Human T-cell-mediated destruction of allogeneic dermal microvessels in a severe combined immunodeficient mouse

Proc.Natd. Acad. Sci. USA

Vol.91,pp.9146-9150, September1994 Immunology

Human T-cell-mediated destruction of allogeneic dermal microvessels in a severe combined immunodeficient mouse

(heterologousgraft/graftrejection/vascularendothelialcells/cytotoicTlymphocyte/adhesionmolecules)

ALLAN G. MURRAY*t4, PETER PETZELBAUER*t§1, CHRISTOPHER C.

W.

HUGHES*t,

JOSE

COSTAt,

PHILIP

ASKENASEt

II, AND

JORDAN

S.

POBER*t§**

*MolecularCardiobiology Program, The Boyer Center for MolecularMedicine, and Departments oftPathology,

I'Medicine,

and§Immunobiology,Yale University School of Medicine, New Haven, CT 06510

Communicated byVincent T.Marchesi, May 18, 1994

ABSTRACT Wedevelopedachimerichuman-severecom- binedimmunodeicentmousemodeltostudy human

allograft

rejection. Mice received first

partial

esshumanskingrafts and then, after anomosis ofthe mouse with

graft

human microvessels,human

lymphocytes alogeneic

totheskin.By 2 weeks, theskingraftsuniformlydeveloped

changes

thatresem- blefirst-set skin rejection in humans. Vascular cell adhesion molecule 1andmajor

histocompatibility

complex class II mol- ecules wereinducedonthe human vascularendotheliumatday 6, prior to signint T-cefl itr . Perivascular human CD4+andCD8+ T-cellInfiltratesweremarkedby day11.Some Tcells, adjacent to injured human vessels,expressed thecy- tolytic granule proteinperforin.The humanmicrovesselswere destroyed by day 16 without sgifcant damage to human keratinocytes oradjacent mouse microvessels. This small animal modelmaypermitevaluationofpotentialtherapeuticreagents thatinhibit human T-cell-mediatedinjury.

Microvascular endothelial cells(ECs)arethoughtto bethe primary targets of acute

allogeneic rejection

reactions in vascularizedallografts (1-3). Consequently, manytherapeu- ticstrategiesarebasedontheconceptofdisruptinginterac- tions between hostTcells andgraftECs.However,manyof themoleculesthatmediate such interactionsare

sufficiently

different between humans and rodents that it has been necessary to useprimates for theinitialscreening of inhibi- tory reagents. The human

peripheral

blood lymphocyte

(PBL)-severe

combined immunodeficient(SCID)mouse was introducedas asmallanimal model for

studying

the human immune system in vivo(4). C.B-17 micehomozygous forthe SCIDmutationareextremely inefficientat

rearranging

T-and B-cell antigen receptor genes (5, 6), and although natural killer (NK) cell function is intact, these T- and B-cell- deficient animals are unabletomediate rejection of human xenogeneic grafts. Upon introduction ofhumanPBLs, such micecanrespondtoantigen bysynthesizinghumanantibod- ies (4, 7-10). However, there has been little evidence that T-cell-mediated inflammatory reactionscandevelop in these animals(11-13), and human Tcells may becomeanergicin the SCIDmouse microenvironment(14).

Wereasoned thatamajorpartofthedifficultyin reconsti- tutingT-cellinflammatory functionin the humanPBL-SCID mousemay arise becausemurine ECs lack critical human cell surface proteins necessary for appropriaterecruitment and activation ofhumanlymphocytes. Wegraftedhumanskin with anintact dermalmicrovascularbed ontoaSCID mouseprior toreconstitutingthemousewithallogeneichuman PBLs. Our model results inareproducible stereotypicreaction in which human T cells infiltrate and destroy the human dermal mi- crovessels, resemblingfirstsetskingraft rejection.

MATERIALS AND METHODS

Animals. C.B-17 SCID mice (Harlan-Sprague-Dawley)

were used atages 5-8 weeks. Theanimalswere housed in microisolatorcagesandwerefed sterilizedwaterandmouse

chow.Allexperimental protocolswereapproved by theYale Animal Care and UseCommittee.

