Enhanced HIV replication in MCSF-treatedmonocytes. J Immunol 146: 298 ± 307.

Kieburtz K, Schiffer RB (1989). Neurologicalmanifestations of human immunode®ciency virusinfections. Neurol Clin 7: 447 ± 468.

Koenig S, Gendelman HE, Orenstein JM, Dal Canto MC,Pezeshkpour GH, Yungbluth M, Janotta F, Aksamit A,Martin MA, Fauci AS (1986). Detection of AIDS virusin macrophages in brain tissue from AIDS patientswith encephalopathy. Science 233: 1089 ± 1093.

Koka P, He K, Zack JA, Kitchen S, Peacock W, Fried I,Tran T, Yashar SS, Marrill JE (1995). Humanimmunode®ciency virus 1 envelope proteins induceinterleukin-1, tumor necrosis factor-a, and nitric oxidein glial cultures derived from fetal, neonatal and adulthuman brain. J Exp Med 182: 941 ± 952.

Korber BTM, Kunstman KJ, Patterson BK, Furtado M,McEvilly MM, Levy R, Wolinsky SM (1994). Geneticdifferences between blood- and brain-derived viralsequences from human immunode®ciency virus type-1 infected patients: Evidence of conserved elements inthe V3 region of the envelope protein of brain-derivedsequences. J Virol 68: 7467 ± 7481.

Lipton SA (1992). Requirement for macrophages inneuronal injury induced by HIV envelope proteingp120. Neuroreport 3: 913 ± 915.

Lipton SA (1994). Neuronal injury associated with HIV-1and potential treatment with calcium-channel andNMDA antagonists. Dev Neurosci 16: 145 ± 151.

Lipton SA, Rosenberg RA (1994). Mechanisms of disease:Excitatory animo acids as ®nal common pathway forneurological disorders. N Engl J Med 330: 613 ± 622.

Lipton SA, Gendelman HE (1995). Dementia associatedwith the acquired immunode®ciency syndrome. NEngl J Med 332: 934 ± 940.

Masliah E, Achim CL, Ge N, DeTeresa R, Terry RD,Wiley CA (1992a). Spectrum of human immuno-de®ciency virus-associated neocortical damage. AnnNeurol 32: 321 ± 329.


415

Masliah E, Ge N, Achim CL, Hansen LA, Wiley CA(1992b). Selective neuronal vulnerability in HIVencephalitis. J Neuropathol Exp Neurol 51: 585 ± 593.

Merrill JE, Martinez-Maza O (1993). Cytokines in AIDS-associated neurons and immune dysfunction. In:Neurobiology of Cytokines. Methods in Neuroscience.de Zouza EB (ed). Academic Press, San Diego: pp.243 ± 249.

Mucke L, Masliah E, Campbell IL (1995). Transgenicanimal models to assess the neuropathogenic potentialof HIV-1 proteins and cytokines. Curr Top MicrobiolImmunol 202: 187 ± 205.

Navia BA, Jordan BD, Price RW (1986a). The AIDSdementia complex. I. Clinical features. Ann Neurol 19:517 ± 524.

Navia BA, Cho E-S, Petito CK, Price RW (1986b). TheAIDS dementia complex: II. Neuropathology. AnnNeurol 19: 525 ± 535.

Nottett HSLM, Jett M, Flanagan CR, Zhai Q-H, PersidskyY, Rizzino A, Bernton EW, Genis P, Baldwin T,Schwartz J, LaBenz CJ, Gendelman HE (1995). Aregulatory role for astrocytes in HIV-1 encephalitis: anoverexpression of eicosanoids, platelet-activatingfactor, and tumor necrosis factor-a by activated HIV-1-infected monocytes is attenuated by primary humanastrocytes. J Immunol 154: 3567 ± 3581.

Parisi A, Stossell M, Pan A, Maserati R, Minoli L (1991).HIV-related encephalitis presenting as convulsantdisease. Clin Electroenceph 22: 1 ± 4.

Persidsky Y, Limoges J, McComb R, Bock P, Baldwin T,Tyor W, Patil A, Nottet HSM, Epstein L, Gelbard H,Flanagan E, Reinhard J, Pirruccello SJ, Gendelman HE(1996). Neuropathological analysis of HIV encephalitisin SCID mice. Am J Pathol 149: 1027 ± 1053.

