Zbtb20 modulates the sequential generation of neuronal layers in developing cortex

Dynamic Zbtb20 expression in telencephalic progenitors

By E11.5 a gradient of Zbtb20-lacZ activity was evident in the VZ of the lateral pallium (LP) of the heterozygous Zbtb20 ^lacZ/+ embryos (Additional file 1: Figure S1A, arrow). Between E12.5 and E13.5, Zbtb20 immunostaining confirmed a strong signal in both the ventral (VP) and lateral (LP) pallium (Fig. 1a, arrow). Double staining for Zbtb20 and the pallial progenitor marker Pax6 revealed a nearly complete co-expression at E13.5 (Fig. 1b1-b6) in LP (101 out of 101 assayed Zbtb20⁺ cells co-expressed Pax6 (100 %, n = 3). In VP, however, only 40 % of the Zbtb20⁺ cells co-expressed Pax6 (59 out of 148 Zbtb20⁺ cells, n = 3). Notably, co-labelling for Zbtb20 and TF CoupTF1, which at this stage is expressed by both pallial and subpallial progenitors, showed a complete co-expression in LP and VP (143 out of 143 Zbtb20+ cells in LP/VP (n = 3) co-expressed CoupTF1 (Fig. 1c1-c6).

By E14.5-E15.5 the expression of Zbtb20 was spread throughout the entire pallial VZ (Fig. 1d; Additional file 1: Figure S1B, arrows). Co-staining for Zbtb20, RGC marker Pax6 and the mitotic marker phosphorylated vimentin (pVim) confirmed that TF Zbtb20 is expressed in dividing RGCs at the apical surface of VZ (Fig. 1e1-e4, arrows; 90 out of 93 assayed pVim⁺ cells on the apical surface were co-labeled for Zbtb20, 97 %, n = 3).

At early postnatal (P) stages (P4), Zbtb20 immunosignal was evident in the most superficial layers of LP/DP (Fig. 1f, arrow), where Zbtb20⁺ cells co-expressed Satb2 (Fig. 1g1-g3, arrows; 102 out of 129 assayed Zbtb20⁺ cells were co-labeled for Satb2, 79 %, n = 3) and Brn2 (Additional file 2: Figure S2A1-A3, arrows; 111 out of 121 assayed Zbtb20⁺ cells were co-labeled for Brn2, 93 %, n = 3), markers of UL neocortical neurons [40, 41]. However, Zbtb20 did not co-localize with neither the L5 marker Ctip2 (Additional file 2: Figure S2B1-B3) nor the L4 marker ROR [42] (Additional file 2: Figure S2C1-C3), suggesting that the Zbtb20 expression in UL neurons is restricted to L2-L3 neurons. The expression pattern in ULs was preserved at P8 but almost disappeared at P12 (data not shown). In the early postnatal SVZ, Zbtb20 also maintained a strong expression in Nestin + RGCs (Fig. 1h1-h3; 90 out of 92 assayed Zbtb20⁺ cells in SVZ were co-labeled for Nestin, 92 %, n = 3), suggesting a possible involvement in postnatal neurogenesis.

In summary, beginning at E11.5 in VZ of VP/LP, the expression of TF Zbtb20 expands into the VZ of the entire pallium at E14.5 and thereafter, suggesting that the timed expression of Zbtb20 may be involved in generation and/or specification of the UL neurons.

