In the mouse embryo, the cranial paraxial mesoderm contributes to the formation of the chondrocranium, the craniofacial musculature, and connective tissues. In the trunk region, the paraxial mesoderm segments into a series of paired somites, which differentiate to form the dermis, the epaxial and hypaxial muscles, and the vertebrae. In addition to being a major source of precursors of mesodermal tissues, the paraxial mesoderm also provides the substrate on which the neural crest cells (NCCs) migrate. In the cranial region, NCCs arise at the junction between the neural plate and the surface ectoderm and migrate to populate the craniofacial primordia. Cranial NCCs differentiate into an assortment of tissues, including bone, connective tissue, pigment cells, teeth, peripheral neurons, glia, and the tissue of the cardiac outflow tract. In the trunk region, NCCs give rise to the spinal ganglia and nerves, the autonomic ganglia and nerves, the adrenal medulla tissues, and the melanocytes of the skin.
Twist encodes a basic helix-loop-helix (bHLH) factor that is expressed in cranial NCCs, the paraxial mesoderm of the head and trunk, and the limb bud mesenchyme of mouse embryos (Fuchtbauer, 1995). Loss of Twist function results in poor growth and abnormal patterning of the limb bud tissue, failure of the cranial neural tube to close, and hypoplastic branchial arches (Chen and Behringer, 1995; Zuniga et al., 2002; Soo et al., 2002; O'Rourke et al., 2002). Although loss of Twist function does not affect the formation of the cranial NCCs or their ability to populate craniofacial primordia, the migration and localization of Sox10-expressing NCCs within the first branchial arch are abnormal (Soo et al., 2002). When the migration of the cranial NCCs was investigated by transplantation of wild-type NCCs to host embryos of different Twist genotypes, normal Twist activity in the cranial mesenchyme was found to be essential for proper deployment of NCCs to the facial mesenchyme and the first branchial arch (Soo et al., 2002). In the wild-type host embryo, Twist-deficient NCCs were sequestered to the core mesenchyme of the arch where myogenic progenitors but not NCCs are normally localized (Chen and Behringer, 1995; Soo et al., 2002). The abnormal positioning of the mutant NCCs might preclude an effective interaction with the surface ectoderm, which is critical for the differentiation of the NCCs (Ferguson et al., 2000). Consistent with this idea, the differentiation of Twist-/- cranial NCCs has been found to be defective. In wild-type embryos, Sox10 is down-regulated and Sox9 and Gsc are up-regulated in the postmigratory NCCs in the branchial arches, but this effect does not occur in Twist-/- embryos. In addition, trabecular bone and tooth differentiate poorly in transplants of Twist-/- branchial arches that were grown in the subcapsular space of the adult kidney (Soo et al., 2002), suggesting that Twist-deficient cranial NCCs have a reduced histogenetic potential. However, the differentiation and patterning of the neural derivatives of the cranial NCCs in Twist-/- embryos have not been examined. In addition, it is not known whether the differentiation of NCCs at other levels of the body axis may be affected by Twist deficiency.
In the somites, Twist is expressed in the dermomyotome, and later in the sclerotome (Wolf et al., 1991; Hebrok et al., 1994). However, Twist is not expressed in the myotome in which Myf5 and MyoD are expressed, signifying early myogenic differentiation. The mutually exclusive expression of Twist and myogenic factors suggests that Twist may act as an inhibitor of muscle differentiation (Sassoon, 1993; Li et al., 1995). Information on the effect of loss of Twist function on the differentiation of the mesoderm is limited. Loss of Twist expression does not seem to have any significant impact on mesoderm formation. However, an early myogenic factor, myogenin, is down-regulated in the somites of mutant embryos (Chen and Behringer, 1995), suggesting that Twist may be required for initiating myogenic differentiation. Because other molecular markers for myogenic and skeletogenic differentiation of the Twist-deficient somites have not been tested, the goal of the present study is to conduct a more extensive analysis of lineage-markers to reveal the impact of loss of Twist function on somite differentiation. In addition, we have tested the histogenetic potency of the Twist-deficient somites by explanting the tissues before the demise of the mutant embryo in vivo and examining the propensity of tissue differentiation after further growth of the tissue by transplantation to the adult kidney.
