Article Text

Download PDFPDF

Original article
The role of Frizzled-4 mutations in familial exudative vitreoretinopathy and Coats disease
  1. Johane M Robitaille1,
  2. Binyou Zheng1,
  3. Karin Wallace2,
  4. M Jill Beis2,
  5. Cuneyt Tatlidil3,
  6. Jenny Yang1,
  7. Tom G Sheidow4,
  8. Lee Siebert4,
  9. Alex V Levin5,
  10. Wai-Ching Lam6,
  11. Brian W Arthur7,
  12. Christopher J Lyons8,
  13. Elisa Jaakkola9,
  14. Ekaterini Tsilou10,
  15. Charles A Williams11,
  16. Richard Grey Weaver Jr12,
  17. Carol L Shields5,
  18. Duane L Guernsey1
  1. 1Dalhousie University, Halifax, Nova Scotia, Canada
  2. 2IWK Health Centre, Halifax, Nova Scotia, Canada
  3. 3Valley Health Regional Hospital, Kentville, Nova Scotia, Canada
  4. 4University of Western Ontario, London, Ontario, Canada
  5. 5Wills Eye Institute, Thomas Jefferson University, Philadelphia, Pennsylvania, USA
  6. 6University of Toronto, Toronto, Ontario, Canada
  7. 7Queen's University, Kingston, Ontario, Canada
  8. 8University of British Columbia, Vancouver, British Columbia, Canada
  9. 9Oulu University, Oulu, Finland
  10. 10National Eye Institute, Bethesda, Maryland, USA
  11. 11University of Florida, Gainesville, Florida, USA
  12. 12Wake Forest University, Winston-Salem, North Carolina, USA
  1. Correspondence to Dr Johane M Robitaille, IWK Health Centre, Eye Care Team, 5850/5980 University Avenue, PO Box 9700, Halifax, Nova Scotia B3K 6R8, Canada; jrobitai{at}

Aim The aim of this study is to assess the role of Frizzled-4 (FZD4) in familial exudative vitreoretinopathy (FEVR) and Coats disease.

Methods Tissue samples were collected for DNA extraction and automated DNA sequencing of the two coding exons of FZD4 in both directions. Cases carrying a FZD4 mutation and demonstrating extreme disease severity were selected for direct automated sequencing of all coding exons of LRP5, NDP and TSPAN12. Clinical data were obtained for the purpose of identifying genotype–phenotype correlations.

Results 68 probands were diagnosed as having autosomal dominant or sporadic FEVR. Eleven FZD4 mutations (five missense, three deletions, one insertion, two nonsense) were identified. Six of these mutations are novel, and none were found in 346 control chromosomes. In 16 cases of Coats disease, one polymorphism combination was found in two samples: no mutations were detected. No genotype–phenotype correlation emerged. Three severely affected cases with FZD4 mutations failed to show additional mutations in the three other FEVR genes.

Conclusion The authors identified 12 FEVR probands with FZD4 mutations. FZD4 mutation screening can be a useful tool especially in mild or atypical cases of FEVR. Germ-line mutations in FZD4 do not appear to be a common cause of Coats disease.

  • FZD4
  • familial exudative vitreoretinopathy
  • genetics
  • coats disease
  • mutation, retina, embryology and development
  • angiogenesis
  • experimental - laboratory
View Full Text

Statistics from


Familial exudative vitreoretinopathy (FEVR) is a developmental anomaly of the retinal vasculature characterised by a failure of peripheral retinal vascularisation. It was first described in 1969 by Criswick and Schepens1 as retinal and vitreous abnormalities resembling retinopathy of prematurity (ROP) in the absence of a history of prematurity and neonatal use of oxygen therapy, Coats disease and peripheral uveitis.

To date, four genes causing FEVR have been described: NDP (X linked recessive FEVR and Norrie disease),2 Frizzled-4 (FZD4) (autosomal dominant FEVR),3 LRP5 (autosomal dominant and recessive FEVR)4 and TSPAN12 (autosomal dominant FEVR).5 6 These genes code for proteins involved in the canonical Wnt and Norrin signalling pathways. LRP5 and FZD4 act as coreceptors, and norrin, the gene product of NDP, as a ligand with strong affinity to FZD4. All three components are necessary to initiate signalling.7 TSPAN12 is a transmembrane protein that mediates the association of the Norrin-receptor complex.8

The purpose of this study is to screen the FZD4 gene for mutations in individuals with FEVR, Coats disease, ROP, non-syndromic retinal dysplasia and/or congenital retinal folds and persistent fetal vasculature, and examine the data for genotype–phenotype correlations. Our findings on the latter four conditions were published separately.9 10 We now present the clinical and genetic results from a group of 84 probands with FEVR or Coats disease.


