Central Nervous System PerivascularCells Are Immunoregulatory Cells

That Connect the CNS With thePeripheral Immune System

KENNETH WILLIAMS,1* XAVIER ALVAREZ,2 AND ANDREW A. LACKNER2

1Division of Viral Pathogenesis, Beth Israel Deaconess Medical Center, Harvard Medical School,Boston, Massachusetts

2Division of Comparative Pathology, Harvard Medical School, New England Regional PrimateResearch Center, Southboro, Massachusetts

KEY WORDS perivascular cells; CNS; PNS; HIV; SIV

ABSTRACT Perivascular cells are a heterogeneous population found in the centralnervous system (CNS) and the peripheral nervous system (PNS). Several terms are usedfor these cells, including perivascular cells, perivascular macrophages, perivascularmicroglia, fluorescent granular perithelial cells (FGP), or Mato cells. Different termi-nology used may reflect subpopulations of perivascular cells within different anatomicregions and experimental paradigms, neuropathological conditions, and species studied.Different terminology also points to the lack of clear consensus of what cells areperivascular cells in different disease states and models, especially with breakdown ofthe blood-brain barrier (BBB). Despite this, there is consensus that perivascular cells,although a minor component of the CNS, are important immunoregulatory cells.Perivascular cells are bone marrow derived, continuously turn over in the CNS, and arefound adjacent to CNS vessels. Thus, they are potential sensors of CNS and peripheralimmune system perturbations; are activated in models of CNS inflammation, autoim-mune disease, neuronal injury and death; and are implicated as phagocytic and pino-cytotic cells in models of stroke and hypertension. Recent evidence from our laboratoryimplicate perivascular cells as primary targets of human immunodeficiency virus (HIV)and simian immunodeficiency virus (SIV) infection in the CNS of humans and ma-caques. This article reviews current knowledge of perivascular cells, including anatomiclocation and nomenclature and putative immunoregulatory roles, and discusses newdata on the infection of these cells by SIV, their accumulation after SIV infection, and apossible role of the immune system in SIV encephalitis.GLIA 36:156–164, 2001.© 2001 Wiley-Liss, Inc.

ANATOMIC LOCATION OFPERIVASCULAR CELLS

Perivascular cells are found within the central ner-vous system (CNS) in the perivascular or Virchow-Robin space (Peters et al., 1976). Various names havebeen used to describe these cells, including perivascu-lar cells, perivascular macrophages, perivascular mi-croglia, and fluorescent granular perithelial cells (Matocells), which reflect heterogeneity within this popula-tion as well as different models of neuropathology, an-

atomic locations, and species studied. Nonetheless,there is consensus regarding their location, phenotype,

Grant sponsor: U.S. Public Health Service; Grant number: NS37654; Grantnumber: NS40237; Grant number: NS35732; Grant number: NS30769; Grantnumber: RR00168; Grant sponsor: National Multiple Sclerosis Society; Grantnumber: RG 2856-A-1.

*Correspondence to: Kenneth C. Williams, Department of Medicine, Divisionof Viral Pathogenesis, RE-113, Beth Israel Deaconess Medical Center, P.O. Box15732. Boston, MA 02215. E-mail: kenneth_williams@hms.harvard.edu

Received 27 February 2001; Accepted 25 April 2001

Published online 00 Month 2001; DOI 10.1002/glia.1105

GLIA 36:156–164 (2001)

© 2001 Wiley-Liss, Inc.

and putative immune functions. Perivascular cells area minor population in the CNS situated adjacent toendothelial cells immediately beyond the basementmembrane of medium to small vessels (Peters et al.,1976; Hickey and Kimura, 1988; Graeber et al., 1992;Streit and Graeber, 1993). Within the perivascularspace, these cells abut CNS endothelial cells and canextend long branching processes that enwrap the ves-sels to which they are apposed (Hickey and Kimura,1988b; Graeber et al., 1992; Streit and Graeber, 1993).Perivascular cells along with astrocytes and foot pro-cesses of parenchymal microglia comprise the blood-brain barrier (BBB) (Lassmann et al., 1991, 1993).Under certain pathologic conditions, including autoim-mune inflammation and viral encephalitides, perivas-cular cells can accumulate transiently and then disap-pear (Lassmann et al., 1993; Bauer et al., 1995), or theycan remain for long periods (Kida et al., 1993) (Fig. 1).Such accumulation of activated perivascular cells andthe interaction of perivascular cells with the immunesystem might contribute to the vasocentricity of lesionsobserved in multiple sclerosis and human immunode-ficiency virus (HIV) infection.

PERIVASCULAR CELLS AREBONE MARROW DERIVED

Perivascular cells are bone marrow derived and con-tinuously replaced by monocytes. Bone marrow chi-mera studies in rodents (Lassmann et al., 1986; Ma-tsumoto and Fuliwara, 1987; Hickey and Kimura,

1988b) and transplantation studies in humans (Ungeret al., 1993) show a steady rate of perivascular cellturnover in the normal noninflamed CNS. Approxi-mately 30% of perivascular cells are replaced during a3-month period in rats (Hickey et al., 1992b). This is incontrast to parenchymal microglia, which in rodentsand humans are replaced at a very low rate (Hickey etal., 1992b; Unger et al., 1993). Less than 1% of paren-chymal microglia are replaced in rats during a 90-dayperiod. This low rate of parenchymal microglia turn-over persists, even a year after bone marrow chimerismin rodents, and years after marrow transplantation inhumans (Hickey et al., 1992a; Unger et al., 1993). Thenormal turnover of perivascular cells may be exploitedby pathogens such as lentiviruses to gain entry into theCNS (Williams and Blakemore, 1990; Williams et al.,2001).

