Mouse Models in Palate Development and Orofacial Cleft Research: Understanding the crucial role and regulation of epithelial integrity in facial and palate morphogenesis

Abstract

Cleft lip and cleft palate are common birth defects resulting from genetic and/or environmental perturbations of facial development in utero. Facial morphogenesis commences during early embryogenesis, with cranial neural crest cells interacting with the surface ectoderm to form initially partly separate facial primordia consisting of the medial and lateral nasal prominences, and paired maxillary and mandibular processes. As these facial primordia grow around the primitive oral cavity and merge towards the ventral midline, the surface ectoderm undergoes a critical differentiation step to form an outer layer of flattened and tightly connected periderm cells with a non-stick apical surface that prevents epithelial adhesion. Formation of the upper lip and palate requires spatiotemporally regulated inter-epithelial adhesions and subsequent dissolution of the intervening epithelial seam between the maxillary and medial/lateral nasal processes and between the palatal shelves. Proper regulation of epithelial integrity plays a paramount role during human facial development, as mutations in genes encoding epithelial adhesion molecules and their regulators have been associated with syndromic and non-syndromic orofacial clefts. In this chapter, we summarize mouse genetic studies that have been instrumental in unraveling the mechanisms regulating epithelial integrity and periderm differentiation during facial and palate development. Since proper epithelial integrity also plays crucial roles in wound healing and cancer, understanding the mechanisms regulating epithelial integrity during facial development have direct implications for improvement in clinical care of craniofacial patients.

1. INTRODUCTION

Craniofacial malformations account for more than one-third of all congenital birth defects in humans (Twigg and Wilkie, 2015). While many distinct classes of craniofacial anomalies exist, the most common congenital craniofacial defects involve disruption of the lip and/or palate, with cleft lip with or without cleft palate (CLP) occurring at 1:500 to 1:1,000 live births, and cleft palate only (CPO) occurring at about 1:2,000 live births (Dixon et al., 2011; Nasreddine et al., 2021; Seto-Salvia and Stanier, 2014). Although not a major cause of mortality in developed countries, children born with CLP or CPO require extensive surgical and dental treatments and experience a variety of health and psychosocial problems (Christensen et al., 2004; Dixon et al., 2011). The causes of orofacial clefts are heterogeneous, involving both genetic and environmental factors. Over 500 Mendelian syndromes, mostly caused by disruption of single gene loci or chromosome abnormalities, have orofacial cleft as a major clinical feature (Dixon et al., 2011; Leslie and Marazita, 2013; Nasreddine et al., 2021; Seto-Salvia and Stanier, 2014). Several environmental factors, such as maternal smoking or exposure to teratogens including retinoic acid, phenytoin, and valproic acid, have been associated with increased risk of orofacial clefts (Dixon et al., 2011; Garland et al., 2020; Leslie and Marazita, 2013; Shi et al., 2007). Furthermore, about 70% of CLP and 50% of CPO patients show no apparent other structural anomalies, known as isolated or non-syndromic CL/P (Bishop et al., 2020; Dixon et al., 2011; Seto-Salvia and Stanier, 2014). A great deal of research efforts, including multiple genome-wide association studies and more recently large-scale whole exome or whole genome sequencing, have focused on identification of genes and genomic variants involved (Basha et al., 2018; Bishop et al., 2020; Bureau et al., 2014; Cox et al., 2018; Leslie et al., 2017; Leslie and Marazita, 2013; Nasreddine et al., 2021; Seto-Salvia and Stanier, 2014; Takahashi et al., 2018; Yu et al., 2017). Whereas these and other studies have identified mutations and variants in coding or non-coding regions in dozens of genes associated with CL/P, definitive causal mutations have accounted for less than 30% of the estimated liability of orofacial clefts (Bishop et al., 2020). For most non-syndromic orofacial cleft cases, a multifactorial model of inheritance is favored, such that variations in multiple genes are etiologically necessary, with gene-environment interactions and epigenetic modifications contributing with a threshold effect influencing the penetrance and expressivity (Dixon et al., 2011; Young et al., 2021).

Whereas significant progress has been made towards understanding the genetic etiology of orofacial clefts, with technological and bioinformatic advances further speeding up the identification of candidate causal mutations, these results have not significantly impacted clinical care and prevention strategies due to very limited understanding of the cellular, molecular, and developmental processes underlying cleft pathogenesis (Dixon et al., 2011; Leslie and Marazita, 2013; Nasreddine et al., 2021; Vieira, 2021). Since lip closure occurs by the end of the sixth week, whereas palatal fusion is complete by the 12th week, of human gestation (Dixon et al., 2011; Jiang et al., 2006), studies using animal models have been necessary and instrumental for understanding the cellular and molecular mechanisms of normal and abnormal craniofacial development (Bush and Jiang, 2012; Chai and Maxson, 2006; Gritli-Linde, 2007, 2008; Ji et al., 2020; Jiang et al., 2006; Juriloff and Harris, 2008; Li et al., 2017; Van Otterloo et al., 2016). The mouse is the pre-eminent model organism for unraveling the developmental mechanisms of orofacial cleft pathogenesis because it is a small mammal whose genome contains ~99% of homologous genes in human, the embryonic facial and palate development processes are strikingly similar in mouse and human, and many mouse strains carrying spontaneous or genetically engineered mutations causing cleft lip and/or cleft palate phenotypes are available (Abbott, 2010; Gritli-Linde, 2008; Jiang et al., 2006; Juriloff and Harris, 2008; Murray, 2011). In particular, the advent of mouse embryonic stem cell and gene targeting technologies in the 1980s enabled direct targeted manipulation of almost every gene in the mouse genome, which revolutionized mammalian biology and biomedical research (Capecchi, 2005). Whereas individual or combinations of null mutations in genes encoding major signaling pathway components or structural proteins often caused early embryonic lethality of homozygous mutant mice, the adaptation of site-specific DNA recombination systems, particularly the Cre/loxP system, enabled systematic genetic analysis of cell type-specific gene function at specific time or developmental stage during the life of a mouse (Jiang and Gridley, 1997; Lan et al., 2007; Rajewsky et al., 1996). Over 600 distinct mutant mouse lines with cleft palate phenotype have been reported (http://www.informatics.jax.org/phenotypes.shtml), of which studies have not only provided excellent candidate genes for analysis in human genetic studies, but also significantly improved the understanding of the molecular and cellular mechanisms of craniofacial and palate development (Bush and Jiang, 2012; Bush et al., 2004; Gritli-Linde, 2007, 2008; Lan et al., 2004; Lan et al., 2015; Li et al., 2017; Reynolds et al., 2020; Suzuki et al., 2018; Xu et al., 2016; Zhou et al., 2013). A number of excellent review articles have described the molecular and cellular mechanisms involving multiple signaling pathways and transcription factors controlling secondary palate development (Bush and Jiang, 2012; Gritli-Linde, 2007; Li et al., 2017). Here we summarize extensive mouse genetic studies that have revealed a paramount role of and mechanisms regulating epithelial integrity during facial and palate morphogenesis. We discuss how these studies have significantly improved the understanding of the pathogenic mechanisms underlying syndromic and non-syndromic orofacial clefting and the implications for improvement in patient care.

