While most of the cases of cleft lip and palate are nonsyndromic (1 in 700–2,500 births), orofacial clefts have currently been associated with over 460 syndromes with known molecular basis [OMIM (https://www.omim.org)]. Van der Woude syndrome (VWS) is among the most common syndromic forms of clefts, affecting 1/35,000 individuals (1). It is an autosomal dominant condition that accounts for 2% of all cases with orofacial clefts (2, 3). About 44% of the cases with VWS display lip pits as their sole phenotype (1, 4, 5), but cases with VWS can also show cleft lip (CL), cleft lip and cleft palate (CLP), or cleft palate (CP), all with or without lip pits demonstrating the marked variable expression. Mutations in the gene IRF6 account for 70% of VWS cases, while mutations in the gene GRHL3 account for another 5%. However, in approximately 25% of VWS cases, the causal mutation and gene remain unknown.

Mixed clefting types are a common observation in different relatives with VWS within the same family (6–9), further elucidating the variable phenotype observed even among cases carrying the same mutations. Other VWS features include greater prevalence of midfacial hypoplasia and hypodontia (10–12). Moreover, it has been shown that patients with VWS are more likely to have wound complications following cleft repair, including fistulae recurrence (13), and are more likely to require pharyngeal flap surgery (10–12).

IRF6 is a transcription factor with a highly conserved helix-turn-helix DNA-binding domain and a less conserved SMIR/IAD protein-binding domain (14). IRF6 is the only member of the IRF gene family involved in craniofacial development, and the mutations that cause VWS are non-randomly distributed, with most occurring in the DNA-binding domain (exons 3 and 4) and the protein-binding domain (exons7–9) (15). Since the discovery of IRF6 as the first causal gene for VWS (8), more than 300 mutations have been identified in cases with VWS and PPS (15–18).

Although most of the causal mutations are classified as missense mutations, nonsense and frameshift mutations have also been reported. The VWS mutations characterized so far exert their effect on the phenotype via haploinsufficiency or a dominant negative effect.

Animal studies have shown that disruption of IRF6 in mice leads to cleft lip and/or palate (CL/P) in addition to oral epithelial adhesions, poor epithelial barrier functions, and improper skin stratification, which suggests that oral epithelium plays an important role in directing palate development (19, 20). In addition, molecular and histologic analyses showed that IRF6-mutated mouse embryos lack periderm cells at the sites of oral adhesions (21), which leads to abnormal epithelial adhesions between the palatal shelves and the lingual, mandibular, and maxillary surfaces, preventing the proper elevation and fusion of the palatal shelves. Besides the abnormal adhesions, the medial edge epithelium (MEE) located at the medial edge of the palatal shelves failed to dissolve for proper palatal fusion also leading to a cleft palate. Similar abnormal bilateral adhesions leading to a cleft palate were observed in mice that were heterozygous for a mutated allele of GRHL3, the second VWS locus (22). Thus, both IRF6 and GRHL3 are essential to develop a normal oral periderm, which is necessary for palatogenesis (22).

While various established mutations in IRF6 account for the presence of the syndrome, they do not explain the variability and different levels of phenotypic severity. Since the first publication of IRF6 being the causal gene, authors have discussed the possible causes of the phenotypic variability observed even among individuals with the same mutation. Studies have suggested the action of stochastic factors or modifier genes on IRF6 function, but epigenetic factors that could play a role remain unexplored. Another important observation is that the mixed clefting phenotype that is common to VWS families is rare in nonsyndromic cleft families. However, this phenotypic variability is also observed in TP63 (23, 24).

Despite decades of intriguing observations of phenotypic discordance, the first etiological genetic variant for VWS was only discovered in 2002 when Kondo et al. (8) reported a unique, and genetically confirmed, pair of monozygotic twins in which one twin is affected with bilateral CLP and lip pits and the other is unaffected. With the use of genetic mapping and DNA sequencing, an IRF6 mutation was identified in the affected twin and was absent in the unaffected co-twin, which confirmed IRF6 as the first causal gene for VWS.

Later in 2011, another pair of monozygotic twins with VWS (25) was reported. In this case, both twins are affected with VWS, but despite carrying the same genetic mutation in IRF6, their phenotypes are markedly different, with one twin presenting bilateral cleft lip, cleft palate, and lower lip pits, while her twin sister has only lower lip pits. The variable expressivity of VWS phenotype and the phenotypic discordance even among monozygotic twins who carry the same mutation suggests the role of modifier factors, which can be genetic or epigenetic.

