A role of chondroitin sulfate glycosaminoglycan binding site in alpha4beta1 integrin-mediated melanoma cell adhesion.

We have previously reported that alpha4beta1 (but not alpha5beta1) integrin-mediated melanoma cell adhesion is inhibited by removal of cell surface chondroitin sulfate glycosaminoglycan (CSGAG), suggesting that melanoma chondroitin sulfate proteoglycan plays a role in modulating the adhesive function of alpha4beta1 integrin. In the current study, we demonstrated that alpha4beta1 integrin binds to CSGAG. We have identified a peptide from within alpha4 integrin termed SG1 (KKEKDIMKKTI) that binds to cell surface melanoma chondroitin sulfate proteoglycan, indicating that SG1 represents a CSGAG binding site within the alpha4 integrin subunit. Soluble SG1 inhibits alpha4beta1 integrin-mediated human melanoma cell adhesion to CS1. Polyclonal antibody generated against the peptide inhibits melanoma cell adhesion to CS1, and the inhibition is reversed by Mn2+ and an activating monoclonal antibody anti-beta1 (8A2). Additionally, pretreatment of cells with anti-SG1 IgG inhibits the expression of the monoclonal antibody 15/7 epitope in the presence of soluble CS1 peptide, suggesting that anti-SG1 IgG prevents ligand binding by alpha4beta1 integrin. These results demonstrate that alpha4beta1 integrin interacts directly with CSGAG through SG1 site, and that this site can affect the ligand binding properties of the integrin.

v-src-induced cell shape changes in rat fibroblasts require new gene transcription and precede loss of focal adhesions.

The mechanism of v-src-induced morphological transformation is still obscure. We compared LA29 rat fibroblasts, which express a temperature-sensitive (ts) v-src mutant, with D1025 rat fibroblasts, transfected with a ts mutant of v-fps. Upon transformation, LA29 cells adopted an elongated shape with reduced focal adhesions and loss of actin stress fibers. In contrast, activation of v-fps in D1025 cells had little effect on morphology. In both cells, paxillin was strongly tyrosine phosphorylated upon activation of the kinases. This indicates that paxillin phosphorylation is not required, or not sufficient, for the v-src-induced disruption of focal adhesions. As previously described by others, v-src activated the ras-MAP kinase (MAPK) pathway, as indicated by tyrosine phosphorylation of the rasGAP-associated proteins p62 and p190 and MAPK phosphorylation. Since MAPK affects transcription, this suggested that novel gene transcription was required. This notion was confirmed using actinomycin D and cycloheximide, which did not impair activation of v-src kinase activity, but completely blocked v-src-induced morphological changes, as demonstrated using image analysis. Furthermore, we observed that v-src-induced changes in cell shape occurred before the reduction in number and size of focal adhesions. We conclude that v-src-induced transformation of rat fibroblasts depends on synthesis of a protein, which induces rapid changes in cell shape that precede the loss of focal adhesions.

Cell surface chondroitin sulfate proteoglycans in tumor cell adhesion, motility and invasion.

Tumor cell invasion and metastasis is highly dependent on dynamic changes in the adhesion and migration of transformed and malignant cells. As with normal cell adhesion, the adhesion of tumor cells influences their cytoskeletal organization, activation of signal transduction pathways within the cell, and nuclear events leading to changes in mRNA transcription and protein synthesis. Furthermore, as tumor cells invade the circulation, they adhere to activated endothelial cells at sites within the vasculature during arrest and extravasation. Studies in the area of tumor cell adhesion and migration have demonstrated that the recognition of extracellular matrix ligands, or adhesion promoting ligands expressed on neighboring cells (i.e. counter-receptors), involves complex molecular recognition mechanisms. The complexity arises, in part, from the multiple recognition sites that are present within adhesion promoting ligands. Some of these structures within ECM components act by binding integrins, whereas others bind additional receptors such as cell surface proteoglycans. In this sense, adhesion promoting ligands may be considered as informational arrays, that function to modulate cell phenotype by engaging specific combinations of adhesion receptors on the cell surface. Understanding the mechanism(s) by which these receptor 'cluster' modify cell adhesion, motility and growth may lead to novel therapeutic strategies to control tumor cell invasion and metastasis formation. This review will highlight the role that cell surface chondroitin sulfate proteoglycans may play in modulating tumor cell adhesion, migration and invasion, with an emphasis on the relationship between cell surface chondroitin sulfate proteoglycans and integrins.

