Insulin immunization of nonobese diabetic mice induces a protective insulitis characterized by diminished intraislet interferon gamma transcription

Insulin immunization of nonobese diabetic mice

induces a protective insulitis characterized by

diminished intraislet interferon-gamma

transcription.

A Muir, … , J Krischer, N Maclaren

J Clin Invest.

1995;

95(2)

:628-634.

https://doi.org/10.1172/JCI117707

.

We reported previously that daily injections of isophane insulin prevented both

hyperglycemia and insulitis in nonobese diabetic (NOD) mice (Atkinson, M., N. Maclaren;

and R. Luchetta. 1990. Diabetes. 39:933-937). The possible mechanisms responsible

include reduced immunogenicity of pancreatic beta-cells from "beta-cell rest" and induced

active immunoregulation to insulin (Aaen, IK., J. Rygaard, K. Josefsen, H. Petersen, C. H.

Brogren, T. Horn, and K. Buschard. 1990. Diabetes. 39:697-701). We report here that

intermittent immunizations with insulin or its metabolically inactive B-chain in incomplete

Freund's adjuvant also prevent diabetes in NOD mice, whereas immunizations with A-chain

insulin or with BSA do not. Adoptive transfer of splenocytes from B-chain insulin-immunized

mice prevented diabetes in recipients co-infused with diabetogenic spleen cells, an effect

that was abolished by prior in vivo elimination of either CD4+ or CD8+ cells. Insulin

immunization did not reduce the extent of intraislet inflammation (insulitis); however, it did

abolish expression of IFN-gamma mRNA within the insulitis lesions. Immunizations with

insulin thus induce an active suppressive response to determinants on the B-chain that

converts the insulitis lesion from one that is destructive to one that is protective.

Research Article

Find the latest version:

Insulin Immunization

of Nonobese Diabetic Mice Induces

a

Protective Insulitis

Characterized

by Diminished Intraislet Interferon-y Transcription

Andrew Muir,** Ammon Peck,* MichaelClare-Salzler,*Yao-Hua Song,* Janet Comelius,* Roberto Luchetta,*

JeffreyKrischer,§and Noel Maclaren*

*Divisions of Clinical Chemistry and Immunology, Department of Pathology and Laboratory Medicine, andtDepartmentof Pediatrics,

Universityof Florida, College of Medicine, Gainesville, Florida 32610; and 'University of South Florida, H. LeeMoffittCancerCenter,

Tampa,Florida 32612

Abstract

We reported previously that daily injections of isophane insulin prevented both hyperglycemia and insulitis in

non-obese diabetic (NOD) mice (Atkinson, M., N. Maclaren; and R. Luchetta.1990. Diabetes. 39:933-937). The possible mechanismsresponsible include reducedimmunogenicityof pancreatic fl-cells from ",B-cell rest" and induced active immunoregulation toinsulin (Aaen, K., J. Rygaard, K. Jo-sefsen, H. Petersen, C. H. Brogren, T. Horn, and K. Buschard.1990.Diabetes.39:697-701). Wereporthere that intermittent immunizations with insulinoritsmetabolically

inactive B-chain inincomplete Freund's adjuvant also

pre-vent diabetes in NOD mice, whereas immunizations with A-chain insulin or with BSA donot. Adoptive transfer of splenocytes from B-chain insulin-immunized mice

pre-vented diabetes in recipients co-infused with diabetogenic spleen cells, an effect that was abolished by priorin vivo

elimination of either CD4+orCD8+ cells. Insulin immuni-zation did not reduce the extent of intraislet inflammation

(insulitis); however, it did abolish expression of IFN-y mRNA within the insulitis lesions. Immunizations with insu-lin thus induce an activesuppressive response to determi-nants onthe B-chain that converts the insulitis lesion from

one that is destructive to one that is protective. (J. Clin. Invest. 1995. 95:628-634.) Key words: diabetes*

immuno-regulation *B-chain* adjuvant * therapy

Introduction

Nonobese diabetic(NOD)'miceregularly developinfiltrations

of theirpancreaticIslets ofLangerhans by mononuclear leuko-cytes(insulitis).The lesion oftenprogressestoan

autoimmune-This workwaspublishedinabstract form(1993.Diabetes. 42[Suppl. 1] :5a).

AddresscorrespondencetoAndrewMuir, M.D., Departmentof

Pa-thologyandLaboratory Medicine, JH Miller HealthCenter, P.O. Box

100275, UniversityofFlorida, Gainesville,FL32610. Phone:

904-392-3741;FAX:904-392-6249.

Receivedfor publication10May1994and inrevisedform13 Octo-ber 1994.

1.Abbreviations usedinthispaper:IFA,incompleteFreund'sadjuvant;

NOD,nonobese diabetic.

mediated destruction of insulin secreting /3-cells and insulin-dependent diabetes, especially in female mice. Although all NOD/Uf mice acquire considerable isletinfiltrates, only 30% of males and 80% of femalesactually develop sustained hyper-glycemia by 40 wk of age. The insulitis and diabetes can be transferredtoirradiated recipients, indicatingthat an active au-toimmune cellular response is involved. Prior eradication of either CD4+ orCD8+ Tlymphocytes from the donor spleno-cytes ablates this ability to transfer diabetes. It has also been reported that immunomodulatory treatments with complete Freund's adjuvant or oralinsulin can prevent or delay the onset of hyperglycemia in NOD mice without eliminating islet infil-tration (1-3). Ithas been proposed thatsuch treatmentsalter thefunctional phenotypes of intraislet leukocytes, such that the insulitis lesions become "protective" rather than destructive (4). We report here that deliberateimmunization of NOD mice

toinsulin also inducesprotective insulitis lesions.