2. OVERVIEW OF MIDFACIAL MORPHOGENSIS AND PERIDERM DIFFERENTIATION

Development of the face involves precisely coordinated directional growth, morphogenesis, and fusion of initially partly separate facial primordia, which first appear in the fourth week of human gestation (about embryonic day E9.5 in mice) and are composed of the frontonasal prominence (FNP) covering the ventral side of the developing forebrain, a pair of maxillary processes (Mx) and a pair of mandibular processes (Mb) forming the lateral and caudal boundaries of the primitive mouth, respectively (Hinrichsen, 1985; Jiang et al., 2006). These facial primordia initially consist of two cell types, the surface epithelium and the cranial neural crest cells that first arise at the anterior neural plate border and migrate ventrally into the space under the surface epithelium, with the FNP populated by neural crest cells from the forebrain and anterior midbrain region whereas the Mx and Mb populated by neural crest cells from the posterior midbrain and anterior hindbrain region (Figure 1A). As development proceeds, signaling interactions involving BMP, FGF, SHH, and Wnt signaling pathways within the facial epithelium as well as between the frontonasal epithelium and the neural crest derived FNP mesenchyme result in the outgrowth of paired medial and lateral nasal processes (MNP and LNP) and formation of the nasal pit, while a midline groove forms between the MNPs as the forebrain forms a pair of telencephalic vesicles in the 5^th week of human gestation (~E10 in mice) (Fig. 1B) (Jiang et al., 2006; Marcucio et al., 2015; Panaliappan et al., 2018; Reid et al., 2011; Szabo-Rogers et al., 2008; Wang et al., 2011; Zhu et al., 2016). Coordinated growth and morphogenesis of the facial processes driven by the signaling interactions between the epithelium and mesenchyme bring the caudal edges of the LNP and MNP into direct contact with the Mx to form the three-way fusion zone named the lambdoidal junction (Depew and Compagnucci, 2008; Li et al., 2019) (Figure 1C). Adhesion and subsequent dissolution of the intervening epithelial seam between these processes lead to completion of lip fusion, which together with continued growth pushing the nose chambers medio-frontally and with mesenchymal proliferation backfilling the midline groove, resulting in formation of the intact upper lip by the end of sixth week of human gestation (~E12.5 in mice) (Figure 1D) (Jiang et al., 2006). Both BMP and Wnt signaling within the epithelium play crucial roles in lip fusion, whereas interactions of FGF, SHH, and Wnt signaling regulate the directional growth while FGF signaling maintains survival of the neural crest-derived facial mesenchyme during facial primordial outgrowth (Ferretti et al., 2011; Jeong et al., 2004; Jiang et al., 2006; Jin et al., 2020; Kurosaka et al., 2014; Lan et al., 2006; Liu et al., 2005; Marchini et al., 2021; Ray et al., 2020; Reynolds et al., 2019; Reynolds et al., 2020).

Figure 1.

Overview of facial development. (A) Lateral view of an E9.5 Sox10Cre;Rosa26^mTmG/+ mouse embryo showing facial primordia formed from cranial neural crest cells populating between the ventral side of the developing brain and surface ectoderm. Green color marks cranial neural crest cells while red color marks all other cells. (B) Frontal view of an E10 mouse embryonic head marked with the pattern of Fgf8 mRNA expression (purple/blue color) in the developing facial epithelium. Red asterisks mark the nasal pits in between the medial and lateral nasal process. Note that the distal regions of the medial nasal process, lateral nasal process, and maxillary process are separate from each other at this stage. (C) Frontal view of an E11 mouse embryonic head marked with the pattern of Bmp4 expression (purple/blue color) in the developing facial epithelium. Note that the caudal region of the medial and lateral nasal processes and the rostral region of the maxillary processes have formed the lambdoidal junction during lip fusion. (D) Frontal view of an E12.5 mouse embryonic head showing the intact upper lip formed from fusion of the medial nasal, lateral nasal, and maxillary processes. Blue staining marks tissues expressing the Osr2 gene. Red asterisks mark the nostrils. Red arrow points to the philtrum and primary palate. anr, anterior neural ridge; ba2, second branchial arch; e, eye; fnp, frontonasal prominence; lnp, lateral nasal process; mb, mandibular arch or mandible; mnp, medial nasal process; mx, maxillary process; ov, otic vesicle; tg, trigeminal ganglion.

The mammalian palate consists of the primary palate derived from the oral side of the MNP, and secondary palate derived from the oral side of the maxillary processes (Bush and Jiang, 2012). Secondary palate development commences in the 6^th week of human gestation, and about E11.5 in mice, with outgrowth arising from the oral side of the maxillary processes and growing downward to form paired palatal shelves while the tongue arises from the mandible. From E12 to E13.5 in mouse embryos, the palatal shelves grow vertically flanking the developing tongue. From E13.5 to E14.5, the bilateral palatal shelves reorient to the horizontal position above the tongue, grow toward the midline, with subsequent palatal fusion involving adhesion, formation of the midline epithelial seam (MES), and MES dissolution, to form the roof of the oral cavity (Figure 2A–C). Palatal fusion is complete by 12th week of human gestation and about E16.5 in mice. Extensive studies of the molecular mechanisms involving major signaling pathways and transcriptional regulatory networks controlling palatal shelf growth, patterning, and fusion, have been reported, and most have been summarized in several excellent review articles (Bush and Jiang, 2012; Gritli-Linde, 2007; Lan et al., 2015; Lane and Kaartinen, 2014; Li et al., 2017; Reynolds et al., 2020).

Figure 2.

Development of the secondary palate and its disruption in the Jag2 mutant mice. (A-C) Frontal sections of embryonic mouse heads through the middle region of the developing secondary palate showing vertically oriented palatal shelves at E13.5 (A), horizontal palatal shelves that have formed the inter-shelf midline epithelial seam (red arrowhead) during palate fusion at E14.5 (B), and fused secondary palate with mesenchymal confluence through the midline in an E15.5 embryo (C). (D-E) Frontal sections through the midpalate regions of a pair of E16.5 control (D) and Jag2^−/− (E) littermates showing aberrant fusion (arrow) of the medial region of the palatal shelves with the lateral sides of the tongue and aberrant adhesions (arrowheads) between the maxilla and mandible in the molar tooth germ regions of the Jag2^−/− embryo. p, palatal shelf; t, tongue.

In addition to providing many of the instructive signals, including members of BMP, FGF, and Wnt families, and Shh, to direct the growth and patterning of the neural crest derived mesenchyme during facial and palatal morphogenesis (Bush and Jiang, 2012; Gritli-Linde, 2007; Jiang et al., 2006; Lan et al., 2015; Lane and Kaartinen, 2014; Li et al., 2017; Reynolds et al., 2020), the surface epithelium undergoes significant growth and differentiation at the same time (Hammond et al., 2019). At the beginning of facial development, the surface ectoderm is a single layer of multipotent squamous epithelial cells that proliferate by symmetric cell divisions to cover the growing embryo body. At about E9.5 in mice, as the multipotent epithelial cells undergo specification and commitment to epidermal cell fate, a subset of the cells extend cellular processes over their neighbors and subsequently lose contact with the basement membrane as they migrate over the basal cells to form the outer layer of flattened epithelial cells called periderm cells (Hammond et al., 2019; M’Boneko and Merker, 1988; McGowan and Coulombe, 1998; Richardson et al., 2014). Periderm formation occurs apparently simultaneously at multiple sites and a continuous layer of periderm cells, connected by tight junctions at their apico-lateral borders, covers the oral epithelium by E12.5 and over the entire mouse embryo by E14.5 (Casey et al., 2006; Hammond et al., 2019; Morita et al., 2002; Richardson et al., 2014; Yoshida et al., 2012). Epidermal cell stratification and differentiation occur under the periderm layer to form the multilayered mature epidermis with a cornified outer layer functioning as a skin barrier, and the periderm is shed from the surface due to loss of adhesion to the underlying keratinocytes by E17.5 (Hammond et al., 2019; Saathoff et al., 2004). The primary function of the periderm is to provide a diffusion barrier to protect the embryo and the developing epidermis from the amnionic fluid until the barrier function of the cornified epidermis is mature (Hayward, 1983; Hu et al., 2018). In the next section, we will summarize studies that have uncovered a crucial role of the embryonic periderm in facial and palatal morphogenesis.

