Kidney International, Vol. 58 (2000), pp. 2331–2340

CELL BIOLOGY – IMMUNOLOGY – PATHOLOGY

Integrin expression and IgA nephropathy: In vitro modulationby IgA with altered glycosylation and macromolecular IgA

LICIA PERUZZI, ALESSANDRO AMORE, PAOLA CIRINA, LIVIO TRUSOLINO,GIUSEPPE BASSO, EMANUELA RICOTTI, STEVEN NOEL EMANCIPATOR,PIER CARLO MARCHISIO, and ROSANNA COPPO

Nephrology and Dialysis Department, Regina Margherita Children’s Hospital, Torino, Department of Biologicaland Technological Research, San Raffaele Scientific Institute, Milano, Department of Biomedical Sciences andHuman Oncology, and Pediatric Clinic, University of Torino, Torino, Italy; and Institute of Pathology, CaseWestern Reserve University, Cleveland, Ohio, USA

Integrin expression and IgA nephropathy: In vitro modulation tion of changes in the cellular environment and the inte-by IgA with altered glycosylation and macromolecular IgA. gration of such signals within MCs evokes enzymatic

Background. Signal transduction by mesangial cell (MC) modifications of the mesangial matrix [2]. The cross-talkintegrins regulates cell growth and survival, extracellular matrixbetween MCs and the surrounding matrix is mediatedproduction, and organization. The aim of the study was to investi-by specific proteins, collectively called integrins. Integ-gate human MC integrin modulation by differently glycosylated

IgA and macromolecular IgA, which are thought to play a patho- rins are heterodimers of a and b subunits that form, ingenetic role in IgA nephropathy (IgAN). various combinations, a large number of receptors that

Methods. MCs were incubated with purified human poly-are capable of binding one or more ligands [3, 4]. Eachmeric IgA, heat-aggregated IgA, IgA glycoforms generated byintegrin displays a ligand-binding site that recognizes spe-enzymatic hydrolysis of saccharide residues and serum fractions

from IgAN patients, and controls isolated by lectin affinity and cific amino acid sequences of several extracellular matrixcontaining IgA with peculiar glycan patterns. Integrins were proteins, including fibrinogen, fibronectin, thrombospon-quantitated by flow cytometry. din, vitronectin, laminin, and the collagens. The bindingResults. Cultured MCs highly expressed avb3 and some a3b1; to the appropriate extracellular matrix activates the cyto-avb3 was up-regulated by matrix components (P , 0.02). In

plasmic domain of the b integrin, producing a cascade ofvitro desialylated and degalactosylated polymeric human IgAenhanced avb3 expression on cultured MCs (P , 0.001). Serum signals that influences a wide range of cellular functions,IgA glycoforms isolated from IgAN patients with high expo- spanning from reorganization of the cytoskeleton to con-sure of internal sugars, GalNAc, Neu5Ac2,6GalNAc, and Man trol of gene expression [reviewed in 5]. In particular,enhanced av expression on cultured MCs more than healthy

integrin-mediated signal transduction modulates the path-controls.ways controlling cell growth and survival. Cell adhesionConclusions. These data support the hypothesis that IgA gly-

cation plays a role in modulating the cell–matrix interaction, and spreading activate mitogen-activated protein kinaseand that this mechanism can be operating in IgAN. via Ras-dependent and -independent pathways, with con-

sequent cell cycle modulation [6, 7]. On the other hand,whereas engagement of integrin with the matrix pro-

Mesangial cell (MC) activity modulates the develop- motes cell proliferation, in the absence of ligand, integ-ment and evolution of IgA nephropathy (IgAN) [1]. The rins block cell proliferation and down-regulate the ex-MC activation, likely triggered by IgA deposition in the pression of c-fos and c-jun early genes [6–8].mesangium, is closely regulated by interactions between Human MCs in culture express a3b1 and avb3 integrinsthese cells and their environment. In turn, the transduc- [2, 9–12]. These transmembrane proteins are likely to

play a pivotal role in several critical functions of MCs,modulating cell growth, survival, and differentiation [13].

Key words: degalactosylated IgA, apoptosis, cell-matrix interaction,The b1-mediated binding of MCs to collagens or fibro-extracellular matrix, transmembrane proteins, mesangial matrix, glo-

merular hemodynamics. nectin may (1) modify the synthetic rate, catabolism, andorganization of mesangial matrix; (2) enhance the syn-Received for publication April 28, 1999thesis of enzymes, which degrade mesangial matrix; andand in revised form March 31, 2000

Accepted for publication June 9, 2000 (3) promote chemoattraction of circulating monocytesand fibroblasts. Finally, by anchoring the membrane and 2000 by the International Society of Nephrology

2331

Peruzzi et al: Integrins in IgAN2332

the cytoskeleton of MCs to the extracellular matrix, in- human to laboratory animal species, which are providedwith marked variation in IgA biology [16].tegrins may participate in linking the contraction of MCs

Herein, we speculate on the possible role of integrinto the regulation of glomerular hemodynamics and themodulation by aberrantly glycosylated IgA in the devel-filtration properties of the glomerular capillaries [2].opment and evolution of IgAN.The expression of several b1 integrins (a1, a2, a3, a5,

and a8) was demonstrated not only in cultured MCs butalso in normal human kidney tissue [13]; however, a3b1 METHODSwas found by immunoelectron microscopy exclusively

Experimental planon visceral and capsular epithelial cells. Similarly, avb3

The expression of integrins on MCs after incubationwas detected in normal kidney mostly along the glomeru-with each of several selected stimuli was compared withlar capillary wall. Since some avb3 reactivity also wasthat on MCs maintained in a standard cultured medium.found in the mesangial area and the avb3 ligand, vitronec-

