Regulated expression of fibronectin, laminin and related integrin receptors during the early chondrocyte differentiation

Reduction in the intercellular spaces and formation of extensive cell-cell contacts between mesenchymal pre-chondrogenic cells, i.e. cell condensation, are the earliest morphogenetic events involved in the development of limb bones. They are necessary for the position-dependent morphogenesis of skeletal structures (Maini and Solursh, 1991). The first chondrogenesis area is found in the central portion of the blastemata and extends peripherally; the periphery of the condensed area is surrounded by elongated fibroblasts (Thorogood and Hinchliffe, 1975). It has been suggested that cell condensation may be mediated by different factors, including the cell adhesion molecules N-CAM and N-cadherin (Jiang et al., 1993; Oberlender and Tuan, 1994; Tavella et al., 1994), or by extracellular matrix (ECM) components. Among the latter, hyaluronic acid (HA), fibronectin (FN), laminin (LN), tenascin, type I collagen and a large chondroitin sulfate proteoglycan (Dessau et al., 1980; Tomaseck et al., 1982; Mackie et al., 1987; Solursh and Jensen, 1988; Frenz et al., 1989a,b; Pacifici et al., 1993; Downie and Newman, 1994; Maleski and Knudson, 1996) are the main molecules that play

a role in the perspective condensation area or in the early stages of condensation. During chondrogenesis, the pattern of ECM components produced by the differentiating chondrocytes, and consequently the organization of the ECM itself, changes depending on the region of the blastemata considered, i.e. central differentiating vs peripheral non-differentiating areas (Tacchetti et al., 1987; Castagnola et al., 1988; Cancedda et al., 1995). It is not clear yet whether matrix changes are determinants or products of cell condensation and chondrocyte differentiation. The definition of a chronological hierarchy of events would help assess this contention. To address this question we have made use of an in vitro model system able to reproduce different stages of chondrocyte differentiation, and we have focused our attention on two ECM molecules, namely FN and LN, and on their integrin receptors.

Chondrocytes interact with FN and LN via integrin receptors. Integrins are αβ heterodimers and the specific combination of different α and β subunits determines both their ligand and signaling specificity (Hynes, 1992; Hemler, 1990). Durr et al. (1993) have shown that human differentiated fetal chondrocytes express several FN (α₅β₁) and LN (α₃β₁ and α₆β₁) receptors. Furthermore, they demonstrated that chondrocyte adhesion to FN-, LN-coated surfaces is completely blocked by anti-β₁ antibodies.

We have previously described a tissue culture model system that allows both condensation and chondrogenic differentiation of dedifferentiated cells (Castagnola et al., 1986; Tacchetti et al., 1987; Cancedda et al., 1995). Cells isolated from chick embryo tibiae and dissociated by enzymatic treatment adhere to tissue culture dishes and display a fibroblastic phenotype. Under these culture conditions, dedifferentiated cells proliferate rapidly and secrete type I collagen into the culture medium. When transferred into suspension cultures, dedifferentiated cells rapidly aggregate (30 minutes –12 hours) and secrete type II collagen and other stage I chondrocyte ECM molecules. In approximately 3 weeks of culture, cell clusters separate into single hypertrophic chondrocytes secreting type X collagen (stage II chondrocytes) (Castagnola et al., 1988; Cancedda et al., 1995). Using this model system, we have shown previously that cell condensation is an obligatory step for the induction of chondrogenic differentiation in dedifferentiated cells (Tacchetti et al., 1992). Secretion of a specialized ECM is probably one of the earliest functional changes occurring at the condensation stage, and it may be crucial for the establishment of a specific microenvironment that drives the onset of chondrogenesis. Here we have studied the chronology of expression of FN, LN and of their receptors during chondrocyte differentiation in vitro, and we have performed functional studies to define their role in cell condensation.

