Kidney International, Vol. 62 (2002), pp. 465–475

Interactions of human mesangial cells with IgA andIgA-containing immune complexes1

JAN NOVAK, HUONG L. VU, LEA NOVAK, BRUCE A. JULIAN, JIRI MESTECKY, andMILAN TOMANA

Departments of Microbiology, Pathology, and Medicine, University of Alabama at Birmingham, Birmingham,Alabama, USA

Interactions of human mesangial cells with IgA and IgA-con- vated levels of IgA1 and IgA1-containing circulating im-taining circulating immune complexes. mune complexes (CIC) are found in sera of most IgAN

Background. IgA nephropathy (IgAN) is characterized by patients [4–7]. IgA1-containing immune deposits are ob-IgA1-containing immune complexes in mesangial deposits andserved in the renal mesangium (usually with C3 comple-in the circulation. The circulating immune complexes (CIC) arement component, and often with IgG, IgM, or both) [8].composed of galactose- (Gal) deficient IgA1 and IgG or IgA1

antibodies specific for the Gal-deficient IgA1; interactions of The progression of the disease is characterized by prolif-these CIC with mesangial cells (MC) were studied. eration of mesangial cells and expansion of extracellular

Methods. Binding, internalization, and catabolic degrada- matrix (ECM), leading ultimately to glomerular sclerosistion of myeloma IgA1 protein as a standard control and the[9]. About 30 to 40% of IgAN patients develop kidneyisolated CIC were studied using human MC, hepatoma cell linefailure within 20 years of clinical onset; these patients re-HepG2 expressing the asialoglycoprotein receptor (ASGP-R),

and monocyte-like cell line U937 expressing the Fc�-R (CD89). quire either dialysis or kidney transplantation [10]. IgANBiochemical and molecular approaches were used to assess recurs in approximately 50 to 60% of patients after kid-expression of CD89 and ASGP-R by MC.

ney transplantation [11]. On the other hand, a kidneyResults. At 4�C, radiolabeled IgA1 bound to MC and HepG2transplanted from a donor with subclinical IgAN to acells in a dose-dependent and saturable manner. The binding

was inhibited by IgA-containing CIC or excess IgA1 or its Fc patient without IgAN clears the immune deposits afterfragment but not by the Fab fragment of IgA1. At 37�C, the several weeks [12]. It is therefore apparent that IgANcell-bound IgA1 was internalized and catabolized. In addition arises from formation and deposition of immune com-to IgA1, HepG2 cells also bound (in a Ca2�-dependent man-

plexes rather than from a defect inherent in the kidneyner), internalized, and catabolized asialoorosomucoid (ASOR),[2, 4–6, 8, 13–16].other asialo-(AS)-glycoproteins, and secretory component (SC).

The binding by MC appeared to be restricted to IgA1 or AS- IgA immune deposits are likely derived from CIC,IgA1 and was not Ca2�-dependent. Furthermore, MC and HepG2 however, the nature of the antigens in the CIC is un-cells internalized and catabolized IgA1-containing CIC. Using known [4, 5, 14]. Based on our experimental data, weRT-PCR with ASGP-R- or CD89-specific primers, mRNAs of

postulated that aberrant glycosylation of IgA1 immuno-the two respective genes were not detected in MC.globulins in the circulation of IgAN patients leads toConclusions. The data showed that the ability of MC to bind

IgA1 and IgA1-containing CIC in vitro was mediated by an formation of CIC [6]. Incomplete galactosylation ofIgA receptor that was different from CD89 or ASGP-R and O-linked glycans in the IgA1 hinge region results in thehad a higher affinity for IgA-CIC than for uncomplexed IgA. exposure of underlying neo-antigenic N-acetylgalactos-

amine (GalNAc) [6, 16] that is recognized by naturallyoccurring antibodies (IgG or IgA1 specific for GalNAc)

Idiopathic IgA nephropathy (IgAN) is the most com-[6]. We further postulated that these CIC deposit onmon primary glomerulonephritis in the world [1–3]. Ele-mesangial cells (MC), resulting in MC proliferation andoverexpression of extracellular matrix [15, 17]. Recently,

1 See Editorial by Gomez-Guerrero, Suzuki, and Egido, p. 715. highly undergalactosylated IgA1 was detected in themesangial deposits in IgAN patients, and this furtherKey words: renal mesangial cells, circulating immune complexes, IgA

nephropathy, glomerulonephritis. supports our hypothesis [18, 19].In this regard, understanding the mechanisms by whichReceived for publication March 26, 2001

CIC bind to MC is of utmost importance [10]. Severaland in revised form March 7, 2002Accepted for publication March 22, 2002 studies evaluating the nature of an IgA receptor ex-

pressed on MC have provided conflicting results. Some 2002 by the International Society of Nephrology

465

Novak et al: Mesangial cells and IgA-containing complexes466

found Fc�R (CD89) or the asialoglycoprotein receptor (a) positive staining for vimentin, and (b) negative stain-(ASGP-R) expressed by cultured MC or in some renal ing for factor VIII-related antigen and cytokeratin (to ex-biopsy samples from IgAN patients [20–23]. Other stud- clude contamination with endothelial and epithelial cells,ies showed binding of IgA1 to MC but did not confirm respectively). Human hepatocarcinoma cell line HepG2expression of CD89 by MC [24–27]. Most of these studies and human monocytic cell line U937 (both obtained fromwere conducted using uncomplexed [monomeric (m), ATCC, Rockville, MD, USA) were maintained in RPMIpolymeric (p)] or heat-aggregated IgA1. However, in light 1640, 10% FCS, and antibiotics [28]. MC for binding ex-of recent findings on the role of CIC containing aberrantly periments were serum starved (medium contained onlyglycosylated IgA1 in the pathogenesis of IgAN [6, 10, 0.5% FCS) 24 hours before the experiment.17–19], it is important to examine interactions of MC

Isolation of IgA1 proteins, their Fab andwith not only free IgA1 and de-galactosylated IgA1, butFc fragments, and IgA1-containing CICalso with galactose-deficient IgA1 bound in CIC.

Our study reports that human MC in primary culture IgA1 and IgA2 myeloma proteins were isolated frombind, internalize, and catabolically degrade IgA1 and plasma of several different patients with multiple mye-IgA1-containing CIC. HepG2 cells expressing ASGP-R loma by precipitation with ammonium sulfate, starch-[28, 29] and U937 expressing Fc�R (CD89) served as a block electrophoresis, and size-exclusion and ion-exchangepositive controls. ASGP-R is a hetero-oligomer of two chromatography [6, 33]. The polymeric forms of IgA1subunits, H1 and H2, encoded by separate genes [30, 31]. (Mce) and IgA2 (Fel) were used [6] and their polymericASGP-R on hepatocytes binds glycoproteins with termi- character was demonstrated by size-exclusion chroma-nal Gal or GalNAc residues, and the bound glycopro- tography and sodium dodecyl sulfate (SDS)-gel electro-teins are subsequently internalized and catabolized [28, phoresis under non-reducing conditions. The Fab frag-29, 32]. A standard probe to detect ASGP-R is asialo- ment of IgA1 myeloma protein (Ste) was prepared byorosomucoid (ASOR) [28, 29] that is rapidly internalized cleavage with IgA1 protease from Haemophilus influen-and catabolized by hepatocytes upon binding [6, 29]. zae (HK50); Fab and Fc fragments of IgA1 (Mce) wereASGP-R also binds secretory component (SC) [28], a generated using IgA1 protease from Neisseria gonor-highly glycosylated fragment of the pIg receptor (pIgR) rhoeae [39, 40]. The resulting Fab and Fc fragments werethat remains bound to secretory IgA or IgM [33–35]. purified by size-exclusion chromatography [33] and theirEventual catabolic degradation of SC or ASOR by MC purity was verified by SDS electrophoresis.thus would confirm the previously postulated presence Fractions rich in Gal-deficient IgA1-containing CICof ASGP-R on MC [21]. Fc�R (CD89) is a glycoprotein

