|Home | About | Journals | Submit | Contact Us | Français|
In humans, rare non-synonymous variants in the planar cell polarity gene VANGL1 are associated with neural tube defects (NTDs). These variants were hypothesized to be pathogenic based mainly on genetic studies in a large cohort of NTD patients. In this study, we validate the potential pathogenic effect of these mutations in vivo by investigating their effect on convergent extension in zebrafish. Knocking down the expression of tri, the ortholog of Vangl2, using an antisense morpholino (MO), as shown previously, led to a defective convergent extension (CE) manifested by a shortened body axis and widened somites. Co-injection of the human VANGL1 with the tri-MO was able to partially rescue the tri-MO induced phenotype in zebrafish. In contrast, co-injection of two human VANGL1 variants, p.Val239Ile and p.Met328Thr, failed to rescue this phenotype. We next carried out overexpression studies where we measured the ability of the human VANGL1 alleles to induce a CE phenotype when injected at high doses in zebrafish embryos. While overexpressing the wild-type allele led to a severely defective CE, overexpression of either p.Val239Ile or p.Met328Thr variant failed to do so. Results from both tri-MO knockdown/rescue results and overexpression assays suggest that these two variants most likely represent “loss-of-function” alleles that affect protein function during embryonic development. Our study demonstrates a high degree of functional conservation of VANGL genes across evolution and provides a model system for studying potential variants identified in human NTDs.
Neurulation is a fundamental embryonic process that leads to the formation of the neural tube, the precursor of the brain and spinal cord. This process is dynamic and involves a series of complex and well-coordinated morphogenetic events controlled by distinct molecular pathways (Bassuk and Kibar, 2009). Failure of fusion of the neural tube early during embryogenesis leads to neural tube defects (NTDs) that represent the most common structural malformations of the central nervous system in humans, affecting 1-2 infants per 1000 births (Bassuk and Kibar, 2009). The most common forms of NTDs include anencephaly and myelomeningocele (spina bifida), which result from the failure of fusion in the cranial and spinal regions of the neural tube, respectively. NTDs in humans are considered to be multifactorial with multiple genes and environmental factors interacting to modulate their incidence and phenotype severity (Bassuk and Kibar, 2009). Despite a long history of etiological studies, the underlying molecular and cellular pathogenic mechanisms causing NTDs remain largely unknown.
Recent years have witnessed a breakthrough in elucidating the role of planar cell polarity pathway (PCP) in neurulation and how molecular lesions in this pathway lead to neural tube defects in animal models and humans (Simons and Mlodzik, 2008; Bassuk and Kibar, 2009). PCP, also called tissue polarity, is the process by which epithelial and mesenchymal cells become polarized within the plane of a tissue. This process has been well studied in the adult wing hairs, abdominal bristles and ommatidia (eye units) in Drosophila. Genetic studies of a wide range of mutants affecting these highly organized structures in the fly have identified a group of so called “core” PCP genes which include Van gogh/Strabismus (Vang/Stbm), Frizzled (Fz), Dishevelled (Dvl), Flamingo (Fmi), Prickle (Pk) and Diego (Dg). Genetic and molecular studies of a wide range of mutants and knockdowns of orthologs of fly PCP genes in zebrafish, frog and mouse models have demonstrated a high level of conservation of this pathway in vertebrates where it has been implicated in controlling the morphogenetic process of convergent extension (CE) during gastrulation and neurulation through the non-canonical Wnt/Frizzled pathway. CE describes the narrowing and lengthening of a tissue that could be the embryonic axis during gastrulation or the neural plate during neural tube closure (Simons and Mlodzik, 2008).
