Abstract
Scribble1 (Scrib1) is a tumor suppressor gene that has long been established as an essential component of apicobasal polarity (ABP). In mouse models, mutations in Scrib1 cause a severe form of neural tube defects (NTDs) as a result of a defective planar cell polarity (PCP) signaling. In this study, we dissected the role of Scrib1 in the pathogenesis of NTDs in its mouse mutant Circletail (Crc), in cell lines and in a human NTD cohort. While there were no obvious defects in ABP in the Scrib1Crc/Crc neuroepihelial cells, we identified an abnormal localization of the apical protein Par-3 and of the PCP protein Vangl2. These results were concordant with those obtained following a partial knockdown of Scrib1 in MDCK II cells. Par-3 was able to rescue the localization defect of Vangl1 (paralog of Vangl2) caused by partial knockdown of Scrib1 suggesting that Scrib1 exerts its effect on Vangl1 localization indirectly through Par-3. This conclusion is supported by our findings of an apical enrichment of Vangl1 following a partial knockdown of Par-3. Re-sequencing analysis of SCRIB1 in 473 NTD patients led to the identification of 5 rare heterozygous missense mutations that were predicted to be pathogenic. Two of these mutations, p.Gly263Ser and p.Gln808His, and 2 mouse NTD mutations, p.Ile285Lys and p.Glu814Gly, affected Scrib1 membrane localization and its modulating role of Par-3 and Vangl1 localization. Our study demonstrates an important role of Scrib1 in the pathogenesis of NTDs through its mediating effect of Par-3 and Vangl1/2 localization and most likely independently of ABP.
Introduction
Primary neurulation is an early embryonic event that takes place during the third and fourth weeks of development and forms the brain and most of the spinal cord till the upper sacral level. A faulty neurulation leads to neural tube defects (NTDs) that are the most common and severe malformations of the central nervous system (affecting 1–2 per 1000 births). The two most common types of NTDs in humans are anencephaly and spina bifida [or myelomeningocele (MMC)] that result from failure of neural tube closure in the cranial and spinal regions, respectively. Anencephaly is invariably lethal while spina bifida is compatible with life but with severe physical and psychological disabilities. NTDs have a complex etiology involving genetic and environmental factors that remain largely undetermined (1). Despite the fact that periconceptional administration of folic acid has reduced their incidence by 50–70%, thousands of families are still affected with NTDs each year urging the need for a better understanding of the underlying pathogenic mechanisms and development of other preventive strategies (1,2).
Neurulation is a highly complex and dynamic process that is tightly regulated by distinct signaling pathways. Planar cell polarity (PCP) signaling, also called the non-canonical Wnt/Frizzled signaling, is a key pathway that controls a morphogenetic process called convergent extension (CE) responsible for narrowing and lengthening of the neural plate (3). PCP was first described in the fly where it controls the planar polarity of highly organized structures like the distal orientation of wing hairs and the complex organization of the ommatidia (eye units). A set of core PCP genes including Frizzled, Dishevelled, Strabismus/Van Gogh, Flamingo, Prickle and Diego, have been well characterized in the fly and demonstrated to be highly conserved in vertebrates (3,4). Mutations in PCP genes have been associated with NTDs in vertebrate models and in a small fraction (1–2%) of human patients (5,6).
Morphogenesis of the embryonic neural plate also implicates another type of polarity called apico-basal polarity (ABP) that is essential for normal epithelial cell shape, function, and proliferation, and for the overall tissue integrity and architecture. ABP is generated and maintained by antagonistic interactions between the apical aPKC–PAR6–PAR-3 and Crumbs–PALS1–PATJ complexes and the basolateral proteins Scribble, Lgl and Dlg. The apical and basal domains of epithelial cells are separated by adherens junctions and tight junctions that provide important adhesive contacts between neighboring epithelial cells and helps define each cell’s apical–basal axis (7,8). Defects in these processes result in many types of cancer (9,10). During neural tube formation, the apical membranes of individual neuroepithelial (NE) cells face the neural tube lumen and their basal membranes define the outer border of the neural tube. These cells adhere to each other to form a tight coherent monolayered NE sheet (11). In addition to this basic and essential role in maintaining the overall integrity of the neural tube tissue, ABP has been tightly linked to the process of apical constriction where cells change in shape from columnar to wedge-shaped during the formation of hinge points that are essential for bending of the neural plate (12).
Scribble1 (Scrib1) is tumor suppressor gene that has long been established as a key player in ABP (7,13). As mentioned above, it is located at the basolateral membrane of epithelial cells. Null mutations in Scrib1 cause disorganized epithelial tissues and neoplastic growth in Drosophila and the expression level of SCRIB1 is decreased in a number of human cancers (9,10). Scrib1 is a large cytoplasmic protein that consists of an N-terminal region composed of 16 leucine-rich repeats (LRR) that are required for membrane and basolateral targeting, 4 PDZ domains that are involved in protein–protein interactions and a C-terminal region that contains three spectrin binding motifs which are essential for SCRIB cortical dynamics and polarity function (7,14). In mice, mutations at Scrib1 cause a severe form of NTDs called Craniorachischisis where the neural tube fails to close along the entire body axis. This severe phenotype is a hallmark of a defective PCP signaling (15). Scrib1 genetically interacts with the core PCP gene Vangl2 to mediate the orientation of ear sensory cells (another PCP-controlled process), CE and neural tube formation (15,16). In mammalian cultured cells, Vangl2 co-localizes with Scrib1 to the basolateral plasma membrane and binds to its PDZ domains (17). Vangl2 and its paralog Vangl1 have very similar roles and are used interchangeably to dissect their roles in PCP signaling and neurulation (18–20).
Scrib1 is the only ABP gene so far that has been associated with defects in PCP signaling providing a possible link between these two types of polarity during neural tube formation. While the role of Scrib1 in ABP and in cancer has been extensively studied and well established, its role in PCP signaling and in the pathogenesis of NTDs remains to be determined. Does Scrib1 cause NTDs through a direct effect on PCP signaling? If yes, is this effect dependent on its interaction with Vangl2? Is it dependent or independent of its role in ABP? Does Scrib1 function in a distinct pathway that is independent of ABP and PCP but that converges with PCP to regulate cell movements during the process of CE? In the present study, we aimed at answering some of these questions and at dissecting the role of Scrib1 in neural tube formation using the mouse Scrib1 mutant called Circletail (Crc) that contains a single base insertion that creates a frameshift and leads to premature termination of the Scrib1 protein (15). We also used shRNA knockdown in MDCK II cells and re-sequencing and functional validation studies in a large cohort of human NTDs. We show that the localization of the apical marker Par-3 and the PCP gene Vangl2 is affected in Scrib1Crc/Crcneural tube while ABP seems to be generally unaffected. Partial knockdown of Scrib1 in MDCK II cells recapitulated these findings. Interestingly, we demonstrate that Par-3 can rescue the localization defect of Vangl1 caused by partial knockdown of Scrib1 and partial knockdown of Par-3 can cause a similar localization defect of Vangl1. We hypothesize that Scrib1 mediates the localization of Vangl1/2 indirectly through Par-3. Re-sequencing analysis of the coding region and exon–intron junctions of SCRIB1 in 473 NTDs identified five rare heterozygous missense mutations that were predicted to pathogenic in silico. Two of these human NTD mutations, p.Gly263Ser and p.Gln808His, and two mouse NTD mutations, p.Ile285Lys (21) and p.Glu814Gly (22), are most likely pathogenic since they affect Scrib1 membrane localization and its modulating role of Par-3 and Vangl1 localization.
