Mouse Models of Rare Craniofacial Disorders

Abstract

A rare disease is defined as a condition that affects less than 1 in 2000 individuals. Currently more than 7000 rare diseases have been described, and most are thought to be of genetic origin. Rare diseases primarily affect children, and congenital craniofacial syndromes and disorders constitute a significant proportion of rare diseases, with over 700 having been described to date. Modelling craniofacial disorders in animal models has been instrumental in uncovering the etiology and pathogenesis of numerous conditions and in some cases has even led to potential therapeutic avenues for their prevention. In this chapter we focus primarily on two general classes of rare disorder, ribosompothies and ciliopathies, and the surprising finding that the disruption of fundamental, global processes can result in tissue specific craniofacial defects. In addition we discuss recent advances in understanding the pathogenesis of an extremely rare and specific craniofacial condition known as syngnathia, based on the first mouse models for this condition. Approximately 1% of all babies are born with a minor or major developmental anomaly, and individuals suffering from rare diseases deserve the same quality of treatment and care and attention to their disease as other patients.

Introduction

Craniofacial development is an intricate process that requires several distinct signaling pathways and elaborate morphogenetic movements encompassing each of the ectoderm, mesoderm, and endoderm germ layers. In addition, a transient, migratory and multipotent cell population called neural crest cells (NCC) that can be considered the “fourth” germ layer also contributes significantly to craniofacial development. In fact, NCC will give rise to the majority of the facial mesenchyme, and subsequently the craniofacial skeleton.

NCC arise during early embryonic development at the border of the neural and non-neural ectoderm. Although derived primarily from the neuroepithelium, the non-neural or surface ectoderm is also able to generate small numbers of NCC. After induction, NCC undergo an epithelial-to-mesenchymal transition, delaminate from the neural plate or tube, and migrate throughout nearly the entire embryo. Upon reaching their destinations, NCC differentiate into a broad range of cell and tissue types, including cranial and spinal ganglia, enteric neurons, melanocytes, endocrine and paraendocrine cells, connective tissue, and craniofacial bone and cartilage (Bronner & LeDouarin, 2012; Santagati & Rijli, 2003). NCC are generally divided into two primary groups, cranial and trunk NCC, based on the axial levels from which they are derived in the embryo. The major difference between these two populations is the ability of the cranial NCC to differentiate into bone, cartilage, and connective tissue, whereas trunk neural crest cells generally lack this ability.

Cranial NCC arise from the diencephalon, mesencephalon and rhombencephalon and generate the majority of the bone and cartilage of the face and skull (Fig 1). The most anterior cranial NCC populate the frontonasal region, which will give rise to the frontal and nasal bones, whereas posterior cranial NCC populate the pharyngeal arches (PAs), which will give rise to the upper and lower jaw, middle ear and structures in the neck. The PAs comprise a reiterated series of outgrowths on the lateral side of the head that are highly conserved throughout vertebrate evolution (Frisdal & Trainor, 2014). Mammals exhibit five pairs of PAs that develop rostrocaudally in a serial manner. Each PA is composed externally of ectoderm and lined internally by endoderm, enveloping a mesenchymal core composed of NC and mesoderm. The juxtaposition of the endoderm and ectoderm forms an internal cleft and an external pouch, respectively, that separates each arch from its neighbor. Complex reciprocal signaling interactions between the ectoderm, endoderm, mesoderm and NC ensures proper development of the craniofacial skeleton (Frisdal & Trainor, 2014).

Figure 1. Contribution of neural crest cells to the craniofacial skeleton.

The cranial neural crest gives rise to the majority of the craniofacial skeleton. All the colored bones in the mouse (left) and human (right) skull are derived from the neural crest.

Considering the multiple and distinct stages of NCC development and their contribution to the craniofacial skeleton, it is not surprising that many of the congenital craniofacial abnormalities, which constitute 1 out of 3 birth defects, are due to defects in NCC development. Such abnormalities are collectively referred to as neurocristopathies. Perturbation of distinct stages of NCC development results in different craniofacial deficiencies and syndromes. For example, Treacher Collins syndrome (TCS) arises due to defects in the generation and proliferation of NCC, whereas syngnathia is associated with defective differentiation of NCC.

Here we describe rare craniofacial neurocristopathies and efforts to understand their genetic and cellular basis as well as the pathogenesis of these disorders using mouse models. Interestingly, many of these tissue-specific disorders are caused by mutations in genes involved in global cellular processes such as ribosome and cilium biogenesis. Termed ribosomopathies and ciliopathies respectively, these congenital disorders have changed the way we think about the composition and function of ribosomes and cilia, as these global processes have surprisingly tissue-specific roles and effects.

Ribosomopathies

The ribosome is a large ribonucleoprotein complex responsible for translating mRNAs into proteins in all living organisms. The mature eukaryotic 80S ribosome is composed of two subunits, the small (40S) and the large (60S) subunit, each of which consists of ribosomal proteins (RP) and ribosomal RNAs (rRNA). The 40S ribosome, comprised of the 18S rRNA and 33 RPs, decodes the mRNA, whereas the 60S ribosome, comprised of the 5S, 5.8S and 28S rRNAs together with 46 RPs, catalyzes the formation of peptide bonds (Lafontaine & Tollervey, 2001). The process of generating ribosomes, termed ribosome biogenesis, is elaborate and utilizes all three RNA Polymerases (Pol I, II, III). The four rRNAs are transcribed by RNA Pol I (18S, 5.8S, 28S) and Pol III (5S), whereas about 80 RPs, accessory proteins and approximately 70 small nucleolar RNAs (snoRNA) are transcribed by Pol II. Additionally, more than 200 proteins are involved in the production, processing and assembling of the rRNA alone (Kressler, Hurt, & Bassler, 2010; Lafontaine & Tollervey, 2001).

Transcription of rDNA genes is the first step in ribosome biogenesis and is considered to be a rate-limiting step (Fig 2). Transcription of the 47S precursor rRNA, which includes the 18S, 5.8S and 28S, by Pol I takes place in the nucleolus whereas transcription of the 5S rRNA by Pol III happens in the nucleus. Altogether rDNA transcription accounts for about 60% of overall cellular transcription (Laferte et al., 2006). Subsequently the 47S rRNA is processed by covalent modifications, bound by RPs and cleaved several times to produce the individual 5.8S, 18S, and 28S rRNAs, which will then - along with the 5S rRNA - assemble into their respective ribosomal subunits together with the appropriate RPs. These subunits are then exported from the nucleus into the cytoplasm to form the mature and active 80S ribosome.

Figure 2. The ribosome biogenesis pathway and ribosomopathies.

Schematic figure summarizing the major stages of ribosome biogenesis. The genes and the corresponding ribosomopathies are depicted. Note that disruption of ribosome biogenesis at different stages leads to different syndromes.

Given the importance of the ribosome in protein synthesis, which is a basal, global process vital for cell survival and proliferation, it is remarkable that mutations in genes disrupting the process of ribosome biogenesis result in congenital disorders with tissue-specific defects. Studies of these unique ribosomopathy conditions in animal systems, and specifically in mammalian models such as mice, have been instrumental in unraveling tissue-specific roles for ribosomal genes during embryogenesis.

Treacher Collins syndrome

Treacher Collins syndrome (TCS, OMIM 154500) is a congenital birth defect also known as mandibulofacial dysostosis or Franceschetti–Zwahlen–Klein syndrome. It occurs with an incidence of about 1 in 50,000 human live births and was described by George Andreas Berry in 1889, followed in more detail by the ophthalmologist Treacher Collins in 1900 (Kadakia, Helman, Badhey, Saman, & Ducic, 2014; Sakai & Trainor, 2009). Later in 1940, ophthalmologists Franceschetti and Klein further delineated and termed the condition mandibulofacial dysostosis (Franceschetti & Klein, 1949).

Clinical Features

TCS originates during the fifth to eighth week of gestation. It affects the proper formation of the facial prominences and first and second pharyngeal arches and consequently their derivatives in the head and neck (Fig 3B) (Kadakia et al., 2014). The malformations and the degree of severity are highly variable in TCS patients, however a characteristic feature is general hypoplasia of the facial bones: mainly the maxilla, the mandible and the zygomatic complex. Other major clinical features include downward slanting of the palpebral fissures with colobomas of the lower eyelids and absence of the medial third of the lower eyelashes due to clefting of the inferolateral orbital, giving TCS patients characteristic downslanting eyes (Trainor & Andrews, 2013). Atresia of the external auditory canals and deformities of the middle ear including anomalies in the auditory ossicles, together with frequent fusion of the rudimentary malleus and incus, partial absence of the stapes and oval window, or even complete absence of the middle ear and epitympanic space can lead to conductive hearing loss (Trainor, Dixon, & Dixon, 2009). Moreover many patients exhibit cleft or high arch palate, dental malocclusion with anterior open bite due to retrognathia, and airway dysfunction such as tracheostoma or choanal stenosis/atresia. Together these anomalies lead to speech and language difficulties, visual impairment, breathing difficulty, and obstructive sleep apnea (Kadakia et al., 2014).

Figure 3. Examples of rare craniofacial disorders.

