effects of interferon-y on primary cultures of human brain

American Journal ofPatbology, Vol. 142, No. 4, April 1993Copyright © American Societyfor Investigative Pathology

Effects of Interferon-y on Primary Cultures ofHuman Brain Microvessel Endothelial Cells

Hanh K. Huynh and Katerina Dorovini-ZisFrom the Department ofPathology, Section ofNeuropathology, Vancouver General Hospital, and TheUniversity of British Columbia, Vancouver, BritishColumbia, Canada

Primary cultures of human brain microvesselendothelial ceUs were used to study the effects ofhuman recombinant interferon-vy (IFN- y) on ce-rebral endothelium in vitro Incubation ofmono-layers with various concentrations ofIFN- y(lOto200 U/ml)for 12 to 96 hours induced surface ex-pression ofclass IImajor histocompatibility com-plex (Ia) antigen in a time- and concentration-dependent manner. In immunogold-stained cul-tures, labeling was observed as early as 12 hours,was maximalafter 48 hours, andpersisted atpla-teau levels in the continuous presence of the cy-tokine. Expression was blocked by coincubationwith anti-IFN- yantibody andwas reversed 4 daysfollowing removal ofIFN- yfrom the culture me-dia. Endothelial ceUs treated with IFN-yfor 3 to 4days became spindle-shaped, extensively over-lapped, and frequently formed celular whorls.These changes did not occur in the presence ofanti-IFN- y antibody and reversed upon removalof IFN- yfrom the media. The morphological al-terations were associated with increasedperme-ability of confluent monolayers to macromole-cules as compared with untreated cultures. Theresults ofthese studies indicate that human brainmicrovessel endothelial ceUs respond to in vitrocytokine stimulation by undergoing profoundmorphological, functional, and permeabilitychanges. We conclude that cerebral endotheliummay play an important role in the initiation andregulation oflymphocyte traffic across the blood-brain barrier in inflammatory disorders of thehuman central nervous systenm (Am J Pathol1993, 142:1265-1278)

Inflammatory and autoimmune demyelinating dis-eases of the central nervous system (CNS), such as

multiple sclerosis (MS) and its prime animal model,experimental autoimmune encephalomyelitis (EAE),are characterized by increased permeability of theblood-brain barrier (BBB) and migration of lympho-cytes from blood into brain through the cerebral vas-culature that normally excludes circulating inflamma-tory cells from the brain. Since brain endothelial cellsare the first native cells of the CNS to encounter cir-culating lymphocytes, membrane interactions be-tween these two cell types may be important in initi-ating a localized immune response. The specificmechanisms that mediate such interactions and areresponsible for the selective adhesion and transmi-gration of lymphocytes across the cerebral endothe-lial barrier have not yet been fully investigated. A largebody of evidence indicates that endothelial cells lin-ing extracerebral blood vessels may be induced toexpress class 11 major histocompatibility complex(MHC 11 or la) antigens following in vitro or in vivostimulation with interferon-y (IFN-y) and may thus actas antigen-presenting cells for circulating CD4+ Tlymphocytes.1-8 In addition, IFN-y-treated endothe-lial cell monolayers increase lymphocyte adhesionand transmigration as compared with untreatedmonolayers.9 13

Endothelial cells lining the cerebral blood vesselsare morphologically and functionally different fromendothelial cells of other vascular beds. The pres-ence of interendothelial tidht junctions that restrict theparacellular movement of macromolecules, a paucityof cytoplasmic vesicles, and specialized membranetransport systems endow these cells with unique bar-rier properties. The role of the cerebral endotheliumin CNS inflammation remains ill defined and rathercontroversial. Recently, molecular changes on the ce-

Supported by grants from the National Institutes Health(R01NS23746), the Medical Research Council of Canada (MA-9823), and the British Columbia Health Care Research Foundation.H. Huynh is the recipient of a studentship from the Multiple Sclero-sis Society of Canada.Accepted for publication September 28, 1992.

Address reprint requests to Katerina Dorovini-Zis, M.D., Depart-ment of Pathology, Section of Neuropathology, Vancouver GeneralHospital, 855 West 12th Avenue, Vancouver, BC, Canada V5Z1 M9.

1265

1266 Huynh and Dorovini-ZisAJP April 1993, Vol. 142, No. 4

rebral endothelium have been increasingly impli-cated as important modulators of the initial stages ofthe localized immune response. Thus, rat and guineapig brain microvessel endothelium can be induced toexpress la antigen by IFN-y in vitro1415, and isolatedmouse brain endothelial cells, treated with con-canavalin A-conditioned media, are able to presentantigen to myelin basic protein-sensitized lymphnode cells.16 Other studies, however, report expres-sion of la antigen by brain microvascular smooth mus-cle or microglia rather than endothelial cells17'18 andindicate that smooth muscle cells can present antigento specific T-cell lines to a much greater extent thanendothelial cells.19 In vivo immunohistochemicalstudies have demonstrated la antigen expression onendothelial cells, in addition to astrocytes, microglia,and macrophages in acute, active chronic, and silentchronic MS lesions,2021 in graft-versus-host dis-ease, 22 and in simian immunodeficiency virus en-

cephalitis.23 Similar studies in EAE, however, arerather conflicting, since la antigen expression bybrain endothelium has been reported by some24,25but not other investigators.26'27

In the present study, the induced expression of laantigen by human cerebral endothelium was inves-tigated in primary cultures of human brain microves-sel endothelial cells (HBMEC) treated with recombi-nant human IFN-y, a cytokine known to specificallyinduce class 11 MHC antigen expression. We havepreviously reported that HBMEC in culture form con-fluent monolayers of Factor VIII and Ulex europaeusagglutinin-positive cells that retain important mor-phological characteristics of their in vivo counter-parts, namely, a paucity of cytoplasmic vesicles andthe presence of tight junctions that restrict thetransendothelial passage of macromolecules.28Treatment of HBMEC with IFN-y resulted in de novoexpression of la antigen in a time- and concentration-dependent manner and induced changes in the mor-phology, organization, and permeability of the mono-layers. Our results indicate that IFN-y inducesprofound morphological and functional alterations onhuman cerebral endothelial cells that may be impor-tant for antigen presentation and initiation of inflam-matory responses within the human CNS.

