Requirements for the Ca2+-independent component in the initial intercellular adhesion of C2 myoblasts

Using a sensitive and quantitative adhesion assay, we have studied the initial stages of the intercellular adhesion of the C2 mouse myoblast line. After dissociation in low levels of trypsin in EDTA, C2 cells can rapidly reaggregate by Ca2+-independent mechanisms to form large multicellular aggregates. If cells are allowed to recover from dissociation by incubation in defined media, this adhesive system is augmented by a Ca2+-dependent mechanism with maximum recovery seen after 4 h incubation. The Ca2+-independent adhesion system is inhibited by preincubation of cell monolayers with cycloheximide before dissociation. Aggregation is also reduced after exposure to monensin, implicating a role for surface-translocated glycoproteins in this mechanism of adhesion. In coaggregation experiments using C2 myoblasts and 3T3 fibroblasts in which the Ca2+-dependent adhesion system was inactivated, no adhesive specificity between the two cell types was seen. Although synthetic peptides containing the RGD sequence are known to inhibit cell-substratum adhesion in various cell types, incubation of C2 myoblasts with the integrin-binding tetrapeptide, RGDS, greatly stimulated the Ca2+-independent aggregation of these cells while control analogs had no effect. These results show that a Ca2+- independent mechanism alone is sufficient to allow for the rapid formation of multicellular aggregates in a mouse myoblast line, and that many of the requirements and perturbants of the Ca2+-independent system of intercellular myoblast adhesion are similar to those of the Ca2+-dependent adhesion mechanisms.     

Role of calcium-dependent and calcium-independent mechanisms in the adhesion of retinal and pigment epithelium cells in the chick embryo

The mechanisms of adhesion of the retinal and pigment epithelium cells, as well of cell interaction within each of these tissues were studied during development. It was shown by means of separation of retina from pigment epithelium in different dissociation media that the adhesion of these tissues in 5-6 day old chick embryos is realized via a Ca2+-independent mechanism. The adhesion of these tissues decreases between days 7 and 16. Starting from day 16, both Ca2+-independent and Ca2+-dependent mechanisms are involved in the interaction of the retinal and pigment epithelium cells. By measuring the output of single cells into the suspension after the treatment of retina and pigment epithelium with different dissociating agents, it was shown that from the 5th day of incubation on the adhesion of pigment epithelium cells is mediated by Ca2+-dependent mechanism. In the retina three types of cells were found: interacting via Ca2+-dependent mechanism only, Ca2+-independent mechanism only, and both the mechanisms. In the course of differentiation, the numbers of the population of cells interacting only via Ca2+-dependent mechanism increase, while those of cells interacting via Ca2+-independent mechanism decrease. It is suggested that at each developmental stage those retinal cell possess Ca2+-dependent mechanism of adhesion which are closest to the definitive state.     

Use of preformed cell aggregates and layers to measure tissue-specific differences in intercellular adhesion

The binding of aggregates formed from various 7-day chick embryo tissues to cultured cell layers was analyzed 24 hr following trypsin dissociation of the tissues. The proprotion of aggregates binding is independent of the number of aggregates added, and changes with time over 60 min in a manner consistent with a first-order process. The adhesive parameter measured, the percentage of aggregates binding to cell layers per unit time, varies slightly with aggregate size but is not dependent upon the probability of collision of the aggregate with the layer. The rate of binding and the effect of modifiers of binding (temperature, inhibitors of oxidative metabolism and glutaraldehyde) are substantially different for neural retina interactions than for liver or heart interactions, suggesting that retina cells may form intercellular bonds via a mechanism distinct from that of liver or heart cells. The rate of binding between like tissue types is, with one exception, greater than between unlike types. Glutaraldehyde treatment of only one of the reactants abolishes this adhesive specificity. Aggregate binding provides a means of quantitatively assessing intercellular adhesion which has the advantage of reducing the effects of trypsinization on measurements of adhesion, and therefore lends itself to the investigation of cellular consequences of adhesion.     

