American journal of Pathology, Vol. 142, No. 2, Febrary 1993

Copyrght © Americani Societv for Investigative Pathology

Expression of Vascular Cell Adhesion Molecule(VCAM-1) in Liver and Pancreas AllograftRejection

Carlos E. Bacchi,* Christopher L. Marsh,tJames D. Perkins,t Robert L. Carithers Jr.,*John P. McVicar,t Kelly L. Hudkins,*Christopher D. Benjamin,§ John M. Harlan,*Roy Lobb,§ and Charles E. Alpers*From the Departments of Pathology,* Surgery,t andMedicine,* University of Washington, Seattle, Washinigton,and Biogen, Inc, Cambridge, MassachusettsO

VCIM-1, a leukocyte adhesion molecule expres-sed by cytokine-activated endothelial cells in cul-ture, may mediate mononuclear leukocyte infil-tration in vessels and interstitium in solid organallograft rejection. Using the avidin-biotin immu-noperoxidase technique and two different anti-bodies-4B9, a murine monoclonal antibody, andrabbitpolyclonal antisera to recombinant humanVC4AM (rVCAM Ab), which work in methacarn-fixed tissues-we studied the expression of thismolecule in biopsies of transplanted liver andpancreas with and withoutfeatures of rejectionas weU as nontransplant control tissues. TherVCAM Ab, but not 4B9, showed a population ofreactive endothelial ceUs limited to sites ofprom-inent subendothelial leukocytic cell infiltration inarteries and veins in rejecting allografts. VCAM-1expression by sinusoidal endothelium in reject-ing liver allografts was also observed. In addi-tion, a population of ceUs (DC) with dendriticmorphology was identified by rVCAM Ab withinsites of lymphoid ceU aggregations in both liverand pancreas allografts. Further evidence thatthese ceUs represent true DC was obtained byidentification of VCAM-1 positive morphologi-cally similar ceUs in both germinal centers andinterfoUicular areas of reactive lymph nodes;and by similar staining ofthese cells in allograftorgans by a monoclonal antibody to nervegrowthfactor receptor, previously shown to recognizeDC DCs were generaly not seen in normal con-trol organs or portions of allografts uninvolvedby lymphoid aggregates. This study provides evi-dence that 1) endothelial cell expression of

VCAM-1 may be important in transplant rejection,2) different epitopes of VCAM-1 may be pre-served in tissue sections and recognized by dif-ferent antibodies, and 3) there isprobably apop-ulation of VCAM-1 expressing DC that partici-pates in the celular rejection progress. (Am JPathol 1993, 142:579-591)

Acute rejection in liver and pancreas allografts, as inother solid organ allografts such as heart and kidney,is characterized by infiltration of the parenchyma byeffector immunocompetent and inflammatory cells,principally lymphocytes and monocytes/macro-phages.1-6 In liver, the pathology of rejection seen inallograft biopsies most often displays features of aportal triaditis, inflammatory injury to interlobular bileducts, and what has been called endothelialitisinvolving the portal or central veins.1-7 In pancreas,the pathologic features of rejection typically includeinflammatory infiltration of the interstitial tissue aswell as the epithelium of the exocrine pancreas.8 10

In both liver and pancreas, a process of significantvascular injury can be identified in many cases aspart of the rejection process. This injury may take theform of a vasculopathy involving muscular arteries,which can be manifest acutely as an endothelialitiswith infiltration of the subendothelial intima by mono-nuclear inflammatory cells. This lesion, like thatoccurring in cardiac and renal allografts, may thenprogress to a chronic arteriopathy with smooth mus-cle migration into the intima and matrix deposition,leading to intimal sclerosis.11-13 The microvascula-ture is also a common participant in solid organallograft rejection, in which the microvascular endo-thelium apparently may serve either as an antigenictarget to immunocompetent cells of the host14 or mayundergo phenotypic changes that enhance the

Supported in part by grants DK40802, HL42270, HL47151, andHL18645 from the National Institutes of Health.Accepted for publication July 27, 1992.

Address reprint requests to Dr. Charles E. Alpers, Department ofPathology, RC-72, University of Washington, Seattle, WA 98195.

579

580 Bacchi et alAJP February 1993, Vol. 142, No. 2

transmigration of host leukocytes into interstitial tis-sues, 15'16 where they can encounter additionaldonor cell types against which a rejection responsemay be mounted.

It is known that these vascular rejection processes

are not uniform and that rejection of solid organs isusually an irregularly distributed process. The mech-anisms by which circulating leukocytes localize toonly certain artery segments or areas of extravascu-lar parenchyma are not known. Nonetheless, an

intriguing mechanism that could explain in part a

process of focal leukocyte recognition and attach-ment to portions of the vasculature corresponding toareas of injury involves the induction or up-regulationof specific leukocyte adhesion molecules on thesurface of vascular endothelium. At present, foursuch endothelial proteins have been molecularlycloned17-20 and shown to be involved in leukocyteadhesion in man21-26; these include endothelial leu-kocyte adhesion molecule-1 (ELAM-1, E-selectin),PADGEM (platelet activation-dependent granule-ex-ternal membrane protein, GMP-140, CD62, P-selec-tin), intercellular adhesion molecule-1 (ICAM-1,CD54), and vascular cell adhesion molecule-1(VCAM-1, INCAM 110). It seems likely that modula-tion of the expression or configuration of such adhe-sion proteins on the endothelium or similar modula-tion of the corresponding ligands on circulatingleukocytes might be especially important in estab-lishing the endothelial and subendothelial injurycharacteristic of vascular rejection in solid organ

allografts, and might also be vital to localization ofthe intraparenchymal inflammatory infiltrates mediat-ing interstitial rejection in such allografts.

Induction of VCAM-1 expression by microvascularendothelium in areas of leukocyte sequestration incardiac transplant rejection has been demon-strated.27'28 The interaction between VCAM-1expressed on endothelium present in inflamed syn-

ovium in patients with rheumatoid arthritis and itscounter-receptor on resting and activated T lympho-cytes has also been shown to be involved in recruit-ment and localization of leukocytes to sites ofimmune injury occurring in disease settings otherthan transplantation.26 In this study, we have usedboth a recently characterized monoclonal antibody(4B9) to VCAM-1, and a rabbit polyclonal antisera torecombinant human VCAM-1 (anti-rVCAM-1) to eval-uate the expression of this molecule in the setting ofliver and pancreas allograft rejection. Both antibod-ies recognize VCAM-1 epitopes that are preserved intissue fixed in methyl-Carnoy's solution. In this study,we show that anti-rVCAM-1, but not 4B9, recognizesup-regulated expression of VCAM-1 on vascular

endothelium at sites of active inflammatory infiltrationcharacteristic of rejection, and that this VCAM-1expression becomes undetectable as rejection epi-sodes are successfully treated by increasedimmunosuppression. We also demonstrate that apopulation of dendritic cells with immunophenotypicfeatures of both follicular and interdigitating dendriticcells of lymph nodes can be identified focally in theinterstitium of occasional livers at the time of organdonation and within interstitial lymphoid aggregatespresent in acutely rejecting liver and pancreaticallografts.

