Hyperacute rejection of vascularized hearttransplants in BALB/c Gal knockout mice

Gock H, Salvaris E, Han W, Mottram P, Murray-Segal L, Pearse MJ,Cowan PJ, Goodman DJ, d'Apice AJF. Hyperacute rejection ofvascularized heart transplants in BALB/c Gal knockout mice.Xenotransplantation 2000; 7: 237±246. # Munksgaard, Copenhagen

Abstract: Pig-to-primate vascularized xenografts undergo hyperacuterejection (HAR). This results from pre-formed xenoreactive antibodiesdirected against galactose-a1,3-galactose (aGal) in the donor organ andactivation of the complement cascade. We describe an in vivo murinemodel of HAR using a BALB/c mice system devoid of histocompatibilityor complement differences between donor and recipient to investigate inisolation, the effects of aGal epitope and anti-aGal antibody interactionsin causing rejection of vascularized heart transplants. Gal KO mice wereimmunized with rabbit red blood cell membranes to induce high anti-aGal antibody titers that were predominantly IgM by ELISA (enzyme-linked immunosorbent assay). When aGal-expressing mice hearts weretransplanted heterotopically into these recipients (n512), 67% of graftsrejected within 24 h, the majority within 16 h with histological featuresof HAR. In contrast, none of the grafts in the non-immunized Gal KOrecipient control group (n511) underwent HAR. Interestingly,approximately 50% of the remaining grafts in both the immunized andnon-immunized Gal KO recipient group were rejected between 7 and27 days by a rejection process characterized by a dense in®ltrate ofmacrophage/monocytes, perivascular cuf®ng and tissue destructionsimilar to recent descriptions of delayed xenograft rejection (DXR). Inaddition, some grafts (21.5%) continued to survive in the immunized GalKO recipients despite the presence of anti-aGal antibody and normalcomplement activity and these showed well-preserved myocardium whenharvested whilst still functioning well at days 30 or 90. No rejection wasseen when Gal KO donors were used in this system (n54), nor whenaGal-expressing BALB/c hearts were transplanted into aGal-expressingBALB/c recipients (n55). This in vivo small animal model offers theopportunity to test a variety of strategies to overcome HAR prior tomore resource intensive pig-to-primate studies, and may provide insightsinto the processes similar to DXR and accommodation.

Hilton Gock,1* Evelyn Salvaris,1*

Lisa Murray-Segal,1

Patricia Mottram,2 Wenruo Han,2

Martin J. Pearse,1

David J. Goodman,1 Peter J. Cowan1

and Anthony J. F. d'Apice1

1Immunology Research Center, St Vincent'sHospital, Fitzroy, Victoria, Australia, and2Department of Surgery, The Royal MelbourneHospital, Melbourne, Victoria, Australia

Key words: anti-aGal antibody ± aGal ± heart ±

xenotransplantation

Address reprint requests to Prof. A. J. F. d'Apice,

Immunology Research Center, St Vincent's Hospital,

41 Victoria Parade, Fitzroy, Victoria 3065, Australia

(E-mail: dapice@svhm.org.au)

Received 21 February 2000;

Accepted 23 May 2000

Introduction

Xenografts are a potential alternative source ofsolid organs for humans and for a variety oflogistical, ethical and physiological reasons, thepig is currently the most favored donor animal[1]. Described here is a small animal model ofhyperacute rejection (HAR) in which the patho-genesis closely re¯ects the pig-to-human com-bination. This is a unique model in which the

rejection processes are solely the result of aGaland/or anti-aGal antibody interactions in anin vivo setting where donor and recipient areotherwise isogeneic. In addition, the model mayprovide valuable insights into the fate of graftsbeyond HAR in this setting.

The most immediate and catastrophic immu-nological barrier in the pig-to-primate com-bination is HAR. This is a rapid process inwhich pre-formed xenoreactive antibodies in therecipient's circulation bind to speci®c antigenson the endothelium of the donor organ, initiat-ing a cascade of events including complement*Gock & Salvaris contributed equally to this paper.

Xenotransplantation 2000: 7: 237±246

Printed in UK. All rights reserved

Copyright # Munksgaard 2000

XENOTRANSPLANTATIONISSN 0908±665X

237

activation, endothelial cell damage and activa-tion, platelet clumping, thrombosis and infarc-tion of the transplanted organ [2]. The majortarget antigen of these xenoantibodies is galac-tose-a1,3-galactose (aGal), a disaccharide resi-due expressed on cell membrane glycoproteinsand glycolipids [3±5]. aGal is produced bya1,3-galactosyltransferase (GalT), an enzymefound in all mammals except humans, apesand Old World monkeys [6]. In these higherprimates, an evolutionary mutation in thegalactosyltransferase gene prevents synthesis ofthe enzyme and so prevents formation of aGal.Anti-aGal antibodies develop as a consequenceof exposure to the antigen expressed on normalgut ¯ora [6]. Approaches to overcome HARhave been to deplete the xenoreactive antibody,block the complement cascade, or reduce aGalexpression in the donor organ [1,5].

