Development 113, 1069-1084 (1991)Printed in Great Britain © The Company of Biologists Limited 1991

1069

Avian neural crest cell attachment to laminin: involvement of divalent

cation dependent and independent integrins

THOMAS LALLIER and MARIANNE BRONNER-FRASER*

Developmental Biology Center, University of California, Irvine, CA 92717 USA

* Author for correspondence

Summary

The mechanisms of neural crest cell interaction withlaminin were explored using a quantitative cell attach-ment assay. With increasing substratum concentrations,an increasing percentage of neural crest cells adhere toInminin. Cell adhesion at all substratum concentrationswas inhibited by the CSAT antibody, which recognizesthe chick /J, subunit of integrin, suggesting that f}rintegrins mediate neural crest cell interactions withlaminin. The HNK-1 antibody, which recognizes acarbohydrate epitope, inhibited neural crest cell attach-ment to laminin at low coating concentrations (>1/jgmP1; Low-LM), but not at high coating concen-tration of laminin (10 /igml"1; High-LM). Attachmentto Low-LM occurred in the absence of divalent cations,whereas attachment to High-LM required >0.1mMCa2+ or Mn2+. Neural crest cell adherence to the E8fragment of laminin, derived from its long arm, wassimilar to that on intact laminin at high and low coating

concentrations, suggesting that this fragment containsthe neural crest cell binding site(s). The HNK-1 antibodyrecognizes a protein of 165 000 Mr which is also found inimmunoprecipitates using antibodies against the ftsubunit of integrin and is likely to be an integrin asubunit or an integrin-associated protein. Our resultssuggest that the HNK-1 epitope on neural crest cells ispresent on or associated with a novel or differentiallyglycosylated form of /?i-integrin, which recognizesInminin in the apparent absence of divalent cations. Weconclude that neural crest cells have at least twofunctionally independent means of attachment to lam-inin which are revealed at different substratum concen-trations and/or conformations of laminin.

Key words: extracellular matrix receptors, cell adhesion,laminin, integrin, neural crest, quail.

Introduction

Interactions between the extracellular matrix (ECM)and cell surface receptors are thought to play pivotalroles in cell movement and axon guidance. Antibodiesthat disrupt adhesion of cells to the extracellular matrixhave been used to define a class of cell surface receptorsin the integrin family. As a class, integrin receptorsmediate adhesion of cells to a variety of extracellularmatrix molecules including fibronectin, laminin, vitro-nectin, various collagens and tenascin (Horwitz et al.1985; Buck et al. 1986; Hynes, 1987; Tomaselli et al.1988; Bourdon and Ruoslahti, 1989). There is evidencethat these transmembrane glycoproteins also interactwith the cytoskeleton, thereby coupling the extracellu-lar matrix to the cell's intracellular machinery (Buckand Horwitz, 1987; Burn et al. 1988). While someintegrin receptors have specificity for a single molecule,several have been shown to bind multiple extracellularligands (Buck and Horwitz, 1987).

An integrin receptor is a heterodimer composed of

two non-covalently linked transmembrane subunits, aand /S. There are at least twelve a and five /? subunitsthat have been wholly or partially characterized.Different a subunits share a significant amount ofsequence identity to each other, as do different /3subunits, suggesting that these are families of relatedmolecules. The a and fi subunits have distinct andunrelated amino acid sequences. Three major classes ofintegrin receptors have been defined by their use of acommon /S subunit. The matching of one a subunitwith one /? subunit is thought to determine thespecificity of the receptor complex, or subunits possessbinding sites for divalent cations that are thought to berequired for function (Cheresh et al. 1987). All a-fiintegrin heterodimers described to date have beenshown to require the divalent cations Ca2+,Mg2"1" and/or Mn2+ for ligand binding (Edwardset al. 1988; Ignatius and Reichardt, 1988; Smith andCheresh, 1988). The fi subunits contain regions ofabundant intrachain disulfide linkages, resulting in anapparent increase in relative molecular mass (Afr) by

1070 T. Lallier and M. Bronner-Fraser

SDS-PAGE under reducing conditions when com-pared with non-reducing conditions (Horwitz et al.1985; Hynes, 1987).

Antibodies that perturb cell-substratum adhesionhave been useful for defining the interactions importantfor cell movement and neurite outgrowth. The CSATmonoclonal antibody recognizes a ft integrin subuniton avian cells (Horwitz et al. 1985), and inhibits chickneurite outgrowth on laminin, fibronectin and type IVcollagen in vitro (Bozyczko and Horwitz, 1986; Hall etal. 1987; Letourneau et al. 1988; Neugebauer et al.1988). In addition, the CSAT antibody disrupts theattachment of neural crest cells to laminin andfibronectin in vitro (Bronner-Fraser, 1985) and themigration of cranial neural crest cells in vivo (Bronner-Fraser, 1986). The HNK-1 antibody can also perturbcell adhesion. The HNK-1 antibody recognizes acarbohydrate epitope present on leukocytes, neuralcrest cells, various neurons and neuron-supporting cells(Abo and Balch, 1981; Tucker et al. 1984). This epitopeis found on several adhesion-related molecules, includ-ing N-CAM, Ng-CAM, myelin-associated glycoprotein,and tenascin/cytotactin (Kruse et al. 1984, 1985;Grumet etal. 1985; Hoffman and Edelman, 1987; Tan etal. 1987). The HNK-1 epitope also has been reported tobe present on the avian ft subunit of integrin (Peshevaet al. 1987). The HNK-1 antibody perturbs cranialneural crest cell migration in vivo (Bronner-Fraser,1987) and inhibits neural crest cell attachment tolaminin in vitro but, unlike CSAT antibodies, it doesnot disrupt attachment to fibronectin (Bronner-Fraser,1987). The HNK-1 epitope may be involved in heparin-dependent neurite outgrowth from dorsal root ganglioncells (Dow et al. 1988) and in cell adhesion and neuriteoutgrowth in the mouse cerebellum (Kunemund et al.1988).

Neural crest cells are a convenient experimentalsystem for studying the cell interactions involved in cellmigration and differentiation because they migratewidely in embryos along pathways containing numerousECM molecules and give rise to diverse cells types(LeDouarin, 1982). Interactions between neural crestcells and their surrounding extracellular matrix havebeen proposed to influence both their movement anddifferentiation (Newgreen and Thiery, 1980; Bronner-Fraser, 1985, 1986; Rogers et al. 1986; Runyan et al.1986; Tan et al. 1987; Perris et al. 1989). One approachfor characterizing cell-matrix interactions that may beimportant in neural crest development is to determinewhich receptors on the neural crest cell surface mediateinteractions with the extracellular matrix. In the presentstudy, we utilize a centrifugation assay (McClay et al.1981) to analyze the adhesive interactions of neuralcrest cells with laminin and fibronectin. In contrast totypical qualitative migration assays, this approachprovides a reproducible and quantitative measure ofcell attachment for small cell numbers to definedsubstratum molecules at varied surface concentrations.The CSAT and HNK-1 antibodies were used to inhibitneural crest cell-matrix attachments at a variety ofsubstratum and divalent cation concentrations. The

results suggest that neural crest cells attach to lamininby at least two different mechanisms: one mediated by adivalent cation-independent integrin(s), which isblocked by the HNK-1 antibody, and the other by adivalent cation-dependent integrin(s), which is notblocked by the HNK-1 antibody.

Materials and methods

MaterialsExtracellular matrix glycoproteins (mouse laminin and humanfibronectin) were purchased from Collaborative ResearchInc. (Bedford, MA). Rat laminin was purchased from Telios(La Jolla, CA). Mouse laminin-nidogen complex and purifiedlaminin fragments (E8, El ' and PI) were the generous gift ofDr Matts Paulsson; these molecules were tested for purity byHPLC (Paulsson et al. 1987). Heparin, dextran sulfate (Mr500000), poly-D-lysine and ovalbumin were purchased fromSigma (St Louis). CSAT antibodies were the generous gift ofDr A. F. Horwitz. The HNK-1 antibody producing hybrid-oma cell line was purchased from ATCC (Rockville, MD).The 20B4 antibody producing hybridoma cell line wasobtained through Developmental Studies Hybridoma Bank(Iowa City, Iowa).

Purification of IgMs using protamine Sepharose beadsPreswollen Sepharose 4B (Pharmacia) was activated usinglOmgmT1 cyanogen bromide at pHll.O for 5min, andthoroughly washed using distilled water, followed by 100 mlborate-buffered saline. Protamine sulfate (Smgml"1) in100 mM NaHCO3, pH8-9, was added to these beads andincubated with gentle stirring for 12 h at 4°C. The beads thenwere washed thoroughly with 100mM PBS, pH7.4 (Hudsonand Hay, 1980).

Before use, these beads were washed with 80 mM PBS,pH7.4, containing 77 mM NaCl. Solution containing IgMs wasadded to the protamine-Sepharose beads and incubated withstirring for 3h at 4°C. The beads then were packed into acolumn, washed thoroughly with 80 mM PBS, pH7.4 contain-ing 77 mM NaCl, and the IgMs were eluted using 80mM PBSpH7.4 containing 1.1M NaCl. IgMs in solution then wereprecipitated by mixing equal volumes of the IgM containingsolution and a 55 % solution of (NH^SC^. The precipitatewas centrifuged at 1500g and the supernatant discarded. Thepellet was resuspended in distilled water of a volume equal tothat of the original sample, and the precipitation wasrepeated. The final pellet was resuspended in a minimalvolume of distilled water and dialyzed against 100 mM PBS,pH7.4, at 4°C for 48 h. The purity of the resulting IgMsolution was ascertained by reducing and non-reducingSDS-PAGE. A single major band was seen with a molecularmass greater than 500000Afr under non-reducing conditions,and three bands were seen of 70000, 25000, and 20000Mrunder reducing conditions by silver stain and on immunoblotsusing an anti-IgM antibody conjugated to horseradishperoxidase. These bands are characteristic of IgMs.

