Nephrol Dial Transplant (2000) 15: 28–33Nephrology

DialysisTransplantationOriginal Article

Humoral immunity against the proline-rich peptide epitope of the IgA1hinge region in IgA nephropathy

Tohru Kokubo, Kenjiro Hashizume, Hitoo Iwase1, Kenji Arai2, Atsushi Tanaka2, Kazunori Toma2,Kyoko Hotta1 and Yutaka Kobayashi

Department of Medicine and 1Department of Biochemistry, School of Medicine, Kitasato University, Sagamihara-City and2Analytical Research Center, Asahi Chemical Industry Co., Ltd, Fuji-City, Japan

Abstract Key words: IgA nephropathy; humoral immunity; IgA1hinge region; proline-rich peptide; O-glycosylationBackground. The human IgA1 hinge region is a unique

mucin-like O-linked proline-rich glycopeptide, and itscore peptide was found to be exposed aberrantly bythe underglycosylation in IgA nephropathy (IgAN ). IntroductionWe describe here the presence of humoral immunityagainst the IgA1 hinge peptide epitope in IgAN and

IgA nephropathy (IgAN) is the most common glomer-evaluate the relationship between the underglycosyl-ular disease which is characterized immunohistolog-ation of the IgA1 hinge region and humoral immunity.ically by the predominant deposition of the IgA1Method. The serum anti-IgA1 hinge peptide antibodysubclass in the mesangial area [1,2]. The structural(anti-a1HP ab) titre was measured and comparedcharacteristics of the IgA1 molecule recently werebetween the IgAN (n=37) and control groups (n=34)associated with the pathogenesis of IgAN, because theby enzyme-linked immunosorbent assay (ELISA) IgA1 molecule is obviously different from the otherusing a synthetic peptide corresponding to the human immunoglobulins in its hinge region which is a uniqueIgA1 hinge region, PVPSTPPTPSPSTPPTPSPS, as an mucin-like glycopeptide. The IgA1 hinge region has aantigen. Next, to evaluate the relationship between the proline-, serine- and threonine-rich amino acid

underglycosylation of the IgA1 hinge region and the sequence in which the serine and threonine residueshumoral immunity, the reactivity of the serum IgG are able to carry O-linked oligosaccharides consistingfrom the patients with IgAN against monoclonal IgA1 of sialic acid (NeuNAc), galactose (Gal ) and N-which had been digested enzymatically to remove the acetylgalactosamine (GalNAc) with microhetero-carbohydrates from the IgA1 hinge region was meas- geneity [3,4]. Some investigators, including us, revealedured by ELISA. the presence of defective O-glycans in the hinge regionResults. The anti-a1HP ab titre was significantly higher of the IgA1 molecules derived from sera of patientsin the IgAN group than in the control group (OD with IgAN [5–9]. Additionally, we showed that thevalue: IgG class, 0.564±0.344 vs 0.331±0.154, P= core peptide of the IgA1 hinge region was exposed by0.0014; IgM class, 0.272±0.148 vs 0.141±0.072, its underglycosylation in IgAN using a rabbit poly-P<0.0001) and it was positive in ~40% of the patients clonal antibody against a synthetic peptide correspond-with IgAN. In addition, the reactivity of the serum ing to the human IgA1 hinge region [10].IgG from the IgAN patients against the monoclonal Regarding the biological significance of O-glycans,IgA1 was found to be increased as the carbohydrates very interesting findings have been reported in studieswere enzymatically removed from the IgA1 hinge of mucin. In patients with several different kinds ofregion (when native=100; asialo, 122±9.5; agalacto, cancer, it was shown that the core peptide of mucin167±11.5; naked, 188±3.9). expressed on the malignant cells was exposed by theConclusion. These results suggested that the peptide underglycosylation, resulting in humoral and cellularepitope of the IgA1 hinge region which was aberrantly immune responses to its peptide epitope [11–13]. Sinceexposed by underglycosylation could induce the hum- the IgA1 hinge region is very similar to mucin inoral immune response in IgAN. proline residue abundance and the presence of O-

glycans [14], the IgA1 hinge peptide core exposed byunderglycosylation may induce some immune responsein patients with IgAN.Correspondence and offprint requests to: Tohru Kokubo MD,