Pig-to-primate xenotransplantation models areresource intensive and there are obvious logisticalproblems in terms of the size and gestation timesof the subjects, thus, small animal models havebeen developed to investigate the pathogenesisand to test strategies to overcome HAR [7±9].Unfortunately, these models either do not involvethe aGal epitope [9], require the addition ofexogenous xenoreactive antibody or complementto cause rejection [7], or in the case of other GalKO mice systems, have signi®cant allograft mis-matches between recipient and donor, reducingthe clarity of rejection processes mediated byaGal and anti-aGal antibody [8].

It was previously thought that rodents wereunable to hyperacutely reject vascularized graftsbecause of a defective complement system, but invivo evidence in this paper and recent in vitroevidence suggests that the assays previously usedwere suboptimal [10]. In this study, we describeanti-aGal antibody-mediated HAR and a DXR-like rejection of heterotopic murine cardiactransplants using aGal expressing wild-type(WT) donors and Gal KO recipients. All micewere on a BALB/c genetic backgrounds allowingthe study of aGal and anti-aGal antibody effectsin the absence of histocompatibility differencesbetween donor and recipient. Although HAR isthe central feature of this model, a smallproportion of grafts survived beyond the HARtime frame only to fail in a fashion analogous toDXR, and several continued to function in thepresence of xenoreactive antibody and a normalcomplement system for the 60-day duration ofthe experiment. These phenomena have beenobserved by other investigators in models whereHAR is averted by intervention [11,12].

Materials and methods

Mice

a1,3-galactosyltransferase knockout (Gal KO)mice on a mixed 129sv/C57Bl6/CBA background[13] were used to establish two experimental lines.Line 1 was generated by mating the original GalKO mice once with BALB/c mice, followed bybrother±sister mating of heterozygotes (Gal+/±).Line 1 mice, therefore, have a mixed geneticbackground consisting of elements from 129sv,C57Bl6, CBA and BALB/c mice. Gal KO (Gal±/±)and WT-aGal-expressing (Gal+/+) mice from line1 were used in a pilot study.

Line 2 was developed to remove any histo-compatibility differences by backcrossing 10times with BALB/c mice. Skin grafts betweennormal BALB/c (Animal Resources Center,Adelaide, Australia) and these mice (n59) werenot rejected (.120 days), in contrast to CBAmice skin, which was rejected between 7 and14 days. Gal KO (Gal ±/±) mice were identi®ed byPCR genotyping of the 10 times BALB/c back-crossed line and subsequently maintained bybrother±sister mating. Normal BALB/c micewere used as donors in this group. All miceused from both lines 1 and 2 in transplantationexperiments were less than 6-months-old.

Induction of anti-aGal antibodies in aGal KO mice

Anti-aGal antibodies were induced by immuniza-tion with rabbit red blood cell membranes(RRBC) in which aGal is a major carbohydratesurface structure [6,14]. Rabbit blood stored inAlsevers solution (20% v/v) was washed threetimes in phosphate buffered saline (PBS). Therabbit red blood cells were lysed in sterile water.Recipient mice were injected intraperitoneallywith two doses of 107 RRBC at 14-day intervals.Peak anti-aGal antibody titer was reached at21 days from the ®rst immunization measured byELISA as below (n55, line 1 mice, data notshown). WT aGal-expressing mice (n520, line 1mice) did not develop anti-aGal antibody titersabove background even with repeated immuniza-tions consisting of a further two weekly boosters(data not shown).

Anti-aGal antibody determination

The anti-aGal antibody titer was determined byenzyme-linked immunosorbent assay (ELISA)using aGal-BSA (Dextra Laboratories, UK) asthe substrate. Flat-bottom 96-well ELISA trays(Dynex, USA) were coated with aGal-BSA at1.25 mg/ml in carbonate buffer, pH 9.5, and keptat 4 uC for 12 h. The non-speci®c binding sites

Gock et al.

238

were blocked for 1 h at 37 uC with 2% BSA inPBS. The plates were washed three times in PBS/0.05% Tween 20/2% BSA. Sera were seriallydiluted with initial dilution at 1/32 and thenadded to the plates and incubated for 45 min atroom temperature. The plates were washed asdescribed above. 50 ml of the secondary antibody,rabbit anti-mouse immunoglobulin-conjugated(Dako, Denmark, 1/500) or rabbit anti-mouseIgM-conjugated horseradish peroxidase (HRP)for the anti-aGal IgM (Zymed, USA, 1/500)isotype titer was added and incubated for 45 minat room temperature. The plates were washed afurther ®ve times and developed using O-pheny-lenediamine dihydrochloride (Sigma Chemicals,St Louis, MO, USA), and the optical densityof the solution was measured at a wavelengthof 492 nm on a microplate reader (Behring,Germany). The anti-aGal antibody titer wasde®ned as the highest dilution at which theoptical density measured 0.1.