Materials and Methods

Isolation and Culture of Human BrainMicrovessel Endothelial Cells

Primary cultures of endothelial cells were estab-lished from normal brains at autopsy and temporallobectomy specimens as previously described.28

The isolated clumps of endothelial cells wereseeded onto fibronectin-coated plastic wells (Corn-ing Plastics, Corning, NY) and maintained in culturein minimum essential medium a (aMEM) (Gibco,Burlington, Ontario) supplemented with 10% horseplasma-derived serum (HyClone Laboratories, Lo-gan, UT), 25 mM (N-2-hydroxyethyl)-1-piperazine-ethanesulfonic acid (pH 7.4), 20 pg/ml endothelialcell growth supplement, 100 pg/ml heparin (all fromSigma Chemical Co., St. Louis, MO), and penicillin(100 pg/ml), streptomycin (100 pg/ml), and ampho-tericin B (2.5 pg/ml) (Gibco) at 37 C in a humidified2.5% C02/97.5% air atmosphere. The endothelialorigin of the cells was confirmed by their stronglypositive, perinuclear immunofluorescence for FactorVIll-related antigen and their binding of U.europaeus-I lectin as described previously.28 Cul-ture media were changed every other day. Conflu-ent monolayers of contact-inhibiting cells were ob-tained 5 to 7 days after plating. Cultures were usedas soon as they reached confluency. There were nodifferences in the growth pattern and cell morphol-ogy between cells derived from postmortem brainsand temporal lobectomy specimens.

Antibodies

Mouse monoclonal antibody against human recom-binant IFN-y (KM48) and mouse anti-humanHLA-DR IgG (DK22) were obtained from DimensionLaboratories, Inc. (Missisauga, Ontario). Goat anti-mouse IgG coupled to 5-nm gold particles (Auro-probe LMGAM IgG) was obtained from Janssen/Cedarlane Labs, Ltd., (Hornby, Ontario), and mouseanti-human pituitary follicle-stimulating hormone IgGwas from Biogenex Laboratories (Dublin, CA).

Treatment of Primary HBMEC Cultures

Human recombinant IFN-y (Collaborative Research,Inc., Bedford, MA) was diluted in complete media toa final concentration of 10, 20, 50, 100, 150, and200 U/ml. Confluent monolayers of HBMEC grownin replicate wells were incubated with different con-centrations of IFN-y for 4 days and with 200 U/ml for12 hours to 4 days at 37 C. Cultures used for theseexperiments were derived from endothelial cells iso-lated from several different brains and temporallobectomy specimens. The specificity of la Ag in-duction by IFN-y was tested in cultures coincu-bated with IFN-y (200 U/ml) and anti-IFN-y mono-clonal antibody (optimal concentration, 10 pg/ml)for 4 days. In order to study the reversibility of la Agexpression, monolayers previously treated with

IFN-y-lnduced Changes on Cerebral Endothelium 1267AJP April 1993, Vol. 142, No. 4

IFN-y (200 U/mI) for 4 days were thoroughlywashed with aMEM to remove the cytokine, thenplaced in complete media and returned to the incu-bator for another 4 days prior to la Ag detection.

Immunocytochemical Localizationof la Ag

Light Microscopy Immunocytochemistry

Following incubation with IFN-y, the monolayerswere washed 3 times with buffer containingphosphate-buffered saline (PBS) (10 mM, pH 7.2),1% bovine serum albumin (BSA), and 1% normalgoat serum (PBS/BSA/NGS) and incubated for 40minutes at room temperature with mouse anti-human HLA-DR monoclonal antibody at 1:30 dilu-tion in carrier buffer containing PBS, 5% BSA, and4% NGS. Following brief washing with PBS/BSA/NGS, the monolayers were incubated with the sec-ondary antibody (Auroprobe LMGAM IgG coupledto 5-nm gold particles) diluted 1:40 in carrier bufferfor 60 minutes at room temperature. At the end ofthe incubation period, the cells were washed withPBS/BSA/NGS, fixed in fresly prepared bufferedformaldehyde-acetone fixative (20 mg Na2HPO4,100 mg KH2PO4, 30 ml distilled H20, 25 ml 37% for-maldehyde, and 45 ml acetone) for 30 seconds,washed with distilled H20, and incubated in silverenhancing solution (IntenseM, Janssen/Cedarlane)for 25 to 35 minutes. After washing with distilledH20, the monolayers were counterstained withGiemsa and coverslipped using JB-4 plus(Polysciences/Analychem, Markham, Ontario) asthe mounting medium.

Controls included intact monolayers grown in theabsence of IFN-y and cultures incubated with 1)normal mouse IgG at the same concentration as theprimary antibody (5.9 pg/ml IgG), 2) carrier buffer,and 3) irrelevant antibody (anti-human pituitaryfollicle-stimulating hormone IgG) instead of the pri-mary antibody.

Stained monolayers were examined under a Ni-kon Labophot light microscope. Quantitation of laAg expression was performed by counting one cen-tral and four peripheral randomly selected fields ofeach culture well with an ocular grid under X200magnification. All counts were performed blindly.Data are expressed as the mean ± SEM.

Immunoelectron Microscopy

Confluent monolayers of HBMEC, treated with200 U/ml IFN-y for 4 days, were washed with buffercontaining PBS, 1% BSA, and 0.2% NaN3 (PBS/

BSA) and incubated with mouse anti-humanHLA-DR monoclonal antibody at 1:30 dilution in car-rier buffer containing PBS, 5% BSA, and 4% NGSfor 30 minutes at room temperature. After washingwith PBS/BSA, the cells were incubated with 5-nmgold particle-conjugated secondary antibody (Auro-probe LMGAMIgG) at 1:40 dilution in carrier bufferfor 45 minutes, washed, and fixed in periodate-lysine-paraformaldehyde fixative29 overnight at 4 C.Following fixation, the cells were washed in PBS,postfixed in 1% buffered OS04, stained en bloc withuranyl magnesium acetate overnight at 4 C, dehy-drated in graded series of methanol, and embed-ded in Epon-Araldite. Blocks cut from the embed-ded cultures were re-embedded for cross-sectioning. Thin sections were examined in a PhilipsEM400 without heavy metal staining. Controls con-sisted of cells maintained in IFN-y-free growth me-dia, and monolayers incubated with normal mouseIgG or carrier buffer instead of the primary antibody.

