blood-brain barrier biology and methodology6)/556-569.pdf · blood-brain barrier biology and...

Review

Blood-brain barrier biology and methodology

William M Pardridge*,1

1Department of Medicine, UCLA School of Medicine, Los Angeles, California, CA 90095-1682, USA

The blood-brain barrier (BBB) is formed by epithelial-like high resistance tightjunctions within the endothelium of capillaries perfusing the vertebrate brain.Because of the presence of the BBB, circulating molecules gain access to braincells only via one of two processes: (i) lipid-mediated transport of smallmolecules through the BBB by free diffusion, or (ii) catalyzed transport. Thelatter includes carrier-mediated transport processes for low molecular weightnutrients and water soluble vitamins or receptor-mediated transport forcirculating peptides (e.g., insulin), plasma proteins (e.g., transferrin), orviruses. While BBB permeability, per se, is controlled by the biochemicalproperties of the plasma membranes of the capillary endothelial cells, overallbrain microvascular biology is a function of the paracrine interactions betweenthe capillary endothelium and the other two major cells comprising themicrocirculation of brain, i.e., the capillary pericyte, which shares thebasement membrane with the endothelial cell, and the astrocyte foot process,which invests 99% of the abluminal surface of the capillary basementmembrane in brain. Microvascular functions frequently ascribed to thecapillary endothelium are actually executed by either the capillary pericyteor the capillary astrocyte foot process. With respect to BBB methodology, thereare a variety of in vivo methods for studying biological transport across thisimportant membrane. The classical physiologic techniques may now becorrelated with modern biochemical and molecular biological approachesusing freshly isolated animal or human brain capillaries. Isolated braincapillary endothelial cells can also be grown in tissue culture to form an `in vitroBBB' model. However, BBB research cannot be performed using only the invitro BBB model, but rather it is necessary to correlate observations made withthe in vitro BBB model with in vivo studies.

Keywords: endothelium; cell culture; pericyte; astrocyte; biological transport

Introduction

Blood-brain barrier (BBB) permeability is regulatedby brain capillary endothelial transport properties.Microvascular biology in the central nervous system(CNS) is regulated by paracrine interactions be-tween the capillary endothelium in brain and itsneighboring cells: the pericyte (which shares thecapillary basement membrane with the endothe-lium), the astrocyte foot process (which invests 99%of the abluminal surface of the brain capillary), thesmooth muscle cell (which invests the endotheliumof pre-capillary arterioles), and neuronal endings(which directly innervate either the capillaryendothelium or the astrocyte foot process investing

the capillary endothelium) (Cohen et al, 1995;Paspalas and Papadopoulos, 1996).

In this overview of BBB methodology andbiology, two general theses are discussed. First,with respect to BBB methodology, it is presently notpossible to accurately investigate BBB transportprocesses using `in vitro BBB models' without invivo correlation. In the in vitro models, braincapillary endothelial cells are grown in tissueculture and transport across the endothelial mono-layer is investigated, with or without astrocyte co-cultures (Dehouck et al, 1990). The in vitro BBBmodel can be a useful adjunct when performed inconjunction with in vivo investigations of the BBB.However, when in vitro BBB models are used inisolation, the limitations of this model reviewedbelow become a limiting factor in interpretation ofthe data. Second, with regard to BBB biology, the

*Correspondence: WM PardridgeReceived 12 February 1999; revised 21 April 1999; accepted26 April 1999

Journal of NeuroVirology (1999) 5, 556 ± 569

ãhttp://www.jneurovirology.com

1999 Journal of NeuroVirology, Inc.

brain microvasculature is comprised of at leastthree cells (endothelia, pericytes, and astrocyte footprocesses), and important brain microvascularfunctions are performed by either the pericyte orthe astrocyte foot process. However, these functionsare frequently assigned to the endothelial cell,because the role played by either the pericyte orthe astrocyte foot process is ignored. While theendothelial cell regulates BBB permeability, per se,overall brain microvascular function is regulated bythe paracrine interactions between the endothelialcell, the pericyte, and the astrocyte foot process, aswell as direct neuronal innervation of the micro-vasculature.

Blood-brain barrier methodology

Until the late 1960's, blood-brain barrier function invivo was studied with vital dyes such as trypanblue. These large molecular weight, highly anionicdyes that are tightly bound by albumin, do not crossthe BBB in vivo with two exceptions. First, there area half dozen tiny areas of the brain, calledcircumventricular organs (CVOs), such as thechoroid plexus, that are perfused by porousfenestrated capillaries. Second, in pathophysiologicconditions, there may be BBB disruption, leading tofocal uptake of vital dyes. Subsequently, animals

were administered horseradish peroxidase (HRP)followed by peroxidase histochemistry of braintissue sections, and visualization of brain at eitherthe light microscopic or electron microscopic level(Figure 1). Around 1970, radioisotopes were intro-duced and these, in conjunction with physiologictechniques, or brain imaging methods (Figure 1),allowed for the quantitative assessment of BBBpermeability in vivo. In the 1970s, techniques forisolation of brain microvessels were introduced andthis allowed for subsequent isolation of BBB plasmamembranes for biochemical assays and radiorecep-tor studies (Pardridge, 1998b). However, braincapillaries, isolated with either a mechanicalhomogenization technique or an enzymatic homo-genization method, are not metabolically viable andhave very low levels of ATP (Lasbennes and Gayet,1983). The low ATP concentration in braincapillaries freshly isolated with an enzymatichomogenization technique is somewhat of anenigma since it is possible to isolate other cells ofthe body with enzymatic homogenization techni-ques and these cells have normal levels of ATP.Isolated brain capillaries provide an intact mem-brane preparation of the BBB and the brainmicrovasculature, and allow for experiments thatcannot be done with other methodologies. In the1980s, brain capillary endothelial cells were platedin tissue culture and this led to the early in vitro

Figure 1 Linear evolution of blood-brain barrier methodology from vital dyes to molecular biology. From Pardridge (1998a).

Blood-brain barrierWM Pardridge

557

BBB models (De Bault et al, 1979; Bowman et al,1983). However, it became apparent that braincapillary endothelial cells grown in tissue culturelack many of the properties of the BBB in vivo andthis gave rise to endothelial/astrocyte co-cultures.Initially, it was thought that soluble factors secretedby cultured astrocytes could be added to endothe-lial cultures to induce BBB properties, but in thelast 20 years, this approach has not led to anysigni®cant progress in the molecular identi®cationof the putative factors. This lack of progress hasgiven rise to the hyopthesis that in order to observeinduction of BBB properties in endothelial cellsgrown in culture, it is necessary for there to bephysical contact between astrocytes and endothelialcells in culture (Meyer et al, 1991). The physicalcontact can arise from either the use of mixedcultures, where endothelial cells and astrocytes areboth plated on the same dish in culture, or in a co-culture system where endothelial cells and astro-cytes are plated on opposite sides of a porous ®lterand the diameter of the pores are suf®ciently largeto allow for protrusion of astrocyte processesthrough the pores so that there can be cell-cellcontact between the astrocyte foot process and theendothelium. These endothelial-astrocyte contactsin tissue culture culture are only focal contacts anddo not approximate the complete investment of theendothelial cell by astrocyte foot processes thattakes place in vivo. In the 1990s, the production oftarget speci®c antibodies and target speci®c cDNAshas allowed for biochemical and molecular biologi-cal investigations of blood-brain barrier transportprocesses (Figure 1). A particularly valuable ap-proach is to use speci®c antibodies or cDNAs toinvestigate the expression of proteins or mRNAs infreshly isolated brain microvessels. As reviewed insubsequent sections of this chapter, the immunocy-tochemistry of cytocentrifuged brain capillariesallows for differentiation between endothelial cells,pericytes, and astrocyte foot processes in a way thatcannot be readily done with immunohistochemistryof brain tissue sections. The isolation of poly(A)+

