Genetic defects of GDF6 in the zebrafish out of sight mutant and in human eye developmental anomalies
© den Hollander et al. 2010
Received: 16 August 2010
Accepted: 11 November 2010
Published: 11 November 2010
Skip to main content
© den Hollander et al. 2010
Received: 16 August 2010
Accepted: 11 November 2010
Published: 11 November 2010
The size of the vertebrate eye and the retina is likely to be controlled at several stages of embryogenesis by mechanisms that affect cell cycle length as well as cell survival. A mutation in the zebrafish out of sight (out) locus results in a particularly severe reduction of eye size. The goal of this study is to characterize the out m233 mutant, and to determine whether mutations in the out gene cause microphthalmia in humans.
In this study, we show that the severe reduction of eye size in the out m233 mutant is caused by a mutation in the zebrafish gdf6a gene. Despite the small eye size, the overall retinal architecture appears largely intact, and immunohistochemical studies confirm that all major cell types are present in out m233 retinae. Subtle cell fate and patterning changes are present predominantly in amacrine interneurons. Acridine orange and TUNEL staining reveal that the levels of apoptosis are abnormally high in out m233 mutant eyes during early neurogenesis. Mutation analysis of the GDF6 gene in 200 patients with microphthalmia revealed amino acid substitutions in four of them. In two patients additional skeletal defects were observed.
This study confirms the essential role of GDF6 in the regulation of vertebrate eye size. The reduced eye size in the zebrafish out m233 mutant is likely to be caused by a transient wave of apoptosis at the onset of neurogenesis. Amino acid substitutions in GDF6 were detected in 4 (2%) of 200 patients with microphthalmia. In two patients different skeletal defects were also observed, suggesting pleitrophic effects of GDF6 variants. Parents carrying these variants are asymptomatic, suggesting that GDF6 sequence alterations are likely to contribute to the phenotype, but are not the sole cause of the disease. Variable expressivity and penetrance suggest a complex non-Mendelian inheritance pattern where other genetic factors may influence the outcome of the phenotype.
Microphthalmia, anophthalmia, and chorioretinal coloboma are ocular malformations that affect 1 in 3000-4000 individuals [1–3]. In microphthalmia and anophthalmia, one or both eyes are abnormally small or clinically absent. Colobomata are clefts caused by absent eye tissue, due to a failure of the optic fissure to close. Colobomata are frequently grouped with microphthalmia and anophthalmia, as they are often associated with a reduction of eye size. The aetiology of these ocular malformations is complex, and a wide variation is seen in phenotypic expression. Recessive, dominant and X-linked modes of inheritance have been described, but often sporadic and non-Mendelian inheritance patterns are seen [4, 5]. In addition, a variety of environmental factors may be causative in certain cases, including viral infection, such as rubella, irradiation and drug intake in pregnancy.
Mutations in the transcription factor SOX2 are the most prevalent monogenic cause of microphthalmia and anophthalmia identified to date . Other genes include the transcription factors PAX6, OTX2, CHX10 and RAX [7–10]. More recently, mutations in three members of the transforming growth factor-β (TGF-β) superfamily (BMP4; GDF6, also known as BMP13; and GDF3) have been associated with microphthalmia/anophthalmia [11–15]. Members of the TGF-β superfamily of secretory signaling molecules play essential roles in embryonic development [16, 17]. Members of this superfamily regulate cell proliferation and apoptosis, and play important roles in various processes such as the creation of dorsal-ventral axes in the embryo, specification of the neural crest, bone formation, and organogenesis [18–20].
