Integrity of the homophilic binding site is required for ...

Integrity of the homophilic binding site is required for the preferential localization of NCAM in intercellular contacts Biochem. Cell Biol. Downloaded from www.nrcresearchpress.com by Renmin University of China on 06/03/13 For personal use only.

Martin Sandig, Yong Rao, Vitauts I. Kalnins, and ChimHung Siu

Abstract: The neural cell adhesion molecule NCAM is a member of the immunoglobulin (Ig) superfamily. NCAM can undergo homophilic binding and heterophilic interactions with cell surface components and is often concentrated at sites of intercellular contact. To investigate the molecular basis of this biased surface distribution, we examined L cell transfectants expressing wildtype or mutant forms of chick NCAM-140 by laser scanning confocal microscopy. Mutant NCAMs that lacked Ig-like domains 1 , 2 , 4 , or 5 were preferentially localized in contact regions. However, the relative concentration of these mutant NCAMs in contact sites was substantially reduced compared with wild-type NCAM. In contrast, NCAM redistribution to intercellular contacts was abolished in cells expressing mutant NCAMs that either lacked Ig-like domain 3 or contained mutations in the homophilic binding site in this domain. In heterotypic contacts between PC12 cells and L cell transfectants, colocalization of rat NCAM and chick NCAM was again dependent on the integrity of the homophilic binding site of the NCAM expressed on L cells. These results provide evidence that homophilic binding is the main mechanism by which NCAM becomes redistributed to intercellular contacts. They also implicate a role for other Ig-like domains in the accumulation of NCAM at cell-cell contacts. Key words: cell-cell adhesion, adhesion molecule, NCAM, homophilic binding, surface distribution.

R6sumB :La molkule d'adhtrence des cellules neurales (NCAM) est un membre de la superfamille des immunoglobulines (Ig).

Il peut y avoir une liaison homophile et des interactions httkrophiles entre la NCAM et des constituants de la membrane cellulaire et la NCAM est souvent concentrke aux sites de jonction intercellulaire. Afin d'ktudier le mkanisme molkculaire de cette distribution superficielle asymktrique, des cellules L transfecttes exprimant la forme naturelle ou une forme mutante de la NCAM-140 depoulet ont ttk observkes parmicroscopie confocale balayage au laser. Les NCAM mutantes ne posskdant pas les domaines 1 , 2 , 4 ou 5 semblables aux Ig sont surtout localistes dans les zones de jonction. Cependant, la concentration relative de ces NCAM mutantes par rapport k la NCAM naturelle est passablement rkduite dans les zones de jonction. Par contre, la NCAM n'est plus redistribuke aux jonctions intercellulaires dans les cellules qui expriment les NCAM mutantes ne posskdant pas le domaine 3 semblable aux Ig ou ayant des mutations dans le site de liaison homophile de ce domaine. Dans les jonctions hktkotypiques entre les cellules PC1 2 et les cellules L transfectkes, la colocalisation de la NCAM de rat et de la NCAM de poulet est Cgalement fonction de l'intkgritt du site de liaison homophile de la NCAM exprimke h la surface des cellules L. Ces rksultats dtmontrent que la NCAM est redistribuke dans les jonctions intercellulaires principalement grsce au mkcanisme de liaison homophile. Ils indiquent tgalement que d'autres domaines semblables aux Ig jouent un r6le dans l'accumulation de la NCAM aux jonctions intercellulaires. Mots clks : aadhrence intercellulaire, molkcule d'adhkrence, NCAM, liaison homophile, distribution superficielle. [Traduit par la rkdaction]

Received October 10, 1995. Revised January 3 1, 1996. Accepted February 8, 1996.

Abbreviations:Ig, immunoglobulin; NCAM, neural cell adhesion molecule; HBSS, Hank's balanced salt solution; BSA, bovine serum albumin; FCS, fetal calf serum; PBS, phosphate-buffered saline; PCR, polymerase chain reaction; LSCM, laser scanning confocal microscopy; ELISA, enzyme-linked immonosorbant assay; mAb, monoclonal antibody. M. Sandig. Banting and Best Department of Medical Research, University of Toronto, Toronto, Ont., Canada. Y. ~ a oand ' C.-H. ~ i u . Banting ' and Best Department of Medical Research and Department of Biochemistry, University of Toronto, Toronto, Ont., Canada. V.I. Kalnins. Department of Anatomy and Cell Biology, University of Toronto, Toronto, Ont.. Canada. Present address: Howard Hughes Medical Institute, UCLA, Los Angeles, CA 90024, U.S.A. Author to whom all correspondence should be addressed: Charles H. Best Institute, University of Toronto, 112 College Street, Toronto, ON MSG 1L6, Canada. Biochem. Cell Biol. 74: 373-381 (1996). Printed in Canada 1 Imprim6 au Canada

Biochem. Cell Biol. Vol. 74, 1996

Biochem. Cell Biol. Downloaded from www.nrcresearchpress.com by Renmin University of China on 06/03/13 For personal use only.

