Gut 1995; 36: 411-418

Increased expression of cell adhesion moleculeP-selectin in active inflammatory bowel disease

GM Schurmann, A E Bishop, P Facer, M Vecchio, J C W Lee, D S Rampton, JM Polak

AbstractThe pathogenic changes of inflammatorybowel disease (IBD) depend on migrationof circulating leucocytes into intestinaltissues. Although leucocyte rolling andtenuous adhesion are probably regulatedby inducible selectins on vascularendothelia, little is known about theexpression of these molecules in Crohn'sdisease and ulcerative colitis. Usingimmunohistochemistry on surgicallyresected specimens, this study investi-gated endothelial P-selectin (CD62, gran-ular membrane protein-140) in frozensections of histologically uninvolvedtissues adjacent to inflammation (Crohn'sdisease= 10; ulcerative colitis= 10), fromhighly inflamed areas (Crohn'sdisease=20; ulcerative colitis=13), andfrom normal bowel (n=20). By lightmicroscopy, two forms of P-selectinimmunoreactivity were detected thatapparently corresponded ultrastruc-turally to stored and released distribu-tions. Compared with the normal gut,there was a 3-7-fold increase of P-selectinimmunoreactivity on veins (p<0.0001),venules (p<0.0001), and capillaries(p<0.05) in the highly inflamed gut, with-out differences between Crohn's diseaseand ulcerative colitis. In the uninvolvedgut, P-selectin expression was similar tothat seen in normal controls, except for afocal increase of P-selectin in the vicinityof small lymphocyte aggregates. Thedramatic upregulation of P-selectin in theinflamed tissue and its potential role inleucocyte trafficking support the conceptof P-selectin blocking therapy for thecontrol of active IBD.(Gut 1995; 36: 411-418)

Keywords: inflammatory bowel disease, P-selectin.

The migration of leucocytes into tissues is thecentral event in inflammation and in animmune response. In inflammatory boweldisease (IBD), there is a dense intestinal infil-trate of inflammatory and activated immunecells with a differential distribution pattern forCrohn's disease and ulcerative colitis. For thedevelopment of the local intestinal cellularinfiltrate, circulating cells must stick to theintestinal vascular endothelium and transmi-grate into the tissue, where the immunoinflam-matory reaction is created.

Recently, a multistep cascade of adhesion ofcirculating cells to endothelial cells has been

proposed, entailing margination from thecentreline of blood flow towards the vascularwall, rolling, tethering to the endothelia, stableadhesion, and finally, transendothelial migra-tion.1 Each of these steps involves specific fam-ilies of adhesion molecules, which areexpressed on endothelial cells and on circulat-ing cells as their counterparts and ligands.2 3The selectin family of adhesion molecules,

which comprises E-selectin, P-selectin, and L-selectin, predominantly mediates the first stepsof cellular adhesion4 5 and several studies haveshown upregulation of E-selectin on activatedendothelial cells in a variety of tissues6-8 includ-ing the gut in patients with IBD.9 10 Littleinvestigation has been made, however, of P-selectin in normal and diseased gut, althoughits DNA was cloned and sequenced in 1989.11

P-selectin (also known as PADGEM,CD62, LECAM-3, or granular membraneprotein-140) is stored in endothelial cells andplatelets and is released after activation bymediators of inflammation,12 allowing thesecells to bind to their receptor/ligands, the car-bohydrate structure of sialyl-Lewis X, presenton neutrophils and monocytes.13-15Furthermore, P-selectin binds to CD4+ lym-phocytes,'6 subpopulations of memory cells,and natural killer cells.17 Expression of P-selectin is upregulated by histamine, thrombin,tumour necrosis factor o, 12 18 19 and by oxygenradicals,20 some of which have been shown tobe present in excess in IBD.21-23 P-selectin isexpressed also on endothelial cells infected byviruses,24 the presence of which has recentlybeen reported in IBD.25

In IBD, we have shown an increased per-centage of P-selectin positive platelets in theperipheral blood of patients with Crohn'sdisease and ulcerative colitis.26 In inflamedintestinal tissue, there is a single report of P-selectin in both Crohn's disease and ulcerativecolitis27 which, however, was confined toadvanced lesions and provided only limitedinformation about the topography and grade ofP-selectin immunoreactivity.

In this study, we hypothesised that P-selectin is upregulated in the development ofthe inflammatory lesion and in active IBD.Upregulated P-selectin could induce theadhesion of circulating inflammatory cells andthus contribute to the genesis of the intestinalcellular infiltrate.The aim of the study therefore was to

investigate qualitatively and quantitatively theexpression of immunoreactive P-selectin onendothelial cells in both uninvolved and highlyinflamed areas of Crohn's disease and ulcera-tive colitis, using light microscopy of operative

Department ofHistochemistry, RoyalPostgraduate MedicalSchool, LondonG M SchurnannA E BishopP FacerM VecchioJ M Polak

Department ofMedicine, St Mark'sHospital, LondonC W Lee

Department ofGastroenterology,Royal LondonHospital, LondonD S Rampton

Correspondence to:Dr G Schurmann,Department ofHistochemistry, RoyalPostgraduate MedicalSchool, Du Cane Road,London W12 ONN.