Antibodies. Rabbit anti-asialo GM1 for mouse NK-cell depletionwaspurchased from Wako Chemicals (Richmond, VA). The following antibodies were used for immunohis- tochemistry: mouse anti-human T cell (anti-CD3), mouse

anti-human monocyte (anti-CD68), mouse anti-human Langerhans cell (anti-CD1a), mouse anti-human platelet- endothelialcelladhesionmolecule(anti-CD31; Dako),ham- steranti-mouseCD31 (gift from SteveBogan, Boston Uni- versity),mouseanti-human intercellular adhesion molecule 1 (anti-ICAM-1, RR 1/1; gift from T. Springer, Institute for Blood Research, Harvard University), mouse anti-human HLA-DR(LB3.1, gift from J. Strominger, Harvard Univer- sity), ^mouse anti-human CD4 and mouse anti-human CD8 (Becton Dickinson), mouse anti-humanvascular cell adhe- sion molecule 1 (anti-VCAM-1, E1/6; gift from M.P. Bev- ilacqua, Amgen Biologicals), and mouse anti-human E-se- lectin [H4/18 (15)], mouse anti-humanperforin (T Cell Di- agnostics, Cambridge, MA), and hamster anti-mouse CD3 (PharMingen). Biotin-conjugated Ulexeuropaeus agglutinin 1(Dako)wasalso used forimmunohistochemistrytoidentify humanmicrovascular endothelium. The following antibodies

were used forfluorescence cytometry: fluorescein isothio- cyanate-conjugatedmouseanti-humanCD3andmouseanti- human CD4, phycoerythrin-conjugated mouse anti-human CD8(Exalpha, Cambridge, MA),andred613-conjugatedrat anti-mouseCD45(GIBCO/BRL).

Peripheral Blood Mononuclear Cell (PBMC) Isolation and

En tinent.PBMCswereisolated from adult volunteer do-

norsby leukopheresis (protocol approved by the Yale Human Investigations Committee)andpurified usinglymphocytesep-

aration medium(OrganonTeknika-Cappel). Approximately 3

x 108 cells wereinjected i.p. into each mouse.Where indi- cated, mice were pretreated by i.v. injection of 50

A.

of anti-asialo GM1antibody 12-24hbeforePBMCengraftment.

Human Immunoglobulinand Circulating Lymphocyte De- termination. HumanIgG concentrations and circulatinghu-

manlymphocyte frequencyweredetermined from heparin- ized blood collected by retroorbital venipuncture. Human Abbreviations:EC,endothelialcell;ICAM-1,intercellular adhesion molecule 1; PBL,peripheralbloodlymphocyte; PBMC,peripheral blood mononuclearcell;SCID,severecombinedimmunodeficiency;

VCAM-1, vascular cell adhesion molecule 1; NK, natural killer;

H&E,hematoxylin/eosin; mAb,monoclonalantibody.

*A.G.M.

and P.P. contributedequallytothis work.

VPresent

address:DepartmentofDermatology,Universityof Vienna MedicalSchool,A-1090Vienna,Austria.

**To whomreprintrequestsshould be addressed.

9146 Thepublicationcosts of this article weredefrayedinpartby page charge payment.Thisarticle musttherefore behereby marked "advertisement"

inaccordance with 18U.S.C.§1734 solelytoindicatethis fact.

Downloaded by guest on October 7, 2021

Immunology: Murray et al.

andmouse IgG levels werequantitatedbyasandwich ELISA (16) using capture reagents and standards from Cappel (Durham, NC); the sensitivity ofthe assay was 100ng/ml.

Micewith detectable murineIgGlevels were excludedfrom the studies. Thefrequency ofcirculatinghuman Tlympho- cytes as aproportion of bloodleukocyteswasdeterminedby fluorescence flow cytometry (FACSort; BectonDickinson) (4). Alivegate wasestablishedin thelymphocyte regionand

-10,000 events wererecordedfor analysis.

