The EA hy926 cell line is a continuous, clonable, human cell line that displays a number of features characteristic of vascular endothelial cells (Edgell et al., 1983). Here we report that when EA hy926 cells (EA cells) are plated on an extracellular matrix material [Matrigel], they undergo a process of morphological re-organization leading to the formation of a complex network of cord or tubelike structures. These events seem to resemble, in some respects, an in vitro process of angiogenesis. The morphological re-arrangement occurs within a 12-16 hr period and seems to require expression of new messenger RNA and protein, since it is completely blocked when actinomycin D or cycloheximide are present at the time the cells are plated on Matrigel. This is not due to overt toxicity of the drugs, since exposure of cells to actinomycin D at 2 hr or more after plating on Matrigel has little effect on the formation of the tubelike structures. The process of Matrigel-induced tube formation also apparently involves a G-protein mediated signal. Treatment of the EA cells with pertussis toxin completely blocks the process and causes the ADP-ribosylation of a 42 kD protein that is recognized by an antibody to Gi-alpha subunits. In contrast, concentrations of pertussis toxin sufficient to block tube formation have only modest effects on the adhesion or motility of EA cells on purified matrix components such as laminin or collagen IV. The process of Matrigel-induced tube formation also involves integrins since monoclonal antibodies to integrin alpha 6 or beta 1 subunits can completely block the process. The concentrations of anti-integrin antibodies needed to block tube formation are much lower than those required to block cell adhesion on purified matrix components and are sufficient to occupy less than 10% of the alpha 6 or beta 1 subunits available at the cell surface. These results suggest that integrins may be involved in this potential model of angiogenesis in processes beyond their usual role in cell adhesion. Based on these results, it seems likely that the EA hy 926 cell line will prove to be a useful model for in vitro study of angiogenic processes.

Abstract. Chinese hamster ovary cell populations were enriched for cells displaying low surface expression of the 140-kD integrin fibronectin receptor (FnR) by means of fluorescence-activated cell sorting using monoclonal anti-FnR antibodies. Selected cells were cloned by limiting dilution, and the resulting clones were screened for low cell surface FnR expression by ELISA. Two multiply sorted populations gave rise to variant clones possessing ,x,20 or 2% FnR expression, respectively, compared with wild-type cells. Growth rates of the "20%" and "2%" clones on serum-coated plastic dishes were similar to that of wild-type cells. Variant cells expressing 20% FnR could attach and spread on substrata coated with purified fibronectin, although somewhat more slowly than wild-type cells, while cells expressing 2% FnR could not attach or spread. Cells from all variant clones attached normally to vitronectin substrata, but some of the 2% clones displayed altered morphology on this type of substratum. Motility assays in blind well chambers showed a correlation of movement with level of expression of FnR. The number of cells migrating in response to fibronectin was greatly reduced compared with wildtype cells for the 20% FnR variant clones, while variant clones with 2% FnR showed virtually no migratory activity.Surface labeling with 125I and immunoaflinity purification of FnR showed reduced levels of intact FnR on the plasma membranes of variants with 20% FnR, while none was detected in variants expressing 2 % FnR. Nevertheless, /3 subunits were detected on the surfaces of all variant clones. Immunoblots of cell lysates from wild-type cells and from both types of variant clones showed substantial amounts of FnR/3 chain as well as enhanced amounts of a pre-/3 moiety in the variants, ct chain was markedly reduced in the 20% variants and essentially absent in the 2% variants, indicating that failure to assemble intact FnR in these variants was due to deficiencies of ct chain production. Dot blots of total mRNA from a representative clone expressing 20% FnR showed reduced levels of material hybridizing to an 0.97-kb hamster FnR ct chain eDNA probe as compared with wild type, while mRNA from a representative clone expressing 2% FnR had no detectable hybridizable RNA; this seems to agree well with the results obtained by immunoblotting. Thus, the defect in the variant clones seems to be due to reduced levels of c~ chain mRNA leading to a deficit of mature FnR and consequent alterations in cell adhesion and motility on fibronectin substrata.T HE cell surface receptor for fibronectin (FnR) ~ plays a major role in mediating adhesive interactions between cells and the extracellular matrix (11,31,45). This intrinsic membrane glycoprotein of ~140 kD is a member of the integrin superfamily of cell adhesion receptors. Integrins share a number of common features; for example, they are composed of noncovalenfly linked beterodimers with large extracellular domains. In mammalian cells, the integrin receptor superfamily is divided into three ...