Perivascular cell turnover and accumulation havebeen studied in rodent models including experimentalallergic encephalomyelitis (EAE) and by injection ofIndia ink into the perivascular space. In rodent EAE,Lassmann demonstrated that there was less than oneperivascular cell per mm2 (n 5 3 rats) in the normalspinal cord (Lassmann et al., 1993). This number in-creased to greater than 1,000 per mm2 (n 5 3 rats) atpeak EAE and then returned to approximately 4 cellsper mm2 (n 5 3) after recovery from EAE (Lassmann etal., 1993). The perivascular cells in this study likelyconsist of both perivascular macrophages proper andmacrophages within the perivascular space that accu-mulate with BBB breakdown. The fate of the recruitedperivascular cells after peak EAE is unknown. Trans-

Fig. 1. Perivascular macrophages accumulate around CNS endo-thelial cells. Multi-label confocal microscopy of CNS tissue from arhesus macaque demonstrating Glut-1-positive endothelial cells (up-per left) and CD14-positive perivascular macrophages (upper right)

and differential interference contrast microscopy (lower left). Theindividual images are combined (lower right) demonstrating CD14-positive perivascular macrophages wrapping around Glut-1-positiveendothelial cells.

157CNS PERIVASCULAR CELLS AND PERIPHERAL IMMUNE SYSTEM

plantation studies using neonatal (Lewis 3 BN) F1 ratdonor tissues placed into the CNS of Brown Norway(BN) rats suggest that populations of transplantedCNS macrophages, including perivascular cells, canleave the CNS (Broadwell et al., 1994). Alternatively,these cells might die within the CNS, as has beendemonstrated with lymphocytes and macrophages inEAE (Nguyen et al., 1994, 1997). In other experimentalparadigms, including stroke, low-density lipoprotein(LDL) uptake, and pinocytosis of India ink, perivascu-lar cells can remain in the CNS for years (Kida et al.,1993; Mato et al., 1996; Kosel et al., 1997). Accumula-tions of HIV- and SIV-infected perivascular cells, andmultinucleated giant cells (MNGC), which most likelyarise from the fusion of perivascular cells, are foundterminally at AIDS. Whether such infected cells accu-mulate for long periods or are recent recruits in theCNS remains open. Evidence in our laboratory suggestthat SIV-infected perivascular macrophages are short-term residents that have recently arrived (Williams etal., unpublished observations). Most likely, all threemechanisms (perivascular cells leaving the CNS,phagocytic cells remaining for long periods, and apo-ptosis of recruited, virus-infected cells) are operative inthe CNS in different pathologies, including transientinflammation in nondemyelinating rat EAE, neuronalinjury and cell death, and lentiviral infection. Differentresponses suggest heterogeneity within the perivascu-lar cells in the CNS in different diseases and diseasemodels.

IMMUNOPHENOTYPE OFPERIVASCULAR CELLS

To this point in the discussion, perivascular cellshave been defined solely by their anatomic locationwithin the perivascular space. Studies in rodents andhumans have defined, in part, the immunophenotype ofperivascular cells. In rats, ED2 is a universally ac-cepted marker of perivascular cells in the normal andinflamed CNS (Dijkstra et al., 1985; Graeber et al.,1989). The ED1-3 antigens define subpopulations of ratmacrophages, where ED1 and ED2 are considered dif-ferentiation antigens and ED1 is detected on most ratmacrophage subpopulations, as well as blood mono-cytes (Dijkstra et al., 1985). Hickey produced monoclo-nal antibodies for populations of brain macrophages byimmunizing mice with rat microglia. Two such mono-clonal antibodies (mAb) TLD-1F5 and TLD-2C1A, reactwith perivascular cells, but not parenchymal microglia(Flaris et al., 1993). 1F5 appears to bind to the ED2antigen, while the 2C1A does not (Elmquist et al.,1997). In addition to ED2, rodent perivascular cells arestrongly CD45-positive by immunohistochemistry andflow cytometry (Hickey and Kimura, 1988b; Graeber etal., 1989; Sedgwick et al., 1991; Lassmann et al., 1993;Ford et al., 1995). Rat perivascular cells are OX-42(CR3)-positive (Hickey and Kimura, 1988b; Graeber etal., 1989; Lassmann et al., 1993), although populations

of these cells are identified that are OX-42 and GSA-I-B4 (ISB-4)-negative (Graeber et al., 1989). In additionto these markers of perivascular cells, CD4 is easilydetected and apparently upregulated on perivascularcells in inflammation (Flaris et al., 1993). Similarly,FcR for the constant region of immunoglobulin is alsofound on perivascular cells (Perry et al., 1985; Mo-rimura et al., 1990; Ulvestad et al., 1994a,c). Graeberand colleagues make a distinction between perivascu-lar cells that can be defined, ultrastructurally, as beingsurrounded by basement membrane and “perivascularmicroglia” in a perivascular location (Graeber and St-reit, 1990). These distinctions are more difficult tomake at the light microscopic level, especially underinflammatory conditions, with BBB disruption. Rodentperivascular cells have constitutive major histocompat-ibility class II (MHC II) expression that is further aug-mented after exposure to interferon-g/tumor necrosisfactor-a (IFN-g/TNF-a) (Lassmann et al., 1991), EAE(Hickey et al., 1987; Hickey and Kimura, 1988a; Lass-mann et al., 1993), neuronal damage (Streit et al.,1989a; Streit and Graeber, 1993), and subclinical orclinical graft-versus-host disease (GvHD) (Hickey andKimura, 1987).

A human analogue of rodent ED2 that labels perivas-cular cells selectively does not exist. In humans andnonhuman primates, perivascular cells can be distin-guished in vivo based on detection of CD14 and CD45(Williams et al., 1992, 2001; Ulvestad et al., 1994a,c).Perivascular cells in humans also have constitutiveMHC II and CD4 expression that distinguishes thesecells from parenchymal microglia (Graeber et al., 1992;Bo et al., 1994; Ulvestad et al., 1994b). In addition toCD4, perivascular cells have detectable chemokine re-ceptors, including CCR3, CCR5, and CXCR4, whichfunction as coreceptors for HIV and SIV (He et al.,1997; Rottman et al., 1997; Westmoreland et al.,1998b). Unlike rodents, perivascular cells in humansare strongly positive for nonspecific esterase, while pa-renchymal microglia are negative (Williams et al.,1992; Ulvestad et al., 1994a). Basal levels of MHC IIantigens, B7, and CD40, and FcR are increased onperivascular cells in inflammatory CNS conditions in-cluding MS (Ulvestad et al., 1994a,b,c; De Simone etal., 1995; Aloisi et al., 2000; Williams et al., 1994). Thepresence of these immune accessory molecules alongwith the location of perivascular cells next to CNSendothelium underscores their importance for phago-cytosis of immune complexes and antigen presentation.However, few immune functions by these cells havebeen demonstrated in vivo.