Introduction The development and maintenance of tissue architecture depends largely on the interactions of cell adhesion molecules with the substratum and with one another (Takeichi 1990; Edelman and Crossin 1991; Reichardt and Tomaselli 1991; Doherty and Walsh 1992). The neural cell adhesion molecule NCAM is one of the first characterized adhesion molecules and it is a member of the Ig superfamily of adhesion receptors (Williams and Barclay 1988). Several alternatively spliced NCAM isoforms have been identified (Cunningham et al. 1987; Walsh and Doherty 1991), and they are expressed in a variety of embryonic and adult tissues, mediating a wide range of neuronal and non-neuronal cell-cell interactions (Edelman and Crossin 1991).In addition to participating in the formation of intercellular contacts, NCAM-mediated cell adhesion has been shown to generate signals that lead to neurite outgrowth from PC12 cells and several primary neurons (Doherty and Walsh 1992; Sandig et al. 1994; Williams et al. 1994). Intercellular contact formation mediated by adhesion molecules is a complex event. Different functional domains of NCAM and their corresponding ligands appear to be involved in NCAM function. Extracellularly, cell surface NCAM can bind heparan sulfate and a heparin binding site has been identified in Ig-like domain 2 (Cole et al. 1985; Cole and Akeson 1989;Reyes et al. 1990).Heparin binding appears to contribute significantly to cell-cell contact formation (Cole et al. 1986; Murray and Jensen 1992). Furthermore, NCAM is capable of undergoing cis-interactionswith the cell adhesion molecule L1 (Thor et al. 1986; Kadmon et al. 1990). This interaction apparently involves a C-type lectin consensus sequence located in Ig-like domain 4 and is believed to augment cell adhesion dependent neurite outgrowth (Horstkorte et al. 1993). NCAM can also undergo homophilic binding with another NCAM in a ca2+-independent manner (Rutishauser et al. 1982; Hoffman and Edelman 1983; Edelman et al. 1987; Pizzey et al. 1989). We have recently identified and characterized a homophilic binding site located in the third Ig-like domain of chick NCAM-140 (Rao et al. 1992, 1993). This binding site consists of 10 amino acids between positions 243 and 252 with the sequence KYSFNYDGSE. This region is capable of undergoing isologous interactions with the same sequence on an apposing molecule (Rao et al. 1994). NCAM is concentrated at sites of intercellular contact in a variety of cell types (Pollerberg et al. 1987; Bloch 1992). However, it is not clear how NCAM becomes localized and enriched at intercellular contact regions after reaching the plasma membrane. Since NCAM contains multiple functional domains and is capable of interacting with different ligands, we performed experiments to investigate the role of the homophilic binding site in the preferential localization of NCAM in cellkell contact regions. The contribution of other structural domains that are involved in heterophilic interactions was also evaluated using NCAM domain deletions. Laser scanning confocal microscopy (LSCM) was used to examine the surface distribution of NCAM in PC12 cells and in L cell transfectants expressing different mutant forms of NCAM. We demonstrate that the concentration of NCAM in intercellular contact regions is mainly dependent on the functional integrity of the homophilic binding site in Ig-like domain 3. Other Iglike domains, however, may also contribute to this process.

Fig. 1. Domain deletion mutations and site-specific mutations created in chick NCAM-140. The five Ig-like domains are numbered from the amino-terminus and are shown as open boxes. The fibronectin type III homology domains are shown as stippled boxes, the transmembrane domains are shown as solid bars, and the cytoplasmic domains are shown as hatched boxes. The small numbers indicate the positions of amino acids flanking each deletion according to the numbering system of Cunningham et al. (1987). Mutations in the decapeptide sequence of the homophilic binding site in Ig-like domain 3 are underlined and in boldface type.

A1 NCAM

2131415H.

Materials and methods Construction of mutant forms of NCAM and transfection of cells Domain deletion mutations and point mutations in the homophilic binding site are shown in Fig. 1. They were generated in the NCAM cDNA encoding chick NCAM-140 by PCR methods as previously described (Rao et al. 1992, 1993). All mutant NCAM constructs were verified by DNA sequencing and then cloned into the expression vector pEC, which used the SV-40 early promoter to drive NCAM expression (Edelman et al. 1987). Standard recombinant DNA methods were followed in the construction of expression vectors (Sambrook et al. 1989). Mouse L (tk-) cells were stably transfected with 20 kg NCAM plasmid DNA, using the calcium phosphate precipitation method. Two days after transfection, cells expressing NCAM were selected with a polyclonal antiNCAM antibody using the cell panning method (Wysocki and Sato 1978) and then subcloned in 96-well plates by limited dilution. The level of expression of mutant forms of NCAM was determined by gel electrophoresis, ELISA, and immunofluorescence microscopy as described previously (Rao et al. 1992; Sandig et al. 1994). L cells transfected with plasmid DNA without the NCAM cDNA insert were used as control cells.

