Article Text


Distinct CDH3 mutations cause ectodermal dysplasia, ectrodactyly, macular dystrophy (EEM syndrome)
  1. K W Kjaer1,
  2. L Hansen1,
  3. G C Schwabe2,7,
  4. A P Marques-de-Faria4,
  5. H Eiberg5,
  6. S Mundlos2,3,
  7. N Tommerup1,
  8. T Rosenberg6
  1. 1Wilhelm Johannsen Centre for Functional Genome Research, Institute of Medical Biochemistry and Genetics, University of Copenhagen, Copenhagen, Denmark
  2. 2Max Planck Institute for Molecular Genetics, Berlin, Germany
  3. 3Institute for Medical Genetics, Charité, Berlin, Germany
  4. 4Department of Medical Genetics, State University of Campinas School of Medicine, Campinas, SP, Brazil
  5. 5Institute of Medical Biochemistry and Genetics, University of Copenhagen, Copenhagen, Denmark
  6. 6Gordon Norrie Centre for Genetic Eye Diseases, National Eye Clinic for the Visually Impaired, Hellerup, Denmark
  7. 7Department of Endocrinology, Children’s Hospital Charité, Berlin, Germany
  1. Correspondence to:
 K W Kjaer
 Wilhelm Johannsen Centre for Functional Genome Research, Institute of Medical Biochemistry and Genetics, Panum Institute 24.4, 2200 Copenhagen N, Denmark;


Background: EEM syndrome is the rare association of ectodermal dysplasia, ectrodactyly, and macular dystrophy (OMIM 225280).

Methods: We here demonstrate through molecular analysis that EEM is caused by distinct homozygous CDH3 mutations in two previously published families.

Results: In family 1, a missense mutation (c.965A→T) causes a change of amino acid 322 from asparagine to isoleucine; this amino acid is located in a highly conserved motif likely to affect Ca2+ binding affecting specificity of the cell-cell binding function. In family 2, a homozygous frameshift deletion (c.829delG) introduces a truncated fusion protein with a premature stop codon at amino acid residue 295, expected to cause a non-functional protein lacking both its intracellular and membrane spanning domains and its extracellular cadherin repeats 3–5. Our mouse in situ expression data demonstrate that Cdh3 is expressed in the apical ectodermal ridge from E10.5 to E12.5, and later in the interdigital mesenchyme, a pattern compatible with the EEM phenotype. Furthermore, we discuss possible explanations for the phenotypic differences between EEM and congenital hypotrichosis with juvenile macular dystrophy (HJMD), which is also caused by CDH3 mutations.

Conclusions: In summary, we have ascertained a third gene associated with ectrodactyly and have demonstrated a hitherto unrecognised role of CDH3 in shaping the human hand.

Statistics from

Referring to a previously published Danish family,1 Ohdo et al first coined the term EEM syndrome to describe the rare association of ectodermal dysplasia, ectrodactyly, and macular dystrophy2 (OMIM 225280). Synonyms are Albrectsen-Svendsen syndrome or Ohdo-Hirayama-Terawaki syndrome. To date five families have been reported in the literature.1–5 The ectodermal defect is characterised by hypotrichosis with sparse and short hair on the scalp, sparse and short eyebrows and eyelashes, and partial anodontia. Different degrees of absence deformities as well as syndactyly have been described, the hands often being more severely affected than the feet. The phenotypic spectrum of limb defects reported in EEM families is shown in table 1. The retinal lesion appears as a central geographic atrophy of the retinal pigment epithelium and choriocapillary layer of the macular area with coarse hyperpigmentations and sparing of the larger choroidal vessels. Autosomal recessive inheritance is supported by the presence of multiple affected individuals in a family with unaffected parents and parental consanguinity as well as an equal distribution between both sexes. However, the pathogenesis remains unknown.