Abstract. Cells are capable of adhering to and migrating on protein components of the extracellular matrix. These cell-matrix interactions are thought to be mediated largely through a family of cell surface receptors termed integrins . However, the manner in which individual integrins are involved in cell adhesion and motility has not been fully determined. To explore this issue, we previously selected a series of CHO variants that are deficient in expression of the integrin a5ß1, the "classical" fibronectin receptor. Two sets of subclones of these variants were defined which respectively express -20% or 2% of fibronectin receptor on the cell surface when compared to wild-type cells (Schreiner, C. L., J. S. Bauer, Y N. Danilov, S. Hussein, M . M. Sczekan, and R . L. Juliano . 1989. J. Cell Biol. 109:3157-3167) . In the current study, the variant clones were tested for haptotactic motility on substrata coated with fibronectin or vitronectin . Data from assays using fibronectin show that cellular motility of the 20% variants was substantially decreased NTEGRINs are transmembrane glycoprotein receptors which participate in a wide variety of cellular functions including cell-cell interactions, and the adhesion of cells to extracellular matrix (ECM) proteins such as fibronectin (Fn), I laminin, collagen, and vitronectin (Vn) (Akiyama et al., 1990;Hemler, 1990;Hynes, 1987;Juliano, 1987;. Integrins exist on the cell surface as heterodimers of noncovalently associated a and ß subunits. Some integrins interact with their respective ligands through RGD sequences contained within certain ECM proteins, while other integrins recognize non-RGD sites in proteins (Gehlsen et al ., 1988;Hall et al ., 1990;Sonnenberg et al ., 1990). Several integrins which share the ßl subunit are known to be receptors for Fn; this would include a5ß1, the "classical" Fn receptor (FnR), as well as 00 1, a4ß 1, and the recently described receptoravß 1 Juliano, 1985, 1987;Carter et al., 1990;Dedhar and Gray, 1990;Elices et al., 1991;Guan and Hynes, 1990; 1. Abbreviations used in this paper: ECM, extracellular matrix ; Fn, FnR, fibronectin, fibronectin receptor ; HpmF, high powered microscope field; Vn, VnR, vitronectin, vitronectin receptor.© The Rockefeller University Press, 0021-9525/92/01/477/11 $2 .00 TheJournal of Cell Biology, Volume 116, Number 2, January 1992477-487 (30-75 % of wild type), while the motility of the 2 variants was nearly abolished (2-20% of wild type) . Surprisingly, a similar pattern was seen for haptotactic motility of both 2% and 20% variants when vitronectin was used (=20-30% of wild type) . The reduced haptotactic motility of the fibronectin receptordeficient variant clones on vitronectin was shown not to be due to reduced vitronectin receptor (avß3) expression nor to a failure of these variants to adhere to vitronectin substrata . Transfection of the deficient variants with a cDNA for the human a5 subunit resulted in normal levels of fibronectin receptor expression (as a human a5/hamster ß1 chimera) and restored th...

