Gene regulation during cartilage differentiation: temporal and spatial expression of link protein and cartilage matrix protein in the developing limb

The developing chicken embryonic limb is an excellent system for studying the temporal and spatial regulation of gene expression. Many morphological and biochemical changes associated with limb development have been described. During chondrogenesis, these changes are characterized by the extracellular matrix (ECM) molecules that are being synthesized (Goetinck et al. 1981). The initial stages of chondrogenesis are characterized by the condensation, or aggregation, of mesenchymal cells in the central core of the developing limb bud. This condensation process may be initiated by a progressive decrease in extracellular hyaluronate (Kosher et al. 1981; Toole, 1973). Other extracellular matrix macromolecules such as fibronectin and type I collagen may also be involved in the cell-cell interaction during the condensation process (Dessau et al. 1980; Kosher et al. 1982; Tomasek et al. 1982). This precartilaginous condensation is first detectable at stage 23 of Hamburger and Hamilton (Fell and Canti, 1934; Hamburger and Hamilton, 1951).

Cartilage differentiation (Zwilling, 1968), has been traditionally measured by the detection of two major cartilage ECM molecules. These markers for chondrogenesis include a switch in collagen synthesis from type I to type II collagen (Linsenmayer et al. 1973) and an increased macromolecular incorporation of ³⁵SO₄ (Searls, 1965) which is a reflection of expression of the cartilage chondroitin sulfate proteoglycan (Goetinck et al. 1981). Not until stage 24 is the synthesis of type II collagen detected biochemically or immunologically (Dessau et al. 1980; Linsenmayer et al. 1973; von der Mark et al. 1976) in the developing limb. Link protein and cartilage matrix protein are also characteristic cartilage extracellular matrix molecules whose regulation has not yet been defined.

The condensed central limb mesoderm forms a histologically detectable Y-shaped pattern at stages 24-25 (Searls et al. 1972) and cartilage cells can first be detected histologically at stage 25 (Fell and Canti, 1934). As differentiation continues, the pattern of the cartilaginous regions formed corresponds to the elements that will become the limb skeleton.

In vitro studies have demonstrated several possible factors involved in regulating cartilage differentiation in the limb (Solursh, 1984). However, the molecular mechanisms governing the changes in gene activity that are necessary for chondrogenesis are not understood. In order to understand the mechanisms of determination and differentiation during chondrogenesis in the developing limb, the temporal and spatial patterns of chondrocyte gene expression have to be defined.

Several ECM molecules are characteristic for chon-drocytes and, therefore, the expression of their genes can be used as a measure of cartilage differentiation. These molecules include: type II collagen (Miller and Matukas, 1969), the major sulfated cartilage proteoglycan (Kimata et al. 1974), link protein (Hardingham, 1979), and cartilage matrix protein (Paulsson and Heinegård, 1979, 1981). The interactions of these macromolecular components play a role in the formation and maintenance of the extracellular matrix of cartilage and, as a result, in the morphology of the cartilaginous skeleton. In this study we have determined the temporal and spatial changes in gene expression during chondrogenesis in the developing chicken embryonic limb from stage 23 through stage 28 using RNA in situ hybridization. We have used cDNA probes encoding four different extracellular matrix proteins to define the regulatory events in chondrogenesis more completely.

RNA extracted from 14-day-old normal embryonic chicken sterna was electrophoresed on an 0-8 % agarose gel containing formaldehyde (Lehrach et al. 1977) and then transferred to a nitrocellulose filter (Schleicher and Schuell). Prehybridization and hybridization with radioactively labeled cDNA inserts was performed using methods described by Maniatis et al. (1982) with the inclusion of 50 % formamide. The filter was then washed at low stringency, 4 × SSC at 42°C (1 × SSC is 150mM-NaCl, 15mM-sodium citrate) and exposed to X-ray film with an intensifying screen at —70°C for 18 to 24 h. For rehybridization of the RNA blot, the radioactive cDNA probe was removed in 0·1 × SSC, 0·1% SDS at 90°C for 10 min.

Fertile chicken eggs were obtained from McIntyre Farms, Lakeside, CA and incubated to achieve embryos of the desired developmental stages (Hamburger and Hamilton, 1951). Stages 21 through 28 were used. Torsos with the attached wing buds were removed from the staged chicken embryos and cut at the midline. The tissues were then oriented and embedded immediately in OCT compound (Tissue-Tek). The embedding process was performed on dry ice and the embedded tissue was stored at —70 °C. The embryonic torsos were embedded ventral side down with the limb oriented, as much as possible, at a right angle to the anterior-posterior axis of the embryo. Frontal serial sections of this embedded tissue were then used for the in situ hybridization. Since, the curvature of the limb changes with development, composite photos of the serial sections were made to represent all areas of hybridization.

