Early loss of Scribble affects cortical development, interhemispheric connectivity and psychomotor activity

Neurodevelopmental disorders arise from combined defects in processes including cell proliferation, differentiation, migration and commissure formation. The evolutionarily conserved tumor-suppressor protein Scribble (Scrib) serves as a nexus to transduce signals for the establishment of apicobasal and planar cell polarity during these processes. Human SCRIB gene mutations are associated with neural tube defects and this gene is located in the minimal critical region deleted in the rare Verheij syndrome. In this study, we generated brain-specific conditional cKO mouse mutants and assessed the impact of the Scrib deletion on brain morphogenesis and behavior. We showed that embryonic deletion of Scrib in the telencephalon leads to cortical thickness reduction (microcephaly) and partial corpus callosum and hippocampal commissure agenesis. We correlated these phenotypes with a disruption in various developmental mechanisms of corticogenesis including neurogenesis, neuronal migration and axonal connectivity. Finally, we show that Scrib cKO mice have psychomotor deficits such as locomotor activity impairment and memory alterations. Altogether, our results show that Scrib is essential for early brain development due to its role in several developmental cellular mechanisms that could underlie some of the deficits observed in complex neurodevelopmental pathologies.

showed a role for Scrib in the brains of adult mice in fine tuning of excitatory synapses and correlated its deletion with some features of ASDs [17][18][19] .
In humans, the absence or mutation of the SCRIB gene is linked to neurodevelopmental disease. Mutation in SCRIB is associated with NTDs (OMIM # 182940) [20][21][22][23] , like other core PCP genes 24,25 . NTDs represent a CNS (Central Nervous System) congenital malformations that affect 1:1000 children 26 . Spina bifida (spinal NTD) remains the commonest congenital CNS defect and is often associated with cerebellum, corpus callosum abnormalities and hydrocephalus 27 . As a result, patients with spina bifida face neurobehavioural alterations that include psychosocial, memory and motor defects 28 . In addition, microdeletions in 8q24.3 (encompassing both PUF60 and SCRIB) were found in children presenting microcephaly 29 . 8q24.3 deletion syndrome, also called Verheij syndrome (VRJS, OMIM #615583), is characterized by complex features such as growth retardation, short stature, dysmorphic facial features as well as renal and cardiac defects 29,30 . The neurological symptoms of this syndrome include delayed psychomotor development, mild intellectual disability and epilepsy that are frequently associated with neurodevelopmental defects such as microcephaly and/or ACC 29,30 . Patients with a large deletion on chromosome 8 show microcephaly and ACC 31 , and a more specific deletion of the 8q24.3 region is associated with ASDs or ADHD 32 . VRJS is considered as a contiguous gene syndrome because it results from the haploinsufficiency of the SCRIB and PUF60 genes, which are located within the minimal critical region 29 . The best support for Scrib participating in VRJS came from morpholino-based knockdown experiments in zebrafish showing altered brain size in either scrib or puf60 morphants 29 . While all these data suggest a role for Scrib in early vertebrate brain development, the potential contribution of Scrib to some of the observed structural and psychomotor deficits has never been assessed in mammals.
To evaluate the structural and psychomotor consequences of early Scrib deletion in the mammalian brain, we developed a conditional gene-targeting strategy to generate two mouse lines. The results show that early deletion of Scrib in the dorsal telencephalon lead to (1) reduced brain cortical size associated with cortical layering defects (2) agenesis of the corpus callosum (CC) and the hippocampal commissure, stemmed from a combination of neurogenesis and neuronal migration defects. Behavioral analysis of Scrib cKO animals show altered psychomotor behavior accompanied by memory deficits. Altogether, our results show that the absence of Scrib during early brain development lead to differential structural and behavioral deficits, often observed in spina bifida or VRJS patients, supporting a role for this gene in these pathologies.