ANTIGEN PRESENTATION BYPERIVASCULAR CELLS

Numerous immune functions have been ascribed toperivascular cells based solely on the expression ofimmune molecules. Assessing the immune function ofperivascular cells in vitro is difficult because they are

158 WILLIAMS ET AL.

low in number and difficult to selectively isolate. Per-haps the most elegant studies of perivascular cell func-tion in vivo were performed by Hickey, who demon-strated that these cells are minimally required antigenpresenting cells (APC) in the CNS (Hickey andKimura, 1988b). This was achieved using bone marrowchimeras where BN rats were reconstituted with(Lewis 3 BN) F1 donor marrow. After allowing 2months for chimerism, perivascular cells in the CNS ofBN rats were F1 derived. When chimeras were injectedwith a Lewis rat myelin basic protein (MBP) reactiveencephalitogenic T-cell line (EP), they developed EAE.Since the perivascular cells in this system were theonly cell type in the CNS with the appropriate MHC IIfor the EP T cells, they were implicated as the residentCNS APCs (Hickey and Kimura, 1987, 1988b).

Since these studies were conducted, others have usedflow cytometry and fluorescence-activated cell sorting(FACS) for immediate isolation of ex vivo parenchymalmicroglia and brain macrophages, including perivascu-lar cells, and confirm that the nonparenchymal micro-glia population, which includes perivascular cells, con-tains potent APCs (Sedgwick et al., 1991; Ford et al.,1995). Lastly, perivascular cells in rodents synthesizeIL-1 that is a critical costimulatory cytokine for TH1CD41 T lymphocytes (Bauer et al., 1993).

PERIVASCULAR CELLS AS SENSORS OFNEURONAL INJURY AND DEATH

Studies described thus far implicate perivascularcells in transient CNS inflammation where elements ofthe peripheral immune system enter the CNS. Perivas-cular cells also become activated, as do perineuronalmicroglia, in response to neuronal injury and death.These observations have been made utilizing periph-eral nerve transection, or nerve transection followed byinjection of toxic ricin (Streit et al., 1989a,b; Streit andGraeber, 1993). Both models result in reproducibleneuronal injury and/or death without significantbreach of the BBB. Results from these studies demon-strate variable and graded responses by parenchymalmicroglia and perivascular cells. Parenchymal micro-glia activation is more rapid with neuronal death, butevident in both paradigms (Streit and Graeber, 1993).Activation of perivascular cells, as demonstrated byincreased MHC II expression, is found in both condi-tions but appears less dependent on neuronal celldeath (Graeber et al., 1990). Perivascular cell and per-ineural microglia activation is observed not onlyaround injured nerve cells, as a result of transection,but near contralateral nerves that were not transected.The observation of perivascular cell and parenchymalmicroglia activation without an inflammatory cell com-ponent led to the notion that within noninflamed CNSthere exists an immune network comprised of paren-chymal microglia and perivascular cells (Streit andGraeber, 1993). Interestingly, in addition to activatedED2 perivascular cells in the nerve transection model,

another population of ED2 perivascular cells that donot express MHC II antigens was evident. These cellswere considered a subpopulation of “immunologicallyinert” perivascular cells (Graeber et al., 1989). Thesestudies point out immune activity of perivascular cellswithout frank inflammation seen in EAE and demon-strate heterogeneity within the perivascular cell re-sponse resulting from neuronal injury and death andperhaps subpopulations of perivascular cells.

PHAGOCYTOSIS AND ACCUMULATION OFDEGRADATION PRODUCTS AND SMALLMOLECULES BY PERIVASCULAR CELLS

In addition to immune activation of perivascularcells in response to inflammation or neuronal injury/death, several reports describe populations of perivas-cular cells that can perform phagocytosis and pinocy-tosis, or respond to cytokines and LPS in the peripheralblood (Franson, 1985; Kida et al., 1993; Elmquist et al.,1997). Kida et al., (1993) showed that perivascular cellstake up India ink injected into the perivascular spacewithin hours postinjection and that ED1/ED2-positivecells containing ink can be detected up to 2 years later(Kida et al., 1993). These perivascular cells are locatedwithin the interstitial space where they can sampleantigens and potentially leave the CNS by travelingalong the perivascular space to cerebrospinal fluid tothe subarachnoid space and eventually into lymphaticsin nasal mucosa and cervical draining lymph nodes(Harling-Berg et al., 1989; Kida et al., 1993). This lo-cation might allow perivascular cells to bring CNSantigens to the peripheral immune system. Similarly,perivascular cells might also preferentially take upviruses or pathogens in the CNS. Whether perivascularcells with CNS antigens or viruses can leave and areinvolved in maintaining immune tolerance to CNS an-tigens remains speculative. Clearly, if the perivascularcells take up large foreign antigens, their ability toleave the CNS is restricted.