Results
The open neural tube of Scrib1Crc/crc mutant mice shows a defect in Par-3 and Vangl2 localization but no overt defects in ABP
To determine whether Scrib1 acts within the ABP and/or the PCP pathway in neural tube development, we analyzed the expression patterns of markers of both types of polarity and of tight and adherens junction formation in E9.0–9.5 neural tube sections from Scrib1Crc/Crc mutants (Fig. 1). The cellular distribution of the basolateral marker Lgl2 as well as the structure of the tight junctions and adherens junction were not affected as ZO-1 and β catenin were normally distributed and detected at cell–cell contacts (Fig. 1A–I). Immunostaining with ZO-1 at the apical side of the NE cells facing the lumen of the neural tube was significantly diminished in mutant sections (P < 0.05) (Fig. 1H and I). In contrast, a prominent disruption of the localization of the apical marker Par-3 was observed where it significantly lost its localization at the apical site (P < 0.05) and changed its pattern from cell borders to mostly cytosolic in mutant tissues (Fig. 1J–L).
Apicobasal polarity, cytoskeletal structure and Vangl2 distribution in Scrib1Crc/Crc neural tube. Immunostaining of transverse cryosections of E9.0–9.5 ventral neural tube shows a normal distribution of β-catenin (B), Lgl2 (E) and ZO-1 (H) in mutant tissues when compared with wild-type tissues (A, D, G, respectively). The expression of ZO-1 is significantly reduced in mutant tissues at the apical site (H, I). The cellular distribution of Par-3 in mutant tissues is severely disrupted where it shifts from the cell borders to a cytosolic pattern (J compared with K, L). Vangl2 is significantly more enriched at the apical side in mutant tissues when compared with the wild type (N compared with M, O). Phalloidin staining showed normal cytoskeletal structure in mutant tissues (Q) when compared with wild-type tissues (P). The expression of Phalloidin is significantly reduced in mutant tissues at the apical site (Q, R). Images were taken by confocal microscopy with a × 63 magnification. The quantification of fluorescence intensity at apical and subapical regions in wild-type and mutant sections is shown (C,F,I,L,O,R). Au, arbitrary units. *P < 0.05; **P < 0.01; ***P < 0.001. Blue is DAPI. S, a diagrammatic representation of a mouse embryo showing the level of sectioning.
We next examined whether the Scrib1Crc mutation affects the localization of the PCP protein Vangl2 in neural tube development. In wild-type sections, Vangl2 was membrane associated with some punctate staining in the cytoplasm and was homogeneously distributed across the NE cells (Fig. 1M). In Scrib1Crc/Crc mutant tissues, the cellular localization of Vangl2 was not affected; however, there was a prominent change in its spatial localization as demonstrated by its significant enrichment at the apical side (P < 0.01) and reduction at the subapical site of the neural tube (P < 0.001) (Fig. 1N and O). Immunostaining with phalloidin that marks the filamentous F-actin network showed a normal cytoskeletal organization in mutant sections (Fig. 1P–R). However, its expression at the apical side of the neural tube was significantly diminished in mutant sections (P < 0.05) (Fig. 1Q and R).
The expression pattern of all these markers was also studied at a later stage of neural tube closure at E11.5 and the results were similar to those obtained at E9.0–9.5 (Supplementary Material, Fig. S1). Changes in Par-3 and Vangl2 localization in mutant tissues at both stages, E9.0–9.5 and E11.5, were not because of excessive cell death since wild-type and mutant neural tissues showed similar expression levels of Caspase 3 (Supplementary Material, Fig. S2).
Suppression of Scrib1 in MDCK II cells affects Par-3 and Vangl1 localization
In parallel to our immunostaining study in Scrib1Crc/Crc mice, we investigated the effect of Scrib1 knockdown on cellular localization of ABP and PCP markers in MDCK II canine kidney epithelial cells that are widely used for studying cell polarity and adhesion. A previously characterized small hairpin RNA (shRNA) called ScrbKD was used to efficiently suppress the expression of Scrib1 by transient transfection in MDCK II cells plated at high densities (23). As shown previously, this shRNA can reduce the expression level of Scrib1 by ∼60% (Supplementary Material, Fig. S3). Suppression of Scrib1 had no obvious effect on cytoskeletal organization, tight junction and adherens junction formation as revealed by immunostaining with phalloidin, ZO-1, and β-catenin, respectively (data not shown). Suppression of Scrib1 also had no effect on the expression pattern of the basolateral marker Lgl2 (Supplementary Material, Fig. S4). In contrast, suppression of Scrib1 had dramatic effects on the expression patterns of both Par-3 and Vangl1 (Fig. 2).
Effect of suppression of Scrib1 on Par-3 and Vangl1 localization. (A) Suppression of Scrib1 shifts the cellular localization of Par-3 from the cell periphery and apical surfaces to a cytosolic pattern across the cells. Human Scrib1 can rescue this defect. MDCK II cells were transfected with ScrbKD alone or with a construct encoding GFP-tagged human Scrib1 (hScrib-GFP). Cells were fixed and stained for endogenous Par-3. (B) Quantification of the Par-3 apical or cytosolic localization upon knockdown of Scrib1 and following the rescue experiment. (C) Suppression of Scrib1 causes an apical enrichment of Vangl1 that was rescued by co-transfection with hScrib-GFP. MDCK II cells stably transfected with c-Myc/HA-tagged VANGL1 were transfected with ScrbKD alone or with hScrib-GFP. Cells were fixed and stained for HA-tagged VANGL1. (D) Quantification of the basolateral only versus basolateral and apical localization of HA-VANGL1 upon knockdown of Scrib1 and following the rescue experiment. Images and Z-stacks shown at the lower panels were taken by confocal microscopy with a × 63 magnification. A total of 135 cells were examined per category. **P < 0.01.
In polarized MDCK II cells, Par-3 was expressed exclusively at the apical surfaces (Fig. 2A and B). Suppression of Scrib1 caused a dramatic alteration of this expression pattern where Par-3 became cytosolic in ∼80% of the cells (Fig. 2A and B). Western blot analysis demonstrated similar levels of Par-3 in Scrib1 knockdown-cells when compared with untransfected cells suggesting an effect on protein localization rather than protein stability (Supplementary Material, Fig. S5). Co-injection of human SCRIB1 with ScrbKD restored the apical distribution of Par-3 (∼80%) demonstrating that the observed phenotype is specific to the suppression of Scrib1 (Fig. 2A and B).
The effect of suppression of Scrib1 on Vangl1 was assessed in MDCK II cells stably transfected with HA-tagged Vangl1 (HA-Vangl1). These cell lines were previously used to study the effect of mutations in Vangl1 and Vangl2 on protein localization and stability. Immunostaining of these cell lines showed a predominant expression of the Vangl1 protein in the basolateral membrane (∼90%) and to a lesser extent at both basolateral and apical sides (∼10%). Suppression of Scrib1 caused a dramatic reversal of this distribution where it was significantly enriched at the apical sides of the cells (∼80%). Co-injection of human SCRIB1 with shRNA-Scrib1 partially restored this effect demonstrating the specificity of observed phenotype (Fig. 2C and D).
Scrib1 modulates the localization of Vangl1 through Par-3
To determine whether the disrupted patterns of expression of Par-3 and Vangl2 caused by Scrib1 suppression are interrelated, we conducted rescue experiments with Vangl2 or Par-3 in cells expressing reduced levels of Scrib1. Overexpression of Vangl2 did not rescue the Par-3 localization defect caused by reduced levels of Scrib1 (Fig. 3A and B). Remarkably, even with very low amounts of Scrib1, the overexpression of Par-3 partially rescues the abnormal localization of Vangl1 at the apical sides (Fig. 3C and D).
