|Home | About | Journals | Submit | Contact Us | Français|
Clefts of the lip and/or palate (CLP) are common birth defects of complex etiology. CLP can occur in isolation or as part of a broad range of chromosomal, Mendelian, or teratogenic syndromes. Although there has been marked progress in identifying genetic and environmental triggers for syndromic CLP, the etiology of the more common non-syndromic (isolated) forms remains poorly characterized. Recently, using a combination of epidemiology, careful phenotyping, genome-wide association studies and analysis of animal models, several distinct genetic and environmental risk factors have been identified and confirmed for non-syndromic CLP. These findings have advanced our understanding of developmental biology and created new opportunities for clinical translation research.
Clefts of the lip and/or palate (CLP) are immediately recognizable disruptions of normal facial structure. Although not a major cause of mortality in developed countries, CLP does cause considerable morbidity to affected children and imposes a substantial financial risk for families with a concomitant societal burden1. Individuals with CLP may experience problems with feeding, speaking, hearing and social integration that can be corrected to varying degrees by surgery, dental treatment, speech therapy and psychosocial intervention. CLP is etiologically heterogeneous and this has critical implications for understanding the biology of facial development, how environmental risks interact with genetic factors and how we can incorporate known etiologic variables to improve clinical care. Recent successes in genome-wide linkage and association studies have identified novel loci significantly associated with CLP2–6. Researchers are currently striving to identify the etiologic variants at these novel loci to understand the developmental disturbances leading to CLP, and this knowledge should eventually result in improved prevention, treatment and prognosis for individuals with these conditions.
Development of the lip and palate is outlined in Figure 1. The common forms of CLP involve disruption of tissue planes above the lip extending into the nares and/or the palate (hard and/or soft) (Figure 2a, 2b). Fogh-Andersen and Fraser noted clefts involving the anterior structures (lip and primary palate) could be separated on both genetic and embryologic grounds from those involving only the secondary palate7, 8. While there are many disruptions affecting the craniofacial complex, the overwhelming majority involve only the upper lip and/or palate. Further, approximately 70% of cases of CLP occur as isolated entities with no other apparent cognitive or structural abnormalities; commonly termed “isolated, non-syndromic CLP”. Because the defects arise early in embryological development, have a complex etiology with both genetic and environmental contributions and modest recurrence rates, it has proven difficult to identify specific etiologic factors. A combination of epidemiologic, candidate gene and genome-wide studies, plus analysis of animal models, has recently provided deeper insights into the causes of non-syndromic CLP.
With the advent of the genomics era, there have been major advances in identifying the causative genetic mutations underlying syndromic forms of CLP (http://www.ncbi.nlm.nih.gov/omim). In contrast, owing to its genetic heterogeneity, departure from Mendelian inheritance patterns, the lack of (and expense of) genomic tools and the necessity for very large datasets, there has been less progress in advancing our understanding of the genetic etiology of non-syndromic CLP. However, the recent development of innovative approaches to phenotyping and powerful genomic tools, together with extrapolation from studies of syndromic forms of CLP, has increased our understanding of non-syndromic CLP. Because of its particular challenges, in this review we focus on non-syndromic CLP and we summarize syndromic forms (which are genetically tractable) only briefly. We discuss important epidemiologic clues, environmental contributions, genetic architecture and issues of phenotyping, gene discovery and insights into molecular pathogenesis. We also speculate about the clinical implications of these findings for recurrence and new clinical associations building on advances in imaging and large databases to examine long-term outcomes.
CLP affects approximately 1/700 live births, with wide variability across geographic origin, racial and ethnic groups, as well as environmental exposures and socioeconomic status. In general, Asian and Amerindian populations have the highest reported birth prevalence rates, often as high as 1/500, European-derived populations have intermediate prevalence rates at about 1/1000, and African-derived populations have the lowest prevalence rates at about 1/2500. These observations suggest the relative contribution of individual susceptibility genes may vary across different populations6, 9, 10. The frequency of CLP also differs by sex and laterality: there is a 2:1 male to female ratio for clefts involving the lip and approximately a 1:2 male to female ratio for clefts of the palate only; and there is a 2:1 ratio of left to right sided clefts among unilateral cleft lip cases.