A segmental deletion encompassing the GDF6 gene, and several amino acid substitutions in GDF6 have been identified in patients with ocular anomalies, including coloboma and microphthalmia [13–15]. In addition, a chromosomal rearrangement and several amino acid substitutions in GDF6 have been detected in patients with Klippel-Feil syndrome, a complex skeletal disorder characterized by congenital fusion of the cervical spine, and in patients with other skeletal defects [21, 14]. Gdf6 is expressed in the dorso-temporal retina, and morpholino (MO) knockdown experiments of Gdf6 in zebrafish and Xenopus result in reduced eye size or even the absence of optic lobes [13, 22]. As these experiments were performed using antisense compounds, it is likely that this variability reflects imperfections of the morpholino knockdown approach. In Xenopus, and to a lesser extent in zebrafish, this phenotype is accompanied by a disorganization of retinal layering [22, 13]. In addition, the presence of skeletal anomalies (curled or kinked tails) was observed in a fraction of MO-treated zebrafish embryos. The loss of Gdf6 in homozygous knockout mice causes abnormalities in joint, ligament and cartilage formation, and variable ocular phenotypes [23, 14]. Finally, a small eye phenotype in the zebrafish mutant dark half s327 was attributed to a nonsense mutation in the gdf6a gene . The authors of this work show that gdf6a establishes dorsal-ventral positional information in the retina and controls the formation of the retinotectal map.
Given conflicting results of studies in different vertebrate model systems, the role of GDF6 in eye development merits further investigation. We performed analysis of the gdf6a gene in the zebrafish out m233 mutant, characterized by a severe reduction of eye size [25, 26]. We found that the out m233 mutation eliminates the initiation codon in the gdf6a gene, disrupting the function of this gene. Despite the reduction of eye size, retinal lamination is normal in most mutant animals, and the optic nerve appears largely unaffected. The small eye size phenotype in out m233 is associated with a wave of apoptosis during early stages of development. Other than eye size reduction, no obvious defects are observed in the external appearance of out m233 mutants, including its craniofacial skeleton. out m233 mutant homozygotes frequently survive to adulthood and display a variable reduction of size. Other than eye defects, we have not observed obvious abnormalities in the appearance of swimming behavior of out m233 mutant homozygotes. To investigate the role of GDF6 in human eye defects, we screened the GDF6 gene in 200 patients suffering from microphthalmia, anophthalmia or related abmormalities, and detected amino acid substitutions in four of them.
To confirm that its defects are indeed responsible for the out m233 mutant phenotype, we overexpressed the wild-type gdf6a in the progeny of crosses between out m233 heterozygotes. This was accomplished by injecting into embryos a DNA construct containing gdf6a under control of a heat-shock promoter. A mutant gdf6a gene that contains the p.Met1Val substitution was used in these experiments as a negative control. The overexpression of the wild-type gdf6a produced a significant decrease (p = 0.003, chi square) in the frequency of phenotypically mutant animals from ca. 20% (34/171) to 8% (13/154). At the same time, we observed that ca. 9% (14/155) of animals featured one small and one normal size eye. This phenotype was entirely absent in animals treated with the control mutant construct (0/171). We interpret the one eye phenotype as a partial rescue of out m233 eye size defect.
To further test whether the p.Met1Val substitution affects gene function, we overexpressed wild-type and mutant Gdf6 in zebrafish using a heat-inducible promoter. In agreement with previous studies , the overexpression of the wild-type Gdf6 construct frequently results in ventral eye defects. In contrast to that, the overexpression of the mutant gene produces ventral eye abnormalities with a much lower frequency (Figure 2D-F). These data are in agreement with the results of out m233 rescue experiments, and indicate that the p.Met1Val substitution affects gene function. Based on these results, we conclude that mutation in the gdf6a gene is responsible for small eye phenotype in the out m233 mutant strain.
A cohort of 200 unrelated probands with coloboma, microphthalmia and/or anophthalmia was screened for mutations in the GDF6 gene. Three heterozygous synonymous changes were identified in 11 patients (c.93C > T/p.Ser31Ser, c.852C > G/p.Ser284Ser, and c.936G > C/p.Ser312Ser). A heterozygous transition (c.955G > A) leading to an amino acid substitution (p.Ala319Thr) was identified in a patient (ID178) with isolated unilateral microphthalmia. This variant was not detected on 362 control chromosomes.
