Idea Transcript
Clin Exp Immunol 1993; 94:478-485
Intercellular adhesion molecule-l (ICAM-l) in Graves' disease: contrast between in vivo and in vitro results A. CIAMPOLILLO, G. NAPOLITANO, R- MIRAK.IAN, A, MIYASAKI, R. GIORGINO* & G. F. BOTTAZZO Department of Immunology. The London Hospital Medical College. London. UK. and "Istituto di Clinica Mediea. Etidocrinologia e Malattie Meiaboliehe. Univer.sita degti Studi. Bari. Italy
(Accepted for publication lO August 1993)
SUMMARY We have reassessed the possible role ofthe adhesion molecule ICAM-l in the pathogenesis t>f thyroid autoimmunity. In order to do that, we have investigated its expression in eight Graves' thyroids both in vivo {i,e. on cryostal seetions and on cell suspensions), and in vitro (i.e. on cells cultured in monolayers for 3 days), and theresultswerecompared with those obtained with similar preparations from lour normal glands. On cryostat seetions. the expression of ICAM-l. and for comparison that of HLA Class I and Class II molecules, was studied by immunolliioreseence (IFL), but the former were also assessed by a distinct immunohistochemical technique. ICAM-I was not detected in tbyrocyles in vivo of both nonnal and Graves* glands, but solely in endothelial ceils and antigenpresenting cells (APC), This selective reaction was confirmed by a four-layer leehnique using specifie markers which idetitify endotiielial cells and thyrocytes, HLA Class II molecules were conlirnied to be inappropriately expressed in thyroeytes of Graves' glands, but there was no co-expression of these prodttcts and ICAM-I in the same cciis. In contrast. ICAM-i appeared de now in a proportion of Graves' and normal tiiyroeytes soon after the attaehment and spreading of these cells in monoiaycr eultures (36 48 h). Graves' thyrocytes showed a quantitivcly higher degree of expression compared with that detected on normal thyroid cells (40 70".n versus l2-2()"''i)- Under these experimental eonditions, the four-layer staining with thyroid microsomai antibodies confirmed that thyroeyles were indeed the positive eelis which expressed ICAM-l. Blocking experiments with cultured thyrocytes from two Graves" glands and MoAbs to tumour necrosis factor-alpha (TNF-«) and interferon-gamma (IFN-/) did not prevent the nceurrencc of iCAM-l expression, Asa resuit ofour study, we fiiiied to demonstrate that Graves' thyrocytes express ICAM-I in vivo. Tiie unexpected ease of inducing ICAM-i on thyroid eells under eertain in vitro conditions remains intriguing. The phenomenon could be the simple consequence of a meebanieai effect rather than exerted by specific bioiogicai processes. Further investigations are. therefore, needed to establish whetiier lCAM-1 is really involved in lhe patbogenesis of Graves" disease. Keywords intercciiuiar adhesion molecule-1 MHC molecules
Graves" disease
adhesion molecules
activated lymphocytes and on APC in general \t-% In addition, this molecule is easily inducibie hi vitro. In as iittie as 4 h, many eell types, including epithelial cells, express ICAM-I in response to eell activation or following lymphokine stimulation [10,1 i]. In particular, tumour necrosis factor-alpiia (TNF-x) anti intcrfcron-gamma (IFN--/) iiave been found, aione or in combination, lo promptly induce ICAM-I in vitro on human endocrine eells, sueb as thyrocytes [12-15] and pancreatic islet cells [ 16J 7]. However, it is stiii not clear whether the expression of this accessory molecule on endocrine cells occurs spontaneously in vivo in glands of patients affected by autoimmunity. This uncertainty is emphasized by the eontrasting results recently presented in tlie thyroid ficid, where some groups have shown that ICAM-i was expressed on thyrocytes of glunds of
fNTRODUCTION Adhesion molecules play a central role in cell-to-celi communications for the ultimate induction of an effective immune response [i 5], In partictilar. ihc interaction between ICAM-I on antigen-presenting cells (APC) and its counter receptor on T lymphocytes triggers a cascade of reactions leading to effector eell induetion and proliferation. This ultimately results in antibody production and direct T cell-mediated actions, lCAM-1, a member of the IgG superfamily, is weaicly expressed on resting lymphocytes but strongly expressed on Correspondence: Professor G, F, Boitaz?o. Department of Immunology, The London Hospital Medical College, 56 76 Ashfieki Street, London El 2AD. UK,
478
479