Historically, CLP has been divided into cleft palate only and cleft lip with or without cleft palate (CL/P).7, 8 However, recent epidemiologic data suggest that cleft lip only may have unique etiologic features, including strong genetic associations, while some individuals with cleft palate only show evidence of sub-clinical cleft lip11–14. Nevertheless, this broad sub-division of anatomical defects is consistent with the distinct developmental origins of the lip/primary palate and the secondary palate. Furthermore, separate cellular and genetic etiologies for CL/P and cleft palate only are consistent with the general observation that these two conditions do not segregate in the same family, although exceptions have been reported for families with etiologic mutations in specific genes (for example, tumor protein p63 (TP63), msh homeobox 1 (MSX1),interferon regulatory factor 6 (IRF6) and fibroblast growth factor receptor 1 (FGFR1)).15–19 Approximately 70% of all cases of CL/P and 50% of cases of cleft palate only are considered to be non-syndromic.20–22 The remaining cases are composed of a wide range of malformation syndromes, including over 500 Mendelian syndromes (see Online Mendelian Inheritance in Man (OMIM) for further information www.ncbi.nlm.nih.gov/omim/), as well as those arising secondary to chromosomal or teratogenic effects. These syndromic forms are somewhat more tractable to genetic analysis, and Table 1 provides a summary of a subset in which the underlying genetic mutation has been identified.
While twin studies and familial clustering studies have provided compelling evidence for a genetic component to non-syndromic CLP,23 few pedigrees show clear-cut Mendelian inheritance and most cases appear sporadic.24 Moreover, CLP is known to be influenced by environmental risk factors;25, 26 consequently, a multifactorial model of inheritance is favored in which genetic risk factors of small individual impact may interact with environmental covariates.12 These combined factors complicate genetic analysis of non-syndromic forms of CLP.
Accurate phenotyping is crucial to understanding both the epidemiology and etiology of any congenital malformation, because the power to detect effects is weakened when heterogeneous groups are treated as a single entity. Although clefts of the lip and palate show a range of phenotypic expression (Figure 2A, 2B), they are generally defined as qualitative traits (i.e. affected or unaffected). Dividing CLP in this simplistic way has the potential to lose important information. For example, different patterns of genome-wide linkage are observed when multiplex families are divided into subgroups depending on the overt CLP phenotypes present in affected individuals, which suggests careful attention to phenotypes will be an important tool in furthering our understanding of the genetic heterogeneity underlying non-syndromic CLP.2 Furthermore, numerous lines of evidence now suggest the spectrum is more complex and should include a variety of sub-clinical phenotypic features observed in either an individual with CLP and/or their “unaffected” relatives.27
Sub-clinical phenotypes can include minor structural variants including lip pits/prints,28 dental anomalies,29 defects of the orbicularis oris muscle,30, 31 3D facial image measurement,27 brain variants as assessed by MRI32, 33 or by surrogate measures34, 35, and speech or cognitive differences such as velopharyngeal insufficiency, reading disability and IQ. Palatal subphenotypes have been less explored but also include bifid uvula, submucous cleft palate, the differentiation of clefts of the hard and soft palate and possibly ankyloglossia. A better understanding of palatal subdivisions by phenotype and pathway will benefit from both human and mouse models. Defects of the orbicularis oris muscle show particular promise for enhancing the search for causative genetic variants and for contributing to clinical risk assessment.30, 36–39 Orbicularis oris defects can be assessed using high-resolution ultrasound of the upper lip (Figure 3A, 3B). Sub-clinical phenotyping therefore holds great promise to enhance the power of family studies and may lead to opportunities for translational research relevant for both clinical care of patients and clinical genetics as a science.
To date, genetic approaches to non-syndromic CLP have included: linkage analysis using large, multiplex families or smaller but inbred families, or analysis of affected relative pairs; association studies using case/parent trios or case-control samples; identification of chromosomal anomalies or micro-deletions in cases; and direct sequencing of affected individuals. These methods can be applied to candidate genes or genome-wide strategies can be used. Each approach has its own advantages and disadvantages, some of which will depend on the underlying genetic architecture of the disease, as well as the realities of economics and technology. We briefly summarize successes using a range of approaches, followed with more detail on the results of recent genome-wide association (GWA) studies. Most studies of non-syndromic clefts to date have focused on cleft lip with or without cleft palate rather than isolated cleft palate. This has been biased perhaps by the larger numbers, easier ascertainment and less confusion from confounding syndromes. Future studies will need to address this gap and also the somewhat counterintuitive observation that more mouse models are available for cleft palate than cleft lip.
