www.bba-direct.com

Biochimica et Biophysica Acta 1643 (2003) 113–123

Glycosylation profile of integrin a3h1 changes with melanoma progression

Ewa Pocheca, Anna Litynskaa,*, Angela Amoresanob, Annarita Casbarrab

a Institute of Zoology, Jagiellonian University, R. Ingardena 6, 30-060 Cracow, PolandbDipartimento di Chimica Organica e Biochimica, Complesso Universitario Monte S’Angelo, Via Cinthya 8, 80126 Naples, Italy

Received 22 July 2003; received in revised form 10 October 2003; accepted 15 October 2003

Abstract

Glycosylation of integrins has been implicated in the modulation of their function. Characterisation of carbohydrate moieties of a3 and h1

subunits from non-metastatic (WM35) and metastatic (A375) human melanoma cell lines was carried out on peptide-N-glycosidase F-released

glycans using matrix-assisted laser desorption ionization mass spectrometry (MALDI-MS). h1 integrin subunit from both cell lines displayed

tri- and tetraantennary oligosaccharides complex type glycans, but only in A375 cell line was the sialylated tetraantennary complex type glycan

(Hex7HexNAc6FucSia4) present. In contrast, only a3 subunit from metastatic cells possessed h1–6 branched structures. Our data indicate thatthe h1 and a3 subunits expressed by the metastatic A375 cell line carry h1–6 branched structures, suggesting that these cancer-associated

glycan chains may modulate tumor cell adhesion by affecting the ligand binding properties of a3h1 integrin. In direct ligand binding assays,

a3h1 integrin from both cell lines binds strongly to fibronectin and to much lesser degree to placental laminin. No binding to collagen IV was

observed. Enzymatic removal of sialic acid residues from purified a3h1 integrin stimulates its adhesion to all examined ECM proteins. Our data

suggest that the glycosylation profile of a3h1 integrin in human melanoma cells correlates with the acquisition of invasive capacity during

melanoma progression.

D 2003 Elsevier B.V. All rights reserved.

Keywords: a3h1 integrin; Glycosylation; MALDI MS; Cell adhesion; ELISA

1. Introduction

Cutaneous melanoma is one of the fastest-rising malig-

nancies in the last several decades. The multi-stage nature of

melanoma development provides an attractive model for

studying the transition from benign lesion to metastatic

cancer [1].

Many studies revealed that tumor cells often show a

decrease in cell–cell and cell–matrix adhesion correlated

with progression of malignancy. Integrins are the major

group of adhesion molecules responsible for mediating

cellular interactions with the extracellular matrix and appear

to have an important role in various aspects of cancer.

Essentially all types of cells show integrin-dependent adhe-

0167-4889/$ - see front matter D 2003 Elsevier B.V. All rights reserved.

doi:10.1016/j.bbamcr.2003.10.004

Abbreviations: MALDI MS, matrix-assisted laser desorption ionization

mass spectrometry; PNGase F, peptide-N-glycanase F; GlcNAc-TV, N-

acetylglucosaminyltransferase; GlcNAc, N-acetylglucosamine; Man, man-

nose; Gal, galactose; Sia, sialic acid; Fuc, fucose; GNA, Galanthus nivalis

agglutinin; DSA, Daturia stramonium agglutinin; SNA, Sambucus nigra

agglutinin; MAA, Maackia amurensis agglutinin; AAA, Aleuria aurantia

agglutinin; PHA-L, Phaseolus vulgaris agglutinin

* Corresponding author. Tel.: +48-12-6336377; fax: +48-12-6343716.

E-mail address: lita@zuk.iz.uj.edu.pl (A. Litynska).

sion based on binding of ECM components to various types

of integrin [2,3]. Most integrins which are a family of a/h-heterodimers are able to bind different ligands with different

affinities. It seems clear that both a and h subunits extracel-

lular domains, which are heavily glycosylated, contribute to

the formation of the binding site. The affinity of integrins for

their ligands may vary depending on the cell type in which

they are expressed or as a result of conformational changes.

Glycosylation of integrins has been implicated in the mod-

ulation of their affinity.

Integrin a3h1 is widely distributed in almost all tissues,

and it has been suggested to be involved in tumor invasion

[4,5]. It has been reported to be a receptor for laminins [6,7],

fibronectin [8] and collagen [9]. However, recent works

suggested that it has a more restricted set of ligands [10].

Now a3h1 is considered primarily as a laminin-5 receptor

[11,12] and as a weak receptor for other ECM components

[13]. In contrast to integrin a6h4, another laminin-5 receptor,

integrin a3h1, is considered to play an active role in cell

spreading and motility on laminin-5 [14]. More recently, it

has been shown that a3h1 can form a stable and specific

association with the transmembrane-4 family member

CD151. This association provides a linkage with phospha-

E. Pochec et al. / Biochimica et Biophysica Acta 1643 (2003) 113–123114

tidylinosytol 4-kinase that regulates cell motility [15]. In

addition, a3h1 can pair with other tetraspanin proteins, which

in turn stimulates the production of MMP-2 [16]. The role of

these associations in melanoma has not been fully explored

but they may be important in tumor cell dissemination. a3h1

has been shown to be up-regulated in several melanoma cell

lines [4]. It is likely that a3h1 plays significant role in

melanoma development and tumor invasion, however, the

mechanism through which it modulates cell motility is not

yet understood. Furthermore, there is evidence suggesting

that a3h1 may also regulate a number of different cell

functions even without direct interaction with ECM, but

influencing cell behaviour via other integrins or via proteo-

litic ECM remodelling [17]. There is a growing amount of

evidence that integrin a3h1 transdominantly regulates integ-

rin a6h1 in mouse keratinocytes [18,19], or negatively

regulates integrin a2h1 in human breast carcinoma cell lines

[20].

Glycosylation is generally altered in tumor cells in

comparison with their normal counterparts. Post-translation-

al modulations of integrins include glycosylation of both a

and h subunits. Several roles for integrin glycosylation have

been suggested. The number and structure of N-linked

oligosaccharides is an important regulator of integrin func-

tion. This was particularly well-illustrated for h1 integrins

function in colon carcinoma [21], melanoma cell metastasis

[22–24], as well as human fibrosarcoma [25]. Multiple

reports have suggested that N-glycosylation of h1 integrins

is essential for optimal function [26–29].

