HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Oleszewski, M. Articles by Altevogt, P. Search for Related Content PubMed PubMed Citation Articles by Oleszewski, M. Articles by Altevogt, P. Social Bookmarking                      What's this?

J Biol Chem, Vol. 274, Issue 35, 24602-24610, August 27, 1999

Integrin and Neurocan Binding to L1 Involves Distinct Ig Domains^*

Matthias Oleszewski, Sandra Beer, Stephanie Katich, Claudia Geiger, Yvonka Zeller, Uwe Rauch, and Peter Altevogt^§

From the Tumor Immunology Programme, G0100, German Cancer Research Center, D-69120 Heidelberg, Germany and the  Department of Experimental Pathology, University of Lund, University Hospital, S-22185 Lund, Sweden

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The cell adhesion molecule L1, a 200-220-kDa type I membrane glycoprotein of the Ig superfamily, mediates many neuronal processes. Originally studied in the nervous system, L1 is expressed by hematopoietic and many epithelial cells, suggesting a more expanded role. L1 supports homophilic L1-L1 and integrin-mediated cell binding and can also bind with high affinity to the neural proteoglycan neurocan; however, the binding site is unknown. We have dissected the L1 molecule and investigated the cell binding ability of Ig domains 1 and 6. We report that RGD sites in domain 6 support 51- or v3-mediated integrin binding and that both RGD sites are essential. Cooperation of RGD sites with neighboring domains are necessary for ₅₁. A T cell hybridoma and activated T cells could bind to L1 in the absence of RGDs. This binding was supported by Ig domain 1 and mediated by cell surface-exposed neurocan. Lymphoid and brain-derived neurocan were structurally similar. We also present evidence that a fusion protein of the Ig 1-like domain of L1 can bind to recombinant neurocan. Our results support the notion that L1 provides distinct cell binding sites that may serve in cell-cell or cell-matrix interactions.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

L1 is the prototype of a neural subfamily of cell adhesion molecules (CAMs)¹ structurally belonging to the Ig superfamily. These CAMs play an important role in the nervous system. In addition to providing mechanical stability for cell-cell or cell-matrix contact, they can transduce signals out of cells and into the cell from the microenvironment and thereby influence cellular properties like shape, adhesion, motility, growth, and differentiation (1-4).

Mouse L1 is a 200-220-kDa transmembrane glycoprotein consisting of six Ig-like domains and five fibronectin-type III repeats followed by a transmembrane region and a highly conserved cytoplasmic tail (5-8). Related molecules exist in several species including mouse (L1) (6), rat (NILE) (9), chick (NgCAM) (10), Drosophila (neuroglian) (11), and human (L1) (12, 13). L1 was originally recognized as a neural CAM shown to be involved in granule neuron migration in the developing mouse cerebellar cortex (14), the fasciculation of neurites (15), and neurite outgrowth on other neurites and Schwann cells (16, 17). Recent work on L1-knockout mice have confirmed that L1 is an important molecule for the development of the nervous system (18-20). Indeed, studies in humans have found a close association of L1 mutations with hereditary brain malformations (for reviews, see Refs. 3 and 21). L1 expression was also found on hematopoietic cells in both mouse and humans (22, 23) and on certain nonhematopoietic normal cells and tumors (24, 25). In the hematopoietic system, L1 seems to play a role in lymph node architecture, which is maintained by the fibroblastic reticular matrix (26).

Multidomain proteins like the members of the L1 family can undergo distinct binding interactions. Indeed, functional studies have shown that mouse L1 can promote binding by several mechanisms: (i) homotypic binding involving L1-L1 interactions (27, 28), (ii) assisted homophilic binding between L1 and L1·NCAM complexes at the surface of adjacent cells (27-29), and (iii) heterotypic binding to several RGD-binding integrins, i.e. ₅₁, _v1, and _v3 as well as the platelet integrin _IIb3 (7, 30-33). Integrin-mediated cell binding and migration involves Ig-like domain 6 of L1 (15, 30-33). NgCAM, the chick homologue for L1, can also bind to the glycosyl-phosphatidylinositol-anchored, axon-associated CAM TAG-1/axonin-1 (34), and mouse L1 can bind to the glycosyl-phosphatidylinositol-anchored molecule HSA/CD24 (35, 36). L1/NgCAM are also high affinity ligands for the nervous tissue-specific chondroitin sulfate proteoglycans neurocan and phosphacan (37, 38). These proteoglycans are believed to be major components of brain extracellular matrix. In addition to its function as a cell surface adhesion molecule, L1 has also been found in the extracellular matrix (31, 39); however, its function in this context is unknown.

A recent publication has questioned the role of L1 RGDs in integrin-mediated binding and has suggested a significant contribution of non-RGD sequences (40). This is an important issue, since it could mean that also other members of the L1 family not having RGDs in their Ig domain 6 might be able to support integrin-mediated binding. In addition, L1 was shown to interact with neurocan; however, the binding site in L1 is unknown. To address these questions we have dissected the L1 molecule and used site-directed mutagenesis of Ig domain 6 of L1 in combination with cell binding assays. We find that RGD sites are absolutely essential for integrin binding of ₅₁ or _v3. We observed also that the T cell hybridoma IH3-1 and activated mouse T lymphocytes can bind to L1 in a non-RGD-dependent fashion. This binding is mediated by proteoglycans (i.e. by cell surface-exposed neurocan) and involves a novel binding site in Ig-like domain 1 of L1. We present evidence that recombinant neurocan can directly bind to Ig-like domain 1.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Chemicals-- Heparin (from porcine intestinal mucosa), heparan sulfate, chondroitin sulfate, NaClO₃, 4-methylumbelliferyl--D-xyloside, heparinase II, heparinase III (EC 4.2.2.8), chondroitinase ABC (EC 4.2.2.4), vitronectin, and bovine fibronectin were obtained from Sigma (Deisenhofen, Germany). The BCA protein determination kit was obtained from Pierce (KMF, St. Augustin, Germany). Laminarin sulfate (41) was a gift from Dr. Israel Vlodavsky, Hadassah Hospital (Jerusalem, Israel).