Small fragments of IgMsAn IgM solution (lmgml"1 in PBS pH7.4) was digested with0.01 mgml"1 TPCK trypsin (Gibco) in 50mM Tris pH8.0containing 150 mM NaCl and 20 mM CaCl2 for 5h at 37 °C(Matthew and Reichardt, 1982; Bidlack and Mabie, 1986).2-Mercaptoethanol was added to the solution to a finalconcentration of 10 mM and incubated for 5min at 37 °C.Soybean trypsin inhibitor was added to a final concentration

Cation independent integrin on neural crest cells 1071

of O.lmgml l and incubated for 5min at 37°C to stop theenzymatic cleavage. Iodacetamide was added to a finalconcentration of 60 mM and incubated for 10 min at 25°C. IgMfragments then were dialyzed to 100 mM PBS at 4°C for 48 h.Fragments were separated by size using a Biogel P-200column calibrated with catalase, IgG, BSA and ovalbumin(232000, 150000, 68000 and 44000Mr respectively). Onlyfragments of >70000Mr were used in the present study.Fragments were concentrated using Amicon centricon tubesto a concentration of lmgrnl"1.

Neural crest cell primary culturesPrimary neural crest cultures were prepared from the neuraltubes of Japanese quail embryos (Coturnix cotumix japonica).Embryos were incubated for 48 h, at which time theirdevelopmental age was comparable to that of chick stages13-15. The region of the trunk consisting of the six to ninemost posterior somites as well as the unsegmented mesen-chyme was dissected away from the embryo. The neural tubeswere isolated by trituration in 160 units ml"1 of collagenase(Worthington Biochemical, Freehold, NJ). After stopping theenzymatic reaction with MEM containing 15 % horse serumand 10 % chick embryo extract, neural tubes were plated ontofibronectin-coated culture dishes. After several hours, theneural crest cells migrated away from the explant.

Neural crest cultures for use in quantitative attachmentassays were labelled by the addition of [3H]leucine (10-50jiCiml"1) to their culture media 4h after explantation. Thecells were allowed to incorporate labelled leucine for 16 hbefore use in the assay. Cultures were scraped to remove theneural tube, notochord and occasional somite cells. Theremaining neural crest cells were rinsed five times withblocking MEM [bMEM; minimal essential media (Gibco)containing O.SmgmF1 ovalbumin (Sigma)]. Cells wereremoved by incubation in 5 mM EDTA in bMEM for 10 min at37 °C.

Cell attachment assayCell-substratum adhesion was measured as described byMcClay et al. (1981). Briefly, fibronectin, laminin orovalbumin were coated at the indicated concentrations incarbonate buffer (100 mM, pH8.0; containing lmM CaCl2 and500 /IM MgCl2) onto polyvinyl chloride (PVC) microtitreplates by adsorption for 12 h at 25 °C. Wells were washedextensively with PBS (5 times) and incubated with bMEM for2h at 25 °C. Ovalbumin was routinely added to block non-specific binding. Wells were rinsed and filled with eitherbMEM containing the antibody to be tested, or blockingphosphate buffer (bPBS; 10 mM phosphate buffer, pH7.4containing 150 mM NaCl and 0.5 mg ml" ovalbumin) contain-ing the divalent cation concentration to be tested. Aliquots of[3H]leucine-labelled neural crest cells, rinsed to removeexcess EDTA, were added to each test well. In those caseswhere cell adhesion was measured in the absense of divalentcations, cells were resuspended in Ca2+/Mg2+-free PBScontaining 0.5 mg ml"1 ovalbumin and 1.5 mM EDTA. Neuralcrest cells were brought into contact with the substratummolecules by a centifugal force of 150 g for 5 min. Chamberswere sealed, and incubated for 15 minutes at 37°C. NeuralCTest cells that failed to adhere strongly to the substratummolecules were removed by a centrifugal force of 50 g for5 min. The chambers were quickly frozen in a methanol/dryice bath. The tops (unbound cells) and bottoms (bound cells)of each well were removed and placed separately intoscintillation vials. Samples were analyzed for 3H cts min"1

using a scintillation counter (Beckman LS 5801). Individualwells contained between 5000-lOOOOctsmin"1 with back-

ground counts on the order of 20-50 cts min '. The percent-age of bound neural crest cells per well was determined asfollows:

% cells bound=% cts min"1 bound=cts min"1 bound to substratum (bottom)* 100%

Total cts min"1 (bottom-Hop)Within these experiments, non-specific binding to ovalbuminranged from 5-15 % of the counts present within a given well.Variability between experiments using identical conditionswas —10%. This variability is likely to be due to slightdifferences within the population of neural crest cells.

Comparisons between two values for percentage of neuralcrest cell attachment were deemed significantly differentwhen the P value, determined by Student's one-sided f-test,was <0.05.

Measurement of substratum attachment to PVC platesRadiolabelling of proteins (fibronectin, laminin and ovalbu-min) was performed by reductive methylation as previouslydescribed by Tack et al. (1980). Briefly, proteins at finalconcentrations between 1 and lOmgml were dialyzedagainst 200mM borate buffer, pH8.9, and 0.5 ml were placedin glass vials on ice for lh. 0.05 ml of 0 .18 M formaldehyde wasadded to the sample, followed by 0.1 ml [3H]NaBH4 (2 mCi in0.1 N NaOH) and allowed to react for 1 h on ice. The unbound[3H]NaBH4 was removed by running the sample through adisposable Sephadex G-25M column (Pharmacia). Proteinconcentrations were assayed using the BioRad protein assayfollowing the manufacturer's instructions. Labelled proteinswere serially diluted in carbonate buffer, pH8.0, applied towells in PVC microassay plates, and allowed to bind for 12 h at25 °C. Wells were washed five times with carbonate buffer, cutinto individual wells and placed into scintillation vials foranalysis.

Phalloidin staining of neural crest cellsNeural crest cells were isolated as described for cellattachment assays. Cells were brought into contact withfibronectin and laminin substrata in 35 mm polystyrene dishes(Falcon) under conditions similar to those employed in thecell attachment assay. Fifteen minutes after cells wereintroduced to the substrata, non-adherent cells were rinsedfrom the substrata with three washes of 100 mM PBS, while theadherent cells were fixed in a solution of 4% paraformal-dehyde in 100mM PBS at 25°C for lh. Cells were labelledwith FTTC-phalloidin (Sigma) in a solution of 100 mM PBScontaining 0.1% saponin and 1% horse serum for lh . Cellsthen were rinsed with 100 mM PBS and viewed using anOlympus Vanox epifluorescence microscope.

ImmunoprecipitationsImmunoprecipitations were performed on cultures of surfacelabelled neural crest cells. Cells were isolated as describedabove (Neural Crest Cell Primary Cultures) and removedfrom their substratum by incubation in 5 mM EDTA for 5 min.Cells then were washed four times in labelling buffer (140 mMNaCl, 8mM KC1, 2mM glucose, 0.8 mM MgSO4, 1.5 mMCaCl2, 6mM NaHCO3, pH 7.4) and labelled with biotin asdescribed by von Boxberg et al. (1990). Briefly, a solution of50/igml"1 biotin-X-NHS (biotin-e-aminocaproic acid-A -hy-droxy-succinimide ester, Calbiochem) in DMSO was added toa volume 5 % of the total volume of the cell suspension andincubated for 5 min at 25 °C. Cells were washed 4 times withlabelling buffer containing 100 mM NH4CI. Proteins then wereextracted from these cells using an extraction buffer (20 mMTris pH7.2, 150mM NaCl, lmM CaCl2, 1% Triton X-1000,

1072 T. Lallier and M. Bronner-Fraser

1 nun PMSF) for 1 h at 4°C. Extracts were cleared twice byincubation with Sepharose 4B for 1 h at 4°C. Antibodies wereadded to the extracts, incubated for 12h at 4°C, followed byincubation with protein A-Sepharose for lh at 4°C.Antibody-antigen-protein A complexes were washed exten-sively with wash buffer (20 mM Tris pH72, 150 nun NaCl,lmM CaCl2, 0.1% Triton X-100, 1 mM PMSF). HNK-1immunoprecipitations were performed using purified anti-body covalently coupled to CNBr-activated Sepharose (Phar-macia) prepared using the manufacturers' instructions. Insome cases, immunoprecipitations were performed in thepresence of 5 mM EDTA and 1 % SDS in order to disruptintegrin complexes. Immunodepletions were performed byrepeating three cycles of immunoprecipitation with a givenantibody on one sample, and then immunoprecipitating with asecond antibody on that same sample. For immunoblots withthe HNK-1 antibody, immunoprecipitations were performedon samples from trunk tissue dissected from 2.5 day quailembryos.

Imrnunoprecipitates then were separated by SDS-polyacrylamide gel electrophoresis (PAGE), carried out usinga vertical slab gel system (Hoeffer; dimensions of83 mm x57mmx.75mm)witha4% stacking gel, utilizing thebuffer system described by Laemmli (1970). Eight percentacrylamide gels were chosen to achieve optimal separation ofproteins with MT in the range of 250000 to 50000. Extracts ofquail tissues were solubilized in buffer containing 4% SDS,10% glycerol, 0.04 M EDTA and 0.1% PMSF in 0.125 MTris-HCl. Some samples included 5 % /3-mercaptoethanol toreduce disulfide bonds before boiling for 5min, though non-reducing conditions were used extensively to resolve integrina subunits better. Proteins were electrophoretically trans-ferred to nitrocellulose membranes using a Tris-glycine buffersystem (Burnette, 1981). Apparent relative molecular masseswere estimated by comparison to pre-stained high rangemolecular weight standards (BRL). Membranes were blockedwith 5 % BSA prior to incubation with 125I- streptavidin(Amersham) for lh in TBST (10mM Tris pH7.2, 150mMNaCl, and 1 % Tween 20) containing 1 % BSA. Membraneswere air dried, and allowed to expose X-ray film (XAR5,Kodak) at —70°C, which was developed using D19 developer(Kodak).