Materials and Methods

Tissue SelectionA total of 38 hepatic and 11 pancreatic biopsies from20 patients who underwent pancreas and liver trans-plants, respectively, were used in this study. All tis-sues were obtained as core needle biopsies. Elevenof the 15 patients undergoing liver transplantationhad at least one episode of biopsy-proven rejection,as did all of the pancreatic transplant patients. Threepatients had two consecutive allograft rejection biop-sies available for study, for a total of 14 rejectionbiopsies. All the liver transplant patients had protocolbiopsies obtained at the time of transplantation, andafter episodes of rejection. The protocol baselinebiopsy, biopsies showing rejection, and at least onefollow-up biopsy in which features of rejection hadresolved were studied in each liver transplant case,and biopsies showing rejection and subsequent res-olution of rejection (three of five cases; one death;one persistent rejection) were studied in pancreastransplant cases. Eleven additional donor liver biop-sies or normal liver obtained from uninvolved por-tions of hepatectomy specimens resected for local-ized malignancy were also studied.

Because VCAM-1 has been identified within den-dritic cells in lymphoid tissues, the reactivity of theantisera used in this investigation was also studied inseven reactive lymph nodes and tonsils, surgicallyremoved for lymphoid hyperplasia.

All tissues were fixed in methacarn fixative (60%methanol, 30% chloroform, 10% acetic acid) for atleast 12 hours and then processed, paraffin embed-ded, and sectioned using conventional techniques.

ImmunohistochemistryBriefly, sections of methyl-Carnoy's fixed tissue weredeparaffinized with xylene and graded ethanol,blocked with 3% hydrogen peroxide, and washed

VCAM-1 in Liver and Pancreas Allografts 581AJP February 1993, Vol. 142, AMo. 2

with phosphate-buffered saline (PBS) (138 mmol/LNaCI, 2.7 mmol/L KCI, 3.2 mmol/L Na2HPO4, 1.5mmol/L KH2PO4, pH 7.3). The tissue was then incu-bated with one of the primary murine monoclonalantibodies (see below) or rabbit polyclonal antiseraand subsequently processed using an avidin-biotinimmunoperoxidase method with 3,3'-diaminobenzi-dine (with nickel chloride enhancement) as the chro-mogen as previously described.29'30 Sections werecounterstained with methyl green or hematoxylin. Forall samples, negative controls for the immunohis-tochemical procedures consisted of substitution ofthe primary antibody with both irrelevant murinemonoclonal antibodies, or nonimmune rabbit sera,and PBS. Positive controls included concurrent stain-ing of fixed human tonsil, a tissue with detectableconstitutive expression of VCAM-1 on dendriticcells,31 (Figure 1A) and fixed normal human kidney,a tissue with detectable constitutive expression ofVCAM-1 on parietal epithelial cells.31

AntibodiesVascular Cell Adhesion Molecule- 1

Two sources of antibodies were utilized for thisstudy. Murine monoclonal antibody 4B9 has beencharacterized previously and its specificity forVCAM-1 established through competitive bindinginhibition studies.25 Reactivity and specificity of thisantibody in fixed tissues was established usingmethyl-Carnoy's fixed-cell aggregates of ChineseHamster Ovary cells either transfected or untrans-fected (negative control) with VCAM cDNA (kindlyprovided by Dr. Margaret Rosa, Biogen, Cambridge,MA), which were then shown to express VCAM-1 atthe cell surface by appropriate binding of leukocytecell lines as well as binding inhibition assays.32We also used a rabbit polyclonal antisera that was

raised against a recombinant form of humanVCAM-1 (rsVCAM) that was purified to homogeneityby immunoaffinity chromatography as previouslydescribed.33 New Zealand White rabbits, 3 to 4 kg,were immunized with purified rsVCAM (1 mg), emul-sified (1:1) in Freund's complete adjuvant (Difco Lab-oratories, Detroit, Ml). The rabbits were boosted withrsVCAM (1 mg) in incomplete adjuvant, at monthly(3x) and then bimonthly intervals. Bleeds were taken7 to 14 days after each boost. Rabbit antisera wasaffinity purified by passage over protein A, with theIgG fraction then passed over an affinity resin ofhuman rsVCAM-1 immobilized on Affigel (8 mg ofrsVCAM-1 per milliliter of resin). The antisera waseluted with buffer at pH 3.0, dialized into PBS, ali-quoted, and stored at -80 C.

Nerve Growth Factor Receptor-5 (NGFR5)

NGFR5 is a monoclonal antibody originally devel-oped to study the expression of nerve growth factorreceptor in tumors and normal tissues.34 Among nor-mal tissues, in addition to expected neural immu-nostaining, NGFR5 has been demonstrated to reactwith several non-neural cell types, including lym-phoid follicular dendritic cells.34'35 In this study weused the NGFR5 as an independent confirmatoryimmunolocalization marker of dendritic cells in lym-phoid and nonlymphoid tissue.

Leukocyte Markers

Immunophenotypic characterization of infiltrat-ing leukocytes was performed as previouslydescribed.13 Commercially available antibodieswere used to identify populations of monocytes/macrophages (anti-CD 68, monoclonal antibodyKP-1, Dako Corporation, Carpinteria, CA),36 T lym-phocytes (anti-CD 43, monoclonal antibody Leu 22,Dako Corporation),37 and B lymphocytes (anti-CD20,monoclonal antibody L26, Dako Corporation).38'39

Endothelial Markers

Rabbit polyclonal anti-human factor VIII relatedantigen/von Willebrand factor (Dako Corporation)was used as previously described.40

Proliferating Cell Nuclear Antigen (PCNA)

19A2, a murine monoclonal antibody to PCNA/cyclin (Coulter Corporation, Hialeah, FL) was usedas previously described.4

Absorption AssayChinese Hamster Ovary (CHO) cells transfected withVCAM-1, ICAM-1, and ELAM-1 and demonstratingsurface expression of each of these molecules aswell as untransfected CHO cells, were maintained inculture as previously described.32 Culture plateswere washed with PBS, scraped, and pelleted aftercentrifugation at 1200 rpm. Cell pellets were fixed inmethyl-Carnoy's solution, and processed andembedded in paraffin for tissue immunohistochem-istry using procedures detailed above. Both the rab-bit polyclonal antisera to recombinant VCAM-1 andthe murine monoclonal antibody 4B9 were demon-strated to be reactive with the cell surface of CHOcells transfected with VCAM-1, but not with untrans-fected cells or those transfected with ICAM-1 orELAM-1 (data not shown).

