Cross-Linking of Brain Endothelial Intercellular Adhesion Molecule (ICAM)-1 Induces Association of ICAM-1 With Detergent-Insoluble Cytoskeletal Fraction

Cross-Linking of Brain Endothelial Intercellular AdhesionMolecule (ICAM)-1 Induces Association of ICAM-1 With

Detergent-Insoluble Cytoskeletal FractionClaire Amos, Ignacio A. Romero, Charlotte Schültze, Jonathan Rousell, Jeremy D. Pearson,

John Greenwood, Peter Adamson

Abstract—Intercellular adhesion molecule (ICAM)-1 plays a vital role in the process of leukocyte transmigration throughendothelial cell (EC) barriers and has been shown to mediate signal transduction events in ECs induced either by itscross-linking or by the binding of T lymphocytes. Immunoblotting of ICAM-1 of Triton X-100 detergent fractionsdemonstrated that the majority of ICAM-1 was contained within the detergent-soluble fraction (noncytoskeletalassociated) under basal conditions. After cross-linking of endothelial ICAM-1 with monoclonal antibody or coculturewith T lymphocytes, EC ICAM-1 was observed to partition with a Triton X-100–insoluble (cytoskeletal associated)fraction in a dose- and time-dependent manner. Redistribution of ICAM-1 was specific, inasmuch as no association withthe Triton X-100–insoluble fraction was observed after cross-linking of vascular cell adhesion molecule-1, nor didcross-linking of ICAM-1 result in a redistribution of the platelet and EC adhesion molecule. ICAM-1 association withthe endothelial cytoskeleton after cross-linking was unaffected after treatment of the cells with cytochalasin D,C3-transferase, removal of extracellular calcium ions, or chelation of intracellular calcium ions. These data show thatICAM-1 colocalizes with the endothelial cytoskeleton and associates with a detergent-insoluble fraction aftercross-linking.(Arterioscler Thromb Vasc Biol. 2001;21:810-816.)

Key Words: endotheliumn intercellular adhesion molecule-1n cytoskeleton

L ymphocytes migrating across the central nervous systemvasculature have to overcome impermeable tight junc-

tions that are characteristic of the endothelial cells (ECs)forming the blood–central nervous system barriers. However,it is becoming apparent that in this context, the barrier is notpassive, inasmuch as brain ECs are intimately and activelyinvolved in facilitating lymphocyte diapedesis into the centralnervous system.1 The expression of intercellular adhesionmolecule (ICAM)-1 on ECs is pivotal in supporting lympho-cyte migration across the vascular endothelium.2–5 Neutral-ization of ICAM-1/lymphocyte function–associated antigen(LFA)-1, but not of vascular cell adhesion cell (VCAM)-1/very late antigen-4, leads to a decreased migration of inter-leukin-2–activated antigen-specific T lymphocytes throughunstimulated brain EC monolayers.4,6

Recent evidence has shown that brain EC ICAM-1 iscapable of evoking signal transduction events after ICAM-1cross-linking or coculture with T lymphocytes bearing theICAM-1 counterreceptor. Mimicking lymphocyte attach-ment, through cross-linking EC ICAM-1 molecules in vitro,leads to an induction of actin stress fibers1 and an enhancedtyrosine phosphorylation of cortactin7 in the rat brain EClines GP8/3.9 and RBE4. In addition, focal adhesion kinase,

paxillin, and p130casalso display enhanced tyrosine phosphor-ylation after cross-linking of EC ICAM-1.8 Src/focal adhe-sion kinase–mediated phosphorylation of paxillin has beenpreviously shown to be involved in the formation of focaladhesions and cytoskeletal rearrangement, and phosphoryla-tion of these proteins suggests the involvement of the cy-toskeleton in ICAM-induced signaling.

Further experiments have implicated a cytoskeletal com-ponent in ICAM-I–induced signaling, in that either ICAM-1cross-linking or coculture of ECs with concanavalin A–stim-ulated peripheral lymph node lymphocytes results in in-creased levels of GTP-loaded endothelial Rho proteins. Treat-ment of ECs with C3-transferase, which specifically ADP-ribosylates and inhibits Rho proteins,9 results in an inhibitionof transendothelial lymphocyte migration. This effect isassociated with inhibition of downstream signaling events.1,8

In addition, agents that disrupt the actin cytoskeleton, such ascytochalasin D, inhibit ICAM-1–mediated signaling path-ways and transendothelial lymphocyte migration.1

ICAM-1 is reported to binda-actinin10 and GAPDH11

through its C-terminal domain.a-Actinin is known to belinked to the endothelial actin cytoskeleton, and GAPDH iscapable of bundling microtubules.12 Therefore, intracellular

Received May 31, 2000; revision accepted January 9, 2001.From the Division of Endothelial and Epithelial Cell Biology, Institute of Ophthalmology, University College London, London, UK, and the Centre

for Cardiovascular Biology and Medicine (J.D.P.), Kings College London, London, UK.Correspondence to Dr Peter Adamson, Division of Endothelial and Epithelial Cell Biology, Institute of Ophthalmology, University College London

11-43 Bath Street, London, UK EC1V 9EL. E-mail [email protected]© 2001 American Heart Association, Inc.