ImmunoblotsImmunoprecipitations of 2.5 day trunk tissue were performedusing the CSAT antibody; proteins were separated by 10 %SDS-PAGE and transferred to nitrocellulose membranes asdescribed above. Membranes were incubated with the HNK-1antibody (l^gml"1) for l h in TBST (10mM Tris pH7.2,150nut NaCl, and 1% Tween 20) containing 1% BSA.Membranes were rinsed extensively, and incubated in 125I-anti-mouse IgM/lgG/IgA secondary antibodies (0.1/igmT1:Amersham) in TBST with 1 % BSA for 1 h. Membranes wereair-dried and allowed to expose X-ray film (XAR5, Kodak) asdescribed above.

Results

To analyze neural crest cell adhesion to various coatingconcentrations of fibronectin and laminin, we have useda quantitative centrifugation attachment assay thatallows reproducible measurement of the adhesiveproperties of small populations of cells. Roughly 105

cells per experiment were isolated from primary neuralcrest cell cultures within 24 h of explantation from the

embryo. This corresponds to a time well before theirdetectable differentiation, thus providing a populationof 'migrating' neural crest cells. The centrifugationattachment assay makes it possible to draw meaningfulconclusions from diverse assay conditions using smallnumbers of cells. Radioactively labelled neural crestcells were placed into fluid-filled chambers to whichantibodies, divalent cations, or heparin could be added.The chambers were sealed and a constant centrifugalforce was used to bring the cells into contact with thefibronectin- or laminin-coated substrata. Non-adherentcells were removed by inverting the chamber andcentrifuging at 50g. Because cells were allowed to settleonto their substratum for 15min prior to removingunbound cells, we define attachment as initial adhesionplus cell spreading.

Effects of laminin and fibronectin coatingconcentration on neural crest cell attachmentTo quantitate^the amount of substratum binding to thepolyvinyl chloride (PVC), radioactively labelled fibro-nectin, laminin or ovalbumin were added to PVC wellsat various coating concentration. The binding of allthree radiolabelled molecules appeared sigmoidal withincreasing coating concentrations, saturating between10 and 100/zgmT1 (Fig. 1).

The percentage of neural crest cells binding tofibronectin or laminin increased with increasing sub-stratum coating concentration, over the range of 0.1 to10 /igml"1 (Fig. 2A and B). The maximal percentage ofadhering cells (~85%) was observed between coatingconcentrations of 1 and 10 /zgml"1 for fibronectin and 3and lOjugml"1 for laminin.

Effects of the monoclonal antibodies CSAT, HNK-1and 20B4 on neural crest cell attachmentThe CSAT antibody, which recognizes the & subunit ofthe integrin a/ ft heterodimer, inhibited neural o-est cellattachment to fibronectin at coating concentrations

CM

U 10-.

o

i

i0.0) •

0X101 0.01 0.1 1 10

Protein coating concentration Q

100

I ml"1)

1000

Fig. 1. Measurement of protein binding to polyvinylchloride wells. Various concentrations of 3H-labelledprotein were coated to wells and their relative binding wasassayed. Points represent the mean of eight experimentsand the error bars represent the standard error of themean (S.E.M). Proteins tested were fibronectin (FN),laminin (LM) and ovalbumin (Ov).

Cation independent integrin on neural crest cells 1073100

Substratum coating concentration (jiq ml )

Fig. 2. Neural crest cell attachment to variousconcentrations of fibronectin and laminin in the presence ofthe CSAT and HNK-1 antibodies. Fibronectin (A) andlaminin (B) were assayed for their abilities to promoteneural crest cell attachment over coating concentrationsranging from 0.1 to 10 fig ml"1. CSAT and HNK-1antibodies (50/igmP1) were added to test their ability toperturb cell adhesion over a range of substratum/coatingconcentrations. Points represent the mean of at least sixexperiments and the error bars represent the S.E.M.

of 0.1 to 10/igmT1 (Fig. 2A; 3A); inhibition wasdependent on antibody concentration, with totalabolition of cell binding to fibronectin at 500/igml"1 ofantibody (Fig. 3A). Neither HNK-1 nor 20B4 anti-bodies, which recognize epitopes on the neural crestcell surface (Tucker et al. 1984; Stem et al. 1989),inhibited neural crest cell adhesion to fibronectin atantibody concentrations of up to lmgmP 1 (Fig. 3A).

Adhesion to all laminin substratum concentrationswas completely abolished by the CSAT antibody(Fig. 2B) at a concentration of 50ngmTl (Fig. 3B andC), suggesting the involvement of -integrin receptors.In contrast, neural crest cells exhibited differentialsensitivity to the HNK-1 antibody at different coatingconcentrations of laminin (Fig. 2B). At low coatingconcentrations of laminin (Low-LM; sSl/igmP1),neural crest cell adhesion was inhibited in a dose-dependent manner with maximal inhibition (70%)occurring at lmgmP 1 of HNK-1 antibody (Fig. 3C).With increasing laminin coating concentrations(>l^gml~1), neural crest cells became progressivelyless sensitive to the HNK-1 antibody and wereunaffected at high coating concentrations of laminin(High-LM; 10/igmP1) by antibody concentrations ofup to 1 mgrnl"1 (Figs 2B and 3B). Because the HNK-1

E

M

I

M

B O - •

B O - •

70-

80- -

SO- -

40-

30- -

20- -

10-

0- -

80-<

80- -

70-

80-

90-

40-

3 0 -

20-

10--

0- -

80-

8 0 -

70-

80-

50-

4O--

30-

20-

10- -

HNK-120B4-

CSAT

f-HNK-1

H-l

L-LM

20B4

040.200 10.000 100.000 1000.000

Antibody concentration (jig ml" )

Fig. 3. Neural crest cell attachment to (A) 10/igml x offibronectin, (B) 10/igmT1 of laminin, and (C) 1/igmP1 oflaminin were assayed in the presence of variedconcentrations of CSAT, HNK-1, fragments of HNK-1(>70000A/r), and 20B4 antibodies. Points represent themean of at least six experiments and the error barsrepresent the S.E.M.

antibody is an IgM, its large size could lead to non-specific steric hindrance. To circumvent this problem,small proteolytic fragments (>70000 MT) were pre-pared (Matthew and Reichardt, 1982). These mono-valent HNK-1 fragments inhibited neural crest celladhesion on Low-LM as efficiently as the intactantibody but had no effect on High-LM (Fig. 3B andC). The 20B4 antibody, which also binds to the surfaceof neural crest cells, had no effect on adhesion to highor low coating concentrations of laminin (Fig. 3B andC). The results suggest that there are at least twomechanisms by which neural crest cells attach tolaminin: one that is HNK-1 insensitive and evident at

1074 T. Lallier and M. Bronner-Fraser

Table 1. Comparison of neural crest cell attachment to laminin from different sources

[LM]/igml

10

1

[HNK-1]/igmr1

OQ

OO

O

QO

O

[Ca]mM

1*1*01

1*1*01

TL-MBF

82% ±5%83%±4%6%±1%

46% ±3%12%±3%44%±4%

Laminin

Mouse LM (EHS)

CR

73%±3%74%±7%8%±3%

69%±4%

51% ±2%12%±2%44%±7%55% ±5%

source

MP

85% ±2%82% ±2%22%±7%89% ±2%

76% ±4%36%±7%47%±7%48% ±7%

RatTelios

81%±5%82%±6%

72%±1%32% ±5%

TL-MBF=Mouse laminin isolated in our laboratory.CR=Mouse laminin purchased from Collaborative Research Inc.MP=Mouse laminin-nidogen generously donated by Mats Paulsson.Telios=Rat laminin purchased from Telios Inc.* = MEM containing lmM Ca2+ and 0.5 mM Mg2"1".

high substratum concentrations, and one that is HNK-1sensitive and active at lower substratum concentrations.

The differential binding of neural crest cells to lowversus high coating concentrations of laminin might berelated to impurities in the mouse laminin. To test thispossibility, we assayed neural crest cell adhesion tolaminin purified in different ways from differentsources. No significant differences were noted betweenmouse EHS laminin and rat laminin with respect toneural crest cell binding efficiency or the degree ofinhibition by CSAT or HNK-1 antibodies at varioussubstratum coating concentrations (Table 1). Further-more, a mouse laminin-nidogen complex tested forpurity by HPLC analysis (Paulsson et al. 1987)supported comparable attachment. Since neural crestcells do not attach to nidogen (Perris et al. 1989), thissuggests that the observed binding is to laminin. Anadditional argument that impurities are unlikely tocontribute to the binding characteristics of neural crestcells to laminin is that we observe 'typical' cell-laminininteractions at high coating concentrations of laminin,which would be expected to contain the highest amountof impurities. At lower concentrations, in which'unusual' cell-laminin interactions are observed, theconcentrations of impurities would be decreased by oneor more orders of magnitude. We also tested for thepossibility that the HNK-1 antibody was blockingadhesion of neural crest cells to laminin by directlybinding to carbohydrate chains on the laminin mol-ecule. In studies using biotinylated HNK-1 antibody,the HNK-1 antibody did not bind significantly tolaminin substrata at either the high or low coatingconcentrations (data not shown).

Divalent cations dependence of neural crest cellattachment to lamininIt has been suggested that integrin receptor hetero-dimers require divalent cations (Ca2+, Mg^+, or Mn2+)for binding to their ligands (Hynes, 1987). Weexamined the divalent cation requirements of neural

crest cell attachment to both High-LM and Low-LMcoating concentrations (10 and 1/igml"1 respectively).Attachment to High-LM required >100mM of eitherCa2+ or Mn2+, both of which were equally efficient atpromoting attachment, while concentrations of up tolmM Mg 4" did not promote adhesion (Fig. 4). In thepresence of divalent cations, neural crest cells attachedto both High-LM and Low-LM, though binding toHigh-LM was significantly greater (P<0.001) than toLow-LM (-80% versus -50% of total cells; Fig. 8Aand B). In contrast, with divalent cation concentrationsbelow 100 fJM, neural crest cells exhibited significantattachment to Low-LM (-50 %; Fig. 4 and 8B) but notto High-LM (Fig. 4 and 8A; P<0.01). Attachment ofneural crest cells to Low-LM in the absence of divalentcations was inhibited significantly by both the HNK-1and CSAT antibodies (P<0.01; Fig. 5), suggesting thatthe adhesion is mediated by ^-integrin receptors. Wedefine the absence of divalent cations as incubationconditions containing 1.5 mM EDTA, with no sup-plementation of divalent cations.