In this study, we examined the antibody productionDepartment of Medicine, School of Medicine, Kitasato University,1-15-1 Kitasato, Sagamihara, Kanagawa 228-8555, Japan. against the peptide epitope of the IgA1 hinge region

© 2000 European Renal Association–European Dialysis and Transplant Association

Anti-IgA1 hinge peptide antibody in IgA nephropathy 29

Mann–Whitney U-test. A P-value of <0.05 was consideredin patients with IgAN by enzyme-linked immunosorb-to be a significant difference.ent assay (ELISA) using a synthetic peptide corres-

Preliminarily, we examined the specific reactivity of theponding to the human IgA1 hinge region as an antigen,IgG from patients with IgAN against the sa1HP. Theand evaluated the relationship between the underglyco-dose-dependent reactivity of the IgG (serum dilution,sylation of the IgA1 hinge region and the humoral 1/1600–1/100) against the plate-coated sa1HP was

immune response. tested using the sera from patients with IgAN (n=6)and controls (n=6) with the above-mentioned method.Each mean±SD of the six cases was expressed, and thestatistical difference between IgAN and the control was

Materials and methods analysed by ANOVA. In addition, the inhibition assays wereperformed using a free sa1HP and an unrelated 21mersynthetic peptide, human b-endorphin (1–5)+(16–31),Sera were collected from 37 patients (17 males, 20 females,YGGFMTLFKNAIIKNAYKKGE (American Peptidemean age 37 years, range 16–56 years) with biopsy-provenCompany, Sunnyvale, CA). Fifty ml of the sera (1/50 dilution)IgAN and 34 control patients (22 males, 12 females, meanfrom the patients with IgAN (n=6) were added with 50 mlage 43 years, range 17–74 years) with other biopsy-provenof the free synthetic peptides (0–20 mg/ml ) to the sa1HP-renal diseases without glomerular IgA deposition (membran-coated wells. After incubation for 3 h and washing, the IgGous nephropathy, n=6; non-IgA mesangial proliferativebinding to the plate-fixed sa1HP was detected with peroxid-glomerulonephritis (GN), n=4, membranoproliferative GN,ase-conjugated anti-human IgG antibody in the same way.n=4; tubulointerstitial nephritis, n=4; focal glomerulo-The reactivity without inhibitors was regarded as 100%, andsclerosis n=3; lupus nephritis, n=3; lipoid nephrosis, n=2;each mean±SD of the percentage reactivity was expressed.nephrosclerosis, n=2; rapidly progressive GN, n=2; amilo-