Disturbances in superficial neocortical layers in Zbtb20 loss-of-function

Cresyl violet (Nissl) staining of P10 coronal brain sections revealed apparent defects in cortical layering in Zbtb20 ^lacZ/lacZ mice (Fig. 2a1-b2). While the LLs (L6, L5) appeared overrepresented in the Zbtb20 ^lacZ/lacZ somatosensory (SS) cortex, the ULs (L4-L2) seemed thinner (Fig. 2b1-b2). This impression was confirmed by NeuN immunoassaying (Fig. 2b3-b4). In order to investigate the UL disturbances in more detail, we performed immunostaining for specific UL neuronal markers. Given the widespread expression of Zbtb20 in VZ of the entire pallium at E14.5, we first investigated the expression of the L4 marker ROR [43]. Intriguingly, in the mutant cortex, the thickness of L4 was greatly augmented (Fig. 2c1-c3). This notion was confirmed by another marker with a strong expression in L4, TF CoupTF1 (see below). Immunoassaying for the global (L2-L4) UL marker Cux1 [18, 19], however, showed a reduction of UL neurons (Fig. 2d1-d3). Additional UL markers, Brn2 (L2, L3, and L5; [40, 41]) and Satb2 (L4-L2; [7, 8]) confirmed the strongly diminished representation of UL subsets in the Zbtb20 ^lacZ/lacZ mice (Fig. 2h1-i3).

The overall reduction of ULs, accompanied by the selective expansion of L4, raised a question on the status of the L2-L3 neurons in the mutant cortex. Therefore, we quantified the L3-L2 Cux1⁺/ROR⁻ population (Fig. 2f2, g2, arrow), which was markedly depleted (Fig. 2g3), in contrast to the L4 Cux1⁺/ROR⁺ population, which was augmented (Fig. 2g4). In order to study whether L3 or L2 neuronal subsets were specifically affected, we made use of the expression patterns of TF FoxP1, which is expressed in L6a and L5-L3 [44] and TF Brn2, which specifically marks L3 and L2 [40, 41]. Double immunostaining for FoxP1 and Brn2 revealed an ectopic expansion of FoxP1 expression into the normal position of L2 in the mutant cortex, while cells specifically fated to L2 identity (Brn2⁺/FoxP1⁻, Fig. 2j1-j2, arrows) were almost completely missing (Fig. 2j3). The depletion of the ULs was not due to an enhanced apoptosis, as studied by the expression of activated Caspase-3 (data not shown). To investigate whether the molecular boundary between L5 and L4 was preserved, we applied Ctip2 (L5)/ROR (L4) double immunohistochemical (IHC) staining, and we found that these two subpopulations of cortical neurons were properly segregated (Fig. 2k1-k2).

To sum up, these findings indicate that Zbtb20 deficiency results in a significant diminishing of L3 and especially L2 neuronal subsets as well as in an augmented and ectopic presence of L4 neurons in UL position.

Enhanced deep layers and normal arealization in Zbtb20 ^lacZ/lacZ cortex

To study quantitatively the apparent enhancement of both L5 and L6 sets in the mutants (Fig. 2b1-b4), we performed IHC staining and counting of cells in the SS cortex on cross sections of both genotypes with antibodies for TFs FoxP2 [44] in L6 (Fig. 3a1-a4), Tbr1 [9] in L6 (Fig. 3b1-b4) and Ctip2 [15, 45] in L5 (Fig. 3c1-c4). Indeed, the results revealed a statistically significant increased number of both L6 and L5 neurons in the mutant as compared with the control Ncx (Fig. 3a5, b5, c5). Furthermore, ISH staining of sagittal P4 brain sections indicated that the mutant cortex displayed enhanced LL neuronal subsets, including Fezf2 ⁺ L5 [15, 46] (Additional file 3: Figure S3A1-A4), Clim1 ⁺ L5 [46] (Additional file 3: Figure S3B1-B4), and Id2 ⁺ L5 and Id2 ⁺ L6 [47] (Additional file 3: Figure S3C1-C4). In order to confirm that the described layering abnormalities in the Zbtb20 ^lacZ/lacZ mice are not restricted to the SS cortex, we studied the patterning of the primary motor cortex and, similarly to SS cortex, we found a decrease of the ULs and enhancement of the LLs (Additional file 4: Figure S4).

