- Email: [email protected]
CELL TYPE-DEPENDENT COLLAGEN-TYPE RECOGNITION BY CELL RECEPTORS
Cell Biology International 2001, Vol. 25, No. 7, 643–648 Article No. cbir.1999.0487, available online at http://www.idealibrary.com on
CELL TYPE-DEPENDENT COLLAGEN-TYPE RECOGNITION BY CELL RECEPTORS L. A. SADOFIEV*, N. S. NIKOLAENKO, N. V. KALMYKOVA and O. I. PODGORNAYA Institute of Cytology RAS, Tikhoretsky pr. 4, 194064, St Petersburg, Russia Received 8 January 1999; accepted 25 November 1999
Affinity chromatography of a number of cell types on collagens I and III reveals three proteins with MR of 250, 170 and 140 kDa. These proteins are able to discriminate between types I and III, but not types III and IV. Collagen-type recognition is therefore characteristic for cells of connective tissue origin. Polyclonal antibodies (Ab) raised against 170 and 140 kDa polypeptides and used in immunofluorescence show membrane localisation for both, with their distribution being similar to each other and to the distribution of the integrin 1 chain. Ab p140 and commercial monoclonal antibodies against 2 chain stain a band of the same molecular mass as from purified collagen binding proteins from liver cells, indicating that the 140 kDa protein is probably the 2 integrin chain. The 2 chain containing integrins are therefore able to discriminate collagen types I and III and collagen type recognition by this receptor is cell-type 2000 Academic Press dependent. K: extracellular matrix; collagen; receptors; integrins; connective tissue.
INTRODUCTION Collagens are a family of extracellular matrix (ECM) proteins, strongly affecting cell proliferation and differentiation (Gerstenfeld et al. 1988). Even structurally very similar proteins, such as collagen type I and III, can affect cell behaviour in different ways (Miller and Rhodes, 1982). The mechanism of difference in action could arise from the cell receptors for collagens. The majority of a number of cell receptors known to be capable of binding collagens belong to integrin family, and exist in many tissue types. In some cases, integrins mediate cell-ECM interactions and cell-cell adhesion (Buck and Horwitz, 1987; Albelda and Buck, 1990; Hynes, 1992), but are also responsible for mediating signals between the ECM to the cell nuclei, even in switching between proliferative and differentiation modes of cell activity (Dedchar et al., 1987; Maniotis et al., 1997; Roskelley et al., 1995; Takeuchi et al., 1996). Integrins are heterodimers formed by one -type subunit and one -type subunit. More than one kind of -subunit can usually combine with each kind of -subunit, forming receptors with different *To whom correspondence should be addressed. 1065–6995/01/070643+06 $35.00/0
functions and ligand specificity (Buck and Horwitz, 1987; Albelda and Buck, 1990; Hynes, 1992). Integrins are the main type of collagen receptors, and data exists which show their ability to recognize different collagen types (Tuckwell and Humphries, 1996). The ability to recognize collagen types has been shown for the other group of collagen receptors, the anchorins. Anchorins are chondrocyte receptors of the annexin family, which have relatively low molecular weights (40 kDa) and can selectively bind cartilage collagen II (Mollenchauer et al., 1984). The aim of the present work was to determine whether there are receptors capable of recognising collagen types of cells other than chondrocytes. The principal method employed involves a comparative analysis of cell membrane proteins binding to affinity columns containing collagen types I, III, IV. Three polypeptides of approximate MR 250, 170, 140 kDa were able to recognize collagens I and III, an ability only found in connective tissue cells, such as fibroblasts and osteoblasts. Polyclonal antibodies raised against these polypeptides localised them to the cell surface in a very similar pattern to that of the integrin -chain. Immunoblots of 2001 Academic Press
644
collagen-binding proteins from liver cells with antibodies against 140 kDa protein and commercial monoclonal antibodies against integrin 2 chain revealed the same zone. MATERIALS AND METHODS Affinity columns Collagens for affinity columns were prepared from human placenta by pepsin digestion (types I, III, IV), or from rat skin without pepsin treatment (types I and III), according to the standard procedure of Miller and Rhodes (1982). The purity of collagens obtained was >90% according to SDS– PAGE (Laemmli, 1970), with only some crosscontamination of collagen types. Coupling to BrCN-activated Sepharose 4B (Pharmacia, Sweden) was carried out according to the manufacturer’s recommendations. We used 10 mg of protein/ml of wet Sepharose; the column’s bead volume was 0.2 ml for analytical purposes and 1 ml for preparative purposes. Sources of membranes The following cell lines were obtained from the Cell Culture Bank, Institute of Cytology, RAS. Cell lines of connective tissue origin: immortalised rat fibroblasts E1A-12S, diploid fibroblasts from rat lung and muscle, rat osteosarcoma ROS 17/2.8, human osteosarcoma HOS. Cell lines of epithelial origin: human epidermoid carcinoma A431, human epithelioid carcinoma HeLa, rat pituitary tumour GH3. All cells were cultured on plastic dishes in DME-F12 (3:1) medium in the presence of 10% fetal calf serum. Membrane extracts of human embryo liver were used to prepare isolated receptors. Isolation and characterization of receptors All procedures were carried out at 4C. Cells were removed from dishes by Versene solution without trypsin. Cells were disrupted by hypotonic shock in 1/10 of normal strength PBS. Membrane vesicles were collected by centrifugation at 4000g, and membrane proteins were extracted in 2 ml TBS containing 0.5% Triton X-100, 5 m CaCl2, 5 m MgCl2, 0.15 NaCl (referred to as receptor extraction buffer, or REB). Extracts were applied to affinity columns after the removal of insoluble material by centrifugation at 4000g. Human embryonic liver was homogenised in Tris–HCl EDTA buffer at pH 7.5. The membrane