Scanning Electron Microscopy

Confluent HBMEC monolayers treated with 200U/ml IFN-y for 3 to 4 days and monolayers incu-bated for the same period of time with 200 U/mlIFN-y with the addition of 10 pg/ml anti-IFN-y mono-clonal antibody, as well as intact control cultures,were processed for scanning electron microscopyas described by Schroeter et al.30 Cultures werewashed in Hanks' balanced salt solution and fixedin 2.5% glutaraldehyde in 0.05 M sodium cacody-late buffer (pH 7.2) for 1 hour at 4 C. Followingwashing in cacodylate buffer, the cells were post-fixed in buffered 1% OS04 for 1 hour, washed inbuffer, and treated with 1% tannic acid for 1 hour.After further washing in cacodylate buffer, themonolayers were dehydrated in graded series ofmethanol up to 70% and block stained with 4% ura-nyl acetate overnight at 4 C. The cells were then de-hydrated with methanol up to 100%, critical pointdried, gold coated, and viewed with a CambridgeStereoscan 250T scanning electron microscope.

Permeability Studies

Confluent HBMEC monolayers treated with IFN-y(200 U/ml) for 4 days were washed with serum-freeaMEM and incubated in aMEM containing 1 mg/mlhorseradish peroxidase (HRP) (Sigma Type VI) for 5to 10 minutes at 37 C as previously described.31 Atthe end of the incubation period, the cells werefixed in 2.5% glutaraldehyde and 2% paraformalde-hyde in 0.1 M sodium cacodylate buffer (pH 7.4) for

1268 Huynh and Dorovini-ZisAJP Apil 1993, Vol. 142, No. 4

1 hour at 4 C. Following washing with buffer, themonolayers were incubated with 3,3'-diamino-benzidine (Sigma) for 1 hour at 4 C, washed withcacodylate buffer, postfixed in 1% buffered OS04,stained en bloc with uranyl magnesium acetate, de-hydrated with graded series of methanol, and em-bedded in Epon-Araldite. Thin plastic sections wereexamined in a Philips EM400 without heavy metalstaining. Controls consisted of identical age-matched primary cultures grown to confluence inIFN-y-free media.

Quantitation of the junctional permeability and pi-nocytotic activity of endothelial cells was performedby counting the number of permeable and imper-meable intercellular contacts and the number ofHRP-labeled and -unlabeled cytoplasmic vesiclesin IFN-y-treated and in untreated cultures.

Growth Studies

Freshly isolated endothelial cells were plated in rep-licate wells of Corning 24-well plates at a density of1 x 105 cells/cm2 on day 0. On day 1, all experi-mental wells were refed with complete medium con-taining IFN-y (150 U/ml). Medium was changed ev-ery other day. Control cultures were maintained ingrowth medium in the absence of IFN-y. The cellswere viewed with a Nikon Diaphot TMD inverted mi-croscope, and photographs of one central and fourperipheral fields of each well were taken daily atx100 magnification. The number of cells in eachphotograph was counted, and the data are ex-pressed as the mean + SEM. Student's t-test wasused to determine significant differences betweenIFN-y-treated'and untreated cultures.

ResultsInduction of la Ag ExpressionTreatment of cultures with IFN-y induced expressionof la Ag by endothelial cells, which was dependentupon the concentration and length of exposure toIFN-y. Surface labeling was observed as early as12 hours following incubation with 200 U/ml in asmall cell population (9.24 ± 0.99%), increased upto 88.35 + 0.18% after 24 hours, and reached100% after 48 hours (Fig. 1). Ia Ag expressionreached plateau levels after 2 days and persistedfor 4 days in the continuous presence of the cyto-kine. Expression was maximal with 100 to 200 U/mlIFN-y (100% of cells) and minimal with 10 U/ml(25.76 + 7%) (Fig. 2). Incubation with 20 U/ml ofIFN-y induced la Ag expression in 68.73 ± 18.5%

100

V:)cn

a)

'a)a)

4-

0

1o-e

80

60

40

20

0 0.5 1 2 3 4Time (days)

Figure 1. Time course of Ia antigen induction on human cerebralendothelium. Confluent HBMEC czlltures uere incubated with 200unitslml IFN-y for 0.5 to 4 days and then stained with the immu-nogold technique for the immunohistochemical demonstration of Iaantigen. Results are expressed as percentage of labeled cells in treatedcultures. Untreated cells were not labeled. Bars represent the mean +SEM of duplicate wells of two separate experiments.

100

C/)

a1)U

a)

o-

80

60

40

20

_n I0 10 20 50 100 150 200

IFN-Y Concentration (U/mi)Figure 2. Dose response of Ia antigen induction by IFN-y on HB-MEC. Confluent monolayers were incubated for 4 days with 10 to200 units/ml IFN- y and then immunostained for the demonstrationof Ia antigen. Results are expressed as percentage of labeled cells intreated and untreated cultures. Bars represent the mean ± SEM ofduplicate wells of three seperate experiments.

of cells, while 90.85 ± 5.5% of cells were labeledafter treatment with 50 U/ml. Endothelial cells ex-pressing la Ag showed diffuse surface staining inthe form of dark brown-black, granular deposits(Fig. 3A). In marked contrast, untreated endothelialcells invariably lacked la Ag expression, as indi-cated by their consistently negative staining with

r-r---7r-

IL-iII-I

F

IFN-y-Induced Changes on Cerebral Endothelium 12694/P April 1993, Vol. 142, No. 4~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~. ....,,....,,;

*~~~~~~~~~~~~~~~~~~~~~~~~~~~~... ...r~~~~~~~~~~~~~~~~~~~~~~~~~~~........... .....