Mato and colleagues have characterized fluorescentgranular perithelial cells (FGP) or Mato cells, whichare identical to perivascular cells in normal rodent andhuman brain (Mato et al., 1986, 1996, 1998). Thesecells were first identified in the CNS by their endoge-nous autofluorescence, most likely occurring from thenormal accumulation of lipid break down products. Inaddition to phagocytosis of CNS constituents, thesecells take up foreign substances injected into the CNS,including horseradish peroxidase (HRP) and ferritin(Mato et al., 1981, 1984, 1986). Mato cells are ED2- andMHC II-positive and have type I and II scavenger re-ceptors (Mato et al., 1996). They accumulate aroundcerebral vessels in hypertensive rodents (Liu et al.,1994) and are detected as “class II-positive lipophages”in the spinal cord of stroke patients (Kosel et al., 1997).They have a honeycomb foam-cell phenotype in ani-mals fed high fat diets (Mato et al., 1996). The idea thatperivascular cells accumulate, and perhaps regulate,

159CNS PERIVASCULAR CELLS AND PERIPHERAL IMMUNE SYSTEM

CNS lipids (Brierley and Brown, 1982) and myelindegradation products (Franson, 1985) and function as“neuronal macrophages” (Angelov et al., 1996) extendstheir putative functions beyond their defined role asCNS APCs. Perhaps most interesting are recent obser-vations that these cells respond to immune activationproducts released by the peripheral immune systeminto the blood, and might therefore function to transmitthese responses to the CNS (Elmquist et al., 1997).

Recent evidence extends the notion of perivascularcells as sensors of CNS and PNS damage to that of cellsthat might be initiators within the CNS of acute phaseimmune responses initiated in the periphery (Elmquistet al., 1997). The production of prostaglandins isthought to be a critical step in transducing acute phasestimuli to the CNS. Cyclooxygenase catalyzes rate-lim-iting steps in the synthesis of prostaglandins after LPSor endotoxin injection. Cyclooxygenase-2 (COX-2) inparticular is an important inducible enzyme in acutephase responses. Elmquist et al., (1997) and othershave observed COX-2 reactivity in ED2 perivascularcells after LPS and endotoxin injection in the blood(Breder et al., 1995; Elmquist et al., 1997). These ob-servations implicate perivascular cells as potentialtransducers signaling acute phase stimuli from theperiphery to the CNS.

Taken together, these results point to important im-mune surveillance functions of perivascular cells: theyfunction as APCs in CNS inflammation, sensors of neu-ronal injury and death, phagocytic and pinocytotic cellstaking up CNS antigens in the perivascular space, andsensory cells sensitive to PNS injury, cytokine levels,and endotoxin outside the CNS.

In addition to the role of perivascular cells in im-mune surveillance described above, recent evidence im-plicates these cells as targets for infection by humanimmunodeficiency virus (HIV) and simian immunode-ficiency virus (SIV). Because of their location , shorthalf-life, and appropriate surface receptors, perivascu-lar cells are ideal candidates as the “Trojan horse”carrying virus to the CNS, as well as targets of produc-tive infection within the brain. Recent evidence fromour laboratory suggests that perivascular cells are themajor target of SIV infection and that the immunesystem controls the accumulation of perivascular cellsin the CNS and ultimately defines the outcome of HIVand SIV infection of the CNS.

HIV AND SIV INFECTION OF THE CNS:ROLE OF PERIVASCULAR CELLS

Approximately 30% of HIV-infected individuals de-velop subcortical dementia (HIV-associated dementia-HAD) (Sacktor et al., 1996; McArthur et al., 1999). Theclinical syndrome is characterized by cognitive, motorand behavioral changes (Price et al., 1988). Correlatesof HAD and HIV in the CNS include neuronal celldeath (Wiley and Achim, 1994) and dendritic pathology(Everall et al., 1992; Masliah et al., 1992). The best

correlate of HAD is the presence and number of inflam-matory CNS macrophages (Glass et al., 1995). Theevidence points increasingly to CNS macrophages asmediators of CNS neuropathogenesis. Data in our lab-oratory point to perivascular cells (perivascular macro-phages) as targets of lentiviral infection. Moreover, theimmune system, in particular CD81 lymphocytes, ap-pear to control the accumulation of infected macro-phages in the SIV/macaque model of neuroAIDS.

Macrophages, along with MNGC are the major com-ponent of lentiviral encephalitis caused by HIV or SIV,referred to as HIV or SIV encephalitis (HIVE andSIVE) (Koenig et al., 1986; Glass et al., 1995). Thesemacrophages, some of which are infected by HIV/SIV,are found within the meninges, choroid plexus, andperivascular space and are scattered throughout thebrain parenchyma. The SIV/HIV-infected population ofbrain macrophages has been referred to as parenchy-mal microglia and perivascular macrophages, but dou-ble-label studies designed for rigorous definition of theinfected cell types have not been performed.

We have used the SIV-infected rhesus macaquemodel of neuroAIDS to study targets of infection at thetime of initial neuroinvasion (2 weeks postinfection)and terminally in animals with AIDS. To colocalizecell-type specific markers and viral proteins and/or nu-cleic acid, we have used in situ hybridization followedby immunohistochemistry on serial sections, double-label immunohistochemistry, and multi-label confocalmicroscopy. Perivascular macrophages were differenti-ated from parenchymal microglia by their expression ofCD14 and CD45. Soon after infection we found in-creased numbers of CD141 perivascular macrophagesthat accumulate and wrap around CNS vessels (Laneet al., 1996; Williams et al., 2001) (Fig. 1).

Detailed analysis of SIV RNA and protein followedby immunohistochemistry for cell-type specific mark-ers consistently showed that CD11b1/CD141 perivas-cular macrophages were the target of SIV infection(Fig. 2). Multinucleated giant cells (which are consis-tently SIV-infected) were also strongly positive forCD14 and CD45, suggesting that they arise from thefusion of perivascular macrophages, and not parenchy-mal microglia. No evidence of SIV RNA or proteinwithin CD11b1, CD142 parenchymal microglia wasfound, nor did we detect productive viral infection ofastrocytes or endothelial cells (Fig. 2).

Infection of perivascular cells, but not of parenchy-mal microglia, has important implications for under-standing the observation that productive SIV and HIVinfection of the brain is present early, at the time ofneuroinvasion, and in terminal disease, but not duringthe intervening “asymptomatic” period (Gosztonyi etal., 1994; Sinclair et al., 1994; Smith et al., 1995).Together, these data suggest that while productivelyinfected perivascular cells are present in the CNS earlyafter infection, they are not productively infected afterpeak viremia. These cells are, however, productivelyinfected with the development of AIDS. A possible ex-planation for this is that perivascular cells that are

160 WILLIAMS ET AL.

infected coincident with viremia are the source of virusobserved in the CNS within perivascular cuffs. Afterpeak viremia and neuroinvasion, the infected CNSperivascular cells either die or exit, and are replacedduring asymptomatic infection, when viral loads aredecreased, with uninfected perivascular cells. Theknown turnover of perivascular cells in rodents sup-

ports this hypothesis. Further studies regarding thisphenomenon clearly warrant examination.