Abnormal Distribution of Sox10-Expressing Cells in the Cranial Mesenchyme and the Paraxial Mesoderm of the Trunk
In the craniofacial region of the embryonic day (E) 9.5 wild-type embryo, Sox10-expressing cells were found in the mesenchyme around the optic evagination of the forebrain and in the upper facial region. In the paraxial mesenchyme at the lower midbrain to upper hindbrain levels, Sox10-expressing cells were localized to the trigeminal ganglion (gV) and facioacoustic (geniculate and vestibulocochlear) ganglion (gVII/VIII; Fig. 1A,C). In the postotic region, Sox10-expressing cells were associated with the glossopharyngeal ganglion (gIX), the vagus (gX) ganglia, and the occipital nerve roots that contribute to the hypoglossal (XII) nerve (Fig. 1A). In contrast, Sox10-expressing cells were distributed more widely in the cranial mesenchyme of E9.5 Twist-/- mutant embryos (Fig. 1B), particularly in the mesenchyme ventral to the rostral part of the open cephalic neural folds and in the maxillary region. Contrary to the congregation of Sox10-expressing cells in the cranial nerve ganglia in the wild-type embryo (Fig. 1A,C), Sox10-expressing cells in the mutant embryo were found not only at the site of the prospective gV and gVII/VIII (Fig. 1B) but were distributed more distally into the first and second branchial arches (Fig. 1D,E). In the mesenchyme proximal to the branchial arches, Sox10-expressing cells were not restricted to the region immediately subjacent to the surface ectoderm but invaded more deeply into the core of the paraxial mesoderm (Fig. 1D,E). Of the 13 Twist-/- embryos studied for the pattern of Sox10 expression, 10 showed fusion of the trigeminal and facioacoustic clusters of the Sox10-expressing cells (Fig. 1B), which led to the obliteration of the NCC-free zone that normally separates these two cranial nerve ganglia in the wild-type embryo (Fig. 1A). In the postotic region of five Twist-/- mutant embryos, Sox10-expressing cells were organized into distinct clusters that are reminiscent of gIX and gX. However, in other mutant embryos, Sox10-expressing cells formed irregular clusters (n = 4, Fig. 1B) or were widely scattered in the postotic paraxial region (n = 4). Our findings show that patterning of the Sox10-expressing NCCs in the cranial region of the embryo was disrupted by the loss of Twist function.
In the trunk region of the E9.5 wild-type embryo, Sox10-expressing cells formed segmental clusters in the paraxial mesoderm heralding the formation of the dorsal root ganglia and spinal nerves (Fig. 1F,H). While Sox10-expressing cells were present in the paraxial mesoderm in the cervical to lower thoracic level in the wild-type embryo (Fig. 1A,F), in Twist-/- embryos of equivalent developmental stages, they were not found in somites as caudal as those in the wild-type embryo (Fig. 1A,B). In the most affected Twist-/- embryo, Sox10-expressing cells in the trunk region were either scattered and not organized into segmental clusters or localized along the dorsal border of the somite (Fig. 1F,G).
During embryonic development, the entry of the NCCs into the paraxial mesoderm follows the craniocaudal progression of somite differentiation. To compare the timing of onset of NCC migration into the paraxial mesoderm, 14 wild-type and 12 Twist-/- embryo at the 20- to 24-somite stage were analysed to determine the axial position of the most caudal somite that was colonized by the NCCs. Our results showed that NCCs tended to be found in somites in more cranial positions in the Twist-/- embryo than the wild-type embryo at equivalent somite-number stages (Table 1). In the wild-type embryo, NCCs have already migrated into somites located more caudally in the body axis, indicating that migration began sooner after somite formation. This finding suggests that in the mutant embryo NCCs commenced migration after a longer lapse of time since somite formation. Onset of neural crest cell migration into the trunk paraxial mesoderm, therefore, may have been delayed in Twist-/- embryos.
Table 1. Onset of NCC Migration as Determined by the Position of Most Posterior Somite in the Craniocaudal Axis That Contains the Sox10-Expressing Cells
The developmental stage of the embryo was determined by the somite number. The most posterior somite in the body axis that contains Sox10-expressing cells was scored for assessing the craniocaudal progression of neural crest cell migration into the paraxial mesoderm; significant difference between wild-type and Twist−/− embryo by Mann–Whitney test (P < 0.002, two-tailed). NCC, neural crest cell.
Wild-type (14 embryos)
Twist−/− (12 embryos)
In the trunk region where Sox10-expressing cells were found in the paraxial mesoderm, the NCCs were distributed along the full dorsoventral distance in the paraxial tissue of both wild-type and Twist-/- embryos (Fig. 1H,I), suggesting that, once migration has commenced, the NCCs can travel along the same path taken by their counterparts in the wild-type embryo. However, in the specimen shown in Figure 1I, the Sox10-expressing cells were not yet incorporated into spinal ganglia and nerves. The delay in the migration and incorporation of the Sox10-expressing cells into dorsal root ganglia is not due to the general retardation in the development of the mutant embryo, because the wild-type and mutant embryos formed similar numbers of somites up until E10.5 (Twist+/+, 34.5 ± 1.4 pairs of somites [n = 25]; Twist-/-, 35.3 ± 0.5 [n = 49]). Development became retarded thereafter such that significantly fewer somites were formed at E11.5 (Twist+/+, 41.7 ± 0.9 [n = 20]; Twist-/-, 36.9 ± 1.0 [n = 17], P < 0.001 by Student's t-test).
