The following primary antibodies were used in this study: rabbit GSK3α and GSK3β antibodies (1:200 CST; #4337; #9315), rat anti-5-bromo-2-deoxyuridine (BrdU) antibody (1:100; Covance; MMS-139S), rabbit anti-Tbr2 (T-brain gene-2) antibody (1:100; Abcam; ab23345); rabbit anti-Cux1 antibody (1:80 Santa Cruz Biotechnology; sc-13024), rabbit anti-c-Myc antibody (1:200; GeneTex GTX103436), rabbit anti-β-catenin antibody (CST; #8480), and rabbit anti-GAPDH antibody (1:1000; Abcam; ab181603). The secondary antibodies conjugated with Alexa fluorophores 488 or 568 were directed against the IgGs of the primary antibody species (1:500; Invitrogen).

The small interfering RNAs (siRNAs) against GSK3α and GSK3β (ON-TARGET plus SMART POOL) were from Thermo Scientific Dharmacon (Chicago, IL, United States). The sequences of GSK3α siRNA duplexes were: 5′-GUA CUA CCG UGC UCC AGA ATT-3′ (forward); 5′-UUC UGG AGC ACG GUA GUA CTT-3′ (reverse); 5′-CGU GAC AGC GGG AAG GUG A TT-3′ (forward); 5′-UCA CCU UCC CGC UGU CAC GTT-3′ (reverse); 5′-GAU UAC ACC UCG UCC AUC GTT-3′ (forward); 5′-CGA UGG ACG AGG UGU AAU CTT-3′ (reverse); 5′-GUG GUC GGC UGG CUG UGU ATT-3′ (forward); 5′-UAC ACA GCC AGC CGA CCA CTT-3′ (reverse); GSK3β siRNA duplexes were: 5′-GGA CCC AAA UGU CAA CUA TT-3′ (forward); 5′-UAG UUU GAC AUU UGG GUC CTT-3′ (reverse); 5′-CCA CAG GAA GUC AGU UAU ATT-3′ (forward); 5′-UAU AAC UGA CUU CCU GUG GTT-3′ (reverse); 5′-UCA GAA GUC UAG CCU AUA UTT-3′ (forward); 5′-AUA UAG GCU AGA CUU CUG ATT-3′ (reverse); 5′-GAU UAC ACG UCC AGU AUA GTT-3′ (forward); 5′-CUA UAC UGG ACG UGU AAU CTT-3′ (reverse). The sequences of scrambled control siRNA were: 5′-UUC UCC GAA CGU GUC AGG UTT-3′ (forward) and 5′-AGG UGA CAC GUU CGG AGA ATT-3′ (reverse).

The mice used in this study were ICR mice. All mice were handled and treated according to the animal care and handling protocols approved by the Institutional Animal Care and Use Committee of Soochow University. In all experiments, the pregnant mice were anesthetized with a mixer of isoflurane (11.5%) and oxygen isoflurane (30%) or 3.6% chloral hydrate.

Embryonic brains were fixed in 4% paraformaldehyde at 4°C for 24 h, followed by dehydration in 20 and 30% sucrose solution (w/v), each for 24 h. Brains were cryosectioned in 14 μm thickness, and the brain sections were washed three times in PBS containing 0.3% Triton X-100, followed by blocking for 1 h in PBS containing 10% fetal bovine serum (FBS). Sections were incubated with the indicated primary antibodies overnight at 4°C, washed in PBS containing 0.3% Triton X-100 and then incubated with the corresponding secondary antibodies for 1 h. Sections were stained with Hoechst (Beyotime, C1022) for 20 min at room temperature. The sections were mounted with mounting medium (Vector Labs, H-1400) after washing in PBS containing 0.3% Triton X-100. For BrdU staining, the sections were treated with 2 N HCl for 20 min in a 60°C thermostat drier before blocking with PBS containing 10% FBS.

In utero electroporation procedure was performed as described (Saito and Nakatsuji, 2001). We used pEX-4 plasmid containing a reporter gene EGFP downstream of CMV promoter in all experiments to visualize transfected cells (C05004, GenePharma Company, China). Embryonic day 14.5 (E14.5) pregnant mice were anesthetized with a mixer of isoflurane (11.5%) and oxygen isoflurane (30%) or 3.6% chloral hydrate. The abdomen was cleaned with 75% ethanol and disinfected with a LIONSER^® compound iodine cotton swab. A 3-cm midline laparotomy was performed, and the uterus was taken out. Microinjection was carried out with PCR micropipettes (Drummond^®; #5-000-1001 × 10). Lateral ventricles of E14.5 embryos were injected with a mixture of 1 μl of pEX-4 (2.5 μg/μl) and 1 μl of siRNA (100 μM). siGSK3α/β and scrambled control siRNA were injected into different sides of lateral ventricles. Electric pulses (voltage: 45 V; pulse length: 50 ms; number: 5; interval: 950 ms) were applied by an ECM 830 Electroporator (BTX, Holliston, MA, United States). The electroporated pregnant mice were placed on a soft bedding with a warming blanket (37°C) until the dams were fully awake. The electroporated mice were sacrificed and analyzed at E18.5. Electroporated mice were from the same litter.