582 Bacchi et alA/P Februlart 1993, Vol. 142, No. 2

Cell pellets, each containing approximately 25 x106 total cells, of VCAM-1, ICAM-1, and untrans-fected CHO cells were also collected in serial dilu-tions of PBS and then incubated for 60 minutes atroom temperature with aliquots of both rabbit poly-clonal anti-VCAM-1 antisera and the monoclonalantibody 4B9. After incubation, the suspensionswere again centrifuged at 1200 rpm and the super-natants collected for incubation on tissue sections ofboth the cell pellets as noted above and on methyl-Carnoy's fixed tissue sections of human tonsil. Iden-tification of tissue binding was determined with theavidin-biotin immunoperoxidase technique detailedabove, and tissues were counterstained with methylgreen.

ResultsLymphoid OrgansThe rabbit polyclonal anti-rs VCAM-1 showed prom-inent reactivity with follicular dendritic cells in thegerminal centers of reactive lymph nodes, and to aless numerous population of interdigitating cells inthe T-cell-dependent interfollicular areas of lymphnodes and tonsils (Figure 1A). The monoclonal anti-body 4B9 showed a similar but less pronounced pat-tern of staining. Absorption of the polyclonal antiserawith VCAM-1 expressing CHO cells, but not ICAM-1transfected or untransfected CHO cells, completelyblocked the reactivity of the antisera with dendriticcells (Figure 1B).A second antibody, NGFR5, was also reactive to

follicular, but not interdigitating, dendritic cells in lym-phoid tissues as previously described.34 The reac-tivity patterns of the NGFR5 and VCAM-1 antiserawith follicular dendritic cells in germinal centers wereconcordant.

Allograft TissueExpression of VCAM-1 in both hepatic and pancre-atic allograft biopsies is summarized in Table 1. Theresults shown and discussed below were obtainedwith the rabbit polyclonal antisera to recombinanthuman VCAM-1. Immunostaining with the 4B9 mon-oclonal antibody occasionally was similar to thatobtained with the polyclonal antisera, but most oftenwas entirely absent in liver and pancreatic tissue.There were no examples of cells reactive with the4B9 antibody but unreactive with the rabbit antisera.

LiverVCAM-1 expression was not detected in normal liver.No VCAM-1 expression was detected in 13 of 15protocol baseline liver biopsies; 1 in 15 casesshowed the presence of portal dendritic cells withVCAM-1 expression (see below) and 1 in 15 biopsiesshowed endothelial expression in a single hepaticartery segment but was otherwise negative. At timesof rejection, diagnosed by established morphologiccriteria,1-6 focal expression of VCAM-1 was demon-strated in 12 of 14 biopsies (Figure 2). This expres-sion was manifest in one of three ways. Focal stain-ing of the endothelial lining of portal veins in areas oftriaditis and endothelialitis (Figure 2E) was the mostcommon finding identified in 10 in 12 of the rejectionbiopsies demonstrating VCAM-1 expression. Portalvein endothelial expression of VCAM-1 was alwaysfocally distributed and associated with either adher-ence of mononuclear leukocytes to/the luminal sur-face of the vessel (rarely seen), or more commonlyassociated with infiltration of the immediately adja-cent tissues by mononuclear leukocytes (Figure 2, Ato C). Immunophenotypic studies of these leuko-cytes showed that virtually all could be identified as

Figure 1. A: Demotnstration ofprominienit reactivity of tonisillarfollicilar denidritic cells with the rabbit polyclonal antibody raised against recom-binianit human VCAM-1. B: Reactivity ofantiserum is abolished afterserum is absorbed against VCAM-1 tranisfected CHO cells. Reactivity ofthe serumis niot abolished after absorptioni againest ICA 1-1 or ELAM-1-transfected CHO cells or nontransfected CI10 cells (data not shoun). The Avidin-biotinimmunoperoxidase technique was uised, tith methyl green counterstain. MtagnYiication X .370

M-.- 4 A. I 1- ... I

VCAM-1 in Liver and Pancreas Allografts 583AJP Febnrary 1993, Vol. 142, No. 2

Table 1. VCAM-1 in Solid Organ Allografts

VCAM-1 expression*Sinusoidal Portal vein Dendritic cells

Diagnosis n endothelium endothelium (lymphoid aggregates) Arteritis

Liver transplantsNormal liver 2 - -Donor biopsies 17Protocol biopsies (no rejection) 5 -

Acute rejectiont 14 8/14 10/14 4/14Post rejection 13 - - 3/14§

Pancreas transplantsAcute rejectiont 6 6 2/2Chronic vascular rejection (autopsy) 1No evidence of rejection 4

Detectable by rabbit polyclonal antisera to recombinant human VCAM-1.t Only 2 of 6 cases had arteritis.t Two of 14 cases of acute rejection were completely negative for VCAM-1 expression.§ One in three cases demonstrated de novo viral hepatitis.

belonging to T-lymphocyte (CD43+) or monocyte/macrophage (CD68+) lineage (Figure 2, B and C) butnot B-lymphocyte (CD20+) lineage. Further immuno-labeling with the antibody to PCNA showed at timessignificant numbers of these infiltrating leukocytesactively proliferating (Figure 2D).A second pattern of VCAM-1 expression in liver

allografts was that of focal, irregularly distributedexpression on sinusoidal endothelium, seen in 8 of14 cases of allograft rejection (Figure 2F). Unlike thepattern seen with portal vein endothelium, areas ofVCAM-1 expression were usually not associated withthe infiltration of adjuvant tissues by leukocytes.However, although VCAM-1 expression at this sitecould apparently be disassociated from concomitantlocalized inflammatory injury, it is emphasized thatthis pattern of expression was encountered only inthose allograft biopsies that had unequivocal fea-tures of rejection, and hence inflammatory cell infil-tration, identified in other sites within the hepaticparenchyma.The third pattern of VCAM-1 expression was the

finding of a population of cells with dendritic mor-

phology within portal triads that showed prominentexpression of NGFR and somewhat less widespreadexpression of VCAM-1. By both morphologic appear-

ance and this immunohistochemical characteriza-tion, these cells could be identified as dendritic cellssimilar to those found in both lymph nodes and tonsil.These cells were always found within prominentaggregates of lymphoid cells, which by immunophe-notypic characterization could be shown to be pre-

dominately T cells (CD43+) with only small numbersof B cells (CD20+). The finding of dendritic cells amidthe portal lymphocyte aggregates was present in 4 of14 cases of rejection. VCAM-1 expressing dendriticcells was found in only a single prerejection liverbiopsy obtained 1 hour after achieving anastomosis;

these cells were identified in a portal tract with aprominent accumulation of leukocytes. Both the por-tal inflammation and presence of dendritic cells inthis patient are of uncertain significance; in no otherbiopsy before rejection was this pattern identified.

