- Email: [email protected]
Evidence for sequential appearance of cartilage matrix proteins in developing mouse limbs and in cultures of mouse mesenchymal cells
Differentiation
Differentiation (1987) 36: 199-210
0 Springer-Verlag 1987
Evidence for sequential appearance of cartilage matrix proteins in developing mouse limbs and in cultures of mouse mesenchymal cells Ahnders Franzen', Dick Heinegard ', and Michael Solursh' Department of Biology, University of Iowa, Iowa City, Iowa 52242 Department of Physiological Chemistry, University of Lund, Lund, Sweden
Abstract. The initiation of synthesis and the accumulation of four cartilage matrix proteins (type I1 collagen and three noncollagenous proteins, one of M, 148, one of M, 59, and an oligometric protein of M, above 500 with 100-kDa subunits, respectively) were studied in developing mouse limbs and in cultures of limb bud mesenchyme by means of immunolocalization. On day 13 of gestation, type I1 collagen was observed throughout the entire humerus, whereas the 148-kDa protein was localized only in the central portion. Neither the 100-kDa-subunit protein nor the 59-kDa protein could be demonstrated in the humerus at that stage. On day 141/2, type I1 collagen and the 148-kDa protein were codistributed throughout the humerus. The 100-kDasubunit protein was detectable in the periphery of the humerus, whereas little 59-kDa protein could yet be demonstrated. On day 18, all four proteins being studied could be detected immunologically in the developing mouse humerus. They differed in immunolocalization. Type I1 collagen, the 148-kDa protein, and the 100-kDa-subunit protein were codistributed throughout the distal and proximal parts of the cartilage. However, the 148-kDa protein could no longer be detected immunochemically in the outermost part of the cartilage in the proximal shoulder joint. The 148-kDa protein codistributed with type I1 collagen and the 100-kDa-subunit protein in the distal cartilaginous region, where joint development was less advanced. On the other hand, the 59-kDa protein was not demonstrated directly within the hyaline cartilaginous structures, but surrounded the entire structure. This protein was also present in the same part of the proximal joint region as that in which the 148-kDa protein was not detected. To develop an in vitro model for studies of skeletogenesis, mesenchymal cells prepared from mouse limb buds were cultured as micromass cultures at high initial cell density to favor chondrogenesis. On day 3 of culture, type I1 collagen was the only protein that could be detected immunochemically in the cultures, whereas on day 6 the 148-kDa protein was demonstrated, and a few chondrocytes in the central portion of each cartilaginous nodule were associated with the 100-kDa-subunit protein. The 59-kDa protein could not yet be immunochemically detected. On day 14, type I1 collagen, the 148-kDa protein, and the 100-kDa-subunit protein could be detected throughout the cartilaginous nodules, whereas the 59-kDa protein was distributed in tissues surrounding each nodule. On day 55, the distribution of the four cartilage matrix proteins was the same on day 14, except that the 59-kDa pro-
tein was also demonstrated near some of the chondrocytes. In conclusion, the four cartilage matrix proteins being studied appeared in sequence. The order of appearance in the developing mouse humerus was identical to that in cultures of mesenchymal cells from the early mouse limb bud. These results indicate that the mesenchymal cell culture system can be used as an in vitro model for studies for skeletogenesis.
Introduction Cartilage, like many other connective tissues, is characterized by its low content of cells. The cartilage cells are scattered throughout the tissue and separated by an abundant extracellular matrix. The major molecular constituents of that matrix are the large aggregating proteoglycans and type I1 collagen (for review, see [5]). However, more recently other molecules representing minor constituents have been isolated from cartilage. Among these are: type IX collagen [19]; a 148-kDa protein [15]; 58-kDa and 59-kDa proteins [7]; an oligomeric protein, of M, greater than 500, that has subunits of 100 kDa (Heinegard, to be published) and is possibly identical to a protein described by Fife and Brandt [4]; fibronectin [21] ; chondronectin [8]; two types of small proteoglycan [6]; and link proteins [21], which stabilize the aggregates of large proteoglycans. Type X collagen is restricted to hypertrophic cartilage [I 71. While a number of the molecular constituents in cartilage are known, we are far from knowing how these components are assembled at the molecular level, forming the functional cartilage matrix; nor is it known whether these components are synthesized at the same time and/or by the same population of chondrocytes. Such questions are of considerable importance for understanding cartilage biology during normal skeletal development, as well as in various disease processes including joint disease. Some relevant information can be obtained from studies of chondrocyte differentiation. Over the years, the developing limb, and also cultures of limb bud mesenchymal cells, have been widely used as models for studies of chondrocyte differentiation (for review, see [18]). Differentiation of chondrocytes initially occurs among clustered mesenchymal cells [3]. The process is preceded by cell-to-cell interaction between certain mesenchymal cells. The newly differen-
200
Fig. 1 A-C. Chondrogenesis and osteogenesis in the developing mouse forelimb. Forelimb buds were prepared from mouse embryos at 13 (A), 141/2 (B), and 18 (C) days of gestation. The structures were stained for cartilage with Alcian blue and for bone with Alizarin red. The arrow in B shows the localization of initial bone formation; R,radius; U , ulna; A x 100; B x 7 5 ; C x 25
tiated chondrocytes, morphologically characterized by their roundish form, start to synthesize an extracellular matrix containing proteoglycans and type I1 collagen. Since none of the other minor cartilage matrix proteins described above has been studied during initial chondrogenesis, the present investigation was undertaken to monitor by immunochemistry the appearance of three noncollagenous proteins, i.e., the 148-kDa protein, the 100-kDa-subunit protein, and the 59-kDa protein, with reference to the appearance of type I1 collagen (as a major marker for cartilage) in the developing mouse limb and in cultures of mesenchymal cells from mouse limb buds. The results showed that the four proteins appeared in a specific sequence during initial chondrogenesis and that these events were the same in the developing mouse limb bud and in mesenchymal cell cultures.
Methods Tissue preparation. Pregnant mice of 11 'I2, 13, 14112, and 18 days of gestation (the day at vaginal plug was noted being day 0) were anesthetized with ether and rapidly killed by cervical dislocation. The embryos were dissected out under sterile conditions and placed in ice-cold Puck's saline G. The day-1 1'i2 embryos were used to establish micromass cultures of mesenchymal cells as described below. Forelimb buds from embryos at day 13, 14'12, and 18 were dissected out, embedded in O.C.T. compound (Miles Scientific) and subjected to frozen sectioning at -25" C in a cryostat. Sections 10 pm thick were picked up on albumin-coated glass slides. The frozen sections were subsequently fixed in icecold acetone and stored at -20" C until being processed further. Some other forelimb buds of various stages were fixed in buffered formalin, stained with Alcian blue (ICN Pharmaceuticals, Inc., New York, USA), cleared in potassium hydroxide and subsequently stained with Alizarin red S (Fisher Sci, Co., New Jersey, USA) as described elsewhere [91.