Candidate gene studies have been at the core of cleft research since Ardinger and colleagues40 suggested a role for TGFA (transforming growth factor, alpha) variants in risk for non-syndromic CL/P. The identification of candidate genes has traditionally relied on gene expression and developmental analyses performed in model organisms, particularly the mouse, either to first identify the candidate genes or to provide biological plausibility for the association. More recently, extrapolation from the study of syndromic forms of CL/P has proven to be a useful adjunct to this approach. As with candidate gene studies of many complex disorders, rigorous confirmatory replication is not common, with only variants in IRF6 (interferon regulatory factor 6) yielding consistent evidence of association across multiple studies13, 41–44 (discussed further below). Analysis of chromosomal anomalies in patients has proven to be a productive route for identification or confirmation of CL/P loci, with recent successes for FGFR2 (fibroblast growth factor receptor 2)45 and SUMO1 (a member of the small ubiquitin-like modifier family).46–48 Candidate gene-based association studies and analysis of chromosomal anomalies have recently been reviewed in detail26, 49.
There have been many attempts to use linkage analysis to identify regions of the genome likely to carry genes controlling pathogenesis of CLP, and the region surrounding the FOXE1 (forkhead box E1) gene reached genome-wide levels of significance with subsequent fine-mapping and replication.2, 50 There have been several resequencing studies of candidate genes to identify specific variants that might underlie statistical associations with clefting, and the best current evidence has been reported for mutations in MSX1,17, 51 FGFR1 and FGF8,52 and BMP4 (bone morphogenetic protein 4).36 Whole exome sequencing has recently been successful in identifying causative genetic variants for Mendelian traits,53, 54 including Miller syndrome55, (which is an autosomal recessive syndrome that can include cleft palate55) and Kabuki syndrome56, (a dominant disorder than can include cleft palate56), but is yet to be successful for complex and heterogeneous traits such as non-syndromic CLP.
As is now apparent for many common complex disorders, GWA studies have provided recent major advances in our understanding of genes and pathways that play a role in the etiology of CLP. To date, there are three published GWA studies for CL/P using the case-control design,3–5 and one case-parent trio study from an international consortium that is part of GENEVA (the gene environment association studies consortium).6, 57 These studies have mostly excluded cases with cleft palate only, based on likely etiologic heterogeneity. Birnbaum and colleagues3 confirmed the impact of IRF6, which had previously been identified in candidate gene studies,13, 41 and discovered a new region on chromosome 8q24 that gave extremely strong evidence of association in their European case-control sample. Grant and colleagues4,58 independently confirmed that this “gene desert” region on chromosome 8q24 was strongly associated with CL/P in a sample of European American cases and controls. Mangold and colleagues5 subsequently used an expanded dataset from Europe and identified additional loci at chromosomes 10q25 (VAX1, ventral anterior homeobox 1) and 17q22 (NOG, noggin) that achieved genome-wide significance.
The GENEVA Cleft Consortium study6 used case-parent trios from multiple populations and reconfirmed the IRF6 findings, as well as replicating the chromosome 8q24 and 10q25 (VAX1) findings. Interestingly, in this consortium study the level of statistical evidence from markers within chromosome 8q24 was much stronger among case-parent trios of European ancestry, whereas the evidence for linkage and association for markers in IRF6 was much stronger in trios of Asian ancestry. This GENEVA study identified at least two new loci (MAFB and ABCA4) not previously associated with CL/P as significant at the genome-wide significance level, with stronger signals in Asian compared to European populations.6 The signals and this population difference were replicated using independent families from multiple populations (see further details below.