ICAM-} in GD patients with Graves'disease (GD) [12,1,1], but others could nol subsequently eontirm these initial observations (lX). CIcariy. if iCAM-1 were expressed in i-ivo on epithelial cells at sites of autoimmune inflammation, interactions between these celis and those of the immune system would be greatly facilitated. This proeess would be particularly relevant in an autoimmune context, where thyroid follicular cells also eo-express HLA Class II molecules. Thus, mechanisms underlying autoantigen presentation to activated Tcells by Class H*/ICAM-l^ thyroeyles would be greatly enhanced. We have re-addressed this important issue in order to clarify whether ICAM-I could indeed play a relevant role in the pathogenesis of thyroid autoimmunity. Thus, we investigated its possible spontaneous expression on thyroeytes from Graves" patients and from eontrol individuals botb in vivo. i.e. on cryostat seetions and on cell suspension, and in vitro on ceii nionolayers soon after tbe attachment of tbe cells. ."^6-48 h after plating the cells. MATERIALS AND M E T H O D S Source of thyroid glands Thyroids from two groups ofpatients were studied. Eight glands wete obtained at partial thyroideetomy from patients with GD (kindly provided by Mr R. C. G. Russeii, tiie Middiesex Hospital. London. UK), and four from patients undergoing surgery for carcinoma of the larynx (kindly provided by Proicssor D, F. N. Harrison, The Royal ENT Hospital, London, UK). The latter glands were used as a source of control thyrocytes. The diagnosis of GD was based on established clinical eriteria and eonventional laboratory parameters. Ali Graves" patients had been treated wiih a course of carbimazole (20-30 mg/day) fora period of at least 6 months, and operated upon beeause they had faiied to respond to the prolonged amithyroid drug treatment. Thyroid antibodies were detected by an established haemagglulinaiion technique (Fujirebo Inc., Japan) and all Graves' sera tested were positive for thyroid microsomai/ thyroperoxidase (Mc-TPO) antibodies (titres ranging from goto 320-) and/or tbyrogiobulin (Tg) antibodies (titres ranging from 80 to i60). Thyroid /i.\.ii/e processing
Each thyroid specimen obtained at operation was divided ibr two types of processing: (i)a portion of the tissue was cut in 1-2 cm blocks which were then snap fro/en in isopentane cooled in dry icc'acetone. Cryostat sections {?• /jm) were cut from each block and these were stained as described below; (ii) fresh tissue was minced and digested with eoliagenase (Worthington, Type IV, USA) (4 mg/ml) for ^ h at 37 C, Ceiis were then filtered through a 200-nm nylon mesh and washed thoroughly several times, as previously described [i9]- Tiie cell yieid so obtained was approximately 10 20 x 10' cells,ml witb a viability ranging between SO'io and 90"/ii, as assessed by staining wilh elhidium bromide/aeridine orange. Immunofiuorescence studies Conventional indirect immunonuorescenee (IFL) was carried out with the MoAbs listed in Table i on the following thyroid cell preparations; (i) cryostat sections; (ii) viable celts in suspension soon after digestion; (iii) monoiaycr eultures after 36-48 h. Aii preparations were read blindly by two independent
Table I. Monoelonai antibodies used in this study Speeilieity
Clone
Isotype
HLA Class I HLA Class II
W6/32 Mid 3
Pan T
UCHTI
IgG2A Dakopatts IgGl Prof. P, Lydyard (Middlesex Hospital, London. UK) IgG I P, Beverley (ICRF, London. UK) IgGl Imnuinotceh IgGl Prof, P, Lydyard
ICAM-I 84HIt) Mouse thyroglobulin P1IH4
Source
observers using a Zciss III photomicroscope equipped with epiillumination. IFL on cryostat .wction.s\ All the preparations were fixed in acetone for 3 min and then incubated for 30 min with MoAbs to HLA Class I and Class U products, ICAM-l and total T lymphocytes. Asa negative control, we used a MoAb to mouse Tg (PI I), a reagent wiiich stains neither tiie surface nor the eytopiasm of human tiiyroid celis (20]. Tiie sections were subsequentiy washed for 15 min in PBS and incubated for 25 min with aHtnily-purified F'lTC-conjugatcd rabbit anti-mouse IgG (Dako Lid, UK). Indirect IFL was performed on at least three sections from two different levels of the same frozen block and on at least two dill'ercnt blocks from each tissue specimen investigated. IFL on thyroid cell su.\pensions. Four dilTerent Graves' tliyroids and two normai glands were used In these experiments. Viable thyroid eeiis (U)'') in suspension were incubated and stained by IFL with the various MoAbs soon after digestion. Eiichincubation was stopped with tlieadditinnon '\, BSS-BSA, IFL on thyroid cell monolayers.
