b1,6-n-acetylglucosamine-bearingn-glycans in human...

[CANCER RESEARCH 60, 134–142, January 1, 2000]

b1,6-N-Acetylglucosamine-bearingN-Glycans in Human Gliomas: Implications for aRole in Regulating Invasivity1

Hirotaka Yamamoto,2 Jason Swoger, Suzanne Greene, Tasuku Saito, Jay Hurh, Charla Sweeley, Jan Leestma,Edward Mkrdichian, Leonard Cerullo, Atsushi Nishikawa, Yoshito Ihara, Naoyuki Taniguchi, and Joseph R. MoskalThe Chicago Institute of Neurosurgery and Neuroresearch, Chicago, Illinois 60614 [H. Y., J. S., S. G., T. S., J. H., C. S., J. L., E. M., L. C., J. R. M.], and Department ofBiochemistry, Osaka University Medical School, Osaka 565, Japan [A. N., Y. I., N. T.]

ABSTRACT

The metastatic potential of tumor cells has been shown to be correlatedwith the expression of tri- and tetra-antennary b1,6-N-acetylglucosamine(b1,6-GlcNAc)-bearing N-glycans, which are recognized byPhaseolusvulgaris leukoagglutinating lectin (L-PHA). The expression ofb1,6-Glc-NAc-bearing N-glycans also has been used as a marker of tumor progres-sion in human breast and colon cancers. In this report, the role ofN-glycan branching in regulating glioma migration and invasion wasexamined. The expression ofb1,6-GlcNAc-bearing N-glycans was foundin human glioma specimens, whereas astrocytes from normal adult brainwere negative. The expression ofN-acetylglucosaminyltransferase V(GnT-V) mRNA, which is responsible for the biosynthesis ofb1,6-Glc-NAc-bearing N-glycans, was high in glioma cell lines with robustets-1expression. To study the molecular mechanism of GnT-V expression inhuman glioma cells, an inducibleets-1gene was stably transfected intoSNB-19 cells using a tetracycline repressor system. GnT-V mRNA expres-sion was increased by the induction of c-ets-1, suggesting that the Ets-1transcription factor directly regulates the transcription of GnT-V. Stabletransfection of GnT-V into human glioma U-373 MG cells resulted inchanges in cell morphology and focal adhesions and a marked increase inglioma invasivity in vitro. L-PHA has little effect on cell migration. On thecontrary, Phaseolus vulgariserythroagglutinating lectin (E-PHA), whichrecognizes bisectingb1,4-GlcNAc-bearingN-glycans, strongly inhibits cellmigration (haptotaxis) on a fibronectin substrate in U-373 MG transfec-tants and other glioma cell lines tested. These results suggest that theincreasedb1,6-GlcNAc-bearing N-glycan expression found in malignantgliomas is modulated by GnT-V through the Ets-1 transcription factor,and that the branching of complex typeN-glycans plays a major role inglioma invasivity.

INTRODUCTION

The carbohydrate moieties of cell surface glycoconjugates play animportant role in cell adhesion and metastasis (1, 2). Aberrant cellsurface glycosylation patterns, a hallmark of oncogenic transforma-tion and malignant phenotypes of tumor cells, lead to alterations incell-cell and cell-matrix interactions by affecting the function ofadhesion molecules such as E-cadherin, integrins, and CD44 (1, 3, 4).Conversely, modifications in tumor cell surface carbohydrate expres-sion by specific inhibitors of glycosylation have been shown to lead todecreases in tumor formation or metastasisin vivo (5–7).

Malignant gliomas, unlike tumors found outside of the centralnervous system, do not metastasize but are highly invasive. A numberof reports have strongly suggested that integrins play a key role inregulating invasivity (8, 9), and that theN-glycans of integrins mod-ulate the function of integrins (10, 11). The two most commonlyobserved aberrantN-glycosylations in experimental tumor models are

an increase in terminal sialylation (2, 12) and a shift to more highlybranchedN-linked oligosaccharides (1, 13–15).a2,3-Linked sialicacids were shown to be expressed on malignant glioma cell surfacesbut were absent in normal human adult astrocytes (16). When alter-ations in the glycosylation patterns of the glioma-associated integrin,a3b1, were introduced by transfection of thea2,6-sialyltransferasegene into a malignant glioma cell line, inhibition of invasivity wasobservedin vitro (17). On the other hand, there have been no studiesexamining whether highly branchedN-linked oligosaccharides play arole in glioma invasivity.

Recent studies demonstrate that branching ofN-linked oligosaccha-rides is dependent upon two distinct enzymes: UDP-GlcNAc3:b-D-mannosideb1,4-N-acetylglucosaminyl transferase III (GnT-III; EC2.4.1.144) and UDP-GlcNAc:b-D-mannoside b1,6-N-acetylglu-cosaminyltransferase V (GnT-V; EC 2.4.1.155) (18–20). GnT-IIIproducesN-glycans with bisecting structures, whereas GnT-V in-creasesb1,6 branching to create tri- and tetra-antennary structures.Increased expression of tri- or tetra-antennaryb1,6-GlcNAc-bearingN-glycans has been correlated with metastatic potential in rodenttumor models (3, 13) and also has been shown to be a marker of tumorprogression in human breast and colon neoplasia (14, 15). GnT-Vexpression appears to be regulated at least in part by the Ets family oftranscription factors because it has been shown that GnT-V expressionis dependent upon Ets-1 in a human bile duct carcinoma cell line (21)and other cell lines (22), and that increased expression of GnT-V bySrc kinase stimulation was abolished by a dominant-negative mutantof Ets-2 in human hepatocarcinoma Hep G2 cells (23).

On the basis of both the histochemical study ofb1,6-linked N-glycan expression in primary glioma specimens using L-PHA andNorthern analyses of primary gliomas and glioma cell lines, we haveexamined the regulation of the branching of complex typeN-glycansin glioma cells. We have also created GnT-III- and GnT-V-transfectedglioma cells to directly evaluate the biological function of the branch-ing of N-glycans in cell adhesion, migration, and invasionin vitro.

MATERIALS AND METHODS

Cell Culture and Brain Tumor Specimens. All established human braintumor cell lines were maintained using DMEM (containing 4.5 g/l glucose)supplemented with 10% heat-inactivated FBS (Whittaker BioProducts, Walk-ersville, MD). The following cell lines were used for Northern analysis: humanglioblastomas, SNB-19 and D-54MG (generously provided by Dr. Paul Korn-blith, University of Pittsburgh, Pittsburgh, PA and Dr. Darrell Bigner, DukeUniversity, Durham, NC, respectively); human glioblastomas, U-87 MG,U-373 MG, U-118 MG, and SW1088 (ATCC, Rockville, MD); human neu-roblastoma cell lines, SKN-SH, SKN-MC, and IMR 32 (ATCC), and LAN-5(generously provided by Dr. Stephan Ladish, Children’s Research Institute,Washington DC); and human hepatocarcinoma, Hep G2 (ATCC) as a positivecontrol for GnT-III and GnT-V. For Northern analysis of GnT-III and GnT-V,a panel of surgical specimens was used that consisted of 13 gliomas: 1astrocytoma grade II, 1 high-grade oligodendroglioma, 1 mixed glioma, 3

Received 2/3/99; accepted 10/18/99.The costs of publication of this article were defrayed in part by the payment of page

charges. This article must therefore be hereby markedadvertisementin accordance with18 U.S.C. Section 1734 solely to indicate this fact.