Anti-CD9 monoclonal antibodies induce homotypic adhesion of pre-B cell lines by a novel mechanism

Anti-CD9 mAb are known agonists of platelet aggregation, but have not been implicated in cell-cell adhesion. We show here in an experimental system that the anti-CD9 mAb 50H.19, ALB6, and BA-2 can induce rapid, and irreversible, homotypic aggregation of the CD9-positive pre-B lymphoblastoid cell lines NALM-6 and HOON, but not of the CD9-negative B cell line Raji. The specificity of the response is indicated by the failure to effect aggregation with mAb directed to CD24, or to HLA class I Ag. The initiation of strong homotypic aggregates of lymphoid cells is a property ascribed to lymphocyte function-associated Ag-1 (LFA-1), a member of the beta 2 subfamily of leukocyte integrins. We show that CD9-induced aggregation is an active process which proceeds at 37 degrees C, but not at 4 degrees C, requires the expenditure of metabolic energy, and a functioning cytoskeleton, and is not inhibited by Arg-Gly-Asp-Ser peptide. These are properties described for LFA-1-mediated aggregation. However, because beta 2-integrins are not expressed on NALM-6 or HOON cells, they are not the mediators of CD9-induced aggregation. In contrast to LFA-1-mediated adhesion which is Mg2+ dependent, CD9-induced adhesion has an absolute requirement for Ca2+, but not Mg2+, indicating that a Ca2(+)-dependent event is sufficient for adhesion. However, Mg2+ enhances adhesion even at optimal concentrations of Ca2+, implicating an additional Mg2(+)-dependent event which requires Ca2+ to be effective. These findings suggest that CD9 Ag regulates a novel mechanism for promoting tight cell-cell adhesion which requires both Ca2+ and Mg2+ for optimal expression.     

Some effects of trypsin dissociation on the inhibition of reaggregation among embryonic chicken neural retina cells by cycloheximide

The effects of cycloheximide on the reaggregation of trypsin-dispersed, embryonic chicken, neural retina cells were investigated. The cells were used either immediately after isolation (F-cells), or after 24 hr prior culture under conditions not permitting reaggregation (V-cells). The parameters of aggregation used were the size of aggregates formed in gyrotatory shaker cultures, and the concentration of single cells in these cultures. At both 24 and 48 hr, following treatment with cycloheximide, the mean diameters of the aggregates of F-cells showed a greater reduction than the V-cells, when compared with their corresponding controls. In addition, cycloheximide resulted in a higher concentration of single cells, that is less cell/cell adhesion, than in comparable controls. A higher proportion of single cells were present in the F-cell cultures in the presence of cycloheximide than in the V-cell cultures. Thus, the degree of inhibition of cell adhesion by one inhibitor of protein synthesis differed in F-cells and V-cells. These experiments may serve to focus attention on some secondary effects of tryptic dissociation, which is an often overlooked factor in subsequent studies of reaggregating embryonic cells, particularly in those experiments involving the use of metabolic inhibitors.     

A role for the Ca2(+)-dependent adhesion molecule, N-cadherin, in myoblast interaction during myogenesis

The formation of multinucleate skeletal muscle cells (myotubes) is a Ca2(+)-dependent process involving the interaction and fusion of mononucleate muscle cells (myoblasts). Specific cell-cell adhesion precedes lipid bilayer union during myoblast fusion and has been shown to involve both Ca2(+)-independent (CI)2 and Ca2(+)-dependent (CD) mechanisms. In this paper we present evidence that CD myoblast adhesion involves a molecule similar or identical to two known CD adhesion glycoproteins, N-cadherin and A-CAM. These molecules were previously identified by other laboratories in brain and cardiac muscle, respectively, and are postulated to be the same molecule. Antibodies to N-cadherin and A-CAM immunoblotted a similar band with a molecular weight of approximately 125,000 in extracts of brain, heart, and pectoral muscle isolated from chick embryos and in extracts of muscle cells grown in vitro at Ca2+ concentrations that either promoted or inhibited myotube formation. In assays designed to measure the interaction of fusion-competent myoblasts in suspension, both polyclonal and monoclonal anti-N-cadherin antibodies inhibited CD myoblast aggregation, suggesting that N-cadherin mediates the CD aspect of myoblast adhesion. Anti-N-cadherin also had a partial inhibitory effect on myotube formation likely due to the effect on myoblast-myoblast adhesion. The results indicate that N-cadherin/A-CAM plays a role in myoblast recognition and adhesion during skeletal myogenesis.     

