Myf5 ^−/− :MyoD ^−/− amyogenic fetuses reveal the importance of early contraction and static loading by striated muscle in mouse skeletogenesis

Abstract

The mechanical loading of striated muscle is thought to play an important role in shaping bones and joints. Here, we examine skeletogenesis in late embryogenesis (embryonic day 18.5) in Myf5 ^−/− :MyoD ^−/− fetuses completely lacking striated muscle. The phenotype includes enlarged and fused cervical vertebrae and postural anomalies, some viscerocranial anomalies, long bone truncation and fusion, absent deltoid tuberosity of the humerus, scapular and clavicular hypoplasia, cleft palate, and cleft sternum. In contrast, neurocranial bone development was essentially normal. While the magnitude of individual effects varied throughout the skeletal system, the results are consistent with skeletal development depending on functional muscles. Novel abnormalities in the amyogenic fetuses relative to less severely paralyzed phenotypes extend our understanding of skeletogenic dependence on embryonic muscle contraction and static loading.

Introduction

Vertebrate striated muscles and bones belong to a functional system and cannot function independently from one another. The muscles provide the mechanical forces, while the bones provide the support. The interconnecting signals that exist between these tissues are poorly understood. However, they both share a response to mechanical stimuli (reviewed by Herring 1994).

Most of our knowledge about the dependence of bone development on muscles arises from experiments performed on embryos and fetuses of other vertebrates. However, experimental procedures that are possible in avian and amphibian embryos are not possible in mammals because of their intrauterine development. Fortunately, mutant mice that have altered embryogenesis allow us to examine various fundamental issues in the developmental biology of mammals. For instance, there is a complete absence of striated muscle formation in Myf5 ^−/− :MyoD ^−/− (or amyogenic) embryos (Rudnicki et al. 1993) and complete absence of secondary myogenesis (e.g., formation of functional muscle fibers) in myogenin null embryos (Brennan et al. 1996).

In mammals, myogenesis occurs as a two-phase event (Harris et al. 1989). In mice, the first phase begins prior to embryonic day (E) 12.5. The primary muscle fibers that form during this phase constitute approximately 20% of the muscle in the newborn. The second phase begins during E14.5 and gives rise to approximately 80% of neonatal muscle through the development of secondary muscle fibers.

The myogenic regulatory factors (MRFs), which are transcription factors including MyoD, myogenin, Myf5, and MRF4, function to regulate striated muscle development (Rudnicki and Jaenisch 1995; Kablar and Rudnicki 2000). The introduction of null mutations in Myf5, MyoD, myogenin, and MRF4 has revealed a hierarchical relationship among the MRFs and established that functional redundancy is a feature of the MRF regulatory network (Rudnicki and Jaenisch 1995).

Mice with a null mutation in myogenin fail to form secondary muscle fibers (Hasty et al. 1993; Nabeshima et al. 1993). Mutations in the other three genes result in essentially normal patterning and striated muscle volume (Braun et al. 1992; Rudnicki et al. 1992; Zhang et al. 1995). Interestingly, in double-mutant embryos and fetuses lacking both Myf5 and MyoD genes, the entire embryonic lineage that gives rise to striated muscle never forms, as evidenced by the absence of myoblasts and myofibers throughout development (Rudnicki et al. 1993; Kablar et al. 1999; Kablar et al. 2003).

Muscles are known to have widespread reciprocal effects on bones. The most usual interpretation of these effects is that they are primarily mechanical (reviewed by Herring 1994). The influence of mechanical effects on bone development was studied extensively by Hall and Herring (1990), using paralyzed or immobilized chicken embryos and, to a much lesser extent, mutant mouse embryos (Herring and Lakars 1982).

Most skeletal elements, except those originating from secondary cartilages, differentiate even in the complete absence of muscle activity (Herring 1994). However, in the absence of muscle activity, some skeletal elements undergo changes in morphogenesis, resulting in alterations in their shape and size. For example, when the innervation of auricular muscles in rat neonates is disturbed, the shape, size, and position of the external ear (i.e., cartilage and bone) are altered too (Chiu et al. 1979). One source of loading produced by an inactive but still present muscle is lost in paralyzed embryos and in muscular dysgenesis (Herring 1994), but the static loading produced by the muscle tissue itself is only eliminated in the double knockout embryos lacking both Myf5 and MyoD since they completely lack the striated muscles (Rudnicki et al. 1993; Kablar et al. 1999, 2003).