Cell culture. Forelimb buds at day 111/, of gestation were used to establish mesenchymal cell cultures according to the method of Owens and Solursh [14]. The cells were plated out as micromass cultures with an initial cell density above
confluency (2 x l o 5 cells in 10 PI). The cultures were established as 10-p1 dots in the middle of 35-mm tissue culture dishes. The cells were cultured in 1.5 ml CMRL medium supplemented with 10% fetal calf serum and antibiotics as described [14]. At certain times (i.e., at 3, 6, 14, and 55 days after plating), cultures were briefly rinsed twice with saline G to remove proteins present in the medium, embedded in O.C.T. compound and subjected to frozen sectioning as described above. Immunojhorescence. The frozen sections prepared from mouse forelimb buds (as described above) and from mesenchymal cell cultures were hydrated for 10 min with 20 m M phosphate-buffered saline (PBS) prior to immunostaining. The sections were then incubated with testicular hyaluronidase, to digest the glycosaminoglycan side chains in the cartilage proteoglycans and thereby enhance the tissue penetration by the antibodies [20]. Testicular hyaluronidase (Worthington Biochemical Co., New Jersey, USA) was solubilized (1 mg/ml) in 0.15 A4 sodium chloride with 20 m M sodium acetate, pH 5.0. From that solution, 50 pl was pipetted onto each section. The enzyme digestion was performed for 30 min at room temperature, followed by three rinses with PBS. Each experiment included at least four consecutive sections. One section was always double-stained with mouse hybridoma supernatant (dilution, 1 : 2) directed against type I1 collagen [13] and with rabbit polyclonal antibodies (dilution, 1 :40) directed against the 148-kDa protein [15]. Other sections were stained with rabbit polyclonal antibodies (dilution, 1 :40) directed against the 100-kDa-subunit protein (Heinegard, to be published) and the 59-kDa protein [7], respectively. Control sections were incubated with normal mouse and rabbit sera and with PBS, the latter for monitoring the nonspecific binding of the second antibody. The sections were incubated with the primary antibodies (or mouse and rabbit sera as indicated) for 30 min at room temperature. After rinsing the sections with PBS, they were incubated with a 1 :300 dilution of a mixture of two secondary antibodies: goat antibodies to mouse IgG, labelled with tetramethylrhodamine isothiocyanate (TRITC) ; and goat antibodies to rabbit IgG, labelled with fluorescein isothiocyanate (FITC ; Cappel Laboratories, Inc., Cochranville, PA). Incubation with the secondary antibodies was performed for 30 min at room temperature.
201
Fig. 2A-C. Immunolocalization of cartilage matrix proteins in the developing mouse forelimb on day 13 of gestation. A sagittal section of a forelimb bud from a mouse embryo at 13 days of gestation was examined by phase-contrast microscopy (A) and subsequently double-stained with rabbit polyclonal antibodies directed against a 148-kDa cartilage matrix protein (B) and mouse monoclonal antibodies directed against type I1 collagen. The primary antibodies were visualized with fluorescein-labelled goat antibodies to rabbit immunoglobulins (B) and with rhodamine-labelled rabbit antibodies to mouse immunoglobulins (C); R, radius; U , ulna. x 70
After rinsing as described above, the sections were mounted with coverslips in PPD: glycerol [lo] and photographed using a Leitz Laborlux microscope with a Ploemopak illuminator and an L2 or N2 filter cube (E. Leitz, Inc., Rockleigh, New Jersey, USA). In some cases, frozen sections were stained for calcium deposits by the Von Kossa method and counterstained with safranin [9]. Results Characterization of the developing mouse limb
Forelimbs from mouse embryos at day 13, 141/2, and 18 of gestation were used to study the development of the humerus. Since the humerus is first formed as a cartilagi-
nous model, which later becomes partly ossified, the different developmental stages were initially characterized by the extent of that cartilage matrix. Whole specimens were therefore stained with Alcian blue in order to demonstrate the glycosaminoglycans (i.e., the proteoglycans) in the cartilage matrix. The limbs were also stained with Alizarin red to localize bone tissue (Fig. 1A-C). The cartilaginous humerus was almost fully formed on day 13 as indicated by the presence of the Alcian blue staining (Fig. IA). The central portion showed more pronounced staining than the distal and proximal portions, indicating that chondrogenesis began in the central portion and then gradually expanded, both distally and proximally. The same feature was also found in the radius and ulna of the same limb (Fig. 1A). With time, the cartilaginous
202
Fig. 3A-I. Immunolocalization of four cartilage matrix proteins in the developing mouse humerus at day 141/2 of gestation. Sagittal sections of a mouse forelimb bud at 141/2days of gestation were examined by phase-contrast microscopy (A) and subjected to immunolocalization for type11 collagen (B, F), the 148-kDa protein (C, G), the 100-kDA-subunit protein (D, H) and the 59-kDa protein (E, I). The arrow in A indicates the initial bone formation. x 50
portion of the humerus increased and reached its final extent around day 14'J2 (Fig. 1 B). The whole humerus consisted of cartilage, except for a thin bone collar which surrounded the diaphysis (Fig. 1 B, arrow). Also, the two easily identifiable epiphyseal structures in the distal and proximal regions of the humerus had started to form joint structures with the shoulder girdle as well as with the radius and ulna. In the mouse humerus on day 18, most of the diaphyseal cartilaginous tissue had been replaced by bone and the cartilage remained in the distal and proximal structures forming the epiphyses (Fig. 1C). Joint development was also more advanced than on day 14'12. For further study, these three stages were selected from a large number of stages, because day 13 represents a stage of initial chondrogenesis, and day 14'12 a stage when the structure has reached its final shape. The day-18 stage was selected because the cartilaginous tissues in the proximal and distal regions have become more mature and special-
ized, resembling articular cartilage, as parts of functional joints. Immunolocalization of cartilage matrix proteins during chondrogenesis in the developing mouse humerus The three stages (i.e., day 13, 14'J2, and 18 of the mouse humerus) were frozen-sectioned and immunostained. Immunolocalization was used to study the distribution of four different cartilage matrix proteins : type I1 collagen, and three noncollagenous proteins - one of M, 148 [15], one with 100-kDa subunits (Heinegard, to be published), and one of M, 59 [7]. During initial experiments we used polyclonal rabbit antibodies directed against the four proteins, isolated from bovine nasal cartilage. Our preliminary results suggested that the proteins appeared in a specific order. However, the difference in distribution of type I1 collagen
203
Fig. 4A-I. Immunolocalization of four cartilage matrix proteins in the mouse humerus on day 18 of gestation, sectioned in the sagittal plane. The sections were examined by phase-contrast microscopy (A) and after incubation with antibodies directed against type I1 collagen (B, F), the 148-kDa protein (C,G), the 100-kDa-subunit protein (D, H) and the 59-kDa protein (E, I). The boxed area3 on the left (proximal) and right (distal) in A were further analyzed by immunolocalization. A x 20; B-J x 25
and the 148-kDa protein was sometimes difficult to visualize in consecutive sections. Double staining for the two proteins on the same section using a monoclonal mouse antibody directed against type I1 collagen [13] and polyclonal rabbit antibodies directed against the 148-kDa protein [16] was then utilized. The staining pattern of type I1 collagen in the cartilage matrix was identical, whether obtained with polyclonal rabbit antibodies or monoclonal mouse antibodies (data not shown). Sagittal sections of forelimb from day-13 mouse embryos were examined. A phase-contrast picture of a representative section of the whole limb is shown in Fig. 2A; the humerus is identified within the marked area, while the radius and ulna are indicated by letters. Consecutive sections were incubated with antibodies directed against the four cartilage matrix proteins described above. One section