were prepared from sera of IgAN patients by size-exclu-expressed on myeloid cells that binds both IgA sub-

sion chromatography on a calibrated Superose 6 columnclasses [36]. Binding of IgA to the Fc�R also may lead[6]; high-molecular-mass fractions reactive with Helixto internalization and catabolic degradation [28]. Usingaspersa (HAA) lectin (which binds terminal GalNAc ofthese control cells, we have demonstrated differences inGal-deficient O-glycans of IgA1; Sigma, Chemical Com-properties and binding specificities of IgA1 receptorspany, St. Louis, MO, USA) and containing IgA and IgGexpressed by human MC, HepG2 and U937 cells. Ourwere pooled [6].observations indicated that the ability of MC to bind

Asialo- (AS) IgA1, asialo-agalacto IgA1 and ASORIgA1 and IgA1-containing CIC in vitro is mediated bywere prepared from native proteins by incubation witha new type of IgA receptor with higher affinity for IgA-neuraminidase (from Vibrio cholerae) and �-galactosi-CIC than for uncomplexed IgA.dase (from bovine testes that preferentially cleaves �1,3linkages) [6, 16, 28, 41]. SC was isolated from human

METHODS milk [33]. IgG was purified from normal human serumby ammonium sulfate precipitation, and subsequent ion-Cellsexchange (DEAE-cellulose) and affinity chromatogra-Human MC were isolated from the normal portionsphy on a Protein-G column [42].of kidney cortexes of tumor nephrectomy specimens (2

To prepare in vitro a complex of IgG bound to Gal-cell cultures) as described before [37, 38] or purchased asdeficient pIgA1, we incubated 125I-labeled degalactosy-primary cells (2 cell cultures) from a commercial sourcelated IgA1 (Mce) [6] with purified human IgG with anti-(Clonetics, San Diego, CA, USA). Cells from passagesGalNAc activity [6]. Free IgA1 was separated from the3 to 5 were used in our experiments. MC were maintainedIgG-IgA1 complexes by size-exclusion chromatographyin RPMI 1640 supplemented with 20% fetal calf serumon a Superose 6 column. The fractions containing IgG(FCS), l-glutamine (2 mmol/L), penicillin G (100 U/mL),bound to the radiolabeled IgA1 were identified by usingstreptomycin (0.1 mg/mL) in humidified 5% CO2 atmo-Protein-G-coated Immulon Removawell strips (Dyna-sphere at 37�C. Purity and identification of MC was based

on cell morphology and immunohistochemical features: tech Laboratories, Alexandria, VA, USA) and a Packard

Novak et al: Mesangial cells and IgA-containing complexes 467

model 5110 gamma spectrometer (Packard Instrument HEPES buffer, lysed with a polyacrylamide gel electro-phoresis loading buffer containing 2% SDS, and ana-Company, Downers Grove, IL, USA).lyzed by SDS-polyacrylamide gel electrophoresis (SDS-

Radioiodination and tetramethylrhodamine PAGE), using 1.5 mm thick vertical slab gels (20 �isothiocyanate (TRITC)-labeling 20 cm) with 5 to 20% polyacrylamide gradient [28]. The

gels were fixed, frozen with dry ice, and sliced with a gelProteins were radiolabeled with carrier-free Na125I byslicer. The radioactivity in 1-mm sections was measuredthe lactoperoxidase method [43]. The excess free Na125Iby a gamma spectrometer.was separated from the protein by size-exclusion chro-

matography on a column of Sephadex G-25 [28]. Confocal microscopyIgA1 (Mce) was TRITC-labeled using 0.02 mg TRITC

Mesangial cells grown on a microscope slide were incu-per mg protein in 2% bicarbonate buffer, pH 8.2. Afterbated with TRITC-labeled IgA1 (Mce; 10 �g) overnightovernight incubation, free TRITC was removed on aat 4�C, extensively washed with RPMI medium containingcolumn of Sephadex G-50 and the TRITC-IgA1 conju-0.5% FCS and then incubated at 37�C for four hours,gate was isolated using ion-exchange chromatographyfollowed by incubation at room temperature for twoon a column of DEAE-cellulose. The TRITC-labeledhours. Imaging was performed on a Leica DMIRBE in-

protein was aliquoted and stored at �70�C. verted epifluorescence/Nomarski microscope outfitted withLeica TCS NT Confocal optics. The system is equippedBinding experimentswith UV, argon ion, krypton ion, and helium/neon lasers

Mesangial cells (or HepG2 cells) grown in 24-well plates for imaging in a wide range of blue, red, and far-red(60-80% confluent) were washed twice with 20 mmol/L fluorescence. The laser was set to optimal TRITC exci-HEPES buffer, pH 7.3, containing 140 mmol/L NaCl, tation wavelength to observe the internalized TRITC-0.8 mmol/L MgCl2, 0.34 mmol/L K2HPO4, 0.34 mmol/L IgA1.KH2PO4, (buffer A), and incubated with a radiolabeled

Reverse transcription-polymerase chainprotein in 0.2 mL buffer B [buffer A supplemented withreaction (RT-PCR)2.7 mmol/L CaCl2 and 1% bovine serum albumin (BSA)]

on ice for one hour [28]. Buffer B without CaCl2 supple- Total RNA isolated from the cells (MC, HepG2, andment also was used in some experiments, as described U937) with RNA-Stat-60 reagent (Tel-Test, Friends-in the text. In the inhibition experiments, MC were pre- wood, TX, USA) was used for RT-PCR. HepG2 and