Cell Biology International, Vol. 25, No. 7, 2001
fraction was obtained by ultra-centrifugation in sucrose gradient and extraction with REB. Extracts were passed through an albumin–Sepharose column to remove proteins which bound nonspecifically, and the flow-through was loaded on to collagen–Sepharose columns. The columns were washed with the same buffer until no proteins were detected in the eluate, after which they were washed with 10 bed-volumes of the same buffer without detergents. Bound proteins were eluted with TBS buffer containing 10 m of octylglucoside, 0.5 NaCl, and 50 m EDTA. Proteins were separated using 7.5% gels by SDS–PAGE (Laemmli, 1970), and analysed following silver-staining. Antibodies Commercial monoclonal anti-chicken integrin 1 chain antibodies were obtained from Sigma (St Louis, MO, U.S.A.) and anti human integrin 2 chain from DAKO (Norway). Collagen-binding proteins of 250, 170 and 140 kDa were purified on collagen affinity columns from Triton X-100 extracts of whole human liver, separated on SDS–PAGE, and the protein zones cut out and injected into mice. Three injections of approximately 5 g of protein per injection were given at weekly intervals. Serum was taken 1 week after the last injection. Immunoblotting Immunoblotting was carried out after SDS–PAGE electrophoresis in 7.5% gel according to standard procedure for nitrocellulose membranes (Millipore, U.S.A.; Towbin et al., 1979). The secondary antibodies were biotinylated goat anti-mouse (Sigma), followed by streptavidin–alkaline phosphatase treatment and NBT-BCIP stain, according to the manufacturer’s recommendations. Immunofluorescence Cells in cold PBS (3000–5000 cells in 0.01 ml) were placed on glass coverslips or on multitest slides (Flow, U.K.) coated with D-polylysine (MR >70 kDa; Sigma). After incubating for 30– 60 min at 4C, unattached cells were removed, and those attached were fixed at 4C formaldehyde or paraformaldehyde (Merck, Germany) in PBS containing MgSO4 (PBS-Mg) or processed without fixation. All antibodies were diluted with PBS–BSA buffer. The working dilutions for antiserum against the 250, 170 and 140 kDa proteins, and against integrins 1
Cell Biology International, Vol. 25, No. 7, 2001
645
Fig. 1. (A) SDS–PAGE of collagen-binding proteins. Elution with EDTA-NaCl buffer. Eluates from collagen columns were concentrated by lyophilization and separated on a 7.5% gel in the presence of -mercaptoethanol. Proteins were visualised by silver staining. The collagen type used for protein isolation is indicated by roman numbers below each line. Brackets indicate the position of proteins with MR70 kDa, not completely separated. Group of lines: (1) murine fibroblasts; (2) relatively poorly differentiated human osteosarcoma cell line, HOS; (3) highly differentiated rat osteosarcoma cell line, ROS; (4) human embryo liver cells. (B) Testing of antibodies raised against 250, 170 and 140 kDa proteins by immunoblotting. The antibody tested indicated at the bottom of each line. For testing, unpurified membrane extract of human embryo liver cells was used. Antibodies against 170 and 140 kDa proteins show good specificity, although some cross-reaction between 170 and 140 kDa proteins was observed. The specificity of antibodies against 250 kDa is poor. Dilution of antisera; p250–1:50, p170–1:50, p140–1:100. (C) Immunoblot staining of collagen-binding proteins with anti-human integrin 2 chain monoclonal antibodies (Dako) (1) and our anti-140 kDa protein antibodies (2). Collagen-binding proteins from human embryo liver were purified on collagen I affinity column. Dilution of both antibodies 1:1000.
chain, were 1:50, and secondary antibodies (RAMFITC, Sigma) were diluted 1:100. Cells were incubated with antibodies for 1 h at 4C (unfixed samples), or at room temperature (fixed samples). The slides were viewed and photographed with an Axioskop microscope (Ziess, Germany).
RESULTS Comparative analysis of collagen receptors Affinity chromatography of a range of cell types on collagen columns identified at least three high MR polypeptides of 250, 170 and 140 kDa which were able to discriminate collagen types I and III, but not types III and IV, under the conditions of the affinity chromatography which was used. These have been referred to as ‘collagen receptors’,
according to the isolation procedure. The comparative analysis of collagen I, III, and IV binding receptors from cells lines of different origins is shown in Fig. 1A. (except collagen IV) and summarised in Table 1. No differences between patterns of collagen III and collagen IV binding proteins were found. The ability of the polypeptides to recognize collagen types in vitro was dependent on the buffer ionic strength and type of cells. In buffers lacking NaCl, all three polypeptides bound the three types of collagen with similar affinities as in fibroblast cells, and all three polypeptides bound collagen I, but not collagen III, in ROS cells. But adjustment of NaCl concentration upwards to the physiological level of 0.15 completely inhibited the binding of the 250 and 140 kDa polypeptides with collagen I in both cases. No salt concentration effect was observed in the case of liver cells.
646
Cell Biology International, Vol. 25, No. 7, 2001
Table 1. Collagen-type recognition by high molecular weight cell receptors Fibroblasts I
250 170 140
* *
III
+ +
Osteogenic cells IV
+ +
HOS
Liver cells ROS
I
III
IV
I
III
IV
+
+ +
+ +
* + *
N/A N/A N/A
I
III
IV
+ + +
+ + +
+ + +
Protein present in eluate (+); protein is absent (); not tested (N/A); *effect is ionic strengthdependent.