In all but three biopsies obtained after clinical res-olution of rejection, no residual expression ofVCAM-1 could be identified. One of these three,obtained 3 months after the rejection episode,showed evidence of a newly diagnosed viralhepatitis. Two biopsies showed persistent portal lym-phocytic infiltrates without other features diagnosticof rejection; within these infiltrates was a persistentpopulation of NGFR+, VCAM-1+ dendritic cells.No examples of acute or chronic arteritis were

present in this series.

PancreasOf 10 pancreatic allograft biopsies from five patients,six showed features of acute rejection according toconventional criteria,e 10 with two of these showingadditional features of an acute arteritis. All six casesof acute rejection showed the presence of lymphoidaggregates within the interstitial tissue. As in the liverallografts, such infiltrates invariably could be char-acterized as predominately composed of T-lympho-cytes (CD43+) by immunohistochemical analysis,and within these aggregates a population of NGFR+and, less commonly, VCAM-1,+ dendritic cells waspresent (Figure 3, A-D). Examples of arteritis, inwhich the persistent presence of the endothelial lin-ing could be demonstrated by immunohistochemicallocalization of the endothelium specific marker factorVIII-related antigen/von Willebrand factor, showedunequivocal endothelial expression of VCAM-1 (Fig-ure 3, E and F). The subendothelial space underlyingthis VCAM-1+ endothelium contained numerous

584 Bacchi et alAJP Febnuary 1993, Vol. 142, No. 2

Figure 2. A: Acute cellular rejection of liver Florid endothelialitis is present in a branch ofa portal vein (PV). (Hematoxryin-cosin; magniificationlX 720.) B: Acute cellular rejection ofliver showingflorid endothelialitis involving a branch ofa portal vein (PV). 7Tere is subendothelial inifiltrationof Tcells expressing CD43 antigen. (Magnification X 720.) C: Acute cellular rejection oflivershowingflorid enzdothelialitis involving portal vein (PV).Immunostaining of CD68-expressing monocytes/macrophages shows nuimerous cells in subendothelial location. (Magniificationi X 720.) D: Aciutecellular rejection of liver. Presence of PCNA-positive immtune cells in the wall of a portal veint (PV) branch uith endothelialitis indicates activereplication of these cells at sites of rejectioni inijuiry. (Magnzification X 720.) E: VCAM-I expression by endotheblial cells is present in a portal vein (Pl')with endothelialitis. (Magnification X370.) F: VCAM-J expression by etndothelial cells linzinig sinutisoidal spaces in a case ofacute celliular rejectioninvolving the liver. (Magnification X3 70.) G: VCAM-I expression by denzdritic cells within a lymphoid aggregate in a liver allograft with rejectioni(Magnification X370.) ForB to G the avidini-biotin immtnoperoxidase technique was uised. For B and C, hematoxylinz counterstain uas uised. ForD to G, methyl green counterstain uas tused.

VCAM-1 in Liver and Pancreas Allografts 585AJP Febnrary 1993, Vol. 142, No. 2

4..

A

Figure 3. A: Follictulardentdritic cells ini agerminal ceniterofa tonsil arestroniglyreactive uith a motoclonialanitibody to nervegrou'thfactorreceptor(.VGFR-5). (MagniVication X 185.) B: VCAM-1 shons similar expression byfollicular denidritic cells in germinal centers of tontsils. (MagnlificationlX 185.) C: Lymphoid aggregate in a panicreatic allograft tissne uhere cellular rejectioni was presetnt. Nnimeronis denidritic cells are strongglyimmnnonostained byG,VFR-5 monoclonal antibody. (Magnification X 720.) D: Lymwphoid aggregate in the samepancreatic allograft tiss2eas illustratedin Figure 1 C uith dendritic cells expressing VCAM-1. There arefewer VICAVM1-expressing cells conipared uith cells expressing NGFR, blit distriblntionloJpositnive cells is simiiilar (Mllagnification x 720.) E: Medium-sized artery ofa palcreatic allograft tissne uwith florid enldothelialitis. Enldothelial cellsare expressing VCAM- I in their limninial snirface (arrouw). (Magniificationt X 185.) F: Detail of the samne field of the Figure 1 E. Endothelial cellsexpressing VCAM-1. (Magniificationi X 720.) G: Panicreatic allograft tissute uith vascular rejection (arteriopathY). NVnimeronts lymphocytes inz thesutbenzdothbclial area anid in the unall of the artery express the T-cell-associated marker CD43. (Magnification X370.) H: Pantcreatic allograft tissnenlith vascular rejection (arteriopathy). Numerous monocytes1macrophagecs expressing the CD68 anitigeni arepresent in a suhbenldothelial locationl anldin the wall ofthe artery. (Magnification X370.) In all, avidin-biotin immuntioperoxidase preparatiotn nas tused. ForA to F, methyl greeni counterstainwas i(csd. For G anid H, heniatoxylin coiunterstain was uised.

586 Bacchi et alAJP February 1993, Vol. 142, No. 2

mononuclear leukocytes, which could be identifiedas T cells (CD43+) and monocytes/macrophages(CD68+) (Figure 3, G and H).One pancreatic allograft studied at autopsy con-

tained areas of chronic arteriopathy identical to thechanges of chronic vascular rejection seen in someallograft kidneys13; VCAM-1 expression could notbe identified in this case. The remaining parenchymain this case was obliterated by neoplastic cellscomprising a post-transplant lymphoproliferativedisorder. In the four pancreatic allograft biopsieswithout evidence of rejection, no expression ofVCAM-1 could be identified.

