Mitotic WNT signalling orchestrates neurogenesis in the developing neocortex

Abstract

The role of WNT/β‐catenin signalling in mouse neocortex development remains ambiguous. Most studies demonstrate that WNT/β‐catenin regulates progenitor self‐renewal but others suggest it can also promote differentiation. Here we explore the role of WNT/STOP signalling, which stabilizes proteins during G2/M by inhibiting glycogen synthase kinase (GSK3)‐mediated protein degradation. We show that mice mutant for cyclin Y and cyclin Y‐like 1 (Ccny/l1), key regulators of WNT/STOP signalling, display reduced neurogenesis in the developing neocortex. Specifically, basal progenitors, which exhibit delayed cell cycle progression, were drastically decreased. Ccny/l1‐deficient apical progenitors show reduced asymmetric division due to an increase in apical–basal astral microtubules. We identify the neurogenic transcription factors Sox4 and Sox11 as direct GSK3 targets that are stabilized by WNT/STOP signalling in basal progenitors during mitosis and that promote neuron generation. Our work reveals that WNT/STOP signalling drives cortical neurogenesis and identifies mitosis as a critical phase for neural progenitor fate.

WNT/STOP (STabilization Of Proteins) signalling is the primary driver of neuronal differentiation of progenitor cells, while WNT/β‐catenin signalling predominantly regulates progenitor cell self‐renewal.

Introduction

During embryonic/foetal development, the mammalian neocortex undergoes a large increase in surface area and a drastic expansion of the number of projection neurons necessary for higher cognitive functions (Caviness et al, 1995; Rakic, 1995; Rakic, 2009; Lui et al, 2011; Florio & Huttner, 2014; Sun & Hevner, 2014). Crucial to this increase are the neural progenitor cells (NPCs) in the developing neocortex. The primary population of NPCs reside in the ventricular zone (VZ), adjacent to the ventricular lumen, and during the initial stages of neocortex development undergo several rounds of self‐amplifying symmetric divisions (Malatesta et al, 2000; Noctor et al, 2001; Götz & Huttner, 2005), commonly referred to as increased self‐renewal (see note on terminology in Methods). With the onset of cortical neurogenesis, these NPCs, known as “apical progenitors” (APs), begin to divide asymmetrically to generate either post‐mitotic neurons (direct neurogenesis, a minor pathway in mammals) or a secondary population of NPCs termed “basal progenitors” (BPs, indirect neurogenesis, the major pathway in mammals) (Kriegstein & Götz, 2003; Götz & Huttner, 2005; Taverna et al, 2014). In contrast to APs that undergo mitosis at the apical surface, BPs undergo mitosis basal to the VZ in a second germinal zone known as the subventricular zone (SVZ). Depending on the mammalian species, BPs either divide to give rise to two neurons (also referred to as neuronal differentiation of BPs), which is typically the case in embryonic mouse neocortex, or first self‐renew to increase their pool size and then generate neurons, which is the case for foetal human neocortex (Lui et al, 2011; Florio & Huttner, 2014; Taverna et al, 2014; Molnár et al, 2019). Maintaining the balance between symmetric self‐renewing AP divisions and asymmetric BP‐genic AP divisions is one critical basis for generating the correct number of neurons in the neocortex (Huttner & Kosodo, 2005; Delaunay et al, 2017). In this context, cell signalling pathways are crucial to regulate neurogenesis, but pinpointing their precise functions (e.g. in NPC self‐renewal vs. generation of differentiated cells, in which cell types, at which developmental stages) has proven challenging, and, at times, controversial (Taverna et al, 2014).

A case in point is WNT signalling, a conserved pathway that is intricately linked with neocortex development. The overall role of WNT signalling in neurogenesis is complex and highly dependent upon the models used, as well as the epistatic level at which the pathway is manipulated (Harrison‐Uy & Pleasure, 2012). For instance, genetic ablation of β‐catenin, the signal transducer driving the transcriptional response of canonical WNT signalling, leads to increased cell cycle exit of NPCs and premature neuronal differentiation (Machon et al, 2003; Woodhead et al, 2006; Mutch et al, 2010). Conversely, overexpression of constitutively active β‐catenin or deletion of glycogen synthase kinase 3 (GSK3), a key negative regulator of canonical WNT signalling, drive increased AP self‐renewal at the expense of BP and post‐mitotic neuron generation (Chenn & Walsh, 2002; Machon et al, 2007; Wrobel et al, 2007; Kim et al, 2009). Hence, it is generally thought that the primary role of canonical WNT signalling is to promote NPC self‐renewal, a phenomenon that is also observed in other parts of the developing nervous system and is mediated by the transcriptional activity of β‐catenin (Zechner et al, 2003; Gulacsi & Anderson, 2008; Draganova et al, 2015). However, in vitro studies have demonstrated that WNT/β‐catenin, through transcriptional regulation of N‐myc and the neurogenic transcription factors Ngn1/2, can also promote differentiation of NPCs (Hirabayashi et al, 2004; Israsena et al, 2004; Kuwahara et al, 2010). Wnt7a and Wnt7b promote NPC proliferation (Viti et al, 2003; Qu et al, 2013). In contrast, mice mutant for LRP6 (low‐density lipoprotein receptor‐related protein 6), the principle co‐receptor for the canonical pathway, exhibit normal NPC proliferation but decreased neuronal differentiation (Zhou et al, 2006). Further complicating issues, expression of Wnt3a in the neocortex of mouse embryos by in utero electroporation leads to both AP self‐renewal and neuronal differentiation of BPs (Munji et al, 2011).

Taken together, the precise role of WNT signalling in mouse neocortex development remains unclear. Does canonical WNT signalling promote self‐renewal or differentiation of NPCs? One solution to the observed discrepancies could be that WNT signalling regulates both progenitor self‐renewal and differentiation but in different populations of NPCs, namely self‐renewal in APs and differentiation in BPs (Munji et al, 2011). However, another possible explanation for the apparent inconsistencies may be that manipulation of WNT signalling components can exert different effects depending on whether the manipulated WNT effectors function upstream or downstream in the pathway. This is because WNT/LRP6 signalling triggers various sub‐pathways that are differentially affected by manipulation at the ligand, receptor, or intracellular levels (Acebron & Niehrs, 2016; García de Herreros & Duñach, 2019).

The main function of LRP6‐dependent WNT signalling is to inhibit GSK3, and it is still widely assumed that the only relevant GSK3 substrate in the WNT pathway is β‐catenin (Nusse & Clevers, 2017). However, in addition to β‐catenin, GSK3 can phosphorylate many other proteins and target them for proteasomal degradation (Taelman et al, 2010; Acebron et al, 2014). In this context, a novel player is the WNT/STOP (WNT‐stabilization of proteins) pathway (see Fig EV1A for a model), which acts post‐transcriptionally, is independent of β‐catenin, peaks during mitosis, and slows down degradation of numerous proteins as cells prepare to divide (Acebron et al, 2014). Key effectors of WNT/STOP signalling are cyclin Y (Ccny) and cyclin Y‐like 1 (Ccnyl1), conserved cyclins that together with their cyclin‐dependent kinases (CDK) 14 and 16 phosphorylate and activate the WNT co‐receptor LRP6 in G2/M. This co‐receptor activation leads to a peak of WNT signalling and GSK3 inhibition in mitosis (Davidson et al, 2009). In vivo, the WNT/STOP pathway plays a role in germ cells (Huang et al, 2015; Koch et al, 2015) and cancer cells (Madan et al, 2018; Hinze et al, 2019).

Figure EV1. Data supporting Fig 1 .

• A
  WNT/LRP6/GSK3 signalling. Model illustrating the WNT/β‐catenin and WNT/STOP signalling pathways. Ccny/l1 (CycY), together with their cyclin‐dependent kinases (CDKs) (not shown), phosphorylate and prime the WNT co‐receptor LRP6 for incoming WNT ligands. Binding of WNT to LRP6 inhibits GSK3, leading to increased β‐catenin‐mediated transcription of target genes. Inhibition of GSK3 also leads to increased protein levels via the stabilization of proteins (STOP) pathway, which peaks during the G2/M phase of the cell cycle (not shown).
• B
  DKO neocortex displays fewer newly formed neurons. IF for Tuj1 in E13.5 control and DKO neocortices. Area above white dashed lines (marked by white open boxes) depicts Tuj1⁺ area. Numbers are quantification of the length (shown as percentage of total cortex thickness) of the Tuj ⁺ area ± SEM (n = 4 embryos, 3 litters). Scale bar 20 μm.
• C
  The radial glial cell scaffold is unaltered in DKO neocortex. IF for the intermediate filament marker nestin to visualize the radial glial scaffold in E13.5 control and DKO neocortices. No obvious differences are detected. Three independent control and DKO neocortices were analysed. Scale bar 20 μm.
• D
  No apical–basal polarity defects are detected in DKO neocortex. No obvious differences in apical–basal polarity are detected in E13.5 DKO neocortex when compared to controls, as demonstrated by IF for the adherens junction protein β‐catenin. Three independent control and DKO embryos were analysed. Scale bar 20 μm.
• E, F
  Validation of Ccny and Ccnyl1 antibodies. IF using homemade polyclonal antibodies against Ccny (E) and Ccnyl1 (F) in control and DKO E13.5 neocortices. Note the absence of Ccny immunoreactivity (indicated by white arrowheads) in DKO neocortex, and the absence of Ccnyl1 immunoreactivity (indicated by white arrowheads) in DKO neocortex. All scale bars 50 μm.
• G, H
  Ccny/l1 localize to the apical membrane in APs and appear as puncta in BPs. (G) Triple IF for Ccny or Ccnyl1 plus Sox2 and GFP in E15.5 embryos electroporated with GFP plasmid at E13.5. White arrows and dashed outlines depict Ccny/l1 immunoreactivity at the apical membrane of APs in the ventricular zone (VZ). Asterisks depict nuclei of cells analysed. Scale bar 5 μm. (H) Triple IF for Ccny or Ccnyl1 plus Tbr2 and GFP in E15.5 embryos electroporated with GFP plasmid at E13.5. White arrows and dashed outlines depict Ccny/l1 immunoreactivity as single puncta in BPs in the subventricular zone (SVZ). Scale bar 2 μm.
• I
  pLRP6 (S1490) is detected in mitotic neural progenitors. Co‐IF for pLRP6 (S1490) and pHH3 in E13.5 wild‐type neocortex. Area in dashed box in left panel is shown at higher magnification in right panel. Mitotic BP (division in basal VZ/SVZ, left panel, arrow and inset) and mitotic APs (division at apical membrane / cortex, right panel, arrow and inset) show high pLRP6 immunoreactivity. Scale bar left panel 50 μm, right panel 20 μm.
• J
  Ccny and Ccnyl1 are detected in mitotic neural progenitors. Ccny and Ccnyl1 co‐IF with pHH3 showing mitotic APs at the ventricular surface of E13.5 wild‐type neocortex. Ccny/l1 immunoreactivity indicated by white arrowheads. Scale bar 2 μm.
• K
  WNT/STOP signalling components are enriched in the VZ. IF for LRP6 (total levels), cyclin‐dependent kinase 14 (CDK14) and glycogen synthase kinase 3β (GSK3β) on the neocortex of control E13.5 embryos. In all cases, immunoreactivity indicated by white arrowheads. Scale bar 40 µm.
• L
  Active β‐Catenin levels are not significantly altered in DKO forebrains. Immunoblot analysis of control and DKO E13.5 forebrain lysates with an antibody against active (dephoshorylated) β‐catenin. Number above blot are quantification of active β‐catenin levels normalized to total β‐catenin and are expressed as fold change vs. controls (n = 4 embryos, 2 litters).

Data information: Unpaired two‐tailed t‐test for all statistical analyses: ns, not significant. P = 0.048 (B, DKO); P = 0.99 (L).Source data are available online for this figure.

Here, we examine the in vivo role of WNT/STOP signalling in NPCs during multiple phases of mouse neocortex development. Genetic ablation of Ccny and Ccnyl1 leads to a thinner cerebral cortex and a reduced number of BPs and post‐mitotic neurons. Importantly, Ccny/l1‐deficient mice display decreased WNT signalling at the receptor level, but not at the transcriptional level. Through a series of in vivo and in vitro analyses, we show that WNT/STOP signalling is essential for asymmetric AP division, cell cycle progression of BPs and neuron generation. Mechanistically, Ccny/l1 stimulate neuronal differentiation of BPs through post‐transcriptional regulation of Sox4 and Sox11, two essential neurogenic transcription factors (Bergsland et al, 2006; Chen et al, 2015) that we identify as direct GSK3 targets. We therefore propose that WNT/STOP signalling is the primary driver of neuronal differentiation of NPCs whereas WNT/β‐catenin signalling predominantly regulates NPC self‐renewal.

Results

WNT/STOP signalling is required for neurogenesis in the embryonic mouse neocortex

To study the role of WNT/STOP signalling in the developing mouse neocortex, we generated embryos deficient for Ccny and Ccnyl1 (hereafter referred to as double knockout (DKO) embryos). In contrast to individual Ccny or Ccnyl1‐deficient embryos, which are both viable (An et al, 2015; Koch et al, 2015), DKO embryos displayed in utero death beginning at embryonic day 14.5 (E14.5). To avoid that the results of our analyses of DKO embryos might reflect non‐specific effects resulting from early lethality, we analysed DKO and littermate control embryos at E13.5. Analysis of haematoxylin–eosin (HE)‐stained DKO forebrains revealed a significantly thinner neocortical wall (−32%, P = 0.0004) compared to littermate controls. Mediolateral neocortex length was not significantly changed (Fig 1A; quantified in Appendix Fig S1A and B).

Figure 1. WNT/STOP signalling is required for neurogenesis in the embryonic mouse neocortex.