Arterioscler Thromb Vasc Biol.is available at http://www.atvbaha.org

810

signal transduction initiated through ICAM-1 in response toleukocyte adhesion may be propagated via the endothelialcytoskeleton. The present study demonstrates that afterICAM-1 ligation, ICAM-1 specifically associates with anendothelial cytoskeletal fraction. We also show that ICAM-1association with the endothelial cytoskeleton after cross-linking is unaffected after inhibition of downstream signalingpathways, suggesting that ICAM-1 redistribution is an earlyevent in the signaling cascade.

MethodsChemicalsRecombinant rat interferon (IFN)-g was obtained from Genzyme.1,2-Bis(2-amino-5-methylphenoxy)ethane-N,N,N9,N9-tetraaceticacid tetraethyl ester (MAPTAM) and BAPTA acetoxymethyl ester(BAPTAM) were obtained from Calbiochem. C3-transferase proteinwas expressed and purified fromEscherichia coli.1

CellsThe rat brain microvascular EC line GP8/3.913 was grown aspreviously described.13 Peripheral lymph node T lymphocytes,which are adhesive but nonmigratory, were isolated and purified aspreviously described.1

AntibodiesAnti-rat ICAM-1 monoclonal antibody (mAb, clone 1A29) andanti-rat major histocompatibility complex (MHC) class I (RT1a)were obtained from Serotec. Anti-rat ICAM-1 mAb (clone 3H8) andanti-rat platelet and EC adhesion molecule (PECAM)-1 mAb (clone3A12) were kind gifts from Dr W. Hickey, Dartmouth MedicalSchool, Hanover, NH. Anti-rat VCAM-1 mAb (5F10) was a giftfrom Dr Roy Lobb, Biogen, Boston, Mass. Anti-caveolin polyclonalantibody was obtained from Transduction Laboratories. FITC andTRITC-conjugated anti-mouse IgG and FITC conjugated anti-rabbitIgG were obtained from Jackson Immunochemicals.

Immunoprecipitation and Western BlottingCell lysates and sucrose density fractions were added to SDS-PAGEsample buffer and resolved on 8% to 12.5% SDS-PAGE undernonreducing conditions as previously described.8

Fluorescence Microscopic Analysis of ICAM-1 andCaveolin-1 ColocalizationIn all cases, GP8/3.9 ECs were grown on 8-well plastic chamberslides and treated with 100 U/mL rat recombinant IFN-g for 48 hoursto upregulate ICAM-1 expression. Cells were fixed, permeabilized,and blocked as previously described.1 Cells were incubated withmouse monoclonal anti–ICAM-1 antibody (IgG, supernatant from3H8 hybridoma) and rabbit polyclonal anti–caveolin-1 (IgG, 1:500)overnight at 4°C. Cells were washed and incubated with secondaryantibody, FITC-conjugated anti-rabbit IgG (1:50), and TRITC-conjugated anti-mouse IgG (1:50). Cells were washed, mounted inVectashield (Vector Laboratories, Burlingame, Calif), and viewed ona scanning confocal microscope. Selected fields were scanned athigh resolution (102431024) with 8 sequential Z sections acquiredper field for TRITC (543 nm) or FITC (488 nm). The intensity ofexpression for each fluorochrome in the field was then plottedagainst the x-y position.

Detergent Fractionation of Cellular ProteinsCells (13106) were washed in RPMI containing 1% FCS andreplaced with ice-cold HBSS containing 5 mmol/L EDTA. Lysateswere centrifuged at 10 000g for 5 minutes, supernatant was removed,and 100mL of ice-cold PHEM buffer (consisting of 60 mmol/LPIPES, pH 6.9, 25 mmol/L HEPES, pH 6.9, 10 mmol/L EGTA,2 mmol/L MgCl2, 1 mmol/L phenylmethylsulfonyl fluoride,20 mmol/L benzamidine, and 5 mmol/L iodoacetamide) containing0.1% Triton X-100 was added to the pellet. Pellets were extracted for2 minutes on ice and recentrifuged at 10 000g for 30 minutes.