Skin Engraftment. Normal adult human breast skin was obtainedasdiscarded tissuethroughtheYaleUniversityDe- partmentofPathology (protocolapprovedbythe Yale Human InvestigationsCommittee).The

superficial

portions ofthe skin wereharvested in500-to

700-pm-thick

sheetsusingaGoulian dermatome knife (gaugesize, 0.016;Weck, Research

Triangle

Park, NC) andcutinapproximately7 x 7mm

pieces,

which werekeptin RPMI 1640medium(GIBCO/BRL) at40Cuntil transplantation. SCID micewereanesthetized

by i.p. injection

ofXylozine (3

mg/kg;

Lloyd

Laboratories, Shenandoah, IA)

and ketamine (3

mg/kg;

Aveco,Fort

Dodge, IA).

The skinwas shaved,two 5 x 5 mmskin segmentswere excised fromthe back of eachmouse,and thedefectswere

immediately

covered with the human skin grafts and fixed with disposable skin staples (3M, St.Paul). Graftswere

individually

harvestedatthe indicatedtimesusing thesameanesthesia

protocol.

Histology and Immunohstology. Each skin

sample

was used to prepare 3-,um

paraffin-embedded

sections stained withhematoxylinandeosin(H&E), and

4-num

cryostatsec- tionswereusedfor immunohistochemical

staining (17).

Con- trolsusedspecies-matched

nonbinding

monoclonalantibod- ies(mAbs) instead of

specific

mAbs.

DataAnalysis. All skin specimens wereevaluated

by

two observers. (i) Thenumber of cells

positive

for human

CD3,

Proc. Natl.Acad. Sci. USA91 (1994) 9147 CD4, CD8, CD1a, and CD68 and mouse CD3 was assessed. A specimen was considered toshowinfiltrationif the numberof humanleukocytes within the skin was at least10-foldover that seenin therespective controlspecimen (i.e.,humanskin graft without reconstituted PBMCs in the same experiment). (ii) Theexpression ofE-selectin,VCAM-1,ICAM-1,or HLA-DR wasscoredsemiquantitatively(-,negative; +,discrete;++, moderate; + ++, strong). Tissue injury due to allogeneic PBMC reconstitution was assessed in paraffin-embedded specimens.Pooled data from threeexperimentsarereported.

RESULTS

Engraftment of Human PBMCs to C.B-17 SCID Mice.

HumanPBMCswereengrafted intoSCIDmiceasindicated byserumhumanIgG concentrations (=1000pg/mlatday 14) andbycirculatinghuman T cells. HumanCD3+lymphocytes could be detectedintheperipheral bloodasearlyas24 hafter introduction into theperitoneal cavity. Preliminary experi- ments established the optimum number of PBMCs to be transferredas3x 108 cells(datanotshown).Pretreatmentof themice withapolyclonalantibodytoasialo-GMl increased the frequency of circulating human T cells at days 3-14 (range, 1-60o ofcirculating leukocytes analyzed) without increasing human IgG concentrations when compared to saline-pretreated animals.

Histology of Human Skin Graftedon C.B-17 SCID Miee.

Adulthuman skin grafts wereroutinely accepted by SCID mice(Figs. 1Aand2A).Acriticalconsiderationwasthat the dermisbe as thinas possible yet retain the human dermal superficial vascular plexus. Seven days after engraftment, granulation tissuewasevidentatthe base of the

graft.

Mouse and humancapillaries (defined by double staining with anti- human CD31 and anti-mouse CD31 mAbs) appeared to

p -

', *.' Asn

~ X". 4Ve*o sr'^^ti

& _ ..#~- qe /jI

'rk~~ ~ ~ ~ ~ ~ ~ '

