ABSTRACT
Mice carrying a targeted mutation in the gene for the myogenic factor Myf-5 fail to form major parts of the ribs, which leads to an unstable thorax and perinatal death. Here, we report that somites of Myf-5-deficient mice lack the expression of FGF-4 and FGF-6 while TGFβ-2 is expressed normally. Early sclerotomal markers, such as Pax-1 reveal no substantial reduction of sclerotome size. At E11.5 the condensing mesenchyme of the rib anlagen is considerably reduced in size in Myf-5 mutant mice. This may be caused by the lack of Myf-5-positive, FGF-expressing cells which normally are in close contact with the lateral sclerotome generating the rib progenitors. The potential role of FGFs and TGFβ on sclerotome formation is demonstrated in micromass cultures of early somites. Combinations of FGF-4 or FGF-6 with TGFβ-2 potentiate chondrogenesis suggesting that these growth factors emanating from early myotomal and dermomyotomal cells may have instructive or permissive effects on differentiation or outgrowth of sclerotomal cells.
INTRODUCTION
Somites, which are progressively generated as condensations of the paraxial mesoderm on both sides of the neural tube along the rostrocaudal axis give rise to several mesodermal cell types. Cells in the ventromedial segment of the somite form the sclerotome which contains the precursor cells for ribs, vertebrae and intervertebral discs (Keynes and Stern, 1988). In the dorsal aspect of the somite the dermomyotome forms as a columnar epithelium from which a myotomal layer separates when somites mature (Christ et al., 1978). Development of the myotomal cell layer is characterized by the expression of a family of myogenic factors which precedes the expression of sarcomeric structural proteins (Arnold and Braun; 1993, Olson and Klein, 1994). We have shown previously that myotome formation in Myf-5 mutant mice is delayed and initiates only when MyoD expression starts at E10.5, followed by the expression of myogenin, Myf-6 and other muscle-specific proteins (Braun et al., 1994; Braun and Arnold, 1994). The most striking phenotype of Myf-5 mutant mice, however, is the lack of the distal parts of the ribs which leads to secondary malformations like an unstable thorax and to perinatal death (Braun et al., 1992, Braun and Arnold, 1995). The occurrence of rib defects in Myf-5 mutant mice was particularly surprising, since Myf-5 expression is confined to the skeletal muscle cell lineage and is not observed in cells of the sclerotome or the skeleton.
Grafting and microsurgical experiments have demonstrated that the fate of somite cells is determined by a number of local signalling events (Christ et al., 1992). Removal of the notochord prior to somitic condensation prevents sclerotome formation, whereas ablation of the notochord at a later stage, when somites have already condensed, does not disrupt this process (Pourquie et al., 1993). These findings have led to the concept that notochord acts either directly or indirectly via induction of floor plate in the neural tube to promote sclerotome formation. Recently, it has been shown that sonic hedgehog alters dorsoventral specification of somitic derivatives by induction of proliferation and patterning of the induced tissue (Johnson et al., 1994). However, it is not yet clear whether other signalling pathways are also involved in somite specification (Fan and Tessier-Lavigne, 1994). In particular, signals from the surface ectoderm or the dermomyotome have not been defined. Likewise, the nature of molecules which may mediate signalling events within the somite are largely unknown.
Potential candidates expressed within the somite during its compartmentalisation include transforming growth factors (TGFs) and fibroblastic growth factors (FGFs; Dickson et al., 1993; deLapeyriere et al., 1993). The family of FGFs comprises 9 members which interact with several different transmembrane tyrosine kinase receptors (FGFRs) exerting different affinities and specificities (Goldfarb, 1990; Givol and Yayon, 1992). During development FGFs and FGFRs are widely expressed and have been implicated in a variety of developmental processes. Expression of FGF-4, FGF-5 and FGF-6 was found in somites and in several mesenchymal derivatives including skeletal muscle (Goldfarb, 1990; deLapeyriere et al., 1993). The role of FGFs in the control of cell proliferation and differentiation has been studied extensively. Gene inactivation experiments in mice demonstrated pivotal roles of FGF-4 and FGF-5 in the control of proliferation of the inner cell mass and the hair follicle, respectively (Hebert et al., 1994; Feldman et al., 1995). In addition, it has been proposed that FGF-4 maintains proliferation of cells in the progress zone of the limb which underlies the apical ectodermal ridge and controls the correct number and order of skeletal elements in limbs (Niswander et al., 1993).