Cell culture methods are extensively described elsewhere (Castagnola et al., 1986). Briefly, to establish primary cultures, stage 28-30 (Hamburger and Hamilton, 1951) chick embryo tibiae were isolated, washed in phosphate buffered saline (PBS), pH 7.2, and digested for 15 minutes at 37°C with 400 U/ml collagenase I (Worthington, Biochemical Corp., NJ, USA) and 0.25% trypsin (Gibco Ltd, Grand Island, NY, USA). After sedimentation the supernatant, containing tissue debris and perichondrium, was discarded and the pellet was digested for an additional 45 minutes, with the above dissociation buffer supplemented with 1,000 U/ml of collagenase II (Worthington, Biochemical Corp.). Cells were grown in anchorage-dependent conditions in tissue culture dishes and, when indicated, transferred after trypsin treatment (0.25% w/v in PBS), to anchorage-independent conditions on a 1% (m/v) agarose layer. The culture medium was Coon’s modified Ham F-12 (Ambesi-Impiombato et al., 1980) with the addition of 10% FCS, 50 U/ml penicillin and 50 μg/ml Streptomycin (Flow Laboratories, Irvine, Ayrshire, Scotland), unless otherwise indicated.

Particular attention was paid to use populations of dedifferentiated cells containing less than 6% differentiated chondrocytes for these studies, as determined by immunofluorescence staining for type II collagen.

Preparation of fibronectin-depleted FCS and elution of FN were obtained by Sepharose/gelatin affinity chromatography, as described by Akiyama and Yamada (1985). When indicated, purified bovine FN (Sigma, St Louis, MO, USA) or FN eluted from the gelatin/Sepharose column were added to the culture medium, at a concentration matching the normal plasma concentration, i.e. 300 μg/ml. GRGDSP or GRGESP peptides (TIB MOLBIOL S.r.l., Genova, Italy) were used at a concentration of 0.05, 0.5, or 1 mg/ml.

In order to compare the percentage of aggregation with the size of the cell aggregates, image processing and analysis was performed using ‘NIH Image’ software as described by Martin et al. (1997).

Adhesion assays were performed as described (Bodary et al., 1989; de Curtis and Reichardt, 1993). Briefly, tissue culture 96-well plates were coated overnight at 4°C either with 75 μl of PBS containing 50 μg/ml laminin, 0.5 mg/ml poly-D-lysine (PDL) (Sigma), or 15 mg/ml BSA in PBS.

All wells, except those coated with PDL, were washed with PBS and blocked with 100 μl PBS containing 15 mg/ml BSA for 3 hours at 37°C. After washing with PBS, 2×10⁵, 1×10⁵ or 0.5×10⁵ cells per well were plated, in 100 μl culture medium, without FCS. Experiments were performed in triplicate.

Plates were then centrifuged at 600 rpm for 2 minutes. After 1 hour incubation at 24°C, wells were washed twice with PBS, fixed for 15 minutes with 100 μl 2% glutaraldehyde, 5% sucrose in PBS, and stained for 10 minutes with crystal violet. Plates were washed twice with distilled water and cells were dissolved with 1% SDS. Adhesion was evaluated by measuring the absorbancy of the solution at 630 nm.

A 250 bp FN specific DNA fragment was synthesized by RT-PCR using RNA extracted from aggregated chondrocytes after 72 hours of suspension culture. The oligonucleotides used were: (a) 5′ TGA GGA TGG AAT CCA TGA GC as forward primer; (b) 5′ TGT AGC CAG TGA GAC GAA CG GG as reverse primer, starting at positions 1,213 and 1,463 of the FN sequence published by Norton and Hynes (1987). A 238 bp laminin β₁ specific DNA fragment was synthesized by RT-PCR using RNA extracted from aggregated chondrocytes after 72 hours of suspension culture. The oligonucleotides used were: (c) 5′ GGA GAT AAC CTC CTG GAT TCG as forward primer; (d) 5′ GTC GCC AAG GTA AGT CAT GG as reverse primer, starting at positions 527 and 744 of the sequence published by O’ Rear (1992). A 688 bp α₆ integrin subunit specific probe was synthesized by RT-PCR using RNA extracted from aggregated chondrocytes after 24 hours of suspension culture. The oligonucleotides used were: (e) 5′ TCA CAT CTT CAG ATA CGC GC as forward primer; (f) 5′ CTC CTT CGA ACA TGT GCT CC as reverse primer, starting at positions 275 and 1,652 of the sequence published by de Curtis et al. (1991). To synthesize a probe for the chicken α_6B integrin subunit, the oligonucleotides used were: (g) 5′ TCC TGC TGC AGC TAA GAA C as forward primer; (h) 5′ GAC GCT GTC ATC GTA CCT G as reverse primer, starting at positions 2,973 and 3,295 of the published sequence (de Curtis et al., 1991). All oligonucleotides used were from TIB MOLBIOL.