In a first series of experiments, MC culture mediumtin, was seen exclusively in the mesangium, interest haswas implemented with 10 to 30 mg/mL of purified humanrecently been focused on MC avb3 integrin [13].fibronectin (Sigma Chemical Co., St. Louis, MO, USA),Several functional roles of mesangial integrins, includ-thus increasing the fibronectin concentration of cultureing the transmission of contractile forces to the glomeru-medium by 0.5 to 1.5 times. The effects of supplementa-lar basement membrane and the regulation of cell prolif-tion of the culture medium with human type IV collageneration and matrix metabolism, may be relevant for the10 mg/mL were examined (Sigma). Several preparationsdevelopment and evolution of IgAN. Accordingly, ourof purified human IgA, including polymeric IgA (Sigmastudy was aimed to explore mesangial integrins in IgAN,from human cholostrum) and heat-aggregated polymericfocusing on factors modulating avb3 expression.IgA, prepared by heating the polymeric IgA 20 mg/mLGlycosylation of the protein chains greatly influencesat 638C for 120 minutes [26], each at a concentration ofthe functional properties of adhesion molecules, including50 mg/mL were employed as test stimuli in a subsequentintegrins and their matrix ligands [14, 15]. On the otherseries of experiments. Moreover, abnormal glycoformshand, IgAs are highly glycosylated molecules [16]. Carbo-of IgA, rich in terminal Gal/Gal NAc or in GalNAc boundhydrates represent 12% in serum IgA and 15% in secre-in 2,6 to Neu5Ac, typical sugars of O-linked chains, or intory IgA. Human IgA1 contains two potential N-linkedMan, typical sugar of N-linked chains, were isolated fromglycosylation sites and five O-linked sites in the hingeIgAN patients’ sera by sequential lectin affinity and gelregion of each chain. Each O-glycan has a core N-acetyl-filtration chromatography (vide infra). In these experi-galactosamine (GalNAc) linked in b1,3 to galactose (Gal)ments, parallel MC cultures were incubated with Neu5Ac,

and with one or two sialic acid (Neu5Ac) units in a2,3 GalNAc, or Man, each at a 0.1 mol/L concentration.linkage with Gal or a2,6 linkage with GalNAc [17]. IgA2 Finally, in a third round of experiments, abnormallacks these O-linked sugars, but the IgA2m allotype has glycoforms of IgA were generated enzymatically by treat-two extra N-linked sites [17]. ment of normal human polymeric IgA with selected endo-

During the last years, several groups have demon- glycosidases (vide infra). The desialylation and degalac-strated an aberrant glycosylation of serum IgA in IgAN tosylation were not done in patient samples because ofpatients [18–20]. We previously found that IgA from the limited amount of serum available. In all cases, stim-patients with IgAN has increased binding to some lectins uli tested by the LAL test (standard Escherichia coli[18] and to matrix components [19], and that this binding III:B4; Kabi, Stockholm, Sweden) to ensure endotoxin-was due to carbohydrate interactions [20]; moreover, free conditions were added to the same supplementedwe found that these properties characterize transplanted culture medium used to maintain the cell in culture (dis-IgAN patients with recurrency of IgAN [21]. In patients cussed later in this article), in a humidified atmospherewith IgAN, a reduced Gal and/or sialic acid content containing 5% CO2 and incubated for four hours. Eachhas been demonstrated by increasing binding to specific experiment included cells similarly cultured and incu-lectins (Vicia Villosa) or increased asialo chains [21–23]. bated in the same medium and atmosphere, with noA defective b1–3 galactosyltransferase activity has been additions (basal conditions).detected in B lymphocytes from IgAN patients [24]. Human IgG (50 mg/mL, Cohn fraction II; Sigma) was

The functional significance of IgA altered glycosyla- tested as well.tion and its potential to create mesangial injury is still

Enzymatic modification of IgA carbohydrate sideunexplored. We chose integrin expression as a clue markerchain residuesof MC–matrix interactions since their reciprocal interac-

tions can regulate cell metabolism, proliferation, and To remove terminal sialic acid and expose the Galdifferentiation [25]. We employed human MCs and hu- residues, which become terminal, human IgA was incu-

bated for six hours at 378C with neuraminidase (fromman IgA to avoid differences in data interpretation from

Peruzzi et al: Integrins in IgAN 2333

Clostridium perfrigens; Sigma), using 0.1 U of enzyme umns (Economopack; Bio-Rad Laboratories, Hercules,per mg of IgA in 25 mmol/L potassium acetate buffer CA, USA) and washed with 0.1 mol/L PBS, pH 7.4.containing 0.1% bovine serum albumin (BSA), pH 4.5 Each of the five columns was loaded with 800 mL of[27]. The effluent from a column of Sepharose cross- serum from four patients and four healthy controls. Thelinked to Sambucus nigra lectin (Vector Laboratories, efflux was clamped, and the column was incubated for oneInc., Burlingame, CA, USA) was deficient in terminal hour at room temperature. The columns were washedNeu5Ac linked to Gal residues. This purification step for approximately three hours with flowing PBS, untilmight leave some residual sialylation of GalNAc. the OD at 280 nm was negligible. Finally, the bound glyco-

A portion of the neuraminidase-treated IgA, nonad- proteins were eluted from the columns by washing withherent to Sambucus nigra column, was subsequently in- solutions of the appropriate monosaccharide specific forcubated for 12 hours at 378C with b-galactosidase (from each lectin: 0.1 mol/L Man, 0.1 mol/L GalNAc, 0.05 mol/Lbovine testis; Sigma) to remove terminal Gal and to Neu5Ac. The eluates were collected in 0.5 mL fractions,expose GalNAc, which becomes terminal, at a ratio of and the absorbance values at 280 nm were considered0.1 U of enzyme per 200 mg of IgA in 0.2 mol/L sodium to isolate the main peak of eluted proteins. The mainphosphate, 0.1 mol/L citrate buffer containing 0.1% peak in each column eluate was fractionated by high-BSA, pH 4.5 [28]. The IgA bearing residual terminal performance liquid chromatography (HPLC) on a BioSilgalactose was removed by affinity chromatography using column (Biorad; Bio-Rad Laboratories), which separatesSepharose cross-linked with Ricinus communis (Vector molecules in the 150 to 500 kD molecular mass range.Laboratories), which recognizes terminal Gal. The isolated peaks at 250 to 500 kD in IgAN patients and

in controls were separately pooled for MC conditioning.Verification of IgA-bound oligosaccharide terminationEluted material was assayed for IgA content by sandwich