Viable ceiis (10^), from all
eight Graves' digested thyroid samples and the four normal glands proeessed in the same way, were plated onto 32-nim glass coverslips. Cells were cultured in tissue culture medium (TCM) composed of RPMI enriched with transferrin (25 /ig/ml), glutamine (2 niM), insulin (0-25 U/ml) and ditTerent concentrations of fetal caif serum (FCS; 1%, 10%, 15%, respectively, and in one experiment by using FCS-free medium)- Ceiis in culture were allowed to attach to the covcrslip and spread, Tiiis step occurred within 24-36 h with i 0"/.> FCS eoneentrations, where;is il took approximately 48 h with lower FCS and 18 h with \5% FCS. Staining with different MoAbs was performed as soon as the thyroid cells were attaehed to the coverslips (24-48 h from the cell seeding). Blocking e.xperiments. These were performed in an attempt to block the potentially inducing effect of cytokines, produced by contaminating lymphocytes, on ICAM-l expression on thyrocytes in culture. Cells (10^) from two Graves" thyroids were plated onto glass coverslips and the following neutralizing antibodies were added to the TCM, either separately or together, in different combinations of concentrations: MoAb lo TNF-a (IO-l()t)O ncutrah/ing U/ml) (Bcihringer, Mannheim, Germany) and MoAb (o JFN-y (100 Wm iictKraJii-iiig (.'ml) (Gen/yme, Boston, MA). The experiments were performed in duplicate. Ai"ter 48 h incubation, the I PL slaining procedure was carried oul with the same MoAbio ICAM-l used in aii thebasai experiments. Immunolmtochemieal studies Immunohistochemical iocalization of positive cells in thegiunds was carried oul on cryostat sections of four Graves" and two
Fig, I, Four layer immunofluoreseenee (IFL) staining for (a) ICAM-l and (b)F-VIII on aeryostat section ofGraves'thyroid gland. A positive staining is observed on endothelial cells (EC) in (a). These eells are also eoneomitantly reaeting for F-VIII. as shown in (b), thus eonfirming that they are EC, Thyroid epithelial eells are elearly negalive.
481
ICAM-l in GD normal thyroids, by ineubating the preparations with MoAbs to ICAM-l and/or PI 1. This step was followed by three alternate cycles of anti-mouse IgG and alkaline phosphatase,anti-alkaline phosphatase complex. The staining was llnaliy accomplished with the addition of the appropriate substrate, as previously described 112], ICAM-I staining was identified by the change in substrate eolour.
Four-layer IFL staining: ICAM-UMc-TPO antibodies: ICAMijFactor VW To assess whether ICAM-I, in addition to the reaetion apparentiy seen on endotheiiai and APC, was also expressed on thyroid epitiieliul cells, a four-iaycr IFL staining procedure was carried out in three dilTerenl Graves' thyroid preparations from four glands: (i) cryosiat seclions; (ii) cell suspensions; and (iii) cultured eell monolayers. In the first set of experiments, the three different preparations were stained with Mo.Ab Io ICAM-l, followed by FITClabclicd affinity-purified rabbit anti-mouse IgG. As third iayer, a patient's serum eontaining a higb titre (320-) of Mc-TPO antibodies was then applied for 30 min at dilutions of 1 :5, i : 10, i:2t). followed by a 30-min incubation with tetramethylrodamine (TRITC)-conjugated rabbit anli-human IgG (Dako), Facli incubation step was followed by a washing with \%> BSS/
BSA. As control, PI I MoAb was used in place of ICAM-l MoAb nnd nonnal human sertim (NHS) (1:40 dilution) rcpiaced the high titre Mc-TPO-positive scrum. In the second set of experiments, the first 30 min addition of MoAb to ICAM-i to the same tliree preparations was foilowed by incubation with FITC-eonjugated goat anli-mouse IgG (Southern Biotechnology. USA), in this case. Mie third layer consisted of rabbit anti-human IgG to Factor Vlll (Dako) followed by a 30-min application of TRlTC-conjugaled goat anti-rabbit immunoglobulin (Nordic Labs, The Netherlands), As control. PI I was again used instead of MoAb to ICAM-I. and in this case normal rabbit serum (NRS) (dilution 1:20) was applied instead of rabbit anti-Factor VUI. Ali these preparations were also read blindly on a Zeiss III photomicroscope equipped with filters for two-colour fluorescence. In the four-layer IFL experiments with celi suspensions and cell monoiayers. the linal assesstnent of ICAM-I molecule cxpressionon thyrocytes was completed by counting at least 200 thyrocytes under alternate green and red filters. RESULTS Ti.'i.iue sections IIL.i Cla.ss I molecule e.xpression. As c-xpcctetl, these products were expressed both in thyrocyiesand in the iymphomono-
I Fig. 2. Alkaline phosphatase teehnique staining for ICAM-1 on a eryostat seetion of Graves" thyroid gianJ, The positive staining of the endothelial eells (EC) eontrasts with the negalive reaction of thyroeyics.