These observations suggest that not only are there multiple genetic variants influencing risk of CL/P, but that some of these genes may be differentially tagged by polymorphic markers in a population-specific manner. For example, in the chromosome 8q24 region the most significant SNP (rs987525) showed similar patterns of over-transmission to the affected child but had a higher minor allele frequency among parents of European ancestry compared to parents of Asian ancestry (0.26 vs. 0.07).6 In fact, the entire region of signal on chromosome 8q24 showed higher rates of heterozygosity among parents of European ancestry compared to those of Asian ancestry, which means that European trios would be far more informative than Asian trios for this region. Therefore, it may be more difficult to identify causal genetic variants in some populations compared to others. Some putative causal genes have been identified through polymorphic markers in most populations (for example, IRF6), while others (for example, 8q24, MAFB, ABCA4) seem to be more population-specific, which could reflect variable coverage by available marker panels or true allelic heterogeneity. True allelic heterogeneity, in which multiple mutations occurred on different background haplotypes, would make it much more difficult to identify causal genes through association studies; however, Dickson and colleagues59 noted there may be mixtures of multiple rare alleles on common haplotypes within a single causal gene for complex and heterogeneous disorders such as CLP.
Below, we provide a short summary of each of the genes confirmed or identified through GWA studies together with insights into the molecular pathogenesis derived from analysis of animal models. In Table 2, we summarize genes with a confirmed role in non-syndromic CLP, those that seem likely to be involved, and some that have been intensively studied but where the supporting data remain less convincing.
While GWA studies will increase the number of CLP loci identified, the move from a GWA study signal to a causative variant will still be demanding. Animal models and gene expression data are powerful tools for identifying candidate genes for complex traits; importantly, they also contribute to our knowledge of normal facial development and the molecular pathogenesis of CLP. The mouse is the pre-eminent model organism for studies of this type as facial development in this species mirrors human craniofacial development, and mouse strains with high rates of CLP are available. A number of excellent reviews have described the cellular and molecular mechanisms underlying normal and abnormal development,60, 61; here we provide examples of how the mouse has impacted our understanding of the molecular pathogenesis of CLP in humans.
Mutations in IRF6 were first identified as etiologic in the autosomal dominant Van der Woude syndrome, which can include CL/P and/or cleft palate only along with dental anomalies and lip fistulas.18 Subsequent research showed common alleles in IRF6 were associated with non-syndromic CL/P.41 This association has been independently replicated in GWA studies as well as in many candidate gene studies3–6, 13, 41–43, 62 with some failures of replication possibly due to population differences.63 Recently, an approach that integrated the identification of cis-regulatory elements using sequence conservation across multiple species, analysis of animal models and biochemical analyses resulted in the identification of one specific sequence variant (rs642961) located within an enhancer ~10 kb upstream of the IRF6 transcription start site that is significantly over-transmitted in non-syndromic cleft lip only.13 Importantly, this apparent risk allele was found to disrupt a binding site for transcription factor AP-2α, which is mutated in the autosomal dominant CLP disorder branchio-oculo-facial syndrome64, therefore strongly suggesting it is a contributory variant.13
A role of IRF6 in CLP is further supported by analysis of animal models. Recent research has shown that Irf6 mutant mice exhibit a hyper-proliferative epidermis that fails to undergo terminal differentiation, which leads to multiple epithelial adhesions that can occlude the oral cavity and result in cleft palate.65, 66 These results demonstrated that IRF6 is a key determinant of the keratinocyte proliferation/differentiation switch and subsequent research indicated that IRF6 also plays a key role in the formation of oral periderm, spatio-temporal regulation of which is essential in ensuring appropriate palatal adhesion.67 Recently, a combination of mouse genetics, expression analyses, chromatin immunoprecipitation and luciferase reporter assays has shown IRF6 is a direct target of p63, which underlies several malformation syndromes that include CLP as a hallmark feature15, 16. p63 activates IRF6 transcription through an enhancer element, variation within which increases susceptibility to cleft lip only.68
