Antigen nonspecific effect of major histocompatibility complex haplotype on autoantibody levels in systemic lupus erythematosus prone lpr mice

Antigen nonspecific effect of major

histocompatibility complex haplotype on

autoantibody levels in systemic lupus

erythematosus-prone lpr mice.

P L Cohen, … , E S Sobel, R A Eisenberg

J Clin Invest.

1993;

91(6)

:2761-2768.

https://doi.org/10.1172/JCI116517

.

MHC-linked genes strongly influence susceptibility to autoimmune diseases and also

regulate responses to exogenous antigens. To begin to understand the mechanism of this

MHC effect on disease, we have investigated MHC-congenic mouse strains that develop

spontaneous autoimmunity because of the lpr gene. C57BL6/lpr (B6/lpr) mice (H-2b) are

known to have substantial levels of autoantibodies to chromatin, single stranded DNA

(ssDNA3), and IgG of different murine subclasses (rheumatoid factor). We have crossed the

H-2d and the H-2bm12 (la mutant) haplotypes onto the B6/lpr background. Surprisingly,

levels of all the autoantibodies were markedly lower in 2d, but levels in

B6/lpr.2bm12 were no different from those in B6/lpr mice. The downregulating influence of the

H-2d allele was dominant, and there was no effect on autoantibody fine specificities. The

genetics of the H-2d effect and its diffuse influence on multiple autoantibody specificities, in

addition to the lack of effect of the bm12 mutation, which modifies the peptide-binding

groove of I-A, together raise the question of whether MHC-linked genes other than classical

(IR) genes may be responsible for MHC disease associations in this model.

Research Article

Antigen Nonspecific Effect of Major

Histocompatibility Complex

Haplotype

on

Autoantibody Levels in Systemic Lupus Erythematosus-prone

Ipr Mice

Philip L. Cohen, ElizabethCreech, DanieleNakul-Aquaronne, Richelle McDaniel, ScottAckler,

Rebecca G. Rapoport, Eric S. Sobel,and Robert A. Eisenberg

DepartmentsofMedicineand Microbiology/Immunology, UniversityofNorth Carolina School ofMedicine,

Chapel Hill, North Carolina27599-7280

Abstract

MHC-linkedgenes stronglyinfluence susceptibility to autoim-mune diseases and also regulate responses to exogenous anti-gens. Tobegin to understand the mechanism of this MHC ef-fect on disease, we have investigated MHC-congenic mouse strains that develop spontaneousautoimmunity because of the Ipr gene. C57BL6/lpr (B6/lpr) mice (H-2b) are known to have substantial levels of autoantibodies to chromatin, single stranded DNA(ssDNA3), and IgG ofdifferent murine sub-classes(rheumatoid factor).We have crossed theH-2dand the H-2bil2 (la mutant) haplotypes onto theB6/lpr background. Surprisingly, levels ofallofthe autoantibodies were markedly lower in B6/lpr.H-2d, but levels in B6/lpr.H-2b l2 were no different fromthose in B6/lprmice. Thedownregulating influ-enceoftheH-2dallele was dominant, and there was no effect on autoantibodyfine specificities.The genetics of theH-2deffect anditsdiffuseinfluenceonmultiple autoantibody specificities, in addition to the lack ofeffectof thebml2 mutation, which modifies the peptide-binding groove of I-A,together raise the question ofwhether MHC-linked genes other than classical (IR) genes may be responsible for MHCdisease associations in this model. (J. Clin. Invest. 1993. 91:2761-2768.) Key words: systemic lupus erythematosus - autoantibody- major

histocompatibilitycomplex-murine models * immune response

genes

Introduction

Susceptibilityto many autoimmunediseases isinfluencedby genesoftheMHC(reviewed in reference 1).For humanSLE, certain alleles of genesencodedby theHLADR, DP,DQ,and C4lociresult inenhancedrisk ofdiseaseand may beassociated with particular autoantibody specificities (1-4). In murine SLE, data also indicate an association between MHC alleles and disease. In the(NZB X NZW) Fl model, H-2 isoneof severalgenetic locithatcontributetodiseaseseverity(5). For NZBmice(H-2d), the substitution ofthe H-2bml2,but not the

H-2b,

haplotype results in production ofgreater levels of anti-body to double- and single-stranded DNA (dsDNA and

E.Creech and D. Nakul-Aquaronne made equalcontributionstothis work.

Addresscorrespondence andreprint requeststo Dr.PhilipL. Co-hen, CB#7280, 932 FLOB, University of North Carolina School of Medicine, Chapel Hill, NC 27599-7280.

Receivedfor publication 14 September 1992 and in revisedform 22 January 1993.

ssDNA, respectively) ',and to a shortenedlifespan due to glo-merulonephritis (6). In addition, the autoimmune disease of the BXSB mouse is less severe in animalsexpressing H-2d in-stead of the H-2b allelenormally present in this strain (7).