Effect of Par-3 on subcellular localization of VANGL1. (A) VANGL2 fails to rescue the abnormal localization of Par-3 caused by suppression of Scrib1. MDCK II cells were transfected with ScrbKD either alone or with a construct encoding human VANGL2. Cells were fixed and stained for endogenous Par-3. (B) Quantification of the Par-3 apical or cytosolic localization upon knockdown of Scrib1 and following the rescue experiment with human VANGL2. (C) Par-3 can partially rescue the apical enrichment of VANGL1 caused by suppression of Scrib1. MDCK II cells were transfected with ScrbKD either alone or with a construct encoding human Par-3. Cells were fixed and stained for HA-tagged VANGL1. (D) Quantification of the basolateral only versus basolateral and apical localization of HA-VANGL1 upon suppression of Scrib1 and following the rescue experiment with hPar-3. (E) Suppression of Par-3 causes an apical enrichment of Vangl1 that was rescued by co-transfection with hPar-3. MDCK II cells stably transfected with c-Myc/HA-tagged VANGL1 were transfected with Par-3 KD either alone or with hPar-3. Cells were fixed and examined for the subcellular localization of VANGL1-GFP. (F) Quantification of the basolateral only versus basolateral and apical localization of HA-VANGL1 upon knockdown of Par-3 and following the rescue experiment. Images and Z-stacks shown at the lower panels were taken by confocal microscopy with a × 63 magnification. A total of 135 cells were examined per category. **P < 0.01.
To further investigate the role of Par-3 in Vangl1 localization, we used a previously characterized shRNA (called Par-3 KD) to knockdown the expression of Par-3 in MDCK II cells plated at high densities (24). As shown previously, this shRNA can reduce the expression level of Par-3 by ∼60% (Supplementary Material, Fig. S3). Suppression of Par-3 caused a significant disruption of the localization pattern of Vangl1 similar to the effect observed upon Scrib1 suppression where Vangl1 became enriched at the apical sides (∼75%) (Fig. 3E and F). Co-injection of Par-3 with Par-3KD was able to partially rescue this effect demonstrating the specificity of observed phenotype (Fig. 3E and F).
These data suggest that Scrib1 mediates the basolateral localization of Vangl1 through its restrictive effect on Par-3 localization at the apical sides. Upon suppression of Scrib1, Par-3 loses its apical localization and become expressed in a cytosolic manner which in turn and through unknown mechanisms leads to enrichment of Vangl1 at the apical sides. This effect seems to be independent of the well characterized effect of Scrib1 on Par-3 apical localization in the process of establishing A-B cell polarity, since the general morphology, the cytoskeletal organization, the structure of the tight junctions and adherens junction and the distribution of the basolateral marker Lgl2 seem to be unaffected upon suppression of Scrib1.
Rare and novel mutations in SCRIB1 are associated with human NTDs
To further investigate the role of SCRIB1 in the pathogenesis of NTDs, we sequenced its coding region and exon–intron junctions in a large and a well-characterized cohort of 473 patients affected with various forms of open and closed NTDs. We identified 9 rare missense mutations, five of which, p.Gly263Ser, p.Pro649His, p.Gln808His, p.Arg1150Gln and p.Thr1422Met, were absent in 400 Italian controls and were predicted to be pathogenic in silico. All five mutations were absent from the dbSNP database. One mutation, p.GLy263Ser, was also absent in the ExAC database, while the other 4 mutations were detected at very small frequencies (Table 1).
Rare mutations identified in SCRIB1 in neural tube defects
| Mutationa | Amino acid changea | Freq. in patients (n=473) | Freq. in controls (n=467) | Freq. in dbSNP | Freq. in ExAC | Polyphenb | SIFTb |
|---|---|---|---|---|---|---|---|
| c.279A>C | p.Asp93Ala | 1 MMC | 1 | 0.0002 | 1.3×10−3 | + | + |
| c.435G>A | p.Gly145Arg | 1 MMC | 1 | 0.0004 | 1.3×10−3 | − | + |
| c.787G>A | p.Gly263Ser | 1 MMC | 0 | 0 | 0 | + | − |
| c.1407G>A | p. Gly469Ser | 1 MMC | 2 | 0 | 9×10−5 | + | − |
| c.1945C>A | p.Pro649His | 1 CA | 0 | 0 | 4×10−4 | + | + |
| c.2423G>T | p.Gln808His | 1 MMC | 0 | 0 | 3×10−4 | + | + |
| c.3198G>A | p.Val1066Met | 1 MMC | 2 | 0 | 2×10−5 | + | + |
| c.3450G>A | p.Arg1150Gln | 1 VS | 0 | 0 | 2×10−4 | − | + |
| c.4267C>T | p.Thr1422Met | 1 MMC | 0 | 0 | 7×10−5 | − | + |
| Mutationa | Amino acid changea | Freq. in patients (n=473) | Freq. in controls (n=467) | Freq. in dbSNP | Freq. in ExAC | Polyphenb | SIFTb |
|---|---|---|---|---|---|---|---|
| c.279A>C | p.Asp93Ala | 1 MMC | 1 | 0.0002 | 1.3×10−3 | + | + |
| c.435G>A | p.Gly145Arg | 1 MMC | 1 | 0.0004 | 1.3×10−3 | − | + |
| c.787G>A | p.Gly263Ser | 1 MMC | 0 | 0 | 0 | + | − |
| c.1407G>A | p. Gly469Ser | 1 MMC | 2 | 0 | 9×10−5 | + | − |
| c.1945C>A | p.Pro649His | 1 CA | 0 | 0 | 4×10−4 | + | + |
| c.2423G>T | p.Gln808His | 1 MMC | 0 | 0 | 3×10−4 | + | + |
| c.3198G>A | p.Val1066Met | 1 MMC | 2 | 0 | 2×10−5 | + | + |
| c.3450G>A | p.Arg1150Gln | 1 VS | 0 | 0 | 2×10−4 | − | + |
| c.4267C>T | p.Thr1422Met | 1 MMC | 0 | 0 | 7×10−5 | − | + |
The five missense SCRIB1 mutations detected in NTDs and absent from ethnically matched controls and prioritized for functional validation are indicated in bold.
The numbering systems of the mutation at the nucleotide level and the amino acid change are based on SCRIB1 cDNA and protein sequence with the accessions NM_015356.4 and AAH44627.1, respectively.
Polyphen: +, probably or possibly damaging; –, benign. SIFT: +, not tolerated; −, tolerated. CA, caudal agenesis; Freq., frequency; MMC, myelomeningocele; VS, vertebral schisis.
Rare mutations identified in SCRIB1 in neural tube defects
| Mutationa | Amino acid changea | Freq. in patients (n=473) | Freq. in controls (n=467) | Freq. in dbSNP | Freq. in ExAC | Polyphenb | SIFTb |
|---|---|---|---|---|---|---|---|
| c.279A>C | p.Asp93Ala | 1 MMC | 1 | 0.0002 | 1.3×10−3 | + | + |
| c.435G>A | p.Gly145Arg | 1 MMC | 1 | 0.0004 | 1.3×10−3 | − | + |
| c.787G>A | p.Gly263Ser | 1 MMC | 0 | 0 | 0 | + | − |
| c.1407G>A | p. Gly469Ser | 1 MMC | 2 | 0 | 9×10−5 | + | − |
| c.1945C>A | p.Pro649His | 1 CA | 0 | 0 | 4×10−4 | + | + |
| c.2423G>T | p.Gln808His | 1 MMC | 0 | 0 | 3×10−4 | + | + |
| c.3198G>A | p.Val1066Met | 1 MMC | 2 | 0 | 2×10−5 | + | + |
| c.3450G>A | p.Arg1150Gln | 1 VS | 0 | 0 | 2×10−4 | − | + |
| c.4267C>T | p.Thr1422Met | 1 MMC | 0 | 0 | 7×10−5 | − | + |
| Mutationa | Amino acid changea | Freq. in patients (n=473) | Freq. in controls (n=467) | Freq. in dbSNP | Freq. in ExAC | Polyphenb | SIFTb |
|---|---|---|---|---|---|---|---|
| c.279A>C | p.Asp93Ala | 1 MMC | 1 | 0.0002 | 1.3×10−3 | + | + |
| c.435G>A | p.Gly145Arg | 1 MMC | 1 | 0.0004 | 1.3×10−3 | − | + |
| c.787G>A | p.Gly263Ser | 1 MMC | 0 | 0 | 0 | + | − |
| c.1407G>A | p. Gly469Ser | 1 MMC | 2 | 0 | 9×10−5 | + | − |
| c.1945C>A | p.Pro649His | 1 CA | 0 | 0 | 4×10−4 | + | + |
| c.2423G>T | p.Gln808His | 1 MMC | 0 | 0 | 3×10−4 | + | + |
| c.3198G>A | p.Val1066Met | 1 MMC | 2 | 0 | 2×10−5 | + | + |
| c.3450G>A | p.Arg1150Gln | 1 VS | 0 | 0 | 2×10−4 | − | + |
| c.4267C>T | p.Thr1422Met | 1 MMC | 0 | 0 | 7×10−5 | − | + |
The five missense SCRIB1 mutations detected in NTDs and absent from ethnically matched controls and prioritized for functional validation are indicated in bold.
