Clinical Presentation and Genetic Correlation of Patients With Mutations Affecting the FZD4 Gene

Objective  To correlate the ophthalmic findings of patients with pediatric vitreoretinopathies with mutations occurring in the FZD4gene.

Methods  A total of 123 patients diagnosed with autosomal-dominant familial exudative vitreoretinopathy (AdFEVR) or retinopathy of prematurity (ROP) and 42 control patients were enrolled in the study. Diagnoses were based on retinal findings at each patient's first examination or during ROP screening. Genomic DNA was isolated and polymerase chain reaction and direct sequencing of the FZD4gene performed.

Results  FZD4gene mutations were discovered in 13 of the 123 (10.6%) patients. Nine of the 63 patients with AdFEVR (14.3%) has mutations in the FZD4gene. Four heterozygous mutations were identified: C117R, C181Y, Q505X, and P33S/P168S. Four of the 60 patients with ROP (6.7%) have a double missense mutation P33S/P168S that was also found in the patients with FEVR. No other FZD4mutations were found in the patients with ROP. Additionally, patients expressing the double mutation had clinical presentations that overlapped, making it difficult to assign a definitive diagnosis. None of the mutations found in the patients with FEVR or ROP were seen in the control chromosomes.

Conclusion  Mutations occurring in the FZD4gene affect patients diagnosed with both FEVR and ROP. The clinical picture often overlaps and may require a detailed birth and family history for diagnosis. Genetic testing confirms inherited vitreoretinopathy and helps direct clinical management.

Clinical Relevance  Patients diagnosed with ROP may have a mutation in the FZD4gene and display characteristics consistent with FEVR. Analysis of the FZD4gene should be considered.

Mutations in the gene encoding the frizzled-4 receptor (FZD4) (gene, FZD4; OMIM 133780) have been described in many patients diagnosed with autosomal-dominant familial exudative retinopathy (AdFEVR)¹and in a smaller number of patients with retinopathy of prematurity (ROP).²In both diseases, the patient is born with enlarged and tortuous retinal vessels and an area of avascular peripheral retina. Additionally, varying degrees of subretinal exudation, vitreoretinal traction, and abnormal extraretinal vessels/neovascularization may occur.

Proper FZD4 signaling is necessary for normal retinal vascular development.³Frizzled-4 is a 537–amino acid, 7-transmembrane receptor that transduces Wnt signaling (Figure 1and Figure 2). Several Wnt ligands, Wnt-3a, Wnt-8, Wnt-5a, and the non-Wnt ligand, Norrin, are known to activate FZD4. There are 6 intracellular paths of Wnt signaling that can occur downstream of frizzled receptor activation: the canonical Wnt/β-catenin pathway, the planar cell polarity pathway, and the Wnt-Ca²⁺pathway. In vitro studies have shown that transduction of the FZD4 signal can occur by any of these pathways. Although it was originally thought that the path taken was determined by the ligand, it is still unclear how the frizzled receptor signals are transduced.

The canonical Wnt/β-catenin pathway is the most thoroughly studied. Activation is initiated by ligand binding to both the FZD4 receptor and its coreceptor, the low-density lipoprotein receptor–related protein 5 (LRP5), ultimately resulting in dephosphorylation of β-catenin and translocation to the nucleus.⁴Nuclear β-catenin participates as a transcriptional activator of the transcription factor and lymphoid enhancer–binding factor family of DNA-binding proteins (Figure 2). Target genes include C-MYC, CYCLIN D1, and VEGF, presumably regulating cell proliferation in specific tissues.

Similar to the canonical pathway, the planar cell polarity pathway also involves activation of disheveled protein subsequent to FZD4 activation (Figure 2). Unlike the canonical pathway, LRP5 is not needed for this signal transduction. Subsequently, 1 of 2 small guanosine triphosphatases (Rho or Rac) is activated, which in turn activates an alternative signal transduction pathway.⁵The planar cell polarity pathway mediates cytoskeletal organization and cell migration.