Frontal sections of about 6 μm were cut on a cryostat at —20°C and placed on microscope slides that were washed in chromic acid (Chromerge) and coated in a solution of poly-L-lysine (50μgml⁻ in 10mM-Tris-HCI, pH8·0). The sections on the slides were allowed to dry for 1-2 h at room temperature before fixing in a solution of 3·7% paraformaldehyde, 5 mM-MgCl₂ in 1 × PBS. After a fixation period of 15 min the slides were washed in 70% ethanol and stored in fresh 70% ethanol at 4°C.

The known sequence characteristics of the four cartilage ECM proteins were considered when selecting the cDNA probes. The four cDNA probes as well as the control plasmid DNA were radioactively labeled with ³⁵S-dCTP (1200 Ci mmole⁻) using the Pharmacia oligolabeling kit (Feinberg and Vogelstein, 1983, 1984) to a specific activity of 10⁷-10⁹ctsmin^-μg⁻ of DNA. 50 ng of probe were boiled for 15 min and then transferred to a 37°C water bath for 10 min. Nucleotides (dATP, dGTP, dTTP), the random primers, BSA, the Klenow fragment of DNA polymerase I and the isotope at 75 to 100 #x03BC;Ci per 50 ng of probe were then added. The oligolabeling reaction was for 2h at room temperature. Disodium EDTAwas added to a final concentration of 0·03 M to stop the reaction.

The ³²P-labeling of the cDNA probes was also by oligolabeling. However, the isotope was used at 50#x03BC;Ci per 50 ng of probe. Labeled probe was separated from the free isotope on a mini-spin column of Sephadex G-50 (Worthington), boiled 10 min and then quick-chilled before adding to the hybridization solution.

The RNA in situ hybridization was based on techniques in Rentrop et al. (1986). Stored sections were rehydrated in 1 × PBS, 5mM-MgCl₂ for 10 min and then incubated in 0·1M-glycine, 0·2M-Tris-HCl pH7·4 for 10 min. Before hybridization, the sections were washed in 2 × SSC. The hybridization conditions were as follows: 50% formamide, 600mM-NaCl, ImM-EDTA, 10 mM-Tris-HCl pH7·4, 10% dextran sulfate, 1 × Denhardts (0·02% Ficoll, 0·02% Polyvinylpyrrolidone, 0·02% Bovine serum albumin), 1 mg ml⁻ yeast tRNA, 50μgml⁻ salmon sperm DNA, 10mM-dithiothreitol (DTT), probe at about 0·1 ngμl⁻¹. Each slide contained 40 μl of hybridization solution overlaid with a siliconized coverslip and sealed with rubber cement. Hybridization was overnight at 42 °C.

Slides were submerged in 4 × SSC, 10ITIM-DTT at room temperature to soak off the coverslips. The sections were then washed as follows: 2 × SSC, 10mM-DTT for 5 min at room temperature; 2 × SSC, 10mM-DTT for 30 min at 42°C; 1 × SSC and 10mM-DTT for 30 min at 42°C, twice. Before coating with emulsion the sections were dehydrated in two changes each of: 70% ethanol, 300mM-ammonium acetate; 95% ethanol, 300mM-ammonium acetate; and 99% ethanol, 300mM-ammonium acetate.

Immediately following the dehydration step after hybridization, the slides were coated with Kodak NTB-2 emulsion (1:1 with 600mM-ammonium acetate) in a darkroom. The coated slides were allowed to dry slowly (overnight) in a light tight, humidified environment. They were then transferred to slide boxes containing Drierite and placed at 4°C for exposure times of 1 to 2 weeks.

The slides were set at room temperature for 15 min to bring them to a temperature of about 15°C. The development and washing procedures were at a constant temperature of 15°C. Development was in D-19 developer (Kodak) diluted 1:1 with deionized H₂O (dH₂O) for 2·5 min. The stop bath was dH₂O for 30s followed by fixing in Kodak fixer for 4–5 min. The slides were then washed for 15 min in dH₂O and then in running tap H₂O for 1 h.

The following steps are from staining, through dehydration and mounting: 0·02% toluidine blue O in 1 × PBS pH 7·0, 30-60 s followed by sequential 30-60 second washes in dH₂O, 30% ethanol, 70% ethanol, 85% ethanol, 95% ethanol, 2 changes of 100% ethanol, and 2 changes of xylene; DPX Mountant (BDH Chemicals). Observation and photography of the slides was under bright-field and dark-field microscopy using a Nikon Optiphot, a preset automatic light meter, and Ilford XP1 400 film.