The anti-aGal IgG antibody titer was deter-mined by a three-step ELISA. The ®rst stagewas as described above, except the plates werecoated with 10 mg/ml of aGal BSA (DextraLaboratories, UK) and the incubation withmouse serum was extended to 1 h. After washing,the plates were incubated with biotinylated goatanti-mouse IgG-Biotin (Southern BiotechnologyAssociates Inc, AL, USA; 1/500) for 1 h atroom temperature. The plates were washed andincubated for 1 h with HRP-conjugated strepta-vidin (LSAB(R)2, Dako) diluted 1:10 in PBS/0.05%Tween 20/2% BSA. The plates were washed®ve times and the color was developed andmeasured as described above.

Transplantation and monitoring

Pilot study using mice from line 1Heterotopic cardiac transplants were performedusing sex- and sibling-matched WT (aGal+/+)donors into RRBC-immunized Gal KO recipi-ents. Non-immunized Gal KO recipients wereused in a control group.

BALB/c study using mice from line 2Normal BALB/c mice were used as heart donorsand RRBC-immunized Gal KO mice from line 2were used as recipients. As controls (1) BALB/chearts were transplanted into non-immunizedGal KO mice and (2) BALB/c hearts weretransplanted into WT (aGal+/+) recipients fromline 2 as ``isograft'' controls.

Pre-transplant anti-aGal antibody titers weremeasured at their expected peak at 21 days after

the primary immunization and transplantationwas performed within the following two days.

The surgery was carried out as previously des-cribed [12]. In brief, mice were anesthetized withintraperitoneal chloral hydrate (0.1 ml of a 3.6%solution/10 g of body weight) and penthraneinhalation. The donor organ was grafted by anend-to-side anastomosis between its aorta and theinfrarenal aorta of the recipient, and an end-to-side anastomosis of the donor pulmonary arterywith the recipient inferior vena cava was createdto produce a venous drainage into the graft.

Heart function was monitored daily byabdominal palpation and graded on the basis ofthe rate and strength of heart beat using a scale of4 to 0, where 4 indicated strong, rapid and regularcontraction and 0 corresponded to completecessation of graft function. Donor hearts wereharvested at the time of rejection or at either30, 60 or 90 days post-transplantation, and therecipient's own heart was harvested at the sametime for comparison.

Histology and immunohistochemistry

Donor and recipient hearts were dissected intothree sections. The base of each heart was ®xedin 4% paraformaldehyde and paraf®n blockswere made using standard procedures. 4-mmsections were transferred onto acetone/3-amino-propyl triethoxysilane coated slides. The mid-portion of the heart was placed in Tissue-TekOCT embedding medium (Miles Inc., IN, USA)and snap frozen in liquid nitrogen. 4-mm frozensections were collected on gelatin-coated slidesand ®xed in 4% paraformaldehyde for 5 min.The paraf®n and frozen section slides werestained with hematoxylin and eosin (H & E)using standard procedures.

For immunohistochemistry, the frozen sectionswere ®xed in 4% paraformaldehyde for 15 min at4 uC and washed in PBS. The slides were then pre-incubated with 10% swine serum (Institute ofVeterinary Sciences, Adelaide, Australia) for10 min. The sections were then stained with arat anti-mouse antibody for either granulocytes(Gr-1 antibody, Pharmingen), macrophage/monocytes (anti-CD11b antibody, Pharmingen),or T cells (KT3 antibody [15]) at 4 uC for 17 h.The slides were then washed in PBS and rabbitanti-rat immunoglobulins (Dako) at 1:250 in 5%mouse serum (Institute of Veterinary Sciences)and 1% swine serum applied at 4 uC for 30 min.After the slides were washed, endogenous perox-idase was blocked by taking the sections througha series of graded alcohol baths of 70%, 90% and100% alcohol followed by a 10-min incubation in

Hyperacute rejection in Gal KO mice

239

1% hydrogen peroxide in methanol. The slideswere then taken through the graded alcohol inreverse and then ®nally washed in PBS for 5 min.A third layer consisting of swine anti-rabbitimmunoglobulins conjugated to HRP (Dako,1/50) was applied for 30 min and then washedoff with PBS. The slides were incubated withrabbit peroxidase±anti-peroxidase (Dako, 1/30)for 30 min and then incubated for 15 min withthe developer, diaminobenzidine. The sectionswere then counter stained with hematoxylin,dehydrated and mounted.

As the anti-CD11b could also stain activatedgranulocytes, macrophage/monocytes were dis-tinguished by excluding cells staining positive foranti-Gr1 and on the basis of nuclear morphologyon serial sections. Where relevant, in®ltrates werequanti®ed by counting the number of cells perhigh power ®eld (HPF) at340 magni®cationexpressed as an average of 10 different areas ofthe section.