Recently, cell signalling mediated by FGFs has been suggested to play an important role in coordinating regular growth of bones (Erlebacher et al., 1995). Mutations in the human FGFR3 receptor seem to result in achondroplasia, a dominantly inherited human skeletal disorder (Rousseau et al., 1994; Shiang et al., 1994). In a recent report, Rosseau et al. (1995) have described stop codon FGFR3 mutations in thanatophoric dwarfism type 1 which is characterized by a cloverleaf scull, bowed long bones, platyspondyly and, most interestingly, short ribs. Likewise, skeletal defects found in Crouzon Syndrome and Jackson-Weiss Syndrome have been correlated to mutations in the FGFR-2 receptor (Reardon et al., 1994; Muenke et al., 1994). Finally, patients with Pfeiffer Syndrome, characterized by synostosis and limb defects, carry mutations in the FGFR-1 receptor (Jabs et al., 1994).
Despite these remarkable effects of FGF or FGF receptor mutations, no functional role has yet been assigned to FGF-expression in somites, particularly in the myotomal layer. Here, we analyzed the effect of the targeted Myf-5 gene inactivation on sclerotome development and the expression of fibroblastic growth factors in somites. We show that somitic expression of FGF-4 and FGF-6 is absent in mutants at E9.5 and E10.5, while expression of TGFβ-2 is not affected.
We find no evidence that the size of the sclerotome or the expression of sclerotomal markers is severely impaired in mutant mice. The first manifestation of the rib defect is detected in the condensing mesenchyme which later forms the cartilage of the ribs. In wild-type mice this condensing mesenchyme is in close apposition with Myf-5-positive cells, while MyoD-expressing cells are located more dorsomedially. We also demonstrate that FGF-4 or FGF-6 together with TGFβ synergistically potentiate chondrogenic nodule formation in cultures of somitic cells. We propose that Myf-5-expressing myotomal cells support development of rib chondrocytes by the release of fibroblastic growth factors.
MATERIALS AND METHODS
Expression of FGF-4 and FGF-6 in Pichia pastoris
To obtain large amounts of biologically active FGF-4 and FGF-6 proteins the methylotrophic yeast strain Pichia pastoris was used. cDNAs of the coding regions for FGF-4 and FGF-6 were inserted into pPIC9 (Invitrogen) which contains the promoter and transcriptional terminator of the native AOX1 gene. A fusion with the leader peptide of the α-factor of S. cereviseae was generated to direct secretion into the culture medium and facilitate purification. Expression plasmids were linearized with SalI and transformed into host strain GS115(his–). Transformants were screened for integration into the chromosomal HIS 4 gene and tested for expression in small scale cultures. Large scale production was performed in a 20 l fermenter as described by Clare et al. (1991). FGF-4 and FGF-6 were recovered from the medium supernatant, purified and analyzed for biological activity. Concentrations of growth factors were determined by analysis of samples from biologically active fractions on SDS polyacrylamide gels and staining with Comassie blue. Only preparations that induced DNA synthesis in quiescent 3T3 fibroblasts were used for cultivation of somitic cells.
Mice
Myf-5 mutant mice have been described previously (Braun et al., 1992; Braun et al., 1994). Analyses were performed with 129Sv/JMyf5m1 mice backcrossed 3 or 4 times to either C57BL6 or BALB/c mice. Staging of embryos was performed by counting the appearance of the vaginal plug as day 0.5 p.c, and by the number of somites. Material for genotyping of the embryos was obtained from the yolk sac. Isolation of DNA and Southern blot analysis was done as described previously (Braun et al., 1992).
In situ hybridization on sections and on whole-mount preparations
Embryos were dissected in PBS, fixed in 4% paraformaldehyde for 12 hours, and embedded in paraffin as described (Bober et al., 1991). Tissue sectioning, prehybridization and hybridization using 35S-UTP-labelled cRNA probes were done exactly as described by Bober et al. (1991). Some embryos were sectioned on a cryostat after fixation in 4% paraformaldehyde and dehydration in 0.5 sucrose in PBS overnight (Braun et al., 1994). The cRNA probes for Pax-1, Pax-3, myogenin, twist, FGF-4, FGF-6 were described by Deutsch et al. (1988); Bober et al. (1994); Hebrok et al. (1994); Niswander and Martin (1992), and deLapeyriere et al. (1993), respectively. Whole-mount hybridizations were performed with digoxigenin labelled cRNA probes for FGF-6 and Pax-1 as described by Wilkinson (1992). Embryos hybridized with Pax-1 were photographed in PBS. Afterwards embryos were embedded in a gelatine/albumen mixture and sectioned with a vibratome as described by Bober et al. (1994).
Immunohistochemistry and histological staining procedures
Immunocytochemistry was performed on cryostat sections fixed in methanol/acetone as described by Braun and Arnold (1994). Simultaneous detection of primary antibodies developed in mouse (anti-MyoD) or rabbit (anti-Myf-5) was performed with secondary antibodies coupled to FITC and rhodamine, respectively (obtained from Sigma Biotechnolgy).