Fig. 1. The reaction of integrin a3 and h1 subunits with digoxigenin-labeled lecti

markers (HMW, Sigma). For immunodetection, 3 Ag GDP peptide-bound materia

The N-linked glycosylation of human a3h1 has also been

analysed. In colon cancer, a3h1 integrin is a sialoglycopro-

tein containing h1–6 branched N-linked oligosaccharides

and short poly-N-acetyllactosamine structures. This is a

major carrier of oncodevelopmental and cancer-associated

carbohydrate epitopes [30]. Recently, we have shown by

matrix-assisted laser desorption ionization mass spectrome-

try (MALDI MS) the presence of complex type oligosac-

charides with a wide heterogeneity in both subunits of a3h1

from human ureter epithelium cells [31]. However, the exact

function of integrin a3h1 oligosaccharides remains to be

determined. In the present study, we found that integrin a3h1

from human melanoma cell lines with different metastatic

abilities differs in glycosylation profiles in a way that is

characteristic of invasive tumor cells. Moreover, in direct

adhesion assay, we provide evidence that enzymatic removal

of sialic acid residues from purified integrin a3h1 stimulates

its adhesion to fibronectin (FN) laminins: mouse and pla-

cental (LN) and collagen IV (Col). Our results suggest that

changes in integrin a3h1 glycosylation profiles may play an

important role in some steps of melanoma progression.

2. Materials and methods

2.1. Chemicals

A Glycan Differentiation Kit containing the digoxigenin

(DIG)-labeled lectins and N-glycosidase F from Flavobac-

ns: GNA, DSA SNA, MAA, AAA and PHA-L. Line S—molecular weight

l was used, and for lectin analysis—15 Ag.

Fig. 2. Western blotting analysis of a3h1 integrin treated with neuramini-

dase. GD6 peptide purified a3h1 integrin was digested with neuraminidase,

separated by SDS-PAGE and analysed by Western blotting with specific

antibodies for a3 (A) and h1 (B) subunit.

Fig. 3. Western blotting analysis of a3h1 integrin treated with PNGase and swainso

for 24 h, harvested in PBS and homogenized by sonification. GD6 peptide purifi

analysed by Western blotting with specific antibodies for a3 (A) and h1 (B) subu

E. Pochec et al. / Biochimica et Biophysica Acta 1643 (2003) 113–123 115

terium meningosepticum were purchased from Boehringer

(Mannheim). Neuraminidase from Arthrobacter ureafaciens

was from Oxford GlycoSciences. PVDF membranes were

from Millipore. Collagen IV, fibronectin, laminin ESH

(mouse Engelbreth–Swarm–Holm), BSA solution, crystal

violet, proteinase inhibitor cocktail, fucose, Phaseolus vul-

garis agglutinin (PHA-L) were from Sigma Chemical Co.

All remaining chemicals were of analytical grade. Rabbit

polyclonal antisera against a3 subunit, mouse monoclonal

antisera against h1 subunit and human placental laminin

(hLN) were obtained from Chemicon.

2.2. Cell lines and culture conditions

Melanoma cell lines were kindly donated by Prof. A.

Mackiewicz from the Department of Cancer Immunology,

University School of Medical Sciences at Great Poland

Cancer Center, Poznan, Poland. The WM35 cell line was

from primary tumor (established by Meenhard Herlyn, The

Wistar Institute, Philadelphia) and A375 (ATCC-CRT(L)-

1619) was from metastatic site.

The cell lines were cultured in medium RPMI 1640

(Sigma) containing 10% fetal bovine serum (FBS)

(GibcoBRLk) and antibiotics (penicillin—100 U/ml, strep-

tomycin—100 Ag/ml, Polfa Tarchomin, Poland). Cells were

grown in monolayers at 95% air/5% CO2 atmosphere at 37

jC in a humidified incubator. After reaching confluence, the

nine. Cells were grown in the presence of swainsonine (10 Ag /1 ml RPMI)

ed a3h1 integrin was digested with PNGase, separated by SDS-PAGE and

nit.

E. Pochec et al. / Biochimica et Biophysica Acta 1643 (2003) 113–123116

cells were washed twice and then harvested in phosphate

buffered saline. The cell pellets were homogenized in 50

mM Tris–HCl containing 15 mM NaCl, 1 mM MgCl2, 1

mM MnCl2, pH 7.4, and protease inhibitor cocktail (Sigma)

by sonification (five times, 5 s each) (Sonies, Vibra-Cell),

extracted for 1 h on ice with the same buffer containing

additionally 30 mM octyl-h-D-glucopyranoside and 3%

protamine sulfate. Finally, the cell extracts were clarified

by centrifugation at 35,000� g for 1 h (L7-65, Beckman).

2.3. Affinity chromatography

GD6 peptide (residues 3011–3032 of the globular

domain of the laminin A chain) [6] was synthesised at

Fig. 4. Positive-ion MALDI MS mass spectrum of N-glycans obtained by PNGase

cell line (B), a3 subunit from A375 cell line (C), h1 subunit from A375 cell line

the Faculty of Chemistry, University of Gdansk, Poland.

This peptide was coupled to activated CH-Sepharose

(Sigma) according to the manufacturer’s instruction (Phar-

macia Fine Chemicals AB). Isolation of a3h1 integrin was

performed as described in detail in Ref. [31]. Briefly,

about 5 mg of protein was loaded onto the column,

allowed to interact with the immobilised ligand overnight

at 4 jC and the bound material was eluted with 20 mM

EDTA in 50 mM Tris–HCl, pH 7.4 containing 30 mM

octyl-h-D-glucopyranoside. Fractions eluted with EDTA

were collected, dialysed against water and concentrated

to 0.5 ml by lyophilisation and then analysed by SDS-

PAGE under nonreducing conditions and probed with

specific antibodies.

F digestion of: a3 subunit from WM35 cell line (A), h1 subunit from WM35

(D).

Fig. 4 (continued).

E. Pochec et al. / Biochimica et Biophysica Acta 1643 (2003) 113–123 117

2.4. SDS-PAGE and immunodetection of a3 and b1 subunits

For deglycosylation of the a3 and h1 subunits prior to

SDS-PAGE, 3 Ag of the material eluted from GD6-Sephar-

ose column was denatured by heating at 100 jC for 2 min in

5 Al of buffer (20 mM sodium phosphate, pH 7.5 containing

50 mM EDTA, 0.5% SDS, 5% mercaptoethanol). After

cooling, 1 Al of 10% non-ionic detergent (Nonidet P40),

followed by 0.35 U of peptide-N-glycanase F (PNGase F)

(Boehringer) were added and the solution was incubated

at 37 jC overnight. Material eluted from GD6-Sepharose

before or after deglycosylation was separated by 8% SDS-

PAGE in nonreducing condition according to Laemmli [32]

and transferred to a PVDF membrane, by electrophoretic

blotting [33] for 1 h at 100 V. The efficiency of protein

transfer was at least 95% as checked by staining the gel with

Coomassie Brilliant Blue R-250. Proteins immobilised on

PVDF membranes were stained with Ponceau S followed by

destaining in H2O. The blots were blocked in TBS/Tween

(0.02 M Tris–HCl, pH 7.6, containing 0.15 M NaCl and

0.1% Tween 20), with 1% bovine serum albumin (BSA).