Cath.-a-differentiated (CAD) cells are a variant of a CNS catecholaminergic cell line that expresses pan-neuronal markers and differentiates to extend processes in serum-free conditions (Qi et al., 1997). CAD cells were cultured in DMEM/F-12 medium supplemented with 10% FBS, 2 mM L-glutamine, and 100 U/ml penicillin and streptomycin (all from Invitrogen). Cells were cultured at 37°C in a CO₂-humidified incubator. Lipofectamine 2000 reagent was used for siRNA transfection according to the instructions provided by the manufacturer.

Primary cortical neurons dissected from E15 mice as previously reported (Jang et al., 2016) with minor modifications. Cerebral cortices of E15 mice were dissected out and digested with trypsin (Gibco; 12604-021) for 5 min at 37°C. Enzyme-digested cortices were washed three times with MEM and dissociated in MEM. The dissociated neurons were centrifuged to remove the supernatant and resuspended in electroporation buffer containing siRNAs against GSK3 or control siRNA. Electroporation was performed immediately using an ECM 830 Electroporator (BTX, Holliston, MA, United States) with electrical pulses: five 50 ms pulses at 112 V with 950 ms interval. After electroporation, cells were immediately mixed with a suitable volume of prewarmed Neurobasal Medium supplemented with GlutaMAX and B27 and the neurons were plated on polylysine-coated plastic dishes. At 4–6 h after electroporation, the medium was changed to remove the electroporation buffer, and then the cells were cultured for 3 days.

For western blot analysis, cells were lysed in RIPA buffer. Protein concentration was determined by using the BCA Kit (Beyotime; P0012). Proteins were separated in a 10% SDS-PAGE gel, and transferred to a PVDF membrane, which was blocked (1 h) with 5% non-fat milk in TBST with 0.05% Tween 20. The membranes were sequentially incubated with the indicated primary antibodies in TBST containing 5% non-fat milk overnight at 4°C, followed by incubation with horseradish peroxidase-conjugated secondary antibodies for 2 h at room temperature. ECL Western Blotting Detection Reagents (Millipore; WBKLS0500) were used to visualize immunoreactive proteins. The band intensity was analyzed by Image J.

For BrdU labeling experiments, either pEX-4 plus siGSK3α/β or pEX-4 plus control siRNA was transfected into E14.5 embryos by in utero electroporation. After 4 days (at E18.5), pregnant mice were injected intraperitoneally with BrdU (50 mg/kg body weight) and 2 h later, animals were sacrificed, and embryos were prepared for immunohistochemical analysis as described above. Cell proliferation was calculated as the percentage of BrdU⁺ GFP⁺ double positive cells among total cells positive for GFP.

Fluorescent images were acquired using an AxioImager M1 epifluorescence microscope (Carl Zeiss). Images were taken with a 10× or a 20× objective. All images were acquired at a 1388 × 1040 pixel resolution. When analyzing fluorescent intensity, fluorescent exposure settings were kept the same, and all images were processed in parallel. Images were analyzed using AxioVision Rel.4.7 software (Carl Zeiss) and image J. To increase clarity, images were processed using Adobe Photoshop CS5 software and processing was applied equally across the entire image.

Data were collected from three independent experiments (for western blots) or three pairs of electroporated brain littermates (N = 3, for immunostaining). For each brain, eight slices (n = 8) were analyzed. For quantification of cells in the brain sections obtained from mice electroporated with either pEX-4 plus siGSK3α/β or pEX-4 plus control siRNA, the number of cells co-expressing a maker of interest (BrdU, Tbr2, or Cux1) together with GFP was counted in the dorsal cortex of telencephalon and presented as a percentage value of total GFP⁺ cells. All data were collected from the three independent experiments. Two-tailed Student’s t-test or one-way ANOVA was performed using SPSS software followed by Bonferroni’s post-test as post hoc test for comparing multiple groups. Statistics was applied to N and values are presented as mean ± SEM, and significance was set at p < 0.05 (^∗p < 0.05; ^∗∗p < 0.01; ^∗∗∗p < 0.001).

We first examined where GSK3 isoforms were expressed in the developing mouse neocortex. We found that GSK3α and GSK3β were highly expressed in NPCs in the germinal zone and IPCs in the SVZ of the dorsal telencephalon. GSK3α and GSK3β-immunoreactive fluorescence signals were detected in the apical wall of the VZ of the dorsal telencephalon, but no immunoreactive fluorescence signals were observed when the coronal sections of the dorsal telencephalon were stained with non-immune control immunoglobulin (data not shown).