The RT-PCR and PCR reactions were carried out using reagents and protocol of the amplitaq DNA Polymerase kit (Perkin Elmer, Norwalk, CT, USA). Samples were analyzed by 1.5% agarose gel electrophoresis. PCR fragment of interest were purified by NA45 DEAE membrane (Schleicher and Schuell, Dassel, Germany) following the manufacturer’s instructions. Briefly the DNA band was intercepted by the membrane during gel electrophoresis. The membrane was then washed three times with 0.15 M NaCl, 20 mM Tris-HCl, pH 8.0, 0.1 mM EDTA. The DNA was eluted from the membrane by incubation in 1 M NaCl, 20 mM Tris-HCl, pH 8.0, 0.1 mM EDTA at 68°C for 30 minutes. DNA was recovered by ethanol precipitation. After purification, fragments were cloned into the pCR II vector using TA cloning kit procedure (Invitrogen, CA, USA), sequenced using sequenase version 2.0 DNA sequencing kit (USB, Ohio, USA), and compared with those deposited in Gene Bank.

Using probes ‘g’ and ‘h’, two DNA fragments were obtained by RT-PCR on a template of RNA from cells at different stages of differentiation. One, of 341 bp, had the size expected for the chick α_6A integrin subunit, while the other, of 211 bp, had the size of the putative chick α_6B integrin subunit. The latter fragment was then sequenced by Amplicycle Sequencing Kit (Perkin Helmer), following the manufacturer’s instructions.

The probe for chicken laminin γ₁ was obtained from a cDNA chicken embryo aorta library and those for α₁ and α₂ laminin chains were obtained by RT/PCR using degenerate oligonucleotides as primers, designed on the basis of the human sequences. The amplified fragments, spanning about 600 bp in the VI domain, were cloned in pGEM-T vector (Promega) and sequenced. Their identity was confirmed by comparison with the corresponding human sequence (I. Bellina et al., unpublished work).

For northern blot analysis, total RNA was extracted from cultured cells by the guanidium thiocyanate method (Chomczynski and Sacchi, 1987). Hypertrophic chondrocytes, after 3 weeks in suspension culture, were filtered through a 42 μm mesh Nitex filter, to avoid contamination by cell clusters. RNAs were separated by electrophoresis in 1% agarose gels, in the presence of formaldehyde, and blotted onto Hybond N membranes (Amersham Buckinghamshire, UK). Blots were hybridized with ³²P-dCTP random priming labeled probes (Boehringer Mannheim, Germany). Hybridization was performed at 65°C in 7% SDS, 0.33 M sodium phosphate monobasic, pH 7.2. Washing conditions were as recommended by Amersham, but at room temperature. Autoradiographs were obtained at −80°C using hyperscreen and hyperfilm (Amersham).