Bothdiabetes and insulitis can be prevented in NOD mice (5), BB rats (6), and perhaps even in humans (7) by daily

prophylactic injections of insulin. An hypothesis of "/3-cell rest" has beenproposedtoexplain thisprotection, wherebythe pancreatic

,/-cells

enter a"quiescent"statein whichexpression of cellularautoantigens is reduced (8, 9). Daily administrations of insulintodiabetic patients, however, is knowntoinducean

active immune response characterizedby insulin antibodies.We

have thereforeinvestigated whether alteredimmunitytoinsulin in NOD mice could have a role in the antidiabetic effect of prophylactic insulin therapy. Our studies indicatethat anactive immunization of NOD mice to an epitope on the B-chain of insulincanpreventdiabetes, probably bystimulatinganactive immunoregulatory response that inhibits intraislet CD8+ cyto-toxicT-lymphocyte effector functions.

Methods

Immunizations ofinsulin and its B-chain. Six units (-220

lsg)

of

purified porcine isophaneinsulin (Novo Nordisk, Copenhagen, Den-mark) inincompleteFreund'sadjuvant(IFA)(GIBCO,GrandIsland,

NY) wereadministered subcutaneously in theinguinal, axillary,and dorsal neckregionsofweanlingNOD mice (n = _{12).Littermate}

con-trolsreceivedcommercial hormone-freeisophaneinsulin diluent(Novo)

in IFA (n = 10). Ina second study, immunizations withequivalent

doses of human recombinant insulinwereperformed(n=_13),whereas littermate controls receivedinjections ofinsulin diluent(n = ₁₂₎ or

equimolaramountsofeither human recombinant insulinA-chain (105

ttg,n = ₁₂₎orB-chain (140_jg, n = 13)in IFA (theinsulins used

werekindly provided by Dr. R.Chance,Eli Lilly,Indianapolis, IN).

Theimmunizingtreatmentsbeganat4wkofage withsubsequent

injec-tionsgivenatages 12, 16, and 22 wk in thefirst study andat 12, 20,

and28 wk in the secondstudy.Bloodglucoselevelsweredetermined withChemstrips bG and diabetes wasdiagnosed whenhyperglycemia

ofover240mg/dlwasfound.

628 Muiretal.

J.Clin. Invest.

© The AmericanSocietyfor ClinicalInvestigation,Inc.

0021-9738/95/02/0628/07 $2.00

Humoraland cellular responses toimmunization. Insulinbinding

by serumantibodies was determined witha liquid-phase competitive radiobinding assay aspreviously described(10). Splenocyte

prolifera-tions were assessed by tritiatedthymidine incorporation (1 kCi/well)

duringthelast 16 h of an 88-h incubation.Groupsof three mice, immu-nized at 4 and 8 wk, hadtheirspleen cells harvested at 9 wk. Using flat-bottomed96-wellplates, 5 x I0O cells/wellwereincubated in tripli-cate with test antigens in Click's media (Eagle High Amino Acid)

supplemented with 0.5% normal mouse serum (11). The cells were

then harvested and the radioactive uptakes were measured using an automated apparatus(Matrix 96, Packard,Meridan, CT).All

prolifera-tionstudieswereperformedon atleast three separateoccasions.

Intravenous splenocyte cotransferstudies. Weanling female mice wereleft untreated or were immunized at 4 and 8 wk of age withIFA

emulsions ofeither B-chaininsulin,A-chaininsulin,ordiluent. At 10 wkof age,20 x 106of their splenocytes were mixed in equal proportion with splenocytes pooledfrom acutely diabeticcodonors and then

in-jected intravenously toirradiated (700 rad)naive male littermates of theimmunizedcodonor mice.Fifteen recipients were co-infused with

mixtures containing splenocytesremoved after immunizations with B-chaininsulin,whereas another 14recipientsreceived control mixtures

of splenocytes (untreated,n =3;diluenttreated,n =8;A-chaininsulin treated, n = 3 ).

In vivo depletion of CD4+ and CD8+ lymphocytes. Synergistic

depletingmonoclonal antibodiesagainstCD4+ (YTS 191.1 and YTA

3.1.2)and CD8+ (YTS 169.4and YTS 156.7)T lymphocytes were kindlyprovided by Dr. Anne Cooke(Cambridge University,UK).

Con-trol micereceivedinjections of either PBS or an irrelevantIgG mono-clonalantibody against influenzaAnucleoprotein (kindly provided by

Dr. Parker Small, University ofFlorida, Gainesville). Starting 1 wk

after completingatwo-doseimmunizingcourseof B-chain insulinat4 and 8 wk of age,oneintravenous dose followedbytwointraperitoneal

monoclonal antibodyor control treatments were administered. These weregiven7, 5,and 3 d before the animalswerekilled. Thedepleting

capacityof thesehybridomashave beenrepeatedlydemonstrated in the

laboratory ofHerman Waldmann atCambridge,U.K., both in vitro, by

Cr5' release assays, and in vivo, by staining ofperipheral blood or

lymphnode cells with targetcell-specificfirst antibodies and biotinyl-ated secondantibody againstratIgGand rat Klightchains. In antibody-treated thymectomized mice,noregeneration ofeither ofthetargeted lymphocytes subsets wasobserved, whereas CD4+ T-lymphocyte re-covery after a single dose of YTS 191.1 in euthymic mice was not

detectable for upto 10 d after the antibodytreatment. In vivo

anti-CD4 antibodytreatmentsalso reducedprimaryandsecondary humoral

responses to sheep erythrocytes andinhibited mitogen-stimulated

IL-2 production. Combined anti-CD4 and anti-CD8 treatment conferred

toleranceto allogeneic skingrafts thatwas more long-livedthan that inducedbyeitherantibody givenalone (12-14).