mRNA from freshly isolated brain capillaries allowsfor direct molecular biological investigations of thecells comprising the BBB in vivo. Since the braincapillary endothelial cell volume is only 1 mL/gbrain, the endothelial volume in brain is only 0.2%of the total cell volume in brain. Therefore, it isgenerally not possible to perform BBB molecularbiological studies from poly(A)+ mRNA isolatedfrom whole brain extracts. Rather, it is necessaryto ®rst isolate brain capillaries and then isolate thecapillary derived poly(A)+ mRNA for BBB molecu-lar biological studies (Boado and Pardridge, 1991).

Despite the array of methodologies available forperforming BBB investigations (Figure 1), there hasbeen a long tradition within BBB research to studythis important biological membrane with a singlemethodology. One lab may exclusively use an

internal carotid artery perfusion method, whereasanother lab may exclusively use an intravenousinjection technique, or another lab may exclusivelyuse quantitative autoradiography (QAR). The latestvariation in this trend is to study BBB transportprocesses exclusively with the in vitro BBB cellculture model. Brain capillary endothelial cells canbe grown as a monolayer in tissue culture (Figure2B) and these cells are grown on ®lters withspeci®ed porosity and then placed in side-by-sidetransport chambers for studies of transport acrossthe monolayer (Figure 2A), in a way that wasoriginally performed with drug studies of skintransport (Audus et al, 1990). A given radiolabeleddrug may be placed on one side of the transportchamber and movement across the endothelialmonolayer may be investigated by sampling the`acceptor chamber.' From these clearance studiesperformed with ®lter plus endothelial monolayer aswell as ®lter alone, the endothelial monolayerpermeability-surface area (PS) product may becomputed (Pardridge et al, 1990). The PS productfor a series of drugs, as well as D-glucose or L-DOPA,with normalization for molecular weight (Mr), areplotted versus the 1-octanol lipid partition coef®-cient (PC) as shown in Figure 2B. This linearcorrelation between PS products across the en-dothelial monolayer and the lipid solubility couldbe taken as evidence for the reliability of the in vitroBBB model for assessing drug transport. However,quantitative analysis of this data reveals thefollowing two caveats. First, the slope of the plotof the in vitro PS product and the lipid PC isapproximately 150-fold smaller than the slopegenerated by measuring the BBB PS product withan in vivo technique, the internal carotid arteryperfusion technique (Pardridge et al, 1990). That is,the in vitro BBB is 4150-fold leakier than the invivo BBB. Accordingly, the transport of drugs acrossthe BBB that occurs via lipid mediated transport isover-estimated by more than a 100-fold by the invitro BBB model (Figure 2B). Similarly, the elec-trical resistance across the endothelial monolayergrown in the absence of astrocytes is approximately80 O.cm2. This is approximately 100-fold less thanthe electrical resistance across the BBB in vivowhich is estimated to be 8000 O.cm2 (Smith andRapoport, 1986). Although the electrical resistanceacross pial vessels on the surface of the brain is ofthe order of 1000 ± 2000 O.cm2, these vessels do nothave a complete BBB, as the vessels are investedonly by a glial limitans. The electrical resistanceacross pial vessels is more than fourfold lower thanthe electrical resistance across parenchymal vesselswith an intact BBB. The electrical resistance acrossan endothelial monolayer in tissue culture is as highas 400 ± 1000 O.cm2 using special substrata such astype 4 collagen and ®bronectin (Tilling et al, 1998),but these values are still a log order lower than theelectrical resistance across the BBB in vivo.


558

The second observation noted in the plot of the invitro PS product versus the lipid PC is that thevalues for D-glucose and L-DOPA, two moleculesthat undergo carrier-mediated transport through theBBB in vivo via the hexose carrier and the largeneutral amino acid carrier, respectively, fall on thelipid solubility trend line (Figure 2B). Conversely,when the PS product for these two molecules ismeasured with an in vivo technique, the internalcarotid perfusion method, the PS product for thesetwo molecules is 4 ± 5 log orders above the lipidsolubility trend line (Pardridge et al, 1990). In otherwords, there is a profound down-regulation of theactivity of the glucose or neutral amino acid carrierin the in vitro BBB model. This down-regulation canalso be seen with molecular biological techniqueswherein the mRNA for the GLUT1 glucose trans-porter in capillary endothelial cells in tissue cultureand in freshly isolated bovine brain capillaries ismeasured in parallel (Boado and Pardridge, 1990).The Northern blot in Figure 2B shows the signalrepresenting the GLUT1 glucose transporter mRNAin isolated brain capillaries is more than 100-foldgreater than the transcript in endothelial cells

grown in tissue culture (lane 1, Figure 2B) or intotal brain (lane 3, Figure 2B). The greater density ofthe GLUT1 glucose transporter in isolated braincapillaries as compared to total brain re¯ects thefact that this transcript is expressed in normal brainprincipally at the endothelium in brain and not inneurons or astrocytes in vivo. The down-regulationof carrier-mediated transport (CMT) systems in thein vitro BBB model means that drug candidates thatcross the BBB via CMT (e.g. L-DOPA) will be missedby drug screens using in vitro BBB models. Inaddition to carrier mediated transport systems,there is also a marked down-regulation of enzymesin the in vitro BBB model, that normally comprisethe `enzymatic BBB' in vivo (van Bree et al, 1988;Thompson and Audus, 1994).

In an attempt to make the in vitro BBB modelmore accurately re¯ect the BBB in vivo, there havebeen several innovations in the in vitro BBB model.One innovation is to use a homologous co-culturesystem (Hurwitz et al, 1993), whereby human fetalbrain primary astrocytes are grown with humanumbilical vein endothelial cells in primary culturein a co-culture format using a three micron pore. It

Figure 2 (A) Bovine brain capillary endothelium in tissue culture (BCEC) is shown in the top panel. The endothelial cells are grownon ®lters which are then placed in side-by-side transport chambers (bottom panel) to establish an `in vitro BBB model.' (B) Thepermeability-surface area (PS) product for glucose, L-DOPA, and radiolabeled drugs is plotted versus the 1-octanol lipid partitioncoef®cient (PC). The PS products were measured with the in vitro BBB model system and normalized for molecular weight (Mr). Thecorrelation line was drawn by eye, whereas the slope and intercept were determined by linear regression analysis. A Northern blotshowing profound down-regulation of the GLUT1 glucose transporter mRNA in the BCEC is shown in lane 1. The GLUT1 glucosetransporter Northern blot signal for freshly isolated bovine brain capillaries or total bovine brain is shown in lanes 2 and 3. FromPardridge et al (1990) and Boado and Pardridge (1990).