U937 cells served as controls for ASGP-R and CD89incubated for one hour with inhibitors on ice, beforeexpression, respectively. First-strand cDNA synthesisradiolabeled IgA1 was added and incubated for anotherwas performed at 42�C for 15 minutes, followed by 37�Cone hour. Following extensive washing with buffer B,for 45 minutes using murine leukemia virus reverse tran-cells were lysed with 0.4 mL 0.3 N NaOH and the radioac-scriptase and oligo(dT)16 as the primer. The cDNA wastivity of the lysate was determined in a � spectrometer.amplified using the following primers that we designatedCell count was determined with cells released by trypsin/based on the sequences of the correspond ing genesethylenediaminetetraacetic acid (EDTA) treatment us-submitted to GeneBank [30, 31, 44] for Fc� receptoring a hemocytometer.(CD89): CD89F 5-AGCACGATGGACCCCAAACAGA-3; CD89R 5-CTGCCTTCACCTCCAGGTGTT-3,Catabolic degradation of internalized proteinsand the following primers for H1 and H2 genes encod-Mesangial cells and HepG2 in tissue culture flasks wereing the two ASGP-R subunits: H1F 5-CTGGACAATincubated with radiolabeled proteins (�1 �g of ASOR,GAGGAGAGTGAC-3; H1R 5-TTGAAGCCCGTCAS-IgA1, or immune complexes) in 1 mL buffer B atTCGTAGTC-3; H2F CCTGCTGCTGGTGGTCATC37�C for four hours. Radioactivity was determined in (a)TG-3; H2R 5-CCCATTTCCAAGAGCCATCAC-3.incubation medium, (b) cells washed and released by A pair of primers specific to �-actin (F 5-TTCCAGCC

trypsinization, and (c) released cells after additional tryp- TTCCTTCCTGG-3; R 5-TTGCGCTCAGGAGGAGsinization to remove protein bound on cell surface. Cata- CAA-3) was used as a control. PCR was performed in abolic degradation was expressed as percentage of total DNA thermal cycler using 94�C melting, 60�C annealing,radioactivity not precipitable by 10% trichloroacetic acid and 72�C extension temperatures for 35 cycles. PCR(TCA; from incubation medium or cell lysate, as speci- amplicons were analyzed on 2% NuSieve 3:1 agarosefied in the text) [28]. Protein in the cell lysate was deter- gels.mined spectrophotometrically using a BioRad assay (Bi-oRad, Hercules, CA, USA) with BSA as the standard.

RESULTSTo determine molecular masses of proteins cataboli-Binding of IgA1 and IgA1-containing CIC to MCcally degraded, MC were incubated in 2 mL HEPES

buffer, pH 7.3 containing 1% BSA with 10 �g 125I-labeled To determine binding characteristics of IgA1, MC wereincubated for one hour on ice with various concentra-IgA1 at 37�C for 16 hours. The cells were washed with

Novak et al: Mesangial cells and IgA-containing complexes468

Fig. 1. Binding of 125I-labeled IgA1 (Mce) myeloma protein (�) and125I-labeled IgA1 (Mce) with partially degalactosylated and desialylatedO-linked glycans ( ) to mesangial cells (MC). MC were incubated forone hour on ice with 1 �g radiolabeled proteins. Cells were washed and

Fig. 2. Isolation of Gal-deficient IgA1 containing CIC from serum ofthe bound radioactivity determined as described in the Methods section.an IgAN patient. Serum (0.5 mL) was fractionated on a Superose 6column (0.9 � 60 cm), 0.25 mL fractions were collected and analyzedby ELISA for IgA (�), IgG (�), and reactivity with HAA lectin (�;

tions of radiolabeled pIgA1 (Mce) myeloma protein. specific for terminal GalNAc, thus reacting with Gal-deficient IgA1).Fractions containing CIC with aberrantly glycosylated IgA1 (molecularConsistent with earlier reports, the binding of the pIgA1mass about 700 to 900 kD) were pooled and used in the experiments.to MC was dose-dependent and saturable. The binding

of radiolabeled pIgA1 was inhibited by 68% using a 150-fold excess of unlabeled pIgA1. Analyses of IgA1 bind-ing to MC using a Scatchard plot (not shown) suggested not by the Fab fragment of IgA1 (that contained a por-a single population of receptors with approximately 5 � tion of the hinge region). Surprisingly, CIC containing105 binding sites per cell and Ka 4.79 � 106 mol/L�1. The aberrantly glycosylated IgA1 appeared to be better in-calculations were based on the assumption that pIgA1 hibitors than uncomplexed monomeric or polymeric IgA1(Mce) was predominantly polymeric, as judged from its (Fig. 3A). The same amount of uncomplexed IgA1 (Gal-elution profile on a calibrated size-exclusion high pres- deficient or normally glycosylated) had no significantsure liquid chromatography (HPLC) column (TSK 5000). inhibitory effect (data not shown), suggesting that prop-Furthermore, we assumed that each receptor bound only erties such as spatial organization of IgA1 in CIC mayone IgA1 molecule. play a role in MC binding. By comparing the inhibitory

Because the mesangial immune deposits in IgAN activity of Fab and Fc portions of IgA1, respectively, wepatients contain aberrantly glycosylated (Gal-deficient concluded that IgA1 bound to a MC IgA receptor by itsO-linked glycans) IgA1 [18, 19], we examined MC bind- Fc portion (Fig. 3B). This is consistent with our findinging of Gal-deficient IgA1 myeloma protein. The binding that both IgA subclasses (IgA1 and IgA2) bound to MCof IgA1 that was modified by treatment with neuramini- (data not shown).dase and �-galactosidase (that removes preferentially The results in the prior section suggested better bind-�1,3 bound Gal) was more than twofold greater com- ing of IgA1 in CIC compared with free IgA1. To verifypared with the unmodified control (Fig. 1). In sera of this finding, sera from three IgAN patients were fraction-IgAN patients, however, the aberrantly glycosylated ated using size-exclusion chromatography on a calibratedIgA1 [16, 45–48] is not free, but rather is complexed Superose 6 column and analyzed for IgA, and for reac-with IgG (or IgA1) in CIC [5, 6, 16]. To prepare IgA1- tivity with HAA. The fractions corresponding to mIgAcontaining CIC, serum from an IgAN patient was frac- and to IgA complexed in CIC were incubated with MCtionated using size-exclusion chromatography. Fractions grown on microscope slides. After three hours of incuba-with Gal-deficient IgA1 were identified with GalNAc- tion at 4�C, MC were washed with phosphate-bufferedspecific lectin (HAA) in ELISA (Fig. 2). The HAA- saline (PBS) and stained with TRITC-conjugated F(ab)2

reactive fractions (CIC of molecular mass 700 to 900 kD), fragment of anti-human IgA antibody and examineddesignated as CIC containing Gal-deficient IgA1, were with a fluorescence microscope. MC incubated with CICpooled and used in further experiments. showed strong IgA binding, while MC incubated with

Binding of radiolabeled Gal-deficient pIgA1 was in- uncomplexed IgA exhibited only background-level fluo-rescence.hibited by unlabeled Gal-deficient pIgA1 (Fig. 3A), but

Novak et al: Mesangial cells and IgA-containing complexes 469

Fig. 3. Inhibition of 125I-labeled Gal-deficient pIgA1 (Mce) binding to MC. MC grown to 80% confluence in a 24-well plate were pre-incubatedwith inhibitors for one hour on ice and then about 1 �g radiolabeled protein was added. Cells were washed and the bound radioactivity wasdetermined as described in Methods. (A) Control is no inhibitor, and inhibitors are Gal-deficient pIgA1 Mce (100 �g), mIgA from serum of anIgAN patient (�50 �g), and Gal-deficient-IgA1-containing CIC from serum of an IgAN patient (�1 �g). (B) Control is no inhibitor, and inhibitorsinclude Gal-deficient pIgA1 Mce (100 �g), Fab fragment of IgA1 (100 �g), and Fc fragment of IgA1 Mce (100 �g).