1. Haematoxylin and eosin (HE) staining on sections of the neocortex from control (Co) and cyclin Y (Ccny) and cyclin Y‐like 1 (Ccnyl1) double knockout (DKO) embryos at E13.5. Neocortex thickness (indicated by yellow bars) is reduced in DKO vs. control embryos while mediolateral neocortex length (indicated by pink arrows) does not display obvious differences. Scale bars 100 μm.
2. Hoechst (DNA) staining and IF for Tbr2 to calculate the thickness (white bars) of the VZ, SVZ and intermediate zone + cortical plate (IZ+CP) in control and DKO E13.5 neocortex. Scale bar 20 μm.
3. Quantification of VZ, SVZ and IZ+CP thickness in control and DKO E13.5 neocortex. Each zone is represented as percentage of total thickness (measured in µm) of the neocortex. Columns are means ± SEM (n = 6 embryos, 4 litters).
4. IF for Pax6 and Tbr2 (not shown) in control and DKO E13.5 neocortex. White bars indicate thickness of Pax6⁺ zone (i.e. VZ). Scale bar 20 μm.
5. Quantification of the number of APs (Pax6⁺Tbr2⁻) from (D). Pax6⁺Tbr2⁺ cells were excluded from final counts. AP numbers are expressed as percentage of total cell number (sum of D, F and H). Columns are means ± SEM (n = 4 embryos, 3 litters).
6. IF for Tbr2 in control and DKO E13.5 neocortex. White bars indicate thickness of Tbr2⁺ zone (i.e. SVZ). Scale bar 20 μm.
7. Quantification of the number of BPs from (F). Columns are means ± SEM (n = 6 embryos, 3 litters).
8. IF for Tbr1 in control and DKO E13.5 neocortex. White bars indicate thickness of Tbr1⁺ zone (i.e. mostly CP). Scale bar 15 μm.
9. Quantification the number of deep‐layer post‐mitotic neurons from (H). Columns are means ± SEM (n = 3 embryos, 3 litters).
10. IF for Ccny (a), Ccnyl1 (b) and LRP6 (c) (low‐density lipoprotein receptor‐related protein 6) phosphorylated at T1479 (pLRP6) in the neocortex of E13.5 control embryos. White arrowheads in (a,b) indicate apical membrane / cortex, shown at higher magnification in the insets. High levels of LRP6 phosphorylation are detected in mitotic APs (c, white arrowheads). Scale bar 50 μm.
11. IF for Ccny(a), Ccnyl1(b) and pLRP6(c), each together with IF for Tbr2, in the neocortex of E12.5 control embryos. White arrowheads in the insets of (a, b) depict Ccny/l1 puncta in Tbr2⁺ BPs. High levels of LRP6 phosphorylation are detected in mitotic BPs (c, inset, white arrowhead). DNA staining for mitotic pLRP6⁺ BP shown in bottom right inset (c, green arrowhead). Scale bar 20 μm.
12. IF for pLRP6 (T1479) in the neocortex of E13.5 control and DKO embryos. White arrowheads depict mitotic APs and purple arrowheads depict mitotic BPs. Scale bar 20 μm.
13. Quantification of relative pLRP6 staining intensity in mitotic APs and BPs from (L). Data are expressed as fold change vs. controls and are means ± SEM (n = 3 embryos, 3 litters).
14. Immunoblot analysis with an antibody against LRP6 phosphorylated at S1490 (pLRP6 (S1490)) of protein lysates from E13.5 control and DKO forebrains. Numbers above blot indicate fold change vs. controls of phospho‐LRP6 (S1490) protein levels normalized to total LRP6 (n = 6 embryos, 3 litters).
15. Expression analysis of the β‐catenin target gene Axin2 by RNAScope in situ hybridization on neocortex sections from E13.5 control and DKO embryos. Scale bar 20 μm.
16. Quantification of (O). Axin2⁺ area was calculated and normalized to total neocortex area. Data are expressed as fold change vs. controls and are means ± SEM (n = 4 embryos, 3 litters).
17. Axin2 and N‐myc qPCR expression analysis on RNA extracted from E13.5 dorsal forebrains. Data are expressed as fold change vs. controls and are means ± SEM (n = 3 embryos, 2 litters).

Data information: Unpaired two‐tailed t‐test for all statistical analyses: ns, not significant; *P < 0.05, **P < 0.01, ***P < 0.001. P = 0.00040 (A); P = 0.016 (C, VZ); P = 0.031 (C, SVZ); P = 0.042 (C, IZ +CP); P = 0.039 (E); P = 0.00060 (G); P = 0.011 (I); P = 0.010 (M); P = 0.0010 (N); P = 0.85 (P); P = 0.95 (Q, Axin2); P = 0.41 (Q, N‐myc)

Source data are available online for this figure.

To further dissect the reduction in the thickness of the neocortical wall observed in E13.5 DKO embryos, we measured VZ, SVZ and intermediate zone (IZ) plus cortical plate (CP) thickness upon DNA staining and immunofluorescence microscopy (IF) for T‐box brain protein 2 (Tbr2), a BP marker that permitted visualization of the SVZ. This revealed that within the thinner neocortical wall of DKO embryos, the proportion of neocortical wall thickness constituted by the VZ was increased by 12%, whereas that constituted by the SVZ and IZ+CP was decreased by 20% and 18%, respectively, when compared to the respective proportions in the thicker neocortical wall of control embryos (Fig 1B and C; Appendix Fig S1C and F).

To corroborate these observations, we quantified the number of APs, BPs and post‐mitotic deep‐layer neurons upon IF for their respective markers Paired box protein 6 (Pax6), Tbr2, and T‐box brain protein 1 (Tbr1). The percentage of Pax6⁺ cells in the thinner neocortex of E13.5 DKO embryos was slightly increased (+11%, P = 0.04) whereas the percentages of Tbr2⁺ and Tbr1⁺ cells were decreased (−38%, P = 0.0006; −25%, P = 0.01; respectively) when compared to the thicker neocortex of control embryos (Fig 1D–I). Consistent with this, IF for βIII‐tubulin (Tuj1), which marks newborn neurons, revealed that in the thinner neocortical wall of E13.5 DKO embryos, the layers containing newborn neurons comprised a lesser proportion of the neocortical wall thickness than in the thicker neocortical wall of control embryos (control 50.9 ± 5.5 vs. DKO 33.2 ± 1.0 (% of total cortex thickness), P = 0.048; Fig EV1B). Together, these data indicated that the reduction in neocortical wall thickness in E13.5 DKO embryos was primarily due to a decrease in the levels of BPs and newborn neurons, which in turn suggested that cortical neurogenesis is reduced in DKO embryos.

Decreased BP and, consequently, post‐mitotic neuron levels can be due, at least in part, to structural defects in neocortical cytoarchitecture, such as improper organization of the radial glial scaffold and disruption of apical‐basal polarity, which may impede migration of BPs to the SVZ and of newborn neurons to the CP (Taverna et al, 2014). However, IF against the radial glia‐specific intermediate filament marker nestin revealed no overt abnormalities in the radial glial scaffold of E13.5 DKO neocortex when compared to control (Fig EV1C). Also, we observed normal enrichment of β‐catenin at the apical cell cortex of DKO forebrains, suggesting that apical–basal polarity was not affected in the absence of Ccny/l1 (Fig EV1D). We conclude that DKO embryos display neurogenesis defects but no major structural abnormalities in the embryonic neocortex.

We performed IF for Ccny and Ccnyl1 in the E12.5‐13.5 mouse neocortex. Interestingly, both Ccny and Ccnyl1 immunoreactivity was concentrated at the apical cell cortex/apical plasma membrane of the VZ of control embryos (Fig 1J‐a,b). No Ccny or Ccnyl1 immunoreactivity was detected in the neocortex of DKO embryos, confirming that the two polyclonal Ccny and Ccnyl1 antibodies used are specific (Fig EV1E and F). Ccny and Ccnyl1 immunoreactivity was also detected in, respectively, 28 ± 0.63% and 24 ± 0.68% of Tbr2⁺ BPs in the SVZ (Fig 1K‐a,b). Ccny/l1 were not detected in post‐mitotic neurons. To further analyse the distribution of Ccny/l1 protein in APs and BPs, we performed in utero electroporation (IUE) in E13.5 embryos with a plasmid coding for green fluorescent protein (GFP) and analysed embryos at E15.5. Triple IF for Ccny/l1, the AP marker Sox2 (SRY (sex determining region Y)‐2), and GFP confirmed Ccny/l1 immunoreactivity at the apical membrane of electroporated APs (Fig EV1G), while triple IF for Ccny/l1, Tbr2 and GFP revealed Ccny/l1 immunoreactivity as single puncta in BPs (Fig EV1H). In light of these observations, we next performed IF with an LRP6 antibody specific for the casein kinase 1 gamma (CK1γ) phosphorylation site T1479, which marks active WNT signalling (Davidson et al, 2005). Similar to the Ccny and Ccnyl1 immunoreactivity, phospho‐T1479 LRP6 immunoreactivity was also found to be concentrated at the apical cell cortex / apical plasma membrane of the VZ of E13.5 control mouse neocortex (Fig 1J‐c). This concentration reflected the specific enrichment of phospho‐LRP6 immunoreactivity in 91% of mitotic APs analysed (Fig 1J‐c, arrowheads, inset). The apical concentration of phospho‐LRP6 immunoreactivity in mitotic APs is consistent with the fact that mitosis of APs typically occurs at the ventricular surface of the neocortex (Taverna et al, 2014). Phospho‐T1479 LRP6 immunoreactivity was also observed in 48% of mitotic BPs of E12.5 control mouse neocortex (Fig 1K‐c, inset). To confirm this pattern of immunostaining, we performed IF on control embryonic mouse neocortex with another phospho‐LRP6 antibody that detects the CDK14 priming phosphorylation site S1490 (Davidson et al, 2009). Again, phospho‐S1490 LRP6 immunoreactivity was enriched in E13.5 mitotic APs (90%) and BPs (88%), with the mitotic stage of these progenitors being confirmed by co‐immunostaining with the mitotic marker phospho‐histone H3 (pHH3) (Fig EV1I, insets and arrowhead). Mitotic APs also showed Ccny (30%) and Ccnyl1 (29%) immunoreactivity (Fig EV1J). IF for total LRP6, CDK14 and GSK3β revealed the greatest concentration of immunoreactivity at the apical cell cortex / apical plasma membrane of the VZ of E13.5 control mouse neocortex (Fig EV1K). Altogether, these data suggest that the core components of the WNT/STOP signalling pathway are expressed in the embryonic mouse neocortex and that WNT/LRP6 signalling peaks during mitosis in APs and BPs.

To determine whether WNT signalling in DKO forebrains was affected at the receptor level, we quantified the immunostaining intensity for active phospho‐LRP6 (T1479) within mitotic APs and BPs of the E13.5 neocortex and detected a decrease (−15%, P = 0.01) when compared to controls (Fig 1L and M). Phosphorylation of LRP6 at S1490 was even more markedly reduced (−45%, P = 0.001) when analysed by immunoblotting of protein lysates extracted from E13.5 DKO and control forebrains (Fig 1N). To monitor canonical WNT signalling, we performed RNAScope analysis on sections of E13.5 neocortex using a probe against the β‐catenin target gene Axin2. Axin2 expression was not significantly changed in the neocortex of DKO embryos when compared to controls (Fig 1O and P). To confirm this result, we extracted RNA from E13.5 dorsal forebrains and performed qPCR analysis. Expression of Axin2 and N‐myc, another WNT target gene in the neocortex (Kuwahara et al, 2010), was not significantly altered in DKO dorsal forebrains (Axin2, P = 0.95; N‐myc, P = 0.41) (Fig 1Q). Furthermore, immunoblot analysis of whole forebrain lysates probed with an antibody against dephosphorylated β‐catenin, which represents its active form, showed no significant change in E13.5 DKO embryos (Fig EV1L). We conclude that combined Ccny/l1 deficiency in the embryonic mouse neocortex leads to decreased LRP6 receptor activation without changes in β‐catenin activity. Together with the reduction in neuron levels in the DKO neocortex, these data are consistent with WNT/STOP signalling being required for neurogenesis in the embryonic mouse neocortex.

DKO embryos show delayed cell cycle progression and increased mitosis length in BPs

In light of the reduced levels of BPs and neurons in the neocortex of DKO embryos, we analysed cell cycle parameters of APs and BPs in control and DKO embryos, as alterations in cell cycle progression have been shown to affect NPC fate and cortical neurogenesis (Götz & Huttner, 2005; Dehay & Kennedy, 2007; Arai et al, 2011; Borrell & Calegari, 2014). To label NPCs in S‐phase in the neocortex of control and DKO embryos, we injected the thymidine analog bromo‐deoxyuridine (BrdU) at E11.5, E12.5 and E13.5 and sacrificed mice 1 h later. Co‐IF for Pax6, Tbr2 and BrdU did not reveal any major difference between control and DKO neocortex in the proportion of APs (Pax6⁺ Tbr2⁻) that were in S‐phase (i.e. BrdU⁺) at E11.5 and E12.5, although a slight increase was detected for DKO neocortex at E13.5 (Fig 2A and B). The average percentage of E13.5 BrdU⁺ APs, i.e. APs in S‐phase, was 31% for control and 39% for DKO neocortex (Fig 2C). The proportion of BPs (Tbr2⁺) that were in S‐phase (BrdU⁺) was moderately decreased in DKO neocortex compared to control at all time points analysed (Fig 2D and E). The average percentage of E13.5 BrdU⁺ BPs, i.e. BPs in S‐phase, was 24% for control and 21% for DKO neocortex (Fig 2F).

