Sustained Pax6 Expression Generates Primate-like Basal Radial Glia in Developing Mouse Neocortex

The evolutionary expansion of the neocortex in mammals has been linked to enlargement of the subventricular zone (SVZ) and increased proliferative capacity of basal progenitors (BPs), notably basal radial glia (bRG). The transcription factor Pax6 is known to be highly expressed in primate, but not mouse, BPs. Here, we demonstrate that sustaining Pax6 expression selectively in BP-genic apical radial glia (aRG) and their BP progeny of embryonic mouse neocortex suffices to induce primate-like progenitor behaviour. Specifically, we conditionally expressed Pax6 by in utero electroporation using a novel, Tis21–CreERT2 mouse line. This expression altered aRG cleavage plane orientation to promote bRG generation, increased cell-cycle re-entry of BPs, and ultimately increased upper-layer neuron production. Upper-layer neuron production was also increased in double-transgenic mouse embryos with sustained Pax6 expression in the neurogenic lineage. Strikingly, increased BPs existed not only in the SVZ but also in the intermediate zone of the neocortex of these double-transgenic mouse embryos. In mutant mouse embryos lacking functional Pax6, the proportion of bRG among BPs was reduced. Our data identify specific Pax6 effects in BPs and imply that sustaining this Pax6 function in BPs could be a key aspect of SVZ enlargement and, consequently, the evolutionary expansion of the neocortex.

Introduction mostly of the proliferative type, and bRG constitute at least half of all BPs [23][24][25][26][27]. Moreover, in these species, both bRG and proliferative bIPs undergo mostly symmetric proliferative rather than neurogenic divisions [23,24,31]. These self-amplifying divisions significantly increase the number of BPs residing in the SVZ, consequently leading to the expansion of the SVZ. Moreover, the SVZ of these animals comprises not only a rodent SVZ-related layer called the inner SVZ (iSVZ) but in addition a novel layer called the outer SVZ (oSVZ) [32]. Importantly, these alterations in the mode of cell division and the resulting increase in BP abundance and formation of an oSVZ have been hypothesized to be major causes underlying the expansion of the neocortex [2][3][4][5][6]8,32].
A key question then is how these differences in BP abundance and mode of cell division between rodents and primates are brought about at the molecular level. A candidate regulatory mechanism is the differential expression of transcription factors. Of particular interest in this regard is Pax6 (accession number: AAH36957), a paired-box transcription factor [36][37][38][39]. Several mouse and rat mutant models have demonstrated that Pax6 is required for normal aRG abundance and mode of cell division [37,[40][41][42][43][44][45][46][47][48][49]. Moreover, although Pax6 mRNA levels are generally lower in BPs than APs, this down-regulation is much greater for mouse than human [50]. Consistent with this, only a minority of mouse and rat BPs (<30%) show Pax6 immunoreactivity (which is of lower level than in APs) [3,51,52], whereas the opposite is the case for primate, notably human, BPs (>80% Pax6-positive), with essentially all bRG and the majority of bIPs containing this transcription factor [3,[23][24][25][26][27]53,54]. Together, these findings raise the possibility that the differences in Pax6 expression between rodent and primate BPs may be responsible, at least in part, for the greater abundance and proliferative or self-renewal capacity of the latter.
We therefore sought to maintain Pax6 expression specifically in newly generated BPs in order to investigate if such expression would increase the abundance of BPs, notably of bRG, and their proliferative or self-renewal capacity. Using a novel approach of conditional Pax6 expression [16,21,55], we find that sustaining elevated Pax6 levels in BP-genic mouse aRG and the BP progeny derived therefrom increases both the proportion of bRG among the newly generated BPs and the self-renewing capacity of BPs.

Results
Expression of a Tamoxifen-Dependent Cre Recombinase in the Neocortex of Tis21-CreER T2 Knock-in Mouse Embryos Is Specific to Neurogenic Progenitors In mouse, the aRG subpopulation that gives rise to BPs, in contrast to self-amplifying aRG, specifically expresses Tis21, a pan-neurogenic progenitor marker [16,21,55]. Thus, as a tool towards maintaining Pax6 expression in mouse BPs, we generated a Tis21-CreER T2 knock-in mouse line. In this mouse line, exon 1 of Tis21 is replaced by CreER T2 containing a herpes simplex virus (HSV) tag at its C-terminus via homologous recombination (Fig 1A; for details, see S1 Fig), in order to limit Cre expression to Tis21-positive cells. To assess the cellular specificity of Cre expression, Tis21-CreER T2 knock-in mice were crossed with Tis21-GFP knock-in mice [16]. Immunofluorescence of the dorsolateral telencephalon of double-transgenic mice at embryonic day (E) 10.5, corresponding to the onset of Tis21 expression, and at E13.5, corresponding to the time point at which the in utero electroporations described below were conducted, showed that Cre was expressed in essentially the same cells as GFP (Fig 1B and 1C), indicating its expression selectively in the neurogenic subpopulations of cortical progenitors. Specifically, quantitation at E10.5 revealed that 97% of the cells containing nuclear Tis21-GFP were also positive for cytoplasmic Cre (Fig 1D), and no Cre was detected in Tis21-GFP-negative cells.
We next ascertained that the Tis21-CreER T2 mouse exhibits tamoxifen-dependent recombination by crossing this mouse line with a conditionally activateable GFP reporter mouse line, RCE:loxP [56] (Fig 1E). In these double-transgenic mice, GFP should be expressed only when CreER T2 has been translocated from the cytoplasm into the nucleus and excised a stop cassette that prevents the transcription of the GFP mRNA; the estrogen analog tamoxifen induces such CreER T2 translocation [57]. Indeed, no GFP-positive cells were observed in the absence of tamoxifen ( Fig 1G). In contrast, when treated with tamoxifen ( Fig 1F), GFP fluorescence was observed throughout the double-transgenic mouse brain (Fig 1I), and GFP-positive cells were found in all layers of the embryonic neocortex (Fig 1I'). This reflected Cre recombinase activity, because no GFP expression was observed when tamoxifen was administered to RCE:loxP offspring lacking the Tis21-CreER T2 allele ( Fig 1H). We conclude that Tis21-CreER T2 mouse embryos can be used to obtain tamoxifen-dependent recombination specifically in the neurogenic subpopulations of cortical progenitors.