482
A. Ciampolillo et al.
Fig. 3. Four layer immunolluoreseenee (IFL) staining for (a) lC'AM-1 and (b) thyroid microsomal/thyroperoxidase (Me/TPO) antibody-positive patient"s serum on Graves" thyroid eells in monoiaycr eulture. A positive reaetion on the surfaee of Ihyroid cells is observed in (a). These eells are concomitantly reacting for Me/TPO antibody, as shown in (b), thus continuing ihat they are thyrocytes.
ICAM-1 in GD nuclear cetls, and the staining pattern was similar in autoimmutie and non-autoimmune glands. In general, the intensity of the reaction was increased in Graves' glands, atthough this was not easy to ascertain because of the quite strong positivity atready observed in thyrocytes from normal glands. Ht.A Class // niolecide e.xpression. De nova expression of Class II molcculeexpression was detected in Graves'thyrocytes, inctuding those surrounded by areas of lymphoeytie infittration. Occasionally, Class Il-positive thyrocytes were observed in patchy distribution in thyrocytes of control glands. ICAM-l expression. This expression was detected solely in endotheiiai celts (FC) and APC-like cells, both in Graves" thyroids and in normal glands. This was confirmed by the two distinct four-layer IFL staining procedures. ICAM-I was constantly negative in the thyroeytcs of Graves" thyroids as well as in normat glands in att the tissue blocks investigated at different levels (Fig. Ia,b and Fig. 2). Particular attention was paid to those ihyrocytes surrounded by areas of active infiltration but, yet again, no specific staining could be observed for ICAM-1 products on these ceils. Graves'samples showed only a tnoderate increment in the number and size of EC in comparison with eontrois, and the intensity of ICAM-I* staining was approximately similar in ihe two scries of specimens. Clearly, only asubpopulation of Factor VIII * cells were atso ICAM-1 '. since this EC marker detects mainly targe and medium sized post-capillary EC, whereas lCAM-1 tends to be restricted to medium sized post-capiltary endothetium [2I|. When we compared on consecutive sections ICAM-1 and HLA Class II expressions on Graves* glands, no co-expression could be detected on thyrocytes.
Cett siispen.-iions and monolayers Cell suspension preparations of thyrocytes from both normat and Graves" gtands were invariably negative for ICAM-t expression when examined soon alter digestion. This was confirmed by the vwo disiii^ct four-layev IFL staining procedure.s. where ECwcre ICAM-I ' but thyrocytes were negative. On the other hand, after 36 48 h in monotayer cultures, soon after the celt attached to ttic glass coverstip, a proportion of thyrocytes derived from normal glands bceatBe cteariy positive for ICAM-1 (range 12-20%). The intensity of the surface staining varied from positive ( + ) to a strong positive reaction I + •¥). This phenomenon was even more pronounced in thyrocytes frotn Graves' patients, where it higher percentage of positivity (range 40 70"/u), and overall a stronger staining, coutd be observed. The identity of ICAM-I * thyroeytes was conlirmed by the four-layer IFL technique, with Mc-TPO' patient serum (Fig. 3a,b). NodifTerencein the nutnber of IC.AM-1 * eeils or in the intensity ofthe reaction was apparent using diflerent concentrations of FCS (nil, 1%, 10%, 15%) added to the TCM. The expression of H LA Class 11 on thyrocytes in monolayer celt culture from normat and Graves" glands showed approximately the same profile as that observed in the sections (() 5";. in normals and 40 80"-^. in Graves' glands). Co-expression of ICAM-t and HLA Class Ilcould thi.s time beclearly observed in the Graves' thyroid monolayer cells. Bloetilttti experiments. Addition of antibodies lo TNF-a and IL'N--/, separately or to both cytokines together, was unable to block the de novo expression of ICAM-l molecules on thyrocytes at any of the neutralizing concentrations used.