In the Wnt-Ca²⁺pathway, FZD receptor activation stimulates an intracellular Ca²⁺release, activating calcium-calmodulin kinase 2 and protein kinase C (Figure 2). This pathway is important in cell adhesion and cell movement during gastrulation.

In this study, we used direct sequencing to screen for mutations in the coding sequence of FZD4in patients with FEVR and ROP. The possible implications of these sequence variations in relation to signal transduction and disease type will be discussed.

Patients were recruited to the study through a protocol approved by the internal review board at William Beaumont Hospital and consented to participation. They were diagnosed with FEVR or ROP based on fundus examination, family history, and gestational age. Participants provided a blood sample from which genomic DNA was isolated from the leucocytes using the Purgene Genomic DNA Purification Kit (Qiagen, Valencia, California). When possible, genomic DNA from the relatives of patients expressing an FZD4mutation were tested for sequence variations.

The coding sequence and flanking splice sites of FZD4were amplified from 100 ng of genomic DNA using Herculase Hotstart PCR Master Mix (Stratagene, La Jolla, California). Four sets of primers were used (Table 1). Amplification conditions were as follows: 1 cycle at 98°C for 1 minute, 40 cycles of 30 seconds at 98°C, 30 seconds at 55°C, and 1 minute at 72°C, and a final extension for 10 minutes at 72°C. Amplified DNA was cleaned using the QIAquick Multiwell PCR Purification Kit (Qiagen). Sequencing reactions were performed using the Beckman Dye Terminator Cycle Sequenc Quick Start Kit (Beckman Coulter, Inc, Fullerton, California) and a Beckman CEQ 8000 autosequencer.

FZD4gene mutations were discovered in 13 of the 123 patients enrolled in this study. Nine of the 63 patients with AdFEVR (14.3%) were found to have mutations in the coding sequence of FZD4(Table 2). Four heterozygous mutations were found: C117R, C181Y, Q505X, and P33S/P168S (eFigure). In addition, 4 of the 60 patients with ROP (6.7%) had the double missense mutation P33S/P168S that was found in the patients with FEVR (Table 2). No other FZD4mutations were found in the patients with ROP. None of the mutations found in the patients with FEVR and ROP were seen in the 84 control chromosomes.

Two novel missense mutations were found in patients with AdFEVR, C117R (2 patients) and C181Y. Both occur in 1 of the 13 conserved cysteine residues of FZD4 ligand–binding domain (Figure 1).

One patient with FEVR had a mutation resulting in early termination of protein translation, Q505X. This mutation has been reported in an Australian family.⁶It is located immediately downstream from the highly conserved canonical Wnt activation motif (disheveled binding; KTXXXW) and causes premature termination prior to the planar cell polarity (c-Jun N-terminal kinase) pathway, PDZ1 binding motif (KTXV) (Figure 1).

A double missense mutation, P33S/P168S, was found in patients with both FEVR (5 patients; 7.9%) and ROP (4 patients; 6.7%). The double mutation was previously reported in a patient with AdFEVR.²Both P33S and P168S have been independently reported in patients with AdFEVR.^6,7AA33 is within the signal sequence and AA168 lies downstream of the cysteine-rich domain (CRD), Wnt-binding domain, and prior to the start of the 7-transmembrane domains (Figure 1). The P168S mutation has been reported in 1 control (of 400) but the P33S and double mutation have not been reported in control patients.

Several relatives of the patients with the double missense mutation P33S/P168S were screened and found to express the mutation as well. Asymptomatic family members demonstrated retinal peripheral avascularity but no abnormal extraretinal vessels or exudates (Figure 3). Interestingly, 1 patient with ROP had a fraternal twin with the mutation but no apparent manifestation of disease, although fluorescein angiography was not performed and retinal avascular zones could not be definitively assessed. The asymptomatic twin is fraternal, had minimal comorbidities, and had much better systemic health, suggesting a role for epigenetics in the severity of disease manifestation.

