Epitope Spreading and Autoimmune Glomerulonephritis in Rats Induced by a T Cell Epitope of Goodpasture's Antigen

Epitope Spreading and Autoimmune Glomerulonephritis inRats Induced by a T Cell Epitope of Goodpasture’s Antigen

Warren Kline Bolton,* Lanlin Chen,* Thomas Hellmark,‡ Jorgen Wieslander,‡ Jay W. Fox†

*Department of Medicine, Division of Nephrology, and †Department of Microbiology, University of Virginia HealthSystem, Charlottesville, Virginia; and ‡Department of Nephrology, Lund University Hospital, Lund, Sweden

An amino-terminal region of �3 chain of type IV collagen noncollagenous domain [�3(IV)NC1] that induces experimentalautoimmune glomerulonephritis (EAG) in rats has been identified. Only recombinant antigens that contain a nine–amino acid(AA) span of �3(IV)NC1, consistent with a T cell epitope, could induce EAG. It was hypothesized that synthetic peptides ofthis region should induce EAG. Human and rat peptides of this region were synthesized and rats were immunized to definethe nephritogenic epitope. A 13-AA rat peptide induced EAG with proteinuria, decreased renal function, and glomerularbasement membrane (GBM)-bound deposits in half of the rats. This peptide induces lymph node cell proliferation anddevelopment of antibodies to epitopes of �3(IV)NC1 external to the peptide immunogen. Carboxy-terminal extension to 21amino acids results in all rats’ demonstrating anti-GBM antibody and severe EAG. Asparagine at position 19 is critical for EAGinduction. None of the 50 rats that were immunized with peptide that contained human sequence with isoleucine at position19 developed EAG, whereas rat sequence with asparagine 19 induced EAG. Truncation of amino terminal AA of the peptideaborts EAG induction. These studies demonstrate that a T cell epitope of �3(IV)NC1 induces lymph node cell proliferation,EAG, and intramolecular epitope spreading; that the length of this peptide influences the formation of anti-GBM antibody;and that the presence of asparagine at position 19 of the peptide is critical to disease induction.

J Am Soc Nephrol 16: 2657–2666, 2005. doi: 10.1681/ASN.2004100823

G oodpasture’s syndrome is one of the few forms ofautoimmune glomerulonephritis in humans forwhich the antigen has been identified (1,2). Epitope

mapping using sera from patients with Goodpasture’s syn-drome has identified the �3 chain of type IV collagen noncol-lagenous domain [�3(IV)NC1] of the glomerular basementmembrane (GBM) as the responsible antigen (3,4). Further map-ping has localized the antibody immunodominant region to theamino-terminal third of the �3(IV)NC1 (5–9). Although thisdisease has been considered an antibody-mediated process,there is persuasive evidence to suggest that cellular immunitymay play an important role in its induction (10). This is basedon phenotypic analysis of kidney biopsies, in vitro evidence ofcellular-mediated immunity (CMI), and relatively poor corre-lation of clinical course with antibody titers (1). Therapeuticresponse occurs with modalities that remove antibodies as wellas those that curb CMI.

Delineation of pathogenic mechanisms is essential to tailor-ing therapeutic intervention. To that end, we and others haveused a model of experimental autoimmune glomerulonephritis(EAG), which recapitulates Goodpasture’s syndrome in hu-mans (11–15). Using this model, it has been demonstrated thatantibodies alone and CMI alone can induce disease (16–24).

EAG develops in animals without antibody production, trans-fer of disease without antibody can be accomplished with cells,and disease in which antibodies are not detectable on the GBMmay be induced (12,22,25,26). We have shown that the respon-sible epitope for EAG in rats is in the same amino-terminalthird of �3(IV)NC1 as the human antibody epitope (27) and thatdiscrete segments of �3(IV) NC1 are responsible for EAG in-duction (25). A critical amino acid (AA) sequence within theamino-terminal region consisting of nine or fewer AA conveysnephritogenicity to the non nephritogenic �1(IV)NC1. Diseaseinduced with this construct is associated with severe glomeru-lonephritis with crescents and fibrin with proteinuria but withminimal or absent IgG on the GBM in many animals (25). Thepresence of T cell proliferation in vitro without in vivo antibodybinding suggested that this region might contain a T cellepitope that is responsible for induction of glomerulonephritis.We therefore hypothesized that a peptide construct of thisregion should induce EAG in rats, possibly without antibodyformation. The purpose of these studies was to examine thishypothesis.

Materials and MethodsPreparation of Immunogens

GBM was isolated from glomeruli by differential sieving, sonication,and collagenase solubilization (cs) to form csGBM (12,14). NC1 do-mains were isolated by column chromatography. Recombinant human�3(IV)NC1 was prepared and isolated as described (25,27). This proteininduces EAG in rats and binds Goodpasture’s serum from patients withGoodpasture’s disease. Sequence analysis was performed by liquidchromatography/mass spectroscopy/mass spectroscopy (LC/MS/MS)

Received October 5, 2004. Accepted June 21, 2005.

Published online ahead of print. Publication date available at www.jasn.org.