was simultaneously incubated with monoclonal mouse antibody directed against type I1 collagen and with polyclonal rabbit antibodies against the 148-kDa protein. Other sections were incubated with polyclonal rabbit antibodies directed against the 100-kDa-subunit and the 59-kDa proteins, respectively. Both the 148-kDa protein (Fig. 219 and type I1 collagen (Fig. 2C) were found in the day-13 mouse forelimb, whereas no immunostaining could be demonstrated for the 100-kDa-subunit protein and the 59-kDa protein (data not shown). The distributions of type I1 collagen and the 148-kDa protein were also quite different. Type I1 collagen was widespread throughout the humerus, even in the distal and proximal ends. The 148-kDa protein, on the other hand, was mainly seen in the center of the structure, where type I1 collagen codistributed. The data also indicated that the appearance of type I1 collagen pre-
204
Figs. 5-8. Sections of mesenchymal cell cultures from mouse limb buds at 3 (Fig. 5), 6 (Fig. 6), 14 (Fig. 7) and 55 (Fig. 8) days after initial plating. The sections were examined by phase-contrast microscopy (A-D) and incubated with antibodies directed against type I1 collagen (E), the 148-kDa protein (F), the 100-kDa-subunit protein ( C ) and the 59-kDa protein (H). Magnifications: Fig. 5 x425; Fig. 6 x425; Fig. 7 x 175; Fig. 8 x 175
ceded the 148-kDa protein during the development of the distal as well a s proximal parts of the mouse humerus. Furthermore, the distribution of type I1 collagen throughout the structure (Fig. 2C) was almost identical to the distribution of proteoglycans demonstrated by Alcian blue staining (Fig. 1A). A longitudinal section of a day-14'/, mouse humerus is shown in Fig. 3A. The structure is composed of cartilagi-
nous epiphyses at either end of a cartilage diaphysis, which in the middle is surrounded by a tiny bone collar (Fig. 3A, arrow). From the phase-contrast picture one can conclude that the epiphyseal cartilage is less mature than that in the more-central region. In the outermost parts of the epiphyseal cartilage structures, there is ongoing appositional cartilage formation (Fig. 3A). The present study focuses entirely on the processes that occur in the distal and proxi-
205
Fig. 6
ma1 regions of the developing humerus. Consecutive sections of the day-14'1, mouse humerus were incubated with antibodies directed against type I1 collagen (Fig. 3 B, F), the 148-kDa protein (Fig. 3C, G), the 100-kDa-subunit protein (Fig. 3 D, H), and the 59-kDa protein (Fig. 3 E, I). All four cartilage matrix proteins were present in the day14'/2 mouse humerus, but in varying amounts. Type I1 collagen (Fig. 3B, F) and the 148-kDa protein (Fig. 3C, G) were immunochemically monitored in the same section. Immunostaining for type I1 collagen was more extensively dis-
tributed than that for the 148-kDa protein. The intensity of the immunostaining of these two proteins was the same in the more-central portion of the structure (Fig. 3B, C, black arrows), whereas it differed in the proximal area (Fig. 3B, C, white arrows). The situation was the same in the distal humerus (Fig. 3F, G). These data indicate that chondrogenesis occurs in a proximal as well as in a distal direction and that the appearance of type I1 collagen precedes the appearance of the 148-kDa protein. Moreover, the 100-kDa-subunit protein was present in the day-l4l/,
206
Fig. 7
mouse humerus. The protein was immunolocalized primarily in the periphery of the humerus (Fig. 3D, H). Weak immunostaining for the 59-kDa protein was seen in the distal region of the humerus (Fig. 31) but not in the corresponding proximal structure (Fig. 3 E). It is hard to understand why the 100-kDa-subunit protein was present only in the periphery of the humerus, whereas type I1 collagen and the 148-kDa protein were widespread throughout the entire cartilaginous structure. Chemical data, however, indicate that the 100-kDa-subunit protein is much enriched in
articular cartilage when compared with other types of cartilage (Heinegard, unpublished). It is thus possible that the differential staining indicated an early commitment to form articular cartilage. The distribution of the four proteins was examined at later stages of development. A phase-contrast micrograph of the day-18 mouse humerus (Fig. 4A) showed two main structural differences when compared to the day-14'1, mouse humerus (Fig. 3A). First, most of the cartilaginous diaphysis had been replaced by bone tissue. Second, the
201
Fig. 8
cartilaginous epiphyses had developed further, and in the proximal epiphysis a primitive joint space had been formed (Fig. 4A). The immunolocalization of type I1 collagen (Fig. 4B, F) and the 148-kDa protein (Fig. 4C, G) was the same in the proximal and distal regions. Furthermore, the intensity of the fluorescence was also the same. However, the only region where they did not codistribute immunochemically was in the outermost proximal epiphysis (Fig. 4C, black arrows). For instance, compare the joint space in Fig. 4B and 4C, where the joint space in the latter
is somewhat wider due mainly to the absence of immunostaining for the 148-kDa protein. Since there was no great difference in distribution between the two proteins in the distal region, the absence of the 148-kDa protein could have occurred because the proximal joint was at a moredeveloped stage than the distal one. The 100-kDa-subunit protein codistributed immunochemically with type I1 collagen and the 148-kDa protein in the proximal epiphysis (Fig. 4D), but had a more scattered and superficial distribution in the distal epiphysis, consistent with preferential lo-
208
Fig. 9. Section of mesenchyme cell culture from mouse limb buds at day 14 stained by the method of von Kossa and countcrstained with safranin; note the calcium deposits associated with hypertrophic chondrocytes and around nodules. It is not yet known whether osteocytes were present in these cultures. x 445
calization to articular cartilage (Fig. 4H). The 59-kDa protein was distributed similarly within the proximal (Fig. 4E) and the distal (Fig. 41) epiphyses. The protein was immunolocalized in structures surrounding the two epiphyses, and was not present within the epiphyseal cartilage. Immunolocalization of cartilage matrix proteins during chondrogenesis by cultured limb bud cells
To establish an in vitro model for skeletogenesis, we compared the sequence in which four cartilage matrix proteins appear in cultured limb bud cells to that in developing mouse limbs in situ. Mesenchymal cells were isolated from mouse forelimb buds and plated out as micromass cultures (i.e., with initial cell density above confluency). The cultures were embedded for frozen sectioning at 3, 6, 14, and 55 days after plating. Consecutive sections obtained at each time were incubated with antibodies directed against type I1 collagen, the 148-kDa protein, the 100-kDa protein subunit and the 59-kDa protein. The distribution of the proteins was then studied by immunofluorescence. On day 3 (Fig. 5, E-H), the mesenchymal cells had started to form nodules. Type I1 collagen was the only protein of the four studied which could be identified in sections of the day-3 mesenchyma1 cell cultures (Fig. 5E). None of the other proteins, i.e., the 148-kDa protein (Fig. 5F), the 100-kDa-subunit protein (Fig. 5G) and the 59-kDa protein (Fig. 5H) could be detected at this stage. The nodules in the day-6 mesenchymal cell cultures were composed of hypertrophied chondrocytes surrounded by a sparse extracellular cartilage matrix (Fig. 6E-H). Type I1 collagen (Fig. 6E) and the 148-kDa protein (Fig. 6F) were immunochemically detected throughout the cartilaginous tissue within the nodule. Type I1 collagen was somewhat more pronounced in the peripheral tissue than in the 148-kDa protein. Since type I1 collagen and the 148-kDa protein were immunochemically monitored in the same section, it was possible to study the accumulation of the two proteins around individual chondrocytes within the nodule.