The study was approved by the Research Ethics Board of the IWK Health Centre and consent obtained according to Canadian Tri-Council guidelines. Participants were initially enrolled locally. The laboratory was later listed on the GeneTests website,11 and more subjects were recruited via their treating physician who contacted the laboratory for testing. Study subjects had a clinical diagnosis of FEVR or Coats disease.

Clinical information was collected either by chart review or prospectively, and included eye examination (best-corrected visual acuity, alignment, slit-lamp examination, dilated fundus examination) and, whenever possible, intravenous fluorescein angiography (IVFA) and fundus photography. A scale from 0 to 6 was used to describe the fundus findings to reflect the potential need for treatment and the predicted effect on visual acuity: 0 for no or unrelated abnormalities, 1 for avascular area only, 2 for peripheral retinal thinning/lattice, 3 when peripheral fibrovascular mass present without traction on macula, 4 for temporal macular dragging, 5 for partial retina detachment and folds, and 6 for total retina detachment. This scale is not meant to be used as a classification system and was devised based on the distribution of clinical descriptions provided by referring clinicians.

Tissue samples (blood or buccal swabs) were obtained from each participant for DNA extraction using standard protocols (QIAGEN QIAamp DNA blood kits and ORAgene DNA self-collection kit). Direct automated sequencing of the two coding exons of the FZD4 gene in both forward and reverse directions was performed. In selected cases manifesting a severe phenotype (defined as blinding complications present within 6 months of birth), direct automated sequencing of the 23 coding exons of LRP5, all three exons of the NDP gene (including the 5 prime untranslated exon and the promoter region) and all seven coding exons of TSPAN12 was performed in both directions including all consensus splice sites (primers for all amplicons can be requested from the corresponding author). Detection of sequence variation was performed both manually and using Mutation Surveyor. All sequence variations were evaluated for functional significance. Using direct sequencing or restriction fragment length polymorphisms, we tested for these sequence changes in 173 random control samples (346 chromosomes) of Caucasian descent. Sequence alignment was performed using the ClustalX program, version 2.0.10 (Conway Institute of Biomolecular and Biomedical Research). Protein prediction programs were used to estimate the pathogenicity of missense changes (SIFT (, PolyPhen (, Align-GVGD (


Sixty-eight unrelated probands were diagnosed as having autosomal dominant FEVR based on pedigree analysis or sporadic FEVR. Eleven heterozygous FZD4 mutations (five missense, three deletions, one insertion, two nonsense) were identified in six singletons and six pedigrees (figure 1), resulting in an 18% mutation detection rate. Six of the 11 mutations are novel: p.C106G, p.M157K, p.Y211fsX, p.D428fsX1, p.T503fsX31 and p.G488V. None were found in the controls. With the exception of mutation p.G488V, from a person who was from India, the novel mutations were identified in Caucasian individuals. The chromatograms of the six novel mutations and alignment sequences of missense mutations are shown in figure 2.

Figure 1

Pedigrees with Frizzled-4 (FZD4) mutations. Square symbols represent males, and round symbols females. Black symbols indicate affected individuals, while symbols with an ‘N’ designate relatives that were examined and lacked features of familial exudative vitreoretinopathy. A dot within a symbol indicates that the individual carries a mutation without manifesting the disease. Pedigree I is the original pedigree that led to the identification of the FZD4 gene as causative of autosomal dominant familial exudative vitreoretinopathy.3 Pedigree II is described elsewhere.9 Of note, documented affected relatives of the probands with the p.G488V and individual IV-1 carrying the P.M157K were not available to participate in the study, and the pedigree information is not provided.

Figure 2

Sequence chromatograms of six novel Frizzled-4 (FZD4) mutations and FZD4 protein sequence alignment of missense mutations. (A) Heterozygous mutation c.316 T>C resulting in a Cys-to-Gly change at the position of 106 in exon 2. (B) Heterozygous mutation c.470T>A causing a Met change to Lys at the position 157 in exon 2. (C) Heterozygous mutation c.633delC resulting in a frameshift in the Tyr codon 211 and creating a STOP codon (p.Y211fsX). (D) Four-base-pair heterozygous deletion of bases 1282–1285 causing a frameshift mutation, p.D428fsX1, yielding a truncated protein. (E) Insertion of a single C base between the C at position 1508 and the G at 1509 position causing a frameshift mutation, p.T503fsX31. (F) Mutation c.1463G>T changing GGC to GTC and altering a glycine to valine at codon 488. (G) Protein sequence alignment of human FZD4 with orthologues (mouse, rat, chick, frog) and paralogues in the vicinity of novel missense mutations p.C106G, p.M157K and p.G488V (each designated by an asterix).