These results indicated that IgA1 bound to MC throughthe Fc part of its molecule because IgA1 or its Fc frag-ment, but not its Fab fragment, inhibited IgA binding.Furthermore, abnormalities of O-linked glycans of IgA1and spatial organization of IgA1 molecules clustered inCIC influenced binding to MC, favoring IgA in CIC overuncomplexed IgA.

Catabolism and internalization of IgA1 by MC

To detect potential degradation products of cell-asso-ciated and internalized IgA1, the radiolabeled proteinwas added to MC and incubated for 16 hours; the cellswere then washed, lysed, and the lysate was analyzed bySDS gel electrophoresis under non-reducing conditions(Fig. 4). Generation of protein fragments smaller than30 kD was observed after incubation with MC, indicatingcatabolic degradation of the 125I-labeled IgA1. This find-ing is consistent with decreased TCA-precipitable radio-activity. Fig. 4. Catabolism of 125I-labeled IgA1 (Mce) by MC. About 10 �g

125I-labeled protein was added to the cells in tissue culture flask andTo visualize internalization of IgA1 by MC, we incu-incubated in 2 mL HEPES buffer with 1% BSA at 37�C for 16 hours.bated MC grown on a microscope slide with TRITC- The cells were then washed, and lysed using 2% SDS buffer and the

labeled IgA1. The cells were then fixed and observed with lysate was analyzed by SDS gel electrophoresis under non-reducingconditions (solid line). The distribution of radioactivity in the gel wasa confocal laser scanning microscope. Many MC showeddetermined by assaying the radioactivity of 1-mm sections of the gel inintracellular fluorescent vesicles, which indicated inter- a gamma counter. Control 125I-labeled IgA1 (dotted line) was electro-

nalization of IgA1 (Fig. 5). phoresed in parallel. Migration of standards is shown by arrows.

Internalization and catabolism of AS-IgA1, ASOR,and SC by HepG2 and MC ASGP-R [28, 29] and served as a positive control. The

cell cultures were then washed, treated with trypsin toAsialoglycoprotein receptor has been reported as onerelease cells from the flasks and surface-bound radiola-of the receptors responsible for binding of IgA1 to MCbeled protein from cells. After further washing, the cells[21]. To verify this report, we incubated 125I-labeled AS-

IgA1 with MC and HepG2 cells. HepG2 cells express were lysed with NaOH. The internalized protein was de-

Novak et al: Mesangial cells and IgA-containing complexes470

Fig. 5. Confocal laser scanning photomicro-graph of IgA1 internalized by a MC. MC grownon a microscope slide were incubated withTRITC-labeled IgA1 (Mce; 10 �g) overnightat 4�C, extensively washed with RPMI 1640medium containing 0.5% FCS and incubatedat 37�C for four hours, followed by incubationat room temperature for two hours. A singleMC is shown with fluorescent vesicles in thecytoplasm. Bar depicts 20 �m.

tected as radioactivity in the trypsin-treated and washed MC was internalized 144-fold more effectively by HepG2than MC (Fig. 6). Likewise, the catabolic degradationcells. AS-IgA1 was internalized by both cell cultureswas more effective in HepG2 cells. Furthermore, it was(Fig. 6). Under the same conditions, HepG2 internalizedalso reported that HepG2 internalize and catabolize SCsixfold more AS-IgA1 per mg cell protein than did MC.[28]. Therefore, we compared catabolism of SC and IgA1To estimate the kinetics of catabolic degradation of inter-in MC (Table 1). More than 99% of original 125I-labelednalized proteins, percentage of the radiolabeled proteinSC incubated with MC remained intact, while the origi-not precipitable with TCA was determined. HepG2 cellsnal 125I-labeled IgA1 was partially (about 13%) cataboli-catabolized 47% of the internalized AS-IgA1 comparedcally degraded during the four-hour incubation. Theseto 39% for MC. Unlike with HepG2, the binding anddata demonstrated that SC, but not IgA1, escaped cata-

internalization of AS-IgA1 by MC was not Ca2�-depen- bolic degradation by MC. Therefore, ASGP-R is missingdent (data not shown). or nonfunctional on MC.

Earlier studies demonstrated that ASGP-R on hepato-Binding and internalization of IgA1-containingcytes or HepG2 cells is responsible for rapid internaliza-CIC by HepG2 and MCtion and catabolic degradation of ASOR [28]. We investi-

gated whether ASOR also is internalized and catabolized Studies described above indicated that IgA1-contain-ing CIC bound to MC more effectively than free IgA1.by MC. 125I-labeled ASOR incubated with HepG2 and

Novak et al: Mesangial cells and IgA-containing complexes 471

Fig. 6. Internalization of 125I-labeled ASOR and 125I-labeled asialo-(AS) IgA1 (Mce) by HepG2 and MC. About 1 �g of each radiolabeled Fig. 7. HepG2 and MC cell-associated (bound and internalized)protein was added to the cells in tissue culture flask and incubated in 1 125I-labeled Gal-deficient IgA1-containing CIC isolated from serum ofmL buffer B at 37�C for four hours. Cells were washed and radioactivity an IgAN patient (�) and CIC from a healthy control ( ). About 1 �gmeasured after trypsinization to release surface-bound molecules. Re- aliquots of the radiolabeled proteins were added to the cells in tissuesults represent an average from experiments conducted in triplicates. culture flasks and incubated in 1 mL buffer B at 37�C for four hours.

Cells were washed before the cell-associated radioactivity was mea-sured. Results are averages from experiments conducted in triplicates.

Table 1. Catabolism of radioiodinated secretory component (SC)and IgA1 (Mce) by human mesangial cells

TCA precipitable protein %

SC IgA1 (Mce)

Before incubation 93.9 95.6After incubation 93.0 82.8

One-microgram aliquots of the SC or IgA1 proteins were added to MC incultivation flasks and incubated for 4 hours at 37�C. Then, the supernatant wascollected, precipitated with TCA, and the intact protein (TCA-precipitable radio-activity measured using gamma counter) was expressed as % of TCA-precipitableradioactivity. The experiment was conducted in triplicate.