Effects of antibody and divalent cation concentrationon neural crest cell attachment to polyanionic andpolycationic substrataNeural crest cells can adhere to some substrata bymeans of charge interactions that are not mediated byspecific receptor-ligand recognition. To test the speci-ficity of the antibodies used in this study, neural crestcells were grown on four charged substratum mol-ecules: heparin, dextran sulfate, polylysine and ovalbu-min. These were chosen because they represent bothpolyanionic and polycationic substrata, as well as theblocking protein used in the present experiments.Neural crest cells were allowed to attach to thesesubstrata in the presence (Fig. 6A) and absence(Fig. 6B) of divalent cations. The HNK-1 and CSATantibodies failed to inhibit neural crest cell attachmentto any of these substrata (P>0.1). The ability of theseantibodies to block attachment to laminin but not to

Cation independent integrin on neural crest cells 1075

s

IM

100- -

90- •

80- •

70-

80- -

BO- -

40- •

30- -

20- -

10- •

0

90- •

80- -

70-

• 0 -

80-

40-

30- •

20-

10 - •

0- -

90-

80-

70-

• 0 - •

80-

40-

30-

20-

10-

0

Ca 2+

H-LM

Ug 2+

L-LM

H-LM

-5 B

Mn 2+

H-LM

10 100

Divalent cation concentration1000

Fig. 4. The dependence of neural crest cell attachment tolaminin on divalent cations. Two coating concentrations oflaminin, 10/igmF1 (H-LM) and 1/igmT1 (L-LM), weretested. Divalent cations, (A) Ca2+, (B) Mg2*, and (C)Mn2+, were tested over a range of concentrations from1 ^M to 1 mM. Points represent the mean of at least sixexperiments and the error bars represent the S.E.M.

polyionic substrata suggests that attachment to lamininis mediated by specific cell surface receptors.

Neural crest cell attachment to fragments of lamininTo determine the region of laminin to which neuralcrest cells bind, we tested proteolytic fragments of theintact laminin molecule for their ability to promoteattachment in the presence or absence of the HNK-1and CSAT antibodies (50/igmT1; Fig. 7). Threefragments were utilized: the E8 fragment, correspond-ing to the C-terminal 'long arm' of laminin including theheparin-binding domain; the El ' fragment, corre-sponding to the N-terminal three 'short arms' oflaminin; and the PI fragment, corresponding to the

boun

d

I

M

BO

BO-

70-

eo

so-

40-

30-

20-

10-

0-

. No divalent cations

JL

ff i •

L-LM

•, * .

No artfcody HNK-1 CSAT

Antibodies present

Fig. 5. Neural crest cells were assayed for their ability toattach to ljugml"1 of laminin (L-LM) in the absence of thedivalent cations. The CSAT and HNK-1 antibodies(50/igmP1) were tested for their ability to inhibit neuralcrest cell attachment to L-LM in the absence of divalentcations. Asterisks (*) indicate significant (P<0.05)differences from adhesion in the absence of divalentcations. Points represent the mean of at least sixexperiments and the error bars represent the S.E.M.

100

ao

80

70

so

90

40

X

20

10

0

90

80

70

M

90

40

30

20

10

0

A

B

JL

1

1_1

1

1

1111

m i MEM• i HNK-1

T ^ CSAT

CD PBSM HNK-1BS CSAT

HS OS PL

Substratum

Fig. 6. Neural crest cells were assayed for their ability, toattach to heparin (HS), dextran sulfate (Mr 500000) (DS),poly-D-lysine (PL), and ovalbumin (Ov) with and withoutthe CSAT or HNK-1 antibodies (SO^gmP1). Adhesion wasmeasured in the presence (A) and absence (B) of divalentcations (lmm Ca2+ and 0.5 mM Mg2"^). Asterisks (*)indicate significant (P<0.05) differences from adhesion inthe absence of antibodies. Points represent the mean of atleast six experiments and the error bars represent theS.E.M.

1076 T. Lallier and M. Bronner-Fraser

100

»o

BO

70

eo

90

40

30

20

10

0

80

BO

70

SO

50

40

X

20

10

0

10/jg mi- 1CD No AbM HNK-1ra CSAT

EB EV

1/ig ml- 1CD NoAbM HNK-1BS CSAT

*

la I jE8 EV

Substratum

Fig. 7. Proteolytic fragments of laminin were compared tointact laminin for their ability to promote neural crest celladhesion in the presence and absence of antibodies.Laminin, the E8 (constituting the long arm of laminin), theEl ' (constituting the three short arms of laminin) and thePI (constituting the cross region of laminin) fragmentswere coated onto the substratum at concentrations ofequivalent to 10 (A) and 1 fig ml"1 (B) of laminin. TheCSAT and HNK-1 antibodies (50/igml"1) were tested fortheir ability to inhibit neural crest cell adhesion to lamininand its proteolytic fragments. Asterisks (*) indicatesignificant (P<0.05) differences from controls. Pointsrepresent the mean of at least six experiments and theerror bars represent the S.E.M.

'cross' region of laminin. The purity of these fragmentswas determined using HPLC analysis (Paulsson et al.1987). Coating concentrations of laminin fragmentswere chosen at molar equivalents to those used forintact laminin; for simplicity, we refer to the concen-trations used for the intact molecule. The E8 fragmentof laminin coated at 1 or 10/xgmP1 equivalentspromoted attachment nearly as well as the intactmolecule ( -45% and - 7 0 % , respectively; P>0.1).The El ' fragment promoted attachment of —35% ofthe cells at 1 or lO/igml"1. The PI fragment promotedno significant attachment at lO/zgml"1. CSAT anti-bodies inhibited neural crest cell adhesion to the E8 andEl ' fragments by - 8 5 % and - 7 5 % (P<0.001 and<0.01), respectively, at a substratum coating concen-tration of lO^gml" . At this substratum concentration,HNK-1 antibodies caused no significant inhibition ofadhesion (P>0.1). In contrast at 1/igml"1 coatingconcentration (Fig. 7B), neural crest cell binding to theE8 fragment of laminin was inhibited by both HNK-1and CSAT antibodies (-60% and -90%, respectively;

Table 2. Summary of statistically significantdifferences of neural crest cell attachment to various

substrata and conditions

(A) Laminin and laminin fragments vs. antibodiesAntibody

JUU3U dlUllJ

one

High-LM*High-E8*High-El"

Low-LM*Low-E8*Low-El"

High-LMHigh-E8High-El'

High-LMHigh-LM

Low-LMLow-LM

two

High-LMHigh-E8High-El'

Low-LMLow-E8Low-El'

Low-LMLow-E8Low-El'

High-E8High-El'

Low-E8Low-El'

None

0

~t0

HNK-1 CSAT

0 +0 +0 +

0 +

+=Significant differences between adhesion of cell populationsby Students (-test P<0.05.

0=No significant differences.•=In the absence of antibodies.

(B) Laminin' and laminin fragments vs. divalent cationsCation

one

High-LM*High-E8*High-El"

Low-LM*Low-E8*Low-El"

High-LMHigh-E8High-El'

High-LMHigh-LM

Low-LMLow-LM

two

High-LMHigh-E8High-El'

Low-LMLow-E8Low-El'

Low-LMLow-E8Low-El'

High-E8High-El'

Low-E8Low-El'

None

0

00

0

Ca2 +

0

000

0

+0

Mg*+

00

000

0

0

0

Mn2+

000

000

+

00

+=Significant differences between adhesion of cell populationsby Students f-test P<0.05.

0=No significant differences.* = In the absence of divalent cations.

F<0.01); adhesion to El ' fragment was inhibited onlyby the CSAT antibody (P<0.01). Table 2A summarizesa statistical analysis of data in Fig. 7. In summary,molar equivalents of the E8 fragment and intact laminindemonstrated similar neural crest cell attachment andantibody inhibition. In contrast, attachment to the El 'fragment of laminin was significantly different from thatobserved for both the E8 fragment and intact lamininwith respect to attachment and antibody sensitivity.

The divalent cation dependence of neural crest cellattachment to the E8 fragment of laminin was similar tothat observed for the intact molecule (Fig. 8). At highcoating concentrations (10/igml equivalents;Fig. 8A), neural crest cells bound avidly in the presenceof lmM Ca2+ or Mn2+. This attachment was signifi-

Cation independent integrin on neural crest cells 1077

100

so

ao

70

60

so

40

30

20

10

0

so

80

70

SO

90

40

30

20

10

0

10 M9 ml- 1 • CU No Ions

Bffl Mg.E 3 Mn

2 +

2+

a &.1 JliEV

No IonsCo 2+M g 2 +

M n 2 +

Substratum

Fig. 8. Proteolytic fragments of laminin were compared tointact laminin for their ability to promote neural crest celladhesion in the presence and absence of divalent cations.Laminin, the E8 (constituting the long arm of laminin),and the El ' (constituting the three short arms of laminin)fragments were coated onto the substratum atconcentrations of 10 (A) and l^gml"1 (B). Neural crestcell attachment to the E8 and El ' fragments was alsotested for dependency on the divalent cations Ca2+, Mg2"1"and Mn2+. Asterisks (*) indicate significant (P<0.05)differences from controls. Points represent the mean of atleast six experiments and the error bars represent theS.E.M.

cantly better than that observed in the absence ofdivalent cations or in the presence of lmM Mg2"1"(P<0.001). On low coating concentrations of the E8fragment of laminin (Fig. 8B), neural crest cell attach-ment was not significantly affected by the presence orabsence of divalent cations (P>0.1). In contrast, neuralcrest cell attachment to high (lO/igmF1 equivalent)concentrations of the El ' fragment required thedivalent cations Mg2"1" or Mn2+, but not Ca2+. Table 2Bsummarizes a statistical analysis of data in Fig. 8. Insummary, neural crest cell attachment to the E8fragment was comparable to that on intact laminin, butwas significantly different from that to the El 'fragment.