To explore the possible role of GSK3α and GSK3β in cerebral cortex development, we used siRNAs targeting either GSK3α (siGSK3α) or GSK3β (siGSK3β). In a catecholaminergic cell line CAD, where transfection efficiency reaches ∼80% with lipid-based transfection (Byun et al., 2012), we confirmed that siGSK3α specifically and effectively (79 ± 3.6%) knocked down the target isoform without affecting GSK3β, and likewise, siGSK3β downregulated GSK3β (81 ± 2.5%) without altering the expression of GSK3α (Figures 1A,B and Supplementary Figure S1). We also confirmed the efficacy and specificity of siGSK3α or siGSK3β in primary cortical neurons (Figures 1C,D and Supplementary Figure S1). In primary cortical neurons, knocking down efficacies were lower than those achieved in CAD cells, probably because of a lower transfection efficiency. Notably, in both CAD cells and primary cortical neurons, the extent of knockdown achieved by siGSK3α and siGSK3β toward the target isoform were similar, providing a basis for comparing the effect of the two siRNAs. Using GSK3 siRNAs, we performed in utero electroporation at E14.5 together with GFP and confirmed that substantial knockdown of the target GSK3 isoform was achieved in the developing cortex (Figures 1E–G and Supplementary Figure S2).

In the E14.5 embryonic cortex, most NPCs in the VZ divide asymmetrically to generate another NPC and either a postmitotic neuron or an IPC, which will then divide again to generate two postmitotic neurons in the SVZ (Haubensak et al., 2004; Noctor et al., 2004). IPCs populate the SVZ, and postmitotic neurons journey through the intermediate zone (IZ) toward the developing CP (Molyneaux et al., 2007). When we electroporated E14.5 embryos with pEX-4 and examined the locations of GFP⁺ cells at E18.5, GFP-expressing cells were present throughout the cortical layers but localized mostly to the IZ and the CP. When GSK3α was depleted by in utero electroporation at E14.5, more cells remained in the VZ and the SVZ, and fewer cells migrated into the CP by E18.5 as compared to control brains that had been transfected with scrambled, control siRNAs (Figures 2A,B and Supplementary Figure S3). We found a much severe migration defect in siGSK3β-transfected cells. When GSK3β was depleted, the percentage of GFP⁺ cells remaining in the VZ was four times higher than that of control brains, and few GFP⁺ cells reached the CP (Figures 2C,D and Supplementary Figure S3). These results show that depletion of GSK3α or GSK3β traps cells in the VZ and prevents cell migration toward the CP layer.

GFP⁺ cells remaining in the VZ and the SVZ after GSK3 depletion likely correspond to cells that maintain a progenitor identity capable of self-renewal and reentering the cell cycle. To test this, we performed in utero electroporation at E14.5 with pEX-4 together with siGSK3α, siGSK3β, or control siRNAs, and then at 4 days after electroporation, we labeled the embryos with BrdU. BrdU incorporates into cells that are in S-phase of the cell cycle and thus marks actively proliferating cells. Two hours after injection of BrdU into the pregnant dam, embryos were dissected, and brain sections were subjected to immunostaining with BrdU antibodies. We detected a higher percentage of GFP⁺ cells co-labeled with BrdU in the dorsal cortex of the mice electroporated with either siGSK3α or siGSK3β, as compared to control siRNAs (Figures 3A–D and Supplementary Figure S4). These results indicate that more cells were undergoing division after depletion of GSK3α or GSK3β, suggesting that both GSK3α and GSK3β inhibit the proliferation of progenitors in the developing cortex.

Our findings show that depletion of GSK3α or GSK3β enhances the proliferation of progenitors (Figures 3A–D). Cortical neurons are generated from IPCs in the SVZ, which are derived from RGCs in the VZ or directly from RGCs (Guillemot et al., 2006). To examine the effect of GSK3 depletion on the genesis of IPCs, we performed in utero electroporation at E14.5 with pEX-4 and siGSK3α, siGSK3β or control siRNAs, and then at E18.5 brain sections were stained for Tbr2, a marker of IPCs. The percentage of Tbr2 and GFP double-positive cells were increased in the siGSK3α-transfected brains, as compared to control brains that had been electroporated with pEX-4 and scrambled siRNA (Figures 4A,B and Supplementary Figure S5). By contrast, brains transfected with pEX-4 and siGSK3β showed a substantial decrease in the percentage of Tbr2⁺ GFP⁺ cells, as compared to control brains (Figures 4C,D and Supplementary Figure S5). Although depletion of GSK3α and GSK3β commonly promoted the proliferation of progenitors in the VZ, knocking down of the two isoforms caused different outcomes in terms of Tbr2 staining. These results show that the conversion of radial progenitors to INPs is enhanced by knocking down GSK3α but suppressed by GSK3β depletion, suggesting distinct roles of the GSK3 isoforms in the developing cortex.