Discussion

Two principal findings of this study link inducedexpression of VCAM-1 to active rejection within liverand pancreas allografts. The first of these is thatVCAM-1 expression on vascular and sinusoidalendothelium can be identified in tissue at the time ofacute rejection, and can be localized to areas ofinfiltrating T cells and monocytes/macrophages. Asshown in these studies, there is no detectable con-

stitutive endothelial expression of VCAM-1 withinsubsequently affected organs when protocol biop-sies obtained at the time of organ harvest or trans-plantation are studied. Furthermore, organs in whichacute rejection and concomitant VCAM-1 expressioncan be demonstrated in general no longer showdetectable VCAM-1 expression after rejection epi-sodes have been successfully treated and follow-upbiopsies reveal no residual features of active inflam-matory injury. Our study of the immunophenotype ofthe infiltrating leukocytes at sites of VCAM-1 expres-

sion by endothelial cells shows that virtually all ofthese cells are T-lymphocytes or monocytes or

macrophages. These cells are known to constitu-tively express the very late activation-4 (VLA-4) inte-grin (CD49d) on their cell surface, and hence wouldbe among the cell populations adhering to cells withsurface VCAM-1 expression. These studies stronglyimplicate, but do not prove, a role for VCAM-1 inattracting specific leukocyte populations to paren-

chymal sites of rejection.However, to prove that VCAM-1 has a primary role

in mediating rejection, it is necessary to exclude thealternate possibility that the up-regulated expressionof VCAM-1 is a result of, rather than a contributoryfactor to, active rejection. In this scenario, it is theinfiltrating leukocytes or damaged parenchymal cellsthat provide signals necessary to induce VCAM-1expression. Distinction between these two possibili-ties cannot be accomplished by relatively static

biopsy studies such as this one, but may becomeamenable to analysis like that employed in recentstudies in primates and in mice on the role of ICAM-1in promoting allograft rejection.4243 Those studiesutilized the effect of functional blocking antibodies toICAM-1 administered in vivo to help establish a pri-mary role for this leukocyte adhesion molecule in therejection process.42'43 The functional activity ofVCAM-1 has been studied in a murine model of car-diac transplantation. In that model, rejecting cardiacallografts demonstrated increased VCAM-1 expres-sion on endothelium. When an anti-VCAM antibody,not otherwise characterized as to its functional abilityto inhibit leukocyte adhesion to endothelium, wasadministered to recipients of cardiac allografts, leu-kocyte infiltration of the allografts was reportedly notinhibited.44 Clearly, further studies will be required toassess the extent to which VCAM-1 may be neces-sary in mediating leukocytic injury solid organtransplantation.

The second principal finding is the identification ofprominent numbers of VCAM-1 expressing dendriticcells within the lymphoid aggregates that can beidentified in some solid organ allografts at the time ofrejection. VCAM-1 expression on both follicular den-dritic cells and the interdigitating cells in T-cell-dependent regions of lymph nodes has been previ-ously described.31 4546 We were able to replicatethese findings with each of the antisera to VCAM-1utilized in this study. We were further able to show byimmunohistochemical studies on replicate tissuesections of lymphoid tissues that the follicular den-dritic cells that express VCAM-1 also expressNGFR-5, which has been shown previously to beuseful phenotypic marker of such cells.34'35 NGFR-5is not identifiable in the normal hepatic or pancreaticparenchyma, and so we were able to utilize itsexpression as an independent marker to furtheridentify and confirm the presence of a transientVCAM-1 expressing population of dendritic cells inthe transplanted organs in the time of rejection. In theB-cell-dependent areas (germinal centers) of lymphnodes and tonsils, there is generally good correlationbetween the expression of VCAM-1 and NGFR-5dendritic cells, as assessed both by staining inten-sity and by distribution of cells reactive with theseantisera. However, our studies of dendritic cellswithin allografted organ shows the population ofNGFR-5-expressing cells within lymphoid aggre-gates to be somewhat greater than that of VCAM-1-expressing cells. Taken together, these data suggestthat there is a population of follicular dendritic cellspresent within some rejecting solid organ allografts,but that only a subpopulation of these cellsexpresses detectable levels of VCAM-1.

VCAM-1 in Liver and Pancreas Allografts 587AJP February 1993, Vol. 142, No. 2

Our studies of transplanted organ tissue both pre-ceding and following episodes of rejection have notrevealed a stable population of NGFR-5- or VCAM-1-expressing dendritic cells that are present in theabsence of inflammatory injury. It would appear sucha cell population either migrates from the host lym-phoid tissues to the allograft at times of rejection,where these cells presumably help present antigento immunocompetent cells as part of the rejectionresponse, or, if constitutively present in donororgans, undergo phenotypic modulation fromNGFR-, VCAM- to a presumably more active statewhere these proteins are expressed. In either sce-nario, the finding of such cells in lymphoid aggre-gates and allografts, as in reactive lymphoid tissue,is indicative of a role for these cells in mediating therejection response.

These findings are of interest in view of recentstudies in the mouse that have demonstrated themigratory nature of dendritic cells and demonstratedtheir role in initiating an allostimulatory response.47These studies suggest that dendritic cells compriseat least part of a population of passenger leukocytesthat are transplanted within the donor organ.48 Aftertransplantation the cells migrate out of the donatedorgan to central lymphoid organs of the host, such asthe spleen, where they serve as the sensitizing stim-ulus for activation of host immunocompetent cells.The finding of dendritic cells at sites of organizedlymphocytic responses in human organ allograftsprovides evidence that their role in recruitment andstimulation of activated lymphocytes is likely to takeplace, at least in part, locally within the graft ratherthan centrally within the host's lymphoid organs.

Studies by Freedman et a145 have shown thatVCAM-1 expression by follicular dendritic cells maypromote adhesion of lymphoid cells, and hence mayhelp further localize immunocompetent cells to sitesof immune injury. It seems likely that the dendritic cellcontaining lymphoid follicles encountered in organallografts are an in vivo correlate to these ex vivostudies. Finally, studies by several groups have dem-onstrated that VCAM-1 interactions with lymphocytesoccurring via binding with the VLA-4 counter recep-tor may result not just in enhanced leukocyte adhe-sion, but may provide additional costimulatory acti-vation of resting T cells beyond that provided byantigen and mixed histocompatibility complex(MHC) class 11 molecules and so may amplify orpromote allograft rejection responses.49-51 Thesestudies suggest another mechanism by whichVCAM-1+ dendritic cells may contribute to liver andpancreas transplant rejection.