The numbering systems of the mutation at the nucleotide level and the amino acid change are based on SCRIB1 cDNA and protein sequence with the accessions NM_015356.4 and AAH44627.1, respectively.
Polyphen: +, probably or possibly damaging; –, benign. SIFT: +, not tolerated; −, tolerated. CA, caudal agenesis; Freq., frequency; MMC, myelomeningocele; VS, vertebral schisis.
The mutation p.Gly263Ser (c.787G>A) was identified in a sporadic case of MMC associated with hydrocephalus and Chiari II malformation. It affects an absolutely conserved glycine residue that maps to the LRR domain implicated in the correct localization of SCRIB1 at the basolateral side during the establishment of the ABP (17,25) (Fig. 4A, Supplementary Material, Fig. S6). It is not conservative as it dramatically reduces the hydrophobicity of the affected glycine residue and replaces it with a hydrophilic serine residue. The mutation p.Pro649His (c.1945C>A) was identified in a sporadic case of a severe form of closed NTDs called caudal agenesis type II with dermal sinus, tethered cord, vesical-ureteral reflux, costal agenesis, right arm hypoplasia, ptosis and growth deficit. This mutation maps to the linker region between the LAPsd and PDZ1 domain and is non-conservative as it changes a moderately conserved proline to a positively charged histidine residue (Fig. 4A, Supplementary Material, Fig. S6).
Rare NTD-associated SCRIB1 mutations and their effect on Scrib1 membrane localization. (A) Schematic diagram of the SCRIB1 protein showing the approximate location of the five missense mutations identified in our cohort that were absent in controls and the two mouse mutations included in functional studies. (B) Western blot analysis shows comparable protein expression levels of the wild-type and various mutant hScrib-GFP constructs in transfected MDCK cells. (C) Top, representative MDCK II cells transfected with hScrib-GFP and showing its localization at plasma membrane (PM) only, cytoplasmic and PM and cytoplasmic only. Bottom, quantification of the three patterns of localization of wild-type or mutant hScrib-GFP cells in MDCK II cells. Two mouse mutations, p.Ile285Lys and p.Glu814Gly, and two human mutations, p.Gly263Ser and p.Gln808His, significantly affected the subcellular localization of SCRIB1. A total of 300 cells were examined per construct. **P < 0.01 when compared with the wild-type hScrib-GFP.
The mutation p.Gln808His (c.2423G>T) was detected in a sporadic case of MMC, associated with hydrocephalus, Chiari II malformation, tethered cord, equinus foot and precocious puberty. It affects a moderately conserved residue that maps to PDZ1 implicated in interaction of Scrib1 with the PCP protein Vangl2 (17) (Fig. 4A, Supplementary Material, Fig. S6). Replacement of a glutamine by a histidine residue is not conservative as it introduces a positive charge at an otherwise neutral position. The mutation p.Arg1150Gln (c.3450G>A) was identified in familial case of vertebral schisis that is a considered as the mildest form of closed NTDs and that reflects a defect in fusion of the posterior vertebral neural arch. The mother of this patient had one apparently healthy child and three terminated pregnancies because of lumbosacral MMC. Unfortunately, no DNA was available from members of this family for segregation analysis of this mutation. This mutation replaces a highly conserved base residue into a neutral one and maps to the fourth PDZ domain also implicated in Scrib1 interaction with Vangl2 (17) (Fig. 4A, Supplementary Material, Fig. S6).The p.Thr1422Met (c.4267C>T) mutation was identified in a sporadic case of MMC, hydrocephalus, Chiari II malformation, hip dislocation and limb dysmetria. It affects a weakly conserved threonine that maps to the C-terminal domain (Fig. 4A, Supplementary Material, Fig. S6).
Rare mutations in SCRIB1 associated with NTDs affect its membrane localization
All five NTD-associated mutations that were absent in ethnically matched controls and were predicted to be pathogenic were prioritized for further functional studies. Previous studies showed that Scrib1 must localize to the cell membrane for its normal function (17,26). We next validated the effect of all five mutations detected in SCRIB1 in our cohort on its membrane localization. For a more comprehensive analysis of the function of Scrib1 in the development of NTDs, we also included in our functional validation assays, two ENU-induced mutations that cause Craniorachischisis and a kinked tail in mouse, p.Glu814Gly that maps to the first PDZ domain (22) and p.Ile285Lys (21) that maps to the LRR region that is implicated in membrane targeting of Scrib1 and that was previously demonstrated to disrupt its membrane localization. All 5 human mutations and both mouse mutations did not affect the protein expression levels in transfected MDCK II cells (Fig. 4B). Wild-type hScrib-GFP exhibited strong membrane localization and as previously reported the mouse mutation p.Ile285Lys led to a significant increase in the number of cells expressing hScrib-GFP in the cytoplasm (P < 0.01) (Fig. 4C). One mouse mutation, p.Glu814Gly and two human mutations, p.Gly263Ser and p.Gln808His, behaved like p.Ile285Lys and significantly affected the subcellular localization of SCRIB1 (P < 0.01) (Fig. 4C). The other human mutations tested, p.Pro649His, p.Thr1422Met and p.Arg1150Gln, exhibited the same localization pattern as the wild-type (Fig. 4C).
Mutations in SCRIB1 associated with NTDs fail to rescue the effect of SCRIB1 suppression on par-3 and Vangl1 localization
In this study, we have demonstrated an abnormal cellular localization pattern of Par-3 and Vangl1/2 in Scrib1Crc/Crc mutant neural tissues and upon suppression of Scrib1 in an MDCK II cell line-based assay. To determine whether mutations in SCRIB1 associated with human or mouse NTDs could affect this modulating effect, we conducted a knockdown/rescue validation assay in MDCK II cells. Co-transfection of each of the two mouse NTD mutations, p.Ile285Lys and p.Glu814Gly and each of the two human NTD-associated mutations, p.Gly263Ser and p.Gln808His, with ScrbKD failed to rescue the localization defect of Par-3 and Vangl1 caused by knockdown of Scrib1 (P < 0.01) (Fig. 5). Co-transfection of the 3 other human NTD-associated mutations, p.Pro649His, p.Thr1422Met, and p.Arg1150Gln, with ScrbKD was able to rescue the localization defect, similar to the wild-type level (Fig. 5). Representatives of raw imaging data for this functional validation assay are provided in Supplementary Material, Figure S7.