Antibodies-- The following mAbs were used and have been described in previous publications (22, 30): mAb 324 against mouse L1 adhesion molecule, mAb 79 against mouse CD24 (heat-stable antigen), mAb HM5-1 against mouse ₅ integrin, mAb RMV-7 against mouse _v integrins, and mAbs HM-1 and HM-3 against mouse ₁ and ₃ integrin chains, respectively. mAbs were used in a purified form or as hybridoma supernatants. The antibody against neurocan was produced in rabbits using rat brain-derived neurocan for the first injections followed by booster injections with recombinant neurocan isolated from tissue culture supernatant of HEK 293 cells (see below). In Western blot, the antibody reacted, aside from intact neurocan, predominantly with the 150-kDa core protein fragment (see Fig. 9).

Cell Culture-- B16F10 melanoma cells were provided by Dr. Margot Zöller, DKFZ. The mouse T cell hybridoma IH3-1 was obtained from Dr. Bruno Kyewski of our department. Spleen cells were collected from 6-8-week-old DBA/2 mice. Erythrocytes were lysed by brief incubation in 155 mM NH₄Cl, 0.1 mM EDTA, 10 mM KHCO₃ solution followed by washing of the cells. T lymphocytes were activated with ConA or immobilized mAb to CD3 (mAb 500A12) as described previously (42). Blasts were isolated by density centrifugation using histopaque 1077 (Sigma). Resting T lymphocytes were isolated from mesenteric lymph nodes by depletion of B lymphocytes with the B220-specific mAb RA36B2 and Dynabeads coupled with sheep anti-rat IgG (Dynal, Hamburg, Germany). The purity of the T cells was assessed by staining with fluorescein isothiocyanate-conjugated anti-mouse CD3 and was approximately 95%. All cells were cultivated at 37 °C, 5% CO₂, and 100% humidity. IH3-1 cells were incubated for 72 h in the presence of 30 mM NaClO₃ for the inhibition of sulfation or in the presence of 1 mM 4-methylumbelliferyl--D-xyloside to interfere with proteoglycan biosynthesis (43).

Cytofluorography-- The surface staining of cells with saturating amounts of mAbs, either hybridoma supernatants or purified antibodies, and phycoerythrin-conjugated goat antibodies to rat Ig or goat anti-rabbit (SERVA, Heidelberg, Germany) or hamster Ig (Dianova, Hamburg, Germany) has been described elsewhere (32). Stained cells were analyzed with a FACScan (Becton & Dickinson, Heidelberg, Germany).

Construction of Recombinant Proteins-- An L1-Fc fusion protein containing the Fc portion of human IgG1 and the ectodomain of mouse L1 (amino acids 1-1117) was constructed by PCR using an L1 cDNA clone as template and the primer 5'-TTAAGCTTGCAGAAAGATGGTCGTGATGCTGCGGTACGTGTG-3' (containing a HindIII site, a Kozak sequence, and the start codon) and the reverse primer 5'-GGGAATTCACTTACCTGTAGTAGAAACTCGCACAGGGCCAGT-3'. The latter primer contained an overhang with an artificial splice donor site and an EcoRI restriction site to allow subsequent directional cloning into the HindIII/EcoRI site of the pIg vector (32). For all PCR reactions, 2 units of Taq polymerase expand long template mixture (Roche Molecular Biochemicals, Germany) was used. The 6.L1-Fc fusion protein containing only Ig-like domain 6 of mouse L1 (amino acids 526-603) fused to the human IgG1 Fc portion was constructed in a similar fashion as the respective human construct (32). For site-directed mutagenesis, the L1-Fc or 6.L1-Fc were excised from the pIg vector and subcloned in Bluescript SK(+). Single or double mutations D555E and D564E, respectively, were introduced using the Chameleon or the Quickchange^TM site-directed mutagenesis kit (Stratagene, Heidelberg, Germany). Mutants on the background of 6.L1-Fc were denoted mutA-C. Mutants derived from L1-Fc were termed mutI-III. All mutations were confirmed by DNA sequencing.

Fc fusion proteins containing variable numbers of Ig-like domains (1-6 domains) were constructed using PCR and mouse L1 cDNA in pCDM8 as template. The T7 promoter primer served as sense primer. The antisense primers were designed as described above to introduce an artificial splice site and an EcoRI site in the 3' overhang. The following primers were used: domain 1, 5'-ATGAATTCACTTACCCTCATGCGACATGGCAGTTCC-3'; domains 1 and 2, 5'-ATGAATTCACTTACCCTTTTGAATGATGGTCCGGGT-3; domains 1-3, 5'-ATGAATTCACTTACCCCGGGCACTGCCCAGCGAGTT-3'; domains 1-4, 5'-ATGAATTCACTTACCCAGGAGCCCATGCTGGTTGCG-3'; domains 1-5, 5'-ATGAATTCACTTACCCAAAATGGTCACATTGTTCTG-3'; domains 1-6, 5'-ATGAATTCACTTACCTGCCCTGCTCTCCACCTCAT-3'. Amplified fragments with the expected size were digested with EcoRI and HindIII and cloned into the pIg vector as described (44). Plasmid-DNA was transfected into COS-7 cells, and supernatants containing the Fc fusion protein were collected.