Accepted for publication24 June 1994

411

Schiurmann, Bishop, Facer, Vecchio, Lee, Rampton, Polak

TABLE I Clinical and histologicalfeatures of 63 patients with inflammatory bowel diseaseand 21 control patients

Crohn 's Ulcerativedisease Controls colitis Controlsileum ileum colon colon

PatientsNumber (male:female) 35 (24:11) 7 (5:2) 28 (16:12) 14 (9:4)Mean age (range) (years) 34 (18-70) 60 (38-81) 39 (28-53) 53 (25-77)

Preoperative therapyGlucocorticosteroids 28 0 23 0Sulphasalazine 4 0 22 0Azathioprine 3 0 10 0

OperationsResection of ileocolonic anastomosis 2 0 0 0Ileal resection 7 0 0 0Ileocaecal resection 26 0 0 0Right sided colectomy 0 7 0 3Left sided or subtotal colectomy 0 0 3 7Total colectomy 0 0 25 4

Tissues*Histological grade of inflammationNone (grade 0) 12 7 18 14Mild inflammation (grade 1) 3 0 8 0Intermediate (grade 2) 11 0 6 0High (grade 3) 29 0 16 0

*=Total numbers of tissue samples studied from study A and Study B.

specimens. We also examined the intracellularexpression of P-selectin in IBD by electronmicroscopy.

Methods

PATIENTS AND TISSUESSurgically resected specimens from a total of35 patients with Crohn's disease and 28patients with ulcerative colitis were obtainedwithin 30 minutes of removal. For the first partof the study, tissue samples were collectedfrom either macroscopically uninvolved areasat a distance of 2-4 cm from the inflamed areaor from the centres of inflammation (study A;one sample per case). In addition, to studythe expression cell adhesion of molecules atdifferent distance from the main lesion withinthe same patient, up to six samples perspecimen were collected from uninvolved,intermediate, and severely affected areas fromeach of a further five Crohn's disease patientsand five ulcerative colitis patients (study B;several samples per case). Indications forsurgery were chronic stenosis or disease refrac-tory to treatment, or both, in Crohn's diseaseand longstanding pancolitis or left sided colitisin ulcerative colitis. Non-inflamed controltissues were taken from hemicolectomy speci-mens resected for cancer, at least 5 cm fromthe malignancy (n= 17), and from total colec-tomy specimens resected for familiar adeno-matosis coli (n=4) (see Table I for furtherdetails). In all cases, diagnosis was confirmedby histopathological examination of theresected specimen.

TABLE II Antibody characteristics

Antibodies to Dilution Type Source

P-selectin (CD62) 1:1000 m/c A Mazurov*PECAM-1 (CD31) 1:8000 m/c A Mazurov*von Willebrand factor 1:200000 m/c Serotec, UKCD45RO (UCHL1) 1:8000 m/c Dako, DenmarkInterleukin 2 receptor (CD25) 1:100 m/c Dako, DenmarkPan-T cell (CD3) 1:800 p/c Dako, DenmarkMacrophages (CD68) 1:200 m/c Dako, Denmark

m/c=Mouse monoclonal; p/c=rabbit polyclonal; *antibodies were kindly donated by AM,Institute of Experimental Cardiology, Cardiology Research Centre, Moscow, Russia.

HISTOLOGICAL ASSESSMENTTissues were fixed by immersion in Zamboni'ssolution (saturated picric acid; 0.1 M phos-phate buffer; 2% w/v formalin pH 7.2) andrinsed in 15% (w/v) sucrose in 0-1 M phos-phate buffered saline with 0 01% (w/v) sodiumazide. Cryostat blocks were prepared and sec-tioned at 6 ,um thickness. One section fromeach sample was stained with haematoxylin andeosin for histological determination of inflam-mation according to a previously publishedmethod,28 grading from '0' (non-inflamed) to'3' (highly inflamed). Only tissues withouthistological signs of inflammation (grade '0')were included as 'uninvolved' (Table I).

IMMUNOCYTOCHEMISTRYTissue sections were immunostained using arange of antibodies (see Table II) by anindirect immunoperoxidase method29: sectionsserial to those used for staining P-selectin, werestained with monoclonal antibody againstplatelet endothelial cell adhesion molecule-i(PECAM-1) for the identification of micro-vessels.30 31 Tissues with P-selectin showing apunctate staining pattern were also immuno-stained for von Willebrand factor on serial sec-tions. Infiltrating mononuclear cells werefurther characterised by immunostaining forCD68 (macrophages), CD3 (T cells), andCD25 (interleukin 2 receptor) and CD45RO(memory cells) as markers ofT cell activation.Sections were counterstained with neutral fastred and mounted with glycerol gelatin.