Theautosomalrecessivelprgene,recently demonstratedto be anonexpressedmutant form of the Fasapoptosis-inducing cellsurfacereceptor(8),causeslymphadenopathyand autoan-tibody formation. The nature and severity ofdiseaseare depen-dent onbackgroundgenes(9),andlpr accelerates any underly-ing autoimmune proclivity. Forexample,

MRL/Mp-lpr/lpr

mice develop massive lymphadenopathy;antibodies to multi-pleautoantigens including dsDNA, ssDNA, chromatin, IgG, Su, ribosomalPprotein,andSm,aswellas severe renal disease. In contrast, C57BL/6-lpr/lpr (B6/lpr) mice develop more modest lymphadenopathy and antibodies to chromatin, ssDNA, andIgG (9),but fail todevelopantibodies to Sm or dsDNA, and donothavevasculitisorrenal disease. Intercross analysis has shown that multiple backgroundgenesare respon-sible for the differencesin thelpr-induced autoimmunity,and ithas thusfarnotbeenpossibletodemonstrateanMHC effect on renal diseaseorautoantibody production (10). We have, therefore, soughtasystem free of the influence of othergenetic differencestostudythe effects of MHC genesonsystemic auto-immunity. We thustook advantage of the existence ofH-2 congenic C57BL/6strainstoderivenewlprstrainsthat differ onlyatMHC loci. The resultsindicated that substitution ofthe H-2d allele forH-2bresultedindiminishedautoimmunity,but that H-2bml2 didnotinfluenceautoantibody production.

Sur-prisingly,

thereductionin

autoimmunity

observed in

B6/lpr-H-2d reflecteda

generalized

loweringofautoantibody produc-tion ratherthananeffectonautoantibodyfinespecificity.

Methods

Mice. C57BL/6 (B6) andC57BL/6-lpr/lpr(B6/lpr) mice

(H-2b);

and HWl9(H-2d-congenic B6) micewereoriginallyobtained from the Jackson Laboratory (Bar Harbor, ME)and maintained in our

breedingcolonyattheUniversityof NorthCarolinaatChapelHill. B6.C-H-2bml2(bm12)micewereoriginallyobtained from Dr.Roger Melvold, NorthwesternUniversity (Evanston, IL)and have been bred inourcolony.

To produce

H-2d-congenic

B6/lprmice, wecrossedB6/lprand

HW19.Fl hybridmicewereintercrossed,andtheF2generationwas screened for

lpr

homozygosity byassessingantichromatin autoantibod-ies and rheumatoidfactors byELISA. H-2phenotypewasdetermined by immunofluorescenceusingAF6-88.5.3

(anti-H-2Kb)

and

SFl-1.1.I

(anti-H-2Kd),bothfromAmericanTypeCultureCollection,

(Rock-1. Abbreviations used in this paper: BBS, borate-buffered saline; dsDNA, double-stranded DNA; EDF, equivalent dilution factor; HGG,human gammaglobulin; IR,immune response;KLH,keyhole limpet hemocyanin;ssDNA,single-strandedDNA.

J.Clin.Invest.

©TheAmericanSocietyfor ClinicalInvestigation, Inc. 0021-9738/93/06/2761/08 $2.00

ville, MD). Toobtain H-2d homozygous foundermice, it was

neces-sary tointercrossone moregeneration. TheB6/lpr.H-2dmicewere

also crossedtoB6/lpr, and the F2generationwastestedfor segregation of H-2 phenotype andautoantibody levels (see text).

Toproduce the B6/lpr.H-2bmI2strain, B6 /lprwasmatedtobm12, and F2mice screened inafashion analogoustothatdescribed above. Typing for I-Ab was accomplished by immunofluorescence using D3137.5(anti-I-Ab); asthereisnoI-Abm12-specificreagentgenerally available, bm 12 micewereidentified by their lack of lab. Additional intercrosses were necessarytogenerate B6/lpr.H-2bmI2 homozygous founder mice. TheidentityofB6/lpr.IH-2lbml2micewasalso confirmed by their characteristic responsetoimmunization with insulin (11).

IgGantichromatin. Chromatin Agwaspurified from chicken eryth-rocytenuclei (12).Polyvinyl chloride microtiter plates (Dynatech Lab-oratories, Inc., Chantilly, VA)werecoated with 100 pJofa3-Mg/ml chromatinsolution in borate-buffered saline (BBS), pH 8.4, for5 h at

4°C. Afterwashingwith BBS andincubatingwith 200,l ofcoating buffer (BBS, 0.5% BSA, 0.4% Tween 80) for 1h at4°C,serumsamples (diluted 1/500 incoatingbuffer)wereadded. Afterovernight incuba-tionat4°C, theplateswerewashedfive times withBBS,anda1/2,000 dilution ofbiotinylated affinity-purified goat anti-mouse IgG (pFc' specific)wasadded for 1 hat4°C.Afterwashing,theplateswere incu-bated withavidin-alkaline phosphatase (ZymedLabs,Inc., South San Francisco, CA) for 1 h at4°C. After furtherwashing, phosphatase substrate(SigmaChemicalCo.,St. Louis,MO)wasadded in0.01 M

diethanolamine, pH9.8, toeach well. Theplateswerereadwithan

automatedmicro-ELISA readerat 1,3, and 24 h after substrate addi-tion. A standard curve wasconstructedusingahightiterMRL/lpr

mouseserum, andEDF(equivalent dilutionfactor)valueswere calcu-latedaccordingtothefollowingformula:EDF=(dilutionof standard referenceserathatgives theequivalentODof thetestserum) x 106. Thestandarddilution curves were usually linear andnotsteep in the OD range tested.

Anti-ssDNA.Calf thymusDNA(Sigma Diagnostics Co, St. Louis, MO)wasdenaturedat100°Cfor 10 min, and 100ML of a 10-,ug/ml solutionof ssDNAwasaddedtoeach well. Theantibodieswere de-tectedwithabiotinylatedaffinity-purifiedratanti-mouseK monoclo-nal antibody. A high titerMRL/lprmouse serum wasused for the standardcurve.