Weperformedflowcytometric analyses of splenocytes fromtreated

orcontrol miceon aFACScan (Becton-Dickinson, SanJose, CA)after

stainingwithGK1.5 (anti-CD4)and clone53-6.7(anti-CD8)antibodies

(Pharmingen,SanDiego, CA). Using splenocytesfrom B-chain immu-nized donors whose T cells were depleted before they were killed,

adoptivecotransferswererepeated. Beforetransfer,six donors received

treatments with anti-CD4 antibodies, eight animals were treated with anti-CD8 antibodies, and sevenreceived injections of either PBS (n =3)ortheirrelevantIgGmonoclonalantibody(n =4).

Histology.FemaleNOD mice immunizedtoeitherA-chain (n =9)

or B-chain (n = ₇₎ insulin ortreated with insulin diluent inWFA (n

=9)at4and 8 wk of age underwentpancreatic histological studies. They were killed at 10 wk of age (the age of established insulitis but before theironsets ofdiabetes), and their pancreases wereremoved, fixed in 10%formalin,and stainedwithhematoxylinand eosinfor light

microscopicexaminations. The slides were coded and an insulitis score

was determined by three independent examiners. The results were

pooled and averaged before the codes were broken. The degree of

insulitis was scored from 0 to 3 using a previously described dual

scale that assessed both the severity of immune infiltration and the

cytoarchitectural disruption (5).

Cytokineproduction withininfiltrated islets. At 4 and 8 wk of age, groupsof threetofourmice each receivedmultiplesubcutaneous

injec-tions of emulsions of IFAcontaining either insulindiluent, A-chain,or B-chain. The studies were also controlled with groups of untreated littermates. At 10 wk of age, islets from nondiabetic mice were hand-picked after pancreatic mincing and collagenase/DNAsedigestion. The islets were dispersed with Trypsin at370Cand the mRNAwasextracted using the Micro-Fast Track mRNA Isolation Kit (InVitrogen, San Diego, CA).Cytokine-specific primers weredesignedasoutlined pre-viously (15). After completing RT-PCR and agarose electrophoresis, theDNA was Southernblotted to nylon membranes (Boehringer-Mann-heim,Indianapolis, IN) in 0.4 N NaOH and their identities confirmed usingdigoxigenin-labeledcytokine-specificinternalprobes and the

Ge-niusdetection system (Boehringer-Mannheim). The numberof islets

from which message was extracted was considered as thestandardfor

comparison between groups. The studieswereperformedintriplicate.

The yield of DNA product using this PCR protocol has a log-linear relationship with the concentration of added template (IFN-y cDNA from Clontech, Palo Alto, CA) over at least sixorders of magnitude (50 aM-50 nM). Theprofiles reported below werereadily reproducible

inreplicate experiments.

Statistics. The method ofKaplan and Meier (16)wasused to con-structlife tables and the logrankChi-square statistic was used to compare them (17).AStudent'sttest orone-way ANOVA was used to compare means. Pvalues are two sided.

Results

Immunizations of insulin and its B-chain protect NOD mice from diabetes. Becausethe twoimmunization protocols yielded

similarresults, only the combined dataarepresentedhere. As shown in Fig. 1, insulin immunizations dramatically lowered the incidence of diabetes over the 40-wk study period. The

protection against diabeteswasreproduced by immunizing with B-chain insulin but, importantly, notby immunizing with

A-chain insulin or with BSA as controls (latter data not pre-sented).

The insulin immunizations stimulated active antigen-spe-cificresponses inboth the humoral andthecellulararmsof the immune system as expected, documenting that immunizations hadnotinducedantigen-specific anergy. Fig. 2A presents the frequency of anti-insulin antibody-positive mice after immuni-zation. In all of eight insulin-immunized mice, specific anti-insulin antibodieswere detected atlevels that were 2 10-fold

above thosetypical of insulin autoantibodiesinuntreatedNOD mice inourradiobinding assay. Similarly, increases of insulin binding antibodieswerealso observedin serafrommice immu-nizedagainst eitherA-orB-chaininsulin. Antigen-specific

pro-liferative responses, although modest, were also consistently

stimulatedasdemonstrated inFig. 2 B.

Protectionfrom diabetes canbe adoptively transferred to

naive recipients. To confirm the active immune basis of the therapeutic effect of B-chain insulinimmunizations, 10-wk-old maleirradiated recipient NOD micewereinjected with

spleno-cytesfrom immunized femalelittermatestogether withanequal number of splenocytes harvested from female diabetic co-donors. Life table analysis revealed a significant delay in the onset of diabetes among the mice that received infusions of splenocytes from B-chain insulin-immunized donors (P

---_{A-CHAIN+IFA B-CHAIN+IFA} -- -DILUENT+IFA

"

-

_

-I _ _ _ _ _ _ _ _

I- - -

-I

1

L_

.I

L

- I

---~ ~ ~ ~ ~ ~ ~~~~~~~~~~~~~iii

0 1 2 3 4 5 6 7 9 1011 12131415161718192021222324252627282930313233343636373394041

Figure1. Subcutaneous insulin peptides in IFA in NOD mice. The survival curve shows the probability of remaining nondiabetic among fe-maleNOD mice who received

WFA

suspensions containing either insulindiluent,isophane insu-lin, insulin A-chain, or insulin B-chain. Immuni-zationwith insulin B-chain provided protection equaltothatconferredbyintactinsulin.

diabetic within 3 wk of cell transfer. Incontrast, diabetes

oc-curred inonly 3 of 15 (20%) recipients of splenocytes from

B-chain-immunized donorsoverthesametimeframe.