Address correspondence to: Dr. W. Kline Bolton, P.O. Box 800133, University ofVirginia Health System, Charlottesville, VA 22908-0133. Phone: 434-924-5125; Fax:434-924-5848; E-mail: [email protected]

Copyright © 2005 by the American Society of Nephrology ISSN: 1046-6673/1609-2657

on peptides that were obtained from purified trypsinized recombi-nant human �3(IV)NC1 using a Thermo Electron LCQ mass spec-trometer (28). The published sequences of human and rat �3(IV)NC1were used to synthesize peptides (29 –31). The synthesized peptidesare shown in Figure 1. The design of the peptides is addressed in theResults section.

Experimental Animals and ImmunizationFemale WKY rats, 4 to 6 wk of age, were obtained from Harlan

(Indianapolis, IN). These rats have been inbred for �60 generations.The protocol was approved by the Animal Care and Use Committeeand adhered to the National Institutes of Health Guidelines for the Careand Use of Laboratory Animals. The immunizing protein antigens wereprepared and administered using complete Freund’s adjuvant as de-scribed previously (14,27). Peptides were administered in PBS by thesame protocol. Negative controls were immunized with CFA alone.

Serum Biochemistries, Urinalysis, and Total Urinary ProteinUrinary protein was determined using 3% sulfosalicylic acid (14).

Urine was also examined for hematuria (0 to 3�; Multistix 10 SG, BayerCorporation, Elkhart, IN). Serum creatinine and blood urea nitrogenwere measured using kits from Sigma (St. Louis, MO; Procedure No.555). Urine protein was determined weekly. Urea nitrogen and creati-nine were quantified every 2 wks after immunization.

Immunofluorescence Studies and Histologic ExaminationKidney tissue obtained at death or killing at 12 wk after immuniza-

tion was stained for rat IgG and fibrinogen (31). The intensity of

deposits was graded semiquantitatively in a masked manner from 0 to4� (12,14). As little as 6 femtograms of IgG per glomerular section canbe detected (12). Tissue fixed in Bouin’s solution was stained withhematoxylin-eosin and examined in a masked manner using a four-point scale (31). mAb 17 was used to identify �3(IV)NC1 (7,25,27).

Elution of Antibodies from KidneyKidneys were eluted by the glycine method (32). Eluate protein

concentration was determined by the BCA method (Pierce, Rockford,IL) and by measuring absorbance at 280 nm.

Antibody StudiesSerum from patients with Goodpasture’s syndrome served as a

source of human autoantibodies. Horseradish peroxidase–conjugatedantibodies, fluorescein-conjugated goat anti-rat IgG and fibrinogen,and antihuman IgG were purchased from ICN/Cappel (Irvine, CA).Tandem adsorption of rat anti-p13 antisera was performed using avi-din-biotin and Sepharose columns. Biotinylated p13 was bound to anavidin-biotin adsorption column (NeurtrAvidin Biotin-Binding Proteincolumn; Pierce). p13 was conjugated to cyanogen bromide–activatedSepharose 4B gel (Sigma-Aldrich, St. Louis, MO). Sera from rats withreactivity to p13 and chimeric/recombinant proteins in ELISA wereadsorbed four times on the avidin-biotin column. Because anti-p13activity remained, they were adsorbed twice more on Sepharose. Thewash-through and eluted antibody to p13 was tested in ELISA versuspeptides and proteins.

ELISA assays for antibody to native and recombinant proteins andpeptides were performed as described (6,12,14,33). SDS-PAGE andimmunoblotting were performed in 12.5% gels under nonreducingconditions (25,27,30).

Lymphocyte-Proliferation AssayLymphocytes isolated from lymph nodes 2 mo after immunization

were stimulated in vitro with different antigens and cultured in 96-wellflat-bottom plates (30,31). Data are expressed as the stimulation index,the ratio of stimulated to medium counts per minute. We consideredstimulation indices of 2.0 or greater as significant (31).

Statistical AnalysesData are expressed as mean � SEM. Statistical differences between

groups were evaluated by the t test and ANOVA (12).

ResultsDesign of Peptides

Our studies using chimeric human proteins and point muta-tions within these constructs identified a sequence in the ami-no-terminal portion of �3(IV)NC1 domain responsible for in-duction of EAG (25,27). The initial human peptide that wesynthesized and tested, p21, is shown in Figure 1. A total of 50rats were immunized with various doses of this peptide undervarious conditions. When that peptide failed to induce EAGunder any conditions, we considered that that peptide lengthmight be too long and affect antigen processing. We elected touse a shorter peptide beginning at the same amino terminal siteusing rat rather than human sequence. Because the regioncontaining AA “TAIPS” was critical in nephritogenicity usingchimeric protein mapping (25), we elected to include this regionin the peptide. When the shorter peptide induced EAG, we nextshortened the peptide by one AA in a sequential manner from

Figure 1. Chart showing the peptides used to immunize ani-mals. p21 is a 21–amino acid (AA) synthetic peptide of human�3(IV)NC1 domain spanning AA 14 to 34 with an isoleucine atposition 19, which did not induce experimental autoimmuneglomerulonephritis (EAG) under multiple immunizationschedules, and with various modifications of the peptide. P1,P2, and P3 are peptides used previously to immunize animalsand span the same region of the human sequence (31). Milddisease ensued with some of these peptides. p13, p12a, p11a,p10a, p12c, p11c, and p21-N are various-length peptide se-quences of this same region of �3(IV)NC1 domain with anasparagine substituted for isoleucine at position 19 and are thesubject of this article.