Anti-dsDNA. Theindirect fluorescent antibody procedure for the detection of anti-dsDNA antibodies was performed with substrate slides ofCrithidia luciliae(KallestadDiagnostics,Inc.,Chaska,MN). Theserumsamples diluted 1 /10wereincubated 30 min inamoist box. Afterwashingwith PBS, anti-dsDNA antibodies boundtotheCrithidia kinetoplast DNA were detected with a rabbit anti-mouse IgG linked to

afluorochrome (FITC). 30 min later, the slides were readusinga

fluorescence microscopeat a magnification of 400. An anti-dsDNA positive MRL/lprhybridomaandananti-dsDNA negative tissue

cul-turesupernatantwereusedascontrols.

Rheumatoidfactorsand total IgM. Rheumatoid factor and total

serumIgMweretested in assays similartothosedescribed above. The plateswerecoatedwith monoclonal reagents rendered free of IgM by previous absorption. Myeloma proteins used as antigen for detection of isotype specific rheumatoid factors were as listed in( 13)except for IgGla (HB57) and the IgG2aa (P1.17). Antibodies to these plate-boundIgGsweredetectedwithbiotinylated affinity-purified Bet-2 rat anti-mouse IgMF(ab')2,followed by avidin-alkaline phosphatase and substrate.

Immunizations. Mice were given 50,gofhuman gamma globulin (HGG)purified from Cohn Fraction II of human serum or keyhole limpet hemocyanin (KLH) (Sigma Chemical Co.) subcutaneously in CFA.Theywereboosted with 50 ,tg of antigen in incomplete Freund's adjuvant10 d later, and were tail-bled 1wklater (17 d after immuniza-tion).

HGGand KLH ELISAs. IgG antibodies to HGG and KLH were quantitatedbycoatingmicrotiterwellswith 3

Mg/ml

ofprotein antigen in BBS. The plates were then processed as above. Serum was added at a

1/200dilution. Bound antibodywasdetected using biotinylated goat anti-mouseIgG.

Immunoblotting. Chickenerythrocytechromatinwasdissolvedat

0.55mg/ml in Laemmli buffer containing bromphenol blue, sodium phosphate,glycerol, SDS, water, and DTT. After heating for 3 minat

1000C, the diluted sample and S Ml of prestained molecularmass pro-tein standards, 3,000-43,000 D, (Gibco BRL, Gaithersburg, MD)were

electrophoresedona 10-20% gradient polyacrylamide gel with SDS. Theelectrophoresiswasperformedat20mA for 2.5 h. The separated polypeptides were transferredtonitrocellulose paper by electroblotting intransfer buffer (25 mMTrishydroxide, 192 mM glycine, 20% metha-nol) at100Vfor 45 min on ice. Nitrocellulose paper was thencutinto strips andpressed overnight. Thenextday, the strips were incubated with Tris-buffered saline 5% milkfor 1 h. Afterwashing, theserum

samples,diluted 3 Ml/ml in TBS,wereadded. After overnight incuba-tion, the antibodies were detected with biotinylated affinity-purified rat anti-mouse Kdiluted 1 / 1,000, followed by avidin-alkaline phospha-tase. Antibody binding was visualized using nitroblue tetrazolium-bromo-chloro-indolyl phosphate, Sigma Chemical Co.

Measurementofproteinuria.Protein wasestimated in spoturines from F2 mice (Albustix; Miles Inc., Elkhart, IN). Protein

concentra-tion in the urinewasclassified accordingtofive categories (0-4+).

Results

Antichromatin in MHCcongeniclprmice. Antichromatin anti-bodies were quantitated by ELISA in sera from B6/lprand B6/lpr.H-2d miceat5moof age.InFig. 1a,it can be seen that antichromatin levelsinthe H-2d micewere lower than those measured in samples from

H-2'

individuals (P < 0.001 for comparison of pooled male and female samplesof each strain by Student's t test). B6/lpr female mice had significantly higherlevels of antichromatin than males (P=0.0001), but there was no sex difference in antichromatin levels for B6/ lpr.H-2d (P =0.8).

Toensurethatthe differences observed between the B6/lpr andB6/lpr.H-2d mice segregated with H-2, we mated B6/lpr andB6/lpr.H-2d mice and intercrossed theFloffspring.In the F2generation, there were 28 H-2 b/b mice, 43 H-2b/d mice, and 24 H-2d/d mice. Asshown in Fig. 1 b, sera from the bb mice contained significantly more antichromatin antibodies than sera fromddindividuals (P<0.001 ). Antichromatin levels in bdmicewere not significantly different fromthose of dd mice

(P

=

_0.71).

Fig. 2 shows a comparison of antichromatin determina-tions on the sera of B6/lpr.H-2bml2 and B6/lpr mice. In con-trast to what wasobserved forB6/lpr.H-2d,there wasno signifi-cantdifference in antichromatin between B6/lpr.H-2bml2 and B6/lpr.

Females Males Females Males

B6/lpr

B6/lpr.H-2

d

b

dd

Females Males

B6/lpr

Females Males

bml2/lpr

Figure1.Antichromatinantibody distribution in B6/ lpr and inB6/lpr-H-2dsera.(a) Eachpoint

repre-sentstheELISAopticaldensity value ofanindividual sample takenat5moof age. The numbers of B6 /lpr andB6/lpr-H-2dmicestudiedwere:49(26males and 23females) and 50 (25 males and 25females),

re-spectively.(b)AntichromatinEDFlevels foracohort ofF2(bb),F2(bd)andF2(dd)mice. The results werestandardized againstahightiterMRL/lpr mouse, sothatantichromatinEDFtiters could be calculated..,females;u,males.