Effects of heparin on neural crest cell attachment tolamininThe adhesion of many cell types to laminin has beenshown to involve heparan sulfate proteoglycans (Charo-nis etal. 1988). Exogenous heparin (1-500 fi% ml"1) wasadded to the assay wells as a competitive inhibitor of

•ac

IcE

cts

80

70-

SO-

SO-

40-

30-

20-

10-

H-LM

- • L-LM

Heparin concentration (jiq ml )

Fig. 9. Addition of exogenous heparin (1 to SOO/xgml"1) toneural crest cell adhering to 10/igml"1 of laminin (H-LM)and 1/igml"1 of laminin (L-LM). Points represent themean of at least six experiments and the error barsrepresent the S.E.M.

heparan sulf ate proteoglycans. Neural crest cell bindingto High-LM and Low-LM was not significantly alteredby exogenous heparin (P>0.1; Fig. 9).

Neural crest cell morphology on fibronectin andlamininNeural crest cell morphology was investigated on threesubstrata; fibronectin, High-LM and Low-LM. Themorphology of the cells was visualized by staining withFITC-phalloidin, which binds to filamentous actin. Onfibronectin, neural crest cells possessed multiple longthin processes and appeared to be sessile (Fig. 10A).On both High-LM and Low-LM (Fig. 10B and C),neural crest cells possessed multiple broad lamellapodiacharacteristic of migratory cells, but lacked long thinprocesses. The distal portion of these lamellapodiacontained abundant microfilaments, adjacent to anactin-free zone. Neural crest cells appeared morpho-logically similar on both types of laminin but differentof fibronectin.

Identification of HNK-1- and CSAT-bearing proteinson neural crest cellsThe HNK-1, CSATand 20B4 antigens were isolated byimmunoprecipitation of extracts from surface biotinyl-ated neural crest cells. The CSAT antibody immuno-precipitated three bands major protein bands of180000, 165000 and 130 000 Mr, as well as a minor65 000 Mr protein separated by SDS-polyacrylamide gelelectrophoresis (PAGE) under non-reducing conditions(Fig. 11; Lane 2). The 180000 and 165000Mr bandsprobably represents a subunits while the 130 000 MTband probably represents the ft subunit. HNK-1immunoprecipitations revealed three broad bands withMr of 180000, 165000 and 130000Afr (Fig. 11; Lane 3);the level of the 130 000 Mr protein was reduced in thepresence of EDTA(Fig. 11; Lane 4). With the additionof 1 % SDS (which causes integrin complexes todissociate), a single band at 165000Mr was detectedwith the HNK-1 antibody (Fig. 11; Lane 1). Immuno-precipitations using the 20B4, which recognizes another

1078 T. Lallier and M. Bronner-Fraser

epitope on neural crest cells, revealed a differentpattern of bands (Fig. 11; Lane 1).

Immunodepletion experiments were performed toshow that the proteins recognized by the CSAT andHNK-1 antibodies were the same. Samples were

Fig. 10. Neural crest cell morphology was examined afteradhesion to flbronectin and laminin substrata. Neural crestcells were labelled with phalloidin to visualize filamentousactin. Neural crest cells adhering to fibronectin (A)possessed multiple focal contacts. Neural crest cellsadhering to either High-LM (B) and Low-LM (C)possessed multiple focal lamellapodia.

3 4 5HNK-1 HNK-1 HNK-1

2+

6CSAT

7HNK-1

8CSAT

Fig. 11. Neural crest cells were surface biotinylated andthe antigens for the 20B4 (Lane 1), CSAT (lane 2) andHNK-1 antibodies (Lane 3) were immunoprecipitated in abuffer containing Ca2+ and 1 % Triton X-100. The HNK-1antigen was also immunoprecipitated in the presence ofEDTA (Lane 4) and 1 % SDS in order to dissociate it fromany proteins with which it might aggregate (Lane 5). Lanes6-8 show the results of immunodepletion experiments, inwhich the HNK-1, CSAT and 20B4 antibodies were used todeplete their respective antigens prior toimmunoprecipitations with the HNK-1 or CSAT antibodies.Dashes indicate calculated molecular masses of theidentified bands, from top to bottom, of 180000, 165000,130000, 90000 and 65000Mr.

immunoprecipitated repeatedly with HNK-1, CSAT or20B4 antibodies in order to immunodeplete them of therespective antigens: (Fig. 11; Lanes 6, 7 and 8,respectively). The samples then were immunoprecipi-tated with a different antibody: the CSAT antibodyafter HNK-1 immunodepletion (Lane 6), the HNK-1antibody after CSAT immunodepletion (Lane 7) or theCSAT antibody after 20B4 immunodepletion (Lane 8).Lanes 6 and 7 show no protein bands, indicating thatthe two antibodies recognize the same proteins. Lane 8shows the 180000, 165000 and 130000Afr bands of theCSAT antigen, indicating that the 20B4 antibody doesnot recognize these integrin proteins. Comparableprotein bands were detected by immunoprecipitationwith the HNK-1 antibody, following immunodepletionby the 20B4 antibody (data not shown).

It is possible that the technique of surface labellingneural crest cells does not recognize all HNK-1 reactive

Cation independent integrin on neural crest cells 1079

Fig. 12. The CSAT antigen was immunoprecipitated fromtrunk tissue extracts of 2.5 day old quail embryos,biotinylated, and detected following non-reducing SDS-PAGE by 125I-streptavidin (Lane 1). Three major bands of180000, 165000 and 130000Mr were detected, along withtwo minor bands of 90 000 and 65 000 Mr. Samples ofunlabelled CSAT antigen (Lane 2), purified CSAT antibody(Lane 3), and whole tissue extracts (Lane 4) wereimmunoblotted with the HNK-1 antibody and visualizedusing 125I-conjugated secondary antibodies. One majorband (probably an integrin a subunit) was observed in theCSAT immunoprecipitate.

glycoproteins. In order to identify whether the HNK-1epitope was associated with integrin receptors onunlabelled neural crest cells, we examined extracts fromthe trunk tissue derived from 2.5 day quail embryos. Inembryos at this stage, neural crest cells are activelymigrating and represent the vast majority of HNK-1immunoreactive cells. Tissue extracts were immuno-precipitated with the CSAT antibody. After separationof the CSAT immunoprecipitates by non-reducingSDS-PAGE, three major bands were detected of180000, 165000 and 130000 MT, as well as two minorbands of 90000 and 65000Mr (Fig. 12; Lane 1). Inimmunoblots of these precipitates, the HNK-1 antibodyrecognized a dominant 165000A/r glycoprotein (Lane2). This band was absent in immunoblots of the CSATantibody along (Lane 3) and greatly enriched overimmunoblots of whole tissue extracts (Lane 4). Thisconfirms that the HNK-1 antibody recognizes a

165 000 MT glycoprotein which is immunoprecipitatedby an antibody to the ft subunit of integrin.

Discussion

Our results demonstrate that neural crest cells possessat least two distinct mechanisms for attachment tolaminin, as revealed by experiments analyzing celladhesion to various laminin-coating concentrations(summarized in Table 3). Both attachment mechanismsare inhibited by the CSAT antibody, which recognizesthe avian ft subunit of integrin. Cell attachment toHigh-LM is unaffected by the HNK-1 antibody and isdivalent cation dependent. In contrast, cell attachmentto Low-LM is inhibited by the HNK-1 antibody anddoes not require divalent cations. The HNK-1 antibodyrecognizes a dominant 165 000 MT glycoprotein on thesurface of neural crest cells which appears to be on orassociated with an integrin. The 20B4 antibody, whichbinds to the surface of neural crest cells, does not blockadhesion to laminin at any coating concentration.

One possible explanation for the distinct bindingmechanisms observed at different coating concen-trations of laminin is that the conformation of lamininmay be altered as a function of coating concentration.For example, physiological concentrations of Ca2+ mayalter the configuration of laminin at high versus lowcoating concentrations. In fact, preliminary resultsindicate that the ratio of laminin to Ca in thesubstratum solution may determine the conformation ofthe molecule (unpublished observation). Given themeasured concentrations of laminin adsorbed onto thePVC plates, one can perform calculations that suggestthat Low-LM may be monolayered whereas High-LMmust be multilayered or aggregated (as depicted inFig. 13). This suggests that different receptor sitesand/or conformations may be exposed at the differentcoating concentrations.

An alternate explanation is that there may bedifferent isoforms of laminin that are preferentiallyexposed at different coating concentrations. Recently,evidence for a laminin gene family has emerged.S-laminin, an analog of the Bl chain of laminin, hasbeen identified in the synaptic cleft (Hunter et al. 1989)and merosin, an A chain variant, is located in thebasement membrane of numerous human tissues (Ehriget al. 1990). Independent of whether different coatingconcentrations result in different conformations or

TableCoatingconcentrationof laminin

3. Summary of results for neuralAntibody

HNK-1 CSAT 20B4

crest

Ca2+

cell attachmentDivalent cations

Mg*+

to laminin

Mn2+ Heparin

10/igmr1

(High-LM)1/igmr1

(Low-LM)

+

a

+

o

+ indicates a requirement for nut concentrations of the molecule.- indicates inhibition by the antibody at 50/igml"1.0 indicates no effect by the molecule over the range of concentrations tested.

1080 T. Lallier and M. Bronner-Fraser

"165 x 10

HNK-1

Low-LM High-LMFig. 13. Schematic diagram illustrating possible mechanisms of neural crest attachment to laminin mediated by /Sj-integrins.At low coating concentrations of laminin (Low-LM), the laminin is likely to be monolayered on the substratum. Neuralcrest cells adhere to Low-LM by means of an integrin which has an HNK-1 epitope either directly attached to it or on anassociated molecule which is involved in receptor binding to laminin; this integrin receptor complex functions in theabsence of divalent cations. At high coating concentrations of laminin (High-LM), the laminin is multilayered andaggregated. This may confer a different conformation to the receptor binding site or may expose a different site on the E8fragment of laminin. Neural crest cell binding at this concentration depends on a ft-integrin that requires Ca2+ or Mn2+,but not the HNK-1 epitope, for function.

forms of laminin, either of these scenarios supports theidea that neural crest cells possess two independentmechanism for attachment to laminin(s).