'If~~~~

;/.2v/ - XX ** :* S 4 2 w / Xr v t

FIG.1. Histologicalappearanceofsplit-thickness human skin grafts harvested from SCID mice after reconstitution with allogeneic human PBMCs.(A)Micrograph ofaday-16 control specimen harvested fromananimal thatreceived salinei.p. (B)Micrograph ofaday-16 specimen harvested fromamousethatreceivedallogeneic humanPBMCsi.p. Each specimen wasstained withH&E. Note themarked mononuclear infiltrateandfocal loss of thedermal-epidermal junctionin B.Themononuclear cellstendtobearrangedin^anangiocentricpattern.(x150.)

e s;W v

*S # e _ _

e!^

^{f - 4X -}

- .. 4's A.

t ,,sF

._ ,. A*,.

g i * j9zE.gh

... ...

b

,....# ...:

;:

+ ,t r

t t ..

z % ! s

,

^{* , # ..I. 1.11} I...". ^{t- :}_;^-..W

0 .0 .0

,of '..

A.

I

Downloaded by guest on October 7, 2021

9148

Immunology: Murray

etat.

A ~~~~~~~~~~~~~~~~~j

Proc. Natl. Acad. Sci. USA 91

(1994)

Li /~~~~~~~~~~~~~~~~~~J^.

F

.Pts'

G~~~~~~~~~~~~~i

4 .)

K

I".

I.

''¼

4g *.

Es

7 $

'4

I p

b.j $

. 0*

a-

p -

HW

L

^"

-4

FIo.2. Histologicalappearanceof human skintransplantedontoSCID mice before andafter introduction of human PBMCsallogeneicto the skin donor.(A)Low-powermicrographof human skin 2 weeks afterengraftmentstained withH&E.Note thatthe humanepidermisis thicker than the mouseepidermis (presentat theleftedgeof thespecimen) and that the humanepidermisshows normalmaturation. There is little inflammation atthe graft base at this time. (Inset) Double stained with anti-human CD31 (blue) and anti-mouse CD31 (red) to document microvascular anastomosisatthegraftbase.(B-E)H&E-stainedmicrographsof human skin harvested fromSCID miceat1, 6, 11,and 16days, respectively,after introductionof human PBMCsallogeneictothe skin donor. Note theprogressiveinfiltration ofthe dermisby miononuclear inflammatory cellsbeginningatday6(C)thatfills the dermisby day 16(E),while theepidermisisessentiallyuninvolved.(D Inset) Fibrin thrombus andperivascularproliferationof endothelial cellsathighpowerinaday-il specimen.(EInset)Destructionofahumanmicrovessel

,6te

I

M

4

I il.I_I

i ,"t

)4,0 -L

I.

i,

:.It i

10% -) V

%,1, 4

P. t

:4 I I.,

I _{I j}

,p .

4

I I

t.,-.,-

Downloaded by guest on October 7, 2021

Proc.Natl. Acad.Sci. USA 91 (1994) 9149 intermingle, with focal evidence of microvessel anastomosis

atthegraft margins (Fig. 2A Inset). Human CD31+ vessels persisted within the human dermis and contained intralumi- nal erythrocytes,indicating perfusion withmouseblood. The engrafted dermis retained small numbers of CD4+ T cells and CD68+ macrophages, and the epidermis retained CD1a+

Langerhans cells. Few if any CD8+ T cellswerepresent.

Fourteendays afterengraftment, thegranulationtissueat the base of the human graft was largely resolved. Graft humanCD31+ microvesselspersistedand occasionalmouse CD31+ vessels could be identified. Theboundarybetween the human and mouse epidermis was sharply demarcated, and the human basal celllayershowed normaldifferentiation, exceptfor focalareaswithintercellular edema andraresingle necrotic keratinocytes. The epidermal thickness was com- parable to thepregraftspecimen,and theretepeg patternwas wellpreserved. Therewaslittleevidenceforimmunological activationin theresting tissue.Human endotheliumdidnot express E-selectin or VCAM-1 molecules. ICAM-1 and HLA-DR molecules wereweakly expressedonthe endothe- lium at levels comparable to pregraft specimens. Kerati- nocytes were HLA-DR-negative and ICAM-1 expression wasdetectable at very low levels on the basalkeratinocytes.