2658 Journal of the American Society of Nephrology J Am Soc Nephrol 16: 2657–2666, 2005

both the amino- (p12a, p11a, and p10a) and carboxy-terminalends (p12c and p11c) and tested the ability of these peptides toinduce EAG. Finally, we went back to our original humanpeptide, p21, confirmed the sequence, and investigated theimpact of converting a single AA from the human to the ratsequence.

Induction of EAGTwenty-one (81%) of 26 animals immunized with p13 rat

sequence developed EAG with a spectrum of disease similar tocsGBM (Figure 2A). Animals with EAG developed proteinuriathat was comparable but of slightly lesser degree than that inanimals immunized with csGBM (Figure 2B), hematuria, andabnormal renal function (Figure 2, C and D). Nephritic rats hadno detectable IgG in 59% of animals that were immunized withp13 (Figure 3). Thirty-two percent of p13-immunized animalshad 1 to 2� linear GBM deposits, and 9% had deposits �2�

intensity, comparable to animals immunized with csGBM. Fi-brinogen was present in glomeruli of all but one rat. Most ratshad fibrinogen of intensity comparable to positive csGBM con-trols.

Lymphocyte Proliferation StudiesLymph node cells from animals immunized with p13 prolif-

erated in response to p13 peptide (Figure 4). p21 inducedlow-grade proliferation, whereas p21-N induced strong prolif-eration. csGBM induced the most marked proliferative re-sponse in p13-immunized rats.

Correlation with IgG Bound to GBM and Histologic ScoreBecause some rats had EAG with GBM-bound IgG but others

did not, we examined the correlation between GBM-bound IgGand the histologic score by hematoxylin and eosin (Figure 5).Fibrinogen deposits were comparable between csGBM-positivecontrols and animals immunized with p13 (Figure 3). However,there was no correlation between GBM-bound IgG and histo-logic score.

Antibody StudiesSerum and Kidney Eluate. p13-immunized rats developed

serum antibody that recognized p13 itself and human�3(IV)NC1 in ELISA (Figure 6). The average amount of anti-body to recombinant human �3(IV)NC1 was low, althoughsome rats had higher levels of reactivity. There was reactivity tohuman �3(IV)NC1 by immunoblot, which was greatly de-creased under reducing conditions (data not shown). There wasno reactivity against rat glomeruli by indirect immunofluores-cence on saline-processed kidney sections. There was, however,antibody binding to human glomerular sections in most ani-mals. Because some animals had linear GBM deposits whereasothers were negative, we pooled negative and positive kidneys,respectively, and eluted glomeruli. Eluates were tested by in-direct fluorescence on normal rat and human kidney processedin saline and after treatment with glycine/urea to expose cryp-tic antigens (34). Serum antibodies were also retested on gly-cine/urea-treated sections. With the use of these methods, nei-ther eluate nor serum of rats that lacked GBM-bound antibody

Figure 2. Histology of rats immunized with collagenase-solubi-lized glomerular basement membrane (csGBM; positive con-trol) and p13 peptide (A) and total urinary protein (B). Rats thatwere immunized with p13 and csGBM had significantly de-creased renal function as measured by urea nitrogen (C) andserum creatinine (D). Histology at death or killing, creatinine,and urea nitrogen at 7 wk.

J Am Soc Nephrol 16: 2657–2666, 2005 Epitope Spreading with T Cell Epitope–Induced GN 2659

fixed to rat GBM. Serum and eluate from rats with GBM-boundIgG, which had absent or barely detectable antibody to GBM onsaline-processed kidney sections, were clearly positive on gly-cine/urea-pretreated human kidney. Kidney eluates but notserum from the GBM-positive rats fixed to rat kidney sections.

Immunosorption Studies. Sera from p13-immunized ratsreactive in ELISA to p13 and human �3(IV)NC1 were subjectedto column immunoadsorption. After multiple adsorptions onp13 columns, antibody activity to domains on human�3(IV)NC1, chimeric proteins of human �3(IV)NC1, and ratNC1 remained despite no anti-p13 activity (Figure 7). Anti-p13eluted antibody reacted only with p13.

Kidney eluates of positive kidneys were likewise subjected to

immunosorption. Anti-p13 adherent eluate did not bind rat orhuman GBM (Figure 8), whereas nonadsorbed kidney eluatefixed in a linear pattern to both rat and human GBM and toessentially all tubular BM in rat sections. The pattern of kidneyeluate binding was the same as the distribution of �3(IV)NC1for both human and rat kidney sections, respectively. Thesedata demonstrate production of antibodies that react with�3(IV)NC1 domains outside the immunizing T cell peptide.These external epitopes were shared between rat and humansequences. Our studies further demonstrate that only antibodyto human GBM was detectible in the circulation by indirectfluorescence despite ELISA reactivity to both human and ratGBM antigens, whereas antibody to both rat and humanepitopes was present in kidney-bound antibodies.

Nephritogenicity of p21 and Amino/Carboxy-TerminalTruncated Peptides

As noted previously, immunization of multiple rats withhuman p21 sequence was unsuccessful in inducing glomerulo-nephritis. Rats did develop high-titer antibody to p21N andp21. Animals immunized with p10a and p11a had no evidenceof EAG. Two of five rats immunized with p12a developed EAG,and all rats immunized with p12c and p11c developed EAG(data not shown).