In mouse, Looptail (Lp) was the first mutant to implicate a role of PCP and CE in neural tube defects (Kibar et al., 2001; Murdoch et al., 2001). Lp is a semi-dominant mutation that produces a characteristic “looped” tail in heterozygotes and a severe NTD called craniorachischisis in homozygotes where the neural tube remains open throughout the spinal cord (Strong and Hollander, 1949). We and others identified the PCP gene Vangl2 as the gene defective in Lp (Kibar et al., 2001; Murdoch et al., 2001). Vangl2 encodes a membrane protein whose predicted features include four transmembrane domains and a PDZ-domain binding motif at the carboxy terminus involved in protein-protein interaction (Kibar et al., 2001). Vangl2 has another homolog in vertebrates, Vangl1, that shares ~68% identity and exhibits an overall similar structure (Torban et al., 2004). Vangl1 shows a dynamic expression in the developing neural tube and genetically interacts with Vangl2 (Torban et al., 2008). We have previously identified in VANGL1 three rare missense mutations in familial (p.Val239Ile, p.Arg274Gln) and sporadic (p.Met328Thr) cases of NTDs, including a de novo mutation (p.Val239Ile) appearing in a familial setting (Kibar et al., 2007). These mutations were not detected in 171 control individuals and affect highly conserved residues in the Vangl family. Using the yeast-two-hybrid system as a measure of the interaction of VANGL1 with Dvl1, Dvl2 and Dvl3, we also demonstrated that the p.Val239Ile mutation abrogated interaction of VANGL1 with all 3 Dvl proteins (Kibar et al., 2007). This suggests that this mutation and others may be pathogenic, though these have not been tested in an in vivo animal model and thus remain unproven and their mechanism of action undetermined.
To fully evaluate the significance of the VANGL1 alleles associated with NTDs, it is critical to determine the extent to which their biological activities are affected in vivo. The zebrafish represents a well-suited model for the study of the molecular and cellular mechanisms mediating gastrulation and neural tube closure in vertebrates. Mutations in PCP genes, particularly the trilobite gene (tri, ortholog of VANGL2), or knockdown of the gene product with antisense oligonucleotides (MO) or overexpression of the gene, cause CE defects including marked shortening of the anteroposterior axis (Jessen et al., 2002; Park and Moon, 2002; Wada et al., 2005; Carreira-Barbosa et al., 2009). The CE phenotype represents a well-established readout for defects in this pathway. Importantly, injection of zebrafish Vangl1 mRNA partially suppresses the CE defect in tri/Vangl2 mutant embryos, suggesting similar biochemical activities for both genes during development (Jessen and Solnica-Krezel, 2004). In this study, we test the evolutionary conservation of the human VANGL genes as key regulators of the process of CE and study the effect of three VANGL1 missense mutations we previously identified in NTD patients (p.Val239Ile, p.Arg274Gln and p.Met328Thr) on this process in zebrafish.
In zebrafish, tri/vangl2 is expressed maternally and zygotically with a peak expression in the neurula stage; in contrast, Vangl1 is not maternally expressed in blastula stage, and is first detected at the 15-somite stage after gastrulation has completed and neurulation has already begun (Park and Moon, 2002; Jessen and Solnica-Krezel, 2004). Since we are assaying for defects in the CE process that occurs during gastrulation and neurulation very early during development, and since it has been shown that Vangl1 can function similarly to Vangl2, we chose to use the well-established tri/Vangl2-MO knockdown model in our study.
In zebrafish embryos, knockdown of the tri gene product with antisense MO causes a well described CE defect depicted by marked shortening of the anterior–posterior axis, a phenotype almost as severe as the tri mutant (Jessen et al., 2002; Park and Moon, 2002). Indeed, when embryos were injected with tri-MO and raised until 2 days post fertilization, a severe (42%) decrease in length of the anterior–posterior axis in tri-MO fish was observed (P < 0.05) (Fig. 1A and B, and Table 1). It has been reported that injection of zebrafish Vangl1 mRNA partially suppresses the CE defect in tri/Vangl2 mutant embryos (Jessen and Solnica-Krezel, 2004). To test for evolutionary conservation of the VANGL genes, we determined whether or not human VANGL1 could also suppress this defect. When co-injecting embryos with tri-MO and human VANGL1 RNA, the average body length of tri-MO/VANGL1 fish was significantly increased (P < 0.05) (Fig. 1C and Table 1). Higher doses of human VANGL1 RNA (50–150 pg) were needed to partially rescue the tri-MO-induced CE defect with the same efficiency as zebrafish Vangl1 RNA (10–25 pg) (Supplementary Fig. S1). Nevertheless, these results indicate that the human VANGL1 protein can interact with zebrafish PCP proteins and function as the zebrafish Vangl1 to compensate for the reduced length associated with loss of tri function. These data demonstrate evolutionary conserved and similar biochemical functions of VANGL1 and VANGL2 during embryogenesis. For the following assays in our study, we chose to use the human rather the zebrafish cDNA because it represented a more stringent test of function of the human mutations, where we wanted to analyze the mutations in the context of the complete natural human backbone onto which they emerged in patients.
