Lymphocyte Migration Through Brain Endothelial CellMonolayers Involves Signaling Through Endothelial ICAM-1Via a Rho-Dependent Pathway1

Peter Adamson,2* Sandrine Etienne,† Pierre-Olivier Couraud, † Virginia Calder,* andJohn Greenwood*

Lymphocyte extravasation into the brain is mediated largely by the Ig superfamily molecule ICAM-1. Several lines of evidenceindicate that at the tight vascular barriers of the central nervous system (CNS), endothelial cell (EC) ICAM-1 not only acts as adocking molecule for circulating lymphocytes, but is also involved in transducing signals to the EC. In this paper, we examine thesignaling pathways in brain EC following Ab ligation of endothelial ICAM-1, which mimics adhesion of lymphocytes to CNSendothelia. ICAM-1 cross-linking results in a reorganization of the endothelial actin cytoskeleton to form stress fibers and acti-vation of the small guanosine triphosphate (GTP)-binding protein Rho. ICAM-1-stimulated tyrosine phosphorylation of theactin-associated molecule cortactin and ICAM-1-mediated, Ag/IL-2-stimulated T lymphocyte migration through EC monolayerswere inhibited following pretreatment of EC with cytochalasin D. Pretreatment of EC with C3 transferase, a specific inhibitor ofRho proteins, significantly inhibited the transmonolayer migration of T lymphocytes, endothelial Rho-GTP loading, and endo-thelial actin reorganization, without affecting either lymphocyte adhesion to EC or cortactin phosphorylation. These data showthat brain vascular EC are actively involved in facilitating T lymphocyte migration through the tight blood-brain barrier of theCNS and that this process involves ICAM-1-stimulated rearrangement of the endothelial actin cytoskeleton and functional EC Rhoproteins. The Journal of Immunology,1999, 162: 2964–2973.

T o fulfill their role in tissue surveillance, lymphocytes mustbe able to leave the circulation and traffic through thesolid tissues of the body. Thus, the migration of lympho-

cytes across the vascular endothelial cell (EC)3 wall is a prereq-uisite in the implementation of lymphocyte function. Lymphocytetransendothelial migration is dependent on lymphocyte activationvia signaling through the T cell and IL-2 receptor (1). The in vivocorrelate of these observations is that only Ag-specfic, IL-2-de-pendent T cells are capable of trafficking to the central nervoussystem (CNS) (2). The vascular recruitment of circulating lym-phocytes and their transvascular migration has been the subject ofsubstantial investigation and has led to a greater understanding ofthe role of lymphocytes under normal and inflammatory condi-

tions. Unlike most vascular endothelia, which are connected to-gether by incomplete, permeable junctions, those of the brain andretina are joined together by impermeable tight junctions formingthe blood-brain and inner blood-retinal barriers, respectively. De-spite these barriers, a low level of leukocyte traffic into the CNSoccurs (2), and this can be dramatically up-regulated during thedevelopment of immune-mediated diseases. To achieve entry,lymphocytes must either penetrate the tight junctions or migratethrough the body of the CNS EC through the formation of pores.Whichever path is utilized during leukocyte diapedesis within theCNS, it is highly likely that this process can only operate with theactive participation of the EC. Although it is clear that the activa-tion state of lymphocytes is central in controlling transvascularmigration (1, 3, 4), additional control may be provided by the CNSendothelia to further facilitate the passage of migratory cells acrossthe impermeable vessel wall.

The mechanisms by which vascular EC capture circulating lym-phocytes are now well characterized and many of the ligands thatcontribute to their adhesion and subsequent transvascular migra-tion identified (5, 6). Previous studies have shown that initial cap-ture of immune cells from the circulation occurs via endothelialselectin molecules that result in weak attachment and, under con-ditions of flow, result in leukocyte rolling along the vessel wall (6).Tighter binding and subsequent migration through the EC wall ismediated predominantly by the LFA-1/ICAM-1 (and possiblyICAM-2) pairing. However, on cytokine-activated endothelia, boththe LFA-1/ICAM-1 and the very late Ag-4/VCAM-1 interactionappear to play a part in T lymphocyte migration (3, 7–9). Recentstudies indicate that the molecules employed during lymphocyteadhesion to and migration through CNS endothelium are essen-tially the same as those governing recruitment at other vascularbeds, with ICAM-1 being the predominant endothelial moleculeinvolved in migration (1, 10, 11). However, the active participation

*Department of Clinical Ophthalmology, Institute of Ophthalmology, University Col-lege London, London, United Kingdom; and†Laboratoire d’Immuno-PharmacologieMoleculaire, Institut Cochin de Ge´netique Moleculaire, Paris, France

Received for publication September 8, 1998. Accepted for publication November23, 1998.

The costs of publication of this article were defrayed in part by the payment of pagecharges. This article must therefore be hereby markedadvertisementin accordancewith 18 U.S.C. Section 1734 solely to indicate this fact.1 This work was supported by the Wellcome Trust (U.K.), Centre National de laRecherche Scientifique, Institut National de la Sante´ et de la Recherche Me´dicale,Association pour le Developpment de la Recherche sur le Cancer, Ligue NationaleFrancaise contre le Cancer, Universite´ de Paris, and the Ministere de la Recherche etde l’Enseignement Superieur.2 Address correspondence and reprint requests to Dr. Peter Adamson, Department ofClinical Ophthalmology, Institute of Ophthalmology, University College London,Bath Street, London, EC1V 9EL, U.K. E-mail address: padamson@hgmp.mrc.ac.uk3 Abbreviations used in this paper: EC, endothelial cell(s); CNS, central nervoussystem; FAK, focal adhesion kinase; PLNC, peripheral lymph node lymphocytes;RAM, rabbit anti-mouse; LPA, lysophosphatidic acid; GTP, guanosine triphosphate;GDP, guanosine diphosphate; ECL, enhanced chemiluminescence; BCA, bicincho-ninic acid; AEBSF, 4-(2-aminoethyl) benzenesulfonyl fluoride; ADP, [32P]adenosine59-diphosphate; HRP, horseradish peroxidase; NAD, nicotinamide adeninedinucleotide.

Copyright © 1999 by The American Association of Immunologists 0022-1767/99/$02.00

of CNS EC in leukocyte extravasation, above that of the provisionof adhesion molecules, has remained largely speculative.

If CNS endothelia play an active part in facilitating lymphocytediapedesis, it is likely that they receive external signals from ad-herent lymphocytes. This has led to the possibility that moleculesintimately involved in lymphocyte diapedesis, such as ECICAM-1, may also be involved in the transduction of extracellularsignals. Proteins of the Ig superfamily, including CD2, CD4, MHCmolecules and Fcg are well-documented as signal transducers inboth lymphocytes (12) and U937 cells (13). In addition, neural celladhesion molecule has also been shown to be capable of signaltransduction in PC12 cells (14). Although it has been demonstratedin leukocytes that ICAM-1 is capable of generating intracellularsignals (15–17), and more recently, signaling via VCAM-1 andplatelet endothelial call adhesion molecule-1 has been demon-strated in platelets (18) and EC (19), the functional effects of sig-naling via Ig superfamily molecules in EC remain unresolved.