The B3D6 mouse anti-avian FN monoclonal antibody was obtained from the Developmental Studies Hybridoma Bank (Iowa City, IA)(Gardner and Fambrough, 1983). Rabbit antisera to the cytoplasmic portion of the α₅ and α₃ integrin subunit was a generous gift from Prof. Guido Tarone (University of Torino, Italy) (DeFilippi et al., 1991). The antiserum 317 raised against the cytoplasmic tail of β₁ (Tomaselli et al., 1988), the rabbit antiserum raised against the extracellular portion of the α₆, and the rabbit antisera against the alternatively spliced region of the α_6A and α_6B integrin subunits (de Curtis et al., 1991; de Curtis and Reichardt, 1993), were a generous gift from Dr Ivan de Curtis (DIBIT, Milano, Italy). While the antibodies to α_6A and α_6B integrin subunits were specific for chicken, those to α₅, α₃, β₁ and α₆ integrin subunit also recognize different species. The rabbit antiserum raised against the laminin from the basement membrane of the Engelbreth-Holm-Swarm sarcoma (EHS laminin, Lam-1) (Gloghini et al., 1989) and the mouse monoclonal antibodies to laminin α₂, β₁ and γ₁ chains were a gift from Dr Roberto Perris (CRO, Aviano, PN, Italy).

To assay for FN, α₃, α₅, α₆, and β₁ integrin subunit expression, cells were lysed in ice-cold 10 mM Tris-HCl, 150 mM NaCl, 1% NP-40, pH 7.5 (Sigma) buffer, containing 2 mM phenylmethylsulfonyl fluoride (PMSF) (Sigma), 2% aprotinin (Sigma), and 2 μg/ml leupeptin (Sigma), as protein inhibitors. Samples were resolved by SDS-PAGE in 6% polyacrylamide gel, under reducing conditions (Laemmli, 1970), followed by transfer to nitrocellulose (Towbin et al., 1979) and the corresponding bands were revealed by incubation alternatively with: B3D6, in the form of hybridoma supernatant, rabbit antisera to the carboxy tails of α₃, α₅, and β₁ diluted 1:200 and α₆, diluted 1:300. As second reagents we used alkaline phosphatase labeled goat anti-mouse (Sigma) or sheep anti-rabbit IgG (Boehringer Mannheim). The immuno-enzymatic reaction was obtained by incubation with NTB/BCiP (4-nitroblue tetrazolium chloride/5-bromo-4-chloro-3-indolyl phosphate-toluidine salt; Bio-Rad, Hercules, CA). To detect the presence of α_6A and α_6B, the samples were first immunoprecipitated with the rabbit antisera to α_6A and α_6B, separated by SDS-PAGE, blotted, and stained with the antiserum to the extracellular portion of α₆.

1 ml of cell lysates, obtained as described in the previous section, were incubated with 50 μl of Sepharose-conjugated Protein A (Sigma), 1 hour at 4°C, after centrifugation the supernatant was treated again with 50 μl of Sepharose conjugated Protein A and 15 μl of control rabbit serum, 1 hour at 4°C. Samples were centrifuged and incubated with 5 μl of either one of the rabbit antisera to α_6A and α_6B integrin subunits. Samples were resolved by SDS-PAGE (Laemmli, 1970) in a 6% polyacrylamide gel under reducing conditions. The corresponding bands were revealed by western blot.

Hind limb buds were removed from stage H.H. 24 chick embryos (staging according to Hamburger amd Hamilton, 1951) and immediately frozen in methyl-butane (Bracco, Milan, Italy) pre-chilled in liquid nitrogen, and embedded in OCT. Cells after 24 and 72 hours of suspension culture were fixed with 3.7% paraformaldehyde in PBS, embedded at 4°C in increasing concentrations of sucrose in PBS, followed by OCT (Tissue-Tek, Miles Inc., Elkhart, IN), and frozen in liquid nitrogen. Cryosections were laid on gelatin coated glass slides and air dried. Dedifferentiated cells were cytocentrifuged on glass slides, washed in PBS and fixed with a 1:1 solution of methanol:acetone, pre-chilled at −20°C. Immunolocalisation was performed using the primary antibodies described above and, as secondary reagents, TRITC-conjugated, affinity purified, goat anti-rabbit or goat anti-mouse IgG antibodies (Jackson Immunoresearch, West Grove, PA).