559

was found that the GLUT1 glucose transporter in theendothelial cell could be induced by astrocytes ifthe ®lter contained a three micron pore that allowedfor the astrocyte foot processes to pass through andcome in contact with the endothelial cells. Whenthe co-cultures were grown on ®lters with 0.45micron pores, there was no induction of the GLUT1glucose transporter because this pore diameter wastoo small to allow for passage of the astrocyte footprocess (Hurwitz et al, 1993). More recently, hollow®ber tubes with diameters450 microns, and poresizes of 0.5 microns have been used to establishendothelial astrocyte co-cultures (Stanness et al,1997). Given the small size of the pore diameter (0.5microns), there may be minimal cell-cell contactbetween the astrocytes grown on the exterior side ofthe cylinder with the endothelial cells grown on theinterior side. This model has unusual propertiesthat are not found at the BBB in vivo. For example,the endothelial permeability coef®cient (Pe) for L-aspartic acid, an acidic amino acid that normallydoes not cross the BBB in vivo, is higher than theBBB Pe value for the theophylline, a lipid solublesmall molecule that is freely transported across theBBB in vivo (Stanness et al, 1997). In this model, thepermeability coef®cient for sucrose is quite low,approximately 2061078 cm/s after a 10 min perfu-sion, but increases 25-fold to 50061078 cm/s after a60 min perfusion. This increase in Pe in the in vitromodel suggests the in vitro BBB becomes leaky witha 60 min perfusion.

In summary, after 20 years, there are still majorlimitations in establishing an in vitro BBB model.This model is a useful adjunct in investigating BBBtransport processes when used in conjunction withone of the many in vivo methodologies available(Figure 1). However, when the in vitro model is usedalone, it is dif®cult to extrapolate the data obtainedwith this model to the BBB permeability propertiesin vivo. In many cases, data obtained with in vitroBBB culture models replicate more the properties of®broblasts in culture rather than the BBB in vivo.

Blood-brain barrier biology

Three cell model of the brain microvasculatureThe three cell model of the brain microvasculatureis depicted in Figure 3, wherein the centralendothelial cell, which regulates BBB permeability,shares intimate relationships with two other cells,the pericyte and the astrocyte foot process. Thepericyte is a phagocytic cell that shares a commonbasement membrane with the capillary endothelialcell (Cancilla et al, 1972). In isolated brain capillarypreparations, there is approximately one pericytefor every three endothelial cells. An enzyme, g-glutamyl transpeptidase, that is frequently used as a`brain endothelial marker,' is also localized topericytes in the brain microvasculature (Risau etal, 1992).

The astrocyte foot process invests 99% of thebrain surface of the capillary basement mem-brane. The intimate relationship between theastrocyte foot process and the endothelial cell inbrain was illuminated by recent confocal micro-scopy studies performed by Kacem et al (1998).Antibodies to either an endothelial marker, theGLUT1 glucose transporter, or to an astrocyte footprocess marker, glial ®brillary acidic protein(GFAP), were used in immuno¯uorescent con-focal microscopy of brain sections with three-dimensional reconstruction of serial confocalsections. The endothelial GLUT1 glucose trans-porter is seen in yellow in panel 1 of Figure 4and the astrocyte foot process marker, GFAP, isseen as red in panel 1 of Figure 4. There is aclear demarcation between the cellular localiza-tion of GLUT1 glucose transporter immunoreac-tivity and astrocyte GFAP immunoreactivity,indicating the GLUT1 glucose transporter isexpressed exclusively within the endothelial celland the GFAP is produced exclusively within theastrocyte foot process. The extensive investmentof a microvessel by astrocyte foot processes isshown in panel 2 of Figure 4 and the outline of amicrovessel can be clearly delineated using anantibody to an astrocyte foot process marker suchas GFAP. Indeed, when performing immunocyto-chemistry of brain tissue sections with GFAPantibodies, those antibodies can frequently givean immunocytochemical picture comparable tothat obtained with an antibody to an endothelialantigen (Stewart, 1994). The analysis by Kacem etal (1998) shows astrocyte foot processes forming

Figure 3 The three cell model of the brain microvasculatureemphasizes the paracrine interactions between the braincapillary endothelium, the pericyte, and the astrocyte footprocess. From Pardridge (1999).


560

rosette-like structures at the albuminal surface ofthe endothelium (panel 3, Figure 4). Astrocyteswere found to send processes to only singleendothelial cells, although one endothelial cellmay receive inputs from multiple astrocytes(Kacem et al, 1998). The rosette-like structureformed by astrocyte foot processes surroundingthe capillary endothelium is seen at low magni-®cation in panel 4 of Figure 4 and at highmagni®cation in panel 5 of Figure 4. A completeastrocyte/endothelial interaction is shown inpanel 6 of Figure 4, wherein a single astrocytefoot process is seen to invest a single endothelialcell.

The distance between an astrocyte foot processand a capillary endothelium is only 200 Angstromsin vivo or 20 nanometers (Paulson and Newman,1987). This space is ®lled by the thickness of thecapillary basement membrane. Astrocytes signalneurons in tissue culture (Nedergaard, 1994), andsignal endothelial cells in culture and most likelysend signals to these cells in vivo. It has yet to beshown that brain endothelial cells send signals toastrocytes or even neurons in vivo. With respect tothe myocardium, it is known that myocardialautoregulation is modulated by cyclic AMP regula-tion of myocardial contractility via endothelial-derived fctors (McClellan et al, 1993). Myocardialcontractile function is in part regulated by themyocardial endothelial cell. The extent to whichneuronal transmission and brain function is regu-lated by signals emanating from capillary endothe-lial cells in brain is at present unknown.

P-glycoprotein in astrocyte foot processesP-glycoprotein is an active ef¯ux system that isresponsible for pumping drugs out of cells. Im-munoreactive p-glycoprotein was found in themicrovasculature in tissue sections of brain (Cor-don-Cardo et al, 1989) and in isolated braincapillaries (Jette et al, 1993). On the basis of thesestudies, it was assumed that immunoreactive p-glycoprotein at the brain microvasculature wasendothelial in origin and regulated drug transportacross the BBB in vivo. While this mechanism mayoperate in rodents, recent studies with isolatedhuman brain microvessels indicate human brainmicrovascular p-glycoprotein immunoreactivity isessentially entirely con®ned to the astrocyte footprocess with minimal, if any, expression in thecapillary endothelium (Figure 5). In these studies,an antibody was used to the GLUT1 glucosetransporter endothelial marker, an antibody wasused to the GFAP astrocyte foot process marker, andthe MRK16 murine monoclonal antibody to humanp-glycoprotein was used to identify this drugtransport system (Pardridge et al, 1997). Confocalmicroscopy of cytocentrifuged isolated humanbrain capillaries was performed to allow for co-localization of either the GLUT1 glucose transporterand GFAP or p-glycoprotein and GFAP. Confocalimmuno¯uorescence microscope images of duallabeled un®xed human brain capillaries is shownin Figure 5 (Golden and Pardridge, 1999). Thediscontinuous staining pattern of MRK16 is demon-strated by the single channel scan for texas red (A),while a single channel scan for ¯uorescein (B)

Figure 4 Immuno¯uorescent confocal microscopy shows the three-dimensional relationship between brain capillary endotheliumand astrocyte foot processes. In panel 1, the endothelial GLUT1 glucose transporter is shown in yellow, and the astrocyte foot processmarker, glial ®brillary acidic protein (GFAP) is shown in red. The monoclonal antibody to GFAP allows for illumination of themicrovascular astrocyte foot processes as shown in panels 2 and 4. Astrocyte foot processes form rosette-like structures at thealbuminal surface of the endothelium as schematized in panel 3 and shown in high magni®cation in panel 5. Panels 1 ± 5 are fromKacem et al (1998) and panel 6 is from Blumcke et al (1995).


