Cell–matrix interactions: influence of noncollagenous proteins from dentin on cultured dental cells

Several experimental approaches have demonstrated the importance of the extracellular matrix in the mediation of epitheliomesenchymal interactions during embryonic development (Reddi, 1984). However, the nature and the mechanisms of the cell-matrix interactions that are involved in the control of organogenesis and cell differentiation are still under investigation (for reviews, see Yamada, 1983; von der Mark et al. 1984; Bernfield, Banerjee, Koda & Rapraeger, 1984). Different models have been used for these studies, including the tooth (Thesleff, Lehtonen & Saxen, 1978; Ruch et al. 1982; Slavkin, 1982), which is representative of such interactions.

The tooth germ consists of two interacting tissues: the enamel organ and the dental papilla. At the epitheliomesenchymal junction, the ectomesenchymal cells in contact with the basement membrane differentiate into functional odontoblasts secreting the predentin. Cells of the inner dental epithelium differentiate into functional ameloblasts, which secrete the enamel. However, preodontoblasts and preameloblasts do not differentiate spontaneously. In order to differentiate, preodontoblasts have to interact with the mesenchymal face of a stage-specific basement membrane (Osman & Ruch, 1981) and preameloblasts require interaction with the predentin (Ruch, Karcher-Djuricic & Gerber, 1972).

Attempts have been made to promote the terminal differentiation of these cells using artificial substrates made of collagens (Thesleff, 1978; Lesot et al. 1985), noncollagenous glycoproteins (Lesot et al. 1985) and glycosaminoglycans (Ruch, 1985). Up to now, the polarization of odontoblast-like cells could only be obtained in the presence of either hyaluronic acid or chondroitin sulphate (Ruch, 1985). Recently, the cytological and functional differentiation of ameloblasts has been observed when cells from the inner dental epithelium were cultured in association with a specific region of isolated dental matrices (Karcher-Djuricic, Staubli, Meyer & Ruch, 1985).

In the present study, we report experiments in which enamel organs and dental papillae have been cultured on Millipore filters coated with noncollagenous matrix proteins extracted from rabbit incisor dentin. The behaviour of the cultured dental cells was examined at the histological and cytological levels.

Dentin was prepared from incisors extracted from New Zealand White rabbits as previously described (Smith & Leaver, 1981) and powdered in a percussion mill cooled with liquid nitrogen.

The powdered dentin was demineralized for 10 days at 4°C with 10% EDTA, pH 7·2, containing 10 mM-N-ethylmaleimide and 5 mM-phenylmethylsulphonyl fluoride as protease inhibitors (Smith & Smith, 1984). The demineralizing solution was changed every 48 h and the EDTA-soluble fraction in the supernatant collected after centrifugation. The combined EDTA-soluble fractions were exhaustively dialysed against distilled water, reduced in volume by vacuum dialysis and dialysed against two changes of 0·05M-tris-HCl, pH7·2, prior to DEAE-cellulose chromatography.

The total EDTA-soluble fraction from demineralization of the dentin was applied to a column (18x1·0 cm) of DEAE-cellulose equlibrated with 0·05 M-tris-HCl, pH 7·2, and after elution of the pregradient material with the same buffer, elution was continued with a salt gradient of 0·0–7M-NaCl in 0·05M-tris-HCl, pH7·2. Column effluents were monitored as previously described (Smith & Leaver, 1979). The contents of the tubes were pooled into fractions 1–5, designated as in Fig. 1, and dialysed against distilled water and lyophilized.

After washing with distilled water, the demineralized EDTA-insoluble dentin residue was incubated with 1000 units of collagenase (Sigma type VII from Clostridium histolyticum, 1600 units per mg) in 0·025M-tris-HCl·0·3M-calcium acetate buffer, pH7·2, at 37°C. After two days, a further 500 units of collagenase were added and the incubation continued for a further three days. The digested material was then centrifuged at 12000g for 20 min to separate the insoluble collagenase-released residue (ICR). The soluble collagenase-released fraction in the supernatant (CRF, ‘fraction 6’) was decanted off, dialysed against distilled water and lyophilized. The ICR fraction was washed with distilled water and extracted with 8M-urea at 4 °C until the absorption of the extracts at 280nm became minimal. The extracts were dialysed against distilled water and lyophilized (ICR-U, ‘fraction 7’). Protease inhibitors (10mM-N-ethylmaleimide, 2 mM-phenylmethylsulphonyl fluoride, 25mM-EDTA) were included at all stages to prevent artefactual degradation.