1 This work was supported in part by grants from the Illinois division of The AmericanCancer Society (to H. Y.), The Buchanan Foundation (to J. M.), The Brach Foundation (toJ. M.), and The Falk Foundation (to J. M.).

2 To whom requests for reprints should be addressed, at The Chicago Institute ofNeurosurgery and Neuroresearch, 2515 North Clark Street, Suite 800, Chicago, IL 60614.Phone: (773) 388-7880; Fax: (773) 935-2132; E-mail: [email protected].

3 The abbreviations used are: GlcNAc,N-acetylglucosamine; GnT,N-acetylglucosami-nyltransferase; L-PHA,Phaseolus vulgarisleukoagglutinating lectin; E-PHA,Phaseolusvulgaris erythroagglutinating lectin; ATCC, American Type Culture Collection.

134

Research. on January 22, 2020. © 2000 American Association for Cancercancerres.aacrjournals.org Downloaded from

http://cancerres.aacrjournals.org/

cases of astrocytoma grade III, and 7 cases of astrocytoma grade IV,i.e.,glioblastoma [WHO Brain Tumor Classification (24)].

Northern Analysis. Human GnT-V cDNA (1.24 kb) was isolated afterEcoRI restriction digestion and used as a cDNA probe for Northern analysis.Human GnT-III cDNA (1.8 kb) was used afterEcoRI andXbaI restrictiondigestion. Human Ets-1 cDNA was cloned by using the reverse transcription-PCR (RT-PCR) and poly(A)1RNA from U-87 MG cells based on thesequence reported previously (25). A sense primer 59-TTGGGAA-GAAAGTCGGATT-39 (bp 2119 to 2101) and an antisense primer 39-CAGGCTGAATTCATTCACAGC-59(bp 270 to 250) were used for reversetranscription-PCR. A 389-bp PCR product was cloned into pT7 Blue T vector(Novagen, Madison, WI), and the sequence of the insert was confirmed by thedideoxy termination method (Sequenase, United States Biochemical, Cleve-land, OH). The cDNA coding for human ets-1 was isolated from the gel afterNdeI andBamHI digestion and was used as the probe.

Surgical specimens were immediately frozen in liquid nitrogen upon resec-tion. Total RNA was isolated from clinical glioma specimens and culturedbrain tumor cells using guanidinium isothiocyanate, followed by CsCl2 cen-trifugation as described previously (16). Thirtymg of total RNA per primarybrain tumor and 20mg of total RNA per tumor cell line per lane wereelectrophoresed in an agarose-formaldehyde gel and transferred to Duralonnylon membranes (Stratagene, La Jolla, CA). After UV cross-linking, the blotswere hybridized with a32P-radiolabeled cDNA probe synthesized by using arandom priming kit (Stratagene) and ExpressHyb solution (Clontech, PaloAlto, CA). The blots were then exposed to X-OMAT film (Kodak, Rochester,NY), and the films were developed appropriately.

Lectin Histochemistry with L-PHA. b1,6-linkedN-glycan expression inprimary glioma specimens was examined using L-PHA (26). Paraffin-embed-ded sections (6mm thick) of formalin-fixed specimens, derived from onemixed glioma case, two cases of astrocytoma grade III, and two cases ofglioblastoma (astrocytoma grade IV), were processed at room temperatureunless otherwise mentioned. The sections were dewaxed and hydrated and thensoaked in Tris-buffered saline (TBS; 150 mM NaCl, 50 mM Tris-HCl, pH 7.5)at 37°C for 1 h or 13 h (according to our preliminary studies with other lectins)to unmask lectin binding sites. Then the sections were rinsed with TBS for 10min and incubated in 0.5% blocking reagent (Boehringer Mannheim, Indian-apolis, IN) in TBS for 45–60 min. After rinsing twice with TBS and once withbuffer 1 (TBS with 1 mM MgCl2, 1 mM MnCl2, and 1 mM CaCl2, pH 7.5) for10 min each, 10mg/ml digoxigenin-labeled L-PHA (Boehringer Mannheim) inbuffer 1 with or without 0.05% Tween 20 and 0.05% Triton X-100 wasoverlaid for 1 h. Rinsing with TBS (33 10 min) was followed by incubationwith anti-digoxigenin Fab fragments conjugated with 0.75 unit/ml alkalinephosphatase (Boehringer Mannheim) in TBS containing 0.05% Tween 20 and0.05% Triton X-100 for 1 h. After rinsing (TBS, 33 10 min), 5-bromo-4-chloro-3-indolyl phosphate/nitroblue tetrazolium solution (Sigma Chemical

Co., St. Louis, MO) was overlaid as chromogen in darkness up to 50 min andrinsed with 10 mM Tris-HCl with 1 mM EDTA. The sections were lightlycounterstained with nuclear fast red and fixed with 10% buffered formalin tolessen fading of reaction product during dehydration and clearing.

To check the specificity of lectin binding, each staining was performedsimultaneously with labeled L-PHA that was preincubated in the presence of9 mM bovine thyroglobulin (Sigma) for 90–120 min prior to lectin incubationas a negative control.

Western and Lectin Blot. Cultured cells were rinsed twice with PBS andlysed in hot cell lysis solution containing 1% SDS, 10 mM Tris-HCl (pH 7.4).To detectb1,6-GlcNAcN-glycans, 30mg of cell lysates were loaded on an 8%SDS-polyacrylamide gel. After electrophoresis, proteins were transferred to apolyvinylidene difluoride membrane, and the membrane was blocked with 5%BSA in PBS. It was then incubated with 0.1mg/ml horseradish peroxidase-conjugated L-PHA (EY Laboratory, CA) in TBS containing 2% BSA and 0.1%Tween 20 for 1 h at room temperature. Next, the membrane was washed withTBS containing 2% BSA and 0.1% Tween 20 for 10 min, followed by washingtwice with 0.1% Tween 20 in TBS. The blot was then developed with the ECLChemiluminescence detection system (Amersham, Buckinghamshire, UnitedKingdom). Protein concentrations were determined using the BCA reagent(Pierce). To detect Ets-1 protein expression in brain tumor cell lines, 20mg ofprotein cell lysates were loaded on a 8% SDS-polyacrylamide gel immediatelyafter boiling each sample in the presence of 2%b-mercaptoethanol. Afterelectrophoresis, proteins were transferred to a polyvinylidene difluoride mem-brane, and the membrane was blocked with 5% BSA in PBS. It was thenincubated with a 1: 10,000 dilution of monoclonal antihuman Ets-1 antibody(Clone 47; Transduction Laboratory, KY) in Tris-buffered saline pH 7.4 (TBS)containing 2% BSA and 0.1% Tween 20 for 1 h at room temperature. Themembrane was then washed with TBS containing 2% BSA and 0.1% Tween20 for 10 min, followed by washing twice with 0.1% Tween 20 in TBS. Next,it was incubated with a 1:10,000 dilution of horseradish peroxidase-conjugatedantimouse IgG (Amersham) for 1 h at room temperature in 2% BSA in TBScontaining 0.1% Tween 20. The membrane was then washed as describedabove and developed with the ECL Chemiluminescence detection system(Amersham) according to manufacturer’s instructions.