ELECTRON MICROSCOPYPre-embedding transmission electron micro-scopic immunohistochemistry was performedon selected cases of both normal and Crohn'sdisease gut to elucidate the cause of the dif-ferential staining pattern of P-selectin seenby light microscopy on particular endothelia.Serial sections of 40 ,um thickness were cut andstained as mentioned above, free floating, in a12 well tissue culture plate. After staining, thesections were fixed in 1% (w/v) osmium tetrox-ide for two hours at 4°C, washed in phosphatebuffered saline, dehydrated in graded ethanols,and flat embedded in epoxy resin on a slidecovered with an acetate sheet.32 After removalof the sheet, the flat embedded sections wereobserved under a light microscope. To analysethe same vessels at both light and electronmicroscopic level, areas of interest, for examplevessels showing a punctate staining pattern,were cut out and re-embedded in plastic cap-sules, modified by trimming off the coneshaped tips of standard EM capsules. Ultrathinsections of silver-gold interference colour werestained with 4% (w/v) uranyl acetate inmethanol followed by Reynolds's lead citratefor one to two minutes and observed in a Zeiss10 CR electron microscope.

EVALUATIONFive visual fields within the mucosa andsubmucosa were chosen randomly from each

412

P-selectin in inflammatory bowel disease

TABLE III Differential expression of P-selectin on vascular endothelia in normal gut and in inflammatory bowel disease

Normal ileum (n= 7) Uninvolved Crohn's disease (n= 10) Highly inflamed Crohn's disease (n=20)P-selectinscore A Aa V Vv Cap A Aa V Vv Cap A Aa V Vv Cap

4 0 0 0 0 0 0 0 0 0 0 0 0 4 13 13 0 0 0 1 0 0 0 0 0 0 0 2 9 5 42 0 0 0 1 0 0 0 0 0 0 0 4 4 2 51 0 1 0 3 2 0 0 1 2 2 3 3 0 0 30 7 6 7 2 5 10 10 9 5 6 17 11 3 0 7

Normal colon(n= 14) Uninvolved ulcerative colitis (n= 10) Highly inflamed ulcerative colitis (n= 13)P-selectinscore A Aa V Vv Cap A Aa V Vv Cap A Aa V Vv Cap

4 0 0 0 0 0 0 0 0 0 0 0 0 0 9 03 0 0 0 3 1 0 0 0 0 0 0 0 10 3 02 0 3 2 2 2 0 0 0 3 2 0 0 3 1 81 0 2 2 4 2 0 0 2 3 2 0 8 0 0 30 14 9 10 5 9 10 10 8 4 6 13 5 0 0 2

A=arteries; Aa=arterioles; V=veins; Vv=venules; Cap=capillaries.

section for semi-quantitative light microscopicanalysis. Two independent observers who didnot know the diagnosis of the tissues assessedthe extent of endothelial expression ofP-selectin separately for each type of vessel.They counted the number of PECAM-1positive vessels and then counted the numberof P-selectin positive vessels on a section serialto that stained for PECAM-1. The percentageofPECAM-1 positive vessels that coexpressedP-selectin was estimated and the results wereexpressed as scores according to criteria similarto those reported elsewhere,33 meaning 0=nostaining; 1=30% of all vessels of a specificvessel type stained; 2=31-60%; 3=61-90%;4=more than 90% stained. Comparisonsbetween groups were made using the Mann-Whitney two tailed test for unpaired samples.Correlations were tested using Spearman'srank correlation test.

Results

P-SELECTIN IN THE NORMAL GUTIn normal gut, there was sporadic endothelialexpression of P-selectin in all layers of theintestinal wall without differences betweenileum and colon. Vascular endothelium wasmore often P-selectin positive in the mucosaand submucosa than in the subserosa or mus-cularis propria. P-selectin immunoreactivitywas mostly a feature of venules and, far less,capillaries (Table III). There was no staining

p = 0-001

c0

Cn

xa)00)COL-

4-

3..

2-

1-

0-

NSI I

lasto S..

o *see0o co

000 @0

00 00000

Normal ileum Uninvolved Highly inflamedCD CD

Figure 1: P-selectin immunoreactivity in venules in normalileum (n= 7) and in uninvolved (n= 10) and highlyinflamed (n=20) gut in Crohn's disease (CD).

on arteries and only very little on arterioles andveins. P-selectin immunoreactivity in anindividual vessel usually was either strong orcould not be detected; only very few vesselsshowed weak staining. The controls were olderthan the disease group but there was no signifi-cant decrease in expression of endothelial P-selectin with age (control ileum rankcorrelation=-0.24 (p=056), control colonrank correlation=-0.10 (p=0 78)). Intra-vascular platelets, as identified by PECAM-1,strongly coexpressed P-selectin without anapparent difference between the groups.

P-SELECTIN IN THE UNINVOLVED GUTADJACENT TO INFLAMMATIONIn the uninvolved areas of Crohn's disease andulcerative colitis, P-selectin showed distribu-tion patterns and staining intensity similar tothose seen in the normal gut. As in controls,immunoreactivity for P-selectin was restrictedto venules, capillaries, and some veins but wasnot detected on arterial vessels (Table III).Statistical evaluation of semiquantitativeanalysis showed no significant differencesbetween uninvolved areas of IBD and thenormal gut (Fig 1 and Fig 2).

In some vessels (<5% of all P-selectinpositive vessels), P-selectin immunoreactivityappeared as small black dots and patches (Fig3) rather than as homogeneous bands, as wasusually seen (Fig 4). The patchy distributionpattern was found on venules and, very rarely,

p =0N0001

NS 0A .-- *.0

cnCOU)CL,

00)

am

~0

CQxD

4-

3-

2-

1.