InB-chain insulin-immunized micethatreceivedaninvivo

course of lymphocyte-depleting antibodies before they were

killed,flowcytometric analysisofsplenocytes suggested

virtu-allycomplete depletion of the respectiveT cell subsets, 3 full dafter thedepleting antibodycoursehad beencompleted,

sup-porting thatspecific Tcelldepletions had been affected. Both the anti-CD4 and anti-CD8treatmentsreduced theirrespective

targetcellmarkersby 96% comparedwithPBS-injectedcontrol mice(Fig. 4A).Theseprofileslasted fordaysaftercompleting

the antibody course, supporting that specific Tcell depletions

had been affected. Depletionofeither CD4+ orCD8+ T

lym-phocytes before transfer abrogatedthe abilityof B-chain-im-munized splenocytes to prevent the transfer of diabetes (P

<0.01).3wk after thesplenocyte transfer,fiveof sixrecipients depleted of CD4+ T cells and four of fiverecipients depleted

ofCD8+ T cells becamehyperglycemic (Fig. 4B).These

ex-periments suggest that B-chain insulin immunizations inhibit isletdestructionbypromotinganactiveantigen-specific

immu-noregulatory lymphocytenetwork withsuppressionof isletcell

autoimmunity.

Histological insulitis grade is not decreased by B-chain insulin immunization. Despitethe lowerfrequencyof diabetes

seenin mice with B-chain insulinimmunization, histologic stud-iesrevealed that theextentofmononuclearleukocyteinfiltration of their islets was not reduced by the treatments. Using an

infiltration scale of 0-3 (1),themeanandSEinsulitis scores

in pancreases of 10-wk-old mice immunized at4 wk of age

were0.9±0.2, 1.3±0.3,and 1.5±0.2 for diluent(n = 9mice), A-chain(n=9mice),and B-chain(n=7mice)treated_groups,

respectively (P = 0.3). The time chosen reflected an early

phaseofthe established lesion.Further,flowcytometric

analy-ses revealed no significant differences in the proportions of

CD3+, CD4+, CD8+, or I-A+ cells in dispersed islet cell

preparationsfromtreatedand control mice(datanotpresented). Immunizations alter cytokine production within infiltrated

islets.Toassesswhether functional differences existed between

the islet-infiltratingimmune cells of treated and control mice, cytokine mRNAprofiles wereexaminedby RT-PCRanalyses (15).Thepanelofcytokine-specific primersweused included

IL-1,8,

IL-2, IL-4, IL-10, TGF-,33, and IFN-y. Thesalient

re-sults are presentedin Fig. 5. High levels ofIL-1O and

TGF-/33 wereobserved in untreated mice but werenotconsistently

increased further by insulin immunizations. Increased levels of

IL-4 mRNAwereobservedafterimmunizations with IFA alone, aneffect thatwasnotseenafterspecific immunizationstoeither of theinsulinpeptide chains. ImmunizationswitheitherA-chain

orB-chainincreased intraisletlevels of IL-If and 11-2mRNA

comparedwith controls. However, themostnotablechange that

was specific toislet infiltrates of thehighly protected B-chain insulin-immunized mice was a consistently reduced level of

IFN-y mRNA.

As demonstrated by in vivo depletion of CD4+ orCD8+

cells from control unimmunized NOD mice, CD8+ cells are

thepredominant source of intraislet IFN-y. Cytokine mRNA

levels were, however, enhanced when bothTcell subsets

re-mained(Fig. 6).Incontrast,depletionofCD4+cells in B-chain

insulin-immunizedmicereducedthelevelsof IFN-ybelowthe

limits of detection, documenting that areduction inCD8+ T

lymphocyte transcriptionofIFN-yhadoccurred after immuni-zations with B-chain insulin(Fig. 6).

Discussion

Anincreasing number ofself-antigens targetedinautoimmune disease states have been identifiedatthemolecularlevel,

per-mittingautoantigen-based immunotherapiestobecomeamajor

research focus (18-24).Todate,theseantigen-based therapies

haveattemptedto restoreimmunological toleranceorto other-wiseinduceunresponsivenesstoself-antigens implicatedas im-portantindiseasepathogenesis. Theabilityofimmunizationto

insulin or its B-chain to prevent IDD in an antigen-specific

fashion is thereforenovel,because itdependson an active im-munesensitizationtotheautoantigenandnotanergyinduction.

Antigen specificity fortheeffectwasdemonstratedbythe

per-sistence ofdiabetes in NOD mice immunized with either

A-chain insulinorBSA in WFAandprotectionwhen immunization

to B-chain was used. The induction of both humoral and in

vitro cellularresponsesspecifictotheimmunizing antigen docu-mented thatanactiveimmunoregulatory processwasinvolved andspecifically ruledoutthepossibilitythatantigen-mediated immunological anergy hadbeen induced. Thishypothesiswas

confirmed in the adoptive cotransfer experiments, through

which diabeteswaspreventedinnaive irradiated micereceiving 630 Muiretal.

100

s0

601I

la

I

40 .

201.

0

~~INSULIN

+IFA

.

