ABSTRACT
In order to gain further understanding of the spatial organization of interstitial and basement membrane matrices, we studied the expression of the interstitial matrix protein, fibronectin, and the basement membrane protein, laminin, in heterokaryons formed by the fusion of normal fibroblasts and teratocarcinoma-derived epithelial PYS-2 cells. These heterokaryons showed various distributions of the matrix proteins depending on the proportions of the different parental cell nuclei within the cytoplasm of the cell. Heterokaryons containing equal numbers of fibroblast and PYS-2 cell nuclei showed an abundant laminin matrix subcellularly and only minor amounts of fibronectin matrix at the periphery of the cells. Similar results were obtained in heterokaryons containing an excess of epithelial cell nuclei. In heterokaryons containing an excess of fibroblast nuclei, on the other hand, laminin matrix was reduced and a fibrillar fibronectin matrix was seen also on top of the cell body. The results suggest a gene dosage-type of effect on the expression of these proteins. Furthermore, extracellular laminin and fibronectin matrices did not codistribute around the heterokaryons but the two proteins were assembled into separate structures. The lack of codistribution of fibronectin and laminin matrices in heterokaryons suggests that the molecular interactions, which determine the assembly of basement membrane and interstitial matrices in these cells are highly type-specific. Similar mechanisms may also operate in the assembly of extracellular matrices in vivo.
INTRODUCTION
The extracellular matrix is composed of two biochemically and morphologically separate entities, the basement membrane (Timpl, 1989) and the interstitial connective tissue matrix (Linsenmayer, 1991). Basement membranes are sheets of extracellular matrix that separate epithelia from mesenchyme. In adult tissues they contain type IV collagen (Timpl et al., 1981; Tryggvason et al., 1984; Timpl, 1989), noncollagenous glycoproteins belonging to the family of laminins (Timpl et al., 1979; Timpl, 1989), entactin (Paulsson et al., 1986; Durkin et al., 1987) and proteoglycans, containing mostly heparan sulfate (Kato et al., 1988; Klein et al., 1988; Soroka and Farquhar, 1991). The extracellular matrix of the interstitial connective tissue, e.g. embryonic mesenchyme and the adult dermis, has a distinctly different composition. This type of matrix contains interstitial collagens of types I, III, V, VI, VII, XII and others (Linsenmayer, 1991), as well as fibronectins (Ruoslahti et al., 1982; Hynes, 1990; Yamada, 1991) and proteoglycans, mostly of the chondroitin sulfate (Brennan et al., 1984; Voss et al., 1986) and heparan sulfate types (Cöster et al., 1986; Heremans et al., 1989).
The two types of matrices, basement membranes and interstitial matrices, form separate although adjacent structures in vivo (Vracko, 1974). Mixed matrices containing molecules of both types have not been reported in normal tissues, with the exception of embryonic basement membranes (Leivo, 1983; Martinez-Hernandez and Amenta, 1983). In order to understand better the nature of spatial organization in the assembly of the different matrix components, we have studied the formation of the different matrices under circumstances where the different components are synthesized within the same cell. As our model system we selected heterokaryons, multinucleated cells formed by the fusion of two different kinds of parental cells. Fibroblasts and teratocarcinoma-derived, epithelial PYS-2 cells were selected as parental cells, since they produce components of only one type of matrix, i.e. fibronectin (Hedman et al., 1979) or laminin (Leivo et al., 1982), respectively.
Our results show that heterokaryons of fibroblasts and PYS-2 cells deposit both laminin and fibronectin in their matrices. The ratio of the different parental cell nuclei determined the type of matrix deposited, suggesting a gene dosage-dependent effect. Laminin and fibronectin are generally not incorporated into the same structures, indicating the separate assembly of basement membrane and interstitial matrices within the extracellular domain of even a single cell.