In order to evaluate the relationship between the undergly-idosis, n=2; acute GN, n=1; diabetic nephropathy, n=1).cosylation of the IgA1 hinge region and the humoral immuneA 21mer peptide corresponding to the amino acidresponse, the reactivity of the IgG from patients with IgANsequence of the human IgA1 hinge region,against a monoclonal IgA1 which was digested enzymaticallyPVPSTPPTPSPSTPPTPSPSC, was synthesized by and pur-to remove the carbohydrates from the IgA1 hinge regionchased from Bio-Synthesis, Inc. (Lewisville, TX ). The puritywas measured by ELISA. Neuraminidase (NA) fromand molecular weight were confirmed by HPLC and MALDI-Streptococcus 6646K (1 U/100 ml; Seikagaku Co. Tokyo,TOFMS. The synthetic IgA1 hinge peptide (sa1HP) wasJapan), b-galactosidase (GA) from bovine testis (1.0 U/ml;conjugated with bovine serum albumin (BSA, SigmaSigma Chemical Co.) and a-N-acetylgalactosaminidaseChemical Co., St Louis, MO) for plate coating. The conjuga-(GNA) from Acremonium sp. (13 U/ml; Seikagaku Co.) weretion of BSA with the peptide was performed by the followingused in this examination according to the following procedureprocedure. A 1.5 ml aliquot of BSA (13.4 mg/ml ) wasbased on our previous study [15]. The monoclonal IgA1mixed with 100 ml of N-e-maleimidocaproyloxy succinimidederived from a patient with IgA1-type multiple myeloma(100 mg/ml, Wako Junyaku Co., Tokyo) and incubated for(Funakoshi Co. Tokyo, Japan) was divided into four samples1 h at room temperature. After desalting, 250 ml of the(1 mg each). The first sample was dissolved in 100 ml ofpeptide (20 mg/ml ) was added and incubated for 36 h at0.2 M sodium acetate buffer, pH 5.0 (SAB) without adding4°C. After desalting, the BSA-conjugated peptide was usedany enzymes and was used as a control. The second sampleas the antigen for ELISA.was dissolved in 95 ml of SAB and 5 ml of NA was added toThe 96-well microtitration plates (Linbro/Titertek, Flowremove NeuNAc. The third sample was dissolved in 75 ml ofGeneral Company, McLean, IL) were coated with 100 ml ofSAB and 5 ml of NA and 20 ml of GA were added to removethe BSA-conjugated sa1HP diluted to a concentration ofNeuNAc and Gal. The fourth sample was dissolved in 65 ml10 mg/ml with 0.015 M carbonate buffer, pH 9.6. Afterof SAB and 5 ml of NA, 20 ml of GA and 10 ml of GNA wereblocking of the unreacted sites with 0.01 M phosphate buffer,added to remove NeuNAc, Gal and GalNAc. All the IgA1pH 7.5, 0.15 M NaCl (PBS) containing 1% BSA, 100 ml ofsamples containing the enzymes were incubated overnight atsera (1/100 dilution in PBS) from patients with IgAN (n=37°C and then dialysed against PBS. The non-treated,37) and controls (n=34) were added to the sa1HP-coatedNA-treated, NA+GA-treated and NA+GA+GNA-treatedand non-coated wells. After incubation for 3 h at roomIgA1 were named ‘native’, ‘asialo’, ‘agalacto’ and ‘naked’temperature and washing with PBS containing 0.1% BSAIgA1, respectively. The efficacy of the enzymatic treatmentsand 0.05% Tween-80 (PBS/BSA/Tween), 100 ml of peroxi-was confirmed by binding assays with peanut agglutinin anddase-conjugated goat anti-human IgG (1/500, Fc portion-Vicia vilosa lectins according to the previously describedspecific, Organon Teknika Corp., West Chester, PA), anti-method [16 ]. The microtitration plates were coated withhuman IgA (1/500, a-chain-specific, Organon Teknika Corp.)100 ml of the IgA1 (10 mg/ml ). After blocking with 1% BSA,and anti-human IgM (1/500, m-chain-specific, Organon100 ml of the sera (1/100 dilution) from the patients withTeknika Corp.) antibodies were added to the wells. AfterIgAN (n=6) was reacted against the coated IgA1. Becauseincubation for 1 h at room temperature and washing withof the non-specificity of NA+GA to O-glycans, we alsoPBS/BSA/Tween, the plates were exposed to an enzymeperformed the same assays using a monoclonal IgA2substrate consisting of 0.4 mg/ml of o-phenylenediamine(Funakoshi Co., Tokyo, Japan). The IgG reactivity wasdihydrochloride (Sigma Chemical Co.) and 0.03% hydrogendetected with peroxidase-conjugated anti-human IgG anti-peroxide in a solution containing 0.1 M disodium hydrogenbody using the above-mentioned ELISA method. The ODphosphate and 0.05 M citric acid monohydrate. Thevalue in the native IgA1 was regarded as 100% and eachdeveloped colour was read at 490 nm with a microplatemean±SD of the percentage reactivity in the deglycosylatedreader (Bio-Rad Laboratories, model 450, Richmond, CA).IgA1 was expressed.Each absorbance level was reduced by that of the non-