Results
Scrib expression is consistent with a developmental function in the forebrain. We evaluated the expression profile of Scrib in the developing brain using in situ hybridization (ISH) (Fig. 1A) and an inhouse Scrib antibody for immunohistochemistry ( Fig. 1B-E). At E16.5, Scrib ISH indicated that it was expressed throughout the dorsal forebrain and enriched in specific regions such as the upper layers of the cortical plate (CP) along the medio-lateral axis (from the cingulate to the piriform cortex), in the sub-ventricular and ventricular zones (SVZ and VZ, respectively) and the Indusium Griseum (IG) at the cortical midline ( Fig. 1A-A″). Right before the neurogenesis peak, at E13, Scrib in expressed in most neural Nestin-positive progenitors throughout the cortical plate and is enriched in the VZ (Fig. 1B). Scrib immunostaining at E16.5 is consistent with its mRNA expression pattern, with an enrichment in the CP and VZ at cell-cell junctions in both apical (labeled by Pals1) and basal domains (Fig. 1C). At birth, Scrib is also found in axonal tracts, especially in the interhemispheric commissural fibers (Fig. 1D). Double staining of Scrib (green) and GFAP (glial fibrillary acidic protein, a marker of mature glial cells, red) revealed that Scrib is also enriched in midline glia cells (Fig. 1E).
Early Scrib deletion results in microcephaly. Scrib spontaneous mouse mutant Circletail causes severe brain and neural tube damages that result in neonatal lethality 8 , precluding the analysis of the role of Scrib during forebrain development 17 (Fig S1). In order to circumvent this issue, we have applied a conditional genetargeting strategy to inactivate Scrib at different developmental stages and in different cellular types in the brain. We crossed floxed Scrib fl/fl mice with either Emx1-Cre mice, which express the Cre recombinase starting at E10.5 in the dorsal telencephalic progenitors, or FoxG1-Cre mice, which express the Cre recombinase as early as E8.5 in the entire telencephalon (hereafter reported as Emx1-Scrib −/− and FoxG1-Scrib −/− cKOs respectively). Specific Scrib excision in conditional mutants was validated ( Fig. 2A-F), and the spatial expression pattern of Cre recombinase was further confirmed by crossing Emx1-Cre mice with the Ai6 reporter mice (Fig. 2G-H). The cerebral hemispheres of Emx1-Scrib −/− cKOs were 7.8% ± 0.5% smaller than those of their control littermates, a mild but significant reduction (Fig. 3A). Histological brain analysis at the rostral and caudal levels, revealed a marked decrease in the thickness of the caudal cerebral cortex (~ 25% throughout the mediolateral axis) (Fig. 3B). This decrease was observed throughout the cingulate (Cg), motor (M), primary somatosensory (S1) and secondary somatosensory (S2) cortices and was maintained in adults (data not shown). This phenotype was absent in more rostral regions of Emx1-Scrib −/− cKO brains (Fig. 3C). FoxG1-Scrib −/− mutant brains displayed a similar but more severe reduction of cortical area and thickness along the entire rostrocaudal axis (Fig S2A-D). We next sought to identify the cellular roots of the microcephaly deficits observed in Scrib mutant mice.

Loss of Scrib disrupts cortical layering.
To evaluate the consequences of Scrib deletion on cortical layering, we examined the expression of neuronal markers such as CuxI (layer II/III), Satb2 (layer II/VI) and Ctip2 (layer V), some of these markers being essential to the formation of the CC 33 . In Emx1-Scrib −/− mutant P0 pups, we observed a reduction in the numbers of CuxI-positive cells (ctrl, 28.2% ± 4.8; mutant, 18.5% ± 3.7; p = 0.04) and Satb2-positive cells (ctrl, 46.6% ± 2.4; mutant, 32.7% ± 3.5; p = 0.002) (Fig. 3E,F). This was particularly pronounced in bins 2-4, corresponding to layers II-III. In contrast, the expression of Ctip2, a marker of early-born neurons, was unchanged (ctrl, 22  and Nestin (a general marker of neural progenitor cells, red) expression pattern by IF on E13 mouse embryo coronal sections. The dashed boxes in (B) are magnified in (B′), then within the insets. Scrib protein is detected throughout the cortical plate (CP) and is enriched in the VZ. Scale bar 0.1 mm (B), 0.05 mm (B′). (C) Representative Scrib (green) and Pals1 (an apical surface marker, red) expression pattern by IF on E16.5 mouse embryo coronal sections. The higher magnification illustrates Scrib accumulation at cell-cell contacts in the entire VZ, next to the apical marker Pals-1 (red). Scale bar 0.1 mm. (D,E) Representative Scrib (green) expression pattern by IF on P0 cortex (C), Corpus Callosum (CC) (D) and midline glia (E). Scrib labeling overlaps with Ephrin-B2 (marker of the CC, red in (D) but is also markedly enriched in GFAP-positive (red in E) structures including