The role of FN in the cell aggregation is investigated in experiments in which dedifferentiated cells are suspended in culture medium in the absence of FCS or in the presence of FN-depleted FCS (Fig. 1). After 6 hours in culture, no aggregation is detected in the absence of serum or in the presence of FN-depleted FCS (Fig. 1B,D), whereas in control cultures (Fig. 1A) and in the presence of purified bovine FN or FN eluted from the gelatin/Sepharose column (Fig. 1C,E,F) about 60% of the cells form clusters. These results indicate that FN is involved in the very early stage of dedifferentiated cell aggregation, however, they give no indication as to direct cell-FN interactions.

To determine whether direct cell-FN interactions are involved in the early cell aggregation, dedifferentiated cells have been maintained in suspension culture in the presence of the peptides GRGESP or GRGDSP. After 5 hours (Fig. 2 and Fig. 3A), approximately 60% of the GRGESP-treated control cells, irrespective of the concentration of peptide used, aggregate, while only 29%, 20% and 12% form clusteres in the presence of GRGDSP, at a concentration of 0.05, 0.5 or 1 mg/ml, respectively. Determinations of the size of cell aggregates also demonstrate that fewer cells participate in the formation of clusters in the presence of GRGDSP, compared to those formed in the presence of GRGESP (Fig. 3B).

Dedifferentiated cells either adherent, or cultured in suspension for 36, 72 hours, 11 days and 3 weeks, have been analyzed by northern blot for the level of FN mRNA (Fig. 4A). The band with the highest intensity has been observed in dedifferentiated cells (lane 1) and the progressively decreasing signal (lane 2-4), disappears in hypertrophic cells (lane 5). Similarly, the amount of FN present in the cell cultures at different stages of differentiation, detected as a 230 kDa band by western blot, decreases progressively, with the highest intensity observed in dedifferentiated cells, and no signal detectable in hypertrophic chondrocytes (Fig. 4B).

The expression of four laminin chain mRNAs, α₁, α₂, β₁, γ₁ (Shaw and Mercurio, 1994), has been analyzed by northern blot analysis during the differentiation of chondrogenic cells (Fig. 5). Dedifferentiated cells express no α₁ and β₁ mRNAs and only low levels of γ₁. After 12 hours in suspension culture, the α₁, β₁ and γ₁ mRNA levels increase dramatically. The β₁ level remains constant from 36 hours to 11 days in suspension culture, and disappears in hypertrophic cells, whereas α₁ expression decreases progressively with time in suspension culture and it is not detectable after 11 days of suspension culture and in hypertrophic chondrocytes. At variance from α₁ and β₁, γ₁ mRNA levels decrease progressively after 24 hours in suspension culture. However, a band is still detectable in hypertrophic chondrocytes. The α₂ chain mRNA has always been undetectable (not shown). The expression of the three laminin chain mRNAs α₁, β₁ and γ₁, suggests that, if translated into secreted proteins, a laminin of the EHS basement membrane type is assembled within the cell aggregates after 12-24 hours in suspension culture.

Sections of cell aggregates at different times in suspension culture are stained by immunofluorescence for FN with B3D6 antibodies and with monoclonal antibodies to the α₂, β₁ and the γ₁ laminin chains, or with polyclonal antibodies to EHS laminin (Fig. 6). FN is deposited extracellularly, and with time in culture segregates at the periphery of the cell aggregates (A,C). Anti-bodies to the α₂ chain give no staining on cells in culture or in limb buds ex vivo (not shown). All the remaining antibodies to LN fail to stain dedifferentiated cells (not shown) and reveal an extracellular fibrillar network after 24 hours in suspension culture. Staining is localized mostly, but non exclusively, in the central regions of the cell aggregates (B). With time in culture the staining increases in intensity and localizes in the central regions of the cell aggregates (D). To investigate whether the EHS type of LN is localized in cell condensation areas also in vivo, immunofluorescence on stage 24 chick embryo hind buds is performed. A sub-ectodermal staining of the basal lamina is observed (Fig. 6E), while no staining is detected in the underlying mesenchymal tissue (Fig. 6E). In contrast, an extracellular fibrillar network in the chondrogenic condensation area is observed in the central region of the same limb bud (Fig. 6F).