Images of patients with (A) Treacher Collins syndrome, (B) Oral-facial-digital syndrome type 1 [adapted from (Gonzalez, Castro, Nieto, & Bouzan, 2014)], and (C) Syngnathia syngnathia. (D) A 3D CT scan image showing a bilateral fusion of the mandible to the maxilla, an underlying feature of syngnathia.

Genetics

Mutations in three genes have been identified in association with TCS: TCOF1, POLR1C, and POLR1D. About 60% of all mutations are thought to arise de novo. The remainder is inherited predominantly in an autosomal dominant fashion, but recessive inheritance has also been recorded. Mutations in TCOF1 are autosomal dominant and account for about 80% of patients with TCS. Mutations in POLR1C are autosomal recessive and mutations in POLR1D are predominantly autosomal dominant but can also be recessive. (Dauwerse et al., 2011; group, 1996; Schaefer et al., 2014; Trainor et al., 2009). Mutations in POLR1C and POLR1D account for about 2% of the patients sequenced to date, which suggests that mutations in additional genes may also be causative for TCS.

Interestingly, TCOF1, POLR1C and POLR1D are part of the ribosomal DNA (rDNA) transcription machinery (Fig 2). TCOF1 encodes the nucleolar phosphoprotein Treacle that physically interacts with upstream binding factor (UBF), a transcription factor required for the initiation of transcription by RNA Polymerase I (PolI) (Russell & Zomerdijk, 2006; Valdez, Henning, So, Dixon, & Dixon, 2004). Not much is known about POLR1C and POLR1D other than that they are subunits of Pol I and Pol III, which are responsible for transcription for all the rRNAs and that they interact strongly with each other and are essential for Pol I and Pol III assembly (Larkin & Guilfoyle, 1997; Mann, Buhler, Treich, & Sentenac, 1987; Yao, Yamamoto, Nishi, Nogi, & Muramatsu, 1996). Mutations in TCOF1, POLR1C and POLR1D together with their roles in ribosome biogenesis are therefore consistent with the description of TCS as a ribosomopathy.

Mouse Models and Mechanism

The Tcof1^+/− mouse is the first and only TCS mouse model to have been described and it has revealed much about the cellular function of Tcof1 and the pathogenesis of TCS (Dixon et al., 2006; Jones et al., 2008). Tcof1 haploinsufficient mice mimic the craniofacial deformities observed in TCS patients, including general hypoplasia of the craniofacial skeleton, with misshapen and hypoplastic maxilla and palatine bones, occasional cleft palate and choanal atresia (Fig 4B) (Dixon et al., 2006). At the cellular level, Treacle, the protein encoded by Tcof1, is required for proper ribosome biogenesis. Haploinsufficiency of Tcof1 causes a reduction in ribosome production, which results in nucleolar stress. Nucleolar stress then leads to stabilization of p53 in the neuroepithelium, the birthplace of NCC. Consequently, elevated cell death is observed in these NCC precursors, resulting in fewer migrating and therefore fewer differentiating NCC (Dixon et al., 2006). A deficiency in the number of NCC is the primary cause of craniofacial deformities observed in the Tcof1^+/− mouse model of TCS and by extrapolation in human patients.

Figure 4. Mouse models of rare craniofacial disorders.

Alizarin red (bone) and alcian blue (cartilage) skeletal staining. (A) Wild-type E18.5 mouse embryo; (B) Tcof1^+/− littermate that resembles Treacher Collins syndrome (TCS). (C) Tcof1^+/−; p53^+/− embryo with normal craniofacial morphology [adapted from (Jones et al., 2008)]. (D) The mandible of a wild-type E18.5 mouse embryo. (E) Mandible of an E18.5 Foxc1^−/− embryo fused to the maxilla, which is consistent with syngnathia.

Spatiotemporal characterization of Tcof1 expression in the developing embryo revealed broad yet dynamic activity in the neuroepithelium and in migrating and differentiating NCC. This pattern coincides with the onset of NCC specification and persists throughout their migration and differentiation consistent with the phenotypes observed in Tcof1 haploinsufficient embryos (Dixon et al., 2006). Furthermore, analysis of Treacle in a cell culture system revealed colocalization with UBF, suggesting its possible function within the rDNA transcription machinery (Valdez et al., 2004). Indeed, upon downregulation of Tcof1 in vitro, rDNA transcription is greatly reduced, consistent with the in vivo reduction in ribosome biogenesis observed in Tcof1^+/− embryos (Valdez et al., 2004). Treacle has also been associated with the early stages of pre-RNA processing. Specifically, Treacle was shown to interact with Nop56-associated pre-ribosomal ribonucleoprotein (pre-rRNPs) complexes, which methylate 2’-O-ribose moieties on pre-rRNA (Hayano et al., 2003). Taken together, these data support a role of Treacle in ribosome biogenesis and defines TCS as a ribosomopathy.

Nucleolar stress caused by insufficient ribosome biogenesis leads to activation and stabilization of p53 and p53-dependent cell death (Rubbi & Milner, 2003). Tcof1^+/− mouse embryos exhibit over two-fold higher levels of p53 than their control littermates. Inhibiting p53 by genetic or pharmacological means prevented the craniofacial defects in Tcof1^+/− embryos. p53 inhibition suppresses cell death in the neuroepithelium, facilitating the proper formation and survival of NCC (Fig 4C) (Jones et al., 2008). These results support a role for p53 downstream of perturbed ribosome biogenesis and implicate p53 in the potential prevention of TCS. Although inactivation of a tumor suppressor is not a viable preventative therapy, these results showcase the power of genetic mouse models in uncovering the cellular mechanisms of congenital disease and their utility in designing potential preventative therapy.

Interestingly, the rescue of Tcof1^+/− embryos via p53 loss of function correlated with suppression of neuroepithelial cell death and the restoration of normal numbers of NCC, but failed to reverse the ribosome biogenesis defects (Jones et al., 2008). This finding implies that Tcof1 may play roles in neural progenitor cells and NCC processes other than ribosome biogenesis. In vivo data has shown that during metaphase, Treacle localizes to the centrosomes of neuroepithelial cells and is important for proper spindle orientation and chromosome segregation (Sakai, Dixon, Dixon, & Trainor, 2012). Additionally, Treacle has been implicated in the regulation of oxidative stress and the DNA damage response (Ciccia et al., 2014; Duan, Kelsen, Clarkson, Ji, & Merali, 2010; Larsen et al., 2014). Treacle is required for the translocation of NBS1, a central regulator of DNA damage response, into the nucleoli upon DNA damage. This in turn triggers pan-nuclear silencing of rDNA transcription by Pol I in order to maintain genomic integrity after DNA damage (Ciccia et al., 2014; Larsen et al., 2014). Taken together these data suggests a protective role of Tcof1 in oxidative stress and the DNA damage response. It has been speculated that in TCS patients, Treacle fails to globally shut down rDNA transcription upon DNA damage, which leads to genome instability and cell death (Ciccia et al., 2014; Larsen et al., 2014). It would be interesting to study the DNA damage response and the oxidative stress pathways in Tcof1^+/− mice in order to gain a better understanding of the function of Treacle and the underlying pathogenesis of TCS.

Furthermore, it will be important to study POLR1C and POLR1D mouse models to determine whether their functions overlap with those of TCOF1 in the pathogenesis of TCS. However, although the majority of traditional knock-out mouse models have been helpful in understanding the cellular function of genes, they do not necessarily mimic the exact pathophysiology of human disease. Human diseases are often not caused by null alleles, but by hypomorphic, dominant negative or dominant positive alleles. Therefore, the function of these gene mutations in humans can be quite different physiologically from total loss of function or null mutations in mouse models. Thus it will likely be invaluable to mimic specific mutations that TCS patients carry in TCOF1, POLR1C and POLR1D by using the TALEN or CRISPR system for genome editing (Doudna & Charpentier, 2014; Sommer, Peters, Baumgart, & Beyer, 2015).

Diamond-Blackfan anemia

Diamond-Blackfan anemia (DBA, OMIM 105650) is a congenital disorder that occurs in about 1 per 200,000 live births and is considered one of the founding ribosomopathies (Narla & Ebert, 2010). It was originally characterized as an anemia in infancy and childhood before being later defined as a congenital hypoplastic anemia (Lipton & Ellis, 2009).

Clinical Features

DBA is usually diagnosed during the first year of life primarily due to symptoms such as pallor and lethargy. Patients exhibit anemia (deficiency of red blood cells), macrocytosis (enlarged erythrocytes), reticulocytopenia and a selective decrease or absence of erythroid precursors in an otherwise cellularly normal bone marrow (Narla & Ebert, 2010). Additionally, approximately half of the patients also exhibit craniofacial abnormalities, including cleft lip and/or palate as well as microtia, or other physical abnormalities like thumb deformities and/or cardiac defects (Gazda et al., 2008; Lipton & Ellis, 2009; Ross & Zarbalis, 2014; Trainor & Merrill, 2014).