MATERIALS AND METHODS
Cell culture
Mouse teratocarcinoma-derived PYS-2 (parietal yolk sac carcinoma) cells used in this study were gifts from Dr. J. Lehman (Lehman et al., 1974). In karyotypic analysis performed at the Department of Pathology, University of Helsinki, the cells were hypotetraploid, in agreement with the original observations (Lehman et al., 1974). PYS-2 cells were cultivated in subconfluent monolayers in Eagle’s Minimum Essential Medium (MEM) supplemented with 10% fetal calf serum (Gibco, Paisley, Scotland) and antibiotics (penicillin 100 i.u./ml and streptomycin 50 mg/ml, Orion Diagnostica, Helsinki, Finland). Mouse fibroblastic NIH-3T3 cells (from Dr. S. Aaronson, Bethesda, MD) and human embryonic body wall fibroblasts (HES) were cultivated in RPMI 1640 medium (Gibco) supplemented with 10% fetal calf serum and antibiotics.
Identification and fusion of cells
For identification of the parental cells and their fused products two different methods were used. First, the cells were labelled with cytoplasmic polystyrene particles (Veomett et al., 1974; Laurila et al., 1979). Fibroblastic cells (HES or NIH-3T3) were incubated with large polystyrene particles (diameter 1.2 μm) and the PYS-2 cells were incubated with small polystyrene particles (diameter 0.765 μm) (Polysciences, Warrington, PA). Nonphagocytized particles were removed by washing the monolayer three times with PBS and by centrifuging (400 g, 3 × 5 min) the cells that had been detached with trypsin. Secondly, differences in nuclear morphology between mouse and human cells allowed the identification of these cells in culture. Staining of the nuclei was carried out using the bisbenzimid fluorochrome (Hoechst 33258, Riedel-de Haen, Hannover, Germany) as described previously (Russell et al., 1975; Laurila et al., 1978). The numbers of the different parental cell nuclei in the heterokaryons were determined using this method.
Fibroblastic cells (HES or NIH-3T3) cultured on coverslips were treated with 3000 haemagglutinating units of Sendai virus (courtesy of Dr. K. Cantell, Public Health Institute, Helsinki, Finland) inactivated by beta-propiolactone (Neff and Enders, 1968), and subsequently PYS-2 cells were allowed to adhere to the virustreated fibroblast monolayer. Cell fusion was induced by incubation of the cells at +37°C for 30 min as previously described (Laurila et al., 1979). Alternatively, polyethyleneglycol (PEG 1500; BDH Chemicals, Poole, England; 50%, v/v, in RPMI 1640) was used to induce cell fusion. The cells were treated for 1 min at +37°C with 50% PEG 1500. Subsequently the PEG was gradually diluted with the culture medium as described previously (Laurila et al., 1979).
After fusion, the cells were cultured in a defined medium containing glutamine (Gibco) and ITS (insulin/transferrin/selenium; Collaborative Research Inc., Waltham, MA; Barnes and Sato, 1980), and supplemented with 30 μg/ml of sodium ascorbate on the first day and 15 μg/ml on subsequent days of the experiment.
Immunocytochemistry
For immunofluorescence analysis of extracellular antigens the cultures were fixed with 3.5% paraformaldehyde in 0.1 M phosphate buffer, pH 7.4, for 30 min at various time points (3 h to 96 h) after fusion. For the study of intracellular structures the fixed cells were treated with the non-ionic detergent Nonidet P40 (0.05%, BDH Chemicals, Poole, England) in phosphate-buffered saline, pH 7.2, for 30 min (Laurila et al., 1978). The guinea pig antibodies against rat L2 tumour laminin (Engvall et al., 1983) and the rabbit antibodies against human plasma fibronectin (Ruoslahti et al., 1982) used in this study have been described previously. The specificities of the antisera, lack of cross-reactivity, as well as immunohistochemical control tests, including blocking experiments, have been described previously (Ruoslahti et al., 1982; Engvall et al., 1983).