Tis21-CreER T2 Mice Allow Conditional Pax6 Expression Specifically in Neurogenic APs and Their Progeny
To conditionally express Pax6 in BP-genic aRG of developing neocortex, we introduced a floxed Pax6 plasmid at midneurogenesis into APs of tamoxifen-treated Tis21-CreER T2 mouse embryos. Specifically, we generated a plasmid (referred to as Pax6-expressing plasmid) containing a constitutive promoter (CAG) followed by a membrane (GAP43)-GFP cassette flanked by two loxP sites, mouse Pax6, an internal ribosome entry site (IRES) sequence, and nuclear RFP (nRFP) (Fig 2A). Upon Cre-mediated recombination, the membrane-GFP cassette would be excised, leading to the simultaneous expression of Pax6 and nRFP. Introduction of this plasmid into APs of tamoxifen-treated Tis21-CreER T2 mouse embryos should ensure maintenance of Pax6 expression as mouse BPs arise from aRG divisions, as well as during their subsequent migration to, and function in, the SVZ. An identical plasmid but lacking the Pax6 and IRES sequences served as control (Fig 2A).
We first validated the Pax6-expressing plasmid by transfection of HEK 293T cells, a cell line in which the endogenous PAX6 gene is not expressed. Transfection with the Pax6-expressing plasmid alone resulted in GFP, but not nRFP, expression. Cotransfection of the Pax6-expressing plasmid and a Cre-expressing plasmid yielded both Pax6 and nRFP expression, whereas only nRFP expression was observed upon cotransfection of the control plasmid and the Creexpressing plasmid (S2 Fig).
We then explored whether the Pax6-expressing plasmid could be used in Tis21-CreER T2 mouse embryos to obtain conditional Pax6 expression specifically in the neurogenic subpopulation of APs and their progeny. To this end, we used the in utero electroporation technique where an electric field is generated across the cortical wall in order to allow for the unidirectional delivery of the negatively charged plasmid DNA, injected into the ventricular lumen, into APs. Dorsolateral telencephalon of tamoxifen-pretreated (E12.5) Tis21-CreER T2 mice was electroporated with Pax6-expressing plasmid at E13.5 and analyzed at E14.5, the peak of BP generation from neurogenic aRG [22] (Fig 2B). For the ease of presentation, we shall refer to this approach from here onwards simply as conditional Pax6 expression. Analysis of the Pax6 expression pattern yielded the following observations. First, analysis of the level of Pax6 immunoreactivity revealed that a subpopulation of cells had higher Pax6 immunoreactivity upon conditional Pax6 expression than in the control (Fig 2C and 2D). Upon closer inspection, all these highly Pax6-immunoreactive cells were RFP-positive, indicating that these cells constituted Pax6-expressing-plasmid-electroporated neurogenic APs and their progeny (Fig 2C,2D and 2F). The level of Pax6 immunoreactivity in these cells in the VZ was approximately 3-fold higher than that of the nonelectroporated APs or control-plasmid-electroporated neurogenic APs and their VZ progeny (Fig 2G), essentially all of which are known to express endogenous Pax6 [37,51,52]. In the SVZ, where mouse BPs normally down-regulate Pax6 expression [3,[50][51][52], this difference was even greater (%6-fold higher) ( Fig 2H).
Second, the appearance of these highly Pax6-immunoreactive and RFP-positive cells upon Pax6-expressing plasmid electroporation was strictly dependent on tamoxifen pretreatment (S3 Fig). Together, these observations allow us to equate the RFP-positive cells with the cells containing Pax6 due to the electroporation. To distinguish these conditionally Pax6-expressing cells from the cells expressing Pax6 endogenously, we shall refer to them from here onwards as exogenous Pax6-(exoPax6-) expressing cells. In addition, considering the results shown in Fig  1, we conclude that these cells constitute specifically the neurogenic subpopulation of APs and their progeny, notably the aRG-derived BPs.
Third, we found that electroporation with Pax6-expressing plasmid did not affect, after 24 h, the distribution of the progeny (RFP+ cells) of the electroporated neurogenic APs between (Fig 2C-2E) and within (S4 Fig) the germinal layers (i.e., VZ and SVZ). This implies that conditional Pax6 expression in neurogenic APs and their progeny, even if this expression exceeds the normal endogenous level, does not cause any overt effects on cell migration within the first 24 h after electroporation. The finding that RFP-positive cells are similarly distributed in control and upon conditional Pax6 expression allows for a valid comparison between germinal layers of the effect of conditional Pax6 expression in subsequent experiments.
Conditional Pax6 expression in aRG has previously been found to induce apoptosis when pan-aRG Cre drivers based on Emx1 and hGFAP promoter and regulatory sequences were used. However, this phenomenon was not observed with a Cre driver based on Ngn2 expression [58], which, similar (but not identical) to Tis21 expression, is characteristic of neurogenic progenitors [59]. It was therefore important to ascertain that conditional expression of Pax6 in Tis21-CreER T2 mice would not induce apoptosis. Indeed, immunofluorescence for the apoptosis marker activated caspase-3 did not reveal any significant difference in the number of caspase-3-positive cells between the progeny of control-plasmid-and Pax6-expressing-plasmid-electroporated neurogenic APs (S5 Fig). We therefore conclude that the present approach of conditional Pax6 expression is suitable to maintain high levels of Pax6 expression specifically in neurogenic APs and their progeny, notably the aRG-derived BPs, thus recapitulating the Pax6 expression pattern observed in BPs of developing primate neocortex.

Conditional Pax6 Expression Increases Cycling BPs
In assessing the functional consequences of sustained Pax6 expression in BPs, we sought to obtain initial clues as to the identity of the progeny of the Pax6-electroporated neurogenic APs. Using the cycling cell marker Ki67, we first investigated whether the exoPax6-expressing cells exhibited the same proportion of progenitors versus neurons as control cells (Fig 2I-2K). Whereas conditional Pax6 expression did not alter the percentage of Ki67-positive cells in the VZ, it did result in a significant increase in Ki67-positive cells in the SVZ (Fig 2K). This suggested that the conditional Pax6 expression increased the population of cycling BPs derived from electroporated aRG.
We noticed in some experiments that in both control and conditional Pax6 expression, more Ki67-positive cells were observed in the basal region of the SVZ, and in particular in the intermediate zone of the electroporated area, but not in the contralateral area nor in nonelectroporated dorsolateral telencephalon. This reflected a previously described side effect of in utero electroporation, that is, the displacement of some Pax6-positive cells towards the cortical plate [60]. Importantly, this side effect does not affect the findings described in the present study for three reasons. First, all our data are comparisons between control and conditional Pax6 expression, both of which involve identical conditions of in utero electroporation. Second, all our quantifications are confined to electroporated, RFP-positive cells, and the electroporation side effect has been reported to affect mainly nonelectroporated cells [60]. Third, our quantifications of cells in the SVZ exclude cells in the intermediate zone.