In layer 5, the rostral limit of expression of TF Id2 outlines the position of the border between the motor/somatosensory (M/SS) cortex, which is also marked by the caudal limit of Id2 expression in L3-L2 [48]. In Zbtb20 ^lacZ/lacZ cortex, the M/SS boundary appeared rostrally displaced into the M field (Additional file 3: Figure S3C1-C2, asterisk). We therefore examined other cortical area-specific markers, including Cadherin-8 (Additional file 3: Figure S3D1-D2, asterisk), Serotonin (Additional file 3: Figure S3E1-E2, asterisk), Bhlhb5 (Additional file 3: Figure S3 F1-F2, asterisk), ROR (Fig. 2c1-c2, asterisk, and data not shown), Cadherin-6 and Lmo3 (data not shown). None of these markers exhibited shifts in the mutants along the antero-posterior axis, so we concluded that the neocortical arealization in Zbtb20 ^lacZ/lacZ mice is grossly not affected. This notion is also supported by the preserved pattern of the graded expression of TFs Pax6, Emx2, Foxg1 and Lhx2 in Zbtb20 ^lacZ/lacZ embryonic pallium at E12.5 [35], a stage at which these TFs are known to have crucial roles in specification of the intrinsic program of cortical arealization encoded in the progenitors [23].

ZBTB20 modulates the temporal onset for generation of distinct neuronal layer identities

To investigate whether Zbtb20 controls the switch to generation of neurons with different layer identities, we performed BrdU birthdating experiments at E12.5, E14.5 and E16.5 when predominantly LLs (L6, L5), L4 or L3-L2 neurons are born, respectively (Fig. 4). Taking advantage of the fact the pattern of NeuN immunostaining allowed distinguishing the location of LL and UL in the postnatal cortex (Fig. 2b3-b4), we calculated the percentage of BrdU⁺/NeuN⁺ cells located to either LLs (L5-L6) or ULs (L2-L4) out of the total BrdU⁺/NeuN⁺ cells in frames spanning the entire cortex (Additional file 5: Figure S5). Analysis of the position of the E12.5-born cells revealed no differences in their laminar location between WT and the mutant (Fig. 4a1-a3; Additional file 5: Figure S5A1-A3). In contrast, birthdating at E14.5, revealed a significantly larger proportion of BrdU⁺ cells in the deep position of the Zbtb20 ^lacZ/lacZ cortex, and on contrary, a smaller fraction of such cells was found in the superficial position of the mutants (Additional file 5: Figure S5B1-B3; also Fig. 4f1-f3). Similarly, BrdU pulse labelling at E16.5 (at the peak of UL generation) showed a reduced distribution of tagged cells at superficial position and significantly more deeply located BrdU⁺ cells in the mutant cortex (bins 4–6; Fig. 4k1-k3), that was also evident on BrdU/NeuN double-stained sections (Additional file 5: Figure S5C1-C3).

Using BrdU co-staining with layer-specific neuronal markers, for L5 (Neurofilament/Smi32 and Ctip2) and L6 (Tbr1), we quantitatively investigated the layer fate of cells generated at stages E12.5, E14.5 or E16.5. The results revealed that the mutant Ncx had significantly increased populations of L6 and L5 neurons born at E12.5 (Fig. 4c1-e3), as well as at E14.5, the peak of generation of L4 neurons during normal corticogenesis (Fig. 4h1-j2). At E16.5 such expansion of L6/L5 fate identities was no more evident (Fig. 4m1-m2). However, the cells born at E16.5 in the mutant Ncx demonstrated a significant increase of the co-labeling with the L4 marker ROR as compared to the control Ncx (Fig. 4o1-o2, arrows, O3). These results suggest that the developmental window for generation of L6-L5-L4 neurons was expanded by at least 2 days for each neuronal type, which will profoundly affects the progenitor pool size for L3-L2 neurons. In a further support of such a scenario were the results after analysis of the fractions of Cux1⁺/ROR⁺ (L4) and Cux1⁺/ROR⁻ (L3-L2) neuronal subsets born at stages E12.5, E14.5 or E16.5 in WT and mutant mice (Additional file 6: Figure S6). We found that at P12, the UL fractions born at E12.5 or E14.5 did not differ significantly between WT and mutant mice (Additional file 6: Figure S6A1-B3). However, changes were observed in the E16.5-born UL neuronal fractions in Zbtb20 LOF as compared to WT: the L4 fraction was larger, while the L3-L2 fraction was diminished (Additional file 6: Figure S6C1-C3).