Genetics

About half of the cases of DBA exhibit autosomal dominant inheritance whereas the remainder arise de novo (Gazda et al., 2008). No matter the nature of the mutation, all DBA patients are heterozygous, suggesting that the homozygous condition is embryonic lethal (Nakhoul et al., 2014). Mutations in RPS19, which encodes a ribosomal protein (RP), were the first to be identified in association with DBA and they occur in 25% of known DBA cases (Gazda et al., 2004; Nakhoul et al., 2014). Another 10–15 small and large ribosomal subunit RPs have also been implicated in DBA and, together, mutations in RPs account for about 50–70% of DBA patients (Gerrard et al., 2013; Nakhoul et al., 2014). Interestingly, mutations in different RPs result in distinct clinical manifestations with specific phenotypes. For example, mutations in RPL5 are more likely to be associated with cleft lip and/or palate, whereas mutations in RPL11 correspond to a higher incidence of thumb abnormalities. In contrast, mutations in RPS19 have a low occurrence of either (Gazda et al., 2008). Nevertheless, the phenotypes characteristic of DBA are variable even within families, similarly to TCS, and therefore environmental factors may also play a role in the manifestation of DBA.

Recently, mutations in the hematopoietic transcription factor GATA1 were identified in two DBA pedigrees, making GATA1 the first non-ribosomal gene to be implicated in DBA (Sankaran et al., 2012). However, specialized ribosomes have previously been shown to selectively control translation of specific proteins, possibly explaining why mutations in RPs can cause tissue specific phenotypes (Kondrashov et al., 2011). Mutations in RP genes may lead to a selective decrease in GATA1 translation, making Gata1 a potentially common factor in the pathogenesis of DBA (Ludwig et al., 2014).

Mouse Models and Mechanism

Several potential mouse models of DBA have been generated, but to date none fully recapitulate the human condition. Several RP mutant mice have been generated that carry deletions, missense, or splicing mutations derived through targeted gene deletion, forward genetic mutagenesis screens, or have arisen spontaneously (Matsson et al., 2004; Terzian & Box, 2013). Homozygosity has typically led to lethality prior to implantation whereas heterozygosity has had no effect (McGowan & Mason, 2011). However, dark skin mice (Dsk) mice with dominant negative and missense mutations in Rps19 exhibit growth retardation and mild anemia (Devlin, Dacosta, Mohandas, Elliott, & Bodine, 2010; McGowan et al., 2008). Dsk mice also display a white belly spot, which is suggestive of defects in NCC derived melanocyte development in humans.

Despite the absence of a complete DBA mouse model, much has been learned from mice and zebrafish carrying mutations in RPs, as well as from primary cells from human patients. A common theme from these studies is that defects in ribosome biogenesis can lead to a decrease in either the 40S, 60S or 80S ribosomal subunits depending on the RP mutated (Narla & Ebert, 2010). Consequently, this decrease results in a global or a selective suppression of protein synthesis, which can lead to nucleolar stress. During nucleolar stress, certain RPs (including RPL11, RPL5 and RPL23) bind to and sequester MDM2, which normally ubiquitinates p53, targeting it for degradation. Without MDM2 function, p53 accumulates and activates the cell death program (Rubbi & Milner, 2003). Similar to the prevention of craniofacial anomalies in Tcof1^+/−;p53^+/− mice, inhibition of p53 prevents anemia and developmental anomalies in rpl11 zebrafish morphants, Rps19 zebrafish, and Rps19 mice, suggesting a similar molecular mechanism underlying the pathogenesis of distinct ribosomopthies that is p53-dependent (Chakraborty, Uechi, Higa, Torihara, & Kenmochi, 2009; Danilova, Sakamoto, & Lin, 2008; Jaako et al., 2011; McGowan et al., 2008).

Another pathway that has been explored for potential treatment and/or prevention of DBA is the mTOR pathway, an important regulator of ribosome biogenesis and consequently cell growth and proliferation. Activation of the mTOR pathway by amino acids such as L-leucine leads to phosphorylation of the RNA Pol I initiation complex, which in turns results in the stimulation of rDNA transcription (Hannan et al., 2003; Mayer, Zhao, Yuan, & Grummt, 2004). Dietary supplementation with L-leucine rescues the anemia in an Rps19 mouse model and the craniofacial abnormalities in rps14 and rps19 morphant zebrafish (Jaako et al., 2012; Payne et al., 2012). Remarkably, dietary supplementation with L-leucine successfully treated the anemia in some DBA patients (Cmejlova et al., 2006; Pospisilova, Cmejlova, Hak, Adam, & Cmejla, 2007). However, mutations in RPs can cause a reduction in the assembly of one of the ribosomal subunits that can then cause a global reduction in protein translation or a more selective reduction in specific proteins. Therefore, although leucine supplementation is exciting as a novel treatment approach, elevating the levels of rDNA transcription by activating the mTOR pathway will not necessarily “fix” proper assembly of the ribosome.

Bowen-Conradi syndrome

Bowen-Conradi syndrome (BCS, OMIM 211180) is an autosomal recessive disorder that occurs almost exclusively within the Hutterite population of the Canadian Prairies at a frequency of 1 in 355 live births and causes death within the first year of life (Lowry et al., 2003). It has been predicted that 1 in 10 is a carrier within the Hutterite population.

Clinical Features

BCS is characterized by pre- and post-natal growth retardation, with a failure to thrive and is currently the only ribosomopathy characterized by severe neurological impairment from birth (Armistead et al., 2015; Lowry et al., 2003). BCS patients also exhibit micrognathia, prominent nose with a lack of glabellar angle, joint anomalies such as camptodactyly, as well as microcephaly and severe psychomotor delay (Lowry et al., 2003). These features overlap with cerebro-oculo-facial-skeletal syndrome and trisomy 18 making diagnosis challenging.

Genetics

In 2009, the identification of c.400A/G, p.D86G mutation in EMG1 was reported as causative of BCS in the Hutterite population (Armistead et al., 2009). EMG1 (essential for mitotic growth 1), which is also known as Nep1 in yeast, is part of the small subunit (SSU) processome, a ribonucleoprotein complex essential for maturation of the 18S rRNA and assembly of the 40S ribosome (Sondalle & Baserga, 2014). EMG1 is highly conserved throughout archaea and eukaryotes, and it is believed to act as a pseudouridine methyltransferase during processing of the 18S rRNA (Wurm et al., 2010).

Mouse Models and Mechanism

Emg1^−/− mouse embryos fail to develop into blastocysts and die prior to implantation (Wu, Sandhu, Patel, Triggs-Raine, & Ding, 2010). This severity is not surprising because the mouse mutation results in null alleles of Emg1. In contrast, the human D86G substitution does not cause a complete deficiency, but rather primarily affects the function of the protein (Armistead et al., 2009; Wu et al., 2010). EMG1 functions as a dimer or a multimer, and the D86G substitution increases the affinity of EMG1 monomers resulting in a 10-fold increase in dimers (Armistead et al., 2009). Interestingly, in yeast, the equivalent D86G substitution does not affect its role as a methyltransferase (Meyer et al., 2011). Therefore it has been suggested that the elevated dimerization of mutant EMG1 prevents its localization to the nucleolus and thus the D86G mutation results in BCS due to a defect in the assembly of the 40S ribosome and not in rRNA modification (Meyer et al., 2011).

Recently, a new mouse model was generated for BCS by knocking-in the D86G substitution in an effort to better mimic the human condition (Armistead et al., 2015). Homozygous mice (Emg^G/G) are smaller than their siblings, exhibit neural tube closure defects, and die between E8.5-E12.5. Emg^G/G embryos exhibit decreased cell proliferation, and this phenotype was also detected in brain tissue and bone marrow autopsies of BCS patients. Furthermore, derivation of Emg^G/G mouse fibroblasts showed a high number of binucleated cells pointing to a defect in mitosis that could be the cause of growth arrest in BCS (Armistead et al., 2015). Interestingly, mitotic arrest and ribosome biogenesis have also previously been linked in other ribosomopathies such as TCS and Shwachman-Diamond syndrome. Thus, further study of this relationship in the context of these diseases will be important for a better understanding of the pathogenesis of ribosomopathies (Austin et al., 2008; Sakai et al., 2012).

Postaxial acrofacial dysostosis

Postaxial acrofacial dysostosis (POADS, OMIM 263750), also known as Miller, Genée-Wiedemann and Wildervanck-Smith syndrome, is an autosomal recessive disorder with an occurrence of 1 in 1,000,000 live births (Wieczorek, 2013).

Clinical Features

The craniofacial deformities observed in POADS are almost identical to the characteristic features of TCS and include micrognathia, orofacial clefts, malar hypoplasia, coloboma of the lower eyelid and cup-shaped ears. Unlike TCS, however, POADS patients also exhibit postaxial limb defects that include absence of the fourth or fifth digits, or both rays of the hands and feet with or without ulnar and fibular hypoplasia (Trainor & Andrews, 2013).

Genetics

POADS, the first genetic disorder for which the causative gene was identified via whole exome sequencing, correlates with mutations in dihydroorotate dehydrogenase (DHODH), a gene associated with the mitochondrial electron transport chain and de novo pyrimidine biosynthesis pathway (Ng et al., 2010). Fourteen different mutations in the coding region of DHODH have been identified to date and, surprisingly, all affected individuals carry compound heterozygous mutations in DHODH, which is atypical of an autosomal dominant disorder (Rainger et al., 2012).