Sandig et al.

375


Fig. 2. Redistribution of wild-type NCAM to contact regions in L cell transfectants. L cells expressing wild-type chick NCAM140 were labeled with rabbit anti-NCAM IgG for LSCM. Both X N (a, d) and XIZ (b, e) images of a single cell (a, b) and cell pair (d, e) are shown. The lines in a and d indicate the location of the X/Zscans in b and e. The pixel profiles (c,n are taken along the lines in b and e, respectively. In a and b, bar represents 5 pm; in d and e, bar represents 10 pm.

Cell cultures Mouse L cells were routinely maintained in 60-mm culture dishes in a-MEM medium supplemented with 10% fetal calf serum (FCS) in a humidified atmosphere containing 5% CO,. PC12 cells were cultured in dishes coated with 0.02% (wlv) poly-L-lysine. They were maintained in RPMI 1640 medium, containing L-glutamate, 5% donor horse serum, and 5% FCS. PC12 cells were also cultured together with L cells by adding PC12 cells to a subconfluent layer of L cells on cover slips. Co-cultures were kept for 3 h in complete RPMI 1640 medium before fixation for imrnunofluorescence labeling. Immunofluorescence labeling Cultures were fixed for 5 min in absolute methanol at -20°C and rehydrated during three 5-min washes with PBS. Alternatively, cells were fixed for 15 min with 3% (wlv) paraformaldehyde in PBS at room temperature. All subsequent procedures were carried out at this temperature. Fc receptors were blocked by treating the cells for 10 min with 20% goat serum in PBS. Cells were incubated first for 1 h with rabbit

IgG directed against chick NCAM (Rao et al. 1992) at a concentration of 10 p g / d and then with biotinylated goat antirabbit secondary antibody at 1:1000 dilution for 1 h. In some experiments, cells were first incubated for 1 h with 2 pg1mL of mAb 5B8 directed against rat NCAM and then with biotinylated horse anti-mouse secondary antibody at 1:1000 dilution for 1 h. All antibodies were diluted in PBS containing 1% (w/v) BSA. This was followed by incubation of the cells for 1 h with Texas Red conjugated streptavidin (1: 1000 dilution in PBS). Cells were washed three times with PBS (5 rnin each) between incubation periods. Cover slips were mounted on microscope slides in 80% (vlv) glycerol in PBS containing 0.1 % (wlv) p-phenelenediarnine to retard photobleaching. Strips cut from plastic cover slips were used as spacers between slide and cover slip to prevent contact of cells with the microscope slide. The preparations were sealed with nail enamel.

Laser scanning confocal microscopy Cells were observed and analyzed with a Nikon Optiphot I1

376

Biochem. Cell Biol. Vol. 74, 1996


Fig. 3. Effects of mutations on NCAM redistribution to intercellular contact sites in L cell transfectants. L cells expressing domain deletion mutants of NCAM were labeled with polyclonal anti-NCAM IgG and analyzed by LSCM to obatin the XIZ images: wild-type NCAM (a), AINCAM (b), A2NCAM (c), A3NCAM (4,A4NCAM (e), and ASNCAM (f). XIZ images of L cells expressing point mutations of NCAM, NCAM(Y,F + A), and NCAM(S,N + A) are shown in g and h, respectively. Bar represents 10 pm.

microscope, equipped with epifluorescenceillumination and a BioRad MRC 600 confocal laser scanning system. Cell pairs with comparable levels of NCAM expression and similar shape were selected for further analysis. Images of contacting cells were recorded on optical disks and then analyzed using the BioRad COMOS software package or NIH Image I. A minimum of three X/Z scans were made at different locations through the contact region between each cell pair. Pixel profiles were obtained from two to three line scans drawn -2 pm apart through the middle of each X/Z image, resulting in at least six pixel profiles for each pair of cells. The fluorescence peak heights of the NCAM label corresponding to the free cell surfaces of either cell in the pair were designated A and C, and the fluorescence intensity corresponding to the contact surfaces between the cells was designated B. The intensity ratio of NCAM label between contact region and free cell surfaces was defined as BI(A + C). The mean ratio values were used as an indicator of the increase in NCAM enrichment in contact regions in comparison with free cell surfaces. Where applicable, data were analyzed using the Student's t test for unpaired samples.