To examine the exposure of the IgA1 hinge peptide coresa1HP-coated well. All of the assays were performed inin IgAN, the reactivity of a rabbit polyclonal anti-synthetictriplicate. Each median was expressed. The statistical differ-

ence between the two groups was analysed using the IgA1 hinge peptide antibody (anti-sa1HP ab), which was

T. Kokubo et al.30

reported in our previous study [10], against the serum IgAwas measured by ELISA and compared in the IgAN andcontrol groups. The microtitration plates were coated with100 ml of goat anti-human IgA antibody (10 mg/ml, Fabfraction, Organon Teknika Corp). After blocking with BSA,100 ml of the sera (1/100 dilution) from the IgAN group (n=24) and control group (n=24) were added to the wells. Afterincubation for 3 h at room temperature and washing, 100 mlof the rabbit anti-sa1HP ab (1/100 dilution) was added tothe wells. After incubation for 3 h at room temperature andwashing, 100 ml of the peroxidase-conjugated goat anti-rabbitIgG antibody (1/500 in PBS, Organon Teknika Corp.) wasplaced in the wells. After incubation for 1 h at room temper-ature and washing, the absorbance levels were measuredusing the above-mentioned method. These assays were per-formed in triplicate. Each median was expressed. The statist-ical difference between the two groups was analysed usingthe Mann–Whitney U test.

Fig. 2. Reactivity of serum IgG from patients with IgAN (n=6)against plate-fixed sa1HP was inhibited by the free sa1HP but not

Results by the control peptide.

Specific reactivity of serum IgG from IgAN patientsagainst sa1HP

As shown in Figure 1, the serum IgG from the patientswith IgAN reacted against the plate-fixed sa1HP in adose-dependent manner. The reactivity was signific-antly higher in the patients with IgAN (n=6) than inthe controls (n=6) (ANOVA, P=0.017). In addition,the binding of the IgG from the patients with IgAN(n=6) to the plate-fixed sa1HP was clearly inhibitedby the free sa1HP but not by a control peptide(Figure 2).

Serum anti-a1HP ab titres in IgAN

As shown in Figure 3, the serum anti-a1HP ab titresof the IgG and IgM classes were significantly higherin the IgAN group (n=37) than in the control group(n=34) (IgG class, 0.564±0.344 vs 0.331±0.154, P=0.0014; IgM class, 0.272±0.148 vs 0.141±0.072,P<0.0001). However, the antibody titre of the IgA

Fig. 3. Serum anti-a1HP ab titres of IgG and IgM classes weresignificantly higher in the IgAN group (n=37) than in the controls(n=34). The titres were positive in ~40% of patients with IgAN(positive level>mean+2 SD of controls).

class was not detectable in all cases of both groups.When more than the mean+2 SDs of the controlgroup was regarded as a positive level of the anti-a1HP ab titres (IgG class, >0.639; IgM class, >0.285),the positive rate was 40.5% (15/37) in the IgG classand 43.2% (16/37) in the IgM class in the IgAN group,and 0% (0/34) in both antibody classes in the controlgroup. There is a significant correlation between theFig. 1. Serum IgG from patients with IgAN (n=6) reacted againstanti-a1HP ab titres of the IgG and IgM classes (R2=sa1HP in a dose-dependent manner. The reactivity was significantly

higher in IgAN patients than in controls (n=6). 0.366, P<0.0001). However, no significant correlation

Anti-IgA1 hinge peptide antibody in IgA nephropathy 31

was found between the anti-a1HP ab titres and theserum concentration of IgG or IgM.