Biochemical Analysis-- Lactoperoxidase-catalyzed iodination of intact cells was carried out as described (45). Following the labeling reaction, the cells were washed three times in PBS and lysed at 4 °C for 15 min in TBS containing 1% Nonidet P-40 in the presence of 2 mM Ca²⁺ and Mg²⁺ ions. Lysates were prepared by centrifugation in an Eppendorf centrifuge at 4 °C for 15 min and precleared with 30 µl of packed rat Ig coupled to Sepharose. Immunoprecipitations were carried out using mAbs preadsorbed to protein-G-Sepharose for 1 h at 4 °C. The precipitates were washed four times in lysis buffer and eluted from the Sepharose by boiling for 2 min in nonreducing SDS sample buffer. SDS-PAGE was performed on 7.5% slab gels. Gels were dried and exposed to x-ray sensitive films (Kodak Biomax-MS) using the Biomax MS intensifying screen.

Purification of Proteins-- Fc fusion proteins were purified by protein A-Sepharose chromatography and analyzed using a capture enzyme-linked immunosorbent assay. This procedure has been described in detail elsewhere (32). Purified proteins were analyzed by SDS-PAGE and Western blot analysis. Western blot analysis following SDS-PAGE, blotting to Immobilon membranes, incubation with specific antibodies, and detection of bound antibodies using the ECL system (Amersham Pharmacia Biotech, Freiburg, Germany) was done as described before (30). A human P-selectin-Fc fusion protein was used in all experiments as control. Native L1 (with cytoplasmic tail) was isolated from mouse brain (30) or from the supernatant of B16F10 cells (only ectodomain).² Neurocan was isolated from mouse brain, activated T blasts, or IH3-1 cells by DEAE-Sephacel chromatography as described (45).

Cell Adhesion Assays-- For the binding of cells to fusion proteins, the purified proteins were diluted in Tris-buffered saline (approximately 10 µg/ml) and used to coat LABTEK glass chamber slides (Nunc, Wiesbaden, Germany) for 16 h at 4 °C. Fusion proteins were also adsorbed to slides precoated with anti-human IgG-Fc (Dianova, Hamburg, Germany) to allow directional coating. Equal coating of the wells was controlled in parallel by performing an anti-human Fc-specific enzyme-linked immunosorbent assay. Following coating, the wells were blocked with 1% bovine serum albumin in PBS for 1 h at room temperature, washed with Hanks' balanced salt solution containing 10 mM Hepes and used for the assay. For binding, cells (5 × 10⁶/ml) were suspended in Hanks' balanced salt solution containing 10 mM Hepes, 0.5 mM Mn²⁺, and 0.2-ml aliquots were added to the coated slides. Where indicated, the Mn²⁺ ions were substituted with 2 mM Ca²⁺ or 2 mM Mg²⁺/EGTA. The binding assay was performed for 30 min at room temperature without shaking, and the slides were fixed in 4% glutaraldehyde in PBS after briefly dipping into PBS. For antibody blocking studies, cells were preincubated with purified antibody (10 µg/ml final concentration) for 10 min at room temperature and then transferred to the chamber slides. Cell binding was measured by counting six independent × 10 fields by video microscopy using IMAGE 1.47 software.

Glycanase Treatment of Cells-- IH3-1 cells were incubated with 1 unit/ml each of heparinase II, heparinase III, and chondroitinase ABC in RPMI 1640 medium for 1 h at 37 °C. As a control, untreated cells were incubated in medium only. The treatment did not significantly reduce the cell viability as tested by trypan blue exclusion. Cells were washed twice in Hanks' balanced salt solution and split, and half of the cells were tested by fluorescence-activated cell sorting analysis for _v integrin expression. The other half were analyzed for binding to L1-Fc mutIII protein or vitronectin.

Isolation of RNA and Reverse Transcriptase-PCR Analysis-- The isolation of total RNA from cells has been described in detail elsewhere (32). Total RNA (6 µg) was transcribed into cDNA using Moloney murine leukemia virus reverse transcriptase (Promega, Heidelberg, Germany) and oligo(dT)₂₀ for priming. The RNA/DNA hybrid was treated with RNase H and used as template for PCR using an anealing temperature of 60 °C and 35 cycles of 80 s. PCR products were separated on a 1% agarose gel containing 0.5 µg/ml ethidium bromide. Primers for rat phosphacan (accession no. U04998) were as follows: sense, 5'-CGGGTATGTCATGTTGATGGATTACTTAC-3'; antisense, 5'-TATGGAAACCATATTTAAGGATTCTGCAGT-3'. Primers for mouse neurocan (accession no. X84727) were as follows: sense, 5'-TCAGAGGCCCTAAGTGCTGTCTCCC-3; antisense, 5'-CGCCTGATGTGGGACTCCCTGGTAA-3'. PCR products were subcloned in pCR-blunt (Invitrogen, Groningen, The Netherlands), and the identity of the expected product was verified by DNA sequencing or restriction mapping.

Analysis of Neurocan Binding-- Recombinant neurocan was produced in HEK 293 cells and purified from the culture supernatant on a mAb 1D1 affinity column as described (45). For the analysis of L1-neurocan interaction, 1 µg of neurocan was adsorbed to Immobilon membranes using a dot blot apparatus. Membranes were incubated with antibodies or Fc fusion proteins at the indicated dilutions and the respective secondary antibodies followed by ECL detection as described above for Western blots.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Integrin-mediated Cell Binding to Mouse L1-- The starting interest of our study was to evaluate the role of RGDs in integrin-mediated cell binding to mouse L1. A series of Fc fusion proteins and RGD mutants of the whole extracellular part of L1 (L1-Fc) or of the single domain 6 (6.L1-Fc) were therefore constructed and are presented below (see Figs. 3A and 4).