Mouse Models and Mechanism

DHODH is one of three enzymes that catalyze the six enzymatic reactions needed for de novo synthesis of pyrimidine. DHODH catalyzes the conversion of dihydroorotate to orotic acid, which is then converted to uridine monophosphate, the RNA nucleotide essential for ribosome biogenesis (Brosnan & Brosnan, 2007; Minet, Dufour, & Lacroute, 1992). There are currently no mouse models for POADS, but inhibition of DHODH by the immunosuppressant leflunomide in zebrafish completely abolished NCC, resulting in a reduction of their derivatives as well as a disruption in the jaw cartilage (White et al., 2011). Moreover, microarray analysis of leflunomide-treated zebrafish embryos showed downregulation of NCC genes. Leflunomide also caused a decrease in the self-renewal capacity of rat NC stem cells. These data suggest defects in NCC development are a potential cause of the pathogenesis of POADS.

Interestingly, suppression of DHODH leads to p53-dependent cell death and, independent of this role, DHODH has also been linked to oxidative stress, illustrating similarities in the pathogenesis of TCS and POADS (Khutornenko, Dalina, Chernyak, Chumakov, & Evstafieva, 2014; White et al., 2011). Generating mouse models for POADS that mimic the mutations identified in humans are necessary to provide a better understanding of this disease and the cellular and tissue abnormalities associated with this condition.

Roberts syndrome

Roberts syndrome (RBS, OMIM 268300) is another rare genetic disorder that is inherited in an autosomal recessive manner and currently only about 150 affected individuals have been reported.

Clinical Features

RBS is characterized by slow pre- and postnatal growth, mental retardation, craniofacial defects including cleft lip and palate, hypertelorism, hypoplastic nasal alae, micrognathia and ear abnormalities, as well as limb defects that include shortened arm and leg bones (Trainor & Merrill, 2014).

Genetics

RBS is caused by mutations in ESCO2 (Establishment of cohesion 1 homolog 2). The yeast homolog, Eco1 is an acetyltransferase crucial in the assembly of cohesin (Vega et al., 2005). Cohesin is a multiprotein complex that holds sister chromatids together from S phase until the start of mitosis, helping to ensure genomic integrity (Haarhuis, Elbatsh, & Rowland, 2014). The acetyltransferase activity of Eco1 is required for the establishment of cohesin-DNA interactions during S phase (Skibbens, Corson, Koshland, & Hieter, 1999; Toth et al., 1999). Mutations in Eco1 result in sensitivity to DNA damage agents that cause double-strand breaks, which is an emerging theme in ribosomopathies (Bose et al., 2012; S. Lu et al., 2010).

Mouse Models and Mechanism

Esco2 morphant zebrafish exhibit smaller heads and eyes, abnormal pigmentation, reduced and deformed craniofacial cartilage as well as defects in pectoral fin development and cardiac problems (Monnich, Kuriger, Print, & Horsfield, 2011); all of which resemble RBS. In contrast, Esco2 homozygous mouse embryos die by the 8-cell stage due to problems in chromatid cohesion. However, conditional knock out of Esco2 specifically in the neuroepithelium of the dorsal telencephalon (Emx1-CRE;Esco2flx/flx) resulted in viable mice with microcephaly, a deformity found in some RBS patients (Whelan et al., 2012). At midgestation, Emx1-CRE;Esco2flx/flx embryos are characterized by agenesis of the neocortical and hippocampal neuroepithelium due to cell death. Given the craniofacial deformities that patients exhibit, it would be interesting to conditionally knock out Esco2 from NCC in mice and examine whether it recapitulates the human condition. Moreover, given the early embryonic requirements for Esco2, it will be important to control deletion of Esco2 temporally to determine if Esco2 has other tissue specific roles later in development. Nonetheless, analysis of RBS mutations in yeast and human patient cell lines revealed defects in ribosome biogenesis, rRNA production and protein translation, as well as a disruption in nucleolar morphology (Bose et al., 2012). It is clear that RBS can be categorized as a ribosomopathy, however additional work using the Esco2 conditional mouse model is needed for a more detailed understanding of RBS.

The identification and characterization of ribosomopathies in humans and their subsequent study in mouse and zebrafish models has revealed new and surprising roles for the molecular machines called ribosomes. The discovery that mutations in genes involved in ribosome biogenesis, are not necessarily embryonic lethal, but instead cause tissue-specific craniofacial defects, has changed our view of the roles and functions of global cellular processes during development and disease. Mouse models have helped reveal the etiology and pathogenesis of ribosomopathies and have also been instrumental in furthering our understanding of other rare craniofacial disorders that arise within ciliopathies, or disorders of cilia development and function.

Ciliopathies

Primary cilia are finger-like projections that extend from the surface of nearly every vertebrate cell in order to receive and process signaling cues. Each cilium is composed of an axoneme, a microtubule based structure; the basal body (centrosomes) that connects the axoneme to the cell body; and the specialized ciliary membrane that covers the axoneme and hosts numerous signaling pathway receptors (Fig 5) (Chang, Schock, Attia, Stottmann, & Brugmann, 2015; Sorokin, 1962). Another critical component is the transition zone, which is found at the base of the axoneme and the distal basal body. The transition zone accommodates Y-shaped protein structures that connect the axonemal microtubules to the ciliary membrane and also serves as the ciliary gate required for establishing the cilium as its own compartment separated from the rest of the cell body (Dawe, Farr, & Gull, 2007). The primary cilium is absolutely required for the transduction of important signaling pathways including the Hedgehog (Hh), platelet-derived growth factor (PDGF) alpha, polycystin, and Wnt pathways (Pedersen & Rosenbaum, 2008).

Figure 5. The primary cilium and ciliopathies.

Schematic representation of a primary cilium detailing the main compartments of the cilium as well as the products of the genes and their localization and involvement in specific ciliopathies.

Ciliogenesis, the formation of the cilium, is an elaborate process that starts with differentiation of the mother centriole into a basal body followed by its docking at the apical membrane. The basal body then provides a scaffold for axonemal microtubules to polymerize giving rise to the axoneme and thus extending the cilium into the extracellular space. This extension requires intraflagellar transport (IFT) machinery for the transport of tubulin and other building blocks of the axoneme. IFT is also critical for the function of the cilium because this microtubule-motor-based process shuttles and localizes proteins including signal transduction components to the mature cilia. This transport is bi-directional along the axoneme with IFT proteins that travel to the distal end of the microtubules being referred to as anterograde, whereas proximal transport towards the cell body is referred to as retrograde (Reiter, Blacque, & Leroux, 2012).

Mutations in genes encoding proteins involved in the formation, structure, maintenance or function of primary cilia can lead to ciliopathies (Fig 5) (Badano, Mitsuma, Beales, & Katsanis, 2006). Many ciliopathies exhibit craniofacial malformations among other deformities, suggesting a role for cilia in NCC development. The following section focuses on the role of cilia in craniofacial structures and rare craniofacial disorders based on mouse models of ciliopathies.

Oral-Facial-Digital syndrome 1

Oral-Facial-Digital syndrome 1 (OFDS1, OMIM 311200) is an X-linked dominant disorder that affects 1 in 50,000 to 250,000 newborns and is the most common subtype among thirteen different oral-facial-digital conditions. All OFDS1 patients are female, as male carriers die during the first or second trimester of pregnancy. Although primarily an inherited disorder, 30% of the cases are sporadic (Gurrieri, Franco, Toriello, & Neri, 2007; Wettke-Schafer & Kantner, 1983).

Clinical Features

OFDS1 is characterized by craniofacial anomalies, digit deformities, central nervous system abnormalities, and polycystic kidney disease (Thauvin-Robinet et al., 2006). Craniofacial malformations occur in over 87% of reported cases and include facial asymmetry, hypertelorism, micrognathia, broadened nasal ridges, cleft palate, high arched palate, lingual hamartomas, hypodontia, and hyperplastic buccal fernula that can lead to clefting (Brugmann, Cordero, & Helms, 2010; Gurrieri et al., 2007; Thauvin-Robinet et al., 2006).

Genetics

OFD1, is the only gene to be associated with OFDS1 thus far, although it still does not account for most of the OFDS1 and other OFD subtypes (Ferrante et al., 2001; Rakkolainen, Ala-Mello, Kristo, Orpana, & Jarvela, 2002; Romio et al., 2003). OFD1 is widely expressed from early stages of development in all tissues affected in OFDS1 in both humans and mice (de Conciliis et al., 1998; Ferrante et al., 2003; Romio et al., 2004; Romio et al., 2003). OFD1 encodes a centrosomal protein that localizes at the basal body of primary cilia. Numerous mutations and deletions have been identified in OFD1 with most of them resulting in truncated protein products that are non-functional (Ferrante et al., 2001; Thauvin-Robinet et al., 2006).

Similar to the ribosomopathy disorders described earlier, there is high phenotypic variability even within the same family. In this case the variability has been attributed to somatic mosaicism due to random × inactivation, but environmental factors may also play a role (Franco & Ballabio, 2006).