Spreading and focal contact formation of human melanoma cells in response to the stimulation of both melanoma-associated proteoglycan (NG2) and alpha 4 beta 1 integrin.

In this study, we evaluated the potential role for a specific melanoma-associated chondroitin sulfate proteoglycan core protein, termed NG2, to collaborate with alpha 4 beta 1 integrin in focal contact formation in human melanoma cells. Although melanoma cells adhered to substrata coated with either the alpha 4 beta 1 integrin binding fibronectin synthetic peptide CS1-OVA or anti-NG2 mAbs, no spreading or focal contact formation was observed on either substratum. However, melanoma cells spread and formed focal contacts on "chimeric substrata" coated with CS1-OVA and the anti-NG2 mAb, 9.2.27, indicating that engaging both adhesion receptors changes the adhesion phenotype of melanoma cells by reorganizing the cytoskeleton. The collaboration between the two receptors is specific to fibronectin, since cells adherent on substrata coated with low concentrations of either laminin and 9.2.27 or type IV collagen and 9.2.27 failed to spread, while cells adherent on low concentrations of fibronectin and 9.2.27 exhibited a fully spread morphology. Two selective tyrosine kinase inhibitors, genistein and herbimycin A, totally inhibited cell spreading on the substrata coated with CS1-OVA and 9.2.27, indicating that tyrosine kinase(s) is important for cell spreading and focal contact formation. When cells were cultured on substrata coated with CS1-OVA and 9.2.27, two proteins (M(r) 130,000 and 120,000) were tyrosine phosphorylated in a genistein- and herbimycin A-sensitive fashion. These proteins were not immunologically related to pp125FAK or alpha 4 beta 1 integrin. Importantly, when melanoma cells were cultured on substrata coated with CS1 and then stimulated with 9.2.27-conjugated microsphere beads, formation of focal contacts and stress fibers was also observed, indicating that NG2 can collaborate with alpha 4 beta 1 integrin when each receptor is engaged on distinct and separate substrata. These results demonstrate that NG2 acts as a coreceptor for spreading and focal contact formation in association with alpha 4 beta 1 integrin in melanoma cells and suggest a model in which the NG2 core protein communicates to alpha 4 beta 1 integrin by an inside-out signaling mechanism.

Adhesion mechanisms in liver metastasis formation.

Our results strongly indicate that integrin mediated adhesion between metastasizing tumour cells and hepatocytes has a decisive role in the formation of liver metastases. However, it is also clear that tumour cells may use different adhesion pathways and furthermore that the adhesion may be modulated by several factors. The role of adhesion has been demonstrated most clearly for LFA1 on T cell hybridomas, which interacts with ICAM1 present in the liver. Alternative pathways must exist, however, given the high invasive capacity of ESb cells, which is apparently LFA1 independent. A possible alternative is adhesion to fibronectin, which is present in abundance on the hepatocyte surface, both in vivo and in vitro. As there is no basement membrane under the endothelium of liver microvessels, so that tumour cells cannot adhere to laminin and collagen type IV as in other organs, adhesion to this fibronectin may be particularly important for metastasis to the liver. Many tumour types can use this pathway, and many different fibronectin receptors may be involved, including VLA-4, VLA-5 and several alpha V integrins. For carcinomas there is another possibility: Fibronectin receptor deficient cells may still adhere using the integrin alpha 6 beta 4, which binds to an unknown ligand present on the hepatocyte surface. Modulation of adhesion can occur in several ways. One example is steric hindrance by mucins that may strongly affect and even abrogate adhesion, despite high levels of appropriate integrins. Another is the activation required for integrins on lymphoma cells, the best known example of which is the activation of LFA1 on T cell hybridomas. It will be evident, therefore, that the role of adhesion in the formation of liver metastases can only be fully understood if the complete set of adhesion molecules on the tumour cells is known as well as their functional status and the possible effects of both cellular and extracellular modulating factors.

