Phenotypes of apolipoprotein B and apolipoprotein E after liver transplantation

Phenotypes of apolipoprotein B and

apolipoprotein E after liver transplantation.

M F Linton, … , M R Wardell, S G Young

J Clin Invest.

1991;

88(1)

:270-281.

https://doi.org/10.1172/JCI115288

.

Apolipoprotein (apo) E and the two B apolipoproteins, apoB48 and apoB100, are important

proteins in human lipoprotein metabolism. Commonly occurring polymorphisms in the

genes for apoE and apoB result in amino acid substitutions that produce readily detectable

phenotypic differences in these proteins. We studied changes in apoE and apoB

phenotypes before and after liver transplantation to gain new insights into apolipoprotein

physiology. In all 29 patients that we studied, the postoperative serum apoE phenotype of

the recipient, as assessed by isoelectric focusing, converted virtually completely to that of

the donor, providing evidence that greater than 90% of the apoE in the plasma is

synthesized by the liver. In contrast, the cerebrospinal fluid apoE phenotype did not change

to the donor's phenotype after liver transplantation, indicating that most of the apoE in CSF

cannot be derived from the plasma pool and therefore must be synthesized locally. The

apoB100 phenotype (assessed with immunoassays using monoclonal antibody MB19, an

antibody that detects a two-allele polymorphism in apoB) invariably converted to the

phenotype of the donor. In four normolipidemic patients, we determined the MB19

phenotype of both the apoB100 and apoB48 in the "chylomicron fraction" isolated from

plasma 3 h after a fat-rich meal. Interestingly, the apoB100 in the chylomicron fraction

invariably had the phenotype of the donor, indicating that the vast majority of […]

Research Article

Find the latest version:

Phenotypes

of

Apolipoprotein

B and

Apolipoprotein

E after Liver Transplantation

MacRae F.Linton,* RobertGish,t Susan T. Hubl,* EstherButler,'CarlosEsquivel,' William 1.Bry,t Janet K.

Boyles,*"

Mark R.Wardell,*andStephen G.Young*'1

*Gladstone FoundationLaboratoriesfor Cardiovascular Disease, Cardiovascular Research Institute, Departments of

1Medicine

and

'Pathology, UniversityofCalifornia, San Francisco, California 94140-0608; tLiver Unit,

Pacific

Presbyterian Medical Center, San Francisco,California 94115; and

lCentral

Laboratory, Blood Transfusion Service ofthe Swiss Red Cross, CH-3000 Berne 22, Switzerland

Abstract Introduction

Apolipoprotein(apo) E and the two B apolipoproteins, apoB48 andapoB100,areimportantproteins in human lipoprotein

me-tabolism.Commonly occurringpolymorphisms in the genes for apoE and apoB result in amino acid substitutions that produce readily detectable phenotypic differences in these proteins. We

studied changes inapoE andapoB phenotypes before and after liver transplantation to gain new insights into apolipoprotein

physiology.Inall 29 patients that we studied, the postoperative serumapoEphenotype oftherecipient, as assessed by

isoelec-tricfocusing,convertedvirtuallycompletely to that of the do-nor,providingevidence that > 90% of the apoE in the plasma

is synthesizedby theliver.Incontrast,thecerebrospinal fluid

apoEphenotype didnotchangetothe donor'sphenotypeafter

liver transplantation, indicatingthat mostof the apoE in CSF cannotbederived from the plasmapool and therefore must be

synthesized locally.TheapoB100phenotype(assessed with im-munoassays using monoclonal antibody MB19, an antibody thatdetectsatwo-allelepolymorphism inapoB) invariably

con-vertedto thephenotype of the donor. In four normolipidemic

patients, we determined the MB19 phenotype of both the

apoB100 and apoB48 in the "chylomicron fraction" isolated

from plasma 3 h after a fat-rich meal. Interestingly, the

apoB100inthe chylomicron fraction invariablyhad the pheno-typeofthedonor, indicatingthat the vast majority of the large,

triglyceride-rich apoB100-containing lipoproteinsthatappear

intheplasmaafterafat-rich meal areactually VLDL of

he-patic origin. The MB19 phenotype of the apoB48 intheplasma

chylomicronfraction didnotchangeafterliver transplantation,

indicating that almost all ofthe apoB48 in plasma chylomi-cronsis derivedfrom the intestine.These results were consis-tentwithourimmunocytochemical studiesonintestinalbiopsy

specimens

oforgandonors;using apoB-specificmonoclonal

an-tibodies, we found evidence for apoB48, but notapoB100,in donor intestinal biopsy specimens. (J. Clin. Invest. 1991.

88:270-281.)

Key words: lipoproteins * cholesterol - genetic

polymorphism-monoclonalantibody

AddresscorrespondencetoM.Linton,GladstoneFoundation

Labora-tories,P.O. Box40608,SanFrancisco,CA94140-0608.

S.Hubl'scurrentaddressisDepartmentofBiochemistry,

Univer-sityofCalifornia, Berkeley,CA94720.

Receivedfor publication26 October1990and in revisedform22

February1991.

Orthotopic liver transplantation-the complete removal ofa

diseased liver and the implantation ofanormal liver froma

donor-has becomeacommon treatmentforend-stage paren-chymalliver disease andbiliarytractdisease(1).Implantation

of the donor liver requires creating the appropriate surgical

anastomoses for the biliary duct, the portal vein, the inferior venacava, and the hepatic artery (1). Liver transplantation has also been used to treat a largevarietyofinborn errors ofmetabo-lism (1),including deficiencies of ahepatic membrane recep-tor, an intracellularenzyme, or a secreted protein. Recently, a familialhypercholesterolemia homozygotewith severe

athero-sclerotic heart disease underwentacombinedliver/heart trans-plantation procedure (2). Familial hypercholesterolemia is causedbyadeficiencyof normal LDL receptors, which retards thehepatic clearance ofLDL.Postoperatively,thepatienthad increasedhepatic clearance of LDL and only moderate hyper-cholesterolemia, indicating that the metabolic defect had been partially corrected by the liver transplantation (2). Apatient

withtyrosinemiahadnearly completecorrection ofthat defect

with livertransplantation (1, 3), and both

hemophilia

Aand

hemophiliaB arecorrectable bylivertransplantation (1, 4,5). The extent towhich inbornerrors ofmetabolismarecorrected after thisprocedurehasfrequently yieldedfreshinsightsinto the extent of theliver's role in the development of disease (1).

Ourlaboratory has focusedonthemetabolismof apolipo-protein E and the two B apolipoproteins, apoB48 and apoB100.Apolipoprotein E is amajor proteincomponent of VLDL(6),isaligand for theLDL receptor, and is important

for the

receptor-mediated

uptakeof

chylomicron

and VLDL remnantsbytheliver. ApolipoproteinEis also found in

rela-tivelyhigh levelsinthecerebrospinalfluid(CSF),1 and investi-gatorshave speculated that itmay

play

an

important

role in the

redistributionoflipidswithin the brain(7-9). Apolipoprotein B48, which containstheamino-terminal2,152amino acids of

apoB100, isacrucialstructuralproteinin

chylomicrons

and is

generally thoughttobesynthesized exclusively bythe human intestine

(10), although

arecentreporthas

suggested

that small

amountsof the

protein

might

be made

by

the humanliver

_(11).

Apolipoprotein B100 isacrucial structural proteininthe

for-mation ofthe

triglyceride-rich

VLDLinthe liver

(12),

but

re-centstudies havesuggestedthat the human intestine may also

synthesize apoB100 (11, 13-17).

Studiesperformedoverthe past decade have demonstrated

that thereare

commonly

occurring polymorphisms

in both the

apoB andapoEgenes(6,

18).

These

polymorphisms

result in

1.Abbreviations usedinthis paper: CSF, cerebrospinalfluid;IEF, iso-electricfocusing.

J.Clin. Invest.

©TheAmericanSocietyfor ClinicalInvestigation,Inc.

0021-9738/91/07/0270/12 $2.00

amino acid substitutions andphenotypic changesin the

pro-teins thatcanbe assessed by simplelaboratorytests.Isoelectric focusingcanbe usedtocharacterize commonlyoccurring

phe-notypesof apoE thatareduetosubstitutions involving charged

amino acidsatresidues 112 and 158 (6).Radioimmunoassays

using the apoB-specific monoclonal antibody MBl9 can be usedtodetect thegeneticallydetermined"MB19phenotype"

in apoB48 and apoB 100(19);the MBl9polymorphismis per-fectlyassociated with andprobablycausedbyanamino acid

substitutionatapoB residue 71 (20).

Weinitiatedastudy of thechangesin apoE andapoB phe-notypesafter liver transplantation with theexpectationthata

study in those patientscouldprovide important insightsinto

apolipoprotein physiology. This expectation wasfulfilled;the study of liver transplant patients was uniquely informative about several topics in human apolipoprotein physiology.

Among theissuesaddressed in thisstudy werethe origin of apoE inserumandCSF,theglycosylationpatternoftheserum

apoE that is derived from extrahepaticsources,theoriginofthe apoE-A-II heterodimers, the origin of apoB48 in fasting plasma, and the origin of the apoB100 that appears in the

plasma "chylomicron fraction" afterafat-rich meal.