Table 1

 Phenotypic spectrum of limb defects in EEM

We carried out a molecular study of two previously published families from Denmark1 and Brazil.4 As CDH3 was recently identified as the cause of congenital hypotrichosis with juvenile macular dystrophy (HJMD), which shares an overlapping hair and eye phenotype with EEM,6 we chose CDH3 as a candidate gene. Here we present a molecular analysis of the CDH3 gene and mouse expression data supporting the suggestion that CDH3 is the cause of EEM and is thus a yet unrecognised ectrodactyly causing gene.



Written and informed consent was obtained from four members of family 1 and five members of family 2. The project and the consent forms were approved by the Danish ethical committee. The remaining affected and unaffected family members could not be traced or they refused to participate in the study. In family 1, two persons displayed limb defects.1 Their parents were first cousins. One individual had bilateral ectrodactyly with absence of middle phalanx 3 and distal phalanx 2, 3, 4 and interstitial polydactyly with two phalangeal bones articulating with metacarpal 4 on one side (fig 1). On the other side, proximal phalanx 2, 3 and middle and distal phalanx 2, 3, 4 were absent, and syndactyly occurred between fingers 4 and 5. On the feet, syndactyly between the third and fourth toe was seen bilaterally, and between toes 1 and 2 (partial) unilaterally. The other person had unilateral syndactyly of fingers 3 and 4 and normal feet. In family 2, the three persons with limb defects displayed different degrees of syndactyly between fingers 2/3, 2/3/4, or 1/2/3/4.4

Figure 1

 EEM phenotype in individual 1.II1. (A) Note sparse hair. (B) Note bilateral syndactyly of toes 3 and 4, and unilateral syndactyly of toes 1 and 2. (C) Note absence defects of fingers 2, 3, and 4, and partial syndactyly. For variability of limb malformations observed in EEM see table 1. (Photographs reproduced with permission)

Molecular testing

DNA was extracted from peripheral blood samples using standard methods. Testing for homozygosity at the CDH3 locus was carried out using microsatellite markers (D16S3107, D16S3025, D16S496, and D16S3141) spanning approximately 1.5 Mb on chromosome 16q22.1. Primers for STS markers were radioactively labelled using γ-[33P]-dATP (Hartmann Analytic, Braunschweig, Germany) and T4-DNA-polynucleotide kinase (Fermentas, Helsingborg, Sweden) according to the manufacturer’s protocols. Polymerase chain reaction (PCR) was carried out using 50 ng DNA template and Taq-DNA polymerase (New England Biolabs, Beverly, MA, USA) under standard conditions according to the manufacturer’s instructions, followed by separation by 5% acrylamide, 7 M urea, 1×TBE gel electrophoresis and overnight exposure to x ray films. All 16 exons and exon-intron border regions in the CDH3 gene were sequenced on both strands; primers were designed in the intron regions (table 2) and amplified by PCR using Taq-DNA polymerase (New England Biolabs) according to the manufacturer’s protocol. PCR conditions included an initial denaturation for 5 min at 94°C, followed by 40 cycles of 30 s at 94°C, 30 s at the individual annealing temperature determined for each primer set (table 2), 45 s at 72°C, and a final extension for 7 min at 72°C. PCR products were separated by 2% agarose, 1×TBE gel electrophoresis. Primers were removed by treatment with 1 U shrimp alkaline phosphatase (USB, Cleveland, OH, USA) and 10 U exonuclease I (New England Biolabs) followed by sequencing using the BigDye Terminator Kit (Applied Biosystems, Foster City, CA, USA) and analysed on an ABI 377 sequencer (Applied Biosystems).

Table 2

 Primers used for sequencing of CDH3

Mutations were confirmed with restriction enzyme digests of DNA from the families and from a minimum of 60 normal individuals (120 alleles) serving as a control group. The mutation c.965A→T disrupted a BsrDI restriction site. Digestion of the PCR product of exon 7 generated by the sequencing primers (table 2) resulted in a 32 bp and a 97 bp fragment from a normal allele and a 129 bp fragment from the mutated allele, which were separated on a 20% acrylamide 1×TBE gel. The mutation c.829delG disrupted a BanI restriction enzyme site. Digestion products of the PCR product of exon 7 generated by the sequencing primers (table 2) were separated on a 2% agarose gel resulting in a 159 bp and a 194 bp fragment from a normal allele and a 353 bp fragment from a mutated allele.