LFA-1 integrin redistribution during T-cell hybridoma invasion of hepatocyte cultures and manganese-induced adhesion to ICAM-1.

We have reported previously that the integrin LFA-1 is essential for metastasis of T-cell hybridomas to the liver. We show here that hepatocytes isolated from normal non-inflamed rat liver express intercellular adhesion molecule-1 (ICAM-1) at the dorsal surface and more prominently at the lateral and substratum-adherent surfaces. Anti-rat ICAM-1 mAb inhibited adhesion of TAM8C4 T-cell hybridoma cells to hepatocytes. Invasion between hepatocytes was not affected, but this is probably due to lack of penetration of the mAb between the hepatocytes. In all hepatocyte-adherent TAM8C4 cells, LFA-1 was concentrated at the adhesion site. Redistribution of ICAM-1 to the interacting hepatocyte membrane was also seen, but only for part of the adherent TAM8C4 cells. LFA-1 was highly concentrated on pseudopods of invading TAM8C4 cells inserted between hepatocytes, and on the upper surface of invaded TAM8C4 cells located under the hepatocytes. ICAM-1 was concentrated in the hepatocyte membrane overlying TAM8C4 cells located underneath the monolayer. These results suggests that ICAM-1 is of major importance for liver invasion by these lymphoma cells. For optimal adhesion to ICAM-1, LFA-1 on T-cell hybridomas requires activation, which apparently occurs upon contact with cell layers that are invaded (G. La Rivière et al., J. Cell Sci. 107, 551-559, 1994). LFA-1 can be activated artificially by Mn2+. To study LFA-1 redistribution upon ICAM-1 interaction with higher resolution, we performed immuno-EM on cells before and after Mn(2+)-induced adhesion and spreading on immobilized ICAM-1. By immune fluorescence, LFA-1 was observed to redistribute to the ICAM-1-adherent surface, and to be concentrated in lamellipodia of spreading TAM8C4 cells. By immuno-EM, LFA-1 was localized in microclusters of approximately 10 gold particles. This was seen in cells fixed in suspension, and the size of these clusters did not change upon adhesion to ICAM-1. LFA-1 was present at high density in thin filopodia, but again in microclusters of similar size. Comparable results were obtained with a cytotoxic T-cell clone. We conclude that Mn(2+)-induced activation of LFA-1 is not associated with the formation or enlargement of LFA-1 clusters.

Immuno-EM localization of the beta 1 integrin subunit in wet-cleaved fibronectin-adherent fibroblasts.

Using immuno-EM, we have studied the distribution of the beta 1 integrin subunit in chicken embryo fibroblasts allowed to adhere and spread for 3 hours on a fibronectin-coated surface in serum-free medium. The cells were wet-cleaved, which removed most of the cell body, yielding ventral plasma membranes with little, and sometimes virtually no, associated cytoskeleton. The beta 1 integrin subunit was detected with antibodies against the cytoplasmic domain. In immune fluorescence, it colocalized with adhesion plaques, in a punctate staining pattern, and often seemed to be at the periphery of the plaque. By immuno-EM, beta 1 was in fact found in discrete clusters, not throughout the plaque. In deep-cleaved cells from which virtually all cytoskeleton was removed, clusters could often be seen to be located on fibronectin fibrils. Furthermore, beta 1 was present in clusters at the cell margins, and isolated or in small groups at the very edge of the cell. When fibronectin synthesis, and consequently fibril formation, was inhibited by cycloheximide, large adhesion plaque-like structures were formed at the cell margin. This phenotype was reversed by addition of soluble fibronectin, which was incorporated into fibrils. As in normal plaques, talin and vinculin were present, the plasma membrane was very close (10-20 nm) to the substratum and the fibronectin layer underneath was removed. These plaques did contain beta 1 integrins but they were not in clusters. These observations indicate that the talin-vinculin network of an adhesion plaque is normally anchored to the substratum at discrete beta 1 integrin clusters that may be located on fibronectin fibrils, and that elsewhere the plaque is not necessarily attached to the substratum by interaction of integrins with matrix proteins. In the absence of fibronectin fibrils, adhesion plaque-like structures can be formed, but these are aberrant in size, location and fine structure.