After hydrolysis in 6 N-HCI, the amino acid compositions of the hydrolysates were determined using an LKB 4400 autoanalyser.

Samples of the fractions were analysed for hexose (Dubois et al. 1956), hexosamine (Blumenkrantz & Asboe-Hansen, 1976), uronic acid (Blumenkrantz & Asboe-Hansen, 1973) and sialic acid (Skoza & Mohos, 1976).

Noncollagenous protein fractions from rabbit incisor dentin were separated by electrophoresis in 0·1 % sodium dodecylsulphate (SDS) on a linear 6/10% polyacrylamide gradient gel (PAGE) using the buffer system described by Laemmli (1970). The proteins were stained by the silver procedure (Oakley, Kirsch & Morris, 1980).

First lower molars of Swiss mouse embryos were used. Tooth germs were removed on day 18 (vaginal plug = day 0) of gestation. The teeth were staged according to morphological features (size, form, vascularization, etc.). These teeth contain dividing preameloblasts and preodontoblasts and the first postmitotic odontoblasts (Karcher-Djuricic et al. 1985).

Enamel organs and dental papillae were isolated from tooth germs enzymically dissociated with 1% trypsin (Difco 1:250) in Hanks’ balanced salt solution as previously described (Ruch, Karcher-Djuricic & Thiebold, 1976). Soybean trypsin inhibitor, 1% in phosphate-buffered saline (PBS), was used to stop the activity of the enzyme.

Day-18 isolated enamel organs and dental papillae were cultured for 4 days on Millipore filters (150 μm thickness and 45 μm pore size) coated or noncoated with 100–150 μg of noncollagenous proteins (applied as solution and dried) isolated as above. The culture medium was Dulbecco’s Minimum Essential Medium (MEM) supplemented with 10% foetal calf serum and was changed every two days. Cultures were performed in a humidified incubator at 37°C in an atmosphere of 5 % CO2 in air.

The tissues were fixed with Bouin’s fixative and paraffin-embedded specimens were cut in 5 μm thick serial sections and stained with Mallory Alun haematoxylin.

Specimens were fixed for 60min at 4°C in a 2% glutaraldehyde solution buffered with 0·1 M cacodylate, pH7·3, and postfixed for 30 min in 1% osmium tetroxide in the same buffer. The specimens were embedded in Epon 812 and processed for electron microscopy.

Seven noncollagenous protein fractions were prepared from rabbit incisor dentin and used as substrates to coat Millipore filters. For each protein fraction, 60–100 dental papillae and enamel organs were cultured on these as well as uncoated control Millipore filters and observed after 4 days in culture. The protein fractions used in these experiments represented three independent isolations from rabbit dentin.

Ion-exchange chromatography on DEAE-cellulose of the EDTA-solubilized material from dentin resulted in fractions 1–5 (Fig. 1). Examination of these fractions on SDS-PAGE followed by silver staining confirmed that the individual peaks demonstrated heterogeneity in their protein content (Fig. 2). Fractions 6 and 7 were extracted, respectively, after collagenase and after 8M-urea treatment of the EDTA-insoluble residue and also demonstrated protein heterogeneity on electrophoresis. The amino acid and sugar compositions of those fractions (2, 5 and 6) producing morphological and cytological changes in the cultured dental cells reported below are shown in Table 1. Appreciable amounts of the anionic amino acids, aspartic acid, glutamic acid and serine are present, although more anionic fractions are present in dentin. The carbohydrate content indicates the presence of glycoproteins in the fractions and the hexosamine and uronic acid contents of fraction 5 are in accord with the presence of glycosaminoglycans in this fraction.

After 4 days in culture, the mesenchymal cells in contact with Millipore filters coated with rabbit dentin protein fraction 2 or fraction 6 demonstrated important changes in their morphology (Fig. 3B) as compared with the same cells cultured in contact with noncoated Millipore filters (Fig. 3A). Cells in contact with Millipore filters coated with one of these two substrates had an epithelial arrangement and appeared elongated and polarized (Fig 3B). The nuclei migrated to the basal pole of the cells (Figs 4C, 5A). When compared with control cultures, these cells demonstrated the widening of the cisternae of the rough endoplasmic reticulum (Fig. 4B) and distended Golgi cisternae (Figs 4C, 5B). Coated vesicles originated from ergastoplasmic cisternae and fused with the Golgi apparatus which produced secretory granules (Fig. 5B,C). No excretion of granules was observed and there was no deposition of an extracellular matrix (Fig. 4C). 55 % of the cultured dental papillae demonstrated this behaviour. After 4 days in culture, in control experiments, the corresponding cells neither polarized nor elongated (Fig. 4A).