For initial evaluation of the zebrafish tri-MO model system, we tested the two known Lp-associated mutants, p.Asp255Glu and p.Ser464Asn, for their ability to rescue the tri-MO phenotype. These two mutations were identified in two independent alleles of Lp, Lp and Lpm1Jus, where they produce identical phenotypes of a looped-tail appearance in heterozygous mice and severe craniorachischisis in homozygous embryos (Kibar et al., 2001). Importantly, the variants p.Asp255Glu and p.Ser464Asn were proven to be pathogenic based on genetic data where they were absent in 36 inbred strains including parental strains and on functional data where they were shown to impair interaction with members of the Dishevelled family (Kibar et al., 2001; Torban et al., 2004). Both variants affect highly conserved residues across evolution where Asp255 is absolutely invariant across all available Stbm/Vang sequences while Ser464 is conservatively substituted to cysteine in the fly and worm homologs (Torban et al., 2004). These two mouse mutations were introduced into a human VANGL1 ORF and each of the resulting transcripts was co-injected with tri-MO. Co-injection of tri-MO with either VANGL1D259E or VANGL1S467N failed to rescue the CE defect (Fig. 1D and E, and Table 1). These results strengthen the evidence that these two mutations are indeed pathogenic and prove the effectiveness of the zebrafish tri-MO model system for evaluation of the potential pathogenic effect of mutations identified in mammals.
To test the in vivo activity of the unique VANGL1 variants identified in NTD patients, VANGL1M328T, VANGL1M328T and VANGL1R274Q, we co-injected tri-MO with RNA transcripts from each one of these mutants. These variants affect highly conserved residues in the Vangl family where Val239 is absolutely conserved across evolution, Arg274 is invariant in all known orthologs except that in worm Stbm, where it is replaced by glutamate; and Met328 is moderately conserved where it is replaced by valine in human, mouse and frog Vangl2 and fly Stbm, by lysine in zebrafish Vangl1 and by leucine in the worm Stbm (Kibar et al., 2007, data not shown). Co-injection of VANGL1M328T or VANGL1M328T failed to rescue the CE phenotype (Fig. 2D and E), whereas injection of VANGL1R274Q RNA with the tri-MO mimicked the effects of VANGL1 when compared to tri-MO injected fish (P < 0.05) (Fig. 2B and F). The body length measurements and the number of embryos analyzed are depicted in Table 1.
Since the tri-MO has been reported to cause an increase in necrosis (Park and Moon, 2002), we sought to determine if the reduced body length phenotype was caused by increased cell death. We labeled zebrafish treated with tri-MO with acridine orange and we observed a slight increase in the levels of cell death in these morphants, whether or not the body length was normal, with the bulk of apoptotic cells being in the yolk, as in untreated larvae (Supplementary Fig. S2). Thus, it is very unlikely that the decrease in axial length was due to a massive increase in cell death.
The severity of the extension defects was also assessed by measuring the body angle formed by the end of the head and the end of the tail in 10–12 somites embryos (Table 1 and Fig. 2G–L). Upon tri-MO injection, the body angle increased significantly (P < 0.001), indicating that extension movements following neural tube closure were impaired in tri morphants (Fig. 2G and H). This phenotype was partially rescued upon injection of VANGL1 RNA (P < 0.05) (Fig. 2I) and VANGL1R274Q RNA (P < 0.05) (Fig. 2L) when compared to tri-MO embryos. This was not the case when co-injecting tri-MO with either VANGL1M328T RNA (Fig. 2J) or VANGL1M328T RNA (Fig. 2K) where the body angle was not significantly different from tri-MO treated embryos.