P...

r* yu, o?. j" cnigven expression ny Himt(u aeiectea ny immunogoia silver staining. A, endotbelial cells incubated witb 200 units/ml IFN-y for 4days demonstrating intenise granular sturface staining for Ia atntigen. B, cotntrol untreated monolayers not expressing Ia antigen. C, endotbelialcells treated uith 200 unitVs/ml IEN-yfor 24 houzrs exhibitinig less denise labeling. Individual cell variation in stainitng intensity is apparent in A andC. In D, cultures coincubated uith anti-IEV-y antibody failed to label u'itb the immuntogold reagent. Bars = 20 pm.

Fifiiirp R I/i lj"tioo" ,Yhrocciri" I,", 14PUPf" 1- I


immunogold (Fig. 3B). The staining intensity variedwith the concentration and length of incubation withIFN-y. Thus, labeling was less intense in cells incu-bated with 10 to 20 U/ml for 4 days or with 200 U/mlfor 12 to 24 hours (Fig. 3C) and most dense in cul-tures treated with higher concentrations for 3 to 4days (Fig. 3A). Within the same culture, the largercells were usually stained most intensely. Therewere no differences in la Ag expression amongHBMEC monolayers originating from different indi-viduals and subjected to identical culture conditionsand IFN-y treatment. Staining was not observed incontrol cultures incubated with normal mouse IgG,carrier buffer, or irrelevant antibody.

In monolayers coincubated with IFN-y and anti-IFN-y monoclonal antibody for 4 days, induction ofla Ag was completely abolished (Fig. 3D), indicat-ing that IFN-y specifically induces expression ofclass 11 MHC molecules on human brain endothelialcells. Treatment of cells with 200 U/ml IFN-y fol-lowed by withdrawal and culture in regular growthmedia resulted in complete reversal of la Ag ex-pression and negative staining of the cultures.

Ultrastructural examination following immunogoldlabeling showed that la Ag was readily detectableon the apical surface of endothelial cells. Gold par-ticles were found at the cell membrane with a ten-dency to localize on or near thin cytoplasmic pro-

cesses (Fig. 4A). The basal cell surface was notlabeled. No labeling was seen in untreated controlcultures (Fig. 4B).

Effects of IFN-,y on Cell Morphology,Organization, and Growth

Primary cultures of HBMEC grown in regular mediain the absence of IFN-y formed highly ordered con-fluent monolayers of elongated, closely associated,contact-inhibiting cells (Fig. 5A). Endothelial cellstreated with 200 U/ml IFN-y for 3 to 4 days acquireda spindlelike shape and long attenuated processes.These markedly elongated cells frequently ar-ranged themselves in ill-defined whorls and exhib-ited prominent overlapping, thus contributing to theunique appearance of the monolayers (Fig. 5B).These changes were most conspicuous underscanning electron microscopic examination. Undernormal culture conditions, elongated HBMEC growin close contact to each other and display distinctmarginal folds in areas of cell-to-cell contact (Fig.6A). In contrast, endothelial cells treated with IFN-ybecame attenuated, and their long, thin processesoften extended over and covered adjacent cells(Fig. 6B). As a result of this rearrangement, intercel-lular contacts and marginal folds became less

Figure 4. Immunogold staining ofHBMECfor the demonstration ofIa antigen. A, endothelial cells incubated with 200 units/ml IFN-y for 4 days.Five-nanometer gold particles focally decorate the apical surface of endothelial cells (arrowheads) with a tendency to localize close to fingerlikecytoplasmic folds. The basal cell surface is not labeled. B, staining absent in untreated cells. Bars = 0.5 jm.


Figure 5. A, HBMEC maintained under standard culture conditions forming highly organized, confluent, contact inhibiting monolayers com-posed of elongated cells. B, endothelial cells incubated with IFN--y (200 units/ml) for 4 days have become spindle shaped, overlap, andfocally ar-range themselves into whorls. Bars = 10 rim.

prominent, and the monolayers lost their highly or-ganized appearance. The above morphologicalchanges were reversed 4 days following withdrawalof the cytokine from the culture media and were notobserved in cultures coincubated with IFN--y andanti-IFN-y antibody.The effect of IFN--y on the growth of HBMEC was

less profound. Thus, the number of cells in primaryHBMEC cultures treated with 150 U/ml IFN--y fromday 1 was slightly less than that in control cultures(Fig. 7). This slight inhibitory growth effect of IFN-yprovides further support to the observation that re-arrangement and overlapping of HBMEC are the di-rect effect of the cytokine and not the result of cellovergrowth.

Permeability of HBMEC Monolayers

In order to examine the effect that the IFN-y-induced morphological changes might have on thepermeability of the monolayers to macromolecules,confluent treated and untreated cultures were incu-bated with HRP, and the labeling of intercellularcontacts and cytoplasmic vesicles was assessedultrastructurally. Intercellular contacts that impededthe tracer entirely or were penetrated for a short dis-

tance from either the apical or basal cell surface byHRP were considered impermeable. In untreatedcultures, 75.2% of interendothelial junctions pre-vented the passage of HRP, in contrast to 36.6% ofimpermeable junctions in cultures incubated withthe cytokine for 4 days (Table 1). In untreatedmonolayers, endothelial cells formed a single celllayer and were bound together by junctions, most ofwhich were not labeled with the tracer (Fig. 8A). Fo-cally, HRP penetrated an intercellular contact for ashort distance from the basal aspect of the mono-layer before being arrested at a junctional complexof an otherwise intact cleft (Fig. 8B). In treated cul-tures, interendothelial clefts were often penetratedby the tracer throughout their entire length (Fig. 8C).Overlapping of endothelial cells resulted in the for-mation of two or more layers. HRP often penetratedthe intercellular clefts between endothelial cells atthe top layer, and extensive deposits were foundbetween adjacent cells at the lower layers (Fig. 8D).The number of cytoplasmic vesicles labeled withHRP was equally low in control and experimentalcultures (Table 1), indicating that, contrary tothe prominent conformational and organizationalchanges, the pinocytotic activity of HBMEC is notaffected by IFN-y treatment.