Collagen binding to the polypeptides seemed to be dependent on divalent cations. The 250 and 140 kDa polypeptides were completely removed from collagen affinity columns by EDTA without an increase in the ionic strength of the buffer. The 170 kDa polypeptide from ROS cells bound to any of the collagens both in the presence or absence of EDTA, and could be eluted only by increasing the ionic strength up to 0.5 NaCl. A zone, possibly two band, with MR in the range of 90 kDa appears in the SDS–PAGE of eluates, and there is a strong correlation with the appearance of three polypeptides (Fig. 1A, arrowheads). Epithelioid cell line membrane extracts showed no high MR proteins by affinity chromatography, even after a 10-fold increase in loading (data not shown). Integrins are shown to be in A431 cells (Genersch et al., 1996), so the inability of the collagen receptors to attach to the collagen types used is probably due to some features of the receptors. Antibodies Mouse polyclonal antiserums against polypeptides 250, 170 and 140 kDa (Ab p250, Ab p170 and Ab p140 respectively) from human liver have been raised and tested by Western blotting (Fig. 1B). They showed good specificity, even when the detergent extract of whole liver was loaded. In each case the respective protein itself gave the strongest reaction with the corresponding antibody, though some cross-reactivity between Ab p170 and Ab p140 was observed. A similar pattern of murine proteins was recognized among the whole membrane proteins from ROS cells (data not shown). Furthermore, the specificity of the Abs in the sera made them suitable for the immunofluorescence experiments without further purification. Immunofluorescence staining of cells Examples of immunofluorescence staining of cells are shown in Fig. 2A–D. In all cases investigated
thus far, the antigens were localised to the cell membrane, with no staining in the cytoplasm or ECM. Furthermore, there were no detectable differences between fixed and non-fixed cells. The localisation of 170 and 140 kDa proteins in all kinds of cells is similar to each other, and to that of integrin 1 chain (Fig. 2A–D). Therefore these collagen-binding polypeptides are cell surface molecules which could possibly form a complex with integrin 1 chain. In the ROS cell line, the antigen distribution was highly polarised, most of the receptors being concentrated in several bright spots. At high resolution, it sometimes became apparent that the spots consisted of groups of fine dots, each group located close to the others. The distribution of 250 kDa protein was different and no spots/clusters were observed. Antibodies obtained which were used in immunofluorescence for comparison with integrin 1 chain confirm the membrane localisation of the antigens. Comparison of integrin 2 chain antibodies and Ab p140 The isolation method, membrane localisation and MR similarities suggested to us that the 140 kDa protein is the integrin 2 chain (also see Albelda and Buck, 1990). Comparison of staining of purified collagen binding proteins from liver cells with integrin 2 chain antibodies and Ab p140 was therefore carried out to establish whether this was true. The result of the immunoblot is shown in Fig. 1C. Both antibodies stain bands of the same MR indicating that the 140 kDa protein was almost certainly the integrin 2 chain. DISCUSSION Interactions between cells and ECM are complex and dynamically regulated. Signals from ECM modu-
Cell Biology International, Vol. 25, No. 7, 2001
647
Fig. 2. Immunofluorescence staining of fixed ROS cells. (A) Anti-chicken 1 integrin chain monoclonal antibodies (Sigma); (B) anti-250 kDa protein antibodies; (C) anti-170 kDa protein antibodies; (D) anti-140 kDa protein antibodies.
late cell behaviour, but cells have mechanisms for signal modification according to cell type and state. Integrins have a key role in signalling pathways. The combination of different subunits in integrin heterodimers allows them to possess different ligand specificities. At least three integrin subunits are able to associate with the 1 integrin chain, and the resulting dimeric receptors are able to bind ECM proteins with different specificity. Some of these receptors are highly specific to collagen, laminin and collagen c-peptides, for example 21 and 11 (Hynes, 1992), whereas others bind not only collagen, but a variety of ECM proteins—vitronectin (v1), laminin, fibronectin, VWF (von Willebrandt factor) (11; Buck and Horwitz, 1987; Albelda and Buck, 1990). It has recently been shown that transformed cell lines overexpressing integrin 2 or 1 chains possess the ability to discriminate collagen types, i.e. 2-bearing cells attached better to the collagen 1, while 1-bearing cells prefer collagen type IV. This ability seems to be dependent on the i-domain, which is a feature of the 1 and 2 chains (Kern and Marcantonio, 1988). Our in vitro observations are
in accordance with in vivo data in the case of ROS cells, but in less differentiated osteoblast-like cells (HOS) and fibroblasts, 2 subunits bind preferably to collagens III and IV (Fig. 1, Table 1). Nevertheless, the different ligand specificity for collagen types shown in vivo for integrins of different chain composition strongly supports the conclusion of this current work, and the possibility that the difference in binding specificity is due to the different cations used (Kern and Marcantonio, 1988). The ligand-specificity of integrins is shown to be a cation-dependent. 21 integrin is able to bind collagen in a divalent cation dependent manner (Elices et al., 1991). Cell-type dependence of the ligand specificity is also known: 21 integrin, described as collagen receptor, does not recognise laminin in some cell types, but is able to recognise it in other types (Elices and Hemler, 1989). The chemical modification of the intracellular domains of integrin subunits is one of the mechanisms causing change in ligand-specificity. Phosphorylation of some amino acid residues of integrin chains affects the ligand specificity without changing recep-