Stable Transfection of Inducible ets-1 Gene into Human GliomaSNB-19 Cells.The 1.4-kb human c-ets-1(full coding sequence) was insertedinto the pcDNA4/TO vector (Invitrogen, San Diego, CA). The pcDNA4/TO/c-ets-1and pcDNA6/TR (Invitrogen) vectors were cotransfected into humanglioma SNB-19 cells using the cationic liposome system, DOTAP (BoehringerMannheim). After 3 weeks of culture in selection medium containing 10mg/mlof Blasticidin and 1 mg/ml of Zeocin, transfected cells were subcloned withcloning rings to isolate individual clones. Individual clones were furthercultured for 4 weeks in the selection medium and then analyzed for theregulated gene expression in the presence of 2mg/ml tetracycline.

Stable Transfection of GnT-V into Human Glioma U-373MG Cells.The 2.4-kb human GnT-V cDNA (full coding sequence) was inserted into thepcDNA3 expression vector (Invitrogen) at theKpnI andXbaI sites, and theorientation of the insert was confirmed byHindIII restriction digestion. ThepcDNA3/GnT-V was then transfected into U-373 MG cells using the cationicliposome system DOTAP, (Boehringer Mannheim) according to the methodsdescribed previously (17). After 3 weeks of culture in selection mediumcontaining 800mg/ml of G418, transfected cells were subcloned with cloningrings to isolate individual clones. Individual clones were further cultured for 4weeks in the selection medium and then analyzed for the gene expression byNorthern analyses and L-PHA lectin blotting to identify successful GnT-Vtransfectants. Stable transfection ofGnT-III gene into the same U-373 MG wasreported previously (27).

Invasion Assay. Invasivity of the GnT-V-transfected subclones was exam-ined using a commercial membrane invasion culture system (9, 28). BiocoatMatrigel Invasion Chambers (Collaborative Research, Bedford, MA) consist oftwo compartments separated by a filter precoated with Matrigel (containslaminin, type IV collagen, entactin, and heparan sulfate). Cell invasion, whichis the result of cell adhesion to the extracellular matrix, degradation of thematrix proteins, and cell migration to the other side of the filter, is measuredby counting the number of cells passing to the opposite side of the filter via8-mm pores. Cells (43 104) were plated into the upper compartment andincubated for 24 h. U-373 MG cell conditioned medium (0.5 ml) was placedin the lower compartment to facilitate chemoattraction (28).

Fig. 1. The expression of GnT-III and GnT-V mRNA in glioma specimens. Thirtymgof total RNA per lane were used for Northern analysis.Lane 1,normal human brain;Lanes 2–14,clinical glioma specimens. Increased GnT-III expression is seen inLanes 3and10 compared with normal brain, whereas other specimens showed similar levels orless than that in normal brain (A). Enhanced GnT-V mRNA expression is seen inLanes3, 4, 7,and10 (B), and other samples showed similar levels or less compared with that innormal brain.C, ethidium bromide (EtBr) staining of total RNA.

135

ABERRANT b1,6-LINKED N-GLYCOSYLATION IN HUMAN GLIOMAS



Cells that migrated through the Matrigel and through the filter were fixed in10% formalin and stained with hematoxylin. The membranes were mounted onglass slides, and the cells were counted (9). Parental U-373 MG- and pcDNA3vector-transfected cells were used as controls.

Immunofluorescence Microscopy.To characterize the morphologicalchange of GnT-V and GnT-III transfectants, immunofluorescence microscopywas performed using monoclonal antihuman vinculin antibody (Sigma; clonehVIN-1) and monoclonal anti-VLA3 antibody (Chemicon; clone M-KD102).Anti-vinculin antibody was used to visualize focal adhesion sites and anti-VLA3 antibody was used to visualizeda3b1 integrin in the transfectants. Cellswere plated on fibronectin-coated (10mg/ml) coverslips and incubated inDMEM supplemented with 10% FBS for 16 h. Cells were gently washed twicewith PBS, then fixed with 4% formalin in PBS for 30 min, followed bywashing with PBS for 3 min. Cells were treated with 1% NP40 in PBS for 10min, followed by washing with PBS three times. After blocking with 10%normal goat serum for 15 min at room temperature, cells were incubated withmonoclonal antihuman vinculin antibody (1:400 dilution) or monoclonal anti-a3b1 integrin antibody (1:200 dilution) in PBS for 30 min at room tempera-ture. They were then washed three times with PBS (5 min each) and thenincubated with FITC-labeled goat antimouse immunoglobulin antibody (1:160dilution; Sigma) for 30 min at room temperature. The cells were washed with

PBS five times to remove unbound secondary antibody and were mounted withVectashield (Vector). Fluorescence microscopy was performed using a NikonModel 401 Fluorescence Microscope.

In Vitro Cell Migration Assays. Directed cell migration on a solid-phasegradient of a fibronectin substrate (haptotaxis) was measured using a Transwell(Costar, Cambridge, MA), which consists of two compartments separated by6.5-mm inserts with 8-mm-pore polycarbonate filters in 24-well culture plates.To establish a solid-phase gradient, only the underside of the filter was coatedwith 10 mg/ml human plasma fibronectin (Life Technologies, Grand Island,NY) in sodium bicarbonate buffer (pH 9.7) overnight at 4°C. It was thenblocked with 1% BSA (fatty acid free; Sigma) in PBS for 45 min at roomtemperature and rinsed three times with PBS.

GnT-V, GnT-III transfected U-373 MG and control cells were gently treatedwith 3 0.5 trypsin-EDTA (Life Technnologies, Inc.) in PBS for;5 min at37°C and then neutralized with DMEM containing 0.2% BSA. After washingwith 0.2% BSA-DMEM, cells were resuspended in protein-free DMEM andwere plated 10,000 cells/100ml/insert. The inserts were moved onto the lowerwells, which contained protein-free DMEM (0.5 ml), and were incubated for6 h at 37°C in a CO2 incubator. For inhibition of cell migration by lectins,L-PHA or E-PHA (Vector Laboratory) at the final concentration of 2 or 10mg/ml was added to both upper and lower compartments. Monoclonal anti-a3

Fig. 2. L-PHA lectin staining of human glioma specimens. L-PHA lectin staining showed variable but typical morphological features found in high-grade astrocytomas.A, cellsurface staining of a specimen of glioblastoma that characteristically contained zonal necrosis and multiple nucleated tumor cells. In another glioblastoma specimen (B), the lectinstaining was found in extracellular matrices between undifferentiated small tumor cells with high cellularity. Cytoplasmic round bodies in gemistocytic astrocytoma cells were alsostained with the lectin (C), whereas normal astrocytes were not stained (D). L-PHA stained the vasculature found in the glioblastoma specimens, which was closely related to thedistribution of tumor, but varied in size and shape: endothelial cells in capillaries, thin-walled vessels with extended lumina, thick-walled larger vessels, and vessels with convolutedlumina (glomeruloid vessels). These vessels were compatible morphologically with the well-described neovascularization typically found in glioblastomas; the staining pattern wasconsistent with the idea that L-PHA binds to the vascular basement membrane produced by the glioblastoma cells.Bar, 20 mm.