..:...: -- '.1 .:..--. .;:.::.f .: .."; :::Awl

1272 Huynh and Dorovini-ZisAJPApril 1993, Vol. 142, No. 4

Figure 6. Scanning electron micrograph ofHBMEC grown in the absence (A) or presence (B) of IFN--y in the culture media. A, endothelial cellsclosely packed u4ithout apparent intercellzular spaces. Marginalfolds (arrous) are present in areas of cell-cell contact. In B, incuibationi u'ith IFN-y(200 uinits/ml) for 3 days induices marked attenuiation of cell cytoplasm and disorganization of the motnolayer due to the tendenzcy of endothelialcell processes to extenid over and untider ad/acetnt cells. Bars = 20 m.

-,I"= .ft;-.'f.,,;A

rl.,

pp--

.- .- Jj.z ,. p-01"


tLCI lt;U Ul 1Ut; IUI lCt WLUILU Vv%.,UI IU LVI1. ILI CX Y LV

these reports, endothelial cells of normal guinea pigCNS display surface MHC in vivo and in vitro,15 andminimal basal expression has been reported in pri-mary cultures of rhesus monkey cerebral endothe-ium,23 while cultured rat heart vascular endothelialcells constitutively express considerably higher lev-els of la Ag.2 It is apparent, from the above studies,that the presence of la antigen on normal, unstimu-lated vascular endothelium may vary among differ-ent species and vascular beds.

Previous studies on la antigen induction by IFN-yon HUVEC report a rapid increase of MHC class 11mRNA that nrens1s Q ijrfqcc exYressinn hv 1 to 2J 11 11 lfr% Li IC1L Pi U%.;UsO OUI ICX%,IUqU p I IILWY LV L_

0 2 4 6 8 10 12 days and rapidly declines to almost undetectable

Time (days) levels following withdrawal of the cytokine, while

Tect of IFN--y upon the growth ofpnimary cultures of HB- surface expression declines slowly after 4 days.1 Invere cultured under standard conditions (-- -) or in HBMEC, removal of IFN-y from the media results in)us presence of IFN-y (150 units/ml) in the culture mediaes represent the mean ± SEM of two experiments, each uniform loss rather than a decrease to lower levelslicate. P = 0.035 for days 4 and 6. Differences are not of class 11 MHC surface expression after 4 days. Ratfter day 6.

heart endothelium, however, behaves in a much dif-ferent way, since withdrawal of IFN-y is not followed

-ion by return of the la antigen expression to basal lev-

ent studies demonstrate that human re-els after 3 days.2

ent stuyindies demonstratexthatshuan r- Previous immunohistochemical studies on MS

itMHFN-C novn epresionmcof and EAE have demonstrated that surface expres-MHC antigen by human brain micro-dothelial cells in primary culture in a time- sion of la antigen on endothelial cells is discontinu-,lentration-dependent manner. In addition, ous along the microvessel lumen, so that la' cells

treated endothelial cells undergo unique are interposed between endothelial cells lacking la20.2-5 Asmlryvralxrsin their morphology and organization, antigen expression. A similarly variable expres-

incide with a considerable increase in the sion of la antigen was observed in vitro whenlity of confluent cultures to macromole- HBMEC were treated with low concentrations of

IFN-y or with higher concentrations for less than 2

ulated HBMEC grown under standard cul- days. Taken together with the in vivo studies, theselitions do not constitutively express la anti- observations may indicate individual cell variation inidicated by the lack of immunogold stain- the regulation of class 11 MHC molecule expression.jht and electron microscopy. Previous in Induction of la antigen expression on HBMECunohistochemical studies have demon- was restricted to the apical portion of the cell mem-ie absence of la antigen expression by brane. Immunogold particles were not identified on

al cells within the normal human CNS with the lateral or basal cell surfaces. Our findings corre-

;of reactivity detected in blood vessels of late with previous immunohistochemical studies invith brain neoplasms, abscesses, autoim- acute EAE demonstrating la expression on the lumi-inective tissue disease, and cerebral inf- nal but not abluminal surface of cerebral microves-in older patients without identifiable CNS sel endothelial cells25 and with similar observations32-34 Although the endothelial cells used in a variety of epithelial cells in mice treated withidies were isolated from normal brains of IFN-y.35 Although the mechanisms responsible for:nors with a wide age distribution, expres- the asymmetrical presentation of la antigen on theLantigen was not observed in any of the cell membrane are not known, polarization of ex-cultures. A similar lack of constitutive expression is probably of functional significance since

)f class 11 molecules has been observed in it would enable circulating T lymphocytes to recog-:ultures of rat brain endothelium14 and in nize antigen in association with class 11 MHC mole-lated35 as well as serially passaged6 hu- cules on the luminal surface of the cerebral endo-ilical vein endothelial cells (HUVEC) main- thelium and then migrate to sites of inflammation.

CNEU

LO)0

x

%.oa)U

.0

Ez

ic

2

C

Figure 7. EfjMEC. Cells uthe continuo( ). Valudone in dupsignificant a

DiscussThe pres(combinanclass 11vessel enand conccytokine-tchangeswhich coipermeabicules.

Unstimtture condgen, as iring on licvivo immstrated thendotheliclow levelspatients Vmune corarcts andlesions.20in our stuseveral d(sion of launtreatedpression cprimary c

freshly iscman umbi

tqinp-ri iinrip-r nr)rmql nidtimm rnnriitinnq- rnntrqrv to

t

1274 Huynh and Dorovini-Zis4/PApril 1993, Vol. 142, No. 4

The antiproliferative effect of IFN-y on primarycultures of HBMEC correlates with previous studiesdemonstrating inhibition of cell growth by IFN-y in-duced on extracerebral large and small vesselendothelial cultures in a dose-dependent man-ner3637-39 and possibly through modulation of theendothelial cell growth factor receptors.36 Lowerconcentrations of IFN-y (10 to 100 U/mI), however,appear to have a stimulating effect on cultured HU-VEC in both the absence and presence of endothe-lial cell growth factor.40 In addition, IFN-y signifi-

cantly inhibits the formation of endothelial tubularstructures in in vitro models of angiogenesis.41'42

Treatment of HBMEC with IFN-y induces markedelongation of endothelial cells, prominent overlap-ping, and frequent arrangement in a whorled pat-tern. A similar alteration of the morphological phe-notype and monolayer organization has beenpreviously reported in cultures of HUVEC3639 andhuman dermal microvascular endothelial cells36treated with IFN-y for 3 to 4 days and has beenshown to be associated with reorganization of thecytoskeletal filaments and considerable loss of thefibronectin matrix.39

