Introduction

Familial exudative vitreoretinopathy (FEVR, OMIM: 133780) represents a rare and intricate inherited ocular condition marked by inadequate development of the retinal vasculature, which significantly impairs retinal angiogenesis1,2,3,4,5,6. Clinical manifestations of this disorder exhibit a broad spectrum, encompassing subclinical peripheral retinal vascular anomalies to complete vision loss. These symptoms may include early-onset neovascularization, the presence of falciform folds, subretinal lipid exudation, macular dragging and tractional retinal detachment5,6,7,8. Although it is established that FEVR is a familial condition, the phenotypic characteristics can vary considerably even among affected family members9.

The prevailing understanding is that the pathogenesis of FEVR is complexly linked to the Norrin/β-catenin signaling pathway, which is essential for retinal vascular development10. The Norrin/β-catenin pathway, commonly known as the Norrin/Frizzled-4 (NDP/FZD4) signaling pathway, is an offshoot of the Wnt-β-catenin cascade11. This pathway is initiated by a non-Wnt ligand that engages with a distinct set of receptors11. The extracellular signaling molecule Norrin, produced by the NDP gene, binds to a receptor assembly that includes FZD4, low-density lipoprotein receptor-related protein-5 (LRP5), and Tetraspanin-12 (TSPAN12)11,12. In the absence of Norrin binding, β-catenin undergoes degradation via the destruction complex, which consist of the adenomatous polyposis coli (APC) product, scaffolding protein Axin, glycogen synthase kinase 3 beta (GSK3β), and casein kinase 1 (CK1)13,14. Both CK1 and GSK3β phosphorylate not only the amino terminal region of β-catenin but also the Axin and APC proteins, thereby enhancing the interaction between β-catenin, APC, and Axin and increasing β-catenin phosphorylation13. The phosphorylated β-catenin is subsequently targeted by β-transduction repeat containing protein (β-Trcp), an E 3 ubiquitin ligase that recognizes specific amino acid sequence of some specific proteins, forming a β-Trcp ubiquitin ligase complex11,13. This complex facilitates the transfer of ubiquitin molecules to β-catenin, resulting in its ubiquitination and subsequent proteasomal degradation14,15. Upon the binding of Norrin ligand to the receptor complex, this event is accompanied by the recruitment of the scaffolding protein Dishevelled (Dvl) and the Axin complex to the receptors, leading to the phosphorylation of LRP5/6 by CK1 and GSK3β13. These processes inhibit the Axin-mediated phosphorylation of β-catenin, stabilizing the cytosolic pool of non-phosphorylated β-catenin11. The accumulated β-catenin is then transported into the nucleus, where it forms a complex with T cell factor/lymphoid enhancer factor 1 (TCF/LEF-1), thereby regulating the expression of target genes11.

Up to now, over 18 genes have been identified as being linked to the development of FEVR16,17,18. These genes include FZD4 (OMIM: 604579)19, LRP5 (OMIM: 603506)20, NDP (OMIM: 300658)21, TSPAN12; OMIM: 613138)22, zinc finger protein 408 (ZNF408; OMIM: 616454)23, kinesin family member 11 (KIF11; OMIM: 148760)24, catenin α-1 (CTNNA1; OMIM: 116805)25, catenin β-1 (CTNNB1; OMIM: 116806)26, p120-catenin (CTNND1; OMIM: 601045)27, jagged 1 (JAG1; OMIM: 601920)28, RCC1 and BTB domain containing protein 1 (RCBTB1; OMIM: 607867)29, ATP binding cassette subfamily A member 4 (ABCA4; OMIM: 601691)16, atonal homolog 7 (ATOH7; OMIM: 609875)30, exudative vitreoretinopathy 3 (EVR3; OMIM: 605750)31, integrin-linked kinase (ILK; OMIM: 602366)32, discs large MAGUK scaffold protein 1 (DLG1; OMIM: 601014)33, low-density lipoprotein receptor-related protein 6 (LRP6; OMIM: 603507)34, transforming growth factor-beta receptor 2 (TGFBR2; OMIM: 190182)35, and ER membrane protein complex subunit 1 (EMC1; OMIM: 616846)36. Notably, four genes, NDP, FZD4, LRP5, and TSPAN12 are primarily associated with FEVR as they serve as critical components of the Wnt/Norrin pathway within the retina37,38,39. Extensive researches have indicated that these four genes represent the most prevalent causative factors for FEVR among the identified pathogenic genes40,41,42,43.

The FZD4 gene (Genbank access number: NM_012193.3) is situated on chromosome 11q14.2 of the human genome and encodes a putative protein comprising 537 amino acids, characterized by a 7-transmembrane domain. This gene is categorized within the Frizzled receptor family, which is a part of the class F subfamily of G protein-coupled receptors (GPCR)44. The FZD4 protein features an N-terminal signal sequence, an extracellular cystine-rich domain (CRD), seven transmembrane helices, three intracellular loops, three extracellular loops, and a C-terminal cytoplasmic domain45,46 (Fig. 2). The N-terminal CRD, which is linked to the first transmembrane helix through a linker composed of 50 amino acids, is essential for the recognition of ligands such as Wnt or Norrin, and this domain is conserved across Frizzled family members47,48. FZD4 functions as a receptor within the Norrin/Wnt signaling pathway and is crucial for the processes of cellular signaling, proliferation, and apoptosis2. Its role is indispensable for the proper formation of the retinal vascular system2.