Afterwards, the membranes were sequentially incubated

with specific antibodies diluted in TBS/Tween with 1%

BSA (1:1000 for a3 and h1 subunits) for 2 and 18 h,

respectively. Mouse monoclonal antibodies to human integ-

rin subunit h1 (clone B3B11) from Chemicon, and rabbit

polyclonal antibodies to human a3 subunit (Chemicon) were

used. After triple wash with TBS/Tween, the blots were

incubated with alkaline phosphatase coupled secondary goat

anti-mouse or anti-rabbit Ig (Boehringer) (1:500 in TBS/

Tween with 1% BSA) for 1 h and the bands were localised

with 4-nitro blue tetrazolium chloride as a substrate.

2.5. Glycan chain analysis

Individual protein bands corresponding to the a3 and h1

subunits were excised from the PVDF membrane and

further glycan analysis was carried out according to the

method described by Kuster et al. [34] as modified by Hoja-

æukowicz et al. [35], or was performed with the use of

E. Pochec et al. / Biochimica et Biophysica Acta 1643 (2003) 113–123118

Glycan Differentiation Kit (Boehringer) according to Hasel-

beck et al. [36] as described in detail earlier [37].

2.5.1. Protein alkylation

The excised PVDF membrane pieces were placed into

Eppendorf tubes and washed twice with 1 ml of 20 mM

NaHCO3, pH 7.0, for 15 min each. The wash was discarded

and dithiothreitol was added and the protein was reduced at

60 jC for 30 min. After cooling to room temperature, 20 Al of100 mM iodoacetamide was added and the protein was

alkylated for 30 min at room temperature in the dark. The

reducing and alkylation reagents, as well as residual SDS

were then removed by incubation in 1:1 acetonitrile/fresh 20

mMNaHCO3, pH 7.0, for 60 h. Subsequently, the membrane

pieces were incubated in blocking solution (Boehringer).

2.5.2. In situ digestion

Prior to deglycosylation, PVDF membrane pieces were

washed three times with 20 mM NaHCO3, pH 7.0, for 15

min each. The washings were discarded and replaced with

0.5 U of PNGase F in 30 Al of 20 mM NaHCO3, pH 7.0, and

incubated at 37 jC for 12–16 h [38].

2.5.3. Sugar extraction

After deglycosylation, the incubation buffer was com-

pletely dried in a SpeedVacR Plus (Savant) and the residue

dissolved in 10 Al of ultrapure water (Milli-Q Plus, Millipore,

Bedford, CA, USA) and applied to microcolumn clean-up.

2.5.4. Microcolumn clean-up of sugars

Prior to MALDI MS, a microcolumn consisting of an

Eppendorf GELoader pipette tip packed with 5 Al each of

AG-3 (OH� form, bottom) and AG-50 (H+ form, top) was

used. The column was washed with 100 Al of water and an

aliquot of the sugar sample was applied. Glycans were eluted

with 100 Al of water and dried in a SpeedVac.

2.5.5. MALDI mass spectrometry

MALDI mass spectra were recorded by using a Voyager

DE-PRO MALDI-TOF mass spectrometer (Applied Biosys-

tem) equipped with delayed extraction. The MALDI matri-

ces were prepared by dissolving 25 mg of 2,5-dihy-

droxybenzoic acid (DHB) in 1 Al of acetonitrile/0.2%

trifluoroacetic acid (70:30 v/v). Typically, 1 Al of the analytewas then added. Mass calibration was performed with the

MH+ ion of insulin set atm/z 5734.6 and a known peptide ion

at m/z 1209.7. Raw data were analysed using the computer

software provided by the manufacturer, and are reported as

average masses.

2.6. Glycosidase digestions

a3h1 integrin GD6-Sepharose purified was diluted in

appropriate buffer and treated with 0.5 mU of PNGase F in

10 Al of 20 mM Na2HPO4, pH 7.5 or 10 mU of neuramin-

idase in 9 Al of 10 mM Tris–HCl, pH 6.5, containing 150

mMNaCl at 37 jC for up to 18 h. As a control, an equivalent

of a3h1 integrin was incubated in the absence of appropriate

enzyme. Following incubation, the neuraminidase buffer

was adjusted to pH 7.4 by dilution with TBS/Mg buffer

(10 mM Tris–HCl, pH 7.4 containing 150 mM NaCl, 2 mM

MgCl2, 1 mMMnCl2) or TBS/EDTA (10 mM Tris–HCl, pH

7.4 containing 150 mM NaCl, 10 mM EDTA).

2.7. Modified ELISA integrin binding assay

Extracellular matrix proteins: hLN, ESH LN, Col, FN (10

Ag/ml) in TBS/Mg buffer were coated onto the microtiter

plate at 4 jC overnight. After being coated, the plate was

washed with the TBS/Mg buffer and non-specific binding

sites were blocked with a 1% solution of BSA in the TBS/Mg

buffer for 1 h in 37 jC. All the following steps were

performed at room temperature. Affinity purified integrin

a3h1 neuraminidase-treated, PNGase-treated or non-treated

were diluted in TBS/Mg buffer (3 Ag/well) and was added toligand-coated microtiter wells and allowed to adhere for 1

h at 37 jC. The wells were then washed three times with

TBS/Mg buffer or TBS/EDTA (non-specific binding). The

anti-a3 polyclonal antibody (Chemicon) was subsequently

added to the wells (1:1000 diluted in TBS/Mg buffer), and

the samples were incubated for additional 90 min at 37 jC.After the wells were washed three times with the TBS/Mg

buffer, secondary goat anti-rabbit IgG antibody coupled to

alkaline phosphatase diluted 1:1000 in TBS/Mg buffer was

applied for 1 h. After three washings with TBS/Mg buffer,

ELISA was developed by adding pNpp (Sigma) and the

absorbance was measured at 405 nm after 18 h of incubation.

Values for non-specific binding, measured in the presence

of 10 mM EDTA were subtracted from total binding to

give values for the specific binding of integrin a3h1 to

immobilised ligands. Each value was determined in triplicate

and was repeated three times in independent experiments.

The significance of the differences between mean values

were computed using Dunkan’s new multiple range test.

*PV 0.05.

2.8. Other methods

Protein content was determined by the dye-binding meth-

od [39] using BSA as a standard. For protein sequencing,

affinity purified integrin was SDS-PAGE separation sub-

jected. The gel was stained with Coomassie Blue R-250,

destained and the proteins bands that migrated with an

apparent size of 168 and 130 kDa were cut out, dried and

then digested with tripsin and subjected to partial protein

sequence analysis by MS/MS.

3. Results

In order to isolate the a3h1 integrin from human mela-

noma cell lines we used affinity chromatography on GD6

E. Pochec et al. / Biochimica et Biophysica Acta 1643 (2003) 113–123 119

peptide. Electrophoretic analysis under nonreducing condi-

tions of the resulting preparation revealed that the major

species migrating with apparent sizes of 168 kDa correspond

to both a3, and 140, 138 kDa correspond to h1 subunits from

WM35 and A375 cell lines, respectively. The identity of both

subunits were further confirmed by Western blotting with

specific anti-human antibodies and by peptide tripsin diges-

tion followed by MS/MS analysis.