Studies aimed at exploring the signal transduction pathways inCNS EC that may lead to lymphocyte extravasation have beengreatly expedited by the recent development of two immortalizedLewis rat brain microvessel EC lines that have been immortalizedwith the E1A adenovirus protein (RBE4 cell line; 20, 21) andSV40 large T Ag (GP8.3 cell line; 22) and that retain in culture thedifferentiated phenotype of brain endothelia. Previous studies withRBE4 cells have shown that Ab cross-linking of endothelialICAM-1 molecules (used to mimic lymphocyte binding to EC viaLFA-1) or coculture of EC with lymphocytes triggers p60src ac-tivity, which appears to be responsible for the enhanced tyrosinephosphorylation of the actin-binding protein cortactin (20). Theenhanced tyrosine phosphorylation of cortactin following cross-linking of EC with ICAM-1 or coculture with encephalitogenic Tlymphocytes was the first indication of an active endothelial in-volvement in controlling transvascular lymphocyte migration fol-lowing T cell binding. Furthermore, it also served to demonstratethat T lymphocyte binding can be mimicked by ligation of ECICAM-1. However, more recently, additional actin cytoskeletal-associated proteins, such as focal adhesion kinase (FAK), paxillin,and p130cas, have also been shown to be tyrosine phosphorylatedin response to ICAM-1 cross-linking (23). In this paper, we havefurther extended this work and present evidence for the first timethat CNS vascular EC are intimately involved in the processescontrolling the transendothelial migration of T lymphocytes acrosscultures of rat brain endothelial monolayers. These processes re-quire an ICAM-1-stimulated rearrangement of the endothelial ac-tin cytoskeleton and functional endothelial Rho proteins.

Materials and MethodsMaterials

[a-32P]Nicotinamide adenine dinucleotide (NAD) was obtained from NewEngland Nuclear (Herts, U.K.). Na[51Cr207], [32P]orthophosphate, peroxi-dase coupled rabbit anti-mouse (RAM) IgG, and enhanced chemilumines-cence (ECL) reagents were obtained from Amersham International (Bucks,U.K.). Recombinant epidermal growth factor was obtained from R&D Sys-tems (Oxon, U.K.). Polyclonal anti-Rho Ab was obtained from Autogen-bioclear (Wilts, U.K.). Anti-transferin receptor mAb (MAB451) was ob-tained from Chemicon (London, U.K.). Mouse anti-rat ICAM-1 (1A29)was obtained azide-free from Serotec (Oxon, U.K.). The mouse mAb top80/85 cortactin was a generous gift from J. T. Parsons (University ofVirginia, Charlottesville, VA). The pGEX-2T-C3 construct was a generousgift from L. A. Fieg (Tufts University, Boston, MA). Unless otherwisestated, all chemicals used were obtained from Sigma (Dorset, U.K.).

Endothelial cells

Primary rat brain endothelium. Cultures were isolated as previously de-scribed (24, 25), and these methods routinely produced primary cultures of.95% purity. Briefly, rat cerebral cortex was dispersed by enzymatic di-

gestion, microvessel fragments separated from other material and singlecells by density-dependent centrifugation and plated onto collagen-coatedplastic. Cultures were maintained in Ham’s F-10 medium supplementedwith 17.5% plasma-derived serum (First Link, West Midlands, U.K.), 7.5mg/ml of EC growth supplement (First Link), 80mg/ml of heparin, 2 mMglutamine, 0.5mg/ml of vitamin C, 100 U/ml of penicillin, and 100mg/mlof streptomycin at 37°C in 5% CO2. Medium was replaced every 3 days.Immortalized EC. The immortalized Lewis rat brain EC line GP8.3 (22)was maintained in Ham’s F-10 medium supplemented with 17.5% FCS, 7.5mg/ml of EC growth supplement, 80mg/ml of heparin, 2 mM glutamine,0.5 mg/ml of vitamin C, 100 U/ml of penicillin, and 100mg/ml of strep-tomycin. The immortalized Lewis rat brain EC line RBE4 was maintainedin 1:1 Hams F-10/aMEM (Life Technologies, Paisley, U.K.) supplementedwith 10% FCS (Life Technologies), as previously described (21). Cultureswere maintained at 37°C in 5% CO2, and medium was replaced every 3days until the formation of monolayers.

Fluorescence microscopy

Actin localization. Cells were fixed in 3.7% paraformaldehyde in PBS for10 min, followed by 50 mM Tris-HCl (pH 7.5)/PBS for 10 min. Cells weresubsequently permeabilized in 0.5% Triton X-100/PBS and incubated with0.1 mg/ml of FITC-phalloidin for 1 h. Cells were exhaustively washed inPBS and viewed on a Leica (Milton Keynes, U.K.) confocal fluorescencemicroscope.ICAM-1 localization. For ICAM-1 localization, cells were fixed as de-scribed above and incubated with 10% normal goat serum/PBS. Cells weresubsequently exposed to 1mg/ml of anti-rat ICAM-1 (1A29) for 1 h atroom temperature, exhaustively washed in PBS, and incubated with anti-mouse-FITC or anti-mouse-Cy3 (1:50; Jackson ImmunoResearch, WestGrove, PA) for 1 h. Cells were exhaustively washed in PBS and viewed ona Leica confocal fluorescence microscope.

Tyrosine phosphorylation of cortactin

Cell cultures (equivalent to 13 105 cells) were lysed with 250ml ofSDS-PAGE buffer (10 mM Tris-HCl (pH 6.8), 4% SDS, 20% glycerol,0.4% dithithreitol, 1 mM sodium orthovanadate, and 0.1% bromophenolblue), boiled for 10 min, and subject to 7.5% SDS-PAGE. Proteins wereelectrophoretically transferred to nitrocellulose membranes (Amersham)and immunoblotted using either 2.5mg/ml of mouse anti-phosphotyrosine(Upstate Biotechnology, Lake Placid, NY) or 1mg/ml of anti-cortactinfollowed by peroxidase-conjugated RAM IgG. Blots were subsequentlydeveloped using the ECL system (Amersham) and exposed to x-ray film.Immunoprecipitation of p80/85 cortactin was achieved following lysis ofcells for 30 min in buffer (10 mM Tris-HCl (pH 7.5), 140 mM NaCl, 1%digitonin, 1 mM phenylmethylsulfonyl fluoride, 1 mM orthovanadate, 50U/ml of aprotinin, 1 mM EDTA, 2mg/ml of pepstatin, and 2mg/ml ofleupeptin), followed by removal of nuclei (13,0003 g for 30 min) andincubation with 1mg/ml of anti-cortactin mAb for 1 h at4°C. Cell lysateswere preincubated with 50ml of RAM-coated pansorbin (Calbiochem,Notts, U.K.) for an additional 30 min at 4°C, and immune complexes werecollected by centrifugation. Immune complexes were extensively washedin lysis buffer and further analyzed by Western blot analysis. p80/85 cor-tactin phosphorylation was quantitated by densitometry and normalized tolevels of immunoprecipitated cortactin.

Soluble (S)-Ag specific CD41 T cell lines

Lewis rat T lymphocyte cell lines specific for purified bovine retinal S-Agwere prepared as previously described (26). Briefly, lymph nodes werecollected from bovine S-Ag immunized rats, and the T lymphocytes werepropagated by periodically alternating Ag activation with IL-2 stimulation.The cell lines express the marker of the CD41 T cell subset, areCD45Rclow, and recognize S-Ag in the molecular context of MHC class IIdeterminants (26). These cells have previously been shown to be highlymigratory across monolayers of primary cultured brain and retinal endo-thelia (1, 4, 11) and represent Ag-stimulated lymphocytes.

Adhesion of peripheral lymph node cells (PLNC) to endothelia

Adhesion assays were conducted as previously described (27, 28) usingcells harvested from Lewis rat peripheral lymph nodes. Briefly, PLNCwere isolated, and T lymphocytes were obtained after purification on nylonwool columns. These cells that represent non-Ag-activated T lymphocytesare therefore nonmigratory but highly adhesive when activated with themitogen Con A (1, 4, 11). PLNC were activated for 24 h with type V ConA, washed twice in HBSS, and cells labeled with 3mCi [51Cr] per 106 cellsin HBSS for 90 min at 37°C. After washing the cells three times withHBSS, they were resuspended in RPMI 1640 medium containing 10%

2965The Journal of Immunology

FCS. Endothelial monolayers grown on 96-well plates were prepared byremoving the culture medium and washing the cells four times with HBSS.[51Cr]-labeled PLNC (200ml) at a concentration of 13 106/ml was thenadded to each well and incubated at 37°C for 1.5 h. In each assay,gemissions from each of six replicate blank wells were determined to pro-vide a value for the total amount of radioactivity added per well and toallow calculation of the specific activity of the cells. After incubation,nonadherent cells were removed with four separate washes from the fourpoles of the well with 37°C HBSS as previously described (27, 29). Ad-herent PLNC were lysed with 2% SDS, the lysate removed, andg emis-sions quantitated by spectrometry. Results are expressed as the fractionaladhesion of PLNC to untreated EC. Results were obtained from at leastthree independent experiments using a minimum of four separate wells pertreatment. The results are expressed as the means6 SEM, and significantdifferences between groups were determined by Student’st test.