While neuroinvasion by HIV and SIV occurs withindays of infection and virtually all macaques infectedwith pathogenic stocks of SIV develop mild perivascu-lar cuffs that contain SIV-infected perivascular cells,only about 25% of HIV/SIV-infected humans or ma-

Fig. 2. Multi-label confocal microscopy demonstrating perivascularmacrophages within the CNS are the major cell type infected by SIV.Multi-label confocal microscopy demonstrating astrocytes (GFAP-yellow), SIV major capsid protein (SIVp28-blue), endothelial cells(Glut-1 red), and brain macrophages including parenchymal microglia(CD11b-green) and perivascular cells (inset, CD14-green). There is anaccumulation of CD11b1 macrophages, some of which are SIV in-

fected as demonstrated by the production of viral protein. Inset (lowerleft corner) perivascular macrophages (CD14-green) surroundingCNS endothelial cells (Glut-1 red), that are productively infected(SIVp28, blue). These data, which are representative of 5 rhesusmacaques with SIVE, consistently demonstrate that CD14-positiveperivascular macrophages, and not parenchymal microglia are themajor target of SIV infection.

161CNS PERIVASCULAR CELLS AND PERIPHERAL IMMUNE SYSTEM

caques develop HIVE/SIVE (Chakrabarti et al., 1991;McArthur et al., 1993, 1999; Lackner et al., 1994;Smith et al., 1995; Westmoreland et al., 1998a). Thisraises the question of what controls accumulation ofinfected perivascular cells in the CNS and the progres-sion to SIVE/HIVE. Historically, it has been consideredthat changes in viral sequences occur within the CNSresulting in “neurotropic” viral strains that are thedeciding factor as to whether HIVE/SIVE will develop.One common observation related to the development ofHIVE and SIVE is that it is coincident with, or follows,dysfunction of the immune system. We have recentlyinvestigated the role of the immune system, particu-larly CD81 T lymphocytes, in the development of SIVE.

For this, we have taken advantage of a humanized,chimeric anti-CD8 mAb to deplete CD81 T cells inrhesus macaques (Schmitz et al., 1999). We found that80% of animals that were CD8 depleted developedSIVE, in contrast to about 25% of untreated animals(Fig. 3). Examination of CD8 depleted versus non-depleted animals with SIVE showed that the histopa-thology was essentially identical. In addition, we foundmuch more extensive development of perivascularcuffs, consisting of perivascular macrophages, in thebrain and MNGC formation in tissues within 21 days ofinfection of CD8 depleted animals (Fig. 3). These ob-servations support a role for the immune system, spe-cifically CD81 lymphocytes, in controlling the develop-

Fig. 4. SIV infection of perivascular macrophages around dorsalroot ganglia in the spinal cord. Within the PNS, populations of satel-lite cells exist that are very similar if not identical to CNS perivascu-lar cells (based on immunophenotype and kinetics of cell turnover).Similar to perivascular cells in the CNS, these satellite cells are SIVinfected in animals with AIDS. A: Immunohistochemistry for CD68

(macrophages) showing that these cells are present and accumulatedin the dorsal root ganglion of an SIV-infected monkey. B: In situhybridization for SIV RNA showing that several of the satellite cellsin a dorsal root ganglion (DRG) are SIV RNA-positive. The accumu-lation of HIV infected macrophages in the DRG of HIV-infected indi-viduals may contribute to the development of neuropathic pain.

Fig. 3. CD8 depletion results in a high percentage of SIVE andearly accumulation of perivascular macrophages. CD8 depletion re-sults in rapid accumulation of SIV infected cells and multi-nucleatedgiant cells within lymph nodes, increased accumulation of perivascu-lar macrophages in the CNS 21 days postinfection, and a high inci-dence of SIVE. A: SIV in situ hybridization demonstrating numerousSIV-RNA-positive cells in the lymph node of a CD8 depleted animal 21

days postinfection. B: Immunohistochemistry for CD68 (macro-phages) demonstrating significant accumulation of perivascular mac-rophages in the CNS of a CD8 depleted animal 21 days postinfection.C: Immunohistochemistry for CD68 demonstrating accumulation ofperivascular macrophages and activated parenchymal microglia inthe CNS of a SIV infected, CD8-depleted animal with SIVE.

162 WILLIAMS ET AL.

ment of SIVE. The mechanism by which this occurswarrants further study.

A final point worth discussing is the role of perivas-cular cells in HIV and SIV infection of the PNS. Withinthe PNS, especially within the dorsal root ganglia andautonomic ganglia, a population of monocytic phago-cytes similar to perivascular cells exists (Vass et al.,1993). Moreover, chimeric rat studies show that theseperivascular cells have rates of turnover similar toperivascular cells within the CNS (Vass et al., 1993)and may facilitate SIV/HIV infection of the PNS. Wehave examined dorsal root ganglia of macaques thatdied with AIDS and SIVE and find an accumulation ofthese cells, some of which are viral RNA positive (Fig.4). These cells warrant further study with regard tosensory neuropathies and autonomic problems experi-enced by some HIV-infected individuals.

ACKNOWLEDGMENTS

The authors thank Ms. Kristen Toohey for graphicassistance and Mr. Douglas Pauley for editorial sug-gestions. This work was supported in part by PublicHealth Service grants NS37654, NS40237, NS35732,NS30769, and RR00168 and by a grant from the Na-tional Multiple Sclerosis Society RG 2856-A-1. A.A.Lackner is the recipient of an Elizabeth Glaser Scien-tist Award.

REFERENCES

Aloisi F, Ria F, Adorini L. 2000. Regulation of T-cell responses by CNSantigen-presenting cells: different roles for microglia and astro-cytes. Immunol Today 21:141–147.