While MZ twins share identical DNA sequence, the fact that they can be phenotypically distinct offers important insights into the role of environmental factors. With the increased number of epigenetic studies, evidence suggests that epigenetic mechanisms may be potential mediators between environment and phenotypic expression.

DNA methylation of cytosines at CpG dinucleotides was first proposed as a mechanism of mammalian gene regulation in 1975 (3, 4), and has been since then the most broadly studied epigenetic mark. It occurs predominantly at the carbon-5 position of symmetrical CpG (cytosine and guanine separated by a phosphate) dinucleotides (5 mC). The state of DNA methylation is mitotically heritable through the activity of DNA methyltransferases (DNMTs), and it is essential for control of gene expression. Epigenetic marks like DNA methylation are essential for cell differentiation and preservation of tissue homogeneity. During development and throughout life, parent cells use epigenetic marks to ensure that their daughter cells will differentiate properly and function appropriately, and this message may persist through thousands of cell divisions for the lifetime of the organism, unless they are actively erased (by demethylase enzymes) or lost through epimutations. Typically, methylation of CpG sites in promoter regions of genes inhibits gene expression either due to the inability of specific transcription factors to bind methylated CpGs or the recruitment of methyl-CpG-binding proteins with transcription repression activity (26–28).

Aberrant DNA methylation patterns are universally recognized as playing an important role in human diseases, including monogenic and complex disorders. Recent studies indicate that abnormal methylation levels of key genes and/or regulatory elements are involved in heart disorders (29–34), depression and anxiety (35–42), several craniofacial syndromes (43–47), as well as nonsyndromic cleft lip and palate (48–53). In addition, several recent twin studies have shown additional evidence that DNA methylation may play an important role in phenotypic discordance (35, 48, 54–58).

In this article, we report the first DNA methylation profiles of monozygotic twins who present with discordant affection status and phenotypic expression for VWS. Twin pair 1 was previously reported in the study that discovered IRF6 as the first causal gene for VWS (8). Even though they have been confirmed as monozygotic twins, one of them has VWS while the other is unaffected. This discordance in affection status is most likely a result of an early post-twinning mutation in the affected twin. Twin pair 2 was also previously reported (25), but both females are affected by VWS and carry the same mutation in IRF6. Despite being monozygotic and having the same mutation, their phenotype is markedly different.

Our goal is to explore the epigenetic contributions to VWS etiology by comparing the DNA methylation profiles of these two pairs of monozygotic discordant twins. While the mutation that causes VWS in these two twin pairs is known, additional mechanisms behind phenotypic risk and expression variability, especially in the second twin pair, remain underexplored.

Since the two twin pairs were of opposite gender and distinct ages, we analyzed both pairs separately with the same analytical pipeline and compared their results. Twin pair 1 is a male monozygotic twin pair; the affected twin (twin 1A) was shown to carry a nonsense mutation (E92X) in exon 4 of the IRF6 gene, which was absent in the unaffected co-twin (twin 1B) and parents. The affected twin in this pair presented with cleft lip and cleft palate and bilateral lower lip pits. Twin Pair 2 is a female monozygotic twin pair; both girls and their affected father carry the same missense mutation (Y97C) in exon 4 of the IRF6 gene. Although all three individuals share the same mutation, twin 2A presents with cleft lip and cleft palate and bilateral lower lip pits, while twin 2B and their father present only bilateral lower lip pits (no epigenetic data were generated for the father).

The results for twin pair 1 are shown below and summarized in Figure 3.

The results for twin pair 2 are explained below and summarized in Figure 4.

Monogenic diseases, like VWS, often feature highly variable phenotypes, despite a usually well-defined genetic cause for the disease. Modifier genes and epigenetic mechanisms are believed to influence phenotypic risk and variable expressivity. In this article, we see the rare event of monozygotic twins carrying the same mutation (twin pair 2), who display extremely divergent phenotypes for VWS.

Monozygotic twins arise from the same fertilized oocyte and are, therefore, believed to have the same DNA sequence. Postzygotic mutations are responsible for a substantial proportion of de novo mutations in humans and have been shown to contribute to disease phenotypic variability, including among twins (65–71). In twins, the timing of postzygotic mutations determines whether they are present in both twins (pre-twinning mutation) or in only one twin (post-twinning mutation) (68). Our twin pair 1 is an example of post-twinning mutation, where only twin 1A became a carrier of the IRF6 mutation, while twin 1B does not have the causal mutation and is, therefore, unaffected. In twin pair 2, both females inherited the IRF6 mutation from their affected father (Figure 2).