Previous studies of both human and experimentalrat liver transplantation have suggested a role for

dendritic cells in mediating the rejection response. Instudies of arteries with features of chronic vascularrejection (alternately termed obliterative arteriopa-thy) in man, Oguma et a11 have identified a popu-lation of dendritic cells recognized by their reactivitywith antibodies to the S-100 protein. That study didnot further characterize the nature of the dendriticcells, nor did it address whether dendritic cells werepresent in the hepatic parenchyma. Within the rat,populations of la+ dendritic-appearing cells havebeen identified with hepatic and pancreatic intersti-tial tissue, but such cells have not been well charac-terized and their distinction from tissue macro-phages is not well established.52 Demetris et a153observed close temporal and spatial clustering ofinfiltrating T-lymphocytes and la+ dendritic-shapedcells in early rat liver allograft rejection, and alsonoted the presence of donor la+ cells in recipientspleen 1 day after transplantation. Their findings pro-vide evidence of both central (host lymphoid system)and local (intragraft) mechanisms of sensitizingimmune cells in this model of graft rejection. None ofthese studies have addressed issues of VCAM-1expression by dendritic cells or that of phenotypicmodulation by dendritic cells. Although we know ofno data that addresses this last issue directly, webelieve it is an important area for future study.Although we cannot currently exclude the possibilitythat a population of dendritic cells may normally existwithin the human pancreas or liver, we and oth-ers31,54 provide evidence that a population ofVCAM+ dendritic cells cannot typically be identifiedin the interstitial tissues of the normal liver, althoughexpression by normal Kupffer cells has been identi-fied in one study.31

Our findings also point to a peculiar paradox in thenature of the dendritic cells that we have identifiedwithin rejecting organ allografts. As revealed byimmunohistochemical studies on human lymphoidtissue, VCAM-1 can be shown to be expressed onboth the follicular dendritic cells present in B-cell-dependent portions of lymphoid tissues, as well ason the interdigitating dendritic cells that are presentin T-cell-dependent areas. It is currently thought thatdespite somewhat similar terminology and morpho-logic appearance, these represent very distinct celltypes. Follicular dendritic cells are thought to be ofnonhematopoietic origin, bind B-lymphocytes, andpromote B-lymphocyte proliferation.46 Interdigitatingdendritic cells are thought to be of hematopoieticorigin, related to monocyte lineage and to Langer-hans cells of the skin, and principally promote T-lym-phocyte activation and proliferation. Immunohis-tochemical staining for the nerve growth factor

588 Bacchi et al/P February 1993, VW!. 142, No. 2

receptor shows that its expression is confined to thefollicular dendritic cells of the B-cell-containing ger-minal centers.34'35 Yet we have observed that withinthe lymphoid aggregates present in rejecting trans-plants, the dendritic cells identified strongly expressnerve growth factor receptor, suggestive of a follicu-lar dendritic cell phenotype. However, the surround-ing lymphoid infiltrate immunophenotypes as a pop-ulation of predominately T-lymphocytes ratherB-lymphocytes. These findings are not explainableat the present time, given the still limited knowledgeavailable about human dendritic cells. They do sug-gest that rigid characterization of interactionsbetween specific types of dendritic cells and someclasses of lymphocytes may be an unduly limitingconceptualization of how immune cells may bestimulated.We must also point out certain findings in this

study that limit our ability to assess the role ofVCAM-1 in liver and pancreas rejection. The expres-sion of this molecule by sinusoidal and venousendothelium was only seen within focal areas of thebiopsies. In no case was VCAM-1 expression seen tobe widespread. Although this may be due to thesensitivity of the antibodies employed, this findingmight instead reflect the fact that VCAM-1 expres-sion may be a dynamic and rather transient processthat is only intermittently detectable by biopsystudies. It also seems likely that a number of cyto-kines and cell-stimulatory molecules may direct localendothelial expression of VCAM-1, and that furtherknowledge of these kinds of stimuli will be requiredbefore we can understand clearly the basis for thefocality of both VCAM-1 expression and of the rejec-tion response.A second area of concern is the basis for the

discrepancy in immunohistochemical findingsbetween the rabbit polyclonal antisera and themurine monoclonal antibody, both of which aredirected against human VCAM-1. Both antibodiesshow similar degrees of sensitivity in detecting den-dritic cells and other structures known to expressVCAM-1 (eg, kidney parietal epithelial cells) in con-trol tissues fixed in a manner identical to the those ofthe transplant biopsies. Yet it is clear that the epitoperecognized by the rabbit polyclonal antisera is notreliably detected by the 4B9 monoclonal antibodywhen the VCAM-1 molecule is expressed onendothelium. We believe the rabbit polyclonal antis-era is a more sensitive reagent, and we are confidentof its specificity for the following reasons: 1) the anti-sera was raised against purified recombinant humanVCAM-1, 2) the antisera was affinity purified andretained all of its reactivity in transplant biopsies as

described above, and 3) the specific staining pat-terns described were abolished when the antiserawas absorbed against VCAM-1 expressing trans-fected cell lines, but not when the antisera wasabsorbed against similar cell lines either transfectedwith ICAM-1 or untransfected. It remains possiblethat the 4B9 monoclonal antibody recognizes anepitope of the VCAM-1 molecule inaccessible to theantibody in our tissue immunohistochemical studies.It is even possible that 4B9 recognizes a VCAM-1epitope not biologically expressed in this transplan-tation setting. While speculative, this last consider-ation derives support from recent observations thatalternatively spliced forms of VCAM-1 can beproduced55 that have different binding sites withrespect to various VCAM-1 antibodies that havecurrently been developed (L. Osborn, personalcommunication).

Finally, although our study demonstrates that up-regulated VCAM-1 expression in specific tissue sitesis associated with the allograft rejection response, itdoes not address the issue whether VCAM-1 medi-ated leukocyte adhesion is essential for such aresponse to develop. Other proinflammatory pro-cesses are certainly involved, and up-regulation ofother leukocyte adhesion molecules, such ICAM-1,which has previously been demonstrated in hepaticallografts and inflammation, may also be critical tothe infiltration of leukocytes into tissues in solid organtransplant rejection. 54'56,57

In summary, while cognizant of the above caveats,we have demonstrated up-regulated expression ofVCAM-1 within liver and pancreas organ allografts attimes of rejection. The sequence and localization ofVCAM-1 expression by endothelial and dendriticcells suggest that VCAM-1 has an important role inrecruitment, localization, and activation of hostimmunocompetent cells at sites of the rejectionresponse.

AcknowledgmentsWe thank Phyllis Davie, Liz Donato, Janice Morihara,and Tracie Evans for performing the immunohis-tochemical studies. We are grateful to Marina Fergu-son, Tom McDonald, Elaine Yamanaka, Irene Dou-gas, and Gloria Chi-Rosso for providing experttechnical assistance. Melinda Ogilvie providedexpert secretarial assistance.