483 DISCUSSION
Since the initial finding that thyrocytes in autoimmune thyroid gtands inappropriately expres.sed HLA Class II molecules (20], concerted efforts have been made to substantiate the hypothesi.s that these cells are indeed able to act as "antigen-presenting cells' of their own surface autoantigeiis [22]. The confirmation soon catne with the demon.stration that helper T cell clones, raised after expanding CD4'^ lymphocytes isolated from autoimmune thyroid glands, proliferated in the presence of autologous Clas.s W thyrocytes separated from the same glands |23 25). Additional evidence has reeently corroborated this concept, when it was demonstrated that autoteaetive tissue recognition and its subsequent destruction could be induced by injecting IFN-yprimed Class II * thyrocytes into syngeneic mice [26). However, experimentat results obtaitied frotn transgcnic mouse models of Type I diabetes repeatedly demonstrated that the ectopic expression of Class 11 molecules alone on target /(cells is not/XT se sufficient to activate potentially autoreactive T eells. These controversial data have been extensively discussed in subsequent reviews on the topic [27,2S|. In human autoimmunity, a major problem remains unresolved, and this refers to the common notion that a "second signal" is normally required for helper T cells to expand after their interaction with conventional HLA Class 11' APC, Iti retation to thyroid autoimmunity, obvious candidates, such as IL-2 or IFN-)'. have been proposed, but these cytokines are not detectable in autoirnniunc thyroeytes ([2y], and reviewed in [30]). In addition, it is controversial whether another potential "sccotid sigtuil', IL-l, is actually secveled by these cells [31.321, More recently, lCAM-1 was found to fultil the functional requirements for an efficient antigen presentation. In fact, it was shown that co-expression of this adhesion molecule with Class II products on a libroblastoid line (but not Class II expression alone) enabled these cells to present relevant antigen to specific T eell clones [33]. When ihc search tor the possible presence of ICAM-l in ihc thyrocytes of autoimmune glands was completed, agreement was reached that ICAM-t is expressed on thyrocytes in Hashimoto's thyroiditis fl3,IS], However, because of the advanced degree of tissue destruction in this condition, it is unclear how speeifie the phenomenon is in relation to earlier pathogenic events, When the investigation tnovcd to GD. discrepant results etnergcd. While ICAM-I was tirst found, both in the cytoplasm [12,13] and on the surface of Graves' thyrocytes [12], others have subsequently failed to eonlirm these findings [18] or to reproduce them in full [34], Our study supports the ticgative data. These negative results were obtained mn'rp. both in cryostat sections (including those with Graves' thyrocytes near areas of lymphdcytic infiltration) and in viable thyroid cells in suspension, freshly prepared from the same glands. The lack of positive findings, obtained by eonvcntional IFL. was also confirmed by the use of the additional immunohistoehemical technique previously shown also to detect lCAM-l on Graves' ttiyrocytes (12]. Possible explanation.s for these discrepancies could be the high background staining with the alkatinc phosphatase technique, or the use of two different batches of 84H10 MoAb uiiti varying antibody speeificity. However, other groups have been unable to doeument any convincing expression of ICAM-1 in vivo in Graves' thyrocytes using various batches of the same