These results indicated that sometimes the two proteins did not accumulate to the same extent around a single chondrocyte. Since immunohistochemistry only detects antigen accumulation, the relative roles of synthesis and stability are not known. A few chondrocytes, mostly in the periphery of the cartilaginous nodule, were also associated with the 100-kDa-subunit protein (Fig. 6G), whereas this protein was not identified in the central portion of the nodule. This differential localization closely resembled that in the developing humerus. Moreover, the 59-kDa protein could not be immunochemically detected in the day-6 culture (Fig. 6H). In the day-I 4 cultures, therc were multiple cartilaginous nodules in close contact with each other (Fig. 7A-D). At this time calcium deposits could be detected in association with many of the cells around nodules as well as with the hypertrophic chondrocytes within nodules (Fig. 9). In addition, all of the four cartilage matrix proteins being studied were detected in the day-14 cultures (Fig. 7E-H). The 148-kDa protein (Fig. 7F) and the 100-kD-subunit protein (Fig. 7G) were present throughout the cartilage matrix, whereas the type I1 collagen (Fig. 7E) molecules were more pronounced among the peripheral cartilaginous cells. The 59-kDa protein was not immunochemically localized within the cartilaginous nodules, but instead surrounded each nodule (Fig. 7H). On day 55, the nodules had increased in size compared to that on day 14. However, the cartilaginous portion of the nodule did not seem to have increased as much. Thus, the growth occurred mainly in tissues surrounding each nodule (Fig. 8A-D). Except for the increased size of the nodules, the distribution of the four proteins (Fig. 8 E-H) was quite similar to that on day 14. However, on day 55 the 59-kDa protein was also localized near some chondrocytes in the cartilaginous portion of the nodule (Fig. 8 H). Discussion
In the present study, evidence is presented supporting the sequential appearance of four cartilage matrix proteins
-
209
type I1 collagen, 148-kDa protein [I 51, 100-kDa-subunit protein (Heinegard, to be published) and 59-kDa protein [7], in developing mouse limbs and in cultures of limb bud mesenchymal cells. The four proteins studied appeared in the following order: type I1 collagen, 148-kDa protein, 100-kDa-subunit protein, and 59-kDa protein. There are several possible mechanisms for the sequential appearance of these proteins. The main regulatory events are likely to occur at the transcriptional level and also the translational level. It has been shown that the biosynthesis of type I1 collagen is regulated at both the transcriptional and translational levels in initial chondrogenesis of the chicken wing. The mRNA for type I1 collagen can be detected about 12 h before the protein is immunochemically detected in the extracellular matrix [12]. Furthermore, there is also sequential expression of the genes for type I1 collagen and proteoglycans during chondrocyte differentiation in the developing chicken wing and in cultures of chicken limb bud cells [Ill. Another mechanism for the sequential appearance of the four cartilage matrix proteins could be that they are synthesized by different populations of chondrocytes, each differentiating at various times during chondrogenesis. This appears unlikely for type I1 collagen and the 148-kDa protein, since these appear to codistribute at some stages of development. Notably, the 148-kDa protein showed limited expression by peripheral chondrocytes. The 100-kDa-subunit protein at times showed preferential localization to peripheral cartilage, subsequently forming the articular surface. The sequential appearance of these proteins indicates that cartilage matrix is assembled in an ordered fashion. There was also a striking similarity between these events occurring in vivo and those occurring in vitro with regard to the sequential appearance and localization of the four proteins studied. This included the progression of chondrogenesis to the point of hypertrophy and finally the formation of mineralized cartilage. This was the case even though the mesenchymal cells were plated out in a randomized fashion after cell dissociation of the whole limb bud. The distribution of the proteins followed that of the developing limb in situ. Each cartilaginous nodule could be regarded as a single skeletal element, with type I1 collagen, 148-kDa protein, and the 100-kDa-subunit protein localized within the cartilaginous structure and the 59-kDa protein surrounding the nodule. The tendency for differential staining of central and peripheral portions, however, appears to have been lost in the cultures. Besides the sequential appearance of these proteins, this study yielded another important finding. Figure 4 C shows that the 148-kDa protein could not be detected irnmunohistochemically in the edge of the proximal epiphysis in the day-18 mouse humerus. The protein was present in the same region of the day-I4l/, mouse humerus (Fig. 3C) as well as in the distal region of day-IS mouse humerus. It is therefore likely that the biosynthesis or accumulation of the 148-kDa protein became inhibited in the edge of the cartilaginous epiphysis, which would later form the articular cartilage. The fact that the 148-kDa protein was absent from the presumptive articular cartilage region fits very well with the observation that the 148-kDa protein is not present in extracts of articular cartilage [16]. The major morphological difference between articular cartilage and other types of cartilage is that the former tissue is not surrounded by a perichondrium and has ceased growth.
A cDNA clone for the 148-kDa protein of chick cartilage has recently been isolated and sequenced [I]. The data show that the 148-kDa protein contains a sequence similar to that of epidermal growth factor. It has been suggested that upon proteolytic cleavage the 148-kDa protein can function as a local growth factor during active chondrogenesis. This protein can be detected immunohistochemically in cartilage that is formed in mesenchymal cell cultures, at all stages studied so far. Apparently articular cartilage does not form in these cultures, since there is no cartilage structure lacking the 148-kDa protein. It is possible that the biosynthesis of the 148-kDa protein is suppressed in the outermost part of the cartilage in the day-18 mouse humerus by the production of a specific substance by some cells in the developing joint. This hypothesis warrants testing. Acknowledgments. The authors thank Dr. D.F. Paulsen for his valuable comments on the manuscript, Virginia Dress and Dwight Moulton for technical assistance, and Karen Kriege for typing the manuscript. This work was supported by NIH grants DE05837 and HD05505.
References 1. Argraves WS, Deak F, Sparks KT, Kiss I, Goetinck P F (1987) Structural features of cartilage matrix protein deduced from cDNA. Proc Natl Acad Sci USA 84:464-468 2. Baker TR, Caterson B (2979) The link proteins as specific components of cartilage proteoglycan aggregates in vivo. J Biol Chem 254 :2394-2399 3. Fell HB (1925) The histogenesis of cartilage and bone in the long bones of the embryonic fowl. J Morphol40:417-451 4. Fife R, Brandt K (1984) Identification of a high-molecularweight ( >400,000) protein in hyaline cartilage. Biochim Biophys Acta 802:506-514 5. Heinegard D, Paulsson M (1987) Cartilage. In: Cunningham LW (ed) Methods in Enzymology, vol 145. Academic Press, Inc, Orlando, pp 336-362 6. Heinegard D, Bjorne-Persson A, Coster L, Franzen A, Gardell S, Malmstrom A, Paulsson M, Sandfalk R, Vogel K (1985) The core proteins of large and small interstitial proteoglycans from various connective tissues form distinct subgroups. Biochem J 230:181-194 7. Heinegard D, Larsson T, Sommarin Y, Franzen A, Paulsson M, Hedbom E (1986) Two novel matrix proteins isolated from articular cartilage show wide distributions among connective tissues. J Biol Chem 261 : 1386613872 8. Hewitt AT, Varner HH, Silver MH, Dessau W, Wilkes CM, Martin GR (1982) The isolation and partial characterization of chondronectin, an attachment factor for chondrocytes. J Biol Chem 257: 233G2334 9. Humason GL (1972) Animal tissue techniques. 3rd ed. WH Freeman and Company, San Francisco 10. Johnson GD, Araujo NC (1981) A simple method of reducing the fading of immunofluorescence during microscopy. J Immunol Methods 43 :349-350 11. Kosher RA, Gay SW, Kamanitz TJ, Kulyk WM, Rodgers BT, Sai S, Tdmaka T, Tanzer ML (1986) Cartilage proteoglycan core protein gene expression during limb cartilage differentiation. Dev Biol 118:112-117 12. Kravis D, Upholt WB (1985) Quantitation of type I1 procollagen mRNA levels during chick limb cartilage differentiation. Dev Biol 108:164-172 13. Linsenmdyer TF, Hendrix MJC (1980) Monoclonal antibodies to connective tissue macromolecules: type I1 collagen. Biochem Biophys Res Comm 92:440-446 14. Owens EM, Solursh M (1981) In vitro histogenic capacities