Reactivity of serum IgG from IgAN patients againstdeglycosylated IgA1 (monoclonal)

The serum IgG from patients with IgAN (n=6) reactedstrongly against the deglycosylated IgA1 comparedwith native IgA1. When the reactivity against nativeIgA1 was regarded as 100%, the values were 122±9.5%(mean±SD) against the asialo IgA1, 167±11.5%against the agalacto IgA1 and 188±3.9% against thenaked IgA1. The reactivity was increased as the carbo-hydrates were removed from the IgA1 hinge region,and the reactivity against naked IgA1 was ~2-foldhigher than that against native IgA1 (Figure 4). Inaddition, there was no significant difference in thereactivity of the patient’s IgG against a monoclonalIgA2 between native and deglycosylated IgA2.

Fig. 5. Reactivity of rabbit polyclonal anti-sa1HP ab against serumIgA was significantly higher in the IgAN group (n=24) than in theReactivity of rabbit polyclonal anti-sa1HP ab againstcontrols (n=24).

serum IgA in IgAN

As shown in Figure 5, the reactivity of the rabbit and it was regarded as positive in ~40% of the patientspolyclonal anti-sa1HP antibody against serum IgA with IgAN. It was also observed that this titre waswas significantly higher in the IgAN group (n=24) increased in both the IgG and IgM classes but notthan in the control group (n=24) (0.321±0.167 vs detectable in the IgA class, indicating that antibody0.189±0.137; P=0.0042). There was no significant production against the peptide epitope of the IgA1correlation between this reactivity and the serum IgA hinge region occurred in the IgG and IgM classes.level. These observations suggested that these humoral

immune responses could result in the formation ofIgA1–IgG and IgA1–IgM immune complexes in the

Discussion circulation of patients with IgAN. Czerkinsky et al.previously reported that IgA-containing circulating

The results of the ELISA using a synthetic peptide immune complexes were found in patients with IgANcorresponding to the human IgA1 hinge region, [17]. The IgA1–IgG and IgA1–IgM interactions medi-PVPSTPPTPSPSTPPTPSPS, as an antigen showed ated by the IgA1 hinge peptide observed in the presentthat the serum anti-a1HP ab titre was significantly study may be one of the mechanisms involved in thehigher in the IgAN group than in the control group, formation of IgA-containing immune complexes in

IgAN.Next, we examined the reactivity of serum IgG from

the patients with IgAN against native and deglycosyl-ated monoclonal IgA1. This reactivity was found tobe greatest in the naked, followed by the agalacto,asialo and native IgA1, in that order; i.e. the morecarbohydrates that were removed from the IgA1 hingeregion, the greater the reactivity. This result suggeststhat the accessibility of the anti-a1HP ab to the peptideepitope of the IgA1 hinge region was increased by theenzymatic removal of the carbohydrates from the IgA1hinge region, and that the carbohydrate moietiesincluding NeuNAc, Gal and GalNAc attached to thehinge region function to protect the peptide epitope.Therefore, if there is some defect in the O-glycans ofthe IgA1 hinge region, the probability of the peptideepitope being recognized by the anti-a1HP ab expressedon B lymphocytes is thought to increase and, con-sequently, the B cells will be activated and produce theanti-a1HP ab.Fig. 4. Reactivity of serum IgG from patients with IgAN (n=6)

Furthermore, we also showed that the reactivity ofagainst a monoclonal IgA1 was increased as carbohydrates wereremoved from the IgA1 hinge region. a rabbit polyclonal anti-sa1HP ab against serum IgA

T. Kokubo et al.32

was significantly higher in the IgAN group than in the mucin. Therefore, if the immune response to the pep-tide epitope of the IgA1 hinge region observed in thecontrol group, suggesting that the peptide core of the