T lymphocyte transendothelial migration

The ability of the immortalized cells to support the transendothelial mi-gration of Ag-specific T lymphocytes was determined using a well-char-acterized assay as previously described (1, 4, 11). Briefly, T lymphocyteswere added (23 105 cells/well) to 24-well plates containing EC mono-layers. Lymphocytes were allowed to settle and migrate over a 4-h period.To evaluate the level of migration, cocultures were placed on the stage ofa phase-contrast inverted microscope housed in a temperature controlled(37°C), 5% CO2 gassed chamber (Zeiss, Herts, U.K.). A 2003 200-mmfield was randomly chosen and recorded for 10 min spanning the 4-h timepoint using a camera linked to a time-lapse video recorder. Recordingswere replayed at 1603 normal speed, and lymphocytes were identified andcounted that had either adhered to the surface of the monolayer or that hadmigrated through the monolayer. Lymphocytes on the surface of the mono-layer were identified by their highly refractive morphology (phase-bright)and rounded or partially spread appearance. In contrast, cells that had mi-grated through the monolayer were phase-dark, highly attenuated, and wereseen to probe under the EC in a distinctive manner (1, 4, 11). Treatment ofEC with cytochalasin D or C3 transferase was conducted before addition oflymphocytes, following extensive washing and replacement into new me-dium. Control data were expressed as the percentage of total lymphocyteswithin a field that had migrated through the monolayer. All other data wereexpressed as a percentage of the control migrations. A minimum of threeindependent experiments using a minimum of wells per assay was per-formed. The results are expressed as the means6 SEM, and significantdifferences between groups are determined by Student’st test.

Expression and purification of C3 transferase

GlutathioneS-transferase-C3 was expressed inEscherichia Colifor 5 husing 1 mM isopropylb-D-thiogalactoside (Life Technologies/BRL). Cellswere harvested by centrifugation at 40003 g for 15 min and sonicatedthree times for 5 min in lysis buffer (50 mM Tris-HCl (pH 8), 50 mM NaCl,5 mM MgCl2, 1 mM DTT, and 1 mM 4-(2-aminoethyl) benzenesulfonylfluoride (AEBSF)). Bacterial lysates were then centrifuged at 10,0003 gfor 30 min, and the supernatant was chromatographed over glutathione-agarose. The column was washed with 5 vol of AEBSF-free lysis bufferand 3 vol of thrombin cleavage buffer (50 mM Tris-HCl (pH 8), 150 mMNaCl, 5 mM MgCl2, 2.5 mM CaCl2, and 1 mM DTT). Bovine plasmathrombin (10 U/ml gel) was then added to the column for 16 h at 4°C. Theeluate from the column was collected and subsequently washed with 3 volof PBS. Thrombin was removed from C3 transferase protein released fromthe column by chromatography overp-aminobezamidine-agarose and dia-lysed into PBS. C3 transferase was then concentrated in ultrafiltration units(Amicon, Beverley, MA). This procedure produced pure C3 transferaseprotein as assessed by SDS-PAGE. Protein concentration was assessedusing bicinchoninic acid (BCA) reagent (Pierce, Chester, U.K.).

[ 32P]adenosine 59-diphosphate (ADP)-ribosylation of endothelialand lymphocyte lysates with C3 transferase

Cells were lysed in 10 mM Tris-HCl (pH 7.4), 1 mM EDTA, 2 mM MgCl2,and 1 mM AEBSF, and protein concentrations were determined using BCAreagent (Pierce). Lysate protein (25mg) was then added to reactions con-taining 50 mM Tris-HCl (pH 7.4), 1 mM EDTA, 2 mM MgCl2, 0.5 mMATP, 0.3 mM guanosine triphosphate (GTP), 1 mM AEBSF, 5mCi/ml of[a-32P]NAD, and 250 ng/ml of recombinant C3 transferase and incubatedat 37°C for 1 h. Reactions were stopped by the addition of 2 vol of 30% w/vtrichloroacetic acid on ice for 30 min. Precipitated proteins were separatedby centrifugation at 14,0003 g for 10 min, and pellets were washed threetimes with ethanol at220°C. Pellets were dried, solubilized in 10 mM

Tris-HCl (pH 6.8), 4% SDS, 20% glycerol, 0.4% dithithreitol, and 0.1%bromophenol blue, and proteins resolved on 15% SDS-PAGE.

Guanine-nucleotide, immunoprecipitation, and immunoblotanalysis of endothelial Rho proteins

Serum-starved GP8.3 (53 106) cells were incubated in phosphate-freeDMEM overnight in the presence of 0.2 mCi/ml of [32P]orthophosphate.Cells were washed in HBSS and recultured in serum-free DMEM. Afteraddition of stimuli, cells were lysed in 100 mM HEPES (pH 7.4), 2%Triton X-100, 1% sodium deoxycholate, 0.1% SDS, 300 mM NaCl, 10 mMMgCl2, 2 mM EGTA, 2 mg/ml of BSA, and protease inhibitors (20 mMbenzamidine, 20mg/ml of leupeptin, 20mg/ml of pepstatin, 20mg/ml ofaprotinin, 20mg/ml of soybean trypsin inhibitor, and 2 mM AEBSF). Ly-sates were centrifuged at 14,0003 g for 5 min to sediment nuclei, andsupernatants were adjusted to 500 mM NaCl. Lysates were immunopre-cipitated by incubation with 4mg rabbit polyclonal anti-Rho (Autogenbio-clear) for 2 h at 4°Cfollowed by incubation with 50ml of 50% proteinG-Sepharose (Pharmacia, Oxon, U.K.) for 2 h at4°C. Immunoprecipitateswere washed eight times with 1 ml of 50 mM HEPES (pH 7.4), 0.005%SDS, 500 mM NaCl, and 0.1% Triton X-100. Immunoprecipitates weresubsequently heated to 68°C for 20 min in the presence of 5 mM EDTA,2 mM DTT, 0.2% SDS, 0.5 mM GTP, and 0.5 mM guanosine diphosphate(GDP) to elute [32P]phosphate-labeled nucleotides. Nucleotides were sep-arated by thin-layer chromatography on 0.1-mm polyethyleneimine-cellu-lose plates (Machery-Nagel Beds, U.K.) in 1.2 M ammonium formate/0.8M HCl (30) and autoradiographed at270°C using Fuji-RX film. Autora-diographs were subsequently used as templates and radioactive areas thatcomigrated with GTP and GDP standards, scraped, and radioactivity de-termined byb-scintillation spectrometry. EC lysates were also resolved on15% SDS-PAGE gels and electrotransferred to nitrocellulose. Membraneswere blocked with 10% dried milk protein for 1 h atroom temperature andincubated in 0.1mg/ml of polyclonal anti-Rho Ab (Autogenbioclear) for1 h at room temperature. Endothelial Rho proteins were visualized follow-ing incubation with horseradish peroxidase (HRP)-conjugated goat anti-rabbit (1:15,000; Promega, Hants, U.K.) and subsequent development byECL (Amersham). [32P]-nucleotides were normalized to levels of immu-noprecipitated Rho proteins.

Protein determination

Protein concentration in cell lysates was determined using BCA reagent(Pierce) with BSA as standard.