The clinical course varied for the individual patients in which a mutation was identified. Patients presented between birth (up to 16 weeks premature) and 7 years of age, although all patients reported poor vision in at least 1 eye since birth. Four of the 13 patients had ROP, and 2 developed aggressive posterior ROP (APROP). Nine patients had phenotypes consistent with FEVR, although 2 of these patients were initially diagnosed with ROP. One patient had an initial diagnosis of persistent fetal vasculature syndrome due to a posterior lens plaque but surgical intervention demonstrated a temporal retinal fold with a posterior lens attachment consistent with FEVR. A single patient with the double mutation in FZD4also expressed a cysteine mutation in the NDPgene, and demonstrated severe bilateral retinal dysgenesis. This patient was diagnosed with Norrie disease and excluded from this study.

All 4 infants with ROP with FZD4mutations had the double mutation (Table 2), with various degrees of disease severity (Figure 4). Three of the 4 had poor outcomes. Two developed APROP (zone 1, stage 3, Plus) and developed bilateral stage 5 retinal detachments. One infant with classic (zone 2, stage 3, Plus) ROP progressed to bilateral stage 4b despite appropriate and timely laser ablation. One of the premature infants demonstrated peripheral avascular retina only and did not progress to neovascular changes. The infants with APROP were both products of multiple births, one with a surviving twin and the other a sole surviving triplet. The twins were fraternal but both have the FZD4double mutation. The twins had very different postnatal courses; the twin with APROP had multiple comorbidities and poor health, whereas the asymptomatic twin had an uneventful course with no systemic maladies. The surviving triplet, similarly, had multiple comorbidities and poor systemic health. Both cases suggest a role for epigenetics in the development and severity of inherited diseases.

The patients with FEVR generally showed asymmetry between the eyes and ranged from stage 1 disease to stage 5 (Table 1and Table 2; Figure 5). Two patients also had rhegmatogenous retinal detachments. All patients had a history of amblyopia or strabismus shortly after birth (Table 2).

The mutations found in this study involve residues that are located in distinct regions of the FZD4 protein. A novel mutation (C117R) was found in 2 patients. It is located in the extracellular Wnt-binding domain. In the wild-type protein, Cys117 forms a disulfide bond with Cys158 (in the CRD of FZD4).⁸Presumably, disruption of this bond causes a change in the FZD4 conformation, therefore affecting its ability to bind ligands. Wu et al⁹showed that similar cysteine mutations in the CRD of Norrin, a FZD4 ligand, resulted in the most severe phenotype.

The other novel mutation found in this study, C181Y, is located within the N-terminal extracellular domain of FZD4as well. This cysteine residue is not known to form an intracellular disulfide bond but it is the 11th of 13 cysteine residues that are conserved in vertebrate and may be required for dimerization.¹⁰Dann et al¹¹observed that other frizzled receptors can dimerize and proposed that this may be relevant to the mechanism of Wnt binding and signaling.

The Q505X mutation falls immediately after the highly conserved Wnt signaling motif (KTXXXW). Presumably, this early termination results in haploinsufficiency due to nonsense-mediated mRNA decay. It is possible that a truncated protein is produced and terminated prior to the PDZ binding motif (ETXV) of FZD4 (c-Jun N-terminal kinase pathway) (Figures 1and 2). Perhaps noncanonical signal transduction via the c-Jun N-terminal kinase pathway is necessary for normal angiogenesis and vascularization of the retina. Alternatively, the mutant allele may form an oligomer with the wild type, trapping both in the endoplasmic reticulum. Kaykas et al¹²demonstrated that a similar FZD4mutation (L501fsX533) resulted in this occurrence. Finally, Robitaille et al¹³demonstrated that the L501fsX533 mutation decreased activation of the Wnt-Ca²⁺pathway. It is therefore difficult to tell which intracellular Wnt pathway is disrupted by this mutation.