Figure 2. Antichromatinantibodydistribution inB6/ lprand

B6/lpr.H-2bmI2

mice.Eachpointrepresents

anindividual mouse, 4-6moold. 42sex-and age-matched ( 16 males and 26females)B6/lprandB6/

lpr.H-2bml2

micewerestudied. Nosignificant

differ-encesbetweenB6/lprand

B6/lpr.H-2bml2

werenoted. 2.5

2

1.5

0.5

0

-200,000

150,000 _

IL

U.

c100,000

50,000

0

1.6

1.4

1.2

0

o

0

8

o

O

80

0

0~~~~

8

0

o

8

8

d_{0.8 F}

0.6

1-0.4

[-0.2

F

a

u

en 4

Got

mn

nV

H-i1-

H-S5-H-3

H-4~~~~~~~~~~~~~~~~~~~~~~~~~~4

r _ _ _ ~~~~~~~~~q -_

H-3

H.4

st _lotO Ct _lOt 0 _Ic- 0 0

%nnIn v) In

.W

bA.Xe~

r-c'n

.n

..

N CD 0' e.

rN In Wn

en en en In V)

In In

.)VI

WI.

..

Figure3. Western blotsof antihistone antibodies.

Re-N sultsarefor individualserafrom5-mo-old mice

(di-S luted1/333).Histones H2B and H2Arunclosely to-gether, justbelowH3.(a)The extreme left lane(lpr 8557cn2)showsserumfroma5-mo-oldMRL/lpr mouse;theremainingseraarefromB6/lpr

individ-uals. (b) B6/lpr.H-2dsera.(c)B6/lpr.H-2bmI2sera.

Amido blackstainingtoidentifytotalproteinshas

beenincluded for the blots inFig.3C.

B6/lprsera,withnoselective reductioninreactivitytoasingle

histoneorgroupof histones. Thesesera wereselectedforhigh antichromatin levels, thus the immunoblots show no

differ-enceinstainingintensitybetweenstrains. Because therewasno

differenceinthe distribution ofantibodydirected toward the histone components of chromatininB6/lpr.H-2dsera,the de-creased overallreactivityof thesesera wascausedbya general-ized reduction in autoantibodies to several components of chromatin.Fig. 3 also shows immunoblot results from

repre-sentativeB6/lpr.H-2bmI2sera.Theseserashowedpatterns simi-lartowhatwasseenwith B6/ lprsera.Therefore,theequivalent

ELISAresults didnotmaskadifferenceinantichromatin

speci-ficity.

Anti-DNA in MHC-congenic lprmice. Aswasseenfor

an-tichromatin, anti-DNA titers were significantly lower in Ipr miceexpressing thedhaplotype.Thiswasascertained by

com-paringanti-ssDNA levels in acohort ofparental B6/lprand B6/lpr.H-2d individuals (not shown). The data were

con-firmed inananalysisof F2 mice(Fig.4a).Those mice which inherited the dd and bdhaplotypeshadsignificantlylower

lev-els of anti-DNA than bb individuals(P<0.05).In contrast to

these results, B6/lpr.H-2bml2 mice showed anti-DNA levels thatweresimilarto those ofB6/lpr.As seeninFig. 4 b, al-thoughtherewas amodestdecrease inthe female B6/lpr.H-2bm12mice,this didnotachievestatistical significance.We also measured anti-dsDNA in B6/lpr.H-2d and B6/lpr.H-2bml2. None of these strains showed any positivity in the Crithidia luciliae immunofluorescenceassay.

Rheumatoidfactor levelsincongeniclprmice.B6/lprmice areknowntoproducesubstantialamountsof rheumatoid

fac-tor( 14).We determined the effect of MHChaplotype substitu-tions on the amount and specificity of this autoantibody. Rheumatoid factor inparental B6/lprmicewaspresent in sig-nificantly higherlevels than inB6/lpr.H-2danimals(datanot

shown). As was done for the other autoantibody

measure-ments,resultswereconfirmed in F2mice.Fig. 5ashows that IgM anti-IgG1blevelsweremarkedlylower in bd and dd mice. Thesamewastrueforanti-IgG2bb (Fig.5b). Levelsinbd mice

werenotsignificantly differentfromthose measuredindd ani-mals. Incontrasttothesedata,therewasnodifference between

c C 2u

0

H.1 -+

H-5

--H-3 -,

-40,000 1

a

30,000

i20,000

10,000 _

0

10,000

U.

a

dd

rW WDmlpr

Figure4.(a) Anti-ssDNAantibodylevels(EDF)inthesamecohort of F2mice used forantichromatinmeasurementsshown inFig. 1. H-2bmice showedagreaterlevel of anti-ssDNA antibody compared

toH-2'(P<0.001)and H-2b/d mice(P<0.001). (b)Anti-ssDNA EDFtitersin B6/lpr and B6/lpr.H-2"''2 mice.The differences ob-servedwerenotsignificant (P>0.05). .,females; *,males.

B6/lpr and B6/lpr.H-2bml2 rheumatoid factor titers against the IgGIb and IgG2b isotypes (Fig. 6). Inanefforttodiscern dif-ferencesinrheumatoid factorfinespecificity betweenB6/lpr andB6/lpr.H-2bml2,wemeasuredIgMantibody against botha

and ballotype myeloma proteins representing all ofthe murine subclasses(Fig. 6). Therewere nodifferencesbetween thetwo

strains.