Importance of Isoleucine to Asparagine Substitution inPosition 19

Even though the human sequence when contained within therecombinant protein, native GBM, and various human chimericconstructs induces EAG (25,27,35), the human peptide se-quence did not. Nonetheless, shorter peptides that containedasparagine at position 19, rat sequence, did induce EAG. There-fore, we re-examined a peptide identical to the original p21 butwith a substitution of asparagine for isoleucine in position 19(p21-N). This is the only position within the nephritogenicamino-terminal region of our recombinant proteins that differsbetween rat and other mammalian species (36,37). This singlesubstitution resulted in conversion of the nonnephritogenic p21peptide to a nephritogenic peptide (Figure 9). All animals de-veloped abnormal urinary protein, with linear GBM deposits ofIgG, fibrinogen deposits, and proliferative EAG by light mi-croscopy. Like p13- and csGBM-immunized animals, there wasno correlation between GBM-bound IgG and histologic score.Mass spectrometric analysis of the recombinant human�3(IV)NC1 confirmed that the AA at position 19 was in factisoleucine, as predicted.

DiscussionPrevious studies identified an amino-terminal immunodom-

inant region of human �3(IV)NC1 responsible for Goodpastureantibody binding and induction of EAG (5,27,38). A segment ofthis region, consistent with a T cell epitope, induced EAG (25).Point mutations within this critical region abrogated antibodybinding and EAG induction. We further explored this regionusing synthetic peptides. Our studies, presented here, demon-strate that a 13-AA peptide rat T cell epitope induces EAG withdevelopment and spreading of antibody response to both rat

Figure 3. Immunofluorescent deposits of IgG and fibrinogen inp13 peptide-immunized rats and csGBM-positive controls. p13-immunized rats had less IgG and similar fibrinogen deposits tocsGBM controls. Negative complete Freund’s adjuvant (CFA)control rats had no immune deposits or histologic abnormali-ties and are not shown.

Figure 4. Antigen-specific lymphocyte proliferation in a p13-immunized animal. CP 333 is recombinant human �3(IV)NC1domain that induces EAG. Scrambled peptide is p13 AA syn-thesized in a random order, negative control.


and human epitopes on �3(IV)NC1 external to the immunizingpeptide. Although a peptide sequence of 13 AA that causes Tcell proliferation might be expected to produce a totally non–IgG-associated disease, nonetheless, 41% of the animals hadlinear IgG on the GBM. The other animals had no IgG presenton the GBM despite the presence of EAG by light microscopy.We do not know why some of these highly inbred rats devel-oped antibodies and others did not. This has been reported byothers and remains to be explained (22). There was no correla-tion between development of EAG and IgG deposits, suggest-ing that CMI induced the EAG in some and perhaps all of theanimals. Support for a role for CMI alone in induction of thedisease derives from a variety of sources that show that T cellsare required for induction of disease, antibody-deficient ani-mals still develop EAG, interruption in T cell activation abro-gates disease, the histologic phenotype is consistent with a CMIresponse, disease may be transferred by cells alone, and inter-vention that blocks CMI blocks development of disease(12,14,18–21,23,26,39–41).

Wu et al. (22,23,42) also demonstrated lack of concordanceamong disease induction, antibody deposition, and EAG. Theyshowed that immunization of rats with peptide of the sameregion as reported here induced EAG (43). In these latter stud-ies, the authors reported that all of their animals developedsevere EAG. Even though they were unable to demonstrate thepresence of circulating antibody to rat GBM or production ofantibody by in vitro techniques, no immunofluorescence find-ings in the kidneys of rats were reported. Thus, it was not clearwhether these animals with peptide-induced EAG also hadantibody along the GBM as observed in some of our animals.They did not use tissue treated to reveal cryptic epitopes,

Figure 5. The relationship between light microscopic (hematoxylin and eosin [H&E]) score and GBM intensity of IgG deposits inp13- and csGBM-immunized animals. There is no correlation between the intensity of GBM-bound IgG and histologic score. Manyanimals had intense glomerulonephritis by light microscopic analysis but minimal or absent IgG deposits.

Figure 6. ELISA of serum and glomerular eluate from ratsimmunized with p13 peptide. Serum (1:200) and eluate (neat)from these rats recognized the immunizing (p13) peptide andhad low activity versus human �3(IV)NC1 protein. Serum fromanimals 7 wk after immunization.

Figure 7. Seven-week serum from a p13 rat with GBM-boundIgG in vivo and circulating antibody to p13 and human�3(IV)NC1. ELISA of serum, six-fold p13 column adsorbedeluted antibody, and nonadsorbed column flowthrough versusp13, recombinant and chimeric human �3(IV)NC1, and ratNC1. Flowthrough nonadsorbed antibody still demonstratesactivity to human �3(IV)NC1, various human chimeric pro-teins, and rat NC1. CP, chimeric proteins consisting of thirds ofhuman �1(IV)NC1 and human �3(IV)NC1 as described (27);rNC1, rat NC1 domain; bGBM, bovine csGBM.


glomerular eluates, or human kidney substrate (43). In subse-quent experiments, they showed that the T cell epitope didindeed induce GBM-bound antibodies in many of their rats(42). Circulating antibody bound only to the immunizing pep-tide, not GBM, but it was not reported whether human kidneysections were used as a substrate. We observed anti-p13 anti-body and low levels of circulating antibody to human�3(IV)NC1 (Figure 6) but could detect minimal or no antibodyto native GBM in tissue sections until the cryptic epitopes wereexposed. Then we could show antibody in the circulation tohuman but not rat GBM in tissue sections, as described previ-ously (25,30). Positive antibody binding to rat antigens byELISA but not by indirect immunofluorescence may relate toconformational versus linear epitopes as postulated by others(42) or to epitopes revealed under the conditions of ELISA butnot by indirect serum binding to cryostat sections.