ResultsCross-linking of endothelial ICAM-1 results in clustering ofICAM-1 and actin stress-fiber formation

Ligation of the external domain of endothelial ICAM-1 moleculeswith mouse anti-rat ICAM-1 (1A29; 20mg/ml for 30 min) fol-lowed by cross-linking with RAM IgG (60mg/ml for 15 min) ledto clustering of endothelial ICAM-1 molecules and a redistributionof the endothelial actin cytoskeleton in both serum-starved RBE4(Fig. 1, c and d) and GP8.3 cells (data not shown), which hadpreviously been incubated with 50 U/ml of IFN-g for 48 h toincrease ICAM-1 expression. In EC, under identical conditions butin the absence of ICAM-1 cross-linking, FITC-phalloidin stainingshowed the actin cytoskeleton to be diffuse with few stress fibersand a cortical concentration of actin in both RBE4 (Fig. 1b), GP8.3(data not shown), and primary culture cells (data not shown).ICAM-1 staining was uniformly distributed within the plasmamembrane (Fig. 1a). However, following ICAM-1 cross-linkingfor 15 min, there was a dramatic increase in the number of actinstress fibers that was not observed after treatment of cells witheither RAM (Fig. 1b) or anti-ICAM-1 alone (data not shown).Identical results were observed in both RBE4 and GP8.3 cells (datanot shown). Exposure of RBE4 (Fig. 1c and j) or GP8.3 (data notshown) cells to 10mM lysophosphatidic acid (LPA) also resultedin a similar induction of actin stress fibers (Fig. 1j). Cross-linkingof the transferin receptor did not induce stress fibers in eitherGP8.3 or RBE4 cells and were identical to control or RAM-treated

2966 LYMPHOCYTE MIGRATION THROUGH BRAIN EC REQUIRES Rho

cells (data not shown). Similar effects on actin stress-fiber forma-tion were also observed in both RBE4 and GP8.3 cells followingICAM-1 cross-linking to basal levels of ICAM-1 (data not shown).

Cross-linking of endothelial ICAM-1 or coculture with Tlymphocytes results in GTP-loaded endothelial Rho

Serum-starved GP8.3 cells induced with 50 U/ml of IFN-g for 48 hwere labeled with 0.2 mCi/ml of [32P]-orthophosphate in phos-phate-free growth medium for 20 h. Cells were washed in phos-

phate containing serum-free growth media and stimulated withRAM (60 mg/ml) alone, anti-ICAM-1 (20mg/ml) followed byRAM (60 mg/ml), anti-transferin receptor (20mg/ml) followed byRAM (60 mg/ml), or a 10:1 ratio of syngeneic peripheral nodelymphocytes. Rho proteins were subsequently immunoprecipitatedfrom cell lysates, bound nucleotides eluted, and analyzed by TLC.Analysis of GTP:GDP ratios of Rho proteins immunoprecipitatedfrom GP8.3 EC showed that following cross-linking of ECICAM-1 for 10 min, or coculture with T lymphocytes for 10 min

FIGURE 1. Effect of ICAM-1 cross-linking Abs onICAM-1 localization and actin distribution: effect of cy-tochalasin D and C3 transferase. RBE4 cells were preincu-bated for 48 h with 50 U/ml of rat IFN-g and exposed to 20mg/ml of mouse anti-rat ICAM-1 (1A29) for 30 min, fol-lowed by RAM (60mg/ml) for 15 min. Cells were then sub-sequently fixed in 3.7% paraformaldehyde for 10 min, fol-lowed by 50 mM Tris-HCl (pH 7.5) for 10 min and stainedfor actin and ICAM-1 localization. Cells were either incu-bated with 1mg/ml of anti-rat ICAM-1 (1A29) or perme-abilized in 0.5% Triton X-100 and incubated in 0.1mg/ml ofFITC-phalloidin for 1 h. For ICAM-1 staining, cells werewashed six times in phosphate buffered saline A and incu-bated with anti-mouse-Cy3 (1:50) for 1 h. Cells were thenexhaustively washed in phosphate buffered saline A and vi-sualized by confocal fluorescence microscopy. Cells wereserum- and growth factor-deprived 48 h before addition ofcross-linking Abs. Identical fields were stained for ICAM-1and actin.a andb, RAM (60 mg/ml) only. c andd, ICAM-1(20 mg/ml) followed by RAM (60mg/ml). e and f, ICAM-1(20 mg/ml) followed by RAM (60mg/ml) in presense of 2mM cytochalasin D.g andh, ICAM-1 (20 mg/ml) followedby RAM (60 mg/ml) following 8 h pretreatment with 50mg/ml of C3 transferase.i andj, 10 mM LPA. a, c, e, g, andi, stained for ICAM-1.b, d, f, h, andj, stained for actin. Scalebar5 25 mm. ICAM-1 stains are shown as projections of alloptical sections. A single identical optical section is shownfor phalloidin staining.

2967The Journal of Immunology

(following initial settling and adhesion of T lymphocytes), therewas an 8.0-fold and 9.5-fold increase in the GTP:GDP ratio, re-spectively, as compared with either unstimulated cells or cellstreated with RAM only (Fig. 2A). The increase in GTP:GDP-bound Rho was transitory following ICAM-1 cross-linking, return-ing to control levels at 30 min. C3 transferase pretreatment of ECshowed an 85% and 77% inhibition of ICAM-stimulated and lym-phocyte-stimulated Rho-GTP loading in EC, respectively. Immu-noprecipitation of transferin receptor from ICAM-1/RAM-treatedcells followed by GTP/GDP analysis revealed no labeled guaninenucleotides demonstrating that32P-labeled nucleotides were de-rived from immunoprecipitated Rho proteins. In addition, immu-noblot analysis of cell lysates demonstrated that similar amounts ofEC Rho proteins were immunoprecipitated following each treat-ment (Fig. 2B).

Cytochalasin D inhibits ICAM-1-mediated stress-fiber formationand cortactin phosphorylation in EC

Since ICAM-1 cross-linking induced the formation of actin stressfibers in EC, the potential role of the actin cytoskeleton in regu-lating ICAM-1-mediated signals was assessed. Therefore, ICAM-1was cross-linked on RBE4 EC as described above in the presenceof 2 mM cytochalasin D or vehicle (0.5% DMSO). p80/85 cortac-tin was immunoprecipitated from EC lysates obtained from eithercontrol or treated cells. Immunoprecipitated proteins were resolvedon 7.5% SDS-PAGE and immunoblotted with anti-phosphoty-rosine Ab. p80/85 cortactin (20) showed a 2.7-fold increase in

tyrosine phosphorylation following cross-linking of ICAM-1,which was abolished (inhibited by 100%) in the presence of 2mMcytochalasin D (Fig. 3). Reprobing of immunoblots with mouseanti-cortactin mAb revealed that equal amounts of p85 cortactinwere immunoprecipitated following each treatment. Treatment ofRBE4 cells with 2mM cytochalasin D for 1 h resulted in thedisruption of all polymerized actin that exhibited a punctate dis-tribution (Fig. 1f). Under these conditions, cross-linking ICAM-1