Conditional Pax6 Expression Increases the Proportion of Cortical Progenitors in S-phase
To gain further insight into a possible regulation of the cell cycle of cortical progenitors by conditional Pax6 expression, we examined specific cell cycle parameters. We first examined the effect of conditional Pax6 expression on the total cell cycle length (Tc) of neurogenic aRG by performing live imaging on E14.5 organotypic slices prepared from control or Pax6-expressing plasmid-electroporated brains. The time period between two successive aRG mitoses was taken to indicate the length of the cell cycle, Tc. In both control and conditional Pax6 expression, we observed no major difference in Tc, although there was a trend for a shorter Tc upon conditional Pax6 expression (control, 21.0 ± 3.3 h, n = 8 cells versus Pax6, 18.5 ± 1.2 h, n = 9 cells, S1 Table top).
To estimate the proportion of the progeny of control-plasmid-and Pax6-expressing-plasmid-electroporated neurogenic APs that were in S-phase, we performed pulse-labeling with the thymidine analog EdU one hour before analyzing the embryos at E14.5. This revealed that a significantly greater proportion of the exoPax6-expressing progeny than of the control progeny was in S-phase, in both the VZ and SVZ (Fig 3A-3C). Given that conditional Pax6 expression did not increase the population size of cycling APs (Fig 2K), nor alter much their Tc  Table top), the increase in the proportion of cells in S-phase in the VZ (Fig 3C) likely reflected a greater share of S-phase in the AP cell cycle, rather than an increase in cycling APs as such.
To address this directly, we performed a dual pulse chase experiment as previously described [61]  We further corroborated this by analyzing the pattern of immunofluorescence of the cycling cell marker proliferating cell nuclear antigen (PCNA). Like other cycling cells, cortical progenitors in S-phase show a punctate nuclear PCNA pattern, whereas progenitors in G1 and G2 show diffuse nuclear PCNA immunoreactivity [23,52,62]. Based on punctate PCNA staining, we observed a proportion of neurogenic APs in S-phase upon control electroporation that was similar to previously published data on E14.5 Tis21-positive APs [52] (S1 Table middle). Conditional Pax6 expression, however, was found to significantly increase the percentage of PCNA-positive nuclei in the VZ that showed a punctate pattern (Fig 3D-3F), i.e., increased the proportion of neurogenic APs that were in S-phase. These findings, together with the Ki67 ( Fig 2K) and EdU ( Fig 3C) data, imply that conditional Pax6 expression increases the relative proportion of S-phase within the AP cell cycle.
As there was no significant difference in Tc but an increase in the proportion of cells in Sphase upon conditional Pax6 expression in Tis21-positive APs, we hypothesized that the G1-phase must have been shortened to compensate for the longer S-phase. Consistent with this hypothesis, a significantly smaller proportion of the exoPax6-expressing progeny in the VZ than of the control progeny of electroporated neurogenic APs was positive for cyclin D1, a cyclin that is expressed from mid-to late-G1 (Fig 3G-3I). To estimate the length of the G1-phase, we combined the data obtained from live imaging with the punctate PCNA staining data (S1 Table bottom). As none of the apical mitoses observed lasted for >1 h and no difference in G2 length was reported between neural progenitors [52], we assumed that the proportion of neurogenic aRG in G2-and M-phase remained unchanged upon conditional Pax6 expression. Similar to the data obtained for cyclin D1 (Fig 3I), we estimated a shorter G1-phase upon conditional Pax6 expression (control 15.6 h versus Pax6 12.8 h, S1 Table bottom).
As to BPs, the increase in the proportion of EdU-positive cells in the SVZ upon conditional Pax6 expression (Fig 3C) was consistent with that of Ki67-positive cells (Fig 2K), corroborating our conclusion that the population size of cycling BPs derived from electroporated aRG was increased under this condition. Further support for this population size increase was provided by immunofluorescence for phosphohistone H3, a marker of cells in late G2-and M-phase, which revealed a significant increase in mitotic BPs derived from electroporated aRG (Fig 3J-3L). Also in the case of BPs, conditional Pax6 expression significantly increased the relative proportion of S-phase within the cell cycle as revealed by the pattern of nuclear PCNA immunoreactivity (Fig 3D-3F), albeit not at the expense of decreasing the relative proportion of G1 ( Fig 3I).

Conditional Pax6 Expression Induces Tis21-Expressing aRG to Increasingly Generate BPs with Radial Glia Characteristics
Our group previously reported a difference in S-phase length between Tis21-positive and Tis21-negative APs [52]. As Tis21-negative and Tis21-positive APs differ in the type of division (symmetric versus asymmetric) and progeny produced (APs versus BPs) [16,55], we wondered whether the increase in the relative proportion of S-phase within the cell cycle of the exoPax6expressing APs (Fig 3F) may be indicative of an alteration in their mode of division.
To explore this possibility, we investigated the nature of the cycling BPs that were increasingly observed upon conditional Pax6 expression (Fig 2I-2K) by examining the expression of two characteristic transcription factors, Tbr2 (Fig 4A-4C) and Sox2 (Fig 4D-4F). Tbr2 is typically expressed by the differentiating progeny of Tis21-expressing aRG fated to become bIPs [22,51,52,63], whereas Sox2 expression is characteristic of aRG and bRG [23,24,26,28,29,31,64]. Upon conditional Pax6 expression, analysis for the abundance of Tbr2-positive cells revealed a significant reduction in the exoPax6-expressing progeny as compared to control (Fig 4A-4C). This reduction was largely accounted for by the decrease in Tbr2-positive cells in the SVZ, most of which presumably were bIPs ( Fig 4C). Conversely, the abundance of Sox2-positive cells was higher in the exoPax6-expressing progeny as compared to the control (Fig 4D-4F). Remarkably, this increase occurred in the SVZ rather than the VZ (Fig 4F). This suggested that conditional Pax6 expression, which increased the population of BPs (Fig 2K), induced Tis21-expressing aRG to increasingly generate BPs with a radial gliacharacteristic transcription factor expression (i.e., bRG), at the expense (at least relatively) of bIP production.
To directly investigate a possible effect of conditional Pax6 expression on the mode of cell division of neurogenic APs, we performed a daughter cell pair assay [65] by analyzing areas of dorsolateral telencephalon that contained only a few RFP-positive cells in the VZ 24 h after electroporation. Tbr2 immunofluorescence allowed us to distinguish three types of RFP + daughter cell pairs: (1) Tbr2-/Tbr2-(no bIP daughter cells), (2) Tbr2+/Tbr2-(1 bIP daughter cell) and (3) Tbr2+/Tbr2+ (2 bIP daughter cells) (Fig 4G). Importantly, virtually all Tbr2daughter cells in the VZ are likely to be radial glia, either aRG or newborn bRG, based on the following considerations. Essentially all daughter cell nuclei in the VZ were PCNA-positive (S7 Fig). This was in line with the findings that >80% and almost 90% of the progeny in the VZ that was derived from electroporated neurogenic APs were Ki67+ ( Fig 2K) and Sox2+ (Fig 4F), respectively. Hence, the Tbr2-daughter cells were radial glial progenitors rather than neurons. Consistent with this, almost all cells in the mouse E14.5 VZ are cycling [52], and very few of them are newborn neurons [52].
Quantification of daughter cell pairs in the VZ showed that in the control, the majority (77%) of these pairs derived from AP divisions that had generated bIPs. Specifically, 56% of divisions were asymmetric (and presumably self-renewing) (Tbr2+/Tbr2-, Fig 4H, red), and 21% symmetric self-consuming (Tbr2+/Tbr2+, Fig 4H, green). These findings were in line with the fact that the progeny specifically of neurogenic APs was analyzed. Of note, only 23% of divisions did not generate any bIPs and hence were presumably symmetric proliferative with regard to the radial glia nature of the daughter cells (Tbr2-/Tbr2-, Fig 4H, blue). In contrast, upon conditional Pax6 expression, the majority (59%) of the daughter cell pairs were derived from neurogenic AP divisions that did not generate bIPs but radial glia (Tbr2-/Tbr2-, Fig 4H, blue). This occurred at the expense of bIP-generating divisions, that is, asymmetric self-renewing divisions (Tbr2+/Tbr2-, reduced to 32%, Fig 4H, red), and symmetric self-consuming divisions (Tbr2+/Tbr2+, reduced to 8%, Fig 4H, green).
The observations that conditional Pax6 expression increased (i) the non-bIP generating divisions (Tbr2-/Tbr2-, Fig 4H, blue) and (ii) the Sox2-positive progeny in the SVZ ( Fig 4F) suggested that the former progeny increasingly consisted of newborn bRG. As bRG are known to delaminate from the ventricular surface [24][25][26][27]31,35], we explored whether the radial glia progeny in the VZ observed upon conditional Pax6 expression increasingly showed signs of delamination. To this end, we measured the distance of the ventricular-most nucleus of each Tbr2-/Tbr2-daughter cell pair from the ventricular surface ( Fig 4I). In light of the observation that the mean distance of the ventricular-most nucleus of the control and exoPax6-expressing Tbr2+/Tbr2-and Tbr2+/Tbr2+ daughter cell pairs was always >40 μm (S8 Fig), whereas that of the Tbr2-/Tbr2-pairs was <26.5 μm (Fig 4I, S8 Fig), we focused our attention on the abundance of the ventricular-most nuclei of Tbr2-/Tbr2-daughter cell pairs with a distance from the ventricular surface of !27 μm (corresponding to >3 nuclear diameters and referred to as abventricular location [16]). Whereas only 1 of the 7 ventricular-most nuclei (14%) of the Tbr2-/Tbr2-daughter cell pairs in the control was found in an abventricular location, 7 of the 15 nuclei (47%) analyzed upon conditional Pax6 expression were abventricular ( Fig 4I). This suggested that conditional Pax6 expression promoted a substantial proportion of the radial glia progeny derived from neurogenic AP divisions to delaminate from the ventricular surface, as would be expected for newborn bRG.
In species with a high abundance of bRG in the SVZ, the radial thickness of the VZ decreases concomitant with bRG generation [8,23,[25][26][27]64]. In light of the findings described above, we investigated a possible reduction in VZ thickness upon conditional Pax6 expression by quantifying the total number of nuclei (both RFP-and RFP+) in the VZ within a 200-μm wide, electroporated region of the dorsolateral telencephalon. Indeed, we observed a significant, approximately 10%, reduction in the number of nuclei in the VZ upon conditional Pax6 expression ( Fig 4J). The magnitude of this reduction was consistent with the efficiency of electroporation and the estimated increase in the proportion of the progeny of electroporated neurogenic APs that delaminated upon conditional Pax6 expression as compared to control ( Fig  4H). Taken together, the findings presented so far strongly suggest that mouse neurogenic APs and their progeny that constitutively express Pax6 increasingly generate bRG at the expense of generating bIPs.