3.1. Glycosylation pattern

The glycosylation patterns of isolated a3 and h1 subunits

were first examined with digoxigenin-labeled lectins (Fig. 1)

as previously described [40]. a3h1 oligosaccharides from

WM35 cells reacted with each used lectin except a3 subunit

with PHA-L. These results suggested the presence of high-

mannose type glycans as well as complex type with di-, tri-

and tetra antenna in case of h1 subunit. Part of the glycans

were also sialylated (reaction with Sambucus nigra aggluti-

nin (SNA) and Maackia amurensis agglutinin (MAA)) and

fucosylated (reaction with Aleuria aurantia agglutinin

(AAA)). In case of A375 cells, a3 subunit reacted with all

lectins except MAA, while h1 subunit reacted with all lectins

except Galanthus nivalis agglutinin (GNA). Comparison of

two cell lines indicated that in more invasive A375 cell line,

the glycosylation profile changed into structures more

Table 1

Hypothetical oligosaccharide structures observed in MALDI MS spectra of a3 an

WM35 A375

a3 Subunit h1 Subunit a3 Subunit

Number of

peak

Mass

(m/z)

Number of

peak

Mass

(m/z)

Number of

peak

1

1 1057.5

2 1569.5

3 1626.5

4 1666.6

2

1 1702.6

(K+, Na+)

2 1744.1

(Na+)

5 1745.5 3

3 1882.1 6 1880.6

4 1995.2

(Na+)

7 1995.2

(Na+)

4

5 2108.7

(K+, Na+)

8 2108.7

(K+, Na+)

5

6 2256.4

(K+, Na+)

9 2256.4

(K+, Na+)

6

2257.4 2257.4

10 2585.3

11 2786.2

12 2825.2

HexNAc—N-acetyl-D-glucosamine, Hex—D-mannose or D-galactose, Sia—sialic

branched (positive reaction of a3 subunit with PHA-L).

The presence or absence of sialic acid moieties as well as

branched oligosaccharides in a3 and h1 subunits was also

confirmed by neuraminidase digestion or cell culture in the

presence of swainsonine. Untreated and neuraminidase-trea-

ted purified a3h1 integrin were resolved by SDS-PAGE, then

Western blotting to detect the a3 and h1 subunits. As shown

in Fig. 2, neuraminidase-treated integrin demonstrated a

more rapid electrophoretic motility, consistent with the loss

of sialic acid. The differences in molecular weights were

8 and 10 kDa for a3, and 6 and 10 kDa for h1 subunits in

WM35 and A375 cell lines, respectively.

In the WM35 cell line, upon swainsonine treatment, only

h1 subunit showed decrease in size about 8 kDa suggesting

that only this subunit of integrin a3h1 possesses branched

oligosaccharides. On the opposite, in A375 line, a3 subunit

decreases in size about 11 kDa, while in h1, by about 18 kDa

showing that both are substrates for N-acetylglucosaminyl-

transferase (GlcNAc-TV) (Fig. 3).

Positive ion spectra were acquired by MALDI MS for the

total glycan pool released from the a3 (Fig. 4A,C), and h1

(Fig. 4B,D) subunits of WM35 and A375 cell lines, respec-

tively. Masses composition and the structures of glycans

released by PNGase F are listed in Table 1. The signals at

m/z 1744.1 and 1882.1 as well as 1719, 1883.4, 1880.6 likely

originate from high-mannose type glycans of the a3 and h1

d h1 subunit from WM35 and A375 cell lines

Hypothetical

h1 Subunitstructure

Mass

(m/z)

Number of

peak

Mass

(m/z)

1477.4

(K+)

1 1477.4

(K+)

Hex5HexNAc3

Hex3HexNAc2Fuc

Hex4HexNAc3Sia

Hex4HexNAc4Fuc

Hex3HexNAc5Fuc

1687.4

(Na+)

Hex3HexNAc5Fuc

Hex5HexNAc4

1719.3 2 1720.2 Hex8HexNAc2

3 1883.4 Hex9HexNAc21995.4

(Na+)

4 1995.4

(Na+)

Hex4HexNAc5Sia

2109.5

(K+, Na+)

5 2109.5

(K+, Na+)

Hex5HexNAc6

2256.4

(K+, Na+)

6 2256.4

(K+, Na+)

Hex5HexNAc6Fuc

2257.4 2257.4 Hex4HexNAc6SiaS

Hex6HexNAc5Sia

Hex6HexNAc6Sia2Hex8HexNAc6Sia

7 3724.2

(K+)

Hex7HexNAc6FucSia4

acid, Fuc–L-fucose, S—sulfate group.

E. Pochec et al. / Biochimica et Bioph120

subunits, respectively. These structures were present in each

subunit and in both cell lines. Them/z value of 1476.4 as well

as 1477.4 in both a3 and h1 subunits from A375 correspond

to hybrid type glycans. In WM35 cells, a3 subunit had a

complex diantennary glycan as was shown by the presence of

signal at m/z 1702.6, while in A375 cells, the signal at m/z

1687.4 could correspond to triantennary complex type gly-

cans. The signals atm/z 1995.4, 2109.5 and 2254.6 of both h1

subunits and a3 from A375 line may be due to complex type

glycans with tri- or tetraantenna or alternatively to a complex

biantennary structures with a polylactosaminyl residues. The

m/z value of 1687.4 is only characteristic for subunit a3 from

A375 cell line and may be attributed to triantennary fucosy-

lated glycan. In fact, as it was shown by PHA-L reaction, only

a3 subunit fromA375 line was PHA-L-positive. For the same

reason, the structure Hex4HexNAc5Sia obtained for WM35

line a3 subunit could be attributed to HexNAch(1 –4)Mana(1–3)Man complex triantennary type structure. Also

m/z value of 3724.2 was only observed in h1 subunit from

A375 cells and could correspond to sialylated tetrantennary

complex type glycan with fucose.

Fig. 5. Ligand binding specificity of purified a3h1 integrin PNGase and

neuraminidase treated from WM35 (A) and A375 (B) cell lines. Ligands

(hLN—placental laminin, ESH LN—laminin ESH frommouse Engelbreth–

Swarm–Holm, Col—collagen IV, FN—fibronectin) were coated at 10 Al /ml

to the microtiter plate. Purified a3h1 integrin (3 Ag) was digested with

PNGase or neuraminidase and allowed to bind to the ligands in the presence

of either divalent cations (2 mMMgCl2 and 1 mMMnCl2) or 10 mM EDTA.

Integrin binding to the ligands was quantitated using amodified ELISA assay

as described in Materials and methods. The average of triplicate deter-

minations are shown: *PV 0.05. The significance of the differences between

mean values were computed using Dunkan’s new multiple range test.