561

demonstrates the identical discontinuous stainingpattern of anti-GFAP labeling of astrocyte footprocesses. Dual channel scanning of the ®eld (C)demonstrates complete co-localization of MRK16,indicated by the yellow signal, where the red andgreen signals overlap. In contrast, discontinuouslabeling of immunoreactive p-glycoprotein byMRK16 (D) is not co-localized with the continuouslabeling provided by an anti-GLUT1 glucose trans-porter antiserum (E). The dual channel scan (F)demonstrates preservation of the texas red signal forMRK16 in a discontinuous pattern. Similar resultswere obtained with capillaries dual-labeled withantibodies against GFAP (G) and GLUT1 glucosetransporter (H). The similarity of the overlappingimage provided by dual-channel screening forGFAP and GLUT1 localization (I) to that observedfor p-glycoprotein and GLUT1 glucose transporter(F) indicates that the discontinuous localizationpatterns for GFAP and p-glycoprotein on theastrocyte foot processes differ from that of the

continuous staining of the endothelial glucosetransporter. Results of this study indicate that theimmunoreactive p-glycoprotein is localized to theastrocyte foot processes of the albuminal surface ofthe human brain microvasculature. The localizationof p-glycoprotein drug ef¯ux at the astrocyte plasmamembrane may alter the distribution of drugsbehind the BBB without directly affecting brainendothelial permeability to drugs. P-glycoprotein isan example of a brain microvascular antigen that islocalized to the astrocyte foot process rather thanthe endothelium.

Does the capillary pericyte in brain have acontractile function?Capillary pericytes in brain are said to have acontractile function and to be the smooth muscleanalog of the capillary. Contractile cells such assmooth muscle cells express the a-actin isoform.The a-actin gene is expressed in brain capillarypericytes grown in tissue culture (Herman andD'Amore, 1985). However, immunoreactive a-actinis not expressed in pericytes in vivo in microvesselsof bovine retina (Nehls and Drenckhahn, 1991).Similarly, the pericytes in vivo of brain capillariesdo not express immunoreactive a-actin, as demon-strated by the immunocytochemical studies usingcytocentrifuged isolated bovine brain capillariesand a monoclonal antibody to a-actin (Figure 6). Asshown in (A), smooth muscle cells of pre-capillaryarterioles in isolated bovine brain microvessels arehighly immunoreactive for a-actin. Higher magni®-cation (B and C) shows immunoreactive a-actin(Boado and Pardridge, 1994) is localized only tosmooth muscle cells of pre-capillary arterioles andis not found at the capillary level in eitherendothelial cells or pericytes. The pericyte nucleiare distinguishable from endothelial nuclei, whichhave a more ovoid shape, and the pericyte nucleiare indicated in B and C by arrows. The absence ofimmunoreactive a-actin in pericytes of either brainor retina microvessels in vivo suggests braincapillary pericytes do not have a contractilefunction in vivo under normal conditions. Inpathophysiologic state such as high ¯ow/shearstates (Liwnicz et al, 1990), it is possible that a-actin is expressed in capillary pericytes, but this hasyet to be demonstrated in vivo. The absence of a-actin expression in capillary pericytes in vivo, andthe abundant expression of a-actin in capillarypericytes grown in tissue culture, again emphasizesthe need for in vitro/in vivo correlations whenexamining brain microvascular cells in tissueculture alone.

Antigen presentation at the brain microvasculaturein vivo: role of the capillary pericyteThe CNS is said to be an immune priviliged site,because of the minimal antigen presentation that isassumed to occur in brain (Pollack and Lund, 1990).

Figure 5 Confocal ¯uorescent microscopic images of duallabeled isolated human brain capillaries cytocentrifuged to aglass slide and stained with antisera to the GLUT1 glucosetransporter (an endothelial marker), to GFAP (an astrocyte footprocess marker), or p-glycoprotein. (A) P-glycoprotein immunor-eactivity, (B) GFAP immunoreactivity, (C) Co-localization of p-glycoprotein and GFAP, (D) P-glycoprotein immunoreactivity,(E) GLUT1 glucose transporter immunoreactivity, (F) Co-localization of p-glycoprotein and GLUT1 glucose transporter,(G) GFAP immunoreactivity, (H) GLUT1 glucose transporterimmunoreactivity, (I) GFAP and GLUT1 glucose transporterimmunoreactivity overlap. Images (A ± C): 20 micron620micron; images (D ± I): 40 micron640 micron. From Goldenand Pardridge (1999).


562

Class II histocompatibility antigens, such as the DR-antigen, are very low in brain (Wong et al, 1984).However, if DR-antigen immunoreactivity is loca-lized primarily to the microvasculature in brain andnot to brain cells, then overall brain DR-immunor-eactivity will be low, owing to the small volume ofbrain represented by microvascular cells. Thepresence of the DR-antigen in human brain micro-vessels was examined with cytocentrifuged humanbrain capillaries using a monoclonal antibody to thehuman DR-antigen (Pardridge et al, 1989). Asshown in A of Figure 8, there is abundant DR-antigen immunoreactivity in the pre-capillaryarteriolar smooth muscle cells of microvesselsisolated from brains of subjects dying withoutneurologic disease. Conversely, when microvesselswere isolated from Multiple Sclerosis (MS) plaquetissue, there was abundant DR-antigen immunor-eactivity at the capillary level, but this was notfound in endothelial cells, but was found in

capillary pericytes, as shown at low magni®cationin B and at high magni®cation in panel C (Figure 7).The expression of the DR-antigen in capillarypericytes, and not in capillary endothelial cells,means that antigen presentation in brain occursbeyond the BBB as shown in D of Figure 7.Activated lymphocytes may undergo diapedesisthrough the capillary endothelial barrier (Vass etal, 1986), and present antigen to either capillarypericytes or arteriolar smooth muscle cells. Theexpression of the DR-antigen in capillary pericytesmay be under the regulation of cytokines as theincubation of isolated rat brain capillaries withinterferon g for 18 h results in an increase inpericyte DR-antigen immunoreactivity (Dore-Duffyet al, 1994).

Multiple sclerosis and the blood-brain barrierThe earliest lesion of MS is around brain bloodvessels and this observation was noted in the early1900's (Broman, 1964; Daniel et al, 1981). Similarly,the earliest lesion is microvascular in experimentalallergic encephalomyelitis (EAE) and in post-vaccinal disease or post-infectious encephalomye-litis. The search for autoantibodies to the endothe-lial cells in MS has been generally negativealthough these searches have employed humanumbilical endothelial cells, instead of capillariesisolated from MS plaque tissue (Tsukada et al,1985). When guinea pigs are immunized with ratbrain capillary membranes, EAE develops and theearliest lesion is macrophages in®ltrating postcapillary venules and demyelination, even thoughthe animals were not initially immunized withmyelin antigens (Tsukada et al, 1987). Consideringthe earliest legion of MS is microvascular andconsidering the ease with which microvessels maybe isolated from MS human autopsy brain plaquetissue (Pardridge et al, 1987; Washington et al,1994), it is somewhat surprising that there are sofew studies on the immunology and biochemistry ofcapillaries isolated from MS plaque tissue.