136




integrin antibody (Chemicon; clone P1B5) was also used to inhibita3b1integrin-mediated cell migration. After thorough absorption of DMEM withcotton swabs, the porous filter was dried with air blow and cut from the plasticsupports. Cells on both sides of the filter were fixed and stained with Diffu-Quick (Baxter, Chicago, IL). The filters were then mounted with Permount(Fisher Scientific, Chicago, IL) on glass slides with 12-mm coverslips. Undermicroscope, cells on both the topside (i.e.,nonmigrated) and underside (i.e.,migrated) of the filters were counted in eight consecutive fields along one filterdiameter (;10% of the entire surface was observed). The percentage ofmigration (migrated cell count/total cell count) was determined based ontriplicate experiments.

RESULTS

The normal brain tissue examined in these studies was found tomarkedly express GnT-III mRNA but only trace amounts of GnT-VmRNA. Gliomas, on the other hand, expressed highly variableamounts of both GnT-III and GnT-V mRNA (Fig. 1). On the basis ofthese results, no obvious correlation could be made between tumorgrade and mRNA expression, but there was a marked difference in theGnT-III and GnT-V mRNA expression patterns in most gliomascompared with normal brain.

To identify the cells expressing glycoproteins bearing high levels ofb1,6-GlcNAc containingN-glycans, lectin histochemistry was per-formed using L-PHA.b1,6-GlcNAc-bearingN-glycans were highlyexpressed on the cell surface of glioma cells as well as in neovascularendothelial cells and their extracellular matrices, whereas normalastrocytes were not stained. In normal brain tissue distant from neo-plastic cells, lectin binding was scarcely found and was restricted tothe subarachnoid space (Fig. 2).

Thus, the variability in GnT-III and GnT-V mRNA expressionfound in the glioma specimens may be attributed to the fact thatb1,6-GlcNAc-bearingN-glycans were expressed on both glioma cellsand neovascular endothelial cells.

The expression of GnT-III and GnT-V mRNA was also studied ina panel of five glioma and four neuroblastoma human brain tumor celllines. These data, along with the expression of the transcription factorets-1, is shown in Fig. 3. Unlike in the clinical specimens examined,marked and consistent GnT-III mRNA expression was found in all ofthe cell lines, whereas the level of GnT-V mRNA expression variedfrom cell line to cell line. It can be seen that the cell lines that highlyexpressed GnT-V mRNA also expressed high levels ofets-1mRNA(ets-2mRNA was undetectable in all of the cell lines examined; datanot shown).

L-PHA lectin blots were done to directly analyze the expression ofb1,6-GlcNAc-bearing glycoproteins in the tumor cell lines (Fig. 4A).A major glycoprotein ofMr 140,000 was revealed in all five gliomacell lines but was absent or barely detectable in the neuroblastoma cell

lines. Moreover, the pattern of L-PHA-reactive glycoprotein expres-sion in the glioma cell lines was different from that in the neuroblas-tomas.

It has been demonstrated previously that Ets-1 controls the tran-scription of GnT-V (21, 22). A Western blot showed that Ets-1 proteinwas expressed uniformly in the entire panel of brain tumor cell linesexamined (Fig. 4B). The induction of c-ets-1mRNA in the gliomatransfectants resulted in the increased expression of GnT-V mRNA asshown in Fig. 5.

Malignant gliomas, unlike tumors found outside of the centralnervous system, do not metastasize but are highly invasive. To eval-uate the effect of increasedb1,6-branching ofN-glycans on gliomainvasivity, the GnT-V gene was stably transfected into the humanglioma cell line U-373 MG, according to methods described previ-ously (17). This cell line expressed low levels of GnT-V mRNA (Fig.3). As shown in Fig. 6, we have isolated five GnT-V stable transfec-tants that expressed the 3.0-kb GnT-V transcript in addition to theendogenous 9.5-kb transcript.

The GnT-V transfectants showed higher invasionin vitro compared

Fig. 3. Expression of GnT-III, GnT-V, and c-ets-1 mRNA inhuman brain tumor cell lines. Twentymg of total RNA per lanewere used for Northern analysis.Left panel: Lanes 1–5arehuman glioma cell lines, andLanes 6–9are human neuroblas-toma cell lines.Lane 10,Hep G2 human hepatocarcinoma as apositive control for GnT-III and GnT-V expression.Right pan-el: Lanes 1–6are human glioma cell lines, andLanes 7–10arehuman neuroblastoma cell lines. All brain tumor cell linesexpressed similar amounts of GnT-III mRNA (A), but GnT-Vexpression varied among the cell lines (B). Brain tumor celllines with high GnT-V expression (D) also showed robustexpression of c-ets-1 mRNA (E).C and F, ethidium bromide(EtBr) staining of total RNA.

Fig. 4. Expression of L-PHA binding proteins and Ets-1 protein in glioma cell lines.A,L-PHA lectin was used to detect glycoproteins carryingb1,6-GlcNAc N-glycan.B,Western blot of Ets-1 protein using monoclonal anti-Ets-1 antibody.Lanes 1–5,humanglioma cell lines SW1088, U-118 MG, U-373 MG, U-87 MG, and D-54MG, respectively.Lanes 6–9,human neuroblastoma cell lines SKN-SH, SKN-MC, LAN-5, and IMR-32,respectively. L-PHA lectin recognized theMr 140,000 glycoprotein (arrow) in all humanglioma cell lines (A).MW, molecular weight.Mr 51,000 Ets-1 protein was detected in allglioma and neuroblastoma cell lines (B).

137




with parental U-373 MG or vector-transfected control cells (Fig. 7).The GnT-V-transfected cells also showed a fan-shaped cell morphol-ogy with a distinct leading edge (Fig. 8), indicating directional cellmigration of the transfectants.

Further characterization of the GnT-V transfectants was undertakenusing an anti-vinculin antibody to visualize focal adhesions and ananti-a3b1 integrin antibody to visualize the leading edge of the cells.As shown in Fig. 9, stronga3b1 integrin staining, the predominantintegrin in U-373 MG cells (17), was observed at the leading lamel-lipodia of the GnT-V transfectants (Fig. 9F), whereas dendritic proc-ess staining was observed in controls. Focal adhesion sites visualizedby anti-vinculin antibody staining radiate toward leading lamellipodiain the GnT-V transfectants (Fig. 9E), whereas vinculin staining wasfound along the edge of parental U-373 MG cells and vector-trans-fected controls (Fig. 9,A andG). In the GnT-III transfectants, focaladhesion sites were randomly distributed (Fig. 9C).