648
tor composition, providing a way to regulation dynamically cell–ECM interactions (Hynes, 1992). There are known mechanisms of regulation of integrin specificity, such as alternative splicing of extra- and intra-cellular domains (Fornaro and Languino, 1997) and phosphorylation of amino acid resides in intracellular domain of integrin chains (Takagi et al., 1997; Yokosaki et al., 1996). The proteins described in the Results section (above) are members of the integrin family for the following reasons. In agreement with the known integrin composition as heterodimer each of these polypeptides in the eluate is accompanied by 90 kDa polypeptide (Fig. 1). This MR corresponds to 1 and 3 integrin subunits (Albelda and Buck; 1990, Hynes, 1992). Localisation of the polypeptides revealed by their corresponding antibodies is similar to that of the integrins. No single collagen-type specific polypeptides have been found in focal contacts in osteogenic cells. In primary osteoblast cultures, collagen-binding integrins have not been found localised in focal contacts so far (Pistone et al., 1996). All three polypeptides are localised on the cell surface and 170 and 140 kDa receptor distributions are very similar to that of integrin 1 chain (Fig. 2; Pistone et al., 1996). Both experimental and control antibodies were raised in mice, so we were unable to prove co-localisation by double-labelling. However, control commercial antibodies against the integrin 2 chain proved that it was p140. Since both antibodies against the integrin 2 chain showed some cross-reactivity with the 170 kDa protein, it is possible that 170 kDa protein is also the integrin; 2 chain-containing integrins are able to discriminate between collagen types I and III, and the collagen type recognition by these receptors is cell-type dependent. ACKNOWLEDGEMENTS We are very grateful to Dr Irene Kazdan and Dr Phebe Leboy (Pennsylvania State University) for the generous gift of the ROS cell line, and Dr Geogy Pinaev (Institute of Cytology, RAS) for patience and very helpful discussion. This work was supported by the Russian Fundamental Research Foundation (grant 96-04-49663), Human Genome Program DOE (U.S.A.) and the Wellcome Trust (U.K.). REFERENCES A SV, B CA, 1990. Integrins and other cell adhesion molecules. FASEB J 4: 2868–2880.
Cell Biology International, Vol. 25, No. 7, 2001
B CF, H AF, 1987. Cell surface receptors for extracellular matrix proteins. Ann Rev Cell Biol 3: 179–205. D S, R E, P MD, 1987. A cell surface receptor complex for collagen type I recognizes the Arg-Gly-Asp sequence. J Cell Biol 104: 585–593. E MJ, H ME, 1989. The human integrin VLA-2 is a collagen receptor on some cells and a collagen/laminin receptor on others. Proc Natl Acad Sci USA 86: 9906–9910. E MJ, U LA, H ME, 1991. Receptor function for the integrin VLA-3: fibronectin, collagen, and laminin binding are differentially influenced by Arg-Gly-Asp peptide and divalent cations. J Cell Biol 112: 169–181. F M, L LR, 1997. Alternatively spliced variants: a new view of the integrin cytoplasmic domain. Matrix Bioll 16: 185–193. G E, S D, L RB, 1996. Signaling by epidermal growth factor differentially affects integrinmediated adhesion of tumor cells to extracellular matrix proteins. J Mol Med 74: 609–616. G LC, C SD, K CM, H KJ, L DD, L WJ, 1988. Collagen expression, ultrastructural assembly, and mineralization in cultures of chicken embryo osteoblasts. J Cell Biol 106: 979–989. H RO, 1992. Integrins: versatility, modulation, and signaling in cell adhesion. Cell 69: 11–25. K A, M EE, 1998. Role of the I-domain in collagen binding specificity and activationof the integrins a1b1. J Cell Physiol 176: 634–641. L UK, 1970. Cleavage of structural proteins during assembly of the head of bacteriophage T4. Nature (London) 22: 680–685. M AJ, C CS, I DE, 1997. Demonstration of mechanical connections between integrins, cytoskeletal filaments, and nucleoplasm that stabilize nuclear structure. Proc Natl Acad Sci USA 94: 849–854. M EJ, R K, 1982. Preparation and characterization of the different types of collagen. Meth Enzymol 82: 33–64. M J, B JA, L MA, M K, 1984. Role of anchorin CII, a 31,000 molecular weight membrane protein, in the interaction of chondrocytes with type II collagen. J Cell Biol 98: 1572–1578. P M, S C, F A, S F, D P, S F, Q G, M PC, M P, 1996. Integrin syntesis and utilization in cultured human osteoblasts. Cell Biol Int 20: 471–479. R CD, S A, B MJ, 1995. A hierarchy of ECM-mediated signalling regulates tissue-specific gene expression. Curr Opin Cell Biol 7: 736–747. T J, K T, M J, P J, ML W, T Y, 1997. Changing ligand specificities of avb1 and avb3 integrins by swapping a short diverse sequence of the beta subunit. J Biol Chem 271: 19,794–19,800. T Y, N K, M T, 1996. Differentiation and cell surface expression of transforming grown factor beta receptors are regulated by interaction with matrix collagen in murine osteoblastic cells. J Biol Chem 271: 3938–3984. T H, S T, G J, 1979. Electrophoretic transfer of proteins from polyacrylamide gels to nitrocellulose sheets: procedure and some application. Proc Natl Acad Sci USA 79: 4350–4354. Y Y, M H, C J, S D, 1996. Differential effects of the integrins a9b1, avb3, and avb6 on cell proliferative responses to tenascin. Roles of the beta subunit extracellular and cytoplasmic domains. J Biol Chem 271: 24,144–24,150.