Results NCAM enrichment in cell-cell contacts of L cell transfectants To investigate the mechanism involved in the preferential association of NCAM with intercellular contacts, we first examined the surface distribution of NCAM in stably transfected L cells that expressed the wild-type NCAM-140. Generally, 70-80% of the transfectants expressed NCAM at similar levels. L cell transfectants expressing wild-type or mutant NCAMs attached equally well to the poly-L-lysine substrate. The expression of wild-type NCAM in transfectants resulted in a two- to three-fold increase in cell-pair formation in comparison with control L cells. X/Y and X/Z images obtained by LSCM showed that NCAM exhibited a relatively even distribution on the cell surface in single stationary cells (Figs. 2u and 2b).Labeled cytoplasmic vesicles were observed primarily in the perinuclear area, characteristic of Golgi vesicles, especially in cells that expressed high levels of NCAM. In cells that formed intercellular contacts, NCAM was enriched in the contact region (Figs. 2d and 2e). Quantitative


Sandig et al. Fig. 4. Quantitative analysis of NCAM enrichment in intercellular contacts. The mean pixel intensity ratios between contact areas and free cell surfaces of L cell pairs expressing different mutant forms of NCAM were calculated as described in the Materials and methods section. A minimum of 6 pixel profiles were obtained for each cell pair, and 5 to 10 cell pairs were analyzed for each case. Data represent the mean ? SD. The asterisks indicate that the intensity ratios of both A3NCAM and NCAM(Y,F + A) contacts were significantly different from that of wild-type NCAM (wtNCAM) contacts, with p < 0.001. They were also significantly different from that of A2NCAM and A4NCAM, withp < 0.005. The intensity ratios of AlNCAM, A2NCAM, A4NCAM, and A5NCAM were significantly different from that of the wild-type NCAM or NCAM(S,N += A), withp < 0.001.

I

NCAM(Y ,F+A)

0

0.5

1

1.5

2

25

3

3.>

4

4.3

Intensity Ratio

analysis of pixel scans of the X / Z images showed a three-fold increase in NCAM staining intensity between the contact region and the non-contact regions of the cell surface (Figs. 2d and 2f). In contrast, intercellular contacts between NCAMexpressing and non-expressing cells did not show an enhancement in NCAM staining (data not shown).

Preferential localization of NCAM in contact regions is dependent on NCAM homophilic binding Because the extracellular Ig-like domains of NCAM have been implicated in heterophilic interactions with different ligands as well as homophilic binding to apposing NCAM molecules, experiments were carried out to determine which extracellular structural domain(s) had a role in the redistribution of NCAM. L cell transfectants expressing mutant NCAMs with different domain deletions (see Fig. 1) were examined by LSCM. X/Z images showed that the patterns of NCAM distribution for AlNCAM, A2NCAM, A4NCAM, and A5NCAM were similar to that of wild-type NCAM (Fig. 3). These mutant NCAMs were also preferentially localized in the intercellular contact zones between cells, though to a lesser extent than wild-type NCAM. In contrast, in cells expressing A3NCAM, NCAM did not redistribute to intercellular contacts (Fig. 3 4 . Quantitative analysis based on pixel scans of W Z images (see Figs. 2c and 2j) was used to estimate the level of NCAM enrichment in intercellular contact regions for all mutant forms of NCAM. The pixel intensity ratios between contact regions and free cell surfaces were determined for cell pairs of each mutant NCAM (Fig. 4). Whereas the wild-type NCAM

showed a three-fold increase in NCAM concentration in the contact zones, A3NCAM did not show significant increase in NCAM staining intensity in similar regions. The intensity ratio actually dropped to 0.6. Therefore, the intensity ratio for wild-type NCAM was about five-fold higher than that of A3NCAM. It was of interest to note that mutant NCAMs containing other domain deletions also exhibited a significant decrease in intensity ratios compared with wild-type NCAM. In comparison with wild-type NCAM, AlNCAM, A4NCAM, and A5NCAM showed a 40-50% reduction in NCAM concentration in the contact regions. Deletion of Ig-like domain 2 appeared to have a more severe effect and a 60% reduction was observed. Nevertheless, concentrations of these four mutant NCAMs at contact regions were still two to three times higher than that of A3NCAM, indicating a significant degree of biased distribution. When subjected to statistical analysis using the Student's t test, the intensity ratio of A3NCAM was significantly different from that of A2NCAM ( p < 0.001). The above results indicated that NCAM redistribution to cell-cell contact regions were affected by all five Ig-domain deletions. However, deletion of Ig-like domain 3 was clearly most deleterious and resulted in the total loss of NCAM concentration in the contact regions. Since the homophilic binding site resides within Ig-like domain 3, these results suggest that NCAM-NCAM homophilic binding plays a crucial role in the preferential localization of NCAM to contact regions. However, it was also possible that deletion of the entire Ig-like domain 3 might have an adverse effect on the conformation of A3NCAM, somehow interfering with the lateral mobility and (or) accumulation of NCAM in contact regions. To distinguish between these possibilities, L cells expressing mutant NCAMs with substitutions in the homophilic binding sequence, rather than deletions of the whole domain, were examined (see Fig. 1). We have previously shown that NCAM(Y,F -+ A), but not NCAM(S,N -+ A), has lost its homophilic binding activity (Rao et al. 1994; Sandig et al. 1994). X/Z images showed that NCAM(Y,F -+ A) failed to accumulate in intercellular contact regions (Fig. 3g), whereas NCAM(S,N -+ A) redistributed to contact regions like the wild-type NCAM (Fig. 3h). The mean intensity ratio of NCAM(Y,F -+ A) was not significantly different from that of A3NCAM ( p > 0.1) but was significantly different from that of A2NCAM ( p < 0.005) (Fig. 4). These results indicate that the enrichment of NCAM in intercellular contact regions is dependent on the integrity of the homophilic binding site.

