Facial development involves a complex and highly integrated pathway of events that requires the coordinated growth and fusion of many specialised tissues. The cues that orchestrate these processes arise from signalling events directed by many different proteins together with environmental factors. Alterations to this complex series of events, resulting from aberrant or insufficient gene expression, may result in a wide variety of birth defects. Treacher Collins syndrome (TCS; OMIM number 154500) is one example of a disorder of facial development. TCS is an autosomal dominant condition that results from mutations in the TCOF1 gene, which encodes the nucleolar phosphoprotein Treacle (The Treacher Collins Syndrome Collaborative Group, 1996; Gladwin et al., 1996; Edwards et al., 1997; Wise et al., 1997, Splendore et al., 2000, 2002). The mutations observed in TCS are predominantly family-specific, and the vast majority result in the introduction of a premature termination codon, suggesting that the developmental anomalies result from haploinsufficiency of TCOF1. The most common clinical signs of TCS include hypoplasia of the mandible and maxilla; downward slanting of the palpebral fissures with deep notches in the lower eyelids; abnormalities of the external ears and middle ear ossicles, resulting in bilateral conductive hearing loss; and cleft palate (Phelps et al., 1981). More than 60% of TCS cases have no previous family history and arise as the result of a de novo mutation (Gorlin et al., 1990). Penetrance of the genetic mutations underlying TCS is thought to be very high; however, extreme inter- and intrafamilial variation in the severity of the phenotype is a striking feature of the condition (Dixon et al., 1994; Marres et al., 1995). At one end of the clinical spectrum, severe cases have resulted in perinatal death due to a compromised airway (Edwards et al., 1996). Conversely, some individuals are so mildly affected that it can be extremely difficult to establish an unequivocal clinical diagnosis. Indeed, it is not unusual for mildly affected TCS patients to be diagnosed retrospectively only after the birth of a more severely affected child. This pattern suggests that nonpenetrance must occur more frequently than has been documented and that as yet undetermined factors, potentially including the effect of the genetic background, environmental factors, and stochastic events, contribute to the clinical variation observed in TCS patients.
To investigate the developmental basis of TCS, our laboratory previously generated chimeric mice that carry a germline mutation of one allele of the murine orthologue of TCOF1. Breeding of male chimeras with female C57BL/6 mice generated heterozygous offspring that exhibited several features reminiscent of the human disorder, including hypoplasia of the mandible and abnormalities of the maxilla (Dixon et al., 2000). However, additional features of the phenotype, not observed in TCS patients, included absence of the eyes and nasal passages, and neural tube defects resulting in exencephaly. We hypothesised that the presence of these features may be due either to species-specific differences in the function of the human and murine genes or that the severity of the Tcof1 phenotype depends on the presence of modifying loci elsewhere in the genome. The results of this study support the latter hypothesis and provide a framework for the identification of genetic factors that influence facial development and phenotypic variability.
Penetrance and Severity of Facial Defects Varies Widely in Different Genetic Backgrounds
Tcof1 heterozygous (Tcof1+/-) embryos were characterised by gross morphologic, histologic, and skeletal analyses at a variety of ages from embryonic day (E) 8.5 to term. Each different inbred strain generated Tcof1+/- embryos that exhibited a highly reproducible, strain-dependent phenotype. In particular, we observed that the penetrance and severity of facial defects was dependent upon the genetic background (Fig. 1; Table 1). The lethality of Tcof1+/- mice derived from the C57BL/6 strain has been documented previously (Dixon et al., 2000). In addition, Tcof1+/- mice derived from CBA/Ca and C3H/HEN females (referred to as CBA and C3H, respectively) also exhibited a lethal phenotype with perinatal death resulting from severe abnormalities of the nasal complex and neural tube defects that resulted in exencephaly. In contrast, the vast majority of Tcof1+/- mice from DBA/1 and BALB/c backgrounds were viable and displayed exencephaly very rarely. For clarity and comparison, a brief summary of the original C57BL/6 Tcof1+/- phenotype is also presented here.
Table 1. Major Abnormalities Exhibited by Tcof1 Heterozygous Mice Derived From Different Inbred Genetic Backgrounds as Indicateda
E, embryonic day.