To examine the possible role of liver cells and ASGP-Rin binding and processing of these CIC, we comparedthe binding and catabolism of these CIC by MC andHepG2. Gal-deficient IgA1-containing CIC were puri- Fig. 8. Internalization of a complex of IgG-Gal-deficient pIgA1 (�)

and free Gal-deficient pIgA1 ( ) by human hepatoma cell line HepG2.fied from serum of an IgAN patient and control CIC125I-labeled degalactosylated IgA1 was incubated with purified human

were isolated from serum of a healthy volunteer using IgG with anti-GalNAc activity. Free IgA1 was separated from the IgG-IgA1 complexes by size-exclusion chromatography on Superose 6 col-size-exclusion chromatography [6]. These CIC (molecu-umn. The fractions containing IgG bound to the radiolabeled IgA1lar mass about 700 to 900 kD) were radioiodinated, andwere detected by capture radioimmunoassay using Protein-G-coated

incubated with MC and HepG2. MC bound and internal- Removawell strips and gamma-counter detection. The proteins wereincubated with HepG2 cells at 37�C for three hours, then the cells wereized more protein from the IgAN-CIC than from controlwashed, treated with trypsin, and the radioactivity was measured withCIC. On the other hand, HepG2 cells bound less proteina gamma counter and expressed per mg of cell protein.

from the IgAN-CIC (Fig. 7).We hypothesized that the lower degree of internal-

ization of IgAN-CIC by hepatoma cells may be due tothe presence of GalNAc-specific IgG [6] that bound to In summary, these experiments indicated that MCIgA1 and thus prevented IgA1 hinge region O-glycans bound and internalized IgA1-containing CIC via a recep-[32] or Fc glycans [49, 50] from binding to ASGP-R on tor different from ASGP-R. MC bound more effectivelyHepG2 cells. To test this hypothesis, we prepared in the CIC from an IgAN patient than that from a healthyvitro 125I-labeled Gal-deficient IgA1 in a free form and control. Furthermore, CIC from an IgAN patient werebound to GalNAc-specific IgG and used HepG2 cells to less efficiently internalized by hepatoma cells (HepG2)assess the effect on internalization of the radiolabeled than control CIC. The IgG bound to Gal-deficient IgA1IgA1. Complexing IgG with the IgA1 reduced the bind- apparently masks the binding sites on IgA1 glycans from

ASGP-R or interferes with an efficient internalization.ing and internalization by HepG2 cells (Fig. 8).

Novak et al: Mesangial cells and IgA-containing complexes472

Fig. 9. RT-PCR of Fc�R (CD89) transcriptsin MC and U937 cells. Total RNA was reverse-transcribed and the cDNA was PCR-amplifiedwith CD89-specific primers. The ampliconswere separated on 2% agarose gel and photo-graphed under UV light. Lane 1, molecularsize standards; lanes 2-5, �-actin RT-PCR inMC; lane 6, �-actin RT-PCR in U937; lane 7,RT-PCR of Fc�R transcripts in U937 withthree signals detected that correspond to thea.1, a.2, and a.3 splicing variants; lanes 8-11,RT-PCR of mRNA from MC grown undervarious conditions failed to reveal any CD89-specific signal (lanes 10, 11, MC were supple-mented with insulin-like growth factor; lanes9, 10, 5% glucose was added to the storagemedium).

CD89 and ASGP-R mRNA expression in MC mune deposits originate from CIC includes: (a) IgA1,but not IgA2, is present in CIC in the circulation of mostTo determine a possible involvement of CD89 in bind-IgAN patients [4, 5] and in their mesangial deposits [55];ing IgA to MC, RT-PCR was used to examine whether(b) shared idiotypic determinants are expressed on CICthe CD89 gene is transcribed in MC. Total RNA isolatedand in mesangial deposits [56], however, without a dis-from MC grown with, or without, insulin-like growth fac-ease-specific idiotype [57]; (c) Gal-deficient IgA1 is pres-tor and from U937 cells served as templates for RTent in CIC [6, 16, 45–48] and mesangial deposits [18, 19]followed by PCR amplification with CD89-specific prim-in IgAN; and (d) Gal-deficient IgA1 is also found in theers. The results did not indicate the presence of CD89circulation of Henoch-Schonlein purpura patients, butmRNA in MC, although all three major splicing prod-only in those with clinical nephritis [58].ucts, a.1, a.2, and a.3, (observed as PCR amplicons of

We have postulated that aberrant glycosylation of theabout 0.9 kb, 0.85 kb, and 0.62 kb, respectively) of CD89hinge region of some IgA1 molecules of IgAN patientsmRNA were detected in samples from CD89-positiveexposes antigenic determinant(s) comprised of GalNAcU937 cells (Fig. 9). Adding insulin-like growth factorlinked to Ser or Thr of the polypeptide chain [16]. The(known to alter gene expression in MC [51] and used inGal-deficient IgA1 is, in turn, recognized by naturallyexperiments by others [20, 52, 53]) to the growth mediumoccurring antibodies (IgG or IgA1 specific for GalNAc)or glucose [54] to the storage medium did not inducethat form CIC [6, 16], some of which deposit in theexpression of CD89 (Fig. 9).mesangium. Indeed, two groups have detected highlyFurthermore, we determined whether MC expressundergalactosylated IgA1 in the kidney mesangial cellsASGP-R. cDNA prepared from total RNA from MCof IgAN patients [18, 19]. While binding of human IgA1and HepG2 cell cultures served as templates for RT-to human and rat MC has been well documented [20,PCR with two sets of primers specific for H1 and H2 sub-24, 26, 27, 53], there had been no such study with CICunits of the ASGP-R. MC did not yield any specific sig-containing undergalactosylated IgA1 isolated from seranals, while samples from HepG2 contained RNA for bothof IgAN patients.ASGP-R subunits, detected on an agarose gel as bands

This study compared uncomplexed Gal-deficient IgA1of about 0.6 kb and 0.5 kb, respectively (data not shown).and CIC containing Gal-deficient IgA1 for the binding,In summary, MC expressed a receptor that boundinternalization, and catabolism by human MC. IntactIgA1 and IgA1-containing CIC, but this receptor did notIgA1 or the Fc portion but not its Fab fragment inhibitedexhibit properties of Fc�R (CD89) or ASGP-R. Greaterbinding of IgA1 to MC. This finding indicates that IgA1binding affinity of this receptor for Gal-deficient IgA1-bound to MC through the Fc portion of the molecule.containing CIC compared with uncomplexed IgA mayMC bound asialo-agalacto-IgA1 better than normallyexplain mesangial deposition of these CIC in IgAN.glycosylated IgA1 [59, 60]. Results of inhibition experi-ments indicated that CIC from IgAN patients bound to

DISCUSSION MC more efficiently than complexes from healthy con-trols, or than normally glycosylated IgA1 or asialo-aga-IgA deposits in the glomerular mesangium in IgANlacto-IgA1. Interestingly, binding of CIC to MC wasare apparently derived from CIC [2, 4, 5, 8, 10, 15, 17];partially inhibited by normally glycosylated IgA1 or asi-however, the nature of antigens and ensuing CIC is un-

known. The evidence suggesting that the mesangial im- alo-agalacto-IgA1 but only marginally by IgG. These

Novak et al: Mesangial cells and IgA-containing complexes 473

findings underscore the importance of IgA receptor(s) [20, 22, 23], we did not detect its mRNA in MC fromnormal kidney tissue or commercial sources. Concordantfor binding of IgA1-CIC and are consistent with observa-

tions that IgG receptors are significantly expressed only with our results, other recent studies also failed to detectCD89 on human MC [24–27]. Insulin-like growth factorafter MC activation or stimulation [61].