FIGURE 2. ICAM-1 cross-linking induces an increase in GTP-loaded Rho proteins.A, Serum-starved GP8.3 cells were cultured in the presence of 50U/ml of rat IFN-g for 48 h and in phosphate-free growth medium supplemented with 0.2 mCi/ml of [32P]orthophosphate for 20 h. Cells were washed innormal growth medium and were stimulated with either RAM (60mg/ml) alone for 10 min, anti-ICAM-1 (20mg/ml) for 30 min followed by RAM (60mg/ml) for 10 min, anti-transferin receptor (20mg/ml) followed by RAM (60mg/ml), or a 10:1 ratio of lymphocytes. In C3 transferase experiments, cellswere treated for 4 h with 50mg/ml of C3 transferase before incubation with Abs or coculture with PLNC. Cells were lysed on ice in 0.5 ml of 100 mMHEPES (pH 7.4), 2% Triton X-100, 1% sodium deoxycholate, 0.1% SDS, 300 mM NaCl, 10 mM MgCl2, 2 mM EGTA, 2 mg/ml of BSA, and proteaseinhibitors. Nuclei were removed and lysates adjusted to 500 mM NaCl. Anti-Rho Ab (4mg) (1) or anti-transferin receptor (control) (1) was added andincubated with lysates for 2 h at 4°Cfollowed by 50ml of 50% protein-G Sepharose. Immunoprecipitates were washed with 50 mM HEPES (pH 7.4),0.005% SDS, 500 mM NaCl, 0.1% Triton X-100 and heated to 68°C for 20 min in the presence of 5 mM EDTA, 2 mM DTT, 0.2% SDS, 0.5 mM GTP,and 0.5 mM GDP. Labeled nucleotides were separated by TLC on polyethyleneimine-cellulose plates in 1.2 M ammonium formate/0.8 M HCl andautoradiographed. Areas containing GTP and GDP as assessed by comigration with GTP and GDP standards were scraped and radioactivity determinedby b-scintillation spectrometry.1, presence of ICAM-1 mAb, RAM transferin-receptor mAb, C3 transferase, or lymphocytes.2, no mAb, C3 transferase,or lymphocytes added. Time points involving the addition of lymphocytes were taken following an initial coculture period of 30 min that was necessaryto allow the majority of the T lymphocytes to settle on the EC monolayer.

FIGURE 3. Effect of cytochalasin D on ICAM-1-stimulated cortactinphosphorylation. Serum-starved RBE4 cells cultured in the presence of 50U/ml of rat IFN-g for 48 h were stimulated with either RAM (60mg/ml)alone (2) or anti-ICAM-1 (20mg/ml), followed by RAM (60mg/ml) (1)in the presence of 2mM cytochalasin D (in 0.5% DMSO) or vehicle(DMSO alone). Cells were lysed and immunoprecipitated with anti-cortactin Ab, and immunoprecipitates were resolved on 7.5% SDS-PAGE.Proteins were transferred to nitrocellulose and immunoblotted with anti-phosphotyrosine Ab (A), followed by stripping and reprobing with anti-cortactin Ab (B). Immunoblots were exposed to anti-mouse HRP-conju-gated secondary Ab and developed using ECL.

2968 LYMPHOCYTE MIGRATION THROUGH BRAIN EC REQUIRES Rho

Abs were unable to induce the appearance of actin stress fibers.Pretreatment of EC with cytochalasin D also appeared to alter thecellular localization of ICAM-1 (Fig. 1e).

ICAM-1-stimulated stress-fiber formation, but not cortactinphosphorylation, is inhibited by pretreatment of EC with C3transferase

In a variety of other cell types, the regulation of actin stress fibersis regulated by members of the Rho family of small GTP-bindingproteins (31). Therefore, to assess the potential role of these pro-teins in ICAM-1-mediated signaling, EC were treated with thebacterial toxin C3 transferase, which is capable of entering cellsand inactivating Rho proteins. Treatment of EC with 50mg/ml ofC3 transferase followed by in vitro ribosylation of EC lysates withC3 transferase and [a-32P]NAD showed a time-dependent inacti-vation of EC Rho proteins (Fig. 4). Inactivation of EC Rho pro-teins is apparent as a reduction in the availability of substrate forsubsequent in vitro [a-32P]ADP ribosylation in EC lysates. Onlyone major C3 transferase substrate was observed in EC or lym-phocyte lysates with an apparent m.w. of;25 kDa, which is con-sistent with Rho proteins that have been correctly posttranslation-ally modified (32). C3 transferase treatment of both RBE4 (Fig.1h) and GP8.3 cells (data not shown) significantly reduced thenumber of actin stress fibers and abolished the induction of stress-fiber formation following subsequent ICAM-1 cross-linking. In asimilar manner to cytochalasin D, C3 transferase also resulted in aredistibution of EC ICAM-1 (Fig. 1g). Immunoprecipitation andWestern blot analysis demonstrated that preincubation of eitherRBE4 (Fig. 4) or GP8.3 (data not shown) EC for 8 h with 50mg/mlof C3 transferase, under which conditions all EC Rho protein wasinactivated, was ineffective in inhibiting the ICAM-1-mediated en-hanced tyrosine phosphorylation of p80/85 cortactin (Fig. 5).Cross-linking of RBE4 cells with ICAM-1/RAM led to a 3.7-foldincrease in tyrosine phosphorylated cortactin, which compared with a3.6-fold induction following preincubation with C3 transferase.

Cytochalasin D inhibits lymphocyte migration through, but notadhesion to, EC monolayers

Rat brain EC monolayers were able to support the ICAM-1-de-pendent transendothelial migration of Ag-specific lymphocytesover a 4-h period with 436 5.5% (n 5 22) of the lymphocytesmigrating through the EC monolayer. The ICAM-1-dependency ofAg-specific T cell migration through CNS EC has previously beendemonstrated in Ab blockade studies (1, 11). Pretreatment ofGP8.3 monolayers with 2mM cytochalasin D for 1 h followed byremoval, exhaustive washing, and replacement with fresh mediumbefore coculture with lymphocytes dramatically reduced transen-dothelial migration of Ag-specific lymphocytes to 61.86 14.3%( p , 0.0001;n 5 6) of the control value (Fig. 6A). Increasing the

pretreatment time of GP8.3 monolayers with 2mM cytochalasin Dto either 3 h or 24 hbrought about a further decrease in migrationto 39.7 6 12.3.% and 35.86 14.8% of control values, respec-tively. Similar results were obtained from identical experimentsusing both primary cultures (data not shown) and RBE4 cells (datanot shown). Increased pretreatment times of EC with cytochalasinD were found to be necessary since cytochalasin D was not presentduring the 4-h lymphocyte coculture, and the effects of cytocha-lasin D treatment on EC stress-fiber formation was observed to bepartially reversible under these conditions. Basal adhesion of ConA-stimulated PLNC, which adhere in an identical manner to Ag-specific T cell lines, but are not capable of transendothelial migra-tion, (4) to GP8.3 EC was 18.06 0.7%. Basal adhesion of PLNCto primary cultures and RBE4 cells was similar to that observed forGP8.3. Pretreatment of EC with either 2–20mM cytochalasin Ddid not affect the adhesion of PLNC to GP8.3 (Fig. 6B), RBE4, orprimary cultures of EC (data not shown). These findings demon-strate that the effects of cytochalasin D in inhibiting lymphocytemigration is due to an effect on the endothelial support of lym-phocyte migration and not to prevention of lymphocyte binding toendothelia. In addition, it was also observed that treatment of ECwith cytochalasin D did not affect the adhesion of cocultured Ag-specific lymphocytes since, during time-lapse video microscopy,lymphocytes displayed normal spreading and motile behavior onthe EC surface. All EC monolayers used were able to excludetrypan blue following pretreatment with cytochalasin D and co-culture with lymphocytes.

FIGURE 4. Time course of C3-mediated ADP ribosylation of Rho proteins in EC and lymphocytes. Confluent cultures of EC were pretreated forspecified times (0–8 h) with 50mg/ml of recombinant C3 transferase. In separate experiments, GP8.3 cells, RBE4 cells, or lymphocytes cocultured withC3-treated GP8.3 EC were washed, lysed, and ADP ribosylated with 250 ng/ml of C3 transferase in the presence of 5mCi/ml of [a-32P]NAD, 50 mMTris-HCl (pH 7.5), 1 mM EDTA, 2 mM MgCl2, 1 mM AEBSF, 300mM GTP, and 0.5 mM ATP. Lysate proteins were then resolved on 15% SDS-PAGEand autoradiographed.