Conditional Pax6 Expression Increases bRG
To corroborate and complement these findings, we next investigated the effect of conditional Pax6 expression on the proportion of bRG in the BP progeny of electroporated aRG. To this end, we analyzed the morphology of mitotic BPs using phosphovimentin immunofluorescence ( Fig 5A-5C), which stains both the cell bodies and processes of mitotic cortical progenitors [66]. bRG characteristically extend basally and/or apically directed processes [23][24][25][26][27][28][29]31,35], whereas bIPs do not [17,18,[21][22][23][24][25][26]28,35]. As the apically directed processes have been reported to be thinner than basal processes and may not be easily detected via phosphovimentin staining [23], we focused our analysis on basal process-bearing mitotic BPs. In the control, the vast majority (91%) of mitotic BPs were nonpolar and only a small minority (9%) extended a basal process (Fig 5C), consistent with the high abundance of bIPs and low abundance of bRG in the embryonic mouse SVZ [28,29,35]. In contrast, upon conditional Pax6 expression, we observed a more than 2-fold increase in the proportion of mitotic BPs with a basal process, i.e. of bRG within the BP population (23%, Fig 5C). These data show that, concomitant with the increase in the proportion of BPs among the aRG progeny ( Fig 2K), conditional Pax6 expression more than doubled the proportion of bRG within these BPs.
As the apically-directed process of bRGs may be harder to detect via phosphovimentin immunofluorescence at mitosis [23], we next investigated the diversity of bRG morphology during interphase. To do this, we made use of the residual membrane-GFP (Fig 2A) expressed presumably due to incomplete Cre recombination (see Materials and Methods, live imaging) (Fig 5D and 5E). To distinguish bRG from migrating neurons, we stained for Sox2, which is expressed in radial glia but not in neurons. In the control, all of the bRG progeny of the electroporated neurogenic APs exhibited a basal process, and 40% of them an apically-directed process as well ( Fig 5F). Upon conditional Pax6 expression, we found an increase in the proportion of bRG exhibiting both basally and apically directed processes (Fig 5E and 5F, 53%) and also observed bRG with an apically directed process only (Fig 5D and 5F, 7%). Interestingly, in the macaque, bRG with both processes and bRG with an apically directed process only have been reported to have a higher self-renewing capacity as compared to bRG with a basal process only [23]. Of note, the basal process of the bRG generated upon conditional Pax6 expression sometimes extended all the way to the pia (S9A Fig

Lack of Pax6 Decreases bRG Abundance
The data presented so far show increased bRG generation upon elevating Pax6 levels in neurogenic aRG and sustaining it in the BPs derived therefrom. We sought to complement these findings by a converse, loss-of-function, approach. To this end, we investigated the proportion of mitotic (phosphovimentin-positive) bRG among BPs in the dorsolateral telencephalon of E14.5 homozygous small eye (Sey) mutant mice, which lack functional Pax6 because of a mutation that generates a premature translational stop codon (Fig 5G-5I). We found a significant reduction in the percentage of bRG as compared to littermates that have at least one copy of the Pax6 gene ( Fig 5I). These data indicate that although Pax6 function is not absolutely required for bRG generation, its level of expression is crucial for determining the abundance of these cells in the developing mouse neocortex.

Conditional Pax6 Expression Increases Cell Cycle Re-entry of BPs
Ferret and primate bRG are known to undergo multiple rounds of self-renewing division [23][24][25][26]31], whereas bIPs in mouse and rat embryonic neocortex typically undergo one round of self-consuming division [16][17][18][20][21][22]. In light of the increase in cycling BPs ( Fig 2K) and bRG ( Fig 5C) upon conditional Pax6 expression, it was therefore of interest to investigate whether conditional Pax6 expression would subsequently lead to increased cell cycle re-entry of the BP progeny derived from electroporated aRG. To this end, a single pulse of EdU was administered at 24 h after electroporation and analyzed after an additional 24 h for the proportion of cycling, Ki67-positive cells among the EdU-labeled progeny of electroporated APs, in order to identify cells that had re-entered the cell cycle ( Fig 6A).
In the control, 75% of such daughter cells present in the VZ, but only 23% of such daughter cells in the SVZ, had re-entered the cell cycle (Fig 6B and 6D). In contrast, upon conditional Pax6 expression, whereas daughter cell cycle re-entry was the same in the VZ, it nearly doubled in the SVZ (41%, Fig 6C and 6D). Again, we used the residual membrane-GFP fluorescence to determine the morphology of daughter cells that had re-entered the cell cycle. Two types of such daughter cells were observed, monopolar cells with a distinct basal process (Fig 6E), i.e., bRG, and multipolar cells with short extensions during interphase (Fig 6F), presumably bIPs.