Effect of rare NTD-associated SCRIB1 mutations on its mediating ability of Par-3 and VANGL1 localization. (A) Quantification of the basolateral only versus basolateral and apical localization of HA-VANGL1 upon suppression of Scrib1 and following co-transfection with either wild-type or mutant hScrib-GFP. Co-transfection of each of the two mouse NTD mutations, p.Ile285Lys and p.Glu814Gly and each of the two human NTD-associated mutations, p.Gly263Ser and p.Gln808His, with ScrbKD failed to rescue the localization defect of Par-3 and Vangl1 caused by suppression of Scrib1. A total of 135 cells were examined per construct. Cells co-transfected with ScribKD and wild-type hScrib-GFP were compared with untransfected cells. Cells co-transfected with ScribKD and mutant hScrib-GFP were compared with cells co-transfected with ScribKD and wild-type hScrib-GFP. (B) Quantification of the Par-3 apical or cytosolic localization upon suppression of Scrib1 and following co-transfection with either wild-type or mutant hScrib-GFP. **P < 0.01.
Variants in SCRIB1 associated with NTDs interrupt its interaction with VANGL2
Scrib1 physically interacts with Vangl2 and plays a direct role in its asymmetric targeting during PCP signaling in the cochlea (16,17). This interaction involves the PDZ domains of Scrib1 and the C-terminal PDZ binding domain of Vangl2. Two of the mutations associated with human NTDs identified in this study, p.Gln808His and p.Arg1150Gln, map to the first and fourth PDZ domains, respectively (Fig. 6A). One mouse mutation, p.Glu814Gly, maps to the fist PDZ domain. Hence, we tested the ability of all three mutations to affect Scrib1 interaction with Vangl2 using the yeast two-hybrid system (Y2H). All SCRIB1 mutants were expressed in yeast cells at comparable levels to that seen for the wild-type protein in the yeast suggesting that they do not affect protein stability in yeast cells (Fig. 6B). All 3 Scrib1 mutations abolished the interaction of Scrib1 with Vangl2 in triple dropout conditions (-Hi/-Trp/-Leu) providing additional evidence of their pathogenic effect on protein function (Fig. 6C and D).
Effect of rare NTD-associated SCRIB1 mutations on its interaction of with VANGL2 using a yeast two-hybrid system (Y2H). (A) Schematic diagram of SCRIB1 PDZ domains showing the location of missense human and mouse mutations tested in the Y2H. (B) Western blot analysis of wild-type and mutant SCRIB1-PDZ (aa 728–1194) with anti-HA shows comparable expression levels in yeast cells. (C) Diploid cells produced by mating hSCRIB1-PDZ-pGADT7 and hVANGL2-Cterm-pGBKT7 expressing yeast cells. Cells were plated on a medium of increasing stringency of selection to detect interaction (growth). The medium labeled DDO (double drop out) lacks leucine and tryptophan and supports the growth of diploid cells without a requirement for protein interaction. The medium labeled TDO (triple drop out) lacks histidine, leucine, and tryptophan and supports the growth of diploid cells only if proteins interact. The 2 human mutations p.Gln808His and p.Arg1150Gln and the mouse mutation p.Glu814Gly disrupt the interaction of SCRIB1-PDZ with VANGL2-C terminal. (D) Interaction between the hSCRIB1-PDZ-pGADT7 and hVANGL2-Cterm-pGBKT7 was confirmed by measuring the activity of α-galactosidase. *P < 0.05.
Discussion
Scrib1 exerts a tissue-specific effect in Scrib1Crc/crc mutant tissues
ABP and cell–cell contacts were generally unaffected in the NE cells in Scrib1Crc/Crc mutants, although we cannot completely rule out subtle alterations. These data are partially consistent with previous studies of Scrib1Crc/Crc mutants where ABP seems to be normal in the developing lung and heart providing strong evidence that the developmental defects caused by loss of function of Scrib1 in all tissues analyzed are independent of its role in ABP (27,28). This finding is in sharp contrast to the severe disruption of ABP in Drosophila Scrib mutants (13,29).
Scrib1Crc/Crc neural tissues displayed a normal subcellular distribution of the adherens junction protein β-catenin suggesting that cell–cell contacts are properly maintained in affected neural tissues when compared with heart and lung tissues (27,28). This distribution was severely disturbed in mutant lung and heart epithelia and following knockdown of Scrib1 in organotypic cultures demonstrating an essential role for Scrib1 in regulating cell adhesion during lumen formation in the developing lung and cardiomyocyte organization within the early heart tube. It is important to mention that adherens junction were not lost in these mutant tissues and it was demonstrated in the lung that mislocalization of junctional proteins resulted in weaker cell–cell interactions and reduced cell cohesion (28). This function seems to be distinct than PCP as defects in cell junctions during lung morphogenesis were not observed in two PCP mutants, Celsr1 or Vangl2. While β-catenin distribution in Scrib1Crc/Crc mutant neuroepithelium seems to be normal, we cannot rule out the possibility of weaker cell–cell interactions in this tissue and we cannot withdraw major conclusions on the requirement of Scrib1 as a major regulator of cell adhesion, in addition to PCP, in neural tube formation.
In this study, we have demonstrated an important role for Scrib1 in proper distribution of Par-3 and Vangl2 in the developing neural tube. While the localization of Par-3 has not been previously investigated in Scrib1Crc/Crc mutants, defects in Vangl2 localization were also detected in Scrib1Crc/Crc in lung epithelium and cardiomyocytes however they were not similar to those observed the neural tube in this study. In the developing lung, Vangl2 was enriched toward the apical side of wild-type airways and this was not observed in Scrib1Crc/Crc mutant lung tissue indicating mild changes in spatial distribution of Vangl2 (28). In the developing myocardium, Vangl2 co-localized with Scrib1 within the membrane compartments of cardiomyocytes and this was significantly altered in mutant cells where it became mainly localized in the cytoplasm (27). In our study, the subcellular localization of Vangl2 was not affected in Scrib1Crc/Crc NE cells where it was always detected at the membrane; however, its tissue distribution was affected where it became significantly enriched at their apical sides suggesting a tissue-specific functional requirement of Scrib1 in spatial distribution of Vangl2 in the developing neural tube.
Scrib1 controls the localization of Vangl1/2 through it effect on Par-3 localization
Scrib1Crc/Crc mutant NE tissues display a significant alteration in Par-3 localization where it was significantly reduced at the apical site of the neural tube and where it shifted from around the cell periphery to a cytosolic pattern. These mutant cells also displayed an apical enrichment of Vangl2 facing the lumen. Suppression of Scrib1 in MDCK II cells caused similar defects in Par-3 and Vangl1 localization that can be rescued by co-transfection of wild-type Scrib1. Following Scrib1 shRNA knockdown, Par-3 can partially rescue the resulting abnormal localization of Vangl1 while Vangl2 failed to rescue the defect in Par-3 localization suggesting that Par-3 and Vangl1/2 do not reciprocally maintain the apical and basal boundaries. Suppression of Par-3 caused a defect in Vangl1 localization that was similar to that observed upon Scrib1 suppression and that can be partially saved by Par-3. These data suggest that Scrib1 controls the apical confinement of Par-3 that in turns and through unknown mechanisms restricts the basolateral localization of Vangl1/2. Upon suppression of Scrib1, Par-3 loses it apical restriction and contributes to a loss of the basolateral restriction of Vangl1/2, a phenotype that is also generated by Par-3 knockdown and that is independent of the role of either Scrib1 or Par-3 in ABP. This model does not exclude the presence of other parallel mechanisms by which Scrib1 mediates the basolateral localization of Vangl1/2, especially that Par-3 does not completely rescue the abnormal apical enrichment of Vangl1 caused by suppression of Scrib1.