Angelov DN, Neiss WF, Streppel M, Walther M, Guntinas-Lichius O,Stennert E. 1996. ED2-positive perivascular cells act as neuronoph-ages during delayed neuronal loss in the facial nucleus of the rat.Glia 16:129–139.

Bauer J, Berkenbosch F, Van Dam AM, Dijkstra CD. 1993. Demon-stration of interleukin-1 beta in Lewis rat brain during experimen-tal allergic encephalomyelitis by immunocytochemistry at the lightand ultrastructural level. J Neuroimmunol 48:13–21.

Bauer J, Huitinga I, Zhao W, Lassmann H, Hickey WF, Dijkstra CD.1995. The role of macrophages, perivascular cells, and microglialcells in the pathogenesis of experimental autoimmune encephalo-myelitis. Glia 15:437–446.

Bo L, Mork S, Kong PA, Nyland H, Pardo CA, Trapp BD. 1994.Detection of MHC class II-antigens on macrophages and microglia,but not on astrocytes and endothelia in active multiple sclerosislesions. J Neuroimmunol 51:135–146.

Breder CD, Dewitt D, Kraig RP. 1995. Characterization of induciblecyclooxygenase in rat brain. J Comp Neurol 355:296–315.

Brierley JB, Brown AW. 1982. The origin of lipid phagocytes in thecentral nervous system. II. The adventitia of blood vessels. J CompNeurol 211:407–417.

Broadwell RD, Baker BJ, Ebert PS, Hickey WF. 1994. Allografts ofCNS tissue possess a blood-brain barrier. III. Neuropathological,methodological, and immunological considerations. Microsc ResTechnol 27:471–494.

Chakrabarti L, Hurtrel M, Maire M, Vazeux R, Dormont D, Montag-nier L, Hurtrel B. 1991. Early viral replication in the brain ofSIV-infected rhesus monkeys. Am J Pathol 139:1273–1280.

De Simone R, Giampaolo A, Giometto B, Gallo P, Levi G, Peschle C,Aloisi F. 1995. The costimulatory molecule B7 is expressed onhuman microglia in culture and in multiple sclerosis acute lesions.J Neuropathol Exp Neurol 54:175–187.

Dijkstra CD, Dopp EA, Joling P, Kraal G. 1985. The heterogeneity ofmononuclear phagocytes in lymphoid organs: distinct macrophagesubpopulations in the rat recognized by monoclonal antibodies ED1,ED2 and ED3. Immunology 54:589–599.

Elmquist JK, Breder CD, Sherin JE, Scammell TE, Hickey WF, De-witt D, Saper CB. 1997. Intravenous lipopolysaccharide inducescyclooxygenase 2-like immunoreactivity in rat brain perivascularmicroglia and meningeal macrophages. J Comp Neurol 381:119–129.

Everall I, Gray F, Barnes H, Durigon M, Luthert P, Lantos P. 1992.Neuronal loss in symptom-free HIV infection. Lancet 340:1413–1413.

Flaris NA, Densmore TL, Molleston MC, Hickey WF. 1993. Charac-terization of microglia and macrophages in the central nervoussystem of rats: definition of the differential expression of moleculesusing standard and novel monoclonal antibodies in normal CNSand in four models of parenchymal reaction. Glia 7:34–40.

Ford AL, Goodsall AL, Hickey WF, Sedgwick JD. 1995. Normal adultramified microglia separated from other central nervous systemmacrophages by flow cytometric sorting. Phenotypic differences de-fined and direct ex vivo antigen presentation to myelin basic pro-tein- reactive CD41 T cells compared. J Immunol 154:4309–4321.

Franson P. 1985. Quantitative electron microscopic observations onthe non-neuronal cells and lipid droplets in the posterior funiculusof the cat after dorsal rhizotomy. J Comp Neurol 231:490–499.

Glass JD, Fedor H, Wesselingh SL, McArthur JC. 1995. Immunocy-tochemical quantitation of human immunodeficiency virus in thebrain: correlations with dementia. Ann Neurol 38:755–762.

Gosztonyi G, Artigas J, Lamperth L, Webster HD. 1994. Humanimmunodeficiency virus (HIV) distribution in HIV encephalitis:study of 19 cases with combined use of in situ hybridization andimmunocytochemistry. J Neuropathol Exp Neurol 53:521–534.

Graeber MB, Streit WJ. 1990. Perivascular microglia defined [letter;comment]. Trends Neurosci 13:366.

Graeber MB, Streit WJ, Kreutzberg GW. 1989. Identity of ED2-positive perivascular cells in rat brain. J Neurosci Res 22:103–106.

Graeber MB, Streit WJ, Kiefer R, Schoen SW, Kreutzberg GW. 1990.New expression of myelomonocytic antigens by microglia andperivascular cells following lethal motor neuron injury. J Neuroim-munol 27:121–132.

Graeber MB, Streit WJ, Buringer D, Sparks DL, Kreutzberg GW.1992. Ultrastructural location of major histocompatibility complex(MHC) class II positive perivascular cells in histologically normalhuman brain. J Neuropathol Exp Neurol 51:303–311.

Harling-Berg CJ, Cserr HF, Knopf PM. 1989. The role of cervicallymph nodes in the immune response to human serum albuminmicroinfused into rat cerebrospinal fluid. J Neuroimmunol 25:93–105.

He J, Chen Y, Farzan M, Choe H, Ohagen A, Gartner S, Busciglio J,Yang X, Hofmann W, Newman W, Hackay CR, Sodroski J, GabuzdaD. 1997. CCR3 and CCR5 are co-receptors for HIV-1 infection ofmicroglia. Nature 385:645–649.

Hickey WF, Kimura H. 1987. Graft-vs.-host disease elicits expressionof class I and class II histocompatibility antigens and the presenceof scattered T lymphocytes in rat central nervous system. Proc NatlAcad Sci USA 84:2082–2086.

Hickey WF, Kimura H. 1988. Perivascular microglial cells of the CNSare bone marrow-derived and present antigen in vivo. Science 239:290–292.