The staining procedure for indirect immunofluorescence using fluorescein-isothiocyanate (FITC)-conjugated anti-rabbit IgG antiserum (Cappel, Cochranville, PA) has been described previously (Laurila et al., 1978). For simultaneous visualization of fibronectin and laminin, the cultures were first treated with guinea-pig antilaminin antibodies (Cappel) at a 1:40 dilution followed by FITCconjugated goat anti-guinea pig IgG antiserum (dilution 1:40, Cappel), and thereafter with rabbit anti-fibronectin antibodies followed by tetramethylrhodamine isothiocyanate (TRITC)-conjugated goat anti-rabbit antibodies (Cappel) at a 1:40 dilution. In control experiments, an identical double staining of the cells was performed, but the first antibodies, either the guinea pig antilaminin antibodies or the rabbit anti-fibronectin antibodies, were omitted. In the first case, no green FITC fluorescence was detected, and in the latter case no or only negligible red TRITC fluorescence was seen.
The expression of fibronectin and laminin was classified according to the distribution and intensity of fluorescence. Fibronectin and laminin fluorescence of the fused cells was compared with that of unfused cells nearby. The fluorescence pattern was classified as “negative or weak” when the cells showed either no extracellular matrix fluorescence or only a few short strands at the cell periphery or in contact regions between cells. The fluorescence was classified as “moderate” if fibronectin or laminin were seen also on or under the cell body, and as “strong” if the matrix fluorescence was as extensive as that of confluent fibroblastic or PYS-2 cells on the coverslip (see also Laurila et al., 1979; Laurila and Stenman, 1982).
Light microscopy
For fluorescence microscopy, a Zeiss Universal microscope with epi-illuminator IIIRS and HBO 200 W lamp for specific fluorescence excitation, together with filters for FITC (490 nm excitation light), TRITC (545 nm excitation light) and ultraviolet light, was used. There was no leakage of the FITC-fluorescence to the TRITC channel, whereas small amounts of pale yellow TRITC-derived fluorescence were observed on the FITC channel. FITC-specific fluorescence, however, appeared as a more-intense bright green colour. The cells were first identified in phase-contrast microscopy, based on the number of nuclei and the cytoplasmic labelling wih polystyrene particles. Thereafter, the same cell was observed in UV-fluorescence for identification of the different nuclei. To compare the distribution of fibronectin and laminin, FITC-fluorescence was photographed first and then the same field was re-exposed for TRITC-fluorescence. The localization of extracellular fibronectin and laminin fluorescence relative to the upper and lower surfaces of the cells and to the substratum was determined by changing the level of focus within the specimen.
RESULTS
Identification of cells
Cells were identified in phase-contrast microscopy by the size of cytoplasmic polystyrene particles and the morphology of the nuclei in UV-fluorescence after bisbenzimide staining. The efficiency of the polystyrene labelling was assayed by studying 200 single cells of both cell types. The fibroblastic cells (HES and NIH-3T3) contained ∼15-50 large particles per cell and less than 2% of the cells were unlabelled. The epithelial PYS-2 cells contained ∼10-40 small particles per cell and 5% of these cells appeared unlabelled. In cocultivation experiments over 96% of mononucleated cells contained particles of only one kind. The rest of the cells were either negative or contained two kinds of particles (2-5%). Occasionally, a few loose contaminating particles were seen on the coverslips. These data are in accordance with our previous results (Laurila et al., 1979, 1982; Laurila, 1981).
Two different types of heterokaryons (mouse × human and mouse × mouse) were studied. In heterokaryons formed between mouse NIH-3T3 cells and mouse PYS-2 cells, the identification of the heterokaryons took place by cytoplasmic polystyrene labelling. Estimation of the number and ratio of the different parental cell nuclei in these heterokaryons was not possible due to the similarity of the nuclear morphology in UV-fluorescence. In mouse × human heterokaryons, on the other hand, clear differences in nuclear and nucleolar morphology allowed us to distinguish between nuclei of the two species in UV-fluorescence after staining with the Hoechst 33258 fluorochrome. Murine nuclei characteristically showed numerous brightly fluorescing dots, whereas fibroblast nuclei typically displayed a more evenly distributed fluorescence pattern with a conspicuous unstained nucleolus (compare Fig. 1B and Fig. 2B). These differences in nuclear morphology also allowed us to determine the number and ratio of the nuclei contributed by the different parental cells. In heterokaryons formed between mouse PYS-2 cells and human fibroblasts the identification of the different cell types using latex labelling and UV-immunofluorescence correlated well (over 95% correlation). These and our previous results (Laurila et al., 1979, 1982; Laurila, 1981) show that the identification of heterokaryons is reliable under our experimental conditions.