Redistribution of NCAM in PC12 cells and in heterotypic contacts between PC12 cells and L cell transfectants To determine the role of the chick NCAM homophilic binding site in interspecies NCAM-NCAM interactions, we studied the redistribution of rat NCAM in PC12 cells on binding with L cell transfectants. First, the distribution of NCAM in single PC12 cells was compared with that in PC12 cell pairs. Similar to the results obtained with L cell transfectants, NCAM was fairly evenly distributed on the entire cell surface of single PC12 cells (Figs. 5a and 5b). In cells that had formed intercellular contacts, however, NCAM was greatly enriched in the contact region (Figs. 5c and 5 4 . Heterotypic interactions between PC12 cells and L cells expressing different mutant NCAMs were then investigated. In co-cultures of PC12 cells

Biochern. Cell Biol. Vol. 74, 1996

378


Fig. 5. Redistribution of NCAM to contact regions between PC12 cells. PC12 cells were labeled with polyclonal anti-NCAM IgG for LSCM. Both X/Y (a, c) and X/Z (b,d) images of a single cell (a, b) and a cell pair (c, 6 ) are shown. The lines in a and c indicate the location of the X/Zscans in b and d. Bars represent 5 p,m.

and L cell transfectants, PC12 cells could be distinguished from the elongated L cells since they were round, smaller, and labeled with an anti-Ll antibody (data not shown). When PC12 cells were added to L cells expressing wild-type chick NCAM, prominent labeling of NCAM was observed within the intercellular contact regions between both cell types (Figs. 6a and 6c). Co-cultures were also labeled with mAb 5B8, which is specific to rat NCAM, and results confirmed an enrichment of rat NCAM in homotypic contacts between PC12 cells and in heterotypic contacts between PC12 cells and L cell transfectants (Fig. 6b). When the pixel intensity ratios of XIZ images were determined (Fig. 7), the mean intensity ratio for heterotypic cell contacts was comparable with that obtained for homotypic contacts between PC12 cell pairs or L cell pairs expressing wild-type NCAM (see Fig. 4). The redistribution of NCAM to contact regions was abolished when PC12 cells were co-cultured with L cell transfectants expressing A3NCAM (Fig. 6d).The mean intensity ratio in PC12 and A3NCAM - L cell pairs was -0.7 (Fig. 7), close to the value obtained for L cell pairs expressing A3NCAM (Fig. 4). Similarly, it was greatly reduced in contact regions between PC12 cells and L cells expressing NCAM(Y,F + A) (Fig. 6e). A redistribution of rat NCAM did not occur in contacts between PC12 cells and control L cells (data not shown). On the other hand, a high concentration of NCAM was observed in contacts

between PC12 cells and L cell transfectants expressing NCAM(S,N + A) (Fig. 68. These results indicate that mobilization of rat NCAM to the heterotypic cell contact sites is dependent on the hemophilic binding site of chick NCAM.

Discussion In this study, we have demonstrated that NCAM is enriched at the intercellular contact sites, consistent with its function as a cell-cell adhesion molecule. We have also assessed the role of the extracellular Ig-like domains of NCAM in its preferential localization to cell-cell contact regions. The localization of NCAM to cell-cell contact sites appears to be a specific phenomenon. Non-specific trapping can be ruled out because several mutant NCAMs fail to redistribute to these sites. Quantitative analysis of LSCM images showed a three-fold increase in NCAM concentration in intercellular contacts between L cell transfectants (Figs. 2 4 ) . A preferential distribution of NCAM to intercellular contact regions has been reported for several NCAM-expressing cell lines and transfected cells (Bloch 1992; Woo et al. 1993). Measurements of the amount of NCAM between these cells showed an average 4.5-fold increase in NCAM concentration (Bloch 1992), comparable with what we have obtained for the L cell transfectants. In epithelial cells, contact formation is largely mediated by cadherins, which also accumulate at cell-cell contact sites

Sandig et al.