Supernatants were removed and added to 33 nonreducing samplebuffer (detergent-soluble fraction). Pellets were washed with PHEMbuffer containing 0.1% Triton X-100 and were added directly to 33nonreducing Laemmli sample buffer (detergent-insoluble fraction).Detergent phases were subsequently vortexed and passed through a21-gauge needle before resolution of proteins on 8% to 10%SDS-PAGE.14,15

Subcellular Fractionation of ECsCells (23107) were washed with ice-cold PBS and scraped into 2 mLof 500 mmol/L sodium carbonate (pH 11). Cells were lysed byhomogenization and sonication (3 times, 20 seconds). Lysates weremixed with an equal volume of 80% sucrose, and 4 mL of thesolution was placed in the bottom of an ultracentrifuge tube. Thesucrose gradient was achieved by the addition of 4 mL of a 30%sucrose solution in sodium carbonate 250 mmol/L (pH 11), followedby 4 mL of 5% sucrose in the same buffer. Sucrose gradients werethen centrifuged at 100 000g for 16 hours at 4°C. Sucrose gradientswere fractionated into 1-mL fractions, and the protein content wasdetermined by the Bradford method (Bio-Rad). Proteins from eachfraction were precipitated with 10% trichloroacetic acid, and sampleswere centrifuged at 10 000g for 15 minutes. Proteins were solubi-lized in sample buffer and resolved on 8% to 12.5% SDS-PAGE.

ResultsICAM-1 Associates With a Triton X-100–InsolubleCytoskeletal Fraction After ICAM-1Cross-Linking or Adhesion of T LymphocytesGP8/3.9 cells were treated with 50 U/mL IFN-g for 48 hoursto upregulate the surface expression of ICAM-1 on ECs.1

Cells were extracted by detergent fractionation to produce aTriton X-100–soluble fraction and a Triton X-100–insolublefraction. These fractions have previously been demonstratedas containing noncytoskeletal and cytoskeletal elements,respectively.14,15Under basal conditions, ICAM-1 was almostexclusively associated with the Triton X-100–soluble frac-tion (Figure 1A). Partitioning of ICAM-1 within the TritonX-100–soluble fraction differed from that in the rat epithelialcell line IE6, in which ICAM-1 was distributed equallybetween phases (data not shown). ICAM-1 was apparent asan immunostaining band at'90 kDa. On ligation of ICAM-1with anti-ICAM-1 and goat anti-mouse IgG (GAM), therewas a significant shift in the solubility of ICAM-1 from theTriton X-100–soluble to the –insoluble fraction (Figure 1A).There was no change in the Triton X-100 solubility ofICAM-1 when cells were incubated with primary or second-ary antibodies alone (Figure 1A). The redistribution ofICAM-1 was dependent on the concentration of primaryanti–ICAM-1 mAb (Figure 1A), with maximal effects occur-ring at '10 mg/mL anti–ICAM-1 mAb. Maximal effectswere also observed to occur at 20mg/mL (data not shown).The effect of cross-linking ICAM-1 with mAb on its TritonX-100 solubility was time dependent but occurred rapidly,being maximal within'5 minutes (Figure 1C). Identicalresults were obtained if the anti–ICAM-1 mAb clone 1A29was substituted with anti–ICAM-1 clone 3H8 (data notshown) or if GAM was substituted with rabbit anti-mouseIgG (data not shown). In addition, identical results were alsoobtained when experiments were conducted in the absence ofIFN-g (data not shown) because of the presence of basalICAM-1 expression. Cross-linking of an alternative immuno-globulin superfamily molecule such as MHC class I (RT1a)did not increase its association with the Triton X-100–

Amos et al Detergent Partitioning of ICAM-1 811

insoluble fraction after cross-linking with anti-RT1a/GAMantibodies (data not shown).

To demonstrate that cross-linking ICAM-1 with anti-ICAM-1/GAM was physiologically significant in mimickingleukocyte adhesion to ECs, the detergent solubility of ECICAM-1 was determined after adhesion of T lymphocytes. Tlymphocytes were incubated with ECs for 45 minutes, fol-lowed by a wash in ice-cold PBS, which is effective inremoving 99.9% of adherent and nonadherent T lymphocytes.Adhesion of T lymphocytes resulted in a significant shift ofEC ICAM-1 to the Triton X-100–insoluble fraction (Figure1B). EC lysates showed no contamination with T-lymphocyteproteins, inasmuch as all lysates were negative for T-cellreceptor protein by Western blot analysis (data not shown).