the indusium griseum (IG) and the midline zipper (MZ). Cg is the cingulate cortex. Higher magnification for selected insets (boxed areas) illustrates strong Scrib expression in glial midline structures (see arrowheads). Scale bar 0.1 mm (D,E). www.nature.com/scientificreports/ many Ctip2-positive neurons were mislocalized in the deeper layers of the motor cortex (bins 8-10), implying migration defects (Fig. 3G,H). Similar results were observed within the somatosensory cortex (data not shown). A comparative analysis in FoxG1-Scrib −/− mutant mice showed a more severe phenotype than that of Emx1-Scrib −/− mutants, with a dramatic reduction in CuxI, Satb2 and Ctip2 expression (Fig S2E-H). Using in utero electroporation approach employing a previously validated shRNA construct 17 , we depleted Scrib in E14.5 brain embryos, and observed a significant increase in the fraction of electroporated cells mislocalized in the VZ/SVZ (ctrl shRNA, 5.4% ± 0.5; Scrib shRNA, 15% ± 1.4) associated with a significant decrease in the fraction of cells reaching the upper layers (UL) (ctrl shRNA, 47.1% ± 1.4; Scrib shRNA, 30.7% ± 1.9) (Fig. 3I,J). Altogether, these results indicate that embryonic deletion of Scrib alters the neuronal composition of the cortical layers of the brain of the mutant mice, a phenotype that could stem from a decrease in neurogenesis and/or a reduction in neuronal migration.

Proliferation defects in cortical progenitors in Emx1-Scrib −/− mutant.
To assess potential progenitor proliferation deficits, we examined the numbers of both apical (using Pax6 as a marker) 34 and basal progenitors (using Tbr2 as a marker) 35 in cortical sections of E13 control and mutant littermates. We found a significant decrease in the numbers of Tbr2-positive neuronal progenitor cells in the VZ (ctrl, 28.6% ± 2.4; mutant, 22.0% ± 5.3; p = 0.027) and Pax6-positive intermediate basal progenitors in the SVZ (ctrl, 65.8% ± 4.5; mutant, 57.5% ± 5.3; p = 0.015) in the Emx1-Scrib −/− mutant neocortex (Fig. 4A,B). We next used an antibody against Ki67 to assess the levels of proliferation in absence of Scrib. As shown in Fig. 4C, the number of Ki67-positive cells was also slightly reduced in cKO neocortices (ctrl, 47.7% ± 2.7; mutant, 41.9% ± 4.3; p = 0.021), suggesting a mild proliferation decrease of progenitors. Finally, we evaluated apoptosis and did not detect significant variation of cell death in in Emx1-Scrib −/− mutant cortices (Fig. 4D). Together, these results demonstrate that the loss of Scrib lead to a mild decrease in progenitors proliferation, leading to a small reduction of neuronal progenitor populations in the VZ/SVZ, but has no impact on apoptosis levels.
Early deletion of Scrib affects corpus callosum and hippocampal commissure development. Histological analysis was performed along the rostrocaudal axis in order to assess the impact of early Scrib loss on the major forebrain commissure (the CC) but also the dorsal and ventral hippocampal commissure (DHC and VHC) and the anterior commissure (AC) (Fig. 5). Coronal histological sections of P0 Emx1-Scrib −/− mutant brains revealed ACC in caudal sections, with callosal axons apparently unable to cross the midline (Fig. 5A,B). Failure of the axons to cross was accompanied by bundles of aberrantly projecting axons near the midline, known as Probst bundles, which are frequently associated with hemispheric fusion defects ( Fig. 5A′-B′). Anterograde axonal tracing studies using DiI staining confirmed the absence of midline-crossing callosal axons ( Fig. 5A″-B″). In contrast, callosal fibers in rostral sections from the brains of Emx1-Scrib −/− mice crossed the midline despite an apparent CC hypoplasia ( Fig. 5C-D″). Of note, a similar phenotype was observed in adult Emx1-Scrib −/− mutant mice using 3D light-sheet microscopy of uDISCO treated brains injected in the caudal cortex with an adenovirus encoding GFP (Fig. 5E,F′) suggesting that ACC is not due to a developmental delay. The deletion of Scrib in the FoxG1-Cre mice led to ACC along the entire rostrocaudal axis (Fig S3). Since Scrib is expressed in glial midline structures (Fig. 1E) that are critical for promoting hemisphere fusion and allowing the CC to cross between hemispheres with the help of axonal guidance signals 36 , we next sought to determine whether CC agenesis observed in Emx1-Scrib −/− mutant mice results from mislocalization of glial structures. Immunostaining of P0 coronal sections for GFAP and the axonal marker L1-CAM revealed that early deletion of Together with the previous results, these findings suggest that the commissural deficits observed in absence of Scrib are due to a reduction in the number of projecting neurons in layer 2/3 or the cortex, and/or a disruption of the midline glial organization.