The MAFB gene encodes a basic leucine zipper transcription factor. Markers near MAFB achieved genome-wide significance in the GENEVA Cleft Consortium study6, with trios of Asian descent providing much stronger statistical evidence. In independent replication samples, 1149 pedigrees of European ancestry showed evidence of linkage and association with a SNP (rs13041247, p=0.0007) located 260 bp from the SNP yielding the strongest signal among Asian families (rs11696257, p=0.0009 in 331 independent pedigrees). A missense mutation, H131Q, in MAFB was found in 3.5% of Filipinos with CL/P but only 0.7% of controls (P<0.0001). This variant occurs in a region of strongly conserved sequence, suggesting there may be a rare variant in MAFB that contributes to the observed GWAS signal. It is noteworthy the gene-poor regions on either side of MAFB include numerous binding sites for transcription factors known to play a role in palate development (including transcription factors in the MSX, IRF, SOX and BACH gene families). In the mouse, Mafb expression was shown to be strong in the epithelium of the palatal shelves and in the medial edge epithelium during palatal fusion.6
ABCA4 encodes an ATP-binding cassette transporter. Multiple markers in ABCA4 and in the 3’ end of the gene gave evidence of linkage and association at the genome-wide significance level in the GENEVA Cleft Consortium GWA study6, again with stronger evidence among Asian samples. Two of the SNPs with strongest signals were replicated in independent family samples, and one of these SNPs (rs560426) gave a far stronger signal in Asian families (p=0.0003 in 331 pedigrees) compared to European families (p=0.005 in 1149 pedigrees). This difference in the strength of statistical evidence again raises the possibility of either an allele common to both groups but with differing frequencies, or multiple risk alleles occurring on different haplotype backgrounds. ABCA4 is known to cause the autosomal recessive retinal degenerative disease Stargardt’s disease and sequencing of the 50 exons of ABCA4 in 190 CL/P cases identified 27 different missense mutations, many of which have been previously reported in Stargardt’s or other ocular disorders (http://www.ncbi.nlm.nih.gov/omim). Since ABCA4 is surrounded by many other genes, the peak signal in ABCA4 may be a surrogate for etiologic variants in another gene nearby. Furthermore, no Abca4 expression has been seen in mouse palatal shelves around the time of palatal fusion.6
In the studies by Mangold et al.5 and the GENEVA Cleft Consortium6, markers in or near the VAX1 gene at chromosome 10q25 yielded evidence approaching genome-wide significance; the same two SNPs in VAX1 (rs7078160 and rs4752028) were over-represented in CL/P cases in both studies.5, 6 VAX1 encodes a transcriptional regulator with a DNA-binding homeobox domain. Mouse knockouts for Vax1 develop cleft palate and this gene is expressed widely in developing craniofacial structures;69 thus variants in VAX1 itself are strong candidates for contributing to CLP.
Although not as yet implicated by GWA studies, variants within WNT genes have been reported to be associated with non-syndromic CL/P70 and mutations in WNT3 underlie autosomal recessive tetra-amelia with cleft lip and palate71. Although the evidence for the involvement of WNT signaling in non-syndromic CL/P is not strong, these findings have led to further analysis of genes in the Wnt signalling pathway as candidates for normal development of the lip and palate. Targeted mutation of Wnt9b in mice leads to CLP; and the A/WySn strain of mice, which have increased incidence of spontaneous CLP, have insertion of a retrotransposon 6.6 kb downstream of the Wnt9b gene (a site known as the clf1 locus)72 These findings suggest Wnt9b plays a key role in development of the lip.72–74 Further support for this hypothesis arises from the observation that canonical Wnt signaling is activated during midfacial morphogenesis in mice75, and genetic inactivation of low density lipoprotein receptor-related protein 6 (Lrp6), a co-receptor of the Wnt/β-catenin signaling pathway, causes CLP76. Intriguingly, Msx1 and Msx2 (see below) have been shown to be downstream targets of this Wnt/β-catenin signaling pathway during lip formation and fusion.76
As in humans, loss-of-function mutations in the homeobox gene Msx1 result in cleft palate in mice.77 Msx1 is a downstream target of BMP signaling in a number of embryonic sites and Msx1 is necessary for expression of Bmp4 and/or Bmp2.78 In mice, loss-of-function of type I Bmp receptor (Bmpr1a) in the craniofacial primordia resulted in CL/P, while deficiency of Bmp4 resulted in cleft lip only79; this shows that Bmp signaling has distinct functions in development of the lip versus the secondary palate. In the context of Bmp4 deficiency, all Bmp4 mutant embryos exhibited bilateral cleft lip at E12, but only 22% still displayed cleft lip at E14 suggesting some in utero repair mechanism.79 These observations parallel the findings that mutations in BMP4 may underlie a subset of cases of subepithelial, microform and overt cleft lip in humans.36
Identification of environmental components of clefting and studies of gene by environment interaction require large (ideally prospective) cohort studies and access to genetic material to be optimally effective. While a few such resources are available (Denmark, Norway, the National Birth Defects Prevention Study in the US)11,14, GENISCA ref they are still primarily in the analysis phase. Nonetheless there are a few studies that have begun to provide data on environmental risks. Since the environment is more malleable identification of environmental risks, particularly if they can be personalized with genetic covariates, afford the best short term opportunities to be applied to prevention.