CELL TYPE-DEPENDENT COLLAGEN-TYPE RECOGNITION BY CELL RECEPTORS L. A. SADOFIEV*, N. S. NIKOLAENKO, N. V. KALMYKOVA and O. I. PODGORNAYA Institute of Cytology RAS, Tikhoretsky pr. 4, 194064, St Petersburg, Russia Received 8 January 1999; accepted 25 November 1999
Affinity chromatography of a number of cell types on collagens I and III reveals three proteins with MR of 250, 170 and 140 kDa. These proteins are able to discriminate between types I and III, but not types III and IV. Collagen-type recognition is therefore characteristic for cells of connective tissue origin. Polyclonal antibodies (Ab) raised against 170 and 140 kDa polypeptides and used in immunofluorescence show membrane localisation for both, with their distribution being similar to each other and to the distribution of the integrin 1 chain. Ab p140 and commercial monoclonal antibodies against 2 chain stain a band of the same molecular mass as from purified collagen binding proteins from liver cells, indicating that the 140 kDa protein is probably the 2 integrin chain. The 2 chain containing integrins are therefore able to discriminate collagen types I and III and collagen type recognition by this receptor is cell-type 2000 Academic Press dependent. K: extracellular matrix; collagen; receptors; integrins; connective tissue.
INTRODUCTION Collagens are a family of extracellular matrix (ECM) proteins, strongly affecting cell proliferation and differentiation (Gerstenfeld et al. 1988). Even structurally very similar proteins, such as collagen type I and III, can affect cell behaviour in different ways (Miller and Rhodes, 1982). The mechanism of difference in action could arise from the cell receptors for collagens. The majority of a number of cell receptors known to be capable of binding collagens belong to integrin family, and exist in many tissue types. In some cases, integrins mediate cell-ECM interactions and cell-cell adhesion (Buck and Horwitz, 1987; Albelda and Buck, 1990; Hynes, 1992), but are also responsible for mediating signals between the ECM to the cell nuclei, even in switching between proliferative and differentiation modes of cell activity (Dedchar et al., 1987; Maniotis et al., 1997; Roskelley et al., 1995; Takeuchi et al., 1996). Integrins are heterodimers formed by one -type subunit and one -type subunit. More than one kind of -subunit can usually combine with each kind of -subunit, forming receptors with different *To whom correspondence should be addressed. 1065–6995/01/070643+06 $35.00/0
functions and ligand specificity (Buck and Horwitz, 1987; Albelda and Buck, 1990; Hynes, 1992). Integrins are the main type of collagen receptors, and data exists which show their ability to recognize different collagen types (Tuckwell and Humphries, 1996). The ability to recognize collagen types has been shown for the other group of collagen receptors, the anchorins. Anchorins are chondrocyte receptors of the annexin family, which have relatively low molecular weights (40 kDa) and can selectively bind cartilage collagen II (Mollenchauer et al., 1984). The aim of the present work was to determine whether there are receptors capable of recognising collagen types of cells other than chondrocytes. The principal method employed involves a comparative analysis of cell membrane proteins binding to affinity columns containing collagen types I, III, IV. Three polypeptides of approximate MR 250, 170, 140 kDa were able to recognize collagens I and III, an ability only found in connective tissue cells, such as fibroblasts and osteoblasts. Polyclonal antibodies raised against these polypeptides localised them to the cell surface in a very similar pattern to that of the integrin -chain. Immunoblots of 2001 Academic Press
644
collagen-binding proteins from liver cells with antibodies against 140 kDa protein and commercial monoclonal antibodies against integrin 2 chain revealed the same zone. MATERIALS AND METHODS Affinity columns Collagens for affinity columns were prepared from human placenta by pepsin digestion (types I, III, IV), or from rat skin without pepsin treatment (types I and III), according to the standard procedure of Miller and Rhodes (1982). The purity of collagens obtained was >90% according to SDS– PAGE (Laemmli, 1970), with only some crosscontamination of collagen types. Coupling to BrCN-activated Sepharose 4B (Pharmacia, Sweden) was carried out according to the manufacturer’s recommendations. We used 10 mg of protein/ml of wet Sepharose; the column’s bead volume was 0.2 ml for analytical purposes and 1 ml for preparative purposes. Sources of membranes The following cell lines were obtained from the Cell Culture Bank, Institute of Cytology, RAS. Cell lines of connective tissue origin: immortalised rat fibroblasts E1A-12S, diploid fibroblasts from rat lung and muscle, rat osteosarcoma ROS 17/2.8, human osteosarcoma HOS. Cell lines of epithelial origin: human epidermoid carcinoma A431, human epithelioid carcinoma HeLa, rat pituitary tumour GH3. All cells were cultured on plastic dishes in DME-F12 (3:1) medium in the presence of 10% fetal calf serum. Membrane extracts of human embryo liver were used to prepare isolated receptors. Isolation and characterization of receptors All procedures were carried out at 4C. Cells were removed from dishes by Versene solution without trypsin. Cells were disrupted by hypotonic shock in 1/10 of normal strength PBS. Membrane vesicles were collected by centrifugation at 4000g, and membrane proteins were extracted in 2 ml TBS containing 0.5% Triton X-100, 5 m CaCl2, 5 m MgCl2, 0.15 NaCl (referred to as receptor extraction buffer, or REB). Extracts were applied to affinity columns after the removal of insoluble material by centrifugation at 4000g. Human embryonic liver was homogenised in Tris–HCl EDTA buffer at pH 7.5. The membrane