FIGURE 5. Effect of C3 transferase on ICAM-1-stimulated cortactinphosphorylation. RBE4 cells cultured in the presence of 50 U/ml of IFN-gfor 48 h and serum-starved for 24 h were stimulated with either RAM aloneor anti-ICAM-1 (20mg/ml), followed by RAM (60mg/ml) (1) in eitheruntreated cells or cells previously treated with 50mg/ml of recombinant C3transferase for 8 h. Cells were lysed and immunoprecipitated with anti-cortactin Ab, and immunoprecipitates were resolved on 7.5% SDS-PAGE.Proteins were transferred to nitrocellulose and immunoblotted with anti-phosphotyrosine Ab (A), followed by stripping and reprobing with anti-cortactin Ab (B). Immunoblots were exposed to anti-mouse HRP-conju-gated secondary Ab and developed using ECL.

2969The Journal of Immunology

Transendothelial lymphocyte migration but not adhesion isinhibited by pretreatment of EC with C3 transferase

Pretreatment of both GP8.3 and RBE4 EC with 50mg/ml of C3transferase resulted in the inhibition of transendothelial lympho-cyte migration (Fig. 7,A andB). Lymphocyte migration throughmonolayers of GP8.3 cells after incubation with C3 transferase for4 h and 8 h was reduced to 60.16 4.2% (p , 0.005) and 18.464.1% (p , 0.005) of control values, respectively. Lymphocytemigration across cultures of RBE4 cells was only significantly in-hibited after preincubation with C3 transferase for 12 h beforecoculture with lymphocytes, which resulted in a reduction to23.46 4.6% (p , 0.0001) of control lymphocyte migration. Thetime of C3 transferase incubation necessary for significant inhibi-tion of lymphocyte migration corresponded to total inactivation ofRBE4 cell Rho protein as assessed by in vitro ADP ribosylation(Fig. 4).

ADP ribosylation of lymphocyte lysates obtained from cells pre-viously cocultured for 4 h with C3 treated EC showed no inacti-vation of lymphocyte Rho proteins (Fig. 4). The ability of lym-phocytes to migrate across untreated endothelial monolayers wasalso not affected by their pre-exposure to C3-treated EC, sincelymphocytes that had previously been cocultured with C3-treatedendothelial monolayers were found to migrate normally when sub-sequently cocultured with untreated EC (data not shown). Thebinding of Con A-activated PLNC to GP8.3 or RBE4 cells wasunaffected by preincubation of EC with 50mg/ml of C3 transferasefor up to 24 h (Fig. 7,C andD). In addition, treatments that areknown to activate Rho function, such as exposure of cells to LPA(10 mM) (31) or epidermal growth factor (100 nM), which is ca-pable of up-regulating RhoB mRNA in GP8.3 cells (P.A. and J.G.,unpublished observations) and rat fibroblasts (33), did not affectthe extent of lymphocyte/EC adhesion in either GP8.3 or RBE4cells (Fig. 7,C andD). Ag-specific lymphocytes cocultured with

C3 transferase-treated EC displayed normal surface spreading andmotility on the EC surface. C3 treatment of EC had no affect on theintegrity of the EC monolayer, and the EC retained their ability toexclude trypan blue. Con A-stimulated PLNC were used in adhe-sion assays to demonstrate that the effects of both cytochalasin Dand C3 transferase were due to effects on EC mechanisms respon-sible for facilitating T lymphocyte migration, and not due to inhi-bition of lymphocyte binding to EC. Con A-stimulated PLNC andAg-specific T lymphocytes have the same adhesive capacity, butPLNC do not have the ability to migrate due to the absence of Agand IL-2 stimulation (1). This approach allows independent as-sessment of factors affecting the separate steps of adhesion andmigration. Using time-lapse video microscopy, no differences wereobserved in the adhesion of Ag-specific T lymphocytes to EC fol-lowing treatment of cells with either cytochalasin D or C3 trans-ferase (data not shown)

DiscussionThe activation state of a circulating T lymphocyte is of paramountimportance in determining its ability to migrate out of the circu-lation (2–4, 7). Nevertheless, using functional assays, our studieshave shown for the first time that brain EC are intimately andactively involved in facilitating lymphocyte diapedesis. It has pre-viously been demonstrated that the expression of the adhesionmolecule ICAM-1 on EC is pivotal in supporting lymphocyte mi-gration across vascular endothelium (1, 3, 7, 11). To date, its rolein this process has been generally interpreted as the provision of anextracellular docking molecule for leukocyte attachment and lo-comotion via ligation with receptor LFA-1. However, despite alack of direct evidence, it has been proposed that the efficient trans-vascular migration of lymphocytes require intracellular signals tobe generated within the EC. This paper demonstrates that the mi-gration of lymphocytes through CNS vascular barriers can only

FIGURE 6. Effect of cytochalasin D on lymphocyte adhesion to and migration through brain endothelial monolayers. GP8.3 cells were treated with eithervehicle (0.5% DMSO) or 2mM cytochalasin D for various times. After removal and vigorous washing,51Cr-labeled Con A-activated (5mg/ml) rat PLNC(adhesion) or Ag-specific T cells (migration) were added and allowed to adhere for 90 min or migrate over a 4-h period, respectively.A, Migration ofAg-specific lymphocytes through GP8.3 cells.B, Adhesion of Con A-stimulated PLNC to GP8.3 cells. Observations are a minimum of three independentexperiments using a minimum of four wells per assay. Data is expressed as mean6 SEM percent of control migration. Significant differences betweengroups were determined by Student’st test.p, p , 0.02. Identical data was also obtained with both primary cultures and RBE4 cells.

2970 LYMPHOCYTE MIGRATION THROUGH BRAIN EC REQUIRES Rho

proceed through the active involvement of intracellular processeswithin the EC, in addition to those elicited within the lymphocytefollowing Ag stimulation. In support of this thesis, we have re-cently shown that brain EC ICAM-1 is capable of eliciting signaltransduction events following ICAM-1 cross-linking or coculturewith T lymphocytes, demonstrating that EC actively respond toleukocyte adhesion (20). Mimicking lymphocyte attachmentthrough cross-linking EC ICAM-1 molecules in vitro, we haveconfirmed in both GP8.3 and RBE4 rat brain EC lines that there isan enhanced tyrosine phosphorylation of cortactin (20), FAK, pax-illin, and p130cas (23). Following ligation of EC ICAM-1, whichwas up-regulated by prior exposure to IFN-g, there is a reorgani-zation of the EC actin cytoskeleton that results in stress-fiber for-mation. However, similar effects were also observed following li-gation to basal ICAM-1 expressed on EC, suggesting thatpretreatment of cells with IFN-g was not involved in ICAM-1-induced signals. Pretreatment of EC with cytochalasin D inhibitedboth ICAM-1-stimulated cortactin phosphorylation and lympho-cyte transendothelial migration, which supports the proposal thatthe endothelial actin cytoskeleton plays a vital role in orchestratingthese events.