3. MOUSE GENETIC STUDIES UNCOVER A CRITICAL ROLE FOR THE EMBRYONIC PERIDERM IN OROFACIAL MORPHOGENESIS

3.1. Proper periderm differentiation is required for normal palatogenesis.

Whereas formation of the upper lip and palate involves adhesion of initially separate primordial tissues to form an epithelial seam which is subsequently dissolved to allow mesenchymal confluence, the epithelial adhesion processes must be precisely regulated spatiotemporally to prevent aberrant epithelial adhesions that could disrupt lip and/or palate morphogenesis (Casey et al., 2006; Hammond et al., 2019; Jiang et al., 1998; Richardson et al., 2014). Aberrant oral adhesion as a cause of cleft palate was first discovered in studies of mice with disruption of the Jag2 gene, which encodes a cell surface ligand for the Notch family receptors (Casey et al., 2006; Jiang et al., 1998). The Jag2⁻ mice exhibit cleft palate at birth, with the palatal shelves stuck in the vertical position due to aberrant fusion of distal regions of the palatal shelves with the lateral sides of the tongue (Figure 2D, E) (Casey et al., 2006; Jiang et al., 1998). Jag2 is expressed throughout the developing oral epithelium and activates the Notch1 receptor in the oral periderm (Casey et al., 2006). Specific ablation of periderm cells in mouse embryos through Krt17-Cre mediated activation of expression of diphtheria toxin subunit-A from the Rosa26^eGFP-DTA allele resulted in frequent intraoral epithelial adhesions and cleft palate in severely affected mouse embryos, indicating that the periderm layer plays a critical role in preventing aberrant epithelial adhesions (Richardson et al., 2014). In the Jag2^−/− embryos, oral periderm cells formed by E12 but appeared more rounded and disorganized, with significantly reduced nuclear Notch1 activity (Casey et al., 2006). By E13.5, aberrant palate-tongue and maxillary-mandibular adhesions were detected in the Jag2^−/− embryos, indicating that Jag2-Notch1 signaling plays a critical role in the regulation of oral periderm differentiation (Casey et al., 2006).

3.2. The IRF6 and GRHL3 transcription factors regulate periderm differentiation to prevent pathological epithelial adhesions that cause cleft palate and ectodermal defects in Van der Woude and popliteal pterygium syndromes.

Aberrant epithelial adhesions have been associated with several CLP syndromes in humans, including Van der Woude syndrome (VWS, OMIM 119300) and popliteal pterygium syndrome (PPS, OMIM 119500) (Hammond et al., 2019; Kalay et al., 2012; Kondo et al., 2002; Kousa and Schutte, 2016). VWS is an autosomal dominant disorder characterized by CLP or CPO with lower lip pits, whereas PPS presents with similar orofacial phenotypes, plus additional features including popliteal webs, syndactyly, oral synechia, and other abnormalities (Hammond et al., 2019; Kondo et al., 2002; Kousa and Schutte, 2016). Kondo et al. (2002) identified mutations in the Interferon Regulatory Factor-6 (IRF6) gene as the cause of both syndromes (Kondo et al., 2002). IRF6 encodes a transcription factor containing a highly conserved helix-turn-helix DNA binding domain (Taniguchi et al., 2001). Gene expression analysis in mouse embryos showed that Irf6 is expressed throughout the embryonic facial and oral epithelium, with highest levels in the periderm cells (Knight et al., 2006; Richardson et al., 2009; Richardson et al., 2014). Mouse embryos homozygous for either of two distinct mutations, a targeted null allele (Irf6^-) and the Irf6^R84C missense mutation that mimics the most common IRF6 mutation in PPS, exhibited severe intraoral epithelial adhesions causing the vertically oriented palatal shelves to be wedged in between the lateral side of the tongue and mandible, in addition to other epidermal defects at birth (Ingraham et al., 2006; Richardson et al., 2006). Subsequent studies showed that IRF6 is required for periderm formation and maintenance (Kousa et al., 2017; Richardson et al., 2009; Richardson et al., 2014). Transgenic expression of Irf6 in the basal epithelial cells using the Keratin-14 gene promoter was able to partially rescue periderm formation, although not completely rescuing cleft palate, in the Irf6^−/− embryos (Kousa et al., 2017). Interestingly, whereas the palatal and tongue epithelia remained continuous in Irf6^−/− and Irf6^R84C/R84C embryos despite the almost complete adhesion of the palatal shelves with the tongue and mandible, the distal regions of the palatal shelves fused with the tongue, with regional dissolution of both the palatal and tongue epithelium, in the Jag2^−/− and K14-Irf6^Tg/+;Irf6^−/− embryos. This is consistent with another study identifying an essential role for IRF6 in the medial edge epithelium of palatal shelves during palatal fusion (Iwata et al., 2013) (further discussed below).

A comprehensive mutation screening of VWS and PPS patients found that mutations in IRF6 account for about 70% of VWS and 97% of PPS cases (de Lima et al., 2009). A second VWS locus was mapped to human Chromosome 1p34, and mutations in the Grainyhead-like-3 (GRHL3) gene were subsequently identified in a subset of IRF6 mutation-negative families (Koillinen et al., 2001; Peyrard-Janvid et al., 2014). GRHL3 encodes a vertebrate homolog of the Drosophila Grainyhead DNA-binding transcription factor that plays an essential role in development and maintenance of the epidermal barrier (Bray and Kafatos, 1991; Sundararajan et al., 2020). Mice deficient in Grhl3 also exhibited defective skin barrier function and deficient wound repair (Ting et al., 2005; Yu et al., 2006). While the initial report of Grhl3 expression in mouse embryos described it as largely restricted to the surface ectoderm from E12.5 onward (Ting et al., 2005), subsequent studies found that Grhl3 is most highly expressed in periderm cells in both zebrafish and mouse embryos, and its expression is lost in the oral epithelium in Irf6^−/− embryos (de la Garza et al., 2013; Kousa et al., 2017). Studies in zebrafish embryos and human keratinocytes found that IRF6 directly regulates GRHL3 gene expression, and that injection of zebrafish embryos with grhl3 mRNAs partially rescued developmental defects caused by a dominant-negative irf6 construct (de la Garza et al., 2013). Furthermore, Grhl3^−/− mouse embryos exhibited defective oral periderm with reduced levels of IRF6 and Notch1 proteins in some regions and aberrant palate-tongue fusion, with some Grhl3^−/− embryos displaying cleft palate (Peyrard-Janvid et al., 2014). Grhl3^−/− embryos also exhibited syndactyly due to aberrant epithelial adhesions in the interdigital regions (Kashgari et al., 2020). The Grhl3^−/− periderm cells displayed ectopic E-cadherin at the apical surface and mis-localization of the tight junction protein ZO-1 at the basolateral side, in contrast to the polarized apico-lateral localization of ZO-1 in between the periderm cells in the wildtype embryos (Kashgari et al., 2020). These results indicate that GRHL3 mediates part of the IRF6 function in regulating periderm integrity. Remarkably, genome wide association studies have found that mutations in IRF6 and GRHL3 contribute significantly to non-syndromic orofacial clefting, with mutations in IRF6 accounting for approximately 2% of all orofacial cleft cases, whereas loss-of-function or dominant-negative mutations in GRHL3 have been detected in both Caucasian and African patients with non-syndromic CPO (Dixon et al., 2011; Eshete et al., 2018; Leslie et al., 2016; Mangold et al., 2016).

How do IRF6 and GRHL3 regulate periderm differentiation? A recent study identified 66 high-confidence IRF6 target genes through combined RNA-seq and enhancer activities in embryonic epidermis from E15.5 Irf6^−/− and wildtype embryos (Oberbeck et al., 2019). In addition to Grhl3, the significantly downregulated IRF6 target genes in Irf6^−/− embryonic epidermis include Occludin (Ocln), encoding a component of tight junctions, and multiple genes encoding components of lipid metabolism pathways. Quantification of lipids in the skin of E16.5 Irf6^−/− and wildtype embryos revealed significant decreases in six classes of lipids, including ceramides and triacylglycerols, and the Irf6^−/− embryos displayed a marked absence of non-polar lipids in the outer layer of the skin (Oberbeck et al., 2019). GRHL3 is also required for lipid processing in the developing epidermis and directly regulates expression of Tgm1, which encodes Transglutaminase-1, involved in crosslinking cornified envelope proteins and lipid envelope (Matsuki et al., 1998; Ting et al., 2005; Yu et al., 2006). Gene expression profiling of embryonic skin from E18.5 Grhl3^−/− and wildtype embryos showed that expression of several genes encoding components of tight junctions, including Ocln, Cldn1, Cldn4, and Cldn23, were significantly downregulated (Yu et al., 2006). Whether IRF6 and GRHL3 regulate periderm differentiation and integrity through directly controlling expression of tight junction proteins and/or lipid metabolism pathways remains to be elucidated.