Mouse Models and Mechanism

Ofd1 knock-out mice recapitulate the main features of OFDS including craniofacial and limb abnormalities (Ferrante et al., 2006). The phenotypes of Ofd1 mice are more severe than the human condition, probably due to differences in the pattern of X-inactivation between humans and mice. Cilia in Ofd1 mutant male embryos are absent from the embryonic node, consistent with a failure of left-right axis specification. Additionally, Ofd1 mutant embryos exhibit abnormal patterning of the neural tube associated with an absence of Shh and considerable reduction in expression of Shh downstream target genes including Ptc1 and Gli1 (Ferrante et al., 2006). The neural tube phenotype resembles that of some IFT mutants suggesting a genetic interaction between Ofd1 and IFT genes (Ferrante et al., 2006). IFT is required for cilia formation and maintenance, and IFT proteins are also involved in the Shh pathway via regulation of Gli protein function (Liu, Wang, & Niswander, 2005). Interestingly, Ofd1 has been hypothesized to functionally interact with the Gli proteins, and, consistent with this idea, loss of Ofd1 disrupts the ciliary processing of Gli3 resulting in phenotypes that resemble Gli3 mutants (Brugmann, Cordero, et al., 2010; Toriello, 2009).

Recently, a link between FGF-related syndromes and ciliopathies was observed in the high arch palate phenotype of Fuzzy (Fuz) mutant mice (Tabler et al., 2013). Fuz, a planar cell polarity effector essential for targeted membrane trafficking, regulates ciliogenesis, although it is not structurally or physically part of the cilium (Gray et al., 2009; Park, Haigo, & Wallingford, 2006). It has been suggested that the high arched phenotype in Fuz mice is due to expanded Fgf8 expression, which results in abnormal NCC migration into the facial prominences and an enlarged maxillary process. Strikingly, Ofd1 mutant mice also exhibit Fgf8 expansion, enlarged maxillary processes, and a high arch palate (Tabler et al., 2013).

Recently, a mutation in C2CD3 was identified in a 4-year-old male that presented with canonical OFDS1 in addition to microcephaly, micropenis and severe intellectual disability, potentially representing a new OFD subtype (Thauvin-Robinet et al., 2014). C2CD3 colocalizes with OFD1 at the distal end of centrioles, which initiate the assembly of cilia. Hearty mutant mice that carry a mutation in C2cd3 exhibit neural tube defects, abnormal left-right asymmetry axis determination and polydactyly (Hoover et al., 2008). C2cd3, similar to Ofd1, is required for ciliogenesis and is also an essential regulator of the intracellular transduction of Hedgehog signaling. These observations add to accumulating data that OFDS should be classified as a ciliopathy.

Bardet-Biedl syndrome

Bardet-Biedl syndrome (BBS, OMIM 209900) is an autosomal recessive condition with a general occurrence of 1 in 140,000 newborns, but up to 1 in 17,000 newborns in some isolated regions (Klysik, 2008).

Clinical Features

BBS is pleiotropic and is characterized by retinal degeneration, obesity, polydactyly, learning difficulties, and renal and genital anomalies (Fig 3A) (M’Hamdi, Ouertani, & Chaabouni-Bouhamed, 2014). A subgroup of BBS patients also exhibit craniofacial deformities including hypertelorism, deep set eyes, downward slanting palpebral fissures, high arched palate, migrognathia, a flat nasal bridge with anteverted nares and prominent nasolabial folds, a long philtrum, and a thin upper lip. Interestingly a small percentage also displays Hirschsprung disease, which is a neurocristopathy affecting the enteric nervous system (Beales, Elcioglu, Woolf, Parker, & Flinter, 1999).

Genetics

To date, mutations in 18 genes have been identified that account for 70–80% of the BBS cases, and they have been classified as BBS1–18. All of the BBS proteins localize to the centrosome/basal bodies or ciliary axoneme, and seven of them form a complex termed the BBSome (BSS1, 2, 4, 5, 7, 8, and 9). Interestingly, mutations in specific BBS genes are more prevalent than others in different geographic regions (Brugmann, Cordero, et al., 2010; M’Hamdi et al., 2014; Nachury et al., 2007).

Mouse Models and Mechanism

BBS proteins can associate with cofactors to promote trafficking of vesicles to the cilium, but their function still remains undetermined. One hypothesis is that BBS proteins indirectly affect cilia by bringing additional proteins to the centrosome, the failure of which leads to centrosome/basal body abnormalities and ciliary dysfunction. Alternatively, BBS proteins may have a role within the cilium, to promote cohesion between IFT subcomplexes (Brugmann, Cordero, et al., 2010; Nachury et al., 2007).

Mouse models have begun to shed some light on the function of BBS proteins. Bbs2, Bbs4 and Bbs6 knock-out mice exhibit some of the human phenotypes including retinal regeneration, but their cilia and basal body structures are not defective suggesting a role in IFT, and not in the assembly of cilia (Fath et al., 2005; Mykytyn et al., 2004; Nishimura et al., 2004). Further studies have shown that the BBSome functions to promote cilia membrane biogenesis and serves as a coat complex for ciliary membrane protein trafficking (Jin et al., 2010; Nachury et al., 2007).

Surprisingly, craniofacial development has not been extensively studied in these BBS mouse models, perhaps because the phenotypes are subtle. Three-dimensional surface modeling of Bbs4^−/− and Bbs6^−/− adult mice, however, revealed a larger mid-face width to height ratio and a shorter snout, which was attributed to premaxillary and maxillary hypoplasia (Tobin et al., 2008). Furthermore, bbs morphant zebrafish exhibit craniofacial malformations including shortening of the anterior neurocranium, migrognathia and cyclopia, phenotypes that were attributed to an inability of the NCC to migrate into the facial prominences. These morphant zebrafish also exhibited Hirschsprung’s disease, which results from the inability of the NCC to form a complete enteric nervous system. Interestingly, disruption of Wnt and Shh signaling, two signaling pathways regulated by cilia, may underlie these NCC defects (Tobin et al., 2008).

More work clearly needs to be done using the currently available mouse models to understand the relationship between primary cilia and CNC. However new models also need to be generated that mimic the severe craniofacial defects characteristic of the human mutations. Moreover, given the number of the different BBS genes mutated, generating compound heterozygous or homozygous mice will help uncover any level of redundancy that exists among BBS genes.

Meckel-Gruber syndrome

Meckel-Gruber syndrome (MKS, OMIM 249000) is an autosomal recessive disorder that occurs in about 1 in 140,000 live births and results in perinatal lethality. It occurs at a higher frequency in more isolated populations and is especially prevalent in Finns (Szymanska, Hartill, & Johnson, 2014).

Clinical Features

MKS is characterized by polycystic kidneys, cardiac defects, polydactyly and numerous craniofacial abnormalities including occipital encephalocele, microcephaly, sloping forehead, cleft lip/palate, macrostomia, micrognathia, micropthalmia and tongue defects (Brugmann, Cordero, et al., 2010; Szymanska et al., 2014). Lethality occurs either in utero or shortly after birth usually due to pulmonary hypoplasia (Szymanska et al., 2014).

Genetics

There are eleven subtypes of MKS and mutations in 11 genes have been identified to date. The first MKS locus to be identified was MKS1 and subsequently mutations in 10 more genes have been identified: TMEM216, TMEM67, CEP290, RPGRIP1L, CC2D2A, NPHP3, TCTN2, B9D1, B9D2, TMEM231 (Szymanska et al., 2014).

Mouse Models and Mechanism

Several mouse models have been generated to study the different MKS subtypes and much has been learned regarding the role of ciliogenesis in the pathogenesis of MKS.

Mks1

Two independent chemical mutagenesis screens recovered mutants that carried mutations in Mks1 (Cui et al., 2011; Weatherbee, Niswander, & Anderson, 2009). MKS1 is a B9-domain containing protein that localizes to the basal body of mammalian cells. MKS1 regulates ciliogenesis by interacting with meckelin and two other B9-domain containing proteins, B9D1 and B9D2, mutations in which have also been identified in MKS patients (Dawe, Smith, et al., 2007; Kyttala et al., 2006; Williams, Winkelbauer, Schafer, Michaud, & Yoder, 2008). Mks1 mutant mice kerouac (krc)/Mks1^krc and Mks1^del64−323 both display several characteristic MKS deformities including polydactyly, cystic kidneys and craniofacial defects such as exencephaly, reduced ossification of craniofacial bones and cleft palate. Both mutants also display a dramatic reduction in the number of primary cilia in many but not all tissues, which may explain the tissue specific phenotypes observed in both mouse and human MKS. Moreover, these mouse models revealed that Mks1 functions in the dorsal-ventral patterning of the neural tube and the anterior-posterior patterning of the limb bud by regulating the Shh pathway (Cui et al., 2011; Weatherbee et al., 2009). A more recently generated mouse model, Mks1^{tm1a(EUCOMM)Wtsi}, exhibits similar defects to Mks^krc and Mks^del64−323 mice (Wheway et al., 2013). Interestingly, in addition to perturbation of Shh signaling, the Wnt and mTOR pathways were also upregulated in association with ectopic cell proliferation. Upregulation of the mTOR pathway may explain the formation of the cysts in the kidney, however further studies are needed to determine how MKS1 alters the Wnt1 and mTOR pathways.