TA3/St, but not TA3/Ha, mammary carcinoma cell adhesion to hepatocytes is mediated by alpha 5 beta 1 interacting with surface-associated fibronectin.

The cell lines TA3/Ha and TA3/St are derived from the same tumor, grow both in suspension and form liver metastases upon intraportal injection. We have studied the interaction of these cell lines with hepatocytes, which is likely to be relevant for liver metastasis formation. Recently we have shown that the integrin alpha 6 beta 4 is involved in adhesion of TA3/Ha cells to hepatocytes. However, we show here that the alpha 6 beta 4-specific antibodies, that inhibit adhesion of TA3/Ha cells, did not affect adhesion of TA3/St cells to hepatocytes. The beta 4 subunit, present at high levels on TA3/Ha cells, was found to be expressed at a much lower level by TA3/St cells. In contrast, TA3/St cells express much more of the beta 1-integrin subunit than TA3/Ha cells. We assessed whether these differences in integrin expression are responsible for the different adhesion mechanisms used by these cell lines. We show that alpha 5 beta 1, which is expressed by TA3/St cells, and not by TA3/Ha cells, is involved in TA3/St adhesion to fibronectin that is associated with the hepatocyte surface. Since fibronectin is the main extracellular matrix component present on the hepatocyte surface underneath the sinusoidal endothelium, alpha 5 beta 1 may be particularly important for metastasis formation in the liver, as compared to other organs.

The role of integrins and integrin activation in liver metastasis.

An overview is presented of our studies on the interaction between blood-borne tumor cells and the tissues where metastases are formed, in particular the liver. Using blocking antibodies and tumor cell mutants, we have identified the adhesion molecules involved, which so far are all integrins. Strikingly, tumor cell lines that are quite similar, and invade in a comparable fashion, use distinct integrins. Lymphomas that invade the liver massively and diffusely use LFA-1 or fibronectin receptors to adhere to hepatocytes. We have obtained evidence that LFA-1 is activated during the interaction by factors that act through G-protein-coupled receptors, and preliminary results suggest that the same may be true for the fibronectin receptors. Whereas TA3/Ha murine mammary carcinoma cells adhere to hepatocytes via alpha 6 beta 4, TA3/St variant cells of the same tumor bind via the fibronectin receptor alpha 5 beta 1. Adhesion of the TA3/Ha cells appears to be impaired by the mucin epiglycanin that is abundantly present on the surface of these cells.

Induction of nuclear lamins A/C during in vitro-induced differentiation of F9 and P19 embryonal carcinoma cells.

Lamin B is the major constituent of the nuclear lamina of undifferentiated mouse embryonal carcinoma cells. The full complement of the three major lamins A, B, and C, found in somatic mammalian cells, is acquired after induction of differentiation in vitro by certain drugs. In this study we have examined the time course of lamin A/C expression in the two embryonal carcinoma cell lines F9 and P19. We show here that lamins A/C are detectable in these cell lines, at the mRNA level and at the protein level, after 3 days of growth in media containing retinoic acid or retinoic acid + 3-isobutyl-1-methylxanthine. The data reported here indicate that the expression of lamins A/C is mainly regulated at the transcriptional level and occurs when the cells, by morphological and functional criteria, have differentiated along their developmental pathway.

Heat shock-induced redistribution of a 160-kDa nuclear matrix protein.