Measurement of total complement activity

Classical complement pathway activity wasassessed using the method described byHarrison and Lachman [16]. In brief, RBBCwere washed twice in 5 volumes of Dulbecco'sPBS and pelleted at 4000 r.p.m. at 4 uC for10 min. The cells were resuspended to 2.5% inPBS, at 4 uC and mixed with an equal volume ofhemolysin (goat anti-serum to rabbit red bloodcell, ICN) diluted 1/50 in CFDa [16] buffer. Thecells were incubated for 15 min by rotation at4 uC and washed in PBS twice and resuspended to2.5% in CFDa. 0.4 ml of these sensitized rabbitred blood cells were added to 2% agarose in

CFDb [16] buffer at 45 uC. The agarose mix wasthen poured onto a 8.3 cm33.5 cm square of gelbond (Pharmacia LKB Biotechnology) andincubated at 4 uC for 16 h. 25 ml of serum wasadded to each well and incubated at 4 uC for 16 h,then at 37 uC for 6 h. The zone of lysis wasmeasured using a radial immunodiffusion reader(RID-Behring, Germany). Normal complementlysis activity was de®ned as a zone of lysis greaterthan the PBS with no serum control well.

Results

Transplantation

Pilot study using mice from line 1When WT (aGal+/+) line 1 hearts were trans-planted into immunized line 1 Gal KO mice(n516), three distinct outcomes were seen. 56% ofgrafts rejected within 48 h, the majority failingwithin 16 h. 13% of grafts failed between day 21or 22, and the remaining 31% continued tofunction well and were sacri®ced at day 30 or60 for histology. To con®rm that early graftrejection was a consequence of induced anti-aGalantibodies in the Gal KO recipients, WT heartswere also transplanted into a group of non-immunized Gal KO recipients (n58). No earlyfailures was observed in this group, but interest-ingly, 25% of grafts failed in an intermediateperiod at between days 16 and 20. The remaindercontinued to function well and were removed atdays 30 or 90 for histology.

BALB/c study using mice from line 2In this study, the effect of histocompatibilitymismatches was eliminated by using BALB/c

Table 1. Heterotopic cardiac transplant outcomes. Donors were aGal expressing BALB/c mice and recipients were (A) BALB/c aGal KO with induced aGal antibodies,(B) BALB/c aGal KO without aGal antibodies induction, or (C) ``WT'' BALB/c mice

(A)

Immunized Gal KO recipients (n512) n Time of failures (days) Macroscopic appearance Histological appearance

Early period rejection 8 ,1, ,1, ,1, ,1, ,1, ,1, ,1, ,1 Patchy red and pale surface to dark and swollen Hyperacute rejection

Intermediate period rejection 2 10, 13 Slightly swollen DXR-like rejection

No rejection 2 Sacri®ced with functional grafts at 30, 90 Pink and beating Preserved myocardium

(B)

Non-immunized Gal KO recipients (n511) n Time of failures (days) Macroscopic appearance Histological appearance

Early period rejection 0 0 Ð Ð

Intermediate period rejection 5 8, 9, 13, 17, 27 Slightly swollen DXR-like rejection

No rejection 6 Sacri®ced with functional grafts at Pink and beating Preserved myocardium

30, 30, 60, 60, 90, 90

(C)

WT recipients (n55) n Sacri®ced with functional grafts at (days) Macroscopic appearance Histological appearance

No rejection 5 30, 30, 60, 60, 60 Pink and beating Well preserved

Gock et al.

240

donors and line 2 (BALB/c backcrossed) recipi-ents. 66% of aGal-expressing grafts in immunizedGal KO line 2 recipients (n512, Table 1) failedwithin 24 h, again the majority within 16 h. 17%failed between 10 and 13 days, and the remaininggrafts continued to function well and wereremoved at days 30 or 90 for histology. Whennon-immunized Gal KO line 2 recipients wereused (Table 1), no early graft failures occurred.Approximately half of these grafts (n55) failed

between 8 and 27 days, while the rest functionedwell until harvest at days 30, 60 or 90. Anadditional control group of WT recipients did notreject their grafts, which were harvested at days30 or 60.

Histology and immunohistochemistry

There was no appreciable difference in appear-ance between grafts from the pilot study using

Fig. 1. (A) H & E section of a recipient's own heart in the HAR group to serve as a control. (B) H & E section showing thetypical appearance of a graft failing in the early period. This one ceased to function within 16 h and shows typical features ofHAR with interstitial hemorrhage, a polymorphonuclear in®ltrate and severe muscle ®ber disruption. (C) H & E section of asmall blood vessel of the same hyperacutely rejected graft. (D) Gr-1 stain for polymorphs in a heart that underwent HAR. (E)H & E section of a recipient's own heart in the DXR group to serve as a control. (F) & (G) H & E sections of a graft failing at3 weeks showing a mononuclear in®ltrate, perivascular cuf®ng, muscle ®ber destruction. (H) Mac-1 stain of the same graft formacrophage/monocytes. (I) H & E section of a recipients own heart from the ``accommodated'' group at 30 days to serve ascontrol. (J) H & E of a graft sacri®ced at 30 days whilst still functioning. (K) Stain for T cells (KT 3) and (L) macrophage/monocytes (mac-1) showing slightly higher than background numbers.