The role of Scrib in confining Par-3 at the apical sides of the epithelial cells has long been established in the fly (13). Two models were hypothesized but still not proven for the underlying mechanisms. The first model represents Scrib as part of a diffusion barrier, similar to vertebrate tight junctions, that physically separates apical and basolateral compartments. The second model suggests that Scrib has a role in the polarized targeting of transport vesicles carrying apical proteins (13). How can Par-3 restrict the localization of Vangl1/2 at the basolateral membrane? One could speculate on the same two models proposed for Scrib1. Par-3 co-localizes with ZO1 at tight junctions and therefore could act as a physical barrier for Vangl1/2. Alternatively, Par-3 might interact through its PDZ domains with the ‘exocyst’, a secretory targeting apparatus localized to the tight junction and involved in polarized segregation of transmembrane proteins. Interestingly, overexpression of Bazooka (Par-3 orthologue in the fly) in wing epithelial cells was shown to prevent polarized accumulation of the PCP protein Flamingo (Fmi) (30). It was suggested that Par-3 could affect microtubule orientation and consequently redirect polarized transport of Fmi-containing vesicles (30). One could hypothesize a similar role for Par-3 in polarized transport of Vangl1/2-containing vesicles during PCP signaling. Knockdown of Par-3 or its ectopic expression caused by a knockdown of Scrib1 would disrupt this polarized accumulation.
Scrib1, Vangl2, Par-3 and CE
Scrib1 is a multifunctional protein that has essential roles in many processes important for neural tube formation including cell migration, cell adhesion, oriented cell divisions and CE (15,23,31). While a direct analysis of CE in the Scrib1Crc/Crc mutant has not been reported, this mutant exhibits several morphological features that are similar to the PCP mutant Looptail (Vangl2Lp/Lp) including widely spaced neural folds, a shortened body axis and widened midline (15). Scrib1 genetically interacts with Lp (15) and we show in this study that Scrib1 mutants display altered tissue distribution of Vangl2 in the developing neural tube. All these data provide strong evidence that the NTD in Scrib1Crc/Crc is caused by a defect in CE rather than other cellular and developmental processes mediated by Scrib1.
To date, Scrib1 was shown to physically and genetically interact with only one core PCP member Vangl2 (15, 17). Scrib1 was shown to play an important role in asymmetric targeting of Vangl2 to specific cell–cell boundaries along the axis of polarization in the developing mammalian cochlea. This asymmetry was lost in both Vangl2 and Scrib1 mutants. Despite the mutation in Scrib1, Vangl2 was still present at the membranes of mutant cochleae cells (16). This is consistent with our findings where Vangl2 was detected at the membranes of Scrib1Crc/Crc mutant NE cells but displayed a strong enrichment at the apical site. In MDCK II cells expressing low levels of Scrib1, Vangl1 was also expressed at the plasma membrane but was not restricted to its basolateral sides. These data suggest that the Scrib1Crc mutation causes a defect in CE through its effect on Vangl1/2 localization that is very essential for its role in CE and PCP signaling.
One unexpected and interesting finding in our study is the requirement for a proper localization of the apical protein Par-3 in this mediating effect of Scrib1 on Vangl1 localization. Several studies have demonstrated an important role of Par-3 in PCP signaling independently of its well established role in ABP and tight junction formation. In fact, while a conditional knockout of Par-3 in the kidney revealed no major defects in ABP and tight junctions’ formation, it resulted in defective mediolateral orientation of epithelial cells leading to defective CE and tubular morphogenesis (32). A more direct implication of Par-3 in PCP signaling was demonstrated in the wing hair cells in the fly where overexpression of Bazooka/Par-3 perturbs development of PCP by failing to restrict Flamingo (Fmi) to the proximal and distal plasma membranes of the wing epithelium (30). Based on our findings of a direct effect of Par-3 on Vangl1 localization in MDCK II cell lines, it would be highly pertinent to investigate whether Bazooka/Par-3 mutants display any defect in Strabismus/Vangl2 subcellular and tissue distribution.
Asymmetric localization of PCP proteins across cells undergoing planar polarization is a crucial requirement for specific interactions among them and proper PCP signaling (33). Based on these previous studies and our current findings, we hypothesize that the Scrib1Crcmutation affects Par-3 apical localization that in turn affects Vangl1/2 localization leading to defects in PCP signaling and CE, which ultimately results in failure of neural tube formation. This novel role of Par-3 is most likely independent of its well established role in ABP. In our study, major alterations in Par-3 localization did not affect either ABP or Lgl2 distribution. This is concordant with the absence of major defects in ABP in the conditional knockout of Par-3 in the kidney, possibly because of functional redundancy by other PAR3-related proteins (e.g. PAR3L/PAR3).
Mutations in SCRIB1 in mouse and human NTDs affect its membrane association, its ability to modulate Par 3 and Vangl1 subcellular localization and its physical interaction with Vangl2
Rare mutations in Scrib1 have been previously associated with NTDs in mouse models and humans (15,21,22,26,34) . Re-sequencing studies of SCRIB1 in 36 fetuses affected with Craniorachischisis and 192 infants affected with MMC identified a total of 7 rare missense mutations associated with NTDs, 4 of which affected subcellular localization of SCRIB1 (26,34). In this study, we extended the genetic screening of SCRIB1 to a larger cohort of 473 patients affected with various forms of open and closed NTDs including 214 cases of MMC and 247 cases of closed NTDs of which 61 were affected with the severe and very rare caudal agenesis. We detected a total of 5 rare missense mutations that were absent in ethnically-matched controls in 5 NTD patients and that were predicted to be pathogenic. One mutation, pGly263Ser, was novel as it was also absent in public databases. This frequency of NTD-associated mutations in SCRIB1 of 1% (5/473) is lower than what was previously reported for this gene (7/228 or 3%), possibly reflecting the distribution of the NTD type in the cohort analysed or just simply because of sampling bias. These potentially pathogenic mutations were detected in 3 MMC cases, one case of caudal agenesis and one case of dermal sinus that represents the mildest form of NTDs. The latter is a familial case where the mother had 3 terminated pregnancies of fetuses affected with lumbosacral MMC. Most likely and as expected for a disease with a complex and genetically heterogeneous disease as an NTD, these mutations act as predisposing factors that must interact with other genetic and environmental factors to modulate the incidence and severity of the NTD phenotype.
We next proceeded with functional validation of these mutations by investigating their effect on Scrib1 membrane targeting, its physical interaction with Vangl2 and on its ability to modulate the subcellular localization of Vangl1/2 through Par-3. We included in our functional analysis two mouse mutations, p.Ile285Lys and p.Glu814Gly, that act in a semi-dominant and highly penetrant manner. This comprehensive analysis of the potential effect of seven SCRIB1 mutations in three functional assays helped us better understand the role of this gene in the molecular pathogenic mechanisms underlying NTDs. Two human mutations, p.Gly263Ser and p.Gln808His, and two mouse mutations, p.Ile285Lys and p.Glu814Gly, affected significantly the membrane association of SCRIB1 and failed to rescue its ability to mediate subcellular localization of Par-3 and Vangl1, providing strong evidence that these 4 mutations are pathogenic. The mutations p.Ile285Lys and p.Gly263Ser map to the LRR of SCRIB1 that is crucial for its subcellular localization and function. The mutations p.Gln808His, p.Glu814Gly along with another mutation p.Arg1150Gln that map to the PDZ domains of SCRIB1 were found to affect its interaction with Vangl2 in an Y2H assay.