3.3. Bartsocas-Papas Syndrome is also a peridermopathy, and Receptor-Interacting Protein Kinase-4 controls epidermal and orofacial morphogenesis by posttranslational regulation of IRF6 function.

Whereas IRF6 functions as a transcription factor in the regulation of GRHL3 gene expression (de la Garza et al., 2013), immunofluorescent detection in mouse embryonic skin and human keratinocytes showed that the IRF6 protein is primarily localized in the cytoplasm (Ingraham et al., 2006; Kwa et al., 2014; Richardson et al., 2006). Kwa et al. (2014) showed that receptor-interacting protein kinase-4 (RIPK4), which is a critical regulator of keratinocyte differentiation (Holland et al., 2002), regulates IRF6 nuclear translocation and transcriptional activity through phosphorylation of its C-terminal domain at the Ser413 and Ser424 residues (Kwa et al., 2014). Loss-of-function mutations in RIPK4 were identified in individuals affected by Bartsocas Papas Syndrome (BPS1, OMIM 263650), an autosomal recessive disorder with significant phenotypic overlap with PPS, including CL/P, popliteal webbing, and cutaneous syndactyly (Kalay et al., 2012; Kondo et al., 2002; Mitchell et al., 2012). Analysis in mouse embryos showed that Ripk4 is expressed throughout the developing epidermis, including periderm cells (De Groote et al., 2015). Ripk4^−/− mice displayed deficient skin barrier at birth and severe pathological epithelial adhesions including partial fusion of the limbs to the body and partial fusion of the mouth and nose orifices (De Groote et al., 2015; Holland et al., 2002). The epithelial lining of the oral cavity in the Ripk4^−/− embryos was almost completely sealed by E13.5, with the vertical palatal shelves wedged between the mandible and tongue, similar as in Irf6^−/− and Irf6^R84C/R84C embryos (De Groote et al., 2015; Ingraham et al., 2006; Richardson et al., 2006). In contrast to the Irf6 mutant embryos which exhibited lack of periderm (Kousa and Schutte, 2016; Richardson et al., 2014), the Ripk4^−/− embryos had a complete periderm layer (De Groote et al., 2015). Whereas the periderm cells are shed from the surface of the epidermis by E17.5 in wildtype mouse embryos due to loss of adhesion to the underlying keratinocytes as the cornified layer forms a competent protective barrier (Hammond et al., 2019; Richardson et al., 2014), the Ripk4^−/− embryos displayed persistent Krt17-positive periderm cells over the skin by birth (De Groote et al., 2015). The periderm cells displayed ectopic E-cadherin at the apical surface in Ripk4^−/− embryos from E13.5 onward, whereas E-cadherin was absent from the apical surface of the flat periderm cells in wildtype embryos (De Groote et al., 2015). Remarkably, epithelial-specific inactivation of the Cdh1 gene, which codes for E-cadherin, partly rescued the Ripk4^−/− embryos, such that aberrant epithelial adhesions only occurred in areas of incomplete Cdh1 deletion (Urwyler-Rosselet et al., 2018). Mice with conditional inactivation of Ripk4 using a K14-Cre driver (Ripk4^EKO) displayed an open mouth with aberrant epithelial adhesions observed between the mandible and tongue only in the anterior region of the oral cavity, and two-thirds of the E18.5 Ripk4^EKO embryos displayed fused palate (Urwyler-Rosselet et al., 2018). Mice with conditional inactivation of Ripk4 or Irf6 using another K14-Cre driver were described as displaying no soft tissue fusions (Oberbeck et al., 2019). It has been reported that another commonly used K14-Cre transgenic mouse line exhibited strong Cre activity in the oral epithelium except in the palatal periderm (Lane et al., 2014). The differences in epithelial adhesion phenotypes in the Ripk4^−/− and Ripk4^EKO mouse embryos suggest that RIPK4 plays a critical cell-autonomous role in the oral periderm in preventing aberrant intraoral epithelial adhesions.

Oberbeck et al (2019) recently generated and analyzed Ripk4^D161N/D161N mice that express only catalytically inactive RIPK4 and showed that they phenocopied Ripk4^−/− mice, indicating that the RIPK4 kinase activity is crucial for periderm and epidermal development (Oberbeck et al., 2019). They compared RNA-seq gene expression profiling of IRF6-dependent genes with that of RIPK4-dependent genes in E15.5 embryonic epidermis, and found that all the significant changes in gene expression seen in the Ripk4^D161N/D161N epidermis were contained within the changes seen in Irf6^−/− embryonic epidermis, suggesting that RIPK4 primarily acts through regulating IRF6 function during epidermal development (Oberbeck et al., 2019). Furthermore, generation and analysis of Irf6^{S413A,S424A/S413A,S424A} mice carrying mutations in both the RIPK4 phosphorylation sites in the IRF6 protein confirmed that phosphorylation at these sites is essential for IRF6 function in vivo. Moreover, Oberbeck et al (2019) generated and analyzed Irf6^{S413E,S424E/S413E,S424E} mice, analysis of which led to identification of an additional RIPK4-dependent phosphorylation site at Ser90 in the IRF6 protein essential for its in vivo function in epidermal development (Oberbeck et al., 2019). A Ser90Gly mutation in IRF6 has been reported in VWS patients whereas a Ser424Leu mutation has been reported in a PPS patient (Kondo et al., 2002; Matsuzawa et al., 2010), indicating that RIPK4 directly regulates IRF6 function in human orofacial and epidermal development.

3.4. IKKα is required for periderm differentiation, and loss of IKKα function causes ectodermal dysplasia with orofacial clefts.

Three independent gene-knockout mutations in Chuk (conserved helix-loop-helix ubiquitous kinase), encoding the Inhibitor of nuclear factor kappa-B kinase subunit alpha (IKKα) (DiDonato et al., 1997), caused similarly severe epidermal developmental defects in homozygous mutants, with rudimentary limbs encased inside a shiny body skin, absent or severely truncated tail, and short rounded face (Hu et al., 1999; Li et al., 1999; Takeda et al., 1999). Richardson et al. (2014) performed direct comparative analysis of Irf6^R84C/R84C and Chuk^−/− mouse embryos at multiple developmental stages and found that they displayed similar defects in periderm formation and aberrant epithelial adhesions (Richardson et al., 2014). A recent report on studies of tongue papillae development showed similar intraoral adhesions, with vertically oriented palatal shelves wedged in between the lateral sides of the tongue and mandible, in Chuk^−/− embryos (Kawasaki et al., 2016). Interestingly, although IKKa was initially identified as a kinase that phosphorylates Inhibitor of kB in response to cytokine signaling (DiDonato et al., 1997), IKKa function in epidermal development does not depend on the kinase activity nor acting through NFkB (Cao et al., 2001; Hu et al., 2001). Sil et al. (2004) observed that induction of keratinocyte differentiation was associated with nuclear localization of IKKa and identified a nuclear localization signal sequence necessary for IKKa to induce keratinocyte differentiation (Sil et al., 2004). Expression of Irf6 and Jag2 mRNAs in the oral epithelium was unaltered in Chuk^−/− embryos in comparison with wildtype embryos, and Chuk^−/−;Irf6^−/− double mutants showed similar intraoral epithelial adhesions as the Chuk^−/− and Irf6^−/− single mutants (Kawasaki et al., 2016). How IKKa and IRF6 converge on regulating periderm differentiation is not known. Xin et al. (2011) showed that primary keratinocytes from E18.5 Chuk^−/− embryos displayed a significantly reduced amount of cleaved Notch1 intracellular domain (NICD), as well as reduced expression of Notch pathway target genes. Furthermore, they showed that lentiviral mediated overexpression of NICD enhanced spinous cell differentiation and tight junction formation in Chuk^−/− keratinocytes (Xin et al., 2011). Whether IKKa regulates periderm differentiation through modulating Notch signaling remains to be determined.