Mesenchymal cells grown in contact with Millipore filters coated with protein fractions 1, 3, 4, 5 or 7 were similar at the histological level to control cultures.

In control cultures where cells from the inner dental epithelium were associated with the filter, an epithelial arrangement of the cells could been seen after 4 days (Fig. 3C, 6A). The nuclei did not demonstrate any preferential localization (Fig. 6A).

In 40% of the cultured enamel organs, cells of the inner dental epithelium demonstrated morphological changes including a slight elongation and also a preferential localization of the nuclei at the basal pole of these cells (Figs 3D, 6B) when filters coated with protein fractions 5 or 6 were used as substrates. However, the elongation of these cells was much less than that of functional ameloblasts in normal teeth. Furthermore, the mitochondria did not migrate to the infranuclear compartment as they normally do during the terminal differentiation of ameloblasts. However, epithelial cells in contact with filters coated with protein fractions 5 or 6 displayed condensation of their chromatin and a development of the ergastoplasm which indicated an increased metabolic activity. In these conditions, as well as for control cultures, there was no evident reconstitution of a basal lamina in between epithelial cells and coated filters.

Reciprocal epitheliomesenchymal interactions control the terminal differentiation of odontoblasts and ameloblasts (Kollar & Baird, 1970; Slavkin, 1978; Thesleff & Hurmerinta, 1981; Ruch et al. 1982). Since these interactions are matrix mediated (Osman & Ruch, 1981; Karcher-Djuricic et al. 1985), attempts have been made to understand the nature and mechanism(s) of cell-matrix interactions. Several artificial substrates have been tested for their ability to replace the physiological requirements for odontoblast and ameloblast differentiation including different collagen types (Thesleff, 1978; Lesot et al. 1985), noncollagenous glycoproteins such as fibronectin and laminin (Lesot et al. 1985) and glycosaminoglycans (Ruch, 1985).

In the present study, substrates comprising noncollagenous proteins extracted from rabbit incisor dentin have been tested for their effects on dental epithelial and mesenchymal cells in culture. Previous studies have shown that noncollagenous proteins extracted from rat dentin possessed bone and cartilage morphogenetic activity (Bang & Urist, 1967; Yoemans & Urist, 1967; Huggins, Wiseman & Reddi, 1970). Structural analysis has shown that three of these proteins could be solubilized after collagenase treatment, two of which were acidic proteins and the third one was a serine-rich phosphoprotein (Butler, Mikulski & Urist, 1977).

The noncollagenous proteins from rabbit incisor dentin isolated in the present study were prepared essentially as previously described (Smith & Leaver, 1979) except that protease inhibitors were included at all stages to prevent artefactual degradation. The fractions 1–5 from DEAE-cellulose chromatography of the EDTA-soluble material have been previously described as less-acidic glycoproteins (fractions 2 and 3), anionic glycoproteins (fraction 4) and a proteoglycan-containing fraction (fraction 5), whilst fraction 1 appears to include collagenous material (Smith & Leaver, 1979). Fraction 6 represents noncollagenous proteins tightly associated, possibly covalently, with the collagen of the tissue matrix and was only released after digestion of the demineralized dentin residue (Smith, Price & Leaver, 1979). The remaining insoluble residue after collagenase digestion could be partially solubilized by 8 M-urea extraction and has been shown to include proteins of the high aspartic acid and serine type similar to the anionic phosphoproteins (Richardson, Beagle, Butler & Munksgaard, 1977) and also acid structural proteins (Smith et al. 1979). The composition of the fractions 2, 5 and 6 reported here to show effects on the cultured dental cells are similar to those described in these earlier studies, although the hexose of fraction 5 indicates that glycoprotein material is present in addition to the proteoglycan material.