In the fly and mammary cell lines, it was shown that alteration of Scrib localization at the membrane often mimics loss-of-function phenotype (35–37). In our study, the mutations that affected SCRIB1 function mapped to its LRR or PDZ domains. Using an allelic series of mutations along with rescuing transgenes in the fly, a two-part mechanism was suggested to localise Scrib to the membrane: the first step involves interactions mediated by the LRR that tether the protein to the plasma membrane and in the second step, PDZ domain interactions enrich membrane-bound Scrib at the future site of the septate junctions (37). Biochemical and morphological studies in polarized MDCK II cells demonstrated that Vangl2 acts as a membrane anchor for the Scrib complex (17). It was suggested that LRR may act in a dominant manner in binding Scrib1 to the membrane and that Vangl2 binding to Scrib PDZ domains might stabilize the membrane-bound Scrib to allow additional proteins to bind to other regions in Scrib. This “less dominant” and stabilizing role of Vangl2 seems to be important but not essential for Scrib1 localization and function and in fact deleting the PDZ domains caused mild polarity defects in the fly and did not affect the membrane binding of Scrib1 in MDCK II cells (17,37). In our study, one human mutation, p.Arg1150Gln, affected the interaction of Scrib1 with Vangl2 but did not disrupt its membrane localization and was able to rescue the effect of Scrib1 knockdown on Par-3 and Vangl2 localization. On the other hand, two mutations that map to the LRR motif, one human mutation, p.Gly263Ser, and one mouse mutation, p.Ile285Lys, affected Scrib1 subcellular localization and completely abolished its ability to rescue its knockdown effect on Par-3 and Vangl1 localization. These findings are concordant with a dominant and essential role of the LRR and a less dominant role of Vangl2 binding in Scrib1 normal function. However, we detected 2 mutations that map to the PDZ domains, one human mutation, p. Gln808His, and one mouse mutation, p.Glu814Gly, and that acted in a pathogenic manner in all three functional assays. While these two mutations do affect the interaction of Scrib1 with Vangl2, we hypothesize that this effect is not sufficient for pathogenesis and that these two mutations either lead to major conformational changes of SCRIB1 or affect its interactions with other unknown proteins essential for its normal localization and function.
Materials and Methods
Maintenance, phenotyping and genotyping of mice
Circletail mice were obtained from the Jackson laboratory following cryorecovery (stock number 006130, CBACa.Cg-ScribCrc/RachJ). A total of seven females (two Scrib1Crc/+ and five wild-types) and two males (one Scrib1Crc/+ and one wild-type) were obtained and used for establishing the mouse colony. The colony was maintained by brother-sister matings and by crosses to CBA/CaJ mice every third generation. Mice were examined macroscopically for the presence of a severely “kinked” or “looped” tail. Embryos were recovered at various stages and examined for the presence of NTDs. Genomic DNA was extracted from mouse tail or embryonic yolk sac using the EZ-10 Spin Column Animal DNA Mini-Preps kit (BS628, Bio Basic, Inc., ON, Canada). Genotypes were confirmed for all mice included in the study. For the Scrib1Crc mutation, genotyping was done by PCR amplification on genomic DNA for Scrib1-exon 21 (183 bp).
The protocol relevant to support this study was approved by the Institutional Committee for Animal Care in Research of the Research Center of Sainte Justine Hospital.
Immunofluorescence labelling of mouse NE tissues
Mouse embryos at E11.5 were post-fixed in 4% PFA/PBS and cryo-protected in 30% sucrose/PBS, embedded in OCT compound and finally sectioned with the cryostat (14 μm) and mounted on culture slides. Slides were incubated in blocking solution with primary antibody overnight at 4°C with the following antibodies: rabbit polyclonal anti-Scrib1 (SANTA CRUZ sc-28737, 1:400), mouse monoclonal anti-Lgl2 (SANTA CRUZ sc-376857, 1:400), Rabbit monoclonal anti-β-Catenin (Abcam ab32572, 1:250), Mouse Monoclonal anti-ZO-1 (Invitrogen 339100, 1:100), goat polyclonal antibody anti Vangl2 (SANTA CRUZ sc-46561, 1:50), Alexa Fluor 488-conjugated phalloidin (Thermo Fisher Scientific A12379, 1:40) and rabbit polyclonal anti-Par-3 (Millipore 07-330, 1:400). After extensive rinsing in PBS, slides were incubated with the following secondary antibody: goat anti-rabbit Alexa fluor 555 (Thermo Fisher Scientific A-21428, 1:400) for Par-3 and Scrib1, donkey anti-goat Alexa fluor 555 (Thermo Fisher Scientific A-21432, 1:400) for Vangl2 and goat anti-mouse Alexa fluor 555 (Thermo Fisher Scientific A-21422, 1:400) for Lgl2 and ZO-1 for 1.5 h. A final incubation was done with DAPI (300 nM)/PBS for 5 min, coverslips were mounted with VECTASHIELD® Mounting Medium (Vector, H1000). Images were acquired with a confocal microscope (Zeiss LSM 710 Axiovert 100 M inverted confocal microscope and Leica, SPE, upright microscopes), using a 63× glycerol objective.
For quantification of protein expression in neural tissues, image analysis was carried out with Image J software. Briefly, mean fluorescence density was calculated for both the apical region (facing the lumen) and the subapical region (taken at a distance of 10–70 μm from the apical region) of the neural tube. After subtracting background, the mean fluorescence density was quantified in five apical or subapical histological sections per embryo. A total of three embryos from each genotype (wild-type or mutant) were analyzed and the average fluorescence was expressed in arbitrary units. Statistical significance was determined using Student’s t-test and was set at P < 0.05. An example of this kind of analysis is shown in Supplementary Material, Figure S8.
Plasmid constructs
For the Y2H experiments, hVANGL2 cDNA segment corresponding to its cytoplasmic domain (Accession # Q9ULK5; nt 720–1560) was cloned into the yeast plasmid pGBKT7 and the cDNA corresponding to the PDZ domains (nt 2181–3585) of SCRIB1 (GenBank Accession No. NM_015356) was cloned into the yeast plasmid pGADT7 (hSCRIB1-PDZ-pGADT7) using the SLIC method. The mutations p. Gln808His, p.Arg.1150Gln and p.Glu814Gly were introduced into the hSCRIB1-PDZ-pGADT7 using site-directed mutagenesis. For the expression constructs, wild-type full length hScrib-GFP was kindly provided by Dr W.J. Van de Ven (University of Leuven), ScrbKD constructs, Par-3-KD and hPar-3 vector were kindly provided by Dr Ian G. Macara (University of Virginia School of Medicine) and finally the HA-Vangl2 was kindly provided by David D. Ginty (Johns Hopkins University). All the five SCRIB1 mutations were introduced in the hScrib-GFP construct by site-directed mutagenesis. Primers sequences for this assay are available upon request. All constructs were verified by Sanger sequencing.
Cell culture, transfection and immunofluorescence
The MDCK II cell line stably transfected with c-Myc/HA-tagged Vangl1 (subcloned in peGFP-C1) was kindly provided by Dr Philippe Gros (McGill University, Montreal, Canada). For the knockdown studies, Vangl1 stable transfectant MDCK II cells were grown in DMEM (Dulbecco’s modified Eagle’s medium) supplemented with 10% fetal bovine serum, 100 units/ml penicillin, and 100 μg/ml streptomycin. Cells (106) were transiently transfected in suspension by electroporation (Gene Pulser Xcell™ Electroporation Systems), using 5 μg ScrbKD or 5 μg Par-3 KD. For rescue experiments, 5 μg ScrbKD was co-transfected with either 5 μg wild-type or with mutant hScrib-GFP or HA-Vangl2 or hPar-3. A total of 5 μg Par-3 KD was co-transfected with 7 μg hPar-3. For localization studies of hScrib-GFP, MDCK II cells were grown on coverslips for 24 hrs before transfection. Transfection with 5 μg wild-type or mutant hScrib-GFP was performed using Lipofectamine 2000 reagent (Invitrogen), according to the manufacturer’s instructions. Cells were examined for localization of hScrib-GFP at 48 h post-transfection.