In our preceding research, we identified a novel heterozygous variant c.A749G (p.Y250C) in the FZD4 gene within a Chinese pedigree, situated in the first intracellular loop (Fig. 2A). This variant may represent the pathogenic mutation associated with FEVR19. Comparative analysis of the H. sapiens FZD4 gene across different species, such as Pan troglodytes, Canis lupus, Bos taurus, Rattus norvegicus, Gallus gallus, and so on, demonstrated a high degree of conservation of the tyrosine (Y) residue19. Additionally, a paralogous comparison of FZD4 with other homologous genes in H. sapiens, specifically FZD1-3, 5–10, also demonstrated a significant conservation of this tyrosine (Y)19. Nevertheless, the impact of this variant on the Norrin/Wnt signaling pathway remains to be elucidated. This study aims to investigate the potential pathogenic mechanisms associated with the novel, missense mutation [c.A749G (p.Y250C)] in the FZD4 gene.

Materials and methods

Ethics statement

All animal experiments were conducted in accordance with national regulations and institutional guidelines. And the experimental protocol was approved by the Human Ethics Committee in Southwest Medical University with Approval No.: KY2021237 and the Animal Ethics Committee in Southwest Medical University with Approval No.: 20220928-003. The reporting of experimental methods follows the Animals in Research: Reporting In Vivo Experiments (ARRIVE) guidelines.

Mouse tissue- and stage-specific assays

Mouse used in this study were maintained in polypropylene cages under controlled environmental conditions (22–25 °C, ~ 55% humidity) with a 12/12-h light/dark cycle and ad libitum access to food and water. Regarding the mouse tissue- and stage-specific assays, eight C57BL/6 wild-type mice, including one female mouse at two weeks of gestation and five male mice of different ages, were obtained from the Chongqing Ensibi Biological Technology Co., Ltd. Mice euthanasia was performed by controlled carbon dioxide overdose followed by cervical dislocation for subsequent experimental procedures. Testicular, liver, splenic, renal, cerebral, small intestinal, muscular, and blood tissues were collected from the male mouse, whereas mammary gland, ovarian, uterine, lens, fetal eyeball tissues were extracted from the other pregnant female mouse. Eye tissues were obtained from various developmental stages, including, one week prior to birth (-1 w), newborns, and two-week-old sacrificed mice; retinal tissues were collected from mice aged 1, 2, 3, 6, and 12 months; and lens, sclera and cornea tissues from three-month-old mouse. These tissues were weighed, individually sliced, and thoroughly homogenized on ice at the ratio of 50 mg of tissue to 1 ml of 1 × EBC buffer. Then samples were then transferred into separate 1.5 ml EP tubes and placed at 4 °C or on ice for 30 min. The tissue lysates underwent centrifugation at a speed of 12,000 rpm at 4 °C for 15 min. The supernatant was carefully pipetted into new clean 1.5 ml EP tubes, and an equal volume of 2 × SDS Loading Buffer was introduced. The extracts were mixed thoroughly and heated at 100 °C for 5 min, post that they were utilized for western blotting or stored at −20 °C till for use.

Construction of plasmid containing c.A749G mutation in FZD4 gene

The c.A749G mutation (FZD4-MUT) identified in the genome DNA of the proband was utilized as template for amplification using site-specifically designed primer (M362-FZD4) (Table 1) to get the mutant DNA fragments. The resultant PCR products underwent agarose gel electrophoresis at a 1.5% concentration, where the distinct target band was excised under ultraviolet illumination for subsequent purification and recovery. The wild-type FZD4 plasmid (GFP-tagged) (FZD4-WT) was sourced from OriGene (USA) (cat. : RG217286). Both the FZD4-WT plasmid and the mutant DNA fragment of PCR products were subjected to digestion with EcoR V (cat. : R6391, Promega, USA). Following digestion, they were analyzed through agarose gel electrophoresis at 1% concentration. The above 8000 bp bands obtained from the initial gel exposure were excised and purified by standard DNA fragment recovery kit (cat. : DP209, TIANGEN, China) to obtain the expression vectors, while the digestion products from mutant DNA PCR were purified using a column method to extract an even shorter mutant DNA fragments to substitute the corresponding wild-type fragments in later FZD4-MUT plasmid construction.

Table 1 PCR primer and PCR product size.
Full size table

A 1 µl aliquot was taken from the purified or recovered products above, and combined with 4 µl ddH2O and 1 µl 6 × DNA loading buffer to create a total volume of 6 µl, which was then subjected to agarose gel electrophoresis at 1% concentration. The necessary amount of DNA for ligation was determined based on the intensity of the DNA bands. A combination of 0.75 µl of vector (pCDNA5) fragments and 1.5 µl the substitution DNA fragments was ligated in the presence of DNA T4 to form the FZD4-MUT plasmid. The recombined products were transformed into DH5α competent bacteria which were cultured for PCR amplification utilizing the primers (M362-FZD4). Plasmids extraction was performed from the corresponding bacterial cultures yielding positive results with appropriate molecular weight as verified by agarose gel electrophoresis. The sequencing of the PCR products was conducted via the dideoxy DNA sequencing technique on an ABI-3500DX sequencer (Applied Biosystems, Inc., Foster City, CA) in our laboratory, utilizing a specific primer known as M362-FZD4-749L.