Disrupted Patterning of the Cranial Nerves and Ganglia
The abnormal pattern of Sox10 expression raises the possibility that the differentiation and morphogenesis of the neural derivatives of the NCCs may be disrupted in the Twist-/- embryo. To examine the development of the cranial nerves and ganglia, whole-mount immunostaining was performed by using an anti-neurofilament monoclonal antibody on E10.5–E11 wild-type and Twist-/- embryos. In the wild-type embryo, two prominent nerve ganglia were formed in the preotic cranial region: the trigeminal (gV) at the base of the maxillary prominence and the mandibular arc, and the facioacoustic (gVII-VIII) ganglia complex in at the base of the first and second branchial arch (Fig. 1J). The trigeminal ganglia projects three branches of nerves: the ophthalmic nerves to the periocular and upper facial tissues, the maxillary nerves to the upper jaw tissues, and the mandibular nerves to the lower jaw tissues (Fig. 1J,L). Immunostaining of Twist-/- embryos (n = 15) revealed an abnormal organization of the cranial ganglia and nerves. The trigeminal and the facioacoustic ganglia were localized ectopically to a more dorsal position near the edge of the open neural plate (Fig. 1K,M). In some mutant embryos, the trigeminal ganglion and the facioacoustic ganglion fused together (Fig. 1K,M) and the facioacoustic ganglion was less tightly packed together (Fig. 1N,O, compare with Fig, 1J,L). The efferent nerves of the trigeminal and facial ganglia displayed abnormal patterns, including the loss or disorganization of the ophthalmic and maxillary branches (Fig. 1M–O), truncated/foreshortened mandibular branch (Fig. 1O), ectoptic projection of the ophthalmic branch dorsally to the rostral region of the open neural tube (Fig. 1M), intermingling of nerves of trigeminal and facial ganglia (Fig. 1M), the diversion of the facial nerve to join the distal segment of the vagus nerve (Fig. 1O), and the more loosely bundled facioacoustic nerves (Fig. 1N,O).
In the postotic region of the wild-type embryo, neurofilament staining revealed the presence of the glossopharyngeal nerves, which innervate the third (hyoid) branchial arch, and the vagus nerve with its jugular branch, which arches anteriorly and then ventrally to join the nodose ganglion and innervates the heart and the gut (Fig, 1P). In the Twist-/- embryo, development of the glossopharyngeal nerve and vagus nerve was affected: the glossopharyngeal nerve merged with the anterior jugular branch of the vagus nerve (Fig. 1K,P), the vagus was foreshortened (Fig. 1Q), and the nodose ganglion expanded in size (Fig. 1K). In the wild-type embryo, nerves originating from the lower hindbrain and the occipital segments of the spinal cord contribute to the hypoglossal nerve, which innervates the lingual tissues (Fig. 1P). In marked contrast, the hypoglossal nerve of the Twist-/- embryo did not form properly and the occipital roots that contribute to this nerve were short or partially absent. The nerve trunk that gathered these occipital roots meandered through the postotic region of the embryo and did not form a proper hypoglossal nerve (Fig. 1Q) that should have projected to the ventral pharynx (Fig. 1P).
Loss of Twist Does not Affect the Patterning of the Trunk NCCs
The developmental lag in the appearance of Sox10-expressing cells in the somites along the craniocaudal axis (Fig. 1A,B,F,G; Table 1) highlights an initial delay in the migration of the NCCs into the paraxial mesoderm. In wild-type embryos, Cited2 is expressed in the dorsal half of the somites (Fig. 2A). In mutant embryos, Cited2 was expressed in a disjointed and irregular pattern in the dorsal part of the sclerotome (Fig. 2B), suggesting that some changes in the local tissue environment may have impeded the migration of the NCCs. Despite the initial delay in migration of the NCCs (Fig. 1G,I), segmental patterning of the peripheral nerves (spinal ganglia and nerves, and sympathetic trunk) in the Twist-/- embryo was indistinguishable from that in the wild-type embryo (Fig. 1R,S). Consistent with this outcome is the normal segmental organization of the somites of the Twist-/- embryo as revealed by metameric pattern of expression of Meox1 (Fig. 2C,D), Foxc1/Mf1, Foxc2/Mfh1, Hoxb8, and Fgf8 (data not shown) and of the regionalized expression within the somite of Uncx4.1 (Fig. 2E,F), Dll3, Lfng, and Jag1 (data not shown) that reflects appropriate specification of the rostrocaudal polarity. In the wild-type embryo, Ephb2/b3 (Fig. 2G) and Efnb2 (Fig. 2I) are expressed in the cranial (anterior) and caudal (posterior) halves of the somites, respectively (De Bellard et al., 2002). In the Twist-/- embryo, the expression Ephb2/b3 (Fig. 2H) was broader but still regionalized to tissues in the posterior two third of the somites. Efnb2 expression was weaker but was localized to the anterior half of the somite (Fig. 2J) as in the wild-type embryo (Fig. 2I). The correct rostrocaudal compartmentalization of the somite may underpin the normal patterning of the spinal nerves and ganglia in the trunk of Twist-/- embryos, despite the initial disruption of cell migration.