0 -

000 000

00 *of

0000 See

00000 @555

Normal colon Uninvolved Highly inflamedUc Uc

Figure 2: P-selectin immunoreactivity in venules in normalcolon (n= 14) and in uninvolved (n= 10) and highlyinflamed (n= 13) gut in ulcerative colitis (UC).

413

Scharmann, Bishop, Facer, Vecchio, Lee, Rampton, Polak

Figure 3: P-selectin on a submucosal venule in normalileum. Note the punctate intracytoplasmatic stainingpattern in the endothelium (arrow). Intraluminal plateletsare positive (original magnification X 460).

on arterioles. Few small veins and venulesexpressed both types of P-selectin immuno-reactivity within the same vessel. Staining forvon Willebrand factor showed a similar andeven more punctate staining pattern than thatoccasionally seen by staining for P-selectin (Fig5) suggesting that the punctate appearance ofP-selectin immunoreactivity could mean colo-calisation of both molecules. On ultrastruc-tural analysis of these areas it seemed that thepunctate staining pattern corresponded to P-selectin stored in Weibel-Palade bodies (Fig6). In contrast, the homogeneous longitudinalimmunoreactive bands seemed to correspondto P-selectin, redistributed along the endothe-lial cell membrane (Fig 7). Surprisingly, bothdistribution patterns of P-selectin immuno-reactivity could sometimes be detected withinan individual endothelial cell (Fig 8).

P-SELECTIN IN THE INFLAMED GUTIn highly inflamed areas of Crohn's disease andulcerative colitis (grade 3), P-selectinimmunoreactivity was upregulated dramati-cally (Table III, Fig 9). In comparison with

Figure 4: P-selectin on a small mucosal (m) vein close tothe submucosal (sm) border in uninvolved ileum fromCrohn 's disease. Note the homogeneous longitudinalimmunoreactive band along the entire endothelium(original magnification X430).

... . , :J '~~~~~~~~-R,:FYs1 ..

Figure 5: Von Willebrandfactor on a submucosal venule innormal ileum on a section serial to that ofFig 3. Thepunctate intracytoplasmatic staining pattern is even morepronounced than that in Fig 3 (original magnificationX51O.)

normal ileum, inflamed lesions of Crohn'sdisease showed a significant increase in P-selectin immunoreactivity score on venules(p=0.001; Fig 1, veins (p<0.0001) and capil-laries (p=0 05). P-selectin was slightly, but notsignificantly upregulated on arteries that werepositive in nine of 20 cases (Table III).Corresponding changes were seen on veins andvenules in ulcerative colitis (Fig 2), whencompared with normal colon. There were nodifferences in P-selectin expression betweenCrohn's disease and ulcerative colitis. In theinflamed gut, only very few venules showed thepunctate immunoreactivity for P-selectinfound in normal controls and in uninvolvedIBD gut, most vessels in inflamed sectionsshowing homogenous expression along theentire endothelial lining.

P-SELECTIN IN RELATION TO DEGREE OFINFLAMMATIONThe intraindividual relation of P-selectin tograde of inflammation was studied in a further25 specimens taken from five patients withCrohn's disease (histograde '0', n=2; '1', n=3;'2', n=11; '3', n=9) and on 24 specimenstaken from five patients with ulcerative colitis(histograde '0', n=8; '1', n=8; '2', n=6; '3',n=2) (study B). P-selectin immunoreactivityseemed to be increased and became more

Figure 6: Electron micrograph ofan endothelial cell innormal ileum on a section serial to that ofFig 3. P-selectinimmunoreactivity is present in a cytoplasmatic granule(arrow) near the abluminal plasma membrane (originalmagnification x 68 000; lu=lumen).

414

P-selectin in inflammatory bowel disease

Figure 7: Electron micrograph ofan endothelial cell inuninvolved ileum from Crohn's disease on a section serial tothat ofFig 5. P-selectin immunoreactivity is present alongthe luminal plasma membrane and in an invagination ofthe plasma membrane (insert). (Original magnificationX20 000; insert: X 63 000; er=erythrocyte.)

homogeneous across sections with the grade ofinflammation. For venules (Table IV), theexpression of P-selectin in a less inflamedtissue was always lower than or equal to itsexpression in a section displaying a highergrade of inflammation.

CELLULAR ENVIRONMENT OF P-SELECTINPOSITIVE VESSELS

Although evaluation of the total section areas

showed clear cut differences between IBDtissues and normal controls, P-selectinimmunoreactivity within an individual sectionwas usually quite heterogeneous. P-selectinwas expressed more frequently on venules,there was excessive cellular infiltrate in activedisease, but also in vessels situated close tosmall cellular aggregates in the uninvolved gut.In some cases, those parts of an individualvessel that were close to surrounding cellularclusters were P-selectin positive, whereas othersegments abutting uninflamed tissue were

P-selectin negative (Fig 10). Aggregatingmononuclear cells at the site of P-selectin posi-tive vessels were mostly CD3 positive T cellsand CD68 positive macrophages. Most of theT cells were CD45RO positive and some of

,,er.- -'

Figure 8: Electron micrograph ofan endothelial cell inuninvolved colon from ulcerative colitis. Note that P-selectin immunoreactivity is present along the outer surfaceofluminal plasma membrane, in folds ofabluminal plasmamembrane, and in a cytoplasmic granule (arrow).(Original magnification X 11 000; nu=nucleolus;lu=lumen.)