Methods

Study population.Weobtained bloodsamplesfrom 29 livertransplant

recipients in the operatingroomonthe day of their operation. Blood samplesfrom respective organ donor patients were obtained

premor-temin theoperatingroombefore the organswereharvested(i.e.,before

lifesupport wasdiscontinued).We alsoobtained blood samples from allthepostoperative recipients;thesesampleswereobtainedatleast 1 moand usually 2-6moafter the operation. Serum and plasma

(con-taining1 mgEDTA/ml)wereseparated from the blood cellsby centrifu-gation at4°C,and the sampleswerefrozenat-70°C.Genomic DNA waspreparedfromthewhite blood cells(21). I mlofCSF,which wascollected by lumbar puncture for diagnostic purposes, was

ob-tainedfrom sixpatients.Thestudies described herewereapprovedby

the Human Experimentation Committee of the Pacific Presbyterian Medical Center.

ApolipoproteinEphenotyping. ApolipoproteinEphenotypingwas

performedbyisoelectric focusing(IEF)(22). Themostcommon apoE isoform is apoE3. Apolipoprotein E4, amorebasic isoform, differs from apoE3 in having anarginine-for-cysteinesubstitution at residue 112. Apolipoprotein E2 is a more acidic isoform and is most com-monly the result ofacysteine-for-arginine substitution at residue 158 (6).Inthe generalpopulation,therearethree homozygous apoE pheno-types(E3/3,E4/4, andE2/2)and threeheterozygousapoEphenotypes (E4/2, E4/3, and E3/2). Because different apoEisoformsdistribute

dif-ferentlywithin the plasmalipoproteinfractions(23-25), we performed apoEphenotypingonwhole serum samplesusingthe Western blot IEF system reported by Menzel and Utermann (22). Only I ulof plasma wasloaded onto each lane of theIEFgel.After electrophoresis, the proteins in the gel were electrophoretically transferred to a sheet of

nitrocellulosemembrane for immunoblotting. Apolipoprotein E-spe-cific antibodies for these studies were kindly provided by Drs. L.

Cur-tiss,K.Weisgraber,andS. Kunitake. Serum from subjects previously characterizedashavingthe apoE3/3, apoE4/4, apoE3/2, and apoE4/3

phenotypeswereincludedas controls on each gel. Phenotyping studies of CSF apoEwereperformedbyusing identical Western blotting tech-niques. However, the CSF was first concentrated by a SpeedVac Con-centrator(Savant Instruments, Farmingdale, NY); the concentrated

proteinsfrom 50 ul ofCSFwereloaded onto each lane of the gel. Detectionofdisulfide-linkedapolipoproteinEcomplexes.To detect the presence ofapoE-A-IIheterodimers(26)anddisulfide-linkedapoE

homodimers

(27)

in_plasma, 1

gl

ofplasmawasmixed with 30Mlof

sample

buffer

(10

mM

_{Tris-HCl, pH}

_8.0,3%SDS)andloadedontoa

10-20%

gradient

SDS-polyacrylamideslabgel (28).Individualsamples

wereloadedontothegelboth in the presence(reduced)andthe absence (nonreduced)of3%fl-mercaptoethanol.After theserumproteinswere

separated by electrophoresis, the samples were

_{electrophoretically}

transferredtoa_{sheet of nitrocellulose membrane for}

_{immunoblotting}

(28). For these studies,weusedarabbit antiserum toapoE, kindly provided by K. Weisgraber, and apolyclonal antibody to apoA-II, kindly provided byS.Kunitake.

Phenotyping of apolipoprotein

B100 with monoclonal

_antibody

MBJ9. TheapoB-specificmonoclonalantibodyMB19,whichwas de-velopedand characterized byCurtiss and Edgington (29), detects a

commonly occurringtwo-allelepolymorphisminapoB (19).Antibody MB 19 bindstoplasma apoBwith three distinct patterns ofimmunore-activity;strong,intermediate,orweak. These threebindingpatternsare

the result of the codominant inheritance oftwocommonapoBalleles that encode forapoB allotypesMB191 and MB192,which havehigh and low

affinity,

respectively,forantibodyMB 19.Thus, subjectswith

strongorweakbindingpatternsare_homozygousforallotypes

MB191

orMB

192,

respectively,whereas those withan_{intermediate pattern}are

heterozygotes.There isaperfectassociation between theMB191 allo-type, the absence ofanApaLI siteatapoBcDNA nucleotide421,and isoleucineatresidue71;theMB192 allotypeis associated with the

pres-enceofanApaLI siteatnucleotide 421 and threonineatresidue 71. Both thechangeintheApaLI site and the amino acid substitutionare

the result ofasinglenucleotide substitution in the fourthexonofthe apoBgene(20).To characterize thispolymorphismin livertransplant patients, we enzymatically amplified the appropriate region of the apoBgene and examined theamplifiedDNA for the presence of the ApaLI site, asdescribed(20). In addition,theplasmaMBl9apoB

phenotypewasassessedby immunoassayofplasmasamples(19,30).

ApolipoproteinB48 isinvariablypresent inonlytraceamounts, rela-tivetothelargeamountsofapoBI00,in normalplasma.Thiswasalso truefor donorplasmaand preoperativeand postoperativerecipient plasma. For practicalpurposes, therefore, immunoassays ofplasma

assess_onlythe MB19phenotypeoftheapoBI00,andarenot

measur-ablyinfluenced_bytheapoB48phenotype.

We also assessed the MB 19phenotypesof LDLsamplesprepared byultracentrifugation.The LDL(d=1.025-1.063g/ml)wereprepared from 15 plasma samples bystandard ultracentrifugationtechniques

(31).Thesesamplesrepresentedcompletesetsofthree plasmasamples

(i.e.,preoperativerecipientplasma, donor plasma, andpostoperative recipient plasma)from five donor-recipient pairs. The apoB content of each LDLsamplewasdeterminedbyanRIAusingtheapoB-specific monoclonal antibody MB3 (32). The MB3 immunoassay was cali-bratedby determiningtheproteinconcentration of the standard LDL bythe method ofLowryetal.(33), usingBSAasthe standard. All ofthe LDLsamplescontained apoB1OO, andmosthad no more thantrace amountsofapoA-IandapoE,asjudged byCoomassie blue-stained

SDS-polyacrylamide gels.Each LDL samplewastested for its MBl9

bindingpattern inacompetitive RIA, exactlyasdescribed (19).Briefly,

serial dilutions ofeach LDL sample, along withanLDL standard from anMB19,/MB191homozygote,wereincubated withafixed and

limit-ing concentration of antibody MBl9 in microtiter plates that had

previouslybeen coated with the standard LDL. After an overnight

in-cubation,theamountofbound antibody MB 19wasquantitated by the addition of radioiodinated second antibody. Results of the RIAwere

plottedasB/Boversus_{the log ofthe apoB concentration ofthe standard} orcompetitor added,where B andBoarethespecific counts per min-ute-bound in the presence and absence of competitor,respectively.

To assess the MB19 phenotype of the apoBlOO and apoB48 in VLDL and "chylomicron fractions," we simultaneously determined the MBl9 phenotypes of the electrophoretically separatedapoB100

and apoB48 proteins byadouble-label Western blot assay (28). We studied four subjects who had undergone transplantationover6mo before the study and had remained clinically well andnormolipidemic.

into tubes

containing

EDTA(1 mg/ml).Theplasmawasimmediately

separated

from the blood cells

by centrifugation,

and protease

inhibi-tors wereaddedtothe

plasma

(34).Thefasting plasma sampleswere

clear,

whereas the

_postprandial

plasma sampleswereturbid in all pa-tients. The VLDL

samples

werepreparedfrom theplasmasamplesof fasted

_{subjects by}

standard

ultracentrifugation

techniques (31). The

plasma

chylomicron

fraction was isolated from the postprandial

plasma

as

_previously

described_(28).

Briefly,

15 mlofplasmawas over-laid with 15mlof PBS and thenultracentrifugedfor 30 minat23,000

rpm inamodelSW28rotor(Beckman Instruments, Inc.,Fullerton,

CA).

Inthe

_{fasting plasma,}

no

_chylomicrons

weredetectableatthe top ofthe

ultracentrifuge

tube,

buta

_thin,

white

layer

of

chylomicrons

was visibleat_{the top of the tubescontaining postprandial plasmasamples.} Each

_chylomicron

fractionwasrecentrifugedfor24hat39,000rpm in

amodeltype40.3rotor

_(Beckman

_{Instruments, Inc.).}

Toexamine the MBl9phenotypeoftheapoB48andapoBI00,the

delipidated apolipoproteins

of the

_fasting

VLDLand thepostprandial

chylomicron

fractionwereseparated by 3-12%gradientSDS-PAGE under

_{nonreducing conditions,}

thenelectrophoreticallytransferredto

asheetofnitrocellulosemembranefor

immunoblotting

(28).The posi-tions ofthe

apoB48

and

_apoBI00

bandsonthe membranewere

identi-fied

_by

_staining

the membrane with Ponceau S. The nitrocellulose membranewasthen

simultaneously

incubated withI07 cpmof

'25I-an-tibody MBl9/ml (to

detect the

_{polymorphism)}

andI07cpm of

'3'1-an-tibody

MB3/ml (to

quantitate

theapoBtransferredtothemembrane).