Laminin preparations typically contain numerouscontaminants including collagen type IV and heparansulfate proteoglycans. Thus, it is conceivable that thedifferential binding of neural crest cells to low versushigh coating concentrations of laminin might be relatedto impurities in the mouse laminin. However, we foundthat laminin derived from a variety of sources, includinga purified laminin-nidogen complex (Paulsson et al.1987) and purified E8 fragment of laminin werecomparable to mouse tumor laminin in supportingneural crest cell attachment. Because neural crest cellsdo not bind to nidogen alone (Penis et al. 1989), theseresults suggest that the observed binding characteristicsrepresent adhesion to laminin.

Our data support the idea that neural crest cellspossess at least two receptors for laminin. Thesereceptors appear to function independently, with oneactive at high coating concentrations of laminin and theother active at low coating concentrations. A simpleavidity model in which neural crest cells possess onehigh and one low affinity receptor for laminin cannotexplain these two receptor systems. According to such amodel, the only detectable binding at low coatingconcentrations of laminin would be mediated by a highaffinity receptor whereas both low and high affinity

receptors would function at higher coating concen-trations of laminin. However, our data suggest that thereceptor that binds to Low-LM is inactive at highlaminin concentrations, since neural crest cell attach-ment to High-LM is significantly less than that to Low-LM in the absence of divalent cations. It is possible thateach receptor system recognizes different confor-mations of the laminin molecule, resulting from thevarious coating concentrations used in our in vitrosystem. Regardless of whether the concentration-dependent attachment observed in culture is analogousto neural crest cell attachment to laminin within theembryo, this assay can now be used to study thesedifferent receptors on neural crest cells.

Both adhesive mechanisms to laminin appear to bemediated by integrin receptors, one requiring divalentcations for binding and the other functioning in the'absence' of divalent cations. The involvement of /3Xcontaining integrins (/^-integrins) was demonstrated bythe finding that the CSAT antibody inhibited neuralcrest cell attachment to laminin at all substratumcoating concentrations. Even in the 'absence' of Ca2+,Mg2"1" and Mn2+, a significant percentage of neural crestcells attached to Low-LM and their adhesion wasinhibited by the CSAT antibody. These results takentogether indicate that at least one integrin can recognizespecific ligands without divalent cations. All previouslyreported integrins require physiological concentrations

Cation independent integrin on neural crest cells 1081

of divalent cations to interact with their ligands (Hynes,1987). Typically, millimolar concentrations of Ca2+

and/ or Mg2* are utilized (Cheresh et al. 1987; Edwardsetal. 1987; Smith and Cheresh, 1988), though Mn2+ canbe employed by some heterodimeric receptors(Edwards et al. 1988; Ignatius and Reichardt, 1988).Consistent with these characteristics, we observed thatneural crest cell attachment to high coating concen-trations of laminin (>10/igml~1) required Ca2+ orMn2+ at lmM concentrations. We cannot rule out thepossibility that extremely low cation concentrationsremain under our assay conditions, in the presence ofEDTA and the absence of exogenous cations. If theintegrin receptors functioning at Low-LM have veryhigh binding affinities for cations, it is conceivable thatresidual cations may fulfill their divalent cation require-ments. However, this would require the integrinreceptors to bind cations in the nanomolar concen-tration range, whereas all reported integrins bindcations in the millimolar range.

Unlike the CSAT antibody, which blocks neural crestcell adhesion to laminin at all coating concentrations,the HNK-1 antibody only inhibits neural crest celladhesion to laminin at low coating concentrations.Even in the absence of divalent cations, neural crest cellattachment to Low-LM was significantly inhibited(70%) by the HNK-1 antibody, whereas no detectableinhibition was noted on High-LM. In contrast, neitherthe CSAT nor the HNK-1 antibodies blocked neuralcrest cell attachment to charged substrata. Thesefindings, in conjunction with the divalent cationsensitivity, suggest that neural crest cells possess at leasttwo mechanisms for attachment to laminin: (i) adivalent cation-dependent integrin(s), which mediatesbinding at high coating concentrations; and (if) anHNK-1 bearing integrin(s), which mediates binding atlow coating concentrations and functions with orwithout exogenous cations (Fig. 13). The presence ofthe latter type of integrin is further supported by thefinding that HNK-1 antibody recognizes bands withcharacteristics of both integrin a- and /3 subunits inimmunoblots of the CSAT antigen. Because we cannotdetermine functionality from our biochemical analyses,we cannot determine whether the HNK-1 epitope onintegrins is functionally active. Therefore, we cannotdistinguish whether the receptor used for binding onLow-LM has an HNK-1 epitope on the /^-integrin itselfor on an associated molecule involved in attachment tolaminin (Fig. 13). Furthermore, we cannot rule out thepossibility that there are other non-integrin HNK-1immunoreactive proteins that are involved in neuralcrest cell adhesion to laminin. The HNK-1 bearing/Jj-integrin receptor system is distinct in cationdependency from that operating at High-LM and islikely to recognize a different binding site or configur-ation on laminin, that may be determined by theconcentration/state of aggregation of laminin on thesubstratum.

Several integrin receptors that bind laminin havebeen described. All reported laminin-binding integrins,however, can also bind to other ligands including

fibronectin and collagens. For example, the CSATantibody blocks chick skeletal myoblast and fibroblastadhesion to fibronectin and laminin (Horwitz et al.1985; Buck et al. 1986), as well as adhesion and neuriteoutgrowth of chick retinal cells on fibronectin, lamininand collagen type IV (Hall et al. 1987). Rat neuronalcell lines possess integrins involved in adhesion tolaminin (Ignatius and Reichardt, 1988) and collagentype IV (Tomaselli et al. 1988). Recently, a number oflaminin-binding integrin receptors have been identifiedin mammals. These integrins have been classified aso-jft (Forsberg et al. 1990; Hall et al. 1990), a^fa(Languino et al. 1989; Elices and Hemler, 1989; Carteret al. 1990; Lotz et al. 1990), a3^ (Gehlsen et al. 1989;Carter et al. 1990), a6px (Sonnenberg et al. 1988, 1990;Hall et al. 1990; Shaw et al. 1990), and a^ft (Lotz et al.1990) integrins. The a^ apparently requires Ca2+ tofunction, while oc$\ functions in the presence of Mg2"1"and Mn2+, but not in the presence of Ca2+ (Sonnenbergetal. 1988). Another laminin/collagen receptor that hasbeen reported to be a /3rintegrin is a receptor on PC12cells that requires Mg2"1" rather than Ca2+ for function(Turner et al. 1987), similar to the divalent cationdependency for human platelet ayf integrin and neuralcrest cell adhesion to the El' fragment of laminin. Therat hepatocyte a-^fii integrin has been shown to adhereto the E8, El and PI fragments of laminin (Forsberg etal. 1990; Hall et al. 1990), while the human a-3ft(Gehlsen et al. 1989) and o^A (Hall et al. 1990;Sonnenberg et al. 1990) integrins adhere to the E8fragment alone. In addition, two as yet unidentifiedmammalian /^-integrins for laminin have been ident-ified. A bovine /Jpintegrin appears to adhere to lamininin an Arg-Gly-Asp (RGD) dependent manner (Bassonet al. 1990), while a human ^-integrin adheres tolaminin in an RGD-independent manner and does notcontain the known a^-ocf, subunits (Kramer et al. 1989).From these studies, laminin-binding integrins arerelatively common and diverse in their subunit compo-sition, ligand specificity and divalent cation require-ments.

Although the integrin heterodimers present onneural crest cells have not yet been characterized, it ispossible to rule out some integrins on the basis of recentimmunohistochemical data. Using antibodies to theavian 05 and (% integrin subunits, it has been shown thatthese proteins are absent from the surface of migratingneural crest cells (Bronner-Fraser et al. 1992). Inaddition, it seems unlikely that <x^\ or a5^ is a lamininreceptor on neural crest cells, since these integrin bindsonly to fibronectin in other species. Other hetero-dimers, including <X\Pi, a^i and/or a^Pi are reasonablecandidate molecules for mediating neural crest cellattachment to laminin.

Our results and those of a previous study (Perris et al.1989) suggest that neural crest cells bind to the E8fragment of laminin, corresponding to the cell bindingsite on the long arm in the vicinity of the heparinbinding domain (Aumailley et al. 1989). In ourattachment assays, neural crest cells bind to the E8fragment in a manner similar to their attachment to

1082 T. Lallier and M. Bronner-Fraser

intact laminin; attachment is blocked by the CSATantibody at all coating concentrations, and by HNK-1antibody at low coating concentrations only. The E8fragment of laminin also mimics the intact moleculewith respect to the requirements for divalent cations(Ca2+/Mn2+) at high substratum coating concen-trations.

Some neural crest cell attachment was observed onthe El ' fragment of laminin. Although unaffected bythe HNK-1 antibody, binding of neural crest cells to theEl ' fragment was inhibited by the CSAT antibody. Thisadhesion to the El ' fragment of laminin utilizeddifferent divalent cations (Mg2+/Mn2+) than the intactmolecule (Ca2+/Mn2+). This is consistent with previoussuggestions that the El ' fragment may possess a crypticRGD site that is not exposed in intact laminin(Nurcombe et al. 1989). The RGD sequence corre-sponds to the attachment sequence in a number ofextracellular matrix molecules, including fibronectin,vitronectin and tenascin, to which integrins bind(Ruoslahti and Pierschbacher, 1987). Laminin alsocontains an RGD sequence (Sakai et al. 1988), thoughRGD-pep tides fail to inhibit neural crest cell attach-ment to laminin (Nurcombe et al. 1989). It is possiblethat an RGD-sensitive integrin, perhaps the 'fibronec-tin receptor', may be utilized for adhesion to the El 'fragment of laminin. Thus, neural crest cell attachmentto the El ' fragment of laminin is likely to be mediatedby a third distinct /Sx-integrin.