The expression of several α integrin subunits involved in FN and LN binding has been assayed. About 95% of the dedifferentiated cells in the populations analyzed in these experiments expressed α₃, α₅, α₆, and β₁ on their surface, as determined by FACS analysis (G. Bellese, unpublished work). By western blot analysis, both the 140 kDa α₃ and the 135 kDa α₅ integrin subunit bands are detectable in dedifferentiated cells and the intensity of the bands decreases after 72 hours in culture (Fig. 7A,B). However, while the α₅ integrin subunit band disappears in hypertrophic cells (Fig. 7B), α₃ remains at low levels. To evaluate the levels of expression of the α₆ integrin subunit, we used both anti-bodies recognizing the extracellular conserved domain of α₆ (Fig. 8A) and antibodies recognizing the cytoplasmic segments of the two isoforms α_6A and α_6B (Fig. 8B). The total amount of α₆ is constant throughout the first 72 hours of cell differentiation. However, α_6A shows an increasing staining intensity from dedifferentiated cells, to cells cultured for 72 hours in suspension (Fig. 8B, lanes 1-3). On the contrary, α_6B, shows a reverse pattern of expression (Fig. 8B, lanes 4-6). Furthermore, α_6A is much more intense than α_6B in hypertrophic cells (Fig. 8B, lanes 7-8).

The mRNA content for the two α₆ isoforms, analyzed by northern blot, reproduces the pattern observed for the proteins (not shown). However, the signal in the dedifferentiated cells is almost undetectable. By RT-PCR a band with the putative size of the α_6B chain has been isolated and sequenced. A comparative analysis with the human sequence shows that the alternative splicing pattern of α_6A and α_6B mRNAs is conserved in chicken; the α_6A specific exon sequence is inserted between nucleotide 3,145 and nucleotide 3,274 of the chicken sequence published by De Curtis et al. (1991), resulting in a mRNA identical to that encoded by the human α_6B mRNA. RT-PCR analysis, using oligonucleotides that are outside the alternatively spliced box that determines the difference between α_6A and α_6B, show the presence of two bands (Fig. 9) of the size expected for α_6A and α_6B. The bands are present both in dedifferentiated and hypertrophic cells, with a lower intensity for α_6A in dedifferentiated cells and α_6B in hypertrophic cells.

As previously reported, all integrin α subunits studied dimerize in chondrocytes with β₁ (Durr et al., 1993). The expression of β₁ during chondrocyte differentiation in vitro, analyzed by western blot (not shown), demonstrates no appreciable differences in intensity at the different time points considered.

The activity of the laminin receptors expressed by chondrogenic cells in supporting EHS-LN binding has been tested by cell adhesion assays. The binding of dedifferentiated cells to EHS-LN is evaluated in comparison to substrata known to either support (i.e. PDL) or prevent (i.e. BSA) dedifferentiated cell adhesion to plastic. Three different concentrations of cells have been used, 0.5×10⁵, 1×10⁵ or 2×10⁵ per ml. The binding to BSA is negligible at all cell concentrations, while an efficient cell concentration-dependent binding to laminin, comparable to that on PDL, is observed (Fig 10).