In this study, we characterized the zebrafish out m233 mutant strain and identified a mutation in the gdf6a gene affecting the translation start site (p.Met1Val). The size of the retina and the lens are grossly reduced in gdf6a/out m233 mutants, but the overall retinal architecture appears largely intact. This is in contrast to other small eye size phenotypes in zebrafish, which often exhibit retinal defects in lamination or retinal degeneration [40, 26, 41]. Despite eye size reduction, all cell classes that we assayed for are present in out m233 mutants, although some display variable abnormalities. In ~50% of out m233 embryos we observed partial absence of the photoreceptor cell layer in the dorsal and/or ventral regions of the retina. In the regions of photoreceptor absence, we observed a displacement of amacrine cells towards and occasionally into the photoreceptor cell layer. Another interesting defect is present in GABA-positive neurons. These cells are present in abnormally large numbers at the retinal margin at 6 dpf. As neurogenesis continues at the margin of the retina throughout the lifetime of the organism and the quantity of GABA-positive cells decreased in the central retina, this phenotype is most likely transient. As postmitotic GABA-positive neurons near the marginal zone become older, their excess is most probably eliminated by apoptosis.
Our studies suggest that a major contributing factor to the small eye and lens size is deregulated apoptosis. Apoptosis is a normal event during the development of the zebrafish retina, normally occurring around 24 to 25 hpf . The increased apoptosis in the out m233 mutant is evident at 26 hpf and persists until at least 48 hpf, suggesting that retinal precursor cells which normally contribute to the neural retina and lens are being eliminated through apoptosis, which results in a reduction of cell numbers in all retinal layers by 72 hpf.
While this manuscript was in preparation, the small eye size in another zebrafish mutant (dark half s327 ), was attributed to a nonsense mutation in the gdf6a gene . The authors of this work show that gdf6a is necessary to induce dorsal fate in the retina. Loss of gdf6a prevents specification of the retinal ganglion cells with dorsal identity and prevents innervation of the ventral tectum. Gdf6a activates the expression of known dorsal markers (bmp4, tbx5, tbx2b, efnb2) and represses the expression of ventral fate determinant, vax2. Similar to our observations in the out m233 mutant, the retina of the dark half s327 homozygotes features a transient increase in cell death . Lamination defects have not been reported.
Previous studies have examined the effect of GDF6 loss of function on eye development by morpholino antisense oligonucleotide (MO) knockdown in Xenopus and zebrafish [13, 22]. A striking reduction of eye size was observed in Xenopus embryos . In zebrafish, MO injection resulted in highly variable ocular anomalies including ventral colobomata, persistent dorsal-retinal groove, microphthalmia, and even anophthalmia . The retina was disorganized after MO knockdown in Xenopus and zebrafish, and immunolabeling with anti-XAP-1 and anti-islet-1 antibodies in Xenopus did not detect any photoreceptor, ganglion or amacrine cells [13, 22]. This is in contrast to our findings in the out m233 mutant, where retinal lamination is largely normal and all cell classes are present.
MO knockdown of gdf6a in zebrafish recently identified the presence of skeletal anomalies (curled or kinked tails), although their prevalence was much lower than that of ocular anomalies . Axial abnormalities were also seen in Xenopus tadpoles following morpholino injection . In these studies, MO knockdown of gdf6a did not always lead to a mutant phenotype, as significantly reduced levels of correctly spliced gdf6a mRNA were identified in phenotypically unaffected morphant embryo pools . In contrast to these findings, other than eye size reduction, no obvious defects are observed in the external appearance of out m233 mutants, including its craniofacial skeleton. The external morphology of dark half s327 mutants was also inconspicuous, with the exception of smaller eyes . Similarly, we also did not observe any defects in circulation and axial vasculature in the out m233 mutant, although these were reported after MO-knockdown of gdf6a in zebrafish [13, 42, 43]. The differences observed between MO-injected embryos and the out m233 and dark half s327 mutant strains might be attributed to a toxic effect of the MOs or to non-specific interference of anti-gdf6a morpholinos with other genes. Non-target related phenotypes including neuronal cell death have been reported for 18% of MOs, even when relatively low MO concentrations are used . Mistargeting by MOs could be caused by the simultaneous inactivation of an essential gene with serendipitous homology, or perhaps by some unexpected chemical contaminants found in a small fraction of MO syntheses . Such nonspecific effects of morpholino knockdown experiments can be usually excluded via rescue experiments [45, 46]. In the case of gdf6, rescue experiments are difficult to perform as the overexpression of this gene results in the ventralization of the entire embryo [22, 21]. An alternative explanation for the discrepancy could be that the out m233 mutant may be a hypomorph that retains some residual gdf6a function. This, however, seems unlikely since the out m233 mutation disrupts the translation start site, which will either abolish protein synthesis (if the ATG at position c.5A is used) or produce a protein lacking the signal peptide (if the ATG at position c.233A is used). Such a truncated protein is unlikely to be transported to the endoplasmic reticulum and Golgi for posttranslational modifications, which include two proteolytic cleavage steps. Another defect in the zebrafish gdf6a gene was previously reported to result in a short body axis, a reduction of head structures, and death by 48 hpf . That mutant, however, carries a large deletion of many genes, which could explain the more severe phenotype.