Tmem67

MKS subtype 3 (MKS3) is caused by mutations in TMEM67,which encodes meckelin, an orphan transmembrane receptor found in the renal epithelium and the primary cilium (Smith et al., 2006). Two mouse models for MKS3 have been described: the bpck mouse that carries a spontaneous deletion including the Tmem67 gene and the Tmem67^tm1(Dgen/H) mouse that carries a targeted allele of Tmem67 (Cook et al., 2009; Garcia-Gonzalo et al., 2011). Both models exhibit phenotypes that resemble MKS3 including polycystic kidneys and hydrocephalus, but do not show other classic MKS phenotypic characteristics such as polydactyly and encephalocele. Recently however, congenic Tmem67^tm1(Dgen/H) embryos revealed some of the MKS craniofacial deformities including exencephaly, prosencephalon dysgenesis, and encephalocele (Abdelhamed et al., 2013). All mutants exhibit dysmorphic and fewer or absent primary cilia in a tissue-specific distribution that is consistent with their phenotypes. The congenic Tmem67^tm1(Dgen/H) mice also revealed de-regulation of the Shh and Wnt pathways in the developing brain. Therefore it was proposed that TMEM67 regulates the trafficking of Gli proteins and Dishevelled in the cilium (Abdelhamed et al., 2013). The congenic mutants revealed a degree of phenotypic severity consistent with human MKS and, together with the other Tmem67 mutant mice, could serve as models to understand the variability in human patients carrying the same genetic mutations (Abdelhamed et al., 2013).

Rpgrip1l

RPGRIP1L contains several protein-protein interaction domains and is localized in the basal-body centrosome complex as well as the transition zone along with nephrocystins 1 and 4 (Arts et al., 2007; Delous et al., 2007; Sang et al., 2011; Vierkotten, Dildrop, Peters, Wang, & Ruther, 2007). Mutations in RPGRIP1L are associated with MKS subtype 5. Rpgrip1l^−/− (Ftm) mice die neonatally and exhibit phenotypes observed in human patients with MKS including exencephaly, polydactyly, laterality defects, micropthalmia, cleft lip and hypoplastic lower jaw (Delous et al., 2007; Vierkotten et al., 2007). Ftm^−/− embryos, although able to assemble cilia, have fewer cilia, which are unable to transduce the Hh signaling properly and thus manifest Hh-deficiency related deformities (Vierkotten et al., 2007).

Cc2d2a

Cc2d2a also contains protein-protein interaction domains and localizes to the basal body (Garcia-Gonzalo et al., 2011; Gorden et al., 2008; Tallila, Jakkula, Peltonen, Salonen, & Kestila, 2008). Mutations in CCSD2A are associated with MKS subtype 6. Cc2d2a^−/− mouse embryos die just prior to birth and exhibit pleiotropic phenotypes resembling MKS including micropthalmia, exencephaly, polydactyly and defects in multiple organs as well as situs inversus (Garcia-Gonzalo et al., 2011; Veleri et al., 2014). Cc2d2a is part of a complex that includes multiple ciliopathy-associated proteins. This complex localizes to the transition zone, includes Mks1, Tmem216, Tmem67, Tmem231, Cep290, B9d1, Tctn1, and Tctn2, and regulates ciliogenesis and ciliary membrane composition (Chih et al., 2012; Garcia-Gonzalo et al., 2011). Cdc2d2a was shown to localize to the subdistal appendages (SDA) in the mother centriole and its loss inhibits the initiation of cilia biogenesis from the basal body. As a consequence, Cc2d2a^−/− embryos have fewer and defective cilia, which leads to defects in Shh signaling, helping to account for the pleiotropic phenotypes (Veleri et al., 2014).

Nphp3

Nephrocystin-3 (Nphp3) interacts with Nphp1 and Nphp2, two proteins involved in nephronophthisis, a group of autosomal recessive cystic kidney diseases, that localize to the ciliary base (Olbrich et al., 2003). Mutations in NPHP3 are associated with MKS subtype 7. Not much in known about Nphp3 other than it contains protein-protein interacting domains and is expressed during mouse embryonic development in the node, neural tissue, kidney tubules, retina, respiratory epithelium, biliary tract and liver (Bergmann et al., 2008; Olbrich et al., 2003). Two mouse models exist bearing mutations in Nephrocystin-3 (Nphp3). Nphp3^pcy is a spontaneous hypomorphic mutation, and Nphp3^ko is a targeted null mutation (Bergmann et al., 2008; Olbrich et al., 2003; Omran et al., 2001; Takahashi et al., 1986). Nphp3^pcy/pcy mice exhibit adolescent nephronophthisis with no other apparent deformities, whereas Nphp3^ko/ko display MKS-like symptoms including midgestational embryonic lethality and defective left-right asymmetry (Bergmann et al., 2008). Interestingly, no craniofacial defects have been described in these mutants.

Tctn2

Tectonic2 (Tctn2) encodes a transmembrane protein that is part of the ciliopathy complex with Cc2d2a that localizes to the transition zone (Garcia-Gonzalo et al., 2011). Mutations in TCTN2 are associated with MKS subtype 8. Tctn2^−/− embryos exhibit neural tube closure defects, exencephaly, microphthalmia, cleft palate, polydactyly, and left-right asymmetry defects (Sang et al., 2011). Similar to other ciliopathy genes, Tctn2 is required for ciliogenesis and, consequently, for proper Hh signaling transduction.

B9d1 and Tmem231

B9d1 and Tmem231 knock out mouse embryos exhibit similar lethality during late gestation and MKS-like phenotypes including micropthalmia and syndactyly (Chih et al., 2012). Consistent with these phenotypes is a loss of cilia that results in misregulation of the Shh signaling pathway. B9d1 and Tmem231 are part of the ciliopathy complex in the transition zone required for ciliogenesis and ciliary membrane composition (Chih et al., 2012; Garcia-Gonzalo et al., 2011). Mutations in B9D1 are associated with MKS subtype 9.

B9d2

The Stumpy mutant mouse carries a conditional deletion of the B9d2 gene (Town et al., 2008). Similar to other ciliopathy mouse mutants, Stumpy mice exhibit perinatal lethality, hydrocephaly, polycystic kidneys and abnormalities in the olfactory bulb, the hippocampus and the cerebellum. They have a dramatic decrease in primary cilia and an absence of ciliary axonemes in affected tissues, which then result in deregulation of the Shh signaling pathway, in association with aforementioned phenotypes. (Breunig et al., 2008; Town et al., 2008). B9d2 is also part of the ciliopathy complex that localizes to the transition zone of the cilium (Chih et al., 2012) and mutations in B9D2 are associated with MKS subtype 10.

Joubert syndrome

Joubert syndrome (JBTS, OMIM 213300) is an autosomal recessive neurodevelopmental condition with ten subtypes and an occurrence of 1:100,000 (Szymanska et al., 2014).

Clinical Features

JBTS patients exhibit a characteristic brain abnormality called the molar tooth sign that can be detected by MRI and is the result of abnormal development of the cerebellar vermis and the brain stem. Common features of JBTS include developmental delay, ataxia, hyperpnea, hypotonia, sleep apnea, abnormal eye and tongue movements, renal/liver deformities, and polydactyly. JBTS patients also have craniofacial malformations including a long face and frontonasal prominence, bitemporal narrowing, ptosis of the eyelid, protruding nasal bridge and tip, prognathism, eyebrow deformities, trapezoid shaped mouth, lower lip eversion, and thick ear lobes (Maria, Boltshauser, Palmer, & Tran, 1999). Based on the mouse models described below, the genes implicated in JBTS are involved in signaling, rather than the actual structure of the cilium, which is likely the reason JBTS has milder defects compared to other ciliopathies.

Genetics

There are currently 22 genes associated with JBTS: INPP5E, TMEM216, AHI1, NPHP1, CEP290, TMEM67, RPGRIP1L, ARL13B, CC2D2A, OFD1, TTC21B, KIF7, TCTN1, TMEM237, CEP41, TMEM138, C5orf42, TCTN3, ZNF423, CSPP1, and PDE6D. Interestingly, TMEM67, RPGRIP1L, CC2D2A, TMEM216, and CEP290 are allelic to MKS (Szymanska et al., 2014).

Mouse Models and Mechanism

Several mouse models of JBTS have been generated revealing the cellular basis of the disease as well as the role of different JBTS associated genes in ciliogenesis.

Inpp5E

Inpp5E encodes lipid phosphatase 5, which localizes to the ciliary axoneme and hydrolyzes the 5-phosphate of PtdIns(3,4,5)P3 and PtdIns(4,5)P2 (Astle, Horan, Ooms, & Mitchell, 2007; Bielas et al., 2009; Jacoby et al., 2009). Innp5E^Δ/Δ mice die soon after birth and exhibit polydactyly, anopthalmia, polycystic kidneys, cleft palate, anencephaly, exencephaly and delayed ossification of metacarpals and phalanges (Jacoby et al., 2009). Inpp5E^Δ/Δ embryos have fewer primary cilia, with altered morphology, whereas Inpp5E^Δ/Δ mouse embryonic fibroblasts are able to assemble cilia but fail to maintain them. It has been hypothesized that cilium stability is regulated by growth factor receptors such as PDGFRα that regulate ciliary PtdIns(3,4,5)P3 production and cilia disassembly. Therefore, in the absence of Inpp5E, cilia disassemble rapidly due to the accumulation of PtdIns(3,4,5)P3 (Jacoby et al., 2009). The phenotypes of the Inpp5E^Δ/Δ embryos are more severe than those of JBTS human patients, which may again reflect variable severity of the causative mutations in humans versus the null mouse (Plotnikova et al., 2015).