Abstract. The purpose of this study was to explore the functional role of the cytoplasmic domain of the ct subunit of the o~5//31 integrin, a fibronectin receptor. Mutant CHO cells that express very low levels of endogenous hamster a5 subunit (CHO clone B2) were transfected with an expression vector containing fulllength or truncated human a5 cDNAs to form chimeric human o~5/hamster/~1 integrins. Three transfectants were examined: B2a27 expresses a full-length human c~5 subunit with 27 amino acids in the cytoplasmic domain; B2al0 expresses an or5 with a 17-amino acid cytoplasmic truncation; B2al expresses an or5 with a 26-amino acid truncation. Levels of ot5//~l surface expression in B2a27 and B2al0 cells were similar to that in wild type CHO cells. The expression of ct5//31 in B2al cells was less, amounting to 15-20% of WT levels, despite message levels that were three to five times greater than those of B2a27. The transfectants were used to examine the role of the or5 cytoplasmic domain in cell adhesion, cell motility, cytoskeletal organization, and integrin-mediated tyrosine phosphorylation.The adhesion characteristics of B2a27 and B2al0 cells on fibronectin substrata were similar to each other and to wild type CHO cells. B2al ceils displayed slight reductions in the strength and rate of adhesion to fibronectin. Cell motility in the presence of fibronectin was similar for B2a27, B2al0, and wild type CHO cells, while the B2al cells were substantially less motile. Comparable degrees of cell spreading and extensive organization of actin filaments were observed for B2a27, B2al0, and wild type CHO cells on fibronectin substrata. The B2al cells spread to a lesser degree, and some organization of actin was observed; the untransfected B2 cells remained round on fibronectin substrata and showed no actin reorganization. Since the reduced motility and cell spreading observed in the B2al cells might be due either to reduced surface expression of ot5//31 or to the truncation in the or5 cytoplasmic domain, we used flow cytometric cell sorting to select populations of B2al and B2a27 cells expressing similar levels of cell surface a5. The deficits in spreading and motility were present in B2al cells expressing high levels of or5. Thus the region of the ix5 cytoplasmic domain adjacent to the membrane seems to play an important role in cytoskeletal organization and cell motility. We also examined whether ot subunit truncation would affect integrin-mediated tyrosine phosphorylation. When B2a27 cells interacted with fibronectin substrata, increased tyrosine phosphorylation was observed in proteins of ~125 kD. A similar pattern of phosphorylation was observed in wild type CHO, B2al0, and B2al cells, but not in B2 cells. Thus, the c~5 cytoplasmic domain does not seem to be essential for integrinmediated tyrosine phosphorylation of intracellular proteins.

We have examined integrin expression and function in the human colon carcinoma cell line HT29, and in clonal sublines derived from the HT29 line. These cells express several different integrin subunits including beta 1, alpha 2, 3, 6 and alpha v, but do not express the classic alpha 5/beta 1 fibronectin receptor. Clonal variation in the pattern of integrin expression was quite limited. The profile of integrin expression correlates well with the adhesive behavior of HT29 cells. Thus the cells adhere well to vitronectin, laminin and type IV collagen, but not at all to fibronectin. Adhesion to collagen was completely blocked by an anti-beta 1 monoclonal antibody, indicating that beta 1 integrins mediate this process. Adhesion to laminin was strongly blocked by anti-beta 1 monoclonal or anti-beta 6 monoclonal, suggesting that the alpha 6/beta 1 complex functions in attachment to laminin; this was somewhat surprising since immunoprecipitation experiments indicate that most of the alpha 6 subunit seems to be associated with the beta 4 subunit. Despite their strong adherence to laminin, collagen and vitronectin, HT29 cells are not very motile and, in response to gradients of these proteins, do not migrate nearly as well as CHO cells tested under similar conditions. Since HT29 cells can undergo an enterocyte-like differentiation in glucose-free medium, we compared integrin expression in HT29 and its subclones during the process of differentiation. There was no correlation between the state of differentiation, as assessed by expression of brush-border hydrolases, and the level of expression of any of the integrin subunits measured. Thus the pattern of integrin expression in these colonic tumor cells seems to be a characteristic of the cell line, and is not readily modified by changes in cell growth or differentiation.