Together, these results strongly suggest that the timed expression of transcriptional repressor Zbtb20 in cortical progenitors (appearing in the entire pallium only after E14.5) could control the transition from early- versus late born neuronal layer identities.

Impaired late neurogenesis and neuronal migration in Zbtb20-deficient cortex

By using 40 min BrdU-pulse labelling in vivo at E16.5 (the peak of production of UL neurons) we found a significant reduction of the progenitor proliferation in the Zbtb20LOF cortex (Fig. 5a1-a3). Additionally, BrdU/Ki67 double-labelling after a 24 h BrdU pulse (E15.5- > E16.5), indicated an increased progenitor exit from the mitotic cycle as measured by the percentage of the BrdU⁺/Ki67⁻ cells versus all BrdU⁺ cells (Fig. 5b1-b3). Consistent with these data, we found at E16.5 in the mutant DP a reduction of Tbr2⁺ IPs (Fig. 5c1-c3), the main neuronal source for generation of neurons with an UL neural fate [49]. Notably, we did not detect changes in the progenitor cell exit during early neurogenesis (time window of LL generation) using a 24 h BrdU pulse at E12.5- > E13.5, followed by BrdU/Ki67 double-labelling (Fig. 5d1-d3).

Our previous expression analysis of Zbtb20 in developing cortex at stage E18.5 suggested a migratory delay of NeuroD1⁺ neurons (Fig. 4h2 of [35]). At the same stage, we showed here a band of Id2⁺, Math2/Nex⁺ neurons in the pallial SVZ in Zbtb20 KO mice (Additional file 7: Figure S7, arrows), suggesting migratory abnormalities of the lately-born neurons. Indeed, in the UL birthdating experiments at E16.5, the mutant Ncx contained significantly more E16.5-born BrdU⁺/Cux1⁺ cells in the L6 (Fig. 4n1-n3) as well as in the subcortical white matter (data not shown). Notably, the retained Cux1⁺ cells were ROR⁻ (Fig. 2e2/f2; arrowheads) indicating that a sub-population of correctly specified L3/L2 neurons exhibits an impaired migration towards their final destination to the CP. Notably, the postmitotic expression of Zbtb20 in cortical plate is confined to the Cux1⁺/ROR⁻ population.

Zbtb20 deficiency affects the expression of CoupTF1 in developing cortex

As noticed, the orphan nuclear receptor CoupTF1, whose strong expression is normally restricted to L4 in the SS area [45, 50] was ectopically expressed as a thick band along the entire AP axis of the Zbtb20 mutant (Fig. 6a1-a4; arrows). Similar to the presented here abnormalities of Ncx in Zbtb20 LOF, overexpression of CoupTF1 in the pallial VZ promotes progenitor exit from mitotic cycle, inhibits the IPs production and causes enhanced generation of early-born at the expense of late generated neuronal fates [51]. At E12.5, both CoupTF1 [50, 52] and Zbtb20 are expressed at the corticostriatal border in faint DV gradients. Indeed, the double IHC at E12.5 showed a co-localization of Zbtb20 and CoupTF1 in the VZ of the control animals (Fig. 1c1-c3), and significant enhancement of CoupTF1 expression in Zbtb20 ^lacZ/lacZ mice as revealed by both ISH (Fig. 6b1-b2) and IHC (Fig. 6C1-C2). Thus, the observed increased generation of early-born neuronal fates (L6,L5) in Zbtb20 LOF might be mediated by modulation of CoupTF1 expression.