In a previous study, a segmental deletion encompassing the GDF6 gene was identified in a patient with bilateral chorioretinal colobomata . The patient exhibited multiple developmental defects, including neurodevelopmental impairment, bilateral soft-tissue syndactyly of the toes, and an atrial septal defect. Besides GDF6, 30 other genes are also lost in the deletion, which could contribute to the range of developmental defects seen in this patient.
Mutations in GDF6 have also been recently reported in patients with Klippel-Feil syndrome, a complex skeletal disorder characterized by congenital fusion of vertebrae within the cervical spine . A chromosomal inversion segregated with the disease in a 5-generation family exhibiting an autosomal dominant inheritance pattern. Besides fusion of the vertebral bodies and laminae, the affected family members had scoliosis of the thoracic and lumbar spine, fusion of the carpal and tarsal joints, and restricted elbow movement. No DNA loss was evident at the inversion boundaries, and no coding genes appeared to be disrupted. The proximal inversion breakpoint was located 623 kb 3'of GDF6. The authors suggest that the disease is associated with GDF6, perhaps by disruption of long-range enhancer elements, since the skeletal anomalies in this family correspond with the phenotype observed in Gdf6 knockout mice . In these mice, defects in joint, ligament, and cartilage formation were seen in the wrist, the ankle, the middle ear, and the coronal suture between bones in the skull . Analysis of GDF6 in 127 individuals with Klippel-Feil syndrome identified two missense variants, p.Ala249Glu and p.Leu289Pro . No ocular defects were described in these patients . The two variants were not identified on 708 control chromosomes .
In a recent study, 489 patients with ocular anomalies and 81 patients with vertebral segmentation defects were screened for GDF6 mutations . The authors identified four amino acid substitutions associated with ocular phenotypes, two with a skeletal phenotype, and one alteration (p.Ala249Glu) was identified in 3 probands who either had ocular or skeletal phenotypes. These variants were not detected on at least 366 control chromosomes. For three variants, segregation analysis showed the presence of the alteration in an unaffected parent indicating incomplete penetrance. The effect of two amino acid substitutions, p.Ala249Glu and p.Lys424Arg, were evaluated with SOX9 reporter gene assays and Western blot analysis. Mutant GDF6 was less effective in activating SOX9 reporter gene activity compared to wild-type GDF6. By Western blot analysis it was shown that the level of secreted mature GDF6 was reduced with p.Ala249Glu and p.Lys424Arg . Screening of 50 patients with ocular anomalies recently identified GDF6 mutations in four (8%) patients. Three of these patients carried the p.Ala249Glu variant.
In this study, we identified two different amino acid substitutions, p.Ala249Glu and p.Ala319Thr, in four (2%) of 200 patients with ocular malformations. Although both variants are conserved among several species, neither is conserved in zebrafish gdf6a. Therefore it was not possible to perform rescue experiments in the out m233 mutant to determine the pathogenic potential of these variants. Interestingly, the p.Ala249Glu variant has been associated both with Klippel-Feil syndrome and with ocular anomalies [21, 14, 15]. The variant was detected on 8 of a total of 1478 chromosomes of patients with ocular malformations [13–15], and not on a total of 1074 control chromosomes [14, 21] (Fisher's exact test, p-value 0.01), strongly suggesting that it is pathogenic. Parents carrying this variant are asymptomatic, suggesting that GDF6 sequence alterations are likely to contribute to the phenotype, but are not the sole cause of the disease. This is confirmed by the fact that the affected sib of one of the patients in our study does not carry the p.Ala249Glu variant, indicating that other factors contribute to the disease. Besides microphthalmia, one patient in our study carrying the p.Ala249Glu variant has malformed ossicles. Interestingly, middle ear defects were also observed in Gdf6 -/- mice, suggesting that the p.Ala249Glu variant could contribute to the ocular and middle ear defects in this patient. Finally, another microphthalmic patient in our study who carries the p.Ala249Glu variant has several skeletal anomalies, including cleft palate, hemivertebrae and talipes. Microphthalmia in combination with cleft lip and palate have recently been described in another patient carrying the p.Ala249Glu variant , suggesting that the extra-ocular defects might also be attributed to the p.Ala249Glu variant.
