Introduction

Craniosynostosis, which always involves the premature fusion of one or more of the neurocranial sutures and can include many associated dysmorphologies of the craniofacial complex, is a relatively common congenital malformation [1]. The incidence of craniosynostosis is estimated to be in the range of 1 in 2,000–2,500 live births and occurs in all ethnic groups [2–5]. Approximately 85 % of cases are nonsyndromic, and 92 % are non-familial. The other 15 % of cases are defined as having one of the more than 180 known craniosynostosis syndromes, at least 50 % of which follow a Mendelian pattern of inheritance [6, 7].

The frequency of fusion of each of the cranial vault sutures varies. Sagittal synostosis, the most common of the isolated craniosynostoses, occurs in 45–58 % of all craniosynostoses with males more often affected than females (M:F ratio of 3.5:1) [3, 8]. The fusion of the midline sagittal suture results in scaphocephaly, a skull shape that is relatively longer than normal along the anterior-posterior axis and narrowed mediolaterally (Fig. 1a). However, there is a great variability in the scaphocephalic morphology of sagittal synostosis that involves not only the cranial vault but also the facial skeleton and cranial base [9]. Little is known about the sources of this heterogeneity.

Fig. 1
figure 1

Variation in cranial vault shape depicted by three-dimensional computed tomography (CT) reconstructions of infants with various types of single-suture craniosynostosis (superior and inferior (i.e., ectocranial) views; anterior aspect of skull at top, posterior aspect of skull at bottom). Those pictured are cases of single-suture craniosynostosis from our archive of 3D CT images and include skulls of infants with a sagittal craniosynostosis; b bicoronal, right unicoronal, and left unicoronal craniosynostoses (from top to bottom); c metopic craniosynostosis; d bilateral, right unilateral, and left unilateral lambdoidal craniosynostoses (from top to bottom); e unaffected individual. The black arrow heads indicate the sutures that are prematurely closed. Though most attention has been focused on cranial vault shape, the cranial base and facial skeleton are also dysmorphic in craniosynostosis conditions. Not to scale

Full size image

Coronal synostosis occurs in 20–30 % of all cases of craniosynostosis with females more often affected than males (M:F 1:2) [10, 11]. The overt cranial dysmorphology of coronal craniosynostosis varies depending upon whether premature closure of the coronal suture occurs bilaterally involving both the right and the left side of the skull (resulting in brachycephaly) or unilaterally (resulting in anterior plagiocephaly or asymmetry) (Fig. 1b). Unilateral coronal synostosis occurs twice as often as bilateral coronal synostosis. Progressive frontal plagiocephaly or flattening sometimes results from the fusion of the fronto-sphenoidal or fronto-zygomatic sutures and is detected by the detailed three-dimensional computed tomography (3D-CT) imaging of the basilar coronal ring sutures involving the ethmoid-sphenoidal sutures [12, 13]. Metopic synostosis, resulting in trigonocephaly (Fig. 1c), had an estimated prevalence of 6–7 in 100,000 live births prior to 2000, but in the past decade its presentation in some medical centers in Europe and the US has increased as much as fourfold for unknown reasons [14, 15]. Approximately 67 % of all metopic cases are nonsyndromic, and 92 % are non-familial [16]. Male to female ratio has been estimated to be about 3:1. Lambdoid synostosis, resulting in posterior plagiocephaly (Fig. 1d), is estimated to represent about 1 % of all craniosynostosis [1].