Hyperacute rejection in Gal KO mice

241

line 1 mice and equivalent grafts in the BALB/cstudy using line 2 mice. H & E stains on bothparaf®n and frozen sections of grafts in theimmunized Gal KO recipients showed three dis-tinct patterns depending on the rejection period.All grafts failing in the early period had featuresof hyperacute rejection with small vessel throm-bosis, an in¯ammatory in®ltrate consisting ofpolymorphonuclear cells, interstitial hemorrhageand edema with severe disruption of muscle ®bers(Fig. 1, B to D).

Grafts into either immunized or non-immu-nized recipient groups from both lines failing inthe intermediate period had a diffuse in®ltrateconsisting of macrophage/monocytes (120 perHPF) with notable perivascular cuf®ng andmuscle ®ber disruption (Fig. 1, F to H). T cellswere present at only 5 per HPF (not shown)representing only a minor proportion of the totalin®ltrate. Granulocytes were present in similarlow numbers (, 5 per HPF). A severe vasculo-pathy was seen in the small to medium bloodvessels in this group. This consisted of initialcuf®ng by mononuclear cells (Fig. 2B) thenproliferation of endothelial and/or myointimalcells leading to luminal occlusion (Fig. 2, C to E).

Surviving grafts from either immunized or non-immunized recipient groups in both lines hadwell preserved myocardium and blood vessels(Fig. 2A) and a subtle in®ltrate on H & E(Fig. 1J). Immunohistochemistry showed this tobe a slightly higher-than-background T-cell (3 perHPF, background 0 per HPF) and macrophage/monocyte staining (3 per HPF, background 0 perHPF). This is shown at low power in Fig. 1(K,L). The BALB/c to line 2 WT control group graftswere well preserved and free of in®ltrate.

Anti-aGal antibody titers

All non-immunized Gal KO mice from both lines1 and 2 had total anti-aGal antibody titersequivalent to background WT or normal BALB/cmice (titer j 1:25) as shown in Fig. 3. Twenty-one days after the primary immunization, allimmunized aGal KO mice had titers of 1:27 to1:215, representing a four to one thousand-foldincrease compared with background non-immu-nized levels. There was no difference betweenmice from lines 1 and 2, and the data are showncollectively in Fig. 3. Although immunizationand an anti-aGal antibody titer above back-ground correlated with a high proportion ofgrafts undergoing HAR compared with no HARin non-immunized recipients, the actual titer didnot predict the ®nal outcome of the graft.

Fig. 2. Small arteries from aGal-expressing grafts in immu-nized Gal KO recipients. (A) Non-rejected graft at 30 days.(B) Graft failing at 8 days with perivascular cuf®ng bymononuclear cells. (C) to (E) Grafts that failed at 12 dayswith varying degrees of rejection showing vasculopathy withprogressive luminal occlusion.

Gock et al.

242

The anti-aGal antibody IgM and IgG pro®lewas determined in the immunized line 2 Gal KOrecipients (n514). The predominant pre-trans-plant anti-aGal subtype was IgM with titersbetween 1:27 and 1:215, but this isotype titer didnot correlate with the ®nal graft outcome. Thetotal and IgM anti-aGal titers fell by up to 4dilutions from pre-transplant to time of harvest in74% of recipients and the IgM titer rose by up to 5dilutions in 26% of recipients. Anti-aGal IgGsubclass was not detectable in the majority (86%)of immunized pre-transplant recipients. Evenwhen detectable, the titers were low and were, atmost, three dilutions above background. Thisanti-aGal IgG titer did not change signi®cantlypost-transplant when measured at various harvesttimes. All immunized recipients with survivinggrafts had no detectable anti-aGal IgG pre- orpost-transplant, however, the numbers were toosmall for statistical analysis. Pre- and post-transplant anti-aGal titers in non-immunizedand WT recipients remained at background levels.

Con®rmation of normal complement activity

A normal complement system was demonstratedin lines 1 and 2 with no difference seen whencompared with normal BALB/c or CBA mice(n55 in each group). Total hemolytic comple-ment activity was unchanged post-transplant attime of graft harvesting in each of the outcomegroups. A zone of lysis in the hemolytic assay wasseen in all mice sera measured. Thus, endogenouscomplement activity was not depleted followingtransplantation.