We expected that such a short AA sequence that induced T

cell proliferation and also EAG would produce antibody-neg-ative EAG. In human glomerulonephritis and in various mod-els of EAG, much evidence points to a purely T cell–mediatedprocess. We were surprised when this short peptide of 13 AAwas still associated with antibody formation in many animals.This suggested that several processes may be ongoing in thepathogenesis of EAG. First, antibodies to p13 might be cross-reactive to GBM epitopes. Antibody from serum and urine ofrats with EAG and mAb derived from rats with EAG induceEAG (16–18,24). Antibody induces proteinuria within hours,with rapid development of glomerulonephritis comparable tocsGBM-induced EAG. However, anti–p13-specific antibody didnot bind to GBM. Second, cells alone can induce EAG duringnative immunization (12,22,25,27) and by transfer of cells fromanimals with EAG to naive recipients (18,23). In this latter case,EAG requires weeks to develop but is not associated with theformation of anti-GBM antibodies (23). Third, T cell–mediated

Figure 8. Indirect immunofluorescence on normal human (A, C, E, and G) and rat (B, D, F, and H) kidney sections. �3(IV)NC1distribution in human kidney is limited to GBM and Bowman’s capsule (arrow), and distal tubule BM (arrowhead; A) as describedpreviously (34) but is present in essentially all BM of rat kidney (B). Kidney eluate fixes in an “�3” pattern characteristic of human(C) and rat (D) �3(IV)NC1. Serum depleted of anti-p13 from p13-immunized rats is positive in an “�3 ” pattern versus humankidney (E) but negative versus rat kidney (F). Negative control serum demonstrates only nonspecific background staining (G, H).Mesangial staining is a normal pattern for rat glomeruli. Tissue that was treated with glycine/urea to reveal cryptic epitopes (34).


tissue damage might result in release of neoantigens and auto-immunization to multiple kidney antigens, as shown by Wu etal. (42). However, we were able to demonstrate only anti-�3(IV)NCI antibodies, not antibodies to diverse GBM antigens.

Our results thus suggest that yet another mechanism may beinvolved: Intramolecular epitope spreading.

We examined the possibility of epitope spreading by frac-tionating serum and kidney eluate into anti-p13 and non–anti-p13 antibody. Repeated adsorption of serum on p13-adsorbentcolumns demonstrated that serum contained antibody to theimmunogen and also additional antibodies to human and rat�3(IV)NC1 epitopes by ELISA. p13 is found only in the amin-oterminus of �3(IV)NC1, yet serum from p13-immunized rats,depleted of anti-p13 antibody, recognized determinants in ratNC1 and other portions of human �3(IV)NC1 and human chi-meric proteins, representing intramolecular epitope spreading.We presume that the reactivity of antibody to both rat andhuman antigen was resultant from cross-reactive antibodies.These findings could occur only by spread of the immuneresponse to antigens external to the p13 immunogen. Thisprovides an explanation for the observation by Wu et al. (42) ofantibody on the GBM after peptide immunization and immu-noprecipitation of multiple GBM antigens by glomerular elu-ates. Their studies were consistent with epitope spreading, butthe precipitated proteins were not identified. It is important tonote that we and Wu et al. both demonstrated antibody activityto both human and rat GBM after immunization with rat pep-tide. Epitope spreading has previously been described in otherexperimental models, including ovaritis, encephalomyelitis,multiple sclerosis, and thyroiditis (44–49). To our knowledge,this is the first documentation of the phenomenon of intramo-lecular epitope spreading in glomerulonephritis. This is illus-trated in the chart in Figure 10. We cannot explain the presenceof circulating antibody to human but not rat GBM by indirectimmunofluorescence. Cross-reactive antibodies were present ineluate by indirect immunofluorescence and by ELISA in serum.It is possible that linear versus conformational epitopes play arole. Circulating antibody might be related to denatured anti-gens, which would not be present in rat kidney eluate, but thenshould react with both human and rat kidney sections. It is alsopossible that there was intermolecular epitope spreading un-detectable in the rat because of the different distribution of�3(IV)NC1 in the rat, i.e., on GBM as well as tubular BM. In thiscase, we might have expected a non-�3(IV)NC1 pattern onhuman kidney (Figure 8E). Different antibody affinities couldalso play a role in being able to detect anti-�3(IV)NC1 antibod-ies on rat versus human kidney sections. Finally, the specificreactivity of circulating antibody in a human �3(IV)NC1 pattern(Figure 8E) but not to rat antigens suggests an alternativeexplanation. We believe that this may demonstrate B cellepitope spreading to recruit B cells that produce antibody spe-cific to epitopes on human rather than rat, �3(IV)NC1.