Figure 9: P-selectin is upregulated on most of thesubmucosal blood vessels in a representative section ofhighly inflamed Crohn 's disease. Intraluminal platelets arepositive (original magnification X280).

them coexpressed the interleukin 2 receptor,according to their state of activation. P-selectinwas induced at sites of mucosal ulceration andwas upregulated throughout the gut wall incases of transmural inflammation and in thevicinity of granulomas.

DiscussionThis is the first comprehensive report on theendothelial expression of P-selectin in Crohn'sdisease and ulcerative colitis. P-selectin washighly upregulated in inflamed areas of IBD,suggesting that P-selectin may participate inthe recruitment of circulating inflammatorycells into the lesion.

Electron microscopic colocalisation studiesusing double labelling experiments with anti-bodies to both von Willebrand factor andP-selectin have shown that, before release, P-selectin is stored, together with von Willebrandfactor, in endothelial intracellular organellesknown as Weibel-Palade bodies. 12 After stimu-lation, intracytoplasmic Weibel-Palade bodiesmigrate to the cell membranes and fuse withthe membrane to degranulate and allow P-selectin to redistribute along the endothelialcell surface.'2 In our tissues, although intenseblack P-selectin immunoreactivity prohibitedthe visualisation of tubular morphologycharacteristic for Weibel-Palade bodies at theelectron microscopic level,34 the typical shapeand intracellular localisation of the P-selectinimmunoreactive organelles left little doubt thatthey were Weibel-Palade bodies. This findingand the electron microscopic demonstration of

TABLE IV Expression of P-selectin on venules in tissueswith different grades ofinflammation taken from the samespecimens

Patient Crohn's disease Patient Ulcerative colitis

1 0*=tO<1<3 1 0=0=0<2=22 1=1=2=2=2=3 2 0<1<2=3=33 2<2<3=3=3 3 1<1=1<14 2<2=2<2=3 4 0=0=0<0=1=15 2=2=3=3<3 5 1=2<2<2

*=Histological grade of inflammation; t=immunoreactivity ofP-selectin is lower (<) or equal (=) in comparison withsections of tissue with the same or increasing grades ofinflammation.

415

Schurmann, Bishop, Facer, Vecchio, Lee, Rampton, Polak

Figure 10: P-selectin on two small veins packed with bloodcells in highly inflamed Crohn 's disease. Only those parts ofthe vessels that are close to clusters of inflammatory cells(ic) are positive (original magnification X280).

P-selectin redistributed on the endothelialsurface (Figs 7 and 8) suggest that, by meansof immunocytochemistry and light micro-scopy, two different forms of P-selectin can bedifferentiated: (a) a stored form (seen as cyto-plasmic, punctate immunoreactivity), (b) areleased form (seen as longitudinal bandsalong the endothelial surface). The storageform was found predominantly in the normaland uninvolved IBD gut whereas the releasedform was mainly detected in highly inflamedtissues.

In IBD, perhaps surprisingly, no significantupregulation of P-selectin was shown in histo-logically non-inflamed tissues, adjacent toinflamed lesions. Given the susceptibility ofany segment of the gastrointestinal tract to beaffected by Crohn's disease and the continuousspread of ulcerative colitis, areas in the vicinityof inflammation will probably becomeinflamed in the later course of disease and thusmay be regarded as 'early lesions'. We andothers have shown that these areas, althoughlacking evidence of inflammation by cell infil-trates, are not entirely normal, displaying, forexample, changed distribution of vasoactiveintestinal polypeptide containing nerves35 andincreased mucus production.36 Furthermore,increased expression of major histocompatibil-ity complex antigens on nerve bundles37 andendothelial cells38 and upregulated lymphocytefunction associated antigen-i on mononuclearcells31 point to immunoactivation in thevicinity of IBD lesions.

Although the regulators of P-selectinexpression are not yet fully understood, most ofthe known stimulants are produced bymacrophages (for example, histamine, tumournecrosis factor a, oxygen radicals), granulocytes(leukotriene C4, oxygen radicals) or T cells(tumour necrosis factor a), none ofwhich occur

in high numbers in uninvolved areas of intestine.Unchanged expression of P-selectin in thesetissues may thus result from a lack of appropri-ate stimulants. Another factor that may explainthe unchanged expression of P-selectin in earlylesions may be the time course of its release. P-selectin appears on the cell surface of endothelialcells five to 30 minutes after stimulation but, incultures, disappears five to 10 minutes later inthe absence of neutrophils.39 Thus, it may nothave been possible to detect increased expres-sion of P-selectin if it happens less often in theearly lesion than in advanced disease.