Enzyme-linked immunosorbent assay (ELISA) plates ELISA. The same peak of purified polymeric IgA waswere coated with 10 mg/mL of unmodified or enzyme- also resolved by HPLC as a control. In all cases, the IgAtreated IgA in 0.015 mol/L carbonate buffer, pH 9.6, and concentration was adjusted before incubating the HPLCunreacted sites were blocked with 1% BSA-phosphate- fractions with MC cultures.buffered saline (PBS). After overnight incubation at 48C,plates were washed with 0.1% BSA, 0.05% Tween 20, and Cell culturesperoxidase-labeled Arachis hypogea (PNA)—specific for Human glomeruli were isolated from human corticalthe complete, galactosylated Galb1,3GalNAc—or Vicia

fragments obtained from uninvolved tissue of kidneysvillosa (VV)—specific for ungalactosylated GalNAc—

removed for cancer, and MCs were cultured as pre-lectins (5 mg/mL) both from Sigma, were added for one

viously described [29]. Cells were grown in RPMI 1640hour at room temperature. Next, after three washes,(Sigma) supplemented with 20% fetal calf serum (Sigma)—the enzyme substrate phenylendiamine dihydrochloridefinal concentration 20 mg/mL fibronectin—2 mmol/L glu-(Sigma) 0.4 mg/mL in 0.03% vol/vol hydrogen peroxide,tamine, 50 U/mL penicillin, and 50 mg/mL streptomycin0.1 mol/L disodium hydrogen phosphate, and 0.05 mol/Lin a 5% CO2 humidified atmosphere. Before employ-citric acid monohydrate were added. The results of thisment, cells were checked for factor VIII and cytokeratincolorimetric reaction were read at 490 nm at the end ofnegativity and actin-positive staining.the linear phase.

Antibodies to integrinsIgA glycoforms from sera of patients with IgANThe primary monoclonal antibodies (mAbs) used inFractions from sera of IgAN patients containing IgA

this study were anti-a3 (F2; donated by L. Zardi, IST,with various glycosylation patterns were isolated by af-Genoa, Italy), anti-b1 (MAR4; donated by S. Menard,finity chromatography using columns of lectins immobi-Istituto Nazionale Tumori, Milan, Italy), anti-b3 (VIPl-2;lized on Sepharose. Sepharose 4B CNBr-activated col-a gift of W. Knapp, Institute of Immunology, Universityumns (Pharmacia, Uppsala, Sweden) were prepared asof Vienna, Austria), anti-b4 (AA3; donated by V. Qua-recommended by the manufacturer. Selected lectins wereranta, Scripps Research Institute, La Jolla, CA USA),cross-linked to Sepharose by incubating in 0.1 mol/Lanti-a2 (Gi9; Immunotech, Marseille, France), anti-a5NaHCO3, 0.5 mol/L NaCl buffer at 48C for 12 hours.(SAM-1; Immunotech), anti-b5 (IA9; from M. Hemler,The lectins employed included the following: CanavaliaDana Farber Cancer Institute, Boston, MA, USA), anti-avensiformis (Con A), specific for mannose (Man) group;(P3G8; Chemicon, Temecula, CA, USA) and anti-avb3Glycine maximus (SBA), specific for N-acetylgalactos-(MS 3 Hu integrin av, recognizing the avb3 receptor;amine/galactose (GalNAc-Gal) group; and Limulus poly-Chemicon). As control, irrelevant antibodies were usedphemus (LPA), specific for sialic acid (Neu5Ac 2,6 Gal

NAc) groups. The gel beads were packed in 5 mL col- (Table 1).

Peruzzi et al: Integrins in IgAN2334

Table 1. Monoclonal antibodies employed in this study

Clone Specificity Source

F2 Anti-a3 L. Zardi, IST, Genoa, ItalyMAR4 Anti-b1 S. Menard, Istituto Nazionale Tumori, Milan, ItalyVIPI-2 Anti-b3 W. Knapp, Immunology Institute, University of Vienna, AustriaAA3 Anti-b4 V. Quaranta, Scripps Research Institute, La Jolla, CA, USAGi9 Anti-a2 Immunotech, Marseille, FranceSAM-1 Anti-a5 Immunotech, Marseille, FranceIA9 Anti-b5 M. Hemler, Dana Farber Cancer Institute, Boston, MA, USAP3G8 Anti-av P3G8, Chemicon, Temecula, CA, USAMS3Hu Anti-avb3 MS3Hu integrin-av, Chemicon, Temecula, CA, USA

Immunofluorescence USA) set to analyze 5000 cells per sample. Data wereelaborated by the Immuno-4 program (Coulter).After detachment with 0.05% trypsin-0.02% ethylene-

diaminetetraacetic acid (EDTA; Sigma), 5 3 104 cellsQuantitation of integrin receptors

were plated into each well of 24-well plates (CostarTo quantitate the average number of integrin recep-Corp., Cambridge, MA, USA) containing 1.4 cm2 round

tors per MC, the Quantum (TM) 27 R-phycoerythrin (PE)-glass cover slips. Cells were incubated for eight hours inconjugated microbead kit (Flow Cytometry Standardcomplete medium; next, after several washes in PBS,Corp., San Juan, Puerto Rico) was used [31]. This containsthey were fixed in 3% paraformaldehyde, 2% sucrose ina set of calibrated standards, with four populations of PE-PBS, pH 7.6, for 10 minutes at room temperature. Afterconjugated microbeads displaying graded and predeter-rinsing in PBS, cells were permeabilized by soaking covermined fluorescence intensity [expressed in terms of theslips for three to five minutes at 08C in HEPES-Tritonnumber of molecules of equivalent soluble fluorochromesX-100 buffer (20 mmol/L HEPES, pH 7.4, 300 mmol/L(MESFs)] and one reference blank population. Using thissucrose, 50 mmol/L NaCl, 3 mmol/L MgCl2, and 0.5%method, the relative channel number obtained by flowTriton X-100). Indirect immunofluorescence was per-cytometry analysis of a cell population is directly trans-formed as previously reported [30]. Briefly, the primaryformed into the number of MESFs. The linear regressionAb (at the concentration of 10 mg/mL) was layered ontoequation, correlating the channel number with the spe-the fixed and permeabilized cells and incubated in acific MESF value, was calculated using specific softwarehumid chamber for 30 minutes at 378C. After rinsing,(Quickcal V2.0; Flow Cytometry Standard Corp.). Resultscover slips were incubated in the appropriate rhodamine-express the mean MESFs 6 SD of four experiments.tagged secondary antibody (Dakopatts, Copenhagen, Den-The number of MESFs corresponds to the number ofmark) for 30 minutes at 378C in the presence of 2 mg/mL offluorescent molecules detected on the surface of MCs,fluorescein-labeled phalloidin (Sigma). Cover slips wereindicating the number of integrin receptors.mounted in Mowiol 4-88 (Hoechst AG, Frankfurt/Main,

Germany) and observed in a Zeiss Axiophot photomi-Indirect immunoperoxidase techniquecroscope equipped with epifluorescence lamp and plana-