IgA1 hinge region is aberrantly exposed by the defect- present study also functions to eliminate the undergly-cosylated IgA1 molecules from the circulation ofive O-glycosylation in IgAN. This result also supports

the hypothesis that the underglycosylation of the IgA1 patients with IgAN, then synthetic peptides corres-ponding to the IgA1 hinge region may be useful forhinge region causes the humoral immune response to

its peptide epitope in IgAN. Tomana et al. recently an ‘immunotherapy’ against IgAN. It was suggestedrecently that IgA1 molecules which have defective O-reported that the Gal deficiency in IgA1 molecules was

related to the formation of IgA1–IgG complexes in glycans in their hinge region could cause direct glomer-ular deposition by non-immunological mechanismsIgAN [8]. Our data are consistent with theirs in terms

of the relationship between the formation of the [5–9,16,29–34]. However, there is also the possibilitythat the immune response to the IgA1 hinge peptideIgA1–IgG immune complex and the aberrant glycosyl-

ation of the IgA1 hinge region. plays a pathogenetic role in IgAN as a result of theIgA1–IgG immune complex formation or otherConcerning the structure and function of the O-

linked glycopeptides, the study of mucin, a typical O- immune mechanisms. Therefore, it is necessary toevaluate carefully the relationship between the anti-linked glycoprotein, has been well developed in the

field of cancer. Kotera et al. showed the presence of a1HP antibody titre and the clinical course of manypatients with IgAN and to reveal the pathophysiolog-humoral immunity against a tandem repeat peptide

epitope of the human mucin MUC-1 in sera from ical implication of this antibody production in IgAN.We previously indicated that the underglycosylationbreast, pancreatic and colon cancer patients [11]. In

that study, it was shown that the anti-MUC-1 antibody of the IgA1 hinge region caused the non-immunologicalIgA1 self-aggregation and adhesion to extracellulartitre was positive in 8–17% of these cancer patients by

ELISA using a synthetic peptide corresponding to the matrix (ECM) proteins such as type IV collagen,fibronectin and laminin, and proposed that these phen-MUC-1, PDTRPAPGSTAPPAHGVTSA. Also, anti-

MUC-1 antibody production was suggested to result omena might be one of the mechanisms of the glomer-ular deposition of IgA1 in IgAN [16 ]. We also inferredfrom the exposure of the MUC-1 core peptide by

underglycosylation of the mucin expressed on the that the IgA1 self-aggregation and adhesion to ECMproteins could result from the exposure of the proline-malignant cells [18,19]. Since the IgA1 hinge region is

very similar to mucin in terms of the abundance of rich peptide core of the IgA1 hinge region due to thedefect of O-linked sugars in IgAN [10,32], because, asproline residues and the presence of O-glycans, it is

likely that the IgA1 hinge region has a structure and Williamson described, the proline-rich peptide canmediate various molecular interactions by non-specificfunction similar to mucin [14,20–22].

In addition, Rughetti et al. showed that a synthetic physicochemical reactions such as hydrogen bondingand/or hydrophobic interaction [21]. However, in thispeptide corresponding to MUC-1 could induce a B-cell

immune response in an ovarian cancer patient [12]. study, we showed that the exposure of the proline-richIgA1 hinge peptide by underglycosylation could causeJerome et al. reported that cytotoxic T lymphocytes

derived from patients with breast adenocarcinoma not only non-immunological, but also immunologicalIgA1-containing protein–protein interactions in IgAN.recognized a peptide core epitope of mucin preferen-

tially expressed by the malignant cells [13]. These In conclusion, it was shown that the serum anti-a1HP ab titre was significantly higher in the IgANfindings indicated that the mucin peptide core epitope

could induce cellular immunity as well as humoral group than in the control group and it was positive in~40% of the patients with IgAN. The reactivity of theimmunity in humans. Therefore, the production of

antibody against the peptide core epitope of the IgA1 serum IgG from the IgAN patients against a mono-clonal IgA1 was increased as the carbohydrates werehinge region may provide a possible explanation not

only for the formation of the IgA1-containing immune enzymatically removed from the IgA1 hinge region.These results suggested that the peptide epitope of thecomplexes, but also for the abnormalities in the activity

of the cellular immune system observed in IgAN [23]. IgA1 hinge region which was aberrantly exposed bythe underglycosylation could induce the humoralRecent studies in this field provided further interes-

ting information. Agrawal et al. described that T cells immune response in IgAN.from multiparous women specifically proliferated in