210 of limb mesenchyme from various stage mouse embryos. Dev Biol 88 :297-31 1 15. Paulsson M, Heinegard D (1981) Purification and structural characterization of a cartilage matrix protein. Biochem J 1971367-375 16. Paulsson M, Heinegard D (1982) Radioimmunoassay of the 148-kilodalton cartilage protein. Distribution of the protein among bovine tissues. Biochem J 207:207-213 17. Schmid TM, Conrad HE (1982) Metabolism of low molecular weight collagen by chondrocytes obtained from histologically distinct zones of the chick embryo tibiotarsus. J Biol Chem 257:12451-12457 18. Solursh M (1986) Environmental regulation of limb chondrogenesis. In: Kuettner K et al. (eds) Articular cartilage biochemistry. Raven Press, New York, pp 145-161
19. Vaughan L, Winterhalter HM, Bruckner P (1985) Proteoglycan Lt from chicken embryo sternum identified as type IX collagen. J Biol Chem 260: 47584763 20. von der Mark H, von der Mark K, Gay S (1976) Study of differential collagen synthesis during development of the chick embryo by immunofluorescence. Dev Biol48: 237-249 21. Wurster NB, Lust G (1982) Fibronectin in osteoarthritic canine articular cartilage. Biochem Biophys Res Comm 109:1094-1101
Received October 1987 / Accepted in revised form November 19, 1987
Differentiation (1987) 36: 199-210
0 Springer-Verlag 1987
Evidence for sequential appearance of cartilage matrix proteins in developing mouse limbs and in cultures of mouse mesenchymal cells Ahnders Franzen', Dick Heinegard ', and Michael Solursh' Department of Biology, University of Iowa, Iowa City, Iowa 52242 Department of Physiological Chemistry, University of Lund, Lund, Sweden
Abstract. The initiation of synthesis and the accumulation of four cartilage matrix proteins (type I1 collagen and three noncollagenous proteins, one of M, 148, one of M, 59, and an oligometric protein of M, above 500 with 100-kDa subunits, respectively) were studied in developing mouse limbs and in cultures of limb bud mesenchyme by means of immunolocalization. On day 13 of gestation, type I1 collagen was observed throughout the entire humerus, whereas the 148-kDa protein was localized only in the central portion. Neither the 100-kDa-subunit protein nor the 59-kDa protein could be demonstrated in the humerus at that stage. On day 141/2, type I1 collagen and the 148-kDa protein were codistributed throughout the humerus. The 100-kDasubunit protein was detectable in the periphery of the humerus, whereas little 59-kDa protein could yet be demonstrated. On day 18, all four proteins being studied could be detected immunologically in the developing mouse humerus. They differed in immunolocalization. Type I1 collagen, the 148-kDa protein, and the 100-kDa-subunit protein were codistributed throughout the distal and proximal parts of the cartilage. However, the 148-kDa protein could no longer be detected immunochemically in the outermost part of the cartilage in the proximal shoulder joint. The 148-kDa protein codistributed with type I1 collagen and the 100-kDa-subunit protein in the distal cartilaginous region, where joint development was less advanced. On the other hand, the 59-kDa protein was not demonstrated directly within the hyaline cartilaginous structures, but surrounded the entire structure. This protein was also present in the same part of the proximal joint region as that in which the 148-kDa protein was not detected. To develop an in vitro model for studies of skeletogenesis, mesenchymal cells prepared from mouse limb buds were cultured as micromass cultures at high initial cell density to favor chondrogenesis. On day 3 of culture, type I1 collagen was the only protein that could be detected immunochemically in the cultures, whereas on day 6 the 148-kDa protein was demonstrated, and a few chondrocytes in the central portion of each cartilaginous nodule were associated with the 100-kDa-subunit protein. The 59-kDa protein could not yet be immunochemically detected. On day 14, type I1 collagen, the 148-kDa protein, and the 100-kDa-subunit protein could be detected throughout the cartilaginous nodules, whereas the 59-kDa protein was distributed in tissues surrounding each nodule. On day 55, the distribution of the four cartilage matrix proteins was the same on day 14, except that the 59-kDa pro-
tein was also demonstrated near some of the chondrocytes. In conclusion, the four cartilage matrix proteins being studied appeared in sequence. The order of appearance in the developing mouse humerus was identical to that in cultures of mesenchymal cells from the early mouse limb bud. These results indicate that the mesenchymal cell culture system can be used as an in vitro model for studies for skeletogenesis.
Introduction Cartilage, like many other connective tissues, is characterized by its low content of cells. The cartilage cells are scattered throughout the tissue and separated by an abundant extracellular matrix. The major molecular constituents of that matrix are the large aggregating proteoglycans and type I1 collagen (for review, see [5]). However, more recently other molecules representing minor constituents have been isolated from cartilage. Among these are: type IX collagen [19]; a 148-kDa protein [15]; 58-kDa and 59-kDa proteins [7]; an oligomeric protein, of M, greater than 500, that has subunits of 100 kDa (Heinegard, to be published) and is possibly identical to a protein described by Fife and Brandt [4]; fibronectin [21] ; chondronectin [8]; two types of small proteoglycan [6]; and link proteins [21], which stabilize the aggregates of large proteoglycans. Type X collagen is restricted to hypertrophic cartilage [I 71. While a number of the molecular constituents in cartilage are known, we are far from knowing how these components are assembled at the molecular level, forming the functional cartilage matrix; nor is it known whether these components are synthesized at the same time and/or by the same population of chondrocytes. Such questions are of considerable importance for understanding cartilage biology during normal skeletal development, as well as in various disease processes including joint disease. Some relevant information can be obtained from studies of chondrocyte differentiation. Over the years, the developing limb, and also cultures of limb bud mesenchymal cells, have been widely used as models for studies of chondrocyte differentiation (for review, see [18]). Differentiation of chondrocytes initially occurs among clustered mesenchymal cells [3]. The process is preceded by cell-to-cell interaction between certain mesenchymal cells. The newly differen-
200
Fig. 1 A-C. Chondrogenesis and osteogenesis in the developing mouse forelimb. Forelimb buds were prepared from mouse embryos at 13 (A), 141/2 (B), and 18 (C) days of gestation. The structures were stained for cartilage with Alcian blue and for bone with Alizarin red. The arrow in B shows the localization of initial bone formation; R,radius; U , ulna; A x 100; B x 7 5 ; C x 25
tiated chondrocytes, morphologically characterized by their roundish form, start to synthesize an extracellular matrix containing proteoglycans and type I1 collagen. Since none of the other minor cartilage matrix proteins described above has been studied during initial chondrogenesis, the present investigation was undertaken to monitor by immunochemistry the appearance of three noncollagenous proteins, i.e., the 148-kDa protein, the 100-kDa-subunit protein, and the 59-kDa protein, with reference to the appearance of type I1 collagen (as a major marker for cartilage) in the developing mouse limb and in cultures of mesenchymal cells from mouse limb buds. The results showed that the four proteins appeared in a specific sequence during initial chondrogenesis and that these events were the same in the developing mouse limb bud and in mesenchymal cell cultures.
Methods Tissue preparation. Pregnant mice of 11 'I2, 13, 14112, and 18 days of gestation (the day at vaginal plug was noted being day 0) were anesthetized with ether and rapidly killed by cervical dislocation. The embryos were dissected out under sterile conditions and placed in ice-cold Puck's saline G. The day-1 1'i2 embryos were used to establish micromass cultures of mesenchymal cells as described below. Forelimb buds from embryos at day 13, 14'12, and 18 were dissected out, embedded in O.C.T. compound (Miles Scientific) and subjected to frozen sectioning at -25" C in a cryostat. Sections 10 pm thick were picked up on albumin-coated glass slides. The frozen sections were subsequently fixed in icecold acetone and stored at -20" C until being processed further. Some other forelimb buds of various stages were fixed in buffered formalin, stained with Alcian blue (ICN Pharmaceuticals, Inc., New York, USA), cleared in potassium hydroxide and subsequently stained with Alizarin red S (Fisher Sci, Co., New Jersey, USA) as described elsewhere [91.