Ahi1

Ahi1 encodes the signaling protein Jouberin (Jbn), which localizes to the basal body of the cilium and was the first gene to be identified in JBT (Dixon-Salazar et al., 2004; Ferland et al., 2004). Ahi1^−/− null mice survive until birth with no apparent embryonic deformities except a smaller cerebellum and underdeveloped vermis that resembles human JBT. These mice are smaller, with slightly reduced brain size, have polycystic kidneys and higher mortality (Lancaster et al., 2011; Lancaster et al., 2009). Ahi1^−/− mice also display rapid loss of photoreceptors in the retina (Louie et al., 2010). Examination of primary cilia revealed unchanged number, length and morphology, suggesting a role of Jbn in signaling rather than ciliogenesis. Indeed Jbn was shown to modulate Wnt signaling downstream of β-catenin and has been hypothesized to facilitate nuclear translocation of β-catenin.

Arl13b

Arl13b is a small GTPase of the Arf/Arl family that localizes to the ciliary axoneme. A mutation in Arl13b was found in the hennin (hnn) mutant mouse, as part of an ENU screen, and has subsequently been used to study JBTS (Garcia-Garcia et al., 2005). hnn embryos exhibit an open neural tube, randomized heart looping, abnormal eyes, polydactyly, and midgestational lethality (Caspary, Larkins, & Anderson, 2007). hnn embryos have short cilia with defective axonemes that perturb the Gli gradient and thus the Shh signaling necessary for the spatial organization of the ventral neural tube. Studies in zebrafish showed that Arl13b functions in ciliary membrane extension to regulate ciliary length (H. Lu et al., 2015). Moreover, conditional Arl13b mice revealed the presence of guidance cue receptors essential for interneuronal migration on the interneuronal primary cilia, whose concentration and dynamics are controlled by Arl13b (Higginbotham et al., 2012; Su, Bay, Mariani, Hillman, & Caspary, 2012).

Ttc21b

The alien (aln) mouse mutant was identified in a chemical mutagenesis screen affecting late embryogenesis and carries a mutation in the Ttc21b gene (Herron et al., 2002; Tran et al., 2008). Ttc21b^aln/aln embryos exhibit polydactyly, delayed eye and forebrain development, and neural tube defects. Ttc2b localizes to the ciliary axoneme and in its absence the axonemes have an abnormal morphology and accumulate IFT proteins due to impairment of IFT. Ttc21b^aln/aln embryos exhibit activated Shh signaling in association with disrupted IFT (Tran et al., 2008). The aln mutation was particularly useful in uncoupling the roles of anterograde and retrograde transport in Shh signaling and it was suggested, although not shown, that anterograde IFT is required for Gli activation, which is then modulated by retrograde IFT.

Kif7

There are currently two mouse models with mutations in Kif7: a targeted Kif7 knock-out, and the ENU-induced mataraki (maki) allele. Similar to other JBTS mouse models, Kif7^−/− and Kif7^maki embryos exhibit polydactyly, exencephaly, and micropthalmia, phenotypes consistent with defects in the Hh pathway (Endoh-Yamagami et al., 2009; Liem, He, Ocbina, & Anderson, 2009). Kif7^−/− embryos display patterning defects in the neural tube and limb buds due to deregulation of the Shh signaling. Kif7, a kinesin family member, localizes to the base of the cilia in the absence of Shh and accumulates at the distal tip of the primary cilia upon activation of the Shh pathway. Based on studies of the Kif7^−/− embryos, Kif7 regulates Shh signaling by controlling the efficient localization of Gli3 to cilia and processing Gli3 to its repressor form (Endoh-Yamagami et al., 2009). Studies of Kif7^maki embryos revealed a role for Kif7 downstream of Smoothened and upstream of Gli2 controlling Shh signaling in both a positive and negative manner (M. He et al., 2014; Liem et al., 2009). It was recently demonstrated that Kif7 does not traffic Gli proteins into the cilia, but instead controls cilium architecture by modulating microtubule dynamics at the tip and creating a single cilium tip compartment where Gli proteins can be regulated (M. He et al., 2014).

Tctn1

Tectonic1 (Tctn1) mouse mutants die during late gestation and display holoprosencephaly, a failure to form the floorplate and to pattern the neural tube properly, as well as defects in left-right asymmetry, all phenotypes typically associated with defective Shh signaling (Mitchell et al., 2001; Reiter & Skarnes, 2006). Tctn1 is part of the ciliopathy complex that localizes to the transition zone of the cilium and regulates ciliogenesis and ciliary membrane composition (Garcia-Gonzalo et al., 2011). Epistasis experiments revealed that Tctn1 modulates Shh signaling downstream of Smoothened (Reiter & Skarnes, 2006). Examination of the node and neural tube shows a dramatic reduction in the number of cilia (Garcia-Gonzalo et al., 2011). In contrast, cilia number in the limb buds is not affected, but these cilia fail to localize membrane-associated proteins such as Smoothened and Arl13b.

Cep41

Cep41 localizes to the basal body and regulates ciliary entry of the polyglutamylase enzyme TTLL6, which modulates tubulin glutamylation that is necessary for axonemal formation (Lee et al., 2012). Cep41^Gt/Gt embryos have a malformed hindbrain, exencephaly, brain hemorrhage, cardiac deformities and embryonic lethality. Similar to other ciliopathy mouse mutants, Cep41^Gt/Gt embryos exhibit phenotypic variability suggesting extragenic phenotypic modifiers. The cause of the phenotypes, as well as the architecture of the cilia in these mutants, have not been examined yet, hence further studies are needed to better understand the role of ciliary glutamylation in mammals. However, studies in primary cells from human patients suggest that CEP41 is not required for ciliogenesis (Lee et al., 2012).

The role of cilia in NCC and craniofacial development

Much has been learned about the role of cilia in NCC and craniofacial development from an additional array of mutant mice exhibiting dysfunctional cilia. The currently favored hypothesis is that primary cilia are not absolutely required for proper NCC specification (Brugmann, Allen, et al., 2010; Chang et al., 2015). Primary cilia do, however, play an important role in NCC migration, because the receptors of several chemotactic molecules essential for NCC migration localize to primary cilia (Schneider et al., 2005; Theveneau & Mayor, 2012). In support of this idea, NCC in zebrafish morphants and mouse embryonic fibroblasts lacking cilia display aberrant migration patterns (Chang et al., 2015). Furthermore, the importance of primary cilia for proliferation of cranial NCC has also been demonstrated. NCC-specific excision of Kif3a, which encodes a ciliary microtubule motor, causes an increase in proliferation that leads to midfacial expansion (Brugmann, Allen, et al., 2010).

Thus, ciliary proteins can have cell- and tissue-specific effects that are consistent with the ciliopathies described here. Many of the phenotypic anomalies are common to many ciliopathies, but there are also ciliopathy-specific defects in individual tissues. Moreover the degree of severity varies from ciliopathy to ciliopathy, which further supports a tissue-specific role for ciliary proteins that was previously not appreciated.

Thus, similar to ribosomopathies, mouse models of ciliopathies have revealed tissue-specific roles for a global cellular structure, in this case, the cilium. Future research is now aimed at understanding how cilia-associated genes regulate signaling pathways in a tissue specific manner. In contrast to ribosomopathies and ciliopathies that encompass rare disorders of craniofacial development but with other pleiotrophic phenotypes, there are a number of rare but specific craniofacial anomalies such as syngnathia, for which mouse models have been instrumental recently in revealing their etiology and pathogenesis.

Syngnathia

Syngnathia (OMIM 119550), the bony fusion of the upper and lower jaw, is a rare congenital disorder with only about 60 cases reported in the literature to date (Fig 3C,D) (Inman, Purcell, Kume, & Trainor, 2013).

Clinical Features

Syngnathia causes restriction of the mouth opening affecting feeding, breathing, and communication. Currently syngnathia is classified into 4 types: Type 1a, simple anterior syngnathia by bony fusion of the alveolar ridge only, with no other craniofacial congenital anomalies; Type 1b, complex anterior syngnathia by bony fusion of the alveolar ridges associated with other craniofacial defects; Type 2a, simple zygomatic-mandibular syngnathia by bony fusion of the mandible to zygoma; and Type 2b, complex zygomatic-mandibular syngnathia by bony fusion of the mandible to the zygoma associated with cleft palate and/or temporomandibular joint ankylosis (Laster, Temkin, Zarfin, & Kushnir, 2001). Syngnathia can also be associated with syndromes such as Van der Woude and cleft palate lateral alveolar synechiae syndrome. In one rare case, syngnathia was associated with maxillary duplication, nasal cleft, and other non-craniofacial defects including situs inversus, left choanal stenosis and dextrocardia (Patel, Porras, & Lypka, 2015). Furthermore, in another rare case, a newborn presented with both syngnathia and craniosynostosis (Ajike et al., 2008). Depending on when the fusion occurs during development, patients can display complete maxillomandibular-zygomatic fusion, or bony fusion of the maxillary ridge to the mandibular ridge.