This study confirms the essential role of GDF6 in the regulation of vertebrate eye size. The reduced eye size in the zebrafish gdf6a/out m233 mutant is likely to be caused by a transient wave of apoptosis at the onset of neurogenesis. Amino acid substitutions in GDF6 were detected in four patients with microphthalmia. In two patients different skeletal defects were observed, suggesting pleitrophic effects of p.Ala249Glu. Variable expressivity and penetrance suggest a complex non-Mendelian inheritance pattern where other genetic factors may influence the outcome of the phenotype.
The maintenance and breeding of zebrafish strains and staging of embryonic development were performed as described previously [49, 50]. All animal protocols were approved by the MEEI and Tufts University animal care committees. Embryos were observed using an Axioscope microscope (Zeiss, Thornwood, NY) or a Leica (Deerfield, IL) MZ12 dissecting microscope. Images were recorded with digital cameras, and processed using Adobe Photoshop software. The out m233 mutant strain was originally recovered in a large-scale N-ethyl-N-nitrosurea (ENU) mutagenesis screen [25, 26]. To visualize ganglion cells, we used the Tg(pou4f3:gap43-GFP) trangenic line .
Zebrafish larvae were raised to the desired age and fixed in 4% paraformaldehyde (PFA) in PBST overnight at 4°C. Embedding, sectioning, and staining were performed as described previously [51, 52]. Sections were examined using an Axioscope Microscope (Zeiss) and images were recorded using an AxioCam digital camera (Zeiss).
A map cross was set up between heterozygous G0 carriers of the out m233 allele (AB genetic background) and wild-type WIK strain homozygotes. To determine the position of the out locus in the genome, we used a panel of 96 F2 diploid embryos obtained via incrossing of F1 animals.
Bulk segregant mapping analysis was performed by the Mutant Mapping Facility at the University of Louisville on DNA isolated from the mutant embryos http://www.biochemistry.louisville.edu/zfmapping/index.html. Positional candidate genes were selected based on the zebrafish genome sequence (Ensembl Database, Sanger Center UK, and UCSC Genome Browser, University of California at Santa Cruz). Search for mutations was performed by RT-PCR of RNA isolated from wild-type and out m233 embryos, and by PCR on genomic DNA. PCR products were purified using the QIAquick PCR purification kit (Qiagen) and analyzed by direct sequencing. The effect of the mutation, which affects the translation start site, was evaluated with NetStart 1.0 .
The wild-type gdf6a gene was previously cloned into the pNG6 vector under the control of a heat-shock promoter . The p.Met1Val mutation was introduced by site-directed mutagenesis of the wild-type gdf6a construct with phusion DNA polymerase (Finnzymes, Espoo, Finland). The wild-type and mutant constructs were injected into zebrafish embryos at the one-cell stage using standard approaches . Expression was activated via a 1 hour heat shock at the 12 somite stage. The eye phenotype was evaluated at 3 and 5 dpf.
To count cells in retinal layers, wild-type and mutant embryos were fixed in 4% PFA at 36 and 96 hpf, and embedded in JB-4 resin (Polysciences), according to the manufacturer's instructions. Subsequently, 4-μm sections were collected on microscope slides, immersed in Hoechst 33258 solution (Molecular Probes; 1 mg/ml in PBS) for 10 min, and washed in PBS for 1 h. Alternatively, YoPro (Molecular Probes) was diluted 1:5,000 in PBST and used to stain frozen sections for 10 min. Sections were analyzed using UV illumination.