Labeling

of

immunopurified

monoclonalantibodiesto high specific activities

_{(15,000 cpm/ng)}

was

performed using

the lactoperoxidase method

_{(Enzymobeads;}

Bio-Rad

Laboratories,

Richmond, CA).After

theblotswere

washed,

autoradiography

was

performed.

TheapoB48 and

apoB

100bandswerethen removedfrom thenitrocellulose

mem-brane and counted ina_{double-channel gamma}counter.Theratio of

1251/131I

countsin each bandwascalculated.Chylomicronand VLDL fractions from three different

normolipidemic

controlsubjectswere includedoneach

gel.

The control

subjects

includedan

_MBl9I/MBl9,

homozygote,

an

MBI91/MB192

_{heterozygote,}andan

_MBl9JMBl92

homozygote. Previously,

wehave demonstrated thatthis double-label

immunoassay

canbeusedtoassesstheMBl9

phenotype

of_apoB48 and

_apoB100;

wefoundthat the

apoB48

andthe

apoB100

bandsof normal

_subjects

invariably

have thesameMBl9

phenotype

(28).This double-label assay has also been showntobeusefulforMBl9

pheno-typing

evenwhendifferent

apoB

species

withina

lipoprotein

sample havedifferent MBl9

phenotypes,

asis thecaseincertainsubjectswith

hypobetalipoproteinemia (32).

Asindicated inResults,we

invariably

found

_apoB100

in each

_chylomicron

fraction. To

verify

that the

apoB

100found in each

_chylomicron

fractionwasnotanartifact

result-ing

from contaminationbyLDL-sizedparticles,thesize oftheparticles

in the

_chylomicron

fractionwasevaluatedbynegative-stainelectron

microscopy,

asdescribed(35).

Ag

phenotyping

of

serum.Agphenotypingofserumwasperformed

by

the

_{passive hemagglutination}

_{inhibition assay,using}human red cells

_(type

0,

Rh-)

coatedwith

_purified

human LDLofanappropriate

phenotype

and human_Agantisera with_{corresponding specificities,}as

described in detail elsewhere

(36).

Totestfor the presence ofanti-apoBantibodies with Agspecificity

in the

_{postoperative}

serumof livertransplantpatients,adirect

hemag-glutination

assaywasused. Eachpatient'sserumwasdiluted 1:4 with

PBS,

and

_aliquots

wereincubated_separatelywithabatteryofsevenred cell

_suspensions

coatedwith LDLof differentAgtypes,coveringthe full spectrum of the 10 knownAgspecificities (36).Thepreparationof the LDL-coated red cells and the reactionconditionswereexactlythe

samefor the

Ag

phenotyping

discussedabove,except that the human

anti-Ag

serum

_(used

asreagent in the

antigen

test)wasreplaced byan

equal

volumeof PBS

(36).

Immunocytochemistry

ofliver and intestinal biopsy specimens.

Liver

biopsy

specimens

aswellasduodenal,

jejunal,

andilealbiopsy

specimens

wereobtainedfrom

eight

organ donorsubjects.Mostofthe

organdonors hadreceived

only

intravenous

feedings

forthe 24 hbefore

the

biopsy

procedure.

The

_specimens

were

_immediately

fixedin

forma-fin.After 4-12 h offixation, the samples were washed in PBS and incubatedovernightat4VCin PBScontaining18%sucrose.The tissue was frozen inliquid nitrogen, and cryostat sectionswerecut, mounted onslides, and incubated inablockingbuffer beforebeingused for immunocytochemistry. The sectionswereincubated withawide range of dilutions of mouse ascites fluidcontaining variousapoB-specific

monoclonal antibodies for 24h, washed, andsubsequentlyincubated withbiotinylated anti-mouseIgGand thenabiotinylatedhorseradish

peroxidasecomplexed toavidin (Vector Laboratories, San Francisco, CA) as described (37). For these studies, we used several different

apoB-specific monoclonal antibodies (29, 32) that bind both apoB48 and apoBl00 (MB24, MBl 1, MB3) andtwoantibodiesspecificfor the

car-boxyterminal portion of apoB100 (MB44 and MB43) that bind apoB100but not apoB48. Peroxidase activity was detected with di-aminobenzidine and nickel, and the sections were lightly

counter-stained with methyl green and then photographed.

Results

We obtained preoperative and postoperative plasma and serum specimens from 29 consecutive liver transplant recipi-ents,andpreoperative plasma andserumspecimens from their respectiveorgandonors. Inthree othercases,preoperativeand postoperative recipient specimenswereavailable, butnodonor specimenswereavailable. - 30% of therecipients underwent

livertransplantation for biliarytractdisease, and 70% for

par-enchymal liver disease. A single recipient with Wilson's disease represented the only caseofaninborn error of metabolism. The mean age of the recipients was 46 yr; only three were

younger than20yrofage.Themeanagefor theorgandonor patients was24, with a rangeof4 to 58. 68% of the organ

donors had sustained major trauma. Of the deaths from trauma, 65% were due to vehicular accidents. Most ofthe other donors diedas aresult ofsubarachnoidorintracerebral hemor-rhages.

The resultsofthe serum apoEphenotyping studiesare

sum-marized in Table I. In 11 of 26 cases, the apoE phenotype of donorserum andpreoperative recipientserum wasidentical,

sothese cases were not informative.However, in 15 cases, the donor andpreoperative recipientserumapoE phenotypeswere

different. In each of these cases, the postoperative recipient

serumapoE phenotype convertedtothatof the donor.Intwo

Table L Apolipoprotein E Phenotypes ofPreoperative Recipient

Serum,DonorSerum,andPostoperativeRecipient Serum

ApoEphenotype

Preoperative Postoperative No.subjects recipient Donor recipient

10 3/3 3/3 3/3

1 4/3 4/3 4/3

5 3/3 3/2 3/2

3 3/2 3/3 3/3

4 3/3 4/3 4/3

1 4/3 3/3 3/3

1 4/3 4/2 4/2

1 4/3 4/4 4/4

Donorserumnotavailable

1 3/3 3/3

1 4/4 3/2

of the three cases where no donor serum was available, a change in the recipient's postoperative apoE phenotype oc-curred. An exampleofan IEFgelshowingachange in apoE phenotype after liver transplantation isshown inFig. 1. The conversion of the recipient serum apoE phenotype to that of thedonorwascompleteinsofar as we could measure. In most cases, the apoE phenotyping system that we used was not suffi-cientlysensitiveto detect inthepostoperative recipientserum the presence of anapoEisoform madebytherecipient's extra-hepatictissuesbut not made by the donorliver.Occasionally, however,whenexamining postoperative recipientserum, pro-longedexposure of theautoradiogram oftheisoelectric focus-inggeldidreveal asmallamountofanapoEisoformthat was presentinthepreoperative recipientplasma,but not made by thedonorliver. Forexample,thepreoperative apoEphenotype

ofoneofthe liver transplant recipientswas4/4.3 mo

postopera-tively, the phenotypewasclearly 3/2, but a veryfaint apoE4

band wasdetectable

(Fig.

2, lane 5). Becausethe half-life of apoEin plasma is onlyafewhours(23), and because the

postop-erativerecipient blood sampleswerealwaystaken more than 1 moaftertheoperation, the apoE4 bandmusthave been dueto thesynthesis ofapoE4byextrahepatic tissues. Inseveral other cases, we wereable todetectasmall amount of extrahepatic

apoE synthesis (i.e., an apoE isoform madeby

extrahepatic

tissues, butnotthe donor

liver)

by

loading

severalmicrograms ofdelipidatedVLDL

proteins,

rather than serum,ontothe IEF

gel(datanotshown).Weestimatethat the

extrahepatic

contri-butiontothe serumapoE pool

invariably

represented signifi-cantly< 10%of the totalserumapoE.

ApolipoproteinEisaneasily detectable proteincomponent

of human and canine CSF (8, 38). The fact that astrocytes

synthesize apoE (7) stronglysuggeststhata

portion

ofthe apoE in CSF is produced locally in the brain. However,many serum

proteins, particularly those with low molecular

weight,

are pres-entintheCSF.Forexample,apoA-I, which hasamolecular

weightonlyslightlylower than thatofapoE(28vs.35kD), is

present in the CSF and appears to enter the CSF from the

Apo-E

Phenotype

4/4

E4-

*

E3

--E2

-3/3

3/3

4/3

3/3

.A

... ...

_wo

Apo-E

Phenotype

4/3

4/4

,

_.