Early loss of Scrib induces hyperlocomotion and memory defects.
To determine the behavioral consequence of the early loss of Scrib we next subjected our mutant model to a variety of tests that require sensory-motor, emotional and cognitive integration. Adult Emx1-Scrib −/− cKO and their control littermates (10-20 weeks) were submitted to tests for anxiety-, locomotor and exploratory activities. The most remarkable behavior phenotype we observed was the effect on locomotor activity. Compared with their controls, Emx1-Scrib −/− mutant mice showed significantly increased activity in the open field (t test: t 17 = 2.35, P < 0.05*; Fig. 7A). The time spent in the center of the arena was not different between genotypes, suggesting that the Emx1-Scrib −/− mutant mice has comparable anxiety level to control littermates (t test: t 17 = 0.4595, n.s; Fig. 7A). Anxiety-like behavior was further examined in elevated plus-maze (Fig. 7B), where Emx1-Scrib −/− and control mice showed comparable performance, confirming that Emx1-Scrib −/− mice have no anxiety-related behaviors (t test: t 17 = 1.741, n.s; Fig. 7B). In addition, we confirmed that the Emx1-Scrib −/− mutant mice were significantly more active in elevated plus-maze (t-test: t 17 = 3.02, P < 0.01**; Fig. 7B), in Y maze (t-test: t 17 = 2.39, P < 0.05*; Fig. 8A), and in a new cage as assessed by actimetry during a 2 h period (genotype effect: F 1,13 = 8.757, P < 0.05*; Fig. 7C). In addition, Emx1-Scrib −/− cKOs mutant mice displayed enhanced locomotor activity in their home cages during nycthemeral activity compared to the control mice, as determined by a 24-h continuous monitoring of locomotor activities (genotype effect: F 1,13 = 6,802, P < 0.05*; Fig. 7D). Hyperactivity of Emx1-Scrib −/− cKOs mutant mice was associated with a weight loss (t test: t 17 = 2.412, P < 0.05*; Fig. 7E). Notably, Emx1-Scrib −/− mice did not display changes in balance, as tested on beam walking (Fig. 7F) and grid handling (Fig. 7G), as well as motor coordination as tested on the accelerating rotarod (Fig. 7H).
Considering that cKOs mice had impaired hippocampal and cortical connectivity, we examined whether the Emx1-Scrib −/− mice were impaired in different types of learning and memory tests. In response to novelty, Emx1-Scrib −/− mutant mice showed a decrease in exploratory activity over time as the context loses its novelty, and ended up no different from control mice, suggesting that this simple form of spatial recognition is preserved  Fig. 8A). To further probe hippocampus-dependent memory, we next analyzed the effect of early Scrib loss on spatial learning and memory using the Morris water maze test. Mice were trained to learn spatial cues around the maze to find a hidden platform under the water during training sessions (training days: F 9,135 = 17.59, P < 0.0001***; Fig. 8B). After training the mice for 15 days and preformed probe tests at day 7, 9 and 16, we observed that Emx1-Scrib −/− mice showed significantly longer latency to find the platform during training sessions, demonstrating that the spatial learning is impaired in the mutants (genotype effect: F 1.15 = 5.53, P < 0.05*; Fig. 8B). Furthermore, in the reversal test the Emx1-Scrib −/− mutant mice showed a delay of spatial learning (interaction effect: F 4.60 = 4.335, P < 0.01**; Day2 t test: t 13 = 3.2, P < 0.01**; Fig. 8B). In all probes test, Emx1-Scrib −/− mice show normal hippocampus-dependent memory. Importantly, although the Emx1-Scrib −/− mutant mice were hyperactive, swimming speeds during the probe tests were not different between genotypes. Finally, hippocampus-dependent context and amygdala-dependent tone associative memory was assessed by using classical fear conditioning paradigm in which the animals have to associate environmental cues to an electric shock (Fig. 8C). Two different groups of Emx1-Scrib −/− and control mice were re-exposed to the same environment 24 h or 7 days after training, respectively showing the recent (24 h) and the remote (7 days) contextual fear memory (Fig. 8D,E). During the acquisition phase (without shock), all groups of mice displayed a normal freezing behaviour and the first 3 min of acquisition period was considered as the baseline