Here, we have investigated the dependence of embryonic bone development on myogenesis using a direct in vivo approach. To that end, Myf5 ^−/− :MyoD ^−/− fetuses have been used, as they represent a unique opportunity to investigate, for the first time, skeletal development in the complete absence of myogenesis. Amyogenic fetuses are particularly useful because of the absence of (1) a muscular vasculature, (2) electric currents from the muscle, (3) inputs to/from the CNS (Kablar and Rudnicki 1999, 2002; Kablar 2002), and (4) actual muscle weight (i.e., instead of muscle, Myf5 ^−/− :MyoD ^−/− fetuses contain very loose connective tissue, adipose tissue, or edema, which all have a significantly lower specific weight; Kablar et al. 2003). The static loading from the striated muscle is often sufficient for normal bone formation (Herring 1994).

We report a phenotype that includes postural anomalies, fused cervical vertebrae, anomalies of the mandible, cleft palate, disturbances of the shape and the size of some bones of the upper and lower extremities and long bone proximity and fusion, truncation of the femur and tibia, cleft sternum, and heteromorphic clavicle. Some of these are novel heteromorphisms not present in paralyzed embryos. The latter embryos have inactive muscle tissue in contact with bone, and spontaneous contraction may occur early in development. Thus, the observed phenotype suggests that early contractions and static loading from muscle tissue are important for skeletogenesis.

Materials and methods

Interbreeding and collection of fetuses

Fetuses lacking both Myf5 and MyoD were derived by a two-generation breeding scheme. First, MyoD ^−/− mice were bred with Myf5 ^+/− mice to generate Myf5 ^+/− :MyoD ^+/− mice. Second, Myf5 ^+/− :MyoD ^+/− mice were interbred to obtain fetuses of nine different genotypes, including Myf5 ^−/− :MyoD ^−/− (Rudnicki et al. 1993; Kablar et al. 1999, 2003).

Tissue processing and genotyping

Fetuses and the fetal portions of the placenta were collected by cesarean section on the required embryonic day (E18.5), and fetuses were prepared for whole-mount alizarin red/alcian blue staining (we analyzed six wild-type, three Myf5 ^−/−, and five Myf5 ^−/− :MyoD ^−/− double mutants), micro-CT scanning (we analyzed two Myf5 ^−/− :MyoD ^−/− double mutants using a Skyscan 1072 100-kV scanner set at 60 kV, with 5 μm isotropic resolution, 19 ms exposure time, four-frame averaging, 0.45° rotation step, and 3D median filtering of reconstructed section stack), and hematoxylin/eosin (H & E) staining (we analyzed two fetuses per genotype). Genomic DNA was isolated from the fetal portion of the placenta by using the procedure of Laird et al. (1991). Fetuses were genotyped by Southern analysis (Sambrook et al. 1989) of placental DNA using Myf5 and MyoD specific probes, as described previously (Rudnicki et al. 1993).

Photography and morphometry

Digital photos (Nikon 4500) were obtained using a stereomicroscope (Leica MZ6) or a compound light microscope (Zeiss Axioplan 2), and panels were generated in Adobe Photoshop 7.0. Long bone measurements (wild type, n=6; Myf5 ^−/− :MyoD ^−/−, n=5) were obtained using a compass and a metric ruler under the stereomicroscope at ×40 magnification. Two-tailed equal-variance t tests (α=0.05) were used to analyze the significance of the difference between mean values. Data were presented as means±standard deviation (means±SD).

Results

Amyogenic fetuses have serious postural anomalies and fused cervical vertebrae

As previously reported, although the Myf5 ^−/− and double-mutant newborns had pink color initially, they became cyanotic and died shortly after birth. Double-mutants did not move in response to stimulation and lacked spontaneous movements due to the absence of striated musculature (Rudnicki et al. 1993; Kablar et al. 2003). Single MyoD ^−/− mutants were not included in this report because their skeleton developed normally and they were perfectly healthy and viable (Rudnicki et al. 1992 and data not shown). In addition to the phenotypic differences, the embryos were genotyped by Southern analysis (Sambrook et al. 1989), and only wild-type, Myf5 ^−/−, and Myf5 ^−/− :MyoD ^−/− fetuses were used in the study.