Figure 2. DKO embryos show delayed cell cycle progression and increased mitosis length in BPs.

1. Triple‐IF for BrdU, Pax6 and Tbr2 (not shown) in E12.5 control and DKO neocortices. BrdU was injected into pregnant dams 1 h before sacrifice.
2. Quantification of the proportion of Pax6⁺Tbr2⁻ cells that are BrdU⁺, in control and DKO neocortices at E11.5, E12.5 and E13.5. Data are expressed as fold change vs. controls, with the control values set to 1.0, and are means ± SEM (E11.5 n = 4 embryos, 3 litters; E12.5 n = 5 embryos, 5 litters and E13.5 n = 5 embryos, 4 litters).
3. Average percentage of APs (Pax6⁺Tbr2⁻) (from B) that are BrdU⁺, in E13.5 control and DKO neocortices. Data are means ± SEM.
4. Co‐IF for Tbr2 and BrdU (double‐positive cells depicted by white arrowheads) in E12.5 control and DKO neocortices. BrdU was injected into pregnant dams 1 h before sacrifice.
5. Quantification of the proportion of Tbr2⁺ cells that are BrdU⁺, in control and DKO neocortices at E11.5, E12.5 and E13.5. Data are expressed as fold change vs controls, with the control values set to 1.0, and are means ± SEM (E11.5 n = 4 embryos, 3 litters; E12.5 n = 5 embryos, 5 litters and E13.5 n = 5 embryos, 4 litters).
6. Average percentage of BPs (Tbr2⁺) (from E) that are BrdU⁺, in E13.5 control and DKO neocortices. Data are means ± SEM.
7. Double IdU/BrdU injection strategy used to calculate S‐phase length. Top, scheme. Bottom, triple IF for IdU/BrdU/Tbr2 to detect APs (Tbr2⁻) and BPs (Tbr2⁺) that have exited S‐phase (IdU⁺/BrdU⁻, white arrowheads) in control and DKO E13.5 neocortices.
8. Average percentages (numbers in bar graphs) of the IdU⁺ cells that are either BrdU⁺ or BrdU⁻, scored as in (G), for control and DKO APs and BPs. Data are means ± SEM (n = 4 embryos, 3 litters).
9. S‐phase length of E13.5 control and DKO APs and BPs, calculated from the data of panel (H, yellow) as described in Appendix Methods. Data are means ± SEM (n = 4 embryos, 3 litters).
10. Total cell cycle length of E13.5 control and DKO APs and BPs, calculated from the data of panels (C), (F) and (I) as described in Appendix Methods. Data are means ± SEM (n = 4 embryos, 3 litters).
11. Co‐IF for the proliferation marker Ki67, the mitotic marker phospho‐histone H3 (pHH3), and Tbr2 to identify mitotic APs (division at apical surface) and BPs (division in basal VZ or SVZ) in control and DKO E13.5 neocortices. Arrowheads depict pHH3⁺Tbr2⁺ cells.
12. Average percentage of Ki67⁺ APs and BPs that are pHH3⁺ in control and DKO E13.5 neocortices. Data are means ± SEM (n = 4 embryos, 3 litters).
13. Mitosis length of E13.5 control and DKO APs and BPs, calculated from the data of panels (J) and (L) as described in Appendix Methods. Data are means ± SEM (n = 4 embryos, 3 litters).
14. IF for cleaved caspase 3 (Casp3) to identify apoptotic cells in E13.5 control and DKO neocortices.
15. Quantification of apoptotic cells in control and DKO neocortices at E11.5, E12.5 and E13.5. Data are represented as number of Casp3⁺ cells per 5000 cells. Data are means ± SEM (E11.5 n = 4 embryos, 3 litters; E12.5 n = 3 embryos, 3 litters, and E13.5 n = 4 embryos, 3 litters).

Data information: All scale bars 20 μm. Unpaired two‐tailed t‐test for all statistical analyses: ns, not significant; *P < 0.05, **P < 0.01. P = 0.77 (B, E11.5); P = 0.16 (B, E12.5); P = 0.030 (B, E13.5); P = 0.014 (C); P = 0.017 (E, E11.5); P = 0.010 (E, E12.5); P = 0.013 (E, E13.5); P = 0.089 (F); P = 0.0069 (H, APs); P = 0.0078 (H, BPs); P = 0.018 (I, AP); P = 0.027 (I, BP); P = 0.43 (J, AP); P = 0.016 (J, BP); P = 0.35 (M, AP); P = 0.012 (M, BP); P = 0.24 (O, E11.5); P = 0.56 (O, E12.5); P = 0.26 (O, E13.5).

Source data are available online for this figure.

We next determined the length of S‐phase of APs and BPs in E13.5 control and DKO neocortex. To this end, we performed timed injections of the thymidine analogs iodo‐deoxyuridine (IdU) and BrdU. Briefly, IdU was injected at T = 0 to label APs and BPs in S‐phase, and BrdU was injected at T = 1.5 h to identify those APs and BPs that were still in S‐phase at this time vs. those that had left S‐phase. Embryos were collected 30 min after BrdU injection (Fig 2G, schematic). We then extrapolated from the percentage of APs (Tbr2⁻) and BPs (Tbr2⁺) of control and DKO neocortex that were IdU⁺ but BrdU⁻, i.e. that had left S‐phase after 1.5 h (Fig 2H, yellow; Appendix Fig S1G and H), and determined the time when all control and DKO APs and BPs would have left S‐phase, which yielded the length of S‐phase for control and DKO APs and BPs (Fig 2I). This revealed a small increase in S‐phase length of E13.5 DKO APs as compared to control APs, and a ≈ 50% increase in S‐phase length of E13.5 DKO BPs as compared to control BPs (Fig 2I).

Knowing the percentage values of E13.5 control and DKO APs and BPs in S‐phase (Fig 2C and F), and the length of S‐phase of these types of NPCs (Fig 2I), allowed us to calculate the total length of the cell cycle of E13.5 control and DKO APs and BPs by dividing the S‐phase length values (Fig 2I) by the percentage values for these NPCs in S‐phase (Fig 2C and F) and then multiplying the resulting numbers with 100 (Fig 2J). This revealed no difference between E13.5 control and DKO APs, but a nearly doubling of total cell cycle length in DKO BPs as compared to control (Fig 2J).

To calculate the length of mitosis, we first performed co‐IF for pHH3, Ki67 and Tbr2 (Fig 2K) to determine the percentage of cycling (Ki67⁺) E13.5 control and DKO APs (Tbr2⁻) and BPs (Tbr2⁺) that were in mitosis (pHH3⁺) (Fig 2L). This revealed no major differences between control and DKO APs, nor between control and DKO BPs (Fig 2L). We then calculated the length of mitosis of E13.5 control and DKO APs and BPs by multiplying the total cell cycle length values for each of these NPCs (Fig 2J) with the percentage values for the respective NPC in mitosis (Fig 2L) and divided the resulting numbers by 100, which yielded the length of mitosis (Fig 2M). While no significant difference in mitosis length was found between control and DKO APs, mitosis length of DKO BPs was found to be more than doubled as compared to control (Fig 2M). Data consistent with these findings were obtained when the proportion of APs and BPs in mitosis, deduced from ventricular and abventricular pHH3⁺ cells, respectively, were compared between E11.5‐E13.5 control and DKO neocortex (Fig EV2A–C). All cell cycle calculations are detailed in the Appendix Methods.

Figure EV2. Data supporting Fig 2 .

• A–C
  Increased proportion of APs and BPs in mitosis in DKO neocortex (A) Co‐IF for pHH3 and Tbr2 (not shown) in E13.5 control and DKO neocortices. Mitotic APs and BPs were identified by their respective locations at the ventricular surface (arrowhead, left panel) and in the SVZ (asterisk, left panel), respectively. Scale bar 20 μm. (B, C) Quantification of the percentages of APs (B) and BPs (C) in mitosis in E11.5, E12.5 and E13.5 control and DKO neocortices. For APs, the percentages of Tbr2⁻ cells in the VZ+SVZ that were pHH3⁺ were determined. For BPs, the percentages of Tbr2⁺ cells in the VZ+SVZ that were pHH3⁺ were determined. Data are expressed as fold change vs. controls, with the control values set to 1.0, and are means ± SEM (E11.5 and E13.5 n = 5 embryos, E12.5 n = 4 embryos (B) n = 5 embryos (C)).
• D
  No changes in apoptosis in DKO neocortex. Quantification of TUNEL⁺ apoptotic cells in control and DKO E13.5 neocortices. The number of TUNEL⁺ cells per 5,000 cells is shown. Data are means ± SEM (n = 4 embryos, 4 litters).

Data information: Unpaired two‐tailed t‐test for all statistical analyses: ns, not significant; *P < 0.05. P = 0.031 (B, E11.5); P = 0.034 (B, E12.5); P = 0.030 (B, E13.5); P = 0.034 (C, E11.5); P = 0.046 (C, E12.5); P = 0.030 (C, E13.5); and P = 0.94 (D).Source data are available online for this figure.

Mitotic delay can lead to cell cycle arrest and increased apoptosis (Chen et al, 2014; Pilaz et al, 2016). To compare apoptosis levels in DKO vs. control forebrains, we performed IF with an antibody against cleaved caspase 3 (Fig 2N). Except for a minor, however statistically not significant, increase at E11.5, apoptosis levels in the E12.5 and E13.5 neocortex of DKO embryos were unchanged when compared to controls (Fig 2O). This was confirmed by TUNEL staining, which at E13.5 was also not significantly altered in the neocortex of DKO embryos (Fig EV2D). Thus, apoptosis is unlikely to explain reduced thickness of the cortical plate in DKO embryos. We conclude that the reduction in DKO embryos of cortical plate thickness involves delayed cell cycle progression and increased mitosis length in BPs.

Lack of Ccny/l1 expression reduces asymmetric AP division and neurogenesis in the embryonic neocortex

The relative increase in the thickness of the VZ and the reduced BP levels in DKO embryos raised the possibility that their thinner neocortex was not only due to a delayed cell cycle progression of BPs. Instead, the results suggested that Ccny/l1 may affect the generation of BPs from APs, and consequently of post‐mitotic neurons. In the mammalian neocortex, the switch of APs from symmetric, proliferative divisions to asymmetric, BP‐genic divisions is often associated with changes in apical membrane distribution, whereby unequal inheritance of the apical plasma membrane by the daughter cells indicates an asymmetric mode of AP division (Kosodo et al, 2004; Delaunay et al, 2017). We quantified the number of symmetrically vs. asymmetrically dividing APs by IF using the “cadherin hole” method, which allows determination of the plane of division in mitotic APs (Fig 3A) (Kosodo et al, 2004). Interestingly, E13.5 DKO neocortex displayed a 36% reduction in asymmetric AP division (from 47 to 30% of all AP division, Fig 3B), consistent with decreased generation of BPs in the absence of Ccny/l1.

Figure 3. Lack of Ccny/l1 expression reduces asymmetric AP division and neurogenesis in the embryonic neocortex.

• A
  IF for pan‐cadherin, combined with DNA staining, to quantify asymmetric vs. symmetric division in VZ APs. Dashed white lines denote the cleavage plane, and solid lines mark the “cadherin hole”. Asterisks mark the apical‐most basolateral membrane flanking the “cadherin hole”. Cell divisions were scored as symmetric or asymmetric when the cleavage plane bisects (left) or bypasses (right) the cadherin hole, respectively. Scale bar 2 μm.
• B
  Quantification of symmetric vs. asymmetric AP division in E13.5 control and DKO neocortices, expressed as percentage of total AP divisions. Four independent embryos per genotype from 3 litters were analysed. A total of 110 and 99 cells were counted for controls and DKOs, respectively. Data are means ± SEM.
• C
  IF for alpha‐tubulin to visualize mitotic spindle and astral microtubules in E13.5 control and DKO neocortices. Central and apical–basal astral microtubules denoted by white and purple arrows, respectively. Scale bar 2 μm.
• D
  Quantification of apical–basal, central and total (sum) number of astral microtubules in mitotic APs of control and DKO E13.5 neocortices. Data are means of microtubules per cell ± SEM (at least 10 cells counted per embryo)(n = 3 embryos, 3 litters).
• E, F
  In utero electroporation (IUE) of plasmids expressing GFP plus shControl (Co) or shCcny/l1 into the lateral ventricles of E13.5 wild‐type embryos, followed by analysis at E15.5 or E17.5. (E) IF for GFP and Tbr2 in the E15.5 neocortex. (F) IF for GFP and the deep‐layer neuron marker CTIP2 in the E17.5 neocortex. (E, F) White arrowheads denote double‐positive cells. Insets of magnified cells (indicated by blue arrowheads) depict representative GFP + Tbr2 (E) or GFP + Ctip2 (F) co‐staining. Scale bars 20 μm.
• G
  Quantification of the percentage of GFP⁺ cells that are Tbr2⁺, in control and shCcny/l1 neocortices at E15.5 upon IUE at E13.5. Data are means ± SEM (n = 3 embryos).
• H
  Quantification of the percentage of GFP⁺ cells that are Ctip2⁺, in control and shCcny/l1 neocortices at E17.5 upon IUE at E13.5. Data are means ± SEM (n = 4 embryos).
• I
  IF for βIII‐tubulin (Tuj1) of control (Ccny ^+/−; Ccnyl1 ^−/−; shRNA Co) and mutant (Ccny ^+/−; Ccnyl1 ^−/−; shRNA Ccny) NPC monolayer cultures grown in differentiation medium for seven days. Insets depict neurons with representative neurite morphology for controls and mutants. Scale bar 20 μm.
• J
  Quantification of the percentage of Tuj1⁺ cells from (I). Data are means ± SEM (experiment performed twice in triplicates, representative experiment shown).
• K
  Immunoblot analysis of protein lysates extracted from control and mutant NPCs prior to culture in differentiation medium, using an antibody against lysine 48‐ubiquitin (K48‐Ub) to label polyubiquitinated proteins targeted for proteasomal degradation. GAPDH, loading control. Two representative samples per genotype are shown.