Cell culture, transfection, and treatment

The human cervical cancer HeLa cell line was obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA) and maintained at a temperature of 37 °C in Dulbecco’s Modified Eagle Medium (DMEM) enriched with 10% Fetal Bovine Serum (FBS) and 1% penicillin/streptomycin. For transient transfections, cells were exposed to expression plasmids for a duration of 32–48 h utilizing Lipofectamine 3000 (cat. : L3000015, TSINGKE, China) in alignment with the manufacturer’s specifications tailored for various experimental conditions. For the purposes of fluorescent inverted microscopy, co-immunoprecipitation and western blot analysis, HeLa cells cultivated in 6-well plates underwent transfection with 150 ng of FZD4-WT/FZD4-MUT plasmids and simultaneous co-transfected with 50 ng NDP-Myc plasmids (cat. : RC207052, OriGene, USA) or treatment with R-Spondin 1 (RSPO1) protein (cat. : TAY1217111, Bio-techne, USA).

Luciferase reporter assay

HeLa cells were subjected to transfection with 100 ng of TOP-Flash/ FOP-Flash reporter constructs (cat. : CLS-018L, Promega, USA) with about 5.1 kb in length, which include multiple copies of β-catenin DNA binding site TCF/LEF-1 within the promotor regain, alongside 30 ng of either wild-type (FZD4-WT)/ mutant (FZD4-MUT) plasmids using Lipofectamine 3000 when the cells reached 40–50% confluence. After 48 h post-transfection, cells were harvested to assess luciferase activity by adding Luciferase Assay Reagent II and measuring the luminescence on a 3010 Luminometer (BD Monolight™, USA). The results were normalized against the control luciferase activity. Each experimental condition was conducted in triplicate and repeated on two separate occasions.

Co-immunoprecipitation (Co-IP) and western blot analysis

Two groups of HeLa cells cultured in 6-well plates were transfected with 150 ng of FZD4-WT/FZD4-MUT plasmids. For the Co-IP group, cells were co-transfected with 50 ng NDP-Myc plasmids using Lipofectamine 3000 for 32 h. Subsequently, these cells were incubated in 2 ml of activated charcoal adsorbed 10% FBS for a duration of 4 h at 37 °C. The other group of cells for western blot were transfected together with 50 ng NDP-Myc plasmids using Lipofectamine 3000 for 32 h, followed by incubation in 2 ml activated charcoal adsorbed 10% FBS for 4 h at 37 °C or treatment with a DMEM: RSPO1 at a ratio of 500:1, also incubating at 37 °C for 4 h. Then cells were lysed in 1 × EBC buffer, which comprises 20 mM Tris–HCl (pH 8.0), 125 mM NaCl, 2 mM EDTA, and 0.5% NP-40, supplemented with a proteinase inhibitor, vortexed, and incubated on ice for 30 min. Subsequent to centrifugation at 12,000 g for a duration of 15 min at a temperature of 4 °C, the resultant supernatants were pre-cleared. A 10% aliquot of the supernatant was mixed with an equal volume of 2 × SDS sample buffer and boiled at 100 °C for 5 min to serve as the Input. The lysates were subjected to immunoprecipitation by the addition of anti-GFP antibody or normal IgG, followed by overnight rotation at 4 °C. Subsequently, Agarose A/G beads were introduced to each tube, and the mixtures were rotated at 4 °C for 1.5–2 h. The immunocomplexes were subjected to four washes with washing buffer, after which the associated proteins were eluted utilizing 2 × SDS sample buffer. The ensuing steps followed standard immunoblotting protocols, wherein the samples underwent separation using SDS-PAGE before being transferred to nitrocellulose membranes. The membranes were incubated with primary antibodies, including FZD4 (cat. : ab277797, Sbcom), LRP5 (D80F2) (cat. : #5731, Cell Signaling), GFP (cat. : TA150041, Origene, USA), TSPAN12 (cat. : ab09964, Abcom, UK), NDP-MYC (cat. : #2278, Cell Signaling, USA), p-β-catenin (cat. : #9566, Cell Signaling, USA), β-catenin (E-5) (cat. : sc-7963, Santa Cruz, USA), VEGF-A (cat. : ab46154, Abcom, UK), tubulin (cat. : 66031–1-Ig, Proteintech, USA), IgG (cat. : TI100062, OriGene, USA), and GAPDH (cat. : 60004–1-Ig, Proteintech, USA). Following this, the membranes were washed and subsequently incubated with either a secondary goat anti-mouse antibody (cat. : BS22356, Bioworld, USA) or goat anti-rabbit antibody (cat. : A0208, Beyotime, CHN). Finally, the membranes were treated with a horse radish peroxidase (HRP) antibody, and immunoreactive proteins were visualized with ChemiDoc XRS + (BIO-RAD, USA). Image J was used to quantify protein expression levels, and Origin 2021 was employed to generate the corresponding bar graph.