Increasingb1,6-GlcNAc-bearingN-glycans by GnT-V gene trans-fection produced a marked increase in the invasivity of the U-373 MGglioma cells in vitro and changes in cell morphology and focaladhesions. These results suggested that the effects of the GnT-Vtransfection may be attributable to a change in cell motility. Cellmigration assays were performed to more directly evaluate the role ofN-glycan branching. E-PHA or L-PHA lectins, which recognize bi-sectingb1,4-GlcNAc structures (the product of GnT-III) and tri- ortetra-anntenaryb1,6-GlcNAc structures (the product of GnT-V), re-spectively, were used (Fig. 10). The effect of E-PHA and L-PHA oncell migration was tested using GnT-III- and GnT-V-transfected

U-373 MG cells, their controls, as well as the glioma cell linesD-54MG, SNB-19, SW1088, and U-87 MG. GnT-III-transfectedU-373 MG clone P9 overexpresses bisectingb1,4-GlcNAc oligosac-charides (27), and GnT-V-transfected clone J13 expresses the highestlevel of GnT-V andb1,6-GlcNAc expression (data not shown). TheGnT-III transfectants showed significantly less cell migration on afibronectin substrate compared with parental U-373 MG or othertransfectants, whereas the GnT-V transfectants did not show a signif-icant increase in cell migration under our experimental conditions.Cell migration on a fibronectin substrate was completely inhibited at10mg/ml E-PHA in all glioma cell lines tested, whereas the inhibitoryeffect of L-PHA is much less than 50% (data not shown). At 2mg/ml,E-PHA showed strong inhibition of cell migration, whereas L-PHAshowed little effect (Fig. 10). The same levels of inhibition were

Fig. 5. Increased expression of GnT-V mRNA by the induction of ets-1 mRNA inglioma cells. The pcDNA4/TO/c-ets-1 and pcDNA6/TR cotransfected SNB-19 cells(clones1 and2) were incubated in the presence of 2mg/ml tetracycline for 24 h to inducethe transfected c-ets-1 expression. Both the induced (1, with tetracycline) and controlcells (2, without tetracycline) were harvested for Northern analyses; 15mg of total RNAper lane were used. The blot was hybridized by both radiolabeled GnT-V and ets-1 cDNAprobes. Induction of ets-1 resulted in increased expression of GnT-V mRNA in bothtransfectants (A).B, total RNA staining by ethidium bromide (EtBr).

Fig. 6. Stable transfection ofGnT-Vgene into human glioma U-373 MG cells. Twentymg of total RNA per lane were used for Northern analyses.Lane 1,parental U-373 MGglioma cells;Lane 2,pcDNA3 vector-transfected U-373 MG;Lanes 3–7,GnT-V-trans-fected U-373 MG clones. GnT-V stable transfectants express the 3.0-kb GnT-V transcriptin addition to the endogenous 9.5-kb transcript (arrow; A). B, ethidium bromide (EtBr)staining of total RNA.

Fig. 7. In vitro invasion assays of GnT-V transfectants. The relative invasivity ofGnT-V-transfected U-373 MG clones was compared with the invasivity of pcDNA3vector-transfected U-373 MG cells (column 7) as 100%.Columns 1–5,GnT-V-transfectedU-373 MG clones;column 6,parental U-373 MG cells;column 7,pcDNA3 vector-transfected U-373 MG. The transfectants were 2–5-fold more invasive than the vector-transfected control cells and 4–10-fold more invasive than parental U-373 MG cells. Thelevels of GnT-V mRNA expression in the transfected clones are mostly, but not always,correlated with the levels of invasivity. The data are the average values of two separateexperiments done in triplicate;bars,SE.

138




observed by 2mg/ml E-PHA and 10mg/ml anti-a3 monoclonalantibody (Chemicon; clone P1B5) in parental U-373 MG cells. Theseresults clearly demonstrate the functional difference between the twodifferent types ofN-glycan branching structures found in gliomas andare consistent with previous observations that cells expressing highlybranchedb1,6-GlcNAc are less adhesive to extracellular matrices (11,29).

DISCUSSION

The experiments presented in this report are predicated on the ideathat alterations in the expression of normal cell surface carbohydratescan modulate the invasive potential of malignant gliomas. Specifi-cally, an examination of the role ofN-linked oligosaccharide branch-ing, found on the glioma-associated glycoproteins such as the integrin,a3b1, was undertaken.

The importance ofN-linked oligosaccharide branching in tumormetastasis was demonstrated in a series of experiments reported byDennis and Laferte (14). Specifically, they created a panel of glyco-sylation mutants in a highly metastatic murine tumor cell line andshowed a strong correlation between the increasedb1,6-linkedbranching of complex type oligosaccharides and metastatic potential.A number of more recent studies have also shown an increasedexpression of highly branchedb1,6-GlcNAc-linkedN-glycans in avariety of tumor models. These have included experiments using cellstransformed by DNA viruses such as polyoma and Rous sarcoma,oncogenes such as H-rasandsrc, as well as human breast and coloncancers (3, 15, 16, 22, 30). Furthermore, increasedb1,6-GlcNAc-linked N-glycans, brought about byGnT-V gene transfection intopremalignant mink lung epithelial cells, resulted in increased tumor-

igenicity because of an increase in cell motility by alterations ina5b1andavb3 integrins (11).

To address the question as to whether changes inN-glycan branch-ing play a role in glioma invasivity, an examination of the expressionof GnT-III and GnT-V mRNA was undertaken. In normal adulthuman brain, robust GnT-III mRNA expression was observed,whereas GnT-V mRNA expression was very low by comparison. Thisis not surprising in light of the fact that the predominantN-linkedoligosaccharides found in normal brain are complex-type bisectingb1,4-GlcNAc structures that are the product of GnT-III (31). In themalignant gliomas examined, both GnT-III and GnT-V mRNAs werevariably expressed. Lectin staining with L-PHA, which recognizesb1,6-GlcNAc containing oligosaccharides, was undertaken to get aclearer picture of where these structures are expressed. L-PHA stain-ing was found in malignant glioma cells, neovascular endothelialcells, and extracellular matrices surrounding the tumor cells but not innormal cells. Most of the clinical specimens used in this study werehigh-grade gliomas. Patients with these tumors have the shortestsurvival (6–12 months upon diagnosis). We did not find a statisticallysignificant positive correlation between the levels of GnT-V mRNAexpression and clinical outcome among the high-grade glioma pa-tients whose tumors were used in these studies, and a larger numberof low-grade gliomas than available would be required to evaluateGnT-V mRNA expression as a prognostic marker.

In glioma cell lines, GnT-III mRNA levels were uniformly high,whereas GnT-V mRNA levels were quite variably expressed. AnL-PHA lectin blot revealed that most glioma cells express a majorL-PHA-reactive glycoprotein with a molecular weight ofMr 140,000,whereas protein extracts from neuroblastoma cells or normal brainshowed different patterns of L-PHA staining, and theMr 140,000

Fig. 8. Cell morphology of GnT-transfectedclones. Phase-contrast photomicrographs of U-373MG cells (A), GnT-III-transfected cells (B), GnT-V-transfected cells (C), and pcDNA3 vector-transfectedcells (D). Parental U-373 MG cells show similar cellmorphology with the vector-transfected control cells.GnT-V-transfected cells have fan-shaped cell mor-phology with a distinct leading edge, whereas GnT-III-transfected cells are well spread.

