null
Dear Editor,
Craniofacial microsomia (CFM, MIM#164210) is a congenital malformation involving the first and second branchial arch derivatives. The phenotype of CFM is highly variable and typically affects the external ear, middle ear, mandible and temporomandibular joint, and facial muscles on the affected side. Accompanied by craniofacial anomalies, cardiac, vertebral, and central nervous system defects may occur. Microtia is considered the minimum diagnostic criterion [
1,
2]. CFM is also reported as hemifacial microsomia, Goldenhar syndrome, or oculo-auriculo-vertebral spectrum (OAVS) owing to its overlapping clinical manifestations. The typical clinical features, such as auricular malformation and underdeveloped mandible on one or both sides, may occur as isolated malformations or present as clinical features in some syndromic brachial anomalies, such as branchiootorenal syndrome (BOR), branchiooculofacial syndrome, and mandibulofacial dysostosis.
CFM is considered a multifactorial disease. Genetic and non-genetic risk factors contribute to the heterogeneous etiology of CFM [
3,
4]. Most cases of CFM are sporadic, and autosomal dominant inheritance has been supported by multigenerational families. Chromosomal imbalances, such as trisomy of 22q, terminal deletion of 5p14, and translocation t(4; 8)(p15.3; q24.1), have been implicated in CFM [
5–
7]. Multiple genes with different functions are responsible for CFM, including the embryologic formation of the ears and mandible involving transcription factors (TFAP2A, SIX1, SIX5, EYA1, EYA3, HOXA10, and HOXA2), chromatin modifiers (CHD7, KMT2D, and KDM6A), growth factors and their receptors (GDF6, FGF3, FGF10, FGFR2, and FGFR3), the machinery of DNA replication (ORC1, ORC4, ORC6, CDC6, and CDT1), transcription (EFTUD2, TXNL4A, SF3B4, and SF3B2), and translation (TCOF1, POL1RC, and POL1RD), and signaling pathway related to retinoic acid (MYT1) [
3,
8,
9]. However, these genes can only explain the limited number of patients with CFM. Moreover, incomplete penetrance and wide phenotypic variability were observed in some families with presumably dominant inheritance, suggesting the possibility of non-Mendelian inheritance involving genetic modifiers. However, no illustrative cases have been reported to support the hypothesis.
A three-generation family with CFM was recruited in the study (Fig.1). The proband, a 7-year-old Han Chinese girl with congenital microtia, was referred to our hospital for ear reconstruction. She had a low-set hypoplastic left ear with atresia of the external auditory canal resulting in conductive hearing loss (Fig.1 and 1C). She also presented with mild facial asymmetry, mandibular retrusion, and malar flattening, implicating maxillary and mandibular hypoplasia on her left side. Physical examination, chest CT, electrocardiogram, echocardiography, and abdominal ultrasound showed no abnormality. Therefore, the diagnosis of the overlapping syndromes of CFM, such as CHARGE syndrome, Miller syndrome, and BORS, was excluded. Family history showed that her mother had congenital microtia on the right side and received ear reconstruction surgery (Fig.1 and 1E). No other family members were affected either from paternal or maternal origin according to her parents’ complain. The pedigree indicated the disease phenotype may have been transmitted through autosomal dominant inheritance between the proband and her mother. Meanwhile, the phenotypic heterogeneity and incomplete penetrance implicate non-Mendelian inheritance in the family. This study was performed according to the Helsinki Declaration and was approved by the institutional review board of the Plastic Surgery Hospital of Chinese Academy of Medical Sciences (Ethic No.2021-7). Each participant signed written informed consent form for clinical examination, genetic analysis, and data release.
Whole genome sequencing was performed by an external sequencing facility (Novogene, Inc, China) to detect variants in the three family members. The average sequencing depths were 36.94×, 30.43×, and 30.55× in the patient, her father, and her mother, respectively. No suspicious structural variations were found. A total of 682 nonsynonymous mutations with population frequency of less than 0.001 and segregated with phenotype were prioritized. Mutation classification based on the American College of Medical Genetics and Genomics (ACMG) guidelines was considered. Genes responsible for or associated with craniofacial development were regarded as the highest priority. Twelve mutations were classified as “pathogenic” and “likely pathogenic” according to the ACMG guidelines [
10]. We found two mutations associated with CFM:
EYA3 (c.A197T p.D66V) and
EFTUD2 (c.A1787G, p.N596S). Both mutations were confirmed in the patient and her affected mother but not in her unaffected father by Sanger sequencing. We further detected the carrying status of the two mutations in unaffected maternal grandparents. Each carried one of the mutations (Fig.1)
Calibrated and uncalibrated in silico algorithms were used for the pathogenicity interpretation of the two mutations. The scores of REVEL, a calibrated tool for predicting the pathogenicity of rare missense variants, were 0.733 and 0.643 for EYA3 c.A197T and EFTUD2 c.A1787G, respectively. We also used the integrated online database of VarCards and considered the ratio of algorithms predicting variants to be deleterious (Table S1). The results showed that 19 of the 23 algorithms predicted EYA3 c.A197T mutations to be deleterious. It is regarded as an extreme variant with a damaging score of 0.83 and allele frequency of less than 0.0001. The variant is not presented in any population exome or genomic data sets in gnomAD. As for the EFTUD2 c.A1787G variant, 20 of the 23 algorithms predicted it to be deleterious with a damaging score of 0.87. The frequency of this variant is 3.655e–05 in gnomAD. It is documented as rs373685919 in the dbSNP database. The alignment of sequences among 46 species showed that both mutated residues were highly conserved during evolution (Fig.1–1N). The 3D molecular structures of the two proteins with mutations were constructed using PyMOL software. The mutated residue caused changes in polar contacts with spatially adjacent amino acids (Fig.1–1L, and 1M). To the best of our knowledge, no direct interaction was reported between these proteins. However, the expression data in the Mouse Genome Informatics database showed that both genes were expressed at the first branchial arch at E9–E10.25, indicating that the genes have similar biological functions at this specific spatiotemporal stage. To further explore the potential interconnection, we queried the STRING database with a set of 41 proteins encoded by genes identified in human syndromes with microtia as a significant feature to build an interconnected network. By consolidating all known and predicted protein–protein interaction information, we found a possible interconnection between EFTUD2 and EYA3 within a network of proteins associated with craniofacial diseases (Fig.1).