The observation of different mechanisms of attach-ment to laminin substrata created at different coatingconcentrations highlights the importance of consideringnot only the presence but also the amount of availablesubstratum molecules. Many studies have used immu-nocytochemical techniques to establish the distributionof ECM molecules in the regions through which neuralcrest cells migrate (Newgreen and Thiery, 1980;Krotoski et al. 1986; Rogers et al. 1986). Our datasuggest that the local conformation, perhaps conferredby a change in concentration, of a particular ECMmolecule may be important for cell recognition andattachment. Although we find that neural crest cellattachment to laminin increases with substratum coat-ing concentration in the presence of divalent cations,the extent of migration, as assayed by the area of celloutgrowth from the neural tube (Perris et al. 1989),actually decreases at higher laminin concentrations.Thus, more avid adhesion may lead to reducedmigration in this system. These results lead to theintriguing possibility that the two distinct adhesionmechanisms for laminin could be differentially involvedin active cell movement (at low laminin concentrations)and cessation of movement (at high laminin concen-trations). Accordingly, the conformation of laminin inthe local environment could dictate the cellularresponse.

The two distinct mechanisms of neural crest cellattachment to laminin are probably mediated byintegrins that differ in their glycosylation, divalentcation requirements and specificity. There are severalpossible explanations that could account for our

observations. For example, there may be two ft-integrins, each of which contains a distinct a subunit.The different a subunits could be the products ofdistinct genes or alternative splicing of the same gene;they may confer the different specificities for lamininconformation upon the heterodimer receptor complex.Alternatively, differential glycosylation of a single aintegrin subunit could lead to the formation offunctionally distinct receptor complexes. The glycosyl-ation could dictate both the divalent cation require-ments and the specificity for laminin conformation.Two pieces of evidence support the idea that the twoattachment mechanisms may differ in glycosylation: (1)the finding that the HNK-1 antibody (which recognizesa carbohydrate epitope) only blocks the function of onemechanism of adhesion to laminin; and (2) thebiochemical evidence demonstrating the HNK-1 epi-tope is present on a molecule which co-immunoprecipi-tates with the ^-subunit. There is some evidence thatglycosylation alters integrin function. Murine integrinssynthesized by a mutant B16 cell line deficient inglycosylation show reduced affinity for fibronectin andlaminin (Oz et al. 1989). Furthermore, the vitronectinreceptor has an associated ganglioside (Cheresh et al.1987), which is required to function in adhesion. Wecannot rule out the possibility that the two /Sj-integrinreceptor systems have different /3 subunits, both ofwhich are recognized by the CSAT antibody. Thesemodels, which are not necessarily mutually exclusive,are readily testable, and warrant further study. The165000Afr glycoprotein detected by the HNK-1 anti-body on neural crest cells may also be a integrin-associated molecule, and not a member of the or integrinfamily. Future experiments will be directed towardidentifying the a subunits and examining the effects ofglycosylation on the function of the integrins on thesurface of neural crest cells.

We thank Drs Scott Fraser and Roberto Perris for helpfulcomments on the manuscript and invaluable discussion. Wealso would like to thank Dr Mats Paulsson for his generousgift of laminin fragments and the laminin-nidogen complex.This study was supported by USPHS Grant HD-15527.

References

ABO, T. AND BALCH, C. M. (1981). A differentiation antigen ofhuman NK and K cells identified by a monoclonal antibodyHNK-1). /. Immunol. 127, 1024-1029.

AUMAJLLEY, M., NURCOMBE, V., EDGAR, D., PAULSSON, M. ANDTIMPL, R. (1989). The cellular interactions of laminin fragments.Cell adhesion correlates with two fragment-specific high affinitybinding sites. J. biol. Chem. 262, 11532-11538.

BASSON, C. T., KNOWLES, W. J., BELL, L., ALBELDA, S. M.,CASTRONOVO, V., LJOTTA, L. A. AND MADRI, J. A. (1990).Spatiotemporal segregation of endothelial cell integrin andnonintegrin extracellular matrix-binding proteins during adhesionevents. /. Cell Biol. 110, 789-801.

BrDLACK, J. M. AND MABIE, P. C. (1986). Preparation of Fabfragments from mouse monoclonal IgM. J. Immun. Meth. 91,157-162.

BOURDON, M. A. AND RUOSLAHTI, E. (1989). Tenascin mediatescell attachment through an RGD-dependent receptor. J. CellBiol. 108, 1149-1155.

Cation independent integrin on neural crest cells 1083

BOZYCZKO, D. AND HORWTTZ, A.F. (1986). The participation of aputative cell surface receptor for laminin and fibronectin inperipheral neurite extension. /. Neurosci. 6, 1241-1251.

BRONNER-FRASER, M. (1985). Alterations in neural crest migrationby a monoclonal antibody that affects cell adhesion. /. Cell Biol.101, 610-617.

BRONNER-FRASER, M. (1986). An antibody to a receptor forfibronectin and laminin perturbs cranial neural crestdevelopment in vivo. Devi Biol. 117, 528-536.

BRONNER-FRASER, M. (1987). Perturbation of cranial neural crestmigration by the HNK-1 antibody. Devi Biol. 123, 321-331.

BRONNER-FRASER, M., ARTINGER, M., MUSCHLER, J. AND HORWITZ,A. F. (1992). Developmental^ regulated expression of a^integrin in avian embryos, (submitted)

BUCK, C. A. AND HORWTTZ, A. F. (1987). Integrin, atransmembrane glycoprotein complex mediating cell-substrateadhesion. J. Cell Sci. Suppl. 8, 231-250.

BUCK, C. A., SHEA, E., DUGGAN, K. AND HORWTTZ, A. F. (1986).Integrin (the CSAT antigen): Functionality requires oligomericintegrity. /. Cell Biol. 103, 2421-2428.

BURN, P., KUPFER, A. AND SINGER, J. C. (1988). Dynamicmembrane-cytoskeletal interactions: specific association ofintegrin and talin arises in vivo after phorbol ester treatment ofperipheral blood lymphocytes. Proc. natn. Acad. Sci. U.S.A. 85,497-501.

BURNETTE, W. M. (1981). Western Blotting: Electrophoretictransfer of protein from sodium dodecyl sulfate polyacrylamidegels to unmodified nitrocellulose and radiographic detection withantibody and radioiodinated protein A. Anal. Biochem. 112,195-302.

CARTER, W. G., WAYNER, E. A., BOUCHARD, T. S. AND KAUR, P.(1990). The role of o ft and 03ft in cell-cell and cell-substratum adhesion of human epidermal cells. J. Cell Biol. 110,1387-1404.

CHARONIS, A. S., SKUBITZ, A. P. N., KOLIAKOS, G. G., REGER, L.A., DEGE, J., VOGEL, A. M., WOHLHUETER, R. AND FURCHT, L.T. (1988). A novel synthetic peptide from the Bl chain oflaminin with heparin-binding and cell adhesion-promotingactivities. /. Cell Biol. 107, 1253-1260.

CHERESH, D. A., PYTELA, R., PIERSCHBACHER, M. D., KUER, F.G., RUOSLAHTI, E. AND REISFELD, R. A. (1987). An arg-gly-asp-directed receptor on the surface of human melanoma cells existsin a divalent cation-dependent functional complex with thedisialoganglioside GD2. J. Cell Biol. 105, 1163-1173.

Dow, K. E., MIRSH, S. E., RODER, J. C. AND RIOPELLE, R. J.(1988). Neuronal proteoglycans: biosynthesis and functionalinteraction with neurons in vitro. J. Neurosci. 8, 3278-3289.

EDWARDS, J. G., HAMEED, H. AND CAMPBELL, G. (1988).Induction of fibroblast spreading by Mn2+: a possible role forunusual binding sites for divalent cations in receptors forproteins containing Arg-Gly-Asp. J. Cell Sci. 89, 507-513.

EDWARDS, J. G., ROBSON, R. T. AND CAMPBELL, G. (1987). Amajor difference between serum and fibronectin in the divalentcation requirement for adhesion and spreading of BHK21 cells.J. Cell Sci. 87, 657-665.

EHRIG, K., LEIVO, I., ARGRAVES, W. S., RUOSLAHTI, R. ANDENGVALL, E. (1990). Merosin, a tissue-specific basementmembrane protein, is a laminin-like protein. Proc. natn. Acad.Sci. U.S.A. 87, 3264-3268.

EUCES, M. J. AND HEMLER, M. E. (1989). The human integrinVLA-2 is a collagen receptor on some cells and acollagen/laminin receptor on others. Proc. natn. Acad. Sci.U.S.A. 86, 9906-9910.

FORSBERG, E., PAULSSON, M., TIMPL, R. AND JOHANSSON, S.(1990). Characterization of a laminin receptor on rathepatocytes. J. biol. Chem. 265, 6376-6381.

GEHLSEN, K. R., DICKERSON, K., ARGRAVES, W. S., ENGVALL, E.AND RUOSLHATI, E. (1989). Subunit structure of a laminin-binding integrin and localization of its binding site on laminin. /.biol. Chem. 264, 19034-19038.

GRUMET, M., HOFFMAN, S., CROSSIN, K. L. AND EDELMAN, G. M.(1985). Cytotactin, an extracellular matrix protein of neural andnon-neural tissues that mediates glia-neuron interaction. Proc.natn. Acad. Sci. U.S.A. 82, 8075-8079.

HALL, D. E., NEUGEBAUER, K. M. AND REICHARDT, L. F. (1987).Embryonic neural retinal cell response to etracellular matrixproteins: Developmental changes and effects of the cellsubstratum attatchment antibody CSAT. J. Cell Biol. 104,623-634.

HALL, D. E., REICHART, L. F., CROWLEY, E., HOLLEY, B.,MOEZZI, H., SONNENBERG, A. AND DAMSKY, C. H. (1990). Thea-i/ft and o^/A integrin heterodimers mediate cell attachmentto distinct sites on laminin. /. Cell Biol. 110, 2175-2184.

HOFFMAN, S. AND EDELMAN, G. M. (1987). A proteoglycan withHNK-1 antigenic determinants is a neuron-associated ligand forcytotactin. Proc. natn. Acad. Sci. U.S.A. 84, 2523-2527.

HORWITZ, A. F., DUGGAN, K., GREGGS, R., DECKER, C. ANDBUCK, C. (1985). The cell substrate attachment (CSAT) antigenhas properties of a receptor for laminin and fibronectin. /. CellBiol. 101, 2134-2144.