We selected suitable cell lines that could bind to L1 in an integrin-dependent fashion. Mouse B16F10 cells and the T cell hybridoma IH3-1 were chosen. As shown in Fig. 1A, these cells express ₅, _v, ₁, and ₃ integrin chains. In addition, B16F10 cells express L1, whereas IH3-1 cells were negative as revealed by fluorescence-activated cell sorting analysis (Fig. 1A) and by immunoprecipitation analysis (not shown). The heterodimer composition was examined by immunoprecipitation analysis. As shown in Fig. 1B, B16F10 cells expressed ₅₁ and _v3, respectively. The IH3-1 cells showed predominantly _v3, little _v1, and no ₅₁. Antibody inhibition studies revealed that the binding of B16F10 cells to L1-Fc was blocked by mAbs to ₅ or ₁ integrins but not by _v-specific antibodies, suggesting that ₅₁ was the dominant integrin in these cells. In contrast, the binding of the T cell hybridoma was reduced by mAbs to _v and ₃ (Fig. 1C). In conclusion, the selected cell lines used either ₅₁ or _v3, respectively, for the binding to mouse L1-Fc.

View larger version (38K): [in this window] [in a new window]   Fig. 1.   Integrin expression of B16F10 and IH3-1 T cell hybridoma cells. A, indirect immunofluorescence staining of the cells used in the present study. Cells were stained using mAb HM-1 (1 integrin subunit), HM5-2 (5 integrin subunit), HM-3 (3 integrin subunit), or RMV-7 (v integrin subunit) followed by phycoerythrin-conjugated goat anti-rat IgG or anti-hamster IgG, respectively. For negative control, the first antibody was omitted; B, immunoprecipitation analysis of heterodimer expression. Cells were surface-labeled with 125I and lysed in the presence of 1% Nonidet P-40 plus 2 mM Ca2+ and Mg2+ ions. Lysates were incubated with the indicated mAbs adsorbed to protein G-Sepharose. All samples were analyzed by SDS-PAGE under nonreducing conditions; C, antibody inhibition of cell binding to L1-Fc. The adhesion assay was done in the presence of 0.5 mM Mn2+ ions to activate integrins. Under these conditions, homophilic L1-L1 binding of B16F10 cells is marginal. The final concentration of blocking mAbs was 10 µg/ml. BSA, bovine serum albumin.

B16F10 cells also express L1 at the cell surface, raising the possibility of homotypic L1-L1 binding as a potential complication of the assay system. In separate experiments, it was however noted that in the presence of Mn²⁺-activated integrins the contribution of L1-L1 binding of B16F10 cells was marginal (data not shown; also see Fig. 4).

Role of Divalent Cations and Individual RGD Sites-- It has been reported that ₅₁ recognition of L1 in humans is strictly dependent on the type of divalent cations (33). We made use of the 6.L1-Fc, which allows analysis of integrin-mediated binding without interference of other sites of the molecule, to investigate the role of divalent cations. As shown in Fig. 2, the adhesion of B16F10 or IH3-1 cells to the 6.L1-Fc protein was distinct in the requirement for divalent cations. The ₅₁-mediated binding of B16F10 was approximately 4-fold better in the presence of Mn²⁺ ions than in the presence of Ca²⁺ or Mg²⁺/EGTA ions. In contrast, the _v3-mediated binding of IH3-1 cells was best in the presence of Mg²⁺/EGTA ions.

View larger version (15K): [in this window] [in a new window]   Fig. 2.   Divalent cation requirement for the 51- and v3-mediated cell binding to 6.L1-Fc. Adhesion of B16F10 melanoma cells and IH3-1 T cell hybridoma cells to 6.L1-Fc coated at 10 µg/ml in the presence of different divalent cations is shown. Final concentrations were 0.5 mM Mn2+, 2 mM Ca2+, and 2 mM Mg2+/EGTA.

To investigate whether cell adhesion was entirely dependent on the RGD motifs, the 6.L1-Fc and its derived RGE mutants were investigated. As shown in Fig. 3B, B16F10 cells were highly dependent on both intact RGD sites. Replacement of one or the other site by RGE reduced the binding by >90%. The level of binding to single mutants was similar to the level seen with double mutants, and the addition of anti-₁ or ₅ specific mAbs did not reduce the binding much further. The analysis of _v3 gave different results. As shown in Fig. 3C, the binding of IH3-1 cells to 6.L1-Fc mutA or -B proteins was each reduced by 50-60%, and only little binding was present to the double mutant, mutC. The addition of _v- or ₃-specific antibodies in each case reduced the binding to background levels.

View larger version (28K): [in this window] [in a new window]   Fig. 3.   Effect of RGD mutations on cell binding to 6.L1-Fc. A, ECL blot of purified fusion proteins. wt, 6.L1-Fc; mutA, 6.L1-Fc with the first RGD mutated to RGE; mutB, 6.L1-Fc with the second RGD mutated to RGE; mutC, 6.L1-Fc mutated in both RGDs. B and C, adhesion of B16F10 melanoma cells or IH3-1 T cell hybridoma cells to the 6.L1-Fc fusion proteins and the derived mutants coated at 10 µg/ml. The assay was performed in the presence of 0.5 mM Mn2+ ions. Antibodies were added to a final concentration of 10 µg/ml

The RGD-mediated binding to domain 6 could be influenced by neighboring domains. To investigate this question, we compared L1-Fc to its RGE-mutated forms. For ₅₁, the results were indeed quite different. As shown with B16F10 cells in Fig. 4, each RGD site contributed about half of the binding, and a complete loss of activity occurred only when both RGD sites were absent. In contrast to the results obtained with the single domain 6, the binding of IH3-1 cells to L1-Fc and the mutant forms was only marginally different, and the cells were still capable of binding to L1-Fc mutIII protein in which both RGDs are absent (Fig. 4).