Hickey WF, Cohen JA, Burns JB. 1987. A quantitative immunohisto-chemical comparison of actively versus adoptively induced experi-mental allergic encephalomyelitis in the Lewis rat. Cell Immunol109:272–281.

Hickey WF, Vass K, Lassmann H. 1992a. Bone marrow-derived ele-ments in the central nervous system: an immunohistochemical andultrastructural survey of rat chimeras. J Neuropathol Exp Neurol51:246–256.

Hickey WF, Vass K, Lassmann H. 1992b. Bone marrow derived ele-ments in the central nervous system: an immunohistochemical andultrastructural survey of rat chimeras. J Neuropathol Exp Neurol5:246–256.

Kida S, Steart PV, Zhang ET, Weller RO. 1993. Perivascular cells actas scavengers in the cerebral perivascular spaces and remain dis-tinct from pericytes, microglia and macrophages. Acta Neuropathol(Berl) 85:646–652.

Koenig S, Gendelman HE, Orenstein JM, Dal Canto MC, PezeshkpourGH, Yungbluth P, Janotta F, Aksamit A, Martin M, Fauci AS. 1986.Detection of AIDS virus in macrophages in brain tissue from AIDSpatients with encephalopathy. Science 233:1089–1093.

Kosel S, Egensperger R, Bise K, Arbogast S, Mehraein P, Graeber MB.1997. Long-lasting perivascular accumulation of major histocom-patibility complex class II-positive lipophages in the spinal cord of

163CNS PERIVASCULAR CELLS AND PERIPHERAL IMMUNE SYSTEM

stroke patients: possible relevance for the immune privilege of thebrain. Acta Neuropathol (Berl) 94:532–538.

Lackner AA, Vogel P, Ramos RA, Kluge JD, Marthas M. 1994. Earlyevents in tissues during infection with pathogenic (SIVmac239) andnonpathogenic (SIVmac1A11) molecular clones of simian immuno-deficiency virus. Am J Pathol 145:428–439.

Lane JH, Sasseville VG, Smith MO, Vogel P, Pauley DR, Heyes MP,Lackner AA. 1996. Neuroinvasion by simian immunodeficiency vi-rus coincides with increased numbers of perivascular macrophages/microglia and intrathecal immune activation. J Neurovirol 2:423–432.

Lassmann H, Vass K, Brunner C, Wisniewski HM. 1986. Peripheralnervous system lesions in experimental allergic encephalomyelitis.Ultrastructural distribution of T cells and Ia- antigen. Acta Neuro-pathol (Berl) 69:193–204.

Lassmann H, Zimprich F, Vass K, Hickey WF. 1991. Microglial cellsare a component of the perivascular glia limitans. J Neurosci Res28:236–243.

Lassmann H, Schmied M, Vass K, Hickey WF. 1993. Bone marrowderived elements and resident microglia in brain inflammation.Glia 7:19–24.

Liu Y, Jacobowitz DM, Barone F, McCarron R, Spatz M, Feuerstein G,Hallenbeck JM, Siren AL. 1994. Quantitation of perivascular mono-cytes and macrophages around cerebral blood vessels of hyperten-sive and aged rats. J Cereb Blood Flow Metab 14:348–352.

Masliah E, Achim CL, Ge N, DeTeresa R, Terry RD, Wiley CA. 1992.Spectrum of human immunodeficiency virus-associated neocorticaldamage. Ann Neurol 32:321–329.

Mato M, Ookawara S, Saito-Taki T. 1986. Serological determinants offluorescent granular perithelial cells along small cerebral bloodvessels in rodent. Acta Neuropathol (Berl) 72:117–123.

Mato M, Ookawara S, Aikawa E, Kawasaki K. 1981. Studies onfluorescent granular perithelium (F.G.P.) of rat cerebral cortex—especially referring to morphological changes in aging. Anat Anz149:486–501.

Mato M, Ookawara S, Sugamata M, Aikawa E. 1984. Evidence for thepossible function of the fluorescent granular perithelial cells inbrain as scavengers of high-molecular-weight waste products. Ex-perientia 40:399–402.

Mato M, Ookawara S, Sakamoto A, Aikawa E, Ogawa T, MitsuhashiU, Masuzawa T, Suzuki H, Honda M, Yazaki Y, Watanabe E,Luoma J, Yla-Herttuala S, Fraser I, Gordon S, Kodama T. 1996.Involvement of specific macrophage-lineage cells surrounding arte-rioles in barrier and scavenger function in brain cortex. Proc NatlAcad Sci USA 93:3269–3274.

Mato M, Sakamoto A, Ookawara S, Takeuchi K, Suzuki K. 1998.Ultrastructural and immunohistochemical changes of fluorescentgranular perithelial cells and the interaction of FGP cells to micro-glia after lipopolysaccharide administration. Anat Rec 251:330–338.

Matsumoto Y, Fuliwara M. 1987. Absence of donor-type major histo-compatibility complex I antigen-bearing microglia in the rat centralnervous system of radiation bone marrow chimeras. J Neuroimmu-nol 17:71–82.

McArthur JC, Hoover DR, Bacellar H, Miller EN, Cohen BA, BeckerJT, Graham N, McArthur JH, Selnes OA, Jacobson LP, VisscherBR, Concha M, Saah A. 1993. Dementia in AIDS patients: incidenceand risk factors. Neurology 43:2245–2252.

McArthur JC, Sacktor N, Selnes O. 1999. Human immunodeficiencyvirus-associated dementia. Semin Neurol 19:129–150.

Morimura T, Neuchrist C, Kitz K, Budka H, Scheiner O, Kraft D,Lassmann H. 1990. Monocyte subpopulations in human gliomas:expression of Fc and complement receptors and correlation withtumor proliferation. Acta Neuropathol (Berl) 80:287–294.

Nguyen KB, McCombe PA, Pender MP. 1994. Macrophage apoptosisin the central nervous system in experimental autoimmune enceph-alomyelitis. J Autoimmun 7:145–152.