Cell condensation is a necessary step for the differentiation of pre-chondrogenic cells in the limb bud mesenchyme. The morphological and biochemical events leading to cell condensation are not yet completely understood (Von der Mark and Von der Mark, 1977; Tacchetti et al., 1987; Maini and Solursh, 1991). Cell-cell interactions (Solursh and Reiter, 1980; Jiang et al., 1993; Tavella et al., 1994), composition of the ECM (Dessau et al., 1980; Tomaseck et al., 1982; Mackie at al., 1987; Maleski and Knudson, 1996), changes in cell shape (Zanetti and Solursh, 1984) and several specific cytokines (Chang et al., 1994) have been considered to play a role in this process. Neverthless, the interplay between all of the factors and the precise sequence of events leading to differentiation has not yet been fully understood. Aim of the present study has been to determine the kinetics of expression and the role of FN, LN and of their related receptors, during chondrogenesis, using primary cultures of dedifferentiated cells, obtained from chick embryo tibial rudiments. These cells, under the appropriate culture conditions, form aggregates (similar to condensation areas) and differentiate into hypertrophic chondrocytes (Castagnola et al., 1986; Tacchetti et al., 1992). Furthermore, dedifferentiated cells, although not identical to the pre-chondrogenic cells of the limb bud, share with these latter a series of morphological and functional similarities, i.e. they are elongated in shape, secrete

FN, LN and type I collagen, undergo the same modulation of cell-cell adhesion molecule expression (N-CAM and N-cadherin), and differentiate into chondrocytes following a cell aggregation stage (Castagnola et al., 1986, 1988; Tavella et al., 1994). Furthermore, and similarly to what occurs in vivo, differentiation starts at the center of the cell aggregates and proceeds peripherally (Tacchetti et al., 1987; Tavella et al., 1994). In comparison with primary micromass cultures of limb bud mesenchymal cells, these dedifferentiated cells are more homogeneous and do not include all potential phenotypes that will eventually be expressed in the limb bud.

Here we show that: (i) FN is maximally expressed by dedifferentiated cells and decreases during the differentiation process; (ii) both FN depletion, as well as inhibition of cell binding to FN, interfere with cell aggregation; (iii) the FN α₅β₁ integrin receptor expression follows the same trend of changes as that of FN; (iv) a complete laminin molecule may be formed only after cell aggregation has already started. Three laminin chains, α₁, β₁, γ₁, are expressed at the onset of differentiation of dedifferentiated cells to chondrocytes, when cell aggregation is maximal (12-24 hours of suspension culture). (v) the α₃β₁ LN receptor expression is maximal in dedifferentiated cells and decreases thereafter, still persisting at low levels in hypertrophic chondrocytes; (vi) a functionally active α₆β₁ LN receptor is expressed from dedifferentiated cells to hypertrophic chondrocytes; however, concomitantly with the maximal of cell aggregation and the onset of the α₁ and β₁, and the rise of γ₁ LN chain expression, the ratio α_6B/α_6A decreases sharply.

These data suggest a time hierarchy in the regulated expression of FN, LN and of their receptors. In particular, the cell clustering phenomenon (cell condensation) may be divided in two steps. In the first step, as shown by the interference with cell aggregation by FN depleted culture medium or by RGD containing peptides, cells, already expressing N-CAM and N-cadherin on their surface (Tavella et al., 1994), aggregate via integrin-FN interactions. In the second step the clustered cells strengthen the cell-cell contacts via N-CAM and N-cadherin (Jiang et al., 1993; Tavella et al., 1994). Later on N-CAM expressing cells segregate to the periphery of the cell aggregates, where non-differentiating cells reside (Tavella et al., 1994). In the inner differentiating area of the cell aggregates the expression of N-cadherin promotes the formation of adherent junctions (Tavella et al., 1994). Nevertheless, both steps are necessary for the condensation to occur, as demonstrated by perturbation experiments both in vivo and in vitro (Frenz et al., 1989a,b; Oberlender and Tuan, 1994). In addition, the cell clustering and the occurrence of cell-cell adhesion could be supported in vivo by the reduction of the intercellular space, possibly due to a reduced hyaluronic acid accumulation in the perspective condensation area, due to its active degradation via endocytosis by the pre-chondrogenic cells (Kulyk and Kosher, 1987; Maleski and Knudson, 1996).