E4-

4-C~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~m.4W_...of,

E-

_

E2

-3/2

4/4

3/2

~~~~~~1.! 2

1

2

3

4

5

Figure 2. A Western blotof anIEF gel demonstrating the apoE iso-forms in serum. The migration positions of E4, E3, and E2areshown

onthe left.(Lane 1) Serum fromacontrol subject with the 4/3 phe-notype. (Lane 2) Serumfrom a control subject, phenotype 4/4. (Lane 3) Serum fromacontrolsubject,phenotype3/2.(Lane 4) Preopera-tive serumfrom a liver transplant recipient, phenotype 4/4. (Lane 5) Postoperative serumfrom the same patient as in lane 4, phenotype 3/2. However, on theheavily exposedautoradiogramshowninlane 5,avery faint E4 bandis detectable. The apoE4in thislaneis from

synthesis of apoE4 in theextrahepatictissuesoftherecipient.

plasmacompartment(8, 38).We reasoned thatexaminingthe CSF of liver transplant patients would enable us to assess whether the apoE in CSForiginated mainlyfrom the plasma

poolor wasproduced locally.Ifboth the serum and CSFpheno-type changed after transplantation, it would suggest that the

CSFapoE was derived primarily from the plasma pool. If,on

the other hand, the CSF apoE phenotype did not change, it

wouldsuggest that most of the apoE was madelocally. Our studies on the CSF of liver transplant patients indicated that mostofthe apoE in CSF is madelocally (Table II).The six CSF samples that we obtained from liver transplant patients were sterile and had normal protein and glucose concentrations.

Studiesin twosubjects (subjects1 and 2 in TableII)were not informative because the apoEphenotype of thepreoperative recipient serum and that of thedonor serum were identical.

However, studiesinfour other

patients (subjects

3-6 in Table II)demonstratedadifference between theapoEphenotypeof the postoperative serum and that of the CSF. For example,

postoperatively, the serum apoE

phenotype

ofsubject 6 was 3/2, but the CSF apoEphenotypewas4/3(Fig. 3, lanes4 and

5), the same asthis subject's preoperativeserum phenotype.

Similarly,

in

subject

4, the serumapoE

phenotype

converted from 4/3to4/2postoperatively (Fig.4,lanes4and5), butthe

CSFphenotype remained 4/3 (Fig. 4,lane6).

1

2

3

4

5

Figure1.AWestern blotof anIEFgel demonstrating the apoE iso-forms in serum. Isoelectric focusing was performed according to the methodof Menzel and Utermann (22). The migration positions of E4, E3, and E2 isoforms are shown on the left. The cathode is at the top, and the anodeisatthe bottom. (Lane 1) A serum sample from

acontrolsubjectwith the 4/4 phenotype. (Lane 2) A serum sample from a control subject with the 3/3 phenotype. (Lane 3) A serum

samplefromaliver transplant donor, phenotype 3/3. (Lane 4) A preoperative recipient serum sample, phenotype 4/3. (Lane5)A postoperative serum sample from the same patient as in lane 4, phe-notype3/3. Minor sialylated apoE isoforms, which have a more acidic

pI,arevisible in each lane below the major apoE isoforms. For ex-ample, the band in lane 1 that migrates in the E3 position is mono-sialo apoE4.

TableII.Cerebrospinal Fluid ApolipoproteinEPhenotype afterLiverTransplantation

Preoperative Postoperative Postoperative recipient Donor recipient recipient

Subject serum serum serum CSF

1 3/3 3/3 3/3 3/3

2 3/3 3/3 3/3 3/3

3 3/3 3/2 3/2 3/3

4 4/3 4/2 4/2 4/3

5 3/3 4/3 4/3 3/3

Apo-E

Phenotype

4/4

4/3

3/2

3/2 4/3 E

4-E

3

E2

* *''§

_~~WI

1

2

3

4

5

Figure 3. A Western blotof an IEF gel demonstrating the apoE iso-forms inserumand CSF. (Lane1)Serumfrom a controlsubject,

phenotype 4/4. (Lane 2) Serumfromacontrolsubject, phenotype 4/3. (Lane 3) Serumfromacontrolsubject, phenotype 3/2. (Lane 4)Postoperative serumfrom liver transplant recipient 6 (see Table

II),phenotype3/2.(Lane5) Postoperative CSFfrom the same

recipi-ent asin lane4,phenotype4/3. The apoE from CSF ismoreheavily sialylated than plasma apoE, which accounts for the increased

amountofimmunoreactivematerial below themajorapoEisoforms (towarhthe anode).Alargeproportion of CSF apoE migrated more slowlyonSDS-polyacrylamidegels,compared with serum apoE (data notshown).Thischaracteristicof CSF apoE, which is due to a heavier

sialylation,has beenpreviouslynotedand characterized by Pitaset

al.(8).

Twoofthe livertransplant recipientswereparticularly

inter-esting becausetheyinvolved subjects with the4/4 phenotype. ApolipoproteinE4containsnocysteine residues and therefore

cannot

participate

ineither the formation of apoE-A-II hetero-dimers (25)or apoE homodimers (27). One liver transplant recipient, who hadserumapoEphenotype 4/4 preoperatively, convertedtophenotype 3/2postoperatively.NoapoE-A-IIor

apoE dimers were present in'theserumpreoperatively (Fig. 5, lane 2), butboth were easily detectable in thepostoperative

Apo-E

Phenotype

^

E4-E3_

4

E2

Figure 4.AWesteri forms inserumand

phenotype3/2. (La 4/3.(Lane3)Serun (Lane4)Preoperati

ent4in TableII),p

thesamerecipienta CSFfrom thesame

theapoEin CSF is acidic isoforms thai

serumapoE2focus monosialo E3 (com theCSF(lane 6)foc

asignificantamoun

cipient 4,it would b (lane 6).

ApoE

Phenotype

0

_3/2

H.ade

ApoE

1.

Homodimer-_

4/4

3/2

ApoE-A-I

I-ApoE

I

_i

1

2

3

Figure5.ApolipoproteinEimmunoblotsof nonreducedserum

sam-pleselectrophoresedon anSDS-polyacrylamide gel. 1 mlofserum

wasmixedwith 30

Al

of a sample buffer thatcontainednoreducing

agents andwasloadedontoa10-209% gradient SDS-polyacrylamide

slabgel. Afterelectrophoresis,theseparatedproteinswere electro-phoretically transferredto asheetof nitrocellulosemembranefor

im-munoblottingwithanapoE-specificantibody. (Lane 1)Serumfroma

controlsubject,phenotype3/2. (Lane2)Preoperativeserumfrom

aliverrecipient,phenotype 4/4.(Lane3) Postoperativeserumfrom thesamerecipient,phenotype3/2.Theisoelectricfocusing gelsfor the threeserumsamples shown in thisfigureareshown in lanes 3-5 ofFig.2. No apoE-A-II heterodimersorapoE homodimerswere ob-served when thesampleswereincubated with 2%,B-mercaptoethanol before electrophoresisontheSDS-polyacrylamide gel (datanot shown). The apparent molecularweightof theapoE-A-IIwasexactly

thesamein lanes 1 and 3 (seeFig.6). Thiswasalso thecasefor the apoE homodimer. The minorbandsabove theapoEhomodimer band in lanes 1 and 3 represent variousoligomersthatarepossible becauseapoE2containstwocysteineresidues(comparewithFig. 6,

lanesIand3, wherenoapoE2exists).Theidentityof theapoE-A-II

complexwasconfirmed with immunoblotsusinganapoA-IIspecific antibody(datanotshown).

serum(Fig. 5, lane3),

indicating

that

hepatic

apoE

participates

3/2

4/3

4/2 4/3 4/2 4/3 in

disulfide-linked

complexes

with

_apoA-II

andwith

itself

_(Fig.

5). Anothersubject had phenotype 4/3 preoperatively and re-ceivedaliverfromadonorwith the4/4

phenotype.

Immuno-blots of the donor serum revealed the complete absence of Bz stV _

apoE-A-II

heterodimers and

apoE

homodimers

(Fig. 6,

lane

1).

Therecipientserum,obtained 6mopostoperatively, had

atypi-_cal

4/4

phenotype,

as

judged by

theimmunoblot-TEFgel

tech-nique.Onimmunoblots

of SDS-polyacrylamide

gels thatwere

performed

under

nonreducing conditions, apoE-A-II

hetero-1 2 3 4 5 6 dimersandapoE homodimerswerevisibleinthe postoperative

recipient

serum

(Fig.

6,

lane

3),

although they

were atleast

nblot ofan IEFgeldemonstratingtheapoE fson r t t i

ICSF. (Lane1)Serum from a controlsubject I0-fold

fainter,

relative

tothe

itensity

ofthe

apoE

monomer, Lne2) Serumfrom a controlsubject, phenotype than the dimer bands in preoperative recipient serum (Fig. 6,

n from a liver transplant donor, phenotype 4/2. lane 2). Because the donor liver synthesized only apoE4, which

ive serumfrom a livertransplantrecipient(recipi- cannot

_participate

in dimer formation, the observed dimers

)henotype 4/3. (Lane 5) Postoperativeserumfrom musthaveinvolved the

apoE3

synthesized by extrahepatic

tis-Isinlane 4, phenotype 4/2. (Lane 6)Postoperative sues.Thesestudies indicate thatextrahepaticapoE can

partici-recipient, phenotype 4/3.Fig.3illustratesthat pate in disulfide-linked dimer formation. Interestingly, in the moreheavily sialylated and contains more minor latterrecipient, both the

apoE-A-II

heterodimers and the apoE plasmaapoE. Fortunately, with this IEF system, homodimers in the postoperative serum had a slightly higher