A double missense mutation, P33S/P168S, was found in 9 of the screened patients. The P33S mutation is in the signal sequence portion of the gene and could impair translocation of FZD4 to the plasma membrane. Additionally, proline is not a common amino acid and has the unique characteristic of rigidity that could be essential to the proper conformation of the protein. Another possibility is that this mutation affects a second transcript of the FZD4gene, FZD4S. This transcript variant retains the intron between exon 1 and exon 2 and is 327 amino acids long. Currently, it is unclear whether FZD4Sis an antagonist or agonist of FZD4signaling.^14,15

New blood vessel formation requires the coordination of endothelial cell division and the morphogenic movement of vessel expansion,¹⁶and there is evidence for both canonical and noncanonical signal transduction in this process. For instance, LRP5is known only to participate in canonical signal transduction. Because mutations in LRP5have been associated with some cases of AdFEVR, it is likely that the canonical pathway plays a role in vascular eye development. Additionally, in vitro testing in STF cells has shown that mutations in either FZD4or LRP5result in decreased Wnt/β-catenin signal compared with wild type.³

There is also evidence to support the noncanonical transmission of the signal. Zeng et al¹⁶demonstrated that perpendicular orientation of endothelial cell division is necessary for elongation of new vessels and suggests that planar cell polarity is required for this proper orientation. The FZD4 ligand, Wnt5a, has been shown to activate signal transduction by a planar cell polarity pathway. Wnt5a^−/−animals have a disrupted polarization of sensory hair cell stereocilia in the cochlea of the inner ear,¹⁷similar to that seen in the FZD4^−/−mice, which also have enlarged and disorganized vessels in both the cochlea and retina.³

Perhaps the FZD4signal transduction, through both canonical and noncanonical pathways, is necessary for normal vessel development. One could envision upregulation of endothelial cell growth stimulated by canonical activated transcription followed by noncanonical stimulated cytoskeletal rearrangement and subsequent expansion of cells to form new vessels.

This study highlights the complexity of categorizing genetic diseases. There is a clinical diagnosis and a genetic diagnosis, neither of which alone sufficiently describes the disease state. Often a diagnosis of ROP or FEVR is determined by family and birth histories in addition to clinical findings. This study indicates that the differentiation between these two diseases is not always distinct because the same FZD4 mutation was highly prevalent in both types of patients but not in controls.

Both ROP and FEVR share similar clinical phenotypes such as avascular peripheral retina, neovascular changes, vascular tortuosity and dilation, exudation, and vitreoretinal fibrosis. However, the presence of a peripheral avascular zone alone does not always lead to secondary complications in either disease. This is shown in this study by the fact that the parents of both types of patients and the twin with ROP who had the double-missense mutation were asymptomatic despite peripheral retinal avascularity (Figure 3). Presumably, many individuals with a peripheral avascular zone and a FZD4mutation are never evaluated.

It has long been suspected that genetics plays a role in ROP, and a recent twin study¹⁸concluded that there is a 70% genetic predisposition to development of ROP. Prior to this study, only 1 FZD4mutation in a patient with ROP had been described.²We suggest that the severity of retinal abnormalities is secondary to both genetic predisposition and epigenetics. It is our goal to investigate this possibility in further studies.

Correspondence: Kimberly A. Drenser, MD, PhD, Associated Retinal Consultants, William Beaumont Hospital, 3535 W 13 Mile Rd, No. 344, Royal Oak, MI 48073 (kadrenser@associatedretinalconsultants.com).

Submitted for Publication: March 13, 2009; final revision received May 29, 2009; accepted June 6, 2009.

Financial Disclosure: None reported.

Funding/Support: This study was supported by the Margaret Walters Research Fund; and The Association for Retinopathy of Prematurity and Related Diseases.