Less Prolific Myogenic and Skeletogenic Differentiation of Twist-Deficient Somites
To examine more fully the differentiation potency of the somitic tissues of the Twist-/- embryos, we have performed an extensive analysis of the expression of molecular markers for the dermomyotome and the myotome. In the wild-type somite, cells in the dorsomedial part of the epithelial dermomyotome expressed Bmp4 (Fig. 3A) and those in the ventrolateral part express Sim1 (Fig. 3C), Alx4 (Fig. 3E), and Wnt6 (not shown). In the Twist-/- dermomyotome, Bmp4 (Fig. 3B) and Alx4 (Fig. 3F) were not expressed, whereas the expression of Sim1 (Fig. 3D) and Wnt6 (not shown) was markedly reduced. In the Twist-/- somites, expression of En1 was absent from the dermomyotome and the sclerotome (Fig. 3G,H). Sox9 expression, which marks chondrocyte differentiation, was reduced in the sclerotome (Fig. 3I,J), suggesting that early sclerotome differentiation may be affected. Loss of Twist activity led to down-regulation of Paraxis in the sclerotome in the mutant embryos (Fig. 3K,L) but did not affect the expression of two related bHLH transcription factors: Scleraxis in the tendon precursor and Dermo1 in the dermomyotome (data not shown). In addition to the reduced expression of myogenin (Chen and Behringer, 1995), another myogenic determinant, Myf5, which signifies the differentiation of the myotome, was also down-regulated in the myotome (Fig. 3M,N). In contrast, Pax3 expression in the myotome newly emerged from the dermomyotome was unchanged in the Twist-/- embryo (Fig. 3O,P).
Altered expression of molecular markers of early somite differentiation in Twist-/- embryos (Fig. 3) suggests that the differentiation of the skeletal tissues and the muscle may be disrupted by the loss of Twist function. To overcome the confounding effect of early lethality of Twist-/- embryos that prevents a complete assessment of tissue differentiation, we tested the histogenetic potency of Twist-/- somites by transplanting the somitic tissues to the subcapsular space of the kidney to extend their developmental life span. Approximately 60% of the grafts produced a tumour-like growth. There were no consistent differences between tumors derived from somites at two axial levels (the cervical and the thoracic region) of the wild-type, Twist+/-, and Twist-/- embryos. The samples, therefore, were pooled into normal (Twist+/+or Twist+/-) and mutant (Twist-/-) groups (Table 2). Tumours of the normal group formed abundant bony and cartilaginous tissues (Fig. 4A), which were organized into skeletal structures resembling vertebrae with elongating processes. Bundles of skeletal muscle fibres were found associated with these skeletal structures in an organotypic manner (Fig. 4C). Tumours derived from Twist-/- somites were smaller (Fig. 4B). Although bone, cartilage, and skeletal muscle were formed, they were not evidently more abundant than other types of tissues in the tumor (Fig. 4D) and they were also formed less frequently amongst the tumours (Table 2). Keratinized epithelial tissues were formed in the tumors, presumably due to the inclusion of some adherent surface ectoderm in the somite transplants. Neural tissues, morphologically resembling aggregates of ganglionic cells and presumed to be derived from NCCs were present in the tumours. The ganglionic mass was often formed between skeletal masses in a manner reminiscent of the anatomical relationship of the dorsal root ganglia and the vertebrae (Fig. 4A and data not shown). Neural ganglionic tissues were less frequently encountered in tumours derived from the somites of mutant embryos (Table 2). Generally, the loss of Twist function does not affect the histogenetic potency of the paraxial mesoderm or the NCCs. However, the lesser growth of tumours derived from mutant tissues suggests that Twist function may influence the growth potential of the tissue.
Table 2. Tissue Composition of Tumours Derived From Somites of E9.5 Twist+/+ or Twist+/− (Normal) Embryos and Twist−/− (Mutant) Embryos After Transplantation to the Subcapsular Space of the Adult Kidney
Normal (Twist+/+ or Twist+/−) (n = 13)
Mutant (Twist−/−) (n = 16)
aThe relative abundance of the tissue was estimated quasiquantitatively by the area it occupied in a histological section of tumor (see text). The number of “+” signs indicates the average relative abundance of different types of tissues for each group of tumours. The number in parentheses shows the number of specimens containing a specific type of tissue. E, embryonic day.
Loose and fibrous connective tissue
Loss of Twist Decreases Cell Proliferation and Elevates Apoptosis in the Branchial Arch and Somites
A possible cause of the less prolific growth of the somites in ectopic sites is that Twist may affect cell proliferation or viability. In mutant embryos, the proportion of cells that had incorporated bromo-2′-doxyuridine (BrdU) into their DNA at S-phase was reduced in the dermomyotome of the lumbar and thoracic somites and in the sclerotome of the lumbar somites (Table 3). The level of BrdU labelling of the mesenchyme of branchial arch of the Twist-/- embryo (50.8 ± 2.13%, n = 3) also appeared to be lower than that of the wild-type embryo (74.7 ± 1.79%, n = 3). The branchial arch epithelium did not show any significant reduction in the level of BrdU labelling, suggesting that the lower proliferative activity is associated with the mesenchyme.