Endothelial receptor-mediated transcytosis systemsCarrier mediated transport (CMT) systems exist fora small molecule nutrients such as glucose, aminoacids, monocarboxylic acids, purine bases, choline,thyroid hormones, and several water solublevitamins. CMT processes use stereospeci®c trans-porters localized within the endothelial luminaland albuminal membranes that operate withinmilliseconds. In addition, certain circulating pep-tides such as insulin, transferrin, leptin, andinsulin-like growth factors (IGF) undergo transportthrough the capillary endothelium via receptormediated transcytosis (RMT) processes which occurwithin minutes. RMT is comprised of three steps:receptor mediated endocytosis of the circulatingpeptide at the luminal side of the capillaryendothelium, movement through the 200 nm of

Figure 6 Light microscopic immunocytochemistry of cytocen-trifuged bovine brain capillariers immunostained with either amouse monoclonal antibody to the a-actin isoform (A, B and C)or a mouse IgG2a isotype control (D). There is speci®c staining ofthe a-actin isoform in the smooth muscle cells of pre-capillaryarterioles, but there is no immunoreactive a-actin detected incapillary pericytes as indicated by the arrows in (B and C).Magni®cation bar equals 93 micron (A), 23 microns (B and D),and 9 microns (C). From Boado and Pardridge (1994).


563

endothelial cytoplasm, and exocytosis of the pep-tide into the brain interstitial space (Pardridge,1991). Circulating transferring undergoes RMTthrough the BBB via the capillary endothelialtransferrin receptor. The ultrastructural basis ofthe transferrin RMT process was investigated with a5 nm gold conjugate of the OX26 murine mono-clonal antibody to the rat transferrin receptor(Bickel et al, 1994). The gold-OX26 conjugate wasinfused into the internal carotid artery of anesthe-tized rats for 10 min followed by saline perfusion ofthe brain microvasculature to clear residual con-jugate from the luminal compartment, followed byin situ perfusion ®xation with 2% glutaraldehyde,embedding, and visualization at the light micro-scopic level with immunogold silver staining (A,Figure 8). At the electron microscopic level, thegold-OX26 conjugate was visible within the en-dothelial cell in 100 nanometer endosomal struc-tures as shown at low magni®cation in B and highmagni®cation in D of Figure 8. The exocytosis of theconjugate into the brain interstitial ¯uid is shown athigh magni®cation in C of Figure 8 and the overallscheme for RMT of transferrin (Tf) via the BBBtransferrin receptor (TfR) is depicted in D of Figure

8. The RMT of transferrin through the BBB is amodel of not only new pathways for drug delivery tothe brain, but also a model for how viruses mayenter the brain from the circulation.

Drug delivery to the brain via blood-brain barrierreceptor-mediated transport systemsSimilar to the transferrin receptor, there is aninsulin receptor at the BBB and this mediates theRMT of insulin through the endothelial barrier invivo. [125I]insulin was infused into the carotid arteryof 1-month old rabbits for 10 min and the animalswere then sacri®ced by decapitation and the brainwas rapidly frozen for thaw-mount autoradiography(Duffy and Pardridge, 1987). A dark ®eld micro-graph of the brain microvasculature is shown in A ofFigure 9 and indicates that insulin is trapped in thecapillary lumen. However, radioactive insulin isalso found well within the brain parenchymafollowing a 10 min carotid artery perfusion. Reversephase HPLC of acid ethanol extracts of brainshowed that the radioactivity within brain parench-yma was unmetabolized insulin rather than aninsulin degradation product. The transport ofinsulin into brain parenchyma was saturable (Duffy

Figure 7 Light microscopic immunocytochemistry of cytocentrifuged human brain capillaries stained with a mouse monoclonalantibody to the human DR-antigen. Microvessels were isolated from either control human brain (A) or multiple sclerosis plaque (B andC). Magni®cation: 256 (A), 1006 (B), 2506 (C). (D) Model for antigen presentation in brain. From Pardridge et al (1989).


564

and Pardridge, 1987). Moreover, the binding of[125I]insulin to human brain capillaries was alsosaturable and the molecular weight of the insulinbinding site was identi®ed by electrophoretictechniques and shown to be identical to themolecular weight of the insulin receptor (Pardridgeet al, 1985). This provides evidence that thetransport of insulin through the BBB in vivo is areceptor-mediated process.

A monoclonal antibody (MAb) to the humaninsulin receptor (HIR) was used to visualize the BBBinsulin receptor in Old World primate brainimmunocytochemically as shown in B of Figure 9(Pardridge et al, 1995). The HIRMAb may be used asa BBB drug delivery vector by targeting the BBBinsulin receptor (IR), as depicted in C of Figure 9.Drugs may be conjugated to brain drug deliveryvectors such as the HIRMAb using avidin-biotintechnology. In this approach, a conjugate of theHIRMAb and streptavidin is prepared in parallelwith monobiotinylation of the drug that is normallynot transported through the BBB (Pardridge, 1991).The 40 amino acid peptide, Ab1 ± 40, is a potential

amyloid imaging agent should this molecule bemade transportable through the BBB (Saito et al,1995). Therefore, biotinylated Ab1 ± 40 was radiola-beled and attached to a conjugate of HIRMAb andstreptavidin and administered intravenously toanesthetized Rhesus monkeys (Wu et al, 1997). Asshown by the QAR scan in D of Figure 9, there isrobust uptake of the Ab1 ± 40 following conjugation tothe HIRMAb BBB drug delivery system. In contrast,no visible brain uptake was observed when thepeptide radiopharmaceutical was administeredwithout attachment to the BBB drug deliverysystem. Vector-mediated brain drug delivery hasbeen demonstrated in several other systems usingthis BBB delivery system. An epidermal growthfactor (EGF) peptide radiopharmaceutical has beendelivered across the BBB for purposes of imagingbrain tumors (Kurihara et al, 1999). A vasoactiveintestinal peptide (VIP) pharmaceutical has beendelivered to brain for purposes of enhancingcerebral blood ¯ow (Wu and Pardridge, 1996). Abrain derived neurotrophic factor (BDNF) pharma-ceutical has been delivered across the BBB with this

Figure 8 (A) Silver-enhanced vibratome section of rat brain perfused with OX26/gold conjugate in vivo. The coronal section is at thelevel of the frontal cortex. The magni®cation bar=10 mm. (B) Electron micrograph of a brain capillary endothelial cell after perfusion ofrat brain in vivo with the OX26/gold conjugate. The study shows dispersed OX26/gold attached to the luminal plasma membrane andclustered in endosome-like vesicles. Magni®cation bar=100 nm. (C) High magni®cation showing clustering of OX26/gold conjugatewithin endosomal structures as well as exocytosis of the conjugate into brain interstitial ¯uid, via transport across the abluminalmembrane. (D) A scheme for receptor-mediated transcytosis (RMT) of transferrin (Tf) through the brain capillary endothelium via thetransferrin receptor (TfR). From Bickel et al (1994).