Cross-Linking of Endothelial VCAM-1 Does NotInduce ICAM-1 RedistributionGP8/3.9 cells display no basal expression of VCAM-1 butcan be induced to express high levels of VCAM-1 aftertreatment with IFN-g.13 IFN-g–treated GP8/3.9 cells, whichwere treated with either 10mg/mL anti–VCAM-1 (clone5F10) mAb alone for 30 minutes or 10mg/mL anti–VCAM-1for 30 minutes followed by GAM (20mg/mL) for 15 minutes,did not cause redistribution of ICAM-1. However, cells thatwere cross-linked with anti–ICAM-1/GAM or cells that weretreated with anti–ICAM-1 and anti–VCAM-1 together andthat were subsequently cross-linked with GAM were able toredistribute ICAM-1 into the Triton X-100–insoluble fraction(Figure 2A). Anti–VCAM-1/GAM cross-linking was able toselectively redistribute VCAM-1 to the Triton X-100–insol-uble fraction (data not shown)

Figure 1. ICAM-1 associates with a Triton X-100 (TX-100)–insol-uble fraction after cross-linking of ICAM-1 and coculture of Tlymphocytes on GP8/3.9 cells. A, GP8/3.9 cells (13106) weretreated with 20 mg/mL anti–ICAM-1 alone (30 minutes) or 10mg/mL GAM alone (15 minutes) or were cross-linked by use ofvarying concentrations of anti–ICAM-1 (1A29) mAb for 30 min-utes, followed by 10 mg/mL GAM for 15 minutes. B, GP8/3.9cells (13106) were untreated or cocultured with 106 T lympho-cytes for 45 minutes. C, GP8/3.9 cells (106) were treated withanti–10 mg/mL ICAM-1 only for 30 minutes, 10 mg/mL GAM for15 minutes, or 10 mg/mL anti–ICAM-1 for 30 minutes, followedby 10 mg/mL GAM for varying times. Cell lysates weredetergent-extracted with TX-100, and TX-100–soluble (S) and–insoluble (I) proteins were resolved by nonreducing SDS-PAGE.The distribution of ICAM-1 between detergent phases was visu-alized after immunoblotting. Position of ICAM-1 is shown byarrow. Lower molecular weight bands represent anti–ICAM-1IgGH bands. Blots are representative of 3 independentexperiments.

Figure 2. A, Cross-linking of VCAM-1 on GP8/3.9 cells does notcause association of ICAM-1 with the TX-100–insoluble fraction.GP8/3.9 cells (106) were treated with 10 mg/mL anti–ICAM-1(1A29) alone for 30 minutes, 10 mg/mL anti–VCAM-1 (5F10)alone for 30 minutes, or 10 mg/mL GAM alone for 15 minutes.Alternatively, cells were cross-linked by using 10 mg/mL anti–ICAM-1 (1A29) for 30 minutes, followed by 10 mg/mL GAM for15 minutes, 10 mg/mL anti–VCAM-1 (5F10) for 30 minutes, fol-lowed by 10 mg/mL GAM for 15 minutes, or 10 mg/mL anti–ICAM-1 plus 10 mg/mL anti–VCAM-1 for 30 minutes, followedby 10 mg/mL GAM for 15 minutes. Cells were detergent-extracted with TX-100, and TX-100–soluble and –insoluble pro-teins were resolved by nonreducing SDS-PAGE. The distributionof ICAM-1 between detergent phases was visualized after im-munoblotting. Position of ICAM-1 is shown by arrow. Lowermolecular weight bands represent anti–ICAM-1 IgGH bands. B,Cross-linking of ICAM-1 on GP8/3.9 cells does not cause asso-ciation of PECAM-1 with the TX-100–insoluble fraction. GP8/3.9cells (106) were treated with 10 mg/mL anti–ICAM-1 (1A29) alonefor 30 minutes, 10 mg/mL isotype-matched control mAb alonefor 30 minutes, 10 mg/mL anti–PECAM-1 (3A12) alone for 30minutes, or 10 mg/mL GAM alone for 15 minutes. Alternatively,cells were cross-linked by using 10 mg/mL anti–ICAM-1 (1A29)for 30 minutes, followed by 10 mg/mL GAM for 15 minutes, 10mg/mL isotype control mAb for 30 minutes, followed by 10mg/mL GAM for 15 minutes, or 10 mg/mL anti–PECAM-1 for 30minutes, followed by 10 mg/mL GAM for 15 minutes. Cells weredetergent-extracted with TX-100, and TX-100–soluble and –in-soluble proteins were resolved by nonreducing SDS-PAGE. Thedistribution of PECAM-1 between detergent phases was visual-ized after immunoblotting.

812 Arterioscler Thromb Vasc Biol. May 2001

Cross-Linking of EC ICAM-1 Does Not InduceRedistribution of PECAM-1GP8/3.9 cells constitutively express PECAM-1, which onimmunoblotting with anti-PECAM mAb (3A12) detected 1isoform with an apparent molecular mass of'120 kDa.Analysis of unstimulated cells showed that PECAM-1 waspredominantly associated with a Triton X-100–soluble frac-tion. Treatment of cells with 10mg/mL anti–ICAM-1 alone,10 mg/mL isotype-matched control IgG alone, or 10mg/mLanti–PECAM-1 (3A12) alone showed no ability to shiftPECAM-1 into a Triton X-100–insoluble fraction (Figure2B). When cells were treated with anti–ICAM-1, isotype-matched control, or anti–PECAM-1 and were subsequentlycross-linked with GAM, only the anti–PECAM-1/GAM com-bination was effective in changing the Triton X-100 solubilityof PECAM-1 (Figure 2B). Taken together, the inability ofVCAM-1/GAM to alter the detergent solubility of ICAM-1and the inability of ICAM-1/GAM to alter the detergentsolubility of PECAM-1 suggests that these cross-linkingstrategies are specific and do not result in large nonspecificaggregations of integral membrane proteins.