The fine specificity of the AA sequence in inducing EAG inthis model is notable. Immunization of WKY rats with GBMfrom a variety of species—rat, rabbit, mouse, human, and bo-vine—induces EAG. Furthermore, recombinant human�3(IV)NC1 induces EAG (25,27,35). Nonetheless, when syn-thetic peptides that contained exactly the same AA as full-length protein were used to immunize rats, a T cell proliferativeresponse was induced but no EAG. Although the rat sequencehas been reported to be identical to the human sequence in this

Figure 9. EAG in p21-N–immunized rats. Substitution of aspar-agine for isoleucine in position 19 converted the nonnephrito-genic p21 peptide, which did not induce EAG under any cir-cumstances, to a nephritogenic peptide. All rats that wereimmunized with p21-N developed abnormal urinary protein,decreased kidney function, associated with linear deposits ofIgG along the GBM, intense fibrinogen (FIB) deposits withinglomeruli, and severe glomerulonephritis. Serum creatinine inp21-N versus p21 or CFA (P � 0.0002); blood urea nitrogen inp21-N versus p21 or CFA (P � 0.005). H&E evaluation was notperformed on p21 and CFA rats because of normal kidneyfunction and lack of immunoreactants in tissue sections.


region (36), more recent reports demonstrated asparaginerather than isoleucine at position 19 (37). Thus, the rat sequencefor this area differs from other mammalian species in one singleAA. Substitution of asparagine for isoleucine in position 19induced florid EAG in the context of both a 13-AA syntheticpeptide and a 21-AA peptide. Finally, we confirmed that posi-tion 19 in the human �3(IV)NC1 nephritogenic protein was infact isoleucine, not asparagine. It is not clear why the humansequence in the intact protein induces EAG but as peptide doesnot. It is possible that intracellular protein processing results inmodification of the isoleucine such that it fits into the MHC ofthe antigen-presenting cell to allow T cell receptor recognition.This processing may be absent or modified with peptide, thusrequiring the specificity of the rat asparagine in the syntheticpeptide. Additional study will be needed to clarify how isoleu-cine 19 in one context but not another can induce EAG.

In summary, we have shown that a rat T cell epitope inducesEAG; that for reasons yet to be clarified, isoleucine 19 of�3(IV)NC1 whole protein is permissive to induction of EAG butasparagine 19 is required for peptide induced EAG; and thatintramolecular antibody spreading, possibly with heterologousintramolecular spreading as well, is induced by the T cellepitope. Delineation of peptide epitopes that induce EAG withboth antibody-negative and -positive phenotype and antibodyepitope spreading provides a basis for further understandingthe pathogenesis of glomerulonephritis. The close associationof Goodpasture antibody binding to these same epitopes sug-gests that similar regions may serve as a focus for furtherinsight into the development of the human disease.

AcknowledgmentsThis work was supported by US Public Health Service Grant

DK55801 from the National Institute of Diabetes and Digestive and

Kidney Diseases/National Institutes of Health; grants K2004-71XD-15152 and K2005-73X-09487 from the Swedish Research Council, theTegger Foundation, and the Swedish Society for Medical Research; TheWM Keck Biomedical Mass Spectrometry Laboratory at the Universityof Virginia; and a grant from the University of Virginia Pratt Commit-tee.

This work was presented in part at the annual meeting of the Amer-ican Society of Nephrology, November 12 to 17, 2003, in San Diego, CA(J Am Soc Nephrol 14: 167A, 2003).

We thank Joyce de Guzman for secretarial assistance and Dr. A.Michael (University of Minnesota) for the gift of mAb 17.

References1. Bolton WK: Goodpasture’s syndrome. Nephrology Forum.

Kidney Int 50: 1753–1766, 19962. Lerner RA, Glassock RJ, Dixon FJ: The role of anti-glomer-

ular basement membrane antibody in the pathogenesis ofhuman glomerulonephritis. J Exp Med 126: 989–1004, 1967

3. Wieslander J, Bygren P, Heinegård D: Isolation of thespecific glomerular basement membrane antigen involvedin Goodpasture syndrome. Proc Natl Acad Sci U S A 81:1544–1548, 1984

4. Pusey CD, Dash A, Kershaw MJ, Morgan A, Reilly A, ReesAJ, Lockwood CM: A single autoantigen in Goodpasture’ssyndrome identified by a monoclonal antibody to humanglomerular basement membrane. Lab Invest 56: 23–31, 1987

5. Hellmark T, Segelmark M, Unger C, Burkhardt H, Saus J,Wieslander J: Identification of a clinically relevant immu-nodominant region of collagen IV in Goodpasture disease.Kidney Int 55: 936–944, 1999

6. Hellmark T, Burkhardt H, Wieslander J: Goodpasture dis-ease: Characterization of a single conformational epitopeas the target of pathogenic autoantibodies. J Biol Chem 274:25862–25868, 1999

Figure 10. Epitope spreading. An epitope, p13, induces an immune response to itself and also to other epitopes on the samemolecule, intramolecular spreading. Some antibodies cross-react with epitopes on other species proteins, whereas others arespecific for the species of origin. Intermolecular epitope spreading to involve proteins external to the protein that contains theimmunizing epitope may occur but was not demonstrated in these studies. B cell activation to nonrat epitopes could also occur.