References1. Snover DC, Sibley RK, Freese DK, Sharp HL, Bloomer

JR, Najarian JS, Ascher NL: Orthotopic liver transplan-tation: a pathological study of 63 serial liver biopsies

VCAM-1 in Liver and Pancreas Allografts 589AJP February 1993, Vol. 142, No. 2

from 17 patients with special reference to the diagnos-tic features and natural history of rejection. Hepatology1984, 4:1212-1222

2. Demetris AJ, Lasky S, Van Thiel DH, Starzl TE, DekkerA: Pathology of hepatic transplantation: A review of 62adult allograft recipients immunosuppressed with acyclosporine/steroid regimen. Am J Pathol 1985,118:151-161

3. Sankary H, Foster P, Hart M, Ashmann M, Schwartz D,Williams JW: An analysis of the determinants of hepaticallograft rejection using stepwise logistic regression.Transplantation 1989, 47:74-77

4. Demetris AJ, Qian S, Sun H, Fung JJ: Liver allograftrejection: An overview of morphologic findings. Am JSurg Pathol 1990, 14(Suppl 1):49-63

5. Hubscher SG: Histological findings in liver allograftrejection-new insights into the pathogenesis of hepato-cellular damage in liver allografts. Histopathology1991, 18:377-383

6. Snover DC: Liver transplantation. In The Pathology ofTransplantation. Edited by GE Sale. Boston, Butter-worths, 1990, pp 103-132

7. Ludwig J, Batts KP, Ploch M, Rakela J, Perkins JD,Wiesner RH: Endothelialitis in hepatic allografts. MayoClin Proc 1989, 64:545-554

8. Sibley RK, Sutherland DER: Pancreas transplantation.An immunohistologic and histopathologic examinationof 100 grafts. Am J Pathol 1987, 128:151-170

9. Carpenter HA, Engen DE, Munn SR, Barr D, Marsh CL,Ludwig J, Perkins JD: Histologic diagnosis of rejectionby using cystocopically directed needle biopsy speci-mens from dysfunctional pancreatoduodenal allograftswith exocrine drainage into the bladder. Am J SurgPathol 1990, 14:837-846

10. Sibley RK: Pancreas transplantation. In The Pathologyof Transplantation. Edited by GE Sale. Boston, Butter-worths, 1990, pp 179-215

11. Oguma S, Zerbe T, Banner B, Belle S, Starzl TE, Dem-etris AJ: Chronic liver allograft rejection and obliterativearteriopathy: possible pathogenetic mechanisms:Transplantation Proc 1989, 21:2203-2207

12. Hruban RH, Beschorner WE, Baumgartner WA, Augus-tine SM, Ren H, Reitz BA, Hutchins GM: Acceleratedarteriosclerosis in heart transplant recipients is associ-ated with a T-lymphocyte-mediated endothelialitis. AmJ Pathol 1990, 137:871-882

13. Alpers CE, Gordon D, Gown AM. Immunophenotype ofvascular rejection in renal transplants. Modern Pathol1990, 3:198-203

14. Bishop GA, Waugh JA, Landers DV, Krensky AM, HallBM: Microvascular destruction in renal transplantrejection. Transplantation 1989, 48:408-414

15. Renkonen R, Turunen JP, Rapola J, Hayry P: Charac-terization of high endothelial-like properties of peritu-bular capillary endothelium during acute renal allograftrejection. Am J Pathol 1990, 137:643-651

16. Bishop GA, Hall BM: Expression of leukocyte and lym-phocyte adhesion molecules in the human kidney. Kid-

ney Int 1989, 36:1078-108517. Bevilacqua MP, Stengelin S, Gimbrone MA Jr, Seed B:

Endothelial leukocyte adhesion molecule 1: An induc-ible receptor for neutrophils related to complement reg-ulatory proteins and lectins. Science 1989, 243:1160-1165

18. Johnston GI, Cook RG, McEver RP: Cloning of GMP-140, a granule membrane protein of platelets andendothelium: sequence similarity to proteins involved incell adhesion and inflammation. Cell 1989, 56:1033-1044

19. Simmons DE, Makgoda MW, Seed B: ICAM, an adhe-sion ligand of LFA-1 is homologous to neural cell adhe-sion molecule NCAM. Nature 1988, 133:624-627

20. Staunton DE, Marlin SD, Stratowa C, Dustin ML,Springer TA: Primary structure of ICAM-1 demonstratesinteraction between members of the immunoglobulinand integrin supergene families. Cell 1988, 52:925-933

21. Osborn L, Hession C, Tizard R, Vassallo C, LuhowskyJS, Chi-Rosso G, Lobb R: Direct expression cloning ofvascular cell adhesion molecule 1, a cytokine-inducedendothelial protein that binds to lymphocytes. Cell1989, 59:1203-1211

22. Bevilacqua MP, Pober JS, Mendrick DL, Cotran RS,Gimbone MA Jr: Identification of an inducible endothe-lial-leukocyte adhesion molecule. Proc Natl Acad SciUSA 1987, 84:9238-9242

23. Geng J-G, Bevilacqua MP, Moore KL, Mcintyre TM,Prescott SM, Kim JM, Bliss GA, Zimmerman GA,McEver RP: Rapid neutrophil adhesion to activatedendothelium mediated by GMP-140. Nature 1990,343:757-760

24. Staunton DE, Dustin ML, Springer TA. Functional clon-ing of ICAM-2, a cell adhesion ligand for LFA-1 homol-ogous to ICAM-1. Nature 1989, 339:61-64

25. Carlos TM, Schwartz BR, Kovach NL, et al: Vascularcell adhesion molecule-1 mediates lymphocyte adher-ence to cytokine-activated cultured human endothelialcells. Blood 1990, 76:965-970

26. van Dinther-Janssen ACHM, Horst E, Koopman G,Newmann W, Scheper RJ, Meijer CJLM, Pals ST: TheVLA-4NCAM-1 pathway is involved in lymphocyteadhesion to endothelium in rheumatoid synovium. JImmunol 1991, 147:4207-4210

27. Briscoe DM, Schoen FJ, Rice GE, Bevilacqua MP, GanzP, Pober JS: Induced expression of endothelial-leuko-cyte adhesion molecules in human cardiac allograft.Transplantation 1991, 51 :537-539

28. Carlos TM, Gordon D, Fishbein D, Himes V, Coday A,Ross R, Allen MD: Vascular cell adhesion molecule-1 isinduced on endothelium during acute rejection inhuman cardiac allografts. J Heart Lung Transplant1992, 11:1103-1109

29. Hsu SM, Raine L, Fanger H: Use of avidin-biotin per-oxidase complex (ABC) in immunoperoxidase tech-niques: A comparison between ABC and unlabeled(PAP) procedures. J Histochem Cytochem 1981,29:577-580

590 Bacchi et alAJP February 1993, Vol. 142, No. 2

30. Gown AM, Vogel AM: Monoclonal antibodies to inter-mediate filaments proteins of human cells. ll. Distribu-tion in normal human tissues. Am J Pathol 1984,114:309-321

31. Rice GE, Munro JM, Corless C, Bevilacqua MP: Vas-cular and nonvascular expression of INCAM-1 10. Am JPathol 1991, 138:385-393

32. Carlos T, Kovach N, Schwartz B, Rosa M, Newman B,Wayner E, Benjamin C, Osborn L, Lobb R, Harlan J:Human monocytes bind to two cytokine-induced adhe-sive ligands on cultured human endothelial cells:endothelial-leukocyte adhesion molecule-1 and vascu-lar cell adhesion molecule-1. Blood 1991, 77:2266-2271