Candidate Gene Mutations Found in Nonsyndromic Craniosynostosis

Causative mutations for craniosynostosis have been primarily identified in coronal craniosynostosis syndromes most often within the FGFR1, FGFR2, FGFR3, TWIST1, and EFNB1 genes, but the list of new genes involved in less common syndromes is growing (Table 1), reviewed by Passos-Bueno et al. [17], Wilkie et al. [18], and Jabs and Lewanda [19], and there is an association between the pattern of facial dysmorphogenesis and causative mutation for some of these syndromes [20∙]. However, the genetic etiology of nonsyndromic craniosynostosis (NSC) remained poorly understood until very recently. Over the last two decades, the search for genetic mutations underlying NSC has focused on "hotspots" of genes that are known to cause syndromic craniosynostosis [21]. Rare mutations in FGFRs, TWIST1, LRIT3, ALX4, IGFR1, EFNA4, RUNX2, and FREM1 have been reported in a minor fraction of NSC cases (Table 1).

Table 1 Genes and pathways in syndromic and nonsyndromic craniosynostosis
Full size table

An example of a successful identification of a single point mutation in a candidate gene is the fibroblast growth factor receptor 3 (FGFR3) Pro250Arg mutation associated with individuals initially diagnosed with isolated coronal craniosynostosis [22, 23]. The identification of the FGFR3 Pro250Arg mutation resulted in the definition of Muenke syndrome [23] characterized by a highly variable phenotype with some individuals appearing phenotypically normal [24] demonstrating a reduced penetrance of the mutation at about 80 % [18]. It has been estimated that the FGFR3 Pro250Arg mutation may account for 4–12 % of isolated unilateral and 30–40 % of isolated bilateral coronal synostosis cases [25, 26] with a population prevalence of about 1 case per 30,000 [18].

Other rare gene mutations have been identified for isolated synostosis [27]. However, some of these mutations were present not only in the affected probands, but also in other members of the family, who had craniofacial dysmorphisms (but not craniosynostosis) or were unaffected suggesting incomplete penetrance [28–32]. In one case of sagittal NSC, the FGFR2 Ala315Thr mutation was reported [33]. Two cases with sagittal NSC were found to carry Ser494Thr and Cys592Tyr mutations in LRIT3, a protein believed to regulate maturation and signaling of FGFR1 [28]; another three sagittal NSC cases had Val7Phe, Lys211Glu, and Pro306Leu mutations in ALX4, a homeobox containing transcription factor regulating calvarial development through interactions with Wnt and bone morphogenetic proteins (BMPs) [31]. TWIST1 mutations, Ala186Thr, Ser201Tyr, and Ser188Leu in the TWIST Box domain, were found in two cases of the isolated sagittal synostosis and in one case of isolated left coronal synostosis, respectively [29, 30]. Insulin-like growth factor 1 receptor (IGF1R) mutations, R406H, N857S, and R595H, were found in two cases of isolated sagittal and one with coronal synostosis, and rare variants P190S and M446V were also detected [34]. For coronal NSC, an FGFR2 Ala315Ser mutation was reported in a patient with unicoronal synostosis and a birth history of breech presentation and skull compression [35]. EFNA4 His60Tyr, Pro117Thr, and Asn157LysfsX45 mutations have been reported in three patients with coronal NSC [32]. An FGFR1 Ile1300Trp mutation was found in one case of metopic NSC with facial skin tags [36]. Recently, a 1.1 Mb duplication encompassing RUNX2 and mutations in FREM1 has been associated with metopic NSC [37, 38]. Apart from IGFR1 and FREM1, the above genes can be linked directly to TWIST1 and the FGF signaling pathway, which ultimately interact to control the entry of mesenchymal cells into osteoblastic differentiation in the developing suture; it is speculated that FREM1 may also bind FGFs to modulate the FGF pathway [38].

Novel Genes Identified in Craniosynostosis Using Genomic Technologies

In the past 2 years, important breakthroughs have been achieved with the identification of new genes associated with sagittal and coronal NSC. Justice et al. [39∙∙] reported susceptibility loci for sagittal NSC near BMP2 and BBS9. Twigg et al. [40∙∙] showed that the reduced dosage of ERF causes sagittal, lambdoid, and multisuture craniosynostoses in cases diagnosed as isolated or syndromic craniosynostosis. Finally, Sharma et al. [41∙∙] identified mutations within TCF12 associated with unilateral and bilateral coronal craniosynostosis in patients with isolated or syndromic craniosynostosis. Here, we review these mutations in genes that may account for a significant number of patients diagnosed with NSC.