3.2. Ligand binding specificity of the soluble a3b1 integrin

With the isolated a3h1 integrin direct protein–protein

binding studies were possible. As indicated in Fig. 5A,

soluble a3h1 integrin from WM35 cell line bound strongly

to FN and to a much lesser degree to hLN. Very weak

adhesion to LN and ESH LN was also observed. Integrin

a3h1 from A375 line bound to hLN and rather weak to ESH

LN and FN (Fig. 5B). No binding to Col IV of a3h1 from

both cell lines were observed. As shown in Fig. 5, desialy-

lation of a3h1 integrin significantly enhanced attachment to

all examined ECM proteins in both cell lines. De-N-glyco-

sylation enhanced attachment but to a lesser extent than

desialylation, while in the cases of WM35 line to FN and

A375 line to hLN de-N-glycosylation were without effect. To

verify the degree of desialylation and deglycosylation, the

subunits were Western blotted to assay for changes in

electrophoretic mobility. As shown in Figs. 2 and 3, neur-

aminidase- and PNGase F-treated subunits had a more rapid

electrophoretic mobility than untreated subunits consisted

with the loss of sialic acid residues and the N-glycan

moieties.

ysica Acta 1643 (2003) 113–123

4. Discussion

In this paper, we have shown that integrin a3h1 from

human melanoma cell lines with different metastatic abili-

ties differ in glycosylation profiles. In noncutaneous meta-

static A375 cell line, highly tumorogenic in nude mouse,

both subunits showed expression of carbohydrate epitopes

(Hex7HexNAc6FucSia4, Hex3HeNAc5Fuc), which may

contribute to cancer progression.

Metastasis of tumor cells is associated with the aberrant

glycosylation profile of cell surface glycoproteins [41–43],

in particular that of terminal sialylation [23] and h1–6branched N-linked oligosaccharides with poly-N-acetyllac-

tosamine structures presence [30,44]. There is increasing

evidence that progression of cancer from a tumorogenic to

metastatic phenotype is directly associated with increased

level of h1–6 branching N-linked oligosaccharides as the

result of hyperactivity of GlcNAc-TV [42,45,46]. These

oligosaccharides which have the structure GlcNAc h(1–6)Mana 1–6Man-R can be characterised by specific binding

to the lectin leucoagglutinin (PHA-L) [47]. Interestingly,

only a small number of glycoproteins are suitable substrates

for GlcNAc-TV activity. Preferred substrates are LAMP-1,

LAMP-2 [48], and CEA isolated from a number of colon

tumors [49–51]. Lastly, there is a growing amount of

evidence that the h1 integrin subunit could also be a substrate

for GlcNAc-TV [30,41,46,52].

In the present study, h1 integrin subunits from both non-

metastatic and metastatic cell lines displayed tri- and tet-

raantennary oligosaccharide complexes, but only in A375

cell line where sialylated tetraantennary complex type glycan

(Hex7HexNAc6FucSia4) was present. Additionally, the pres-

E. Pochec et al. / Biochimica et Biophysica Acta 1643 (2003) 113–123 121

ence of h1–6 branched structures on h1 subunit from both

cell lines was confirmed by reaction with PHA-L and by the

fact that from swainsonine-treated cells h1 subunit moved

slightly faster on SDS-PAGE. In contrast, a3 subunit only

from metastatic cell line A375 showed reaction with PHA-L

and moved faster on SDS-PAGE following swainsonine

treatment, indicating that it could also be a substrate for

GlcNAc-TV. a3 subunit from WM35 cell line did not react

with PHA-L, lectin-specific for h1–6 linked oligosacchar-

ides, and swainsonine did not influence its electrophoretic

mobility. Bearing in mind these results, it is more probable

that signals at m/z 2107.2 and 2254.6 correspond to N-

acetyllactosaminyl moieties than complex tetraantennary

structures. Also positive reaction of this subunit withDaturia

stramonium agglutinin (DSA)—lectin specific for short N-

acetyllactosamin residues—appears to support this observa-

tion. In contrast, in case of a3 subunit from A375 cell line, all

glycoforms at m/z 2107.2 and 2254.6 are possible. Our data

indicate that the a3 and h1 subunits expressed by the human

melanoma cell line (A375) carries h1–6 branched structures,suggesting that these cancer-associated glycan chains may

modulate tumor cell adhesion by affecting the ligand binding

properties of a3h1 integrin.

A number of questions concerning the relationship be-

tween malignant phenotype and h1–6 branching remains to

be answered, but the growing line of evidence shows that

the large Asn-linked oligosaccharides may reduce cell

adhesion and facilitate cell migration and invasion. Expres-

sion of h1–6 GlcNAc-branched oligosaccharides on integ-

rins or other adhesion receptors may facilitate the turnover

of cell–cell and ECM contacts to accelerate cell motility

[42]. Studies on several adhesion-mediated glycoproteins

support this view [41,48]. EJ-ras-transformed NIH-3T3

fibroblasts acquired migratory phenotype accompanied by

overexpression of a6h1 integrin bearing oligosaccharides

recognised by PHA-L [41]. In human hepatocarcinoma

cells, increases of h1–6 GlcNAc branching by overexpres-

sion of GlcNAc-TV significantly affect the function of

integrins leading to increased cell migration and invasion

through matrigel [46].

Although the question of whether the addition of h1–6branching on integrin a3h1 glycans can modulate its function

is not known, there is evidence suggesting that N-linked

carbohydrates can interfere with protein–protein interac-

tions. Various extracellular matrix proteins have been pro-

posed to be ligands for the a3h1 integrin but it was often

done on the basis of cell adhesion studies. The role of VLA-3

in adhesion to ECM proteins has been difficult to assess

since its specificity for ligands depends on the presence of

other integrins that may display a higher activity for the same

ligands [10,19]. Here with the purified a3h1 integrin, their

adhesion properties to LN, FN, and Col in direct binding

studies were possible. In order to understand the relationship

between carbohydrate structure and adhesion properties, we

have investigated the effect of desialylation and de-N-gly-

cosylation of a3h1 integrin.

Interestingly, most of the suspected ligands showed very

weak or no binding to purified from human melanoma cell

lines a3h1 integrin. It did not recognise mouse laminin or

collagen IV and minimally recognised human laminin. Our

results are partly consistent with adhesion study demonstrat-

ing that recombinant soluble a3h1 integrin bound strongly to

laminin-5 and placental laminin, but did not bind to collagen

IV and fibronectin [53]. Also, Weitzman et al. [54] have

found that a3 transfected cell lines did not adhere to FN, LN

and Col and suggested that relatively weak adhesion to these

proteins may be suitable for migration of these cells. Our

results are consistent with cell adhesion studies which

previously demonstrated that a3h1 integrin is involved in

invasion and migration of melanoma cells, not in adhesion.