Data information: Unpaired two‐tailed t‐test for all statistical analyses: ns, not significant; *P < 0.05, **P < 0.01. P = 0.034 (both asterisks) (B); P = 0.0062 (D, apical‐basal); P = 0.22 (D, central); P = 0.0022 (D, sum); P = 0.0030 (G); P = 0.030 (H); P = 0.0053 (J).

Source data are available online for this figure.

Mitotic spindle orientation is critical for determining symmetric vs. asymmetric cell division of NPCs in the developing neocortex (Konno et al, 2008; Yingling et al, 2008; Lizarraga et al, 2010; Asami et al, 2011; LaMonica et al, 2013; Xie et al, 2013; Mora‐Bermúdez & Huttner, 2015), with a key role of astral microtubules (aMTs) (Mora‐Bermúdez et al, 2014). A specific subpopulation of aMTs, which reach the apical or basal cell cortex and are referred to as apical‐basal aMTs, are more abundant in symmetrically dividing APs than in asymmetrically dividing APs (Mora‐Bermúdez et al, 2014). These apical–basal aMTs promote a mitotic spindle orientation perpendicular to the apical–basal axis of APs and reduce the variability of mitotic spindle orientation, which in turn favours symmetric AP division (Mora‐Bermúdez et al, 2014). Perturbations specifically of these apical–basal aMTs increases asymmetric BP‐genic APs divisions and neurogenesis (Mora‐Bermúdez et al, 2014). To examine aMT abundance, we performed IF for α‐tubulin, acquired Z‐stack images of dividing APs and quantified the number of apical–basal aMTs vs. the aMTs reaching the central cell cortex (central aMTs) (Fig 3C). Strikingly, E13.5 DKO APs exhibited an increase specifically in apical‐basal, but not central, aMTs (Fig 3D), providing a mechanistic explanation for the increase in symmetric AP division (Fig 3B). We conclude that in dividing APs, lack of Ccny/l1 expression results in an increase in the number of apical–basal aMTs, which reduces asymmetric, BP‐genic cell division and hence neurogenesis in the embryonic neocortex.

DKO embryos display premature embryonic lethality, limiting our analysis of neurogenesis to early stages of development in which the cortical plate is less developed. To analyse the effect of Ccny/l1 knockdown in NPCs at later time points, we performed IUE of mouse embryos at E13.5 with plasmids coding for either a control shRNA (Co) or shRNAs against Ccny and Ccnyl1 (shCcny/l1), along with a plasmid coding for GFP. At E15.5, knockdown of Ccny/l1 greatly reduced the proportion of the GFP⁺ progeny of the electroporated cells that were Tbr2⁺, i.e. BPs (Fig 3E and G). Similarly, at E17.5, the proportion of the GFP⁺ progeny of the electroporated cells that expressed the deep‐layer neuron marker Ctip2 was depleted (Fig 3F and H). Altogether, these data corroborate that lack of Ccny/l1 expression reduces neurogenesis in the embryonic mouse neocortex and show that this effect is consistent throughout embryonic development. The IUE results also exclude the possibility that the neurogenesis defect observed in DKO embryos is due to an overall delay in embryonic development, or results from indirect effects related to global deletion of Ccny/l1.

We next asked whether NPCs deficient for Ccny/l1 also display a reduced capacity to generate neurons in vitro. To this end, we isolated NPCs from E13.5 Ccny ^+/−; Ccnyl1 ^−/− forebrains and transduced them with a lentivirus expressing shRNA against Ccny (hereafter referred to as mutant NPCs) to induce acute Ccny/l1 deficiency. NPCs were then cultured in differentiation media to promote neurogenesis. IF for Tuj1 revealed a drastic decrease in the percentage of cells expressing this marker of newborn neurons derived from the mutant NPCs, as well as a significant reduction in neurite length (control 47.7 ± 1.4 µm vs. mutant 23.8 ± 1.8 µm, P = 0.0006; Fig 3I and J). The defect in neuron generation was confirmed by qPCR analysis, which demonstrated a major decrease of the neuronal markers Tuj1 and doublecortin in mutant NPC cultures (Fig EV3A and B). The neural stem cell marker Sox2 showed a slight but not significant increase in mutant cells (Fig EV3C). Levels of phospho‐LRP6 (S1490), marking active WNT receptor signalling, were drastically reduced, whereas Axin2 expression was not significantly altered in the mutant NPC cultures, confirming that the effects of Ccny/l1 during in vitro differentiation are independent of canonical WNT/β‐catenin signalling (Fig EV3D and E). Furthermore, immunoblotting of total protein lysates from mutant NPCs showed overall strongly increased lysine‐48 ubiquitination (Fig 3K), indicating globally enhanced protein degradation, a hallmark of impaired WNT/STOP signalling (Taelman et al, 2010; Acebron et al, 2014). The results provide further support that Ccny/l1 and WNT/STOP signalling promote asymmetric AP division and neurogenesis in the embryonic neocortex.

Figure EV3. Data supporting Fig 3 .

• A–C
  Mutant NPCs display neurogenesis defects. qPCR expression analysis of the neuronal markers βIII‐tubulin (A) and doublecortin (B) and of the stem cell marker Sox2 (C) using RNA extracted from control (Ccny ^+/−; Ccnyl1 ^−/−; shRNA Co) and mutant (Ccny ^+/−; Ccnyl1 ^−/−; shRNA Ccny) NPC cultures after seven days in differentiation medium. Data are expressed as fold change vs. controls and are means ± SEM (experiment performed twice in quadruplicates, representative experiment shown).
• D
  Decreased LRP6 phosphorylation in mutant NPCs. Immunoblot analysis using an antibody against pLRP6 (S1490) of protein lysates extracted from control and mutant NPCs prior to culture in differentiation medium. Numbers above blot represent pLRP6 levels (fold change) normalized to total LRP6 (middle immunoblot). Experiment repeated three times and representative samples shown.
• E
  No changes in Axin2 expression in mutant NPCs. Axin2 qPCR analysis using RNA extracted from control and mutant NPCs prior to culture in differentiation medium. Data are expressed as fold change vs. controls and are means ± SEM (n = 3 biological replicates).

Data information: Unpaired two‐tailed t‐test for all statistical analyses: ns, not significant; ***P < 0.001. P = 0.000050 (A); P = 0.00030 (B); P = 0.12 (C); P = 0.61 (E).

Sox4/11 are direct GSK3 targets

WNT/STOP signalling acts by protecting proteins carrying GSK3 phosphodegrons from proteasomal degradation. To identify downstream targets of WNT/STOP signalling, we screened for effector proteins with a role in neurogenesis and neuronal differentiation that harbour potential GSK3 phosphorylation sites. Two candidates were the SRY‐related high‐mobility‐group box proteins Sox4 and Sox11, SoxC subclass transcription factors. Sox4/11 display pan neuronal activity and are indispensable for neurogenesis and neuronal differentiation (Bergsland et al, 2006). Sox4 is highly expressed by BPs, and Sox11 is expressed both by BPs and newly formed neurons (Chen et al, 2015). Conditional inactivation of Sox4 in the neuroepithelium leads to a strong reduction of BPs while Sox11 deletion leads to decreased post‐mitotic neurons (Chen et al, 2015). Combined phenotypes from loss of Sox4/11 thus resemble the neurogenesis defects observed in DKO embryos. Mouse Sox4 contains a putative GSK3 phosphorylation site consisting of three serines spaced by three amino acids (SxxSSxxSS(316)P) (Fig 4A), a common feature of GSK3 motifs (Beurel et al, 2015). Mouse Sox11 contains two putative GSK3 phosphorylation sites: (S(244)PxxS) and (S(289)PxxSxxxS) (Fig 4B). The Sox4/11 GSK3 sites are highly conserved in other mammals such as chimps and humans. Moreover, S315 and S316 of the Sox4 motif, and S244 and S289 of the Sox11 motifs, are phosphorylated according to PhosphoSitePlus (Hornbeck et al, 2012) (Fig 4A and B). Sox11 S244 and S289 phosphorylation was also reported elsewhere (Balta et al, 2018a). Finally, in silico analysis by NetPhos 3.1 (Blom et al, 1999) predicted GSK3 to be a top candidate for phosphorylating S316 of the Sox4 motif and both S244 and S289 of the Sox11 motifs (Fig 4A and B).

Figure 4. Sox4/11 are direct GSK3 targets.

1. Amino acid sequence alignment of Sox4 GSK3 phospho‐motif from mouse, rat, human and chimp. Mouse serine 316 (S316) (highlighted in green) and S315 (highlighted in yellow) predicted by PhosphoSitePlus to be phosphorylated. For S316 mutant, serine 316 mutated to alanine and for S316all mutant all highlighted serines mutated to alanine. Red highlight indicates proline residue following mutated serine.
2. Amino acid sequence alignment of Sox11 GSK3 phospho‐motifs from mouse, rat, human and chimp. Mouse S244 and S289 (highlighted in green) predicted by PhosphoSitePlus to be phosphorylated. For S244S289 mutant, serines 244 and 289 mutated to alanine and for S244allS289all mutant all highlighted serines mutated to alanine. Red highlights indicate proline residues following mutated serines.
3. Analysis of Flag‐tagged Sox4 wild‐type (WT) protein overexpressed in HEK293T (293T) cells and detected with FLAG antibody. Lysates were treated with or without lambda‐phosphatase (λ‐Phos). Antibody against phosphorylated β‐catenin (pβCAT) was used to validate λ‐Phos treatment. Purple arrowheads indicate Sox4 phosho‐shift from 70 to 65 kDa. ERK blot shows similar loading of samples.
4. Analysis of Flag‐tagged Sox11 WT protein overexpressed in 293T cells and detected with FLAG antibody. Lysates treated with or without λ‐Phos. Upper smear (red arrowhead) represents phosphorylated Sox11. Antibody for pβCAT was used to validate λ‐Phos treatment.
5. Analysis comparing molecular masses of WT Sox4 to S316 and S316all mutants. ERK blot shows similar loading of samples.
6. Analysis comparing molecular masses of WT Sox11 to S244S289 and S244allS289all mutants. Red arrowheads indicate downward shifts of phosphorylated Sox11. ERK blot shows similar loading of samples.
7. Co‐transfection of myc‐tagged GSK3β and Flag‐Sox4 in 293T cells. BIO treatment (0.5 µM) performed for 1 h prior to harvest. Lower band (65 kDa) represents non‐phosphorylated Sox4. High and low exposure (exp) blots shown to better visualize non‐phosphorylated Sox4. Numbers above blot are quantification of Sox4 protein levels normalized to GFP and are expressed as fold change vs. controls. The sum of upper and lower bands was quantified (n = 8 biological replicates).
8. Myc‐GSK3β and S316all mutant co‐transfection in 293T cells plus 0.5 µM BIO treatment 1 h prior to harvest. Protein levels quantified as in (G) (n = 4 biological replicates).
9. Myc‐GSK3β and Flag‐Sox11 co‐transfection in 293T cells. BIO treatment (0.5 µM) performed for 1 h prior to harvest. Upper band represents phosphorylated Sox11. Numbers above blot are quantification of Sox11 protein levels normalized to GFP and are expressed as fold change vs. controls. The sum of lower and upper bands was quantified (n = 14, experiment performed 4× in triplicate and once in duplicate).
10. Myc‐GSK3β and S244allS289all mutant co‐transfection in 293T cells plus 0.5 µM BIO treatment 1 h prior to harvest. Protein levels quantified as in (I) (n = 6, experiment performed 2x in triplicate).
11. In vitro kinase assay on immunopurified Sox4WT and S316all mutant using recombinant GSK3β protein and radiolabelled ATP. Samples were treated with λ‐Phos prior to assay and run on Phos‐tag gel. Coomassie Blue staining shows equal loading of WT and mutant proteins.
12. In vitro kinase assay of purified Sox11WT and S244allS289all mutant using recombinant GSK3β protein and radiolabelled ATP. Experiment performed as in (K).

Data information: (C‐J) Immunoblot analyses. Protein ladder numbers in italics indicate estimated size. Unpaired two‐tailed t‐test for all statistical analyses: ns, not significant; *P < 0.05, ***P < 0.001. P = 0.00010 (G, 2^nd lane); P = 0.049 (G, 3^rd lane); P = 0.76 (H, 2^nd lane); P = 0.83 (H, 3^rd lane); P = 0.00030 (I, 2^nd lane); P = 0.00020 (I, 3^rd lane); P = 0.36 (J, 2^nd lane); P = 0.67 (J, 3^rd lane).

Source data are available online for this figure.

To determine whether Sox4/11 are indeed phosphorylated, we overexpressed N‐terminally Flag‐tagged Sox4/11 in HEK293T (293T) cells and treated protein lysates with λ‐phosphatase. Immunoblot analysis revealed a downshift of Sox4 from 70 to 65 kDa (Fig 4C), and a downshift of a higher molecular weight smear representing phosphorylated Sox11 (Fig 4D). Endogenous Sox4 was also downshifted upon λ‐phosphatase treatment (Fig EV4A). By IF, overexpressed Sox4/11 mainly localized to the nucleus with only minor cytoplasmic staining, consistent with their role as transcription factors, validating our constructs (Fig EV4B and C). Next, we tested the effects of mutating the potential GSK3 phospho‐motifs of Sox4/11. For Sox4, we generated a mutant in which only S316 was mutated to alanine and one in which all five serines were mutated (S316all) (Fig 4A). For Sox11, we generated a mutant in which S244 and S289 were mutated (S244S289) and one where all serines of both motifs were mutated (S244allS289all) (Fig 4B). Immunoblot analysis revealed a downshift in the Sox4 S316 mutant, with multiple bands migrating lower than the expected size of wild‐type Sox4. The S316all mutant revealed an even more drastic downshift, mimicking the effect of phosphatase treatment on wild‐type Sox4 (Fig 4E). For Sox11, a minor downward shift of phosphorylated Sox11 was detected in the S244S289 mutant, while a more obvious shift was detected in S244allS289all mutants, with concomitant increase in the major Sox11 band, corresponding to unphosphorylated Sox11 (Fig 4F).