Employing whole-mount E18.5 cleared mouse fetuses of three genotypes stained with alcian blue (cartilage stain) and alizarin red (bone stain, Fig. 1) (Kaufman 1999), we observed that in comparison with the wild-type fetuses, Myf5 ^−/− fetuses had minor restriction of head movements and somewhat diminished cervical lordosis and thoracic kyphosis. These findings, together with the obvious absence of the rib cage, are in accordance with previous reports (Braun et al. 1992). Here, we primarily used Myf5 ^−/− fetuses as an additional control to our double mutants because Myf5 ^−/− fetuses had essentially normal musculature but had no rib cage and had disorganized intercostal musculature (Braun et al. 1992). On the other hand, Myf5 ^−/− :MyoD ^−/− fetuses completely lacked an entire skeletal musculature and, in addition, had no rib cage, like Myf5 ^−/− fetuses (Rudnicki et al. 1993). In comparison with wild-type and Myf5 ^−/− fetuses, Myf5 ^−/− :MyoD ^−/− double mutants were completely unable to tilt the head back, and their vertebral column acquired almost a “C” shape (as opposed to the normal “S” shape) in the lateral view. Therefore, while the wild-type head could easily tilt back, this was slightly restricted in Myf5 ^−/− fetuses and impossible in double-mutant fetuses, which are shown to have the transverse plane of the head perpendicular to the vertebral column (Fig. 1). Further examination of the double mutants revealed enlarged cervical vertebrae (primarily C1 and C2) and five sites of cervical intervertebral fusion (Fig. 2).

Fig. 1

Amyogenic fetuses have postural anomalies. Lateral views of the skeletal systems of whole-mount E18.5 cleared mouse fetuses of wild-type (a), Myf5 ^−/− (b), and Myf5 ^−/− :MyoD ^−/− (c) genotypes stained with alcian blue (cartilage stain) and alizarin red (bone stain). While wild-type (a) fetuses have normal cervical lordosis and normal thoracic kyphosis, giving the vertebral column the typical “S” shape in the lateral view, the Myf5 ^−/− (b) mutants have an intermediate phenotype, exhibiting absent costal structures, some restriction of head movements, and slight cervical lordosis and thoracic kyphosis. Finally, Myf5 ^−/− :MyoD ^−/− (c) fetuses are unable to tilt the head back, and their vertebral column acquires almost the “C” shape in the lateral view. Arrow in c indicates bifurcated (or “V”-shaped) sternum, only observed in Myf5 ^−/− :MyoD ^−/− fetuses. Bar line, 5 mm

Fig. 2

Amyogenic embryos have fused cervical vertebrae. E18.5 wild-type (a), Myf5 ^−/− (b), and Myf5 ^−/− :MyoD ^−/− (c) fetuses were stained with alcian blue (cartilage) and alizarin red (bone). Some cervical vertebrae (C1–C6) are indicated for comparison between the genotypes. While no vertebral fusion is visible in wild-type (a) fetuses, one site of vertebral fusion is identified in Myf5 ^−/− (arrowhead in b) fetuses between the second and the third cervical vertebrae (i.e., C2 and C3). In the double-mutant (c) fetuses, five sites of fusion (arrowheads in c) are visible between the first six cervical vertebrae (i.e., C1–C6). Bar line, 1 mm

Taken together, we conclude that the striking postural anomalies observed in double-mutant fetuses are likely a consequence of the absence of the musculature rather than a direct consequence of the absence of Myf5 and MyoD. In fact, Myf5 and MyoD are never expressed in the normally developing skeletal system (Kablar et al. 1999; Kablar and Rudnicki 2000; and data not shown).

Neurocranium and viscerocranium are essentially normal, with the exception of the mandible and the palate

The neurocranium consists primarily of the flat membrane bones protecting the brain and the chondrocranial elements at the base of the skull, which develop by endochondral ossification (Sadler 2003). Careful examination of wild-type, Myf5 ^−/−, and double-mutant fetuses revealed no overt abnormalities of the neurocranial skeleton in whole-mount preparations (Fig. 3). These findings are confirmed for double mutants in CT scans (data not shown).