Susceptibility Loci for Sagittal NSC Near BMP2 and BBS9

Though sagittal NSC is the most frequent form of craniosynostosis, the genetic basis for most cases is unknown, and only rare gene mutations have been identified until recently [27]. Justice et al. [39∙∙] conducted a genome wide association study (GWAS) of 130 non-Hispanic case-parent trios of European ancestry followed by the replication analysis of 172 unrelated non-Hispanic Caucasian cases and 548 controls to identify susceptibility loci for sagittal NSC near BMP2 and within BBS9. The discovery/replication meta-analysis demonstrated the combined odds ratios of 4.38 (95 % CI 3.51–5.45; P = 1.1 × 10−39) and 0.24 (95 % CI 0.17–0.32; P = 5.6 × 10−20), respectively. BMP2 is a member of the TGF-β superfamily and a key growth factor regulating osteoblast development [42]. The BMP and FGF pathways interact and are important in skull growth [43–45]. BBS9 is a member of the BBSome, a multiprotein complex localized in the primary cilium that is involved in coordinating many developmentally important signaling pathways including platelet-derived growth factor receptor α, sonic hedgehog, and Wnt [46, 47]. The BBSome is also implicated in intraflagellar transport [48]. BBS9 loss-of-function mutations have been found in Bardet–Biedl syndrome (BBS) patients [49]. Although BBS-affected individuals do not present with suture phenotypes, there are ciliopathy conditions that have craniosynostosis as a feature such as cranioectodermal dysplasia (Sensenbrenner syndrome) [50–53].

As of this writing, no study phenotypically characterizing sagittal NSC cases with or without BMP2 or BBS9 variations has been published. Although the morphology of the cranial vault has been observed as a defining characteristic in craniosynostosis, qualitative assessments of calvarial dysmorphology have shown consistent variability in NSC (Fig. 1) [54], and the exact source of this variation remains unknown. A recent quantitative study of craniofacial shape in 43 infants with nonsyndromic sagittal synostosis using 3D-CT reconstruction and morphometric methods confirmed variation in cranial vault morphology [9]. In all cases studied, the central portion of the sagittal suture was the first to fuse (probably prenatally), and at least two different developmental paths toward complete fusion of the sagittal suture exist either in the anterior section or in the posterior section of the sagittal suture being the second to fuse. The analyses showed association between the variation in craniofacial shape and the exact path of fusion of the sagittal suture. Comparable morphometric studies should be completed using cases carrying either BMP2 or BBS9 NSC-associated variants to determine whether or not these genetic influences correspond with specific phenotypes.

ERF and TCF12 Mutations in Patients with Craniosynostosis

Using whole exome sequencing of seven unrelated individuals with bilateral coronal synostosis and negative for previously described mutations [55], two additional genes, ERF and TCF12, were found to have mutations in two or more patients with craniosynostosis [40∙∙, 41∙∙]. Heterozygous mutations in ERF, an inhibitory ETS family transcription factor that is negatively regulated by the extracellular signal-related kinases 1 and 2 (ERK1/2) of the mitogen-activated protein kinase (MAPK) signaling pathway, were found [40∙∙]. Some of the patients with the ERF mutations were syndromic with sagittal or multisuture synostosis, craniofacial dysmorphology, Chiari malformation, and language delay, and others were nonsyndromic with sagittal, unilateral or bilateral lambdoid, and multisuture synostosis. ERF was found to bind close to regulatory sites recognized by RUNX2, an essential regulator of osteoblast differentiation, and could interfere with transcriptional activation by RUNX2. Thus, loss-of-function mutations of ERF in these patients have a similar effect to FGFR-phosphorylated ERK activation observed in FGFR-related craniosynostoses. Reduced ERF function in these conditions can result in the upregulation of RUNX2 activity, leading to changes in osteoblast differentiation and potential premature ossification of cranial sutures.