Fluorescent inverted microscopy and image analysis

In the section dedicated to fluorescent inverted microscopy and image analysis, HeLa cells grown in 6-well plates underwent transfection with 150 ng of FZD4-WT/FZD4-MUT plasmids and 50 ng NDP-Myc plasmids for 36 h. The cells were subsequently maintained in 10% FBS that had been absorbed with activated charcoal for 4 h at 37 °C. Each well received 1 ml of 4% formaldehyde for fixation in dark for 15 min. Cells were incubated with or without Calnexin antibody (cat. #81938–1-RR, Proteintech, USA) overnight at 4 °C, respectively. Subsequently, the cells were incubated with a fluorescently-labeled goat anti-rabbit secondary antibody (cat. #A0468, Beyotime, China) for 2 h at room temperature in the dark. Then 1 ml of Hoechst 33258 (C0020, Solarbio, CHN) staining solution was introduced into each well and allowed to incubate at 37 °C for a duration of 2 h. The cells were examined under Zeiss AxioImager.Z2 (Carl Zeiss, Germany) to capture the fluorescence images.

Statistical analysis

The statistical evaluation of the data was conducted using GraphPad Prism 8. 0. 1 software. The differences in distribution were examined using Student’s t test and the chi-squared (χ2) test, with a significance threshold set at P < 0.05.

Results

Spatiotemporal expression analysis of Fzd4 protein

To understand the differential expression patterns of Fzd4 across various tissues and developmental stages within the mouse retina, we performed western blot to detect the expression of Fzd4 protein. Western blot results showed that Fzd4 was predominantly expressed in the mouse brain, retina (specifically in fetal mouse eye tissue collected one week prior to birth), as well as in uterus, kidney, and liver tissues; however, it was absent in other tissues (Fig. 1A). Additionally, Fzd4 expression was detectable in the ocular tissues of both the fetal mouse at one week before birth and newborn mouse, while no expression was observed in other developmental stages of the retina or in other ocular tissues (Fig. 1B). These results suggest that Fzd4 protein is primarily expressed during the early phases of retinal and ocular development.

Fig. 1
Fig. 1The alternative text for this image may have been generated using AI.
Full size image

The tissue-specific and stage-specific expressions of Fzd4 protein in retinal tissue. The expression of Fzd4 protein in the indicated tissues (A) and at the specified developmental stages or time points in retinal tissue (B) derived from wild-type C57 mice. The retina samples were obtained from embryonic ocular tissues collected one week before birth (-1 w) as depicted in panel A. Whole eyeballs were sourced from embryos at both one week before parturition (-1 w) and at the time of birth (0 d) as illustrating in panel B. d, day (s); w, week (s); m, month (s); muscle, skeletal muscle; blood, peripheral blood. * indicates non-specific bands. In panel A, Fzd4 and GAPDH were from one membrane; in panel B, they were from another membrane. Full-length blots are provided in the Supplementary Fig. 1.

Construction of plasmid containing c.A749G mutation in the FZD4 gene

After the acquisition of PCR products containing the identified c.A749G mutation within the FZD4 gene from the proband, the FZD4-WT plasmid and the target DNA fragments harboring the c.A749G mutation were digested with EcoR V. All digestion products from the FZD4-WT plasmid and the target DNA fragments were subjected to agarose gel electrophoresis, confirming the successful retrieval of predicted DNA fragments (Fig. 2B). For details, lanes 1–3 contained the complete digestion products of the FZD4-WT plasmid (totaling 40 µl), with 8179 bp in length, which yielded DNA fragments measuring about 8000 bp and 179 bp post-digestion. In lane 4, the electrophoresis results of a 1 µl aliquot of target DNA fragment demonstrated that the substitution DNA fragment achieved a size of 179 bp. The 8000 bp bands observed in lanes 1–3 were subsequently purified from agarose gel, yielding the expression vector. And the remaining digestion products of target DNA fragments were purified by column purification to acquire substitution DNA fragment. A 1 µl sample from each was analyzed using agarose gel electrophoresis to quantify the DNA based on the intensity of DNA bands (Fig. 2C).

Fig. 2
Fig. 2The alternative text for this image may have been generated using AI.
Full size image

Construction and verification of mutant FZD4 plasmids. (A) A schematic illustration of human FZD4 protein is presented, with numbers donating the amino acid positions within the specified domains of FZD4. The mutation located in the first intracellular loop, as the arrow indicated. SPs: Signal peptides. CRD: Cysteine-rich domain, which serves as the ligand-binding domain. TM: Single pass of the transmembrane domain. (B) The electrophoresis results reveal the FZD4 amplified plasmid and target DNA fragment after enzyme digestion. Lanes 1–3 lanes exhibit all digestion products from the FZD4 plasmid, with DNA fragment sizes of 8000 bp (the black arrow indicated) and 179 bp (the white arrow indicated) after digestion, respectively; lane 4 displays the electrophoresis outcome of 1 µl target DNA fragment, confirming that the substitution DNA fragment obtained post-enzyme digestion was 179 bp (the white arrow indicated) in size. (C) The results of electrophoresis for 1 µl of each expression vector DNA fragment (the black arrow indicated) in lane 1 and the substitution DNA fragment (the white arrow indicated) in lane 2 before connection serves as a reference for the preparation of the ligation system. (D and E) Partial PCR identification results of the bacterial solutions. Three positive clones were identified from the PCR electrophoresis results: No. 9, No. 104 and No. 107, as indicated by arrows. Sequencing results in (F and G) confirm that the positive clone plasmids No. 9 and No. 104 contain reverse substitution sequences carrying mutation sites. (H) The positive clone plasmid No. 107 possesses a forward substitution sequence harboring the mutation. (M) DNA molecular size marker DL5000. The arrows signify the mutation sites. Full-length blots are included in the Supplementary Fig. 2.