Genetics

To date, specific mutations in humans have not yet been identified in association with syngnathia. However, inheritance has been reported in a few cases. In one case, the patient was born to consanguineous parents and had a sibling with a similar defect (Ugurlu et al., 1999), consistent with an autosomal recessive mode of inheritance. Another case of a newborn with syngnathia who was also born of consanguineous parents further supports autosomal recessive inheritance (Villanueva-Garcia, Contreras-Masse, Villa-Guillen, Ramon-Garcia, & Murguia-De Sierra, 2009).

Mouse Models and Mechanism

Although no causative loci have been identified in human patients with syngnathia, three mutant mouse models have been described exhibiting syngnathia: Foxc1^−/−, Fgf8^Δ2,3/neo, and Wnt1Cre;pMes-Bmp4 (F. He et al., 2014; Inman et al., 2013). Foxc1 is a member of the forkhead box winged helix transcription factor family and in mice Foxc1 plays a role in meningeal, calvarial, ocular, somitic, and renal development (Carlsson & Mahlapuu, 2002; Hannenhalli & Kaestner, 2009). Foxc1^−/− null mouse embryos have defects in maxillary and mandibular structures, along with bony syngnathia and agenesis of the temporomandibular joint (TMJ) which is the functional jaw joint essential for jaw articulation (Fig 4 D,E) (Inman et al., 2013). More specifically, syngnathia is present in the form of bilateral fusion of the maxillary zygomatic complex to the dentary bone. Foxc1^−/− embryos additionally exhibit defects in palate formation with a slightly higher arch and malformed rugae, as well as facial muscle patterning abnormalities. These phenotypes are consistent with defects in the development of the NCC derived mesenchyme of PA1. Indeed, Foxc1^−/− embryos display defects in the osteogenic differentiation of the NC.

Fgf and Bmp signaling pathways play critical roles in PA1 patterning and development. Fgf8 expression in PA1 is reduced in the absence of Foxc1 and severely hypomorphic Fgf8^null/neo mice also exhibit syngnathia (Inman et al., 2013). Moreover, Foxc1 genetically interacts with Fgf8 such that decreasing Fgf8 dosage results in increasingly severe syngnathic phenotypes. Downstream of Focx1-Fgf8 signaling, localization of Dlx2 and Dlx5, two members of the Dlx transcription factor family that are necessary in the mesenchyme of PA1 for the proper pattern and morphology of the jaw, is altered in Foxc1^−/− embryos. Taken together, syngnathia and TMJ agenesis in Foxc1^−/− embryos are both due to perturbed Fgf8 signaling in PA1. This defect leads to alterations in Dlx2 and Dlx5 activity, resulting in regionalized patterning changes in the maxillary and mandibular mesenchyme that ultimately cause defects in development of the upper and lower jaw (Inman et al., 2013).

The third mouse model of syngnathia, Wnt1Cre;pMes-Bmp4 mice, results from transgenic overexpression of Bmp4 in the NCC (F. He et al., 2014). Bone morphogenetic proteins (BMPs) are secreted proteins associated with the extracellular matrix and regulate several developmental processes including bone and cartilage formation (Brazil, Church, Surae, Godson, & Martin, 2015). Wnt1Cre;pMes-Bmp4 mice exhibit bony syngnathia and TMJ agenesis in addition to several other craniofacial deformities including a shortened snout, open eyelids, absence of external ears and cleft palate (F. He et al., 2014). Further examination revealed enhanced osteoblast differentiation that could explain the jaw fusion. Similar to Foxc1^−/− mutant embryos, expression of patterning genes such as Dlx2, Hand2 and Msx1 is altered in PA1, likely due to reduction of Fgf8 expression in the oral ectoderm.

Much has been learned from these mouse models regarding the molecular mechanisms underlying the pathogenesis of syngnathia. However, the spatiotemporal requirement for Foxc1 and Bmp4 in musculoskeletal development still needs to be examined. Recently a Foxc1 conditional allele was generated that could potentially address this issue (Seo et al., 2012). In addition, it will be informative to determine which genes Foxc1 regulates in the cranial mesenchyme and the oral ectoderm of PA1 using an unbiased approach like RNA sequencing to further our understanding of jaw development and the pathogenesis of jaw disorders. The rarity of syngnathia and especially isolated syngnathia make it challenging for human geneticists to isolate responsible loci. However, analysis of mutations in Foxc1, Bmp4 and Fgf8 and their downstream targets provide a promising starting point.

Conclusions and Perspectives

During the past fifteen years, completion of the first human genome sequence followed by rapid advances in sequencing has accelerated the discovery of the underlying etiology of human genetic disorders. In parallel, mouse models of human congenital disorders have been an instrumental source of knowledge and discovery and resulted in extraordinary findings regarding the cellular basis and pathogenesis of rare craniofacial conditions. Ribosomopathies, the class of diseases caused by disruption in ribosome biogenesis, serve as a classic example of a global cellular process with tissue specific effects and functions. Mutations in different ribosomal genes manifest in distinct phenotypes and are classified as unique human syndromes, which is extraordinary given how critical ribosomes are to the basic functioning of every cell. Even more surprising is the fact that humans exhibiting defective ribosome biogenesis survive with malformations restricted to certain parts of their bodies, with the craniofacial skeleton being an important example. The existence of specialized ribosomes could help account for the spatiotemporal specificity of the deformities observed in ribosomopathies. There are more than 50 RPs and 4 different rRNAs that make up the ribosome, and it was recently suggested that the composition of ribosomes in different tissues could result from different combinations of RPs (Xue & Barna, 2012). Knocking-out specific RPs results in tissue specific developmental defects, demonstrating that not all ribosomes are the same (Xue et al., 2015; Kondrashov et al., 2011). Moreover, RPs translate specific mRNAs adding yet another level of gene expression regulation not previously appreciated. Therefore, mutations in RPs associated with DBA could elicit their tissue specific effects by failing to translate their target mRNAs. Although, current DBA mouse models do not represent the human condition very well, it would be interesting to determine the ribosome profiles in these mutants and explore whether translation of only specific mRNAs is affected.

TCS is a bit more challenging to explain because Tcof1 and presumably Polr1c and Polr1d play a primary role during the transcription of rRNAs, which is the first step in ribosome biogenesis. One would expect that failure to transcribe rRNA would be incompatible with life. However RNA Pol I could also be “specialized” and its composition, like the ribosome, may also be tissue-specific. Another possibility, although less likely, could be that the rRNA composition of ribosomes, similarly to the RP composition, is distinct in each tissue. Treacle is part of the Pol I transcription initiation machinery, but not Pol III. Thus, presumably transcription of only the 5.8S, 18S, and 28S are affected when Treacle is disrupted, and not 5S which is transcribed by Pol III. This disruption is likely to change the rRNA composition of the ribosome. A simpler hypothesis is that NCC are simply more sensitive to protein translation levels than other tissues. The generation of mice carrying mutations in Polr1c and Polr1d will be helpful in elucidating the role of rRNA transcription in ribosomopathies and craniofacial development.

Although more pleiotropic, ciliopathies fall into the same category of disruption of a global process with tissue-specific effects, because cilia are found in nearly every cell in the human body. Although significant discoveries have been made in the last decade regarding cilia and the cellular basis of ciliopathies using mouse models, much remains to be learned regarding their composition, the signaling pathways they help transduce between cells, and their uniqueness that translates into tissue specificity. Moreover, the role of cilia in the NCC needs to be further investigated to pinpoint at which stage(s) of CNC development each ciliary protein or process is needed. This analysis is now possible with the availability of several mouse models of human ciliopathies and other ciliopathic mouse mutants, many of which carry conditional alleles that can be used in a tissue-specific or a temporal manner.

It is also remarkable that some rare syndromes are caused by mutations in the same genes and are thus allelic. Meckel-Gruber (MKS) and Joubert (JBTS) syndromes are a classic example of allelic syndromes and several mouse models of these conditions exist. The use of CRISPR/Cas9 technology to generate mouse mutants carrying the exact mutations that human patients have will be helpful in elucidating the distinct effects that different mutations in the same gene have on cilia and ultimately in the cellular pathogenesis of these ciliopathies.

In summary, the use of mouse models has been invaluable in decoding the cellular mechanisms and pathogenesis of rare craniofacial disorders in humans including ribosomopathies, ciliopathies, and the rarest of the congenital conditions like syngnathia. The advantage of mouse models in their application to understanding human disease resides in their comparative mammalian anatomy and physiology, together with a sophisticated array of genetic tools for exploring gene function in a precise spatiotemporal manner. Much is still to be learned, but the future looks promising with improved phenotyping in humans and new gene editing approaches for mouse models.