Mesangial cells, grown in chamber slides, were stainedpochromatic oil immersion lenses. Fluorescence imagesusing a standard protocol. MCs in basal conditions andwere recorded on Kodak T-Max 400 film exposed atconditioned by aberrantly glycosylated IgA were snap1000 ISO and developed in T-Max Developer for 10frozen in isopentane precooled in liquid nitrogen. En-minutes at 208C.dogenous peroxidase activity was inhibited by incubation

Flow cytometry analysis for 30 minutes in 0.025% H2O2 in methanol. Endogenousbiotin activity was inhibited by sequential 30-minute ex-For this analysis, 5 3 104 cells were allowed to adhereposures to avidin D and biotin blocking solutions (Vectorto 75 cm2 flasks for eight hours at 378C; then the mediumLaboratories Inc.). Following another wash, cells werewas substituted with complete medium containing oneincubated in 0.005% biotin in PBS, washed again, andof the various stimuli noted previously in this article.then overlaid with 10 mg/mL primary antibody (anti-a5,Flasks were incubated for four hours at 378C, and thenanti-b1, anti-b5, and anti-avb3) for 30 minutes in a mois-cells were detached in 1 mmol/L EDTA, washed in PBS,ture chamber at room temperature. Slides were rinsedand incubated with the primary Ab (10 mg/mL) at 48Cin PBS, incubated for 30 minutes at room temperaturefor 30 minutes. After two washes with PBS, MCs werewith biotinylated rabbit anti-mouse IgG, rinsed again,incubated at 48C for 30 minutes with a FITC-rabbit anti-and incubated in avidin-biotin-peroxidase complex (Vec-mouse antibody (Biosoft, Varilhes, France), rinsed intor Laboratories Inc.). Finally, slides were developed forPBS, and analyzed after fixation with 1% paraformalde-

hyde, in an EPICS XL counter (Coulter, Hialeah, FL, 5 to 10 minutes with a solution of 0.075% diaminobenzi-

Peruzzi et al: Integrins in IgAN 2335

Fig. 1. Immunofluorescence staining of cul-tured human mesangial cells (MCs). The cellswere stained with mouse monoclonal antibod-ies to av, followed by incubation with an appro-priate rhodamine-tagged secondary antibody.The integrin chains were found in patches atthe end of actin fibrils, counterstained withfluoresceinated phalloidin. These clusters ofav correspond to focal contacts. In other views(not shown), av colocalizes with vinculin. Orig-inal magnification 3630.

dine and 0.015% H2O2 in 0.05 mol/L Tris. Cells were mg/mL aggregated IgA. We observed that av expressionpeaked between four and six hours of incubation withcounterstained with Meyer’s hematoxylin (Zymed, San

Francisco, CA, USA), mounted in Cleariumt (Surgi- all stimuli, and no modifications were observed in thefollowing hours until 24 hours. Hence, an incubationpath, Canada), and examined with a Zeiss Axiophot

photomicroscope equipped with 316 and 363 plana- time of four hours was adopted in each experiment.Integrins in cultured MCs were quantitated in flowpochromatic lenses. To control for specificity of the im-

munoperoxidase staining, parallel MC preparations were cytometry by means of calibration beads, correlating thefluorescence intensity with the number of fluorochromesoverlaid with 10 mg/mL of a murine monoclonal IgG

antibody specific for Ewing’s sarcoma cells (Signet Labo- (MESFs), hence expressing the number of cell surfacemolecules. This technique has been proposed to quanti-ratories, Dedham, MA, USA) as a primary antibody and

subsequently were incubated with all of the same steps tate the receptor molecules/cell, providing data superim-posable to conventional radiobinding assays [31]. Evenused to detect the specific primary antibody.considering the basal value of MESFs in resting MCs as

Statistical analysis an approximation of the true number of integrins/cell,the data allowed the evaluation of the modulation in-Values are expressed as means 6 SD. A comparison

between group means was performed using Student’s duced by aberrantly glycosylated and macromolecularIgA, which was the aim of the present study.t-test. P values of less than 0.05 were considered statisti-

cally significant. Pearson’s correlation and linear regres- The intensity of av expression on MCs increased aftersion analysis were used to calibrate receptor analysis by incubation with mesangial matrix components. After in-flow cytometry, as detailed previously in this article. cubation with fibronectin or type IV collagen, the modal

fluorescence intensity and the mean channel numbershifted to the right on a log scale (Fig. 2A) relative toRESULTSthe modal fluorescence intensity and the mean channel

Integrin expression in cultured MCs and modulation number seen with cells in medium alone (basal). Simi-of the av chain by purified matrix proteins and IgA larly, the addition of either form of IgA to the culture

medium intensified the expression of av on MCs (Fig.Immunofluorescence staining of cultured human MCsshowed strong expression of av (Fig. 1), b3, and a3 integrin 2B). The calculated number of av receptors per cell (ex-

pressed in MESFs) was nearly tripled after incubationchains, and to a lesser extent, b1, trace amounts of a5

were detected. Conversely, MCs were negative for a2, with fibronectin or collagen: basal MCs had 822 6 55MESFs/cell; MCs conditioned with fibronectin 0.5 timesb4, and b5 (data not shown). Flow cytometry revealed

that 95 6 0.7% of cultured MCs were positive for av in the basal conditions had 2426 6 498 MESFs/cell (P ,0.025 vs. basal), and MCs incubated with type IV collagenassociation with b3. In preliminary studies, we performed

a time course of av expression on MCs after incubation had 2217 6 401 MESFs/cell (P , 0.025 vs. basal). Thesupplementation of culture medium with fibronectinwith 10 mg/mL of mesangial matrix components or 50

Peruzzi et al: Integrins in IgAN2336

Fig. 2. Flow cytometry analysis of av expres-sion on human cultured mesangial cells (MCs).(A) Cells were studied in unconditioned me-dium (basal) and after supplementation of cul-ture media with type IV collagen (Coll IV) orfibronectin (both 10 mg/mL). (B) Cells underbasal conditions are compared with cells pre-incubated with either polymeric IgA (pIgA)or aggregated polymeric IgA (aIgA). For eachof the three superimposed histograms, thenumber of cells at each level of fluorescenceis plotted on an ordinate against the same ab-scissas log fluorescence intensity. The mostfrequent (peak) fluorescence intensity underbasal conditions is 0.66 (mean channel number1.08). Cells preincubated with either extracel-lular matrix component (A) showed peak flu-orescence levels of 1.9 (collagen IV) or 1.8 (fi-bronectin; mean channel number 2.25 and 2.38,respectively), whereas incubation with eitherpreparation of IgA (B) resulted in peak fluo-rescence levels of 2.0 (pIgA) or 1.4 (aIgA; meanchannel number 1.85 and 2.04, respectively).