Gross morphological defects
Exencephaly (E9 or older)
Developmental delay (E8 or older)
Hypoplasia of mandible/maxilla (E9 or older)
6/6 exencephalic embryos
Eye abnormalities (E10 or older)
Unilateral anophthalmia/microphthalmia in 33/33 embryos
Anophthalmia observed in 102/102 embryos
Exophthalmia/small orbit in 29/29 embryos
Unilateral microphthalmia/anophthalmia in 2/6 exencephalic embryos
Digit abnormalities (E13 or older)
Defects observed by skeletal analysis and histology
Abnormalities of the nasal complex
Midline nasal “spear” present in 4/4 embryos
Midline nasal “spear” present in 14/14 embryos
Asymmetry of the nasal passages in 6/6 exencephalic embryos only
Flexion of spine/abnormal posture
Fusions of the cervical or thoracic vertebrae
Delayed ossification in long bones
Defects observed by in situ hybridisation/immunohistochemistry
Abnormal/hypoplastic cranial and dorsal root ganglia
Abnormal neural crest migration
C57BL/6 and CBA Tcof1 Heterozygotes
Analysis of 33 CBA Tcof1+/- embryos indicated that this strain exhibited a phenotype with many similar features to that of the original C57BL/6 strain. Gross morphologic examination indicated that both C57BL/6- and CBA-derived embryos exhibited a developmental delay of approximately 0.5–1.0 gestational day (Fig. 2E,F). Exencephaly of the forebrain and midbrain was present in all C57BL/6 Tcof1+/- embryos and the vast majority of CBA Tcof1+/- embryos. Ocular development was comparable in these two strains: anophthalmia was observed in all C57BL/6 Tcof1+/- embryos, and either anophthalmia or microphthalmia was present in CBA Tcof1+/- embryos. Skeletal analyses, however, revealed a distinct difference between the two strains. While embryos derived from both C57BL/6 and CBA backgrounds exhibited many common abnormalities of the craniofacial and axial skeleton, the CBA Tcof1+/- mice exhibited a profound bone phenotype with delayed ossification of all the long bones and absence of the primary centres of ossification within the digits (data not shown). This aspect of the phenotype was not present in C57BL/6 Tcof1+/- embryos. Embryos from both strains displayed absence of the bones comprising the vault of the skull, including the frontal, parietal, interparietal, and supraoccipital bones (Fig. 3E,F), and abnormalities of the skull base. The development of the nasal complex was poor in Tcof1+/- embryos from both strains; the nasal septum was invariably replaced by a central cartilaginous mass, with no evidence of nasal passages or olfactory sensory structures (Fig. 3H,I). Other skeletal abnormalities included fusions of the inter-costal cartilages and ribs; severe abnormalities and fusions of the cervical and thoracic vertebrae and digit anomalies that included polydactyly of the first hind-digit. Histologic analysis revealed that both strains displayed marked disorganisation of the brain and facial complex, including a narrow and highly arched oral cavity, frequently with a cleft of the secondary palate and abnormally positioned tooth germs (Fig. 2K,L).
C3H Tcof1 Heterozygotes
C3H Tcof1+/- embryos were also predominantly exencephalic. Of the 29 Tcof1+/- embryos examined, only three had fully formed cranial vaults, the remaining 26 exhibited a characteristic exencephaly of the forebrain and midbrain with extremely well-formed, distinct telencephalic vesicles that protruded through the open neural tube. The overall body size of C3H Tcof1+/- mice was comparable to that of their wild-type littermates (Fig. 2D). Ocular development was present in this strain, although this was not visible externally due to the presence of the overlying neural tissue. Histologic sections revealed that the eyes were well-developed, but compressed and abnormally shaped, with little evidence of eyelid development (Fig. 2J). Skeletal analysis indicated that, in contrast to the C57BL/6 and CBA Tcof1+/- embryos, the nasal passages of C3H Tcof1+/- embryos were well-modelled (Fig. 3D,G). In addition, all of the exencephalic heterozygotes exhibited absence of the bones comprising the vault of the skull and abnormalities of the skull base. No other skeletal abnormalities were observed in Tcof1+/- mice derived from this strain.