Figure EV4. Data supporting Fig 4 .

• A
  Endogenous Sox4 is phosphorylated. Immunoblot analysis using a Sox4 antibody of 293T protein lysates treated without (−) or with (+) λ‐Phosphatase (λ‐Phos) (top). ERK (middle) and phospho‐β‐catenin (pβCAT, bottom) immunoblots were used to validate equal loading and λ‐Phos treatment, respectively. Protein ladder numbers in italics indicate estimated size.
• B
  Flag‐Sox4 is nuclear. IF using a FLAG antibody on 293T cells transfected with Flag‐Sox4 plasmid. Asterisks and arrow denote nuclear and cytoplasmic staining, respectively. Scale bar 5 μm.
• C
  Flag‐Sox11 is nuclear. IF using a FLAG antibody on 293T cells transfected with Flag‐Sox11 plasmid. Asterisks and arrow denote nuclear and cytoplasmic staining, respectively. Scale bar 5 μm.
• D, E
  Sox4 and Sox11 are ubiquitinated. 293T cells were subjected to Flag‐Sox4 (D) or Flag‐Sox11 (E) overexpression as indicated, each together with HA‐ubiquitin overexpression, followed by 4‐h MG132 (20 µM) treatment to block proteasomal degradation and immunoprecipitation (IP) for Flag‐tagged proteins. Immunoblot analysis using an HA antibody was performed to detect ubiquitin conjugation to Flag‐Sox4 (D, upper panel) and Flag‐Sox11 (E, upper panel). (D, E lower panels) are anti FLAG immunoblots. Experiment performed twice and representative results shown. Protein ladder numbers in italics indicate estimated size.

Next, we tested the effect of GSK3β overexpression on Flag‐Sox4/11 by co‐transfection in 293T cells. GSK3β overexpression led to a slight but significant decrease in total Sox4 protein levels, with a greater decrease observed in non‐phosphorylated Sox4 (65 kDa band) (Fig 4G). Importantly, this effect was reversed by 1‐h treatment with the GSK3 inhibitor 6‐bromoindirubin‐3'‐oxime (BIO) (Fig 4G). GSK3β overexpression and BIO treatment had no effect on the S316all mutant form of Sox4, confirming this site is a GSKβ phospho‐motif (Fig 4H). GSK3β overexpression also reduced Sox11 protein levels, especially in non‐phosphorylated Sox11, and this effect was reversed by 1 h BIO treatment (Fig 4I). GSK3β overexpression and BIO treatment had no effect on the S244allS289all mutant form of Sox11 (Fig 4J).

To test whether Sox4/11 are directly phosphorylated by GSK3, we immunopurified wild‐type and mutant (S316all and S244allS289all) Sox4 and Sox11, treated the proteins with λ‐phosphatase and then performed in vitro kinase assays by incubating with recombinant GSK3β and gamma‐³²P‐labelled ATP. Wild‐type Sox4/11 were both highly phosphorylated by GSK3β, while only minimal phosphorylation of the corresponding mutants could be detected (Fig 4K and L). Because WNT/STOP signalling inhibits GSK3β to protect target proteins from proteasomal degradation (Taelman et al, 2010; Acebron et al, 2014), we tested whether Sox4/11 are ubiquitinated by co‐transfecting HA‐tagged ubiquitin with Flag‐Sox4/11 and then briefly treating with MG132 to block proteasomal degradation of ubiquitin‐conjugated polypeptides. Pull‐down of Sox proteins with FLAG followed by immunoblot against HA‐ubiquitin revealed a characteristic polyubiquitin‐smear in both Sox4 and Sox11‐transfected cells, indicating that both Sox proteins are ubiquitin‐conjugated (Fig EV4D and E). We conclude that Sox4/11 are (i) directly phosphorylated by GSK3β and (ii) regulated by proteasomal degradation.

Sox4/11 protein levels are decreased in DKO NPCs, predominantly during mitosis

We next tested whether Sox4/11 are regulated by GSK3 in NPCs, focusing first on Sox4. We treated cultured NPCs with BIO for 24 h and analysed endogenous Sox4 protein levels by immunoblot analysis. BIO treatment led to an overall increase in Sox4, with an even greater increase in the lower, non‐phosphorylated band (Fig EV5A). Importantly, mutant Ccny/l1 NPCs showed significantly reduced Sox4 protein levels (Fig 5A), even though Sox4 mRNA levels where slightly elevated (Fig EV5B). We next analysed Sox4 protein levels in the neocortex of E13.5 embryos in vivo. By IF, Sox4 co‐localized with Ccny and pLRP6 (S1490) in the SVZ (Fig EV5C), and was concentrated in mitotic BPs (Fig EV5D). In APs, Sox4 was enriched at mitosis (Fig EV5D). In DKO embryos, we could not detect an obvious decrease in Sox4 staining intensity in Tbr2⁺ cells (Fig EV5E). Likewise, immunoblot analysis of DKO forebrain lysates showed no reduction in total Sox4 protein levels (Fig EV5F), and qPCR analysis of RNA extracted from dorsal forebrains revealed no significant changes in Sox4 expression (P = 0.87; Fig EV5G). In contrast, focusing on mitotic (pHH3⁺) BPs by IF, Sox4 protein levels were greatly reduced in DKO neocortex (−42%, P = 0.0002; Fig 5B and C). To further investigate WNT/STOP regulation of Sox4 during mitosis, we treated NPC cultures with nocodazole to arrest cells in G2/M. Interestingly, nocodazole‐treated mutant NPCs displayed a greater reduction of Sox4 compared to non‐treated cells (−45%, P = 0.02 vs. −25%, P = 0.001) (Fig EV5H). G2/M arrest by nocodazole treatment was validated by FACS analysis of NPCs (Fig EV5I).

Figure EV5. Data supporting Fig 5 .

• A
  BIO treatment stabilizes non‐phosphorylated Sox4 in NPCs. Sox 4 immunoblot analysis of protein lysates from E13.5 wild‐type NPCs treated without (−) or with (+) 1 µM BIO for 24 h. Low (top) and high (middle) exposures (exp) are shown. Numbers above blot represent the difference in the ratio of non‐phospho Sox4 band (65 kDa) to phospho‐Sox4 band (70 kDa), with this ratio for control cells being set to 1.0 and the ratio for BIO‐treated cells being expressed relative to this (n = 3 biological replicates). GAPDH immunoblot to document equal loading.
• B
  Sox4 mRNA levels are slightly increased in mutant NPCs. Sox4 qPCR analysis of RNA extracted from control (Ccny ^+/−; Ccnyl1 ^−/−; shRNA Co) and mutant (Ccny ^+/−; Ccnyl1 ^−/−; shRNA Ccny) E13.5 NPCs. Data are expressed as fold change vs. controls and are means ± SEM (n = 3 biological replicates).
• C
  Sox4 co‐localizes with Ccny and pLRP6 in SVZ cells. Co‐IF for Sox4 and either Ccny or pLRP6 (S1490) in E12.5 wild‐type neocortex. Areas in dashed boxes are shown at higher magnification in the insets. Arrows indicate Ccny protein (left inset). Sox4 co‐IF with pLRP6 detected in mitotic BP (right inset). Scale bar 10 μm.
• D
  Sox4 in APs is enriched at mitosis and is highly expressed in interphase and mitotic BPs. Triple‐IF for Sox4, pHH3 and Tbr2 in E13.5 wild‐type neocortex. Sox4 is highly expressed in interphase and mitotic BPs (upper insets), and in APs is enriched at mitosis (lower insets, indicated by asterisks). Scale bar 20 μm.
• E–G
  Sox4 protein and relative mRNA levels are not altered in DKO forebrains. (E) Quantification of Sox4 relative staining intensity as determined by IF in non‐mitotic BPs in the neocortex of control and DKO E13.5 embryos. Data are means ± SEM (n = 4 embryos, 3 litters). (F) Sox4 immunoblot analysis of protein lysates from E13.5 control and DKO forebrains (top). Numbers above blot represent fold change vs. controls of Sox4 immunoreactivity upon normalization to Tbr2 immunoreactivity (middle) (n = 5 embryos, 4 litters). ERK1/2 immunoblot (bottom) shows equal loading of samples. (G) Sox4 qPCR expression analysis of RNA extracted from E13.5 control and DKO dorsal forebrains. Sox4 mRNA levels were normalized to Tbr2, also determined by qPCR expression analysis. Data are expressed as fold change vs. controls and are means ± SEM (n = 3 embryos, 2 litters).
• H, I
  Sox4 reduction in mutant NPCs is greater upon G2/M arrest. (H) Sox4 immunoblot analysis of protein lysates from E13.5 control and mutant NPCs treated without (−) or with (+) 100 ng/ml nocodazole overnight (Noc). Numbers above blot represent fold change vs. controls (−Noc) of Sox4 protein levels normalized to GAPDH. (I) FACS analysis of propium iodide (PI)‐stained E13.5 NPCs treated without (−Noc) or with (+Noc) nocodazole to determine the percentage of NPCs in G2/M.
• J, K
  Validation of Sox4 phospho‐antibody. (J) Immunoblot analysis using the Sox4 phospho‐antibody (pSox4) of protein lysates from 293T cells transfected with Flag‐Sox4 (top). Protein lysates were treated without (−) or with (+) λ‐phosphatase for 1 h prior immunoblot analysis. Anti‐FLAG immunoblot to demonstrate the presence of dephosphorylated Sox4 (middle). ERK1/2 immunoblot (bottom) shows equal loading of samples. (K) Immunoblot analysis using the pSox4 antibody of protein lysates from 293T cells transfected with WT Flag‐Sox4 or S316‐mutant plasmids as indicated (top). Anti‐Flag immunoblot to demonstrate the presence of S316‐mutant Sox4 (middle). GFP immunoblot (bottom) shows equal loading of samples.
• L
  Increase in Sox11 mRNA levels in mutant NPCs is not statistically significant. Sox11 qPCR analysis of RNA extracted from control and mutant E13.5 NPCs. Data are expressed as fold change vs. controls and are means ± SEM (n = 3 biological replicates).
• M
  Relative Sox11 mRNA levels are not altered in DKO dorsal forebrains. Sox11 qPCR analysis of RNA extracted from E13.5 control and DKO dorsal forebrains. Sox11 mRNA expression was normalized to βIII‐tubulin, also determined by qPCR expression analysis. Data are expressed as fold change vs. controls and are means ± SEM (n = 3 embryos, 2 litters).
• N
  Sox11 co‐localizes with Ccny and pLRP6 in SVZ cells. Co‐IF for Sox11 and either Ccny or pLRP6 (S1490) in E12.5 wild‐type neocortex. Areas in dashed boxes are shown at higher magnification in the insets. Arrow indicates Ccny protein (left inset). Sox4 co‐immunostaining with pLRP6 detected in mitotic BP (right inset). Scale bar 10 μm.

Data information: Protein ladder numbers in italics indicate estimated size. Unpaired two‐tailed t‐test for all statistical analyses: ns, not significant; *P < 0.05, **P < 0.01. P = 0.044 (A); P = 0.042 (B); P = 0.32 (E); P = 0.70 (F); P = 0.87 (G); P = 0.0013 (H, 2^nd lane); P = 0.049 (H, 3^rd lane); P = 0.018 (H, 4^th lane); P = 0.062 (L); and P = 0.46 (M).

Source data are available online for this figure.

Figure 5. Sox4/11 protein levels are decreased in DKO NPCs, predominantly during mitosis.