Maternal smoking has been associated repeatedly with increased risk of CLP and meta-analysis strongly supports an overall odds ratio (OR) for having CLP of ~1.3 among offspring of mothers who smoke.80–82 Increased risks from exposure to maternal smoking during the peri-conceptual period raises the possibility that genes in certain metabolic pathways may play a role in the development of CLP. Specifically, markers in the GSTT1 (glutathione S-transferase theta) or NOS3 (nitric oxide synthase 3) genes appear to influence risk of CL/P in the presence of maternal smoking.81, 83–85 The GSTT1 markers are gene deletion variants, which suggests deficiencies in detoxification pathways may underlie some of this susceptibility. Smoking has also been recently associated with a joint risk with variants in the IRF6 gene86 and the same study reported interactions between multivitamins and IRF6 variants. These findings provide evidence that gene-environment interactions are important in CLP. In addition, some specific teratogens25,26 ADD ABBOTT HERE, for example valproic acid, have yielded evidence of association with cleft palate.87
Exposure to maternal alcohol consumption has also been suggested as a risk factor, but the evidence has been more inconsistent.26. Studies also suggest that ‘binge’ drinking patterns (high doses of alcohol in short periods of time) increase risk88, and this is supported by associations with variation in the ADH1C alcohol dehydrogenase gene.89 However, these links to alcohol consumption remain to be confirmed. Nutritional factors, such as folate deficiency, have also been suggested to influence risk of CL/P, based on both observational studies and interventional trials using folate supplementation to prevent recurrences of CL/P in families.90 However, the studies of vitamin supplementation with folate remain controversial1, 91 and recent studies of levels of folate receptor antibodies did not find an association with CL/P92. Furthermore, food fortification programs using folic acid have shown detectable decreases in the rates of clefting in some93, 94 but not all95, 96 studies. In the future, other nutrient and micronutrient studies will need to be expanded to look for evidence of effects. For example, there are some data to support roles for zinc deficiency in risk of oral clefts in populations in which zinc status is highly compromised97, for cholesterol deficiency in facial clefting,98 as well for as multivitamins in general in cleft prevention.94
Besides nutrients and toxins other environmental exposures have been, and should continue to be, assessed for possible roles in clefting. These exposures include hyperthermia,99 stress, maternal obesity, occupational exposures, ionizing radiation and infection10. Pregnancy planning has been shown to have a protective effect and the basis of this observation needs to be more deeply explored.100 Nonetheless there is no consensus yet on the harmful effects of these factors and prospective cohort studies large enough to measure effects on a relatively rare disorder such as clefting may be required. A particular challenge will be to determine the specificity of the role of an exposure in contributing to clefting, as many exposures will have both identifiable but also unidentifiable coincident risks. Analytic approaches such as Mendelian randomization will be helpful in making these determinations.101
Despite the recent identification of genes likely to influence the risk of non-syndromic CLP, these results have yet to have any direct impact on genetic counseling or clinical management. Improved epidemiologic information does, however, allow for better point estimates for familial recurrence risks14, and it seems likely that genotypic information for apparent risk alleles associated with higher risk of oral clefts could be useful in clinical assessment (once we have a better definition of the full number of causal genes and their potential interactions with one another and with environmental risk factors). The next critical phase of statistical analyses will be to examine the heterogeneity underlying the etiology of oral clefts and to investigate the gene-gene and gene-environment interactions that control risk. A range of study designs will be needed to achieve this level of documentation, including family studies, case-control studies and eventually prospective cohort data. Importantly, incorporating information from sub-clinical phenotypes, such as orbicularis oris defects or dental anomalies, may also allow us to identify etiologically homogeneous sub-groups of cleft cases, and thus should enhance family studies and estimates of recurrence risk.39 New array-based copy number variant analysis and whole exome or even whole genome resequencing could also afford future opportunities for improved molecular diagnostics, as does the increasingly better ultrasound analysis of the fetus for the presence and severity of cleft type prior to birth.