Fibronectin and laminin in homokaryons
In both human and mouse fibroblastic cells and their homokaryons extracellular matrix-associated fibronectin was seen as a typical fibrillar network. No differences in the distribution of the fibronectin matrix were detected between the fibroblastic cells and their homokaryons. Strong intracellular fibronectin fluorescence was present in both the single fibroblasts and the homokaryons after permeabilization of the cells with NP40 (see also Laurila, 1981). No laminin positivity was detected in these cells (Fig. 1).
No extracellular matrix fibronectin was seen on or around PYS-2 cells and their homokaryons (Fig. 2). Neither was intracellular fibronectin detected in these cells. Extracellular laminin, on the other hand, was seen as a typical patchy subcellular matrix, clearly distinct from the fibrillar fibronectin matrix seen in fibroblast cultures (compare Fig. 1D with Fig. 2C). With time, increasing amounts of subcellular laminin matrix were detected and at 4-5 days of culture the cells showed an extensive subcellular matrix (Fig. 2). No difference in the expression of the laminin matrix was detected between the PYS-2 cells and their homokaryons. A strong intracellular laminin fluorescence was seen in these cells throughout the experiment. Latex labelling of PYS-2 cells or fibroblasts did not have any detectable effect on the expression of laminin or fibronectin.
Extracellular matrix in heterokaryons
One day after fusion, heterokaryons between epithelial PYS-2 cells, on the one hand, and fibroblastic NIH-3T3 cells or human HES cells, on the other hand, expressed varying amounts of both laminin and fibronectin matrices. Typically, the laminin matrix was seen as a subcellular amorphous matrix reminiscent of the one seen in the PYS-2 cell homokaryons. Usually these cells showed only a few peripheral strands of fibronectin matrix most of which were not codistributed with the laminin matrix. Some heterokaryons, however, expressed only a little subcellular laminin matrix, whereas a strong fibrillar fibronectin matrix was present on the cell body. In these cells, the laminin and fibronectin matrices were not codistributed on the cell body.
In order to analyze the possible effect of gene dosage on the expression of the different matrix proteins, heterokaryons between mouse PYS-2 and human HES cells were studied. In these cells the number and ratio of the parental cell nuclei could be determined by nuclear morphology in UV fluorescence. Three hours after fusion all the heterokaryons, regardless of the ratio of the different types of nuclei, expressed only a few strands of fibronectin matrix at the periphery of the cells and no or only minimal amounts of spotty laminin matrix was seen under the cell bodies (not shown).
In heterokaryons with equal numbers of both fibroblast and PYS-2 cell nuclei the deposition of laminin onto subcellular matrix increased considerably with time and a confluent matrix was seen at 48 hours of culture (Figs 3C and 6B). In these heterokaryons, the deposition of extracellular fibronectin was minimal with only a few strands seen at the periphery of the cells (Figs 3D and 6A). No fibronectin matrix was seen on top of the cell bodies, although the surrounding fibroblasts and their homokaryons showed an abundant fibrillar fibronectin matrix (Fig. 3D). In heterokaryons with an excess of PYS-2 nuclei extracellular fibronectin and laminin displayed a similar distribution to that seen in heterokaryons with equal numbers of both nuclei (Figs 4 and 6). Typically, the laminin and fibronectin matrices were morphologically distinct, with fibronectin being distributed pericellularly in a fibrillar pattern and laminin showing a subcellular patchy pattern (Figs 3 and 4). Occasionally, close apposition or minor codistribution of the two matrices was observed along the cell periphery.