1. Immunoblot analysis with Sox4 antibody of protein lysates extracted from control (Co) (Ccny ^+/−; Ccnyl1 ^−/−; shRNA Co) and mutant (Ccny ^+/−; Ccnyl1 ^−/−; shRNA Ccny) NPCs. Numbers are fold change vs. controls of Sox4 protein levels normalized to GAPDH (n = 3). Immunoblot image cropped from Fig EV5H. Quantification performed on immunoblot shown in Fig EV5H and two additional immunoblots (not shown).
2. IF for Sox4 and pHH3 on neocortex sections of E13.5 control and DKO embryos. The level of Sox4 immunoreactivity is reduced in pHH3⁺ cells (white arrowheads).
3. Quantification of Sox4 staining in SVZ mitotic cells from E13.5 control and DKO embryos. Mitotic Sox4 staining was defined as high if the pixel intensity was similar to, or higher than, non‐mitotic cells, or low if the levels were below those observed in non‐mitotic cells. Data are means ± SEM for ratios of cells with high vs. low Sox4, and are displayed as fold change vs. controls (n = 4 embryos, 3 litters).
4. Strategy to generate Sox4 S316 phospho‐antibody (pSox4) by phosphopeptide injection into rabbits.
5. Flag‐Sox4 transfection in 293T cells followed by 24‐h nocodazole (100 ng/ml) and BIO (1 µM) treatments. Samples analysed by immunoblot using pSox4 antibody; numbers above blot are quantification of Sox4 phospho‐band normalized to total Sox4 (FLAG) for nocodazole‐treated and non‐treated cells and are expressed as fold change vs. controls (n = 6, experiment performed 2× in triplicate).
6. Immunoblot analysis with pSox4 antibody of cell lysates extracted from E13.5 control and DKO forebrains. Red arrowheads indicate Sox4 phosphorylation band; red asterisks are non‐specific bands. Numbers above blot are quantification of Sox4 phosphorylation levels normalized to total Sox4 and are expressed as fold change vs. controls (n = 4 embryos, 3 litters).
7. Immunoblot analysis with a Sox11 antibody of protein lysates extracted from control and mutant E13.5 NPCs. Numbers are fold change vs. controls of Sox11 protein levels normalized to GAPDH (n = 3 biological replicates).
8. IF for Sox11 and pHH3 in E13.5 control and DKO neocortices. Sox11 staining intensity is decreased in mitotic cells (white arrowheads, insets).
9. Quantification of Sox11 protein levels in SVZ mitotic cells from E13.5 control and DKO embryos. Quantification performed as in (C). Data are means ± SEM for ratios of high vs. low Sox11 and are displayed as fold change vs. controls (n = 5 embryos, 3 litters).
10. Immunoblot analysis with a Sox11 antibody of protein lysates extracted from E13.5 control and DKO forebrains. Numbers are fold change vs. controls of Sox11 protein levels normalized to Tuj1 (n = 5 embryos, 4 litters). ERK1/2 immunoblot shown to document equal loading.
11. Co‐IF for Sox11 and Tbr2 in sections of neocortex of E13.5 control and DKO embryos. Sox11 immunoreactivity is reduced in some Tbr2⁺ cells (asterisks, insets).
12. Quantification of percentage of Tbr2⁺ cells that show with high Sox11 staining intensity, in control and DKO E13.5 neocortices. High or low staining intensity was judged based on visual inspection, taking background staining levels into account. Data are means ± SEM and are shown as fold change vs. controls (n = 5 embryos, 3 litters).

Data information: All scale bars 20 µm. Protein ladder numbers in italics indicate estimated size. Unpaired two‐tailed t‐test for all statistical analyses: ns, not significant; *P < 0.05, **P < 0.01 ***P < 0.001. P = 0.0012 (A); P = 0.00020 (C); P = 0.0031 (E, 2^nd lane); P = 0.0010 (E, 4^th lane); P = 0.047 (F, 2^nd lane); P = 0.047 (G); P = 0.045 (I); P = 0.031 (J); P = 0.0010 (L).

Source data are available online for this figure.

To further investigate Sox4 regulation in NPCs, we raised a phospho‐specific antibody targeting the S316 site of Sox4 (pSox4) (Fig 5D, schematic). Immunoblot analysis of overexpressed Flag‐Sox4 in 293T cell lysates treated with λ‐phosphatase revealed a specific band at 70 kDa in the non‐treated samples only, demonstrating specificity of the antibody towards phosphorylated Sox4 (Fig EV5J). The pSox4 antibody did not detect the S316 mutant upon immunoblotting (Fig EV5K). Treatment of Flag‐Sox4‐transfected 293T cells with BIO led to decreased Sox4 phosphorylation, and this effect was greater in nocodazole‐treated cells (Fig 5E). Moreover, immunoblot analysis of forebrains revealed a specific band at 70 kDa, representing phosphorylated Sox4, that was increased in DKO embryos (Fig 5F), consistent with elevated phosphorylation by GSK3.

For Sox11, Ccny/l1 mutant NPCs revealed strongly reduced protein levels and elevated (not significant) mRNA levels, similar to Sox4 (Figs 5G and EV5L). Moreover, IF analysis on E13.5 neocortex sections revealed significantly decreased Sox11 protein levels in mitotic cells within the SVZ (−23%, P = 0.04) (Fig 5H and I). Sox11 protein levels were also decreased in newborn neurons and non‐mitotic BPs, as evidenced by immunoblot analysis of forebrain protein lysates and co‐IF with Tbr2 (−10%, P = 0.01), respectively (Fig 5J–L). Expression of Sox11 was not significantly altered in DKO dorsal forebrains (P = 0.46), confirming the regulation of Sox11 by Ccny/l1 is post‐transcriptional (Fig EV5M). Finally, Sox11 co‐localized with Ccny and pLRP6 in the SVZ (Fig EV5N). In summary, Ccny/l1 deficiency leads to decreased Sox4/11 protein levels in NPCs and predominantly in mitotic cells, consistent with Sox4/11 being novel WNT/STOP targets in the developing neocortex.

Sox4/11 overexpression and GSK3 inhibition both rescue differentiation defects of DKO NPCs

To test whether Sox4/11 misregulation accounts for the decreased cortical neurogenesis observed in DKO embryos, we attempted to rescue the neurogenesis defect in mutant NPCs by overexpressing both proteins using lentiviral transduction of Flag‐Sox4 and Flag‐Sox11 in the pLenti‐CAG‐IRES‐EGFP vector (Fig 6A, schematic). To avoid possible toxic effects resulting from excessive viral load, we isolated NPCs from DKO and control (Ccny ^+/− Ccnyl1 ^+/−) E13.5 neocortex instead of performing shRNA knockdown of Ccny. DKO NPCs exhibited similar neurogenesis defects as shCcny‐treated Ccnyl1 mutants, showing reduced Tuj1 staining (Fig 6B and C). For rescue experiments, we transduced DKO and control NPCs with Sox4 and Sox11‐overexpressing lentiviruses, or empty pLenti‐CAG‐IRES‐EGFP vector lentivirus as a control, incubated the cultures in differentiation medium, and then monitored neuronal output by IF for Tuj1 and Map2 (microtubule‐associated protein 2, an additional neuron marker). IF for GFP or FLAG was used to identify NPCs infected with control or Sox4/11 lentiviruses, respectively. Overexpression of Sox4/11 was monitored by qPCR and IF, which revealed, respectively, increased Sox4/11 mRNA levels and high transduction efficiency in NPCs (Appendix Fig S2A–F). Strikingly, Sox4/11 overexpression in DKO NPCs led to an almost complete rescue in the number of Tuj1⁺ and Map2⁺ cells generated when compared to DKO NPCs transduced with empty vector (Fig 6B and C; Appendix Fig S3A and B). Importantly, control cells transduced with Sox4/11 lentiviruses showed no significant increase in the number of newly formed Tuj1⁺/Map2⁺ neurons (Fig 6B and C; Appendix Fig S3A and B). Interestingly, many of the rescued Tuj1⁺/Map2⁺ cells lacked mature neurites (Fig 6B, asterisks; Appendix Fig S3A, asterisks; quantified in Appendix Fig S3C), suggesting that Sox4/11 overexpression can rescue initial stages of neurogenesis but not full neuronal differentiation in DKO NPCs.

Figure 6. Sox4/11 overexpression and GSK3 inhibition both rescue differentiation defects of DKO NPCs.

1. Schematic illustrating strategy to rescue DKO NPC differentiation defect by simultaneous lentiviral overexpression of Flag‐Sox4 and Flag‐Sox11 with two lentiviruses. Created with Biorender.com.
2. IF for Tuj1 and Map2 (not shown) to identify newly formed neurons in NPCs after seven‐day culture in differentiation medium. Empty CAG‐IRES‐EGFP vector was used as control in cells not transduced with Sox4/11. Transduced cells were identified by IF for GFP (Co and DKO) or FLAG (Co + Sox4/11 and DKO + Sox4/11). Arrowheads indicate mature neurons, and asterisks denote immature neurons lacking neurite extension. Scale bar 20 μm.
3. Quantification of transduced Tuj1⁺ cells from (B). Note the decrease in neuron number in DKO cultures compared to control, and the rescue in DKO cultures upon Sox4/11 expression (experiment performed twice in triplicates, representative experiment shown). Data are means ± SEM.
4. Schematic illustrating the rescue DKO NPC differentiation defect with the GSK3 inhibitor CHIR99021 (chemical name: 6‐[2‐[[4‐(2,4‐dichlorophenyl)‐5‐(5‐methyl‐1H‐imidazol‐2‐yl)pyrimidin‐2‐yl]amino]ethylamino]pyridine‐3‐carbonitrile). Created with Biorender.com.
5. IF for Tuj1 to identify newly formed neurons in NPCs after seven‐day culture in differentiation medium. Arrowheads indicate mature neurons, and asterisks denote immature neurons lacking neurite extension. Scale bar 20 μm.
6. Quantification of Tuj1⁺ cells in control, control + CHIR (1 µM, 3 µM), DKO, and DKO + CHIR NPC cultures. Data are means ± SEM (experiment performed twice in triplicates, representative experiment is shown). Numbers below asterisks show fold change.

Data information: One‐way ANOVA test for all statistical analyses: ns, not significant; *P < 0.05, **P < 0.01 ***P < 0.001. P = 0.97 (C, yellow bar); P = 0.00010 (C, red bar); P = 0.0023 (C, purple bar); P = 0.031 (F, yellow bar); P = 5 × 10⁻⁶ (F, grey bar); P = 2 × 10⁻⁵ (F, purple bar); and P = 1 × 10⁻⁶ (F, green bar).

Source data are available online for this figure.

Finally, to corroborate that the DKO differentiation phenotype in cultured NPCs is indeed due to diminished WNT signalling, we carried out a rescue experiment. We treated control and DKO NPCs with the GSK3 inhibitor CHIR99021 (CHIR) and monitored neuronal output by IF for Tuj1 (Fig 6D, schematic). Strikingly, 1 µm CHIR treatment in DKO NPCs led to a significant increase in the number of, mostly, immature neurons, while 3 µm CHIR treatment restored the level of mature neurons to that of controls, thereby leading to a complete rescue (Fig 6E and F). CHIR treatment also promoted neurogenesis in control NPCs, which is consistent with previous reports (Rosenbloom et al, 2020), and however, the increase in neurogenesis observed was significantly lower when compared to CHIR‐treated DKO NPCs (Fig 6F). Hence, we conclude that Ccny/l1 promote differentiation of cultured NPCs through the WNT signalling pathway.

Discussion

We set out to determine the role of WNT/STOP signalling in neocortex development by analysing mice mutant for Ccny and Ccnyl1, key regulators of the pathway. We find that WNT/STOP orchestrates the process of neurogenesis by (i) regulating symmetric vs. asymmetric AP division; (ii) controlling the length of the cell cycle, notably of mitosis, of BPs; and (iii) promoting neuron generation through Sox4 and Sox11 stabilization in BPs. The results suggest that WNT/STOP rather than WNT/β‐catenin signalling is the primary driver of cortical neurogenesis. Our study reconciles seemingly contradictory reports by revealing a division of labour in neocortex development, whereby WNT/STOP promotes a differentiative process, i.e. the generation of NPCs committed to neurogenesis, whereas WNT/β‐catenin predominantly promotes NPC self‐renewal.

A division of labour between WNT/STOP and WNT/β‐catenin signalling during neocortical neurogenesis

The precise biological role and molecular function of WNT signalling in the mouse neocortex has remained controversial. Most in vivo studies suggest that the primary role of canonical WNT signalling is to promote self‐renewal at the expense of differentiation. This is particularly evident in β‐catenin mutant mice, which display increased AP cell cycle exit and premature neurogenesis (Mutch et al, 2010; Draganova et al, 2015). Although some studies have shown that β‐catenin can also promote neuronal differentiation of NPCs (Hirabayashi et al, 2004; Israsena et al, 2004; Kuwahara et al, 2010), these studies were mostly performed in vitro, and are not supported by in vivo evidence. More recently, electroporation of Wnt3a into the neocortex of developing mouse embryos increased both, proliferation of APs and differentiation of BPs, seemingly resolving the controversy and leading the authors to conclude that canonical WNT can indeed promote both processes in a cell type‐specific manner (Munji et al, 2011). However, this study relied on manipulating WNT signalling at the ligand level, which not only affects WNT/β‐catenin but also WNT/STOP signalling, which was not examined. WNT/STOP acts through the same WNT/LRP6/GSK3 axis but bifurcates upstream‐ and is independent of β‐catenin. Its primary output is not transcription but post‐transcriptional protein stabilization (Acebron & Niehrs, 2016).

We now propose a new model (Fig 6D), whereby WNT/STOP promotes neuron generation from BPs through stabilization of Sox4/11 while canonical WNT/β‐catenin signalling promotes AP self‐renewal. Our conclusions based on the analysis of E13.5 DKO embryos are further supported by the results of Ccny/l1 knockdown at later stages of neocortex development in vivo and of NPC culture experiments in vitro. Together, these data show reduced generation of BPs, and consequently of neurons, in the absence of Ccny/l1. We therefore conclude that DKO embryos exhibit reduced neurogenesis through decreased generation of BPs from APs and of post‐mitotic neurons from BPs. However, while our data demonstrate that the knockout of Ccny/l1 does not affect WNT/β‐catenin signalling, they do not rule out the possibility that WNT/β‐catenin signalling can also contribute to these processes.

LRP6 is a key regulator in WNT/STOP and its phosphorylation at G2/M by the Ccny/CDK complex leads to GSK3 inhibition and activation of the pathway notably during mitosis, when transcription is low (Davidson et al, 2009). Interestingly, analysis of Lrp6 mutant mice reveals a thinner neocortex, reduced neuronal differentiation, and only minor changes in proliferation (Zhou et al, 2006). Similarly, deficiency in actin‐binding protein filamin A induces neurogenesis defects within the mouse cerebral cortex and impairs LRP6/GSK3 signalling and asymmetric cell division (Lian et al, 2016; Lian et al, 2019). These phenotypes are remarkably similar to those observed here in Ccny/l1 DKO embryos, suggesting that Lrp6 mutants primarily present a WNT/STOP signalling defect. The reason why Lrp6 mutants do not also manifest the proliferation defects of β‐catenin mutants is unclear, but may be due to compensation from LRP5. Functional redundancy between LRP5 and LRP6 in WNT/β‐catenin signalling is well documented (Kelly et al, 2004; Goel et al, 2012; Zhong et al, 2012; Liu et al, 2019).