Global approaches to expression analysis of genes in craniofacial structures have already provided a broad view of gene expression. For example, the COGENE project http://humgen.wustl.edu/COGENE/ provides public web access to human gene expression data for 24 craniofacial specific human tissues isolated from day 26 to day 60 human embryos. In zebrafish, mRNA sequencing and microRNA analysis have been informative for understanding palate development, so it would be useful to build on this knowledge102. Similarly, the ability to analyze tissues in their correct three-dimensional orientation is central to understanding biological processes, particularly when tissues undergo a complex and intricate series of movements relative to each other as occurs in the developing craniofacial region. The mapping of gene and protein expression patterns within these complex shapes can provide important clues about their biological functions and also indicates which genes and/or proteins may interact with one another. The expression of genes relative to each other in both time and space can be visually represented using Optical Projection Tomography (OPT)103 and an atlas of craniofacial gene expression patterns is available online http://genex.hgu.mrc.ac.uk/emage/home.php.
Much of the genetic variation underlying complex disorders (such as non-syndromic CLP) is likely to occur in regulatory elements outside coding sequences of genes. These elements are challenging to identify as they often regulate genes across substantial genomic distances. Although evolutionary sequence conservation can facilitate discovery of regulatory elements, this technique does not predict their spatio-temporal pattern of activity in vivo.104 Recently, chromatin immunoprecipitation followed by Next Generation Sequence analysis (ChIP-seq) for the enhancer-associated protein p300 has been demonstrated to be a highly sensitive method to accurately identify enhancer elements and their associated activities.105 Clearly, detailed mapping of regulatory elements will provide additional (and functionally relevant) targets for sequence analysis, particularly where they fall within regions of the genome implicated by GWA studies or other approaches. The power of integrating association studies in well-characterized patient populations with identification of cis-regulatory elements, analysis of animal models and biochemical analyses is amply illustrated by the example of IRF6 noted above.
Biological roles outside the craniofacial complex are known for some of the candidate genes associated with CLP, increasing the importance of CLP gene-finding endeavors. One recent publication on a small dataset suggests a role for IRF6 in wound healing, at least in the autosomal dominant Van der Woude syndrome.106 Long-term outcomes of individuals born with clefts may include risks for higher overall mortality rates, mental health problems,107 a higher risk of cancer (particularly breast cancer) in affected individuals108 and their family members109 plus alterations in child bearing patterns.110 Identifying long-term adverse outcomes (e.g. cancer and psychiatric disorders) that are seemingly unrelated to a common birth defect may eventually result in decreasing lifelong health burden by recognizing risks at their early, pre-symptomatic stages. Studies into the etiology of clefts may well enhance our understanding of other common, complex traits and allow us to move beyond the attitude that CLP is only a structural birth defect, but instead is a lifelong disorder for which therapies and prevention can promise a fuller and healthier lifespan.
Future advances in our understanding of the molecular pathogenesis of CLP will require strategies that increasingly integrate genetic analysis of precisely phenotyped cohorts of patients, global approaches for the identification and ranking of candidate genes, and improved methods for delineating and analyzing functional elements controlling gene expression. Integration of genetic and environmental risk using epigenetics, systems biology, gene expression and epidemiology all will be required to generate a synthesis that will both better characterize etiologies, as well as provide access to better clinical care and prevention.
We should like to thank many colleagues who have collaborated with us over the years and in particular Kaare Christensen, Rolv Lie, Anil Jugessur, Andrew Lidral, Jacqui Hecht, Alexandre Vieira, Peter Jezewski, David Fitzpatrick, Ron Munger, Paul Trainor, Paul Romitti, Peg Nopoulos, John Canady, John Manak, Brian Schutte, George Wehby, Sandy Daack-Hirsch, Seth Weinberg as well as many students. We apologize for being unable to cite all the relevant papers. We gratefully acknowledge generous funding sources including the Medical Research Council (G0901539), Wellcome Trust (082868), National Institutes of Health (P50-DE016215, R01-DE08559, R01-DE016148, R01-DE014581, U01-DE018993), U01 DE-20057, the Healing Foundation, and the Manchester NIHR Biomedical Research Centre.