View larger version (25K): [in this window] [in a new window]   Fig. 4.   Effect of RGD mutations on cell binding to L1-Fc. Comparison of adhesion between B16F10 and IH3-1 cells to the L1-Fc and its RGE mutants coated at 10 µg/ml. The assay was carried out in the presence of 0.5 mM Mn2+ ions. The inset shows the purified fusion proteins. wt, L1-Fc; I, L1-Fc with the first RGD mutated to RGE; II, L1-Fc with the second RGD mutated to RGE; III, L1-Fc mutated in both RGDs. BSA, bovine serum albumin.

RGD-independent Cell Binding to L1-- This observation raised the possibility that IH3-1 cells used an additional binding site outside domain 6. Indeed, binding to this new site was not blocked in the presence of antibodies to _v and ₃ (Fig. 5A), was seen both at 4 and 20 °C (Fig. 5B), and was independent of divalent cations (not shown).

View larger version (20K): [in this window] [in a new window]   Fig. 5.   RGD-independent binding to mouse L1. A, effect of antibody-mediated blocking of IH3-1 cells to L1-Fc or the RGD-deficient L1-Fc mutIII protein. Note that binding to the mutant protein is not blocked by mAbs to v and 3 integrins. Antibodies were used at 10 µg/ml, and the assay was performed in the presence of 0.5 mM Mn2+ ions. B, comparison of IH3-1 cell binding with L1-Fc and L1-Fc mutIII at room temperature (20 °C) or 4 °C. C, binding of IH3-1 cells to native L1 derived from mouse brain or to the shed ectodomain form derived from B16F10 melanoma cells in the presence or absence of laminarin sulfate at the indicated concentration.

To rule out the possibility that the observed binding was an artifact of the Fc fusion proteins, native L1 was tested. As shown in Fig. 5B, IH3-1 cells could bind to native L1 isolated from the cell surface (surface L1) or to a soluble L1 form (shed L1) containing the whole ectodomain of L1. Also on native L1 forms, the binding was insensitive to low temperature. These findings suggested that it was not mediated by integrins and, since IH3-1 cells are negative for L1, was not due to an L1-L1 homophilic interaction.

Evidence for a Role of Cell Surface Proteoglycans-- To further characterize the RGD-independent binding to L1, inhibition studies were carried out. As shown in Fig. 6A, the binding of IH3-1 cells to L1-Fc mutIII or L1-Fc was blocked by heparin in a dose-dependent fashion, whereas the binding to 6.L1-Fc, which is solely integrin-mediated, was not affected. Based on this observation, we considered that the additional binding to L1 was mediated by proteoglycans.

View larger version (32K): [in this window] [in a new window]   Fig. 6.   Integrin-independent IH3-1 cell binding is inhibited by heparin and glycanase treatment. A, adhesion of IH3-1 T cell hybridoma cells to L1-Fc, L1-Fc mutIII, or 6.L1-Fc fusion proteins coated at 10 µg/ml. The assay was performed in the presence of 0.5 mM Mn2+ ions. Concentrations of heparin are indicated. B, cells were treated with heparinase II and III and chondroitinase ABC in medium or with medium alone and analyzed for binding on L1-Fc mutIII or vitronectin coated at 10 µg/ml. C, cells were also stained and analyzed by FACScan for expression of v integrins.

Other sulfated polysaccharides have been identified that to some extent can mimic the biological effects of heparin. One of the compounds is laminarin sulfate, which is a linear polymer consisting of 1,3-glucan derived from the cell wall of seaweed and modified by chemical O-sulfation (9). Laminarin sulfate could inhibit the binding of IH3-1 cells to L1-Fc mutIII (IC₅₀ < 5 µg/ml) as well as to the native L1 forms (Fig. 5C). Other polysulfated substances like heparan sulfate and chondroitin sulfate were able to inhibit but were less potent (IC₅₀ approximately 600-800 µg/ml). All substances did not interfere with the ₅₁-mediated binding of B16F10 cells to L1-Fc (not shown).

As shown in Fig. 6B, a reduction of IH3-1 cell binding to L1-Fc mutIII was seen following cell surface removal of GAGs by treatment with a mixture of heparinases and chondroitinase ABC as described under "Materials and Methods." The treatment did not affect cell viability significantly (not shown). The expression of _v integrin at the cell surface used as control was not altered, and the ability of the cells to bind to vitronectin was not affected (Fig. 6, B and C). A reduction in cell binding was also observed following treatment of IH3-1 cells with sodium chlorate, an inhibitor of sulfation, or 4-methylumbelliferyl--D-xyloside, which interferes with proteoglycan synthesis. The binding was reduced by 50 or 46%, respectively. Collectively, these data suggested that the binding of cells to RGD-deficient L1 was mediated by cell surface-expressed or -bound proteoglycans.

Binding of Activated T Lymphocytes-- The observation that the T cell hybridoma IH3-1 could bind to L1-Fc mutIII protein prompted us to investigate activated T lymphocytes. Indeed, as shown in Fig. 7, ConA- or anti-CD3-activated T blasts, but not resting T lymphocytes, were able to bind to L1-Fc mutIII efficiently and, similar to IH3-1 cells, the binding was blocked by laminarin sulfate.

View larger version (14K): [in this window] [in a new window]   Fig. 7.   Integrin-independent binding of activated mouse T lymphocytes to L1. Mouse mesenteric lymph node cells were activated with ConA, and blast cells were isolated by Ficoll centrifugation. T cell blasts (act T) or resting T lymphocytes isolated from the same source (rest T) were allowed to bind to the indicated Fc fusion proteins or fibronectin coated at 10 µg/ml. The assay was performed in the presence of 0.5 mM Mn2+ ions. Laminarin sulfate was used at a final concentration of 10 µg/ml. Note that similar results were obtained when the cells were activated with immobilized CD3 antibody instead of ConA.