Redistribution of ICAM-1 Is Not Dependent onActin Polymerization or Activation ofRho ProteinsTreatment of GP8/3.9 cells with 2mmol/L cytochalasin D for30 minutes, which has previously been shown to inhibit actinpolymerization and inhibit ICAM-1–mediated signalingevents, including transendothelial lymphocyte migration,1 didnot inhibit the ability of ICAM-1/GAM to shift the TritonX-100 solubility of ICAM-1 (Figure 3B). Similarly, treatmentof cells with 50mg/mL C3-transferase for 8 hours (which has

previously been shown to be effective in inhibiting cellularRho proteins in these cells1) was also incapable of affectingthe ICAM-1/GAM–induced redistribution of ICAM-1 to theTriton X-100–insoluble fraction (Figure 3A).

Redistribution of ICAM-1 to Cytoskeletal FractionIs Not Inhibited After Removal of ExtracellularCalcium or Chelation of Cytosolic CalciumRecent data from this laboratory have suggested that intra-cellular calcium fluxes resulting from ICAM-1 cross-linkingare important for transendothelial lymphocyte migration andthat the treatment of cells with intracellular calcium chelatorsis effective in inhibiting transendothelial lymphocyte migra-tion.16 The presence of 2 mmol/L EGTA (which will chelatethe available calcium ions in the extracellular medium) hadno effect on the altered Triton X-100 solubility of ICAM-1caused by cross-linking with ICAM-1/RAM (Figure 4). In asimilar manner, cells treated with 20mmol/L BAPTAM or 20mmol/L MAPTAM for 30 minutes, which inhibits transendo-thelial lymphocyte migration,16 had no effect on Triton X-100solubility of ICAM-1 either basally or when cross-linked(Figure 4).

ICAM-1 and Caveolin-1 Cofractionate on SucroseDensity Gradients but Do Not Colocalize inGP8/3.9 CellsAlthough Triton X-100–soluble and –insoluble fractionshave previously been interpreted as noncytoskeletal and

Figure 3. Association of ICAM-1 with the TX-100–insoluble frac-tion is not dependent on polymerized actin or functional Rhoproteins. A, Cells (106) were untreated or treated with 10 mg/mLC3-transferase (C3) for 16 hours. B, Cells (106) were eitheruntreated or treated with 2 mg/mL cytochalasin D (Cyto D) for 1hour. Cells were subsequently cross-linked by using 10 mg/mLanti–ICAM-1 (1A29) mAb and 10 mg/mL GAM for 30 minutes.Cell lysates were detergent-extracted with TX-100, and TX-100–soluble (S) and –insoluble (I) proteins were resolved by nonre-ducing SDS-PAGE. The distribution of ICAM-1 between deter-gent phases was visualized after immunoblotting. Position ofICAM-1 is shown by arrow.

Figure 4. Association of ICAM-1 with the TX-100–insoluble frac-tion is unaffected after clamping of intracellular calcium concen-tration or removal of extracellular calcium. GP8/3.9 cells (106)were cultured in normal media containing normal 2 mmol/LEDTA or containing 20 mmol/L MAPTAM or 20 mmol/L BAPTAMfor 30 minutes. Cell lysates were subsequently cross-linked with10 mg/mL anti–ICAM-1 (1A29) mAb and 10 mg/mL GAM for 30minutes. Cells were detergent-extracted with TX-100, andTX-100–soluble (S) and –insoluble (I) proteins were resolved bynonreducing SDS-PAGE. The distribution of ICAM-1 betweendetergent phases was visualized after immunoblotting. Positionof ICAM-1 is shown by arrow.


cytoskeletal cellular fractions, respectively,14,15 the insolublefraction also contains distinct membrane domains formed bythe clustering of glycosphingolipids and cholesterol, alongwith glycosyl-phosphatidylinositol–anchored proteins andacylated proteins. Caveolae are low density membrane mi-crodomains that have been implicated in initiating the signal-ing events induced by growth factors, such as epidermalgrowth factor and endothelin-1, and receptors for theseligands are enriched within caveolae.17 As expected,caveolin-1, a characteristic caveolar marker protein in thebrain,18 appeared to be retained in the Triton X-100–insolu-ble fraction under basal conditions and after ICAM-1 cross-linking (Figure 5D). Under basal (Figure 5A) and ICAM-1cross-linking conditions (data not shown), ICAM-1 andcaveolin-1 showed no significant colocalization in GP8/3.9cells, although data from sucrose density gradient fraction-ation of GP8/3.9 cell lysates showed that caveolin-1 andICAM-1 cofractionated (Figure 5C). Cofractionation was alsonot affected after ICAM-1 cross-linking. Western blottinganalysis showed that ICAM-1 and caveolin-1 were containedwithin fractions 5 to 8, which correspond to 30% sucrose(Figure 5D).