Acknowledgements. The authors thank Ms M. Saitoh for her excellentresponse to the MUC-1 peptide epitope, indicatingtechnical assistance. Part of this study was supported by the Newthat there is a natural immunization against it duringEnergy and Industrial Technology Development Organization.pregnancy [24,25]. These observations provided an

explanation for a negative correlation with the occur-rence of breast cancer in maltiparous women [26 ] and Referencesled to the potential use of synthetic peptides corres-

1. Berger J. IgA glomerular deposits in renal disease. Transplantponding to MUC-1 as ‘vaccines’ for therapy of breastProc 1969; 1: 939–944cancer [27,28]. Thus, it is clear that the immune

2. Conley ME, Cooper MD, Michael AF. Selective deposition ofresponses to the MUC-1 peptide epitope serve as a immunoglobulin A1 in immunoglobulin A nephropathy, ana-preventive factor against breast cancer by eliminating phylactoid purpura nephritis and systemic lupus erythematosus.

J Clin Invest 1980; 66: 1432–1436the cancer cells expressing the underglycosylated

Anti-IgA1 hinge peptide antibody in IgA nephropathy 33

3. Kerr MA. The structure and function of human IgA. Biochem 18. Burchel J, Gendler S, Taylor-Papadimitriou J et al. DevelopmentJ 1990; 271: 285–296 and characterization of breast cancer reactive monoclonal anti-

4. Mattu TS, Pleass RJ, Willis AC et al. The glycosylation and bodies directed to the core protein of the human milk mucin.structure of human serum IgA1, Fab, and Fc regions and the Cancer Res 1987; 47: 5476–5482role of N-glycosylation on Fca receptor interactions. J Biol 19. Brockhausen I, Yang J, Burchell J, Whitehouse C, Taylor-Chem 1998; 273: 2260–2272 Papadimitriou J. Mechanisms underlying aberrant glycosylation

5. Hiki Y, Iwase H, Saitoh M et al. Reactivity of glomerular and of MUC1 mucin in breast cancer cells. Eur J Biochem 1995;serum IgA1 to jacalin in IgA nephropathy. Nephron 1996; 233: 607–61772: 429–435 20. Jentoft N. Why are proteins O-glycosylated? Trends Biochem

6. Hiki Y, Iwase H, Kokubo T et al. Association of asialo- Sci 1990; 15: 291–294galactosylb1-3N-acetylgalactosamine on the hinge with a con- 21. Williamson MP. The structure and function of proline-richformational instability of jacalin-reactive immunoglobulin A1 in regions in proteins. Biochem J 1994; 297: 249–260immunoglobulin A nephropathy. J Am Soc Nephrol 1996; 7: 22. MacArthur MW, Thornton JM. Influence of proline residues955–960 on protein conformation. J Mol Biol 1991; 218: 397–412

7. Hiki Y, Tanaka A, Kokubo T et al. Analyses of IgA1 hinge 23. Galla JH. IgA nephropathy. Kidney Int 1995; 47: 377–387glycopeptides in IgA nephropathy by matrix-assisted laser 24. Agrawal B, Reddish MA, Krants MJ, Longenecker BM. Doesdesorption/ionization time-of-flight mass spectrometry. J Am pregnancy immunize against breast cancer? Cancer Res 1995;Soc Nephrol 1998; 9: 577–582 55: 2257–2261

8. Tomana M, Matousovic K, Julian B, Radl J, Konecny K, 25. Agrawal B, Reddish MA, Longenecker BM. In vitro inductionMestecky J. Galactose-deficient IgA1 in sera of IgA nephropathy of MUC-1 peptide-specific type 1 T lymphocyte and cytotoxicpatients is present in complex with IgG. Kidney Int 1997; T lymphocyte responses from healthy multiparous donors.52: 509–516 J Immunol 1996; 157: 2089–2095