Antibodies to the a3 subunit did not inhibit adhesion of

melanoma cells to LN, FN or Col IV, in contrast, the same

antibodies were good inhibitors of migration of the cells,

suggesting a special role for a3h1 integrin in the migration

and invasion of melanoma cells [4]. In our study, a3h1

integrin did not adhere to Col IV despite a previous report

that a3h1 integrin from melanoma and breast cancer directly

bind to a1(IV) 531–543 peptide of Col IV [55]. However,

enzymatic removal of sialic acid restored adhesion of a3h1

integrin to Col IV and enhanced adhesion to LN and FN.

The presence of sialic acid on a glycoconjugate has two

important properties that make it likely to be involved in the

regulation of binding. Sialic acid is located only in a

terminal position on the glycan and it is negatively charged.

Because of their terminal localisation and negative charge,

sialic acids have the potential to inhibit interactions between

molecules and even cells. In N-linked glycoproteins, sialic

acids are attached to Gal with a2,6 and a2,3 linkages. In our

study, from lectin probing it was shown that h1 subunit form

both cell lines and a3 subunit from WM35 line has both

types of sialic acid linkage, while a3 subunit from A375 line

had only a2,6 linkage. Tumor-associated alteration in a2,6

sialylation of N-glycans have been described in a number of

studies. Lin et al. [56] have shown that cell surface a2,6

sialylation contributes to cell–cell and cell–ECM adhesion

in breast carcinoma cells. The cells carrying increased

amounts of sialic acid adhered preferentially to collagen

IV and showed reduced cell–cell adhesion and enhanced

invasion capacity. Accumulating evidence suggests that,

sialylation of cell surface receptors is inversely correlated

with cell adhesion to extracellular matrix [57], although this

has not been universally observed. Praetorius and Spring

[57] have shown by confocal epifluorescence microscopy

that fibronectin did not bind to sialylated h1 integrin in

MDCK cells. It was hypothesized that the negatively

charged sialic acid could play an important role in masking

h1 integrin, and that fibronectin binds only to desialylated

h1 integrin on MDCK cells. Sialic acids are closely associ-

ated with cell recognition phenomena [58,59]. They could

potentially inhibit intermolecular and intercellular interac-

tions by their negative charge. However, they could also act

as critical components of ligands recognised by a variety of

E. Pochec et al. / Biochimica et Biophysica Acta 1643 (2003) 113–123122

proteins [58]. Nadanaka et al. [60] have found that in in-

tegrin a5h1, a major fibronectin receptor occurs a2–8 linked

oligosialic acid. After the loss of oligosialic acid, integrin

a5h1 failed to bind to fibronectin-conjugated Sepharose,

indicating that oligosialic acid on the a5 subunit plays

an important role in cell adhesion to fibronectin probably

by bringing the conformation into the high affinity form to

the ligand. In contrast Morgenthaler et al. [61] could not

observe altered adhesion of human colon carcinoma cells to

LN and FN after treatment with neuraminidase. Treatment

of breast carcinoma cells with neuraminidase reduces their

adhesion to collagen IV. Thus, sialic acid in particular a2–6

linked to components of the cell membrane promote cell

adhesion to Col IV but not to LN and FN [56]. In our study,

we have demonstrated that following neuraminidase treat-

ment, integrin a3h1 binding to FN, LN and Col significantly

increased. In agreement with the results presented here are

results of Pretzlaff et al. [62] and Semel et al. [63]. In both

cases, enzymatic removal of sialic acid residues from

purified integrin a5h1 [63] or h1 integrin on HL60 cells

[62] increases binding to FN.

Our current results may suggest the involvement of a3h1

integrin in the acquisition of invasive capacity during mel-

anoma cell progression. However, a number of questions

concerning the modification of cancer cell behaviour by

glycosylation remains to be answered.

Acknowledgements

This work was supported by a grant from the State

Committee for Scientific Research (KBN): 0287/P04/2001/

21.

References

[1] G. Li, K. Satyamoorthy, M. Herlyn, Dynamics of cell interaction and

communications during melanoma development, Crit. Rev. Oral Biol.

Med. 13 (2002) 62–70.

[2] A. Howe, A.E. Aplin, S.K. Alahari, R.L. Juliano, Integrin signalling

and cell growth control, Curr. Opin. Cell Biol. 10 (1998) 220–231.

[3] M.G. Coppolino, S. Dedhar, Ligand-specific, transient interaction be-

tween integrins and calreticulin during cell adhesion to extracellular

matrix proteins is dependent upon phosphorylation/dephosphorylation

events, Int. J. Biochem. Cell Biol. 32 (2000) 171–188.

[4] A. Melchiori, R. Mortarini, S. Carlone, P.C. Marchisio, A. Anichini,

D.M. Noonan, A. Albini, The a3h1 integrin is involved in melanoma

cell migration and invasion, Exp. Cell Res. 219 (1995) 233–242.

[5] G.E. Plopper, S.Z. Domanico, V. Cirulli, W.B. Kiosses, V. Quaranta,

Migration of breast epithelial cells on laminin-5: differential role of

integrins in normal and transformed cell types, Breast Cancer Res.

Treat. 51 (1998) 57–69.

[6] K.R. Gehlsen, P. Sriramarao, L.T. Furtcht, A.P.N. Skubitz, A synthetic

peptide derived from the carboxy terminus of the laminin a chain

represents a binding site for the a3h1 integrin, J. Cell Biol. 117

(1992) 449–459.

[7] A.M. Belkin, M.A. Stepp, Integrins are receptors for laminins, Mi-

crosc. Res. Tech. 51 (2000) 280–301.

[8] N.J. Tawil, V. Gowri, M. Djoneidi, J. Nip, S. Carbonetto, P. Brodt,

Integrin a3h1 can promote adhesion and spreading of metastatic

breast carcinoma cells on the lymph node stroma, Int. J. Cancer 66

(1996) 703–710.

[9] M.J. Elices, L.A. Urry, M.E. Hemler, Receptor functions for the in-

tegrin VLA-3: fibronectin, collagen, and laminin binding are differ-

entially influenced by ARG-GLY-ASP peptide and by divalent

cations, J. Cell Biol. 112 (1991) 169–181.

[10] J.A. Kreidberg, Functions of a3h1 integrin, Curr. Opin. Cell Biol. 12

(2000) 548–553.

[11] W.G. Carter, M.C. Ryan, Epiligrin, a new cell adhesion ligand for

integrin a3h1 in epithelial basement membrane, P.J. Gahr, Cell 65

(1991) 599–610.

[12] T. Tsuji, Y. Kawada, M. Kai-Murozono, S. Komatsu, S.A. Han, K.

Mizushima, H. Mizushima, K. Miyazaki, T. Irimura, Regulation of

melanoma cell migration and invasion by laminin-5 and a3h1 integrin

(VLA-3), Clin. Exp. Metastasis 19 (2002) 127–134.