DBA/1 Tcof1 Heterozygotes
Analysis of 48 DBA/1 Tcof1+/- embryos indicated that this strain exhibited only a moderate craniofacial phenotype. Six DBA/1 Tcof1+/- embryos displayed exencephaly, predominantly of the midbrain, and, to a lesser extent, the forebrain. Mandibular hypoplasia was also observed in these exencephalic embryos, together with occasional abnormal ocular development in the form of microphthalmia and unilateral anophthalmia (two of six exencephalic embryos). As observed in all other exencephalic strains, the bones comprising the skull vault were absent and the majority of bones in the skull base were abnormal. The nasal passages and nasal bone were poorly developed and asymmetrical in these exencephalic DBA/1 Tcof1+/- embryos. In contrast, however, the majority of the DBA/1 Tcof1+/- embryos appeared normal and were phenotypically indistinguishable from their wild-type littermates at a gross morphologic level (Figs. 2B, 3B). Minor abnormalities of the axial skeleton were occasionally present in both exencephalic and nonexencephalic embryos. These included fusions of the ribs (present in four embryos) and abnormalities of the digits (present in one embryo). Nonexencephalic DBA/1 Tcof1+/- mice are viable, fertile, and have a normal lifespan.
BALB/c Tcof1 Heterozygotes
Seventy-six BALB/c Tcof1+/- embryos were examined, five of which exhibited exencephaly. Skeletal analyses indicated that these embryos displayed absence of the bones of the vault of the skull, as well as abnormalities of the skull base. No other skeletal abnormalities were observed. The remaining 71 Tcof1+/- embryos were indistinguishable from their wild-type littermates by skeletal, histologic, and gross morphologic analyses (Figs. 2C, 3C). BALB/c Tcof1+/- mice are viable, fertile, and have a normal lifespan.
Neural Crest Cell Anomalies Are Present in the C57BL/6 and CBA Tcof1+/- Embryos
The structures affected in the Tcof1+/- embryos are derived in part from the neural crest cells, which have numerous derivatives, including pigment cells, autonomic and sensory ganglia, and most of the facial skeleton (Le Douarin and Kalcheim, 1999). To examine the neural crest cells in the different inbred strains of Tcof1+/- embryos, the expression patterns of Sox10, which is expressed in migrating neural crest cells and their neurogenic derivatives, were analysed in E8.5 to E10.5 Tcof1+/- embryos by using whole-mount in situ hybridisation. These methods indicated that the expression of Sox10 was normal at all stages examined in embryos derived from the DBA/1, BALB/c, and C3H strains (Fig. 4B–D,H–J). In contrast, there was very little expression of Sox10 in E8.5 C57BL/6 and CBA Tcof1+/- embryos. This finding was due to the developmental delay observed in these two strains and indicated that neural crest cell migration had not commenced at this developmental stage (data not shown). We therefore examined E9.5 C57BL/6 and CBA Tcof1+/- embryos, which were morphologically similar to wild-type E8.5 embryos. These embryos exhibited Sox10 expression, indicating that the neural crest cells had commenced migration (Fig. 4E,F). However, analysis of Sox10 expression in E10.5 C57BL/6 and CBA Tcof1+/- embryos revealed that the domains corresponding to the cranial ganglia were markedly smaller than in wild-type embryos of a similar morphologic stage. In addition, the dorsal root ganglia were extremely disorganised and exhibited spindly nerve fibres (Fig. 4K,L). Consistent with this observation, examination of the staining patterns obtained with the anti-neurofilament marker 2H3 indicated that all of the cranial and dorsal root ganglia were severely underdeveloped in C57BL/6 and CBA Tcof1+/- embryos (Fig. 4Q,R). In particular, there was complete absence of the ophthalmic branch of the trigeminal ganglion and the petrosal ganglion. These data indicate that the levels of neural crest cell migration in all strains of Tcof1+/- embryos were initially comparable to wild-type controls. Subsequently, however, in Tcof1+/- embryos derived from C57BL/6 and CBA females, the population of cells that would normally contribute to the cranial and dorsal root ganglia were markedly depleted, suggesting severe neural crest cell loss.