379


Fig. 6. Enrichment of rat and chick NCAMs in contacts between PC12 cells and L cell transfectants. PC12 cells (P) were co-cultured with L cells (L) expressing wild-type NCAM (a-c), A3NCAM (d),NCAM(Y,F + A) (e), or NCAM(S,N + A) (f). Cells were labeled with anti-chick NCAM IgG (a, c-j) or mAB 5B8 anti-rat NCAM (b) and analyzed by LSCM. a and b are X/Y images, and c to f are XIZ images. Bars represent 5 p,m.

Fig. 7. Quantitative analysis of NCAM enrichment in intercellular contacts between PC12 cells and L cell transfectants. The mean pixel intensity ratios between contact areas and free cell surfaces of pairs of PC12 cell and L cell expressing different mutant forms of NCAM were calculated as described in the Materials and methods section. Data represent the mean + SD. The asterisks indicate that the intensity ratios of PC12 + A3NCAM-expressing L cell and PC12 + NCAM(Y,F + A) expressing L cell contacts were significantly different from that of PC12 + wildtype NCAM contacts, with p < 0.001. PC12 Pairs

PC12+wtNCAM-L

Intensity Ratio

(Hirano et al. 1987). Cadherins have been shown to become anchored to the cytoskeleton upon stabilization of the contact regions (McNeill et al. 1993). However, associations of NCAM with cytoskeletal proteins in these NCAM-rich contact regions could not be demonstrated (Bloch 1992; M. Sandig and C.-H. Siu, unpublished results), suggesting that NCAM redistribution to the contact regions may rely primarily on the extracellular domain of NCAM. We have demonstrated in this report that the concentration of NCAM in intercellular contacts has an absolute dependence on its homophilic binding site. The homophilic binding site of NCAM consists of a 10 amino acid sequence (243-KYSFNYDGSE-252) that lies within the third Ig-like domain (Rao et al. 1992). Mutant NCAMs with either the homophilic binding site deleted (A3NCAM) or mutated (NCAM(Y,F + A)), so as to abolish the homophilic binding activity, fail to redistribute to the contact regions between cell pairs (Fig. 3). Although the fluorescence intensity ratio was expected to be about 1 if there was no preferential localization of NCAM in contact regions, both A3NCAM and NCAM(Y,F + A) had values between 0.5 and 0.7. It is possible that access of antibodies to molecules on


Biochem. Cell Biol. Vol. 74, 1996 membranes in close apposition was more difficult than to those on the free surfaces. This would result in an underestimation of the fluorescence intensity in cell contact regions, thus a lower value for the fluorescence intensity ratio. Our results suggest that Ig-like domains, other than domain 3, may play a role in the accumulation of NCAM to cell-cell contact sites, since deletion of any one of these Ig-like domains leads to a 35 to 60% decrease in the accumulation of NCAM in contact regions. However, unlike Ig-like domain 3, deletion of any one of the other Ig-like domains was not sufficient to abolish NCAM redistribution to contact regions, suggesting that they play only a secondary role in this event. It is conceivable that deletion of an Ig-like domain may induce conformational changes in NCAM, resulting in long-range effects on the homophilic binding site. Alternatively, NCAM homophilic interactions may involve secondary binding sites along the length of the extracellular segment of NCAM. A possible model is that initial interactions centered at Ig-like domain 3 of two apposing molecules may elicit subsequent interactions between Ig-like domain 1 with Ig-like domain 5, Ig-like domain 2 with Ig-like domain 4, and vice versa. These binding sites, if present, must be relatively weak, since they are unable to promote NCAM-NCAM binding in the absence of the homophilic binding site in Ig-like domain 3 (Rao et al. 1992, 1993). However, together they may contribute to a higher affinity in NCAM-NCAM binding. Therefore, deletion in any one of the Ig-like domains may destabilize the binding between two apposing NCAM molecules. Futhermore, heterophilic interactions mediated by Ig-like domains may have a stabilizingeffect on the NCAM complex. The interaction of NCAM with heparan sulfate via Ig-like domain 2 is known to contribute to NCAM-NCAM binding (Cole et al. 1985, 1986). This may account for the more severe effects of A2NCAM. Chick NCAM is also capable of eliciting an enrichment of rat NCAM in heterotypic contacts between L cell transfectants and PC12 cells (Figs. 5 and 6), indicating that NCAM originating from two different vertebrate species can interact with one another, a finding that is consistent with several previous reports (Hall and Rutishauser 1985; Rao et al. 1994). More importantly, the interaction is dependent on the presence of a functional homophilic binding site in chick NCAM. L cells expressing a mutant NCAM that has lost its homophilic binding activity fail to elicit a redistribution of the rat NCAM on PC12 cells (Figs. 5 and 6). These results thus demonstrate the key role of the chick homophilic binding site in interspecies NCAM-NCAM interactions. Since L cell transfectants expressing an inactive NCAM are capable of forming cell pairs, albeit to a much lesser extent than calls expressing wild-type NCAM, the formation of cell-cell contacts does not seem to depend solely on NCAM. It is conceivable that many other cell surface components are capable of undergoing weak interactions with ligands present on adjacent cells. The mobilization of a higher concentration of NCAM to intercellular contact sites probably augments the adhesiveness between the surfaces of adjacent cells, leading to the formation of larger cell aggregates (Edelman et al. 1987; Pizzey et al. 1989; Rao et al. 1992). The recruitment of NCAM into contact regions appears to be drawn from the existing pool in the plasma membrane. Since all three major NCAM isoforms can be con-