The following primary antibodies were used: anti-Myf-5 (C-20) raised against amino acids 236-255 of human Myf-5 (supplied by Santa Cruz Biotechnology), mAb 5.8A against mouse MyoD (obtained from Medac) and mAb MF20 raised against sarcomeric MyHC (kindly provided by D. Fishman, New York).
Peanut agglutinin (PNA) binding sites were detected by incubating cryostat sections with biotinylated PNA (Vector labs) at a concentration of 10 ng/ml after fixation in 4% paraformaldehyde. Bound PNA was visualized using a vectastain elite kit.
Micromass cultures
For micromass cultures, E10.5 mouse embryos were dissected in cold Hepes-buffered saline (HBS) and treated with dispase. Isolated somites were digested in 0.1% collagenase, 0.1% trypsin for 20 minutes at 37°C with occassional shaking. Cells were plated in DMEM, 2% FCS in 20 μl drops at 2×107 cells/ml in 6-well primaria dishes (Falcon). Staining of micromass cultures for chondrogenic nodules was achieved by fixation in 4% buffered formalin and incubation overnight in 1 mg/ml Alcian Blue in 0.1 N HCl at 37°C. Bound Alcian Blue was extracted with 0.5 ml guanidine HCl and quantified by reading the absorbtion at 550 nm.
RESULTS
Lack of FGF-4 and FGF-6 expression in somites of E9.5 and E10.5 Myf-5-deficient mice
Expression studies using anti-Myf-5 antibodies and cRNA probes as well as lineage tracing with a lacZ reporter gene integrated into the Myf-5 locus have demonstrated that the Myf-5 gene is not active in sclerotomal cells or their progenitors (Tajbakhsh and Buckingham, 1994; Braun et al., unpublished data). The rib defect in Myf-5-deficient mice is therefore unlikely to be caused by a cell autonomous mechanism but rather by lack of signals emanating from myotomal cells and acting upon parts of the sclerotome. We therefore decided to investigate potential signalling mechanisms which would be absent in Myf-5 mutant embryos. Growth factor genes that are expressed during somite development, such as FGFs, appeared to be good candidates to mediate these critical interactions.
In situ hybridization of sections from wild-type and Myf-5 mutant embryos revealed that FGF-4 and FGF-6 transcripts were absent in somites of mutant embryos at E9.5 (23 somites; Fig. 1C,G). At this stage FGF-4 and FGF-6 mRNAs were detected at low levels in thoracic somites of wild-type embryos (Fig. 1A,E). Control hybridizations with myogenin on adjacent sections demonstrated high level expression in wild-type embryos and no expression in the mutants (data not shown). In contrast, expression of FGF-4 in limb buds was not affected. Pax-3 which serves as marker for the dermomyotome was unchanged in somites of both mutant and wild-type littermates (Fig. 1I,K).
Comparison of FGF-4, FGF-6, and Pax-3 expression in genotyped wild-type (A,E,I) and Myf-5 mutant (C,G,K) embryos at E9.5 (23 somites). Adjacent transverse sections of somites of the thoracic region were hybridized with FGF-4 (A,C), FGF-6 (E,G) and Pax-3 (I,K) probes. FGF-4 (A) and FGF-6 (E) is detected in myotomes of wild-type embryos. Homozygous mutant embryos do not express FGF-4 (C) and FGF-6 (G) mRNA, while Pax-3 mRNA is observed in the dermomyotome at normal levels (K). Corresponding bright-field illuminations are shown to the right of each dark-field panel. Magnifications are 320×.
Comparison of FGF-4, FGF-6, and Pax-3 expression in genotyped wild-type (A,E,I) and Myf-5 mutant (C,G,K) embryos at E9.5 (23 somites). Adjacent transverse sections of somites of the thoracic region were hybridized with FGF-4 (A,C), FGF-6 (E,G) and Pax-3 (I,K) probes. FGF-4 (A) and FGF-6 (E) is detected in myotomes of wild-type embryos. Homozygous mutant embryos do not express FGF-4 (C) and FGF-6 (G) mRNA, while Pax-3 mRNA is observed in the dermomyotome at normal levels (K). Corresponding bright-field illuminations are shown to the right of each dark-field panel. Magnifications are 320×.
At the 30 somite stage (E10.5) expression of FGF-6 increased considerably in wild-type embryos and was now detectable by whole-mount in situ hybridization. While FGF-6 mRNA was found in somites of E10.5 embryos along its rostrocaudal axis, reaching relatively high expression levels in thoracic somites, we were unable to detect FGF-6 in mutants (data not shown).