Heterokaryons containing an excess of fibroblast nuclei over PYS-2 cell nuclei (3:1 or more) expressed a fibrillar fibronectin matrix within 48 hours, whereas little or no laminin matrix was seen under or around these cells (Fig. 5). Representative cells from six different fusion experiments at 24 to 48 h of post-fusion culture were scored. The data are presented in Fig. 6. The extracellular matrix expression remained similar even after 96 h of culture.
Intracellular fibronectin and laminin in heterokaryons
All the heterokaryons, regardless of the ratio of parental cell nuclei, expressed an intense cytoplasmic fibronectin fluorescence, which was evenly distributed around all the nuclei. Similarly, all the heterokaryons showed a strong perinuclear laminin fluorescence. The intracellular fluorescence remained strong for both laminin and fibronectin throughout the four days of the experiment (Fig. 7).
DISCUSSION
Regulation of the formation of different types of extracellular matrices is poorly understood (Ruoslahti, 1991). To gain further knowledge on the assembly of different types of extracellular matrices and on the interactions of their constituent molecules, in the present work we have studied the expression of the fibroblastic and epithelial cell matrices in a system where both types of matrices would be expressed by the same cell. Heterokaryons and continuously growing hybrid cells have been widely used in studies on phenotypic regulation and cell differentiation (Ringertz and Savage, 1976; Blau, 1988; Watt, 1991). We selected heterokaryons as our model system, since they offer several advantages over continuously growing hybrid cells. First, in heterokaryon experiments no selective pressures related to culture media are exerted on the parental or fused cells. Secondly, continuously growing hybrid cells tend to loose chromosomes while no chromosome losses occur in heterokaryons maintained in a non-dividing state (Weiss and Green, 1967; Norum and Migeon, 1974). Finally, in heterokaryons the nuclei remain intact as separate compartments, thus allowing for studies on the presence and effects of trans-acting factors (see e.g. Baron and Maniatis, 1986; Bergman and Ringertz, 1990).
Fibroblasts and PYS-2 cells were selected as fusion partners because of their exclusive expression of interstitial and basement membrane matrices, respectively. Cultured fibroblasts form an extensive network of fibronectin-containing extracellular matrix, which does not contain significant amounts of laminin (Hedman et al., 1979). The epithelial PYS-2 cells, on the other hand, express a plaque-like subcellular matrix containing laminin (Leivo et al., 1982) and no fibronectin is detected in these matrices (Leivo et al., 1986).
The fusion process as such or the method of cell fusion seemed to have no effect on the expression of either fibronectin or laminin. After both Sendai virus- and PEG-induced cell fusions, homokaryons of both fibroblasts and PYS-2 cells expressed either fibronectin or laminin in a mutually exclusive manner, with patterns identical to those of their parental cells (see also Laurila et al., 1979; Laurila, 1981; Laurila and Stenman, 1982). In heterokaryons, on the other hand, both fibronectin and laminin were abundant intracellularly, suggesting cosynthesis of the two matrix proteins in these cells. Thus the expression of neither protein was suppressed.
The composition and morphology of the extracellular matrix varied in the different heterokaryons according to the numbers of the different parental cell nuclei present within the cytoplasm of the cells. In heterokaryons with an equal number of the two types of parental cell nuclei, laminin formed an abundant subcellular matrix while only minor amounts of fibronectin were present, indicating that the epithelial matrix phenotype prevailed in these cells. The epithelial matrix phenotype dominated also in heterokaryons with an excess of PYS-2 cell nuclei. These findings are in line with our previous results showing that the fibronectin matrix is lost from heterokaryons of human fibroblasts and epithelial MDCK cells (Laurila, 1981). In the present study, on the other hand, heterokaryons with an excess of fibroblast nuclei expressed increasing amounts of fibrillar fibronectin matrix in proportion to the number of fibroblast nuclei. The amount of the laminin matrix was reduced in these cells. Thus, the structure and composition of these matrices was dependent on the relative numbers of the parental cell nuclei in a gene dosage-dependent manner (see Harris, 1970; Pavlath and Blau, 1986; Blau, 1988). Such a gene dosage could involve, e.g., the genes of different matrix proteins, their receptors or other matrix assembly factors. In addition, the various amounts of matrix receptors and presynthesized matrix proteins derived from parental cell membranes and cytoplasmic components may contribute to the amounts of epithelial and fibroblastic matrices retained by these cells. However, considering the length of our experiments (up to 4 days) it seems unlikely that presynthesized matrix proteins alone could account for the continuous deposition of the abundant extracellular matrices of our heterokaryons.