Two methods were used to evaluate apoptotic cell death: acridine orange staining and TUNEL. To apply the first method, embryos were dechorionated, and placed in 5 mg/ml acridine orange (acridinium chloride hemi-zinc chloride; Sigma) in E3 medium  for 30 min. Embryos were then washed in E3 medium and viewed using UV illumination on an upright microscope. For TUNEL detection of cell death, zebrafish larvae were fixed in 4% PFA, cryoprotected in sucrose, and cryosectioned at 14 μm. Following two washes, 10 min each, in 50 mM PBS (pH 7.4), sections were treated with Proteinase K (Roche Applied Sciences, Indianapolis, IN) for 10 min at the concentration of 2 μg/ml in 50 mM Tris/HCl buffer, pH 8. Subsequently, sections were rinsed 3 times in 50 mM Tris/HCl buffer, incubated with 70% ethanol/30% acetic acid solution at -20°, washed 5 times, 5 min each, in 50 mM Tris/HCl buffer, and treated with blocking solution, containing (v/v in PBST) 10% normal goat serum, and 0.5% Triton X-100 (30 min, room temperature). Finally, sections were incubated with TUNEL reaction mixture (Roche) at 37°for several hours according to manufacturer's instructions. To detect apoptotic nuclei, sections were rinsed in 50 mM PBS for 10 min, coverslipped, and analyzed on a Leica SP2 confocal microscope equipped with a 40x lens. Digital images were processed with Adobe Photoshop Software and used to obtain counts of apoptotic cells.
Antibody staining was performed on frozen sections as described in previous publications [51, 52, 29]. The following primary antibodies and dilutions were used: rabbit anti-phospho-H3 histone (1:200, Upstate Biotechnology); mouse monoclonal anti-neurolin (1:25, Zebrafish International Resource Center); rabbit polyclonal anti-carbonic anhydrase (1:250, gift of Paul Linser, Whitney Laboratory, St. Augustine, FL); mouse monoclonal anti-parvalbumin (1:250, Chemicon); mouse monoclonal anti-HuC/HuD (1:20, Molecular Probes); mouse monoclonal Zpr1 (1:250, Zebrafish International Resource Center); rabbit polyclonal anti-serotonin (1:250, Sigma); rabbit polyclonal anti-tyrosine hydroxylase (1:250, Chemicon); rabbit polyclonal anti-GABA (1:500, Sigma); and rabbit polyclonal anti-PAX6 (1:200, Covance). The anti-serotonin and anti-Pax6 antibodies required citrate treatment prior to blocking in normal goat serum . After staining, sections were imaged using Leica SP2 confocal microscope equipped with a 40x lens. Digital images were processed with Adobe Photoshop Software.
Informed consent was obtained from all participants in this study in accordance with the local Ethic Boards in the respective institutions where they were examined. Blood samples were collected, and genomic DNA was isolated by standard protocols. The coding region and splice junctions of the two exons of GDF6 were screened in 200 patients with coloboma, microphthalmia or anophthalmia by automated sequencing . PCR products were treated with ExoSAP-IT (USB) or purified with 96-wells NucleoFast purification plates (Machery-Nagel) before automated sequencing. All mutations and polymorphisms were confirmed by a second round of PCR amplification.
The authors are grateful to Dr. Zac Pujic for help with the initial characterization of the out mutant phenotype, to Krysta Voesenek for assisting the cloning experiments, to Dr. Paul Linser for providing the anti-carbonic anhydrase antibody, and to Dr. Herwig Baier for providing the pNG6-gdf6a expression construct. This work was supported by NEI R21 award EY018427 to JM. AIdH was supported by the Netherlands Organisation for Scientific Research (916.56.160) and a Ter Meulen Fund Stipend of the Royal Dutch Academy of Arts and Sciences. EIT is supported by an unrestricted grant from Research to Prevent Blindness. The Anophthalmia Registry at Albert Einstein Medical Center is funded by the Albert B. Millett Memorial Fund, A Mellon Mid-Atlantic Charitable Trust, Rae S. Uber Trust, A Mellon Mid-Atlantic Charitable Trust and Gustavus and Louis Pfeiffer Research Foundation. NR is supported by an Academy of Medical Sciences/The Health Foundation Senior Surgical Scientist Fellowship and by grants from VICTA (Visually Impaired Children Taking Action) and the late Ernest Henri Polak Trust.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.