To confirm the successful identification of the FZD4-MUT positive plasmid post-ligation, the outcomes of agarose gel electrophoresis conducted on the PCR products revealed that the clones of No. 9, No. 104, and No. 107 exhibited the appropriate molecular weight, 285 bp (Fig. 2D & E). Sequencing analyses of these three positive clones indicated that the plasmids No. 9 and No. 104 (Fig. 2F-G) contained reverse insertion sequences with mutation sites, while the plasmid No. 107 possessed a forward insertion sequence carrying correct mutation. This confirmed the successful cloning of the mutant FZD4 plasmid (FZD4-MUT) (Fig. 2H). The FZD4-MUT plasmid was constructed to harbor the c.A749G mutation, differing from the FZD4-WT plasmid only at the mutated nucleotide, thus enabling a direct comparison of their functional effects in the following experiments.

The mutant FZD4 affects the Norrin/β-catenin downstream signaling pathway

To evaluate the biological activity of FZD4-MUT, a TOP-Flash/FOP-Flash reporter assay was performed. HeLa cells were transfected with TOP-Flash/ FOP-Flash reporter constructs along with wild-type (FZD4-WT) or mutant (FZD4-MUT) plasmids. The resulting relative light units (RLU) for FZD4-WT and FZD4-MUT were (0.0992 ± 0.0160) and (0.0260 ± 0.0096), respectively (Fig. 3), revealing that the mutant FZD4 proteins deceased the β-catenin level which further attenuating the activity of the Norrin-β-catenin signaling pathway. These results demonstrated that this variant may have a deleterious affection on FZD4 protein functionality and could be implicated in the pathogenesis of FEVR.

Fig. 3
Fig. 3The alternative text for this image may have been generated using AI.
Full size image

Detection of wild-type and mutant FZD4 on downstream signal proteins by luciferase reporter gene assay. FZD4-wild-type (FZD4-WT) and FZD4-mutant (FZD4-MU) represent the relative luciferase activity ascertained from the transfection of wild-type FZD4 plasmid and mutant FZD4 plasmid into Hela cells, respectively. Data are representative of three independent experiments. The experimental data are presented as the mean ± standard deviation (mean ± SD). Differences between groups were assessed using the Student’s t-test, and *P < 0.05.

Detection of the interaction between mutant FZD4 and LRP5

Within the context of the Norrin signaling pathway, FZD4 is known to interact with LRP5 to form complexes. In order to investigate whether the interaction between mutant FZD4 and LRP5 was changed, we conducted Co-IP and western blot analysis. The results showed that there was no discernible difference in the levels of FZD4 and NDP in the two groups; furthermore, the interactions between FZD4-WT or FZD4-MUT proteins with LRP5 were consistent, suggesting that the integrity of the receptor protein complex remained unaffected (Fig. 4). However, TSPAN12 exhibited no interaction with either of them (Fig. 4), indicating that FDZ4 forms complexes with both LRP5 and NDP (FZD4/LRP5/NDP complex), not with TSPAN12.

Fig. 4
Fig. 4The alternative text for this image may have been generated using AI.
Full size image

Examination of interaction between mutant FZD4 and proteins via Co-IP and western blot. Cells transfected with FZD4-WT or FZD4-MUT plasmids were incubated with FZD4-GFP antibody prior to the immunoprecipitation of the binding complex. Western blot analysis was conducted to observe its binding affinity with LRP5, TSPAN12 and NDP, with IgG serving as immunoprecipitation control. Full-length blots are included in the Supplementary Fig. 4. The results for LRP5, FZD4-GFP, TSPAN12, and NDP-MYC were from a same nitrocellulose membrane.

Subcellular localization analysis of the mutant FZD4 protein

Given that the interaction between the mutant FZD4 and LRP5 remained largely unchanged, we aimed to determine whether the subcellular localization of the mutant FZD4 has been altered. Therefore, we performed a subcellular localization analysis of both wide-type and mutant FZD4 proteins by overexpression experiments. Previous studies have consistently shown that mutant FZD4 proteins are often retained in the endoplasmic reticulum (ER). So, we want confirm that weather this FZD4 mutation also contain the same characteristics. Cells were transfected with NDP-Myc plasmids together with FZD4-WT or FZD4-MUT plasmids, which facilitated the expression of a fusion protein that produces green fluorescence upon expression. After 36 h of transfection, the cells were incubated with or without Calnexin antibody, then subjected to Hoechst 33258 staining, which highlighted the nucleus with blue fluorescence. The immunostaining result showed both wild-type FZD4 and mutant FZD4 can co-localized with Calnexin, while mutant FZD4 has a tendency to accumulate in the nucleus and nuclear membrane (Fig. 5A–H), and this phenomenon may different from previous reported ER congregation. When cells were only treated with FZD4 plasmid, the highlighted green fluorescence associated with the FZD4-MUT protein also exhibited nuclear membrane and nucleus aggregation (Fig. 6J–L), whereas the FZD4-WT protein displayed a greater abundance of brighter, spot-like green fluorescence that was predominantly located within the cell membrane and cytoplasm (Fig. 6D–F). The cells without predominantly localization of FZD4-WT or FZD4-MUT protein were showed in Fig. 6A–C and Fig. 6G–I, respectively.