period (basal). Every group exhibited an increase in freezing during the context and tone presentation tested 24 h (test effect: F 1.28 = 75.69, P < 0.0001***; Fig. 8D) and 7 days after training (test effect: F 1.28 = 75.82, P < 0.0001***; Fig. 8E). The freezing levels were comparable between genotypes when the memory was tested 24 h after training, showing that the recent contextual and tone fear memories are intact in Emx1-Scrib −/− mice (no genotype effect: F 1.56 = 1.351, n.s; Fig. 8D). When mice were tested 7 days after training, Emx1-Scrib −/− mice froze less than their control littermate during the tone indicating an acceleration of the extinction of the long term cued fear memory (genotype effect: F 1.56 = 8.459, P < 0.01**; Fig. 8E). Here, Emx1-Scrib −/− mutant mice exhibited no significant prolonged latency to paw to licking/jumping in the hot plate test than controls indicating normal nociceptive reactivity (no genotype effect: F 134 = 0.37, n.s; Fig. 8F). Altogether, these results support the idea that Emx1-Scrib −/− mutant mice display hyperactivity in novel and familiar environments, which is compatible with the altered psychomotor behavior observed in VRJS patients (see "Discussion"). In addition, the remote memory deficits we observed in this mouse model have not been reported in patients with VRJS and should guide memory evaluation in older patients.

Discussion
Our present study demonstrates that Scrib is essential for embryonic brain development and function. We showed that early deletion of Scrib leads to microcephaly and cortex layering defects associated with corpus callosum and hippocampal commissure agenesis. Behavioral analysis showed an increased locomotor activity accompanied by memory defects as a consequence. Our integrative work supports the participation of Scrib to various congenital neurodevelopmental deficits. Both conditional mutants used in this study and with early deletion of Scrib (FoxG1-Scrib −/− and Emx1-Scrib −/− mutant mice) display microcephaly in a range that is comparable to well-established microcephalic www.nature.com/scientificreports/ mouse models 37 . Microcephaly observed in mutants with embryonic Scrib deletion is most likely the result of a combination of disrupted neurogenesis and migration processes, as deficits in any of these critical mechanisms can participate in brain malformation 2 . This multifactor effect of Scrib is consistent with its expression profile. It is expressed throughout the cerebral cortex, both in neuronal progenitors and in the radial glia, also supporting a role in migration. In addition, we also observed Scrib enrichment in the VZ and SVZ at the time of neurogenesis that prompted us to analyze the impact of Scrib loss on progenitors. We observed in Emx1-Scrib −/− mutant mice a decrease in Ki67-positive proliferating cell population (− 12.2% ± 4.9) associated with a reduction of neuronal progenitor populations: Pax6 + cells (− 12.4% ± 6.0), Tbr2 + cells (− 23.3% ± 6.9). We believe that these reductions www.nature.com/scientificreports/ in progenitor populations, which generate cortical neurons, are responsible for the thinning of the cerebral cortex and associated layering alteration. Finally, we cannot rule out the possibility that it also contributes to asymmetric cell division (ACD), as observed in other systems 6,7 , but we can rule out apotosis deficits. The other main phenotype we observed after early Scrib deletion was the CC and DHC agenesis. These commissural defects may arise from neurogenesis, migration, and/or axonal outgrowth/guidance alterations. Scrib inactivation resulted in complete ACC in both rostral and caudal domains in FoxG1-Scrib −/− cKOs, while Emx1-Scrib −/− cKOs display ACC only in the caudal part. The most likely explanation for this restricted phenotype at the caudal level is that Emx1 is expressed on a high-caudal to low-rostral gradient 38 . In the rostral telencephalon, Scrib-dependent ACC is associated with the formation of abnormal swirls of axons called Probst bundles 36 . Such phenotype is associated with failure of the callosal axons to cross the midline rather than an outgrowth problem towards the midline. Although we cannot completely exclude an autonomous