Whole-mount in situ hybridization data was confirmed by in situ hybridization of cryosections of mutant embryos (data not shown). Analysis of MyoD expression in this series of experiments revealed that somites lacking FGF-6 were also devoid of MyoD mRNA, whereas Pax-3 was present in wild-type and mutant mice at comparable levels. In mutant somites that already expressed MyoD, FGF-6 transcripts were detected indicating that MyoD can rescue FGF-6 expression like that of other myotomal markers in the absence of Myf-5 (Braun et al., 1994).
Members of the TGFβ superfamily have also been suggested to be involved in development of cartilaginous structures (for a review see Erlebacher et al., 1995). TGFβ-2, a member of this family is expressed during somite development. We analyzed the expression of TGFβ-2 by in situ hybridization on sections of wild-type and mutant embryos at E9.5 and E10.5. In contrast to the lack of FGF expression in somites of mutant mice, we found no alterations in TGFβ-2 expression (Fig. 2A,B). TGFβ-2 transcripts were located in the dorsal aspects of somites inside the area of the dermomyotome (Fig. 2). Within the limits of resolution obtainable with radioactive in situ hybridization, TGFβ-2 expression patterns were identical in wild-type and mutant embryos. We suggest that TGFβ-2 is expressed in the dermomyotomal cell layer apparently unaffected by the Myf-5 mutation (see also the normal expression of Pax-3 in Fig. 1).
Myf-5 mutant mice express normal levels of TGF-β mRNA during early somite development. In situ hybridization with TGFβ-2 antisense and FGF-6 antisense probes to parallel sections of thoracic somites. Similar levels of TGFβ-2 mRNA are detected in wild-type (A) and Myf-5 mutant embryos (B) at E10.5 (34 somites). In contrast, FGF-6 mRNA is discovered in wild-type (E) but not in mutant embryos (F). TGFβ-2 mRNA appears in the dermomyotomal layer of the somite while FGF-6 is located in the myotomal layer. (A,B,E,F) Dark-field illumination; (C,D) Phase contrast picture. Magnifications are 500×.
Myf-5 mutant mice express normal levels of TGF-β mRNA during early somite development. In situ hybridization with TGFβ-2 antisense and FGF-6 antisense probes to parallel sections of thoracic somites. Similar levels of TGFβ-2 mRNA are detected in wild-type (A) and Myf-5 mutant embryos (B) at E10.5 (34 somites). In contrast, FGF-6 mRNA is discovered in wild-type (E) but not in mutant embryos (F). TGFβ-2 mRNA appears in the dermomyotomal layer of the somite while FGF-6 is located in the myotomal layer. (A,B,E,F) Dark-field illumination; (C,D) Phase contrast picture. Magnifications are 500×.
Expression of sclerotomal markers appears normal in Myf-5-deficient mice
We next asked whether the lack of FGFs in myotomal cells may have any direct effect on the size or patterning of the sclerotome as may be supposed by the immediate vicinity of these two compartments. Useful markers to identify sclerotomal cells include Pax-1, HoxA7, and twist, a basic helix loop helix protein expressed in mesodermal cells with the notable exception of dermomyotomal cells. In addition, expression of several other markers, such as Cart-1, a homeobox gene product expressed in developing cartilage, snail, a zinc-finger protein, or Mox-2, the product of another homeobox containing gene, were studied. However, these markers were either expressed too late (Cart-1) or expression was too low at early stages (snail, Mox-2) to be able to derive reliable data. Therefore, we concentrated on the analysis of Pax-1, HoxA7 and twist.
Whole-mount preparations of wild-type (Fig. 3A) and mutant (Fig. 3B) E9.5 embryos (24 somites) were hybridized with Pax-1 probe. Both wild-type and mutant embryos expressed high levels of Pax-1 mRNA in somites at the fore limb bud and at the interlimb level. Hybridized embryos were cut with a vibratome in transverse (Fig. 3C,D) and frontal (Fig. 3E,F) planes to reveal the distribution of Pax-1-expressing cells in wild-type and mutant somites.
Pax-1 expression in E9.5 wild-type and Myf-5 mutant embryos (24 somites). Whole-mount preparations of genotyped wild-type (A) and mutant (B) embryos hybridized with a Pax-1 probe. Transverse (C,D) and frontal (E,F) sections of wild-type (C,E) and Myf-5 mutant (D,F) embryos from the future interlimb region. Expression of Pax-1 is not reduced in mutant embryos. Note that Pax-1 expression reaches the dermomyotomal epithelium in mutant mice. Sections were photographed with DIC optics at a magnification of 400×.