Modulation of Ca2(+)-dependent intercellular adhesion in bovine aortic and human umbilical vein endothelial cells by heparin-binding growth factors

Cultured endothelial cells have been shown to possess two mechanisms of intercellular adhesion: Ca2(+)-dependent and Ca2(+)-independent. We report here that growth of bovine aortic endothelial cells (BAEC) in complete medium containing purified basic fibroblast growth factor (bFGF, 6 ng/ml) results in loss of Ca2(+)-dependent intercellular adhesion. In the presence of heparin (90 micrograms/ml), this effect is reproduced upon treatment with acidic fibroblast growth factor (aFGF, 6 ng/ml) or endothelial cell growth supplement (ECGS, 100 micrograms/ml), in both human umbilical vein endothelial cells (HUVEC) and BAEC. Treatment at these doses with aFGF in the absence of heparin or with heparin alone is without significant effect. Loss of Ca2(+)-dependent adhesion following treatment of cells with heparin-binding growth factors (HBGFs) is prevented by pre-treatment of cell layers with cycloheximide. The Ca2(+)-independent adhesion mechanism is unaffected by HBGF treatment. Exposure of endothelial cells to HBGFs, moreover, prevents the eventual establishment of quiescence in growing cultures and restimulates replication in confluent cultures that have reached a final density-inhibited state. Addition of bFGF alone or aFGF + heparin at these doses results in a 4-fold increase in DNA synthesis over untreated control cultures at saturation density as reflected by thymidine index. A single addition of bFGF (6 ng/ml) to untreated quiescent confluent BAEC monolayers results in an increase in 3H-TdR incorporation reaching a peak at 22 hours with a parallel loss of Ca2(+)-dependent adhesiveness. Fluorescent staining with rhodamine-phalloidin demonstrates an altered distribution of polymerized F-actin in the bFGF-treated monolayers, marked by disruption of the dense peripheral microfilament bands retained by untreated confluent monolayers. Together, these results indicate that the mitogenic effect of HBGFs in cultured endothelial cells is associated with a "morphogenic" set of responses, perhaps dependent on breakdown of calcium-dependent cell-cell contacts.     

A kinetic study of embryonic cell adhesion

A new assay is described for measuring the kinetics of adhesion of cells to collecting aggregates. The parameter measured is the percentage of cells in suspension adhering to a large number of collecting aggregates per unit time. Using this assay on trypsinized cells it is shown that several hours of recovery are required before the maximum adhesion rate is reached. The recovery period is sensitive to cycloheximide and to low temperature (4°C). As the cells approach the maximum adhesion rate, their collection to aggregates becomes increasingly insensitive to the addition of cycloheximide; this adhesion remains sensitive to low temperature. Specificity studies show that the rate of adhesion of embryonic cells to collecting aggregates is highest between cells and aggregates of the same histotype. This rate is affected by the embryonic age of the cells. Finally, when cells of a tissue are fractionated on Ficoll gradients, it is shown that subpopulations of cells adhere to collecting aggregates at different rates.     

CELL ASSOCIATION PATTERN IN AGGREGATES CONTROLLED BY MULTIPLE CELL-CELL ADHESION MECHANISMS

We have previously assumed the presence of two mechanisms for the aggregation of Chinese hamster V79 cells, the Ca²⁺-dependent one and the Ca²⁺-independent one. In order to examine if each of these mechanisms contributed differently to the various aspects of cell aggregation, the morphology of V79 cell aggregates, pretreated so that they were provided with only one of the two adhesion mechanisms, was compared by light and electron microscopy. The adhesion among cells with only the Ca²⁺-dependent mechanism was very tight, with the formation of gap and intermediate junctions. Cells were arranged in a rod or dendric shape in aggregates. In aggregates of cells with only the Ca²⁺-independent mechanism, cells were loosely attached to each other without the formation of specialized junctions and the aggregates were of globular shape. In aggregates of cells with both mechanisms, both characteristics of the above two aggregates were found. Four clones of V79 cells, which formed colonies with different morphology when they were grown in soft agar, were isolated. It was found that such differences were due to the different activity of the Ca²⁺-independent mechanism among these clones. These results suggested that the two adhesion mechanisms play different roles in the cell arrangement in aggregates.     