role of Scrib in axon guidance as shown during axonal misrouting at the chiasm 39  www.nature.com/scientificreports/ expression in the midline glia and the disorganization of that structure in Emx1-Scrib −/− mutant mice suggest that a non-autonomous Scrib-dependent mechanism affects the crossing of the axons of the CC. Mutations in guidance cues expressed by the midline glia typically lead to a Probst bundles phenotype; thus, we can envision that Scrib deletion affects midline glial cells maturation and/or guidance cues secretion 41 as seen in other systems 42,43 . Alternatively, by impacting these astroglial structures that are essential during interhemispheric remodeling 44 , Scrib may promote tissue fusion 45,46 , which in turn allow the callosal fibers crossing. Tissue fusion appears as a common theme during neural tube closure 47 and during interhemispheric remodeling in CC formation 48 . Of note, one third to half of patients with spina bifida also have CC abnormalities 27 that may explain some behavioral deficits due to improper interhemispheric transfer of information 28 . By impairing neurogenesis, migration and commissure formation, we show that the absence of Scrib during development causes profound defects in the cortical layering and interhemispheric connectivity that underlie cognitive disabilities that are typical of neurodevelopmental disorders and some human syndromes. Edwards and collaborators discriminate two categories of human syndromes: on one hand the ones that display only a microcephalic phenotype and, on the other hand, the ones that do encompass both ACC and microcephaly 3 . SCRIB may also fall into the latter category, reflecting a concomitant function for this gene in cortical organization and axonal guidance during development. Our results support the theory that the microcephaly and ACC observed in VRJS patients is the result, at least partially, of SCRIB haploinsufficiency.
Behavioral analysis of the Scrib cKO mice revealed alterations in psychomotor behavior. Specifically, Emx1-Scrib −/− mutant mice had an increased locomotor behavior that is comparable to established hyperactive models 49 . Because of the early and broad expression of Emx1 in the brain, the pathological origin for the altered locomotor behavior may stem from the cortex and/or the hippocampus, through their connections. Exploration of cognitive performance shows that Emx1-Scrib −/− display memory deficit for a cortex-dependent remote-cued fear memory recall. The CC primary function is to integrate motor, sensory, and cognitive activity between the two hemispheres 3 . The DHC provides interhemispheric connections between hippocampi and despite few studies of the function of the DHC, the ability of the hippocampus to communicate effectively with contralateral homologous regions via the DHC may be important for cognitive performance, compounded by the other deficits observed in the mutants. This lack of communication and/or the microcephaly and/or layering defects could be the cause of these cognitive deficits. At the molecular level, it is tempting to speculate that Scrib may control some aspects of brain architecture and behavior, through known partner such as GIT1 50 or Vangl2 51 . Similar to our model, mice lacking GIT1 have microcephaly 52 and display a hyperactivity phenotype combined with learning and memory deficits 53 . Like Emx1-Scrib −/− mice, Emx1-Vangl2 −/− display partial hippocampal and CC agenesis (but no microcephaly) that is caused by abnormal axonal outgrowth 54 . However, a conditional deletion of Vangl2 in postmitotic hippocampal granule cells does not alter spatial memory 55 , highlighting the complexity of such molecular integration at the behavioral level. Additional work from our group support a critical role for Scrib in synaptic dysfunction and human psychiatric disorders 13,17,19 , at least in part, by the regulation of the glutamatergic signaling 18 . Remarkably, the function of this pathway in the process of learning and memory seems to extend toward the invertebrate phylum where Scrib is pivotal to the regulation of active forgetting 56 . Future studies exploring these possibilities are needed to define the detailed molecular mechanisms.