,esatadistinctivelymorebasic position than apparent molecular weight compared with those in the

preoper-ipare lanes 3 and5).The monosialo E3 band in

usesmore acidically than asialo apoE2. Thus, if

ative

sample

(Fig.

6, lane 2 vs.

lane

3). The higher molecular

Lt ofasialo apoE2 were present in the CSF of re- weight of these disulfide-linked apoE complexes, compared havebeendetectable above the monosialo E3 band with those found in the preoperative sample and other control

samples,

was also observed in two

subsequent experiments.

ApoE

Phenotype

_

4/4

4/3

4/4

ApoE

-

a

Homodimer

ApoE-A-I I

ApoEU

1

1

2

3

Figure 6. Apolipoprotein E immunoblots of nonreduced serum sam-ples electrophoresed on a 10-20% gradient SDS-polyacrylamide gel. See legend toFig. 5 for conditions. (Lane 1) Serum from the donor, phenotype 4/4, as assessed by isoelectric focusing. (Lane 2) Preopera-tive serum of the liver transplant recipient, phenotype 4/3. (Lane 3) Serum from the same recipient obtained 6 mo postoperatively. By isoelectric focusing, this serum sample had a typical 4/4 phenotype. The slightly higher migration of the apoE-A-II heterodimer and the apoE homodimers in the postoperative serum (lane 3), compared with thepreoperative serum (lane 2), was a consistentfindingin multiple experiments. The higher migration of the dimers was not observed after the serum was treated with neuraminidase (data not shown).

The slowermigration of these disulfide-linked complexes in the postoperative serum was eliminated by neuraminidase treatment of the serum (data notshown).Thisfindingsuggests that the serum apoE that isproducedinextrahepatic tissuesis moreheavily glycosylated than the serum apoE that is derived from the liver.

The MB19 phenotype of the apoB 100 in plasma was

as-sessedby RIAusing the apoB-specific antibody MB 19. Because apoB48is present in only trace amounts in theplasma,relative toapoB100, RIAs of plasma samples assess only the apoB 100 phenotype. The MB 19phenotype oftheapoB100 in postopera-tive recipient plasma converted to thephenotypeof the donor inallfour subjectsstudied(TableIII). Competitive RIAs that

demonstrate changes inthe MB19 phenotype ofLDL samples areshown inFig. 7.

Prior studies have proven that allotypes MB 191and MB 1 92 can bedetected by humanantisera specific fortheAg(c)and Ag(g)epitopes, respectively(39). The latter twoepitopes repre-sent twoof10differentAgepitopesthat canbedetectedusing

antisera frommultiplytransfused thalassemic patients (36).All 10antisera were used toanalyzethe changes in apoB100 pheno-type in 15livertransplantpatients. We foundthat the

postoper-ative recipient apoB phenotype, asjudged byall 10 markers, wasinvariably identical to that of the donor (Table IV). Be-cause manyliver transplant recipientswereexposedto new Ag

antigens producedby the donor liver, we testedwhetherany of the 15liver transplant recipientshaddeveloped antibodies

spe-cificfortheforeign Agallotypes produced by the donor liver. However, we did not detect anyAg-specific antibodiesinany ofthe postoperative recipient serum samples, each of which wastaken more than 3 mopostoperatively.

To determine whether the MB19 phenotypes of apoB48 changed after liver transplantation, we used a double-label

Western blotimmunoassay(28, 32). We obtained plasma from fourlivertransplant recipients after they hadfasted for 14 h,

TableIII. The MBJ9 Phenotype of thePreoperativeRecipient,

Donor,andPostoperative RecipientPlasmaApolipoprotein

BJ00

from

28Liver

_{Transplant Operations}

MB19phenotype

Preoperative Postoperative No.subjects recipient Donor recipient

8 2/2 2/2 2/2

1 1/2 1/2 1/2

8 1/2 2/2 2/2

7 2/2 1/2 1/2

3 1/1 1/2 1/2

1 1/2 1/1 1/1

Phenotypingofplasma sampleswasperformed bydetermining their MBl9 binding pattern by radioimmunoassay (17). Allotype I

(MB191)hasahigh affinityfor antibody MB 19; allotype 2 (MB192)

hasalowaffinityforantibodyMB19. Plasmas yieldingaweak pattern ofimmunoreactivity with antibody MB 19 have phenotype 2/2;an intermediate pattern, phenotype 1/2;astrong pattern, phenotype 1/1

(17).The MB 19phenotypeof the preoperative recipient plasma and thedonor plasma correlated perfectly with those predicted froman ApaLIdigestion ofanapoB genefiagment amplified from the geno-micDNA of these subjects (18). The 28 complete sets of plasma samples (pre- and postoperative recipient samples and donor sample) shown in this table include the 26setsofplasma samples shown in Table I, plusacomplete set of samples from two other liver transplant operationsnotincluded in the original 29. The postoperative apoB phenotypes were, in each case, unequivocal. The assays revealedno

evidence ofasmallamountof contamination from thepreoperative

MB 19 phenotype suchaswould occur if there were a significant

amountof extrahepatic synthesis of apoB 100. The assay thatweused is sufficiently sensitivetodetect easilya5-10% extrahepatic synthesis of anMB191/MB 192 LDL in a subject who had receivedaliver from anMB 192/MB 192 donor.

thenpreparedtheVLDLby ultracentrifugation.The apoB100

andapoB48wereseparatedfrom the otherVLDL

apolipopro-teinsby SDS-PAGE.Examination of the postoperative VLDL samples from four normolipidemic liver transplant patients

confirmed that thephenotypeofthe apoB 100 convertedtothat ofthe donor liver (Table V). In contrast, the postoperative

MBl9 phenotype of the apoB48 wasinvariablythe same as

that of the preoperative plasma of the recipient and didnot changetothephenotypeofthe donor (TableV).

We then evaluated the "chylomicrons" that were isolated from plasmaobtainedfrom thesamefour subjects 2 h aftera

fat-rich meal. (None of the subjects had detectable

chylomi-cronsunder fastingconditions.)Examination of the

postpran-dial chylomicrons by Coomassie blue-stained

SDS-polyacryl-amide slab gels revealed that the apoB 100 band stained slightly

moreintenselythan the apoB48 band (data notshown). This

findingwas inkeepingwith the results that wepreviously ob-servedinnormolipidemic control subjects (28). To exclude the possibility that the apoB100 in this fraction was an artifact resulting from contamination by particles of the LDL size range, weexamined each of the chylomicronpreparationsby negative-stain electronmicroscopy.Thelipoproteins were het-erogeneous in size, with most particles havingadiameter of

1.1 1.0 0.9 0.8 o 0.7 I- 0.6

Im

0.5 0.4 0.3 0.2 0.1

1.1 1.0 0.9 0.8

a0

0.7 o _0.6

cos

0.5

0.4 0.3 0.2 0.1

1.1 1.0 0.9 0.8

0 _0.7

Im

0.6

m

0.5

0.4 0.3 0.2 0.1

10 100 1000

ng

Apo-B Added

per

Well

10000

Figure7. Competitive RIAs demonstrating a change in theMBl9

phenotype of LDL samples that were prepared from plasma of liver transplant donors and recipients. The results of each assay are plotted

asB/Bovs.logof the apoB protein added per well, where B andBo representspecific counts bound in the presence and absence of

com-petitor, respectively.Ineach panel, (.. ._{*) represents a control LDL}

with a "strong"MBl9binding pattern, or theMB19,/MB191 pheno-type. This controlLDLwasisolated by ultracentrifugation from the plasma of a laboratory employee.Ineach panel,(o o) represents theLDLof thepreoperative recipientLDL;(. n), postoperative

recipientLDL;(A A), donor LDL.In AandB,the donor pheno-typewasMB191/MB192(an "intermediate" binding pattern, the re-sult ofanequal concentration of apoB allotypesMB191and MB192)

and thepreoperative recipient phenotypes wereMBI9JMB192 (a

"weak"binding pattern). The phenotype of therecipient's postopera-tive LDLwas

MB19,/MB192.

Lowdensity lipoproteins having phe-notypeMB191/MB192aredisplaced approximately twofold to the

right,compared with the phenotypeMB191/MB191standard LDL (see reference 19).InC,thedonorphenotype wasMBl9JMB192; the pre- andpostoperative recipient phenotypes were

MB191/MB192

andMB199JMB192,respectively. An Apa LI digestion of an apoB genefragment amplifiedfrom thegenomicDNA(20) of the donor shown in Cpredicted that the donor plasma would have phenotype

MBl9JMBl92by immunoassay. The immunoassay shown in C sup-ports thisprediction,although the pattern of immunoreactivity is

slightlystronger than usually seen withLDLsamples with phenotype MBl9JMBl92.Thatis,thedonorLDL

competition

curveisshifted

TableIV. AgPhenotypesof theLiverDonor,Preoperative Recipient, and Postoperative Recipient Serum

Ag

Subject x y a, d c g t z h i