Cell Biology International, Vol. 25, No. 7, 2001
fraction was obtained by ultra-centrifugation in sucrose gradient and extraction with REB. Extracts were passed through an albumin–Sepharose column to remove proteins which bound nonspecifically, and the flow-through was loaded on to collagen–Sepharose columns. The columns were washed with the same buffer until no proteins were detected in the eluate, after which they were washed with 10 bed-volumes of the same buffer without detergents. Bound proteins were eluted with TBS buffer containing 10 m of octylglucoside, 0.5 NaCl, and 50 m EDTA. Proteins were separated using 7.5% gels by SDS–PAGE (Laemmli, 1970), and analysed following silver-staining. Antibodies Commercial monoclonal anti-chicken integrin 1 chain antibodies were obtained from Sigma (St Louis, MO, U.S.A.) and anti human integrin 2 chain from DAKO (Norway). Collagen-binding proteins of 250, 170 and 140 kDa were purified on collagen affinity columns from Triton X-100 extracts of whole human liver, separated on SDS–PAGE, and the protein zones cut out and injected into mice. Three injections of approximately 5 g of protein per injection were given at weekly intervals. Serum was taken 1 week after the last injection. Immunoblotting Immunoblotting was carried out after SDS–PAGE electrophoresis in 7.5% gel according to standard procedure for nitrocellulose membranes (Millipore, U.S.A.; Towbin et al., 1979). The secondary antibodies were biotinylated goat anti-mouse (Sigma), followed by streptavidin–alkaline phosphatase treatment and NBT-BCIP stain, according to the manufacturer’s recommendations. Immunofluorescence Cells in cold PBS (3000–5000 cells in 0.01 ml) were placed on glass coverslips or on multitest slides (Flow, U.K.) coated with D-polylysine (MR >70 kDa; Sigma). After incubating for 30– 60 min at 4C, unattached cells were removed, and those attached were fixed at 4C formaldehyde or paraformaldehyde (Merck, Germany) in PBS containing MgSO4 (PBS-Mg) or processed without fixation. All antibodies were diluted with PBS–BSA buffer. The working dilutions for antiserum against the 250, 170 and 140 kDa proteins, and against integrins 1
Cell Biology International, Vol. 25, No. 7, 2001
645
Fig. 1. (A) SDS–PAGE of collagen-binding proteins. Elution with EDTA-NaCl buffer. Eluates from collagen columns were concentrated by lyophilization and separated on a 7.5% gel in the presence of -mercaptoethanol. Proteins were visualised by silver staining. The collagen type used for protein isolation is indicated by roman numbers below each line. Brackets indicate the position of proteins with MR70 kDa, not completely separated. Group of lines: (1) murine fibroblasts; (2) relatively poorly differentiated human osteosarcoma cell line, HOS; (3) highly differentiated rat osteosarcoma cell line, ROS; (4) human embryo liver cells. (B) Testing of antibodies raised against 250, 170 and 140 kDa proteins by immunoblotting. The antibody tested indicated at the bottom of each line. For testing, unpurified membrane extract of human embryo liver cells was used. Antibodies against 170 and 140 kDa proteins show good specificity, although some cross-reaction between 170 and 140 kDa proteins was observed. The specificity of antibodies against 250 kDa is poor. Dilution of antisera; p250–1:50, p170–1:50, p140–1:100. (C) Immunoblot staining of collagen-binding proteins with anti-human integrin 2 chain monoclonal antibodies (Dako) (1) and our anti-140 kDa protein antibodies (2). Collagen-binding proteins from human embryo liver were purified on collagen I affinity column. Dilution of both antibodies 1:1000.
chain, were 1:50, and secondary antibodies (RAMFITC, Sigma) were diluted 1:100. Cells were incubated with antibodies for 1 h at 4C (unfixed samples), or at room temperature (fixed samples). The slides were viewed and photographed with an Axioskop microscope (Ziess, Germany).
RESULTS Comparative analysis of collagen receptors Affinity chromatography of a range of cell types on collagen columns identified at least three high MR polypeptides of 250, 170 and 140 kDa which were able to discriminate collagen types I and III, but not types III and IV, under the conditions of the affinity chromatography which was used. These have been referred to as ‘collagen receptors’,
according to the isolation procedure. The comparative analysis of collagen I, III, and IV binding receptors from cells lines of different origins is shown in Fig. 1A. (except collagen IV) and summarised in Table 1. No differences between patterns of collagen III and collagen IV binding proteins were found. The ability of the polypeptides to recognize collagen types in vitro was dependent on the buffer ionic strength and type of cells. In buffers lacking NaCl, all three polypeptides bound the three types of collagen with similar affinities as in fibroblast cells, and all three polypeptides bound collagen I, but not collagen III, in ROS cells. But adjustment of NaCl concentration upwards to the physiological level of 0.15 completely inhibited the binding of the 250 and 140 kDa polypeptides with collagen I in both cases. No salt concentration effect was observed in the case of liver cells.
646
Cell Biology International, Vol. 25, No. 7, 2001
Table 1. Collagen-type recognition by high molecular weight cell receptors Fibroblasts I
250 170 140
* *
III
+ +
Osteogenic cells IV
+ +
HOS
Liver cells ROS
I
III
IV
I
III
IV
+
+ +
+ +
* + *
N/A N/A N/A
I
III
IV
+ + +
+ + +
+ + +
Protein present in eluate (+); protein is absent (); not tested (N/A); *effect is ionic strengthdependent.