565

system for neuroprotection of hippocampal CA1neurons following transient forebrain ischemia (Wuand Pardridge, 1999)

Subsequent to transport through the BBB, theconjugate may undergo receptor-mediated endocy-tosis into brain cells expressing the TfR or HIR,which are ubiquitously expressed throughout thebrain (Mash et al, 1990). Following endocytosis intobrain cells, the conjugate is transferred to thelysosomal compartment and is degraded. If a radio-nuclide is attached, this is subsequently exportedfrom brain to blood with a t1/2 in primates, in thecase of [125I], of 16 h (Wu et al, 1997).

Virus delivery through the blood-brain barrierMeasles and encephalomyelitis, like MS or EAE,have an early peri-venular in®ltration of mono-nuclear cells with peri-venular demyelination(Gendelman et al, 1984). Canine distemper virus

(CDV) infects brain via transport across the BBB inan analogous fashion with the measles virus(Axthelm and Krakowa, 1987). The measles virusinfection of the CNS may be widespread as 30 ± 50%of children infected with measles have CNSinfection (Axthelm and Krakowa, 1987). In thisevent, capillaries contain virus before there isin®ltration of brain by circulating white blood cells.Circulating viruses likely gain access to braincapillary endothelial cells via receptor mediatedendocytosis of the virus from the plasma compart-ment. Virus coat proteins may act as ligands thattrigger a receptor mediated endocytosis-like processat the luminal membrane of the capillary endothe-lium. For example, the basic amino acid transporterwas originally identi®ed as a murine retrovirusreceptor (Kim et al, 1991). Once inside the capillaryendothelial cell, the virus may enhance the progres-sion of disease by causing BBB disruption which

Figure 9 (A) Thaw-mount autoradiogram with a dark®eld illumination of 1 month old rabbit brain obtained after a 10 min carotidartery perfusion of [125I]-insulin. Silver grains trapped within the vessels are shown, and silver grains are also distributed throughoutthe brain parenchyma, indicative of transport of [125I]-insulin through the BBB and into brain in vivo. From Duffy and Pardridge(1987). (B) Light microscopic immunocytochemistry of frozen sections of Rhesus monkey brain immunostained with the 83-14monoclonal antibody to the human insulin receptor (HIR). A continuous immunostaining is indicative of endothelial origin of the BBBinsulin receptor. From Pardridge et al (1995). (C) A scheme for drug delivery of a peptide radiopharmaceutical, Ab1 ± 40, which ismonobiotinylated and attached to a conjugate of the HIR monoclonal antibody (MAb) and streptavidin (SA). The complex undergoes areceptor-mediated transcytosis through the BBB via the insulin receptor (IR). The HIRMAb binds to an exofacial epitope on the IR thatis removed from the binding site for insulin (Ins). (D) Phosphorimager scan of the occipital hemisphere of brain of a Rhesus monkeyobtained 3 h after the intravenous injection of [125I]-Ab1 ± 40 that was monobiotinylated and attached to the HIRMAb/SA BBB drugdelivery system. There is robust brain uptake of the peptide radiopharmaceutical attached to the drug delivery system and the areas ofgray matter are clearly delineated from the white matter tracks. From Wu et al (1997).


566

then facilitates the entry into brain of circulatingplasma proteins, which are toxic to astrocytes(Nadal et al, 1995), and circulating immune cells(Soilu-Hanninen et al, 1994).

Cytokine-mediated blood-brain barrier disruptionIntra-cerebral injection of interleukin-1b causesBBB disruption with loss of occludin and Z0-1immunoreactivity at endothelial tight junctions.This is associated with increased adherence ofpolymorphornuclear leukocytes to the endothelialcell and increased phosphotyrosine immunoreac-tivty in endothelial cells (Bolton et al, 1998). Theprecise molecular mechanisms by which cytokinescause BBB disruption is not known. In culturedendothelial cells there is a release of nitric oxide(NO) following administration of cytokines such asinterferon g or tumor necrosis factor (TNF)-a(Durieu-Trautman et al, 1993). The release of NOmay lead to an increased perioxynitrite which thenleads to nitration of protein tyrosine residues(Hooper et al, 1997). This provides the basis fornew forms of MS treatment that use perioxynitritescavengers such as uric acid, as well as inhibitors ofnitric oxide synthase. NO synthase inhibitors blockhistamine-mediated BBB disruption of pial vesselsin vivo (Mayhan, 1996). Chemokines may also playa role in BBB disruption. The intra-striatal injectionof the CXC chemokine causes BBB disruption andthe effect is greater in 21 day old mice as opposed to90 day mice (Anthony et al, 1998). Whetherchemokines play a role in virus induced BBBdisruption is at present not known. One experi-mental model for this is the Semliki forest virus(SFV) which enhances the demyelination of EAE byinfecting the brain capillary endothelium andcausing BBB disruption (Soilu-Hanninen et al,1994).

Conclusions

Overall CNS microvascular biology is regulated byparacrine interactions between the capillary en-

dothelial cell, the pericyte, and the astrocyte footprocess. Functions frequently ascribed to thecapillary endothelium are actually performed byeither the capillary pericyte or the microvascularastrocyte foot process, which is separated from thecapillary endothelium by a very short distance ofonly 20 nanometers. Pre-capillary arteriolarsmooth muscle cells and capillary pericytesexpress DR-antigen, indicating antigen presenta-tion in the brain occurs distal to the capillaryendothelium. The brain capillary endothelial cellis highly enriched in specialized transport pro-cesses including carrier-mediated transport sys-tems and receptor-mediated transcytosisprocesses. The latter may account for uptake ofcirculating viruses by brain endothelium. If certainviral coat proteins act as ligands for BBB RMTprocesses, then this will allow for BBB uptake ofthe circulating virus. With respect to BBBmethodologies, there is a wide array of techniquesavailable for studying this important membrane(Figure 1).

Physiologic techniques may now be correlatedwith biochemical and molecular biological meth-ods using freshly isolated laboratory animal orhuman brain capillaries. Capillary cells may begrown in tissue culture to form an `in vitro BBB'model. However, owing to the marked de-differ-entiation of brain capillary endothelial cellsgrown in culture, even in the presence ofastrocyte co-cultures, it is dif®cult to extrapolatedata obtained with the in vitro BBB model to thein vivo situation. Whenever possible, it isessential in BBB research to have in vitro/in vivocorrelations whereby studies with the in vitroBBB model are performed in parallel with in vivoinvestigations.

Acknowledgements

Daniel Jeong skilfully prepared the manuscript.This work was supported by NIH grant NS-25554.

References

Anthony D, Dempster R, Fearn S, Clements J, Wells G,Perry VH, Walker K (1998). CXC chemokines generateage-related increases in neutrophil mediated brainin¯ammation and blood-brain barrier breakdown.Current Biology 8: 923 ± 926.

Audus KL, Bartel RL, Hidalgo IJ, Borchardt RT ± PCR(1990). The use of cultured epithelial and endothelialcells for drug transport and metabolism studies. PharmRes 7: 435 ± 451.

Axthelm MK, Krakowka S (1987). Canine distempervirus: the early blood-brain barrier lesion. Acta.Neuropatholog 75: 27 ± 33.

Bickel U, Kang YS, Yoshikawa T, Pardridge WM (1994).In vivo demonstration of subcellular localization ofanti-transferrin receptor monoclonal antibody-colloidalgold conjugate within brain capillary endothelium. JHistochem Cytochem 42: 1493 ± 1497.