Cell culture. Forelimb buds at day 111/, of gestation were used to establish mesenchymal cell cultures according to the method of Owens and Solursh [14]. The cells were plated out as micromass cultures with an initial cell density above
confluency (2 x l o 5 cells in 10 PI). The cultures were established as 10-p1 dots in the middle of 35-mm tissue culture dishes. The cells were cultured in 1.5 ml CMRL medium supplemented with 10% fetal calf serum and antibiotics as described [14]. At certain times (i.e., at 3, 6, 14, and 55 days after plating), cultures were briefly rinsed twice with saline G to remove proteins present in the medium, embedded in O.C.T. compound and subjected to frozen sectioning as described above. Immunojhorescence. The frozen sections prepared from mouse forelimb buds (as described above) and from mesenchymal cell cultures were hydrated for 10 min with 20 m M phosphate-buffered saline (PBS) prior to immunostaining. The sections were then incubated with testicular hyaluronidase, to digest the glycosaminoglycan side chains in the cartilage proteoglycans and thereby enhance the tissue penetration by the antibodies [20]. Testicular hyaluronidase (Worthington Biochemical Co., New Jersey, USA) was solubilized (1 mg/ml) in 0.15 A4 sodium chloride with 20 m M sodium acetate, pH 5.0. From that solution, 50 pl was pipetted onto each section. The enzyme digestion was performed for 30 min at room temperature, followed by three rinses with PBS. Each experiment included at least four consecutive sections. One section was always double-stained with mouse hybridoma supernatant (dilution, 1 : 2) directed against type I1 collagen [13] and with rabbit polyclonal antibodies (dilution, 1 :40) directed against the 148-kDa protein [15]. Other sections were stained with rabbit polyclonal antibodies (dilution, 1 :40) directed against the 100-kDa-subunit protein (Heinegard, to be published) and the 59-kDa protein [7], respectively. Control sections were incubated with normal mouse and rabbit sera and with PBS, the latter for monitoring the nonspecific binding of the second antibody. The sections were incubated with the primary antibodies (or mouse and rabbit sera as indicated) for 30 min at room temperature. After rinsing the sections with PBS, they were incubated with a 1 :300 dilution of a mixture of two secondary antibodies: goat antibodies to mouse IgG, labelled with tetramethylrhodamine isothiocyanate (TRITC) ; and goat antibodies to rabbit IgG, labelled with fluorescein isothiocyanate (FITC ; Cappel Laboratories, Inc., Cochranville, PA). Incubation with the secondary antibodies was performed for 30 min at room temperature.
201
Fig. 2A-C. Immunolocalization of cartilage matrix proteins in the developing mouse forelimb on day 13 of gestation. A sagittal section of a forelimb bud from a mouse embryo at 13 days of gestation was examined by phase-contrast microscopy (A) and subsequently double-stained with rabbit polyclonal antibodies directed against a 148-kDa cartilage matrix protein (B) and mouse monoclonal antibodies directed against type I1 collagen. The primary antibodies were visualized with fluorescein-labelled goat antibodies to rabbit immunoglobulins (B) and with rhodamine-labelled rabbit antibodies to mouse immunoglobulins (C); R, radius; U , ulna. x 70
After rinsing as described above, the sections were mounted with coverslips in PPD: glycerol [lo] and photographed using a Leitz Laborlux microscope with a Ploemopak illuminator and an L2 or N2 filter cube (E. Leitz, Inc., Rockleigh, New Jersey, USA). In some cases, frozen sections were stained for calcium deposits by the Von Kossa method and counterstained with safranin [9]. Results Characterization of the developing mouse limb
Forelimbs from mouse embryos at day 13, 141/2, and 18 of gestation were used to study the development of the humerus. Since the humerus is first formed as a cartilagi-
nous model, which later becomes partly ossified, the different developmental stages were initially characterized by the extent of that cartilage matrix. Whole specimens were therefore stained with Alcian blue in order to demonstrate the glycosaminoglycans (i.e., the proteoglycans) in the cartilage matrix. The limbs were also stained with Alizarin red to localize bone tissue (Fig. 1A-C). The cartilaginous humerus was almost fully formed on day 13 as indicated by the presence of the Alcian blue staining (Fig. IA). The central portion showed more pronounced staining than the distal and proximal portions, indicating that chondrogenesis began in the central portion and then gradually expanded, both distally and proximally. The same feature was also found in the radius and ulna of the same limb (Fig. 1A). With time, the cartilaginous
202
Fig. 3A-I. Immunolocalization of four cartilage matrix proteins in the developing mouse humerus at day 141/2 of gestation. Sagittal sections of a mouse forelimb bud at 141/2days of gestation were examined by phase-contrast microscopy (A) and subjected to immunolocalization for type11 collagen (B, F), the 148-kDa protein (C, G), the 100-kDA-subunit protein (D, H) and the 59-kDa protein (E, I). The arrow in A indicates the initial bone formation. x 50
portion of the humerus increased and reached its final extent around day 14'J2 (Fig. 1 B). The whole humerus consisted of cartilage, except for a thin bone collar which surrounded the diaphysis (Fig. 1 B, arrow). Also, the two easily identifiable epiphyseal structures in the distal and proximal regions of the humerus had started to form joint structures with the shoulder girdle as well as with the radius and ulna. In the mouse humerus on day 18, most of the diaphyseal cartilaginous tissue had been replaced by bone and the cartilage remained in the distal and proximal structures forming the epiphyses (Fig. 1C). Joint development was also more advanced than on day 14'12. For further study, these three stages were selected from a large number of stages, because day 13 represents a stage of initial chondrogenesis, and day 14'12 a stage when the structure has reached its final shape. The day-18 stage was selected because the cartilaginous tissues in the proximal and distal regions have become more mature and special-
ized, resembling articular cartilage, as parts of functional joints. Immunolocalization of cartilage matrix proteins during chondrogenesis in the developing mouse humerus The three stages (i.e., day 13, 14'J2, and 18 of the mouse humerus) were frozen-sectioned and immunostained. Immunolocalization was used to study the distribution of four different cartilage matrix proteins : type I1 collagen, and three noncollagenous proteins - one of M, 148 [15], one with 100-kDa subunits (Heinegard, to be published), and one of M, 59 [7]. During initial experiments we used polyclonal rabbit antibodies directed against the four proteins, isolated from bovine nasal cartilage. Our preliminary results suggested that the proteins appeared in a specific order. However, the difference in distribution of type I1 collagen
203
Fig. 4A-I. Immunolocalization of four cartilage matrix proteins in the mouse humerus on day 18 of gestation, sectioned in the sagittal plane. The sections were examined by phase-contrast microscopy (A) and after incubation with antibodies directed against type I1 collagen (B, F), the 148-kDa protein (C,G), the 100-kDa-subunit protein (D, H) and the 59-kDa protein (E, I). The boxed area3 on the left (proximal) and right (distal) in A were further analyzed by immunolocalization. A x 20; B-J x 25
and the 148-kDa protein was sometimes difficult to visualize in consecutive sections. Double staining for the two proteins on the same section using a monoclonal mouse antibody directed against type I1 collagen [13] and polyclonal rabbit antibodies directed against the 148-kDa protein [16] was then utilized. The staining pattern of type I1 collagen in the cartilage matrix was identical, whether obtained with polyclonal rabbit antibodies or monoclonal mouse antibodies (data not shown). Sagittal sections of forelimb from day-13 mouse embryos were examined. A phase-contrast picture of a representative section of the whole limb is shown in Fig. 2A; the humerus is identified within the marked area, while the radius and ulna are indicated by letters. Consecutive sections were incubated with antibodies directed against the four cartilage matrix proteins described above. One section