In situ hybridisation

Whole mount and section in situ hybridisation were performed as described previously using murine wildtype embryos of stages E10.5–E13.5.7 For hybridisation a 678 bp antisense probe specific for murine Cdh3 (Ensembl Transcript ID ENSMUST00000034383) was generated with PCR using murine embryonal cDNA and the following primers: forward 5′-TGC TGA CTA GGG GGA CAG TT-3, reverse 5′-CCC TCT CCA TCC ATG TCT GT-3′. The PCR product was cloned into a TOPO TA Cloning Kit (Invitrogen, Carlsbad, CA, USA), sequenced, and transcribed using a DIG RNA labelling kit (Roche, Mannheim, Germany).


Mutation analysis

The initial test for homozygosity at the CDH3 locus showed that all EEM patients (1.II1, 1.II2, 2.II1) displayed a homozygous haplotype, whereas all other tested family members were heterozygous (data not shown). Direct sequencing revealed homozygous mutations in both families. Patients from family 1 (fig 2A) had a missense mutation in exon 8 (c.965A→T) (fig 2C) causing a change from asparagine to isoleucine at amino acid 322 (N322I) (fig 2E). The unaffected parents were both heterozygous for the mutation. This was confirmed by a BsrDI restriction site assay (fig 2A), which also failed to identify any mutations in 120 control alleles.

Figure 2

 Pedigree, digest results (A,B), nucleotide sequences (C,D), and protein sequences (E,F) for family 1 (A,C,E) and family 2 (B,D,F). Conservation of amino acid residue 322 between species (G) and between human cadherins (H). In family 1 (A), digestion with the restriction enzyme BsrDI resulted in a 32 bp and a 97 bp fragment from a normal allele and a 129 bp fragment from the mutated allele. In family 2, digestion with BanI resulted in a 159 bp and a 194 bp fragment from a normal allele and a 353 bp fragment from a mutated allele.

The Brazilian proband (family 2) (fig 2B) was homozygous for a deletion mutation in exon 7 (c.829delG) (fig 2D). The mutation is predicted to introduce a frameshift and a premature stop codon at amino acid residue 295 (fig 2F). Provided this mRNA is translated, it is expected to cause a non-functional protein lacking both its intracellular and membrane spanning domains and its extracellular domains 3–5 (fig 3A, B). The parents, a brother, and a son of the affected were all heterozygous for the mutation. The findings were confirmed by a BanI restriction site assay (fig 2B).

Figure 3

 (A) Schematic protein structure of CDH3. (B) Crystal structure of the five cadherin repeat (EC1–5) modules in Xenopus C-cadherin ectodomain. Correlating sites for EEM mutations and HJMD (OMIM 601553) missense mutations are shown as predicted from alignment of CDH3 and C-cadherin. Arrows: mutations in EEM families; arrowheads: missense mutations in HJMD families. (C) Organisation of CDH8, CDH11, CDH5, CDH16, CDH3, and CDH1 in a cluster on chromosome 16q.

We identified different alleles in patients from family 1 and 2 by confirming two known SNPs in introns 8 and 10, and four SNPs in exons 12, 13, and 15, which were present in one family but not the other (table 3). Three of the latter had previously been reported (c.1623T→C, c.1689G→C, c.1965G→A), while c.2241C→A represents a novel SNP.

Table 3

 SNPs with different alleles in Danish and Brazilian EEM patients

Structure of CDH3 and localisation of mutations

CDH3 is a classical cadherin with five extracellular cadherin repeats (EC1–5), a membrane spanning section, and a cytoplasmic domain (fig 3A,B). It is located on the long arm of chromosome 16 as part of a cluster containing five other cadherins (CDH1, CDH5, CDH8, CDH11, and CDH16) spanning ∼7.2 Mb (fig 3C). By aligning the protein sequences of CDH3 and Xenopus C-cadherin and identifying the correlating residues in the recently published C-cadherin structure,8 we found that amino acid residue 322 affected in family 1 is part of a Ca2+ binding domain connecting EC2 and EC3 (fig 3B). The deletion in family 2 is located in EC2.