concentrations up to 1.5 times the usual medium content value was higher than what was observed with serumfractions isolated from healthy controls (536 6 11 MESFs/led to a further increase in av receptors per cell (2850 6

432 MESFs/cells). cell). It should be considered that the LPA affinity chro-matography separates Neu5Aca2,6GalNAc-rich mole-Mesangial cells conditioned with polymeric IgA had

1715 6 233 MESFs/cell, and with aggregated IgA 1961 6 cules, hence also IgA molecules truncated because ofdeficient b1,3 Gal transferase [24].224 MESFs/cell (both P , 0.01 vs. basal; Fig. 3). The

Serum fractions from patients with IgAN eluted fromculture of MCs with 50 mg/mL aggregated IgA in addi-Con A columns by Man, and therefore rich in IgA glyco-tion to 10 mg/mL fibronectin further increased the integ-form presenting terminal Man residues, increased therin expression in comparison to aggregated IgA aloneexpression of av integrins by cultured MCs (1934 6 261(2552 6 235 MESFs/cell).MESFs/cell, P , 0.02 vs. basal) and provided data sig-

Differential modulation of av by distinct nificantly higher than serum fractions similarly isolatedglycoforms of IgA from healthy controls (1004 6 22 MESFs/cell, P , 0.05

vs. IgAN data). The coincubation of MCs with Man alsoSerum IgA glycoforms isolated in IgAN patients bySBA affinity chromatography and hence exposing termi- augmented av density (1690 6 153 MESFs/cell, P , 0.02

vs. basal; Fig. 4).nal GalNAc residues increased the expression of av in-tegrin on cultured MCs (1741 6 335 MESFs/cell, P , To examine the role of the protein-bound oligosaccha-

rides in the modulation of av integrin expression by IgA0.05 vs. basal; Fig. 4). This effect was significantly higherthan what was observed after incubation of serum fractions further, we generated distinct glycoforms of IgA by enzy-

matic hydrolysis of selected saccharide residues fromfrom healthy controls (953 6 33 MESFs/cell vs. IgAN,P , 0.05). Free GalNAc had an enhancing effect (2660 6 purified normal polymeric human IgA.

The effects of enzymatic hydrolysis of sialic acid and171 MESFs/cell, P , 0.005 vs. basal; Fig. 4). Conversely,free sialic acid strongly depressed av expression on MCs galactose to generate desialylated and desialylated/dega-

lactosylated IgA were confirmed by reciprocal changes(256 6 6 MESFs/cell, P , 0.05). The fractions isolated inIgAN patients by LPA, specific for terminal Neu5Ac2,6 in the binding of the treated IgA to PNA and to VV

lectins. Neuraminidase-treated IgA bound 2.3-fold morebound to GalNAc, failed to significantly modify the num-ber of integrins/MCs in comparison to basal conditions PNA and 2.6-fold less VV than IgA treated sequentially

with neuraminidase and b-galactosidase.(752 6 254 MESFs/cell, P 5 NS; Fig. 4); however, this

Peruzzi et al: Integrins in IgAN 2337

Fig. 3. Quantitation of av integrin chains oncultured human MCs by flow cytometry. Cellsmaintained in unconditioned medium (basal)or after incubation with polymeric IgA (pIgA),aggregated IgA (aIgA), in vitro desialylated(deSia) IgA, or desialylated/degalactosylated(deSia/deGal) IgA were studied by flow cy-tometry, and the number of MESFs/cell, indi-cating the number of sites/cell calculated asdetailed in the text. Data are the mean 6SD of four separate experiments. Note thatpolymeric IgA promote a doubling of av den-sity on the cell surface, while the macromolec-ular aggregation does not further increase theintegrin expression. Desialylated IgA inducean increase in av density similar to polymericIgA. Conversely, desialylated/degalactosylatedIgA promote a striking increase in av expres-sion, which was significantly higher than afterincubation with polymeric IgA (†P , 0.01).*P , 0.025 vs. basal conditions. **P , 0.001vs. basal conditions.

IgA desialylated in vitro increased the expression ofav on cultured MCs, as observed by immunoperoxidasestaining, and desialylated/degalactosylated IgA stronglyenhanced such expression (Fig. 5). The number of av

receptors per cell significantly increased to 3850 6 143MESFs after incubation of MCs with desialylated/dega-lactosylated IgA (P , 0.001 vs. basal conditions and P ,0.01 vs. untreated polymeric IgA; Fig. 3).

The coincubation of MCs with 50 mg/mL of desialy-lated IgA or desialylated/degalactosylated IgA in addi-tion to 10 mg/mL fibronectin failed to increase integrinexpression further, as it was blunted in comparison towhat found by incubating one probe at once (desialylatedIgA plus fibronectin, 16% decrease of desialylated IgAalone; desialylated/degalactosylated IgA in addition tofibronectin, 49% decrease of desialylated/degalactosy-lated IgA alone).

Fig. 4. Effect of glycan on the density of av integrin chains on cultured Human IgG was ineffective in modifying the integrinhuman MCs by flow cytometry. Cells were studied in unconditioned expression on MCs.medium (basal) or with fractions of IgA from the serum of IgANpatients and healthy controls isolated by affinity chromatography onGlycine maximus lectin (IgA-SBA) or lectin from Limulus polyphemus

DISCUSSION(IgA-LPA), or on Canavalia ensiformis (IgA-ConA) or with an appro-priate simple sugar such as N-AcetylGalactosamine (GalNAc), N-acetyl Human MCs in culture strongly express avb3 and, at aneuraminic acid (Neu5Ac), or mannose (Man). Exposure of MCs to

lower intensity, a3b1 receptors for extracellular matrix [2].IgA, which contains terminal GalNAc (IgA-SBA), or terminal Man(IgA-ConA) or to the corresponding simple sugars, elicits a doubling Shikata et al reported markedly increased expression ofor trebling of the expression of av receptors on MCs, compared with

b1, av, and b3 chains in renal tissue from IgAN patientscells under basal conditions (data are expressed in MESF/cell as in Fig. 3).relative to normal kidneys, in which only trace amounts ofFractions in patients were more effective in eliciting av integrin expres-

sion than those similarly isolated in controls. Statistical differences these integrins are detectable by immunofluorescence [12].versus *basal conditions and †fractions isolated in healthy controls, as Unpublished observations of our group were largely con-detailed in the text. Symbols are: (h) sugar; ( ) IgAN; (j) healthycontrols. firmatory (abstract; Emancipator et al, J Am Soc Nephrol