Lahtela et al. (2010) studied aborted fetuses with a lethal popliteal pterygia condition, termed cocoon syndrome, that showed similar phenotypes as the Chuk^−/− mice, and discovered a homozygous nonsense mutation (p.Gln422*) in CHUK exon-12 (Lahtela et al., 2010). Richardson et al. (2014) showed that the CHUK^G422*/G422* human fetal epidermal tissues lacked periderm formation and displayed pathological epithelial adhesions similar to Chuk^−/− mouse embryos, confirming that loss of IKKa function causes peridermopathy in human as well (Richardson et al., 2014). Furthermore, Leslie et al. (2015) performed whole exome sequencing of a patient diagnosed with BPS1 with bilateral CLP but did not carry a mutation in IRF6 or RIPK4; they identified a homozygous mutation in the splice acceptor site of CHUK exon-10 (Leslie et al., 2015). Khandelwal et al. (2017) reported a de novo missense mutation in CHUK (c.425A>G, p.His142Arg) in a 10-year-old patient who showed complex ectodermal dysplasia with cleft palate and oral synechia (Khandelwal et al., 2017). More recently, Cadieux-Dion et al. (2018) reported trans-heterozygous pathogenic mutations in CHUK in a patient initially diagnosed as having Ankyloblepharon-Ectodermal defects-Cleft lip/palate (AEC) syndrome (Cadieux-Dion et al., 2018). Thus, mutations in CHUK have been associated with several related ectodermal dysplasia syndromes with orofacial clefting.

3.5. P63 is a key regulator of periderm differentiation, and TP63 mutations have been associated with syndromic and non-syndromic orofacial clefts.

The mammalian Trp63 gene, encoding transformation related protein 63 (p63), was first identified in 1997 from a unexpected PCR product in a study amplifying the rat homolog of the mouse Glut4 gene, that showed high sequence homology to Trp53 (Schmale and Bamberger, 1997). In contrast to the ubiquitous expression of Trp53, Trp63 expression was found highly restricted in epithelial tissues, particularly basal epithelial cells, and the gene was initially named keratinocyte transcription factor (KET) (Schmale and Bamberger, 1997). Yang et al. (1998) reported that Trp63 produces at least six different transcripts due to the use of two distinct transcription start sites (TA and ΔN) and three alternative splicing routes (α, β, γ) in mouse and human tissues (Soares and Zhou, 2018; Yang et al., 1998). RT-PCR and immunohistochemical analyses revealed that ΔNp63α is the predominant isoform expressed in epithelial tissues (Yang et al., 1998). Mice deficient in p63 exhibited a thin aplastic epidermis and lack of epidermal derivatives (Mills et al., 1999; Yang et al., 1999). Mice with specific deletion of the ΔNp63 isoforms largely phenocopied the epithelial defects in Trp63^−/− mice, including lack of oral epithelial stratification, indicating that ΔNp63 is responsible for most p63 function in epithelial lineages (Romano et al., 2012). Subsequent studies have shown that p63 plays critical roles in multiple stages of epidermal development, including promoting proliferation, cell adhesion, and stratification while suppressing terminal differentiation (Carroll et al., 2006; Koster et al., 2007; Nguyen et al., 2006; Pellegrini et al., 2001; Senoo et al., 2007; Soares and Zhou, 2018; Truong et al., 2006).

The Trp63^−/− mice displayed complete penetrance of bilateral cleft lip and palate (Thomason et al., 2008). At E11.5, Trp63^−/− embryos exhibited disorganized facial epithelium with increased apoptosis. While the outer surface of the facial prominences in wildtype embryos appeared smooth and covered by the newly formed flat periderm cells, Trp63^−/− embryos exhibited rounded cells that were sloughing off from the lateral nasal and maxillary processes (Thomason et al., 2008). At E12.5, when wildtype embryos formed a layer of Krt17+ periderm cells with characteristic flattened morphology at the surface of the palatal shelves, Trp63^−/− palatal shelves displayed patchy rounded cells above a disorganized basal epithelium with no detectable Krt17 expression, indicating that p63 is required for periderm formation (Richardson et al., 2017). Thomason et al. showed that p63 is involved in activation of Irf6 gene expression through binding to an upstream enhancer, and that Irf6 mRNA expression was significantly reduced in the oral epithelium in Trp63^−/− embryos (Thomason et al., 2010). In the Irf6^R84C/R84C, Ripk4^−/−, and Chuk^−/− embryos, high levels of p63 protein persisted in the suprabasal cells in the epidermis, in contrast to the down-regulation of p63 in differentiating suprabasal epidermal cells in wildtype embryos (De Groote et al., 2015; Richardson et al., 2006; Richardson et al., 2014). Moretti et al (2010) reported that IRF6 promoted proteosome-mediated p63 protein degradation in human keratinocytes (Moretti et al., 2010). In addition, Notch signaling has been shown to repress p63 expression in the surface ectoderm at the time of periderm specification (Tadeu and Horsley, 2013). Notch signaling could also activate Irf6 expression and IRF6 mediated p63 down-regulation downstream of Notch in cultured keratinocytes (Restivo et al., 2011). Together, these data suggest that IRF6 converges with Notch signaling to drive periderm and epidermal cell differentiation through feedback downregulation of p63.

Mice homozygous for a classic radiation-induced mutation, repeated epilation (Er), which was later found to be a 1-bp insertion in the coding sequence of the Stratifin (Sfn) gene, displayed very similar intraoral epithelial adhesions and epidermal defects as in the Irf6^R84C/R84C and Ripk4^−/− mutants (De Groote et al., 2015; Herron et al., 2005; Li et al., 2005; Richardson et al., 2006; Tassin et al., 1983). Sfn encodes a cell cycle regulator, and is highly expressed throughout the developing epidermis (Richardson et al., 2006). However, two independent gene-knockout alleles, including one deleting the entire Sfn gene, resulted in surviving homozygous mice with no apparent epidermal development defect except disorganized hair (Su et al., 2011; Winter et al., 2016), indicating that disruption of periderm and epidermal development in Sfn^Er/Er mice resulted from a neomorphic, unnatural function of the mutant gene product. One study suggested that Sfn could interact with ΔNp63 and mediate its nuclear export for proteosome-mediated degradation upon DNA damage in cancer cells (Fomenkov et al., 2004). The persistence of high p63 levels in suprabasal epidermal cells in the Sfn^Er/Er embryos (Richardson et al., 2014) suggests that a truncated Sfn product caused aberrant ΔNp63 stabilization during periderm and epidermal development, and may serve as a tool for dissecting the biochemical mechanism of IRF6-mediated p63 regulation.

Through a forward genetic screen, Lee et al. (2013) discovered that mice deficient in Kdf1 (keratinocyte differentiation factor-1) displayed pathological adhesions and epidermal development defects similar to the Sfn^Er/Er mice (Lee et al., 2013). Kdf1 encodes a novel cytoplasmic protein expressed throughout the developing epidermis. Kdf1^−/− embryos displayed strong p63 protein expression throughout the epidermal layers and failed to develop a functional epidermal barrier (Lee et al., 2013). Epidermal development was largely rescued in Kdf1^−/−;Trp63^+/−embryos (Lee et al. 2013), suggesting that Kdf1 also regulates periderm and epidermal development through modulating p63 expression or stabilization.

Germline mutations in TP63 have been associated with five autosomal dominant ectodermal dysplasia syndromes with orofacial cleft as a hallmark feature in four of them, including AEC (also called Hay-Wells syndrome, OMIM 106260), Ectrodacdyly, Ectodermal dysplasia, and Cleft lip/palate syndrome (EEC, OMIM 604292), Limb-Mammary Syndrome (LMS, OMIM 603543), and Rapp-Hodgkin Syndrome (RHS, OMIM 129400) (Osterburg et al., 2021; Soares and Zhou, 2018). The TP63 mutations, mostly missense, that cause these diseases are believed to act via a dominant-negative or gain-of-function mechanism, and analyses of mouse models with knock-in missense mutations found in AEC or EEC patients resembled the human phenotypes (Celli et al., 1999; Ferone et al., 2012; McGrath et al., 2001; Rinne et al., 2009; Vernersson Lindahl et al., 2013). However, several heterozygous deletions and frame-shift or nonsense mutations have recently been identified in multiple families with orofacial clefting and mild ectodermal defects (Khandelwal et al., 2019). Interestingly, Khandelwal et al. showed that those TP63 loss-of-fucntion variants were also present in a unaffected parent, indicating that TP63 loss-of-function is a predisposing risk factor for orofacial clefting (Khandelwal et al., 2019). Another study identified a heterozygous 2-bp insertion that caused a frameshift in the p63 protein in a father and son with non-syndromic CLP (Basha et al., 2018). A genome-wide meta-analysis of CLP reported strong association of a SNP in an enhancer sequence in the last intron of the TP63 gene with non-syndromic CLP (Leslie et al., 2017). Thus, mutations in TP63 are a significant contributor to syndromic and non-syndromic CLP.