Fig. 3

Neurocranium and viscerocranium are essentially normal, with the exception of the mandible. Whole-mount alcian blue (cartilage) and alizarin red (bone) staining of E18.5 wild-type (a), Myf5 ^−/− (b), and Myf5 ^−/− :MyoD ^−/− (c) fetuses show the following structures: 1 external naris, 2 membranous ossification in nasal bone, 3 membranous ossification in frontal bone, 4 parietal bone and cartilage primordium, 5 interparietal bone and cartilage primordium, 6a supraoccipital bone and cartilage primordium, 6b exoccipital bone and cartilage primordium, 6c basioccipital bone and cartilage primordium, 7a squamous part of temporal bone and cartilage primordium, 7b cartilage primordium of the petrous part of temporal bone (adjacent are malleus, incus, and stapes), 8 os tympanicum, 9 membranous ossification in premaxilla, 10 membranous ossification in maxilla, 11a membranous ossification in the body of mandible (process and ramus to the left), 11b lower incisor tooth, 11c anterior extremity of Meckel's cartilage (blue), and 12 ossification center in cartilage primordium of the body of hyoid bone. Even though most of the identified structures are essentially normal among the three examined genotypes, the mandible is significantly affected in Myf5 ^−/− :MyoD ^−/− (c) fetuses only. The mandible is more sharply bent at roughly the anterior limit of the molar region, while the posterior portion of the mandible is displaced dorsally and posteriorly, resulting in noticeable retrognathia. C1 atlas, C2 axis. Bar line, 2 mm

The viscerocranium refers to the skeleton of the face (Sadler 2003). In comparison with the wild-type and Myf5 ^−/− fetuses, the double mutant zygomatic arch and the mandible were displaced posteriorly, as observed in mdg fetuses (Herring and Lakars 1982). The mandible was bent, while its shorter and posteriorized appearance produced a significant retrognathia.

Surprisingly, fusion of the primary and secondary palates did not occur in double-mutant fetuses (Fig. 4 and data not shown). Sagittal head sections revealed that in comparison with wild-type (and Myf5 ^−/− fetuses), double-mutant fetuses had no tongue muscle. A cleft between the primary and secondary palates in the double mutant, but not in the controls at the same stage, was also observed (Sadler 2003). Therefore, it appears that the tongue muscle, together with other palatal shelf muscles, plays an important role in palatogenesis. Since double-mutant fetuses lacked the tongue muscle, the mechanical obstruction by the tongue could not be the reason for the cleft palate formation, as suggested in the literature (reviewed by Herring 1994). By contrast, it appears that the lack of fusion of the palates, as observed in Myf5 ^−/− :MyoD ^−/− embryos, is a consequence of the absence of mechanical stimulation from the adjacent muscles, and therefore the absence of the muscular mechanical forces for fusion of the palates appears as a reasonable alternative.

Fig. 4

Fusion of primary and secondary palates does not occur in double mutants. Serial sagittal sections of E18.5 wild-type (a) and Myf5 ^−/− :MyoD ^−/− (b) fetuses were stained with H & E. The tongue muscles are absent (compare the arrows in a vs b) and the palates have failed to fuse (compare the arrowheads in a vs b) in the double mutants. Original magnification, ×100

The size of the scapula, deltoid tuberosity of the humerus, shape of the femur, and proximity of long bones were affected in double mutants

Compared with wild-type and Myf5 ^−/− fetuses, in amyogenic fetuses we observed a significantly smaller scapula, a complete absence of the deltoid tuberosity of the humerus, and a noticeable decrease in the separation between the radius and the ulna (Fig. 5). However, the length differences between wild-type and double-mutant humerus, radius, and ulna were not significant based on mean values (Table 1), even though all bones were 1–2 mm shorter in double mutants. Normal hand bone development was found.