RI + - + - - + + - - +

Dl - + + + - + + - - +

R1F/U - + + + - + + - - +

R2 + + + + + + + - + +

D2 + + - + + - + - - +

R2 F/U + + - + + - + - - +

R3 + + + + + + + - - +

D3 - + + + - + + + - +

R3F/U - + + + - + + + - +

R4 + + + + - + + + - +

D4 - + - + + + + - - +

R4F/U - + - + + + + - - +

R5 + + + + - + + + + +

D5 - + + + + + + - - +

R5F/U - + + + + + + - - +

R6 - + - + + + + - - +

D6 - + - + + + + + + +

R6 F/U - + - + + + + + + +

R7 - + + + - + + + - ₊

D7 - + - + + + + - + +

R7F/U - + - + + + + - + +

R8 + + + + + + + + - +

D8 + + + - - + + - - +

R8F/U + + + - - + + - - +

R9 - + + + - + + - - +

D9 - ₊ + + + + + + - +

R9F/U - _{+ +} + + + + + - +

The presence (+)orabsence(-) of 10 different Agantigenswas

as-sessed in thedonor,preoperative recipient,andpostoperative recipi-entserumaccordingtopublished techniques (34).Donorsare desig-natedDl-D9;preoperative recipientsaredesignatedRl-R9; postop-erativerecipientsaredesignated RI F/U-R9 F/U.Shown hereare results from 9 of 15 livertransplantation operationsthatwereassessed by Agphenotyping.Inall 15livertransplantation operationsthat wereassessed, the Ag phenotype of the liver transplantrecipient

con-vertedtothatof the donor.

examined by the double-labelWesternblotassay. Ineach of

thefourrecipients, theMB19phenotype of theapoB

100

was

identicalto the

MBl9

phenotype of the donor liver, whereas

the MB 19phenotype oftheapoB48wasidenticaltothat pres-ent in the preoperative plasma ofthe recipient (Table VI). Thesedataindicate thatthe vastmajority oftheapoB100that appears in the plasma chylomicron fraction after a fat-rich mealisof hepatic, notintestinal,

origin.

Our results with the

1251/131I

Western blot assay indicate

that thelarge majority of apoB48 ismadein the intestineand that the large majority of apoB100 (eventheapoB100 inthe

somewhattotheleft, comparedwith thecurves

usually

observed with LDLsampleshavingthe MB19JMB

192.

Thisobservationismost

likely explained bythe fact that the donor in C hadreceived10 Uof

plasmaorwhole blood within 12hof the

_{liver-harvesting operation,}

atwhich time the bloodsamplewasobtained.

Table V. TheMBI 9Phenotype ofApolipoprotein BJ00andApolipoproteinB48 in thePostoperative VLDL FractionofFour Liver

Transplant Recipients,asAssessed byaDouble-LabelWestern Blot Assay

Plasmaradioimmunoassay:MBl 9 phenotype of the plasma apoB100 Double-label Western blot assay: Ratio of'"I-MB19cpmto'3II-MB3cpm Control Preoperative Postoperative

subjects recipient Donor recipient ApoBlOO ApoB48 Experiment I

Control subjects

I 1/1 - - 6.8 6.6

2 1/2 4.0 3.8

3 2/2 - 1.2 1.8

Liver transplant recipients

I 1/2 2/2 2/2 1.2(2/2)* 4.2 (1/2)*

2 1/2 2/2 2/2 1.4(2/2) 4.7(1/2)

Experiment 2 Controls

1 1/1 - 4.3 4.3

2 1/2 2.9 2.3

3 2/2 1.2 1.0

Liver transplant recipients

2 1/2 2/2 2/2 0.8(2/2)* 3.0(1/2)*

3 2/2 1/2 1/2 3.1(1/2) 1.2 (2/2)

4 2/2 1/2 1/2 2.9(1/2) 0.9(2/2)

Shownontheleftside of thistablearetheMB19apoB100phenotypesof eachsubjectstudied. The MB 19phenotypeof theplasmaapoB100was assessedby RIAofplasma, exactlyaspreviously described (17) (see Methods). Shownontherightsidearethe results of the double-label Westernblot assayof the apoB100andapoB48 in the VLDLfraction. TheVLDLfractionswerepreparedbyultracentrifugationfrom theplasma

offourpostoperative normolipidemiclivertransplantpatientsand controlsubjects.Eachplasma samplewasobtainedaftera14hfast.*The MBl9phenotypeof theapoBl00andapoB48bandsofthepostoperativelivertransplantrecipients'VLDLis shown inparentheses.The MBl9 phenotype of these bandswasdeterminedbycomparingthe 1251/13'Iratios in theapoB48andapoB100 bands with those observed in control

subjects.

TableVI. TheMBI 9PhenotypeofApolipoprotein

BJ00

andApolipoproteinB48 inthe Postoperative ChylomicronFractionofFour Liver

Transplant Recipients,asAssessedbyaDouble-LabelWesternBlotAssay

Plasmaradioimmunoassay:MBl9 phenotypeofthe plasma apoBI00 Double-label Western blot assay: Ratio of'"'I-MB19cpm to'31I-MB3 cpm Control Preoperative Postoperative

subjects recipient Donor recipient ApoBlOO ApoB48

Controlsubjects

I 1/1 6.4 6.7

2 1/2 - 4.7 4.5

3 2/2 1.3 1.0

Liver transplant

recipients

1 - 1/2 2/2 2/2 1.0 (2/2)* 5.5 (1/2)*

2 1/2 2/2 2/2 0.9 (2/2) 5.5 (1/2)

3 2/2 1/2 1/2 5.8 (1/2) 1.6 (2/2)

4 - 2/2 1/2 1/2 5.2 (1/2) 1.1 (2/2)

Shownonthe_{left side of this table are the MB 19 apoB 100 phenotypes of each subject studied. The MB 19 phenotype of the plasma apoB 100 was}

assessed byRIAof plasma(17)(see Methods). Shown on the right side are the results from the double-label Western blot assay of the apoB100

and apoB48 in the chylomicron fractions. The chylomicrons were prepared by ultracentrifugation from the plasma of postoperative liver trans-plant_{patients. The plasma samples were obtained 2 h after a fat-rich meal (see Methods). Liver transplant recipients 1-4 in this table are identical}

torecipients 1-4 in Table V.*_{TheMBI9phenotype of the apoB100 and apoB48 bands of the postoperative liver transplant recipients'}

"chylomicron fraction") is made in the liver. However, it is

importantto add acautionarystatement regarding the sensitiv-ity of the

1251/1311

IWestern blot MBl9 phenotyping assay. This assay system is probably not sufficiently sensitive to detect a 5%

hepatic contribution to the apoB48 pool or a 5% intestinal

contributiontothe apoB 100 pool. However, control LDL and

postprandialVLDLmixing experiments(data not shown) have

indicatedthat the Western blot assay is capable of detecting a 15% contribution ofthe liverto the apoB48 pool or a 15%

contribution oftheintestinetothe apoB 100 pool.

Toexaminedirectly thepossibility ofapoB100 synthesis by

the intestine,weusedimmunocytochemical techniques to

ex-amine the duodenal, jejunal,andileal biopsy specimensas well asliver biopsy specimens from eightorgan donors.

Monoclo-nal antibodies that bind to the amino-terminal portion of

apoB100 (MB3,MBI 1, and MB24) bound strongly to the liver

biopsy specimensaswellas to alloftheintestinalbiopsy speci-mens. Antibodies bindingto the carboxy-terminalportion of

apoB100 (MB43 and MB44) bound to the apoB in hepato-cytes.These latterantibodiesdetected apoB100 in the vascular

submucosa oftheintestinalenterocytes; however, even when

the carboxy-terminal antibodies were incubated with the

biopsy specimensathigh concentrations,nospecific stainingof

intestinalenterocytes was observed. Examples of our

immuno-cytochemical

studiesareshown in Fig. 8.

Discussion

Plasmaandserumsampleswerecollected from liver transplant

donors and recipients because we were convinced that the

study of these samples would yield unique insights into apoli-poprotein physiology. One of the questions thatwewantedto

addresswastheproportion ofserumapoE thatwassynthesized

inthe liverversusextrahepatic tissues. Prior studies had

demon-strated that apoEwassynthesized inawidevariety

ofextrahe-patic tissues and cells, including kidney, adrenal, skeletal

mus-cle, monocyte-macrophages, and the central and peripheral

nervoussystems(40-47). We believed thatwecould address

the issue of the proportion ofserumapoE that originates in the

liver byexamining the postoperativeserumof liver transplant

patients who had receivedaliver fromasubject withadifferent

apoE phenotype. This ideawasnotuniquetoour group;just

months afterwebegantocollectsamples forourstudy, Kraftet

al.reportedaretrospective study ofchangesinapoE,apoA-IV,

andLp(a) phenotypes in 18 liver transplant patients (48). Al-though donorserumwasunavailable formostof thecases

ex-aminedby these investigators, they nevertheless demonstrated convincingly that the serum apoE phenotype frequently

changes after liver transplantation. Our study, performed on

donorandrecipientserum samples collected inaprospective

fashion, confirmed this finding. We found that the

postopera-MB3

_MGolgi

_apparatus

* MB44 Figure 8. Binding of two_{clonal antibodies, MB3 and}apoB-specific

_MB44,

_{to liver and}

mono-appat ,us/ duodenal

biopsy specimens

fromanorgan

do-A , _nor.* The^ */epitopeforantibody MB3,which binds

tobothapoB100andapoB48 (29),is located 4

#tt;1 #? > betweenapoB100aminoacids 994 and 1084

jt

a

M

s.

_Vascular

. u f

v =

(59); the

epitope

for MB44, which binds

to

*LVfl,core --W Y Y

_tapoBIOO

but not apoB48 (34), is located between

Va