Blumcke I, Eggli P, Celio MR (1995). Relationshipbetween astrocyte foot processes and `perineuronalnets' in rat neocortex. GLIA 15: 131 ± 140.

Boado RJ, Pardridge WM (1990). Molecular cloning of thebovine blood-brain barrier glucose transporter cDNAand demonstration of phylogenetic conservation of the5'-untranslated region. Mol Cell Neurosci 1: 224 ± 232.


567

Boado RJ, Pardridge WM (1991). A one-step procedurefor isolation of poly A+mRNA from isolated braincapillaries and endothelial cells in culture. J Neuro-chem 57: 2136 ± 2139.

Boado RJ, Pardridge WM (1994). Differential expressionof a-actin mRNA and immunoreactive protein in brainmicrovascular pericytes and smooth muscle cells. JNeurosci Res 39: 430 ± 435.

Bolton SJ, Anthony DC, Perry VH (1998). Loss of the tightjunction proteins occludin and zonula occludens-1from cerebral vascular endothelium during neutrophil-induced blood-brain barrier breakdown in vivo.Neurosci 86: 1245 ± 1257.

Bowman PD, Ennis SR, Rarey KE, Betz AL, Goldstein GW(1983). Brain microvessel endothelial cells in tissueculture: a model for study of blood-brain barrierpermeability. Annal Neurol 14: 296 ± 402.

Broman T (1964). Blood-brain barrier damage in multiplesclerosis supravital test-observations. Acta NeurolScand 40: 21 ± 24.

Cancilla PA, Baker RN, Pollack PS, Frommes SP (1972).The reaction of pericytes of the central nervous systemto exogenous protein. Lab Invest 26: 376 ± 383.

Cohen Z, Ehret M, Maitre M, Hamel E (1995). Ultra-structural analysis of tryptophan hydroxylase immu-noreactive nerve terminals in the rat cerebral cortexand hippocampus: their associations with local bloodvessels. Neurosci 66: 555 ± 569.

Cordon-Cardo C, O'Brien JP, Casals D, Rittman-Grauer L,Biedler JL, Melamed MR, Bertino JR (1989). Multi-drugresistance gene (P-glycoprotein) is expressed byendothelial cells at blood-brain barrier sites. Proc NatlAcad Sci USA 86: 695 ± 698.

Daniel PM, Lam DK, Pratt OE (1981). Changes in theeffectiveness of the blood-brain and blood-spinal cordbarriers in experimental allergic encephalomyelitis. JNeurol Sci 52: 211 ± 219.

DeBault LE, Kahn LE, Frommes SP, Cancilla PA (1979).Cerebral microvessels and derived cells in tissueculture: isolation and preliminary characterization. InVitro 7: 473 ± 487.

Dehouck MP, Meresse S, Delorme P, Fruchart JC,Cecchelli R (1990). An easier, reproducible, andmass-production method to study the blood-brainbarrier in vitro. J Neurochem 54: 1798 ± 1801.

Dore-Duffy P, Washington RA, Balabanov R (1994).Cytokine-mediated activation of cultured CNS micro-vessels: as system for examining antigenic modulationof CNS endothelial cells, and evidence for long-termexpression of the adhesion protein E-selectin. J CerebBlood Flow Metab 14: 837 ± 844.

Duffy KR, Pardridge WM (1987). Blood-brain barriertranscytosis of insulin in developing rabbits. BrainRes 420: 32 ± 38.

Durieu-Trautmann O, Federici C, Creminon C, Foignant-Chaverot N, Roux F, Claire M, Strosberg AD, CouraudPO (1993). Nitric oxide and endothelin secretion bybrain microvessel endothelial cells: regulation bycyclic nucleotides. J Cell Physiol 155: 104 ± 111.

Gendelman HE, Wolinsky JS, Johnson RT, Pressman NJ,Pezeshkpour GH, Boisset GF (1984). Measles encepha-lomyelitis. Lack of evidence of viral invasion of thecentral nervous system and quantitative study of thenature of demyelination. Ann Neurol 15: 353 ± 360.

Golden PL, Pardridge WM (1999). P-glycoprotein onastrocyte foot processes of un®xed isolated humanbrain capillaries. Brain Res 819: 143 ± 146.

Herman IM, D'Amore PA (1985). Microvascular pericytescontain muscle and nonmuscle actins. J Cell Biol 101:43 ± 52.

Hooper DC, Bagasra O, Marini JC, Zborek A, Ohnishi ST,Kean R, Champion JM, Sarker AB, Bobroski L, FarberJL, Akaike T, Maeda H, Koprowski H (1997).Prevention of experimental allergic encephalomyelitisby targeting nitric oxide and peroxynitrite: implica-tions for the treatment of multiple sclerosis. Proc NatlAcad Sci USA 94: 2528 ± 2533.

Hurwitz AA, Berman JW, Rashbaum WK, Lyman WD(1993). Human fetal astrocytes induce the expressionof blood-brain barrier speci®c proteins by autologousendothelial cells. Brain Res 625: 238 ± 243.

Jette L, Tetu B, Beliveau R (1993). High levels of P-glycoprotein detected in isolated brain capillaries.Biochim Biophys Acta 1150: 147 ± 154.

Kacem K, LaCombe P, Seylaz J, Bonvento G (1998).Structural organization of the perivascular astrocyteendfeet and their relationship with the endothelialglucose transporter: a confocal microscopy study.GLIA 23: 1 ± 10.

Kim JW, Closs EI, Albritton LM, Cunningham JM (1991).Transport of cationic amino acids by the mouseecotropic retrovirus receptor. Nature 352: 725 ± 728.

Kurihara A, Deguchi Y, Pardridge WM (1999). Epidermalgrowth factor radiopharmaceuticals: 111In chelation,conjugation to a blood-brain barrier delivery vector viaa biotin-polyethylene linker, pharmacokinetics, and invivo imaging of experimental brain tumors. BioconjChem 10: 502 ± 511.

Lasbennes R, Gayet J (1983). Capacity for energymetabolism in microvessels isolated from rat brain.Neurochem Res 9: 1 ± 9.

Liwnicz BH, Leach JL, Yeh HS, Privitera M (1990).Pericyte degeneration and thickening of basementmembranes of cerebral microvessels in complex partialseizures: electron microscopic study of surgicallyremoved tissue. Neurosurg 26: 409 ± 420.

Mash DC, Pablo J, Flynn DD, Efange SM, Weiner WJ(1990). Characterization and distribution of transferrinreceptors in the rat brain. J Neurochem 55: 1972 ±1979.

Mayan WG (1996). Role of nitric oxide in histamine-induced increases in permeability of the blood-brainbarrier. Brain Res 743: 70 ± 76.

McClellan G, Weisberg A, Lin LE, Rose D, Ramaciotti C,Winegrad S (1993). Endothelial cells are required forthe cAMP regulation of cardiac contractile proteins.Proc Natl Acad Sci USA 90: 2885 ± 2889.

Meyer J, Rauh J, Galla HJ (1991). The susceptibility ofcerebral endothelial cells to astroglial induction ofblood-brain barrier enzymes depends on their prolif-erative state. J Neurochem 57: 1971 ± 1977.