WNT/STOP signalling promotes neurogenesis via Sox4 and Sox11 stabilization

An important finding of this study is the identification of the neurogenic transcription factors Sox4 and Sox11 as bona fide WNT/STOP targets. Several links between SoxC transcription factors and WNT signalling are known: Sox4 and Sox11 interact with the β‐catenin destruction complex and, in lung cancer cells, Sox4 is a direct target of β‐catenin (Bhattaram et al, 2014; Melnik et al, 2018). Our data support and further develop these links by showing that Sox4/11 are directly phosphorylated by GSK3 and thereby targeted for proteasomal degradation. We also show that the majority of Sox4 is phosphorylated at a single GSK3 phospho‐motif, while Sox11 contains two motifs and is only partially phosphorylated. GSK3 phosphorylation is known to create phosphodegrons for recognition by E3 ubiquitin ligases leading to proteasomal degradation (Taelman et al, 2010). Consistent with this possibility, phosphorylation by GSK3β induced ubiquitination and decreased overall Sox4 and Sox11 levels. Previous reports have shown that, in the neocortex, Sox11 cellular localization and protein stabilization are regulated by phosphorylation and ubiquitination, respectively, supporting that post‐transcriptional regulation of Sox11 is essential for its function in vivo (Balta et al, 2018a; Balta et al, 2018b; Chiang et al, 2021).

WNT/STOP mediated inhibition of GSK3 predominantly occurs in G2/M (Acebron et al, 2014). Consistently, we find that the stabilization of Sox4/11 by WNT/STOP is cell cycle‐dependent, with peak levels in mitotic cells. WNT/STOP signalling in G2/M of mother cells is thought to endow daughter cells with a growth advantage, by stabilizing bulk protein. Deficiency in WNT/STOP signalling reduces G1 growth and delays cell cycle progression (Acebron et al, 2014; Huang et al, 2015). Analogously, WNT/STOP signalling in G2/M of NPCs may promote differentiation of daughter cells via elevated levels of Sox4/11 during G1/S‐phase. An attractive hypothesis is that Sox4/11 are bookmarking transcription factors, that remain bound to chromosomes in mitosis to enable target gene reactivation in a timely fashion upon mitotic exit (Palozola et al, 2019).

The importance of decreased Sox4/11 protein levels for the neurogenesis defect of DKO embryos is highlighted by the fact that their overexpression can rescue the differentiation phenotype. Interestingly, the rescue is incomplete: Sox4/11 increase the number of Tuj1⁺ cells in differentiating DKO NPCs, but only partially restore their morphology, leading to immature neurons. The incomplete rescue may be due to unphysiologically high levels of Sox4/11 from lentiviral transduction since Sox11 overexpression inhibits dendritic morphogenesis (Hoshiba et al, 2016). Alternatively, WNT/STOP signalling may have additional targets required for neuronal maturation and morphogenesis. Indeed, proteomic analysis indicates hundreds of potential WNT/STOP target proteins in HeLa cells (Acebron et al, 2014).

WNT/STOP signalling promotes asymmetric AP division

Another important insight of this work is that Ccny/l1 are required to promote asymmetric division of APs, which is in keeping with their physiological role as the starting point of cortical neurogenesis, since more asymmetric AP divisions lead to more BPs and hence post‐mitotic neurons. In C. elegans embryos, WNT signalling is long known to regulate spindle asymmetry (Sugioka et al, 2011), and in embryonic stem cells, a Wnt3a point source can polarise the mitotic spindle to give rise to asymmetric cell division (Habib et al, 2013). Taken together, these studies support that WNT/STOP signalling promotes asymmetric AP division.

Although we have not identified the molecular targets responsible for the decreased asymmetric AP division in DKO embryos, we narrowed down the mechanism to an increase in apical‐basal aMTs, which can directly stabilize the positioning of the mitotic spindle and thus promote symmetric AP division (Mora‐Bermúdez et al, 2014). This finding is consistent with recent work in which GSK3 overexpression and inhibition of WNT secretion in HCT116 cells increase MT polymerization rates during mitosis (Lin et al, 2020). Interestingly, WNT/STOP inhibition has been also shown to induce mitotic spindle defects and improper chromosome segregation (Stolz et al, 2015; Lin et al, 2020), pointing towards a more general role of WNT/STOP signalling in promoting mitotic spindle assembly. One possibility is that WNT/STOP signalling modulates MT plus ends, which are key determinants for spindle orientation (Lu & Johnston, 2013). Misregulation of MT plus ends may also explain the lengthened mitosis observed in DKO BPs. A candidate WNT/STOP target protein is Kif2a (Taelman et al, 2010), a MT plus end regulator that is required for cortical neurogenesis (Sun et al, 2017; Ding et al, 2019). On the other hand, we cannot exclude cross‐talk between WNT/STOP and the WNT/PCP‐pathway, which is also implicated in asymmetric division in cortical progenitors (Delaunay et al, 2014).

Mitosis is a critical time window for determining NPC fate

Our study suggests that β‐catenin‐independent WNT signalling impacts neurogenesis preferentially during mitosis: (i) LRP6 is maximally phosphorylated and activated in mitotic NPCs, (ii) Sox4/11 protein levels peak in mitotic NPCs, (iii) apical‐basal aMTs and symmetric cell division are both increased in DKO APs, and (iv) mitosis is retarded in BPs. These observations are consistent with the facts that WNT/STOP signalling stabilizes proteins specifically in mitosis (Acebron et al, 2014) and regulates cell cycle progression of mitotic cells (Huang et al, 2015; Stolz et al, 2015). The process of neurogenesis is intricately connected to the cell cycle length of NPCs (Götz & Huttner, 2005; Dehay & Kennedy, 2007; Borrell & Calegari, 2014). Thus, lengthening the cell cycle of APs induces premature neurogenesis (Calegari & Huttner, 2003). Conversely, shortening the cell cycle of APs by reducing G1 delays neurogenesis and promotes generation and expansion of BPs (Lange, et al, 2009). It is unclear at which point in the cell cycle NPCs commit to neuronal differentiation. However, given that Ccny/l1 deficiency reduces neuronal output and that the regulation of neurogenesis by WNT/STOP occurs primarily during cell division, we conclude that mitosis represents a key phase of the cell cycle when NPCs commit to a neurogenic fate. This conclusion reverberates the observation that many human microcephaly associated genes encode mitotic regulators (Hu et al, 2014).

In summary, our findings resolve the controversy regarding the role of WNT signalling in neocortex development by demonstrating that WNT/STOP is the primary driver of a differentiative process in vivo, i.e. the generation of neurons from NPCs, whereas WNT/β‐catenin promotes NPC self‐renewal. They also emphasize the importance of mitosis as a critical determinant of NPC fate, when asymmetric AP division and Sox4/11 protein stabilization in BPs are orchestrated by post‐transcriptional WNT signalling.

Materials and Methods

Animals

Mice were bred in the Central Animal Laboratory in the DKFZ, Heidelberg, under standardized hygienic conditions. All animal work was conducted according to national and international guidelines and was approved by the state review board of Baden‐Württemberg (protocol no. G‐123/18). Sperm from mice carrying a flanked by loxP (floxed) allele of cyclin Y (Ccny^{tm1(flox)Smoc}) was obtained from the lab of Arial Zeng (Shanghai, China; An et al, 2015) and used for in vitro fertilization of wild‐type C57BL/6N oocytes. Heterozygous Ccny‐flox mice were bred with transgenic animals expressing Cre recombinase under the control of the CMV promoter to achieve organism‐wide gene knockout (CcnyKO). Generation of cyclin Y‐like 1‐deficient (Ccnyl1tm1a^(EUCOMM)Wtsi/H; Ccnyl1KO) mice has been described previously (Koch et al, 2015). Ccny and Ccnyl1 double knockout embryos (DKO) were generated by incrossing Ccny ^−/− Ccnyl1 ^+/− males with Ccny ^+/− Ccnyl1 ^−/− females. Ccny ^+/−; Ccnyl1 ^+/− embryos were used as controls. Embryos were analysed at various time points (E11.5, E12.5 and E13.5), and gender was not taken into consideration. For proliferation assays, BrdU (Sigma‐Aldrich, B5002) and IdU (Sigma‐Aldrich, I0050000) were dissolved in 0.9% NaCl and administered to pregnant dams via intraperitoneal (IP) injection at a dose of 50 mg/kg. Injection strategies are indicated in the figure legends. Adult mice were sacrificed by cervical dislocation.

For in utero electroporation experiments, C57BL/6J mice were bred at the Biomedical Services Facility of the MPI‐CBG under standardized hygienic conditions and all procedures were conducted in agreement with the German Animal Welfare Legislation after approval by the Landesdirektion Sachsen (licenses: mouse TVV 5/2015 and TVV 20/2020).

Inutero electroporation

Wild‐type C57BL/6J pregnant mice carrying E13.5 embryos were anesthetized using initially 4% isoflurane (Baxter, HDG9623), followed by 2–3% isoflurane during the in utero electroporation (IUE) procedure. The animals were injected subcutaneously with the analgesic (0.1 ml of metamizol, 200 mg/kg). The peritoneal cavity was surgically opened and the uterus exposed. Using borosilicate microcapillary (Sutter instruments, BF120‐69‐10), the embryos were injected intraventricularly with a solution containing 0.1% Fast green (Sigma‐Aldrich, F7252) in sterile PBS, 2 µg/µl of pSuper plasmid (either 2 µg/µl of pSuper‐shcon or 1 µg/µl of pSuper‐shCcny and 1 µg/µl of pSuper‐shCcnyl1), 0.4 µg/µl of pCAGGS GFP. The electroporations (six 50‐msec pulses of 28V at 1 s intervals) were performed using a 3‐mm diameter electrode (BTX genetronics Inc., 45‐0052INT). After surgery, mice received Metamizol in drinking water (1.33 mg/ml).

Pregnant mice were sacrificed by cervical dislocation at the indicated time points (E15.5‐E17.5), and embryonic brains were dissected, fixed in 4% PFA, overnight at 4°C and processed for cryo‐sectioning.

Cell culture

HEK293T (293T) cells were cultured in DMEM (Gibco, 11960‐044) supplemented with 10% v/v foetal bovine serum (Capricorn, FBS‐12A9), 2 mM glutamine (Sigma‐Aldrich, G7513) and 1% v/v penicillin/streptomycin (Sigma‐Aldrich, P0781). Cells were grown at 37°C and 10% CO₂ in a humidified chamber. Transfections were performed using XtremeGENE 9 DNA transfection reagent (Roche, 06366244001) according to the manufacturer’s instructions. Plasmids were transfected at either 200 ng/ml (Flag‐Sox4/Sox11, GFP, HA‐Ubiquitin) or 400 ng/ml (Myc‐GSK3β) in 24‐well plates coated with poly‐D‐lysine (Sigma‐Aldrich, P6407). Where indicated, cells were treated with bromoindirubin‐3’‐oxime (BIO) (Cayman Chemical, 13123) or nocodazole (BioTrend, BN0389). Doses and duration of treatments are indicated in the figure legends. DMSO was used as a control for BIO and nocodazole treatments.

NPC isolation and culturing

NPCs were obtained by incrossing Ccny ^+/− Ccnyl1 ^+/− animals and collecting embryos at E13.5. The forebrain cerebral cortex was dissected from individual embryos, dissociated into a single‐cell suspension by repetitive pipetting and then filtered through a 70 µm cell strainer (Corning, 431751). The resulting neurospheres were cultured in NPC media (DMEM/F12 (Invitrogen, p5780), B27 (LIFE, 1074547), glucose (Sigma‐Aldrich, s5761) hepes, progesterone (Sigma‐Aldrich, 12587010) putrescine (Sigma‐Aldrich, p7556), heparin (Sigma‐Aldrich, E4127) penicillin/streptomycin, insulin‐transferrin‐sodium selenite supplements (Roche, H3149), sodium bicarbonate (Sigma‐Aldrich, p5780) and 20 ng/ml EGF (Thermo Fisher, PHG0313) for 5–7 days before being passaged.

For passaging, cells were treated with accutase (Capricorn, ACC‐1B) and then pipetted repetitively to obtain a single‐cell suspension. Cells were passaged at least two times before performing experiments. Where indicated, NPCs were treated with BIO, CHIR99021 (Millipore, 361559), or nocodazole. Doses and duration of treatments are indicated in the figure legends.

For NPC differentiation assays, 2 × 10⁵/cells were plated on poly‐D‐lysine‐coated 24‐well plates for 48 h, and then, media was switched to NPC differentiation media (DMEM/F12, B27, glucose, hepes, progesterone, putrescine, heparin, penicillin/streptomycin, insulin–transferrin–sodium selenite supplements, and sodium bicarbonate). Cells were grown for 7–9 days to a confluency ˜ 80% before being harvested for RNA, protein or IF analysis. Media was replaced every 48 h. Knockdown of Ccny was performed by transducing shCcny into NPCs 24 h prior to differentiation assays. For Sox4/11 rescue experiments, Flag‐Sox4 and Flag‐Sox11‐overexpressing lentiviruses were transduced simultaneously into NPCs 24 h prior to differentiation assays. 4 μg/ml polybrene (Sigma‐Aldrich, TR‐1003) was added to media for all transductions. For rescue experiments by GSK3 inhibition, CHIR99021 was added to the differentiation media at the indicated concentrations for the entire duration of the differentiation protocol. DMSO was used as a control.