Table 3. BrdU Labelling of Cells in the Paraxial Mesoderm of Twist+/+ or Twist+/− Embryo and Twist−/− Embryos
BrdU labelling: % total cell count (mean ± standard error of the mean)
Significant difference?(Two-tailed Mann–Whitney test for unrelated scores)
Normal (Twist+/+ or Twist+/) (14 embryos)
Mutant (Twist−/−) (16 embryos)
56.9 ± 3.2
47.8 ± 2.3
54.2 ± 2.8
43.6 ± 4.2
62.1 ± 2.8
49.7 ± 2.3
53.9 ± 2.6
43.6 ± 3.3
54.1 ± 4.5
45.3 ± 3.8
Previous studies using acridine orange staining showed that many cells in the cranial mesenchyme and the sclerotome of the Twist-/- embryo are necrotic (Chen and Behringer, 1995). Nile Blue staining of embryos in the present study confirmed that necrotic cells were more numerous in the branchial arches and the somites of E9.5 Twist-/- embryo than the wild-type embryo (Fig. 5A,B). There was a marked increase in Nile Blue-stained cells in the cranial mesenchyme of the mutant embryo by E10.5 (Fig. 5C, wild-type embryos displaying staining pattern similar to E9.5 embryos, data not shown). Terminal deoxynucleotidyl transferase-mediated biotinylated UTP nick end labelling (TUNEL) analysis revealed apoptosis in the cranial paraxial mesoderm, the branchial arch mesenchyme, and the somites. In the branchial arch, TUNEL-positive cells were observed in both the peripheral region and the core mesenchyme (Fig. 5D,E). In the somite, TUNEL-positive cells were found mostly in the sclerotome (Fig. 5F,G). By Nile Blue staining, it was shown that necrotic cells were first detected in the sclerotome in the differentiating somite, which is coincidental with the up-regulation of Twist activity (data not shown).
It was previously reported that Twist may maintain cell proliferation and suppress apoptosis by antagonizing p53 and down-regulating p53 target genes (Maestro et al., 1999). To examine this possibility in the context of embryogenesis, we used reverse transcriptase-polymerase chain reaction (RT-PCR) to analyze the expression of genes related to p53-dependent cell cycle arrest and cell death in Twist-/- branchial arch tissues. Expression of p53 was similar in wild-type and Twist-/- embryos (Fig. 6). In contrast, expression level of p21/WAF1, an inhibitor of cyclin-dependent kinases and regulator of the G1/S transition under the transcriptional control of p53 (Deng et al., 1995; Del Sal et al., 1996), was elevated in Twist-/- branchial arches. MDM2, which regulates p53 function, has been reported previously to be up-regulated in calvarial osteoblast cell lines with Twist haploinsuffiency (Yousfi et al., 2002). In some experiments, we observed a small difference in Mdm2 transcript levels between normal and mutant branchial arches (Fig. 6). In vitro studies have shown that Twist can reduce the p53-dependent induction of Bax, a proapoptotic member of the Bcl2 family (Maestro et al., 1999), and that Twist deficiency can result in the up-regulation of Bax (Yousfi et al., 2002). However, we observed no difference in Bax expression between normal and mutant branchial arches in vivo. It has been suggested that the anti-apoptotic role of Twist in vitro may be mediated by transcriptional control of Tnfα (Yousfi et al., 2002). We observed a strong increase in the abundance of Tnfα (Fig. 6), suggesting that Twist may act by means of this pathway to exert its antiapoptotic activity in vivo.
Twist Is Required for Normal Patterning and Differentiation of the Neural Crest
The first cranial NCCs to leave the hindbrain are destined to form the bones, cartilage, and connective tissues of the craniofacial structures. The NCCs that migrate later form the cranial nerves and ganglia. Analysis of Sox10 expression indicates that it is the later phase of NCC migration that is affected in Twist-/- embryos. This defect may underpin disorganization of the cranial nerve and ganglia in the mutant embryos. Migration and positional identity of NCCs requires information from the surrounding mesenchymal cell population, as revealed by the impact of loss of function of neuregulin or its receptors ErbB2, ErbB3, and ErbB4 on NCC migration and the patterning of the cranial ganglia and nerves (Britsch et al., 1998; Golding et al., 2000). Previous cell transplantation experiments demonstrated that correct NCC migration requires Twist function in both the NCCs and the cranial mesenchyme (Soo et al., 2002), suggesting that Twist has a critical role in regulating both the production of the instructive signals in the mesenchyme and the ability of the NCCs to correctly respond to these signals. The abnormal pattern of the cranial ganglia and nerves may be attributed to defects in guiding the migration and regulating neurogenic differentiation of the NCCs. However, in addition to the NCCs, ectodermal placodes in the cranial region also contribute cells for the sensory neurons of the cranial nerves (Golding et al., 2000; Begbie and Graham, 2001; Baker and Bronner-Fraser, 2001; Baker et al., 2002). Because the development of the cranial placodes in the Twist-/- embryo has not been investigated, it is not known if abnormal development of the placode might have contributed to the disruption of the cranial nerve pattern.
Impact of Loss of Twist Function on Tissue Patterning Is Region Specific
Results of this study show that loss of Twist function has a differential impact on the patterning and differentiation of the neural crest derivatives at different axial positions of the embryo. Disorganization of the nerves and ganglia was evident mainly in the cranial region, whereas in the trunk neural crest migration is delayed but patterning of the peripheral nervous system appears normal. The differential impact on the neural crest at different axial levels might be related to differential expression of Twist in the cranial and trunk NCCs. Twist has been shown to be expressed in the pre- and early migratory cranial NCCs and in the nerve ganglia (Fuchtbauer, 1995; Gitelman, 1997; Soo et al., 2002). Expression of Twist has not been described in the NCCs of the trunk. In the somite, Twist is expressed most strongly in the sclerotome in the caudal part of the somite (Fuchtbauer, 1995; Gitelman, 1997; Loebel et al., 2002) where normally NCCs would not be found.