Mapping of the Proteoglycan Binding Site in L1-- The temperature-independent binding of IH3-1 cells was also seen on a fusion protein consisting of Ig domains 1-6 of L1 (1-6.L1-Fc) (not shown). To map the respective site, Fc proteins containing successive deletions of domains at the C terminus were constructed and are presented in Fig. 8A. IH3-1 cells showed comparable binding to the fusion proteins containing L1 domains 1, 1-2, 1-3, 1-4, or 1-5 (data not shown). These results suggested that the binding site was located most likely in the N-terminal domain 1. As shown in Fig. 8B, the domain 1 fusion protein (1.L1-Fc) could support the binding of IH3-1 cells. To further corroborate this observation, the 1.L1-Fc protein was used in competition experiments. Fig. 8B presents evidence that in the presence of 1.L1-Fc the binding of IH3-1 cells to L1-Fc or L1-Fc mutIII substrates was blocked by approximately 30 or 53%, respectively. The 1.L1-Fc did not, however, interfere with the integrin-mediated binding of the cells to 6.L1-Fc (Fig. 8B). Additional experiments shown in Fig. 8C indicated that the 1.L1-Fc protein was able to block the binding of IH3-1 cells to L1 mutIII in a dose-dependent fashion, whereas the 6.L1-Fc protein used for control did not have a blocking effect.

View larger version (31K): [in this window] [in a new window]   Fig. 8.   Mapping of the proteoglycan binding site in mouse L1. A, SDS-PAGE and Western blot analysis of L1-Fc fusion proteins with variable numbers of Ig-domains. B, binding of IH3-1 T cell hybridoma adhesion to the indicated Fc fusion proteins coated at 10 µg/ml in the presence or absence of 4 µg/ml 1.L1-Fc. C, serial dilutions of 1.L1-Fc or 6.L1-Fc were tested for inhibition of IH3-1 T cell binding to L1-Fc mutIII protein coated at 10 µg/ml. The assay was performed in the presence of 0.5 mM Mn2+ ions.

Role of Phosphacan and Neurocan-- NgCAM/L1 can bind to the neural proteoglycans phosphacan and neurocan (43, 46). To test a possible role of these proteoglycans in the binding of cells to L1-Fc mutIII, we performed reverse transcriptase-PCR analysis using primer pairs specific for the core proteins of phosphacan or neurocan. A phosphacan-specific band of the expected size was only detected in mouse brain but not in IH3-1 cells or T cell blasts (not shown). A similar analysis for neurocan expression revealed an expected band of 459 base pairs in mouse brain, activated T lymphocytes, the T lymphoma cell line EL-4, and by nested PCR also in IH3-1 cells (not shown).

This prompted us to investigate the presence of neurocan at the cell surface using specific antibodies. As revealed by fluorescence-activated cell sorting analysis, IH3-1 cells and activated T cells showed appreciable cell surface staining for neurocan (Fig. 9A). Resting T lymphocytes revealed low but detectable levels of neurocan expression (mean fluorescence 8 versus 22 for activated T cells).

View larger version (35K): [in this window] [in a new window]   Fig. 9.   Detection of cell surface-bound neurocan. A, indirect immunofluorescent staining of IH3-1 cells or T cell blasts with the neurocan-specific antibody. B, inhibition of IH3-1 cell binding to 1.L1-Fc by the neurocan-specific antibody. Final dilution of anti-neurocan and control serum was at 1:100. C, SDS-PAGE analysis of recombinant neurocan purified from the tissue culture supernatant of HEK 293 cells. Neurocan was treated or nontreated with chondroitinase ABC (Ch'ase) for 1 h before analysis. D, Western blot analysis of neurocan isolated from IH3-1 cells, T blasts, or mouse brain derived from animals of different ages.

To analyze whether antibody reactivity was dependent on intact proteoglycans, IH3-1 cells were pretreated with sodium chlorate, 4-methylumbelliferyl--D-xyloside or treated with chondroitinase ABC, respectively. All treatments reduced the staining intensity by approximately 50% (not shown). We also studied whether the antibody was able to inhibit the binding of IH3-1 cells to 1.L1-Fc protein. Indeed, an inhibition of approximately 68% in comparison with normal serum or the medium control was observed on 1.L1-Fc (Fig. 9B). A similar degree of inhibition was seen on L1-Fc mutIII or 1-6.L1-Fc substrates (not shown).

To examine whether the lymphoid cells expressed authentic neurocan, we carried out biochemical analysis. As shown in Fig. 9C, purified recombinant neurocan migrated as a diffuse band at the top of the gel at approximately 300 kDa. Chondroitinase ABC digestion revealed a band at 250 kDa (intact neurocan) and, similar to brain-derived neurocan, two protease cleavage products at 150 and 130 kDa, respectively (38). The cleavage is due to the presence of endogenous proteases in the conditioned medium of 293 cells.

We examined neurocan from mouse brain, activated T cells, or IH3-1 by Western blot analysis, and the results are presented in Fig. 9D. The neurocan-specific antibody did not detect the intact 250-kDa band, indicating that all of the material had been cleaved. Instead, the antiserum reacted strongly with the 150-kDa cleavage product. Thus, by molecular weight and appearance before and after chondroitinase treatment, the brain- and lymphocyte-derived forms of neurocan appeared to be similarly if not identically processed. We concluded that activated T cells and IH3-1 cells expressed authentic neurocan and that the binding of cells to L1 was mediated most likely by cell surface-exposed neurocan.