Peruzzi et al: Integrins in IgAN2338

on the modulation of MC av chain, which coprecipitateswith b3 to form the major receptor for vitronectin, fibro-nectin and fibrinogen, thrombospondin, and von Wille-brand factor, recognizing a specific RGD sequence [3, 4].Integrin-ligand binding and activation initiate the intra-cellular signals that regulate adhesion, spreading, and mi-gration of several distinct cell lineages, including MCs [5–8].Coincubation of MCs with supraphysiological amountsof matrix components enhanced a significant up-regula-tion of av expression. A different reactivity could havebeen expected for fibronectin, which is a natural ligandfor avb3 and collagen IV, which selectively binds b1 integ-rins [4]. However, integrins can be modulated by path-ways independent from receptor ligation, as by cytokines[34, 35]. Hence, the binding of collagen IV to b1 integrins,by activating the cytoplasmic tail, is expected to regulatethe formation of focal contacts and stress fibers for sus-tained adhesion, which might lead to hyperexpressionof other integrins [34].

Focusing on the modulation of MC integrins by aber-rantly glycosylated IgA and macromolecular IgA, weemployed human cells and immunoglobulins in order toavoid variations in IgA biology from laboratory animalsto humans [reviewed in 16]. We used human secretoryIgA isolated from cholostrum, which is the major sourceof almost pure polymeric IgA (96% of secretory IgA arepolymeric) [16]. Polymeric IgAs are increased in circula-tion in IgAN patients and are dominant in mesangialIgA deposits [36]. Moreover, polymeric IgA in form ofcomplexes reproduce in experimental rodents the IgANnephritogenesis [37]. Polymeric IgA and heat-aggregatedIgA were found to enhance strongly the reactivity ofmurine MCs [38, 39] and enhance in human MC nuclearfactor-kB (NF-kB) transcription [40]. Secretory IgAshave several properties of polymeric IgA, including thehigh capacity to aggregate [41] and to activate the com-plement alternative pathway [42]. Cholostral IgA is pro-

Fig. 5. Immunoperoxidase staining of av in cultured human MCs. Com- vided with additional glycated chain-linked sugars, whichpared with cells cultured under basal conditions (a) and cells preincu-might further enhance the consequences of IgA glycationbated with desialylated IgA (b), cells preincubated with enzymatically

generated asialo-agalactosyl-IgA (c ) exhibited clearly more intense [16]. On the other hand, it is unlikely that the use ofstaining for av. For these preparations, the appropriate reciprocal changes secretory IgA modify the potential significance of ourin IgA glycosylation were verified by lectin affinity assay: asialo-IgA

results in giving new insight into this aspect of IgANbound 2.3-fold better to Arachis hypogea lectin (specific for asialo-oligosaccharides) than doubly enzyme-treated IgA, whereas binding to nephritogenesis.Vicia Villosa lectin (specific for terminal GalNAc) was 2.6-fold less. We also tested heat-aggregated polymeric IgA, which

are likely to be of a very large size. These polymeric IgAaggregates possibly mimic large immune complexes, whichescape the hepatic clearance because they fail to enter8:84, 1997). Expanded mesangial regions were strongly

positive for avb3 and its ligands, fibronectin, and vitronec- the space of Disse and mimic deposited IgA rearrangedelectron-dense deposits.tin. Immunoelectron microscopy confirmed a selective

MC av expression associated with mesangial deposits We detected a clear enhancement of av by the interac-tion between macromolecular IgA and MCs. Besides,[12]. Other reports indicated increased avb3 mesangial

expression in highly proliferative IgAN [32], suggesting this effect was elicited also by the oligosaccharide com-position of the IgA. Fca receptors expressed on rat MCsa role of this integrin in phagocytosis of apoptotic cells

in the inflamed glomerulus [33]. are heavily glycosylated [38]. Moreover, in rodents, IgAinteracts with asialoglycoprotein receptors on MCs [43],Based on its possible involvement in IgAN, we focused

Peruzzi et al: Integrins in IgAN 2339

a surface lectin specific for galactose residues, suggesting It is tempting to speculate that the binding of aber-rantly glycosylated IgA with MC IgA-specific receptorthe possibility of interactions mediated by carbohydrate

bindings, including desialylated IgA. Indeed, in some enhances integrin expression as a result of a broad-spec-trum cell activation with production of transcriptionalstudies, desialylated IgA were found to bind more effi-

ciently than native IgA to rat MCs enhancing cell activa- factors, leading to the release of different mediators(Amore et al, manuscript submitted). The interaction oftion with release of oxygen free radicals [38]. In humans,

the matter is controversial, since human MCs constitu- IgA with Fca receptors of human MCs activates tran-scription factor NF-kB leading to enhanced proinflam-tively express Fca receptor, but this is not coincident

with the CD89 molecule expressed on human monocytes, matory cytokine production and superoxide generation[40]. In turn, several cytokines, including transformingand human MCs lack ASGPR [44]. Hence, the possible

interferences of aberrantly glycosylated IgA with its re- growth factor-b, basic fibroblast growth factor, platelet-derived growth factor, and epidermal growth factor, haveceptor are still speculative.

We found that some sugars represented in the inner been shown to modulate integrin expression in differentcell types [34, 47], hence supporting the possibility thatpart of the carbohydrate side chains of IgA (Man and

GalNAc) enhance av expression, while the external ter- the increased av integrin expression in MCs coincubatedwith aberrantly glycosydated IgA or macromolecularminal residue (Neu5Ac) depressed it. These observa-

tions were supported by the finding that in vitro desialy- IgA could be mediated as well by cytokine release.In conclusion, we found that truncation of terminallated/degalactosylated human polymeric IgA strongly

enhanced integrin expression. Of interest, the effect of sugars of IgA glycan chains as well as macromolecularIgA up-regulated the expression of avb3 receptors inthe removal of both sialic acid and Gal was much more

evident than the simple desialylation. As reported pre- cultured MCs. We speculated that integrin expressionmay modulate cell proliferation, apoptosis with regres-viously in this article, the recent literature suggests an

increased circulation in IgAN of IgA bearing O-linked sion of cell infiltration, and phagocytosis of IgA deposits,regulating mesangial hypercellularity and matrix accu-carbohydrate chains with greater exposure of GalNAc.