Our study showed for the first time that: (a) the IgAN was used by others as a supplement in the culture me-dium [20] and, therefore, we also tested whether thisCIC containing Gal-deficient IgA1 bound to MC more

efficiently than uncomplexed IgA; (b) a greater amount growth factor would induce CD89 expression. However,no induction was detected. Fc�R (CD89) has severalof CIC from an IgAN patient bound to MC than CIC

from a healthy control; and (c) a novel IgA Fc receptor isoforms that originate from alternative splicing [77].Because some reports have shown conflicting data aboutwas important for CIC binding to MC. These findings

suggest a direct role for aberrant IgA1 glycosylation in the expression of CD89 on MC, we considered that anexpressed variant of CD89 may be a splicing version thatthe formation of CIC and their binding to MC. Prelimi-

nary experiments suggested reduced binding of CIC was not detected by the primers. Therefore, we designedprimers that could detect spliced variants, but found nofrom an IgAN patient to HepG2, implying that these

CIC may more easily escape hepatic catabolism. This such transcripts in MC. The reasons some investigatorsdetected CD89 are unclear, but may include differencescharacteristic may be one of the factors responsible for

increased circulating IgA1 levels in IgAN patients. in cultivation techniques, contamination with other celltypes (such as macrophages) known to express CD89,While MC in vitro bind IgA1 in a saturable manner

and the binding is inhibited by an excess of unlabeled or valid induction of CD89 or its variant [27].Two other receptors have recently emerged as novelIgA1 [20, 21, 24, 26, 27, 52], the nature of the recep-

tor(s) has remained controversial. Several IgA receptors candidate IgA receptors on MC: CD71 (transferrin re-ceptor) [65] and Fc�/� receptor [66, 67]. We confirmedhave been identified on human cells: ASGP-R on he-

patocytes [28, 32]; pIgR on epithelial cells [62]; Fc�R the expression of CD71 mRNA by proliferating MCin culture (not shown). However, MC bind both IgA(CD89) on monocytes, neutrophils, and eosinophils [29,

44, 63, 64]; CD71 (transferrin receptor) [65]; and Fc�/� subclasses [27], while CD71 binds only IgA1 [65]. MCbind pIgA1 better than mIgA [27], in contrast to thereceptor [66, 67]. The pIgR [26] and surface-bound ga-

lactosyltransferase can be excluded as possible candi- strong preference of CD71 for mIgA1 [65]. Furthermore,CD71 is expressed by many proliferating cells, some ofdates because the bound proteins are not catabolic-

ally degraded [68–75] and pIgR binds polymeric but not which do not bind IgA1. Obviously the issues concerningthe role of CD71 on MC in binding Gal-deficient IgAmonomeric IgA.

Asialoglycoprotein receptor is a hetero-oligomer of or IgA-CIC remain to be addressed.The other newly described candidate, Fc�/� receptor,two homologous subunits, H1 and H2, encoded by sepa-

rate genes [30, 31]. ASGP-R on hepatocytes binds and appears to be transcribed by MC in vitro [67]. The ex-pression of this receptor by MC or in renal tissue andinternalizes some glycans or glycoproteins with terminal

Gal and GalNAc residues. The internalized proteins are its role in binding IgA-CIC remain to be investigated,although IgM does not inhibit IgA1 binding to MC [27].then catabolically degraded [28]. ASOR and IgA1 are ex-

cellent probes for ASGP-R as they are efficiently bound, This binding pattern may be explained if the variant ofthe Fc�/� receptor expressed by MC preferentially bindsinternalized, and catabolized by human hepatocytes and

the hepatoma cell line HepG2 [28, 29, 76]. To examine IgA. Clearly, the properties of this receptor need to beinvestigated. A well-defined system, such as COS-7 cellsthe postulated presence of ASGP-R on human MC [21],

we compared binding, internalization, and catabolism of expressing the receptor [67], may be appropriate forsuch a task.radiolabeled AS-IgA1 and ASOR by MC in primary cul-

ture and by a HepG2 cell line. Our experiments showed In conclusion, we detected binding, internalization, andcatabolism of IgA1 and IgA1-containing CIC by humanthat only HepG2 bound, internalized, and catabolized

both AS-IgA1 and ASOR, while MC bound and degra- MC, which was apparently independent of CD89 andASGP-R. While IgA1 is likely to bind to a single popula-ded AS-IgA1, but not ASOR. This finding was consistent

with our observation that only HepG2 cells exhibited tion of receptors, CIC can interact with MC in a morecomplex manner that may involve other Ig-specific ormRNAs encoding the two ASGP-R subunits. Therefore,

it was unlikely that MC, under the conditions of our complement receptors. However, the hallmark of IgANis mesangial deposition of IgA1, and because purifiedexperiments, expressed ASGP-R. Other investigators re-

cently reached the same conclusion [26]. IgA1 was shown to bind to MC, it is very likely that aspecific, not yet biochemically characterized, IgA recep-Fc�R (CD89) is a glycoprotein expressed on myeloid

cells that binds both IgA subclasses [36]. Some investiga- tor is involved. Its similarity to a recently described novelIgA receptor on human intestinal epithelial cells [78]tors have postulated that this receptor accounts for IgA1

binding to MC [20, 22, 23]. In contrast to earlier reports and the role of other IgA receptors [65, 67] remain to

Novak et al: Mesangial cells and IgA-containing complexes474

IgA1 in sera of IgA nephropathy patients is present in complexesbe investigated. Understanding the nature of the interac-with IgG. Kidney Int 52:509–516, 1997

tion of IgA1 and its mesangial receptor may not only 17. Novak J, Julian BA, Tomana M, Mestecky J: Progress in molecu-lar and genetic studies of IgA nephropathy. J Clin Immunol 21:310–explain how IgA1 and IgA1-containing CIC deposit in327, 2001glomeruli and cause IgAN, but may also provide a theo-

18. Allen AC, Bailey EM, Brenchley PEC, et al: Mesangial IgA1retical basis for developing disease-specific therapeutic in IgA nephropathy exhibits aberrant O-glycosylation: Observa-

tion in three patients. Kidney Int 60:969–973, 2001inhibitors.19. Hiki Y, Odani H, Takahashi M, et al: Mass spectrometry proves

under-O-glycosylation of glomerular IgA1 in IgA nephropathy.ACKNOWLEDGMENTS Kidney Int 59:1077–1085, 2001

20. Gomez-Guerrero C, Gonzales E, Egido J: Evidence for a specificThis work was supported by grants DK 49358, DK 57750, andIgA receptor in rat and human mesangial cells. J Immunol 151:DK61525 from the National Institutes of Health. The authors express7172–7181, 1993their appreciation to Ms. R. Brown, Ms. R. Kulhavy, and Ms. C. Barker

21. Gomez-Guerrero C, Duque N, Egido J: Mesangial cells possessfor technical assistance and Ms. L.R. Brewer and Dr. K. Matousovican asialoglycoprotein receptor with affinity for human immuno-for critically reading the manuscript.globulin A. J Am Soc Nephrol 9:568–576, 1998

22. Bagheri N, Chintalacharuvu SR, Emancipator SN: Proinflam-Reprint requests to Dr. Jan Novak, Department of Microbiology,matory cytokines regulate Fc�R expression by human mesangialUniversity of Alabama at Birmingham, 845 19th Street S, BBRB 734,cells in vitro. Clin Exp Immunol 107:404–409, 1997Birmingham, Alabama 35294, USA.

23. Kashem A, Endoh M, Yano N, et al: Glomerular Fc�R expressionE-mail: Jan_Novak@microbio.uab.eduand disease activity in IgA nephropathy. Am J Kidney Dis 30:389–396, 1997

24. Diven SC, Caflisch CR, Hammond DK, et al: IgA induced activa-REFERENCEStion of human mesangial cells: Independent of Fc�R1 (CD 89).