Disruption of the gene encoding the cell adhesion molecule DdCAD-1 leads to aberrant cell sorting and cell-type proportioning during Dictyostelium development

The cadA gene in Dictyostelium encodes the Ca2+-dependent cell adhesion molecule DdCAD-1, which is expressed soon after the initiation of development. To investigate the biological role of DdCAD-1, the cadA gene was disrupted by homologous recombination. The cadA-null cells showed a 50% reduction in EDTA-sensitive cell adhesion. The remaining EDTA-sensitive adhesion sites were resistant to dissociation by anti-DdCAD-1 antibody, suggesting that they were distinct adhesion sites. Cells that lacked DdCAD-1 were able to complete development and form fruiting bodies. However, they displayed abnormal slug morphology and culmination was delayed by approximately 6 hours. The yield of spores was reduced by approximately 50%. The proportion of prestalk cells in cadA(-) slugs showed a 2.5-fold increase over the parental strain. When cadA(-) cells were transfected with pcotB::GFP to label prespore cells, aberrant cell-sorting patterns in slugs became apparent. When mutant prestalk cells were mixed with wild-type prespore cells, mutant prestalk cells were unable to return to the anterior position of chimeric slugs, suggesting defects in the sorting mechanism. The wild-type phenotype was restored when cadA(-) cells were transfected with a cadA-expression vector. These results indicate that, in addition to cell-cell adhesion, DdCAD-1 plays a role in cell type proportioning and pattern formation.     

Experimental manipulation of cell surface to affect cellular recognition mechanisms.

The mechanism of selective cell adhesion was studied using Chinese hamster V79 and chick embryonic neural retinal cells. Both of these cell types have been shown to have two experimentally separable mechanisms of adhesion; Ca²⁺-dependent and Ca²⁺-independent. Cells can be dispersed so that either or both of the mechanisms remain intact by use of different treatments. A method of labeling cells with FITC was devised to identify one of the two types of cells in a binary cell population. When cells with one of the two adhesion mechanisms were mixed with cells with the other mechanism, they segregated completely, forming independent aggregates, not only in the heterotypic combination of these cell types but also in the homotypic combination of each cell type. In contrast, when cells were mixed with others with the same adhesion mechanism, either Ca²⁺-dependent or -independent, they formed chimeric aggregates, even in the heterotypic cell combination. These results suggest that the specificity in each of those two mechanisms of cell adhesion plays an important role in cellular recognition processes.     

The Differential Interfacial Tension Hypothesis (DITH): a comprehensive theory for the self-rearrangement of embryonic cells and tissues

A comprehensive theory, herein named the Differential Interfacial Tension Hypothesis for the self-rearrangement of embryonic cells and tissues is presented. These rearrangements include sorting, mixing and formation of checkerboard patterns in heterotypic aggregates of embryonic cells, and total or partial engulfment, separation and dissociation of tissues. This broadly-based theory accounts for the action of all currently known cytoskeletal components and cell adhesion mechanisms. The theory is used to derive conditions for the cell and tissue rearrangements named above. Finite element-based computer simulations involving two or more cell types confirm these conditions.     

Drosophila fasciclin I is a novel homophilic adhesion molecule that along with fasciclin III can mediate cell sorting

Fasciclin I is a membrane-associated glycoprotein that is regionally expressed on a subset of fasciculating axons during neuronal development in insects; it is expressed on apposing cell surfaces, suggesting a role in specific cell adhesion. In this paper we show that Drosophila fasciclin I is a novel homophilic cell adhesion molecule. When the nonadhesive Drosophila S2 cells are transfected with the fasciclin I cDNA, they form aggregates that are blocked by antisera against fasciclin I. When cells expressing fasciclin I are mixed with cells expressing fasciclin III, another Drosophila homophilic adhesion molecule, the mixture sorts into aggregates homogeneous for either fasciclin I- or fasciclin III-expressing cells. The ability of these two novel adhesion molecules to mediate cell sorting in vitro suggests that they might play a similar role during neuronal development.     

Lifetime measurements reveal kinetic differences between homophilic cadherin bonds

Cadherins are multidomain adhesion proteins whose interactions direct cell sorting during histogenesis. They determine cell adhesion specificity, but prior studies failed to identify physical differences that could underlie cell sorting. These single molecule studies identify kinetic and strength differences between different cadherins. They further demonstrate that the modular extracellular architecture of cleavage stage C-cadherin supports a multistate binding mechanism. These multiple bonds exhibit a kinetic hierarchy of strengths that map to the different cadherin domains. The outer two N-terminal domains of C-cadherin form two bound states with dissociation rates of 3.9 and 0.02 s(-1). The latter is 25-fold slower than between the corresponding epithelial cadherin segments. In addition to the two fast bonds, the five-domain fragment (CEC1-5) forms two additional stronger, longer-lived bonds with dissociation rates of 0.00039 and 0.00001 s(-1). We further quantified the lifetimes of bonds subject to a constant force, and thus identified multiple dissociation events with rates that agree quantitatively with the force spectroscopy data. The qualitative features are similar to those reported for epithelial cadherin. However, the significant differences in the dissociation rates of the outer domains, which include the specificity-determining region, suggest that kinetic differences may determine cadherin discrimination, rather than adhesion energies.     