Live Imaging of the Fate of the Progeny Arising from Divisions of bRG
We extended these data by analyzing the fate of the progeny derived from divisions of bRG, using live time-lapse imaging for at least 48 h of organotypic slices prepared from control and Pax6-expressing plasmid-electroporated E14.5 neocortex (Fig 7A). Despite the rare occurrence of bRG in mouse neocortex, we were able to identify several RFP-positive bRG, to image their divisions, and to track their progeny for at least an additional 20 h (i.e., for a time period longer  S2 Table)). In the control, in two out of the seven cases analyzed, the mitotic bRG underwent an asymmetric self-renewing division, as one of the daughter cells was observed to re-enter the cell cycle (S10 Fig). In the other five cases, similar to what has been previously reported for the embryonic mouse brain [28], both daughters did not enter mitosis during the time of our observations (S10 Fig).
Upon conditional Pax6 expression, half of the mitotic exoPax6-expressing bRG (three out of six) gave rise to progeny that subsequently underwent another round of cell division (S10 Fig). In two cases, these bRG divisions were asymmetric self-renewing (S10 and S11 Figs, S1 Movie). Remarkably, we also observed a bRG undergoing a symmetric proliferative division (Fig 7, S10 Fig, S2 Movie), with both daughters undergoing another round of cell division. These live imaging data are consistent with the notion that bRG generated upon conditional Pax6 expression and their progeny are endowed with greater proliferative potential as compared to control. Moreover, together with the cell cycle re-entry analysis (Fig 6), these data suggest that BPs show an increased proliferative capacity upon conditional Pax6 expression.

Conditional Pax6 Expression Increases Oblique Cleavage Plane Orientation of Neurogenic APs
It has been reported that a nonvertical (i.e., oblique or horizontal) cleavage plane orientation in relation to the ventricular surface of dividing APs (for examples, see Fig 8A) increases the probability that daughter cells become bRG [24,31,35,64]. We investigated whether the increased generation of bRG upon conditional Pax6 expression involved such alterations in cleavage plane orientation. In the control, the vast majority (91%) of mitotic neurogenic APs showed a vertical, and only a small minority (9%) an oblique, cleavage plane orientation (Fig 8B), consistent with previous data on Tis21-expressing APs [21,67,68]. Strikingly, conditional Pax6 expression resulted in a significant increase in nonvertical cleavage planes in mitotic neurogenic APs (19%, Fig 8B). As this doubling matched the doubling in the proportion of bRG among BPs (Fig 5C), our observations suggest that the increase in nonvertical cleavage plane orientations of neurogenic APs upon conditional Pax6 expression (Fig 8B) causally contributed to the increased generation of bRG (Fig 4F and 4H, Fig 5C).

Conditional Pax6 Expression Increases Horizontal Cleavage Plane Orientation of BPs
The doubling in cell cycle re-entry of BPs upon conditional Pax6 expression (Fig 6D) matched the doubling of bRG (Fig 5C), which in primates are endowed with constitutive cell cycle reentry capacity [23,24,26]. However, the morphology of the BPs that had re-entered the cell cycle (Fig 6E and 6F, Fig 7) raised the possibility that the increased cell cycle re-entry of BPs upon conditional Pax6 expression (Fig 6D) may not only be due to the increase in the proportion of bRG (Fig 5C) but may in addition reflect an increased cell cycle re-entry of bIPs (Fig 6F,  Fig 7). Moreover, inducing mouse Tbr2-positive BPs to re-enter the cell cycle by Sustained Pax6 Expression Generates Primate-like bRG premature expression of the transcription factor Insm1 has been shown to be associated with an alteration in their cleavage planes from the normal near-random [21,24,63] to mostly horizontal orientations [63]. Finally, not only human bRG are thought to divide preferentially with a near-horizontal cleavage plane [24] but also Tbr2-positive progenitors in the human SVZ, which, like their macaque counterpart [23] and in contrast to mouse bIPs, are endowed with proliferative capacity [26] and show a near-horizontal cleavage plane orientation in the Membrane-GFP fluorescence, single optical sections. 00:00 (hh:mm) denotes the start of mitosis. Yellow arrowheads, mother bRG; white arrowheads, basal process; green and red arrowheads, bIP daughter 1 and bIP daughter 2, respectively, of mother bRG; green and red arrows, daughter cells of bIP daughter 1 and bIP daughter 2, respectively.   [25]. These considerations prompted us to investigate whether conditional Pax6 expression, concomitant with increasing the cell cycle re-entry of the mouse BPs derived from electroporated aRG, would perhaps increase the proportion of horizontal cleavages of these BPs (for examples of vertical, oblique, and horizontal BP cleavage planes in relation to the ventricular surface, see Fig 8C).
In the control, the BP progeny derived from neurogenic aRG showed a random cleavage plane orientation at mitosis (Fig 8D), consistent with previously published data [21,63]. Interestingly, conditional Pax6 expression caused an increase (albeit not statistically significant) in the proportion of the BP progeny that divided with a horizontal cleavage plane, decreasing the proportion of oblique cleavage planes (Fig 8D). Given that conditional Pax6 expression increased not only the proportion of bRG among the BP progeny derived from electroporated neurogenic aRG (Fig 5C) but also the proliferative capacity of this progeny in general (Fig 2K,  Fig 6D), our cleavage plane data are consistent with the notion that a horizontal cleavage plane may be a hallmark of BPs endowed with self-renewal capacity, that is, bRG and proliferative bIPs [24,31,63].

Conditional Pax6 Expression Increases Upper-Layer Neuron Generation at Midneurogenesis
As conditional Pax6 expression increased the proliferative capacity of the BP progeny of neurogenic aRG and the proportion of bRG among these BPs, we finally investigated the consequences for cortical neurogenesis. To this end, we administered EdU 10 h after electroporation, at %E14.0, i.e., around the start of exo-Pax6 expression, in order to label the neuronal progeny born at this midneurogenesis stage, followed by their analysis in the cortical wall at E17.5 ( Fig 9A).
We first quantified the population size of the total progeny at E17.5. Compared to E14.5, this progeny population size was increased in both the control (1.6-fold) and upon conditional Pax6 expression (2.1-fold), with the latter increase being significantly greater than the former (Fig 9D and 9E). This indicated that conditional Pax6 expression increased the total cell output observed at E17.5.
Of note, in the control, the majority (68%) of the progeny had migrated to the cortical plate (S12A and S12D Fig). In contrast, in the case of conditional Pax6 expression, this was observed for only approximately one third (31%) of the progeny, the majority of which exhibited heterotopia in the intermediate zone (S12B and S12D Fig). Strikingly, most of the heterotopia cells had a much higher level of Pax6 immunoreactivity than those that had reached the cortical plate (S12C and S12E Fig). These observations are consistent with previous findings in Pax6-overexpressing mouse models, in which aggregates of Pax6-overexpressing cells in the developing cortical wall have been described [58]. Further characterization of the progeny exhibiting heterotopia showed that these cells were immature neurons (S13 Fig). Next, we analyzed the neuronal fate of the progeny that had migrated to the cortical plate. To distinguish between deep-layer and upper-layer neurons, we made use of established markers, the transcription factor Tbr1, which labels layer V and VI neurons, and the transcriptional regulators Satb2 and Brn2, which label layer II-IV neurons [69]. Conditional Pax6   (Fig 9B, 9C and 9F). Conversely, the proportion of EdU-labeled neurons that expressed Satb2 was increased (Fig 9G-9I). Similarly, the percentage of Brn2-positive cells among the neuronal progeny was significantly increased upon conditional Pax6 expression (Fig 9J). Together with the overall increase in progeny observed at E17.5 (Fig 9D  and 9E), these data show that conditional Pax6 expression at midneurogenesis increases the generation of upper-layer neurons. This likely reflected the increase in BP proliferative capacity (Fig 2K, Fig 6D, Fig 7, S10 Fig) and relative bRG abundance (Fig 4F, Fig 5C) upon conditional Pax6 expression.