I

A 100

* * (11)

80 (12) (9)

*

p<O.0003

(0

E 0 40

(0~~~~01

0.. 20

(10)

0

MO)eum IFA AsChain s Chain Insuin

B

E/

0.1500

1.000/

500

0

MEDIA A20 A.2 8-20 8-2 MEIAA*20 A-2 8.20 82

A-Chain-Inmmuized S-ChainImunze

Figure2.Antigen-specifichumoral and cellular responsesto immuniza-tion.(A)SerumIAAlevelsweremeasured insamplesthatwereobtained

before the second immunization at12 wkof age. The number of mice testedateachtime is presented in parentheses above the bar.Alow

frequencyof spontaneously appearing insulin autoantibodieswas ob-served in diluent and diluent + IFAinjectedmice. Those thatwere

immunizedtoinsulinorits component peptides had highfrequencies

and absolute levelsof anti-insulin antibodies. (B)Invitroproliferative

responses ofspleen cells to 2 and 20

pg/ml

of A-chain and B-chain insulins are shown. 9-wk-old micewerekilled after they had received immunizationsof A-orB-chaininsulin in IFAat4and 8 wk. Error bars represent 1 SDof thetriplicatemean.Antigen-specificresponses wereobserved thatweredosedependent.Similar resultswereobtained infour separate experiments.

infusions of splenocytes from diabetic donors but only when cotransfers ofsplenocytesfrom B-chain insulin-immunized

lit-termates were simultaneously performed. The requirement for

bothCD4+ andCD8+ Tlymphocytesto successfully transfer

the antidiabetic effect of B-chain insulin immunization suggests

thatthe treatment actsbyinducing acomplex lymphocyte

im-munoregulatorynetwork thathasprotective properties (25, 26).

Both CD4+

_Th,

andnoncytotoxic CD8+ T-lymphocyteclones have been isolated from NOD mouse islets that can prevent

diabetes when injected back into susceptible NOD recipients (27, 28). Furthermore, similarcotransfers ofspleencellsfrom diabetic andyoungnondiabetic donors result in protection from diabetes in therecipients (29).

ImmunizationtoA-chain insulin inducedmanyofthe same

immunologic

responses observed after B-chain insulin

treat-ments, yet it didnotprotectagainst diabetes. Specific humoral

and cellular responses were induced and there were parallel changes in intraisletproduction ofIL-1, IL-2, and IL-4 using

immunizationswith either chainof insulin. Whereas these im-munization-induced responses mayin factrepresentirrelevant changes,theprotective CD4+/CD8+Tcell-dependent

immu-noregulatory network induced by B-chain insulin immuniza-tions wasassociatedwith reduced intraislet CD8+ lymphocyte production of IFN-y. Reductions ofIFN-ymRNAlevels inthe

CD4+ T-lymphocyte population may also have been induced

by immunization with insulin B-chain, but this remains

un-provenand is afocus for further study.

Insulin isone ofmany/8-cell autoantigens targeted by the autoimmuneresponse thatleads to diabetesinNOD mice,but

autoimmuneresponsesagainstthe hormone havenotbeen

con-siderednecessary for the induction of diabetes (30). Thenear

complete protection from disease after therapeutic immuniza-tion with B-chain insulin is therefore consistentwith the possi-bilitythat "bystanderimmunosuppression" has been induced. This termdescribes the antigen specificimmunoregulatory cell releaseof cytokines by activatedTcells,upon reencounterwith the inciting antigen. The paracrine actions of such cells then

down-regulateotherautoreactiveactivated immunecellsin the

isletinflammatory lesion, irrespective of theircognateantigens

(31-33). The effect has been studied most in experimental

allergic encephalomyelitis, induced in rodents by

immuniza-tions with myelin basic protein. The release of TGF-/3 and/

or IL-4frommyelin basicprotein-specific suppressorcells is believed to ameliorate the central nervous system disease of experimental allergic encephalomyelitis byinhibitingthe

func-tionsofproinflammatory cells. IL-10 is anotherregulatory

cy-tokine that could have such actions; however, bothbeneficial and detrimental effects of IL-10 on islet viability have been described in NOD mice(34, 35). The intraisletcytokineprofiles

induced by B-chain insulin immunization, however, did not

demonstrateelevations ofanyofthese cytokines, whichwere,

however, abundant both before and after the immunizations. Reductions of intraislet IFN-ymRNA wererepeatedly dem-onstrated in animals who received B-chain insulin

immuniza-tions but were not observed in control mice receiving any of thenonprotectivetherapies. Recently,infiltratingCD8+T

lym-phocytes have been reported to be the predominant source of

IFN-y in NOD mouseislets (15). In agreementwith this, we

found that in vivo antibody treatments thatdepleted CD8+ T

cells of normal NOD mice reduced intraislet IFN-y mRNA more than did depletions of CD4+ T cells. Thus there was

considerabletranscription of IFN-ymRNAfrom CD8+Tcells

in normal NOD mice. In contrast, islets from B-chain insulin-immunized mice demonstrated marked reductions of IFN-y transcripts and their virtual eliminationwasobservedafter

sub-sequent anti-CD4 treatments in vivo, whereas the respective effects from anti-CD8+ treatment were much less obvious.

These results indicate that B-chain insulin immunizations

re-duced IFN-y productionatleastby CD8+ Tcells.We suspect that the immunization-induced process that protects islets is probably complex, and one that will require further study to unravel.

Reduction of intraislet IFN-y may spare

/3-cells

directly because the cytokine itselfappears to have important effector

actions in islet cell autoimmunity. In combination with other

cytokines suchas TNF-a or IL-1, IFN-y has beenreported to

ACHAIN DIWUff 3CHAIN

I

a

100

SO

40

o~~~~~~

. . . .