The observation that both ICAM-1 cross-linking and cocultureof EC with Con A-stimulated PLNC results in increased levels ofGTP-loaded endothelial Rho proteins strongly suggests that endo-

thelial Rho proteins are involved in ICAM-1-mediated signalingevents. As previously reported (34), Rho itself may be activatedthrough cell-surface signals propagated through the actin cytoskel-eton. In support of this, pretreatment of EC with C3 transferase,which is able to specifically ADP ribosylate (35) and inhibit Rhoproteins, resulted in a substantial inhibition of transendotheliallymphocyte migration, inhibition of both ICAM-1 and lympho-cyte-stimulated Rho-GTP loading, and inhibition of ICAM-1-stim-ulated stress-fiber formation, but was unable to inhibit ICAM-1-stimulated cortactin phosphorylation. Other studies from theselaboratories have also demonstrated that ICAM-1-induced in-creases in the tyrosine phosphorylation of FAK, paxillin, andp130cas and increases in the activity of Jun-kinase, are all effec-tively inhibited following pretreatment of GP8.3 and RBE4 cellswith C3 transferase (23). Treatment of EC monolayers with 50mg/ml of recombinant C3 transferase for up to 8 h resulted in thecomplete inactivation of EC Rho proteins, as assessed by the in-ability of C3 transferase to incorporate [a-32P]ADP-ribose intolysates of EC monolayers previously exposed to C3 transferase inculture (Fig. 4). It was also observed that the inactivation of Rhoproteins with C3 transferase was more rapid in GP8.3 cells than inRBE4 cells, which may reflect either a reduced uptake of C3 trans-ferase into RBE4 cells or, as suggested in Fig. 4, increased levels

FIGURE 7. Effect of C3 transferase on lympho-cyte adhesion to and migration through EC mono-layers. Endothelial monolayers were pretreated with50 mg/ml of recombinant C3 transferase for speci-fied times between 0–12 h. After removal and vig-orous washing,51Cr-labeled Con A-activated (5mg/ml) rat PLNC (adhesion) or Ag-specific T cells(migration) were added and allowed to adhere for 90min or migrate over a 4-h period, respectively.A,Migration of Ag-specific lymphocytes throughGP8.3 cells.B, Migration of Ag-specific lympho-cytes through RBE4 cells.C, Adhesion of Con A-stimulated PLNC to GP8.3 cells.D, Adhesion ofCon A-stimulated PLNC to RBE4 cells. Observa-tions are a minimum of three independent experi-ments using a minimum of four wells per assay.Data is expressed as mean6 SEM percent of con-trol migration. Significant differences betweengroups were determined by Student’st test.p, p ,0.02.

2971The Journal of Immunology

of C3 transferase substrate. The reduced efficiency of C3 trans-ferase in ribosylating Rho proteins in RBE4 cells, compared withGP8.3 cells, correlated with the reduced effectiveness of C3 trans-ferase in inhibiting transendothelial lymphocyte migration throughmonolayers of RBE4 cells. These studies have also demonstratedthat the reduction in the ability of lymphocytes to migrate throughEC monolayers is not due to C3-mediated inactivation of lympho-cyte Rho proteins, since cells cocultured with C3-treated ECmonolayers are able to migrate normally when transferred to un-treated EC monolayers (data not shown). In addition, previousstudies have shown that pretreatment of leukocytes with C3 trans-ferase prevents their adhesion to vascular endothelia (36), whichwas not observed following treatment of EC with C3 transferase.Taken together, these data suggest the effect of C3 transferasetreatment is mediated through inactivation of EC Rho proteins.

The inability of C3 transferase to inhibit tyrosine phosphoryla-tion of cortactin, which is effectively inhibited following pretreat-ment of cells with cytochalasin D, suggests that the role of both theactin cytoskeleton and cortactin phosphorylation may be upstreamof endothelial Rho proteins. Thus, it would appear that ICAM-1-mediated signals within EC are propagated via the actin cytoskel-eton, which results in the subsequent activation of Rho proteins.This scheme has previously been suggested in studies on fibro-blasts, in which constitutively activated Rho, when introduced di-rectly into cells, was able to restore Rho mediated focal adhesionsin the presence of cytochalasin D (33). Recent studies have dem-onstrated that tyrosine phosphorylation of cortactin, which wehave shown to be phosphorylated by p60src (20), is responsible foran increase in organized actin (37). The observation that increasedcortactin phosphorylation following ICAM-1 cross-linking is in-sensitive to pretreatment of EC with C3 transferase implies thatcortactin may be intimately associated with the propagation ofICAM-1-induced signals within EC, which results in the subse-quent activation of Rho proteins and is therefore upstream of Rhoactivation. Alternatively, tyrosine phosphorylation of cortactinmay lie on an independent ICAM-1-stimulated pathway.

The ability of Con A-stimulated PLNC to stimulate endothelialRho-GTP loading also demonstrates that the transduction of sig-naling events within the EC is not dependent on the ability of thelymphocytes to migrate through the EC monolayer and is thereforelikely to occur through cell-cell adhesion. As previously sug-gested, the ability of T lymphocytes to migrate through EC mono-layers therefore resides in an ability to effectively signal to ECcoupled with signal transduction events elicited within T lympho-cytes via Ag stimulation (1). The finding that preincubation of Tlymphocytes with anti-LFA-1 Abs is able to effectively inhibitboth T lymphocyte adhesion and transendothelial migration (11)also supports the view that signals induced following coculture ofT cells with EC are mediated through endothelial ICAMmolecules.

The physiological mechanisms within EC that allow lympho-cytes to migrate through the physical barrier afforded by CNS EC,and that lead to either pore formation or disaggregation of tightjunctions, remains unresolved. However, in polarized epithelialcells, RhoA has been shown to be present in the cytosol of MadinDarby canine kidney (MDCK) cells where cell-cell contacts aredisrupted, but translocate to cell margins when cell-cell contactsare restored (38). In addition, Rho has been reported to regulateboth perijunctional actin organization and tight junctions in bothT84 and Caco2 epithelial cells (39). Alternatively, Rho proteinshave been shown to regulate cell membrane invaginations inXe-nopusoocytes (40). A similar role for Rho proteins in regulatingjunctional organization or pore formation within CNS EC maytherefore provide a mechanism by which lymphocytes are able to

penetrate the tight vascular barriers of the CNS. In conclusion, wepropose that T lymphocyte diapedesis through CNS microvesselsis actively facilitated by mechanisms involving adhesion-depen-dent signaling within the endothelium by a process that requires anintact endothelial actin cytoskeleton and functional endothelialRho proteins.

AcknowledgmentsWe thank L. Fieg for the pGEX-2T-C3 construct, J. T. Parsons for anti-cortactin mAb, and O. Durieu-Trautmann for her contribution to the initialphase of this study.

References1. Pryce, G., D. K. Male, I. Campbell, and J. Greenwood. 1997. Factors controlling

T-cell migration across rat cerebral endothelium in vitro.J. Neuroimmunol. 75:84.

2. Hickey, W. F., B. L. Hsu, and H. Kimura. 1991. T-lymphocyte entry into thecentral nervous system.J. Neurosci. Res. 28:254.

3. Oppenheimer-Marks, N., L. S. Davis, and P. E. Lipsky. 1990. Human T-lym-phocyte adhesion to endothelial cells and transendothelial cell migration: alter-ation of receptor use relates to the activation status of both the T-cell and theendothelial cell.J. Immunol. 145:140.

4. Greenwood, J., and V. Calder. 1993. Lymphocyte migration through culturedendothelial cell monolayers derived from the blood-retinal barrier.Immunology80:401.

5. Osborn, L. 1990. Leukocyte adhesion to endothelium in inflammation.Cell 62:3.6. Springer, T. A. 1994. Traffic signals for lymphocyte recirculation and leukocyte

emigration: the multistep paradigm.Cell 76:301.7. Oppenheimer-Marks, N., L. S. Davis, D. T. Bogue, J. Ramberg, and P. E. Lipsky.

1991. Differential utilization of ICAM-1 and VCAM-1 during the adhesion andtransendothelial migration of human T lymphocytes.J. Immunol. 147:2913.

8. Kavanaugh, A. F., E. Lightfoot, P. E. Lipsky, and N. Oppenheimer-Marks. 1991.Role of CD11/CD18 in adhesion and transendothelial migration of T cells.J. Im-munol. 146:4149.

9. Yednock, T .A., C. Cannon, I. C. Fritz, F. Sanchez-Madrid, L. Steinman, andN. Karin. 1992. Prevention of experimental autoimmune encephalitis by antibod-ies againsta4b1 integrin.Nature 356:63.

10. Male, D., J. Rahman, G. Pryce, T. Tamatani, and M. Miyasaka. 1994. Lympho-cyte migration into the CNS modeled in vitro: roles of LFA-1, ICAM-1 andVLA-4. Immunology 81:366.