In the present heterokaryons, laminin and fibronectin matrices were not generally incorporated into the same extracellular structures. Yet, it is known that PYS-2 matrices bind exogenously added plasma fibronectin, presumably due to interactions with other matrix components, e.g. proteoglycans (Leivo et al., 1986). This phenomenon, in addition to the fact that minor amounts of fibronectin are secreted by the fibroblasts in our mixed cultures, could explain the occasional codistribution of laminin and fibronectin at the edges of heterokaryons in our serum-free cultures. Our findings raise questions on the molecular mechanisms responsible for the deposition of separate extracellular matrices by one cell, a phenomenon that to our knowledge has not been described in other cultured cells. Protein chemical studies have shown that isolated basement membrane components interact stoichiometrically with one another and form basement membrane-like complexes in vitro (Kleinman et al., 1986). Also different types of interstitial collagens can interact and form fibrils spontaneously (Fleischmajer et al., 1990; Birk et al., 1991). Soluble laminin, on the other hand, does not show significant binding to soluble fibronectin or interstitial collagens (Sakashita et al., 1980). These preferential binding interactions of the various matrix proteins may partly explain the segregation of the different matrices in our heterokaryons.
The deposition of extracellular matrix components into a matrix seems to require active participation of a cell (Roman et al., 1989). Our results, which show the separate deposition of basement membrane and interstitial matrices, suggest that the cellular mechanisms leading to the assembly of matrix structures are specific for a given type of extracellular matrix. The assembly process may depend on the existence of specialized plasma membrane compartments or pockets as shown previously for the assembly of collagen fibrils (Birk and Trelstad, 1985). Therefore, the deposition of two separate extracellular matrices may reflect a polarization or segregation of parental cell-derived plasma membrane domains in our heterokaryons. The interactions of extracellular matrix molecules with cells are thought to occur via cell adhesion receptors of integrin and non-integrin types (Akiyama et al., 1990; Albelda et al., 1990; Ruoslahti, 1991). Although the different integrins share a variety of common matrix ligands, our results support the notion that the interactions of matrix components with the functioning cell are matrix type-specific. It has been suggested that factors, such as the “matrix assembly receptor” (Ruoslahti, 1991) or “disintegrins” (Barletta et al., 1991) confer specificity on these interactions.
In our study, the fusion of a non-polar fibroblastic cell with an epithelial (PYS-2) cell expressing a polarized subcellular matrix resulted in heterokaryons which maintained the polarity of the laminin matrix even in a moderate excess of fibroblast nuclei. Our system, therefore, differs from previous fusion experiments, where parental cell-derived membrane-associated proteins became evenly distributed in heterokaryons (Edidin et al., 1982). It remains to be determined whether the subcellular location of the laminin matrix in our heterokaryons is associated with a corresponding polarization of laminin receptors (see e.g. De Luca et al., 1990; Sonnenberg et al., 1990) and whether the separate deposition of the two matrices is parallelled by a segregation of laminin and fibronectin receptors. The mechanisms leading to the separate assembly of basement membrane and interstitial matrices in our heterokaryons may also be operative in the assembly of extracellular matrices in tissues in vivo and influence their specific properties.
ACKNOWLEDGEMENTS
The authors are grateful to Dr. Eva Engvall and Dr. Erkki Ruoslahti of La Jolla Cancer Research Foundation, La Jolla, CA, USA, for the use of specific antisera to laminin and fibronectin. Ms. Hannele Laaksonen is gratefully thanked for technical assistance. This study was supported by grants from the Cancer Society of Finland and the Finnish Culture Foundation.