Mouse in situ expression data

At stages E10.5 and E11.5 Cdh3 is expressed during mouse embryonal development in craniofacial structures including the orofacial region, and the pharyngeal arches (fig 4A, B). Specific analysis of expression in the retina was not carried out. Cdh3 expression was also observed in the gut, the presomitic mesoderm, and to a lesser extent in the most caudal somites. At these stages Cdh3 is also expressed in the fore and hind limbs in the limb ectoderm including the apical ectodermal ridge (AER), and to some extent in the mesenchyme (fig 4C). When chondrogenic condensations of phalanges start to form at E12.5, Cdh3 expression is detected in the interdigital web, where it continues to be expressed at E13.5 (fig 4D, E).

Figure 4

 In situ expression analysis of Cdh3 during mouse embryonal development. (A) Expression at E10.5, (B) at E11.5, and (C) at E12.5 (expression in blue). (D) Expression at E13.5 (expression in white); (E) Toluidine blue counter staining. Expression in the AER is seen up to E12.5 and interdigital expression at E12.5 and E13.5 when digital condensations have occurred. Expression in limb buds, AER, and interdigital mesenchyme is marked by arrows.


Ectrodactyly occurs as an isolated trait with at least five known loci and as part of many syndromes. Mutations in TP63 have been shown in isolated split hand foot malformation (SHFM4; OMIM 605289) as well as in ectrodactyly, ectodermal dysplasia, and cleft-lip-palate (EEC3; OMIM 604292). In patients with SHFM3 (OMIM 600095), a tandem ∼400–550 kb duplication of 10q24 involves several genes including Dactylin (FBXW4), a gene known to cause ectrodactyly in mice, LBX1, and FBW1A.9,10 In the present study we have identified a third gene associated with ectrodactyly and show that mutations in a cadherin can cause limb malformations in humans.

Classical cadherins are Ca2+ dependent adhesion molecules acting as dimers in homophilic binding.11,12 Their specific spatial and temporal expression patterns during embryogenesis suggest an important role in normal development.11,13,14 In this study we have identified two distinct homozygous mutations in two families predicted to lead to loss of CDH3 function or defective Ca2+ binding. In family 2, a nonsense mutation (c.829delG) introduced a preterminal stop codon, predicting a protein lacking both its intracellular and membrane spanning sections and its extracellular domains 3–5. Thus, this mutation very likely causes loss of function of CDH3. The missense mutation identified in both affected sibs in family 1 (c.965A→T) results in substitution of the polar asparagine at position 322 with a hydrophobic isoleucine supporting a change of function. Amino acid 322 is highly conserved in different species (fig 2G) and in different human cadherins (fig 2H). Functionally it is part of a motif known to bind Ca2+ located between EC2 and EC3.15 In other cadherins the binding of Ca2+ is assumed to stabilise the extracellular structure necessary for specific cell-cell binding function and thereby affect the angle between different EC modules and the configuration of the extracellular section.8 The cytoplasmic tail of cadherins binds β catenin, which was recently identified as linking transcription and cell adhesion by changing its distribution in the cell upon extracellular signalling from Wnts and BMP inhibitors.16 If mutations in CDH3 cause abnormal distribution of β catenin in the cell, this may also be part of the pathomechanism in EEM.