1. Julian BA, Waldo FB, Rifai A, Mestecky J: IgA nephropathy, Kidney Int 54:837–847, 1998the most common glomerulonephritis worldwide. A neglected dis- 25. Westerhuis R, Van Zandbergen G, Verhagen NA, et al: Humanease in the United States? Am J Med 84:129–132, 1988 mesangial cells in culture and in kidney sections fail to express Fc

2. Emancipator SN, Mestecky J, Lamm ME: IgA nephropathy and alpha receptor (CD89). J Am Soc Nephrol 10:770–778, 1999related diseases, in Mucosal Immunology, edited by Ogra PL, 26. Leung JCK, Tsang AWL, Chan DTM, Lai KN: Absence of CD89,Mestecky J, Lamm ME, et al, San Diego, Academic Press, 1999, polymeric immunoglobulin receptor, and asialoglycoprotein recep-pp 1365–1380 tor on human mesangial cells. J Am Soc Nephrol 11:241–249, 2000

3. D’Amico G: The commonest glomerulonephritis in the world: IgA 27. Barrat J, Greer MR, Pawluczyk IZ, et al: Identification of anephropathy. Quart J Med 64:709–727, 1987 novel Fc� receptor expressed by mesangial cells. Kidney Int 57:

4. Coppo R, Basolo B, Martina G, et al: Circulating immune com- 1936–1948, 2000plexes containing IgA, IgG and IgM in patients with primary IgA 28. Tomana M, Kulhavy R, Mestecky J: Receptor-mediated bindingnephropathy and with Henoch-Schonlein nephritis. Correlation and uptake of immunoglobulin A by human liver. Gastroenterolwith clinical and histologic signs of activity. Clin Nephrol 18:230– 94:887–892, 1988239, 1982 29. Baenziger JU, Maynard Y: Human hepatic lectin. Physicochemi-

5. Czerkinsky C, Koopman WJ, Jackson S, et al: Circulating immune cal properties and specificity. J Biol Chem 255:4607–4613, 1980complexes and immunoglobulin A rheumatoid factor in patients 30. Spiess M, Schwartz AL, Lodish HF: Sequence of human asialogly-with mesangial immunoglobulin A nephropathies. J Clin Invest coprotein receptor cDNA. J Biol Chem 260:1979–1982, 1985

31. Spiess M, Lodish HF: Sequence of a second human asialoglyco-77:1931–1938, 19866. Tomana M, Novak J, Julian BA, et al: Circulating immune com- protein receptor: Conservation of two receptor genes during evolu-

tion. Proc Natl Acad Sci USA 82:6465–6469, 1985plexes in IgA nephropathy consist of IgA1 with galactose-deficienthinge region and antiglycan antibodies. J Clin Invest 104:73–81, 1999 32. Baenziger JU, Fiete D: Galactose and N-acetylgalactosamine-

specific endocytosis of glycopeptides by isolated rat hepatocytes.7. Schena FP, Pastore A, Ludovico N, et al: Increased serum levelsof IgA1-IgG immune complexes and anti-F(ab’)2 antibodies in Cell 22:611–620, 1980

33. Mestecky J, Kilian M: Immunoglobulin A (IgA). Methods Enzy-patients with primary IgA nephropathy. Clin Exp Immunol 77:15–20, 1989 mol 116:37–75, 1985

34. Mestecky J, McGhee JR: Immunoglobulin A (IgA): Molecular8. Emancipator SN: IgA nephropathy and Henoch-Schonlein syn-drome, in Heptinstall’s Pathology of the Kidney, edited by Jennette and cellular interactions involved in IgA biosynthesis and immune

response. Adv Immunol 40:153–245, 1987JC, Olson JL, Schwartz MM, Silva FG, Philadelphia, Lippincott-Raven Publishers, 1998, pp 479–539 35. Mestecky J: Immunobiology of IgA. Am J Kidney Dis 12:378–

383, 19889. Emancipator SN, Lamm ME: Biology of disease. IgA nephropathy:Pathogenesis of the most common form of glomerulonephritis. 36. Monteiro RC, Kubagawa H, Cooper MD: Cellular distribution,

regulation, and biochemical nature of an Fc� receptor in humans.Lab Invest 60:168–183, 198910. Julian BA, Tomana M, Novak J, Mestecky J: Progress in the J Exp Med 171:597–613, 1990

37. Sterzel RB, Lovett DH, Foellmer HG, et al: Mesangial cellpathogenesis of IgA nephropathy. Adv Nephrol 29:53–72, 199911. Odum J, Peh CA, Clarkson AR, et al: Recurrent mesangial IgA hillocks. Nodular foci of exaggerated growth of cells and matrix

in prolonged culture. Am J Pathol 125:130–140, 1986nephritis following renal transplantation. Nephrol Dial Transplant9:309–312, 1994 38. Davies M: The mesangial cell: A tissue culture view. Kidney Int

45:320–327, 199412. Silva FG, Chander P, Pirani CL, Hardy MA: Disappearanceof glomerular mesangial IgA deposits after renal allograft trans- 39. Novak J, Tomana M, Kilian M, et al: Heterogeneity of O-glycosyl-

ation in the hinge region of human IgA1. Mol Immunol 37:1047–plantation. Transplantation 33:241–246, 198213. Coppo R, Basolo B, Piccoli G, et al: IgA1 and IgA2 immune 1056, 2000

40. Symersky J, Novak J, McPherson DT, et al: Expression of thecomplexes in primary IgA nephropathy and Henoch-Schonleinnephritis. Clin Exp Immunol 57:583–590, 1984 recombinant human immunoglobulin J chain in Escherichia coli.

Mol Immunol 37:133–140, 200014. Coppo R, Emancipator S: Pathogenesis of IgA nephropathy: Estab-lished observations, new insights and perspectives in treatment. 41. Nikolova EB, Tomana M, Russell MW: The role of the carbohy-

drate chains in complement (C3) fixation by solid-phase-boundJ Nephrol 7:5–15, 199415. Couser WG: Glomerulonephritis. Lancet 353:1509–1515, 1999 human IgA. Immunology 82:321–327, 1994

42. Tomana M, Schrohenloher RE, Koopman WJ, et al: Abnormal16. Tomana M, Matousovic K, Julian BA, et al: Galactose-deficient

Novak et al: Mesangial cells and IgA-containing complexes 475

glycosylation of serum IgG from patients with chronic inflamma- lower number of O-linked sugar chains on the hinge portion. J Chro-matogr 724:1–7, 1999tory diseases. Arthr Rheumat 31:333–338, 1988

60. Hiki Y, Kokubo T, Iwase H, et al: Underglycosylation of IgA143. Marchalonis JJ: An enzymic method for the trace iodination ofhinge plays a certain role for its glomerular deposition in IgAimmunoglobulins and other proteins. Biochem J 113:229–305, 1969nephropathy. J Am Soc Nephrol 10:760–769, 199944. Maliszewski CR, March CJ, Schoenborn MA, et al: Expres-