Nguyen KB, McCombe PA, Pender MP. 1997. Increased apoptosis of Tlymphocytes and macrophages in the central and peripheral ner-vous systems of Lewis rats with experimental autoimmune enceph-alomyelitis treated with dexamethasone. J Neuropathol Exp Neurol56:58–69.

Perry VH, Hume DA, Gordon S. 1985. Immunohistochemical localiza-tion of macrophages and microglia in the adult and developingmouse brain. Neuroscience 15:313–326.

Peters A, Palay SL, Webster HD. 1976. The fine structure of thenervous system: the neurons and supporting cells. Philadelphia:WB Saunders.

Price RW, Brew B, Sidtis J, Rosenblum M, Scheck AC, Cleary P. 1988.The brain in AIDS: Central nervous system HIV-1 infection andAIDS dementia complex. Science 239:586–592.

Rottman JB, Ganley KP, Williams K, Wu L, Mackay CR, Ringler DJ.1997. Cellular localization of the chemokine receptor CCR5. Corre-lation to cellular targets of HIV-1 infection. Am J Pathol 151:1341–1351.

Sacktor NC, Bacellar H, Hoover DR, Nance-Sproson TE, Selnes OA,Miller EN, Dal Pan GJ, Kleeberger C, Brown A, Saah A, McArthurJC. 1996. Psychomotor slowing in HIV infection: a predictor ofdementia, AIDS and death. J Neurovirol 2:404–410.

Schmitz JE, Kuroda MJ, Santra S, Sasseville VG, Simon MA, LiftonMA, Racz P, Tenner-Racz K, Dalesandro M, Scallon BJ, Ghrayeb J,Forman MA, Montefiori DC, Rieber EP, Letvin NL, Reimann KA.1999. Control of viremia in simian immunodeficiency virus infectionof CD81 lymphocytes. Science 283:857–860.

Sedgwick JD, Schwender S, Imrich H, Dorries R, Butcher GW, terMeulen V. 1991. Isolation and direct characterization of residentmicroglial cells from the normal and inflamed central nervous sys-tem. Proc Natl Acad Sci USA 88:7438–7442.

Sinclair E, Gray F, Ciardi A, Scaravilli F. 1994. Immunohistochemicalchanges and PCR detection of HIV provirus DNA in brains ofasymptomatic HIV-positive patients. J Neuropathol Exp Neurol53:43–50.

Smith MO, Heyes MP, Lackner AA. 1995. Early intrathecal events inrhesus macaques (Macaca mulatta) infected with pathogenic ornonpathogenic molecular clones of simian immunodeficiency virus.Lab Invest 72:547–558.

Streit WJ, Graeber MB. 1993. Heterogeneity of microglial andperivascular cell populations: insights gained from the facial nu-cleus paradigm. Glia 7:68–74.

Streit WJ, Graeber MB, Kreutzberg GW. 1989a. Expression of Iaantigen on perivascular and microglial cells after sublethal andlethal motor neuron injury. Exp Neurol 105:115–126.

Streit WJ, Graeber MB, Kreutzberg GW. 1989b. Peripheral nervelesion produces increased levels of major histocompatibility com-plex antigens in the central nervous system. J Neuroimmunol 21:117–123.

Ulvestad E, Williams K, Mork S, Antel J, Nyland H. 1994a. Pheno-typic differences between human monocytes/macrophages and mi-croglial cells studied in situ and in vitro. J Neuropathol Exp Neurol53:492–501.

Ulvestad E, Williams K, Bo L, Trapp B, Antel J, Mork S. 1994b. HLAclass II molecules (HLA-DR, DP, DQ) on cells in the human CNSstudied in situ and in vitro. Immunology 82:535–541.

Ulvestad E, Williams K, Vedeler C, Antel J, Nyland H, Mork S, MatreR. 1994c. Reactive microglia in multiple sclerosis lesions have anincreased expression of receptors for the Fc part of IgG. J Neurol Sci121:125–131.

Unger ER, Sung JH, Manivel JC, Chenggis ML, Blazar BR, Krivit W.1993. Male donor-derived cells in the brains of female sex-mis-matched bone marrow transplant recipients: A Y chromosome spe-cific in situ hybridization study. J Neuropathol Exp Neurol 52:460–470.

Vass K, Hickey WF, Schmidt RE, Lassmann H. 1993. Bone marrow-derived elements in the peripheral nervous system. An immunohis-tochemical and ultrastructural investigation in chimeric rats. LabInvest 69:275–282.

Westmoreland SV, Halpern E, Lackner AA. 1998a. Simian immuno-deficiency virus encephalitis in rhesus macaques is associated withrapid disease progression. J Neurovirol 4:260–268.

Westmoreland SV, Rottman JB, Williams KC, Lackner AA, SassevilleVG. 1998b. Chemokine receptor expression on resident and inflam-matory cells in the brain of macaques with simian immunodefi-ciency virus encephalitis. Am J Pathol 152:659–665.

Wiley CA, Achim C. 1994. Human immunodeficiency virus encepha-litis is the pathological correlate of dementia in acquired immuno-deficiency syndrome. Ann Neurol 36:673–676.

Williams AE, Blakemore WF. 1990. Monocyte-mediated entry ofpathogens into the central nervous system. Neuropathol Appl Neu-robiol 16:377–392.

Williams K, Barr-Or A, Ulvestad E, Olivier A, Antl JP, Yong WV.1992. Biology of adult human microglia in culture: comparisonswith peripheral blood monocytes and astrocytes. J Neuropathol ExpNeurol 51:538–549.

Williams K, Ulvestad E, Antel JP. 1994. B7/BB-1 expression on adulthuman microglia studied in vitro and in situ. Eur J Immunol24:3031–3037.

Williams KC, Corey S, Westmoreland SV, Pauley D, Knight H, de-Baker C, Alvarez X, Lackner AA. 2001. Perivascular macrophagesare the primary cell type productively infected by simian immuno-deficiency virus in the brains. J Exp Med 193:905–916.

164 WILLIAMS ET AL.