33. Lobb R, Chi-Rosso G, Leone D, Rosa M, Newman B,Luhowskyj S, Osborn L, Schiffer S, Benjamin C, Dou-gas 1, Hession C, Chow P: Expression and functionalcharacterization of a soluble form of vascular cell adhe-sion molecule 1. Biochem Biophys Res Commun 1991,178:1498-1504

34. Thompson SJ, Schatteman GC, Gown AM, Bothwell M:A monoclonal antibody against nerve growth factorreceptor. Am J Clin Pathol 1989, 92:415-423

35. Strobach RS, Nakamine H, Masih AS, Linder J, Weisen-burger DD: Nerve growth factor receptor expression ondendritic reticulum cells in follicular lymphoidproliferations. Hum Pathol 1991, 22:481-485

36. Warnke RA, Pulford KAF, Pallesen G, Ralfkiaer E, BrownDC, Gatter KC, Mason DY: Diagnosis of myelomono-cytic and macrophage neoplasms in routinely pro-cessed tissue biopsies with monoclonal antibody KP1.Am J Pathol 1989, 135:1089-1095

37. Said JW, Stoll PN, Shintaku P, Bindl JM, Butmare JR,Pinkus GS: Leu-22: A preferential marker for T-lympho-cytes in paraffin sections. Staining profile in T- andB-cell lymphomas, Hodgkin's disease, other lympho-proliferative disorders, myeloproliferative diseases,myeloproliferative diseases and various neoplasticprocesses. Am J Clin Pathol 1989, 91:542-549

38. Norton AJ, Isaacson PG: Monoclonal antibody L26: Anantibody that is reactive with normal and neoplastic Blymphocytes in routinely fixed and paraffin wax embed-ded tissues. J Clin Pathol 1987, 40:1405-1412

39. Mason DY, Comans-Bitter WM, Cordell JL, VerhoevenMJ, van Dongen JJM: Antibody L26 recognizes anintracellular epitope on the B-cell-associated CD20antigen. Am J Pathol 1990, 136:1215-1222

40. Turner RR, Beckstead JH, Warnke RA, Wood GS:Endothelial cell phenotypic diversity. Am J Clin Pathol1987, 87:569-575

41. Johnson RJ, Garcia RL, Pritzl P, Alpers CE: Plateletsmediate glomerular cell proliferation in immune com-plex nephritis induced by anti-mesangial cell antibod-ies in the rat. Am J Pathol 1990, 136:369-374

42. Cosimi AB, Conti D, Delmonico FL, Preffer Fl, Wee S-L,Rothlein R, Faanes R, Colvin RB: In vivo effects of mon-oclonal antibody to ICAM-1 (CD54) in nonhuman pri-mates with renal allografts. J Immunol 1990,144:4604-4612

43. Isobe M, Yagita H, Okumura K, Ihara A: Specific accep-tance of cardiac allograft after treatment with antibod-ies to ICAM-1 and LFA-1. Science 1992, 255:1125-1127

44. Orosz CG, van Buskirk A, Sedmak DD, Kincade P, Miy-ake K, Pelletier RP: Role of the endothelial adhesionmolecule VCAM in murine cardiac allograft rejection.Immunol Lett 1992, 32:7-12

45. Freedman AS, Munro JM, Rice GE, Bevilacqua MP,Morimoto C, Mcintyre BW, Rhynhart K, Pober JS,Nadler LM: Adhesion of human B cells to germinalcenters in vitro involves VLA-4 and INCAM-110. Sci-ence 1990, 249:1030-1033

46. Clark EA, Grabstein KH, Shu GL: Cultured human fol-licular dendritic cells: Growth characteristics and inter-actions with B lymphocytes. J Immunol 1992,148:3327-3335

47. Larsen CP, Morris PJ, Austyn JM: Migration of dendriticleukocytes from cardiac allografts into host spleens: Anovel pathway for initiation of rejection. J Exp Med1990, 171:307-314

48. Austyn JM, Larsen CP: Maturation, migration and func-tion of dendritic leukocytes after transplantation. InDendritic cells in lymphoid tissues. Edited by Y Imai,JG Tew, and ECM Hoefsmit. New York, Elsevier SciencePublishers, 1991, pp 209-215

49. van Seventer GA, Newman W, Shimizu Y, Nutman TB,Tanaka Y, Horgan KJ, Gopal TV, Ennis E, O'Sullivan D,Grey H, Shaw S: Analysis of T cell stimulation by super-antigen plus major histocompatibility complex class 11molecules or by CD3 monoclonal antibody: costimula-tion by purufied adhesion ligands VCAM-1, ICAM-1,but not ELAM-1. J Exp Med 1991, 174:901-913

50. Damle NK, Aruffo A: Vascular cell adhesion molecule 1induces T-cell antigen receptor-dependent activation ofCD4+T lymphocytes. Proc Natl Acad Sci USA 1991,88:6403-6407

51. Burkly LC, Jakubowski A, Newman BM, Rosa MD, Chi-Rosso G, Lobb RR: Signaling by vascular cell adhesionmolecule-1 (VCAM-1) through VLA-4 promotes CD3-dependent T cell proliferation. Eur J Immunol 1991,21:2871-2875

52. Steiniger B, Klempnauer J, Wonigeit: Phenotype andhistological distribution of interstitial dendritic cells inthe rat pancreas, liver, heart, and kidney. Transplanta-tion 1984, 38:169-175

53. Demetris AJ, Qian S, Sun H, Fung JJ, Yagihashi A,Murase N, Iwaki Y, Gambrell B, Starzl TE: Early eventsin liver allograft rejection. Am J Pathol 1991, 138:609-618

VCAM-1 in Liver and Pancreas Allografts 591AJP February 1993, Vol. 142, No. 2

54. Volpes R, van den Oord JJ, Desmet VJ: Vascular adhe-sion molecules in acute and chronic liver inflammation.Hepatology 1991, 15:269-275

55. Cybulsky Ml, Fries JWU, Williams AJ, Sultan P, DavisVM, Gimbrone MA Jr, Collins T: Alternative splicing ofhuman VCAM-1 in activated vascular endothelium. AmJ Pathol 1991, 138:815-820

56. Adams DH, Shaw J, Hubscher SG, Rothlein R, Neu-berger JM: Intercellular adhesion molecule 1 on liverallografts during rejection. Lancet 1989, 11:1122-1125

57. Steinhoff G, Behrend M, Pichlmayr R: Induction ofICAM-1 on hepatocyte membranes during liverallograft rejection and infection. Transplant Proc 1990,22:2308-2309