EYA3 and
EFTUD2 function in the embryologic formation of the ears and mandible.
EYA3 is associated with OAVS, one of the other terms of CFM. Tingaud-Sequeira
et al. [
11] found a recurrent missense variant (Asn358Ser) in two unrelated families with OAVS from New Zealand and France EYA3 belongs to the family encoding Eyes absent tyrosine phosphatase, which plays a critical role in repair versus apoptosis decision after damage. Interestingly,
EYA1, a paralog of
EYA3, has more than 240 mutations reported in the Human Gene Mutation Database (HGMD) and is responsible for the known overlapping syndromes with OAVS, such as banchiootic syndrome and BOR. The identified D66V mutation in EYA3 is located at the N-terminal part (Fig.1), which presents a threonine phosphatase activity and is partially conserved among EYA proteins during evolution.
EFTUD2 encodes a highly conserved spliceosomal GTPase, the U5-116kD protein [
12]. Heterozygous mutations in
EFTUD2 lead to mandibulofacial dysostosis with microcephaly (MFDGA, MIM 610536), which clinically overlaps with CFM or OAVS [
13]. To date, 128 mutations are documented in the HGMD: 44 point mutations (missense and nonsense mutations), 33 splicing mutations, 36 small indels (insertions and deletions), and 10 genomic rearrangements. These mutations result in haploinsufficiency by the loss-of-function of the mutated allele. The identified N596S mutation is in a highly conserved functional domain, elongation factor G, III domain (Fig.1). The pathogenic missense mutation L620N, located in the same domain as N596S, can lead to exon skipping.
The presence of two pathogenic mutations in the same family have three explanations. (1) Two functional mutations are transmitted in a typical digenic inheritance (True DI). A patient shows clinical manifestation only when he or she carries two non-allelic mutations simultaneously. (2) One mutation is a driver mutation, whereas the other is a modifier (Pseudo-DI). Both mutations follow the digenic inheritance pattern, but the clinical manifestation may vary in the presence or absent of modifier mutations. (3) Coincidental independent segregation of two mutations occurs, and each follows a classic Mendelian mode of inheritance and may be responsible for a separate disease entity. An example is the intrafamily co-inheritance of polycystic kidney disease and Marfan syndrome [
14]. In the present CFM family, the patient and her mother both showed microtia but on a different side. Thus, the possibility of a genetic modifier was not ruled out. Furthermore, we did not find other apparent abnormalities in the patients, except CFM, especially the additional phenotypes of MFDGA syndrome caused by
EFTUD2 mutation. No evidence of overlapping among different disease entities was obtained. Moreover, we were unable to determine the existence of an addictive effect given that no study on phenotypic differences in patients with OAVS had been carried out. Data supporting the proposed digenic inheritance were limited because the family was small for segregation analysis and no direct supporting function data were obtained. Nonetheless, our finding indicated that complex inheritance is involved in CFM.
We observed incomplete penetrance in carriers (I:1 and I:2) of each single mutation in the family. The general explanation of incomplete penetrance is the presence of many additive modifiers with small effects and environmental factors, that is, multifactorial inheritance. No environmental risk factors were reported in the family, such as diabetes mellitus during pregnancy, maternal hypothyroidism, drug use, and smoking. Genetic factors may play a major role because CFM is transmitted from the mother to the daughter in this family. Complex traits controlled by multiple genes are usually much more common than Mendelian disorders and are likely to be due to the interactions among these genes. The possibility of two genes or a small number of gene loci being implicated in one disorder has given rise to the concept of digenic or oligogenic inheritance. It refers to a disorder’s situation due to the additive effects of heterozygous mutations at two or more different gene loci. To some extent, digenic inheritance is the simplest form of polygenic etiology. The pathogenicity of the two variants involved is stronger than that of many modifiers in polygenic inheritance. A growing list of human diseases or congenital malformations are underlying digenic inheritance, such as neural tube defects, congenital heart diseases, and retinitis pigmentosa. Causative genes under digenic inheritance have similar biological functions or co-exist in the same regulatory network. For example, ROM1 and PRPH2, both encoding proteins in photoreceptors, cause retinitis pigmentosa in digenic inheritance [
15]. All genes responsible for digenic familial hypercholesterolemia regulate lipid metabolism [
16]. In this study, we found that EYA3 and EFTUD2 were intact in a functional regulatory network that functioned in craniofacial development. The co-occurrence of the heterozygous mutations of the
EFTUD2 and
EYA3 may lead to CFM. This speculation requires replications in additional cases and additional functional evidence.
The genetic etiology of CFM may be complicated on top of the known Mendelian inheritance. Identifying cases carrying multiple CFM risk mutations is expected to increase through the increased use of whole-genome sequencing. Herein, we reported a CFM family with two likely pathogenic mutations in EYA3 and EFTUD2, two genes known to be associated with CFM. These findings may provide possible cases with digenic inheritance in CFM and further support the role of a complex gene regulation network in craniofacial development. They also provide information for understanding phenotypic variability and genetic heterogeneity in HFM.
PDF
(1291KB)