From our experiments, the exposure of GalNAc after the mulation. Moreover, it may favor MC contraction andtransmission of contractile forces to the capillary wall,treatment of various glycosydases was mostly effective in

increasing the number of avb3 receptors on MCs. modifying intraglomerular hemodynamics. Such proc-esses may play an important role in the pathogenesisWe observed that the coincubation of polymeric IgA

with fibronectin increased the av expression. Conversely, and evolution of IgAN.we detected a blunted effect when desialylated and desia-lylated/degalactosilated IgA were coincubated with fi- ACKNOWLEDGMENTSbronectin in MC culture medium. We hypothesized that This work was supported by MURST 40% and 60%, Telethon and

AIRC grants to P.C.M., and MURST 40% and 60% and AIRC tothe interaction in vitro of aberrantly glycosylated IgAG.B. The authors are greatly indebted to Dr. Gloria Gallo (Renalwith fibronectin, which is carbohydrate-dependent [45],Immunopathology, NY University Medical Center, New York, NY,

reduced the reactive carbohydrate residues available, USA) for making renal biopsy samples available in preliminary studiesand for critically reviewing the manuscript. The technical assistancehence reducing the effects on cultured MCs. The rele-of Ms. Nancy U. Nagy is gratefully acknowledged.vance of our observation should be further investigated

in IgAN since fibronectin is thought to favor the targeting Reprint requests to Rosanna Coppo, M.D., Divisione di Nefrologiae Dialisi, Ospedale Infantile Regina Margherita, Piazza Polonia 94,of IgA to the mesangium [46].10126 Torino, Italy.However, it should be considered that in vitro deglyco-E-mail: nefrologia@oirmsantanna.piemonte.it

sylation results in gross alterations in sialylation and ga-lactosylation, which are unlikely to occur in patients. For

REFERENCESthis reason, we also used a different approach, testing

1. Coppo R, Emancipator SN: Pathogenesis of IgA nephropathy:serum fractions rich in different IgA glycoforms isolatedEstablished observations, new insight and perspectives in treat-

in IgAN patients and controls. Even considering that the ment. J Nephrol 7:5–15, 19942. Cosio FG, Sedmak DD, Nahman NS: Cellular receptors for matrixfinal serum fraction isolated by lectin affinity chromatog-

proteins in normal human kidney and human MC. Kidney Intraphy and HPLC could contain other proteins complexed38:886–895, 1990

with IgA (for instance, fibronectin [45])—the meaning 3. Hynes RO: Integrins: Versatility, modulation, and signaling in celladhesion. Cell 69:11–25, 1992of our results remain unaffected, since the comparison

4. Ruoslahti E, Noble N, Kagami S, Border W: Integrins. Kidneyof data in IgAN patients and controls indicates a majorInt 45(Suppl 44):S17–S22, 1994

effect in enhancing integrin expression of IgA glyco- 5. Clark EA, Brugge JS: Integrins and signal transduction pathways:The road taken. Science 268:233–239, 1995forms isolated in patients.

6. Chen Q, Lin TH, Der JC, Juliano RL: Integrin mediated activa-It is clear that each investigational approach carriestion of MAPK or MEK is independent of Ras. J Biol Chem

some limitations; however, the convergent results more 271:18122–18127, 19967. Fang F, Orend G, Watanabe N, Hunter T, Ruoslathi E: Depen-convincingly support our conclusions.

Peruzzi et al: Integrins in IgAN2340

dence of cyclin E-cdk2 kinase activity on cell anchorage. Science 29. Amore A, Cavallo F, Bocchietto E, Bussolino F, Gianoglio B,271:499–502, 1996 Peruzzi L, Porcellini MG, Coppo R: Cytokine mRNA expression

8. Zheng DQ, Fornaro M, Bofetiado CJM, Tallini G, Bosari S, by cultured rat mesangial cells after contact with environmentalLanguino L: Modulation of cell proliferation by the integrin cyto- lectins. Kidney Int 43(Suppl 39):S41–S46, 1993plasmic domain. Kidney Int 51:1434–1440, 1997 30. Peruzzi L, Trusolino L, Amore A, Gianoglio B, Cirina P, Basso

9. Sterzel RB, Schulze-Lohoff E, Weber M, Goodmann SL: Inter- G, Emancipator SN, Marchisio PC, Coppo R: Tubulo-interstitialactions between glomerular mesangial cells, cytokines, and extra- responses in the progression of glomerular diseases: Albuminuriacellular matrix. J Am Soc Nephrol 2(Suppl 10):S126–S131, 1992 modulates avb5 integrin. Kidney Int 50:1310–1320, 1996

10. Bruijn JA, Roos A, De Geus B, De Heer E: Transforming growth 31. Stacchini A, Fubini L, Aglietta M: Flow cytometric detectionfactor-b and the glomerular extracellular matrix in renal pathology. and quantitative analysis of the GM-CSF receptor in human granu-J Lab Clin Med 123:34–47, 1994 locytes and comparison with the radioligand binding assay. Cytom-

11. Coppo R, Peruzzi L, Cirina P, Amore A, Trusolino L, Basso etry 24:374–381, 1996G, Linari C, Shen Y, Marchisio PC: The role of integrins in IgA 32. Hillis GS, Roy-Chudhury P, Duthie LA, Stewart KN, Brownnephropathy. Nephrology 3:73–78, 1997 PA, Simpson JG, McLeod AM: Expression of b1 integrins in IgA

12. Shikata K, Makino H, Morioka S, Kashitani T, Hirata K, Ota nephropathy. Nephrol Dial Transplant 12:1137–1142, 1997Z, Wada J, Kanwar YS: Distribution of extracellular matrix recep- 33. Savill J, Hogg N, Ren Y, Haslett C: Vitronectin receptor medi-tors in various forms of glomerulonephritis. Am J Kidney Dis

ated phagocytosis of cells undergoing apoptosis. Nature 343:170–25:680–688, 1995173, 199013. Gauer S, Jao J, Schoecklmann HO, Sterzel RB: Adhesion mole-

34. Kagami S, Kondo S, Loster K, Reutter W, Kuhara T, Yasumotocules in the glomerular mesangium. Kidney Int 51:1447–1453, 1997K, Kuroda Y: Alpha1 beta1 integrin-mediated collagen matrix re-14. Zheng M, Fang H, Hakomori S: Functional role of N-glycosylationmodeling by rat mesangial cells is differentially regulated by trans-in alpha5 beta 1 integrin receptor: De-N-glycosylation inducesforming growth factor beta and platelet growth factor BB. J Amdissociation or altered association on alpha5 and beta 1 subunitsSoc Nephrol 10:779–789, 1999and concomitant loss of fibronectin binding activity. J Biol Chem