Heterozygous mutations in TCF12, transcription factor 12, were also identified in syndromic and nonsyndromic patients with unilateral and bilateral coronal and multisuture synostoses [41∙∙, 56]. Syndromic patients had additional features of craniofacial dysmorphism and external ear and minor limb anomalies. TCF12 mutations were found in 32 % of subjects with bilateral and 10 % with unilateral coronal synostosis. TCF12 heterodimerizes with class II basic helix-loop-helix transcription factors including TWIST1. Loss-of-function mutations in TWIST1 have previously been shown to cause the Saethre–Chotzen syndrome, a craniosynostosis condition with coronal fusion and minor limb anomalies [57, 58]. The TCF12-TWIST1 heterodimer is likely to regulate the specification of the coronal suture between the neural crest-derived frontal bones and mesoderm-derived parietal bones [32, 59]. As in the case of ERF, these dimers may inhibit osteogenic differentiation via actions on RUNX2 and FGFR signaling pathways [60].

Genetic Risk

To estimate the proportion of craniosynostosis patients that screen positive for a gene mutation, a study was conducted on 326 children, who were born from 1993 to 2002 and required surgical treatment in a craniofacial unit in Oxford, England [18]. Genetic diagnoses were made for 21 % of all craniosynostosis cases, and the FGFR3 P250R mutation was the single most common mutation, accounting for 24 % of cases with genetic diagnoses (5 % of all cases). Those with genetic diagnoses were associated with increased rates of many complications. Children with the clinical diagnosis of nonsyndromic unicoronal or bicoronal synostosis were more likely to have an identified causative mutation than those with other sutural involvement. In the extended Oxford birth cohort (cases born from 1998 to 2006), TCF12 mutations were identified in approximately 1.0 % of craniosynostosis cases [41]. While the patients with TCF12 mutations had a more benign course than patients with FGFR3 P250R or TWIST1 mutations, 14 % had developmental delay or learning disabilities and two were diagnosed with autism. ERF mutations explain an additional 1.2 % of etiology for the cohort [40].

Cellular and Animal Model Investigations of Craniosynostosis

Both the genome wide expression analysis of primary osteoblasts derived from craniosynostosis patients and the creation of mutant mouse models allow experimental analysis of craniosynostosis phenotypes and the roles of newly-discovered genes in craniosynostosis. Gene expression profiling of human craniosynostosis samples has been recently reviewed [61]. As an example, a recent survey of 199 NSC patient-derived osteoblasts, including sagittal, metopic, and coronal cases, suggested the common involvement of FGF7, SFRP4, and VCAM1 and the role of extracellular matrix interactions in the craniosynostosis phenotypes [62].

Mouse models of activating Fgfr mutations and of Twist1 loss-of-function have been invaluable in understanding the coronal synostosis and the role of the neural crest/mesoderm boundary forming this suture [63∙, 64]. Erf was shown to be expressed within calvarial sutures in the mouse, and the conditional deletion of Erf demonstrated that loss of Erf was causative for craniosynostosis [40]. While heterozygous null Tcf12 mice alone did not show craniosynostosis, reduction of Tcf12 significantly increased the incidence and severity of craniosynostosis in Twist1 heterozygous null mice, supporting the model of Tcf12 interaction with Twist1 [41]. Gene expression within sutures can be readily determined in mice, and this knowledge is crucial in understanding the connection between gene mutations and specific patterns of suture fusion or other resulting craniofacial dysmorphologies, exemplified by Fgfrs1–3 and Twist1, which have distinct expression patterns within sutures [65]. The restriction of craniosynostosis to specific sutures may reflect the suture-specific expression of some genes. For example, in the mouse model of Greig cephalopolysyndactyly syndrome, the Gli3 (Xt-J/Xt-J) mouse, the lambdoid sutures fuse and interfrontal suture development is anomalous, and these are the sites of strong embryonic calvarial Gli3 expression [66]. A comprehensive knowledge of murine sutural gene expression, particularly at embryonic stages, would facilitate the identification of human craniosynostosis candidate genes. Furthermore, the mouse provides a model for the potential therapeutic amelioration of craniosynostosis. Chemical inhibition of Fgfr tyrosine kinase activity, or of the activity of effector kinases downstream of Fgfrs, results in reduced craniosynostosis in mouse models of Crouzon, Apert, and Beare-Stevenson syndromes [67–70].