* 1 2 3 4 S 6 7 S

insulitis and diabetes have been demonstrated in mice selec-tively expressing intraislet IFN-y through a transgene linked to an insulin gene promoter (39). Conversely, IFN-y anti-bodytreatmentshave been showntoimprove disease outcomes (35, 40). That immunizations by B-chain insulin could so pro-foundlylower intraisletIFN-y may wellaccountforthebenefit

Figure3. Protection from diabetes can be adoptively transferred to naive recipients. Weanlingfemalemice were left untreated or were immunized at 4 and 8 wk of age with EFAemulsions of either B-chain insulin, A-chaininsulin,ordiluent. At 10 wk of age, theirsplenocytesweremixed inequal

pro-portion

with

_{splenocytes pooled}

from

_acutely

diabetic codonors and then

_injected

intrave-nously

toirradiated naive male littermates of L theimmunized codonors (untreated,n= 3;

diluenttreated,n=8;Achaintreated, n = 3; Bchaintreated, n= 15). Protectionfrom

theadoptivetransfer of diabeteswas

con-10 11 12 ferred bestbythesplenocytesof B

chain-treated codonors(P<0.007).

induced; however, the intermediate steps that promote this result need further study.

Ourstudies suggest that the role of autoimmunitytoinsulin in /3 _{cell autoimmunity leading} to insulin-dependent diabetes should be reconsidered. Insulin is an abundant autoantigen in islets that is specific to / cells. A high proportion of

islet-A

0ff I-C4 flU

B

100

0

I

I

o D

0

wtricot INi

-BeB ---. ByCNA4CD4S SCHAN

0 1 2 S 4 S 6 7

-.mI

S S 10 11 12

Figure4. Successful transfer of protection requires

both CD4+and CD8+Tlymphocytes.TheB-chain

immunizationadoptive cotransfer protocolwas

re-peatedexcept that thetreatedcodonors received anti-CD4 oranti-CD8depletingmonoclonal

anti-bodiesbeforetheywerekilled,whereascontrolmice receivedinjectionsof either PBS or an anti-influenza Anucleoprotein IgG murine monoclonal (anti-CD4,

n =6; anti-CD8,n= _{6; control,}n=_{7). (A)}Flow

cytometric analysisofsplenocytesfrom

antibody-orcontrol-treated micerevealednearly complete depletion bythe monoclonal pairs. Stainingof

spleencells from mice treated with the monoclonal controlantibodyareshownontherightmostaxisof

eachplot.Theleftplotshowsdepletionof CD4+ Tlymphocytes (96%)inanti-CD4-treatedmice

comparedwithanti-CD8-treated orcontrolmice. TherightplotdemonstratesdepletionofCD8+T

lymphocytes (96%)inanti-CD8-treated micebut

notanti-CD4-treatedorcontrolmice.(B)Depletion

of eitherCD4+ (dashed line)orCD8+(dotted line)lymphocytesallowedsuccessfuladoptive

transferofdiabetestoB-chain immunizedmice,

whereasB-chain immunization withoutlymphocyte depletion delayedthetransferof disease(solid line)

(P < 0.01).

500

_bp

=.0 .. .. ..