9. Allen AC, Harper SJ, Feehally J. Galactosylation of N-and O- 26. Kalache A, Maguire A, Thompson SG. Age at last full-termlinked carbohydrate moieties of IgA1 and IgG in IgA nephro- pregnancy and risk of breast cancer. Lancet 1993; 341: 33–36pathy. Clin Exp Immunol 1995; 100: 470–474 27. Apostolopoulos V, Xing P, McKenzie IFC. Murine immune

10. Kokubo T, Hiki Y, Iwase H et al. Exposed peptide core of IgA1 response to cells transfected with human MUC1: immunizationhinge region in IgA nephropathy. Nephrol Dial Transplant 1999; with cellular and synthetic antigens. Cancer Res 1994; 54:14: 81–85 5186–5193

11. Kotera Y, Fontenot JD, Pecher G, Metzgar RS, Finn 28. Karanikas V, Hwang L, Pearson J et al. Antibody and T cellOJ. Humoral immunity against a tandem repeat epitope of responses of patients with adenocarcinoma immunized withhuman mucin MUC-1 in sera from breast, pancreatic, and colon mannan–MUC1 fusion protein. J Clin Invest 1997; 100:cancer patients. Cancer Res 1994; 54: 2856–2860

2782–279212. Rughetti A, Turchi V, Ghetti CA et al. Human B-cell immune29. Hiki Y, Kokubo T, Iwase H et al. Under-glycosylation of IgA1response to the polymorphic epithelial mucin. Cancer Res 1993;

hinge plays a certain role for its glomerular deposition in IgA53: 2457–2459nephropathy. J Am Soc Nephrol 1999; in press13. Jerome KR, Barnd DL, Bendt KM et al. Cytotoxic

30. Iwase H, Tanaka A, Hiki Y, Kokubo T, Ishii-Karakasa I,T-lymphocytes derived from patients with breast adenocarcin-Kobayashi Y, Hotta K. Abundance of Gal b1–3 GalNAc inoma recognize an epitope present on the protein core of a mucinO-linked oligosaccharide on hinge region of polymerized IgA1molecule preferentially expressed by malignant cells. Cancer Resand heat-aggregated IgA1 from normal human serum. J Biochem1991; 51: 2908–29161996; 120: 92–9714. Wu AM, Csako G, Herp A. Structure, biosynthesis, and function

31. Iwase H, Tanaka A, Hiki Y et al. Aggregated human serumof salivary mucins. Mol Cell Biochem 1994; 137: 39–55immunoglobulin A1 induced by neuraminidase treatment had a15. Iwase H, Tanaka A, Hiki Y et al. Estimation of the number oflower number of O-linked sugar chains on the hinge portion.O-linked oligosaccharides per heavy chain of human serum IgA1J Chromatogr 1999; B724: 1–7by matrix-assisted laser desorption ionization time-of-flight mass

32. Kokubo T, Hiki Y, Iwase Y et al. Evidence for involvement ofspectrometry (MALDI-TOFMS) analysis of the hinge glycopep-IgA1 hinge glycopeptide in the IgA1–IgA1 interaction in IgAtide. J Biochem 1996; 120: 393–397nephropathy. J Am Soc Nephrol 1997; 8: 915–91916. Kokubo T, Hiki Y, Iwase H et al. Protective role of IgA1

33. Mestecky J, Hashim OH, Tomana M. Alterations in the IgAglycans against IgA1 self-aggregation and adhesion to extracellu-carbohydrate chains influence the cellular distribution of IgA1.lar matrix proteins. J Am Soc Nephrol 1998; 9: 2048–2054Contrib Nephrol 1995; 111: 66–7217. Czerkinsky C, Koopman WJ, Jackson S et al. Circulating

34. Coppo R, Amore A, Gianoglio B et al. Macromolecular IgAimmune complexes and immunoglobulin A rheumatoid factorand abnormal IgA reactivity in sera from children with IgAin patients with mesangial immunoglobulin A nephropathies.

J Clin Invest 1986; 77: 1931–1938 nephropathy. Clin Nephrol 1995; 43: 1–13

Received for publication: 24.2.99Accepted in revised form: 26.8.99