Fig. 5

In the forelimb, the scapular size, the deltoid tuberosity of humerus, and the distance between radius and ulna are altered in double mutants. A comparison is shown between the forelimbs of stage E18.5 wild-type (a), Myf5 ^−/− (b), and double-mutant (c) fetuses stained with alcian blue and alizarin red. While the shape and the size of all the forelimb bones are identical between wild-type (a) and Myf5 ^−/− (b) fetuses (e.g., S scapula; arrowhead pointing towards the tuberosity of humerus; arrow indicating the distance between ulna and radius), in Myf5 ^−/− :MyoD ^−/− (c) fetuses, scapula is significantly smaller (S in c), the tuberosity of humerus is completely absent (arrowhead in c), and finally, the distance between ulna and radius is reduced (arrow in c). Bar line, 1 mm

Table 1 Mean long bone length measurements (in centimeters) at stage E18.5 of Myf5 ^−/− :MyoD ^−/− fetuses compared with those of the wild-type

The deltoid tuberosity of the humerus is the normal site of attachment of the spinodeltoideus muscle, which also expands onto the scapula (Greene 1935). Normal contraction of this muscle would exert a mechanical force on both bones. In the absence of striated muscle, not only is the deltoid tuberosity absent, but the scapula fails to develop its proper size and shape. Greene (1935) illustrates that the wild-type rat scapula is characterized by a curved outer surface, whereas the double-mutant scapula does not appear to have the same magnitude of curvature. Thus, we suggest that the scapula may be molded through the attachment and contractile activity of its associated musculature and that the deltoid tuberosity is highly dependent on muscular stimulation for its proper formation. The absence of the deltoid tuberosity of the humerus was expected since it is also reported absent in mdg mutant fetuses (Pai 1965).

In the hind limb, in comparison with wild-type and Myf5 ^−/− fetuses, the double-mutant femur was rectangular in shape and shorter, while a reduction in separation was observed between the tibia and fibula, which were fused distally (Fig. 6). The reduced longitudinal bone extension of the femur and tibia (Table 1) may suggest muscle-dependent primary bone extension in this region. Normal foot bone development was found.

Fig. 6

In the hind limb, the length and shape of femur and the distance between tibia and fibula are altered in double mutants. A comparison is shown between the hind limbs of stage E18.5 wild-type (a), Myf5 ^−/− (b), and double-mutant (c) fetuses stained with alcian blue and alizarin red. While the shape and the size of all the hind limb bones are identical between wild-type (a) and Myf5 ^−/− (b) fetuses (e.g., F femur; asterisk indicating the distance between ulna and radius), in Myf5 ^−/− :MyoD ^−/− (c) fetuses, the femur has a rectangular shape (F in c) and the distance between the tibia and fibula is almost nonexistent (asterisk in c). Bar line, 2 mm

Amyogenic fetuses have heteromorphic clavicles and bifurcating sternum

Like the primary and the secondary palates, the double-mutant sternum did not completely fuse before birth and was bifurcated, significantly shorter, and “V” shaped (Figs. 1c and 7). This is in contrast to the Myf5 ^−/− sternum, which clearly fused into a single midline structure but acquired the “J” shape (Braun et al. 1992; Fig. 7b). The wild-type sternum was fused and had straight appearance, with easily identifiable sternebrae.

Fig. 7

Bifurcated and smaller sternum and heteromorphic clavicles are visible in Myf5 ^−/− :MyoD ^−/− fetuses. E18.5 wild-type (a), Myf5 ^−/− (b), and Myf5 ^−/− :MyoD ^−/− (c) thoraxes were stained with alcian blue and alizarin red. While the shape and the size of clavicles are essentially identical between wild-type (arrows in a) and Myf5 ^−/− (arrows in b) fetuses, the double-mutant clavicles are deformed, smaller, and bent (arrows in c). While the wild-type fused sternum (S in a and asterisks indicating sternebrae) is normally developed (and somewhat obscured by the extensive costal structures), the Myf5 ^−/− fused sternum (S in b and asterisks indicating sternebrae) is bent at its xiphoid end, resulting in the “J”-shaped sternum. Finally, Myf5 ^−/− :MyoD ^−/− sternum does not develop normally. It remains bifurcated (double S in c) and it is very short and “V” shaped (also see Fig. 1, arrowhead in c). Bar line, 2 mm

In addition, the shape and the size of the clavicles significantly changed in double mutants only (Fig. 7), in comparison with wild-type and Myf5 ^−/− fetuses. The double-mutant clavicles were much smaller and bent, suggesting that skeletal musculature plays a role in proper clavicular formation.