Aflogeneic Lymphocyte Responseto Skin Grafts. Athree- stepprotocolwasdevelopedtoexamineinteractionsbetween circulating human T cells and graft microvessels lined by humanvascularendothelialcells.(i)Humanskinwasgrafted onto SCIDmice and allowedto healfor14days. (ii) Mice weretreated withanti-asialoGM1todepletemouseNKcells.

(iiI) Twenty-fourhours later,human PBMCs,

allogeneic

to the skin, were introduced i.p. This protocol produced a perivascularlymphocytic infiltratein the skin

grafts

of >95%

of theanimalswhenstudiedat16days aftertheinjectionof humanPBMCs(Fig. 1B). Introduction of PBMCs beforeor atthesametimeasskinengraftmentledto aninflammatory infiltrateatthebase of thegraftthatpreventedmouse-human microvessel anastomosis.

Three time courseexperimentswereperformedwith sim- ilar results. Skingrafts examined 1 day after PBMCrecon- stitutionwereindistinguishablefrom controlspecimens (Fig.

2B). By 6 days after PBMC reconstitution, three of five specimens showedfocally damagedvesselscharacterizedby dilated lumens, fibrin thrombi, and extravasated erythro- cytes (Fig. 2C). Clusters of intraluminal human CD4+ cells wereadherent to theendothelium andsome(score, +) CD4+

andCD8+ Tcells werefound inaperivascular distribution (Fig. 2C).The numbersofhumanCD68+ macrophagesand CD1a+ Langerhans cells were indistinguishable from con- trols.Strikingly, in five of five specimens,HLA-DR(score, +++;Fig.2FandG)andVCAM-1(score, ++;Fig. 2Jand K)werestrongly up-regulatedon mostofthe human dermal vessels. Endothelial ICAM-1 expression remained un- changed and E-selectin expression was not induced. Most humankeratinocytes strongly(score, + ++;Fig.2 FandG) expressed HLA-DR, but keratinocyte ICAM-1 expression was notup-regulated on these cells.

By 11 days after PBMC reconstitution, the epidermis showed areas of marked acanthosis, although individual keratinocyte morphology was indistinguishable from day 1

(Fig. 2D). Five of fivespecimens showed infiltrates of human leukocytes. In three of five specimens, the infiltrate was moderate (score, ++) and in two offive, it was discrete (score, +).Infiltratesconsisted ofapproximately equalnum- bersof CD8+ cells (Fig. 2H) and CD4+(Fig.2L) T cells, most of which expressed CD25 (data not shown). Macrophages and Langerhans cells appeared unaltered. Human blood vesselswithin the infiltrate were dilated; fibrin thrombiand extravasated erythrocytes were noted. Perforin-positive T cells were found in apposition to damaged ECs (Fig. 2H Inset). Adjacent areascontained knotsof disorganized ECs, suggestiveofproliferation (Fig. 2DInset). VCAM-1 (score, + ++) and HLA-DR (score, + ++) on ECs were more strongly expressed comparedtovesselsobserved onday6.

Furthermore, ICAM-1 (score, ++) was now strongly up- regulated onvascular ECs, but E-selectin was not detected.

KeratinocyteHLA-DRwasequalordiminishedcomparedto the6-dayspecimens(score, + +to+ ++),whereasICAM-1 wasnowup-regulated (score, ++).

By day 16 afterPBMC reconstitution, the epidermis re- mained acanthotic and there was a focal epidermotropic T-cellinfiltrate anddisorganization ofthe basal celllayer with rarekeratinocyte necrosis (Fig. 2E). Fifteen of16specimens showed strong(score, + ++) human CD3+ T-cellinfiltrates, composed of equal numbers of CD4+ andCD8+cells. Mouse CD3+ T cells were not seen. Intact human blood vessels couldnolonger be identified(Fig.2IandM). In theresidual human microvessels, the vessel wall was disrupted, with increased fibrinthrombiand extravasation oferythrocytes.