In the highly inflamed gut, we found anincreased expression of endothelial P-selectinin confirmation of preliminary findings byNakamura et al.27 Although our study wasbased on morphological findings, somefunctional implications can be drawn.Upregulation of P-selectin immunoreactivitywas mostly a feature of (postcapillary) venules- that is, the important sites of leucocytemigration in inflammation.40 As the mediationof cell adhesion is the only function of P-selectin established so far, increased P-selectinis probably involved in the recruitment ofvarious circulating inflammatory cells intothe IBD lesion; neutrophils, basophils,eosinophils, natural killer cells, and monocytesall have been shown to bind to endothelial P-selectin.13-15 17 Specifically, as T lymphocytesrepresent a major proportion of the cellularinfiltrate in IBD and bind to P-selectin,16 P-selectin could mediate recirculation of thesecells, which, after local intestinal antigen stim-ulation, proliferate elsewhere and migrate backto the intestinal lesion. Subsequent reciprocalinteraction between endothelium and circulat-ing cells - that is, induction of P-selectin onendothelial cells by stimulants released frominflammatory cells and recruitment of inflam-matory cells into the lesion by endothelial P-selectin - could result in a self perpetuatingvicious circle leading to the development ofthe inflammatory infiltrate. Further indirectevidence for the participation of endothelial P-selectin in intestinal cell adhesion in IBDderives from its coexpression with intercellularadhesion molecule-i (ICAM-1) found in ourprevious study31 and by others.2741 ICAM-1mediates steps of the adhesion cascade subse-quent to those mediated by P-selectin.4 Giventhe high constitutive expression of endothelialICAM-1 in the gut,3' however, upregulation ofICAM-1 in IBD is less dramatic in comparisonwith that seen for P-selectin. Thus, in activedisease, continuous migration of circulatingcells into the inflamed tissue is essentiallymediated by P-selectin and may contribute tothe maintenance and spread of inflammation.

Recent evidence exists for possible addi-tional functions of P-selectin. Blocking themolecule's action showed that P-selectin medi-ates vascular permeability and haemorrhageafter intravenous administration of cobravenom factors in rats42 and participates intissue necrosis and oedema after transectionand replantation of the rabbit ear.43 P-selectinalso contributes to the pulmonary micro-vascular dysfunction seen after intestinal

416

P-selectin in inflammatory bowel disease 417

ischaemia/reperfusion.44 Although some ofthese effects may be caused by P-selectinreleased from activated platelets,12 the dramaticupregulation of endothelial P-selectin seen inactive disease may contribute to various patho-logical events in IBD, either by local action, or,systemically, by shedding of the molecule.45

Interrupting cellular recruitment into thelesion in IBD is an important therapeutic aim.Preliminary experimental data show thatcellular adhesion can be blocked by mono-clonal antibodies against adhesion molecules.For example, local systemic application of anti-bodies against lymphocyte function associatedantigen-i in animals significantly reduces thecellular infiltrate and tissue damage in car-diac46 and intestinal47 ischaemia reperfusioninjury and in experimental rat colitis.48 ForP-selectin, application of a monoclonal anti-body significantly decreases adherence of poly-morphonuclear cells to stimulated endothelialcells and protects feline heart in myocardialischaemia and reperfusion injury.49 In rats,endotoxin induced neutropenia and poly-morphonuclear cell accumulation in tissuescan be blocked by treatment with antibodies toP-selectin.50 Infusion of sialyl-Lewis X, aligand for P-selectin, significantly reduces lunginjury and diminishes the tissue accumulationof neutrophils in a P-selectin dependent modelof rat lung injury.51 The dramatic increase ofP-selectin in active IBD shown in this studysupports the concept that blockade of P-selectin may have a therapeutic use in Crohn'sdisease and ulcerative colitis.Supported by Deutsche Forschungsgemeinschaft grant Schu720/2-4.

1 Shimizu Y, Newman W, Tanaka Y, Shaw S. Lymphocyteinteractions with endothelial cells. Immunol Today 1992;13: 106-12.

2 Dustin ML, Springer TA. Role of lymphocyte adhesionreceptors in transient interactions and cell locomotion.Ann Rev Immunol 1991; 9: 27-66.

3 Hogg N. Roll, roll, roll your leucocyte gently down thevein.... Immunol Today 1992; 13: 113-5.

4 Lawrence MB, Springer TA. Leukocytes roll on a selectinat physiologic flow rates: distinction from and pre-requisite for adhesion through integrins. Cell 1991; 65:859-73.

5 Lorant DE, Topham MK, Whatley RE, McEver RP,McIntyre TM, Prescott SM, et al. Inflammatory roles ofP-selectin. J Clin Invest 1993; 92: 559-70.

6 Taylor PM, Rose ML, Yacoub MH, Pigott R. Induction ofvascular adhesion molecules during rejection of humancardiac allografts. Transplantation 1992; 54: 451-7.

7 Ferren C, Peuchmaur M, Desruennes M, Ghoussoub JJ,Brousse N, Cabrol C, et al. Implications of de novoELAM- 1 and VCAM-1 expression in human cardiac allo-graft rejection. Transplantation 1993; 55: 605-9.

8 Steinhoff G, Behrend M, Schrader B, Duijvestijn AM,Wonigeit K. Expression patterns of leukocyte adhesionligand molecules on human liver endothelia. Am Jf Pathol1993; 142: 481-8.

9 Ohtani H, Nakamura S, Watanabe Y, Fukushima K, MizoiT, Kimura M, et al. Light and electron microscopicimmunolocalization of endothelial leukocyte adhesionmolecule-i in inflammatory bowel disease. VirchowsArchivA PatholAnat 1992; 420: 403-9.