Discussion

Using Myf5 ^−/− :MyoD ^−/− fetuses, which completely lack striated muscle (Rudnicki et al. 1993), we were able to examine the developmental dependence of the embryonic skeletal system on muscle loading and contraction. We report a number of skeletal anomalies, some of which are consistent with those in well-documented myopathies and others that are unique to the amyogenic phenotype. These include enlarged and fused cervical vertebrae, general postural anomalies, an essentially normal neurocranium, viscerocranial abnormalities including the heteromorphic mandible, long bone truncation and fusion, absent deltoid tuberosity of the humerus, scapular and clavicular hypoplasia, and cleft palate and sternum. However, hand and foot bones developed normally in the absence of muscle.

The restricted dorsal head movements in double-mutant fetuses have previously been reported in muscular dysgenic (mdg homozygous) fetuses (Pai 1965), which have striated muscle inactivity similar to the mutant fetuses in question. Pai (1965) suggests that the proximity and the enlargement of the cervical vertebrae are the probable causes for the murine mdg postural limitations, both of which were observed in the double mutant in addition to five sites of intervertebral fusion in the cervical vertebral column. Considering that Myf5 ^−/− :MyoD ^−/− fetuses completely lack striated muscle, as opposed to the degrading of partially developed muscle in the mdg myopathy, the associations between the conditions of the cervical vertebrae and the postural appearance may apply even more strongly in our model. Therefore, a likely relationship between muscle contraction, neurocranial morphology, and overall posture is suggested. However, the opposite phenotype of flatter and more elongated skulls in mdg fetuses contradicts the trend of these findings (Lightfoot and German 1998). With the well-established notion that cranial vault geometry depends largely on brain growth, which is mostly completed before the dystrophic phenotype becomes apparent (Lightfoot and German 1998), it is clear that the dystrophic phenotype is not an appropriate neurocranial comparison to the double-mutant fetuses. This is likely because the dystrophic fetuses experience normal muscular stimulation in early development, whereas the amyogenic fetuses do not.

Finally, it is possible that the MRFs directly affect primary bone growth. This is most clearly evidenced by the apparent lack of ribs in Myf5 ^−/− and double-mutant fetuses. The effect is obviously much less extreme for other skeletal components; hence, direct MRF effects appear to be isolated only in the skeletal system.

Strikingly, although the tongue muscle is completely absent from the earliest stages of Myf5 ^−/− :MyoD ^−/− embryo development, lack of fusion of the secondary palatal shelves, palatoschisis (i.e., the cleft palate), is clearly visible in double-mutant term embryos. Therefore, we suggest that the cleft palate is not a consequence of tongue obstruction but of a complete absence of any mechanical help from the facial musculature during the process of merging and fusion of the palatal shelves. Findings in paralyzing conditions that result in a high frequency of cleft palate support this hypothesis (Herring and Lakars 1982). Moreover, cleft palate is normally observed in humans with micrognathia (i.e., Pierre Robin syndrome; Sperber 2001), and we have now observed it in Myf5 ^−/− :MyoD ^−/− fetuses with a similar mandible condition. Finally, epithelial–mesenchymal transformation at the medial edges of the lateral shelves is also essential to shelf fusion (Sperber 2001). In this respect, TGF-β ligands are known to play an important role in palatogenesis (Hall 1994). TGF-β3 knockout fetuses have cleft palates, and this model system is useful in that it provides a specific genetic source for the complication (Hall 1994; Koo et al. 2001). Therefore, the examination of cell communication events at the shelf fusion midline in the double-mutant vs wild-type embryos is our future goal.

While a decrease in the size of the scapula has been consistently found in mdg fetuses (Pai 1965), peripheral curvature abnormality has not been reported. One reasonable possibility is that the mechanical loading of the spinodeltoideus muscle (Greene 1935) and other shoulder muscles is responsible for the surface curvatures of the scapula. Without these forces, as in the double mutant, the bone may not be molded correctly. Furthermore, this defect may not have been observable in mdg fetuses because of the influence of early muscle contraction. Finally, scapular defects have been reported to be associated with brachial plexus birth palsy and associated muscular imbalance (Waters et al. 1998), giving further credence to the importance of muscle contraction in scapular form.