The first mutation, p.C106G (c.316 T>C, reference sequence cDNA NM_012193), was identified in a singleton (individual III-1, figure 3A). The parents were not available for genetic testing, and there were no other affected relatives by history. The mutation changes the polar amino acid cysteine to glycine, a non-polar amino acid. The cysteine 106 position is found in the conserved cysteine-rich domain (CRD) of FZD4 and is evolutionarily conserved among orthologues as well as within other human frizzled proteins (paralogues). The C106G mutation was predicted by SIFT to affect protein function and anticipated to be deleterious, while PolyPhen predicted the variant to be probably damaging, and Align-GVGD predicted the variant to most likely interfere with protein function.

Figure 3

Fundus photos of affected individuals with Frizzled-4 (FZD4) mutations and intravenous fluorescein angiography from a patient with Coats disease. (A) Right posterior pole showing dragging of the retina. A macular fold is present in the left eye. (B) Right fundus showing peripheral pigmentary changes and non-perfusion with leucocoria left eye secondary to total retinal detachment. (C) Peripheral white fibrovascular lesion with macular preretinal fibrosis and dragging present in the right eye. The left eye shows an extensive exudative retinal detachment. (D) Asymmetrical disease with left eye manifesting mild temporal straightening of vessels without any evidence of avascularity. The right eye (not shown) presented with a haemangiomatous lesion with exudation and peripheral avascular retina. (E) Tractional retinal detachment. The left fundus is normal (not shown). (F) Left macula dragged (photo on left) towards a yellowish-white subretinal lesion peripherally (photo on right). The right fundus is normal (not shown). (G) Bilateral dragging of the macula with peripherally straightened blood vessels and thickened fibrous areas and avascular retina. (H) Mild straightening of the vascular pattern, seen in the left eye of the affected mother. (I) Peripheral fibrotic mass temporally with overlying vitreous detachment present in the right eye. The left fundus is normal. (J) Bilateral dragging of the macula. (K) Left eye manifesting a total retina detachment and the right a treated partial retina detachment with peripheral fibrovascular scar tissue. (L) Very large elevated tight fold extending through the inferotemporal quadrant of the right eye, with the left eye showing minimal dragging and fibrosis with temporal avascularity. (M) Peripheral retina showing areas previously treated in the right eye. The left eye shows a large falciform fold extending inferotemporally with lenticular changes visible through the dilated pupil. (N) Inferotemporal dragging with retinal pigmentary changes extending into macula. The left eye is within normal limits. (O) intravenous fluorescein angiography from a male with unilateral Coats disease showing tortuous vessels with aneurysmal dilation and peripheral areas of capillary non-perfusion typical of Coats disease.

Mutation p.M157K (c.470T>A) was found in two individuals, one with a positive family history (IV-1, figure 3B), although affected relatives were not available to participate in the study, and the other in a singleton (V-1, figure 3C whose parents declined participation. Both individuals are from a separate area of the same Canadian province. There is no evidence of a familial relationship. The mutation occurs in the cysteine-rich domain (CRD) and changes a neutral and hydrophobic methionine to a basic lysine. The methionine residue at position 157 is evolutionarily conserved among orthologues but not among paralogues. SIFT predicted this variant to affect protein function and to be deleterious. PolyPhen predicted the variant to be probably damaging to protein function, and AGVGD predicted p.M157K to most likely interfere with protein function.

A heterozygous deletion of base pair C at position 633 in codon 211 (c.633delC) located in the extracellular topological domain of FZD4 was identified in a female singleton (VI-1, figure 3D). Her parents were not available for eye examination or blood samples. This deletion causes a frameshift in the Tyr codon (TAC) 211 that creates a STOP codon (TAA) at the same codon position (p.Y211fsX). This results in a premature termination of the translation of the FZD4 protein and thus a truncated protein.

Mutation p.D428fsX1 was found in a female singleton (VII-1, figure 3F). The family relocated and was not available for further testing. The 4-base-pair heterozygous deletion of bases 1282–1285 causes a frameshift in the mutated allele and leads to a STOP codon at position 429, lysine (AAG) > STOP (TAG). This results in a premature termination of translation of the FZD4 protein and thus a truncated protein.