Discussion

This model offers the unique opportunity tostudy HAR mediated by anti-aGal antibody in

an in vivo small animal model, and to study theDXR-like rejection in the grafts that failed in anintermediate period, in a situation in which theonly mismatch between donor and recipientmice is aGal antigen and anti-aGal antibodies.In addition, aGal-expressing grafts surviving inthe presence of anti-aGal antibodies and afunctional complement system could provideinsights to the process of accommodation.

The similarity in pathogenesis of HAR in thismodel to the pig-to-primate situation will allowit to be used as a cost-effective and readilyaccessible method for testing strategies to over-come the process before moving to moreresource-intensive pre-clinical studies using pigsand non-human primates. The similar HAR ratebetween line 1 and line 2 mice suggests thatHAR in this system occurs independently ofgenetic background. This is an importantobservation because previous studies of cardiacallotransplantation using Gal KO mice haveused mice with a mixed genetic background[7,8]. In the model of HAR described byMcKenzie et al. [7], RRBC immunized GalKO recipients did not hyperacutely reject aGal-expressing cardiac allografts. Our data suggestthat the failure to observe HAR in that study islikely to be the result of a less vigorous immun-ization protocol in terms of dose and timingrather than the genetic background of the mice.

Strategies to overcome HAR that could be test-ed with this system include donor organ mod-i®cation, complement inhibition, and immunemodulation of the recipient.

Donor organ modi®cation by aGal epitopereduction has been shown to be effective in anex vivo murine model [17,18]. In our in vivomodel, Gal KO hearts transplanted into immu-nized Gal KO recipients were not rejected (n54,

Fig. 3. Anti-aGal antibody titres of rabbit red blood cell immunized mice compared with non-immunized mice.

Hyperacute rejection in Gal KO mice

243

survival . 60 days). This optimal strategy ofcompletely eliminating aGal expression bytargeted deletion of the GalT gene is not yetpossible in the pig. An alternative approach is toreduce the substrate for aGal production in acompetitive manner by transgenic expression ofhuman a1,2-fucosyltransferase (H-transferase)[17], N-acetylglucosaminyl-transferase, a1,6-sialyltransferase or b1,4-sialyltransferase [19].Reduction in aGal expression using thisapproach is not complete, for example, H-transferase expression in mice reduces aGalexpression by only 80 to 90% [17] and appearsto be even less effective in transgenic pigs [20]. Itis likely that transgenic expression of a combi-nation of these enzymes will be required foreffective aGal reduction [21]. The optimalcombination is currently unknown and com-parative data of mouse vs. pig tissue in theeffectiveness of reducing aGal-expression bythese means is still needed. However, the HARmodel described in this paper could potentiallyallow rapid testing of organs with various com-binations of transgenically expressed enzymesthat reduce aGal expression in an in vivo setting.This would also allow the study of neo-antigensproduced by substrate diversion that could con-tribute to rejection [22].

Complement inhibition is at least partlyprotective against HAR. The administration ofcobra venom factor or soluble complementreceptor type 1 has been shown to be effectivein non-aGal mediated small animal models orpig-to-primate studies [1]. More recently, andperhaps more promising in terms of minimizingrecipient morbidity, is donor organ modi®cationby transgenic expression of cell surface comple-ment regulatory factors (CRFs) [1]. Cowan et al.in this laboratory have demonstrated an effectivecombination of transgenic human CD55 andCD59 in pig kidney xenografts to baboon recip-ients [23]. Complement manipulation at single ormultiple levels along the cascade can be tested inour mouse model of HAR.

In addition to strategies of epitope reductionand complement inhibition, anti-aGal antibodydepletion maintained with short-term immuno-suppression may facilitate ``accommodation''whereby the graft continues to function despitethe presence of a functional recipient complementsystem and the return of xenoreactive antibody.Again existing evidence is derived from xenograftmodels where aGal is not the xeno-antigen [11].Our model can be used to evaluate approaches toinduce accommodation before proceeding to pig-to-human xenografts.

A novel approach of immune modulation bysome investigators using Gal KO mice is theinduction of mixed aGal+/+ and aGal±/± hemato-poietic chimerism that can lead to simultaneoustolerance of aGal reactive B cells and T cells thatcould react with histocompatibility antigens [8].Here, aGal-expressing bone marrow transplantedare allograft mismatched with the Gal KO recip-ient and tolerance of the same mismatchedcardiac allografts is achieved. With the modeldescribed in this paper, the speci®c effects ofaGal-mediated rejection processes can now bestudied in isolation without the histocompatibil-ity differences.