DiscussionWe have previously demonstrated that ICAM-1 is pivotal incontrolling the transendothelial migration of T lymphocytes4

and have shown that intracellular signals initiated throughlymphocyte binding to ECs are mediated through the LFA-1/ICAM-1 interaction.8,1 Indeed, in a large number of exper-iments, ligation of ICAM-1 using cross-linking antibodieshave been confirmed by T-lymphocyte adhesion.7,1

T-lymphocyte adhesion–mediated events have also beendemonstrated to be LFA-1 dependent (data not shown). Suchobservations suggest that ICAM-1 cross-linking experimentshave a physiological relevance. Almost all the ICAM-1–mediated signals so far identified,8 as well as transendotheliallymphocyte migration,1 are inhibited after treatment of cellswith either cytochalasin D or C3-transferase. These agents areeffective in inhibiting transendothelial migration withoutaffecting leukocyte-endothelium adhesion.1 Therefore, it islikely that the endothelial actin cytoskeleton and endothelialRho proteins are essential in transducing intracellular signalsin ECs initiated through leukocyte adhesion to ICAM-1 andare necessary for the efficient migration of lymphocytesthrough the EC barrier.

Figure 5. ICAM-1 and caveolin-1 do notcolocalize in GP8/3.9 cells. A and B,Immunocytochemical detection ofICAM-1 and caveolin-1 (FITC). Imageshows the localization of ICAM-1 (red)and caveolin-1 (green) plotted at pixellevel along an arbitrary line drawnthrough the image. Scale bar 10 mm. C,ICAM-1 and caveolin-1 colocalize andcosediment in GP8/3.9 ECs. Cells(23107) were lysed and centrifuged onsucrose at 100 000g for 16 hours at 4°C.Sucrose gradients were then separatedinto 1-mL fractions. Equal amounts ofprotein for each fraction were precipi-tated with 10% trichloroacetic acid andresolved on 8% to 12.5% SDS-PAGE.The distribution of ICAM-1 and caveolin-1was visualized after immunoblotting ofsucrose density gradient fractions. D,Association of caveolin-1 with theTX-100–insoluble fraction. Cells wereeither untreated or cross-linked by using10 mg/mL anti–ICAM-1 (1A29) mAb and10 mg/mL GAM for 15 minutes. Celllysates were detergent-extracted withTX-100, and TX-100–soluble (S) and –in-soluble (I) proteins were resolved by non-reducing SDS-PAGE. The distribution ofcaveolin-1 between detergent phaseswas visualized after immunoblotting. XLindicates ICAM-1 cross-linked cells.


After endothelial ICAM-1 cross-linking strategies,ICAM-1 was found to specifically associate with a TritonX-100–insoluble fraction, whereas in non–cross-linked cells,ICAM-1 was primarily associated with the Triton X-100–soluble fraction. These detergent fractions have previouslybeen shown to contain noncytoskeletal (soluble) and cy-toskeletal (insoluble) cellular fractions.14,15The association ofICAM-1 with the Triton X-100 fraction was time dependentand dependent on the concentration of primary and secondaryantibodies. The association of ICAM-1 with the TritonX-100–insoluble fraction occurred only when ICAM-1 wascross-linked and not when either isotype-matched control orantibodies directed against other immunoglobulin superfam-ily molecules were used (eg, anti–VCAM-1, Figure 2A).Studies in which anti–PECAM-1 mAb was used to cross-linkPECAM resulted in a similar shift of PECAM-1, but noredistribution of PECAM-1 was observed when anti–ICAM-1was used as the primary antibody (Figure 2B). Thus, shiftingof ICAM-1 or PECAM-1 to the detergent-insoluble fractionis a specific effect of cross-linking and not the result ofnonspecific membrane aggregates formed through antibodycross-linking. In support of this, cross-linking MHC class I(RT1a) with a specific mAb did not result in the redistributionof this molecule.