Our findings uncover an essential role for Scrib in mammalian forebrain development and connectivity, both of which ultimately affect animal behavior. Given that several aspects of the neurological manifestations of VRJS were recapitulated in Scrib cKO mice, we suggest that Scrib may participate in this syndrome. The minimal common deletion found encompassed 3 genes including SCRIB and PUF60 and displayed most of the cardinal features of VRJS 29 . Although PUF60 appears to be a major driver of VRJS syndrome 29,57-64 , neurological features were reported in a much lower proportion, indicating that PUF60 CNVs may not be their sole cause. From the human genetics standpoint, any VRJS-related phenotype due to SCRIB loss may prove difficult to observe because most SCRIB mutations lead to NTDs so deleterious that they may obscure more "subtle" phenotypes [20][21][22][23] . The scarce number of NTD cases carrying SCRIB variants or VRJS patients implies the possibility that their true phenotype spectrum may be wider than indicated by publications providing few or no detailed neuropsychiatric evaluation. Patients with spina bifida tend to show altered cognitive abilities with language, memory, motor and psychosocial difficulties 28 . Neurological features for VRJS patients include mild intellectual disability, delay of developmental milestones such as standing upright and walking, delayed speech, feeding issues and generalized seizures 29,30 . All of these features, combined with neuroanatomical features such as microcephaly and CC agenesis can fall under the umbrella of a neurodevelopmental disorders whose affected individuals develop psychomotor deficits reminiscent as those observed in ASDs, ADHD and schizophrenia. Interestingly, a case for a patient with ADHD revealed a chromosomal translocation breakpoint at 8q24.3 has been reported 65 and this region appears to overlap with ASDs as well 32 . Further studies are warranted in order to determine whether other behavioral phenotypes such as seizure susceptibility or deficits of attention/impulsivity, are also recapitulated in Scrib −/− cKOs. Our study demonstrates that Scrib mutant mice can provide an entry point to study its forebrain contribution in neuroanatomical and behavioral deficits observed in NTD and VRJS patients. Future investigation using both heterozygous Scrib KO and Puf60 KO mice (alone or in combination) as a model is warranted to address their respective contribution and potential interaction in this syndrome at a more systemic level.

Materials and methods
Animal care and use. All mice were housed in the animal facility of the Neurocentre Magendie, in polypropylene cages under controlled conditions (at 23 ± 1 °C, with lights on from 7:00 A.M. to 7:00 P.M.). Food and water were available ad libitum. For timed pregnancy, the morning in detection of vaginal plug was designated as embryonic day E0. 5 Generation of brain-specific Scrib conditional knock-outs. Scrib spontaneous mouse mutant Circletail causes severe brain and neural tube damages that result in neonatal lethality 8 , precluding the analysis of the role of Scrib during forebrain development 17 . In order to circumvent this issue, we have applied a conditional gene-targeting strategy to inactivate Scrib at different developmental stages and in different cellular types in the brain. See Supplementary material and methods for details.
Histology, in situ hybridization analysis and immunofluorescence. For histology, heads (E16.5 embryos) or brains (P0 new-born) were harvested and fixed in Bouin's or Carnoy's fixative (Electron Microscopy Sciences) overnight as previously described 66 . Samples were dehydrated in ethanol, paraffin-embedded, and sectioned (20 µm) coronally, horizontally or sagitally. Sections were collected onto Superfrost plus Gold slides (Thermo Scientific), stained with hematoxylin and mounted with Entellan (Millipore). The brains sections were examined using Leica MZ-16 stereomicroscope, imaged in the Bordeaux Imaging Center (http:// www. bic.u-borde aux. fr) using the NanoZoomer 2.0-HT slide scanner and analyzed using the Hamamatsu NDP viewer software (Hamamatsu). For in-situ hybridization, E16.5 and E17.5 embryos brains were dissected out, transferred in OCT solution and placed into the dry ice for storage at − 80 °C before sectioning. In situ hybridization was performed using previously validated digoxigenin-labeled cRNA probes 17 . 16 µm coronal embryonic brain sections of were postfixed in 4% paraformaldehyde/0.2% glutaraldehyde for 10 min at room temperature (RT), bleached with 6% H 2 O 2 , digested in Proteinase K (5 µg/ml in PBS) for 2.5 min and postfixed in 4% paraformaldehyde/0.2% glutaraldehyde for 10 min at RT. Then, slides were acetylated into freshly prepared 0.1 M triethanolamine/PBS, pH 8, for 10 min at room temperature; 0.25% acetic anhydride acid was added for the last 5 min. Between each step, slides were rinsed with PBS. All subsequent steps were performed as previously described in 17 . Images were acquired with Nanozoomer (Hamamatsu Four weeks after surgery, the animals were perfused transcardially with PB followed by 4% PFA in PBS; the brains were removed and postfixed in 4% PFA for 24 h at 4 °C and maintained in PBS. The entire brains were cleared using the uDISCO technique as described 67 . The ultramicroscopy was performed using the system from LaVision BioTec (Bielefeld, Germany) equipped with a Fianium white laser, a sCMOS Andor camera, and a 0.5 NA 2X objective with a deeping lens. A zoom from 0.63 to 6.3 could be applied. Images and 3D reconstruction were analyzed with Image J and Imaris software.