To study a possible regulation of CoupTF1 expression by Zbtb20 at molecular level, we investigated whether Zbtb20 binds to the promoter of CoupTF1 [53] (Fig. 6d-d2, Additional file 8: Figure S8). In ChIP assays, we used neural stem cultures (NSC) from E15.5 cortices from WT and Zbtb20KO embryos and Zbtb20 antibody. We found that Zbtb20 binds with low affinities with fragments at locations of −172/-364 (ChIP_3) and +53/+256 (ChIP_5) of the CoupTF1 promoter and the first intron (Additional file 8: Figure S8). Zbtb20 occupies the CoupTF1 promoter with highest affinity at the location of −153/+49 (ChIP_4), which contains multiple DNA binding motifs of Zbtb20 [33] (Fig. 6d1-d2, Additional file 8: Figure S8). These data suggests that TF Zbtb20 is a regulator of the expression of CoupTF1 most probably acting as a repressor. In a further support, ISH with Zbtb20 in situ probe on E15.5 brain sections from WT and CoupTF1 ^−/− mice [51], showed a decreased mRNA signal (Additional file 9: Figure S9; arrows) suggesting a cross regulatory loop between these two TFs.

Zbtb20 is a regulator of timed neurogenesis in developing neocortex

During corticogenesis, the timed generation of layer-specific fates depends on intrinsic and extrinsic cues acting at a given developmental stage [1]. This process includes consecutive steps of repression of cell fates generated during earlier stage. Here we showed that in a lack of TF Zbtb20 in the cortical progenitors, neurons with L6-L5-L4 identity continue to be produced heterochronicly, in an expanded by at least 2 days time window, thus substantially shortening the time for production of L3 and especially of L2 neurons (Fig. 7). We consider that this temporal alteration of the schedule for layer production together with the disclosed enhanced exit from mitosis of E16.5 progenitors are the main factors leading to L6-L5-L4 enlargement in the Zbtb20LOF. Because of its dynamics of expression, Zbtb20 exerts different roles at distinct developmental stages: at E14.5, the time at which the full extent of Zbtb20 expression in the pallial VZ progenitors is achieved, this TF restricts L6-L5 versus L4 fate, while at E16.5, the peak of UL generation, Zbtb20 restricts L4 versus L3-L2 fate. In addition, in Zbtb20 LOF cohorts of progenitors leave the cell cycle prematurely, diminishing the progenitor pool that remains for the latest-born neuronal types.

Genes known to regulate the sequential transitions between cortical subtypes include CoupTF1 [51, 54, 55], FoxG1 [14, 56], Gli3 [57] and Brn2 [58]. The expression of neither FoxG1 [35] nor Gli3 (this study, data not shown) were altered in Zbtb20 ^lacZ/lacZ mice. Interestingly, both Zbtb20 and CoupTF1 exerted a similar expression dynamics, starting in progenitors of VP, LP at E12.5 and progressively expanding at later stages in VZ/SVZ of the entire pallium, and even continue to be co-localized postnatally in UL neurons. As shown here, Zbtb20 binds the promoter of CoupTF1, and in Zbtb20 LOF the expression of CoupTF1 was elevated as early as E12.5, indicating that Zbtb20 controls directly the expression of CoupTF1. Notably, several aspects of the neocortical phenotype in Zbtb20KO mice are reminiscent to abnormalities observed after in vivo overexpression of CoupTF1 [51], including: (i) expansion of early born neuronal sets, (ii) changed balance between early and late born neurons, (iii) enhanced cell cycle exit of progenitors and (iv) diminished production of IPs. On the contrary, in the pallium of CoupTF1 KO mice, the expression of Zbtb20 was reduced, making possible the existence of a feedback loop between Zbtb20 and CoupTF1. The lack of birthdating and cell cycle abnormalities during L6-L5 neurogenesis in the mutant pallium during the early neurogenesis indicates that Zbtb20 most probably exerts effects on temporal specification of the LLs via the ectopic CoupTF1 expression.