When compared to control cultures, cells from the inner dental epithelium cultured in contact with Millipore filters coated with protein fractions 5 or 6 demonstrated slight changes in the cell shape (elongation) and a preferential localization of the nuclei at the basal pole. However, it was not possible to observe the differentiation of functional ameloblasts. When enamel organs were grown on noncoated Millipore filters there was no reconstitution of a basal lamina. This observation agrees with previous data from Thesleff et al. (1978). However, when enamel organs were cultured on agar or plasma, basal lamina was deposited after a few hours while no polarization occurred (Osman & Ruch, 1980). The presence of protein fractions 5 or 6 led to modification in the cell shape but not to the deposition of a new basal lamina. The concomitant existence of polarized mammalian ameloblasts and a continuous basal lamina has been documented for tooth germs cultured in a serum-less chemically defined medium (Slavkin et al. 1983). In physiological conditions, the degradation of the basal lamina precedes and might be a prerequisite for the terminal differentiation of ameloblasts. Karcher-Djuricic et al. (1985) have shown that functional ameloblasts differentiated when enamel organs from stage 18 were cultured for 4 days in contact with the epithelial side of predentin of isolated dentin matrices. The extraction of these isolated dental matrices, either with 4 M-guanidine-EDTA or 0· 5M-acetic acid, resulted in the loss of the morphogenetic properties. Collagen type I, a major constituent of predentin-dentin, does not seem to have morphogenetic properties since isolated molecules of type I collagen are not able to promote the differentiation of ameloblasts. However, protein fraction 6 (and possibly fraction 5) is tightly associated with collagen type I and the possible physiological properties of protein fractions 5 and 6 might depend on interactions with collagen as far as their configuration and spatial distribution is concerned. Such a possibility is in agreement with previous observations showing that bone matrix possessed inductive properties which (1) were lost after 4M-guanidine extraction and (2) could be restored after reconstitution of the matrix (Sampath & Reddi, 1981; Sampath, Nathanson & Reddi, 1984). It is possible that intimate relationships exist between several components of the matrix that are important for morphogenesis and cytodifferentiation. To investigate this, we propose to use new substrates including both type I collagens and noncollagenous protein fractions 5 or 6 in order to see whether it might improve the differentiation process of ameloblasts.

When dental papillae are cultured on noncoated Millipore filters, elongated and polarized cells are never observed. On the other hand, mesenchymal cells in contact with Millipore filters coated with protein fractions 2 or 6 elongated, polarized and demonstrated widening of cisternae of the rough endoplasmic reticulum and of the Golgi, suggesting a high metabolic activity. However, these cells did not deposit an extracellular matrix. Polarized, but nonfunctional, odontoblasts have been recently reported by Sakakura, lida, Ishizeki & Nawa (1984) when developing mouse tooth germs were treated with calcitonin in vitro. In physiological conditions, the terminal differentiation of odontoblast requires interaction of the differentiating cells with a specific basement membrane (Osman & Ruch, 1981). Immunochemical studies are in progress in order to determine if proteins from fractions 2 and 6 are also constituents of this extracellular matrix.

The elongation and polarization of mesenchymal cells interacting with protein fractions 2 and 6 as well as the more discrete effects of protein fractions 5 and 6 on cultured epithelial cells are related to changes in the organization and, or, the activity of the cytoskeleton. A direct effect of matrix molecules such as collagens, fibronectin or laminin on the activity of the cytoskeleton has been reported for different cell types in culture (Sugrue & Hay, 1981; Tomasek & Fugiwara, 1982) and these effects were suggested to be mediated by membrane receptors (Sugrue & Hay, 1982). Membrane proteins interacting specifically with collagens (Chiang & Kang, 1982; Kurkinen, Taylor, Garrels & Hogan, 1984; Mollenhauer & von der Mark, 1983), fibronectin (Hughes, Butters & Aplin, 1981; Pytela, Pierschbacher & Ruoslahti, 1985; Lesot et al. 1985) or laminin (Malinoff & Wicha, 1983; Lesot, Kuhl & von der Mark, 1983) have been identified. Similar investigations will have to be performed with the noncollagenous proteins tested here in order to know whether these interact directly or indirectly with the plasma membranes of cultured dental cells.

We are grateful to Ms S. Brewer for performing the amino acid analyses. The authors also would like to thank Dr Michel Fabre for his helpful comments on the ultrastructural observations and Dr Vera Karcher-Djuricic, who performed the tissue cultures. We are also grateful to Mr Alain Ackermann for technical support. This work was supported by grants from the CNRS Jeune Equipe no. 034329, INSERM no. 854021 and Fondation pour la Recherche Medicale.