139




glycoprotein was rarely found. The expression of the L-PHA-reactiveglycoprotein was high in SW1088 and U-87 MG glioma cell lines,which show high levels of GnT-V expression, whereas a small amountof L-PHA reactivity was found in U-118 MG glioma cells, despite itshigh GnT-V mRNA expression. Furthermore, neuroblastoma celllines with high GnT-V mRNA expression (LAN-5) show little or noMr 140,000 staining. These results suggest that the levels ofb1,6-GlcNAc-bearingN-glycans in gliomas are controlled by mechanismsthat regulate both GnT-V expression and the availability of its proteinsubstrates. Data obtained from immunoprecipitation studies usinganti-a3 integrin antibodies showed that the major glycoprotein rec-ognized by L-PHA in gliomas isa3b1 integrin (data not shown), themost predominant integrin found in clinical glioma specimens (8) and

the U-373 MG glioma cell line used in these studies (17). A veryrecent study has identified thata3 integrin mRNA expression appearsto be quantitatively correlated with the grade of malignancy of glio-mas and medulloblastomas (32).

Thus, b1,6-GlcNAc-bearing oligosaccharides were found on thea3b1 integrin and appeared to be associated specifically with gliomasand not normal astrocytes. Furthermore, aberrant up-regulation ofGnT-V expression, as opposed to decreased GnT-III expression, ap-pears to be responsible for their expression. Because GnT-III andGnT-V are the two enzymes that regulate the type of branchingstructures found withinN-linked oligosaccharides and compete for thesame substrates, the results suggest that a mechanism exists to shiftthe integrin oligosaccharides from bisectingb1,4-GlcNAc to highly

Fig. 9. Immunofluorescence microscopy ofGnT-transfected cells using monoclonal antibodiesagainsta3b1 integrin and vinculin.A, C, E,andG,cells stained with anti-vinculin antibody.B, D, F,and H, cells stained with anti-a3b1 integrin anti-body. U-373 MG cells (Aand B), GnT-III-trans-fected cells (CandD), GnT-V-transfected cells (EandF), and pcDNA3 vector-transfected controls (GandH).

140




branchedb1,6-GlcNAc during the transformation of glia into gliomasor noninvading glioma cells into invasive ones.

GnT-V expression appears to be regulated at least in part by the Etsfamily of transcription factors (21–23). Ets-1 is expressed duringneural crest cell migration (33), which is a physiological example ofcell invasion. Ets-1 is also found within neovascular endothelial cellsto promote neovascularization and in stromal fibroblasts adjacent tocarcinoma cells to promote tumor invasion (33). We have found theexpression ofb1,6-branchedN-glycans in both glioma cells andneovascular endothelial cells. We also found Ets-1 protein in allglioma cells tested, whereas Ets-1 protein was reported to be absentwithin carcinoma cells (34). This difference in terms of Ets-1 proteinexpression within tumors may be attributable to the fact that thestromal reaction by fibroblasts plays an important role in carcinomainvasion, whereas there is little stromal reaction in malignant gliomas.Unlike carcinomas, Ets-1 expression in glioma cells may play a directrole in promoting glioma invasion. It also has been reported that theexpression of Ets-1 can be modulated by growth factors and proteinkinase C activators (34), such as phorbol ester, through its interactionwith other transcription factors, such as AP-1 (35, 36). Those studiessuggest that the mitogen-activated protein kinase pathway and Ets-1play a role in the expression of GnT-V in glioma cells. To examine amolecular mechanism that increases the expression of GnT-V ingliomas, we chose to investigate the possible involvement of the Ets-1transcription factor in glioma cells.

High GnT-V mRNA expression was found in brain tumor cell lineswith robust c-ets-1 mRNA expression, whereas no ets-2 mRNA wasdetected. Induction of c-ets-1 resulted in the increased expression ofGnT-V mRNA in the glioma cells, suggesting that the Ets-1 transcrip-tion factor directly controls the transcription of GnT-V in glioma cells.Thus, the data presented here add further support to the idea that Ets-1plays a pivotal role in modulating glioma invasivity via coordinatedexpression of aberrantb1,6-GlcNAc N-glycans on the glioma-asso-ciateda3b1 integrin and expression of metalloproteases (36).

To study the biological effects of aberrantb1,6-GlcNAc-bearingN-glycan in gliomas, theGnT-V gene was stably transfected intoU-373 MG glioma cells that express very low levels of this mRNA.As predicted from the results discussed above,GnT-V transfectantswere more invasive than controls. These transfectants showed thedistinct fan-shaped morphologies indicative of directional cell migra-tion with a distinct leading edge. It has been reported that smallnumbers of glycoproteins, particularly those involved in adhesion, canbe found at the leading lamellipodia in locomoting cells (37). In the

results reported here,a3b1 integrin was found to be localized on theleading lamellipodia of the GnT-V-transfected cells and focal adhe-sion sites radiated toward leading lamellipodia, whereas parental cellsor vector-transfected controls did not show characteristics of migrat-ing cells.

In contrast, GnT-III stable transfectants displayed decreased cellmigration under the conditions described above (data not shown).Although the data were not presented, this is likely because of anincrease in their adhesion to the fibronectin substratum used in thesestudies.

Thus, when all of the data presented here are taken in whole, itsuggests that: (a) cell surface expressed glycoproteins bearing “brain-type” bisectingb1,4-GlcNAc structures, the products of GnT-III, maybe directly involved in cell adhesion and migration; and (b) the shiftof N-glycans from bisecting to highly branchedb1,6-GlcNAc struc-tures on the glycoproteins may function to reduce adhesivity andincrease migration, thus increasing cell invasivity. The increasedinvasivity found in GnT-V-transfected clones may be attributable toaltered interaction betweena3b1 integrin and its laminin substrate,which is a matrix component in the invasion assays. The interactionbetweena3b1 integrin and appropriate substrata, such as laminin andfibronectin, may be dependent on theN-glycans.

To test this hypothesis,in vitro migration assays were performedusing E-PHA and L-PHA lectins, which bind to bisectingb1,4-GlcNAc or highly branchedb1,6-GlcNAc-bearingN-glycans on gly-coproteins, respectively. We have reported previously that E-PHAlectin had a marked effect on adhesion in U-373 MG cells (38). On theother hand, L-PHA lectin showed no effect on either cell adhesion(38) or cytotoxicity in glioma cells; cytotoxicity was seen in highlymetastatic tumor cell lines (11, 12). In solid phase cell migration(haptotaxis) studies, E-PHA lectin completely abolished glioma cellmigration on fibronectin substrata, regardless of the levels ofb1,6-GlcNAc expression in both U-373 MG transfectants and other gliomacell lines, whereas migration of glioma cells with high levels ofb1,6-GlcNAc N-glycans was weakly inhibited by L-PHA. Further-more, the inhibitory effect by E-PHA was comparable with that ofanti-a3 integrin monoclonal antibody. These data suggest thatb1,4-GlcNAc N-glycans play a direct role ina3b1 integrin-mediated celladhesion, whereas in gliomas, the observed shift to more highlybranchedb1,6-GlcNAc N-glycan reduces cell adhesivity and in-creases invasivity by replacing functionalb1,4-GlcNAc-bearingN-glycans on the adhesion molecules. The binding of E-PHA tob1,4-GlcNAc-bearingN-glycans interferes with cell adhesion (38), thus

Fig. 10. Inhibition of glioma cell migration onfibronectin substrate byP. vulgarisisolectins. Twomg/ml of E-PHA strongly inhibited cell migrationof parental U-373 MG cells on a fibronectin sub-stratum and completely abolished the migration ofthe transfectants. The inhibitory effect was similarwith 10 mg/ml of monoclonal anti-a3 integrin an-tibody (Chemicon; clone P1B5), whereas L-PHAshowed little effect. Twomg/ml of E-PHA alsoinhibited cell migration of D-54MG, SNB-19,SW1088, and U-87 MG glioma cells. At the sameconcentration, L-PHA showed little effect on cellmigration. The data are average values of two sep-arate experiments done in triplicate;bars,SD.