RNA in situ hybridization gave a direct determination of the temporal and spatial regulation of gene expression during chondrogenesis in the developing chicken embryonic limb. Fig. 1 contains schematic diagrams of the four ECM proteins investigated. Also shown are the regions covered by the cDNA probes used in the study. The type II collagen probe encodes primarily the C-propeptide region and contains 3’ untranslated sequences as well. The probe for the proteoglycan core protein covers about 24 % of the coding region at the 3’ end of the transcript. Link protein contains a tandem repeat that functions as the hyaluronic acid binding region (Goetinck et al. 1987; Périn et al. 1987). These domains share 41 % and 33 % homology with the proteoglycan core protein at the amino acid level in the rat (Doege el al. 1987). The remaining third domain also shares a homology of 31 % with the proteoglycan core protein sequence. To eliminate the region of greatest homology in the link protein probe, a fragment from the pLPG2 cDNA insert was isolated and purified for use in the hybridization experiments. This probe corresponds to 55 % of the coding region at the 5’ end of the transcript. The cartilage matrix protein cDNA probe covers 51 % of the transcript coding region at the 3’ end and does not contain the region sharing homology with an epidermal growth factor-like repeat.

The conditions for RNA in situ hybridization were not as stringent as the conditions normally used for filter hybridization. Therefore, low stringency RNA blots were performed to detect possible cross hybridizations (Fig. 2). By comparing this RNA blot to a more stringently treated RNA blot (Stirpe et al. 1987), it was evident that no prominent cross hybridizing transcript species were present. The faint band of about 7·2 kb in Fig. 2, lane 1 most likely corresponds to the low level 7·2 kb type II collagen transcript (Young et al. 1984). No cross-hybridization between the link protein probe and the core protein mRNA is seen.

Bright-field photographs of the chicken embryonic limb tissue examined are shown in Fig. 3. Each stage of development studied is represented. Concentrated hybridization in the developing limb is first seen using the type II collagen probe at stage 23 (Fig. 4A). A condensation of the grains shows a localization of hybridization in the central region of the limb bud. Also evident is hybridization in the somite region (Fig. 4A). Using the core protein probe, there is no positive signal of hybridization in the limb; however, there is hybridization to the notochord and somite tissues (Fig. 4B). Hybridization with the link protein and cartilage matrix protein probes is also negative at this stage in the limb (Fig. 4C and 4D). The somite and notochord tissues are not represented in the tissue sections seen in Fig, 4C and 4D. Whenever the cartilaginous notochord and somite regions, which will form the axial skeleton, do appear in the tissue sections examined, all four cDNA probes show positive hybridization signals in these tissues. At stage 24, the pattern of the signal of hybridization for type II collagen transcripts is Y-shaped (Fig. 5A). Core protein (Fig. 5B), link protein (Fig. 5C), and cartilage matrix protein (Fig. 5D) still show no hybridization. Not until stage 25 are positive signals visible using the proteoglycan core protein and link protein probes (Fig. 6B and 6C). Although the signal is not strong it already appears in a split pattern. The hybridization of the type II collagen probe is concentrated in areas representing the cartilaginous skeletal elements of the limb which include the humerus, the radius, the ulna and the carpals (Fig. 6A). Cartilage matrix protein probe hybridization is still negative (Fig. 6D). Finally, at stage 26 all the probes, including cartilage matrix protein, give hybridization signals in areas corresponding to the skeletal elements of the humerus, radius and ulna (Fig. 7). Hybridization is also evident in the carpal regions using the type II collagen probe (Fig. 7A), the proteoglycan core protein probe (Fig. 7B) and the link protein probe (Fig. 7C), but the cartilage matrix protein hybridization signal is localized in the more proximal regions of the limb as compared to the other three probes. With each additional stage of development examined the signal of hybridization increases and is specific for regions which will form the cartilaginous skeletal elements of the limb (Figs 8 and 9). Control experiments using pGEM-3 plasmid DNA sequences as probe show uniform background levels of hybridization (Fig. 10).

The temporal and spatial changes in gene expression during chondrogenesis in the embryonic chick wing bud for four genes characteristically expressed by chondrocytes were investigated using RNA in situ hybridization. All four probes used were specific for their transcripts under the conditions used. In the tissue sections used for the limb study, hybridization was also seen in the somites at the level of the wing. The first apparent signal in the somites was observed at stage 22. The cartilaginous notochord is also positive for all four ECM protein transcripts at all developmental stages examined from stage 21. Complete systematic studies were not performed on the expression of the cartilage ECM genes during development in the notochord and somites. Therefore, the data do not provide any information when each of the four extracellular matrix genes initiates expression in these tissues. However, hybridization signals are detectable in these tissues prior to their appearance in the limb tissue. The detected temporal and spatial appearance of the transcripts for all four ECM proteins is summarized in Table 1.