484
A. Ciampolillo et al.
MoAb iind npplying Ihcm loa hirgc number of thyroid samples [18.34]. The same applies to MoAb RRl/1 (34). a different antiICAM-l reagent known to recognize a spatiiilly distinct epitope from that seen by MoAb 84HI0 on the same molccule [3]. All the studies arc in agreement thai APC and F,C invariably express ICAM-1. both in normal und in Graves' ihyroid glands. In the case of EC. it was only the introduction ofthe four-layer technique, with the addition of specific markers lo identify Ihese cells (i,e. anli-Hacuir Vlll antibodies) or thyrocyles (i.e. antiMc-TPO antibodies) which conclusively excluded the positive reaction for ICAM-t previously assigned to the vascular pole of ihyrwyics fL^J. und inslead ciilribufcd il cxdusivdy now lo the EC surrounding them. This issue is further complicated by our in vivo rer.\us in vitro observations. Surprisingly, thyrocyles spontaneously expressed ICAM-I during the prolonged period in culture, LIndcr the conditions employed, ihyrocytes from control glands were as readily indtieible as those isolated from Graves' glands, even though the intensity of their staining was not so pronounced and the number of positive cells was fewer compared with Graves' thyrocytes (12 20"/,, ver.sus 40-70"-'ij). One could ;irgue that cylokinessecrclcd b> lymphocytes, concomilanlly present in the cuJlure. were responsible for eliciting the phenomenon, bul ihis was excluded by blocking experiments with neutralizing MoAbs to IFN-y and TNF-a, which did noi decrease the spontaneous expression of ICAM-I, Likewise, the potential inducing role of FCS in the TCM was excluded. On these grounds, the present demonstration that thyroid cells spontaneously and fully express this adhesion molecule only when in culture, casts some doubt on the relevance of ICAM-l to disease palhogenesis. Furlbermore. it leaves open to interpretation ibe resulls obtained with IFN-v [12,14] and other cyiokincs [15). immunomodulators suggested to play a prime role on ICAM-I exprt\s.sioi] on Ihyrocytes in vivo. The lull expression in vitro of ICAM-I on thyrocytes. regardless of whether iheir source derived from normal or Graves' glands, appears lo be merely the consequence of a physical phenomenon. The cITect could be artificially induced by contact of the cells with glass coverslips or through cell-lo-eell contact. We think the latter possibility is unlikely, because our Culture condition of low FCS concentrations in the TCM should have allowed thyroid follicles lo reconstitute with the light physiological polarity, as previously shown [35], On the argument that Graves' thyrocytes might express very low levels of ICAM-I. only in situ hybridization studies might help to clarify this complicated issue. The failure to deteet ICAM-I on autoimmune thyroid cells in vivo indicates thai the density of Class II molecules and the completeness of their sublocus (i,e. D R > DP> DQ) and invariant chain expressions [36) on the cell surface ofthese larget cells may be adequate for a potentially efficient autoanligen presenting ability, as shown in ecrtain experimenta! systems [37|, In other words, in thyroid autoimmunity it would be uneconomical for epithelial/endocrine cells to exhibit 'redundancy' of adhesion molecules, when the complete and anchored set of Class 11 molecules could apparently serve the purpose by themselves. Whether a 'second signal' can ultimately be detected remains the subject of further investigation, but the evidence produeed so far tends to exclude several conventional adhesion molecules as a major candidate for it. mainly because they cannot be detected, at least at the level of thyrocytes affected by GD [21]. However,
ihe ro]e of ICAM-I oncndolhcJial cells could stiJI bccrucia] for the recruitment of autoreactive T cells in the target gland. ACKNOWLEDGMENTS We arc grateful to all the scientists who have donated the various MoAb reagents and to lhe surgeons for providing human ihyrouis. We Ihank Paul Whittick for patiently editing and typing lhe manuscripl, A,M. was supported by The llchara Memorial I-oundalion (Tokyo. Jupani. The Byotaitaisy;) Foundation (Tokyo. Japan) and Novo Japan, A,C', was a fellow of The Juvenile Diabetes Foundation (USA) and G,N, was a feiiow of the Centro Nazionale delle Ricerche (Italy).
REFERENCES 1 Makgoha MW, Sanders ME, Shaw S, The CD2-LFA3 and LFAIiCAM-l pathways—relevance to T-cell recognition. Immunol Today I'»S9; 10,-4f7 22. 2 Marlin SD, Springer T, Purified intercellular adhesion molecule (ICAM-1) is a ligand Ibr lymphocyte function-associated antigen (LFA-i), Cell 1987;51;S13 5 3 Makgoba MW. Sanders ME, Ginlhcr Luce (iK. Gugel F.A, Dusiin ML. Springer TA. Shaw S. t unclional evidence that intercellular adhesion molcciilc-I (K'AM-I) is