7. Netzer KO, Leinonen A, Boutaud AA, Borza DB, Todd P,Gunwar S, Langeveld JPM, Hudson BG: The Goodpastureantigen: Mapping the major conformational epitope(s) ofalpha3(IV) collagen to residues 17-31 and 127-141 of theNC1 domain. J Biol Chem 274: 11267–11274, 1999

8. Borza DB, Netzer KO, Leinonen A, Todd P, Cervera J, SausJ, Hudson BG: The Goodpasture autoantigen: Identifica-tion of multiple cryptic epitopes on the NC1 domain of thealpha3 (IV) collagen chain. J Biol Chem 275: 6030–6037,2000

9. Ryan J, Mason PJ, Pusey CD, Turner N: Recombinant al-pha-chains of type IV collagen demonstrate that the aminoterminal of the Goodpasture autoantigen is crucial forantibody recognition. Clin Exp Immunol 113: 17–27, 1998

10. Bolton WK: What sensitized cells just might be doing inglomerulonephritis. J Clin Invest 109: 713–714, 2002

11. Steblay RW: Glomerulonephritis induced in sheep by in-jections of heterologous glomerular basement membraneand Freund’s complete adjuvant. J Exp Med 116: 253–271,1962

12. Bolton WK, Tucker FL, Sturgill BC: New avian model ofexperimental glomerulonephritis consistent with media-tion by cellular immunity. J Clin Invest 73: 1263–1276, 1984

13. Sado Y, Okigaki T, Takamiya H, Seno S: Experimentalautoimmune glomerulonephritis with pulmonary hemor-rhage in rats. The dose-effect relationship of the nephrito-genic antigen from bovine glomerular basement mem-brane. J Clin Lab Immunol 15: 199–204, 1984

14. Bolton WK, May WJ, Sturgill BC: Proliferative glomerulo-nephritis in rats: A model for autoimmune glomerulone-phritis in humans. Kidney Int 44: 294–306, 1993

15. Pusey CD, Holland MJ, Cashman SJ, Sinico RA, Lloveras JJ,Evans DJ, Lockwood CM: Experimental autoimmune glo-merulonephritis induced by homologous and isologousglomerular basement membrane in Brown-Norway rats.Nephrol Dial Transplant 6: 457–465, 1991

16. Sado Y, Naito I, Okigaki T: Transfer of anti-glomerularbasement membrane antibody-induced glomerulonephri-tis in inbred rats with isologous antibodies from the urineof nephritic rats. J Pathol 158: 325–332, 1989

17. Sado Y, Kagawa M, Rauf S, Naito I, Moritoh C, Okigaki T:Isologous monoclonal antibodies can induce anti-GBMglomerulonephritis in rats. J Pathol 168: 221–227, 1992

18. Kalluri R, Danoff TM, Okada H, Neilson EG: Susceptibilityto anti-glomerular basement membrane disease and Good-pasture syndrome is linked to MHC class II genes and theemergence of T cell-mediated immunity in mice. J ClinInvest 100: 2263–2275, 1997

19. Bolton WK, Chandra M, Tyson TM, Kirkpatrick PR, Sa-dovnic J, Sturgill BC: Transfer of experimental glomerulo-nephritis in chickens by mononuclear cells. Kidney Int 34:598–610, 1988

20. Bolton WK, Benton FR, Lobo PI: Requirement of functionalT-cells in the production of autoimmune glomerulotubularnephropathy in mice. Clin Exp Immunol 33: 474–477, 1978

21. Reynolds J, Sallie BA, Syrganis C, Pusey CD: The role ofT-helper lymphocytes in priming for experimental autoim-mune glomerulonephritis in the BN rats. J Autoimmun 6:571–585, 1993

22. Wu J, Hicks J, Ou C-N, Singleton D, Borillo J, Lou Y-H:Glomerulonephritis induced by recombinant collagenIValpha3 chain noncollagen domain 1 is not associated

with glomerular basement membrane antibody: A poten-tial T cell-mediated mechanism. J Immunol 167: 2388–2395,2001

23. Wu J, Hicks J, Borillo J, Glass WF III, Lou Y-H: CD4� Tcells specific to a glomerular basement membrane antigenmediate glomerulonephritis. J Clin Invest 109: 517–524,2002

24. Kohda T, Okada S, Hayashi A, Kanzaki S, Ninomiya Y,Taki M, Sado Y: High nephritogenicity of monoclonal an-tibodies belonging to IgG2a and IgG2b subclasses in ratanti-GBM nephritis. Kidney Int 66: 177–186, 2004

25. Hellmark T, Chen L, Ohlsson S, Wieslander J, Bolton WK:Point mutations of single amino acids abolish ability ofalpha3 NC1 domain to elicit experimental autoimmuneglomerulonephritis in rats. J Biol Chem 278: 46516–46522,2003

26. Dean EG, Wilson GR, Li M, Edgtton KL, O’Sullivan KM,Hudson BG, Holdsworth SR, Kitching AR: Experimentalautoimmune Goodpasture’s disease: A pathogenetic rolefor both effector cells and antibody in injury. Kidney Int 67:566–575, 2005