The FZD4 mutation, p.T503fsX31, was identified in a female singleton (VIII-1, figure 3G). The parents were not available to participate in the study. The insertion of a single C base between the C at position 1508 and the G at 1509 position causes a frameshift in one allele that creates a STOP codon 31 amino acids downstream. Translation stops four amino acids prematurely, which is expected to affect the function of the intracellular signalling domain.

A proband (IX-1) from India with a known family history of FEVR was found to carry p.G488V. She had a history of retina detachment in the left eye with multiple breaks and ‘proliferative vitreoretinopathy,’ and had multiple cryotherapy scars peripherally in the right eye. Her family was not available to participate in the study, but she had an affected brother and father. The mutation c.1463G>T changes a glycine to valine at codon 488. Both amino acids are non-polar and hydrophobic, and no change in properties results from this base pair change. This mutation is located in the seventh transmembrane domain and is highly conserved among orthologues and paralogues. We did not have a random population sample of Indian origin to test the hypothesis that this might represent a polymorphism in the Indian population. SIFT predicts the change to be deleterious to protein function. PolyPhen predicts this variant to be possibly damaging to protein function. AGVGD predicts this variant would most likely interfere with protein function.

In addition to the above FZD4 mutations, we found three mutations that were previously published by others: p.M105V,12 p.W496X13 and p.W226X.14 p.M105V was found in eight individuals of one pedigree of Finnish descent (Pedigree X, figure 3H–K), p.W496X in a Caucasian singleton (XI-1, figure 3E), and p.W226X in a multiplex Japanese pedigree (Pedigree XII, figure 3L–N). We previously published mutations p.M493_W494del3 and p.I114T9 in Pedigrees I and II, respectively.

The clinical data for the 56 affected relatives are summarised in the supplementary table and figure 3. In Pedigree I, the median age at last examination of the 33 affected individuals was 38 years (range 2–78 years). The median refractive error in the right eye was −0.75 spherical equivalent (range +2.67 to −12.50) and −1.12 in the left (range +1.67 to −11.87). There were 13 cases of strabismus (four esotropia, nine exotropia) and two with pseudoexotropia. Cataract was present in 11 right eyes and 10 left eyes. The distribution of fundus anomalies and visual acuities is depicted in figure 4. There was a wide range of intrafamilial variability with visual acuity ranging from 6/6 in both eyes to bilateral no light perception, and retinal abnormalities ranging from none (confirmed with IVFA) to severe (total retinal detachment).

Figure 4

Summary of familial exudative vitreoretinopathy phenotype in Pedigree I. (A) Distribution of best-corrected visual acuity (VA) in the right eye (OD) of 32 affected relatives. (B) Distribution of retina features using 0–6 scale (see Methods) in the right eye (OD) of affected members. NLP, no light perception.

Comparison of the phenotype in each pedigree falls within the spectrum of FEVR reported in the literature and that found in Pedigree I. No clear genotype–phenotype association emerged. The phenotypic range in the presence of a positive family history overlapped with that found in the singletons. For a selected number of cases with particularly severe disease (Pedigree I, individual I-33, Pedigree II, individuals II-1 and II-2), we previously reported9 mutation screening of LRP5 and NDP as negative. Screening of TSPAN12 in the same individuals in this study failed to reveal additional mutations.

Among the 56 affected individuals with FZD4 mutations, unilateral disease confirmed with IVFA was present in two individuals, and a further eight were found with unilateral features based on fundus examination alone. Both eyes were normal in one mutation carrier who did have an IVFA and two mutation carriers who were not screened with IVFA (supplementary table).

In 16 cases of Coats disease of various ethnic backgrounds (10 Caucasian, two African–American, one mixed Caucasian/Hispanic, two Asian, one unknown), the single FZD4 polymorphism combination p.P33S/p.P168S (rs61735304/rs61735303) was found in two Caucasian samples. No pathological mutations were detected. Figure 3O shows an IVFA from one of the cases highlighting some of the overlapping features with FEVR.


We identified 68 FEVR probands from which we detected 12 with mutations in FZD4, of which six have not been previously described. Among the six pedigrees and six singletons with mutations, we did not detect a genotype–phenotype correlation.