4. CONNECTIONS BETWEEN EPITHELIAL INTEGRITY AND EPITHELIAL-MESENCHYMAL INTERACTIONS REGULATING OROFACIAL MORPHOGENESIS

Studies of mice deficient in Irf6, Chuk, Jag2, and Ripk4 have clearly demonstrated that the embryonic periderm forms a protective barrier to prevent pathological epithelial adhesions that may disrupt normal orofacial morphogenesis (Casey et al., 2006; Hammond et al., 2019; Richardson et al., 2014). The CLP phenotype of the Trp63^−/− mutant mice, on the other hand, indicates that disruption of facial epithelial integrity prior to formation of the periderm is also detrimental to facial and palate morphogenesis (Thomason et al., 2008). A recent single-cell RNA-seq analysis of the dorsal skin from E13.5 Trp63^−/− and wildtype embryos showed that a portion of the embryonic epithelial cells displayed strong enrichment of expression of mesenchymal genes, reminiscent of epithelial-to-mesenchymal transition signatures observed in squamous cell carcinoma during tumorigenic epithelial-mesenchymal transition (EMT) (Fan et al., 2018). By studying epidermal development from specification of multipotent simple epithelium to basal stratified epithelia to mature epidermis using human induced pluripotent stem cells (iPSCs), Soares et al. found that EEC iPSCs displayed defects during the specification switch from simple epithelium to stratified epithelia and aberrant activation of mesodermal gene expression (Soares et al., 2019), consistent with the data from mutant mouse studies. These results indicate that p63 plays a critical role in maintaining epithelial identity and stability of the embryonic surface epithelium, consistent with recent finding that p63 directly regulates expression of a number of epithelial cell adhesion molecules during palate development (Richardson et al., 2017). An important role of p63-mediated regulation of epithelial cell adhesion in facial morphogenesis is also supported by the identification of homozygous nonsense or frameshift mutations in NECTIN1 (also known as PVRL1, poliovirus receptor like-1), a p63 direct target gene that encodes an immunoglobulin-like adhesion molecule important in desmosome assembly, as the cause of cleft lip/palate-ectodermal dysplasia syndrome (CLPED1, OMIM 225060) (Barron et al., 2008; Mollo et al., 2015; Suzuki et al., 2000). NECTIN1 heterozygosity is also a significant risk factor for non-syndromic CLP (Sozen et al., 2001).

The p63 transcription factor also plays essential roles in regulating epithelial-mesenchymal interactions controlling facial primordial growth. Trp63^−/− embryos exhibited marked changes in expression of several key signaling molecules, including Bmp4 and Fgf8, in the facial epithelium, accompanied by increased apoptosis of the epithelium and significantly decreased proliferation of the mesenchyme in the lateral nasal and maxillary processes (Thomason et al., 2008). By ChIP-seq and RNA-seq analysis of Trp63^−/− and wildtype mouse embryonic epidermal cells, Fan et al. showed that p63 directly regulates expression of several components of the BMP, FGF, Notch, and Wnt pathways at the onset of epidermal fate specification, including Bmp7, Fgfr2, Fgfr3, Jag2, Notch1, Wnt3a, Wnt10b, and Fzd6 (Fan et al., 2018). These signaling pathways play essential roles in early facial primordial growth (Jiang et al., 2006; Lan et al., 2006; Reynolds et al., 2020). In addition, Ferreti et al. showed that expression of Trp63 in the early facial ectoderm is regulated by Wnt signaling, and suggested a Pbx-Wnt-p63-Irf6 regulatory module controlling facial morphogenesis (Ferretti et al., 2011).

Ferone et al. showed that the Trp63^+/L514F mice, heterozygous for the same missense mutation as found in AEC patients, were born with cleft palate and several other ectodermal tissue defects including thinner epidermis, and blistering skin lesions in the basal and suprabasal layers of the epidermis (Ferone et al., 2012). At E13, the Trp63^+/L514F embryos displayed significantly smaller palatal shelves with disorganized palatal epithelium containing patchy Krt17+ periderm cells intermingled with Krt14+ basal epithelial cells, and the periderm cells lacked the characteristic flattened morphology observed in wildtype embryos (Ferone et al., 2012). The Trp63^+/L514F embryos exhibited significantly reduced proliferation of both the palatal epithelium and mesenchyme, resembling that in Fgfr2b^−/− and Fgf10^−/− mouse embryos, which also displayed defective palatal periderm and reduced palatal shelf growth (Alappat et al., 2005; Rice et al., 2004). Ferone et al. (2012) further demonstrated that p63 directly regulated Fgfr2 and Fgfr3 expression through binding to their cis-regulatory elements (Ferone et al., 2012). Fgfr2b signaling, likely through regulating epithelial cell proliferation and integrity, maintains Shh expression in the palatal epithelium, and Shh signaling regulates expression of Fgf10 and Fgf18 in the palatal mesenchyme to coordinate palatal shelf growth (Lan and Jiang, 2009; Rice et al., 2004; Xu et al., 2016). In addition, Fgf10-Fgfr2 signaling also regulates palatal epithelial integrity, as Fgf10^−/− embryos exhibited ectopic expression of Tgfb3 in the lateral side of the palatal epithelium with subsequent fusion of the palatal shelves to the mandible (Alappat et al., 2005). Interestingly, mouse embryos lacking the BMP antagonist Noggin displayed a similar palate-mandible fusion defect without affecting Fgf10 expression, suggesting that Fgf10-Fgfr2 signaling may regulate palatal epithelial integrity through modulating BMP signaling (He et al., 2010). Thus, in addition to directly regulating expression of cell adhesion molecules in the palatal epithelium (Richardson et al., 2017), p63 also regulates both palatal epithelial integrity and palatal shelf growth through Fgfr2b signaling.

Another example that maintaining epithelial integrity plays a key role in facial primordial growth and morphogenesis has been provided by recent studies of Grhl2 mutant mice. Grhl2 is expressed in the surface ectoderm starting at E8.5 and directly regulates expression of a number of genes essential for epithelial fate, including epithelial adherens and tight junction molecules E-cadherin and Cldn4, and the epithelial-specific splicing factors Esrp1 and Esrp2 (Carpinelli et al., 2020; Ray and Niswander, 2016; Tanimizu and Mitaka, 2013; Werth et al., 2010). Grhl2^−/− mouse embryos exhibited a neural tube closure defect and a cleft face, with significantly smaller maxillary processes by E10.5 (Carpinelli et al., 2020; Pyrgaki et al., 2011). Through RNA-seq analysis, Carpinelli et al. found that the Grhl2^−/− embryos displayed significantly increased expression of Zeb1 and many mesenchymal-specific transcripts, in addition to significant downregulation of epithelial genes (Carpinelli et al., 2020). Zeb1 is a major regulator of EMT and is a repressor of Cdh1 expression (Nieto et al., 2016; Vannier et al., 2013). Remarkably, in contrast to the mid-gestation lethality of and severe craniofacial defects in Grhl2^−/− embryos, the Grhl2^−/−;Zeb1^−/− embryos displayed a nearly normal nasal septum and fused secondary palate (Carpinelli et al., 2020), indicating that a key role for GRHL2 in facial and palate development is to maintain epithelial integrity through suppression of EMT. Together with a previous study identifying a single nucleotide mutation in the first intron of the Zeb1 gene as the cause of cleft lip and palate in the Twirler mutant mice (Kurima et al., 2011), these results indicate that gain of function of Zeb1 is a risk factor for orofacial clefting through disruption of epithelial integrity during facial morphogenesis.