61. Uciechowski P, Schwarz M, Gessner JE, et al: IFN-� inducession cloning of a human Fc receptor for IgA. J Exp Med 172:1665–the high-affinity Fc receptor I for IgG (CD64) on human glomerular1672, 1990mesangial cells. Eur J Immunol 28:2928–2935, 199845. Andre PM, Pogamp P, Chevet P: Impairment of jacalin binding

62. Mestecky J, Moro I, Underdown BJ: Mucosal immunoglobulins,to serum IgA in IgA nephropathy. J Clin Lab Anal 4:115–119, 1990in Mucosal Immunology, edited by Ogra PL, Mestecky J, Lamm46. Mestecky J, Tomana M, Crowley-Nowick PA, et al: DefectiveME, et al, San Diego, Academic Press, 1999, pp 133–352galactosylation and clearance of IgA1 molecules as a possible etio-

63. Hexham JM, White KD, Carayannopoulos LN, et al: A humanpathogenic factor in IgA nephropathy. Contrib Nephrol 104:172– immunoglobulin (Ig)A C�3 domain motif directs polymeric Ig182, 1993 receptor-mediated secretion. J Exp Med 189:747–751, 1999

47. Allen AC, Harper SJ, Feehally J: Galactosylation of N- and 64. Crago SS, Kulhavy R, Prince RJ, Mestecky J: Secretory compo-O-linked carbohydrate moieties of IgA1 and IgG in IgA nephropa- nent on epithelial cells is a surface receptor for polymeric immuno-thy. Clin Exp Immunol 100:470–474, 1995 globulins. J Exp Med 147:1832–1837, 1978

48. Hiki Y, Horii A, Iwase H, et al: O-linked oligosaccharide on IgA1 65. Moura IC, Centelles MN, Arcos-Fajardo M, et al: Identificationhinge region in IgA nephropathy. Fundamental study for precise of the transferrin receptor as a novel immunoglobulin (Ig)A1 re-structure and possible role. Contrib Nephrol 111:73–84, 1995 ceptor and its enhanced expression on mesangial cells in IgA ne-

49. Phillips JO, Stohrer R, Russell MW, et al: Analysis of the hepa- phropathy. J Exp Med 194:417–425, 2001tobiliary transport of IgA with monoclonal anti-idiotype and anti- 66. Shibuya A, Sakamoto N, Shimizu Y, et al: Fc�/� receptor mediatesallotype antibodies. Mol Immunol 23:339–346, 1986 endocytosis of IgM-coated microbes. Nat Immunol 1:441–446, 2000

67. McDonald KJ, Cameron AJM, Allen JM, Jardine AG: Expres-50. Rifai A, Fadden K, Morrison SL, Chintalacharuvu KR: Thesion of Fc �/� receptor by human mesangial cells: A candidateN-glycans determine the differential blood clearance and hepaticreceptor for immune complex deposition in IgA nephropathy. Bio-uptake of human immunoglobulin (Ig)A1 and IgA2 isotypes. J Expchem Biophys Res Commun 290:438–442, 2002Med 191:2171–2182, 2000

68. Tomana M, Zikan J, Moldoveanu Z, et al: Interactions of cell-51. Grellier P, Sabbah M, Fouqueray B, et al: Characterization ofsurface galactosyltransferase with immunoglobulins. Mol Immunolinsulin-like growth factor binding proteins and regulation of30:265–275, 1993IGFBP3 in human mesangial cells. Kidney Int 49:1071–1078, 1996

69. Moldoveanu Z, Epps JM, Thorpe SR, Mestecky J: The sites of52. Gomez-Guerrero C, Gonzalez E, Hernando P, et al: Interac-catabolism of murine monomeric IgA. J Immunol 141:208–213, 1988tion of mesangial cells with IgA and IgG immune complexes:

70. Moldoveanu Z, Moro I, Radl J, et al: Site of catabolism of autolo-a possible mechanism of glomerular injury in IgA nephropathy.gous and heterologous IgA in non-human primates. Scand J Immu-Contrib Nephrol 104:127–137, 1993nol 32:577–583, 199053. Gomez-Guerrero C, Lopez-Armada MJ, Gonzalez E, Egido J: 71. Phillips JO, Russell MW, Brown TA, Mestecky J: Selective

Soluble IgA and IgG aggregates are catabolized by cultured rat hepatobiliary transport of human polymeric IgA in mice. Mol Im-mesangial cells and induce production of TNF-� and IL-6, and munol 21:907–914, 1984proliferation. J Immunol 153:5247–5255, 1994 72. Phillips JO, Komiyama K, Epps JM, et al: Role of hepatocytes in

54. Pugliese G, Pricci F, Locuratolo N, et al: Increased activity of the uptake of IgA and IgA-containing immune complexes in mice.the insulin-like growth factor system in mesangial cells cultured Mol Immunol 25:873–879, 1988in high glucose conditions. Relation to glucose-enhanced extracel- 73. Tomana M, Phillips JO, Kulhavy R, Mestecky J: Carbohydrate-lular matrix production. Diabetologia 39:775–784, 1996 mediated clearance of secretory IgA from the circulation. Mol

55. Conley ME, Cooper MD, Michael AF: Selective deposition of Immunol 22:887–892, 198574. Russell MW, Brown TA, Mestecky J: Role of serum IgA:immunoglobulin A1 in immunoglobulin A nephropathy, anaphy-

Hepatobiliary transport of circulating antigen. J Exp Med 153:968–lactoid purpura nephritis, and systemic lupus erythematosus. J Clin976, 1981Invest 66:1432–1436, 1980

75. Russell MW, Brown TA, Kulhavy R, Mestecky J: IgA-mediated56. Gonzales-Cabrero J, Egido J, Mampaso F, et al: Characterizationhepatobiliary clearance of bacterial antigens. Ann N Y Acad Sciof circulating idiotypes containing immune complexes and their409:871–972, 1983presence in the glomerular mesangium in patients with IgA ne-

76. Stockert RJ, Kressner MS, Collins JD, et al: IgA interactionsphropathy. Clin Exp Immunol 76:204–209, 1989with the asialoglycoprotein receptor. Proc Natl Acad Sci USA 79:57. van den Wall Bake AWL, Bruijn JA, Accavitti MA, et al: 6229–6231, 1982

Shared idiotypes in mesangial deposits in IgA nephropathy are 77. Patry C, Sibille Y, Lehuen A, Monteiro RC: Identification ofnot disease-specific. Kidney Int 44:65–74, 1993 Fc� receptor (CD89) isoforms generated by alternative splicing

58. Allen AC, Willis FR, Beattie TJ, Feehally J: Abnormal IgA that are differentially expressed between blood monocytes andglycosylation in Henoch-Schonlein purpura restricted to patients alveolar macrophages. J Immunol 156:4442–4448, 1996with clinical nephritis. Nephrol Dial Transplant 13:930–934, 1998 78. Kitamura T, Garofalo RP, Kamijo A, et al: Human intestinal

59. Iwase H, Tanaka A, Hiki Y, et al: Aggregated human serum epithelial cells express a novel receptor for IgA. J Immunol 164:5029–5034, 2000immunoglobulin Al induced by neuraminidase treatment had a