Increased Upper-Layer Neuron Production and Occurrence of Progenitors Residing in the Intermediate Zone of Neocortex of Transgenic Mouse Embryos Sustaining Pax6 Expression in the Neurogenic Lineage
We complemented and extended the data obtained by the conditional Pax6 expression using in utero electroporation by taking a double-transgenic approach. Specifically, we crossed the Tis21-CreER T2 mice with JoP6 mice [58]. Like the Pax6-expressing plasmid, JoP6 mice contain, under a constitutive promoter, a floxed GFP-stop cassette followed by Pax6, an IRES sequence and a reporter [58]. Upon Cre recombination induced by tamoxifen administration at E13.5 (Fig 10A), Pax6 will be expressed at elevated levels in neurogenic aRG, and this expression sustained in their progeny throughout the embryonic neocortex.
With this approach, similar to the results obtained upon the conditional Pax6 expression via in utero electroporation of Tis21-CreER T2 mouse embryos (Fig 9), upper-layer neurons as identified by Satb2 and Brn2 expression were significantly increased when compared to control littermates (Fig 10D and 10E). By contrast, deep-layer neurons as identified by Tbr1 expression were not affected ( Fig 10F). Strikingly, with this double-transgenic approach (Fig 10A), we did not observe the heterotopia seen upon conditional Pax6 expression using in utero electroporation (S12B Fig). This may reflect the more standardized way in which elevated and sustained Pax6 expression is achieved in the double-transgenic embryos.
In ferret and primate neocortex, the increase in proliferating BPs is accompanied by an expansion of the SVZ in the basal direction, that is, an increase in BPs residing in the oSVZ, the key basal-most germinal layer characterized by lesser cell density [4][5][6]8,32]. To explore whether the Tis21-CreER T2 : JoP6 double-transgenic approach resulted in an increase in BPs residing in cortical low-cell-density layers basal to the mouse SVZ proper, that is, the intermediate zone and subplate, we examined the distribution of the cell proliferation marker Ki67. Whereas there was no significant difference in the abundance of Ki67-positive cells between control (Tis21-CreER T2 +/-: JoP6-/-) and Pax6-overexpressing (Tis21-CreER T2 +/-: JoP6+/-) neocortex in the VZ, we observed a significant increase in Ki67-positive cells not only in the SVZ, but also in the intermediate zone and subplate of Pax6-overexpressing mouse neocortex (Fig 10B, 10C and 10G). It is therefore interesting to note that the increase in progenitors residing in cortical low-cell-density layers basal to the SVZ of embryonic mouse neocortex observed fold-increase at E17.5 as compared to E14.5 (E), upon control (Con, white) and Pax6 (black) electroporation. (D) E14.5, mean of three independent experiments, each being the average of two or three embryos (eight in total); E17.5, mean of eight embryos from at least two independent experiments; (E) mean of eight E17. 5

A Novel Approach to Conditionally Express Pax6 Selectively in BP-Genic aRG and Their Progeny
BPs endowed with proliferative capacity, notably bRG, are a hallmark of the developing primate neocortex [1][2][3][4][5][6]8,23,34,70]. Here we show that a single transcription factor, Pax6, when specifically sustained in the aRG-to-BP lineage, is sufficient to generate such BPs (Fig 11). Our study differs in key aspects from previous studies in which Pax6 expression was increased in APs of dorsolateral telencephalon, as the latter either did not observe or address effects on BPs [48,58,71], or obtained opposite results [46]. Specifically, increased Pax6 expression was previously found to increase the mRNA levels for the bIP marker Tbr2 in the VZ and SVZ [46]. In contrast, in the present study, we found a decrease of Tbr2-positive BPs in the SVZ and nascent BPs in the VZ upon conditional Pax6 expression. Our observations are consistent with the increased generation of proliferative BPs, notably bRG. These differences in results presumably reflect the fact that in the previous study, Pax6 expression was increased in all APs, whereas in the present study, conditional Pax6 expression was confined to Tis21-positive, that is, neurogenic and BP-genic, APs and their progeny.

Increasing Pax6 Expression in aRG Is Sufficient to Alter Their Cleavage Plane Orientation Towards Promoting bRG-Generating Divisions
Our findings have three significant implications for elucidating the evolutionary expansion of the neocortex. First, they reveal a key role of Pax6 in the generation of a primate-like SVZ, that is, of proliferative BPs from aRG. We find that the effects elicited by increased Pax6 levels on aRG mitosis and daughter cell fate in embryonic mouse neocortex reproduce the normal situation in fetal human neocortex, which is characterized by higher Pax6 levels in human than mouse aRG (S14 Fig). Specifically, increasing Pax6 levels in mouse BP-genic aRG increased their oblique cleavage plane orientation at the expense of vertical cleavage plane orientation (Fig 11), consistent with previous studies reporting a greater proportion of oblique and horizontal cleavages in human [24] than mouse [19,35,67,68,72,73] aRG. This alteration in cleavage plane orientation may well have been promoted by the fact that conditional Pax6 expression was selective for BP-genic aRG, which are more susceptible to spindle orientation variability due to the reduction of apical and basal astral microtubules as compared to proliferative aRG [67]. The increased oblique cleavage plane orientation of BP-genic aRG likely caused, in line with previous findings [35], the observed increase in (i) self-consuming bRG-genic divisions of mouse aRG at the expense of self-renewing bIP-genic divisions and (ii) aRG daughter cell delamination, and consequently (iii) the decrease in mouse VZ thickness. Taken together, our findings provide a mechanistic explanation for the reduction in VZ thickness that occurs concomitant with the growth of the oSVZ during the progression of cortical neurogenesis in species with an enlarged neocortex [8,23,[25][26][27]32,64]. As to the mechanism how increased Pax6 levels in BP-genic aRG promote oblique cleavage plane orientation, previous work has identified an intriguing Pax6 target gene-the mitotic spindle-and kinetochore-associated protein Spag5 [47]. An increase in nonvertical cleavage plane orientation of mouse aRG has been observed both upon knock-down of Spag5 and when Spag5 mRNA and protein levels in Pax6 mutant mice at midneurogenesis are elevated [47], suggesting that either too low or too high Spag5 levels perturb the normal, horizontal spindle orientation that is required for aRG vertical cleavage plane orientation. This is in line with the concept that for aRG, a horizontal spindle orientation is thought to reflect the active state of the mitotic spindle orientation machinery, and nonhorizontal spindle orientations can occur upon perturbation of this machinery [19,35,67,72,73].
By contrast, for mouse BPs, a default state of the mitotic spindle orientation machinery, with random cleavage plane orientations, is thought to be the normal situation [21,24,63], and activation of this machinery is thought to promote horizontal cleavage plane orientation. Such orientation prevails in primate BPs, notably bRG, which are endowed with much greater proliferative capacity than mouse BPs [23][24][25][26] and is increasingly observed when mouse BPs are induced to proliferate [63]. In this context, it is interesting to note that (i) the relative Spag5 mRNA levels are much higher in the human iSVZ and oSVZ than the mouse SVZ [50], and (ii) increasing the Pax6 level in mouse BPs, which likely results in increased Spag5 levels, was found here to increase their horizontal cleavage plane orientation. Taken together, the concept emerges that Pax6, via its downstream targets including Spag5, increases oblique, self-consuming aRG divisions generating proliferative BPs, notably bRG, and horizontal BP divisions promoting their proliferation or self-renewal.