Perforin-positivelymphocytes were again noted adjacent to theresidual ECs. ManyinfiltratingTcellsappearedasblasts, and multiple mitotic figures were evident. There was an increase in perivascular CD68+ macrophages, but CDla+

Langerhans cells remained unchanged. CD31+ endothelial remnantsstaining for VCAM-1, ICAM-1,orHLA-DRcould not be readily distinguished from staining of infiltrating leukocytes. Keratinocyte expression of HLA-DR and ICAM-1 variedwidelyamongspecimens(score,+to+++).

DISCUSSION

Thisreport describes asmall animalmodelthat may prove useful for the in vivo analysis ofthe human cell-mediated immuneresponse to humanallografts. Wehave introduced technical improvements into SCID mouse transplantation that(i) leadtoperfusion of skin graft microvessels lined by humanECs and (ii) produce sustained circulating levels of human T cells. Specifically, we find that the use of split thickness adult skin, rather than neonatal foreskin speci- mens,leadstogreaterretention ofthe humanmicrovessels (18). In addition, pretreatment of the mice with anti-asialo GM1 in

conjunction

with a

large

PBMC inoculum leads to higher levels of circulatinghuman lymphocytes. When we combine thesetechniques, someof thecirculatinghuman T cellsspecifically localizetoallogeneicskingrafts inaperivas- cular distribution but not to xenogeneic mouse skin. The human

graft

dermal microvasculature becomes activated coincidentwiththeonsetofT-cellinfiltration,asindicatedby the expression of VCAM-1 and

major

histocompatibility complexclass II molecules. A subsetofthe humanCD3+T

andproliferationof perivascular inflammatory cells in a day-16 specimen. (F and G)Stainingwith anti-HLA-DR antibody of specimens harvested atdays1and 6,respectively,after the introduction of human PBMCs. Note that HLA-DR staining of dermal microvessels, some perivascular cells,andepidermalLangerhans cells are present on day 1. The staining intensity is increased by day 6 and HLA-DR is additionally expressed onkeratinocytes. (J and K)Staining with anti-human VCAM-1 antibody at days 1 and 6, respectively. Note that VCAM-1 is absent at day 1 andis expressed on human dermal microvessels by day 6. (H and L) Stained for human CD8 and CD4, respectively, on day 11 specimens. (H Inset)Stained for human perforin. Perforin-positive cells are localized adjacent to human microvessels. (I and M) Double-stained with Ulex europaeusagglutinin 1 (red) to detect human endothelium and anti-human CD3 (blue) to detect T cells. (M) Specimen harvested 16 days after introductionof PBMCs. (I) Paired specimen from a mouse that did not receive PBMCs. Note that the inflammatory infiltrate in M is largely composed of human T cells and that at this time point, human endothelial cells have largely disappeared compared to the control specimen (I).

(A, x45; B-M, x90;insets, x 180.)

Immunology: Murray

^etal.

Proc.Natl. Acad. Sci. USA 91(1994) cells differentiate to display a cytolytic phenotype marked by

expression of the granule proteinperforin,and by -2 weeks after reconstitutionwith allogeneic human PBMCs, the hu- man microvasculature of the skin

graft

is destroyed.

Kawamura et al. (11) have described "chronic" human skin rejection in a SCID mouse model. In this system, skin

grafts

developed sclerosis and lymphocyte infiltration 3-4 weeks after adoptive transfer of presensitized, but not naive, human lymphocytes. Our experience differs in several re- spects. (i) We demonstrate95% of the skingrafts areinfil- trated by unsensitized allogeneic human Tcellswithinaweek of PBMC reconstitution. (ii) We demonstrate destruction of human vascular elements within the skin consistent with the time course seen in first set skin rejection in humans (1). (iii) We provide evidence suggesting that the infiltrating T cells are activated in response to alloantigen. Differences in the experimental protocols likely account for these discrepan- cies. Most critically, we allowed the skin grafts to heal prior to the introduction ofallogeneic human PBMCs. This pro- motes perfusion of the skin graft through anastomosed human microvessels. In addition we treated our mice with anti-asialo GM1antibodiestopromote

engraftment

of the human PBMCs (10, 12, 19) and we transferred large numbers (3 x 108 cells) of freshly harvested unsensitized PBMCs to each mouse.