10 Koizumi M, King N, Lobb R, Benjamin C, Podolsky DK.Expression of vascular adhesion molecules in inflamma-tory bowel disease. Gastroenterology 1992; 103: 840-7.

11 Johnston GI, Cook RG, McEver RP. Cloning ofGMP-140,a granule membrane protein of platelets and endothelium:sequence similarity to proteins involved in cell adhesionand inflammation. Cell 1989; 56: 1033-44.

12 McEver RP, Beckstead JH, Moore KL, Marshall-CarlsonL, Bainton DF. GMP-140, a platelet a-granulemembrane protein, is also synthesized by vascularendothelial cells and is localized in Weibel-Palade bodies.J Clin Invest 1989; 84: 92-9.

13 Polley MJ, Phillips ML, Wayner E, Nudelmann E, SinghlaAK, Hakamori 5, et al. CD62 and endothelial cell-leuko-cyte adhesion molecule-i (EL.AM-1) recognize the samecarbohydrate ligand, sialyl-Lewis x. Proc Nat Acad SCiUSA 1991; 88: 6224-8.

14 Picker U, Wamock RA, Bums AR, Doershuk CM, BergEL, Butcher EC. The neutrophil selectin LECAM-1presents carbohydrate ligands to the vascular selectinsELAM-1 and GMP-140. Cell 1991; 66: 921-33.

15 Erbe DV, Watson SR, Presta LG, Wolitzky BA, Foxall C,Brandley BK, et al. P-selectin and E-selectin use commonsites for carbohydrate ligand recognition and celladhesion. J CellBiol 1993; 120: 1227-35.

16 Damle NK, Klaussmann K, Dietsch MT, MohagheghpourN, Aruffo A. GMP-140 (P-selection/CD62) binds tochronically stimulated but not resting CD4+ T lympho-cytes and regulates their production of proinflammatorycytokines. EurJImmunol 1992; 22: 1789-93.

17 Moore KL, Thompson LF. P-selectin (CD62) binds tosubpopulations of human memory T lymphocytes andnatural killer cells. Biochem Biophys Res Commun 1992;186: 173-81.

18 Weller A, Isenmann S, Vestweber D. Cloning of the mouseendothelial selectins. Expression ofboth E- and P-selectinis inducible by tumour necrosis factor alpha. Jf Biol Chem1992; 267: 15176-83.

19 Collins PW, Macey MG, Cahill MR, Newland AC. VonWillebrand factor release and P-selectin expression isstimulated by thrombin and trypsin but not IL-1 in cul-tured human endothelial cells. Thromb Haemost 1993; 70:346-50.

20 Patel KD, Zimmermann GA, Prescott SM, McEver R,McIntyre TM. Oxygen radicals induce human endothelialcells to express GMP-140 and bind neutrophils. Jf Cell Biol1991; 112: 749-59.

21 Rampton DS, Murdoch RD, Sladen GE. Rectal mucosalhistamine release in ulcerative colitis. Clin Sci 1980; 59:389-91.

22 Simmonds NJ, Allen RE, Stevens TR, Van Someren RN,Blake DR, Rampton DS. Chemiluminescence assay ofmucosal reactive mediator metabolites in inflammatorybowel disease. Gastroenterology 1992; 103: 186-96.

23 Murch SH, Braeger CP, Walker-Smith JA, MacDonaldIT. Location of tumour necrosis factor-ct immunohisto-chemistry in chronic inflammatory bowel disease. Gut1993; 34: 1705-9.

24 Etingin OR, Silverstein RL, Hajjar DP. Identification of amonocyte receptor on herpes virus-infected endothelialcells. Proc Natl Acad Sci USA 1991; 88: 7200-3.

25 Wakefield AJ, Pittilo RM, Siu R, Cosby SL, Stephenson JR,Dhillon AP, et al. Evidence of persistent measles virusinfection in Crohn's disease. JMed Virol 1993; 39: 345-53.

26 Collins CE, Cahill MR, Newland AC, Rampton DS.Circulating platelets are activated in inflammatory boweldisease. Gastroenterology 1994; 106: 840-5.

27 Nakamura S, Ohtani H, Watanabe Y, Fukushima K,Matsumoto T, Kitano A, et al. In situ expression of thecell adhesion molecules in inflammatory bowel disease.Lab Invest 1993; 69: 77-85.

28 Saverymuttu SH, Camilleri M, Rees H, Lavender JP,Hodgson HJF, Chadwick VS. Indium 11 1-granulocytescanning in the assessment of disease activity in inflam-matory bowel disease: a comparison with colonoscopy,histology and fecal indium 11 1-granulocyte excretion.Gastroenterology 1986; 90: 1121-8.

29 Hsu SM, Raine L, Fanger H. Use of avidin-biotin-peroxi-dase complex (ABC) in immunoperoxidase techniques. JHistochem Cytochem 1981; 29: 577-80.

30 Horak ER, Leek R, Klenk N, Lejeune S, Smith K, Stuart N,et al. Angiogenesis, assessed by platelet/endothelial celladhesion molecule antibodies, as indicator of node metas-tases and survival in breast cancer. Lancet 1992; 340:1120-4.