Mice lacking Esrp1 exhibited complete penetrance of cleft lip and palate (Bebee et al., 2015). Lee et al. showed that the Esrp1^−/− embryos displayed significantly reduced cell proliferation in both the epithelium and mesenchyme of the medial and lateral nasal processes, accompanied by alterations in expression of genes in several signaling pathways, particularly multiple Wnt family ligands and Shh, at E10.5 (Lee et al., 2020). They performed RNA-seq analysis of E12.5 facial epithelium and mesenchyme, and identified 1467 alternative splicing changes in Esrp1^−/− embryos, with many RNAs switched from epithelial to mesenchymal isoforms. The validated epithelial-to-mesenchymal isoform switching in the Esrp1^−/− embryos disrupted multiple molecules crucial for adherens and tight junction assembly or maintenance, including Ctnnd1 (p120 catenin), Arhgap17, Magi1, and Scrib (Lee et al., 2020; Warzecha and Carstens, 2012). Of these, Ctnnd1 directly binds to E-cadherin and is a major component of the adherens junction complex (Pieters et al., 2012). A crucial role for maintaining epithelial adherens junction stability in facial morphogenesis is supported by human genetic studies identifying mutations in CDH1 and CTNND1 in several syndromes with cleft lip or cleft palate as a variable feature (Alharatani et al., 2020; Frebourg et al., 2006; Ghoumid et al., 2017; Selvanathan et al., 2020). Loss of function variants in CDH1 and CTNND1, respectively, have also been associated with non-syndromic CLP (Cox et al., 2018; Dixon et al., 2011; Song and Zhang, 2011). In addition, conditional inactivation of Ctnnd1 in the embryonic facial ectoderm caused cleft lip or palate in mice (Cox et al., 2018). Thus, Esrp1 mediated epithelial-specific splicing plays crucial roles in facial morphogenesis at least in part though maintaining epithelial integrity.

5. THE ROLES OF THE PERIDERM AND THE IRF6-P63 PATHWAY IN PALATAL FUSION

Whereas the periderm acts as a protective barrier to prevent epithelial adhesion during palatal shelf growth, the bilateral palatal shelves must adhere to each other at the midline, and the resultant inter-shelf midline epithelial seam (MES) has to be dissolved, during palatal fusion to form the intact roof of the oral cavity (Lan et al., 2015; Li et al., 2017). Palatal fusion is driven by TGFβ3, whose expression is specifically activated in the palatal shelf medial edge epithelium (MEE) prior to palatal shelf adhesion (Fitzpatrick et al., 1990; Kaartinen et al., 1995; Pelton et al., 1990; Proetzel et al., 1995). In Tgfb3^−/− embryos, the palatal shelf growth and reorientation to the horizontal position occurred normally, but the bilateral palatal shelves adhered poorly and exhibited intact periderm layer over their surface (Hu et al., 2015; Kaartinen et al., 1997; Richardson et al., 2017). Interestingly, the Snai1^+/−;Snai2^−/− mouse embryos exhibited cleft palate with an intact periderm layer over the palatal shelves, similar to the Tgfb3^−/− embryos, but they displayed apparently normal patterns of expression of Tgfb3 mRNAs in the MEE (Murray et al., 2007). Snai1 and Snai2 have been implicated in driving EMT by downregulating expression of cell adhesion molecules including E-cadherin (Carver et al., 2001; Liu et al., 2012; Peinado et al., 2007). Although the role of EMT in palatal fusion has long been controversial since no genetically labelled palatal epithelial cells contributed to the post-fusion palatal mesenchyme structures (Bush and Jiang, 2012; Gritli-Linde, 2007; Jiang et al., 2006; Nawshad, 2008; Xu et al., 2006), the palatal defect in the Snai1^+/−Snai2^−/− embryos suggest that Snai1 and Snai2 act partly redundantly downstream of TGFβ3 signaling to drive palatal fusion.

How do the periderm cells respond to TGFβ3-Snai1/2 signaling, and do they play an active role in palatal fusion? Whereas some studies have suggested that the medial edge periderm cells undergo programmed cell death and are sloughed off immediately prior to palatal shelf adhesion (Fitchett and Hay, 1989; Hu et al., 2015), other studies showed that the medial edge periderm cells change shape and extend filapodial or lamellipodial processes at the beginning of palatal shelf adhesion, and migrate away from the MES to the oral and nasal surfaces during palatal fusion (Richardson et al., 2017; Taya et al., 1999; Waterman et al., 1973). Li et al. recently reported that periderm cells were also present and changed shape in the fusion zone between the medial and lateral nasal processes during lip fusion in mouse embryos (Li et al., 2019). Using live imaging of mouse palatal explants, Kim et al. (2015) demonstrated that MEE cells extend protrusions to contact the opposite shelf to establish cellular bridges that give rise to a transient multilayered epithelial seam where periderm cells were still present (Kim et al., 2015). Subsequently, palatal fusion proceeded through concurrent cellular intercalation and displacement in the oronasal axis by convergent extension driven by an actomyosin pathway involving non-muscle myosin-IIA activation (Kim et al., 2015). While it was an elegant study that has provided unprecedented insight into the cellular mechanisms of palatal fusion, the study used K14-Cre mediated lineage labeling, which does not label the medial edge periderm cells (Lane et al., 2014) and, hence, did not address specifically periderm cell behavior during palatal fusion. Richardson et al. examined periderm cell fate during palatal fusion by using confocal imaging of palatal shelf explants from mouse embryos carrying a Krt17-GFP transgenic reporter and observed migration of periderm cells out of the MES into the oral and nasal epithelia of the wildtype palatal shelves but not of the Tgfb3^−/− palatal shelf explants, despite forced contact of the palatal shelves (Richardson et al., 2017). Furthermore, Kim et al. observed that Tgfb3^Cre/+;Myh9^lox/lox embryos displayed failure of cell-intercalation at the MES by E15.5 when the MES had almost completely dissolved in wildtype embryos, whereas the K14-Cre;Myh9^lox/lox embryos also exhibited significant MES dissolution (Kim et al., 2015). Since the K14-Cre mice exhibit robust Cre activity in basal epithelial cells but lack Cre-mediated recombination in palatal periderm cells, whereas Tgfb3^Cre induces recombination in both the basal and periderm layers of the MEE (Lane et al., 2014), could the difference in MES dissolution observed in between the Tgfb3^Cre/+;Myh9^lox/lox and K14-Cre;Myh9^lox/lox embryos be due to a contribution of the periderm cells in facilitating MES formation? In this regard, it is worth noting that eyelid closure in the mouse fetus also requires myosin-IIA driven cell intercalation and a Krt17-Cre lentivirus-mediated selective inactivation of myosin-IIA in the eyelid front cells in Myh9^lox/lox embryos caused failure of eyelid closure (Heller et al., 2014).