Conclusions

The etiology of approximately three quarters of patients diagnosed with craniosynostosis is not known. Thus, much work is needed in the elucidation of causal mutations proximate to the BMP2 and BBS9 gene loci and in identifying new genes involved in craniosynostosis. Since craniosynostosis is a complex heterogeneous condition, the interplay between genetic variants and environmental exposures may explain the low heritability for NSC, and their combined action may elucidate processes underlying the variable dysmorphogenesis of the facial skeleton and cranial base, portions of the skull that do not encompass the cranial vault sutures. These factors may also contribute to the lack of distinct disease phenotypes across diagnostic groups and the current lack of identified molecular causes in some cases. Mutations in genes initially identified in syndromic cases may contribute to causation in milder phenotypes including nonsyndromic or non-penetrant cases. The large population of patients with nonsyndromic craniosynostosis embodies a fundamental need for more work, as well as a fertile research area for the discovery of novel genetic, maternal, and environmental factors. Harnessing next generation sequencing technology and bioinformatics analysis with our understanding of the genome, transcriptome, and epigenome will help to elucidate the etiology of craniosynostosis, both syndromic and nonsyndromic.

As noted previously, although sutural fusion is the most frequent feature studied and treated, craniosynostosis also refers to the abnormal development of the bones of the skull associated with dysmorphic skull shape. In animal models for human craniosynostosis syndromes, the abnormal skull shape can be detected before the premature closure of cranial vault sutures 71∙, 72∙]. The development of animal models for craniosynostosis [70, 73, 74] has already revealed many molecularly driven three-dimensional morphological changes in soft tissues of the head and skull that were not apparent in humans [71∙, 75, 76∙, 77]. These changes are more difficult to evaluate quantitatively in humans where observations are routinely made postnatally and there is a lack of appropriate morphological control data sets to make meaningful comparisons to abnormal phenotypes. Human cases provide access to population-based molecular screens, and more recently the chance to link genotype with phenotype [20], but do not provide easy access to the molecularly-based processes that result in the highly variable, abnormal cranial morphology of syndromic and nonsyndromic cases of craniosynostosis. This access will be necessary in order to understand what unites the various molecular causes of craniosynostosis (Table 1) at the genomic and phenotypic levels. Mouse models provide access to both the processes that underlie these integrated sets of anomalies and the networks that produce them. Emerging technologies (e.g., optical projection tomography [78]) allow direct study of the correspondence between the spatiotemporal dynamics of gene expression patterns, morphogenesis, and morphological diversity.

The coordinated assimilation of results from human- and animal model-based research that build on respective discoveries is crucial to understand the variation in the integrated anomalies that together define craniosynostosis conditions. These associated anomalies contribute to additional health issues critical to effective clinical care of people with craniosynostosis conditions including type and timing of surgery, treatment of comorbidities, and long-term effects on neuropsychological aspects and quality of life. Future research applied to large molecular datasets, analysis of pathways and networks, and the complexity of the craniosynostosis phenotype will require integrative analyses by multidisciplinary teams of physicians and scientists including system and developmental biologists, quantitative anatomists, epidemiologists, geneticists, medical specialists, and surgeons.