Collagen binding to the polypeptides seemed to be dependent on divalent cations. The 250 and 140 kDa polypeptides were completely removed from collagen affinity columns by EDTA without an increase in the ionic strength of the buffer. The 170 kDa polypeptide from ROS cells bound to any of the collagens both in the presence or absence of EDTA, and could be eluted only by increasing the ionic strength up to 0.5 NaCl. A zone, possibly two band, with MR in the range of 90 kDa appears in the SDS–PAGE of eluates, and there is a strong correlation with the appearance of three polypeptides (Fig. 1A, arrowheads). Epithelioid cell line membrane extracts showed no high MR proteins by affinity chromatography, even after a 10-fold increase in loading (data not shown). Integrins are shown to be in A431 cells (Genersch et al., 1996), so the inability of the collagen receptors to attach to the collagen types used is probably due to some features of the receptors. Antibodies Mouse polyclonal antiserums against polypeptides 250, 170 and 140 kDa (Ab p250, Ab p170 and Ab p140 respectively) from human liver have been raised and tested by Western blotting (Fig. 1B). They showed good specificity, even when the detergent extract of whole liver was loaded. In each case the respective protein itself gave the strongest reaction with the corresponding antibody, though some cross-reactivity between Ab p170 and Ab p140 was observed. A similar pattern of murine proteins was recognized among the whole membrane proteins from ROS cells (data not shown). Furthermore, the specificity of the Abs in the sera made them suitable for the immunofluorescence experiments without further purification. Immunofluorescence staining of cells Examples of immunofluorescence staining of cells are shown in Fig. 2A–D. In all cases investigated
thus far, the antigens were localised to the cell membrane, with no staining in the cytoplasm or ECM. Furthermore, there were no detectable differences between fixed and non-fixed cells. The localisation of 170 and 140 kDa proteins in all kinds of cells is similar to each other, and to that of integrin 1 chain (Fig. 2A–D). Therefore these collagen-binding polypeptides are cell surface molecules which could possibly form a complex with integrin 1 chain. In the ROS cell line, the antigen distribution was highly polarised, most of the receptors being concentrated in several bright spots. At high resolution, it sometimes became apparent that the spots consisted of groups of fine dots, each group located close to the others. The distribution of 250 kDa protein was different and no spots/clusters were observed. Antibodies obtained which were used in immunofluorescence for comparison with integrin 1 chain confirm the membrane localisation of the antigens. Comparison of integrin 2 chain antibodies and Ab p140 The isolation method, membrane localisation and MR similarities suggested to us that the 140 kDa protein is the integrin 2 chain (also see Albelda and Buck, 1990). Comparison of staining of purified collagen binding proteins from liver cells with integrin 2 chain antibodies and Ab p140 was therefore carried out to establish whether this was true. The result of the immunoblot is shown in Fig. 1C. Both antibodies stain bands of the same MR indicating that the 140 kDa protein was almost certainly the integrin 2 chain. DISCUSSION Interactions between cells and ECM are complex and dynamically regulated. Signals from ECM modu-
Cell Biology International, Vol. 25, No. 7, 2001
647
Fig. 2. Immunofluorescence staining of fixed ROS cells. (A) Anti-chicken 1 integrin chain monoclonal antibodies (Sigma); (B) anti-250 kDa protein antibodies; (C) anti-170 kDa protein antibodies; (D) anti-140 kDa protein antibodies.
late cell behaviour, but cells have mechanisms for signal modification according to cell type and state. Integrins have a key role in signalling pathways. The combination of different subunits in integrin heterodimers allows them to possess different ligand specificities. At least three integrin subunits are able to associate with the 1 integrin chain, and the resulting dimeric receptors are able to bind ECM proteins with different specificity. Some of these receptors are highly specific to collagen, laminin and collagen c-peptides, for example 21 and 11 (Hynes, 1992), whereas others bind not only collagen, but a variety of ECM proteins—vitronectin (v1), laminin, fibronectin, VWF (von Willebrandt factor) (11; Buck and Horwitz, 1987; Albelda and Buck, 1990). It has recently been shown that transformed cell lines overexpressing integrin 2 or 1 chains possess the ability to discriminate collagen types, i.e. 2-bearing cells attached better to the collagen 1, while 1-bearing cells prefer collagen type IV. This ability seems to be dependent on the i-domain, which is a feature of the 1 and 2 chains (Kern and Marcantonio, 1988). Our in vitro observations are
in accordance with in vivo data in the case of ROS cells, but in less differentiated osteoblast-like cells (HOS) and fibroblasts, 2 subunits bind preferably to collagens III and IV (Fig. 1, Table 1). Nevertheless, the different ligand specificity for collagen types shown in vivo for integrins of different chain composition strongly supports the conclusion of this current work, and the possibility that the difference in binding specificity is due to the different cations used (Kern and Marcantonio, 1988). The ligand-specificity of integrins is shown to be a cation-dependent. 21 integrin is able to bind collagen in a divalent cation dependent manner (Elices et al., 1991). Cell-type dependence of the ligand specificity is also known: 21 integrin, described as collagen receptor, does not recognise laminin in some cell types, but is able to recognise it in other types (Elices and Hemler, 1989). The chemical modification of the intracellular domains of integrin subunits is one of the mechanisms causing change in ligand-specificity. Phosphorylation of some amino acid residues of integrin chains affects the ligand specificity without changing recep-