Fig. 5
Fig. 5The alternative text for this image may have been generated using AI.
Full size image

Immunostaining for FZD4 and Calnexin. The immunostaining for wild-type or mutant FZD4 and Calnexin in Hela cells. FZD4 protein shows in the green image on the left side of the panel; Calnexin displays with red image; nucleus shows in the blue image; the merge of them shows on the right side. Scale bar: 10 μm. Magnification: × 50.

Fig. 6
Fig. 6The alternative text for this image may have been generated using AI.
Full size image

Subcellular localizations and nuclear entry ratios of wild-type and mutant FZD4 proteins. (A–C) Subcellular localization of wild-type FZD4 protein express throughout the cytoplasm in HeLa cells, without highlights. (D–F) Subcellular localization of wild-type FZD4 protein aggregate on cell membrane and cytoplasm of HeLa cells, with highlights. (G–I) Subcellular localization of mutant FZD4 protein express throughout the cytoplasm in HeLa cells, without highlights. (J–L) Subcellular localization of mutant FZD4 protein aggregate on nucleic membrane and nucleus of HeLa cells, with highlights. The highlights represent the aggregation of GFP-labeled FZD4 protein. FZD4 protein shows in the green image on the left side of the panel; nucleus shows in the blue image in the middle; the merge of them shows on the right side. M: The ratio of wild-type FZD4 and mutant FZD4 proteins entering the nucleus. FZD4-WT represents the proportion of wild-type FZD4 protein entering the nucleus, while FZD4-MUT denotes the proportion of mutant FZD4 protein entering the nucleus. *P < 0.05. Scale bar: 5 μm. Magnification: × 50.

To further elucidate these findings, we selected cells exhibiting distinct green fluorescence signals and clear localizations patterns, to quantify the number of wild-type and mutant FZD4 protein fluorescence signals that entered the nucleus in Hela cells. A comparison of subcellular localization differences between wild-type and mutant FZD4 proteins was conducted. A total of 55 wild-type FZD4 positive cells were analyzed, resulting in the identification of 13 cells with fluorescence entering nucleus cells, yielding a ratio of 0.236. In contrast, among 30 cells transfected with the FZD4-MUT plasmids, 20 cells demonstrated fluorescence entering the nucleus, leading to a ratio of 0.667. As illustrated in the bar chart (Fig. 6M), the proportion of mutant FZD4 proteins entering the nucleus was markedly higher than that of wild-type FZD4 protein. This observation suggests that the FZD4-MUT variant may indeed alter the subcellular localization. Consequently, the aggregation of mutant FZD4 protein at perinuclear membrane and within the nucleus may have implications for Norrin/β-catenin signaling pathway.

Effects of mutant FZD4 on the downstream Norrin/β-catenin signaling pathway

In order to investigate the influence of FZD4-MUT on Norrin/β-catenin signaling cascade, we conducted western blot analysis to evaluate the expression levels of downstream protein. It is noteworthy that the phosphorylation of distinct residues on β-catenin exerts varied effects on the Norrin/β-catenin signaling pathway. Specifically, phosphorylation at serine residues 33, 37, and 45, along with the phosphorylation of tyrosine at position 41, contributes to a reduction in the stability of β-catenin protein. Conversely, phosphorylation at position 552 promotes the aggregation of β-catenin in the nucleus, thereby enhancing its transcriptional activity. In our study, we utilized a phospho-specific antibody targeting β-catenin at position 552. Additionally, Norrin and RSPO1 proteins were identified as agonists of the Norrin/β-catenin signaling pathway. The results indicated that the expressions of β-catenin level were consistent across both the two treatment groups; while the expressions of p-β-catenin were obviously decreased in the FZD4-MUT conditions (Fig. 7A). This observation implies that the mutation may influence the phosphorylation state of β-catenin. Compared to the wild-type FZD4, the expression of VEGF-A, a vascular endothelial growth factor, was diminished in both FZD4-MUT conditions. The relative expression level of each protein in these two groups was showed as bar graph in Fig. 7B and C, respectively. From this, we deduced that the FZD4-MUT protein decreased the phosphorylation level of β-catenin rather than the total β-catenin level, thereby regulating the Norrin/β-catenin signaling pathway through attenuating VEGF-A expression. This investigation further elucidated at the molecular level that the c.A749G (p.Y250C) variant in the FZD4 gene is one of the primary genetic determinants of FEVR disease in the family we previously reported19.