Cerebral microvessel endothelial cells in primaryculture are bound together by tight junctions andhave a paucity of cytoplasmic vesicles, two impor-tant morphological characteristics of their in vivocounterparts.26'31 Under standard culture condi-tions, the great majority of interendothelial junctionsrestrict the passage of HRP. In cultures incubatedwith IFN-y, an increase in the permeability of themonolayers was observed that coincided tempo-rally with changes in morphology and rearrange-

ment of the cells. The number of labeled cytoplas-mic vesicles was not increased in IFN-y-treatedmonolayers, indicating that increased junctionalpermeability is primarily responsible for the perme-

ability changes of the monolayers. The mecha-nism(s) responsible for the increased junctional per-

meability are not known at present. Recent in vitrostudies have demonstrated that tumor necrosisfactor-treated aortic endothelial cell cultures un-

dergo prominent cytoskeletal changes similar tothose induced by IFN-y alone or in combination withtumor necrosis factor, which are temporally relatedto an increase in the permeability of the monolayersto macromolecules and are regulated by G pro-tein.43 The fact that leakiness of intercellular con-tacts appears concomittantly with the morphologi-cal changes of the endothelium following IFN-ytreatment may indicate that physiologically "tight"junctional complexes fail to form during the exten-sive rearrangement of the cells and their cytoskele-ton. However, other mechanisms, such as modula-tion of regulatory proteins or cell surface moleculesby IFN-y, cannot be excluded. Disruption of theBBB has been previously described as an earlyand critical event in the evolution of EAE.44-7Recent electron microscopic studies indicate thatincreased junctional permeability as well as in-creased interendothelial space and migration ofinflammatory cells are primarily responsible for theincreased permeability of the BBB to macromole-cules in this disease.46 The functional significanceof the in vitro morphological and permeabilitychanges of HBMEC, observed in our studies, is cur-rently unknown. If, however, similar changes are in-duced in situ on cerebral endothelial cells by cyto-kines released locally by activated T lymphocytes,they would provide an additional mechanism for theopening of the BBB and could facilitate the transmi-gration of inflammatory cells from blood into brainacross the endothelial barrier.

Expression of la antigen in situ by cerebral vas-cular endothelium has been previously demon-strated in autoimmune demyelinating CNS disor-ders. Thus, class 11 MHC molecules have beenlocalized on the surface of endothelial cells liningmicrovessels at the edge of demyelinating plaquesas well as within the adjacent white matter in acute,active, and silent chronic MS lesions.20,21 The pres-

ence of la-positive endothelial cells has also beendocumented in acute EAE,2425 while expression ofla antigen by cerebral endothelium in chronic re-

lapsing EAE appears to coincide with the appear-

Table 1. Permeability ofHBMEC Monolayers to HRP

No. of interendothelialNo. of labeled tight junctionst

cytoplasmic vesicles* Permeable Impermeable

Control 2.0 ± 1.7 vesicles/cell 24.8 ± 2.7% 75.2 ± 2.7%Experimental 2.4 ± 2.1 vesicles/cell 63.4 ± 5.2% 36.6 ± 5.2%

* Numbers represent mean ± SD of labeled vesicles in 100 control and 100 IFN-y-treated cells from one experiment. P > 0.05.t Numbers represent mean ± SD of 400 junctions (200 treated and 200 untreated) from two experiments using two different isolates. P <

0.05.

IFN-y-Induced Changes on Cerebral Endothelium 1275AJPApril 1993, Vol. 142, No. 4

,+ 4

-IF.

.. ..

Figure 8. HRP localization in untreated (A) and B) and IFN-y-treated (C and D) confluent HBMEC monolayers. A, under standard culture con-ditions, tight junctions at intercellular contacts (between arrowheads) impeding the passage ofHRP. B, HRP penetrating a short segment ofan in-tercellular cleftfrom the basal cell surface, forming small deposits at the basal aspect of the cleft (arrowheads) and stopping at a junctional com-plex (arrow). The remaining interendothelial cleft is free of HRP. In C, following 4 days' incubation with IFN-y (200 units/ml), heavy deposits ofHRP are seen under the basal cell surface, and the tracerpermeates the entire length ofa long intercellular cleft. The proximal portion of the cleftisfocally dilated (*). There is no increase in the pinocytotic activity of the endothelium. D, in monolayers treated with IFN-y, HRP penetrating theintercellular clefts andforming extensive deposits between the layers of overlapping cells. Bars = 0.5 ,m.

r-mm

>E :.

-i ::...... .-Y .-

1276 Huynh and Dorovini-ZisAJPApil 1993, Vol. 142, No. 4

ance of inflammatory cell infiltrates and diminisheswhen inflammation subsides.49 In addition, murinecerebral endothelial cells isolated from SJL micewith EAE are able to present antigen to sensitizedsyngeneic lymph node cells following incubationwith IFN-y in vitro.50 Contrary to these observations,cultured rat brain endothelial cells are not effectiveat stimulating T-cell division and therefore have notbeen considered important as antigen-presentingcells.51 It is now well accepted that endothelial cellsderived from different species and vascular bedsvary greatly in their function, morphology, and anti-genic properties, so that results should not be ex-trapolated from one system to the other.52 Thepresent work demonstrates that class 11 MHC mole-cules are not detectable on intact endothelial cellslining human brain microvessels by the methodsemployed in our study but can be specifically in-duced in vitro by human recombinant IFN-y in asso-ciation with prominent alterations in the morphology,organization, and permeability of the monolayers tomacromolecules. Although the ability to present an-tigen by HBMEC has yet to be unequivocallyproved, our findings indicate a possibly importantrole of the human cerebral endothelium inlymphocyte-endothelial interactions, lymphocyte re-cruitment, and alteration of blood-brain barrier per-meability in immune-mediated CNS inflammation.

AcknowledgmentsThe authors wish to acknowledge the expert techni-cal assistance of Mrs. R. Prameya in the isolationand culture of endothelial cells and Mr. D. Wong forassistance with the statistical analysis of the data.