SCU:

tar

_,-

_.apoBIOO

amino acids 2488 and 2658(59). (A)

Vascurar

'i

:/S.

v

/

The

m binding of

MB3

to a

_{duodenal biopsy}

speci-core t t -jimen;T (B)the binding of MB44 to a duodenal

biopsyspecimen; (C)thebindingofMB3 toa ;.,;ti'.y liver biopsy specimen;(b)the bindingofMB44

z

v , e *a toaliver biopsy specimen.Forthebiopsy speci-itle a > A^ #mens shown in each panel,adilute solution of

theapoB-specificmonoclonalantibody(1/

A

' B 20,000dilution of ascitesfluidfor MB3 anda

1/1,000

_dilution

of ascites fluid for MB44) was MB44

},.,GoIgi

incubated with the tissue samples.

MB3

binds

'4

*

k

^

apparafus

strongly

to the Golgi

apparatus

of

intestinal

epi-thelial cells

_(but

not tothe

mucus-secreting

Golgi

apparatus

cells).

It also

binds to the

Golgi apparatus

of

liver

..,

". parenchymalcells.Bindingis also present in the vascularcoreof theintestinal villi and along the

Blood

sinus * blood sinusesoftheliver.

MB44

bindstothe

Golgi apparatus of liver parenchymal cells; it

-?_$,~ / *\ \ ' bindstothe vascular intestinalvilli core, butnot tointestinal

_epithelial

_cells.

Withtwoother

Blood\

monoclonal antibodies that bindtothe

S* \fBlood s _{amino-terminal}inus _portion

_ofapoB100

_(MB 1and MB24),weobtained results identicaltothose shown here withantibody MB3. One other anti-CGll~rr|_0>4*~gD bodybindingto thecarboxy-terminalportion of

apoB100 (MB43) yieldedresultsidenticalto

thoseshown withantibodyMB44. Inparallel experiments,wewereunabletodemonstratebindingof eitherantibodies MB43orMB44tointestinal enterocytes whenappliedtothosebiopsy specimensinhighconcentrations(1/50dilution of ascitesfluid).Duodenalbiopsyspecimensareshown inpanelAand B;wewerealsounable

todetect apoB100in thejejunalandilealbiopsy specimens.

tive serumapoEphenotypeof therecipientnotonly changes,

but invariably converts to that of the donor (Table I). The change in serumapoEphenotype,asjudged byIEFgels,was almostalwayscomplete;weoccasionally found evidence foran

isoform synthesized by extrahepatic tissues of the recipient (Fig.2). We agreewiththeconclusion of Kraftetal.(48) thatat least 90% ofthe serum apoE is synthesized by the liver. By investigating apoE mRNA levels in avariety of tissues of non-humanprimates, Newman et al.(46) concluded thatfrom 20

to 40% of apoEmaybe

synthesized

extrahepatically

in these species. It remains conceivable that 20-40% ofapoEis

synthe-sizedin extrahepatictissues in humansbut, because of local

catabolism ofthe

protein,

thisrelatively high degree of

extrahe-paticsynthesis isnotreflectedintheserumapoE pool. Roheimandco-workerspointedoutin 1979 that relatively high levels ofapoE are present in normal humanCSF (38).

Theyrecognizedtwopossiblewaysthatapolipoproteinscould enterthe CSF: through local production and then diffusion

intotheCSForthroughcrossing the blood-brain barrier from the plasma,asis the case for certain other serumproteins. Sub-sequentstudies demonstrated that brain tissue

(43,

44), and

specifically theastrocyticglia ofthebrain (7),

synthesizes

apoE. Pitaset al.(8) recently characterized the

apolipoproteins

and

lipoproteins ofhuman and canine CSF, and, liketheearlier studiesof Roheimetal.(38), they foundthatapoEwaspresent

inarelativelyhigh concentration in CSF. They also found that CSF apoEwas more

heavily

sialylated

than

plasma

apoE

(8).

Thesefacts suggestedthat at least a

portion

oftheapoE in CSF isproducedlocally. Ourstudy on the CSFoflivertransplant

patients addressedtheissue oftheorigin ofCSFapoEinanovel way: by examining the CSF of four liver transplant patients in whom we had observed a virtually complete change in serum apoEphenotype. Wefound thattheCSFapoEphenotype did

notchangetothe phenotypeof the donor liver, implying that

the majority of CSF apoE must be synthesized locally

(Ta-bleII).

WeisgraberandMahleyreportedin 1978 that human apoE can form disulfide-linked heterodimers with

apoA-II

(26). Theydemonstratedthattheseheterodimersare presentinthe plasma lipoproteinsand are not an artifact produced during the preparation ofplasma lipoproteins. Apolipoprotein E4 cannot participate in apoE-A-II formation because itlacks a

cysteine residue (25). Recently, Weisgraber andShintohave

demonstrated that apoE3andapoE2, butnotapoE4, canform disulfide-linked homodimers (27, 48a).Inthis study,we were able todemonstrate that bothhepaticapoE andextrahepatic apoEcanparticipatein theformation of these disulfide-linked complexes

(Figs.

5and6).Intheir

original

description

ofapoE-A-II

heterodimers, Mahley and Weisgraberpointed out that

uncertainty

remainedabout whetherthis disulfide-linked com-plex was formed intheplasma or wasformed intracellularly

andsecreteddirectly fromcells(26). Ourexperiments

demon-strating the participation of extrahepatic apoE in

apoE-A-II

formationsupport theformer possibility. AlthoughbothapoE and

apoA-II

are synthesized in extrahepatictissues, they are

synthesizedin

different

extrahepatictissues.Theextrahepatic synthesis of

apoA-II

occursprimarily inintestinalenterocytes

(49),atissue thatmakeslittleor noapoE (6, 44, 50).

ApolipoproteinEisknown to be posttranslationally

glyco-sylated

atthreonine 194

(5 1).

Inhumanserum,

only

20%of

theapoE is

glycosylated

(51).Thereissomeevidence that the

apoE

produced by

some

extrahepatic

tissueculture

cells,

such

as astrocytes andmonocyte-macrophages, is moreextensively glycosylated than the apoEinserum(47, 52). However,ithas

remained unclearwhetherthe apoE in serum that originates in

extrahepatictissues hasadistinctglycosylationpattern,

com-paredwiththe bulkofthe serum apoE,which originates in the liver. Data presented in this study suggest that the apoE in

serumthat originatesinextrahepatictissues may indeed havea distinct glycosylationpattern.We examined a liver transplant recipient (preoperative phenotype 4/3) who received a liver from a donor with the 4/4 phenotype. Postoperatively, the serumapoE phenotypehad a typical 4/4 pattern, as judgedby

IEF. Nevertheless,there were very small amounts ofapoE-A-II heterodimers and apoE homodimers in his plasma; because apoE4 cannot participate in dimer formation (25, 27), these dimers must have contained apoE3 produced by extrahepatic tissues. Both disulfide-linked complexesin the postoperative serum were of a slightly higher apparent molecular weight, compared with the dimers found in the preoperative serum. However,the sizedifference disappeared after neuraminidase treatmentof thepostoperative serum. These findings strongly imply that extrahepatic apoE is moreheavily glycosylatedthan hepatic apoE, and that the different glycosylation pattern

per-sistsin thecirculation.Ifthecarbohydrateofextrahepatic apoE

were processed in the circulation(eitherenzymaticallyor

non-enzymatically)sothat theglycosylationpattern conformed to the glycosylation pattern observed for the bulk ofserumapoE,

one would not expect to observe thehigherapparent molecular weight of the disulfide-linked complexescontaining

extrahe-patic apoE.

We alsostudied changesin thephenotype ofapoB100 after liver transplantation, both by immunoassays using the

allo-type-specific monoclonal antibodyMBl9 andimmunoassays usingAg-specific antisera. Usingthe MB19

immunoassays,

we

demonstrated in 28 livertransplantpatientsthat theapoB100

phenotypeinvariably changes completelytothat of thedonor

(Table III). Thisfinding, which wasconfirmed byAg

pheno-typing studies (Table IV), is inkeepingwith thegenerally ac-cepted notion that the liver synthesizes the vast majority of apoB 100 in humanplasma(12,53).TheAg-specificantibodies

that we used for apoB

phenotyping

were derived from the serum ofmultiply transfused thalassemic patients. We tested the hypothesis that some liver transplant recipients might

makeantibodiestonovelAgantigens synthesized bythe donor

liver; however, no such antibodies were detectable, perhaps

because of the therapeutic immunosuppression regimen pre-scribed for livertransplant recipients.

Apolipoprotein B48is synthesizedin the intestineas a re-sult ofthe editing ofasingle nucleotide of the apoB mRNA, which creates a premature stop codon (54, 55). The editing

process in humans isdevelopmentally regulated; inorgan

cul-tures, fetal intestine

synthesizes

primarily apoB100, whereas