The observation that the association of ICAM-1 with theTriton X-100–insoluble fraction still occurred in the presenceof cytochalasin D or C3-transferase demonstrates that thepartitioning of ICAM-1 is not dependent on actin polymer-ization or the activity of endothelial Rho proteins. Althoughboth of these treatments result in a dramatic inhibition ofICAM-1 signaling, how ICAM-1 is retained with the actincytoskeleton, particularly after treatment of the cells withcytochalasin D, is presently unknown. However, experimentshave shown that the cytoskeletal linker protein ezrin colocal-izes with F-actin and ICAM-1 in control and cytochalasinD–treated cells (data not shown). The data also stronglysuggest that the association of ICAM-1 with the TritonX-100–insoluble fraction occurs before the induction of actinstress fibers that follows ICAM-1 cross-linking in thesecells.1 The inability of C3-transferase to prevent cytoskeletalassociation of ICAM-1 also implies that this association is anupstream event, occurring before Rho protein activation. It isalso interesting to note that C3-transferase does not inhibitICAM-1–induced tyrosine phosphorylation of cortactin,whereas this is effectively inhibited after pretreatment of thecells with cytochalasin.1 This supports the idea that alter-ations in the endothelial cytoskeleton after ICAM-1 cross-linking is an early event and upstream from Rho activation.

It has also been demonstrated that intracellular calcium-chelating agents are effective in inhibiting ICAM-1–generatedsignal transduction events in ECs with concomitant abrogationof transendothelial lymphocyte migration.16 In addition, thesedata show that the activation of endothelial Rho proteins andcalcium mobilization within ECs after ICAM-1 cross-linking aredistinctly separate pathways that are both induced after ICAM-1ligation. Induction of src activation, by ICAM-1 ligation, is acalcium-dependent event, which is inhibited by cytochalasin D,demonstrating that cytoskeletal integrity is essential for calcium-mediated signals after ICAM-1 cross-linking.16 The observationthat intracellular calcium-chelating agents or removal of extra-cellular calcium ions is ineffective in preventing the cross-

linking–induced association of ICAM-1 with the Triton X-100–insoluble fraction also indicates that this association is an earlyevent in the mechanism of ICAM-1–induced signal transduc-tion, with cortactin phosphorylation and Rho protein and intra-cellular calcium mobilization occurring consequentially.

Caveolin-1 is retained in the detergent-soluble fractionunder basal and cross-linking conditions, demonstrating thatcaveolae do not associate with the cytoskeleton after ICAM-1cross-linking. Confocal microscopy confirmed that despitethe cosedimentation of ICAM-1 and caveolin-1 in the mem-brane fraction of the cells, there is no obvious association ofICAM-1 and caveolin-1. In contrast, confocal images havepreviously demonstrated that under control and ICAM-1–cross-linked conditions, there is significant colocalization ofICAM-1 and F-actin.1

ICAM-1 has previously been reported to interact with thecytoskeletal linker protein ezrin, and this interaction is enhancedafter exposure to phosphatidylinositol 4,5-bisphosphate.19 Thesestudies suggest that ICAM-1 signal transduction events may bepropagated via ezrin to the actin cytoskeleton. In addition,because ICAM-1 signaling is highly dependent on functionalendothelial Rho proteins, it is interesting to note that ezrin andother ERM proteins are effectors of Rho signaling pathways,20

controlling stress fiber and focal adhesions. Moreover, Rhoproteins have been shown to regulate phosphatidylinositol4-phosphate 5-kinase,21 which may therefore affect the interac-tion of ICAM-1 and ezrin, thereby altering ICAM-1–mediatedsignaling responses. Our results suggest that the endothelialcytoskeleton is an important initial component in propagatingsignals within ECs after the adhesion of leukocytes and istherefore consistent with the translocation of ICAM-1 to adetergent-insoluble fraction on ICAM-1 cross-linking.

AcknowledgmentsThis study was funded by the Wellcome Trust, British HeartFoundation, and the Biotechnology and Biological Sciences Re-search Council. We are grateful to Dr W. Hickey (DartmouthMedical School, Hanover, NH) for the gift of antibodies.

References1. Adamson P, Etienne S, Couraud P-O, Calder V, Greenwood J. Lym-

phocyte migration through brain endothelial cell monolayers involvessignalling through endothelial ICAM-1 via a Rho-dependent pathway.J Immunol. 1999;162:2964–2973.

2. Oppenheimer-Marks N, Davis LS, Lipsky PE. Human T-lymphocyteadhesion to endothelial cells and transendothelial cell migration:alteration of receptor use relates to the activation status of both the T-celland the endothelial cell.J. Immunol. 1990;145:140–148.

3. Oppenheimer-Marks N, Davis LS, Bogue DT, Ramberg J, Lipsky PE.Differential utilization of ICAM-1 and VCAM-1 during the adhesion andtransendothelial migration of human T lymphocytes.J Immunol. 1991;147:2913–2921.

4. Greenwood J, Wang Y, Calder V. Lymphocyte adhesion and transendo-thelial migration in the CNS: the role of LFA-1, ICAM-1, VLA-4 andVCAM-1. Immunology. 1995;86:408–415.