Acting at the border between the archi- and isocortex, Zbtb20 repression seems to be important to inhibit neocortical cell fate [33]. Upon overexpression in progenitors of medial pallium, Zbtb20 directly binds and represses the activity of genes that are specifically expressed in LL neurons (Tbr1, FoxP2, Fezf2, Ctip2, Sox5) or ULs (e.g. Rorb, Satb2, Cux1/Cux2) [33]. However, in developing Ncx, Zbtb20 co-exists with, and thus might directly repress, only few of those TFs: Fezf2 and FoxP2 (in mitotic VZ precursors), Cux1/2 (in mitotic cells in VZ/SVZ as well as in postmitotic cells in ULs), and Satb2/Brn2 (in postmitotic ULs). The enhancement of markers which do not co-exist in the same cells with Zbtb20, such as Ctip2 (L6 and L6) and ROR (L4), can be explained by action of Zbtb20 at the level of the progenitors of these neurons. The expansion of Ctip2⁺ L5 subsets in Zbtb20 LOF of neurons most probably involves a de-repression of the Fezf2 gene in VZ/SVZ. Together with the results in [33], our data suggest that TF Zbtb20 acts as a modulator of the Fezf2/Ctip2/Tbr1/Satb2 network mostly in pallial VZ progenitors.

The mutant phenotype of the latest-born Zbtb20⁺ L3-L2 cells suggests also an involvement of Zbtb20 at the postmitotic level, probably influencing their proper migration towards the cortical plate. The involvement of Zbtb20 in the regulation of neuronal migration of late-born UL cells was supported by our findings indicating: (i) retention of Cux1⁺/ROR⁻ (Fig. 2f2, arrowheads) and Bhlhb5⁺ (Additional file 3: Figure S3 F2, inset) cells below the ULs of the mutant at P12; (ii) their birth date at E16.5 (Fig. 4n2); (iii) accumulation of postmitotic cells in the mutant SVZ at E18.5 (Additional file 7: Figure S7). Thus, the prenatal reduction of UL precursor pool is possibly exacerbated by a postnatal migrational deficit of L3-L2 neurons. Potential target genes for this deficit are TFs Cux1 which is expressed in both mitotic and postmitotic progenitors [19], and Brn2 [40] which was recently shown as a possible Zbtb20 downstream target [59].

Animal experiments

Animals were handled in accordance with the German Animal Protection Law, after an approval of the experiments by the Niedersächsische Landesamt für Verbraucherschutz und Lebesmittelsicherheit (LAVES)/Oldenburg, contract No 33.9-42502-04-11/0622 from 07.12.2011. Experiments were completed prior to January 1^st 2013. All surgical procedures were performed under isoflurane/N₂O anesthesia, and all efforts were made to minimize suffering. The Zbtb20 gene targeting and the generation of the transgenic mice have been previously described [35]. The Zbtb20 knock out (KO) mice lack the functionally important BTB/POZ domain of the protein as well as the first of five zinc fingers, which were replaced by a lacZ-neomycin cassette. Therefore, homozygous mutants will be referred to as Zbtb20 ^lacZ/lacZ mice within this study. The specificity of the deletion and the complete loss of Zbtb20 protein have been confirmed as described [35].

ChIP analysis

ChIP analyses were performed as described previously [60] with some modifications. Cortical progenitors were prepared from E15.5 WT and Zbtb20KO mice [35] and cultured as described [61]. ChIP experiments were performed using an EZ ChIP assay kit (Millipore), according to the supplier’s instructions with Zbtb20 antibody (HPA016815, Sigma). A genomic fragment of the Trim11 gene and a GFP antibody plus IgG were used as negative DNA and antibody controls, respectively [60]. Primer sequences are listed Additional file 10: Table S1.