For FACS analysis, NPCs were fixed in 70% ethanol for 10 min, washed three times with PBS and then stained with 40 µg/ml propidium iodide (Thermo Fisher, BMS500PI) in FACS staining buffer (0.1% Triton X‐100, 0.1% sodium citrate, in PBS) at 37°C for 30 min. Cells were then analysed according to Davidson et al, 2009 on a BD FACS Canto, and data were processed using FlowJo software.

Lentivirus preparation

The Ccny and Ccnyl1 shRNA plasmids were obtained from the laboratory of Arial Zeng and are described in Zeng et al, 2016. For the Sox4 and Sox11 overexpression plasmids, full‐length Flag‐Sox4 and Flag‐Sox11 were excised from their respective pCS2⁺ plasmids (see below) and each ligated into the pLenti‐CAG‐IRES‐EGFP plasmid using BamHI and BsrGI restriction sites to produce two separate lentiviruses. EGFP was removed with this cloning strategy for both the Sox4‐ and Sox11‐overexpressing lentiviruses. All lentiviruses were packaged in 293T cells according to Lois et al, 2002.

Sox4/Sox11 cloning and mutagenesis

GoTaq (Promega, M7841) DNA polymerase was used to amplify full‐length Sox4 and Sox11 using cDNA obtained from whole E13.5 embryos. N‐terminal Flag tags were added via forward primers (see Appendix Table S1). BamHI and XhoI restriction sites were introduced by PCR to clone into the pCS2⁺ plasmid, and sequencing was performed to verify the integrity of the constructs. Mutagenesis of the Sox4/11 GSK3 phospho‐motifs was performed by amplifying full‐length Flag‐tagged plasmids with Phusion™ DNA polymerase (NEB, M0530L) using primers designed to mutate selected serines into alanines. Extension times of 5 min and 18 cycles were carried out for each PCR. Amplification products were PCR purified (Machery Nagel, 740609.250), digested with Dpn1 restriction enzyme (NEB, R0176) for 1 h at 37°C, PCR purified again, and then transformed into electro‐competent bacteria. Colonies were screened for mutations by sequencing. Double GSK3‐motif mutations (e.g. for Sox11) were performed sequentially.

Real‐time quantitative PCR

mRNA from NPCs and embryonic dorsal forebrains was extracted using the Nucleospin RNA XS kit (Machery‐Nagel, 740902.50) and the Nucleospin RNA kit (Machery‐Nagel, 740955.250), respectively, according to the manufacturer’s instructions. Extracted mRNA was transcribed to cDNA using random hexamer primers. PCR was performed on a Roche Light Cycler 480 using the Universal ProbeLibrary system. Gapdh was used as a housekeeping gene unless otherwise stated. See Appendix Table S1 for primer sequences.

Immunoblotting

Cells were adjusted to equal numbers, washed with PBS, resuspended in triton lysis buffer (20 mM Tris–HCl, 150 M NaCl, 1% Triton X‐100, 1 mM EDTA, 1 mM EGTA, 1 mM b‐glycerolphosphate, 2.5 mM sodium pyrophosphate and 1 mM sodium orthovanadate), incubated for 30min on ice and then spun down at full speed for 7 min to clear lysates. For brain tissue samples, forebrains were dissected, resuspended in triton lysis buffer and sonicated in a water bath for 15 min before being processing as described above. Lysates were heated at 70°C in NuPage LDS buffer (Thermo Fisher, NP0007) with 50 mM DTT. Samples were separated on 7.5% polyacrylamide gels, transferred to nitrocellulose and blocked with 5% skim‐milk powder or 5% BSA in Tris‐buffered saline with 0.05% Tween‐20 (TBST) for 1 h at room temperature. Primary antibodies were diluted in blocking buffer and incubated overnight at 4°C. After 3 washes in TBST, membranes were incubated with peroxidase‐linked secondary antibodies for 1h at RT. Following an additional 3 washes, membranes were treated with Supersignal West Pico solution (Thermo Scientific, 34579). Images were acquired on an LAS‐3000 system (Fuji Film).

Lambda‐phosphatase treatment

Cells were lysed in modified RIPA lysis buffer 50 mM HEPES pH 8.0, 300 mM NaCl, 1% Triton X‐100, 0.2% sodium deoxycholate, 0.05% SDS, 5 mM MgCl₂ supplemented with EDTA‐free protease inhibitor tablet (Pierce, A32965) for 30 min on ice, spun down full speed at 4°C for 5 min and supernatants collected. Lysates were incubated with lambda phosphatase (NEB, P0753S) for 30 or 60 min (indicated in figure legends) at 30°C according to the manufacturer’s instructions.

Ubiquitination assays

HA‐tagged ubiquitin plasmid was co‐transfected with Flag‐tagged Sox4/Sox11 or PCS2⁺ empty vector into 293T cells in 6 cm dishes for 48 h. Cells were treated for 4 h with 20 µM MG132 (Sigma‐Aldrich, C2211) before being harvested in triton lysis buffer. A total of 200 µg protein was incubated with 20 µl FLAG beads (Sigma‐Aldrich, A2220) overnight at 4°C with rotation. Beads were washed 4x with triton lysis buffer, resuspended in 20 mM Tris–HCl pH 7.5 buffer containing 0.1% SDS and heated for 5 min at 95°C to dissociate Sox binding partners. Cells were spun down, resuspended in triton lysis buffer and incubated a second time with 20 µl FLAG beads overnight at 4°C with rotation. Following 4 additional washes, NuPage LDS buffer supplemented with 50mM DTT was added to beads and samples were boiled at 70°C for 10 min.

Invitro kinase assays

Ten micrograms of Sox4/Sox11 wild‐type and mutant plasmids were transfected into 293T cells in 10‐cm dishes for 48 h. Cells were lysed in 2 ml modified RIPA lysis buffer and incubated with 20 µl FLAG beads overnight at 4°C with rotation. Cells were then washed once with modified RIPA buffer, followed by two washes with high salt wash buffer (30 mM HEPES pH 8.0, 500 mM NaCl, 5 mM MgCl₂ supplemented with EDTA‐free protease inhibitor tablet) and one wash with low salt wash buffer (30 mM HEPES pH 8.0, 500 mM NaCl, 5 mM MgCl₂ supplemented with EDTA‐free protease inhibitor tablet). Cells were then resuspended in low salt wash buffer and treated with lambda phosphatase as described above.

For kinase assay, beads were washed once in wash buffer (30 mM HEPES‐KOH pH 7.7, 10 mM MgCl₂, 0.2 mM β‐mercaptoethanol) and equally divided in two 1.5‐ml tubes. 10 nM GSK3β (Millipore, 14‐306) or reaction buffer was added to the corresponding tube, and the reaction was started by adding 50 µM ATP containing 1 µCi of radiolabeled ³²P‐γATP (Permin‐Elmer, NEG502A001MC). Phosphorylation assays were performed at 37°C for 15 min in 30 µl of kinase buffer (30 mM HEPES‐KOH pH 7.7, 10 mM MgCl₂, 1mM DTT, 0.2% BRIJ‐35). Reactions were stopped by directly adding 10 µl of 4xSDS laemmli buffer, and samples were heated at 99°C for 7 min. 15 µl was loaded into a 10% SDS–PAGE and a 7.5% Phos‐Tag gel (Alpha laboratories, 304‐93526) followed by staining with Quick Coomassie^® (Protein Ark, GEN‐QC‐STAIN‐1L) and imaging of the dried gel with phosphorimager (Sapphire™ Biomolecular Imager, Azure Biosystems).

Antibodies

Rabbit polyclonal antibodies against Ccny and Ccnyl1 were raised against synthetic peptides and affinity‐purified as previously described (Davidson et al, 2009; Koch et al, 2015). No cross‐reactivity with Ccnyl1 was detected for the anti‐Ccny antibody and vice versa. The rabbit polyclonal pLRP6 T1479 antibody is described in Davidson et al (2005). The Sox4 and Sox11 polyclonal guinea pig antibodies are described in Hoser et al, 2008. The Sox4‐phospho polyclonal rabbit antibody is described below. All other antibodies used in this study are commercial and are described in the Appendix Table S2.

Sox4 phospho‐antibody preparation

Rabbits were injected bi‐weekly with AASpPAAGRC peptide conjugated to Imject Maleimide‐Activated Blue Carrier Protein (Thermo Fisher, 77664), and serum was collected after 4 months. Rabbit antibody was purified by first passing through phosphopeptide immobilized to Sulfolink beads (Thermo Fisher, 20401) and then subtracted using non‐phosphorylated peptide. Serum from 3 rabbits was collected and purified, and the serum with the highest antibody yield was used for subsequent experiments. Rabbit housing, injections and serum collection was performed by Pineda Antikörper‐Service (Berlin, Germany).

Immunofluorescence

Paraffin

Tissues were fixed overnight in 4% paraformaldehyde at 4°C, progressively dehydrated and embedded in paraffin. 7‐µm thick sections were rehydrated, boiled in a pressure cooker for 2 min with citrate/EDTA buffer (10 mM sodium citrate, 5 mM Tris–HCl, 2 mM EDTA, pH 8.0) and blocked in blocking buffer (PBS solution containing 10% normal donkey serum, 1% BSA and 0.1% Triton X‐100) for 30 min at room temperature. All primary antibodies were diluted in blocking buffer and applied overnight at 4°C. Secondary antibodies were diluted 1:500 in blocking buffer containing Hoechst 33258 dye (1:1,000)(Sigma‐Aldrich, 861405) to stain DNA and applied at room temperature for 1 h. For histological analysis, 7‐µm thick sections were stained with haematoxylin and eosin according to standard procedures.

Frozen sections

Tissues were fixed for 2 h in 4% paraformaldehyde at 4°C, incubated in 30% sucrose overnight, embedded in Tissue‐Tek (OCT, Sakura, 4583) and frozen at −20°C. Sections of 8 µm thickness were washed briefly in PBS and then heated in a microwave for 5 min in sodium citrate buffer (10mM sodium citrate, 0.05% Tween pH 6.0). Blocking and antibody applications were performed as described above.

Cell culture

Cells were cultured on coverslips coated with poly D‐lysine and fixed with 4% PFA at room temperature for 10 min. Following fixation cells were washed twice in PBS and then blocked and stained with indicated antibodies as described above.

TUNEL staining

TUNEL staining was performed with the Click‐IT Plus TUNEL assay (Thermo Fisher, C10617) according to the manufacturer’s instructions.

RNAScope in situ hybridization

Paraffin‐embedded forebrain sections were processed for RNA in situ using the RNAScope 2.5 HD assay‐RED kit (Advanced cell diagnostics, 322360) (Chromogenic and Fluorogenic) according to the manufacturer’s instructions. The RNAscope probe used was Axin2 (NM 015732, region 330‐1287).

Quantification and statistical analysis

Sample sizes (individual embryos, litter numbers and wells (in vitro experiments)) are reported in each figure legend. All cell counts were performed in standardized microscopic fields (additional information in Appendix Methods) using either the Fiji cell counter plug in (quantifications done blindly) or user‐defined macros (no blinding for quantification). All statistical analyses were conducted using GraphPad Prism. Data normality was tested by Shapiro–Wilk normality test, and variances between groups were tested using F‐test. Means between two groups were compared using two‐tailed unpaired Student’s t‐test, and means between multiple groups were compared using one‐way or two‐way analysis of variance (ANOVA) followed by Tukey’s multiple comparison tests. Statistical outliers were calculated using Grubb’s test. Results are displayed as arithmetic mean ± standard error of mean (SEM). Where indicated results are shown as fold change vs. controls. Statistically significant data are indicated as: *P < 0.05, **P < 0.01, and ***P < 0.001. Non‐significant data are indicated as ns.

Note on terminology

The term "self‐renewal" has been commonly used not only for asymmetric NPC divisions that result in the maintenance of the pool size of a given neural progenitor type (e.g. 1 AP − > 1 AP + 1 BP), but also for symmetric divisions of NPCs that increase the pool size of a given NPC type (e.g. 1 AP − > 2 APs, 1 BP − > 2 BPs). To clarify matters, we refer to the latter type of cell division as "increased self‐renewal".

Author contributions

FDS and KZ conceived, performed and analysed experiments. JH supervised animal husbandry and assisted with experiments. EF performed the kinase assays. AP performed IUE experiments and cortical layer thickness measurements. MWB performed astral microtubule quantification. VV performed the RNAScope experiments, and AS helped with manuscript revision. WBH contributed by planning experiments and revising the manuscript. CN supervised all aspects of the project. FDS wrote the manuscript with input from all authors.

Conflict of interest

The authors declare that they have no conflict of interest.

Figure 7. Division of labour model between WNT/STOP and WNT/β‐catenin signalling during neocortical neurogenesis.

Ccny/l1 stimulate cyclin‐dependent kinases (CDK) 14 and 16 predominantly during G2/M (not shown) to phosphorylate and activate LRP6. This co‐receptor activation leads to a peak of WNT/STOP signalling and GSK3 inhibition in mitosis. Mitotic WNT signalling has two consequences: First, it increases asymmetric AP division via unknown effectors, which leads to more basal progenitors and hence post‐mitotic neurons; second, it protects Sox proteins from proteasomal degradation. Sox4 stabilization in mitotic APs leads to increased BP generation, and its stabilization in mitotic BPs may promote BP self‐renewal and/or differentiation. Sox11 is stabilized in both mitotic and non‐mitotic cells of the SVZ, which leads to increased post‐mitotic neuron generation. WNT/β‐catenin signalling primarily promotes AP self‐renewal.

Supporting information

Appendix

Expanded View Figures PDF

Source Data for Expanded View/Appendix

Source Data for Figure 1

Source Data for Figure 2

Source Data for Figure 3

Source Data for Figure 4

Source Data for Figure 5

Source Data for Figure 6

The EMBO Journal (2021) 40: e108041.

Data availability

This study includes no data deposited in external repositories.