35. Prols F, Hartner A, Schocklamann HO, Sterzel RB: Mesangial269:12325–12331, 1994cells and their adhesive properties. Exp Nephrol 7:137–146, 199915. Anderson SS, Kim Y, Tsilibary EC: Effects of matrix glycation

36. Harper SJ, Feehally J: The pathogenic role of immunoglobulin Aon mesangial cell adhesion, spreading and proliferation. Kidneypolymers in immunoglobulin A nephropathy. Nephron 65:337–345,Int 46:1359–1367, 1994199316. Kerr MA: The structure and function of human IgA. Biochem J

37. Rifai A: Characteristics of nephritogenic IgA immune complexes.271:285–296, 1990Am J Kidney Dis 12:402–405, 198817. Field MC, Duck RA, Edge CJ, Rademacher TW: O-linked oligo-

saccharide from human serum immunoglobulin A. Biochem Soc 38. Gomez Guerrero C, Gonzalez E, Hernando P, Ruiz-Ortega M,Trans 17:1034–1035, 1989 Egido J: Interaction of mesangial cells with IgA and IgG immune

18. Coppo R, Amore A, Roccatello D, Gianoglio B, Molino A, complexes: A possible mechanism of glomerular injury in IgAPiccoli G, Clarkson AR, Woodroffe AJ, Sakai H, Tomino Y: nephropathy. Contrib Nephrol 104:127–137, 1993IgA antibodies to dietary antigens and lectin-binding IgA in sera 39. van Den Dobbelsteen MEA, Van Der Woude FJ, Schroeijersfrom Italian, Australian and Japanese IgA nephropathy patients. WEM, Van Den Wall Bake AWL, van Es LA, Daha M: BindingAm J Kidney Dis 4:480–487, 1991 of dimeric and polymeric IgA to rat renal mesangial cells enhances

19. Coppo R, Amore A, Gianoglio B, Reyna A, Peruzzi L, Rocca- the release of interleukin 6. Kidney Int 46:512–519, 1994tello D, Alessi D, Sena LM: Serum IgA and macromolecular 40. Duque N, Gomez-Guerrero C, Egido J: Interaction of IgA withIgA reacting with mesangial matrix components. Contrib Nephrol Fc alpha receptors of human mesangial cells activates transcription104:162–171, 1993 factor nuclear factor-kappa B and induces expression and synthesis

20. Coppo R, Amore A, Cirina P, Messina M, Basolo B, Segoloni of monocyte chemoattractant protein-1, IL-8, and IFN-inducibleG, Berthoux F, Boulharouz R, Egido J, Alcazar R, Clarkson protein 10. J Immunol 159:3474–3482, 1997AR, Woodroffe A: IgA serology in recurrent and non-recurrent 41. Kokubo T, Hiki Y, Iwase H: Protective role of IgA1 glycansIgA nephropathy after renal transplantation. Nephrol Dial Trans- against IgA1 self-aggregation and adhesion to extracellular matrixplant 10:2310–2315, 1995 proteins. J Am Soc Nephrol 9:2048–2054, 1998

21. Allen AC, Harper SJ, Feehally J: Galactosylation of N- and O- 42. Nikolova EB, Tomana M, Russel MW: The role of the carbohy-linked carbohydrate moieties of IgA1 and IgG in IgA nephropathy. drate chains in complement (C3) fixation by solid-phase-boundClin Exp Immunol 100:470–474, 1995 human IgA. Immunology 82:321–327, 199422. Hiki Y, Iwase H, Kokubo T, Horii A, Tanaka A, Hishikido J,

43. Gomez-Guerrero C, Dunque N, Egido J: Mesangial cells possessHotta K, Kobayashi Y: Association of asialo-galactosyl b1-3N-an asialoglycoprotein receptor with affinity for human immuno-acetylgalactosamine on the hinge with a conformational instabilityglobulin A. J Am Soc Nephrol 4:568–576, 1998of jacalin-reactive immunoglobulin A1 in immunoglobulin A ne-

44. Leung JC, Tsang AW, Chan DT, Lai KN: Absence of CD89,phropathy. J Am Soc Nephrol 7:955–960, 1996polymeric immunoglobulin receptor, and asialoglycoprotein recep-23. Tomana M, Matousovic K, Julian BA, Radl J, Konecny K,tor on human mesangial cells. J Am Soc Nephrol 11:241–249, 2000Mestecky J: Galactose-deficient IgA1 in sera of IgA nephropathy

45. Davin JC, LiVecchi M, Nagy J, Foidart JM, Foidart JB, Barba-patients is present in complexes with IgG. Kidney Int 52:509–516,gallo Sangiorgi G, Malaise M, Mahieu P: Evidence that the1997interaction between circulating IgA and fibronectin is a normal24. Allen AC, Topham PS, Harper SJ, Feehally J: Leukocyte b1,3process enhanced in primary IgA nephropathy. J Clin Immunolgalactosyltrasferase activity in IgA nephropathy. Nephrol Dial11:78–94, 1991Transplant 12:701–706, 1997

46. Eitner F, Schulze M, Brunkhorst R, Koch KM, Floege J: On25. Meredith J Jr, Takada Y, Fornaro M, Languino LR, Schwartzthe specificity of assays to detect circulating immunoglobulin A-MA: Inhibition of cell cycle progression by the alternatively splicedfibronectin complexes: Implications for the study of serology phe-integrin b1C. Science 269:1570–1572, 1996nomena in patients with immunoglobulin A nephropathy. J Am26. Emancipator SN, Rao CS, Amore A, Coppo R, Nedrud J: Macro-Soc Nephrol 5:1400–1406, 1994molecular properties that promote mesangial binding and mesangi-

47. Kirchberg K, Lange TS, Klein EC, Jungtaubl H, Heinen G,opathic nephritis. J Am Soc Nephrol 2(Suppl):S149–S158, 1992Meyer Ingold W, Scharffetter Kochanek K: Induction of beta27. Cassidy J, Jourdian G, Roseman S: The sialic acids: VI purification1 integrin synthesis by recombinant platelet derived growth factorand properties sialidase from Clostridium perfringens. J Biol Chem(PDGF-AB) correlates with an enhanced migratory response of240:3501–3506, 1965human dermal fibroblasts to various extracellular matrix proteins.28. Distler J, Jourdian G: The purification and properties of b galac-

tosidase from bovine testes. J Biol Chem 248:6772–6780, 1973 Exp Cell Res 220:29–35, 1995