This protocol enhances the probability that human T cells introduced into the peritoneal cavity of the mouse will circulate and contact the humanskin graft microvasculature.

The interpretation that the destruction of the human dermal microvasculature is mediated by allogeneic T cells is sup- ported by the following observations. (i) We find evidence of destruction of the human dermal microvasculature only in mice reconstituted with allogeneic human lymphocytes. This observation indicates that the vascular damage is the result of neither effects of graft ischemia nor of a xenogeneic immune response mediated by mouse NK cells or otherleukocytes. (ii) A human T-cell infiltrate precedes the loss of the human endothelium in a temporal profile consistent with the activa- tion and differentiation of T cells in the allogeneic setting.

Activation of the dermal T cells is inferred from the expression of thehigh-affinity interleukin 2 receptor, CD25, a marker of T-cell activation. Moreover, a subset of infiltrating T cells demonstrates perforin-positive cytolytic granules consistent with classicT-cell-mediated cytotoxicity directed against the allogeneic endothelium. Furthermore, the functional effects of cytokines produced by activated T cells are evident by the induction of major histocompatibility complex class II and VCAM-1 on the vascular endothelium and by the induction of ICAM-1 and class II molecules on the keratinocytes in the epidermis. Finally, T-cell blasts with numerous mitotic figures are seen in the skin harvested at day 16, presumably prolif- erating in response toalloantigen stimulation. In a preliminary experiment, T cells recovered from an infiltrated skin graft proliferated in vitro in response to dermal microvascular endothelial cells cultured from the skin

graft

donor but not from a third-party donor.

Interestingly, in this system we do not observe frank necrosis of the epidermis. In this model, unlike traditional allogeneic skin allograft models, the graft enjoys a dual blood supply, derived from both the anastomosis ofthe remnant human microvascular arcades with mousemicrovessels and from the ingrowth of new mouse vessels. As the human immune response develops, selective destruction of the hu- man vascular elements ensues, but the epidermis remains viable, presumably nourished by the remaining mouse cap- illaries. Necrotic keratinocytes are observed only rarely, and Langerhans cells are preserved.

Our findings arehistologically similarto thecell-mediated immune response that was reported after transplantation of

human skin to an allogeneic donor (1). Inthoseexperiments, T cellsaccumulated inthedermis of theallogeneic skin

graft

in a perivasculardistribution. This was accompanied by en- dothelial cell activation asjudged bymorphologic criteria and followedby destruction ofthemicrovasculature.Incontrast to what is seen in the current model, foci of skin necrosis developed,coalescedover =24h, andculminated in sloughing ofthegraft. Significantly,onlyfocalkeratinocyte necrosis was observed. The authorsinterpretedtheseresults to suggest that the

graft

losswas

primarily

theeffect of

specific

destructionof the graftmicrovasculature andsecondarily a combination of ischemia andgranulocyteinfiltration.

Our model provides theopportunity toobserve the role of EC adhesionmolecules inthedevelopment of human

allograft

rejection. As the immune response to the graft evolves, VCAM-1 is up-regulatedearly (by day 6), whereasICAM-1 expression isinducedlater inthe response,andE-selectinwas not induced. Theseobservations suggestthat VCAM-1 may serve torecruit Tcells to sites of inammation in the skin.

In conclusion, wehave developed a model of humanmi- crovascularcell injuryinduced byallogeneic T lymphocytes in a SCID mouse. Histological examination suggests that the injury is T-cell mediated,consistent with acute cellular rejec- tion. This model will be useful in studies of endothelial- dependent human T-cell recruitment and activation. Itmay offer the opportunity to testantiinflammatory reagents, such as mAbs orchimericproteinsthat may be species restricted.