centrated in cell-cell contact regions (Edelman et al. 1987; Pizzey et al. 1989), the presence of a cytoplasmic domain is not a requirement for NCAM clustering at these sites. The clustering of NCAM in contact regions may result from the diffusion of molecules to an initial "nucleation" site where a few NCAM molecules are undergoing homophilic interactions. Our finding that the disruption of the homophilic binding site by mutagenesis was able to abolish NCAM clustering at contact regions is consistent with this view and with the mutual capping - co-capping mechanism proposed by Singer (1992). How the lattice of NCAM molecules is organized in these contact regions is not known. It would be of interest to determine whether cisinteractions between NCAM molecules on the same membrane exist in contact regions. Such interactions should increase the avidity of NCAM-NCAM trans-interactions between apposing membrane surfaces and further stabilize the contact region.

Acknowledgements The authors thank Dr. Y. Zhou, W. Arnold, and D. Vidgen for assistance and advice on confocal microscopy. The authors also thank members of their laboratory for discussion and comments on the manuscript. The mAb 5B8 was obtained from the Developmental Studies Hybridoma Bank maintained at University of Iowa, Iowa City, IA 52242, U.S.A. This work was supported by an operating grant from the Medical Research Council of Canada. Y. Rao was supported by a studentship from the Medical Research Council of Canada.

References Bloch, R.J. 1992. Clusters of neural cell adhesion molecule at sites of cell-cell contacts. J. Cell Biol. 116: 449463. Cole, G.J., and Akeson, R. 1989. Identification of a heparin binding domain of the neural cell adhesion molecule N-CAM using synthetic peptides. Neuron, 2: 1157-1 165. Cole, G.J., Schubert, D., and Glaser, L. 1985. Cell-substratum adhesion in chick neural retina depends upon protein-heparan sulfate interaction. J. Cell Biol. 100: 1192-1199. Cole, G.J., Loewy, A., and Glaser, L. 1986. Neuronal cell-cell adhesion depends on interactions of N-CAM with heparin-like molecules. Nature (London), 320: 445447. Cunningham, B.A., Hemperly, J.J., Murray, B.A., Prediger, E.A., Brackenbury, R., and Edelman, G.M. 1987. Neural cell adhesion molecule: structure, immunoglobulin-like domains, cell surface modulation, and alternative RNA splicing. Science (Washington, D.C.), 236: 799-806. Doherty, P., and Walsh, ES. 1992. CAMS, second messengers, and axonal growth. Curr. Opin. Neurobiol. 2: 595-601. Edelman, G.M., and Crossin, K.L. 1991. Cell adhesion molecules: implications for a molecular histology. Annu. Rev. Biochem. 60: 155-190. Edelman, G.M., Murray, B.A., Mege, R.M., Cunningham, B.A., and Gallin, W.J. 1987. Cellular expression of liver and neural cell adhesion molecules after transfection with their cDNAs results in specific cell-cell binding. Proc. Natl. Acad. Sci. U.S.A. 84: 8502-8506. Hall, A.K., and Rutishauser, U. 1985. Phylogeny of a neural cell adhesion molecule. Dev. Biol. 110: 39-46. Hirano, S., Nose, A., Hatta, K., Kawakami, A., and Takeichi, M. 1987. Calcium-dependent cell-cell adhesion molecules (cad-