It has been suggested (Frenz et al., 1989b) that local accumulation of FN could drive cells in the condensation area and influence the expression of differentiation markers, such as highly sulfated proteoglycans, in micromass cultures of limb bud mesenchymal cells (Frenz et al., 1989a). We show that FN plays this role by promoting cell-cell interactions, which in turn determines the formation of a differentiation promoting microenviroment (Tacchetti et al., 1992). Furthermore, in other cell model systems, α₅β₁ binding to FN has been shown to inhibit cell migration (Giancotti and Ruoslahti, 1990). Therefore, it is possible to speculate that the high level of FN and of its receptor in the condensation area may also participate in trapping pre-chondrogenic cells in the prospective condensation area. However, since FN seems to inhibit chondrogenic differentiation (reviewed by Maini and Solursh, 1991), the reduction of FN and of its α₅β₁ receptor, following cell aggregation, is necessary for the progression of cartilage growth and differentiation. The areas of the cell aggregates where FN is still present, correspond to areas where differentiation is impaired, i.e. the periphery of the cell aggregates.

Since in our cell system: (i) the lack of FN interferes with cell aggregation in cells expressing N-CAM and N-cadherin on their surface, (ii) after 5-6 hours in suspension culture, even in the presence of FN-depleted medium or GRGDSP, dedifferentiated cells start to aggregate, it is conceivable that FN plays the role of facilitating and accelerating the formation of initial cell aggregates.

What is the role of LN in this process? Our data show that the α₁, β₁, and γ₁ mRNAs (constituents of the EHS laminin, Lam-1; Burgeson et al., 1994) are maximally expressed at the higher level of cell aggregation, and progressively decrease during the differentiation to stage I chondrocytes; stage II hypertrophic chondrocytes express only low levels of γ₁. The lack of LN in hypertrophic chondrocytes in our culture system, is in line with recently published data obtained in human cartilage (Durr et al., 1996). We do not exclude the possibility that other forms of LN, for which chick specific probes are not yet available, may also be expressed. The increase in α₁, β₁, and γ₁, LN mRNAs is paralleled by a switch in the isotype of α₆ integrin subunits synthesized, from B to A. Therefore, LN EHS does not participate in the initial cell aggregation, but it is involved in events occurring at later stages. The presence of LN predominantly in the central areas of the cell aggregates, that correspond to differentiating cells, supports this hypothesis. Considering that, because of different signal transduction pathways, α_6B, at variance of α_6A, is unable to induce cell migration on LN substrates (Shaw and Mercurio, 1994; Shaw et al., 1995), one could speculate that the increase in the A form may help movement of differentiating cells toward, and/or the segregation within, the central regions of the cell aggregates, keeping them away from the periphery of the cell aggregate where differentiation does not occur. In conclusion, we speculate that cell/FN and cell/LN interactions, concomitantly with other mechanisms (N-CAM and N-cadherin expression; Tavella et al., 1994), participate in the segregation of different cell populations, i.e. inner differentiating from outer non differentiating cells, the latter eventually forming the perichondrium (Tacchetti et al., 1987; Quarto et al., 1990).

The monoclonal antibody B3D6 developed by Dr D. Fambrough, Johns Hopkins University, has been obtained from the Developmental Studies Hybridoma Bank maintained by the Department of Pharmacology and Molecular Sciences, Johns Hopkins University School of Medicine, Baltimore, MD 21205, and the Department of Biological Sciences, University of Iowa, Iowa City, IA 52242, under contract NO1-HD-2-3144 from the NICHD. The authors thank Guido Tarone Roberto Perris, and Ivan de Curtis for the generous gift of antibodies and for helpful discussions. We also thank Simona Montalti and Francesca Fiscella for technical assistance and Marco Arvigo for the photographic work. This work was supported by funds from the Consiglio Nazionale delle Ricerche, from MURST (Ministero Università e Ricerca Scientifica e Tecnologica), from AIRC (Italian Association for Cancer Research) and from the Human Capital and Mobility Programme of the European Community.