was simultaneously incubated with monoclonal mouse antibody directed against type I1 collagen and with polyclonal rabbit antibodies against the 148-kDa protein. Other sections were incubated with polyclonal rabbit antibodies directed against the 100-kDa-subunit and the 59-kDa proteins, respectively. Both the 148-kDa protein (Fig. 219 and type I1 collagen (Fig. 2C) were found in the day-13 mouse forelimb, whereas no immunostaining could be demonstrated for the 100-kDa-subunit protein and the 59-kDa protein (data not shown). The distributions of type I1 collagen and the 148-kDa protein were also quite different. Type I1 collagen was widespread throughout the humerus, even in the distal and proximal ends. The 148-kDa protein, on the other hand, was mainly seen in the center of the structure, where type I1 collagen codistributed. The data also indicated that the appearance of type I1 collagen pre-
204
Figs. 5-8. Sections of mesenchymal cell cultures from mouse limb buds at 3 (Fig. 5), 6 (Fig. 6), 14 (Fig. 7) and 55 (Fig. 8) days after initial plating. The sections were examined by phase-contrast microscopy (A-D) and incubated with antibodies directed against type I1 collagen (E), the 148-kDa protein (F), the 100-kDa-subunit protein ( C ) and the 59-kDa protein (H). Magnifications: Fig. 5 x425; Fig. 6 x425; Fig. 7 x 175; Fig. 8 x 175
ceded the 148-kDa protein during the development of the distal as well a s proximal parts of the mouse humerus. Furthermore, the distribution of type I1 collagen throughout the structure (Fig. 2C) was almost identical to the distribution of proteoglycans demonstrated by Alcian blue staining (Fig. 1A). A longitudinal section of a day-14'/, mouse humerus is shown in Fig. 3A. The structure is composed of cartilagi-
nous epiphyses at either end of a cartilage diaphysis, which in the middle is surrounded by a tiny bone collar (Fig. 3A, arrow). From the phase-contrast picture one can conclude that the epiphyseal cartilage is less mature than that in the more-central region. In the outermost parts of the epiphyseal cartilage structures, there is ongoing appositional cartilage formation (Fig. 3A). The present study focuses entirely on the processes that occur in the distal and proxi-
205
Fig. 6
ma1 regions of the developing humerus. Consecutive sections of the day-14'1, mouse humerus were incubated with antibodies directed against type I1 collagen (Fig. 3 B, F), the 148-kDa protein (Fig. 3C, G), the 100-kDa-subunit protein (Fig. 3 D, H), and the 59-kDa protein (Fig. 3 E, I). All four cartilage matrix proteins were present in the day14'/2 mouse humerus, but in varying amounts. Type I1 collagen (Fig. 3B, F) and the 148-kDa protein (Fig. 3C, G) were immunochemically monitored in the same section. Immunostaining for type I1 collagen was more extensively dis-
tributed than that for the 148-kDa protein. The intensity of the immunostaining of these two proteins was the same in the more-central portion of the structure (Fig. 3B, C, black arrows), whereas it differed in the proximal area (Fig. 3B, C, white arrows). The situation was the same in the distal humerus (Fig. 3F, G). These data indicate that chondrogenesis occurs in a proximal as well as in a distal direction and that the appearance of type I1 collagen precedes the appearance of the 148-kDa protein. Moreover, the 100-kDa-subunit protein was present in the day-l4l/,
206
Fig. 7
mouse humerus. The protein was immunolocalized primarily in the periphery of the humerus (Fig. 3D, H). Weak immunostaining for the 59-kDa protein was seen in the distal region of the humerus (Fig. 31) but not in the corresponding proximal structure (Fig. 3 E). It is hard to understand why the 100-kDa-subunit protein was present only in the periphery of the humerus, whereas type I1 collagen and the 148-kDa protein were widespread throughout the entire cartilaginous structure. Chemical data, however, indicate that the 100-kDa-subunit protein is much enriched in
articular cartilage when compared with other types of cartilage (Heinegard, unpublished). It is thus possible that the differential staining indicated an early commitment to form articular cartilage. The distribution of the four proteins was examined at later stages of development. A phase-contrast micrograph of the day-18 mouse humerus (Fig. 4A) showed two main structural differences when compared to the day-14'1, mouse humerus (Fig. 3A). First, most of the cartilaginous diaphysis had been replaced by bone tissue. Second, the
201
Fig. 8
cartilaginous epiphyses had developed further, and in the proximal epiphysis a primitive joint space had been formed (Fig. 4A). The immunolocalization of type I1 collagen (Fig. 4B, F) and the 148-kDa protein (Fig. 4C, G) was the same in the proximal and distal regions. Furthermore, the intensity of the fluorescence was also the same. However, the only region where they did not codistribute immunochemically was in the outermost proximal epiphysis (Fig. 4C, black arrows). For instance, compare the joint space in Fig. 4B and 4C, where the joint space in the latter
is somewhat wider due mainly to the absence of immunostaining for the 148-kDa protein. Since there was no great difference in distribution between the two proteins in the distal region, the absence of the 148-kDa protein could have occurred because the proximal joint was at a moredeveloped stage than the distal one. The 100-kDa-subunit protein codistributed immunochemically with type I1 collagen and the 148-kDa protein in the proximal epiphysis (Fig. 4D), but had a more scattered and superficial distribution in the distal epiphysis, consistent with preferential lo-
208
Fig. 9. Section of mesenchyme cell culture from mouse limb buds at day 14 stained by the method of von Kossa and countcrstained with safranin; note the calcium deposits associated with hypertrophic chondrocytes and around nodules. It is not yet known whether osteocytes were present in these cultures. x 445
calization to articular cartilage (Fig. 4H). The 59-kDa protein was distributed similarly within the proximal (Fig. 4E) and the distal (Fig. 41) epiphyses. The protein was immunolocalized in structures surrounding the two epiphyses, and was not present within the epiphyseal cartilage. Immunolocalization of cartilage matrix proteins during chondrogenesis by cultured limb bud cells
To establish an in vitro model for skeletogenesis, we compared the sequence in which four cartilage matrix proteins appear in cultured limb bud cells to that in developing mouse limbs in situ. Mesenchymal cells were isolated from mouse forelimb buds and plated out as micromass cultures (i.e., with initial cell density above confluency). The cultures were embedded for frozen sectioning at 3, 6, 14, and 55 days after plating. Consecutive sections obtained at each time were incubated with antibodies directed against type I1 collagen, the 148-kDa protein, the 100-kDa protein subunit and the 59-kDa protein. The distribution of the proteins was then studied by immunofluorescence. On day 3 (Fig. 5, E-H), the mesenchymal cells had started to form nodules. Type I1 collagen was the only protein of the four studied which could be identified in sections of the day-3 mesenchyma1 cell cultures (Fig. 5E). None of the other proteins, i.e., the 148-kDa protein (Fig. 5F), the 100-kDa-subunit protein (Fig. 5G) and the 59-kDa protein (Fig. 5H) could be detected at this stage. The nodules in the day-6 mesenchymal cell cultures were composed of hypertrophied chondrocytes surrounded by a sparse extracellular cartilage matrix (Fig. 6E-H). Type I1 collagen (Fig. 6E) and the 148-kDa protein (Fig. 6F) were immunochemically detected throughout the cartilaginous tissue within the nodule. Type I1 collagen was somewhat more pronounced in the peripheral tissue than in the 148-kDa protein. Since type I1 collagen and the 148-kDa protein were immunochemically monitored in the same section, it was possible to study the accumulation of the two proteins around individual chondrocytes within the nodule.