Fig. 7
Fig. 7The alternative text for this image may have been generated using AI.
Full size image

Western blot detection of the regulation of downstream signaling proteins by mutant FZD4. (A) The protein expression levels of p-β-catenin, β-catenin, and VEGF-A after HeLa cells were transfected with FZD4-WT/FZD4-MUT plasmids alongside NDP-Myc plasmids (FZD4 WT + NDP and FZD4 MUT + NDP) groups, or treated with FZD4-WT/FZD4-MUT plasmids and RSPO1 protein (FZD4 WT + RSPO1 and FZD4 MUT + RSPO1) groups, respectively. The results for p-β-catenin and Tubulin were from the same membrane, while the results for β-catenin and VEGF-A were from another separate membrane. Since the molecular sizes of GAPDH and VEGF-A are similar and cannot be well separated, Tubulin was used as the internal reference here. (B and C) Bar graphs demonstrated corresponding protein levels. The results are from three independent experiments performed in triplicate. # represents non-specific bands. *P < 0.05. NS represents not significant. Full-length blots are displayed in the Supplementary Fig. 7.

Discussion

FEVR is a hereditary condition affecting the vitreous and retina, characterized by retinal vascularization abnormalities and diverse clinical manifestations49. It features a peripheral retina lacking vascularization, which can be asymptomatic early on but may lead to complications like vitreoretinal traction and retinal detachment in advanced stages, causing severe vision impairment50,51. Our preceding research identified a novel heterozygous variant c.A749G (p.Y250C) in the FZD4 gene within a Chinese pedigree of FEVR19. Several types of mutations in various genes are associated with FEVR. As of April 14, 2025, a comprehensive analysis of the Human Gene Mutation Database (HGMD Professional 2024.2) has identified 133 pathogenic variants within the FZD4 gene associated with FEVR. This includes a variety of mutation types: 25 deletions, 12 insertions, 1 complex rearrangement, and the 88 single nucleotide substitutions.

Among all the FEVR pathogenic genes, Li et al. and Seo et al. demonstrated that patients with variants in the NDP gene exhibited a more severe clinical manifestation of FEVR compared to those with variants in the FZD4 gene52,53. Furthermore, Chen et al. reported that among asymptomatic mild FEVR cases, pathogenic mutations in six major FEVR-related genes were identified in 48.4% of individuals, with FZD4 mutations accounting for a large proportion (21.0% of all mutations)54. Concurrently, Lahteenoja et al. announced that the vast majority (94%, 33/35) of carriers of a pathogenic FZD4 variant were visually functional without meeting WHO criteria for moderate or severe impairment55, implying that FZD4 mutations are likely resulted in milder FEVR symptoms. In contrast, Wang et al. found that patients with mutations in either the NDP or FZD4 genes exhibit more pronounced symptoms, with 21.4% of these patients classified as mild FEVR (stage 1–2) and 78.6% categorized as severe (stage 3–5)56. Collectively, prior studies have indicated that FZD4 mutations represent the largest proportion of FEVR cases52,53,57. Despite the clinical importance of establishing a genotype–phenotype correlation for FZD4-related FEVR, studies have not yielded consistent evidence. Lu et al.'s large-cohort analysis found no significant correlation between the specific domains of FZD4 mutations (including the CRD) and disease severity, retinal folds, detachment, or other ocular abnormalities58,59. Similarly, Robitaille et al. reported that the mutation type (missense, frameshift deletion, etc.) did not influence the clinical phenotype60; however, a contrasting study by Yaylacioglu et al. observed a higher incidence of retinal detachment in patients with truncating variants compared to those with nucleotide substitutions61.

Norrin, a cysteine knot protein, exclusively interacts with one of the ten human FZD receptors, specifically FZD447,62. In contrast to the canonical Wnt/FZD pathway, the signaling mediated by Norrin and FZD4 involves an additional four transmembrane proteins, notably TSPAN1263. TSPAN12 not only enhances Norrin-induced signaling64, but also plays a crucial role in mediating the ligand selectivity of FZD4 receptor complexes63. Several previous studies have shown that TSPAN12 could combine with Norrin-FZD4 complexes63,65, a process that appears to hinge on the large extracellular loop (LEL) of TSPAN1263,64, and finally forming a ternary complex together with LRP566. However, the specifics of how TSPAN12 interact with FZD4, including the precise binding sites on FZD4, remain partially elucidated.

In our study, we observed that TSPAN12 was not expressed in either the wild-type or mutant FZD4 plasmids post-transfected, and the expressions levels of LRP5 were notably low in Co-IP assay. This may be attributed to the time of the experiment, conducted 36 h post-transfection, during which the exogenous expression of FZD4 may not have been sufficiently transported to the cell membrane, thus precluding the co-precipitation of TSPAN12. An alternative explanation could be that the binding of TSPAN12’s LEL to pre-existing Norrine may have reduced the likelihood of TSPAN12 binding to FZD4, which is necessary for its transport to the cell membrane. Interestingly, Bruguera et al. proposed that the presence of FZD4 may diminish the affinity between Norrin and TSPAN12, despite the fact that TSPAN12 and FZD4 can form a heterodimer that binds a single Norrin protomer65. Moreover, TSPAN12 can transfer captured Norrin to nearby FZD4 receptors competing with LRP5/6 for Norrine binding before the FZD4-Norrin-LRP5/6 complex formation for signaling purposes65. This suggests that TSPAN12 does not remain associated with Norrin within a quaternary complex involvingFZD4 and LRP5/6 co-receptors65. However, biochemical evidence supporting this part was insufficiency, and further investigation are warranted to elucidate the assembly of accessory protein into receptor complexes.