Responsiveness of congenic lpr miceto immunization. It

waspossible that the lower autoantibody levels in B6/lpr.H-2d

mice reflectageneralized decreased responsivenesstoantigenic stimulation. Thisquestion wasapproached by assessing

anti-body responsesto HGG andKLH. Asseen inTable I, there was nodifference between B6/lpr and B6/lpr.H-2d mice in

antibody levels after immunization.

Otherstudies inMHC-congenic lprstrains. We compared the total spleen and lymph node weights of 7-mo-old MHC-congenic lprmice(Table II). Spleen and lymph node weights

ofF2 bb homozygous mice were significantly greater than thoseobservedfor ddmice(P=0.006 and0.02 forspleenand

lymph nodes, respectively). For bbvsbdmice, differences

ap-proached statisticalsignificance for spleen (P=0.07) but_not

for lymph node (P = 0.21). When bd mice werecompared

with ddhomozygotes, onthe other hand, lymph node weights

12,000

F

10,000

K 8,000 _

i 6,000

4,000 _

2,000 _

0

8,000 r

6,000 F

IL

al c4,000

2,000

0

bb bd dd

Figure 5. Isotype-specific rheumatoid factors directed against IgGl b

(a)andIgG2bb (b)inF2(bb), F2(bd), andF2(dd) mice. H-2b

ho-mozygousmiceweresignificantly different fromB6/lpr-H-2dand

fromF2(bd) for both specificities(P<0.001). .,females; .,males.

weresignificantlygreaterinthe bd mice(P=0.04),yetspleen

weightswerenotsignificantly different (P=0. 14).

We also tested for proteinuria in female mice of these strainsusing the semiquantitative Albustix methodandfound

16,000

14,000 12,000 10,000 8,000

6,000

4,000

2,000

0

IgG1a 1gGlb IgG2aa IgG2ab IgG2ba IgG2bb IgG3 Toal1gM

Figure6. IgGsubclass andallotype-specificrheumatoidfactors and

totalIgMinB6/lprvs

B6/lpr.H-2"1'2

mice. For each rheumatoid

factorspecificity,B6/lprandB6/lpr.H-2bmI2micewerenotdifferent (P> 0.1). .,B6/lpr; *, bml2/lpr.

b

TableI.ResponsesofB6/lprandB6/lpr.H-2dMice to Immunization

No. of Prebleed

Strain mice Antigen OD(SD) 4-wkOD

B6/lpr 7 HGG 0.024(0.022) 0.409(0.102) B6/lpr.H-2d 5 HGG 0.076(0.042) 0.464(0.068) B6/lpr 7 KLH 0.046(0.009) 0.160(0.047) B6/lpr.H-2d 6 KLH 0.039(0.005) 0.146(0.020)

Threemonth old miceofbothsexes wereimmunized subcutaneously with 50ggofantigenin CFA,followed10 d laterbyabooster im-munizationof50,g of antigeninincompleteFreund'sadjuvant. Mice were bled 17 d after initialimmunization,andseraweretested

at adilution of1/200inELISA.

nosignificant urinary protein in any of the mice. Finally, we measuredanti-Sm autoantibodies by ELISA, and foundno pos-itive sera.

Discussion

We have created two congenic B6/lpr mouse strains: B6/ lpr.H-2d, which expressesthe d alleleof theH-2 complex in placeof the b allele expressed by B6/lpr; and B6/lpr.H-2bml2, which is identicalto B6 /lpr, except thatitexpressesthe bm12 mutation atI-A13. B6/lpr.H-2d mice hadsubstantially lower levels of spontaneous autoantibodies to chromatin, single stranded DNA, and mouseIgG. In contrast, B6/lpr.H-2bml2 micewereidenticaltoB6/lpr mice in theseassays.

The present data areconsistent with earlierreportsinthe NZB/NZW and NZB/SWR models implicating an MHC gene effect on

systemic autoimmunity

(5, 15). The results

differ,

however, from

findings

ofan

exacerbating

effect of the bm 12 I-Aphenotypeon NZBautoimmune disease (6).It has been suggested that the acceleration of autoimmunity imposed by the bm 12 haplotypemaybe caused by an amino acid substi-tutionatposition72of

_I-A#3,

a position also implicated as the possible locus ofthe NZW contribution to the autoimmunity of NZB/NZW(6). The present results indicate that the bm 12 effecton NZBisnotgeneralizable to all strains. Our data also differ inthat the NZB

(H-2d)

andNZB.H-2bshowed similar levelsof autoantibodies. The differences between our findings and theeffects ofdifferent H-2 haplotypes on NZB are likely to

TableII. Lymph Node andSpleen Weights inMHCCongenic

lpr

Mice

H-2 Number Lymph Node Spleen

haplotype of Mice Weight (SD) weight (SD)

g

bb 27 1.310(1.191) 0.315 (0.196)

bd 43 1.022 (0.698) 0.221 (0.208)

dd 23 0.669 (0.554) 0.154 (0.077)

(B6/lpr.H-2d x B6/lpr) F2 mice were typed by immunofluorescence staining. At 7moof age, they were sacrificed and the weight of the spleenandthe total weight of the cervical, axillary, paraaortic, in-guinal,andmesenteric lymphnodes weredetermined.

becaused by the many genetic differences between NZB and the B6/lpr model, for instance at the level of T cell receptor repertoire.

H-2 haplotype has recentlybeen shown to affect the sys-temic autoimmune disease of BXSB mice (7). The substitu-tion of H-2d for the H-2" haplotype led to decreased autoanti-body production, diminished glomerulonephritis, and en-hanced survival. These results areanalogous to ours, and may reflect the similarities in background genes between B6 and the related BXSB strain.