31 Schiirmann G, Aber-Bishop AE, Facer P, Lee JC, RamptonDS, Dore C, et al. Altered expression of cell adhesionmolecules in uninvolved gut in inflammatory boweldisease. Clin Exp Immunol 1993; 94: 341-7.

32 Priestley JV, Alvarez El, Averill S. Pre-embedding electronmicroscopic immunohistochemistry. In: Polak JM,Priestley JV, eds. Electron microscopic immunohistochemistry.Oxford: Oxford University Press, 1992: 89-122.

33 Wiedenmann B, Waldherr R, Buhr H, Hille H, Rosa P,Huttner W. Identification of gastroenteropancreaticneuroendocrine cells in normal and neoplastic humantissue with antibodies against synaptophysin, chromo-granin A, secretogranin I (chromogranin B), and secre-togranin II. Gastroenterology 1988; 95: 1364-74.

34 Bibbo M. Comprehensive cytopathology. Philadelphia: WBSaunders, 1992.

35 Bishop AE, Polak JM, Bryant MG, Bloom SR, Hamilton S.Abnormalities of vasoactive intestinal polypeptide con-taining nerves in Crohn's disease. Gastroenterology 1980;79: 853-60.

36 Dvorak AM. Ultrastructural pathology of Crohn's disease.In: Goebell H, Peskar BM, Malchow H, eds. Inflammatorybowel disease - basic research and clinical implications.Lancaster: MTP Press, 1988: 3-41.

37 Geboes K, Rutgeerts P, Ectors N, Mebis J, Penninckx F,Vantrappen G, et al. Major histocompatibility class IIexpression on the small intestinal nervous system inCrohn's disease. Gastroenterology 1992; 103: 439-47.

38 Koretz K, Momburg F, Otto HF, Moeller P. Sequentialinduction of MHC antigens on autochthonous cells ofileum affected by Crohn's disease. Am J Pathol 1987; 129:493-502.

39 Hattori R, Hamilton KK, Fugate RP, McEver RP, Sims PJ.Stimulated secretion of endothelial von Willebrand factoris accompanied by rapid redistribution to the cell surfaceof the intracellular granule membrane protein GMP-140.JfBiolChem 1989; 264: 7768-71.

418 Schurmann, Bishop, Facer, Vecchio, Lee, Rampton, Polak

40 Cotran RS, Pober JS. Endothelial activation: its role ininflammatory and immune reactions. In: Simionescu N,Simionescu M, eds. Endothelial cell biology in health anddisease. New York: Plenum Publishing, 1988: 335-48.

41 Malizia G, Calabrese A, Cottone M, Raimondo M,Trejdosiewicz T, Smart CJ, et al. Expression of leukocyteadhesion molecules by mononuclear phagocytes in inflam-matory bowel disease. Gastroenterology 1991; 100: 150-9.

42 Mulligan MS, Polley MJ, Bayer RJ, Nunn MF, Paulson JC,Ward PA. Neutrophil-dependent acute lung injury.Requirement for P-selectin (GMP-140). Jf Clin Invest1992; 90: 1600-7.

43 Winn RK, Harlan JM. CD-18-independent neutrophil andmononuclear leukocyte emigration into the peritoneum ofrabbits. J Clin Invest 1993; 92: 1168-73.

44 Carden DL, Young JA, Granger DN. Pulmonary microvas-cular injury after intestinal ischemia-reperfusion - role ofP-selectin.7ApplPhysiol 1993; 75: 2529-34.

45 Ushiyama S, Laue T, Moore K, Erickson H, McEver R.Structural and functional characterization of monomericsoluble P-selectin and comparison with membrane P-selectin. J Biol Chem 1993; 268: 15229-37.

46 Yamazaki T, Seko Y, Tamatni T. Expression of intercellu-lar adhesion molecule-1 in rat heart with ischemia/

reperfusion and limitation of infarct size by treatment withantibodies against cell adhesion molecules. Am 7 Pathol1993; 143: 410-8.

47 Zimmermann BJ, Holt JW, Anderson DC, Miyasaka M,Granger ND. Role of adhesion glycoproteins in lipidmediator-induced leukocyte adhesion and emigration inrat mesenteric venules. Gastroenterology 1993; 104:A807.

48 Arndt H, Yamada T, Palitzsch K, Grisham M, GrangerND. Leukocyte endothelial cell adhesion in a model ofchronic intestinal inflammation. Gastroenterology 1993;104: A662.

49 Weyrich AS, Ma XL, Lefer DJ, Albertine KH, Lefer AM. Invivo neutralization of P-selectin protects feline heart andendothelium in myocardial ischemia and reperfusioninjury. J Clin Invest 1993; 91: 2620-9.

50 coughlan AF, Hau H, Dunlop LC, Berndt MC, HancockWW. P-selectin and platelet-activating factor mediateinitial endotoxin-induced neutropenia. Y Exp Med 1994;179: 329-34.

51 Mulligan MS, Paulson JC, Frees SD, Zheng ZL, Lowe JB,Ward PA. Protective effects of oligosaccharides inP-selectin-dependent lung injury. Nature 1993; 364:149-51.