adultintestine synthesizes primarily apoB48 (56). Kane etal.

(10)havedemonstratedthatapoB48isvirtuallytheonly apoB species inchyle, obtained from thoracic duct drainage, clearly implicatingtheintestineas amajorcontributorto theapoB48

observed intriglyceride-richlipoproteins afterafat-richmeal. Recently, however,Higuchiet al.(11)havereported thatavery small percentage ofhuman hepatic apoB cDNA clones have the edited nucleotide, raising the possibility that the human liver might synthesize small amounts ofapoB48. Ifa small percentageof the liver'sapoBsynthesiswereapoB48, it would

of hepatic tissue,asignificantfraction of the apo-B48 infasting plasma could be derived from the liver. To address directly the

question of the tissue originofthe apoB48 in fasting plasma, we

simultaneously determined the MBl9 phenotype of the apoB48 and apoB 100 in the fasting VLDL of four

normolipi-demic liver transplant patients using a double-label Western blot assay (28, 32). Each patient had received a liver from a

donor with a different MB19 phenotype. We demonstrated

that the MBl9 phenotype of the apoB100 infasting VLDL

changed to the phenotype of the donor. However, it was

equally clear thattheapoB48phenotype did not change (Table V), demonstratingthat the liver is not asignificant source of

apoB48 in fasting plasma.Itis importantto point out that our

conclusionsarebasedupon theanalysisof a small number of

patients,and it remains conceivable that the liver could make a

contributiontoapoB48 synthesisin someindividuals. OurMB19 phenotypingstudies on the postprandial plasma

chylomicron fraction of liver transplant patients were quite

informative. First, these studies confirmed that the apoB48 presentinthe buoyantlipoproteins aftera meal issynthesized by the intestine; ifthere were ahepatic contributionto post-prandial apoB48 production in humans, it would have to be

quite small. Second, and equally interesting,wefoundthatthe

apoB 100 that is contained withinthechylomicronfraction is of

hepatic origin. Previously,thefactthat the large, triglyceride-rich lipoproteins in postprandial plasma contained a signifi-cant amountofapoB 100 had suggested to some investigators

the possibility that the intestine might synthesize some apoB100. Recently, Cohn andco-workers reported that the apoB100 contentofhuman VLDL rapidly increases after a

fat-rich meal; however, these investigators wereunable to be

certain abouttheorigin of thisapoB 100 (57). Our data in liver

transplant patientsshowthatthe apoB100inthepostprandial

plasma"chylomicronfraction" is derived from the liver. Our

finding is consistent withthe hypothesis that the human liver

rapidlysecretes large, buoyant,triglyceride-richapoB

100-con-taining lipoproteins

into the plasma inresponse to the uptake

of theremnantlipoproteins of intestinal origin(57, 58).

Ourgenetic evidencethattheapoB100 inthe plasma

chylo-micron fraction is ofhepatic origin (Table VI) isconsistent with theresultsofourimmunocytochemistrystudies (Fig. 8).

Using

severalmonoclonal antibodies that bindto apoB 100 but notapoB48,we were unable to detect apoB100in the intestinal enterocytesoforgan donors. In contrast to our results,

David-son etal. (15),Hoeg et al.(16), and Levy et al. (17) recently

reported small amountsofapoB100 synthesisbynormal hu-manintestinal biopsy specimens in metaboliclabeling studies. Also,Dullaartet al.(13),Hoeg et al.(16),and Levy et al.(17)

have reportedimmunocytochemical evidencefor apoB100

syn-thesis innormal intestinal biopsy specimens. Thereason for

the

discrepancy

between their findingsand oursisnot clear. Wedidnot do anymetaboliclabeling studies on biopsy speci-mens. In ourimmunochemicalstudies, we used different (but

nevertheless thoroughly characterized) apoB-specific monoclo-nalantibodies.One possibilitythatmightexplain the discre-pant findings is that our immunochemical studieswere not

sufficiently

sensitivetodetect smallamountsofapoB100 in the intestinalbiopsy specimens.However,itisnoteworthy that we were unable to detect apoB

100

in the intestinal mucosal cells even when theintestinal biopsy specimenswere incubated with veryhigh concentrations ofseveraldifferentapoB

100-specific

antibodies; with dilute concentrations ofthese antibodies, we

could detect apoB100 intheliveras well as in thesubmucosal tissues ofthe intestine. Another possibilityis that there may be asignificant natural variationinthe amount of apoB100

pro-duction by enterocytes of differentindividuals; it is possible

thattheintestinal enterocytes ofouradultorgan donors, unlike

thenormal subjects studied by other investigators (13, 16, 17), simply didnotmakeapoB100. It is conceivable that the adult human intestine respondsto the stress of hospitalization, life

support measures,andtreatmentofmajortrauma or a

cerebro-vascularaccident (a virtually universal situation with organ

donors) byreducingor"shutting off" theverylow level synthe-sis of apoB100 thatoccurs under normal conditions.

Acknowledgments

Wewould like to acknowledge M.Hardimanfor technical assistance, K.Weisgraber, S. Kunitake, and L. Curtiss forapoE-specific antibod-ies, L. Curtiss and J. Witztum forapoB-specificmonoclonal antibod-ies, W. Concepcion and P. Nakagato for help in obtaining serum and plasma samples from donors and recipients, M. Macawile and J. Lim forprocessing ofblood andtissuesamples,T.Gridleyformanuscript

preparation, C. Benedict and T. Rolain for graphics, and A. Averbach andS.Seehaferforeditorialassistance.

Supported bygrants from theNational Institutesof Health

(HL-01672 andHL-41633)andaSyntexScholars AwardtoS. G.Young.

J. K. Boyles isanEstablished Investigatorofthe American Heart

Asso-ciation. M. F. Linton is therecipientofafellowshipaward from the American HeartAssociation,CaliforniaAffiliate.

References

1.Starzl, T. E., A. J. Demetris, and D. V. Thiel. 1989. Liver transplantation.

N.Engl. J.Med.321:1014-1022 and 1092-1099.

2.Bilheimer,D.W., J.L.Goldstein,S. M.Grundy, T. E. Starzl, and M. S. Brown. 1984. Liver transplantation to provide low-density lipoprotein receptors and lowerplasma cholesterol in achildwith homozygousfamilial hypercholester-olemia.N.Engl. J.Med.311:1658-1664.

3.Starzl, T. E., B. J.Zitelli,B. W.Shaw, Jr., S.Iwatsuki,J.C. Gartner, R. D. Gordon, J. J. Malatack, I. J. Fox, A. H. Urbach, and D. H. V. Thiel. 1985. Changingconcepts: liver replacement for hereditary tyrosinemia and hepatoma. J.Pediatr. 106:604-606.

4.Lewis, J. H., F. A. Bontempo, J. A. Spero, M. V. Ragni, and T. E. Starzl. 1985.Livertransplantationin ahemophiliac. N. Engl. J.Med.312:1189-1190. 5.Merion,R.M., R. E. Delius, D.A.Campbell, Jr., and J. G. Turcotte. 1988. OrthotopiclivertransplantationtotallycorrectsfactorIXdeficiencyin hemo-philiaB.Surgery (St. Louis). 104:929-931.

6.Mahley, R. W., and S. C. Rall, Jr. 1989. TheMetabolicBasis of Inherited Disease. 6th Ed. C. R.Scriver,A.L.Beaudet, W. S. Sly, and D. Valle, editors. McGraw-HillBookCo.,NewYork. 1195-1213.

7.Boyles, J. K., R. E. Pitas, E.Wilson,R. W.Mahley, and J. M. Taylor. 1985. ApolipoproteinEassociated withastrocytic gliaofthe centralnervoussystem and with nonmyelinating glia ofthe peripheral nervous system. J. Clin. Invest. 76:1501-1513.

8.Pitas,R.E., J. K. Boyles, S.H.Lee, D. Y.Hui,andK.H.Weisgraber.1987. Lipoproteinsandtheir receptors in the centralnervoussystem:characterization of thelipoproteinsincerebrospinalfluid andidentificationofapolipoprotein B,E(LDL) receptors in the brain.J.Biol. Chem. 262:14352-14360.

9.Mahley, R. W., J.K.Boyles, andR.E.Pitas. 1989.High Density Lipopro-teins andAtherosclerosisII. N.E.Miller, editor. Elsevier SciencePublishers, Amsterdam.263-268.

10.Kane,J.P., D.A.Hardman, andH. E.Paulus. 1980.Heterogeneity of apolipoproteinB:isolation ofanewspeciesfrom humanchylomicrons.Proc. Natl.Acad. Sci. USA. 77:2465-2469.

11.Higuchi,K.,A.V.Hospattankar,S. W. Law,N.Meglin,J.Cortright,and

H.B.Brewer,Jr.1988. HumanapolipoproteinB(apoB)mRNA:identification of

twodistinctapoBmRNAs,anmRNA with theapoB-I00sequence andanapoB

mRNAcontainingaprematurein-frametranslational stopcodon,inboth liver

andintestine.Proc.Natl.Acad.Sci. USA.85:1772-1776.

12.Havel, R. J., and J. P. Kane. 1989. TheMetabolicBasisofInherited Disease.6thed.C. R. Scriver,A.L. Beaudet,W.S.Sly, and D. Valle, editors. McGraw-HillBookCo.,New York. 1129-1138.