Itis possible that the observeddifferences inautoantibody production bylpr strains with differentMHChaplotypesare caused by the allelic differencesin class II MHC genefunction, analogous to the classic experiments demonstrating Ta control ofthe response to exogenous antigens. This interpretation would fit well intocurrent paradigms ofimmunology. Like exogenousantigens,autoantigens couldberegardedas polypep-tides thatundergo processing and ultimately bind to surface class II MHCantigensforpresentationto Tcells. Infact,in the immune response(IR)gene literature, there is precedentfor immuneresponses tomodified selfantigens suchasalbuminor IgAundersuch MHC-linked gene control( 16).Allelic Ta poly-morphisms might thus reflect the different efficiencies with whichcertain Ia antigens bindpeptidesprocessedfrom particu-larautoantigens. The molecular basis

of

this phenomenon is widely believed to be the three-dimensional structure ofan antigen-bindinggroovedemonstratedonthe exteriorof MHC molecules.Itis alsopossiblethat theMHC-determined differ-encesinautoantibodyproduction reflect differencesin the T cell repertoire that develops in the presence of the MHC b haplotype versusthe d haplotype,orthat theMHCdifferences result in altered immuneresponse because ofsuppressor cell activity. Bothofthese lattermechanismshave beenproposed as explanations of classic MHC-linked IR gene phenomena ( 17). The greaterlymph nodeandspleen size of the b haplo-typemice, comparedtothe dhaplotype

individuals, might

re-flecta moreintenseTcellimmuneresponse in theformer, with consequentgeneration of increased numbers of double nega-tive cells,ormightbe causedby interstrain differences inTcell repertoire.

Theobservations concerningthe d haplotypemight alterna-tivelybe explained by the expression in thishaplotypeofI-E molecules, which are not expressed in B6/lpr or

B6/lpr.H-2bm12

becauseofthenonfunctionalEa geneoftheb haplotype. Such a mechanism would imply a suppressive effect ofI-E expression on autoimmunity, similar tothe effect ofI-E on autoimmune diabetes in the non-obese diabetic model (18, 19),and onmercuricchloride-induced antinucleolar antibod-ies(20). Apotentialfinding of importanceto thesestudies is the recent observation that I-E-derived peptides may be de-tected bound topurifiedI-Amolecules (21 ). Miceexpressing I-Emay thus generate peptidesthatbindto theirsurfaceI-A and undermine its capacity tobind otherpeptides, perhaps reducing immune responses. Any suppressive effect ofI-E, while ofpotential importance in the preceding models, clearly would not besufficientto alleviate systemic autoimmune dis-easein all cases, as severeautoimmunity is seenin theI-E+ (NZB x NZW) and (NZB X SWR)

Fl

mice, as well as in MRL/lpr.

repertoire or providing suppressive signals. In the BXSB model,itwasof interestthat theexpression ofI-EinH-2d mice led, asexpected,to thedeletion ofTcellsexpressing

V#5

and

V#I

1 (7). Nevertheless, it appeared unlikely that the loss of these T cells was responsible for the ameliorating effect of H-2d, because H-2bidmice developed a syndrome resembling that ofH-2b mice, despite the reduction in VI35-and Vf I1-bearingTcells. These results alsoargued againstageneralized protective effect of I-E expression in that system. We are currentlyexploring the influence ofI-Einourmodel by exam-ining other MHC-congenic B6/lpr micethat express I-E, and by assessing autoimmunity in

_I-E.

transgenic

lpr mice.

Our datashow thatfemale individuals, particularly B6/lpr mice, produce largeramountsof autoantibodies than males. Thisobservation is consistent with heightened systemic autoim-munity observed in females. It is possible that the increased levelsof autoantibodies inthefemalesarerelated toincreased numbersof autoreactiveTcells,particularly in the H-2" strain. Infemales,estrogens mayfavor extrathymic differentiation in theliver ofTcells with autoreactivecharacteristics, and thus may lead toenhanced

production

of autoantibodies (22).

Another interestingaspectofourdata isthelack of influ-enceof thebm12 mutation onautoantibody levels. Although only threeresidues in I-AI3 differ between

H-2b"l2

and H-2", thesetwo alleles aregenerally associated with major differences in immune

responsiveness

to

foreign

antigens (11)andin sus-ceptibility to

experimental myasthenia gravis

(23), probably because the alteredaminoacid residues lie within the

peptide

binding groove of the I-A(3 chain. In addition, as discussed previously, NZB.Iml2 mice have considerably

higher

titers of autoantibodies than NZB.H-2" mice(6). It is thus

surprising

that

B6/lpr.H-2lIml2

mice and B6/lpr mice havenodetectable differencesin

autoantibody

levelsorinthespecificities recog-nized by antichromatin antibodiesorby

rheumatoid

factor.

Yetanotherstrikingaspect

of

thepresentresultsis the re-cessive nature of the

high responder phenotype.

In

general,

MHC-linked IR genes are

dominant,

with Fl animals

mani-festing

responses close tohomozygous responders,or atleast intermediatebetweenresponders and nonresponders.Wehave found that the responsesof H-2b/d

lpr

micewere not

signifi-cantly different from those of low

responder

B6/lpr.H-2d

ani-mals. This suggeststhat the influence of H-2-linkedgenes on

autoantibody production

may not be caused

by

classical IR geneeffects mediated

by

allele-specific binding

toclassII mole-culesof

antigenic peptides.