In addition to playing an essential role in estimating phenotypic variability, twin studies offer an opportunity to study epigenetic variation as a quantitative trait. The monozygotic twin discordance rate observed in OFCs (≥50%) suggests that unexplained environmental and epigenetic factors play an etiological role. Epigenetic studies have shown that changes in DNA methylation play a role in nonsyndromic (48–51, 53, 72, 73) and syndromic forms of OFCs. Among the cleft syndromes most commonly associated with differences in DNA methylation are Kabuki and Charge syndromes (43, 44, 46, 47, 74, 75), both of which have specific DNA methylation signatures. We describe the first whole genome methylation profiling of MZ twins discordant for VWS and discuss the possible roles that differences in DNAm may have in phenotypic discordance.

Gene ontology and enrichment analysis of the thousands of CpG sites with differential methylation identified for each pair revealed a list of enriched genes that included, but were not limited to, known cleft-associated genes. The top genes that showed different levels of DNAm between each twin pair include genes like HOXA5, LEPREL1, ARHGEF10, OR4E2, DAD1, AZU1, and TP63, among others (please refer to Tables 1–4). From these, TP63 is known to cause syndromic forms of OFCs and is associated with nonsyndromic clefts; in addition, TP63 is directly involved in the activation of IRF6.

Mutations in the gene encoding the transcription factor interferon regulatory factor 6 (IRF6), cause VWS and PPS, both characterized by ectodermal anomalies and CL/P or CP; mutations in the transcription factor p63 cause autosomal dominant ectodermal dysplasia syndromes such as ectrodactyly ectodermal dysplasia-clefting (EEC), which is also characterized by CL/P or CP and defects in ectoderm-derived tissues, such as the epidermis, hair, teeth, and glands (76). Moreover, mutations in both genes have been implicated in nonsyndromic OFCs (77).

Studies have shown that IRF6 is transcriptionally activated by TP63 and, in turn, induces the TP63 proteasome-mediated downregulation, thereby limiting epithelial cell proliferative potential (78), including in the medial epithelial seam around the time of palate closure. Such process is critical for normal palatal fusion (79–82).

Among the cleft-associated genes in our dataset, TP63 was the most significant hit (P = 7.82E-12). Both twin pairs presented differential DNAm levels in CpG sites in/near TP63 (Twin 1A > Twin 1B and Twin 2A < Twin 2B). Interestingly, in twin pair 2, the twin with only lip pits (twin 2B) presents higher levels of methylation in the promoter region of the TP63 gene. It has been shown that TP63 downregulation necessitates normal IRF6 function, and that this task is absent in irf6 mutant mice, who remain p63 positive in the palatal medial edge epithelial (83). Given that TP63 and IRF6 work together in a regulatory loop to coordinate epithelial proliferations, it is plausible that DNAm can modify the effects of the etiological variant and lead to markedly distinct phenotypes, even among individuals carrying the same mutation. Since methylation of promoters is often associated with gene silencing, it is possible that this epigenetic mechanism compensates for the inability of the mutated IRF6 to downregulate TP63, therefore contributing to a less severe phenotype in twin 2B.

There are some limitations in our study. First, our study includes two pairs of MZ twins discordant for affection status and phenotype, but given the rarity of the condition, we consider the presented data of importance to better understand the additional factors that may contribute to phenotypic discordance. We are aware that the control groups are relatively small and the age among the control individuals varied, and we also acknowledge the fact that the DNAm profiles were obtained from blood DNA and that epigenetic marks can be cell type specific. We did our best to correct for cell type heterogeneity by using well-established bioinformatic tools. Finally, it is known that the cause of the VWS in the affected twins is the mutation reported by Kondo et al. and Jobling et al. (2011); however, this is the first study reporting a genome-wide epigenetic profiling of the syndrome, especially using the powerful discordant twin design.

Despite its limitations, the in silico functional analysis methods that we used, help us gain insight into the biology underlying the regions in which we detected differential methylation and allow us to explore whether pathways or processes are enriched among our best hits. We used over-representation analysis methods to identify potential pathways that could be affected by the observed changes in DNA methylation.

Clinical divergence between patients complicates diagnosis and genetic counseling, this is especially true when syndromic and nonsyndromic cases have overlapping clinical features, like VWS and nonsyndromic clefts. In addition to clinical implications and translational potential, studies of etiological factors of syndromic forms of clefting have greatly contributed to the understanding of the much more complex etiology of nonsyndromic cases. Therefore, our study emphasizes the need to understand the molecular mechanisms underlying phenotypic variability and the role of epigenetic factors in disease etiology.