Pax-1 expression in E9.5 wild-type and Myf-5 mutant embryos (24 somites). Whole-mount preparations of genotyped wild-type (A) and mutant (B) embryos hybridized with a Pax-1 probe. Transverse (C,D) and frontal (E,F) sections of wild-type (C,E) and Myf-5 mutant (D,F) embryos from the future interlimb region. Expression of Pax-1 is not reduced in mutant embryos. Note that Pax-1 expression reaches the dermomyotomal epithelium in mutant mice. Sections were photographed with DIC optics at a magnification of 400×.
No major difference in the distribution of sclerotomal cells marked by Pax-1 were detected in wild-type (Fig. 3C,E) and mutant (Fig. 3D,F) embryos of this early stage. It should be stressed that in mutants Pax-1-expressing cells were detected abutting the dermomyotomal layer without any evidence for reduction of the Pax-1-expressing region.
In E10.5 embryos (32 somites) the position of the myotomal layer within wild-type somites was first marked by hybridization of sections with a probe for myogenin. We obtained strong hybridizations in wild-type and but no expression in Myf-5 mutant embryos. Sclerotomal cells were then identified on adjacent sections of the same embryos by hybridization with Pax-1 which is uniformly expressed throughout the sclerotome at this developmental stage in both wild-type and mutant embryos. Clearly, there was no apparent reduction of Pax-1-positive cells but rather a moderate extension of this region towards the dermomyotome in mutants (data not shown). In addition, comparative in situ hybridizations of wild-type and mutant embryos at E10.5 were performed using a twist probe. In both embryos the mesenchyme hybridized intensively including the sclerotome but excluding the dermomyotome. Neither the expression level of twist in the sclerotome nor the area covered by twist-positive cells appeared to be altered in Myf-5 deficient embryos at this stage. In particular the twistnegative area within somites representing the dermomyotome was not enlarged in mutant mice (data not shown). Further more, crosses of Myf-5 mutant mice with the transgenic mouse strain L53 which expresses the lacZ reporter from a HoxA7 promoter construct in sclerotomal cells revealed no differences between wild-type and mutant mice (data not shown).
In conclusion, our analysis of sclerotomal development in Myf-5 mutant mice did not reveal any significant reduction of the sclerotome size. On the contrary, the lack of myotomal cells seemed to permit a moderate extension of the sclerotome towards the dermomyotomal layer.
A defect in the condensing mesenchyme which forms the rib anlagen is detected by staining with peanut agglutinin at E11.5
To study when the first phenotypic abnormality of rib mesenchyme formation was detectable, we stained transversal sections of wild-type and mutant embryos with peanut agglutinin and Alcian Blue. Peanut-agglutinin binds specifically to Galβ1-3Gal-Nac (Lotan et al., 1975) which is preferentially expressed in the posterior sclerotome and in developing cartilaginous structures. As shown in Fig. 4E, a pronounced reduction in the size of the developing rib mesenchyme was observed in Myf-5 mutant embryos. In wild-type animals the rib blastema projected far more ventrally (Fig. 4D), whereas in mutant embryos only the area close to the future vertebral column was stained.
Myf-5 but no MyoD-expressing cells are located close to the distal parts of the developing rib blastema. (A) Survey picture of a section from the thoracic region of an E11.5 wild-type embryo stained with peanut agglutinin to reveal condensing rib mesenchyme. The black frame in A indicates the area shown in B and C at high magnification after double immunofluorescence staining. Adjacent sections to A were double-stained with a polyclonal antiserum against Myf-5 (B) and a monoclonal antibody against MyoD (C) using immunofluorescence. At this stage only cells expressing Myf-5 are found close to the distal end of the rib blastema.Transverse sections of E11.5 wild-type (D) and Myf-5 mutant embryos (E) were stained with peanut agglutinin to mark developing cartilage. Arrows indicate the borders of cartilaginous condensations. Note the ventrolateral reduction of the rib blastema in the mutant mice (E). The left side of each picture corresponds to the dorsal side of the embryo where the rib blastema contacts the region forming vertebrae. NT, neural tube; CM, condensing mesenchyme; V, vertebrae anlagen; A, aorta; LB, lung buds; L, liver. Magnifications were 25× in A and 500× in B-E.
Myf-5 but no MyoD-expressing cells are located close to the distal parts of the developing rib blastema. (A) Survey picture of a section from the thoracic region of an E11.5 wild-type embryo stained with peanut agglutinin to reveal condensing rib mesenchyme. The black frame in A indicates the area shown in B and C at high magnification after double immunofluorescence staining. Adjacent sections to A were double-stained with a polyclonal antiserum against Myf-5 (B) and a monoclonal antibody against MyoD (C) using immunofluorescence. At this stage only cells expressing Myf-5 are found close to the distal end of the rib blastema.Transverse sections of E11.5 wild-type (D) and Myf-5 mutant embryos (E) were stained with peanut agglutinin to mark developing cartilage. Arrows indicate the borders of cartilaginous condensations. Note the ventrolateral reduction of the rib blastema in the mutant mice (E). The left side of each picture corresponds to the dorsal side of the embryo where the rib blastema contacts the region forming vertebrae. NT, neural tube; CM, condensing mesenchyme; V, vertebrae anlagen; A, aorta; LB, lung buds; L, liver. Magnifications were 25× in A and 500× in B-E.