HUDSON, L. AND HAY, F. C. (1980). Practical Immunology,Oxford: Blackwell Scientific Publications.

HUNTER, D., SHAH, V., MERLIE, J. AND SANES, J. (1989). Alaminin-like adhesive protein concentrated in the synaptic cleftof the neuromuscular junction. Nature 338, 229-234.

HYNES, R. O. (1987). Integrins: a family of cell surface receptors.Cell 48, 549-554.

IGNATIUS, M. J. AND REICHARDT, L. F. (1988). Identification of aneuronal laminin receptor: An Mr 200K/120K intergrinheterodimer that binds laminin in a divalent cation-dependentmanner. Neuron 1, 713-725.

KRAMER, R. H., MCDONALD, K. A. AND VU, M. P. (1989).Human melanoma cells express a novel integrin receptor forlaminin. /. biol. Chem. 264, 15642-15 649.

KROTOSKI, D., DOMINGO, C. AND BRONNER-FRASER, M. (1986).Distribution of a putative cell surface receptor for fibronectinand laminin in avian embryo. J. Cell Biol. 103, 1061-1071.

KRUSE, J., MAILHAMMER, R., WENECKE, H., FAISSNER, A.,SOMMER, I., GORIDIS, C. AND SCHACHNER, M. (1984). Neural celladhesion molecules and myelin-associated glycoprotein share acommon carbohydrate moiety recognized by monoclonalantibodies L2 and HNK-1. Nature 311, 153-155.

KRUSE, J., KEILHAUER, G., FAISSNER, A., TIMPL, R. ANDSCHACHNER, M. (1985). The Jl glycoprotein - a novel nervoussystem cell adhesion molecule of the L2/HNK-1 family. Nature316, 146-148.

KUNEMTJND, V . , JUNGALWALA, F . B . , FISCHER, G . , C H O U , D . K.H., KEILHAUER, G. AND SCHACHNER, M. (1988). The L2/HNK-1carbohydrate on neural cell adhesion molecules is involved incell interactions. J. Cell Biol. 106, 213-223.

LAEMMLI, U. K. (1970). Cleavage of structural proteins duringthe assembly of the head of bacteriophage T4. Nature 227,680-685.

LANGUINO, L. R., GEHLSEN, K. R., WAYNER, E., CARTER, W. C ,ENGVALL, E. AND RUOSLAHTI, E. (1989). Endothelial cells usea ^ integrin as a laminin receptor. /. Cell Biol. 109, 2455-2462.

LEDOUARIN, N. M. (1982). The Neural Crest. Cambridge Univ.Press: Cambridge.

LETOURNEAU, P. C , PECH, I. V., ROGERS, S. L., PALM, S. L.,MCCARTHY, J. B. AND FURCHT, L. T. (1988). Growth conemigration across extracellular matrix components depends onintegrin, but migration across glioma cells does not. J. Neurosci.Res. 21, 286-297.

LOTZ, M. M., KORZELJUS, C. A. AND MERCURIO, A. M. (1990).Human colon carcinoma cells use multipe receptors to adhere tolaminin: involvement of ay34 and <x$\ integrins. Cell Regul. 1,249-257.

MATTHEW, W. D. AND REICHARDT, L. F. (1982). Developmentand application of an efficient procedure for converting mouseIgM into small, active fragments. /. Immun. Meth. 50,239-253.

MCCLAY, D. R., WESSEL, G. M. AND MARCHASE, R. B. (1981).Intercellular recognition: Quantitation of initial binding events.Proc. natn. Acad. Sci. U.S.A. 78, 4975-4979.

NEUGEBAUER, K. M., TOMASELLI, K. J., LIUEN, J. AND REICHARDT,L. F. (1988). N-Cadherin, NCAM, and integrins promote retinalneurite outgrowth on astrocytes in vitro. J. Cell Biol. 107,1177-1187.

1084 T. Lallier and M. Bronner-Fraser

NEWGREEN, D. F. AND THLERY, J.-P. (1980). Fibronectin in earlyavian embryos: Synthesis and distribution along the migrationpathways of neural crest cells. Cell Tissue Res. 211, 269-291.

NGO, T. T. AND LENHOFF, H. M. (1980). A sensitive and versitilechromogenic assay for peroxidase and peroxidase-coupledreactions. Anal. Biochem. 105, 389-397.

NURCOMBE, V . , AUMAILLEY, M . , TlMPL, R. AND EDGAR, D . (1989).The high-affinity binding of laminin to cells. Assignation of amajor cell-binding site to the long arm of laminin and of a latentcell-binding site to its short arms. Eur. J. Biochem. 180, 9-14.

Oz, O. K., CAMPBELL, A. AND TAO, T. (1989). Reduced celladhesion to fibronectin and laminin is associated with alteredglycosylation of pi integrins in a weakly metastatic glycosylationmutant. Int. J. Cancer 44, 343-347.

PAULSSON, M., AUMAILLEY, M., DEUTZMANN, R., TIMPL, R.,BECK, K. AND ENGEL, J. (1987). Laminin-nidogen complex:Extraction with chelating agents and structural characterization.Eur. J. Biochem. 166, 11-19.

PERWS, R., PAULSSON, M. AND BRONNER-FRASER, M. (1989).Molecular mechansims of avian neural crest cell migration onfibronectin and laminin. Devi Biol. 136, 222-239.

PESHEVA, P., HORWTTZ, A. F. AND SCHACHNER, M. (1987).Integrin, the cell surface receptor for fibronectin and laminin,expresses the L2/HNK-1 and L3 carbohydrate structures sharedby adhesion molecules. Neurosa. Letts. 83, 303-306.

ROGERS, S. L., EDSON, K. J., LETOURNEAU, P. C. AND MCLOON, S.C. (1986). Distribution of laminin in the developing peripheralnervous system of the chick. Devi Biol. 113, 429-435.

RUNYAN, R. B., MAXWELL, G. D. AND SHUR, B. B. (1986).Evidence for a novel mechanism of neural crest cell migrationon extracellular glycoconjugate matrices. J. Cell Biol. 102,432-441.

RUOSLAHTI, E. AND PIERSCHBACHER, M. (1987). New perspectivesin cell adhesion: RGD and integrins. Science 238, 491-497.

SAKAI, M., KLEINMAN, H. K., HUBER, H., DEUTZMANN, R. ANDYAMADA, Y. (1988). Laminin, a multidomain protein. The Achain has a unique glubular domain and homology with thebasement membrane proteoglycan and the laminin B chains. J.biol. Chem. 262,, 16536-16544.

SHAW, L. M., MESSIER, J. M. AND MERCURIO, A. M. (1990). Theactivation dependent adhesion of macrophages to laminininvolves cytoskeletal anchoring and phosphorylation of the 06BIintegrin. J. Celt Biol. 110, 2167-2174.

SMITH, J. W. AND CHERESH, D. A. (1988). The arg-gly-asp bindingdomain of the vitronectin receptor. Photoaffinity cross-linkingimplicates amino acid residues 61-203 of the B subunit. J. biol.Chem. 263, 18726-18731.

SONNENBERG, A . , MODDERMAN, P . W . AND HOGERVORST, F . (1988).Laminin receptor on platelets is the integrin VLA-6. Nature 336,487-489.

SONNENBERG, A . , LlNDERS, C. J . T . , MODDERMAN, P. W . , D A M S K Y ,C. H., AUMAILLEY, M. AND TIMPL, R. (1990). Integrinrecognition of different cell-binding fragments and laminin (PI,E3, E8) and evidence that a^Bi but not afe/?4 functions as amajor receptor fragment E8. J. Cell Biol. 110, 2145-2155.

STERN, C. D., NORMS, W. E., BRONNER-FRASER, M., CARLSON, G.J., PRMMETT, D., FAISSNER, A., KEYNES, R. J. AND SCHACHNER,M. (1989). Jl/tenascin-related molecules are not responsible forthe segmented pattern of neural crest cells or motor axons inchick embryo. Development 107, 309-320.

TACK, B. F., DEAN, J., EILAT, D., LORENZ, P. E. AND SCHECHTER,A. N. (1980). Tritium labeling of proteins to high specificradioactivity by reductive methylation. J. biol. Chem. 255,8842-8847.

TAN, S. S., CROSSIN, K. L., HOFFMAN, S. AND EDELMAN, G. M.(1987). Asymmetric expression in somites of cytotactin and itsproteoglycan ligand is correlated with neural crest celldistribution. Proc. natn. Acad. Sci. U.S.A. 84, 7977-7981.

TOMASELU, K. J., DAMSKY, C. H. AND REICHARDT, L. F. (1988).Purification and characterization of mammalian integrinsexpressed by a rat neural cell line (PC12): Evidence that theyfunction as a/fi heterodimeric receptors for laminin and type IVcollagen. J. Cell Biol. 107, 1241-1252.

TURNER, D. C , FLIER, L. A. AND CAXBONETTO, S. (1987).Magnesium-dependent attachment and neurite outgrowth byPC12 cells on collagen and laminin substrata. Devi Biol. 121,510-525.

TUCKER, G. C , AOYAMA, H., LIPINSKJ, M., TURSZ, T. ANDTHIERY, J.-P. (1984). Identical reactivity of monoclonalantibodies HNK-1 and NC-1, conservation in vertebrates oncells derived from neural primordium and on some leukocytes.Cell Diffn. 14, 223-230.

VON BOXBERG, Y., WUTZ, R. AND SCHWAJITZ, U. (1990). Use ofthe biotin-avidin system for the labelling, isolation andcharacterization of neural cell surface proteins. Eur. J. Biochem.190, 249-256.

{Accepted 28 August 1991)

Note added in proofWhile this paper was in press, we discovered that a polyclonalantiserum to the chicken <X\ subunit of integrin inhibitedattachment of neural crest cells to laminin in the absence ofdivalent cations, paralleling the effects of the HNK-1antibody. Chicken at integrin previously has been shown tobe a 165000Afr protein, making it a good candidate for theHNK-1 bearing molecule on neural crest cells.