Functional correlation between cell adhesive properties and some cell surface proteins

The adhesive properties of Chinese hamster V79 cells were analyzed and characterized by various cell dissociation treatments. The comparisons of aggregability among cells dissociated with EDTA, trypsin + Ca2+, and trypsin + EDTA, revealed that these cells have two adhesion mechanisms, a Ca2+-independent and a Ca2+-dependent one. The former did not depend on temperature, whereas the latter occurred only at physiological temperatures. Both mechanisms were trypsin sensitive, but the Ca2+- dependent one was protected by Ca2+ against trypsinization. In morphological studies, the Ca2+-independent adhesion appeared to be a simple agglutination or flocculation of cells, whereas the Ca2+- dependent adhesion seemed to be more physiological, being accompanied by cell deformation resulting in the increase of contact area between adjacent cells. Lactoperoxidase-catalyzed iodination of cell surface proteins revealed that several proteins are more intensely labeled in cells with Ca2+-independent adhesiveness than in cells without that property. It was also found that a cell surface protein with a molecular weight of approximately 150,000 is present only in cells with Ca2+-dependent adhesiveness. The iodination and trypsinization of this protein were protected by Ca2+, suggesting its reactivity to Ca2+. Possible mechanisms for each adhesion property are discussed, taking into account the correlation of these proteins with cell adhesiveness.     

Constitutive and stimulus-induced phosphorylation of CD11/CD18 leukocyte adhesion molecules

The leukocyte CD11/CD18 adhesion molecules (beta 2 integrins) are a family of three heterodimeric glycoproteins each with a distinct alpha subunit (CD11a, b, or c) and a common beta subunit (CD18). CD11/CD18 mediate crucial leukocyte adhesion functions such as chemotaxis, phagocytosis, adhesion to endothelium, aggregation, and cell-mediated cytotoxicity. The enhanced cell adhesion observed upon activation of leukocytes is associated with increased surface membrane expression of CD11/CD18, as well as a qualitative upregulation of CD11/CD18 functions. To elucidate the nature of the qualitative modifications that occur, we examined the phosphorylation status of these molecules in resting human leukocytes and upon activation with PMA or with the chemotactic peptide F-met-leu-phe (FMLP). In unstimulated cells, all three CD11 subunits were found to be constitutively phosphorylated. In contrast, phosphorylation of the common CD18 subunit was minimal. PMA induced rapid and sustained phosphorylation of CD18 that occurred at high stoichiometry, but had only minimal effects on phosphorylation of the associated CD11 subunits. FMLP also induced rapid phosphorylation of CD18, but the effect was of short duration. FMLP-induced phosphorylation of CD18 was not related to its Ca++-mobilizing effect, as CD18 phosphorylation was not observed upon treatment of leukocytes with the Ca++ ionophore, ionomycin. Phosphoamino acid analysis of CD11/CD18 in PMA- or FMLP-treated monocytes revealed a predominance of phosphoserine residues in all CD11/CD18 subunits. A small component of phosphothreonine was present in CD11c and CD18 and a minor component of phosphotyrosine was also detected in CD18 upon leukocyte activation may regulate the adhesion functions mediated by the CD11/CD18 family of molecules.     

Regulation of cell-cell adhesion during Dictyostelium development

During development of Dictyostelium, four adhesion systems have been identified and adherens junction-like structures have been discovered in the fruiting body. The temporal and spatial expression of cell adhesion molecules (CAMs) is under stringent developmental control, corresponding to major shifts in morphological complexity. Genetic manipulations, including over-expression and knockout mutations, of the adhesion genes, cadA (encoding DdCAD-1), csaA (gp80) and lagC (gp150), have shed light on new roles for cell adhesion molecules in aggregate size regulation, cell-type proportioning, cell differentiation and cell sorting. As cell-cell interactions remain highly dynamic within cell streams and aggregates, mechanisms must exist to facilitate the rapid assembly and disassembly of adhesion complexes. Studies on gp80 have led to a model for the rapid assembly of adhesion complexes via lipid rafts.     