Our results differfrom the

findings

intheBXSBmodel,where

Fl

H-2b/dmice

expressed

the

high

autoantibody

levels

characteristic

of their H-2bparents

(7).

The general nature ofthe H-2d effect is also difficult to understand as a manifestation of classicIR geneeffects. The lowresponderdhaplotypeappearstohave diminished

autoan-tibody

responsesto anarray of

autoantigens,

andnot

merely

a

single autoantigen: chromatin

(consisting

of

multiple

histones and DNA);

single-stranded

DNA;andrheumatoid factor di-rectedagainst

IgGl

andagainst IgG2b.In contrast,formostIR genecontrolledresponses, the target

antigens

areeither

simple

synthetic

polypeptides

withlimitednumbersof

epitopes,

or are modified selfantigens.IR genephenomenaareusually discrete and

reciprocal;

forinstance, H-2b mice respondtothe

synthetic

amino acid copolymer (T,G)-A-Lbut not totheclosely re-lated(H,G)-A-L; the reverse istrue for

H-2"

mice

(24).

IR gene

polymorphisms

wouldnotbeexpectedtoresultin

appar-entlyglobal changes inspontaneousautoantibodyproduction wehave observed in our H-2 congenic

lpr

mice, but rather in changes inthe fine specificity of certain autoantibodies, with perhaps increases insomespecificities,alongwithdecreasesin others. We thussuggest thatother genesinthe MHC complex mayaffectthe

lpr

syndrome.Known genes include class I anti-gens,whicharesurface glycoproteinsinvolved in peptide anti-genbinding and interactionwithCD8+ T cells; tumor necrosis factors a and

_f3,

which have been reported to have a potent influence on autoimmune and inflammatory phenomena (25); C2andC4genes, alsolinkedtodisease susceptibility(2); theHSP 70 heat shockproteingene(26);aseries ofsinglecopy genesof uncertain function (27);and an arrayof highly

inter-esting

recently discovered class II-linkedgenesthatare respon-siblefor the intracellulartransportofpeptides and for the degra-dation ofprotein antigens into small peptides for bindingto class Iand class IImolecules(28).Itis

possible

that the sup-pressive effect on autoantibodies resulting from the substitu-tionof the d for the b MHC allele might involveany one

of

these non-Ia genes. For instance, autoantibody formation could be influenced by general differences in autoantigen pep-tide

binding

or

binding

ofan

autoantigen

peptide,

which plays abroad role inT-Bcollaboration; classI

mediated suppressive

effects;

variations inpeptidetransporter genes oringenes

con-trolling protein

degradation (29, 30);oreffects

ofyet-undiscov-eredgenesthatmay also beencoded in the MHC.Recently, it has been shown thatpeptidetransporter gene

polymorphisms

mayinfluence the peptidesbound byMHCgenes(31 ). Such MHC-linked

polymorphisms

mayprove tohave clinical rele-vance by exerting a broad effect on antigen presentation in autoimmunedisease.

Acknowledgments

We thank Robert L. Cheek for assistance with immunoblot assays. This work was supported by U.S. Public Health Service grants AR34156, AR40620, AR26574, AR33887, AR30701, and T32 A107273. E. Sobelis a Fellow of the Arthritis Foundation, and E. CreechisaHolderness Fellow. D.Nakul-Aquaronne received support from thePhilippeFoundation.

References

1.Goldstein,R., and F. C. Arnett. 1987. Thegeneticsof rheumatic disease in

man.Rheum. Dis.Clin. NorthAm. 13:487-510.

2.Olsen,M.L.,R.Goldstein,F.C. Arnett,M.Duvic,M.Pollack, andJ.D. Reveille. 1989. C4Agenedeletion and HLA associations in black Americans with

systemic lupuserythematosus. Immunogenetics. 30:27-33.

3.Fujisaku,A., M. B.Frank,B.Neas, M.Reichlin,andJ.B.Harley. 1990.

HLA-DQgenecomplementationand otherhistocompatibility relationships in

manwith theanti-Ro/SSAautoantibodyresponseofsystemiclupus

erythemato-sus.J.Clin.Invest.86:606-611.

4.Reveille,J.D., K. L. Anderson,R.E.Schroehenloher, R. T. Acton, and B.0.Barger. 1991.Restrictionfragmentlengthpolymorphism analysisof HLA-DR,DQ, DP,and C4 alleles inCaucasianswithsystemic lupus erythematosus.J. Rheumatol. 18:14-18.

5.Knight,J.G.,and D. D.Adams.1978. Threegenesforlupus nephritisin NZBx NZW mice. J.Exp.Med. 147:1653-1660.

6.Chiang,B. L.,E.Bearer,A.Ansari,K.Dorshkind,andM. E.Gershwin. 1990. The bm 12 mutation and autoantibodiestodsDNA inNZB.H-2bm'2mice. J. Immunol. 145:94-101.

7.Merino, R.,L.Fossati,M.Lacour,R.Lemoine,M.Higaki,and S. Izui. 1992.H-2-linked control of theYaagene-inducedacceleration oflupus-like

au-toimmunedisease in BXSB mice. Eur. J. Immunol. 22:295-299.

9.Cohen,P.L., and R.A.Eisenberg. 1991. Lpr andgid:single gene models of systemicautoimmunity and lymphoproliferative disease. Annu. Rev. Immunol. 9:243-269.