Region-specific defects have also been observed in the development of the neural tube and the limb. In the Twist-/- embryo, neural tube closure defects and disruption of dorsoventral tissue patterning are found only in the cranial region (Chen and Behringer, 1995; Soo et al., 2002). Growth and morphogenesis of the forelimb is severely disrupted in Twist-/- embryos, whereas the hindlimb is less affected. In Twist+/- embryos, the hindlimb develops preaxial polydactyly but the forelimb does not (El Ghouzzi et al., 1997; O'Rourke et al., 2002). The region-specific effect of Twist on patterning and differentiation suggests the existence of downstream targets with expression restricted to a subset of Twist-expressing tissues. In a differential screen for downstream targets of Twist, we isolated a novel gene that is expressed exclusively in tissues in the rostral half of the body at E10.5. Expression of this gene was absent from these tissues in Twist-/- embryos, hinting that this downstream gene might be a critical factor the localized defects of development in the upper body of the mouse embryo (Loebel et al., 2002).
Role for Twist in the Maintenance of Mesodermal Progenitors
Our data show that Twist is required for the maintenance of cell viability and proliferation in somitic and branchial arch tissues. We have observed an up-regulation of p21/WAF1 in mutant tissues concomitant with reduced cell proliferation. Previously, Twist was shown to be capable of repressing p21/WAF1 transcription in vitro, possibly by preventing its activation by E2A and CBP (Funato et al., 2001). Twist may also repress p21/WAF1 transcription and affect cell proliferation in vitro through interference with p53-dependent transcriptional activation (Maestro et al., 1999). Our study provides the first demonstration that Twist may operate as part of a similar mechanism in developing embryonic tissues.
Twist is also required to maintain the viability of cells in the sclerotome and the mesenchyme of the first branchial arch. Human TWIST has been shown to be able to counteract the p53-dependent induction of cell death by regulating the tumour suppressing activity of p19ARF and MDM2 (Maestro et al., 1999). Twist haploinsufficiency also induces calvarial osteoblast apoptosis in vitro. A microarray study revealed increased transcription of apoptosis related genes, including MDM2 and BAX and revealed a possible mechanism of cell death induction by overexpression of TNFα (Yousfi et al., 2002). In our study, we observed a small increase in Mdm2 transcript level but no change in Bax transcription between the wild-type and mutant tissues. However, we did observe an increased level of Tnfα transcript in mutant tissues, indicating that the previously suggested role for Twist in regulating apoptosis in vitro may also be present in embryos.
Several lines of evidence point to a role for Twist as an inhibitor of differentiation. Overexpression of Twist in embryonic stem cells or myogenic cells impairs muscle differentiation and myotube formation (Hebrok et al., 1994; Rohwedel et al., 1995). Twist has been shown to repress muscle development by inhibiting trans-activation by Mef2 by blocking DNA binding of MyoD, and preventing the formation of active E protein-MyoD heterodimers (Spicer et al., 1996). Studies with human and mouse cell lines, again by examining the effect of overexpression, implicate Twist in preventing premature osteoblast differentiation. During mouse calvarial bone and suture development, it was found that Twist expression is confined to the osteogenic progenitor and is strongest in cells that are about to differentiate into bone cells (Rice et al., 2000). As these osteoprogenitor cells differentiate, the expression of Twist dramatically decreases until it disappears completely in the mature osteoblasts. Twist directly up-regulates periostin in undifferentiated preosteoblasts, which may be related to a role in maintaining the progenitor cell state (Oshima et al., 2002). Overexpression of TWIST in a human osteoblast-like cell line (HsaOS-2) was found to block osteoblast differentiation and caused these cells to de-differentiate back to a stem cell-like state (Lee et al., 1999).
Results of the present study on the Twist-/- embryo show that loss of Twist function does not affect the specification of the major lineages of the somite nor does it prevent their subsequent differentiation into muscle, cartilage, and bone. However, the absence of Twist does not enhance myogenic or chondrogenic differentiation of the mesodermal tissues of the mutant mouse embryo, as may be predicted if Twist acts primarily as an inhibitor of differentiation. Twist-deficient somites grew less prolifically and displayed a reduced capacity for differentiation than the wild-type somites after transplantation to the adult kidney. This finding may be best explained by the inability to maintain the population of viable progenitor to sustain differentiation cells in the absence of Twist function. Twist may play a complex role in a dose-dependent manner in maintaining the balance of proliferation, viability, and induction of differentiation of progenitor cells.
Twist Mutant Embryos
Mice carrying a targeted mutation of the Twist gene (Chen and Behringer, 1995) were maintained as heterozygotes on a C57BL6x129 background. These mice were interbred, and the resulting embryos were genotyped by PCR by using the primer sequences and parameters previously reported (Chen and Behringer, 1995; Soo et al., 2002). Twist+/+ and Twist+/- embryos show no detectable differences in phenotype except for hindlimb morphology (O'Rourke et al., 2002) and, therefore, were pooled as “normal” (+/+) embryos for this study.