In-utero electroporation and tissue processing. In utero electroporation experiments were performed according to protocols previously described 68 . The Animal Care and Use Committee (Bordeaux) has approved the experimental procedure under the number 5012015-A. Pregnant Swiss CD-1 mice were anesthetized using 4% isoflurane in an anesthesia induction chamber, maintained with 2% isoflurane with an anesthetic mask and injected before surgery with buprenorphine. Mice were subjected to abdominal incision; uterine horns were exposed and E14.5 embryos were placed on humidified gauze pads. Plasmid DNA was purified on Qiagen columns (EndoFree Plasmid Maxi Kit), resuspended in sterile endotoxin-free buffers (Qiagen) and mixed with Fast Green (Sigma). mCherry plasmid, together with pSuper or validated pSuper-Scrib shRNA construct (0.5 μg/μl) 17  www.nature.com/scientificreports/ (Tweezertrode 450165, Harvard Apparatus) connected to an electroporator (ECM830, BTX). Surgical procedure was completed with suture of the abdomen wall and skin. E18.5 embryos or P0 pups were processed for tissue analysis and immunostaining as described in the histology section. Subregions of the cerebral cortex (VZ/SVZ, IZ, LL and UL) were identified based on cell density using DAPI staining (Life Technologies; 1:20,000). For each condition, sections from three embryos obtained from three separate litters were quantified. Quantification of mCherry-positive cells was performed using the cell counter plugin for ImageJ (http:// rsbweb. nih. gov/ ij/ plugi ns/ cell-count er. html). Data are given as a percentage of the total of cells positive for mCherry in each cortical subregion (mean ± SD).
Behavioral testing. Behavioral experiments were conducted with Emx1-Scrib −/− cKOs and their control littermate male mice of 10-11 weeks of age at the start of behavioral tests. All behavioral experiments were performed during the light phase (between 9:00 A.M. and 7.00 P.M.) of a 12 h light/dark cycle, under conditions of dim light and low noise. One week before starting the experiments, mice were housed in individual cage. Several cohorts of animals and multiple behavioral tests were used. Whenever possible, naïve animals were employed for behavioral testing; when the same cohort was used for multiple tests, the most stressful assays were performed last, to minimize between-test interference. To look for behavioral abnormalities mice were tested in activity cages (to measure locomotor activity); in elevated plus maze, open field and Y-maze (to measure exploratory activity and anxiety-like behavior); in rotarod, hot plate, beam walking and grid handling test (to measure sensory-motor activity); in the Morris water maze to test the spatial memory and in the fear conditioning test maze (to measure early and remote contextual end cued memory performance). All experimental apparatuses were cleaned with hydroalcoholic solution (Phagospray-DM) between subjects to remove odor residuals. See Supplementary material and methods for details.
Quantification, statistical analysis and data representation. Details of statistical analyses and n values are provided in the figure legends subsections referring to individual assays. Statistical analyses were carried out using the GraphPad Prism statistical package (GraphPad). Normality of distribution and homogeneity of variance were validated and unpaired Student's two-tailed t tests for two data sets were used to compare groups with similar variance and are indicated along the P values in figures. P ≤ 0.05 was considered as statistically significant. Statistics were derived from at least three independent experiments and not from technical replicates. For behavior analyses, two-way ANOVA testing or repeated measure ANOVA was used for the evaluation of the effect of genotype and time in actimetry, motor activation, rotarod, water maze and fear conditioning behavioral test. The Bonferroni posthoc test was used when appropriate. The Student's t test was used for comparing genotype in other behavior tests. Whenever adequate, individual data points were reported as scatterplots to provide full information about the variability of data sets.