We thank the Department ofPathology and the Comprehensive Cancer Center, Yale New HavenHospital,forproviding skin speci- mens,andthe PheresisService of theYaleBlood Bank forperforming the leukopheresis of human volunteers. Financial support for this work wasprovided in part through a grant from American Cyanamid to the Molecular Cardiobiology Program ^at the BoyerCenter for MolecularMedicine, from National Institute of AllergyandInfectious Diseases (GrantR37-AI-12211) to P.A., and fromapilot grant (toP.A.

and J.S.P.) from the Yale Skin Diseases Research Center (Grant P30-AR414942). A.G.M. is the recipient ofafellowship award from theNationalKidneyFoundationofCanada.P.P. is therecipientof a fellowship awardfrom the MaxKade Foundationof Austria.

1. Dvorak, H. F., Mihm, M. C., Dvorak, A. M., Barnes, B. A.,Manseau, E. J.& Galli, S. J. (1979)J.Exp.Med. 150, 322-337.

2. Leszczynski, D.,Lasczynska,M.,Halttunen& Hiyry, P.(1987)Kidney Int.31, 1311-1316.

3. Matsumoto, Y.,McCaughan, G. W., Painter, D. M. &Bishop, G.A.

(1993)Transplantation56,69-75.

4. Mosier, D. E., Gulizia, R. J., Baird, S.M. & Wilson, D. B. (1988) Nature (London) 335,256-259.

5. Dorschkind, K., Keller, G. M.,Phillips,R.A., Miller, R.G., Bosma, G. C.,O'Toole,M.& Bosma, M. J.(1984)J.Immunol.132,1804-1808.

6. Fulop,G. M.&Phillips,R. A.(1990) Nature(London) 347, 479-482.

7. Carlsson, R.,Martensson,C., Kallioi,S.,Ohlin, M. &Borrebaeck, C. A. (1992)J.Immunol. 148,1065-1071.

8. Leader, K. A.,Macht,L.M.,Steers, F.,Kumpel,B. M.&Elson,C.J.

(1992)Immunology76, 229-234.

9. Hesselton, R. M., Koup, R. A., Cromwell, M.A., Graham, B.S., Johns, M. &Sullivan, J. L. (1993) J.Infect.Dis.168,630-640.

10. Shpitz, B.,Chambers, C.A.,Singhal,A.B., Hozumi,N., Fernandes, B. J., Roifman, C.M., Weiner, L.M., Roder, J.C. & Gallinger, S.

(1994)J.Immunol. Methods169, 1-15.

11. Kawamura, T.,Niguma,T.,Fechner, J. J.,Wolber,R.,Beeskau,M.A., Hullett, D. A., Sollinger, H. W. &Burlingham,W. J.(1992)Transplan- tation 53, 659-665.

12. Murphy, W.J., Conlon, K.C., Sayers, T. J., Wiltrout, R.H., Back, T. C.,Ortaldo,J. R. &Longo, D. L.(1993)J.Immunol.15O,3634-3642.

13. Rendt, K. E., Barry, T. S., Jones, D. M.,Richter, C. B.,McCachren, S. S. & Haynes, B. F. (1993) J.Immunol. 151, 7324-7336.

14. Tary-Lehmann, M.&Saxon,A.(1992)J.Exp. Med. 175,503-516.

15. Pober, J. S., Bevilacqua, M. P., Mendrick, D. L., Lapierre, L.A., Fiers, W. &Gimbrone,M. A.(1986) J. Immunol. 136,1680-1687.

16. Macy, E., Kemeny, M. & Saxon, A.(1988)FASEBJ.2,3003-3009.

17. Petzelbauer, P., Bender, J. R., Wilson, J. & Pober, J.S. (1993) J.

lmmunol. 1S1,5062-5071.

18. Yan,H.-C.,Juhasz, I., Pilewski, J., Murphy, G.,Herlyn,M.&Albelda, S. M. (1993)1 . Clin. Invest. 91,986-9%.

19. Murphy, W. J., Bennett, M., Anver, M. J., Baseler, M. &Longo, D.

(1992) Eur.J. Immunol. 22, 1421-1427.

9150

Immunology: Murray

etal.

Downloaded by guest on October 7, 2021