100 bp - |ig

as

co

_Figure

_{6. IFN-y}

mRNA 2 0 0O isdepleted primarily

C

c

-fromCD8+ T

lympho-O <; cytes. To determine the

I

I

@ @ sourceofIFN-y,female

Unimmunized + + NOD micewereeither

* _ left untreatedor

immu-w a

~~~nized

with B-chain

insu-lin

in

_FA

at4 and 8 wk of age. Both groupswere

depleted ofeither their

B-chain Insulin + CD4+ orCD8+ T

lym-immunized phocytes as described.

IFN-ytranscriptlevelsin

unimmunized mice were

highestinpreparations

depletedoftheirCD4+ but nottheir CD8+ T

lymphocytes.Among B-chain insulin-immunized mice, IFN-y mRNA could not be detected in theislet infiltrates of micedepletedof their CD8+ cells. ThusB-chain

insulin immunization had eliminated CD8+ Tlymphocyte production

ofIFN-y.

I LL A 4 I

1000 bp

Ij

500bp _

100bp -L

IFN-

7

+ 4 4

iI

1000bp 500bp

100bp

Figure 5. Immunizationsalter cytokine production within infiltrated

islets. At 4 and 8 wkofage, groupsof threetofourmice eachreceived

multiple subcutaneous injections of emulsionsof[FA containingeither insulindiluent, A-chain, B-chain,orelsewereleftuntreated. At 10wk ofage,the mRNA fromsingleislet cellsuspensionswaspurifiedfrom

nondiabetic mice. TheRT-PCRproductswereSouthern blotted and

their identitiesconfirmedusing digoxigenin-labeled internal probes. In-traislet IL-113 and IL-2mRNAswereincreased aftereither A-chainor

B-chain insulinimmunizations, whereas the level ofIL-4 mRNAwas

increasedonly aftertreatmentsof diluent + IFAalone. Immunization with either A-orB-chainwasassociated with diminished levelsof IL-4mRNA. Selectivedepression of IFN-y withrespect tolevels in

un-treatedmice(asshownwitharrows)wasobserved after B-chain insulin immunization. The blotsshownaretypicalof thoseseeninthreeseparate

experiments.

derived NOD T cell clones arereported to bereactive to

epi-topeson the B-chain of insulin. Some of these clonesactually

appearabletomediatediabetes(Wegmann, D., personal

com-munication). Diabetesmay eventuallyarise onlyafter

pancre-atic /3-cells sufferprogressive damagefromacascade of autoim-muneresponses against multiple autoantigens.Insulin B-chain immunizationtherapy mayinhibitprogressionof the

autoanti-gendeterminant cascade(23).

Although the exact immune mechanism by which insulin

immunizationprotectsagainst diabetes remains under investiga-tion, we suggest thatimmunizations to insulin B-chain stimu-latesanimmunoregulatorynetworkcomplexthatinducesa

pro-tective immuneresponse, characterizedbyareduction of IFN-y by intraislet CD8+ T lymphocytes. This unique approach

of therapeutic immunostimulation stands in stark contrast to

currently proposedtreatmentsthat aimtoinhibitautoimmunity through induction ofanergyto self-antigens. Further

develop-mentofthis modelmay well lead to human therapeutic trials forinsulin-dependentdiabetes ifnotother autoimmunediseases, especially wheninvestigationsfocusonpeptidesthatare

with-outmetabolicactivity.

Acknowledgments

This workwassupported bygrantsfromthe National Institutesof Health (ROl HD19469-09), the Juvenile Diabetes Foundation, International

(1929523), theChildren's Miracle NetworkTelethon,andagenerous

gift from SidneyKriser.

References

1.Shehadeh, N., F. Calcinaro, B. J. Bradley, L. Bruchlim,P.Vardi, and K. J. Lafferty. 1994. Effect of adjuvant therapyondevelopment of diabetes inmouse

andman.Lancet. 343:706-707.

2.Sadelain, M.W., H. Y. Qin, J. Lauzon,and B.Singh.1990.Prevention of

typeI diabetes in NOD miceby adjuvant immunotherapy. Diabetes. 39:583-589. 3.Muir, A., D. Schatz, and N. Maclaren. 1993. Antigen specific

immunother-apy.Oral tolerance and subcutaneousimmunization in thetreatmentof

insulin-dependent diabetes. Diabetes Metab. Rev. 9:279-287.

4.Shehadeh, N.N.,F. LaRosa,and K. J.Lafferty. 1993. Alteredcytokine activity in adjuvant inhibition of autoimmune diabetes. Autoimmunity. 6:291

-300.

5.Atkinson, M., N. Maclaren, and R. Luchetta. 1990. Insulitis and diabetes inNOD mice reducedbyprophylactic insulin therapy. Diabetes. 39:933-937.

6. Gotfredsen,C.F.,K.Buschard, andE. K. Frandsen. 1985.Reduction of

diabetes incidence of BB Wistarratsby early prophylactic insulintreatmentof diabetes-prone animals. Diabetologia. 28:933-935.

7.Keller,R.J.,R. A.Jackson, and G. S. Eisenbarth. 1993. Insulin prophylaxis

inindividualsathigh risk for the development oftype1 diabetesmellitus. Lancet. 341:927-928.

8.Aaen, K.,J.Rygaard,K.Josefsen, H. Petersen, C.H.Brogren,T.Horn,

andK.Buschard. 1990.Dependenceofantigen expressiononfunctionalstateof beta-cells. Diabetes. 39:697-701.

IL-1

8

IL-2

1000 bp 500bp

100bp

9.Bjork, E., 0. Kampe, J. Grawe, A. Hallberg, I. Norheim, and F. A. Karlsson. 1993. Modulation of beta-cell activity and its influence on islet cell antibody (ICA)and islet cell surface antibody (ICSA) reactivity. Autoimmunity. 16:181

-188.

10.Palmer, J. P., C. M. Asplin, P. Clemons, K. Lyen,0.Tatpati, P. K. Raghu, and T. L. Paquetto. 1983. Insulin antibodies in insulin-dependent diabetics before insulin treatment. Science (Wash. DC). 222:1337-1339.

11.Peck, A. B., and F. H. Bach. 1973. A miniaturized mouse mixed leukocyte culture in serum-free and mouse serum supplemented media.J.Immunol. Meth-ods. 3:147-164.

12. Qin, S., S. Cobbold, H. Tighe, R. Benjamin, and H. Waldmann. 1987. CD4 monoclonal antibody pairs for suppression and tolerance induction. Eur.J.

Immunol. 17:1159-1165.

13.Cobbold,S. P., A. Jayasuriya, A. Nash, T. D. Prospero, and H. Waldmann. 1984. Therapy with monoclonal antibodies by elimination of T-cell subsets 24 in vivo. Nature (Lond.). 312:548-551.