These results indicated that sometimes the two proteins did not accumulate to the same extent around a single chondrocyte. Since immunohistochemistry only detects antigen accumulation, the relative roles of synthesis and stability are not known. A few chondrocytes, mostly in the periphery of the cartilaginous nodule, were also associated with the 100-kDa-subunit protein (Fig. 6G), whereas this protein was not identified in the central portion of the nodule. This differential localization closely resembled that in the developing humerus. Moreover, the 59-kDa protein could not be immunochemically detected in the day-6 culture (Fig. 6H). In the day-I 4 cultures, therc were multiple cartilaginous nodules in close contact with each other (Fig. 7A-D). At this time calcium deposits could be detected in association with many of the cells around nodules as well as with the hypertrophic chondrocytes within nodules (Fig. 9). In addition, all of the four cartilage matrix proteins being studied were detected in the day-14 cultures (Fig. 7E-H). The 148-kDa protein (Fig. 7F) and the 100-kD-subunit protein (Fig. 7G) were present throughout the cartilage matrix, whereas the type I1 collagen (Fig. 7E) molecules were more pronounced among the peripheral cartilaginous cells. The 59-kDa protein was not immunochemically localized within the cartilaginous nodules, but instead surrounded each nodule (Fig. 7H). On day 55, the nodules had increased in size compared to that on day 14. However, the cartilaginous portion of the nodule did not seem to have increased as much. Thus, the growth occurred mainly in tissues surrounding each nodule (Fig. 8A-D). Except for the increased size of the nodules, the distribution of the four proteins (Fig. 8 E-H) was quite similar to that on day 14. However, on day 55 the 59-kDa protein was also localized near some chondrocytes in the cartilaginous portion of the nodule (Fig. 8 H). Discussion
In the present study, evidence is presented supporting the sequential appearance of four cartilage matrix proteins
-
209
type I1 collagen, 148-kDa protein [I 51, 100-kDa-subunit protein (Heinegard, to be published) and 59-kDa protein [7], in developing mouse limbs and in cultures of limb bud mesenchymal cells. The four proteins studied appeared in the following order: type I1 collagen, 148-kDa protein, 100-kDa-subunit protein, and 59-kDa protein. There are several possible mechanisms for the sequential appearance of these proteins. The main regulatory events are likely to occur at the transcriptional level and also the translational level. It has been shown that the biosynthesis of type I1 collagen is regulated at both the transcriptional and translational levels in initial chondrogenesis of the chicken wing. The mRNA for type I1 collagen can be detected about 12 h before the protein is immunochemically detected in the extracellular matrix [12]. Furthermore, there is also sequential expression of the genes for type I1 collagen and proteoglycans during chondrocyte differentiation in the developing chicken wing and in cultures of chicken limb bud cells [Ill. Another mechanism for the sequential appearance of the four cartilage matrix proteins could be that they are synthesized by different populations of chondrocytes, each differentiating at various times during chondrogenesis. This appears unlikely for type I1 collagen and the 148-kDa protein, since these appear to codistribute at some stages of development. Notably, the 148-kDa protein showed limited expression by peripheral chondrocytes. The 100-kDa-subunit protein at times showed preferential localization to peripheral cartilage, subsequently forming the articular surface. The sequential appearance of these proteins indicates that cartilage matrix is assembled in an ordered fashion. There was also a striking similarity between these events occurring in vivo and those occurring in vitro with regard to the sequential appearance and localization of the four proteins studied. This included the progression of chondrogenesis to the point of hypertrophy and finally the formation of mineralized cartilage. This was the case even though the mesenchymal cells were plated out in a randomized fashion after cell dissociation of the whole limb bud. The distribution of the proteins followed that of the developing limb in situ. Each cartilaginous nodule could be regarded as a single skeletal element, with type I1 collagen, 148-kDa protein, and the 100-kDa-subunit protein localized within the cartilaginous structure and the 59-kDa protein surrounding the nodule. The tendency for differential staining of central and peripheral portions, however, appears to have been lost in the cultures. Besides the sequential appearance of these proteins, this study yielded another important finding. Figure 4 C shows that the 148-kDa protein could not be detected irnmunohistochemically in the edge of the proximal epiphysis in the day-18 mouse humerus. The protein was present in the same region of the day-I4l/, mouse humerus (Fig. 3C) as well as in the distal region of day-IS mouse humerus. It is therefore likely that the biosynthesis or accumulation of the 148-kDa protein became inhibited in the edge of the cartilaginous epiphysis, which would later form the articular cartilage. The fact that the 148-kDa protein was absent from the presumptive articular cartilage region fits very well with the observation that the 148-kDa protein is not present in extracts of articular cartilage [16]. The major morphological difference between articular cartilage and other types of cartilage is that the former tissue is not surrounded by a perichondrium and has ceased growth.
A cDNA clone for the 148-kDa protein of chick cartilage has recently been isolated and sequenced [I]. The data show that the 148-kDa protein contains a sequence similar to that of epidermal growth factor. It has been suggested that upon proteolytic cleavage the 148-kDa protein can function as a local growth factor during active chondrogenesis. This protein can be detected immunohistochemically in cartilage that is formed in mesenchymal cell cultures, at all stages studied so far. Apparently articular cartilage does not form in these cultures, since there is no cartilage structure lacking the 148-kDa protein. It is possible that the biosynthesis of the 148-kDa protein is suppressed in the outermost part of the cartilage in the day-18 mouse humerus by the production of a specific substance by some cells in the developing joint. This hypothesis warrants testing. Acknowledgments. The authors thank Dr. D.F. Paulsen for his valuable comments on the manuscript, Virginia Dress and Dwight Moulton for technical assistance, and Karen Kriege for typing the manuscript. This work was supported by NIH grants DE05837 and HD05505.
References 1. Argraves WS, Deak F, Sparks KT, Kiss I, Goetinck P F (1987) Structural features of cartilage matrix protein deduced from cDNA. Proc Natl Acad Sci USA 84:464-468 2. Baker TR, Caterson B (2979) The link proteins as specific components of cartilage proteoglycan aggregates in vivo. J Biol Chem 254 :2394-2399 3. Fell HB (1925) The histogenesis of cartilage and bone in the long bones of the embryonic fowl. J Morphol40:417-451 4. Fife R, Brandt K (1984) Identification of a high-molecularweight ( >400,000) protein in hyaline cartilage. Biochim Biophys Acta 802:506-514 5. Heinegard D, Paulsson M (1987) Cartilage. In: Cunningham LW (ed) Methods in Enzymology, vol 145. Academic Press, Inc, Orlando, pp 336-362 6. Heinegard D, Bjorne-Persson A, Coster L, Franzen A, Gardell S, Malmstrom A, Paulsson M, Sandfalk R, Vogel K (1985) The core proteins of large and small interstitial proteoglycans from various connective tissues form distinct subgroups. Biochem J 230:181-194 7. Heinegard D, Larsson T, Sommarin Y, Franzen A, Paulsson M, Hedbom E (1986) Two novel matrix proteins isolated from articular cartilage show wide distributions among connective tissues. J Biol Chem 261 : 1386613872 8. Hewitt AT, Varner HH, Silver MH, Dessau W, Wilkes CM, Martin GR (1982) The isolation and partial characterization of chondronectin, an attachment factor for chondrocytes. J Biol Chem 257: 233G2334 9. Humason GL (1972) Animal tissue techniques. 3rd ed. WH Freeman and Company, San Francisco 10. Johnson GD, Araujo NC (1981) A simple method of reducing the fading of immunofluorescence during microscopy. J Immunol Methods 43 :349-350 11. Kosher RA, Gay SW, Kamanitz TJ, Kulyk WM, Rodgers BT, Sai S, Tdmaka T, Tanzer ML (1986) Cartilage proteoglycan core protein gene expression during limb cartilage differentiation. Dev Biol 118:112-117 12. Kravis D, Upholt WB (1985) Quantitation of type I1 procollagen mRNA levels during chick limb cartilage differentiation. Dev Biol 108:164-172 13. Linsenmdyer TF, Hendrix MJC (1980) Monoclonal antibodies to connective tissue macromolecules: type I1 collagen. Biochem Biophys Res Comm 92:440-446 14. Owens EM, Solursh M (1981) In vitro histogenic capacities
210 of limb mesenchyme from various stage mouse embryos. Dev Biol 88 :297-31 1 15. Paulsson M, Heinegard D (1981) Purification and structural characterization of a cartilage matrix protein. Biochem J 1971367-375 16. Paulsson M, Heinegard D (1982) Radioimmunoassay of the 148-kilodalton cartilage protein. Distribution of the protein among bovine tissues. Biochem J 207:207-213 17. Schmid TM, Conrad HE (1982) Metabolism of low molecular weight collagen by chondrocytes obtained from histologically distinct zones of the chick embryo tibiotarsus. J Biol Chem 257:12451-12457 18. Solursh M (1986) Environmental regulation of limb chondrogenesis. In: Kuettner K et al. (eds) Articular cartilage biochemistry. Raven Press, New York, pp 145-161
19. Vaughan L, Winterhalter HM, Bruckner P (1985) Proteoglycan Lt from chicken embryo sternum identified as type IX collagen. J Biol Chem 260: 47584763 20. von der Mark H, von der Mark K, Gay S (1976) Study of differential collagen synthesis during development of the chick embryo by immunofluorescence. Dev Biol48: 237-249 21. Wurster NB, Lust G (1982) Fibronectin in osteoarthritic canine articular cartilage. Biochem Biophys Res Comm 109:1094-1101
Received October 1987 / Accepted in revised form November 19, 1987