It has long been known that chondrocyte growth plates are present at each end of a long bone (Farnum 1994, 2005). They are the principal sources of longitudinal bone growth following endochondral expansion from the primary ossification center of the diaphysis. However, the degrees of vascular and mechanical influences are not well understood (Moss-Salentijn 1992) and deserve further attention. The MRF double-mutant fetuses serve as a useful model for evaluating the mechanical influences, as has been done previously (Moss-Salentijn 1981 in Moss-Salentijn 1992), but with less drastic phenotypes. The complete lack of striated muscle is expected to significantly reduce these forces, which are known to be required for metaphyseal growth plate development (Letts 1988 in Moss-Salentijn 1992). Thus, it is expected that long bone development would be reduced in Myf5 ^−/− :MyoD ^−/− fetuses.

However, only the length of the femur and tibia was significantly reduced in the double mutants, with the radius, ulna, and humerus all within statistical bounds of similarity between the two genotypes. Virtually all growth plates present unique growth rates despite similar cellular organization and sequential events, from a chondrocyte stem cell population to hypertrophy and bone extension (Farnum 1994), and this applies even between two ends of the same bone (Farnum 2005; Hall 2005). Thus, it is not surprising that within an embryo different bones are differentially affected by muscular agenesis.

A slight decrease in the separation between the radius and the ulna was also apparent in the double mutants, but this type of effect has not been accounted for in the literature. We suggest that long bone separation in normal development is maintained by muscular forces, which are reinforced by more extreme observations in the tibia and fibula of the double-mutant mouse. As with the radius and ulna, the separation between the tibia and fibula in the hind limb was vastly reduced in amyogenic fetuses, and these bones appear fused distally. Note that the fusion itself between the bones of the lower leg is considered normal (Greene 1935); only the longitudinal extent of fusion is heteromorphic. Finally, the normal curvature of the fibula is also absent in the double mutants, allowing the tibia to come closer. In fact, all long bones normally have a curvature that is a consequence of muscle action but that is also mechanically more stable than a straight cylinder (Bertram and Biewener 1990).

While Herring (1994) indicates that synovial joint formation requires embryonic movement, our results show that the hand and foot bone and joint formation process appears normal in both the wild-type and double mutant. There are at least two possible explanations. The first, although less likely, is that the double-mutant embryos receive sufficient intrauterine stimulation for normal synovial joint formation. The second, and more likely in light of these results, is that synovial joint formation in the hands and feet does not necessarily depend on early embryonic movement. Indeed, this latter suggestion is consistent with the fact that some other double-mutant bones form incorrectly, but they form nonetheless. It may be that these bones would not develop normally postnatally, and Herring (1994) does acknowledge that joint cavitation can begin normally despite failure of complete development in paralyzed embryos. We were unable to study this phenotype after birth because of the 100% postpartum lethality of the newborns.

In addition to palatal clefts, the double mutants also exhibited sternal clefts. While this might conceivably be associated with the lack of ribs in the double mutant, the sternum is known to develop independently of the ribs, forming two cartilaginous bands of somatic mesodermal origin, which fuse and undergo endochondral ossification in normal development (Sadler 2003). A medical condition known as cleft sternum results from the failure of mesodermal fusion into a single sternal rod, and this often produces visceral herniation at the abdomen (Sadler 2003). We have suggested that the presence of all other musculature in Myf5 ^−/− fetuses may allow for normal sternal fusion despite vastly reduced ribs. Alternatively, MyoD may play a direct role in sternal fusion and rib formation, thus allowing the Myf5 ^−/− fetuses to develop a fused sternum. In fact, recent results indicate that the functional redundancy of the MRFs extends to their effects on rib and general axial skeleton development. For example, knock-in replacement of Myf5 with myogenin allows for normal rib development (Wang et al. 1996). Somitic communication between the myotome and sclerotome is also known to be defective in the absence of Myf5, but a MyoD knock-in replacement for Myf5 appears to restore chondrogenic gene expression (Tallquist et al. 2000), suggesting that it may function redundantly with Myf5 in endochondral ossification processes like that at the sternum. Remarkably, a case has been reported in which an individual had cleft lip, cleft sternum, and a mandibular cleft (Seyhan and Kylynr 2002), a combination of abnormalities affecting many of the same structures reported heteromorphic in the double-mutant mouse, but which could be due to failure of midline fusion of the entire facial area rather than muscle action.