Whole-Mount Immunostaining of Neurofilament
Embryos were dissected and fixed overnight in methanol:dimethyl sulfoxide (DSMO; 4:1) at 4°C. The embryos were bleached in methanol:DSMO:H2O2 (4:1:1) for 1 hr at room temperature to block endogenous peroxidases, then incubated in PBSMT (phosphate buffered saline, 2% skim milk, and 0.1% Triton X-100) twice for 1 hr each at 4°C, and overnight with 2H3 mouse monoclonal antibody (Hybridoma Bank) in PBSMT (1:100) at 4°C. After washing twice (1 hr each) in PBSMT at 4°C, followed by three washes at room temperature, embryos were incubated overnight with anti-mouse IgG biotinylated antibody in PBSMT (1:500) at 4°C. The reagents were prepared according to the instructions of the Vectastain ABC kit (Vector laboratories). The embryos were washed twice (1 hr each) in PBSMT at 4°C, followed by three washes at room temperature, and a final wash in PBS at room temperature for 30 min. The embryos were washed three times (10 min each) with PBS at room temperature. Diaminobenzidine (Sigma, D-6190) and the activator (Sigma, D6065) were added. Colour development was monitored under the dissecting microscope.
Whole-Mount In Situ Hybridisation
Whole-mount in situ hybridization was carried out essentially as described by (Wilkinson and Nieto, 1993). Digoxigenin-labelled riboprobes used for in situ hybridization were synthesized by using the AmpliScribe Kit (Epicentre). The following riboprobes were used in this study: Alx4 (Qu et al., 1999), Cited2 (Dunwoodie et al., 1998), Efnb2 (Liebl et al., 2003), En1 (Davis et al., 1991), Ephb2/3 (Krull et al., 1997), Meox1 (Candia et al., 1992), Myf5 (Ott et al., 1991), Paraxis (Burgess et al., 1995), Pax1 (Deutsch et al., 1988), Pax3 (Goulding et al., 1991), Sim1 (Ema et al., 1996), Sox10 (Pusch et al., 1998), Sox9 (Wright et al., 1995), Twist (Chen and Behringer, 1995), Uncx4.1 (Mansouri et al., 1997), Wnt6 (Parr et al., 1993). Whole-mount in situ hybridization was performed as previously described (Soo et al., 2002). After in situ hybridization, specimens were photographed and processed for wax histology.
Assessment of Histogenetic Potency by Teratoma Induction
A segment of the paraxial mesoderm containing four to five somites was dissected mechanically by using needles made from polished orthodontic wire from the cervical–upper thoracic region immediately rostral to the forelimb bud and the mid- to lower thoracic region caudal to the forelimb bud of E9.5 Twist+/+ or Twist+/- and Twist-/- embryos. The paraxial mesoderm was freed from the adjacent neural tube and the underlying endoderm. As much as possible of the adherent surface ectoderm was also removed by dissection with finely drawn glass needles. The explants were transplanted under the kidney capsule of adult ARC/s female mice (Tam, 1993), which were anaesthetized by ketamine (100 mg/kg) and xylazine (15 mg/kg). In each recipient mouse, somites of Twist+/+ or Twist+/- embryo and those of Twist-/- embryos were transplanted to contralateral kidneys. The tumour-like growth were harvested 14 days after transplantation, fixed in Bouin picro-formol solution (Gurr), and processed for wax histology. Serial 8-μm sections of the specimens were stained with hematoxylin and eosin. Every 12–15th section in the series was examined under the light and phase contrast microscope. The number of tumor samples that contained a specific type of tissue was also recorded during histological examination. The relative abundance of each type of tissues was estimated by the area they occupied in a histological section of tumor and scored on a quasiquantitative scale of + (<10% of total section area), ++ (10–50% of total section area), and +++ (>50% of total tissue area, see Fig. 4). Between 12 and 32 sections per specimen were scored for each specimen, depending on its size. An overall relative abundance of each tissue type for the tumors of the same genotype was estimated from the average of the quasiquantitative score of the specimens.
Cell proliferation was analyzed by BrdU (Sigma, B-9285) labelling. Pregnant mice were injected intraperitoneally with 0.01 ml/g body weight of BrdU solution. The embryos were collected 1 hr after injection and fixed in 4% paraformaldehyde overnight, washed in PBS and water, and embedded in paraffin wax. Seven-micrometer-thick sections were treated for 30 min with DAKO target retrieval solution in a 95°C water bath, washed in PBS and incubated for 30 min in 2 N HCl. Sections were washed in PBS and blocked by 10% fetal calf serum at room temperature for 30 min. Subsequently, the sections were then incubated with mouse monoclonal anti-BrdU antibody (Becton Dickinson) at room temperature for 1 hr in a dark humidified container, treated with rabbit anti-mouse AP-antibody for 45 min and washed in PBS and then NTMT. The sections were then incubated with 100 μl of nitroblue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate (NBT/BCIP) in 5 ml of NTMT overnight at 4°C. Sections were counterstained with nuclear fast red and mounted in Canada Balsam for microscopy. In selected sections of the sclerotome and the dermomyotome of thoracic and lumbar somites, and the mesenchyme of the presomitic mesoderm, the number of BrdU-labelling cells and the total number of cells within a delimited area were scored. The level of BrdU labelling of cell of sclerotome and dermomyotome of thoracic and lumbar somites and the presomitic mesoderm was determined in 16 wild-type or Twist+/- embryos and 14 Twist-/- embryos. For each embryo, an average of 220 cells (range, 153–374 cells) were counted in three to four different sections of the thoracic and lumbar somites and the presomitic mesoderm. The differential counts of BrdU-labelled and unlabelled cells were then used to compute the percentage of BrdU-labelled cell in the total population. The percentage scores were then tested for statistical differences by Mann–Whitney two-tailed test for unrelated scores (Table 3).