[13] C.M. Di Persio, K.M. Hodivala-Dilke, R. Jaenich, J.A. Kreidberg,

R.O. Hynes, a3h1 integrin is required for normal development of the

epidermal basement membrane, J. Cell Biol. 137 (1997) 729–742.

[14] X.-P. Zhang, W. Puzon-McLaughin, A. Irie, N. Kovach, N.L. Proko-

pishyn, S. Laferte, K. Takeuchi, T. Tsuji, Y. Takada, a3h1 adhesion to

laminin-5 and invasion: critical and differential role of integrin resi-

dues clustered at the boundary between a3 N-terminal repeats 2 and 3,

Biochemistry 38 (1999) 14424–14431.

[15] R.L. Yauch, F. Berditchevski, M.B. Harler, J. Reichner, M.E. Hemler,

Highly stoichiometric, stable, and specific association of integrin a3h1

with CD151 provides a major link to phosphatidylinositol 4-kinase,

and may regulate cell migration, Mol. Biol. Cell 9 (1998) 2751–2765.

[16] T. Suigura, F. Berditchevsky, Function of a3h1-tetraspanin protein

complexes in tumor cell invasion. Evidence for the role of the com-

plexes in tumor cell invasion. Evidence for the role of the complexes

in production of matrix metalloproteinase, J. Cell Biol. 146 (1999)

1375–1389.

[17] G. Giannelli, S. Astigiano, S. Antonaci, M. Morini, O. Barbieri, D.M.

Noonan, A. Albini, Role of the a3h1 and a6h4 integrins in tumor

invasion, Clin. Exp. Metastasis 19 (2002) 217–223.

[18] K.M. Hodivala-Dilke, M. DiPersio, J.A. Kreidberg, R.O. Hynes, Nov-

el roles for a3h1 integrin as a regulator of cytoskeletal assembly and

as a trans-dominant inhibitor of integrin receptor in mouse keratino-

cytes, J. Cell Biol. 142 (1998) 1357–1369.

[19] E. Laplantine, L. Vallar, K. Mann, N. Kieffer, M. Aumailley, Inter-

action between the cytodomains of the a3 and h1 integrin subunits

regulates remodelling of adhesion complexes on laminin, J. Cell Sci.

113 (2000) 1167–1176.

[20] R.B. Lichtner, A.R. Howlett, M. Lerch, J.-A. Xuan, J. Brink, B.

Langton-Webster, M.R. Schneider, Negative cooperativity between

a3h1 and a2h1 integrins in human mammary carcinoma MDA MB

231 cells, Exp. Cell Res. 240 (1998) 368–376.

[21] S. Ringeard, J. Harb, F. Gautier, J. Menanteau, K. Meflah, Altered

glycosylation of a(s)h1 integrins from rat colon carcinoma cells de-

creased their interaction with fibronectin, J. Cell. Biochem. 62 (1996)

40–49.

[22] O.K. Oz, A. Campbell, T. Tao, Reduced cell adhesion to fibronectin

and laminin is associated with altered glycosylation of h1 integrins in

a weakly metastatic glycosylation mutant, Int. J. Cancer 44 (1989)

343–347.

[23] T. Kawano, S. Takasaki, T.-W. Tao, A. Kobata, Altered glycosylation

of h1 integrins associated with reduced adhesiveness to fibronectin

and laminin, Int. J. Cancer 53 (1993) 91–96.

[24] S.S. Veiga, R. Chammas, N. Cella, R.R. Brentani, Glycosylation of h1

integrins in B16–F10 mouse melanoma cells as determinant of differ-

ential binding and acquisition of biological activity, Int. J. Cancer 61

(1995) 420–424.

[25] H.-B. Guo, I. Lee, M. Kamar, S.K. Akiyama, M. Pierce, Aberrant N-

glycosylation of h1 integrin causes reduced a5h1 integrin clustering

and stimulates cell migration, Cancer Res. 62 (2002) 6837–6845.

[26] S.K. Akiyama, S.S. Yamada, K.M. Yamada, Analysis of the role

E. Pochec et al. / Biochimica et Biophysica Acta 1643 (2003) 113–123 123

glycosylation of the human fibronectin receptor, J. Biol. Chem. 264

(1989) 18011–18018.

[27] R. Chammas, S.S. Veiga, P. Potocnjak, R.R. Brentani, Asn-linked

oligosaccharide-dependent interaction between laminin and gp 120/

140, J. Biol. Chem. 266 (1991) 3349–3355.

[28] M. Zheng, H. Fang, S. Hakomori, Functional role of N-glycosylation

in a5h1 integrin receptor, J. Biol. Chem. 269 (1994) 12325–12331.

[29] M. Zheng, S. Hakomori, Soluble fibronectin interaction with cell sur-

face and extracellular matrix is mediated by carbohydrate-to-carbohy-

drate interaction, Arch. Biochem. Biophys. 373 (2000) 93–99.

[30] N.L. Prokopishyn, W. Puzon-McLaughlin, Y. Takada, S. Laferte, In-

tegrin a3h1 expressed by human colon cancer cells is a major carrier

of oncodevelopmental carbohydrate epitopes, J. Cell. Biochem. 72

(1999) 189–209.

[31] A. Litynska, E. Pochec, D. Hoja-æukowicz, E. Kremser, P. Laidler, A.

Amoresano, Ch Monti, The structure of the oligosaccharides of a3h1

integrin from human ureter epithelium (HCV29) cell line, Acta Bio-

chim. Pol. 49 (2002) 491–500.

[32] U.K. Laemmli, M. Favre, Maturation of the head of bacteriophage T4,

J. Mol. Biol. 80 (1970) 575–599.

[33] H. Towbin, T. Staechelin, J. Gordon, Electroforetic transfer of proteins

from polyacrylamide gels to nitrocellulose sheets; procedure and some

applications, Proc. Natl. Acad. Sci. U. S. A. 76 (1979) 4350–4355.

[34] B. Kun ster, S.F. Wheeler, A.P. Hunter, R.A. Dwek, D.J. Harvey, Se-

quencing of N-linked oligosaccharides directly from protein gels: in-

gel deglycosylation followed by matrix-assisted laser desorption/ion-

ization mass spectrometry and normal-phase high-performance liquid

chromatography, Anal. Biochem. 250 (1997) 82–101.

[35] D. Hoja-Lukowicz, D. Ciolczyk, J. Bergquist, A. Litynska, P.

Laidler, High-mannose-type oligosaccharides from human pla-

cental arylosulfatase A are core fucosylated as confirmed by

MALDI MS, Glycobiology 10 (2000) 551–557.

[36] A. Haselbeck, E. Schickanader, H. Eitz, W. Hosel, Structural charac-

terization of glycoprotein carbohydrate chains by using digoxigenin-

labeled lectins on blot, Anal. Biochem. 191 (1990) 25–30.

[37] A. Litynska, M. Przybylo, Does glycosylation of lysosomal pro-

teins show age-related changes in rat liver? Mech. Ageing Dev.