In the current study, we have demonstrated that the severity and penetrance of facial defects in Tcof1 heterozygous mice are dependent upon the genetic background. Lethal phenotypes, which included profound facial defects, were observed in CBA, C57BL/6, and C3H Tcof1+/- embryos. In contrast, the vast majority of DBA/1 and BALB/c Tcof1+/- embryos appeared to be indistinguishable from their wild-type littermates and were viable. These widely varying phenotypes must result in part from complex interactions specified by the different genetic backgrounds on which the mutation resides. One particular aspect of the phenotype, the difference in susceptibility to neural tube defects, may be related to differences in the timing and position of neural tube closure at the forebrain–midbrain boundary, which has been found to be polymorphic in different inbred strains (Juriloff et al., 1991). In this regard, the variation in Tcof1 heterozygous mice is similar, but not identical, to that observed in jumonji mice, which exhibit a phenotype that includes neural tube defects in a C3H/HeJ background, but not in BALB/cA, C57BL/6J, or DBA/2J backgrounds (Motoyama et al., 1997, Takeuchi et al., 1999).
To date, several reports have documented extreme inter- and intrafamilial variation in the severity of the clinical presentation of TCS patients (Dixon et al., 1994; Marres et al., 1995). Due to the small numbers of children within a sibship, however, it is extremely hard to determine the contribution of genetic factors to phenotypic variability in this facial disorder. The availability of an appropriate mouse model, and in particular, the viability of the BALB/c and DBA/1 Tcof1+/- mice, now provide the resources to map and ultimately identify factors that may contribute to phenotypic variability in TCS by interacting with, or modifying the function of, the protein. Current strategies to identify such loci involve generating large numbers of heterozygous animals in an F2 backcross to provide sufficient numbers of informative recombination events near the loci of interest (Hide et al., 2002). Alternatively, the genetically heterogeneous stocks (HS) provide an exciting resource that can be used for fine mapping and identification of modifier loci in mice (Mott and Flint, 2002). The HS lines are a series of breeder pairs derived from founder inbred strains that have been outbred to greater than 60 generations, resulting in a high level of recombination, with the average distance between recombinants of 1.7 cM (Mott and Flint, 2002). As the chromosomes of each HS animal, are a complex genetic mosaic of each founder haplotype, it is impossible to predict what phenotypes would result from breeding of the HS males with congenic Tcof1+/- females. However, given the highly variable and wide-ranging phenotypes identified in the current study, breeding of the HS and Tcof1+/- lines would likely result in novel and uncharacterised phenotypes. It is feasible, therefore, that certain aspects of the resulting phenotypes would be amenable to further study and would provide a means of identifying genetic factors that modify the phenotypic severity. In the longer term, the identification of such factors will help to dissect the molecular mechanisms governing facial development and may help us to understand the wide clinical variation observed in TCS families. Nevertheless, although the phenotypes observed in the various inbred strains in the current study are largely reproducible, we have observed some intralitter variation. This finding suggests that factors additional to the genetic background, such as environmental factors or stochastic events, may also contribute to the phenotype.
The identification of loci that modify the severity of the Tcof1 phenotype may also help to dissect the function of this nucleolar protein and to determine why it is essential to the normal development of the facial complex. Searches of protein databases have indicated that Treacle is a very simple protein containing few motifs of known function; it consists of three distinct domains, unique amino and carboxy termini and a characteristic central repeat domain (Dixon et al., 1997; Wise et al., 1997). The carboxy terminus of Treacle contains multiple nuclear localisation signals, which have been shown to drive nuclear import of the protein (Marsh et al., 1998; Winokur and Shiang, 1998). Biochemical assays have determined that Treacle is extremely highly phosphorylated, and immunofluorescence studies have localised anti-Treacle antibodies to the dense fibrillar component of the nucleolus (Isaac et al., 2000). More recently, Treacle has been identified as a constituent of human Nop56p-associated preribosomal ribonucleoprotein (pre-rRNPs) complexes by proteomic analysis (Hayano et al., 2003). Nop56p is a component of the box C/D small RNP complexes that direct 2′O-methylation of pre-rRNA during its maturation (Gautier et al., 1997). Analysis of Nop56p-associated pre-rRNPs indicated that Treacle remained associated with hNop56p even after RNase treatment, suggesting that it binds directly to hNop56p in a manner that is independent of rRNA integrity (Hayano et al., 2003). These data indicate that Treacle is contained within an RNP complex in the nucleolus and may be involved in certain stages of ribosome biogenesis. Analysis of the proteomic profiles of the hNop56p- and Treacle-associated pre-rRNPs obtained from the heterozygous mice may help to further elucidate the biochemical role of Treacle.