TUNEL Assay and Nile Blue Staining
The distribution of dead cells was revealed by Nile Blue staining. E8.5–E9.5 embryos were dissected out of the decidua and freed from the Reichert's membrane and amnion. They were cultured in DR50 medium (Sturm and Tam, 1993) containing a 1:400 aqueous dilution of 1.5% Nile Blue (Sigma), for 30 min in 5% CO2 at 37°C. The embryos were then washed twice briefly with PBS, and the staining pattern was examined under a dissecting microscope.
E9.0–E10.5 embryos were assessed for whole-mount cell death by Nile Blue staining as previously described (Trainor and Krumlauf, 2002). Briefly, embryos were cultured in DR50 (containing a 1:400 dilution of 1.5% Nile Blue in water) for 20–30 min in the 5% CO2 incubator, then washed with Dulbecco's PBS to see apoptotic cells stained intensely blue. TUNEL staining was performed according to the manufacturer's manual (Roche). Paraffin sections were incubated for 8 min in freshly prepared 0.1% Triton X-100, 0.1% sodium citrate to permeabilize the tissue before the TUNEL reaction. Apoptosis in the embryo was assessed by the TUNEL procedure as previously described (O'Rourke et al., 2002). In brief, paraffin sections of embryonic specimens were de-waxed and re-hydrated then incubated in freshly prepared 0.1% Triton X-100, 0.1% sodium citrate for 8 min at room temperature. The sections were rinsed twice with PBS and incubated for 60 min with the TUNEL reaction reagents (Roche). The sections were rinsed three times with PBS to reduce background, and the slides were mounted with anti-fade mounting media (2.33% 1,4-diazabicyclo[2.2.2]octane (DABCO, Sigma, catalog no. D2522] in 90% glycerol, 20 mM Tris-HCl, pH 8.0). The fluorescein isothiocyanate–labelled cells were examined under an Olympus BX50WI microscope with excitation at 488 nm and visualized through a green filter (BP 515–560 nm).
RNA was extracted from pooled first branchial arch tissues by using a small scale RNA isolation kit (Qiagen). Reverse transcription was carried out on 1 μg of RNA extracted from wild-type and mutant tissues by using the Advantage RT for PCR kit (Clontech). Primers for PCR were designed to span multiple exons as follows: P21F, atgtccgacctgttccgcaca; P21R, agatggggaagaggcctcctga; P53, gcatgaaccgccgacctatcctt; P53R, gtgagatttcattgtaggtgccaggg; Mdm2F, ccgcaggtccctgtcctttgat; Mdm2R, tcctcagcacatggctctttagcatc; BaxF, cgtggttgcccttcttctactttgcta; BaxR, ccaccattcccacccctcccaata; TnfαF, tgagcacagaaagcatgatccgc; TnfαR, atgagatagcaaatcggctgacggtg; GapdF, ccacagtccatgccatctc; GapdR, tccaccaccctgttgctgta.
Cycling conditions were 94°C for 45 sec, 55°C(p21), 60°C (p53, Mdm2, Bax, Gapd), or 65°C (Tnfα) for 45 sec, and 72°C for 1 min for 25 cycles. Amplification was performed by using 1 μl of the undiluted template and 1 μl of a 1 in 10 dilution of template. Experiments were performed multiple times on independently reverse-transcribed templates to confirm the reproducibility of the results.
We thank Kenneth Soo, Poh-Lynn Khoo, and Kirsten Steiner for technical assistance, the BioServices Unit for mouse husbandry, and Peter Rowe for comments on the manuscript. Riboprobes were kindly provided by Alexandra Joyner (En1), Peter Koopman (Sox9, Sox10), Margaret Buckingham (Myf5), Janet Rossant (Meox1), Frits Meijlink (Alx4), Peter Gruss (Pax1 and -3), Chen-Ming Fan (Sim1), Andrew McMahon (Wnt6), Sally Dunwoodie (Cited2, Uncx4.1), Maria de Bellard (Ephb2/b3, Efnb2), Eric Olson (Scleraxis, Dermo1), and Michael Blanar (Paraxis). The neurofilament antibody developed by T. Jessell and J. Dodd was obtained from the Developmental Studies Hybridoma Bank maintained by the University of Iowa, Department of Biological Sciences, Iowa City, IA 52242. M.O. is a Monbushu Fellow of Japan, D.A.F.L. is the Kimberly-Clark Research Fellow, M.P.O'R. is a NHMRC Dental Postgraduate Scholar, and P.P.L.T. is a NHMRC Senior Principal Research Fellow.