In this paper we describe a 160-kDa protein (p160) which is present in the nuclear matrix of rat, mouse, and human cells. Biochemical and ultrastructural analysis shows that p160 is associated with the internal matrix and is not present in the lamina-pore complex. Immunoelectron microscopy shows that the protein is part of the extranucleolar, fibrogranular network of the nuclear matrix. During an in vivo 42 degrees C heat treatment of HeLa cells, A431 human epidermoid cells, and T24 human bladder carcinoma cells, p160 transiently formed large clusters inside the nucleus. These p160 clusters are associated with the nuclear matrix network, as judged by immunolabeling on isolated nuclear matrices. The percentage of cells showing p160 clusters increased proportionally with longer heat treatments, reaching a maximum after a period of 3 h. At this time 70 +/- 5% of the cells displayed these clusters. Clustering decreased after longer heat treatments and the anti-p160 staining pattern became diffuse granular again. Other nuclear components, such as the A1 antigen of hnRNP (ribonucleoprotein), the Sm antigen of snRNPs, and lamins A and C, did not cluster during the 42 degrees C treatment, indicating that this reallocation is characteristic for the p160 matrix protein. These results demonstrate that p160 is an internal nuclear matrix element with a dynamic spatial distribution.

Ultrastructural localization of nuclear matrix proteins in HeLa cells using silver-enhanced ultra-small gold probes.

We describe a method for immunogold staining of nuclear matrix proteins using ultra-small gold particles. The nuclear matrix of HeLa cells is obtained by two fractionation steps: (a) cell permeabilization with Triton X-100 to isolate the cytoskeleton, and (b) nuclease digestion followed by an incubation in 0.25 M ammonium sulfate to isolate the nuclear matrix. To prevent redistribution of internal matrix proteins during nuclear matrix preparation, pre-fixation with 0.1% acrolein was performed. Under this condition up to 80% of protein and 90% of DNA and RNA could be removed on nuclear matrix isolation, without redistribution of internal nuclear matrix proteins. For immunogold labeling, 1-nm gold probes appeared to be required to obtain optimal penetration into the nucleus. These particles can be visualized after silver enhancement. After gold labeling the matrices are stained, embedded in Epon, and ultra-thin sections are prepared for examination in the electron microscope. The applicability of this method is examplified by the localization of a 125 KD internal nuclear matrix protein and the lamins A and C in nuclear matrix preparations of HeLa cells.

The nuclear matrix from cells of different origin. Evidence for a common set of matrix proteins.

We compared the protein composition of the nuclear matrix isolated from several murine embryonal carcinoma cells and mature tissues by two-dimensional gel electrophoresis. Two nuclear matrix fractions were investigated: the "peripheral" nuclear matrix (matrix proteins that remain insoluble after reduction), and the "internal" nuclear matrix (matrix proteins released by reduction). The two subfractions have completely different protein compositions. Although numerous differences in nuclear matrix protein composition among different cell types were observed, a limited set of polypeptides common to all mouse cell types was identified. A majority of these common proteins was also present in cells from other mammalian species (i.e. rat and human). For this set of proteins, we coin the term "minimal matrix." As expected, lamin B, known to be expressed throughout differentiation, is part of the common set of peripheral nuclear matrix proteins. Lamins A and C are not because these proteins were absent from undifferentiated embryonal carcinoma cells. Since these common nuclear matrix proteins occur in all mammalian nuclear matrices analyzed so far, it is likely that they have a basic role in nuclear organization and function.

The protein composition of the nuclear matrix of murine P19 embryonal carcinoma cells is differentiation-stage dependent.

The protein composition of the nuclear matrix of murine P19 embryonal carcinoma (EC) cells was compared with that of clonal derivatives of P19 EC differentiated in vitro, and with that of P19 EC cells induced to differentiate with retinoic acid (RA). Several major differences in nuclear matrix protein composition were found between the cell lines tested. Some polypeptides were found to occur only in EC cells, whereas others proved to be restricted to one or more of the differentiated derivatives. During RA treatment of EC cells a transient expression of some matrix proteins was observed. Several new proteins appeared, and others disappeared. Our data indicate that the protein composition of the nuclear matrix is a sensitive gauge for the differentiation state of cells.