Histological processing

Isolated embryos or brains at defined stages were washed in cold phosphate-buffered saline (PBS) and fixed in 4 % paraformaldehyde (PFA) overnight at 4 °C. Tissues were rinsed in PBS, and processed for standard cryo-embedding. Cryosections (16 μm thick) were washed and blocked for 1 h in blocking solution containing a normal serum. Primary antibodies were incubated overnight at 4 °C in the blocking solution. After washing, the sections were incubated with species-specific secondary antibodies from the Alexa series (Invitrogen) in blocking solution for 2 h at room temperature (RT), washed again and mounted with Vectashield mounting-medium (Vector Labs) containing DAPI. We used the following primary antibodies/dilutions: mouse anti-β-galactosidase (1:200; Promega, Madison, WI), rat anti-BrdU (1:200; Abcam), mouse anti-BrdU (1:50; BD Bioscience), goat anti-Brn2 (1:50, Santa Cruz Biotech, Santa Cruz, CA), rabbit anti-Caspase-3 (1:200; Cell Signalling, Cambridge, UK), mouse anti-CoupTF1 (1:1000; Perseus Proteomics, Tokyo, Japan), rat anti-Ctip2 (1:200; Abcam), rabbit anti-Cux1 (1:250; Santa Cruz), rabbit anti-FoxP1 (1:500, Abcam), rabbit anti-FoxP2 (1:500, Abcam), rat anti-Ki67 (1:100; Dako), mouse anti-Nestin (1:100; Millipore, Billerica, MA), mouse anti-NeuN (1:100; Millipore), mouse anti-Neurofilament (1:100; Abcam), rabbit anti-Pax6 (1:300; Covance), mouse anti-Pax6 (1:100; Developmental Studies Hybridoma Bank, DHSB), mouse anti-phospho-Histone H3 (1:50; Cell Signalling), mouse anti-phospho-Vimentin (1:300; MBL), mouse anti-ROR (1:100; Perseus Proteomics), mouse anti-Satb2 (1:200; Abcam), mouse anti-Sox2 (1:50; R&D systems), rabbit anti-Tbr1 (1:300; Abcam), rabbit anti-Tbr2 (1:200; Chemicon), rabbit anti-Zbtb20 (1:25–1:100, Sigma). The anti-BrdU antibodies were visualized after pre-treatment of tissues in 2 N HCl at 37 °C for 30 min. The anti-Zbtb20 antibody was used after an antigen retrieval by heating in a microwave (800 W, 3 times, 5 min each) in a citrate buffer (pH 6.0).

In situ hybridization

Whole heads from E12.5 or whole brains from E18.5 or P4 mice were dissected in ice-cold DEPC-treated PBS, fixed in 4 % PFA/PBS for 3 h at 4 °C, washed in PBS, and incubated in 25 % sucrose overnight at 4 °C. Specimens were sectioned at 16 μm after embedding and freezing in OCT cryomatrix (Leica Microsystems Nussloch GmbH, Wetzlar, Germany). Nonradioactive in situ hybridization was done as described [27].

Image analysis and quantification

Images were captured with an Olympus BX60 microscope, a Leica DM6000 epifluorescent system or a laser confocal microscope (Leica Sp5). For cell counts in sections from wild type (WT) and homozygous brains, we blindly counted the positive cells within equally sized frames (size of frames provided in the figure legends) on coded cross sections of somatosensory cortex in WT and mutant mice (n ≥ 3 per genotype). For BrdU birthdating experiments, frames spanning the entire cortex on cross brain sections were divided into 10 equally-sized bins, the BrdU⁺ cells in every bin were counted and divided by the total number of BrdU⁺ cells in all 10 bins. The size of the counting frames for BrdU/marker co-localization are provided in the figure legends.

Laser confocal microscopy was used to verify co-localization of multiple fluorescent signals. We performed Z sectioning at 0.5-1 μm intervals and optical stacks of at least 10 images were used for analysis, by the Leica Advanced Fluorescence software version 2.3.6. All images were processed with Adobe Photoshop (Version CS2) by overlaying the pictures, adjusting brightness, contrast and size.

Statistical analysis

Statistical evaluation was performed by Student’s T-test or one-way ANOVA followed by Tukey-Kramer’s post hoc analysis. Statistical significance between control and experimental condition was considered if p < 0.05. Data are presented as means ± s.e.m.