Sustaining Pax6 Expression in BPs Is Sufficient to Promote Their Proliferation/Self-Renewal
Second, we observed that sustaining high Pax6 expression in BPs increases their cell cycle re-entry (Fig 6) and their abundance not only in the SVZ but even in the layers basal to the SVZ, the intermediate zone and subplate (Fig 10). This finding implies that the maintenance of expression of Pax6 in primate, but not mouse and rat, BPs, notably bRG, is a key feature of the machinery underlying their greater proliferative or self-renewal capacity [23][24][25][26]. It thus appears that Pax6 has the potential to promote proliferation and self-renewal of cortical progenitors in general, that is, for both APs [38,39,74] and BPs (this study). Conversely, as we observed a marked decrease in bRG in the dorsolateral telencephalon of Sey mouse embryos (Fig 5G-5I), it would be interesting to explore whether a similar decrease in bRG is observed in human embryonic stem cell-derived organoids [75] upon PAX6 knockdown after establishment of the SVZ. As a corollary, the molecular mechanisms underlying the sustained Pax6 expression in BPs, at the level of mRNA and protein generation and stability [76][77][78][79][80][81][82][83], then become the crucial issue for SVZ enlargement and neocortex expansion.
The increased cell cycle re-entry of BPs observed here upon sustained Pax6 expression is in contrast to the previously reported increase in cell cycle exit of cortical progenitors in Pax6 overexpressing (PAX77) mice [48]. Again, this discrepancy presumably reflects the difference between conditional Pax6 overexpression selectively in BP-genic APs (present study) and constitutive Pax6 overexpression in all APs [48].
As to the downstream targets of Pax6 that promote BP proliferation or self-renewal, at least two candidates exist. One is the transcription factor Sox2, a well-known stimulator of stem and progenitor cell proliferation and self-renewal [38,59,84]. Pax6 has been shown to induce Sox2 expression [85], and consistent with this, we observed that sustaining Pax6 expression in the aRG-BP lineage indeed increases the proportion of Sox2-positive BPs. The other class of candidates are extracellular matrix (ECM) constituents and their receptors, the integrins, which have been implicated in BP proliferation and self-renewal [25,50,52,62,86]. Interestingly, Pax6 induces the expression of ECM constituents such as tenascin-C [87] and integrin α 5 β 1 [88]. Hence, the increased cell cycle re-entry of BPs upon sustained Pax6 expression may well reflect, at least in part, an altered, more human-like, microenvironment in the mouse SVZ that is now more conducive to BP proliferation and self-renewal.

Enhancing the Proliferative Capacity of BPs by Sustained Pax6 Expression Increases Upper-Layer Neuron Generation
Third, the increased proliferative capacity of mouse BPs achieved by sustained Pax6 expression resulted in a phenotypic change in the cortical plate that is characteristic of primates-an increase in upper-layer neurons (Figs 9 and 10). Also, this aspect of the present phenotype is in contrast to previous findings which showed, concomitant with increased progenitor cell cycle exit, an increase in deep-layer neurons at the expense of upper-layer neurons in the constitutively Pax6 overexpressing PAX77 mice [48]. It should be noted that conditional Pax6 expression in neurogenic aRG resulted in an increase in Pax6 levels that was substantially greater than that in human as compared to mouse APs (compare Fig 2G and S14 Fig). Moreover, upon the present approach of conditional expression, which used a constitutive promoter, Pax6 was found to be present even in neurons (S12 Fig). It is therefore comprehensible that the present approach of conditional Pax6 expression via in utero electroporation in embryonic mouse neocortex has phenotypic consequences, some of which are not observed in fetal human neocortex and upon more controlled Pax6 expression in the double-transgenic mouse (Fig 10), such as the heterotopia which consisted mostly of highly Pax6-positive immature neurons (S12 and S13 Figs).
Hence, considering all aspects of the present phenotype together, sustaining Pax6 expression in BP-genic aRG and the BPs derived therefrom, as is characteristically the case in fetal primate neocortex [23][24][25][26][27]50], is sufficient to induce primate-like progenitor behaviour in embryonic mouse neocortex, that is, (i) translocation of progenitors from the VZ to the SVZ, (ii) an increased proportion of bRG among the BPs generated, (iii) sustained proliferation or self-renewal of BPs in the SVZ, and (iv) an increased upper-layer neuron production. The differential regulation of Pax6 expression in cortical progenitors during development across mammals therefore emerges as a key issue of future studies aiming to understand the evolutionary expansion of the SVZ, and consequently the neocortex.
Although sustained Pax6 expression sufficed to generate primate-like bRG in developing mouse neocortex, it was insufficient to induce cortical folding (Figs 9 and 10). This is in contrast to previous studies in which the expression of specific genes implicated in neocortex expansion led not only to the expansion of BPs but also to folding of the mouse neocortex [89,90]. In these studies, the expansion of BPs comprised an increase in both bRG and bIPs. Expansion of bIPs alone has been reported to be insufficient to induce cortical folding in the mouse neocortex [91]. Moreover, the presence of bRG is essential for tangential dispersion of neurons [27] in order for the basal surface to expand more than the apical surface, and ultimately for cortical folding [4,5,8,27]. Hence, to increase the ratio of basal to apical surface, it appears to be critical to increase the proportion of bRG among the BPs in the SVZ above a certain level. This would increase the divergence of radial fibers emanating from the SVZ, allowing for a broader dispersion of migrating neurons. Our data suggest that a mere doubling of bRG abundance in the embryonic mouse neocortex (from 10% to 20% of all BPs), as was achieved by sustaining Pax6 expression, is still insufficient to result in cortical folding.
On a more general note, human-specific aspects of neocortex expansion can be considered to be caused by (i) the presence of a relevant gene in the human as well as nonhuman genomes, but with differential regulation of expression between human and nonhuman species [50,89,92]; (ii) the presence of a relevant gene in the human as well as nonhuman genomes, but with human-specific alterations in the coding sequence [93,94]; and (iii) the presence of a relevant gene in the human, but not nonhuman, genomes [90]. The present study demonstrates that Pax6, a central player in corticogenesis, can be regarded as a key example of the first scenario.