An active role for the medial edge periderm cells in MES formation in response to TGFβ3 signaling could also be deduced from previous studies of the K14-Cre;Tgfbr1^f/f and K14-Cre;Tgfbr2^f/f mice. In contrast to the lack of proper adhesion of the bilateral palatal shelves in the Tgfb3^−/− embryos resulting in complete cleft palate at birth, the palatal shelves in the K14-Cre;Tgfbr1^f/f and K14-Cre;Tgfbr2^f/f embryos adhered and formed the MES spanning about 90% of the anteroposterior length of the secondary palate by E14.5, but the MES persisted in these conditional mutant embryos resulting in submucous cleft in the anterior two-thirds and complete cleft of the posterior part of the palate (Dudas et al., 2006; Xu et al., 2006). Whereas the milder palatal phenotype in the K14-Cre;Tgfbr1^f/f and K14-Cre;Tgfbr2^f/f mice than in the Tgfb3^−/− mice was likely due to remaining TGFβ signaling in the periderm in the K14-Cre;Tgfbr1^f/f and K14-Cre;Tgfbr2^f/f embryos (Lane et al., 2014), the K14-Cre;Tgfbr1^f/f and K14-Cre;Tgfbr2^f/f embryos have not been analyzed to specifically address the role of the periderm in palatal fusion. Interestingly, while the lack of contribution of K14-Cre lineage cells to the post-fusion palatal mesenchyme and the lack of apoptosis in the MES in the K14-Cre;Tgfbr1^f/f and K14-Cre;Tgfbr2^f/f embryos were presented as strong evidence supporting TGFβ signaling induced apoptosis, rather than EMT, as the predominant cellular mechanism of palatal fusion (Gritli-Linde, 2007; Vaziri Sani et al., 2005), no apoptosis was detected in the periderm in either K14-Cre;Tgfbr1^f/f or the K14-Cre;Tgfbr2^f/f embryos despite the belief that formation of the MES in these embryos, as opposed to the failure of MES formation in the Tgfb3^−/− embryos, was probably due to TGFβ-induced shedding of the medial edge periderm cells (Dudas et al., 2006; Lane et al., 2014; Xu et al., 2006). Furthermore, both K14-Cre;Tgfbr1^f/f and K14-Cre;Tgfbr2^f/f embryos displayed an epithelial bridge in between the palatal shelves that was progressively thinner from the midpalate toward the posterior palate until the gap in the extreme posterior region, but there was no observation of rupture of the MES or discontinuity of the periderm across the midline bridge (Dudas et al., 2006; Xu et al., 2006). Altogether, the existing data do not support the hypothesis that periderm undergoes apoptosis and slough off prior to palatal shelf adhesion, but rather are consistent with the periderm playing an active role in initiating palatal shelf adhesion in response to TGFβ3 signaling and in maintaining palatal epithelial integrity through the palatal fusion processes.

While the exact role of and the mechanism involving the medial edge periderm cells in palatal fusion remain to be resolved, it is remarkable that the mammalian embryo re-deploys the IRF6-P63 pathway used for periderm differentiation to facilitate dissolution of the MES during palatal fusion. Elegant mouse genetic studies have shown that activation of Irf6 expression in the basal cells of the MEE is essential for TGFβ3-induced palatal fusion, and that IRF6 acts downstream of TGFβ-Smad4 signaling to shut down the p63-mediated proliferation and cell-adhesion regulatory network in the MES cells to facilitate MES breakdown (Iwata et al., 2013; Richardson et al., 2017). Interestingly, the failure of MES breakdown in the Tgfb3^Cre/+;Myh9^lox/lox embryos was accompanied by persistent high levels of p63 protein in the MES cell nuclei (Kim et al., 2015), suggesting that biomechanical forces generated by myosin-IIA driven cell intercalation movement play an important role in regulating IRF6-mediated downregulation of p63 in the MES cells. How these biomechanical, biochemical, and cellular mechanisms are integrated downstream of TGFβ3 signaling to achieve palatal fusion while maintaining epithelial tissue integrity remains to be resolved.

6. CONCLUDING REMARKS

Formation of the intact upper lip and palate involves not only coordinated directional growth and patterning of the facial primordia but also timely adhesion and fusion of initially separate primordia, which make facial development particularly susceptible to the influence of any deficiency in epithelial integrity. A critical role of epithelial integrity in orofacial development is demonstrated by findings that loss-of-function mutations in CDH1 and CTNND1 are associated with syndromic and non-syndromic CLP in human patients, and that conditional inactivation of Ctnnd1 in the embryonic facial ectoderm resulted in cleft lip and palate in mice (Alharatani et al., 2020; Cox et al., 2018; Frebourg et al., 2006; Ghoumid et al., 2017; Selvanathan et al., 2020). In addition to providing the structural foundation for the spatiotemporal regulation of many signaling molecules for directing the growth and patterning of the facial primordia, epithelial integrity is critical for generating the biomechanical forces in the surface epithelium to push the maxillary and lateral nasal processes medio-frontally toward meeting the medial nasal process to initiate contact, fusion and filling of the midline groove between the bilateral medial nasal processes. A recent study showed that Grhl2-mediated regulation of surface ectoderm integrity is also required for the cranial neural folds to move toward the midline for neural tube closure (Ray and Niswander, 2016). Previous epidemiological studies suggested that there is a significant etiological overlap between neural tube closure defects and CLP, with the presence of one of these two traits in a family seeming to increase risk of occurrence of the other trait in other individuals (Juriloff and Harris, 2008). GHRL2 regulates the integrity of the surface epithelium during both neural tube closure and facial morphogenesis through activating expression of genes encoding epithelial cell-cell adhesion molecules, as well as suppressing expression of genes driving EMT (Carpinelli et al., 2020; Ray and Niswander, 2016). The rescue of facial and palate development in the Ghrl2^−/−;Zeb1^−/− embryos indicate that multiple molecular pathways, including other transcription factors and epithelial-specific splicing factors, converge on regulating epithelial cell adhesion to ensure epithelial integrity during facial morphogenesis (Carpinelli et al., 2020). Both GHRL2 and p63 directly regulate expression of genes encoding E-cadherin and other epithelial cell junction molecules as well as Esrp1, whereas Esrp1 function is important in generating many components and regulators of epithelial cell junction assembly (Carpinelli et al., 2020; Lin-Shiao et al., 2019; Richardson et al., 2017; Warzecha and Carstens, 2012; Werth et al., 2010). The expression and function of these regulators of epithelial cell-cell adhesion are further modulated by complex regulatory processes including epigenetic modifications such as histone modifications, DNA methylation, and miRNAs, as well as additional post-transcriptional and post-translational modifications (Nieto et al., 2016). While current orofacial cleft research is largely focused on identifying the genetic variants that cause the cleft phenotype, elucidating the regulatory networks that control epithelial integrity is necessary in order to understand the complex etiology and pathogenic mechanisms of orofacial clefting.

A key event for maintaining integrity of the surface epithelium during facial development is the formation of the periderm. The periderm provides a diffusion barrier and non-stick apical surface covering the entire embryo. As studies of the Irf6^−/−, Jag2^−/−, Ripk4^−/− mutant mice and of the mouse embryos with specific genetic ablation of periderm have demonstrated, defects in periderm formation and/or maintenance cause pathological epithelial adhesions that disrupt orofacial morphogenesis. Whereas these studies have identified several key regulators of periderm formation or maintenance, our understanding of periderm differentiation and maintenance is still quite primitive. How is the periderm cell fate specified, and what regulates the stratification process of periderm layer formation? How do the IRF6-GRHL3 and Jag2-Notch pathways regulate periderm differentiation and integrity? Does the periderm layer expand as the embryo grows by proliferation within the periderm layer or by addition of new periderm cells differentiated from the basal epithelial cells? Whereas several studies have implicated TFAP2A as another crucial transcription factor in the transcriptional regulation of Irf6 expression, and other studies have shown that mice deficient in several cytoplasmic proteins, such as ARHGAP29 and SPECC1L, exhibited aberrant oral epithelial adhesions (Hall et al., 2020; McDade et al., 2012; Paul et al., 2017; Rahimov et al., 2008), more research efforts and innovative technologies directed at elucidating the molecular network and cellular dynamics regulating periderm differentiation and maintenance are needed to understand pathogenic mechanisms involving periderm dysfunction. Furthermore, the role and regulation of periderm behavior in palatal fusion remain to be resolved.

Lastly, understanding the significant contribution of epithelial instability to orofacial cleft pathogenesis has strong implications in patient care. Defects in epithelial differentiation and/or integrity significantly affect wound healing and cancer metastasis (Frisch et al., 2017; Nieto et al., 2016; Sundararajan et al., 2020). It has been reported that children with VWS have an increased risk of wound complications after cleft repair (Jones et al., 2010). Both Grhl2 and Grhl3 have been shown to play critical roles in embryonic wound repair in mice (Boglev et al., 2011). Variants in many regulators of epithelial cell differentiation, as well as epithelial cell adhesion molecules such as E-cadherin, have been associated with cancer (Nieto et al., 2016). Epidemiological studies have shown that orofacial cleft patients have increased cancer risk (Leslie and Marazita, 2013; Seto-Salvia and Stanier, 2014). Thus, translating the understanding of the contributions of epithelial instability to orofacial cleft pathogenesis will help improve clinical management and family counseling in craniofacial patient care.