14.Kong, Y.-C. M., A. G. Alvaro, H.Waldmann, S. P. Cobbold, and B. E. Fuller.1989. Resistance to experimentalautoimmunethyroiditis: L3T4+ cells as mediators of both thyroglobulin-activated and TSH-induced suppression. Clin. Immunol.Immunopathol. 51:38-54.

15.Anderson, J. T., J. G. Cornelius, A. J. Jarpe, W. E. Winter, and A. B. Peck. 1993. Insulin-dependent diabetes in the NOD mouse model. II. 63 cell destruction inautoimmunediabetes is aTH2and not a THI event. Autoimmunity. 15:113- 122.

16.Kaplan, E. L., and P. Meier. 1958. Nonparametric estimation from

incom-pleteobservations. J. Am. Stat. Assoc. 53:457-481.

17.Mantel, N. 1988. Evaluation of survival data and two new rank order statisticsarising inits consideration. Cancer Chemother. Rep. 50:163-170.

18. Lider, O., L. M. B. Santos, C. S. Y. Lee, P. J. Higgins, and H. L. Weiner. 1989. Suppressionof experimental autoimmune encephalomyelitis by oraladministration of myelin basic protein. II. Suppression of disease and in vitro immune responses is mediated by antigen specific CD8+ T lymphocytes. J. Immunol. 142:748-752.

19. Zhang, Z. J., S.Shoelson,A.Miller, and H. L. Weiner. 1992. Insulitis is suppressed inNOD mice by oral administration of insulin peptides and glucagon. FASEB(Fed. Am. Soc. Exp. Biol.) J. 6:1693.

20.Elias,D., and I. R.Cohen. 1994.Peptidetherapyfor diabetes in NOD mice. Lancet. 343:704-706.

21.Adorini, L., V. Barnaba, C. Bona, F.Celada,A.Lanzavecchia,E.Sercarz,

N.Suciu Foca, and H. Wekerle. 1990. New perspectives on immunointervention inautoimmunediseases [news]. Immunol. Today. 11:383-386.

22.Steinman,L. 1990.Developmentofantigen-specific therapiesin autoim-munedisease. Mol. Biol. Med. 7:333-339.

23.Kaufman,D.L.,M.Clare-Salzler,J.Tian,T.Forsthuber,G. S. P.Ting,

P. Robinson,M. A. Atkinson,E. E.Sercarz,A.J.Tobin,and P. V. Lehmann. 1993.Spontaneous loss of T cell self tolerance toglutamate decarboxylaseisakey

eventin thepathogenesisof murineinsulin-dependentdiabetes. Nature(Lond.).

366:69-71.

24.Tisch,R., X.-D.Yang,S. M.Singer,R.S.Liblau,L.Fugger,and H. 0.

McDevitt.1993. Immune response toglutamicaciddecarboxylasecorrelateswith insulitis in non-obese diabetic mice. Nature(Lond.).366:72-75.

25.Kawamura,T.,M.Nagata,T.Utsugi,and J.-W. Yoon. 1993. Prevention of autoimmune type 1 diabetesbyCD4+ suppressor T cells in

superantigen-treatednon-obese diabetic mice. J.Immunol. 151:4362-4370.

26.Serreze,D.V.,K.Hamaguchi,andE. H. Leiter.1989. Immunostimulation circumvents diabetes in NOD/Lt mice. J. Autoimmun. 2:759-776.

27.Reich,E.P.,D.Scaringe,J.Yagi,R. S.Sherwin,and C. A.Janeway. 1989. Prevention of diabetes in NOD miceby injectionofautoreactiveT-lymphocytes.

Diabetes. 38:1647-1651.

28.Pankewycz, O.,0.Strom,and V. E.Rubin-Kelley.1991.Islet-infiltrating

T cell clones from non-obese diabetic mice that promote or prevent accelerated

onsetofdiabetes. Eur. J.Immunol. 21:873-879.

29.Boitard, C.,R. Yasunami,M.Dardeene, and J. F. Bach. 1989. T cell-mediatedinhibition of the transfer of autoimmune diabetes in NOD mice. J.Exp.

Med. 169:1669-1680.

30.Atkinson, M. A.,and N. K. Maclaren. 1993. Islet cell autoantigens in

insulin-dependentdiabetes. J. Clin. Invest. 92:1608-1616.

31.Miller,A., 0.Lider,A. B. Roberts, M. B. Sporn, and H. L. Weiner. 1992.

SuppressorTcellsgeneratedbyoraltolerizationtomyelinbasicproteinbothin vitro and in vivo.Immuneresponsesbytherelease of transforming growth factor

,8afterantigen-specific triggering.Proc.Natl. Acad. Sci. USA. 89:421-425. 32. Khoury,S.J.,W. W.Hancock,and H. L. Weiner. 1992. Oral tolerance

to myelinbasic protein and natural recovery fromexperimental autoimmune

encephalomyletisareassociated withdownregulationofinflammatory cytokines

anddifferential upregulationoftransforming growthfactorbeta,interleukin4,

andprostaglandinE expression in the brain. J. Exp. Med. 176:1355-1364. 33. Karpus, W. J., and R. L. Swanborg. 1991. CD4+ suppressor cells inhibit the function of effector cells of experimental autoimmune encephalomyelitis

through a mechanism involving transforming growth factor-fl. J. Immunol. 146:1163-1168.

34.Wogensen, L.,M. S.Lee,and N.Sarvetnick. 1994. Production of interleu-kin 10by islet cells accelerates immune-mediated destruction of beta cells in nonobese diabetic mice. J. Exp. Med. 179:1379-1384.

35.Pennline,K.J.,E. Roque-Gaffney, and M. Monahan. 1994. Recombinant human IL-10 prevents the onset of diabetes in the nonobese diabetic mouse. Immunol. Immunopathol. 71:169-175.

36. Rabinovitch, A.,W.Sumoski,R.V. Rajotte, and G. L. Warnock. 1990. Cytotoxic effects ofcytokinesonhumanpancreaticislet cells inmonolayer cul-ture.J. Clin. Endocrinol. Metab.71:151-156.

37. Foulis, A. K., M. McGill, and M. A. Farquharson. 1991. Insulitis in type I(insulin-dependent) diabetes mellitus inman-macrophages, lymphocytes, and

interferon-gamma containingcells. J.Pathol.165:97-103.

38. Campbell, I.L.,A. Iscaro, and L. C. Harrison. 1988.IFN-gammaand tumor necrosisfactor-alpha.Cytotoxicitytomurineislets ofLangerhans. J. Immunol. 141:2325-2329.

39.Sarvetnick,N., J.Shizuru,D.Liggitt,L.Martin,B.McIntyre,A. Gregory, T.Parslow, and T. Stewart. 1990. Loss of pancreatic islet tolerance inducedby ,1-cellexpression ofinterferon-gamma.Nature(Lond.). 346:844-847.

40.Campbell, I., T. W. H. Kay, L.Oxbrow,and L. C. Harrison. 1991. Essential roleforinterferon-gamma and interleukin-6 in autoimmuneinsulin-dependent

diabetes in NOD/Wehi mice. J. Clin. Invest. 87:739-742.