TCS has long been considered a neurocristopathy of an undetermined basis (Poswillo, 1975). Sox10, which is expressed in migrating neural crest cells and their neurogenic derivatives (Southard-Smith et al., 1998), was observed in all of the Tcof1+/- embryos, regardless of their genetic background, in regions corresponding to the neural crest cells migrating from the prefusion neural folds. At later stages, however, the size and appearance of the Sox10 expression domains in the cranial and dorsal root ganglia, together with data derived using the anti-neurofilament marker 2H3, strongly suggests that the population of neural crest cells that normally migrate to these ganglia was severely depleted in CBA and C57BL/6 Tcof1+/- embryos. Currently, it is impossible to say whether these defects result from altered migration, differentiation or indeed survival of the neural crest cells in Tcof1+/- embryos. The viability of the BALB/c and the DBA/1 Tcof1+/- lines now provide a valuable resource for comparative studies of neural crest cell migration and a means with which to address the developmental basis of TCS. In this regard, focal injections of the vital dye DiI into the neural folds of Tcof1+/- embryos, in combination with in vitro whole embryo culture methods, as described by Trainor and Tam (1995), will permit us to characterise neural crest cell migration in more detail.
In summary, we have demonstrated that different inbred strains of mice, all of which carry an identical mutation in the murine orthologue of the gene underlying TCS, exhibited markedly variable strain-dependent phenotypes that ranged from lethal in the perinatal period to apparently normal and viable. Our results indicate that the genetic background on which the mutation resides is a major determinant of phenotypic variability. The viable Tcof1 heterozygous mice now provide a valuable resource that can be used to identify and ultimately isolate genetic factors that contribute to phenotypic variability in facial development, both in mice and in humans.
Mouse Breeding and Genotyping
The Tcof1 mutation was originally generated in 129-derived R1 embryonic stem (ES) cells by replacement of exon 1 with a neomycin resistance cassette. Correctly targeted ES cell lines were injected into C57BL/6 blastocysts to generate 129/C57BL/6 Tcof1 heterozygous (Tcof1+/-) chimeric male mice (Dixon et al., 2000). These chimeric males were bred with female mice from C57BL/6, BALB/c, DBA/1, CBA, and C3H backgrounds to generate Tcof1+/- offspring. Embryos were obtained by Caesarean section after cervical dislocation of the mother. In addition, some offspring were allowed to develop to term so that postnatal viability could be determined. Genotyping was performed on DNA extracted from the yolk sac, placenta, or embryonic tissue using previously described methods (Dixon et al., 2000).
Skeletal Analysis, Histology, Whole-Mount In Situ Hybridisation, and Whole-Mount Immunohistochemistry
For analysis of the craniofacial skeleton, E16 to term embryos were killed, eviscerated, and fixed in 99% ethanol for 24 hr. Subsequently, the samples were processed as described previously (Wallin et al., 1994). For histology, samples from E10.5 onward were fixed overnight in Bouin's fixative, equilibrated in 50% ethanol, and photographed. Subsequently, samples were embedded in paraffin wax and 5- to 7-μm sections obtained. Sections were stained by using hematoxylin, eosin, and Alcian blue using established protocols. Whole-mount in situ hybridisation was performed as described previously (Kondo et al., 2002), with the modification that all hybridisations were performed at 60°C for 3–4 days. Whole-mount immunohistochemistry was performed according to published methods (Wallin et al., 1994).
We thank Cord Brakebusch and Reinhard Fässler for their help and advice during the generation of the Tcof1 mutant mice. Paul Trainor kindly donated probes for whole-mount in situ hybridisation studies. The monoclonal antibody 2H3, developed by T.M. Jessell and J. Dodd, was obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD and maintained by the University of Iowa, Department of Biological Sciences. J.D. is an MRC Research Training Fellow.