References1. Collins T, Korman AJ, Wake CT, Boss JM, Kappes DJ,

Fiers W, Ault KA, Gimbrone MA Jr, Strominger JL,Pober JS: Immune IFN activates multiple class 11 majorhistocompatibility complex genes and the associatedinvariant chain gene in human endothelial cells anddermal fibroblasts. Proc Natl Acad Sci USA 1984, 81:4917-4921

2. Ferry B, Halttunen J, Leszczynski D, Schellekens H,Meide PHVD, Hayry P: Impact of class 11 major histo-compatibility complex antigen expression on the im-munogenic potential of isolated rat vascular endothe-lial cells. Transplantation 1987, 44:499-503

3. Pober JS, Gimbrone MA Jr: Expression of la-like anti-gens by human vascular endothelial cells is induciblein vitro: Demonstration by monoclonal antibody bind-

ing and immunoprecipitation. Proc Natl Acad Sci USA1982, 79:6641-6645

4. Pober JS, Collins T, Gimbrone MA Jr, Cotran RS, GitlinJD, Fiers W, Clayberger C, Krensky AM, Burakoff SJ,Reiss C: Lymphocytes recognize human vascularendothelial and dermal fibroblast la antigens inducedby recombinant immune interferon. Nature 1983, 305:726-729

5. Pober JS, Gimbrone MA Jr, Cotran RS, Reiss CS,Burakoff SJ, Fiers W, Ault KA: la expression by vascu-lar endothelium is inducible by activated T cells andby human y interferon. J Exp Med 1983, 157:1339-1353

6. Wagner CR, Vetto RM, Burger DR: Expression ofI-region-associated antigen (la) and interleukin 1 bysubcultured human endothelial cells. Cell Immunol1985, 93:91-104

7. Messadi DV, Pober JS, Murphy GF: Effects of recom-binant ry interferon on HLA-DR and DQ expression byskin cells in short term organ culture. Lab Invest 1988,58:61-67

8. Munro JM, Pober JS, Cotran RS: Tumor necrosis factorand interferon-y induce distinct patterns of endothelialactivation and leukocyte accumulation in skin of Papioanubis. Am J Pathol 1989, 135:121-133

9. Haskard DO, Cavender D, Fleck RM, Sontheimer R,Ziff M: Human dermal microvascular endothelial cellsbehave like umbilical vein endothelial cells in T-celladhesion studies. J Invest Dermatol 1987, 88:340-344

10. lssekutz TB: Effects of six different cytokines on lym-phocyte adherence to microvascular endothelium andin vivo lymphocyte migration in the rat. J Immunol1990, 144:2140-2146

11. Masuyama J-l, Minato N, Kano S: Mechanisms of lym-phocyte adhesion to human vascular endothelial cellsin culture. J Clin Invest 1986, 77:1596-1605

12. Oppenheimer-Marks N, Ziff M: Migration of lympho-cytes through endothelial cell monolayers: augmenta-tion by interferon-y. Cell Immunol 1988, 114:307-323

13. Thornhill MH, Kyan-Aung U, Lee TH, Haskard DO: Tcells and neutrophils exhibit differential adhesion tocytokine-stimulated endothelial cells. Immunology1990, 69:287-292

14. Male DK, Pryce G, Hughes CCW: Antigen presenta-tion in brain: MHC induction on brain endothelium andastrocytes compared. Immunology 1987, 60:453-459

15. Wilcox CE, Baker D, Butter C, Willoughby DA, Turk JL:Differential expression of guinea pig class 11 major his-tocompatibility complex antigens on vascular endo-thelial cells in vitro and in experimental allergic en-cephalomyelitis. Cell Immunol 1989, 120:82-91

16. McCarron RM, Kempsi 0, Spatz M, McFarlin DE: Pre-sentation of myelin basic protein by murine cerebralvascular endothelial cells. J Immunol 1985, 134:3100-3103

17. Pardridge WM, Yang J, Buciak J, Tourtellotte WW: Hu-man brain microvascular DR-antigen. J Neurosci Res

IFN-y-Induced Changes on Cerebral Endothelium 1277AJPApril 1993, Vol. 142, No. 4

1989, 23:337-34118. Sethna MP, Lampson LA: Immune modulation within

the brain: recruitment of inflammatory cells and in-creased major histocompatibility antigen expressionfollowing intracerebral injection of interferon-y. J Neu-roimmunol 1991, 34:121-132

19. Fabry Z, Waldschmidt MM, Moore SA, Hart MN: Anti-gen presentation by brain microvessel smooth muscleand endothelium. J Neuroimmunol 1990, 28:63-71

20. Traugott U, Scheinberg LC, Raine CS: On the pres-ence of la-positive endothelial cells and astrocytes inmultiple sclerosis lesions and its relevance to antigenpresentation. J Neuroimmunol 1985, 8:1-14

21. Traugott U, Raine CS: Multiple sclerosis. Evidence forantigen presentation in situ by endothelial cells andastrocytes. J Neurol Sci 1985, 69:365-370

22. Kajiwara K, Hirozane A, Fukumoto T, Orita T, NishizakiT, Kamiryo T, Ito H: Major histocompatibility complexexpression in brain of rats with graft-versus-host dis-ease. J Neuroimmunol 1991, 32:191-198

23. Beilke MA, Riding In D, Hamilton R, Stone GA, JordanEK, Brashears G, Nusbaum W, Huddleston D, GibbsCJ Jr, Gravell M: HLA-DR expression in macaque neu-roendothelial cells in vitro and during SIV encephalitis.J Neuroimmunol 1991, 33:129-143

24. Craggs RI, Webster HdeF: la antigens in the normalrat nervous system and in lesions of experimental al-lergic encephalomyelitis. Acta Neuropathol (Berl)1985, 68:263-272

25. Sobel RA, Natale JM, Schneeberger EE: The immuno-pathology of acute experimental allergic encephalo-myelitis. J Neuropathol Exp Neurol 1987, 46:239-249

26. Vass K, Lassmann H, Wekerle H, Wisniewski HM: Thedistribution of la antigen in the lesions of rat acute ex-perimental allergic encephalomyelitis. Acta Neuro-pathol (Berl) 1986, 70:149-160