Neural tube closure in the zebrafish embryo is paralleled by convergent extension of the somites. In wild-type embryos at the 8–14 somites stage (Fig. 2M), the somites had converged and extended along the closed neural tube, as revealed by whole-mount in situ hybridization (ISH) using the MyoD probe. In tri-MO injected embryos (Fig. 2N), the somites were wider laterally indicating a perturbed convergent extension. The defective CE of somites was partially rescued by co-injection of tri-MO with VANGL1 (Fig. 2O) or VANGL1R274Q RNA (Fig. 2R), but not by VANGL1M328T (Fig. 2P) or VANGL1M328T RNA (Fig. 2Q).
Gain-of-function studies have shown that an increase in the activity of PCP genes also impairs CE in zebrafish and frogs and more specifically that overexpressing tri in zebrafish embryos causes a reduction in axial length similar to that obtained with the tri-MO (Park and Moon, 2002; Carreira-Barbosa et al., 2003; Takeuchi et al., 2003). To substantiate our knock-down/rescue data and better understand the mode of action of the three human VANGL1 variants, we carried out overexpression studies where we assessed the ability of these variants to induce CE when injected at high doses in zebrafish embryos. We first sought to determine whether overexpression of the wild-type human VANGL1 induces a CE defect similarly to zebrafish Vangl2. Indeed, when VANGL1 was overexpressed, a severe reduction in axial length (P < 0.001) was observed (Fig. 3A and B), further confirming that the human VANGL1 protein was fully functional in zebrafish and supporting the evolutionary conservation of functionality of the VANGL genes in vertebrates. Overexpressing the two mouse variants, VANGL1D259E or VANGL1S467N, failed to induce a CE defect, as compared to wild-type VANGL1 (P < 0.001) (Fig. 3C and D) supporting the hypothesis that they act as loss-of-function alleles within a dosage-sensitive signaling pathway. When overexpressing the 3 human VANGL1 variants, VANGL1M328T or VANGL1M328T failed to induce CE defects (P < 0.001, P < 0.01, respectively) (Fig. 3E and F), while VANGL1R274Q behaved as the wild-type allele (P > 0.05) (Fig. 3B and G). Whole-mount ISH analysis of somites with MyoD generated similar results, where overexpression of either wild-type VANGL1 (Fig. 3I) or VANGL1R274Q (Fig. 3L) caused a perturbed CE of the somites while overexpression of either VANGL1M328T (Fig. 3J) or VANGL1M328T (Fig. 3K) did not affect this phenotype. The body length measurements and the number of embryos analyzed are depicted in Table 2.
All of our observations in zebrafish are consistent with a failure of the human VANGL1M328T and VANGL1M328T mutations, but not the VANGL1R274Q mutation, to rescue the tri-MO phenotype or to induce a CE defect upon overexpression. These data indicate that the human RNA containing the p.Val239Ile or the p.Met328Thr mutation fails to function as well as wild-type RNA, suggesting that these two mutations impair the protein function during embryogenesis. The variant p.Val239Ile is a de novo mutation that occurred in a familial setting, strongly favoring the hypothesis of a major genetic contribution to the development of NTD in this family. As previously demonstrated, p.Val239Ile most likely affects the function of VANGL1 by inhibiting its interaction with Dishevelled, an aspect that is absolutely essential for normal PCP signaling (Kibar et al., 2007). The variant p.Met328Thr occurred in a sporadic case and did not affect the interaction of VANGL1 with the three Dishevelled proteins (Kibar et al., 2007). Additional biochemical tests addressing other functional aspects of PCP signaling are needed to investigate its hypothesized pathogenic effect. On the other hand, the variant p.Arg274Gln mutant that occurred in a familial case of NTDs did not affect the interaction of VANGL1 with the Dishevelled proteins and behaved similarly to the wild-type RNA in the tri-MO zebrafish rescue and overexpression experiments. This suggests that p.Arg274Gln is either not pathogenic or that it represents a weaker hypomorphic allele with enough levels of protein function required for normal PCP signaling in zebrafish but that the p.Arg274Gln variant must interact with other major genetic or environmental factors to cause the NTD phenotype in humans. Another interpretation would be assay-based where injection of high doses of the mutant VANGL1R274Q in zebrafish would reach an above-threshold dose required for normal protein function, while this is not the case in human patients. Of course, all our conclusions in this study are based on cooperative interactions between the exogenous human protein and its zebrafish co-partners, which forms the basis of both knockdown/rescue and overexpression assays.