141




inhibiting cell migration as shown in this study. On the other hand,L-PHA binding tob1,6-GlcNAc-bearingN-glycans does not interferewith integrin function and, therefore, has little effect on cell migration.The results presented here are consistent with previous studies that:(a) N-glycans ona5b1 integrins are required for the functionalheterodimerization of integrina andb subunits (10); and (b) a shift ofintegrin N-glycans to highly branchedb1,6-GlcNAc leads to de-creased cell adhesion, resulting in an increase in cell motility byaltering the function ofa5b1 andavb3 integrins (11).

In conclusion, the data presented here show that a shift in theexpression of normal “brain type” bisectingb1,4-GlcNAc to highlybranchedb1,6-GlcNAc N-glycans plays an important role in modu-lating the function of cell surface glycoproteins involved in gliomainvasivity. A recent study suggests that the knock-out of theGnT-Vgene results in the suppression of both breast tumor formation andlung metastases in the null mouse (39). Likewise, the expression ofbisectingb1,4-GlcNAc N-glycans byGnT-III gene transfection hasbeen reported to suppress lung metastasis of B16 melanoma (40). Itwill be interesting to examine whether reversion from aberrantb1,6-GlcNAc-expressingN-glycans to normalb1,4-GlcNAc-bearingN-glycans can retard glioma invasivityin vivo.

ACKNOWLEDGMENTS

We gratefully acknowledge Donna Kersey and Michael McLone for tech-nical assistance and Kevin Cramer for assistance in obtaining fresh tumorspecimens.

REFERENCES

1. Hakomori, S. Tumor malignancy defined by aberrant glycosylation and sphingo(gly-co)lipid metabolism. Cancer Res.,56: 5309–5318, 1996.

2. Varki, A. Biological roles of oligosaccharides: all of theories are correct. Glycobi-ology, 3: 97–130, 1993.

3. Asada, M., Furukawa, K., Segawa, K., Endo, T., and Kobata, A. Increased expressionof highly branchedN-glycans at cell surface is correlated with the malignant pheno-types of mouse tumor cells. Cancer Res.,57: 1073–1080, 1997.

4. Yoshimura, M., Ihara, Y., Matsuzawa, Y., and Taniguchi, N. Aberrant glycosylationof E-cadherin enhances cell-cell binding to suppress metastasis. J. Biol. Chem.,271:13811–13815, 1996.

5. Dennis, J. W. Effects of swainsonine and polyinosinic:polycytidylic acid on murinetumor cell growth and metastasis. Cancer Res.,46: 5131–5136, 1986.

6. Humphries, M. J., Matsumoto, K., White, S. L., and Olden, K. Inhibition of exper-imental metastasis by castanospermine in mice: blockage of two distinct stages oftumor colonization by oligosaccharide processing inhibitors. Cancer Res.,46: 5215–5222, 1986.

7. Dennis, J. W., Koch, K., Yousefi, S., and Vander Elst, I. Growth inhibition of humanmelanoma tumor xenografts in athymic nude mice by swainsonine. Cancer Res.,50:1867–1872, 1990.

8. Paulus, W., Baur, I., Schuppan, D., and Roggerndorf, W. Characterization of integrinreceptors in normal and neoplastic human brain. Am. J. Pathol.,143: 154–161, 1993.

9. Paulus, W., and Tomm, J. C. Basement membrane invasion of glioma cells mediatedby integrin receptors. J. Neurosurg.,80: 515–519, 1994.

10. Zheng, M., Fang, H., and Hakomori, S. Functional role ofN-glycosylation ina5b1integrin receptor. J. Biol. Chem.,269: 12325–12331, 1994.

11. Demetriou, M., Nabi, I. R., Coppolino, M., Dedhar, S., and Dennis, J. W. Reducedcontact-inhibition and substratum adhesion in epithelial cells expressing GlcNAc-transferase V. J. Cell Biol.,130: 383–392, 1995.

12. Dennis, J., Waller, C., Timpl, R., and Schirrmacher, V. Surface sialic acid reducesattachment of metastatic tumor cells to collagen type IV and fibronectin. Nature(Lond.), 300: 274–276, 1982.

13. Dennis, J. W., Laferte, S., Waghorne, C., Breitman, M. L., and Kerbel, R. S.b1-6branching of Asn-linked oligosaccharides is directly associated with metastasis.Science (Washington DC),236: 582–585, 1987.

14. Dennis, J. D., and Laferte, S. Oncodevelopmental expression of -GlcNAcb1–6Mana1–6Manb1-branched asparagine-linked oligosaccharides in murine tissuesand human breast carcinomas. Cancer Res.,49: 945–950, 1989.

15. Fernandes, B., Sagman, U., Auger, M., Demetrio, M., and Dennis, J. W.b1-6branched oligosaccharides as a marker of tumor progression in human breast andcolon neoplasia. Cancer Res.,51: 718–723, 1991.

16. Yamamoto, H., Saito, T., Kaneko, Y., Kersey, D., Yong, V. W., Bremer, E. G.,Mkrdichian E., Cerullo, L., Leestma, J., and Moskal, J. R.a2,3-Sialyltransferase

mRNA anda2,3-linked glycoprotein sialylation are increased in malignant gliomas.Brain Res.,755: 175–179, 1997.

17. Yamamoto, H., Kaneko, Y., Rebbaa, A., Bremer, E. G., and Moskal, J. R.a2,6-Sialyltransferase gene transfection into a human glioma cell line (U-373MG) resultsin decreased invasivity. J. Neurochem.,68: 2566–2576, 1997.

18. Ihara, Y., Nishikawa, A., Tohma, T., Soejima, H., Niikawa, N., and Taniguchi, N.cDNA cloning, expression, and chromosomal localization of humanN-acetylglu-cosaminyltransferase III (GnT-III). J. Biochem.,113: 692–698, 1993.

19. Shoreibah, M., Perng, G-S., Adler, B., Weinstein, J., Basu, R., Cupple, R., Wen, D.,Browne, J. K., Buckhaults, P., Fregien, N., and Pierce, M. Isolation, characterization,and expression of a cDNA encodingN-acetylglucosaminyltransferase V. J. Biol.Chem.,268: 15381–15385, 1993.