Nadal A, Fuentes E, Pastor J, McNaughton PA (1995).Plasma albumin is a potent trigger of calcium signalsand DNA synthesis in astrocytes. Proc Natl Acad SciUSA 92: 1426 ± 1430.

Nedergaard M (1994). Direct signaling from astrocytes toneurons in cultures of mammalian brain cells. Science263: 1768 ± 1772.


568

Nehls V, Drenckhahn D (1991). Heterogeneity of micro-vascular pericytes for smooth muscle type alpha-actin.J Cell Biol 113: 147 ± 154.

Pardridge WM (1991). In: Peptide Drug Delivery to theBrain. Pardridge WM (ed) Raven Press: New York, pp1 ± 357.

Pardridge WM (1998a). Blood-brain barrier methodologyand biology. In: Introduction to the Blood-brainBarrier: Methodology, Biology, and Pathology. Par-dridge WM (ed). Cambridge University Press: Cam-bridge, UK, pp 1 ± 8.

Pardridge WM (1998b). Isolated brain capillaries: an invitro model of blood-brain barrier research. In:Introduction to the Blood-Brain Barrier: Methodology,Biology, and Pathology. Pardridge WM (ed). Cam-bridge University Press: Cambridge, UK, pp 49 ± 61.

Pardridge WM (1999). A morphological approach to theanalysis of blood-brain barrier transport function. In:Brain Barrier Systems. A¯red Benzon Symposium 45,O. Paulson et al, (eds.). Munksgaard: Copenhagen. Inpress.

Pardridge WM, Eisenberg J, Yang J (1985). Human blood-brain barrier insulin receptor. J Neurochem 44: 1541 ±1550.

Pardridge WM, Eisenberg J, Yang J, Tourtelotte WW(1987). Isolation of intact capillaries and capillaryplasma membranes from frozen human brain. JNeurosci Res 18: 352 ± 257.

Pardridge WM, Golden PL, Kang YS, Bickel U (1997).Brain microvascular and astrocyte localization of p-glycoprotein. J Neurochem 68: 1278 ± 1285.

Pardridge WM, Kang YS, Buciak JL, Yang J (1995).Human insulin receptor monoclonal antibody under-goes high af®nity binding to human brain capillariesin vitro and rapid transcytosis through the blood-brainbarrier in vivo in the primate. Pharm Res 12: 807 ±816.

Pardridge WM, Triguero D, Yang J, Cancilla PA (1990).Comparison of in vitro and in vivo models of drugtranscytosis through the blood-brain barrier. J Pharma-col Exp Ther 253: 884 ± 891.

Pardridge WM, Yang J, Buciak J, Tourtellotte WW (1989).Human brain microvascular DR antigen. J NeurosciRes 23: 337 ± 341.

Paspalas CD, Papadopoulos GC (1996). Ultrastructuralrelationships between noradrenergic nerve ®bers andnon-neuronal elements in the rat cerebral cortex. GLIA17: 133 ± 146.

Paulson OB, Newman EA (1987). Does the release ofpotassium from astrocyte endfeet regulate cerebralblood ¯ow? Science 237: 896 ± 898.

Pollack IF, Lund RD (1990). The blood-brain barrierprotects foreign antigens in the brain from immuneattack. Exper Neurol 108: 114 ± 121.

Risau W, Dingler A, Albrecht U, Dehouck MP, CecchelliR (1992). Blood-brain barrier pericytes are the mainsource of g-glutamyltranspeptidase activity in braincapillaries. J Neurochem 58: 667 ± 672.

Saito Y, Buciak J, Pardridge WM (1995). Vector-mediateddelivery of [125I]-labeled b-amyloid peptide Ab1 ± 40

through the blood-brain barrier and binding toAlzheimer's disease amyloid of the Ab1 ± 40 vectorcomplex. Proc Natl Acad Sci USA 90: 10227 ± 10231.

Smith QR, Rapoport SI (1986). Cerebrovascular perme-ability coef®cients to sodium, potassium and chloride.J Neurochem 46: 1732 ± 1742.

Soilu-Hanninen M, Eralinna J-H, Hukkanen V, Roytta M,Salmi AA, Salonen R (1994). Semliki forest virusinfects mouse brain endothelial cells and causesblood-brain barrier damage. J Virol 68: 6291 ± 6298.

Stanness KA, Westrum LE, Fornaciari E, Mascagni P,Nelson JA, Stenglein SG, Myers T, Janigro D (1997).Morphological and functional characterization of an invitro blood-brain barrier model. Brain Res 771: 329 ±342.

Stewart PA (1994). Glial-vascular relations, In: KMAWelch, LR Caplan, DL Reis (eds). Primer on Cere-brovascular Diseases. Academic Press: New York, pp17 ± 20.

Thompson SE, Audus KL (1994). Leucine-enkephalinmetabolism in brain microvessel endothelial cells.Peptides 15: 109 ± 116.

Tilling T, Korte D, Hoheisel D, Galla HJ (1998). Basementmembrane proteins in¯uence brain capillary endothe-lial barrier function in vitro. J Neurochem 71: 1151 ±1157.

Tsukada N, Behan WM, Behan PO (1985). Search forautoantibodies to endothelial and smooth muscle cellsin patients with multiple sclerosis. Acta Neuropath 66:134 ± 139.

Tsukada N, Koh CS, Yanagisawa N, Okano A, BehanWM, Behan PO (1987). A new model for multiplesclerosis: chronic experimental allergic encephalomye-litis induced by immunization with cerebral endothe-lial cell membrane. Acta Neuropath 73: 259 ± 266.

VanBree JB, de Boer AG, Verhoef JC, Danhof M, BreimerDD (1988). Transport of vasopressin fragments acrossthe blood-brain barrier: In vitro studies using mono-layer cultures of bovine brain endothelial cells. JPharm Exp Ther 249: 901 ± 905.

Vass K, Lassmann H, Wekerle H, Wisniewski HM (1986).The distribution of Ia antigen in the lesions of ratacute experimental allergic encephalomyelitis. ActaNeuropathol 70: 149 ± 160.

Washington R, Burton J, Todd RF, Newman W, DragovicL, Dore-Duffy P (1994). Expression of immunologicallyrelevant endothelial cell activation antigens on iso-lated central nervous system microvessels frompatients with multiple sclerosis. Ann Neurol 35: 89 ±97.

Wong GH, Bartlett PF, Clark-Lewis I, Battye F, SchraderJW (1984). Inducible expression of H-2 and Ia antigenson brain cells. Nature 310: 688 ± 691.

Wu D, Pardridge WM (1996). CNS pharmacologic effectin conscious rats after intravenous injection of abiotinylated vasoactive intestinal peptide analoguecoupled to a blood-brain barrier drug delivery system.J Pharmacol Exp Ther 279: 77 ± 83.

Wu D, Pardridge WM (1999). Neuroprotection with non-invasive neurotrophin delivery to brain. Proc NatlAcad Sci USA 96: 254 ± 259.

Wu D, Yang J, Pardridge WM (1997). Drug targeting of apeptide radiopharmaceutical through the primateblood-brain barrier in vivo with a monoclonal anti-body to the human insulin receptor. J Clin Invest 100:2842 ± 2848.


569

blood-brain barrier biology and methodology6)/556-569.pdf · blood-brain barrier biology and...

Documents