It is important to note that we are not using the zebrafish model to mimic the pathophysiology of NTDs in humans. Instead, we are aiming at defining initial functional properties of putative mutations in the VANGL1 gene and investigating the biological process they affect. We show that they affect the process of CE in zebrafish leading to the extrapolation that they most likely affect this highly conserved process in humans, that when interacting with other environmental and genetic factors leads to failure of neural tube closure and causes NTDs. Both mutations, p.Val239Ile and p.Met328Thr, that were shown to affect the protein function, will be prioritized for further functional analysis using cell-based systems or transgenic mouse studies.
Our data demonstrate a high level of conservation of the VANGL genes between zebrafish, mice and humans and give considerable incentive to use the zebrafish model for functional validation of potential NTD mutations identified in the VANGL genes. Other members of the PCP pathway represent excellent candidates for involvement in NTDs and this model would also be important for validating the pathogenicity of mutations yet to-be-identified in these genes, particularly those that were shown to modulate CE in zebrafish, including Prickle1, Diversin, CELSR1 and Scribble 1 (Veeman et al., 2003; Moeller et al., 2006; Wada et al., 2005; Carreira-Barbosa et al., 2009). Functional validation of PCP mutants represents an essential step for dissecting the complexity of the human NTD phenotype and investigating the underlying molecular and cellular pathogenic mechanisms.
Our study demonstrates similar biochemical activities for Vangl1 and Vangl2 during development. Little is known about the role of Vangl1 during development; however several lines of evidence support its involvement in convergent extension. During zebrafish development, Vangl1 can function similarly to tri/Vangl2 and can partially suppress the tri gastrulation phenotype. In our study, we confirm these rescue results with human VANGL1 and we demonstrate that overexpression of VANGL1 in zebrafish causes a severe CE defect similar to that obtained with Vangl2. In mouse, Vangl1 knockouts show mild PCP defects in the inner hair cells of the cochlea and, importantly, Vangl1 genetically interacts with Vangl2 causing a severe NTD in compound heterozygotes. Mouse Vangl1, like Vangl2, physically interacts with Dvl1, Dvl2 and Dvl3, and the Lp Vangl2 mutations, p.Asp255Glu and p.Ser464Asn, introduced in either Vangl1 or Vangl2 cDNA, abrogated this interaction. In humans, pathogenic mutations in VANGL1 have been identified in NTDs, and in this study, we show that two of these mutations, p.Val239Ile and p.Met328Thr, affect convergent extension in zebrafish and we hypothesize that they most likely affect a similar process in humans. No mutations in VANGL2 in human NTDs have been identified yet; however, molecular genetic studies of this gene in a larger cohort and/or detailed analysis of its regulatory regions in human patients are needed before drawing final conclusions on its role in the pathogenesis of human NTDs. To further dissect the roles of both Vangl1 and Vangl2 genes during development, additional studies such as mouse conditional knockouts are needed.