After electroporation, 106 MDCK II cells were plated on glass coverslips in a 10-mm culture plate. Thirty-six hours later, coverslips were fixed in; 4% formaldehyde/PBS (phosphate-buffered saline) for 20 min permeabilized by 0.1% Triton X-100/PBS for 10 min and blocked in 2% goat serum/PBS for 60 min (for Par-3). For Scrib1, Lgl2, β-Catenin, ZO-1 and VANGL1-HA, coverslips were fixed in 4% PFA (1 min) on ice and permeabilized with ice cold 100% methanol for 10 min and rehydrated in PBS for 5 min and blocked in 1% BSA/PBS for 30 min at RT. Coverslips were incubated with the primary antibody in a blocking solution overnight at RT for Par-3 and VANGL1-HA and 1 h for the others. Primary antibodies were as described in the immunofluorescence studies of mouse neural tissues. For VANGL1-HA, mouse monoclonal anti-HA (abcam ab1424, 1:200) was used. After extensive rinsing in PBS and incubation with appropriate secondary antibody for 1.5 h, coverslips were mounted with VECTASHIELD® Mounting Medium (H1000). Images and z-stacks were acquired with a confocal microscope (Zeiss LSM 710 Axiovert 100M inverted confocal microscope and Leica, SPE, upright microscopes), using a 63× glycerol objective. Apical and basolateral sides of MDCK II cells were analyzed using confocal Z stacks acquired at 0.5 μm intervals.
For all localization studies, the transfection was performed three times for each construct and the identity of the construct (wild-type or mutant) was blinded to the experimenter. For subcellular localization of hScrib1-GFP, a total of 100 cells were examined per transfection. For apical and basolateral localization studies, a total of 45 cells were examined per transfection (3 different areas were selected and 15 cells were examined per area).
SDS-PAGE and western blotting
MDCK II cells harvested 48 h post-transfection were washed in PBS and lysed in RIPA Buffer (50 mM Tris, 150 mM Nacl, 0.1% SDS, 0.5% Na-deoxycholate, 1% Triton (X-100), 1 mM EDTA × complete protease inhibitor cocktail, in PBS) for 30 min. Whole E11.5 wild-type, heterozygous and homozygous Crc mouse embryos were washed briefly in PBS and lysed in RIPA buffer (150 µl). Lysates were then centrifuged for 1 min at 17 000 rpm and the supernatant was collected and kept at −80 until use.
A total of 40 μg of protein extracts was boiled in loading buffer, separated by SDS-PAGE, and transferred during 45 min on a PVDF membrane. Membrane was blocked in 5% milk/TBS for 30 min and incubated for 1 h 30 min at room temperature with the primary antibodies against Scrib1 (1:400), HA-Vangl1 (1:5000), Par-3 (1:1000), Lgl2 (1:500), β-Catenin (1:5000), ZO-1 (1:1000) and GFP (Life Technologies A6455, 1:200) in 5% milk/TBS/0.1% tween 20. The membrane was washed in TBST, followed by a 1 h incubation at room temperature with the appropriate secondary HRP-conjugated antibody (Abcam, Cambridge, MA, USA), followed by 3 ×5 min washes with TBST. Finally, the signal was detected by chemiluminescence on a film using Amersham ECL prime western blotting detection reagent (GE Healthcare).
Patients and controls
Our cohort consisted of 473 patients, including 391 patients recruited from the Spina Bifida Center of Gaslini Hospital in Genova, Italy and 82 patients recruited from the Spina Bifida Center of Sainte Justine Hospital in Montreal, QC, Canada. Clinical details of our cohort are as described in De Marco et al. (38). Briefly, 97% of this cohort were white Caucasians, 44% were male and the average age was 2.6 years. The distribution of the major NTD subtypes was 3% cranial, 52% open spinal (of which 99% were MMC) and 45% closed spinal NTDs. We also used a control cohort of 467 individuals consisting of children randomly selected and admitted to Gaslini Hospital for diseases other than NTDs, and young adults in good health who have contributed to the blood bank of the Gaslini Institute. All blood or saliva samples from patients and controls were collected with the approval of the local ethics committees and all patients, parents and control individuals gave their written informed consent.
Sequencing of SCRIBBLE1
The genomic structure of SCRIB1 was determined using the NCBI (GenBank Accession No. NM_015356) and Ensembl (SCRIB1 Transcript ID: ENST00000356994). Our sequencing analysis of this gene focused on its coding region (4965 bp) composed of 37 exons. Primers were developed manually to amplify this coding region along with the exon–intron junctions in a total of 21 amplicons. Sequencing was done using the Big Dye Terminator Ready Reaction Mix (ABI, Carlsbad, CA, USA) at the McGill and Genome Quebec Innovation Center sample and obtained sequences were analyzed using the SeqMan sequence assembly and SNP Discovery software (DNASTAR, Madison, WI, USA). All rare mutations (minor allele frequency <1%) discovered in patients were confirmed by a new PCR. Confirmed mutations were queried against two public databases: dbSNP (http://www.ncbi.nlm.nih.gov/snp) and the Exome Variant Server (http://evs.gs.washington.edu/EVS/). Two software programs: PolyPhen (Polymorphism Phenotyping: http://genetics.bwh.harvard.edu/pph/) and SIFT (Sorting Intolerant from Tolerant; http://sift.jcvi.org/) were used to predict the potential pathogenic effect of the identified mutations on the Scib1 protein function. CLUSTAL W method (ExPASY; http://www.ch.embnet.org/software/ClustalW.html) was used for multiple alignments of the various Scrib1 orthologs.
Yeast two hybrid assay
The commercially available yeast two hybrid system MatchMaker system 3 (#K1612–1, Clontech) was used to study the interactions between hVANGL2-Cterm and wild-type or mutant hSCRIB1-PDZ. The hVANGL2-Cterm-pGBKT7 and each of the wild-type or mutant hSCRIB1-PDZ-pGADT7 constructs were introduced into Saccharomyces cerevisiae yeast strain AH109 (MATa) strain and Y187 (MATα) strain (hSCRIB1-PDZ-pGADT7 in AH109 and hVANGL2-Cterm-pGBKT7 in Y187). In this system, AH109 strain will express hSCRIB1-PDZ protein with the activation domain and the Y187 will express the hVANGL2-Cterm with the binding domain. Saccharomyces cerevisiae AH109 and Y187 clones were mated and the GAL4 activity was tested as previously described in the matchmaker system 3. Reconstituted GAL4 activity (interaction between fusions to the GAL4 DNA binding and activation domains) is tested by plating diploid cells on different selection media: -Leu/-Trp (low stringency lacking leucine and tryptophan) double drop out (DDO), -His/-Leu/-Trp (intermediate stringency lacking histidine, leucine and tryptophan) and triple drop out (TDO). Protein–protein interactions were evaluated by assessing growth of diploid cells in solid medium after 96 h at 30°C. To confirm our results, diploid cells were grown in DDO liquid medium to test for the secretion of α-galactosidase.
Statistical tests
Each experiment was repeated 3 times. Statistical significance between experiments and between groups was calculated with the z-test for 2 population proportions. For the fluorescence density measurements and the Y2H quantification assay, statistical significance was calculated using Student’s t-test. The significance threshold was set at 0.05.
Supplementary Material
Supplementary Material is available at HMG online.
Acknowledgements
We thank all patients who made this study possible. We thank Abdullah Al Mahmud and Patricia Awad for technical help in immunofluorescence studies and Dr. Guy A. Rouleau and Patrick A. Dion for the use of their confocal microscope. We also thank Ruiu Ilaria and Mara Uglietta for technical assistance. We are grateful for Drs WJ, Van de Ven, Ian G. Macara and David D. Ginty for providing constructs and Dr P. Gros for providing MDCK II cell lines stably transfected with c-Myc/HA-tagged Vangl1. We thank Farida Kibar for her help in statistical analyses.
Conflict of Interest statement. None declared.
Funding
This work was supported by the Canadian Institutes for Health Research and the Natural Sciences and Engineering Research Council of Canada (NSERC). Z.K. was funded by the ‘Fonds de Recherche du Québec – Santé’ and PDM and EM have been funded by Trust Aletti and Fondazione Gerolamo Gaslini.