20. Saito, H., Nishikawa, A., Gu, J., Ihara, Y., Soejima, H., Wada, Y., Sekiya, C.,Niikawa, C., and Taniguchi, N. cDNA cloning and chromosomal mapping of humanN-acetylglucosaminyltransferase V. Biochem. Biophys. Res. Commun.,198: 318–327, 1994.

21. Kang, R., Saito, H., Ihara, Y., Miyoshi, E., Koyama, N., Sheng, Y., and Taniguchi, N.Transcriptional regulation of theN-acetylglucosaminyl-transferase V gene in humanbile duct carcinoma cells (HuCC-T1) is mediated by Ets-1. J. Biol. Chem.,271:26706–26712, 1996.

22. Ko, J. H., Miyoshi, E., Noda, K., Ekuni, A., Kang, R., Ikeda, Y., and Taniguchi, N.Regulation of the GnT-V promoter by transcription factor Ets-1 in various cancer celllines. J. Biol. Chem.,274: 22941–22948, 1999.

23. Buckhaults, P., Chen, L., Fregien, N., and Pierce, M. Transcriptional regulation ofN-acetylglucosaminyltransferase V by the src oncogene. J. Biol. Chem.,272: 19575–19581, 1997.

24. Kepes, J. J. Review of the WHO’s proposed new classification of brain tumors.Proceedings of the XIth International Congress of Neuropathology, Japanese Societyof Neuropathology, Kyoto, Japan, September 2–8, 1990.

25. Watson, D. K., McWilliams, M. J., Lapis, P., Lautenberger, J. A., Schweinfest, C. W.,and Papas, T. S. Mammalianets-1andets-2genes encode highly conserved proteins.Proc. Natl. Acad. Sci. USA,85: 7862–7866, 1988.

26. Li, W-P., Zuber, C., and Roth, J. Use ofPhaseolus vulgarisleukoagglutinating lectinin histochemical and blotting techniques: a comparison of digoxigenin-and biotin-labelled lectins. Histochemistry,100: 347–356, 1993.

27. Rebbaa, A., Yamamoto, H., Saito, T., Meuillet, E., Kim, P., Kersey, D. S., Bremer,E. G., Taniguchi, N., and Moskal, J. R. Gene transfection-mediated overexpression ofb1,4-N-acetylglucosamine bisecting oligosaccharides in glioma cell line U373 MGinhibits epidermal growth factor receptor function. J. Biol. Chem.,272: 9275–9279,1997.

28. Hendrix, M. J. C., Seftor, E. A., Seftor, R. E. B., Misiorowski, R. L., Saba, P. Z.,Sundareshan, P., and Welch, D. R. Comparison of tumor cell invasion assays: humanamnionversusreconstituted basement membrane barriers. Invasion Metastasis,82:278–297, 1989.

29. Kawano, T., Takasaki, S., Tao, T. W., and Kobata, A. Altered glycosylation ofb1integrins associated with reduced adhesiveness to fibronectin and laminin. Int. J.Cancer,53: 91–96, 1993.

30. Yousefi, S., Higgins, E., Daoling, Z., Pollex-Kruger, A., Hindsgaul, O., and Dennis,J. W. Increased UDP-GlcNAc. Galb1–3GalNAc-R (GlcNAc to GalNAc)b1,6-N-acetylglucosaminyltransferase activity in metastatic murine tumor cell lines. J. Biol.Chem.,266: 1772–1782, 1991.

31. Hoffmann, A., Nimtz, M., Wurster, U., and Conradt, H. S. Carbohydrate structures ofb-trace protein from cerebrospinal fluid: evidence for “brain-type”N-glycosylation.J. Neurochem.,63: 2185–2196, 1994.

32. Kishima, H., Shimizu, K., Tamura, K., Miyao, Y., Mabuchi, E., Tominaga, E.,Matsuzaki, J., and Hayakawa, T. Monoclonal antibody ONS-M21 recognizes integrina3 in gliomas and medulloblastomas. Br. J. Cancer,79: 333–339, 1999.

33. Kola, I., Brookes, S., Green, A. R., Garber, R, R., Tymms, M., Papas, T. S., and Seth,A. The Ets1 transcriptional factor is widely expressed during murine embryo devel-opment and is associated with mesodermal cells involved in morphogenic processessuch as organ formation. Proc. Natl. Acad. Sci. USA,90: 7588–7592, 1993.

34. Wernert, N., Gilles, F., Fafeur, V., Bouali, F., Raes, M-B., Pyke, C., Dupressoir, T.,Seitz, G., Vandenbunder, B., and Stehelin, D. Stromal expression of c-Ets1 transcrip-tional factor correlates with tumor invasion. Cancer Res.,54: 5683–5688, 1994.

35. Wasylyk, B., Wasylyk, C., Flores, P., Begue, A., Leprince, D., and Stehelin, D. Thec-ets proto-oncogenes encode transcriptional factors that cooperate with c-fos andc-jun for transcriptional activation. Nature (Lond.),346: 191–193, 1990.

36. Uhm, J. H., Dooley, N. P., Villemure, J-G., and Yong, V. W. Mechanisms of gliomainvasion: role of matrix-metalloproteinases. Can. J. Neurol. Sci.,24: 1–15, 1997.

37. Kucik, D. F., Kuo, S. C., Elson, E. L., and Sheetz, M. P. Preferential attachment ofmembrane glycoproteins to the cytoskeleton at the leading edge of lamella. J. CellBiol., 114: 1029–1036, 1991.

38. Rebbaa, A., Yamamoto, H., Moskal, J. R., and Bremer, E. G. Binding of erythroag-glutinating phytohemagglutinin lectin fromPhaseolus vulgaristo the epidermalgrowth factor receptor inhibits receptor function in the human glioma cell line,U373MG. J. Neurochem.,67: 2265–2272, 1996.

39. Granovsky, M., Fata, J., Khokha, R., and Dennis, J. W.N-Acetylglucosaminyltrans-ferase V null mice show hypersensitivity to T cell mitogens and intestinal hyperplasia.Glycobiology,8: A156, 1998.

40. Yoshimura, M., Nishikawa, A., Ihara, Y., Taniguchi, S., and Taniguchi, N. Suppres-sion of lung metastasis of B16 mouse melanoma byN-acetylglucosaminyltransferaseIII gene transfection. Proc. Natl. Acad. Sci. USA,92: 8754–8758, 1995.

142




2000;60:134-142. Cancer Res Hirotaka Yamamoto, Jason Swoger, Suzanne Greene, et al. Gliomas: Implications for a Role in Regulating Invasivity

-Glycans in HumanN-Acetylglucosamine-bearing N1,6-β

Updated version

http://cancerres.aacrjournals.org/content/60/1/134

Access the most recent version of this article at:

Cited articles

http://cancerres.aacrjournals.org/content/60/1/134.full#ref-list-1

This article cites 38 articles, 22 of which you can access for free at:

Citing articles

http://cancerres.aacrjournals.org/content/60/1/134.full#related-urls

This article has been cited by 25 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/60/1/134To request permission to re-use all or part of this article, use this link



http://cancerres.aacrjournals.org/content/60/1/134.full#ref-list-1

http://cancerres.aacrjournals.org/content/60/1/134.full#related-urls

http://cancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]



b1,6-n-acetylglucosamine-bearingn-glycans in human...

Documents