We also wanted to know where Myf-5-positive cells would be located in relation to the condensing mesenchyme forming the rib blastema. Since at this developmental stage (E11.5) both Myf-5 and MyoD-positive cells are present in mouse embryos, we performed double stainings with anti-Myf-5 and anti-MyoD antibodies on the same section. The low amount of Myf-5-positive cells and a relatively weak staining intensity of MyoD and Myf-5 antibodies precluded the detection of Myf-5 and MyoD-positive cells at low magnifications. At high magnification however, Myf-5-expressing cells contacting the condensing mesenchyme were readily detactable in wild-type embryos, whereas MyoD-expressing cells were not found in this area (Fig. 4C). A lower power adjacent section stained with peanut agglutinin (Fig. 4A) shows the condensing rib mesenchyme (black frame in Fig. 4A). MyoD-positive cells are found more dorsally at this stage (data not shown). Only later during development do MyoD-positive cells also populate the region close to the ribs.
A combination of TGFβ-2 and FGF-4 or FGF-6 synergistically stimulates chondrogenic nodule formation in somitic micromass cultures
In order to demonstrate the capability of growth factors expressed in the dermomyotome to direct differentiation of adjacent sclerotomal cells, we isolated somitic cells from E10.5 somites and treated them in micromass cultures with different combinations of growth factors. Addition of FGF-4 or FGF-6 alone had only a small effect on the formation of chondrogenic nodules (Fig 5C,D). Likewise, treatment of somitic cultures with TGFβ-2 or TGFβ1 at different concentrations had only small effects on cartilage differentiation (Fig. 5B).
Synergistic stimulation of chondrogenic differentiation in micromass cultures of somitic cells by TGFβ-2 and FGF-4/FGF-6. Alcian Blue-positive chondrogenic condensations in control cultures (A) or in the presence of TGFβ-2 (B), FGF-4 (C), FGF-6 (D) or combinations of TGFβ-2 and FGF-4 (E), or TGFβ-2 and FGF-6 (F) are shown. Note the drastic increase of chondrocyte differentiation in cultures treated with combinations of TGFβ-2 and FGF-4/FGF-6. All cultures contained 2% FCS. Growth factor concentrations were 10 ng/ml for TGFβ-2, FGF-4, and FGF-6. Pictures were taken under a stereo microscope.
Synergistic stimulation of chondrogenic differentiation in micromass cultures of somitic cells by TGFβ-2 and FGF-4/FGF-6. Alcian Blue-positive chondrogenic condensations in control cultures (A) or in the presence of TGFβ-2 (B), FGF-4 (C), FGF-6 (D) or combinations of TGFβ-2 and FGF-4 (E), or TGFβ-2 and FGF-6 (F) are shown. Note the drastic increase of chondrocyte differentiation in cultures treated with combinations of TGFβ-2 and FGF-4/FGF-6. All cultures contained 2% FCS. Growth factor concentrations were 10 ng/ml for TGFβ-2, FGF-4, and FGF-6. Pictures were taken under a stereo microscope.
Because various growth factors are present in the developing somite, it appeared possible that a combination of growth factors may mediate control of somitic chondrogenesis. To test this possibility, we added FGF-4 or FGF-6 in combination with TGFβ-2 to high density cultures of E10.5 somitic mesenchyme. In contrast to the minute effects observed with individual growth factors, the combined addition of TGFβ-2 and FGF-4 or FGF-6 markedly potentiated chondrogenesis (Fig. 5E,F). Stimulation of chondrogenic nodule formation required concentrations of FGF-4 and FGF-6 above 1 ng/ml and TGFβ-2 between 1 and 10 ng/ml. No further enhancement of chondrogenesis was observed with growth factor concentrations of 10 ng/ml or more. A specific event of chondrogenesis is the accumulation of S-GAGs in chondrogenic nodules. Therefore, we examined the production of S-GAGs in response to FGF-4, FGF-6 and TGFβ-2 or combinations of those and TGFβ-2 by measuring the accumulation of Alcian-blue-positive matrix. As shown in Fig. 6, the addition of the above combinations of growth factors resulted in a synergistic increase of S-GAG deposition in micromass cultures. Although the very high concentration of cells in micromass cultures makes an increase in cell proliferation in response to growth factors unlikely, we determined cell density by measuring the DNA content in treated and untreated cultures. The obtained values revealed similar DNA concentrations under both culture conditions (data not shown).