Peterson PK, Gekker G, Hu S, Schoolov Y, Balfour HH,Chao CC (1992). Microglial cells upregulation of HIV-1 expression in the chronically infected promonocyticcell line U1: The role of tumor necrosis factor-a. JNeuroimmunol 14: 81 ± 88.

Petito CK, Cho E-S, Lemann W, Navia BA, Price RW(1986). Neuropathology of acquired immunode®ciencysyndrome (AIDS): an autopsy review. J NeuropatholExp Neurol 45: 635 ± 646.

Piani D, Spranger M, Frei K, Schaffner A, Fontana A(1992). Macrophage-induced cytotoxicity of N-methyl-D-Aspartate receptor positive neurons involvesexcitatory amino acids rather than reactive oxygenintermediates and cytokines. Eur J Immunol 22:2429 ± 2436.

Poli G, Kinter A, Justement JS (1990). Tumor necrosisfactor-a functions in an autocrine manner in theinduction of human immunode®ciency virusexpression. Proc Natl Acad Sci USA 87: 782 ± 785.

Power C, McArthur JC, Johnson RT, Grif®n DE, Glass JD,Perryman S, Chesebro B (1994). Demented andnondemented patients with AIDS differ in brain-derived human immunode®ciency virus type 1envelope sequences. J Virol 68: 4643 ± 4649.

Price RW, Brew B, Sidhs J, Rosenblum M, Sheck AC,Cleary P (1988a). The brain in AIDS: CNS HIV-1infection and AIDS dementia complex. Science 239:586 ± 592.

Price RW, Sidtis J, Rosenblum M (1988b). The AIDSdementia complex: some current questions. AnnNeurol 23: S27 ± S33.

Sharer LR, Cho ES, Epstein LG (1985). Multinucleatedgiant cells and HTLV-III in AIDS encephalopathy.Hum Pathol 16: 760.

Toggas SM, Masliah E, Rockenstein EM, Rall GF,Abraham CR, Mucke L (1994). Central nervoussystem damage produced by expression of the HIV-1coat protein gp120 in transgenic mice. Nature 367:188 ± 193.

Toggas SM, Masliah E, Mucke L (1996). Prevention ofHIV-1 gp120-induced neuronal damage in the centralnervous system of transgenic mice by the NMDAreceptor antagonist memantine. Brain Res 706: 303 ±307.

Tyor WR, Glass JD, Grif®n JW, Becker PS, McArthur JC,Bezman L, Grif®n DE (1992). Cytokine expression inthe brain during the acquired immunode®ciencydisease syndrome. Ann Neurol 31: 349 ± 360.

Wahl LM, Corcoran ML, Pyle SW, Arthur LO, Harel-Bellan A, Farrar WL (1989). Humanimmunode®ciency virus glycoprotein (gp120)induction of monocyte arachidonic acid metabolitesand interleukin-1. Proc Natl Acad Sci USA 86: 621 ±625.

Weeks BS, Lieberman DM, Johnson B (1995).Neurotoxicity of the human immunode®ciency virustype 1 tat transactivator to PC12 cells requires the tatamino acid 49 ± 58 basic domain. J Neurosc Res 42:34 ± 40.

Wesselingh SL, Power C, Glass JD, Tylor WR, McArthurJC, Farber JM, Grif®n JW, Grif®n DE (1993).Intracerebral cytokine mRNA expression in AIDS.Ann Neurol 33: 576 ± 582.

Wiley CA, Schrier RD, Nelson JA, Lampert PW, OldstoneMBA (1986). Cellular localization of humanimmunode®ciency virus infection within brains ofacquired immune de®ciency syndrome patients. ProcNatl Acad Sci USA 83: 7089 ± 7093.

Wiley CA, Achim C (1994). Human immunode®ciencyvirus encephalitis is the pathological correlate ofdementia in acquired immunode®ciency diseasesyndrome. Ann Neurol 36: 673 ± 676.

Wyss-Coray T, Masliah E, Toggas S (1996). Dysregulationof signal transduction pathways as a potentialmechanism of nervous system alterations in HIV-1gp120 transgenic mice and humans with HIV-1encephalitis. J Clin Invest 97: 789 ± 798.


416

Documents

Mechanical Weathering - CurrTech Integration