Our model is also providing other insightsinto the pathogenesis of delayed xenograft-typerejection. It is clear that HAR is dependent oninduced anti-aGal antibodies in the recipientand that background levels are not suf®cient toinitiate the process. The role of anti-aGalantibodies in the DXR-like rejection is lessclear. While the anti-aGal titer induced byRRBC in the Gal KO recipients in this modeldid not correlate with the development of DXR-like rejection in aGal expressing cardiac grafts, atiter ``dependency'' was previously demonstratedwith recipient Gal KO mice immunized withLeishmania major promastigotes [12]. Further-more, we have shown that Gal KO mice in ourfacility can naturally acquire anti-aGal anti-bodies when greater than six months old andthat these antibodies may play a role in DXR-like rejection of aGal-expressing cardiac grafts[24]. These differences suggest that there may bevariation in the speci®city of anti-aGal that isdetermined by the immunogen. Variation inspeci®city of anti-aGal has been demonstrated inhigher primates [25] and we are currentlyperforming similar studies on induced andnaturally acquired anti-aGal antibodies in GalKO mice.

What determines the outcome of the graftmay ultimately be related to a combination ofthe anti-aGal antibody immunoglobulin isotypepro®le in the recipient and graft tissue resistanceto their effects. Some investigators have sug-gested the expression of anti-apoptotic genes inendothelial cells associated with a host TH2cytokine environment in the graft as a possibleprotective mechanism in small animal xenograftmodels, where rejection is mediated by peptidemismatches rather than aGal [26]. We will betesting whether this holds true for aGalmismatched grafts in BALB/c mice, which areknown to have a tendency to produce TH2responses to some antigen stimuli containingaGal [27,28]. It is possible that the subtle T-cell

Gock et al.

244

in®ltrate in the non-rejecting grafts are afavorable T-cell subset and bene®cial to graftsurvival given their numbers do not increasefrom surviving grafts at days 30, 60 and 90.These cells may have predominated over a more``destructive'' in®ltrate, and detailed time coursestudy is currently in progress to test thishypothesis and the extended long-term effectsof these cells.

The DXR-like rejection in non-immunizedmice raises a further question of whether anti-aGal antibody is required at all for this type ofrejection. One possibility is that cellular innateimmunity, directly recognizing aGal glycopro-tein or glycolipid, causes DXR. Recent evidencesuggests that CD1 molecules present on all bonemarrow-derived cells can present glycolipidantigens to NKT-cells leading to TH2 cytokinesynthesis [29]. We are currently producing B-cellde®cient/aGal KO mice and CD1 KO/aGal KOmice recipients to use in our system to answerthese critical questions.

Thus, this in vivo murine model of anti-aGalantibody-mediated HAR is useful in not onlyinvestigating strategies to overcome HAR priorto resource intensive pig-to-primate studies, butalso in examining DXR-like rejection andaccommodation solely dependent on aGal oranti-aGal antibody.

Acknowledgments

This project is funded in part by the NationalHealth and Medical Research Council ofAustralia.

References

1. D'APICE AJF, GOODMAN DJ, PEARSE MJ. Xenotransplant-ation: an update. Trend Cardiovasc Med 1998; 8: 319.

2. PLATT JL, FISCHEL RJ, MATAS AJ, REIF SA, BOLMAN RM,BACH FH. Immunopathology of hyperacute xenograftrejection in a swine-to-primate model. Transplantation1991; 52: 214.

3. COOPER DKC, GOOD AH, KOREN E, et al. Identi®cation ofa-galactosyl and other carbohydrate epitopes that arebound by human anti-pig antibodies: relevance to dis-cordant xenografting in man. Transplant Immunol 1993;1: 198.

4. SANDRIN MS, VAUGHAN HA, MCKENZIE IFC. Identi-®cation of Gal-a(1, 3)-Gal as the major epitope forpig-to-human vascularised xenografts. TransplantationReviews 1994; 8: 134.

5. SANDRIN MS, MCKENZIE IFC. Gal-a(1, 3)-Gal, the majorxenoantigen (s) recognised in pigs by human naturalantibodies. Immunol Rev 1994; 141: 169.

6. GALILI U, CLARK MR, SHOHET SB, BUEHLER J, MACHER BA.Evolutionary relationship between natural anti-Gal anti-

body and the Gal a1±3 Gal epitope in primates. Proc NatlAcad Sci USA 1987; 84: 1369.

7. MCKENZIE IFC, LI YQ, PATTON K, THALL AD, SANDRIN

MS. A murine model of antibody-mediated hyperacuterejection by galactose-a (1, 3)-galactose antibodies in Galo/o mice. Transplantation 1998; 66: 754.

8. OHDAN H, YANG YG, SHIMIZU A, SWENSON KG, SYKES M.Mixed chimerism induced without lethal conditioningprevents T-cell and anti-Gala1, 3Gal-mediated graftrejection. J Clin Invest 1999; 104: 281.

9. SOARES M, LU XS, HAVAUX X, et al. In vivo IgM depletionby anti-mu monoclonal antibody therapy ± the role ofIgM in hyperactive vascular rejection of discordantxenografts. Transplantation 1994; 57: 1003.