Quantification of chondrogenesis after addition of growth factors to micromass cultures of somitic cells from E10.5 mouse embryos. Alcian Blue, specifically bound to S-GAGs was eluted from cultures treated with FGF-4 and/or TGFβ-2 (A) and FGF-6 and/or TGFβ-2 (B) and used as marker for chondrocyte formation. Cultures at different time points were fixed and eluted. Alcian Blue was measured at 550 nm. Symbols indicate the various growth factors. The growth factors were added at 10 ng/ml. Each point represents the mean of 3 independent cultures. All cultures contained 2% FCS.
Quantification of chondrogenesis after addition of growth factors to micromass cultures of somitic cells from E10.5 mouse embryos. Alcian Blue, specifically bound to S-GAGs was eluted from cultures treated with FGF-4 and/or TGFβ-2 (A) and FGF-6 and/or TGFβ-2 (B) and used as marker for chondrocyte formation. Cultures at different time points were fixed and eluted. Alcian Blue was measured at 550 nm. Symbols indicate the various growth factors. The growth factors were added at 10 ng/ml. Each point represents the mean of 3 independent cultures. All cultures contained 2% FCS.
DISCUSSION
The unexpected lack of rib structures in Myf-5 mutant mice has prompted us to study alterations in somites of Myf-5-deficient embryos. Since expression studies and cell lineage analysis have not supplied evidence for a cell autonomous mechanisms which would explain the Myf-5 phenotype, we concentrated here on the analysis of possible interactions between myotomal and sclerotomal cells in mutant embryos. We analyzed the expression of potential mediators of tissue interactions in somites and found that molecules of the FGF family, i. e. FGF-4 and FGF-6, are not expressed in somites of Myf-5 mutant mice until MyoD expression starts. Another growth factor, TGFβ-2, was present at wild-type levels in mutant somites. Although the sclerotome of mutants was not significantly reduced, we found that the area of the condensing mesenchyme which emerges from the sclerotome and gives rise to rib cartilage was considerably smaller in mutant than in wild-type embryos. Furthermore, we were able to show that FGF-4 and FGF-6 together with TGFβ-2 exerted a synergistic effect on cartilage differentiation of somitic cells in micromass cultures in vitro.
The observation that neither TGFβ-2 nor FGF-4 or FGF-6 alone are sufficient to markedly promote chondrogenesis of somitic cells but a combination of growth factors has a synergistic effect on chondrogenic nodule formation indicates the importance of the complex network of factors for the control of morphogenetic events in somites.
Synergy of members of the FGF and the TGFβ families during embryonic development has been found in several instances. Outgrowth of chick limb buds is stimulated by FGF-4 and BMP-2 (Niswander et al., 1993) and mesoderm is induced in animal cap assays by bFGF and TGFβ (Kimelman and Kirschner, 1987). A recent example shows the synergistic induction of differentiation of the cartilaginous otic capsule by bFGF and TGFβ-1 (Frenz et al., 1994). In this case the otic epithelium, which expresses bFGF and TGFβ-1, serves as an inducer of cartilage growth and differentiation, whereas antibodies against these growth factors suppress chondrogenesis induced by the epithelium (Frenz et al., 1994). Alterations of the balance of growth factors secreted by somitic cells may have a detrimental effect on the local control of bone morphogenesis. The lack of FGF expression in Myf-5 mutant cells most likely causes such an imbalance. Although not all important signals appear to be disturbed in Myf-5 mutant mice, the absence of one critical pathway may be sufficient to abrogate correct pattern formation. The importance of paracrine factors in regulating mesenchyme condensation has been emphasized recently in the mouse mutants short ear (se) and brachypodism (bp). In these mice the genes for BMP5 and GDF-5, two members of the TGFβ superfamily, are mutated (Kingsley et al., 1992; Storm et al., 1994).
In contrast to the previously described effect of TGFβ alone on chick limb bud cells, we did not observe significant and consistent increase of cartilage formation in somite cultures (Kulyk et al., 1989). Based on conflicting results from several laboratories, cell typespecific TGFβ responses have been suggested (reviewed by Centrella et al., 1994).
At present, no valid model is available to explain the synergistic functions of TGFβ and FGFs on chondrocyte development. Previous studies suggested that TGFβ1 positively modulates bFGF activity by increasing its binding (Lefebvre et al., 1991). Alternatively, TGFβ may change the composition of the extracellular matrix and the competence of somitic cells to respond to it (Massague, 1987). Synergistic effects on growth of cultured bone cells mediated by bFGF and TGFβ have been reported (Globus et al., 1988). This however, does not apply to our culture system, since we do not see enhanced growth.