102 (1998) 33–43.

[38] A.L. Tarentino, C.M. Gomez, T.H. Plummer, Deglycosylation of as-

paragines-linked glycans by peptide: N-glycosidase F, Biochemistry

24 (1985) 4665–4671.

[39] M.M. Bradford, A rapid and sensitive method for the quantisation of

microgram quantities of protein utilizing the principle of protein–dye

binding, Anal. Biochem. 72 (1976) 248–254.

[40] A. Litynska, M. Przybyo, E. Pochec, D. Hoja-æukowicz, D. Cioczyk,P. Laidler, D. Gil, Comparison of the lectin-binding pattern in different

human melanoma cell lines, Melanoma Res. 11 (2001) 1–8.

[41] M.G. Jasiulionis, R. Chammas, A.M. Ventura, L.R. Travassos, R.R.

Brentani, a6h1-integrin, a major cell surface carrier of h1–6-bra-nched oligosaccharides, mediates migration of EJ-ras-transformed

fibroblasts on laminin-1 independently of its glycosylation state, Can-

cer Res. 56 (1996) 1682–1689.

[42] J.W. Dennis, M. Granovsky, Ch.E. Warren, Protein glycosylation in

development and disease, BioEssays 21 (1999) 412–421.

[43] D. Bironaite, J.M. Nesland, H. Dalen, B. Risberg, M. Bryne, Protein

glycosylation in development and disease, Tumor Biol. 21 (2000)

165–175.

[44] M. Hummel, H.C. Hedrich, A. Hasilic, Elongation of N-acetyllactos-

amine repeats in diantennary oligosaccharides, Eur. J. Biochem. 245

(1997) 428–433.

[45] M. Demetriou, I.R. Nabi, M. Coppolino, S. Dedhar, J.W. Dennis,

Reduced contact-inhibition and substratum adhesion in epithelial cells

expression GlcNAc-transferase V, J. Cell Biol. 130 (1995) 383–392.

[46] H.-B. Guo, Y. Zhang, H.-L. Chen, Relationship between metastasis-

associated phenotypes and N-glycan structure of surface glycopro-

teins in human hepatocarcinoma cells, J. Cancer Res. Clin. Oncol.

127 (2001) 231–236.

[47] R.D. Cummings, S. Kornfeld, Fractionation of asparagine-linked oli-

gosaccharides by serial lectin-affinity chromatography, J. Biol. Chem.

257 (1982) 11230–11234.

[48] S. Laferte, L.C. Loh, Characterization of a family of structurally

related glycoproteins expressing h1–6-branched asparagine-linked

oligosaccharides in human colon carcinoma cells, Biochem. J. 283

(1992) 193–201.

[49] E.V. Chandrasekaran, M. Davila, D.W. Nixon, M. Goldfarb, J. Men-

dicino, Isolation and structures of the oligosaccharide units of carci-

noembryonic antigen, J. Biol. Chem. 258 (1983) 7213–7222.

[50] D.C. Wojciechowicz, R.G. Mallon, A. Picon, P.B. Paty, Character-

ization of lectin resistant cell populations derived from human colon

carcinoma: correlation of K-Ras with h1–6 branching of N-linked

carbohydrate and CEA production, Biochem. Biophys. Res. Com-

mun. 259 (1999) 588–593.

[51] D.C. Wojciechowicz, P.Y. Park, R.V. Datta, P.B. Paty, CEA is the

major PHA-L-reactive glycoprotein in colon carcinoma lines and tu-

mors: relationship between K-ras activation and h1–6 branching of

N-linked carbohydrate on CEA, Biochem. Biophys. Res. Commun.

273 (2000) 147–153.

[52] A.K. Chakraborty, J. Pawelek, Y. Ikeda, E. Miyoshi, N. Kolesnikova,

Y. Funasaka, M. Ichihashi, N. Taniguchi, Fusion hybrids with macro-

phage and melanoma cells up-regulate N-acetylglucosaminyltransfer-

ase V, h1–6 branching, and metastasis, Cell Growth Differ. 12 (2001)

623–630.

[53] J.A. Eble, K.W. Wucherpfennig, L. Gauthier, P. Dersch, E. Krukonis,

R.R. Isberg, M.E. Hemler, Recombinant soluble human a3h1 integrin:

purification, processing, regulation, and specific binding to laminin-5

and invasion in a mutually exclusive manner, Biochemistry 37 (1998)

10945–10955.

[54] J.B. Weitzman, R. Pasqualini, Y. Takada, M. Hemler, The function

and distinctive regulation of the integrin VLA-3 in cell adhesion,

spreading, and homotypic cell aggregation, J. Biol. Chem. 268

(1993) 8651–8657.

[55] A.J. Miles, J.K. Knutson, A.P.N. Skubitz, L.T. Furcht, J.B. McCarthy,

G.B. Fields, A peptide model of basement membrane collagen

a1(IV)531–543 binds the a3h1 integrin, J. Biol. Chem. 270 (1995)

29047–29050.

[56] S. Lin, W. Kemmner, S. Grigull, P.M. Schlag, Cell surface a2,6-

sialylation affects adhesion of breast carcinoma cells, Exp. Cell

Res. 276 (2002) 101–110.

[57] J. Praetorius, K.R. Spring, Specific lectins map the distribution of

fibronectin and h1-integrin on living MDCK cells, Exp. Cell Res.

276 (2002) 52–62.

[58] A. Varki, Sialic acids as ligands in recognition phenomena, FASEB J.

11 (1997) 248–255.

[59] F. Dall’Olio, The biology and the biochemistry of the sialyl-a2,6-

lactosaminyl-linkage, Curr. Top. Biochem. Res. 2 (2000) 63–75.

[60] S. Nadanaka, Ch. Sato, K. Kitajima, K. Katagiri, S. Irie, T. Yamagata,

Occurrence of oligosialic acids on integrin a5 subunit and their in-

volvement in cell adhesion to fibronectin, J. Biol. Chem. 276 (2001)

33657–33664.

[61] J. Morgenthaler, W. Kemmer, R. Brossmer, Sialic acid dependent cell

adhesion to collagen IV correlates with in vivo tumorigenicity of the

human colon carcinoma sublines HCT116, HCT116a and HCT116b,

Biochem. Biophys. Res. Commun. 171 (1990) 860–866.

[62] R.K. Pretzlaff, V.W. Xue, M.E. Rowin, Sialidase treatment exposes

the h1-integrin active ligand binding site on HL60 cells and in-

creases binding to fibronectin, Cell Adhes. Commun. 7 (2000)

491–500.

[63] A.C. Semel, E.C. Seales, A. Singhal, E.A. Eklund, K.J. Colley, S.L.

Bellis, Hyposialylation of integrins stimulates the activity of myeloid

fibronectin receptors, J. Biol. Chem. 277 (2002) 32830–32836.