5. Pryce G, Male DK, Campbell I, Greenwood. J. Factors controlling T-cellmigration across rat cerebral endothelium in vitro.J Neuroimmunol.1997;75:84–94.

6. Engelhardt B, Laschinger M, Schulz M, Samulowitz U, Vestweber D,Hoch G. The development of experimental autoimmune encephalomy-elitis in the mouse requires alpha4-integrin but not alpha4beta7-integrin.J Clin Invest. 1998;102: 2096–2105.

7. Durieu-Trautmann O, Chaverot N, Cazaubon S, Strosberg AD, CouraudPO. Intracellular adhesion molecule 1 activation induces tyrosine phos-phorylation of the cytoskeleton-associated protein cortactin in brainmicrovessel endothelial cells.J Biol Chem. 1994;269:12536–12540.


8. Etienne S, Adamson P, Greenwood, J Strosberg AD, Cazaubon S,Couraud P-O. Rho-dependent signaling pathways coupled to ICAM-1 inmicrovascular brain endothelial cells.J Immunol. 1998;161:5755–5761.

9. Aktories K, Braun S, Rosener S, Just I, Hall A. The rho gene productexpressed in E. coli is a substrate for botulinum ADP-ribosyltransferaseC3. Biochem Biophys Res Commun. 1989;158:209–213.

10. Carpen O, Pallai P, Staunton DE, Springer TA. Association of inter-cellular adhesion molecule-1 (ICAM-1) with the actin cytoskeleton anda-actinin.J Cell Biol. 1992;118:1223–1234.

11. Federici C, Camoin L, Hattab M, Strosberg AD, Couraud P-O. Asso-ciation of the cytoplasmic domain of intercellular adhesion molecule-1with glyceraldehyde-3-phosphate dehydrogenase andb-tubulin. EurJ Biochem. 1996;238:173–180.

12. Huitorel P, Pantaloni D. Bundling of microtubules by glyceraldehyde-3-phosphate dehydrogenase and its modulation by ATP.Eur J Biochem.1985;150:265–269.

13. Greenwood J, Pryce G, Devine L, Male DK, dos Santos WLC, CalderVL, Adamson P. SV40 large T immortalised cell lines of the ratblood-brain and blood-retinal barriers retain their phenotypic and immu-nological characteristics. J Neuroimmunol. 1996;71:51–63.

14. Schliwa M. Action of cytochalasin D on cytoskeletal networks.J CellBiol. 1982;92:79–91.

15. Yoshida M, Westlin WF, Wang N, Ingber DE, Rosenweig A, Resnick N,Gimbrone MA. Leukocyte adhesion to vascular endothelium induces

E-selectin linkage to the actin cytoskeleton.J Cell Biol. 1996;133:445–455.

16. Etienne-Manneville S, Manneville JB, Adamson P, Wilbourn B,Greenwood J, Couraud PO. ICAM-1-coupled cytoskeletal rear-rangements and transendothelial lymphocyte migration across brain en-dothelium involve intracellular calcium signaling in brain endothelial celllines.J Immunol. 2000;165:3375–3383.

17. Teixeira A, Chaverot N, Schroder C, Strosber AD, Couraud P-O,Cazaubon S. Requirement of caveolae microdomains in extracellularsignal regulated kinase and focal adhesion kinase activation induced byendothelin-1 in primary astrocytes.J Neurochem. 1999;72:120–128.

18. Ikezu T, Ueda H, Trapp BD, Nishiyama K, Sha JJ, Volonte D, Galbiati F,Byrd AL, Bassell G, Serizawa H, et al. Affinity purification and charac-terisation of caveolins from the brain: differential expression ofcaveolin-1, 2, and 3 in brain endothelial and astroglial cell types.BrainRes. 1998;804:177–192.

19. Heiska L, Alfthan K, Gronholm M, Vilja P, Vaheri A, Carpen O. Asso-ciation of ezrin with intercellular adhesion molecule-1 and -2 (ICAM-1and ICAM-2).J Biol Chem. 1998;273:21893–21900.

20. Mackay DJG, Esch F, Furthmayr H, Hall A. Rho- and rac-dependentassembly of focal adhesion complexes and actin filaments: an essentialrole for ezrin/radixin/moesin protein.J Cell Biol. 1997;138:927–938.

21. Chong LD, Traynor-Kaplan A, Bokoch GM, Schwartz MA. The smallGTP-binding protein Rho regulates a phosphatidylinositol 4-phosphate5-kinase in mammalian cells.Cell. 1994;79:507–513.


Documents

Cross-Linking of Brain Endothelial Intercellular Adhesion Molecule (ICAM)-1 Induces Association of ICAM-1 With Detergent-Insoluble Cytoskeletal Fraction