Local disruption of the AER has been suggested as the pathomechanism of ectrodactyly in patients with TP63 mutations.17 We show that Cdh3 is expressed in the AER from E10.5 to E12.5. Animal studies copying the EEM mutation and the phenotype will be necessary to see if loss of normal CDH3 function results in an abnormal AER or if EEM is caused by a different and as yet undefined pathogenetic mechanism. From E10.5 to E12.5 Cdh3 expression is observed to some extent in the mesenchyme, but from E13.5 it is expressed solely in the interdigital region where coordinated apoptosis will later separate the digits. The occurrence of syndactyly in hands and feet in EEM patients can be interpreted as a consequence of abnormal interdigital apoptosis. The observed expression pattern suggests that CDH3 plays an important and as yet unrecognised role in shaping the human hand. It is interesting that the deletion mutation in members of family 2 resulted in syndactyly but not ectrodactyly, whereas the missense mutation in family 1 resulted in both phenotypes. A study of more EEM families is necessary to determine if the variability of limb defects can be explained by a genotype-phenotype correlation. Remarkably, one individual out of four heterozygous for 829delG (2.I1, 2.I2, 2.II2, and 2.III1) presented with minimal symptoms, namely a mild bilateral syndactyly between fingers 1/2/3/4 and small and widely spaced teeth, while ophthalmological and hair examination were unremarkable.4 This suggests that subtle limb and dental defects may represent the mildest phenotype in EEM families.

Mutations in CDH3 were recently reported in HJMD sharing the same hair and eye phenotype as EEM6,18 (table 4). The mutations in HJMD are homozygous single base pair deletions introducing premature stop codons or missense mutations; a single family was compound heterozygous. The mutations are assumed to cause a loss of CDH3 function. Interestingly, the R503H mutation in HJMD also affects a highly conserved amino acid directly involved in Ca2+ binding, but between EC4 and EC5. Functional differences between the cell-cell binding properties of this mutation and N322I between EC2 and EC3 seen in EEM remain to be determined.

Table 4

CDH3 mutations and phenotypic variability

The hair anomalies and retinal pigmentary alterations appear to be identical in HJMD and EEM, but EEM patients are additionally characterised by ectrodactyly or syndactyly (table 1) and oligodontia, enamel hypoplasia, and widely spaced teeth.2,4 The identified deletion (829delG) was previously found in a Turkish HJMD family with associated keratosis pilaris19 indicating that a simple genotype-phenotype correlation is not sufficient. EEM and HJMD may instead belong to a single broad phenotypic spectrum where pleiotropic differences are determined by stochastic factors. However, the fact that no EEM family members display the limited HJMD phenotype may suggest that genetic factors and not stochastics are responsible.

Though speculative, it is possible that CDH3 loss of function causes HJMD, which is then modified by another gene to cause EEM. Since no EEM family member displays HJMD, the designated modifier has to segregate together with CDH3. A candidate gene for this is CDH1 (E-cadherin) located only 38 kb telomeric from CDH3 and known to have redundant functions. The peculiar hair phenotype occurs in hair matrix cells where only CDH3 and not CDH1 is expressed and no defects occur in tissues where the two cadherins are expressed together.6

In this context it is notable that animal models indicate the importance of Cdh3 and redundance of other genes during epithelial morphogenesis and integrity. For example, Cdh3 knock out mice display perturbation of mammary gland development, but no defects in hair, limb, or eye morphogenesis.20 In Caenorhabditis elegans loss of Cdh3 function results in defects at the tip of the nematode tail.21 Such phenotypic differences compared to humans carrying mutations in orthologous genes are not uncommon, and may be caused by expression differences and redundant functions of other cadherins.

In conclusion, CDH3 represents a third gene associated with ectrodactyly and plays an important and hitherto unrecognised role in shaping the human hand. Future studies to better define the role of CDH3 in epithelial mesenchymal interactions may help to better characterise its function during development of the ectoderm and limb morphogenesis.


We thank the families for participating in this study, and Dr Jordão Correa for establishing contact with the Brazilian family. We are indebted to Flemming Skovby, Department of Clinical Genetics, Copenhagen University Hospital for permission to publish the clinical pictures.

View Abstract


  • The Wilhelm Johannsen Centre for Functional Genome Research is funded by the National Danish Research Foundation.

  • Competing interests: none declared

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.