648
tor composition, providing a way to regulation dynamically cell–ECM interactions (Hynes, 1992). There are known mechanisms of regulation of integrin specificity, such as alternative splicing of extra- and intra-cellular domains (Fornaro and Languino, 1997) and phosphorylation of amino acid resides in intracellular domain of integrin chains (Takagi et al., 1997; Yokosaki et al., 1996). The proteins described in the Results section (above) are members of the integrin family for the following reasons. In agreement with the known integrin composition as heterodimer each of these polypeptides in the eluate is accompanied by 90 kDa polypeptide (Fig. 1). This MR corresponds to 1 and 3 integrin subunits (Albelda and Buck; 1990, Hynes, 1992). Localisation of the polypeptides revealed by their corresponding antibodies is similar to that of the integrins. No single collagen-type specific polypeptides have been found in focal contacts in osteogenic cells. In primary osteoblast cultures, collagen-binding integrins have not been found localised in focal contacts so far (Pistone et al., 1996). All three polypeptides are localised on the cell surface and 170 and 140 kDa receptor distributions are very similar to that of integrin 1 chain (Fig. 2; Pistone et al., 1996). Both experimental and control antibodies were raised in mice, so we were unable to prove co-localisation by double-labelling. However, control commercial antibodies against the integrin 2 chain proved that it was p140. Since both antibodies against the integrin 2 chain showed some cross-reactivity with the 170 kDa protein, it is possible that 170 kDa protein is also the integrin; 2 chain-containing integrins are able to discriminate between collagen types I and III, and the collagen type recognition by these receptors is cell-type dependent. ACKNOWLEDGEMENTS We are very grateful to Dr Irene Kazdan and Dr Phebe Leboy (Pennsylvania State University) for the generous gift of the ROS cell line, and Dr Geogy Pinaev (Institute of Cytology, RAS) for patience and very helpful discussion. This work was supported by the Russian Fundamental Research Foundation (grant 96-04-49663), Human Genome Program DOE (U.S.A.) and the Wellcome Trust (U.K.). REFERENCES A SV, B CA, 1990. Integrins and other cell adhesion molecules. FASEB J 4: 2868–2880.
Cell Biology International, Vol. 25, No. 7, 2001
B CF, H AF, 1987. Cell surface receptors for extracellular matrix proteins. Ann Rev Cell Biol 3: 179–205. D S, R E, P MD, 1987. A cell surface receptor complex for collagen type I recognizes the Arg-Gly-Asp sequence. J Cell Biol 104: 585–593. E MJ, H ME, 1989. The human integrin VLA-2 is a collagen receptor on some cells and a collagen/laminin receptor on others. Proc Natl Acad Sci USA 86: 9906–9910. E MJ, U LA, H ME, 1991. Receptor function for the integrin VLA-3: fibronectin, collagen, and laminin binding are differentially influenced by Arg-Gly-Asp peptide and divalent cations. J Cell Biol 112: 169–181. F M, L LR, 1997. Alternatively spliced variants: a new view of the integrin cytoplasmic domain. Matrix Bioll 16: 185–193. G E, S D, L RB, 1996. Signaling by epidermal growth factor differentially affects integrinmediated adhesion of tumor cells to extracellular matrix proteins. J Mol Med 74: 609–616. G LC, C SD, K CM, H KJ, L DD, L WJ, 1988. Collagen expression, ultrastructural assembly, and mineralization in cultures of chicken embryo osteoblasts. J Cell Biol 106: 979–989. H RO, 1992. Integrins: versatility, modulation, and signaling in cell adhesion. Cell 69: 11–25. K A, M EE, 1998. Role of the I-domain in collagen binding specificity and activationof the integrins a1b1. J Cell Physiol 176: 634–641. L UK, 1970. Cleavage of structural proteins during assembly of the head of bacteriophage T4. Nature (London) 22: 680–685. M AJ, C CS, I DE, 1997. Demonstration of mechanical connections between integrins, cytoskeletal filaments, and nucleoplasm that stabilize nuclear structure. Proc Natl Acad Sci USA 94: 849–854. M EJ, R K, 1982. Preparation and characterization of the different types of collagen. Meth Enzymol 82: 33–64. M J, B JA, L MA, M K, 1984. Role of anchorin CII, a 31,000 molecular weight membrane protein, in the interaction of chondrocytes with type II collagen. J Cell Biol 98: 1572–1578. P M, S C, F A, S F, D P, S F, Q G, M PC, M P, 1996. Integrin syntesis and utilization in cultured human osteoblasts. Cell Biol Int 20: 471–479. R CD, S A, B MJ, 1995. A hierarchy of ECM-mediated signalling regulates tissue-specific gene expression. Curr Opin Cell Biol 7: 736–747. T J, K T, M J, P J, ML W, T Y, 1997. Changing ligand specificities of avb1 and avb3 integrins by swapping a short diverse sequence of the beta subunit. J Biol Chem 271: 19,794–19,800. T Y, N K, M T, 1996. Differentiation and cell surface expression of transforming grown factor beta receptors are regulated by interaction with matrix collagen in murine osteoblastic cells. J Biol Chem 271: 3938–3984. T H, S T, G J, 1979. Electrophoretic transfer of proteins from polyacrylamide gels to nitrocellulose sheets: procedure and some application. Proc Natl Acad Sci USA 79: 4350–4354. Y Y, M H, C J, S D, 1996. Differential effects of the integrins a9b1, avb3, and avb6 on cell proliferative responses to tenascin. Roles of the beta subunit extracellular and cytoplasmic domains. J Biol Chem 271: 24,144–24,150.