10.Eisenberg, R. A., Craven, S. Y.,Fisher, C. L., Morris, S. C., Rapoport, R., Pisetsky, D. S., and P. L. Cohen. 1989. Thegenetics of autoantibody production inMRL/lpr lupus mice. Clin. Exp. Rheumatol. 7(Suppl. 3):535-540.

11. Hansen, T. H., W. D.Walsh, R. J. Rubocki, and J. A. Kapp. 1986. Responsesof B6.C-H-2bm 12 to heterologousinsulins show no correlation with theputative geneconversionbutdefineIab"'2asfunctionally unique. J. Mol.

Cell.Immunol. 2:359-368.

12.Fisher, C. L., R. A.Eisenberg, and P. L. Cohen. 1988. Quantitation and IgG subclass distribution ofantichromatin autoantibodies in SLE mice. Clin. Immunol. Immunopathol. 46:205-213.

13.Morris, S. C., R. L. Cheek, P. L. Cohen, and R. A. Eisenberg. 1990. Autoantibodies in chronic graft versus hosts result from cognate T-B interactions. J.Exp.Med. 171:503-517.

14.Warren, R. W., S. A. Caster, J. B. Roths, E. D. Murphy, and D.S. Pisetsky. 1984. Theinfluenceof the lpr gene on B cellactivation: differential antibody expression in lpr congenic mouse strains.Clin.Immunol.Immunopathol. 31:65-77.

15. Ghatak, S., K.Sainis, F. L. Owen, and S. K. Datta. 1987.

T-cell-receptorj8-and I-Aj3-chaingenes of normalSWR mice are linked with the developmentof lupusnephritis in NZB x SWR crosses. Proc.Nadl. Acad.Sci. USA. 84:6850-6853.

16.Cohen, P. L., J. J. Ellner, and E. M. Shevach. 1977. Induction of guinea pig antibody responses in vitro. J.Immunol. 118:2034-2039.

17. Paul,W.E.1984. Immune response genes. In Fundamental Immunology. W. E. Paul,editor. Raven Press, Ltd. New York. 439-455.

18.Nishimoto,H., H.Kikutani, K.Yamamura,and T. Kishimoto. 1987.

Prevention ofautoimmuneinsulitis by expressionof I-E molecules in NOD mice. Nature(Lond.).328:432-434.

19.Reich,E.P., R. S.Sherwin,0.Kanagawa, andC. A. Janeway, Jr. 1989. An

explanationfor theprotective effectof the MHC classIII-Emolecule in murine

diabetes.Nature(Lond.). 341:326-328.

20.Mirtcheva, J., C. Pfeiffer, J.A.DeBruijn,F.Jacquesmart, and E. Gleich-mann. 1989.Immunologicalalterations inducible by mercury compounds. III.

H-2A acts asan immune response and H-2E as an immune "suppression" locus forHgCl2-induced antinucleolar autoantibodies. Eur. J. Immunol. 19:2257-2261.

21.Rudensky, A., P. Preston-Hurlburt, B. K. Al-Ramadi, J. Rothbard, and C.A.Janeway, Jr. 1992. Truncation variants ofpeptides isolated from MHC class II molecules suggest sequence motifs. Nature(Lond.).359:429-431.

22.Okuyama, R., T. Abo, S. Shuhji, T. Ohteki, K. Sugiura, A. Kusumi, and K.Kumagai. 1992. Estrogen administration activates extrathymic T cell

differen-tiationin theliver.J.Exp.Med.175:661-669.

23.Bellone, M., N.Ostlie, S. J. Lei, X. D. Wu, and B. M. Conti Tronconi. 1991. TheI-A""I2mutation,which confersresistance toexperimental

myasthe-niagravis,drasticallyaffects theepitope repertoireof murine CD4+ cells sensi-tized tonicotinicacetylcholinereceptor. J. Immunol. 147:1484-1491.

24.McDevitt,H. O.,and B. Benacerraf. 1969. Genetic controlof specific

immune responses.Adv. Immunol. 11:31-74.

25.Inoko,H., and J. Trowsdale. 1987.Linkageof TNF genestothe HLA-B locus.NucleicAcids Res. 15:8957-8962.

26.Milner, C. M., and R. D. Campbell. 1990. Structure and expressionofthe three MHC-linkedHSP70 genes.Immunogenetics.32:242-251.

27.Cho,S.,M.Attaya, M. G.Brown, and J. J. Monaco. 1991.Aclusterof transcribed sequences between the Pb and Ob genes of the murinemajor histo-compatibility complex. Proc.Nall. Acad.Sci. USA. 88:5197-5201.

28.Glynne, R., S. H.Powiss,S. Beck,A.Kelly,L.A.Kerr, and J. Trowsdale. 1991. Aproteasome-relatedgenebetween thetwoABC transporter loci in the classIIregion ofthe humanMHC. Nature(Lond.).353:357-360.

29. Brown, M.G., J.Driscoll, and J. J. Monaco. 1991. Structural and

serologi-calsimilarityof MHC-linked LMP and proteasome(multicatalytic proteinase)

complexes Nature(Lond.). 353:355-357.

30.Martinez,C. K., and J. J. Monaco. 1991.Homologyofproteasome sub-unitsto amajor histocompatibility complex-linkedLMP gene. Nature(Lond.).

353:664-667.

31.Powiss, S. J., E. V. Deverson, W. J. Coadwell, A. Ciruela, N. S.Huskisson,

H.Smith, G. W. Butcher, and J. C. Howard. 1992.Effect ofpolymorphism ofan

MHC-linkedtransporter on thepeptides assembled inaclass I molecule. Nature