The relatively normal expression of Pax-1, twist, and HoxA7 in the sclerotome of Myf-5 mutant mice seems to suggest that it is not the formation of sclerotomal cells that is disturbed but their local induction and/or the patterning. It also appears possible that FGFs promote survival of rib blastema cells acting as maintenance factors. Since we have not seen excessive cell death in Myf-5 mutants, however, we consider this an unlikely mechanism. FGF-4/FGF-6 and TGFβ may therefore act on committed tissue which needs additional instructive signals to form vertebrae, neural arches, and ribs. The requirement of such signals is probably limited to a narrow time window, since FGF-6 expression in the later appearing MyoD-positive cells cannot rescue the rib defect. The location of MyoD-expressing cells in a more dorsal position probably prevents close contacts with the rib blastema, while in wild-type embryos Myf-5-expressing cells are conjugated to the rib anlagen. We do not know for how long these contacts may be required and whether already outgrowing rib mesenchyme may still depend on it. (For localization of cells expressing the lacZ marker driven by a MyoD promoter/enhancer construct in transgenic mice see Gold hammer et al., 1995).
It should be pointed out that the distal parts of the ribs are always most severely affected in Myf-5 mutant mice. At present, several possibilities can be proposed to try to explain this issue. (a) Proximal and distal regions of the ribs may be derived from distinct cell populations which respond differently to the same growth signals. (b) progenitor cells for the distal parts of the ribs may be anatomically closer to the myotome than progenitor cells for proximal parts, thereby receiving higher concentrations of myotomal growth signals. Therefore, these cells might be more affected by lack of signals. (c) If no discrete cell population for distal parts of the ribs exists or this population is plastic, the lack of myotomal growth factors may simply influence the number of progenitor cells and those that must divide most to form the distal parts of ribs are most affected. Although none of these possibilities can be excluded, the presence of discrete rib progenitor populations appears the least likely. The localization of precursor cells for distal parts of the ribs within the sclerotome is presently unclear, although a recent report (Huang et al., 1994) demonstrated that cells of the somitocoele in epithelial somites are rib progenitors. We favour the anatomical explanation, that distal rib progenitor cells receive growth signals from adjacent myotome and that the cells at the tip of the rib blastema are most dependent on myotomal growth factors.
The experiments presented in this report do not prove definitively but strongly suggest that the rib phenotype in Myf-5 mutant mice is caused by the lack of secreted FGFs in somites of mutant mice. The absence of FGFs correlates well to the time period during which sclerotomal cells are normally induced to form the rib blastema. The synergistic effect of FGF-4/FGF-6 and TGFβ-2 on cartilage differentiation of somitic cells furthermore demonstrates the capacity of these growth factors to induce cartilage development. Analysis of other mouse mutations, such as short ear and brachypodism, clearly demonstrated that alteration of paracrine controls can cause specific skeletal abnormalities. Finally, mutations in FGFR1, FGFR2 and FGFR3 lead to malformations of bones indicating that the FGF signalling pathway may be important to coordinate bone growth. Most intriguingly stop mutations in the FGFR3 gene in humans causes thanotophoric dwarfism, a phenotype which is characterized by the appearence of short ribs, beside other malformations (Rousseau et al., 1995).
To prove our hypothesis in vivo, however, will probably be difficult. Knock-out experiments to inactivate FGF-4 and FGF-6 are in progress but may suffer from possible functional redundancy among individual FGFs or from early defects which make the analysis of somite development impossible. The latter has already been shown in FGF-4 knock-out mice which exert an arrest of inner cell mass development at the blastocyst stage (Feldman et al., 1995). Complex conditional knock-out protocols would be necessary to overcome these difficulties. The expression of a dominant-negative FGF-receptor in the sclerotome of mice or the artificial expression of FGFs in somites by targeted expression may present attractive alternatives to the knock-out approach and may supply in vivo functional insights.
ACKNOWLEDGEMENTS
The excellent technical assistance of S. Heymann is gratefully acknowledged. We would like to thank G. R. Martin for FGF-4 and FGF-6 cDNAs, P. Gruss for Pax-1 and Pax-3 probes, E.-M. Füchtbauer for a twist probe, C. Wright for a Mox-2 probe, D. Wilkinson for a sna probe, B. deCrombrugghe for a Cart-1 probe and Eva Bober for critically reading the manuscript. We are grateful to T. Wagner and D. Voelkel for help with fermentation. This work was supported by the SFB 271: ‘Molekulare Mechanismen morphoregulatorischer Prozesse’ project B2 and by DFG grant Br 1413 to T. B.