Binding of 1.L1-Fc to Neurocan-- Finally, we investigated the interaction of L1-Fc with purified recombinant neurocan at the molecular level. As shown in Fig. 10, using a dot blot immunoassay with immobilized neurocan, strong binding of L1-Fc, L1-Fc mutIII, and the truncated forms 1.L1-Fc and 1-2.L1-Fc to neurocan was observed. In contrast, no binding was seen with 6.L1-Fc or with the control P-selectin-Fc fusion protein. These results strongly supported the cell binding data, showing that the first Ig-like domain of mouse L1 carries a neurocan binding site.

View larger version (45K): [in this window] [in a new window]   Fig. 10.   Binding of neurocan to L1. Recombinant neurocan (1 µg) was adsorbed to Immobilon membranes using a dot blot apparatus. Membranes containing the adsorbed material were incubated with Fc fusion proteins or antibodies to neurocan at the indicated dilutions followed by the respective secondary antibodies and ECL detection as described above. Note that equal concentration of Fc proteins in the supernatant was monitored by enzyme-linked immunosorbent assay.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

In the present study, we have investigated Ig domains 1 and 6 of mouse L1 for their ability to support cell binding. Domain 6 contains the two RGD sites previously implicated in integrin-mediated binding (15, 30-33, 40). While analyzing mutant L1 fusion proteins devoid of RGDs, we noticed that certain cell types were able to bind in a non-integrin-dependent fashion. This binding was proteoglycan-mediated and was directed against domain 1 of L1. It is convenient to discuss these results separately.

Analysis of the role of RGD sites in integrin binding to L1 was initiated by mutants derived from the 6.L1-Fc. The binding of cells via ₅₁ or _v3 to the fusion proteins required different divalent cations. The ₅₁ integrin was strongly dependent on the presence of Mn²⁺ ions, whereas _v3 preferred Mg²⁺ ions (Fig. 2). The recognition of RGDs in L1 also appeared to be distinct for the two integrins. For ₅₁, both RGD sites were necessary to support binding to domain 6 (Fig. 3B); however, in the context of the whole L1 molecule, each single RGD was active (Fig. 4). Thus, in the whole molecule, each RGD site was independent and contributed equally, whereas in domain 6 both RGDs were cooperating in creating the cell binding site. The binding specificity of integrin receptors is determined by the RGD site that collaborates with specific nonhomologous flanking residues and spatially separate "synergy" sequences (47). It may be that the RGDs in domain 6 of L1 are dependent on synergistic sites outside of domain 6. Synergy sites for ₅₁ are well studied in fibronectin (48). The RGD in the 10th type III repeat is supported by the synergistic site PHSRN in the ninth repeat (47, 48). The adhesive activity of ₅₁ to fibronectin can further be modulated by the alternatively spliced EDA segment (46). The _v3 integrin recognition of RGD appears to be more promiscuous and does not require the presence of the synergy site in the ninth repeat (49). The _v3-mediated binding to 6.L1-Fc was only decreased to half when each RGD site was mutated (Fig. 3C). There was only a residual _v3 binding left when both sites were mutated to RGE. The binding could not be studied in the context of the whole L1 molecule using IH3-1 cells due to the interference of GAG-mediated binding. However, the non-RGD binding of _v3 to L1 must be very small, since we did not observe any blocking effect of _v- or ₃-specific mAbs in the binding of IH3-1 cells to L1-Fc mutIII protein (Fig. 5A). In conclusion, our data support the previous notion that RGD motifs in domain 6 support ₅₁ or _v3 integrin binding to L1 and leave little room for a non-RGD-mediated contribution.

In a recent study, Blaess et al. have performed similar mutagenesis studies using a bacterial expressed domain 6 of L1 and solid phase immunoassay with purified integrins (40). The authors demonstrated that only the adhesion of _v3 but not ₅₁ or _IIb3 to domain 6 could be blocked specifically by a linear RGD-containing peptide. Moreover, it was shown that only the binding to a D555E mutant of domain 6 (equivalent to our 6.L1-Fc mutA) but not on the D564E mutant (equivalent to our 6.L1-Fc mutB) was blocked by the peptide. The authors concluded that an RGD-independent mechanism was responsible for the interactions of domain 6 with _II3 and ₅₁ and that only RGD 562-564 was involved in binding to _v3 (40). The results from our study do not support these conclusions. The reasons for the discrepancies between the previous study and our data could be due to several reasons. (i) The failure of Blaess et al. (40) to block ₅₁ binding efficiently could reflect that a short linear peptide does not have high enough affinity to compete against two adjacent RGD sites. Unfortunately, the study did not investigate the binding of ₅₁ to RGD mutant proteins. (ii) Blaess et al. performed their study with purified human integrins. Such an approach may not be adequate to mimic the situation in the living cell. In addition to the recognition of target sequences, integrin affinity is regulated by cytoskeletal movement, leading to clustering of integrin molecules and the formation of focal adhesion sites, which strengthen the binding to the ligand. (iii) The bacterial expressed domain 6 of Blaess et al. and the fusion proteins expressed in eukaryotic cells used in our study may be different. The bacterial fusion protein is a monomer, whereas the Fc fusion proteins are dimers. In addition, it is known that the folding of Ig-like globular domains can be influenced by glycosylation and interaction with neighboring domains. Indeed, a putative N-glycosylation site (position 587) is present in domain 6 adjacent to the second Cys residue. Differences in domain folding could alter the interaction with integrin receptors.