Type II collagen transcripts are first detected at stage 23 in the developing wing by in situ hybridization. This result is coincident with the precartilaginous condensation of mesenchymal cells (Fell and Canti, 1934), signifying the initial stages of chondrogenesis. Previous results of solution hybridization experiments demonstrated low levels of type II procollagen mRNA from whole chicken embryonic limbs as early as stage 20 (Kravis and Upholt, 1985) and as early as stage 18/19 using dot blot hybridization (Kosher et al. 1986). However, these levels are barely detectable above background using RNA in situ hybridization (Swalla et al. 1988). In addition, even though type II collagen mRNA is being transcribed in these prechondrogenic limb mesenchymal cells, type II collagen protein is below the level of detection. Spatially, type II collagen transcripts are the first to be detected in situ in all prechondrogenic regions during the stages of limb development studied. Since the type II collagen mRNA is present at low levels in the chondrogenic progenitor cells it has been proposed to be a marker for cells determined to become chondrocytes (Kosher et al. 1986). It has also been suggested that type II collagen may be involved in promoting tissue interactions that would define a chondrogenic fate (Thorogood et al. 1986). Whereas the presence of type II collagen transcripts may be a marker for chondrogenic determination in the limb, type II collagen is not necessarily an exclusive marker for cartilage since the protein has been localized at early embryonic stages in regions that are nonchondrogenic (Fitch et al. 1989; Kosher and Solursh, 1989).

Stage 25, when cartilage differentiation can be identified histologically, is also the stage where transcripts for both proteoglycan core protein and link protein are first detected in situ. The Y-shaped pattern of hybridization observed is coincident with the pattern of condensation of chondrogenic cells and prechondrogenic mesenchymal cells seen morphologically at this stage. This pattern of hybridization for cartilage proteoglycan core protein was also observed by Mallein-Gerin et al. (1988). Transcripts for cartilage matrix protein are not detectable until stage 26 after overt cartilage differentiation has already occurred and after the presence of the transcripts for the other three genes can be demonstrated. This finding is consistent with immunolocalization data for the cartilaginous mouse humerus which demonstrated that the appearance of type II collagen protein preceded that of the 148×10³M_r protein (cartilage matrix protein) (Franzen et al. 1987). Furthermore, using a chick cartilage matrix protein polyclonal antiserum demonstrates that this protein is first detected at stage 26 in the developing chicken embryonic limb (unpublished data). In subsequent developmental stages mRNA accumulation parallels cartilage matrix accumulation in these regions. From the stage of first detection of transcripts for proteoglycan core protein, link protein, and cartilage matrix protein, the observed hybridization corresponds to the pattern of the cartilaginous skeletal elements forming in the limb.

Chondrification during limb skeletal development was first shown by Johnson (1883) to proceed in a proximodistal direction. Therefore, the skeletal elements in the proximal regions of the limb are more developmentally advanced. The RNA in situ hybridization data parallel this pattern of development. Spat’ally, chondrocyte-specific transcripts for the four genes under investigation are first detected in the limb regions that correspond to the proximal skeletal elements. Temporally, however, the initial appearance of cartilage matrix protein hybridization signal lags behind in this proximodistal pattern. From the RNA in situ hybridization studies, it is evident that type II collagen and cartilage matrix protein are independently regulated. Furthermore, proteoglycan core protein and link protein transcription is regulated independently of type II collagen and cartilage matrix protein. In a separate study, we report that link protein transcripts can be detected in the mesonephros (manuscript in preparation) without any detectable proteoglycan core protein transcripts. This finding provides evidence that these two genes can also be regulated independently.

The temporal and spatial map of gene expression of several chondrocyte molecular markers has provided evidence of non-coordinate regulation of these genes. In addition, the comparison of the stages at which initiation of gene expression occurs for each marker should be helpful to elucidate the factors involved in determining and maintaining the chondrogenic state. Observations in vivo have demonstrated that factors involved in chondrogenesis exist in certain regions of the limb bud (Saunders, 1977). However, no diffusible morphogens have been identified. For example, the apical ectodermal ridge, required for distal outgrowth (Saunders, 1948), is itself believed to depend on a diffusible factor from the mesoderm (Saunders and Gasseling, 1963; Zwilling, 1961). These studies on the regulation of chondrocyte cell differentiation as well as studies on the stability of the chondrogenic state have also been extended to in vitro systems. In both systems the knowledge gained from the present work can be applied in further studies on the regulation of gene expression during chondrogenesis.

This work was supported by grant HD22016 of the NICHD. The authors thank Ms Cindy L. Clatterbuck for manuscript preparation.