Phosphatidylserine-dependent adhesion of T cells to endothelial cells

Phosphatidylserine (PS) was exposed at the surface of human umbilical vein endothelial cells (HUVECs) and cultured cell lines by agonists that increase cytosolic Ca(2+), and factors governing the adhesion of T cells to the treated cells were investigated. Thrombin, ionophore A23187 and the Ca(2+)-ATPase inhibitor 2, 5-di-tert-butyl-1,4-benzohydroquinone each induced a PS-dependent adhesion of Jurkat T cells. A23187, which was the most effective agonist in releasing PS-bearing microvesicles, was the least effective in inducing the PS-dependent adhesion of Jurkat cells. Treatment of ECV304 and EA.hy926 cells with EGTA, followed by a return to normal medium, resulted in an influx of Ca(2+) and an increase in adhering Jurkat cells. Oxidised low-density lipoprotein induced a procoagulant response in cultured ECV304 cells and increased the number of adhering Jurkat cells, but adhesion was not inhibited by pretreating ECV304 cells with annexin V. PS was not significantly exposed on untreated Jurkat cells, as determined by flow cytometry with annexin V-FITC. However, after adhesion to thrombin-treated ECV304 cells for 10 min followed by detachment in 1 mM EDTA, there was a marked exposure of PS on the Jurkat cells. Binding of annexin V-FITC to the detached cells was inhibited by pretreating them with unlabelled annexin V. Contact with thrombin-treated ECV304 cells thus induced the exposure of PS on Jurkat cells and, as Jurkat cells were unable to adhere to thrombin-treated ECV304 cells in the presence of EGTA, the adhesion of the two cell types may involve a Ca(2+) bridge between PS on both cell surfaces. The number of T cells from normal, human peripheral blood that adhered to ECV304 cells was not increased by treating the latter with thrombin. However, findings made with several T cell lines were generally, but not completely, consistent with the possibility that adhesion to surface PS on endothelial cells may be a feature of T cells that express both CD4(+) and CD8(+) antigens. Possible implications for PS-dependent adhesion of T cells to endothelial cells in metastasis, and early in atherogenesis, are discussed.     

A role for fibronectin-leucine-rich transmembrane cell-surface proteins in homotypic cell adhesion

The fibronectin-leucine-rich transmembrane (FLRT) family of leucine-rich repeat (LRR) proteins is implicated in fibroblast growth factor (FGF) signalling, early embryonic development and neurite outgrowth. Here, we have analysed whether FLRTs may also function in cell adhesion. We find that FLRT proteins can physically interact and that FLRT-transfected cultured cells sort out from non-transfected cells, suggesting a change in adhesive properties. A similar sorting effect is also observed in Xenopus embryos and tissue aggregates. FLRT-mediated cell sorting is calcium dependent and substrate independent. Deletion analysis indicates that cell sorting requires the LRR domains, which are dispensable for FLRT-mediated FGF signalling. Conversely, sorting is independent of the cytoplasmic domain, which is essential for FLRT-induced signalling. Therefore, FLRT-mediated FGF signal transduction and homotypic cell sorting can be molecularly uncoupled. The results indicate that FLRT proteins have a dual role, promoting FGF signalling and modulating homotypic cell adhesion.     

Specificity of intercellular adhesion mediated by various members of the immunoglobulin supergene family

The immunoglobulin supergene family members have been shown to be involved in cell-cell recognition and interaction during cell growth and differentiation. Neural cell adhesion molecule, myelin-associated glycoprotein, and carcinoembryonic antigen (CEA) are immunoglobulin supergene family members which can mediate cell adhesion. We show here that nonspecific cross-reacting antigen (NCA), a closely related CEA family member, is found on the surface of rodent cells transfected with functional NCA complementary DNA in different glycosylated forms, all of which can be deglycosylated to an Mr 35,000 core protein. Furthermore, NCA can mediate Ca2(+)-independent, homotypic aggregation of these NCA-producing transfectant cells. Since CEA has three internal repeated C2-set, immunoglobulin-like domains, whereas NCA has one, only one such domain is required for the intercellular adhesive function. We also demonstrate that NCA- and CEA-producing transfectants can form heterotypic aggregates, whereas mixtures of CEA or NCA transfectants and neural cell adhesion molecule or long form-myelin-associated glycoprotein transfectants sort themselves out into homotypic aggregates. The results suggest that subsets of the immunoglobulin superfamily, such as the CEA family, can be used in both homotypic and heterotypic cellular interactions, whereas less closely related members of the family can be used to separate different cell types by strictly homotypic interactions.