27. Butter C, O'Neill JK, Baker D, Gschmeissner SE, TurkJL: An immunoelectron microscopical study of the ex-pression of class 11 major histocompatibility complexduring chronic relapsing experimental allergic en-cephalomyelitis in Biozzi AB/H mice. J Neuroimmunol1991, 33:37-42

28. Dorovini-Zis K, Prameya R, Bowman PD: Culture andcharacterization of microvascular endothelial cells de-rived from human brain. Lab Invest 1991, 64:425-436

29. McLean IW, Nakane PK: Periodate-lysine-parafor-maldehyde fixative. A new fixative for immunoelectronmicroscopy. J Histochem Cytochem 1974, 22:1077-1083

30. Schroeter D, Spiess E, Paweletz N, Benke R: A proce-dure for rupture-free preparation of confluently grownmonolayer cells for scanning electron microscopy. JElectron Microsc Tech 1984, 1:219-225

31. Dorovini-Zis K, Bowman PD, Betz AL, Goldstein GW:Hyperosmotic urea reversibly opens the tight junctionsbetween brain capillary endothelial cells in cell cul-ture. J Neuropathol Exp Neurol 1987, 46:130-140

32. Sobel RA, Ames MB: Major histocompatibility complexmolecule expression in the human central nervoussystem: immunohistochemical analysis of 40 patients.J Neuropathol Exp Neurol 1988, 47:19-28

33. Lampson LA, Hickey WF: Monoclonal antibody analy-sis of MHC expression in human brain biopsies: tissueranging from "histologically normal" to that showingdifferent levels of glial tumor involvement. J Immunol1986, 136:4054-4062

34. Frank E, Pulver M, de Tribolet N: Expression of class 11major histocompatibility antigens on reactive astro-cytes and endothelial cells within the gliosis surround-ing metastases and abscesses. J Neuroimmunol1986, 12:29-36

35. Skoskiewicz MJ, Colvin RB, Schneeberger EE, RussellPS: Widespread and selective induction of major his-tocompatibility complex-determined antigens in vivoby y interferon. J Exp Med 1985, 162:1645-1664

36. Ruszczak Z, Detmar M, lmcke E, Orfanos CE: Effectsof recombinant IFN Alpha, Beta and Gamma on themorphology, proliferation, and cell surface antigen ex-pression of human dermal microvascular endothelialcells in vitro. J Invest Dermatol 1990, 95:693-699

37. Heyns A du P, Eldor A, Vlodavsky I, Kaiser N, FridmanR, Panet A: The antiproliferative effect of interferonand the mitogenic activity of growth factors are inde-pendent cell cycle events. Exp Cell Res 1985, 161:297-306

38. Friesel R, Komoriya A, Maciag T: Inhibition of endothe-lial cell proliferation by y interferon. J Cell Biol 1987,104:689-696

39. Stolpen AH, Guinan EC, Fiers W, Pober JS: Recombi-nant tumor necrosis factor and immune interferon actsingly and in combination to reorganize human vascu-lar endothelial cell monolayers. Am J Pathol 1986,123:16-24

40. Saegusa Y, Ziff M, Welkovich L, Cavender D: Effect ofinflammatory cytokines on human endothelial cell pro-liferation. J Cell Physiol 1990, 142:488-495

41. Maheshwari RK, Srikantan V, Bhartiya D, KleinmanHK, Grant DS: Differential effects of interferon y and aon in vitro model of angiogenesis. J Cell Physiol 1991,146:164-169

42. Tsuruoka N, Sugiyama M, Tawaragi Y, Tsujimoto M,Nishihara T, Goto T, Sato N: Inhibition of in vitro angio-genesis by lymphotoxin and interferon-y. BiochemBiophys Res Commun 1988, 155:429-435

43. Brett J, Gerlach H, Nawroth P, Steinberg S, GodmanG, Stern D: Tumor necrosis factor/cachectin increasespermeability of endothelial cell monolayers by amechanism involving regulatory G proteins. J ExpMed 1989, 169:1977-1991

44. Vulpe M, Hawkins A, Rozdilsky B: Permeability of ce-rebral blood vessels in experimental allergic encepha-lomyelitis studied by radioactive iodinated bovine al-bumin. Neurology 1960, 10:171-177


45. Hirano A, Dembitzer HM, Becker NH, Levine S, Zim-merman HM: Fine structural alterations of the blood-brain barrier in experimental allergic encephalomyeli-tis. J Neuropathol Exp Neurol 1970, 29:432-440

46. Kristensson K, Wisniewski HM: Chronic relapsing ex-perimental allergic encephalomyelitis. Studies in vas-cular permeability changes. Acta Neuropathol (Berl)1977, 39:189-194

47. Juhler M, Barry Dl, Offner H, Konat G, Klinken L, Paul-son OB: Blood-brain and blood-spinal cord barrierpermeability during the course of experimental aller-gic encephalomyelitis in the rat. Brain Res 1984, 302:347-355

48. Claudio L, Kress Y, Factor J, Brosnan CF: Mecha-nisms of edema formation in experimental autoim-mune encephalomyelitis. Am J Pathol 1990, 137:1033-1045

49. Sakai K, Tabira T, Endoh M, Steinman L: la expressionin chronic relapsing experimental allergic encephalo-myelitis induced by long-term cultured T cell lines inmice. Lab Invest 1986, 54:345-352

50. McCarron RM, Spatz M, Kempski 0, Hogan RN,Muehl L, Mc Farlin DE: Interaction between myelin ba-sic protein-sensitized T lymphocytes and murine cere-bral vascular endothelial cells. J Immunol 1986, 137:3428-3435

51. Pryce G, Male D, Sedgwick J: Antigen presentation inbrain: brain endothelial cells are poor stimulators ofT-cell proliferation. Immunology 1989, 66:207-212

52. Zetter BR: Endothelial heterogeneity: influence of ves-sel size, organ localization and species specificity onthe properties of cultured EC. Endothelial Cells, vol 2.Edited by US Ryan. Boca Raton, FL, CRC Press, Inc,1988, pp 63-79

effects of interferon-y on primary cultures of human brain

Documents