Sandig et al. herins): subclass specificities and possible involvement of actin bundles. J. Cell Biol. 105: 2501-2510. Hoffman, S., and Edelman, G.M. 1983. Kinetics of homophilic binding by embryonic and adult forms of the neural cell adhesion molecule. Proc. Natl. Acad. Sci. U.S.A. 80: 5762-5766. Horstkorte, R., Schachner, M., Magyar, J.P., Vorherr, T., and Schmitz, M. 1993. The fourth immunoglobulin-like domain of NCAM contains a carbohydrate recognition domain for oligomannosidic glycans implicated in association with L1 in neurite outgrowth. J. Cell Biol. 121: 1409-1421. Kadmon, G.A., Kowitz, A., Altevoigt, P., and Schachner, M. 1990. The neural cell adhesion molecule N-CAM enhances Ll-dependent cell-cell interactions. J. Cell Biol. 110: 193-208. McNeill, H., Ryan, T.A., Smith, S.J., and Nelson, W.J. 1993. Spatial and temporal dissection of immediate and early events following cadherin-mediated epithelial cell adhesion. J. Cell Biol. 120: 1217-1226. Murray, B.A., and Jensen, J.J. 1992. Evidence for heterophilic adhesion of embryonic retinal cells and neuroblastoma cells to substratum adsorbed NCAM. J. Cell Biol. 117: 1311-1320. Pizzey, J.A., Rowett, L.H., Barton, C.H., Dickson, G., and Walsh, F.S. 1989. Intercellular adhesion mediated by human muscle cell adhesion molecule: effects of alternative exon use. J. Cell Biol. 109: 3465-3476. Pollerberg, G.E., Burridge, K., Krebs, K.E., Goodman, S.R., and Schachner, M. 1987. The 180-kD component of the neural cell adhesion molecule NCAM is involved in cell-cell contact and cytoskeleton-membrane interactions. Cell Tiss. Res. 250: 227233. Rao, Y., Wu, X.-F., Gariepy, J., Rutishauser, U., and Siu, C.-H. 1992. Identification of a peptide sequence involved in homophilic binding in the neural cell adhesion molecule NCAM. J. Cell Biol. 118: 937-949. Rao, Y., Wu, X.-F., Yip, P., Gariepy, J., and Siu, C.-H. 1993. Structural characterization of a homophilic binding site in the neural cell adhesion molecule NCAM. J. Biol. Chem. 268: 20 630 20 638. Rao, Y., Zhao, X., and Siu, C.-H. 1994. Mechanism of homophilic binding mediated by neural cell adhesion molecule NCAM: evidence for isologous interaction. J. Biol. Chem. 269: 27 540 27 548.

Reichardt, L.F., and Tomaselli, K.J. 1991. Extracellular matrix molecules and their receptors: function in neural development. Annu. Rev. Neurosci. 14: 531-570. Reyes, A.A., Akeson, R., Brezina, L., and Cole, G.J. 1990. Structural requirements for neural cell adhesion molecule-heparin interaction. Cell Regul. 1: 567-576. Rutishauser, U., Hoffman, S., and Edelman, G.M. 1982. Binding properties of a cell adhesion molecule from neural tissue. Proc. Natl. Acad. Sci. U.S.A. 79: 685-689. Sambrook, J., Fritsch, E.F., and Maniatis, T. 1989. Molecular cloning: a laboratory manual. 2nd ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y. Sandig, M., Rao, Y., and Siu, C.-H. 1994. The homophilic binding site of the neural cell adhesion molecule NCAM is directly involved in promoting neurite outgrowth from cultured neural retinal cells. J. Biol. Chem. 269: 14 841 - 14 848. Singer, S.J. 1992. Intercellular communication and cell-cell adhesion. Science (Washington, D.C.), 255: 1671-1677. Takeichi, M. 1990. Cadherins: a molecular family important in selective cell-cell adhesion. Annu. Rev. Biochem. 59: 237-252. Thor, G., Pollerberg, G.E., and Schachner,M. 1986. Molecular association of two neural cell adhesion molecules within the surface membrane of cultured neuroblastoma cells. Neurosci. Lett. 66: 121-126. Walsh, F.S., and Doherty, P. 1991. NCAM gene structure and function. Semin. Neurosci. 3: 271-284. Williams, A.F., and Barclay, A.N. 1988. The immunoglobulin superfamily: domains for cell surface recognition. Annu. Rev. Immunol. 6: 381-405. Williams, E.J., Fumess, J., Walsh, F.S., and Doherty, P. 1994. Activation of the FGF receptor underlies neurite outgrowth stimulated by L1, N-CAM, and N-cadherin. Neuron, 13: 583-594. Woo, M.K., Kil, S.-H., Adarns, L.D., Nguyen, T.N., and Murray, B.A. 1993. The large cytoplasmic domain is not required for concentration of N-CAM at cell-cell contacts in transfected mouse neuroblastoma cells. Exp. Cell Res. 205: 331-336. Wysocki, L.J., and Sato, V.L. 1978. Panning for lymphocytes: a method for cell selection. Proc. Natl. Acad. Sci. U.S.A. 75: 2844-2848.