The RGD-independent adhesion of IH3-1 cells and activated T lymphocytes to L1 led us to study a second mechanism that appeared to be dependent on sequences outside domain 6. The binding of the cells to L1 was blocked by polysulfated glycans and was significantly reduced by treatment of the cells with glycanases or metabolic modulators of proteoglycan synthesis. A recent study has readdressed the binding of heparin to CD31 and comes to the conclusion that this interaction was the result of an artifact due to the usage of recombinant Fc proteins (50). According to this study, artificial binding sites for heparin can be created if basic amino acids accidentally present in the expressed C-terminal portion of CD31 come into the vicinity of basic amino acids in the adjacent hinge region of the human Fc portion (50). To exclude this possibility for L1-Fc proteins, we used native L1 as substrate. Native L1 supported IH3-1 cell binding with similar features like temperature insensitivity and the ability to inhibit the interaction in the presence of laminarin sulfate as seen with recombinant L1-Fc mutIII (Fig. 5, B and C). These findings argued against the possibility of an Fc fusion protein artifact. Using domain deletion forms of L1, we have mapped the binding site to the first Ig-like domain of L1 (Fig. 8). At present, we cannot determine whether the first domain is the only domain involved in GAG binding.

The GAG-mediated binding of cells to L1 was not too surprising. Previous studies have indicated that NgCAM and rat L1 can bind with high affinity to the neural tissue-specific proteoglycans phosphacan and neurocan (37, 38). To analyze whether phosphacan or neurocan were involved in the binding of IH3-1 cells or activated T lymphocytes to L1, their expression was first analyzed by reverse transcriptase-PCR. Phosphacan mRNA was only detected in brain tissue. Neurocan mRNA was observed in the brain as expected but surprisingly also in activated T lymphocytes. Since also the T lymphoma cell line EL-4 and the T hybridoma IH3-1 cells expressed the neurocan transcript, it is likely that the signal seen in activated T cells was indeed derived from these cells and not from residual contaminating cells. Neurocan belongs to the group of structurally related proteoglycans of the aggrecan family that are characterized by an N-terminal hyaluronan binding domain, a mucin-like middle part of variable length containing several chondroitin sulfate and oligosaccaride side chains, and a C terminus with epidermal growth factor-like, C-type lectin-like, and complement regulatory-like domains (51). Using a specific antibody, we were indeed able to detect neurocan at the cell surface of T cell blasts and IH3-1 cells. Biochemically, the neurocan forms isolated from mouse brain or from the T cells were structurally similar if not identical (Fig. 9D). Thus, an important observation of the present study is that lymphoid cells express authentic neurocan at the cell surface. Cell surface-associated GAGs have been implicated in several biological functions including the binding of chemokines (52, 53) and mediators of cellular adhesion (54). A coordinate role of cell surface chondroitin sulfate proteoglycans and integrins in mediating melanoma cell adhesion to fibronectin and type I collagen has been demonstrated (55, 56). Also, in our experiments it appeared that in addition to integrins the cell surface-expressed neurocan contributed to the binding of IH3-1 cells to the L1 substrate. This is supported by the results of antibody-blocking studies, enzymatic GAG removal from the cell surface, and (although indirectly) by the observed effects of the proteoglycan synthesis modulators sodiumchlorate and 4-methylumbelliferyl--D-xyloside.

The localization of the GAG binding site in the first Ig-like domain of L1, to which neurocan could bind, was astonishing. Previous studies have mapped the homophilic binding site for L1-L1 interaction to the second Ig-like domain (57, 58). Moreover, the functional characterization of neurocan had shown the ability to bind to neurons and to inhibit the homophilic adhesion of NgCAM and NCAM (59). Thus, based on the findings presented here, it is reasonable to assume that the observed inhibition was caused by steric hindrance due to the close proximity of the two domains. In this context, it is interesting to note that the well known heparin binding site in NCAM has been mapped to the first and second Ig-like domains of the molecule (44, 60-62). Several studies have shown that also NCAM can bind with high affinity to neurocan (37, 45, 59) and that the binding activity was reduced after enzymatic removal of glycosaminoglycan chains (37). As shown here, this is similar to the interaction of L1 with neurocan, which also seems to be sensitive to enzymatic removal of glycosaminoglycan chains. Although the neurocan binding site in NCAM is not known, it is tempting to speculate that it may coincide with the heparin binding site. During brain development, the interaction of neurocan or phosphacan with neural CAMs plays a major role in modulating cell adhesion as well as axonal growth and guidance (38, 63, 64). The localization of a neurocan binding site at the tip of the molecule as shown here for L1 would be well suited for a role as a sensory molecule in the path-finding activities of growing axons or migrating cells during brain development.

	ACKNOWLEDGEMENTS

We thank Dr. Volker Schirrmacher for support and Dr. Israel Vlodovsky for a gift of laminarin sulfate. We are also thankful to Dr. Guni Kadmon for many discussions, Dr. Reinhard Schwarz-Albiez for critically reading the manuscript, and Dr. Hideo Yagita for monoclonal antibodies.

	FOOTNOTES

^* This work was supported by a grant from the German-Israeli Cooperation in Cancer Research (to P. A.).The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

^§ To whom correspondence should be addressed: Tumor Immunology Programme, G0100, German Cancer Research Center, Im Neuenheimer Feld 280, D-69120 Heidelberg, Germany. Tel.: 06221-423714; Fax: 06221-423702; E-mail: P.Altevogt@dkfz-heidelberg.de.

² Beer, S., Oleszewski, M., Gutwein, P., Geiger, C., and Altevogt, P. (1999) J. Cell Sci. 112, in press.

	ABBREVIATIONS

The abbreviations used are: CAM, cell adhesion molecule; NCAM, neural CAM; GAG, glucosaminoglycan; mAb, monoclonal antibody; PBS, phosphate-buffered saline (lacking Ca²⁺ and Mg²⁺); PCR, polymerase chain reaction; PAGE, polyacrylamide gel electrophoresis.

	REFERENCES

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES