Over the last years, the critical role of cytoskeletal proteins in cortical development including neuronal migration as well as in neuronal morphology has been well established. Inputs from genetic studies were provided through the identification of several mutated genes encoding either proteins associated with microtubules (DCX, LIS1, KIF2A, KIF5C, DYNC1H1) or tubulin subunits (TUBA1A, TUBB2B, TUBB5 and TUBG1), in malformations of cortical development (MCD). We also reported the identification of missense mutations in TUBB3, the postmitotic neuronal specific tubulin, in six different families presenting either polymicrogyria or gyral disorganization in combination with cerebellar and basal ganglial abnormalities. Here, we investigate further the association between TUBB3 mutations and MCDs by analyzing the consequences of Tubb3 knockdown on cortical development in mice. Using the in utero-electroporation approach, we demonstrate that Tubb3 knockdown leads to delayed bipolar morphology and radial migration with evidence, suggesting that the neuronal arrest is a transient phenomenon overcome after birth. Silenced blocked cells display a round-shape and decreased number of processes and a delay in the acquisition of the bipolar morphology. Also, more Tbr2 positive cells are observed, although less cells express the proliferation marker Ki67, suggesting that Tubb3 inactivation might have an indirect effect on intermediate progenitor proliferation. Furthermore, we show by rescue experiments the non-interchangeability of other beta-tubulins which are unable to rescue the phenotype. Our study highlights the critical and specific role of Tubb3 on the stereotyped morphological changes and polarization processes that are required for initiating radial migration to the cortical plate.
During brain development, the radial migration of newborn pyramidal neurons is critical for the formation of a laminar cortex in mammals. Post-mitotic neurons exiting from the proliferative ventricular zone (VZ) start their journey by acquiring a succession of bipolar, multipolar and then again bipolar morphologies in the subventricular (SVZ) and intermediate zones (IZ). They migrate long distances to reach their final destination in the cortical plate (CP) forming an ‘inside-out’ layered cortex (1–3). These steps are followed by dendritic and axonal extensions and the implementation of neuronal circuitry.
Over the last years, the critical role of cytoskeletal proteins in cortical development including steps of neuronal migration and neuronal morphology has been well established. Genetic studies demonstrated the implication of several genes encoding either microtubule-associated proteins (MAPs), other microtubule-binding proteins or motors, or tubulin subunits in malformations of cortical development (MCD). Mutations in the DCX and LIS1 genes, encoding proteins that modulate microtubule (MT) homeostasis, were shown to be associated with a spectrum of lissencephaly neuronal migration disorders encompassing agyria, pachygyria and laminar heterotopias (4–8). More recently, mutations in genes encoding MT-related motor proteins, including kinesins (KIF2A and KIF5C) and dynein (DYNC1H1) were also found in MCD patients (9). Moreover, the importance of MT related functions were further demonstrated by the direct association of mutations in tubulin genes and these cortical developmental disorders (10). Indeed, alpha-tubulin (TUBA1A and TUBA8), beta-tubulin (TUBB2B) and gamma-tubulin (TUBG1) genes were shown to be involved in lissencephalies and polymicrogyria (11–14), while mutations in TUBB5 were identified in cases presenting microcephaly associated with cortical gyration abnormalities (15). Furthermore, we identified missense mutations in TUBB3, a postmitotic neuronal tubulin, in six different families presenting either polymicrogyria or gyral disorganization in combination with cerebellar and basal ganglial abnormalities (16) (OMIM#614039). In this study, we also reported a de novo heterozygous p.Met388Val mutation in a fetal case with microlissencephaly. TUBB2B and TUBB3 were also found mutated in congenital fibrosis of the extraocular muscle disorders (17,18). Finally, mutations in the TUBB4A gene were recently reported in families presenting whispering dysphonia associated with dystonia (19,20) and in a leukoencephalopathy hypomyelination syndrome (21).
Tubulin subunits are highly conserved during evolution and alpha–beta tubulin heterodimers are the structural components of MTs. There are multiple copies of the tubulin genes in human and rodent genomes. Alpha- and beta-tubulin subunits are thus encoded by 19 genes (10 alpha and 9 beta) in human and eight alpha- and eight beta-tubulin genes have been isolated in mice (http://genome.ucsc.edu/). Each alpha/beta-tubulin shows >90% amino acid identity with other alpha/beta-tubulins (22). Tubulin isotypes expression and distribution in different cells or tissues is highly regulated, and their apparent redundancy more probably underlies cellular and functional diversity requirements (23). Whereas any of the isotypes could ensure basic structural functions, such as forming an interphase cytoskeleton network (24), convergent studies tend to confirm the functional specificity of the different tubulin isoforms. For instance, specific knockdowns of a set of beta-tubulins in cell cultures showed, for each of the targeted subunits, different cellular phenotypes in differentiating neuronal cells (25). Though further in vivo studies are required, these results highlighting the importance of specific combinatorial expression of different tubulin isoforms, are in favor of the much debated multi-tubulin hypothesis (26).
Here, we investigated the association between TUBB3 mutations and MCDs by analyzing the consequences of either Tubb3 knockdown or overexpression of TUBB3 mutations on radial neuronal migration in the mouse cortex, using an in utero-electroporation approach. We also demonstrate, by rescue experiments, the non-interchangeability of certain beta-tubulins during brain development, confirming their functional specificity.
Tubb3 inactivation perturbs radial migration
In order to assess the consequences of Tubb3 loss of function on the cortical development, we used an in utero-electroporation approach to deliver specific short hairpin RNAs (shRNAs) into cortical progenitor cells of the mouse brain VZ. Two specific anti-Tubb3 shRNAs targeting the 3′UTR sequence were cloned in a psiStrike vector under the control of the U6 promoter. Cellular experiments in P19 neuroblastoma cells showed an extinction of Tubb3 at the transcript (sh1:−44%; sh2: −67%) and protein (sh1:−34%; sh2: −67%) levels and only a minor effect of a scrambled sequence compared with the untransfected control (Fig. 1A and B). The validated shRNAs and scramble constructs were then individually electroporated in utero, together with an RFP reporter construct into the mouse cortical VZ at embryonic day E14.5, a time coincident with extensive neuronal production and migration. Control electroporated cells gave rise to red fluorescent young neurons migrating toward the CP. Embryos were harvested 4 days later (E18.5) and the migration pattern of the electroporated cells was determined by the quantification of RFP+ cells in the different cortical regions: VZ/SVZ, IZ and CP. We checked the efficiency of the shRNAs in vivo by immunolabeling the Tubb3 protein in brain sections and confirmed the large reduction of Tubb3 expression in electroporated cells (Fig. 1C). In cortices transfected only with the RFP plasmid, nearly all the neurons were found as expected in the CP (VZ: 5.6 ± 3.4%, IZ: 6 ± 3.3%, CP: 88.4 ± 6.7%), while the scrambled construct showed slightly lower number of cells in the CP with a proportion still remaining in the IZ, although this did not reach significance (VZ: 8.86 ± 3.6%, IZ: 18.8 ± 5.1%, CP: 74.1 ± 8.3%) (RFP versus scramble Mann–Whitney test; P-value: VZ 0.59; IZ 0.11; CP 0.26). In contrast, neurons transfected with the shRNAs against Tubb3 displayed a severe and significant arrest in radial migration. A majority of them were blocked in the IZ with only a few neurons found in the CP (Sh1; VZ: 11.5 ± 5.2%, IZ: 78.2 ± 7.8%, CP: 10.5 ± 4.5%; Sh2; VZ: 23.3 ± 3.5%, IZ: 53.7 ± 3.5%, CP: 22.3 ± 0.2%) (sh1 versus scramble P-value: VZ 0.35; IZ**; CP**; sh2 versus scramble P-value: VZ*; IZ**; CP**). To validate the specificity of this result, we carried out a rescue experiment by co-electroporation of the anti-Tubb3 sh1 construct and the sh-resistant human TUBB3-coding sequence (NM_006086.3) and revealed a significant derepression of the shRNA blocking effect (VZ: 1.4 ± 0.7%, IZ: 16.3 ± 6.3%, CP: 82.3 ± 6.8%) (sh1 versus sh1+cTUBB3 P-value: VZ 0.06; IZ*; CP*) confirming the specificity of the shRNAs and the importance of Tubb3 for radial migration to the cortex (Fig. 1D–F).
TUBB1, TUBB2B and TUBB4A do not fully rescue the Tubb3 knockdown migration defect
To assess the possible functional redundancy of tubulin subunits and the capacity of other beta-tubulins to substitute for the Tubb3 loss of function, we co-electroporated the Tubb3 shRNA with the human-coding sequence of three other beta-tubulins. We selected one that is weakly expressed in human fetal brain (TUBB1, NM_030773) and two which are highly represented in this tissue (TUBB2B, NM_178012 and TUBB4A, NM_006088) (data not shown). TUBB1 cDNA was not able to rescue the Tubb3 knockdown phenotype as a large majority of electroporated cells remained blocked in the VZ and IZ (VZ: 21.5 ± 6%, IZ: 68.2 ± 5.2%, CP: 10.3 ± 4.1%) (sh1 versus sh1+cTUBB1 P-value: VZ: 0.29; IZ: 0.41; CP: 0.73). On the other hand, TUBB2B and TUBB4A expression led to a partial rescue as two equal populations of neurons were observed in the IZ and CP at day E18.5 (sh1+cTUBB2B VZ: 18 ± 3.1%, IZ: 47.1 ± 8.4%, CP: 34.9 ± 9.6%; sh1+TUBB4A VZ: 13.8 ± 3.7%, IZ: 44.5 ± 9.7%, CP: 41.7 ± 12.5%) (sh1 versus sh1+cTUBB2B P-value: VZ 0.24; IZ*; CP 0.07; sh1 versus sh+cTUBB4A P-value: VZ: 0.60; IZ: 0.09; CP: 0.26). In view of this result, we checked the expression efficiency of these beta-tubulins after electroporation. Indeed, expression analysis of the transgenes by real-time RT–PCR experiments after in utero electroporation and dissection of the electroporated cortical region, showed an efficient and significant expression of these beta-tubulin transgenes (i.e., 177–564 times higher in electroporated regions compared with non-electroporated brains, data not shown).
Although a partial rescue of the migration phenotype was obtained by these two beta-tubulin subunits, the diffused distribution of electroporated cells still remained substantially different from that of the organized cells expressing scramble RNA and RFP alone (Fig. 2A and B), supporting the non-interchangeability of these beta tubulins during cortical development and the functional specificity of Tubb3.
TUBB3 mutants do not rescue the migration impairment
We sought to test whether the expression of MCD-related TUBB3 mutants could complement the phenotype caused by knockdown of Tubb3. We selected therefore TUBB3 variants associated with variable phenotypes in patients and tested p.Gly82Arg and p.Met388Val TUBB3 mutations, responsible for multifocal polymicrogyria in one patient and fetal microlissencephaly, respectively (16). We electroporated each of these two constructs, in combination with the Tubb3-shRNA-1 and analyzed the position of electroporated cells within the cortex. In both experiments, the mutants did not suppress the shRNA effect: while some of the cells reached the CP (sh1+p.Gly82Arg: 26.5 ± 11.4%; p.Met388Val: 23.6 ± 4.5%), a majority of them were blocked in the VZ and IZ (p.Gly82Arg VZ: 20.3 ± 6.6%, IZ: 53.2 ± 7.5%; p.Met388Val VZ: 21.5 ± 4.8%, IZ: 55 ± 5.1%), quite similar to the shRNA-1 alone and significantly different from the scramble shRNA (Fig. 2C and D) (sh1 versus sh1+ p.Gly82Arg P-value: VZ: 0.35; IZ*; CP: 0.48; sh1 versus sh1+ p.Met388Val P-value: VZ: 0.3; IZ*; CP: 0.16) (scramble versus sh1+ p.Gly82Arg P-value: VZ: 0.09; IZ*; CP*; sh1 versus sh1+ p.Met388Val P-value: VZ:*; IZ**; CP**).
We also overexpressed the wild type and the two TUBB3 mutants alone and observed that the electroporated neurons reached the CP normally. However, when examining more precisely neuronal position it seemed that while control and p.Gly82Arg electroporated neurons reached the upper layers of the CP in a comparable way, the p.Met388Val transfected neurons were arranged in a more diffuse manner in the CP compared with the wild type which may suggest a minor dominant effect on migration (Fig. 2E and F).
The migrating defect related to Tubb3 knockdown is a transient phenomenon
We looked at the effect of the Tubb3 shRNA electroporated at E14.5, on neuron positioning at different embryonic and postnatal stages (E16.5, E17.5, E18.5, P0, P2 and P7). At E16.5 the distribution of RFP+ neurons was quite similar comparing sh and scramble conditions, most of them were found in the VZ and the lower IZ, except for a small proportion of neurons that had already reached the upper IZ. The migration delay of the sh electroporated neurons appeared more evident at E17.5, the effect culminating at E18.5. At the neonatal stage P0 some of the sh electroporated neurons reached the CP and at postnatal stages P2 and P7, no particular differences between the scramble and sh conditions were seen and fluorescent neurons formed a thin layer in the upper part of the cortex (Fig. 3A). In order to verify this result, we checked by immunolabeling of postnatal (P2) brain sections that silencing by shRNA is not transient in neuronal cells that reach the CP. Anti-Tubb3 antibodies showed the presence in the CP of neuronal cells positive for the RFP reporter marker showing a downregulation of Tubb3 (data not shown). Furthermore, brain slices were also stained with an antibody against caspase 3 to assess for an apoptosis of cells arrested in the IZ by immunolabeling of brain sections at E16.5, E18.5 and P2 stages (2, 4 and 6 days after electroporation). No differences were observed between scramble and shRNA conditions (data not shown). These data suggest that the migration phenotype of the neuronal population deficient for Tubb3 is a transient phenomenon and arrested cells maintained their ability to reach the CP 6 days after electroporation.
Tubb3 inactivated cells displayed an abnormal morphology
To better define the defect observed during radial migration, we examined the morphology of cells accumulated in the SVZ/lower IZ in control and sh1 conditions at E16.5. While the migration pattern did not seem to be widely altered, the shape of Tubb3 deficient cells appeared to be different when compared with control. Indeed, a great majority of control cells showed a multipolar morphology with visible processes (scramble: 60 ± 5.5%, sh: 42.4 ± 2.5%; t-test, P* < 0.05), compared with the inactivated cells which showed more round shape with no or very short processes (scramble: 8.1 ± 2.6%, sh: 26 ± 3.8%; t-test, P* < 0.01) (Fig. 4A and B). Around 30% (scramble: 31.9 ± 3.6%, sh: 31.6 ± 3.1%) of cells in both control and knockdown conditions displayed a typical migrating uni- or bipolar morphology. Similar morphological states were still visible at E18.5 in shRNA electroporated cells blocked in the lower IZ, whereas the control cells had already reached the CP. Approximately 40% of shRNA cells remained multipolar (20.4 ± 8.6%) or displayed round shapes with no or short processes (20.3 ± 9%) (Fig. 4B). Thus, Tubb3 shRNA cells may be retarded in their morphological development.
In addition, we examined the orientation of the main process of cells harboring a bipolar shape at E16.5 and E18.5 days. We divided them into three main categories according to the orientation of the main process with respect to the surface of the ventricle (radially, tangentially or inversely oriented). shRNA and scramble conditions showed similar distributions at E16.5 with a majority of cells oriented radially with an angle of migration between 60° and 90° (scr: 65.1 ± 5%, sh: 51.4 ± 8.4%), a small proportion of cells moved tangentially with an angle between 0° and 30° (scr: 10.1 ± 2.4%, sh: 14.1 ± 4%) and ∼20% of them migrated towards the ventricle (scr: 15.2 ± 4.9%, sh: 22.9 ± 3.6%) (n: scr = 262, sh v = 206). At E18.5, whereas control cells had already reached the CP, knockdown cells displayed similar orientations as at E16.5. Half of them (48.3 ± 4.8%) were radially oriented, 5.8 ± 4.2% were tangentially oriented and 35 ± 4.5% moved toward the ventricle (Fig. 4C). Altogether, these results showed that Tubb3 inactivation has an impact on neurons morphological changes at E16.5 but not on the orientation of their main process.
Effect of Tubb3 inactivation on intermediate progenitors
We then sought to determine whether Tubb3 inactivation has an impact on the intermediate progenitor (IP) population in the SVZ. We therefore examined the number of these progenitors by labeling with the IP specific marker, Tbr2, at the E16.5 stage in brains electroporated 2 days earlier with shRNA and scramble constructs. We found a significant increase of Tbr2+ electroporated cells in the shRNA condition compared with the control (scramble: 27.8 ± 1.7%; sh1: 36.5 ± 2%; t-test, P* < 0.05) (Fig. 5A). This result suggesting a transition defect and/or a failure of these cells to differentiate into postmitotic neurons prompted us to further investigate consequences of Tubb3 downregulation on IP proliferation.
We checked the expression of Tubb3 in IPs by immunolabeling embryonic brain sections and primary cultures of dissociated progenitors. Double labeling of brain sections suggests that Tubb3 is expressed in a subset of Tbr2 positive cells which are enriched in the basal layer of the SVZ. Because Tbr2 and Tubb3 are localized in different cell compartments, it remained difficult to unambiguously ascertain that these cells express both markers. To further confirm this result, we generated cultures of neuronal progenitor cells and assessed the expression of the two markers. Analysis of the results showed a minor population co-expressing Tubb3 and Tbr2, corresponding to ∼10% of the total population of IPs (data not shown).
We also performed quantitative comparative analysis of the proliferation of these cell populations following in utero electroporation of scramble and Tubb3-targeting shRNAs. In the VZ no quantitative difference for PH3+, nor for KI67+ cells (RFP electroporated cells) was observed. However, in the SVZ, we found a reproducible significantly reduced number of Ki67+ cells for the Tubb3 shRNA in comparison with the scramble control. For PH3+ cells in the SVZ, no difference between the scramble and shRNA was observed, the scarcity of PH3-positive cells did not allowed reliable statistical comparison (Fig. 5B and C). In independent subsequent in utero-electroporation experiments and quantitative comparative analysis using double immunolabeling of brain sections with anti-Tbr2 and anti-Ki67 antibodies, we confirmed the observed decrease of cells in the SVZ co-expressing Tbr2 and Ki67 in the shRNA condition compared with the control (scr: 4.9 ± 1%, sh: 2.4 ± 0.9%; n: scr = 453, sh = 372, χ2-test, *P < 0.05; data not shown).
Altogether these data show the critical role of Tubb3 in cortical development, especially in the morphological changes undertaken by cells at the beginning of migration and in the regulation of the IP transition towards their differentiation into neuronal cells.
The subunit tubulin beta 3 has been considered as a specific marker of neurons for decades and its implication in brain development was more recently emphasized by its involvement in MCDs (16). Here, we used an in utero-electroporation approach to explore more precisely the role of Tubb3 in mouse brain during cortical development. We describe a delay of migration that culminates at E18.5 in neurons electroporated at E14.5 with shRNA constructs suppressing Tubb3 expression. These results show that Tubb3 plays an important role in the correct migration of pyramidal neurons in fitting with the MCD phenotypes. Similar studies previously demonstrated a deleterious effect of shRNAs against Tubb2b and Tubb5 during cortical radial migration in the rodent. While the knockdown of Tubb2b disrupt the radial migration in rat (13), anti-Tubb5 shRNA led to a blocking of the migration in mouse cortices with a specific impact on the length of the neurogenic cell cycling (15). We show here delayed initiation of migration in Tubb3 mutant cells which take a longer time to develop a bipolar morphology. These combined results confirm the importance of the MT during various steps of cortical development.
Interestingly, in this experimental design the Tubb3-related migration defect seems to be a phenomenon that disappears 4–5 days later in postnatal stages. The result showing a continued downregulation of Tubb3 in neuronal cells that had reached the CP suggest that Tubb3-related migration defect is transient. Though additional investigations are required to address this observation, these data suggest that neurons that do not express Tubb3 maintain migratory potential in early stages after birth. This has been also shown in Dcx knockdown condition, in which blocked neurons in the VZ/ IZ resumed their migration after overexpression of Doublecortine transcript at P0 (27). Also, as expression of TUBB2B, TUBB1 and TUBB4A cDNAs did not rescue the migration phenotype, it suggests that other tubulins could be involved in the adaptive phenomenon and compensate the lack of Tubb3 allowing neuronal cells to recover their migration ability. Similarly, expression profiles during neuronal development of MT-interacting proteins could also contribute to the observed recuperation of migration. Tubb3 downregulation leads to a significant reduction of multipolar processes in E16.5 SVZ. Moreover, silenced cells still display multipolar or round shapes at E18.5 (morphologies which are not observed at this stage in control cells). These results are reminiscent of previous studies showing that RNAi in utero-electroporations-targeting Filamin A, Lis1 and Dcx, three other genes responsible for MCDs and interacting directly or indirectly with the cytoskeleton, lead to an accumulation of multipolar cells in the SVZ or the IZ (28). They also suggest that Tubb3 appears necessary, through its structural role, for proper MT dynamics and therefore neuronal morphology organization and process elongation.
Concerning the other beta-tubulins tested here (TUBB1, TUBB2B and TUBB4A), TUBB1 that is weakly expressed in human fetal brain is not able to replace Tubb3 function, leading to the hypothesis that these two beta subunits ensure completely different functions, perhaps involving different protein partners. Beta-tubulin-2B and -4A, two subunits expressed in human fetal brain, overcome partially the Tubb3 related migration defect showing that they share some, but not all brain specific functions. This result tends to support the idea of the non-interchangeability of tubulin subunits. Our data suggest that even if they share significant homology and common functions, each tubulin subunit may have a specific role that cannot be replaced by other subunits during cortical development. These findings are also consistent with the existence of specific syndromes in human caused by mutations in different tubulin subunits. The complexity is also such that different mutations in one tubulin gene can lead also to several different phenotypes (13,16–18).
In this in utero-electroporation-based study, we tested two TUBB3 MCD-related mutations: (i) p.Gly82Arg, responsible for a moderate form of MDC within the TUBB3-related spectrum, and predicted to be adjacent to the lateral beta–beta interaction domain of tubulins and (ii) p.Met388Val, which leads to a dramatic fetal form of microlissencephaly. This latter mutation resides on the outer surface of the MT and could interact with MAPs and motor proteins or be involved in protein folding (29). We established that the two mutations fail to complement the Tubb3 shRNA phenotype, although the two mutations could act in slightly different ways in an overexpression context. While the p.Gly82Arg mutant had no major overexpression effect, p.Met388Val led to a mild dispersion of neurons in the cortex that could have repercussions in the establishment of a proper neuronal connectivity. These combined observations may suggest that the first mutation acts as a loss of function and the second could have potentially a dominant-negative effect during mouse radial migration, consistent with the different phenotype observed in patients. Given that all mutations identified in the tubulin isotypes (TUBA1A, TUBB2B, TUBB3 and TUBB5) are missense mutations, it is likely that some may act in a dominant-negative fashion via an altered protein function, thus in addition to haploinsufficiency, contributing to the mechanisms underlying MCDs. Though recurrent mutations are consistently associated with almost identical MCDs, at the current level of our understanding of tubulin functions and their cellular roles, and the potential multi-level effects of tubulin mutations assessed by different assays, it is difficult to predict molecular mechanisms (haploinsufficiency or dominant-negative effect) and/or disease phenotype resulting from a given mutated residue or isoform.
Finally, we showed that Tubb3 downregulation is associated with a significant reduction of Ki67+ cells in the SVZ. This finding that is apparently in contradiction with the increased number of Tbr2 cells is also difficult to reconcile with the assumption that Tubb3 is mainly expressed in postmitotic neuronal cells. However, we can hypothesize that Tubb3 inactivation might have an indirect effect on Tbr2 progenitor proliferation. Indeed, Tubb3 inactivation and the subsequent disruption of early differentiation and migration steps of cells that are still Tbr2+ and their accumulation in the SVZ might indirectly trigger through cell–cell contacts a signal that leads to a reduction of proliferation of IPs (Fig. 6). In view of these findings, we are also attempted to hypothesize that normal cortical development may also involve a cross talk between CP and VZ/SVZ. It is well established that position of pyramidal neurons in the CP is dependent of their date of birth, and here we propose that cells that reach their final position in the CP ‘talk’ to the VZ, via radial glia fibers, to allow a fine tuning of progenitor proliferation.
Although many questions remain to be addressed, our study emphasizes the crucial role of the MT network in different steps of neuronal migration and especially in the critical stereotyped morphological changes and polarization processes that are required for triggering and regulating radial migration towards the CP. From our study, it appears clear that beta-tubulins are non-interchangeable in the context of this neuronal radial migration assay, providing therefore important information concerning the debate of their potential redundancy.
MATERIALS AND METHODS
We designed two different shRNA oligonucleotides targeting the 3′UTR of the mouse Tubb3 transcript (Sh1 specifically recognizes nucleotides 1427–1451, Sh2, nucleotides 1435–1459, NM_023279). Annealed oligonucleotides were cloned in the psiSTRIKE U6 promoter-driven vector. As control, we used a scrambled oligonucleotide corresponding to Sh2. Human beta-tubulin cDNAs were inserted in the pcDNA vector (Life Technologies, Carlsbad, CA, USA) under the CAG (chicken beta-actin promoter with CMV enhancer) promoter for the rescue experiments.
Quantitative PCR and western blot
Plasmids used for in utero-electroporation were prepared using the Endotoxin-Free Plasmid kit (Macherey Nagel, Duren, Germany). Efficiency of the shRNAs against Tubb3 was assessed at the transcript and protein levels by transfection in murine P19 cells with the Neon System (Life Technologies). RNAs were purified by the RNAeasy Kit (Qiagen, Venlo, Netherlands), cDNA were prepared with Maxima First Strand cDNA synthesis Kit (Thermo Fisher Scientific, Watham, MA, USA), and then RT–PCR products amplified from cDNA were quantified with the LightCycler 480 FastStart DNA SYBR Green I Master kit in a LightCycler 480 System (Roche, Manheim, Germany). Protein explorations were performed by loading equal quantities of lysates on polyacrylamide gels, transferred to nitrocellulose membrane. After blocking with a 5% milk solution (w/v), blots were incubated with the primary antibody overnight at 4°C. Mouse Tubb3 antibody from Covance (Princeton, NJ) was used at 1:1000 and goat Actin antibody (SC-1616, Santa Cruz Biotechnologies, Dallas, TX, USA) was diluted at 1:1000. HRP (horseradish peroxidase) secondary antibodies from Dako-Agilent Technologies (Santa Clata, CA, USA) were used diluted at 1:5000.
In utero electroporation
In utero electroporations were performed as described previously (30) using Swiss timed pregnant mice. Briefly, 1–2 µg of plasmid DNA combined with fast green (2 mg/ml; Sigma) were injected into the lateral ventricles of E14.5 brains and then electroporated into VZ progenitor cells by delivering five electric pulses at 45 V for 50 ms at 950 ms intervals through the uterine wall using a CUY21EDIT System (Nepagene, Chiba, Japan). Embryos or pups were allowed to develop for 2–13 days, brains were harvested, and brain sections of 80 µm were prepared using a vibratome (VT1000S, Leica, Solms, Germany). The positions and morphology of transfected cells were analyzed by fluorescence microscopy after fixation of the sections with 4% paraformaldehyde using a DMRA2 microscope (Leica Microsystems, Solms, Germany).
Free-floating sections were blocked with 2% goat serum in phosphate buffered saline (PBS) and 0.2% Triton-X 100. The primary antibodies used were anti-TBR2 (rabbit polyclonal, 1:200, Abcam, Cambridge, UK), anti-beta 3 tubulin Tuj1 (mouse, 1:500, Covance, Princeton, NJ, USA), anti-phospho Histone H3 (rabbit, 1:500, Millipore, Billerica, MA, USA), anti-Ki67 (mouse, 1:200, Bethyl Laboratories, Montgomery, TX, USA) overnight at 4°C. After three washes in PBS, sections were incubated with either FITC- or Cy5-conjugated secondary antibody (Jackson Immunoresearch, West Grove, PA, USA) in a 1:300 dilution at room temperature, washed and mounted with Fluoromount (Sigma-Aldrich, St. Louis, MO, USA). Images were acquired using a spinning disk system (Leica Microsystems, Solms, Germany) for morphological analyses or a TCS SP5 II (Leica Microsystems, Solms, Germany) for immunolabeling analysis. All cells were counted for statistical analyses using ImageJ software (NIH).
This work was supported by fundings from Institut National de la Santé Et de la Recherche Médicale, the Fondation pour la Recherche Medical (J Chelly—Equipe FRM 2013: DEQ2000326477), the Fondation JED, the Agence Nationale pour la Recherche (ANR Blanc1103-01 projet R11039KK, ANR E-RARE-012-01 projet E10107KP, ANR 08-MNP-013) and the EU-FP7 project GENECODYS (grant number 241995). We thank the INSERM Avenir program and the Fondation Bettencourt Schueller for grants awarded to F.F., the Région Ile-de-France for the NERF postdoctoral grant awarded to EBJ/FF and for their support of animal house and imaging facilities. The team of F.F. is part of the Ecole des Neurosciences de Paris Ile-de-France network.
We thank Katia Boutourlinsky, Reham Khalaf and Audrey Roumegous for their technical support and Pierre Billuart for his helpful comments and discussions.
Conflict of Interest statement. None declared.