27. Chen L, Hellmark T, Wieslander J, Bolton WK: Immuno-dominant epitopes of alpha3(IV)NC1 induce autoimmuneglomerulonephritis in rats. Kidney Int 64: 2108–2120, 2003

28. Bogdanov B, Smith RD: Proteomics by FTICR mass spec-trometry: Top down and bottom up. Mass Spectrom Rev 24:168–200, 2005

29. Turner N, Mason PJ, Brown R, Fox M, Povey S, Rees AJ,Pusey CD: Molecular cloning of the human Goodpastureantigen demonstrates it to be the alpha3 chain of the typeIV collagen. J Clin Invest 89: 592–601, 1992

30. Bolton WK, Luo AM, Fox P, May WJ, Fox J: Goodpasture’sepitope in the development of experimental autoimmuneglomerulonephritis in rats. Kidney Int 49: 327–334, 1996

31. Luo A-M, Fox J, Chen L, Bolton WK: Synthetic peptides ofGoodpasture’s antigen in antiglomerular basement mem-brane nephritis in rats. J Lab Clin Med 139: 303–310, 2002

32. Saxena R, Bygren P, Butkowski RJ, Wieslander J: Specific-ity of kidney-bound antibodies in Goodpasture’s syn-drome. Clin Exp Immunol 78: 31–36, 1989

33. Gunnarsson A, Hellmark T, Wieslander J: Molecular prop-erties of the Goodpasture epitope. J Biol Chem 275: 30844–30848, 2000

34. Kleppel MM, Santi PA, Cameron JD, Wieslander J, MichaelAF: Human tissue distribution of novel basement mem-brane collagen. Am J Pathol 134: 813–825, 1989

35. Sado Y, Boutaud AA, Kagawa M, Naito I, Ninomiya Y,Hudson BG: Induction of anti-GBM nephritis in rats byrecombinant alpha3(IV) NC1 and alpha4(IV) NC1 of typeIV collagen. Kidney Int 53: 664–671, 1998

36. Ryan J, Katbamna I, Mason PJ, Pusey CD, Turner AN:Sequence analysis of the “Goodpasture antigen” of mam-mals. Nephrol Dial Transplant 13: 602–607, 1998

37. Ryan J, Reynolds J, Norgan VA, Pusey CD: Expression andcharacterization of recombinant rat alpha3(IV)NC1 and itsuse in induction of experimental autoimmune glomerulo-nephritis. Nephrol Dial Transplant 16: 253–261, 2001

38. Hellmark T, Johansson C, Wieslander J: Characterization ofanti-GBM antibodies involved in Goodpasture’s syn-drome. Kidney Int 46: 823–829, 1994

39. Reynolds J, Tam FWK, Chandraker A, Smith J, Karkar AM,Cross J, Peach R, Sayegh MH, Pusey CD: CD28–B7 block-


ade prevents the development of experimental autoim-mune glomerulonephritis. J Clin Invest 105: 643–651, 2000

40. Reynolds J, Cashman SJ, Evans DJ, Pusey CD: CyclosporinA in the prevention and treatment of experimental auto-immune glomerulonephritis in the brown Norway rat. ClinExp Immunol 85: 28–32, 1991

41. Reynolds J, Khan SB, Allen AR, Benjamin CD, Pusey CD:Blockade of the CD154-CD40 costimulatory pathway pre-vents the development of experimental autoimmune glo-merulonephritis. Kidney Int 66: 1444–1452, 2004

42. Wu J, Arends J, Borillo J, Zhou C, Merszei J, McMahon J,Lou YH: A self T cell epitope induces autoantibody re-sponse: Mechanism for production of antibodies to diverseglomerular basement membrane antigens. J Immunol 172:4567–4574, 2004

43. Wu J, Borillo J, Glass II WF, Hicks J, Ou C-N, Lou Y-H:T-cell epitope of alpha3 chain of type IV collagen inducessevere glomerulonephritis. Kidney Int 64: 1292–1301, 2003

44. Tung KS: Mechanism of self-tolerance and events leading

to autoimmune disease and autoantibody response. ClinImmunol Immunopathol 73: 275–282, 1994

45. Steinman L: Escape of “horror autotoxicus”: Pathogenesisand treatment of autoimmune disease. Cell 80: 7–10, 1995

46. Thrasyvoulides A, Lymberi P: Evidence for intramolecularB-cell epitope spreading during experimental immuniza-tion with an immunogenic thyroglobulin peptide. Clin ExpImmunol 132: 401–407, 2003

47. Kumar V: Determinant spreading during experimental au-toimmune encephalomyelitis: Is it potentiating, protectingor participating in the disease? Immunol Rev 164: 73–80,1998

48. Tuohy VK, Yu M, Yin L, Kawczak JA, Johnson JM, Mathi-sen PM, Weinstock-Guttman B, Kinkel RP: The epitopespreading cascade during progression of experimental au-toimmune encephalomyelitis and multiple sclerosis. Immu-nol Rev 164: 93–100, 1998

49. James JA, Harley JB: B-cell epitope spreading in autoim-munity. Immunol Rev 164: 185–200, 1998


Epitope Spreading and Autoimmune Glomerulonephritis in Rats Induced by a T Cell Epitope of Goodpasture's Antigen

Documents