The open reading frame (ORF) of VANGL1 (Accession No. NM_138959) was amplified from total human RNA by RT-PCR and the resulting fragment was cloned into pCS2+ using EcoRI and XbaI sites built in the PCR primers. A consensus Kozak sequence, GCCACC, was added to the ORF to enhance levels of protein expression. The two mouse Lp-associated mutations, p.Asp255Glu and p.Ser464Asn, and the 3 human mutations, p.Val239Ile, p.Arg274Gln and p.Met328Thr, were introduced in the human VANGL1 ORF cloned into pCS2+. The mutagenic primers used for VANGL1 mutations, p.Val239Ile and p.Met328Thr, were generated using the following mutagenic primers: 5′-CTGCCTGAGCTCCAGCAGGATGAT-3′ (p.Val239Ile) and 5′-GCCAGTCCCGGGCCACGATTGC-3′ (p.Met328Thr). For the 3 other mutations, p.Asp259Glu, p.Arg274Gln, and p.Ser467Asn, the substitutions were created by PCR overlap extension mutagenesis using the following mutagenic primers in forward and reverse orientations: CA CCGAAGGCGAGTCCCGCTTC (p.Asp259Glu), GAGTATCCAGCAAGCAGCATTGG (p.Arg274Gln) and CTTGTCAATGATGAGGCTGTGAC (p.Ser467Asn). All substitutions are noted in bold underlined. The resulting PCR fragments were next subcloned into the VANGL1–pCS2+ construct using EcoRI-SacI (p.Val239Ile), SacI-XmaI (p.Asp259Glu, p.Arg274Gln), XmaI-XbaI (p.Met328Thr, p.Ser467Asn) cassettes. All constructs were verified by sequencing. Sense capped RNA was synthesized using the mMESSAGE mMACHINE transcription kit (Ambion) as described by the manufacturer.
Longfin zebrafish were raised from a colony maintained according to established procedures in compliance with guidelines set out by the Canadian Council for Animal Care. Injections were performed in oocytes at the one to four cell stage. The vital dye fast-green was incorporated to the injection vehicle to monitor injection quality (0.1%; Sigma). To knockdown translation of the zebrafish tri/Vangl2, ~8 ng of the previously described tri-MO antisense morpholino oligonucleotide was injected. To rescue the phenotype, 50–150 pg of human VANGL1 RNA and 10–25 pg of zebrafish Vangl1 RNA were co-injected with the tri-MO. For the overexpression assay, 300–400 pg of human VANGL1 RNA was used. Body length was assessed at 2 days post fertilization and body angle was measured in 10–12 somites embryos. Because the tri-MO has previously been reported to induce necrosis, cell death was assessed in tri-MO injected embryos by incubating 2 day-old larvae in 0.5 mg/ml acridine orange for 25 min (Sigma).
In situ hybridization was carried out in 8–14 somites zebrafish embryos using the digoxigenin-labeled antisense MyoD probe. Standard protocols were used with minor modifications (Thisse and Thisse, 2008). Briefly, an acetylation treatment (1.25% triethanolamin and 0.27% acetic anhydride) was carried out after proteinase K digestion and 4% paraformaldehyde fixation. The anti-digoxigenin antibody was pre-adsorbed overnight at 4 °C with 2 mg/ml BSA in PBT before overnight incubation with the hybridized embryos. After washing in PBT, the embryos were incubated 3 × 5 min in a levamisol solution (100 mM Tris–HCl, pH 9.5, 50 mM MgCl2, 100 mM NaCl, 0.1% Tween 20 and 1 mM levamisol) instead of the alkaline Tris buffer. This solution was also used to stain the embryos with NBT and BCIP.
Since the measurements of zebrafish body length and body angle were not normally distributed, the Kruskal–Wallis test was used to compare differences between experimental groups, followed by Dunn's comparison as a post hoc test (SigmaPlot). We considered groups statistically significantly different from the control group if the P value was less than 0.05.
We thank L. Solnica-Krezel for the zebrafish Vangl1 construct and A. Bassuk for the MyoD probe. This work was supported by the Canadian Institutes for Health Research (P.G. MT-13425) (Z.K.), the SickKids Foundation (Z.K.), the Fonds de la Recherche en Santé du Québec (Z.K.), the Natural Science and Engineering Research Council, Genome Quebec and Genome Canada (P.D.) and the Gaslini Foundation and Telethon-Italy (Grant No. GGP08051) (V.C.). We also thank A.S.B.I. (Associazione Spina Bifida Italia).
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.mod.2009.12.002.