Our investigation found that the aggregation of the FZD4-MUT protein at the perinuclear membrane and nucleus was significantly increased than that of wild-type FZD4 protein. There was limited research on the subcellular localizations and functional impacts of pathogenic mutations in FZD4. Several mutant FZD4 proteins in Reham’s study were mislocalized, exhibiting a distinct perinuclear reticular pattern instead of localizing to the plasma membrane67. Robitaille et al. discovered a deletion of two nucleotides 1501_1502delCT in FZD4 gene, which resulted in L501fsX533 frameshift mutation. This mutation prevented the protein from aggregating at the cell membrane, leading to its retention within the endoplasmic reticulum (ER). Furthermore, these mutated FZD4 proteins were capable of interacting with wild-type FZD4 to form heterodimers in the ER, thereby obstructing their signaling pathway68,69. The improperly folded proteins were subsequently exported from the ER and directly transferred to lysosomes for degradation, which prevented them from reaching their intended cellular destination and resulting in a loss of function67. This mechanism sheds light on how certain mutated FZD4 can lead to retinal dysfunction67, and may also offers us the reason for the changes of subcellular localization. Although the mislocalization of mutant FZD4 to the nuclear membrane and nucleus observed in this study may partly result from the reasons mentioned above, it could also stem from protein misfolding and subsequent transport defects. The precise mechanism underlying the enhanced nuclear accumulation of FZD4-MUT relative to the wild-type protein, however, requires further investigation.

Norrin signaling pathway is essential for retinal vascular development. Genetic defects in this pathway, specifically involving NDP, FZD4, and LRP5, underlie the retinal hypovascularization observed in FEVR70. These mutations prevent β-catenin activation, impaire endothelial cell migration, and lead to a loss of vascular dendrites71. Consequently, FZD4 deficiency disrupts critical processes such as endothelial migration and the programmed regression of vitreous vessel71. During early embryonic development, inactivation of the Norrin/FZD4/LRP5 signaling pathway reduces the expression of essential factors that promote retinal blood vessel formation. At this stage, the function of VEGF signaling, which normally drives proper vascular growth, is relatively weakened or ineffective72, which cause peripheral retinal avascularity73. Subsequently, pathological hypoxia develops in retinal tissue due to insufficient vascularization, stimulating VEGF-A expression and leading to the formation of abnormal and fragile new blood vessels (neovascularization)74. These vessels not only grow in incorrect locations, such as the vitreous cavity, but are also prone to leakage and hemorrhage, which can cause serious complications like tractional retinal detachment74. Therefore, anti-VEGF therapy is commonly used to treat neovascular growth in such diseases73. In this study, the observed decrease in VEGF-A expression in FZD4 mutant group may be attributed to the shorter duration of FZD4-MU plasmid treatment, which models the early stage of disease progression before hypoxia has been induced.

Yang et al. observed that family-conserved amino acids (Y2502.39f.) adopted different conformations in the CRD-FZD4 when compared to other class-F receptors62. This specific residue has been implicated in downstream signaling and in maintaining structural integrity, indicating that this crucial residue has evolved to adopt distinct conformations in various proteins to fulfill their respective functions75. Collectively, this evidence indirectly proved the theoretical pathogenicity of this residue mutation. Our research has further established the actual occurrence of this mutation in human FEVR disease, thereby substantiating the pathogenic variant c.A749G (p.Y250C) in the FZD4 gene.

This study provides mechanistic insight into the pathogenicity of the FZD4 p.Y250C mutation. However, several limitations should be acknowledged. The use of HeLa cells, rather than retinal-specific models, limits the pathological relevance of the findings. Although disrupted molecular signaling and aberrant protein localization were identified, the study did not functionally validate impaired angiogenesis, for example, by conducting tube formation assays. Furthermore, the localization analysis had restricted statistical power due to a small sample size. The mechanism behind the mutant protein’s abnormal nuclear accumulation remains unknown, which constrains understanding of the mutation’s full impact.

Conclusion

This study demonstrates that the novel FZD4 missense mutation (c.A749G, p.Y250C), identified in a FEVR pedigree, exhibits pathogenic potential through a dual mechanism. Firstly, the mutant FZD4 protein displays altered subcellular localization, with a significant tendency to accumulate at the nuclear membrane and within the nucleus, in contrast to the predominant cytoplasmic and plasma membrane localization of the wild-type protein. Secondly, despite preserving its ability to bind the coreceptor LRP5 and form a complex with Norrin, the mutant protein significantly attenuates the Norrin/β-catenin signaling pathway. This is evidenced by a marked reduction in downstream β-catenin reporter activity, decreased levels of phosphorylated β-catenin (Ser552), and reduced expression of a key angiogenic factor, VEGF-A. Furthermore, the broad expression of Fzd4 in murine neural and other tissues, along with its notable presence during early retinal development, underscores its critical role in ocular vascular formation. Collectively, these findings suggest that the p.Y250C mutation contributes to FEVR pathogenesis by impairing canonical Norrin/β-catenin signaling and disrupting the normal trafficking or stability of the FZD4 protein.