IPCs in the SVZ migrate to the upper layers of the cerebral cortex and undergo differentiation during development (Kwan et al., 2012). We thus next examined how depletion of the GSK3 isoforms affected further migration and differentiation steps. When we immunostained the brain sections for Cux1, which is expressed in excitatory neurons that populate layers II–IV of the neocortex, we found that the percentages of Cux1⁺ GFP⁺ cells were markedly reduced by siGSK3β but unaffected by siGSK3α (Figures 5A–D and Supplementary Figure S6). These results again show distinct outcomes caused by knocking down of either GSK3α or GSK3β in the developing cortex.

Next, we examined how knocking down both isoforms together affected cerebral cortex development. For this purpose, E14.5 embryos were electroporated in utero with a mixture containing pEX-4 and both of the siRNAs (siGSK3α and siGSK3β) or pEX-4 and scrambled siRNA as a control. When brain sections were prepared from E18.5 embryos, we detected substantially higher percentages of GFP⁺ cells in the VZ and the IZ but fewer cells in the SVZ and the CP, as compared to control group (Figures 6A,B and Supplementary Figure S7). Knocking down of both siGSK3α and siGSK3β together markedly increased the percentage of BrdU⁺ cells (Figures 6C,D and Supplementary Figure S7) but decreased Tbr2⁺ (Figures 7A,B and Supplementary Figure S8) or Cux1⁺ cells (Figures 7C,D and Supplementary Figure S8), suggesting that GSK3 activity was required for the conversion of radial progenitor cells into IPCs and upper-layer cortical neurons. These results were consistent with a previous study from double knockout mice (Gsk3a^-/-; Gsk3b^loxP/loxP; nestin-cre), which specifically deleted Gsk3b in the progenitor cells in a Gsk3a null background (Kim et al., 2009). Notably, the changes in the distribution of GFP⁺ cells, as well as the effects on BrdU⁺, Tbr2⁺, and Cux1⁺ cells induced by depletion of both GSK3α and GSK3β were recapitulated by knocking down GSK3β alone (see Figures 2D, 3D, 4D, 5D). Given that siGSK3α and siGSK3β downregulated the target gene to similar extents (see Figure 1), these results suggest that isoform GSK3β might play a dominant role in the regulation of cell migration and fate during cortical development.

Myc transcription factors, including c-Myc, N-Myc, and L-Myc, have been studied extensively as regulators of cell cycle progression and proliferation. In the developing nervous system, c-Myc and N-Myc are expressed, especially in proliferating cells, and Myc transcription factors have been shown to control the capacity of neural stem cells to self-renew and differentiate (Nagao et al., 2009). Notably, it is well documented that GSK3β phosphorylates and regulates the stability of c-Myc and N-Myc (Gregory et al., 2003; Sjostrom et al., 2005). Therefore, Myc proteins are obvious candidates to mediate GSK3 regulation of migration. In primary cortical neurons, we found that knocking down either GSK3α or GSK3β augmented the protein level of c-Myc (Figures 8A,B and Supplementary Figure S1). Interestingly, siGSK3β produced a much prominent enhancement, and the extent of increase induced by depletion of both GSK3α and GSK3β together was comparable to that caused by siGSK3β alone. In the developing cortex, we observed similar effects after in utero electroporation of siGSK3α, siGSK3β, or both, showing increased c-Myc protein level and a stronger elevation induced by siGSK3β (Figures 8C,D and Supplementary Figure S9) These results are consistent with the migration data in that a much severe migration defect was induced by siGSK3β as compared to siGSK3α (see Figure 2) and that the effect of knocking down both GSK3α and GSK3β could be recapitulated by depletion of GSK3β alone (see Figures 2, 6B).

Wnt/β-catenin signaling is also known to control migration of neural progenitors and β-catenin is a well-established substrate of GSK3 (Woodhead et al., 2006). We thus examined the effects of siGSK3α and siGSK3β on the level of β-catenin. Differential effects of knocking down GSK3α and GSK3β was much more evident in that only siGSK3β but not siGSK3α led to a significant elevation in the protein level of β-catenin both in cell culture (Figures 8A,B and Supplementary Figure S1) and the developing cortex (Figures 8E,F and Supplementary Figure S9). Together, these findings provide further support for the distinct roles of GSK3α and GSK3β and suggest that the differential effects of knocking down GSK3α or GSK3β on cell migration and fate in the developing cortex might have resulted from differential activation of the downstream signaling pathways.