11. Greenwood, J., Y. Wang, and V. Calder. 1995. Lymphocyte adhesion and trans-endothelial migration in the CNS: the role of LFA-1, ICAM-1, VLA-4 andVCAM-1. Immunology 86:408.

12. Veillette, A., M. A. Bookman, E. M. Horak, and J. B. Bolan. 1988. The CD4 andCD8 T cell surface antigens are associated with the internal membrane tyrosineprotein kinase p56lck. Cell 55:301.

13. Liao, F., H. S. Shin, and S. G. Rhee. 1992. Tyrosine phosphorylation of phos-pholipase C-g 1 induced by cross-linking the high affinity or low affinity Fcreceptor for IgG in U937 cells.Proc. Natl. Acad. Sci. USA 89:3659.

14. Doherty, P., and F. S. Walsh. 1992. Cell adhesion molecules, second messengersand axonal growth.Curr. Opin. Neurobiol. 2:595.

15. Rothlein, R., T. K. Kishimoto, and E. Mainolfi. 1994. Cross-linking of ICAM-1induces co-signaling of an oxidative burst from mononuclear leukocytes.J. Im-munol. 152:2488.

16. Chintana, C., S. A. Tibbetts, M. A. Chan, and S. H. Benedict. 1995. Cross-linkingof ICAM-1 on T-cells induces transient tyrosine phosphorylation and inactivationof cdc2 kinase.J. Immunol. 155:5479.

17. van Horssen, M., S. Loman, G. T. Rijkers, S. E. Bloom, and A. C. Bloem. 1995.Co-ligation of ICAM-1 (CD54) and membrane IgM negatively affects B-cellreceptor signaling.Eur. J. Immunol. 25:154.

18. Jackson, D. E., C. M. Ward, R. Wang, and P. J. Mewman. 1997. The proteintyrosine phosphatase SHP-2 binds platelet/endothelial cell adhesion molecule-1(PECAM-1) and forms a distinct signaling complex during platelet aggregation.J. Biol. Chem. 272:6986.

19. Ricard, I., M. D. Payet, and G. Dupuis. 1997. Clustering the adhesion moleculesVLA-4 (CD49d)/(CD29) in Jurkat T-cells or VCAM-1 (CD106) in endothelialcells (ECV304) activates the phosphoinositide pathway and triggers Ca21 mo-bilization. Eur. J. Immunol. 27:1530.

20. Durieu-Trautmann, O., N. Chaverot, S. Cazaubon, A. D. Strosberg, andP-O. Couraud. 1994. Intercellular adhesion molecule 1 activation induces ty-rosine phosphorylation of the cytoskeleton-associated protein cortactin in brainmicrovessel endothelial cells.J. Biol. Chem. 269:12536.

21. Roux, F., O. Durieu-Trautmann, N. Foignant-Chaverot, M. Claire, P. Mailly,J. M. Bourre, A. D. Strosberg, and P-O. Couraud. 1994. Regulation ofg-glutamyltranspeptidase and alkaline phosphatase activities in immortalized rat brain mi-crovessel endothelial cells.J. Cell. Physiol. 159:101.

22. Greenwood, J., G. Pryce, L. Devine, D. K. Male, W. L. C. dos Santos,V. L. Calder, and P. Adamson. 1996. SV40 large T immortalized cell lines of therat blood-brain and blood-retinal barriers retain their phenotypic and immuno-logical characteristics.J. Neuroimmunol. 71:51.

2972 LYMPHOCYTE MIGRATION THROUGH BRAIN EC REQUIRES Rho

23. Etienne S., P. Adamson, J. Greenwood, A. D. Strosberg, S. Cazaubon, andP-O. Couraud. 1998. Rho-dependent signaling pathways coupled to ICAM-1 inmicrovascular brain endothelial cells.J. Immunol. 161:5755.

24. Abbott, N. J., C. C. W. Hughes, P. A. Revest, and J. Greenwood. 1992. Devel-opment and characterization of a rat brain capillary endothelial cell culture: to-wards an in vitro blood-brain barrier.J. Cell Sci. 103:23.

25. Greenwood, J. 1992. Characterization of a rat retinal endothelial cell culture andthe expression of P-glycoprotein in brain and retinal endothelium.J. Neuroim-munol. 39:123.

26. Zhao, Z-S., V. L. Calder, M. McLauchlan, and S. L. Lightman. 1994. Differentiallymphokine expression by rat antigen-specific CD41 T cell lines with antigen andmitogen.Cell. Immunol. 159:220.

27. Hughes, C. C. W., D. K. Male, and P. L. Lantos. 1988. Adhesion of lymphocytesto cerebral microvascular cells: effects of interferon-g, tumor necrosis factor andinterleukin-1.Immunology 64:677.

28. Male, D., G. Pryce, C. Hughes, and P. Lantos. 1990. Lymphocyte migration intobrain modeled in vitro: control by lymphocyte activation, cytokines, and antigen.Cell. Immunol. 127:1.

29. Wang, Y., V. Calder, J. Greenwood, and S. Lightman. 1993. Lymphocyte adhe-sion to cultured endothelial cells of the blood-retinal barrier.J. Neuroimmunol.48:161.

30. Downward, J., J. D. Graved, P. H. Warne, S. Rayter, and D. A. Cantrell. 1990.Stimulation of p21ras upon T-cell activation.Nature 346:719.

31. Ridley, A. J., and A. Hall. 1992. The small GTP-binding protein Rho regulatesthe assembly of focal adhesions and actin stress fibers in response to growthfactors.Cell 70:389.

32. Adamson, P., C. J. Marshall, A. Hall, and P. A. Tilbrook. 1992. Post-translationalmodifications of the p21rho proteins.J. Biol. Chem. 267:20033.

33. Jahner, D., and T. Hunter. 1991. The ras related gene rhoB is an immediate earlygene inducible by v-fps, epidermal growth factor and platelet derived growthfactor in rat fibroblasts.Mol. Cell. Biol. 11:3682.

34. Nobes, C. D., and A. Hall. 1995. Rho, rac and cdc42 regulate the assembly ofmulti-molecular focal complexes associated with actin stress fibers, lamellipodiaand filopodia.Cell 81:53.

35. Aktories, K., S. Braun, S. Rosener, I. Just, and A. Hall. 1989. The rho geneproduct expressed in E. Coli is a substrate for botulinum ADP-ribosyltransferaseC3. Biochem. Biophys. Res. Commun. 158:209.

36. Laudanna, C., J. J. Campbell, and E. C. Butcher. 1996. The role of rho inchemoattractant-activated leukocyte adhesion through integrins.Science 271:981.

37. Huang C., Y. Ni, T. Wang, Y. Gao, G. C. Haudenschild, and X. Zhan. 1997.Down regulation of the filamentous actin cross-linking activity of cortactin byc-src mediated tyrosine phosphorylation.J. Biol. Chem. 272:13911.

38. Takaishi, K., T. Sasaki, T. Kameyama, S. Tsukita, S. Tsukita, and Y. Takai. 1995.Translocation of activated rho from the cytoplasm to membrane ruffling area,cell-cell adhesion sites and cleavage furrows.Oncogene 11:39.

39. Nusrat, A., M. Giry, J. R. Turner, S. P. Colgan, C.A. Parkos, D. Carnes,E. Lemichez, P. Boquet, and J. L. Madara. 1995. Rho protein regulates tightjunctions and peripheral actin organization in polarized epithelia.Proc. Natl.Acad. Sci. USA 92:10629.

40. Schmalzing, G., H. P. Richter, A. Hansen, W. Schwartz, I. Just, and K. Aktories.1995. Involvement of the GTP-binding protein rho in constitutive endocytosis inXenopus laeviaoocytes.J. Cell. Biol. 130:1319.

2973The Journal of Immunology