Two reports reviewed the manifestations of FEVR in pedigrees with known FZD4 mutations consisting of 18,15 12 and nine16 affected relatives each. Other pedigrees reported with FZD4 mutations included six affected individuals in one pedigree,13 five affected in two pedigrees,13 17 four affected relatives in three pedigrees15 16 18 and three affected in three pedigrees.16 18 19 Virtually all pedigrees reported high intrafamilial variability with at least one severely affected member and many previously unsuspected, asymptomatic relatives. Our large pedigree with 33 affected members confirms these findings in addition to providing risks for vision loss. However, such risks must be interpreted with caution during prenatal counselling, as illustrated by the two severely affected siblings (II-1 and II-2, supplementary table) of an asymptomatic mother (II-3, supplementary table) in Pedigree II.9

Our mutation detection rate of 18% is in keeping with other reports that range from 14% to 40%.13 16 20 21 The mutation type (missense, deletion, insertion, stop) did not affect the FEVR phenotype. In three children with bilateral severe FEVR manifestations presenting soon after birth, no additional FEVR gene mutations were identified.9 Such intrafamilial variability and lack of genotype–phenotype correlation suggests that other genetic, environmental or stochastic factors play a role in determining the disease severity in individuals carrying a FEVR gene mutation. The only ocular phenotype possibly related to the FZD4 genotype is reported by Ells et al,10 who described two infants from a group of 71 with severe ROP that carried FZD4 mutations outside the functional FZD4 domains that appeared non-pathogenic for FEVR. These milder mutations were hypothesised to impart a risk for the development of severe ROP in the context of prematurity.

Three of the six novel mutations we detected are missense mutations. Further evidence supporting the pathogenicity of the p.M157K mutation is provided by Toomes et al,16 who identified a p.M157V (methionine to valine) that segregated with FEVR in a large American family known to be linked to EVR1. Although we did not have a random population of Indian origin to screen for p.G488V, a mutation in the same codon, p.G488D, was reported in a Japanese pedigree,12 supporting the hypothesis that the FZD4 p.G488V mutation is causative of FEVR in our pedigree of Indian origin. The missense mutation affecting codon 105 in our Finnish pedigree was previously reported in a pedigree of Japanese descent,12 and a different mutation altering the same codon (p.M105T) was found in a British patient with a significant family history compatible with FEVR.16 Both our pedigree and the previously reported pedigrees had features on examination and/or by history consistent with high intrafamilial variability. Finally, Shastry14 mentions mutations p.D428fsX and W226X, but no clinical data are provided. The insertion, deletion and nonsense mutations were predicted to truncate the FZD4 protein and are therefore expected to affect intracellular signalling.

No FZD4 mutation was detected in 16 cases diagnosed as having Coats disease. One study by Black et al22 provides evidence that Coats disease, a condition that shares phenotypic overlap with FEVR, can be caused by a somatic mutation in the NDP gene. Because of the known intrafamilial variability of FEVR and the possibility of a high degree of disease asymmetry, we were interested in determining whether some cases of Coats were in fact highly asymmetrical FEVR (ie, FEVR presenting as Coats disease). In this series, although unilateral FEVR confirmed with IVFA is rare, unilateral presentation on fundus examination alone is common and underscores the need for a thorough evaluation of both eyes in patients with suspected Coats disease. The polymorphisms p.P33S and p.P168S found in two affected Caucasians were previously identified either in combination or as isolated p.P168S in 7% of a random Caucasian population in one study10 and the p.P168S as an isolated variant in 0.3% of Caucasians in another.16 Interestingly, cases of FEVR have been reported with the variation in isolation or in combination,13 17 21 23 but in one large pedigree, the variant combination failed to segregate with the disease.13 Examination of affected globes to screen for somatic FEVR gene mutations may reveal more clues in the aetiology of Coats disease. Our study suggests that germ-line FZD4 mutations are not a common cause of Coats disease.

In individuals presenting a phenotype compatible with FEVR, especially in the absence of a family history and in atypical or subtle cases, mutation screening of the known FEVR genes can be an invaluable tool to provide a diagnosis, aid genetic counselling and determine recurrence risks. As more FEVR genes are identified and their role in developmental retinal vascular disease established, we can optimise further the management of these conditions


We wish to acknowledge the support and effort of the participating families.


View Abstract

Supplementary materials


  • ET is an NIH employee.

  • Funding March of Dimes Birth Defects Foundation: United States; IWK Health Centre: Halifax, Nova Scotia, Canada.

  • Competing interests None.

  • Ethics approval Ethics approval was provided by the IWK Health Centre.

  • Provenance and peer review Not commissioned; externally peer reviewed.

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.