10. ROMANELLA M, AMINIAN A, ADAM WR, PEARSE MJ, DAPICE

AJF. Involvement of both the classical and alternatepathways of complement in an ex vivo model of xenograftrejection. Transplantation 1997; 63: 1021.

11. BACH FH, FERRAN C, HECHENLEITNER P, et al. Accommo-dation of vascularized xenografts: expression of ``protec-tive genes'' by donor endothelial cells in a host Th2cytokine environment. Nature Med 1997; 3: 196.

12. PEARSE MJ, WITORT E, MOTTRAM P, et al. Anti-Galantibody-mediated allograft rejection in a1, 3±galacto-syltransferase gene knockout mice ± a model of delayedxenograft rejection. Transplantation 1998; 66: 748.

13. TEARLE RG, TANGE MJ, ZANNETTINO ZL, et al. The a-1,3-galactosyltransferase knockout mouse ± implicationsfor xenotransplantation. Transplantation 1996; 61: 13.

14. EGGE H, KORDOWICZ M, PETER-KATALINIC J, HANFLAND

P. Immunochemistry of I/i-active oligo- and poly-glycosylceramides from rabbit erythrocyte membranes.Charac-terization of linear, di-, and triantennaryneolactoglycosphingolipids. J Biol Chemistry 1985;260: 4927.

15. TOMONARI K. A rat antibody against a structure func-tionally related to the mouse T-cell receptor/T3 complex.Immunogenetics 1988; 28: 455.

16. HARRISON RA, LACHMAN PJ. Handbook of ExperimentalImmunology and Immunochemistry. Oxford: BlackwellScienti®c Publications, 1986.

17. CHEN CG, SALVARIS E, ROMANELLA M et al., Transgenicexpression of human a1, 2±fucosyltransferase (H±trans-ferase) prolongs mouse heart survival in an ex vivo modelof xenograft rejection. Transplantation 1998; 65: 832.

18. TANGE MJ, TEARLE RG, AMINIAN A, et al. Demonstrationof the functional importance of the Gal epitope in anex vivo model of xenotransplantation. Xenotransplant-ation 1997; 4: 20.

19. MIYAGAWA S, TANEMURA M, KOYOTA S, et al. Masking andreduction of the galactose-a1, 3-galactose epitope, themajor xenoantigen in swine, by the glycosyltransferasegene transfection. Biochem Biophys Res Commun 1999;264: 611.

20. SHARMA A, OKABE J, BIRCH P, et al. Reduction in the levelof Gal (a1,3) Gal in transgenic mice and pigs by theexpression of an a (1, 2) fucosyltransferase. Proc NatlAcad Sci USA 1996; 93: 7190.

21. OSMAN N, MCKENZIE IFC, OSTENRIED K, IOANNOU YA,DESNICK RJ, SANDRIN MS. Combined transgenicexpression of a-galactosidase and a1,2-fucosyltransfer-ase leads to optimal reduction in the major xenoepi-tope Gala (1, 3) Gal. Proc Natl Acad Sci USA 1997;94: 14677.

22. SHINKEL TA, CHEN CG, SALVARIS E, et al. Changes incell surface glycosylation in a1,3-galactosyltransferase

Hyperacute rejection in Gal KO mice

245

knock-out and a1, 2±fucosyltransferase transgenic mice.Transplantation 1997; 64: 197.

23. COWAN PJ, AMINIAN A, BARLOW H, et al. Renal xenograftsfrom triple-transgenic pigs are not hyperacutely rejectedbut cause coagulopathy in non-immunosuppressedbaboons. Transplantation 2000; 69: 2504.

24. SALVARIS E, GOCK H, HAN W, et al. Naturally acquiredanti-alphaGal antibodies in a murine allograft modelsimilar to delayed xenograft rejection. Xenotransplant-ation 2000; 7: 42.

25. NEETHLING FA, COOPER DKC. Serum cytotoxicity to pigcells and anti-aGal antibody level and speci®city inhumans and baboons. Transplantation 1999; 67: 658.

26. LIN Y, SOARES MP, SATO K, et al. Accommodated

xeno-grafts survive in the presence of anti-donorantibodies and complement that precipitate rejectionof naive xenografts. J Immunol 1999; 163: 2850.

27. AVILA JL, ROJAS M, GALILI U. Immunogenic Gala1±3Galcarbohydrate epitopes are present on pathogenicAmerican trypanosoma and leishmania. J Immunol1989; 142: 2828.

28. COFFMAN RL, VARKILA K, SCOTT P, CHATELAIN R. Role ofcytokines in the differentiation of CD4+ T-cell subsets invivo. Immunol Rev 1991; 123: 189.

29. BURDIN N, BROSSAY L, KRONENBERG M. Immunizationwith a-galactosylceramide polarizes CD1-reactive NKT cells towards Th2 cytokine synthesis. European JImmunol 1999; 29: 2014.

Gock et al.

246