Materials and Methods Ethics
Human fetal brain tissue was obtained from the Klinik und Poliklinik für Frauenheilkunde und Geburtshilfe, Universitätsklinikum Carl Gustav Carus of the Technische Universität Dresden, following elective pregnancy termination and informed written maternal consents, and with approval of the local University Hospital Ethical Review Committees. All human fetal brain samples were anonymized. All animal experiments were performed in accordance with the German Animal Welfare legislation ("Tierschutzgesetz"). All procedures pertaining to animal experiments were approved by the Governmental IACUC ("Landesdirektion Sachsen") and overseen by the Institutional Animal Welfare Officer(s). Mice were anaesthetised using isofluorane during the in utero electroporation procedure. Mice were killed via cervical dislocation. The license numbers concerned by the present experiments with mice are: 24-9168.11-9/ 2009-2 (in utero work, tamoxifen, BrdU) and 24-9168.24-9/2012-1 (tissue collection without prior in vivo experimentation).

Mice
Mice were maintained in strict pathogen-free conditions in the animal facility of the Max Planck Institute of Molecular Cell Biology and Genetics, Dresden, Germany. To characterize the Tis21-CreER T2 mouse line described below, females were crossed with either the Tis21-GFP knock-in homozygous males [16] or the RCE:LoxP line [56]. To perform the in utero electroporation experiments with heterozygous Tis21-CreER T2 embryos as described below, homozogyous Tis21-CreER T2 males were crossed with wildtype C57BL/6JOlaHSD females. For the double-transgenic mice, homozogyous Tis21-CreER T2 mice were crossed with heterozygous JoP6 mice. To study the loss of function of Pax6 on bRG generation, heterozygous Sey mice were crossed with one another. The day of the vaginal plug was defined as E0.5. Tis21-GFP [16], hACTB-Flpe [95], RCE:LoxP [56], and JoP6 [58] mouse lines were genotyped as previously described. Offspring from the above crossings were genotyped for the Tis21-CreER T2 allele by PCR using standard procedure as described below.
Generation of the Tis21-CreER T2 Mouse Line Targeting construct. The bacterial artificial chromosome (BAC) bMQ-284G14 (%130 kb, 129S7/SvEv mouse strain, Sanger institute), containing the mouse Tis21 gene (%0.2 kb exon 1, %1.2 kb intron 1, %2.5 kb exon 2) flanked by %80 kb 5' and %31 kb 3' [65], was used to create the targeting construct p15A-Tis21-CreER T2 -neo (S1Ai Fig) for the generation of mouse embryonic stem (ES) cells in which the entire coding sequence of exon1 of one Tis21 allele (S1Aii Fig) was replaced by a CreER T2 cassette followed by a neomycin cassette flanked by FRT sites (S1Aiii Fig, referred to as Tis21-CreER T2 -neo). The targeting construct was cloned using the Red/ET homologous recombination technology [96]. At each of the cloning steps described below, plasmids were verified by sequencing.
First, towards obtaining the intermediate plasmid R6K-CreER T2 -neo, we generated two PCR products using PGK-gb2-neo (neomycin cassette, Genebridges) and R6K-ampicillin (Genebridges) as templates, (i) Tis21-PGK-gb2-neo, which contains a neomycin selection cassette flanked by two FRT sites; and (ii) Tis21-R6K-ori, which contains the R6K replication origin. For PCR amplification of Tis21-PGK-gb2-neo, we used the primer pair neo-forward (for all primer sequences, see below), consisting of a 35 bp sequence identical to the 3' end of the CreER T2 cassette, followed by a BamHI site, an FRT site, and 22 bp of the 5' flanking sequence of the neomycin cassette; and neo-reverse, consisting of 40 bp of the 5' flanking sequence of the Tis21 intron, followed by a HindIII and an EcoRI site, an FRT site, and 22 bp of the 3' flanking sequence of the neomycin cassette. For PCR amplification of Tis21-R6K-ori, we used the primer pair R6K-forward, consisting of an EcoRI and a HindIII site contiguous with a 40 bp sequence identical to the most 5' sequence of the Tis21 intron, followed by a PacI and an NdeI site, and 21 bp of the 5' flanking sequence of the R6K replication origin; and R6K-reverse, consisting of a 35 bp sequence identical to the 5' flanking sequence of the CreER T2 cassette, followed by a 39 bp sequence identical to the sequence 3' to the Tis21 translational start codon, an NdeI and a PacI site, and 21 bp of the 3' flanking sequence of the R6K replication origin.
Second, the BAC bMQ-284G14-Tis21-CreER T2 -neo was derived from the BAC bMQ-284G14 by replacing the entire coding sequence of the Tis21 exon1 with a CreER T2 cassette followed by a neomycin cassette flanked by FRT sites, carrying out a second round of homologous recombination as follows. The CreER T2 -neo fragment released from R6K-CreER T2 -neo by NdeI and PacI digestion was electroporated into E. coli strain DH10B containing the BAC bMQ-284G14 and pSC101-BAD-gbaA for homologous recombination, followed by kanamycin selection to obtain the BAC bMQ-284G14-Tis21-CreER T2 -neo.
Third, the final targeting construct, p15A-Tis21-CreER T2 -neo, was generated from the BAC bMQ-284G14-Tis21-CreER T2 -neo as follows. Using pACYC177 (New England Biolabs) as a template, we generated the PCR product p15A-ori-ampicillin, which consists of the p15A replication origin and the ampicillin selection cassette from pACYC177, and 40 bp each of Tis21 homologous sequence located %4.8 kb 5' and %7.4 kb 3' to the Tis21 translational start codon, respectively. For this purpose, we used the primer pair p15A-forward, consisting of 40 bp of the 5' Tis21 homologous sequence, followed by an NdeI and a PacI site, and 21 bp of 5' flanking sequence of the p15A replication origin; and p15A-reverse, consisting of 40 bp of the 3' Tis21 homologous sequence, followed by a PacI and an NdeI site, and 21 bp of 3' flanking sequence of the ampicillin selection cassette. p15A-ori-ampicillin was electroporated into E. coli strain DH10B containing the BAC bMQ-284G14-Tis21-CreER T2 -neo and the plasmid pSC101-BAD-gbaA for homologous recombination, followed by ampicillin selection to obtain p15A-Tis21-CreER T2 -neo (S1Ai Fig), which contains %12.2 kb genomic sequence harbouring the Tis21 locus with %4.8 kb 5' and %7.4 kb 3' to the Tis21 translational start codon.
Homologous recombination in embryonic stem cells. Homologous recombination of the targeting construct p15A-Tis21-CreER T2 -neo in mouse ES cells was performed essentially as described previously [65]. Southern blot analysis of individual clones was carried out as described [65], using antisense RNA probes for the 5' and 3' regions of Tis21 [65] and Cre [97] (S1Aii and S1Aiii Fig).

Primers
Primers for BAC modification.

Cell Culture
HEK293T cells were plated at 5x10 4 cells on 24-well plates and kept in culture in DMEM supplemented with 10% fetal calf serum. At 24 h after plating, cells were transfected, using 1 μl of Lipofectamine2000 (Invitrogen), with 250 ng of pCAGGs-Cre [101] and/or 250 ng of either control or Pax6-expressing plasmids diluted with serum-free DMEM. Cells were incubated for 48 h, followed by fixation with 4% paraformaldehyde in 120 mM phosphate buffer pH 7.4 for 10 mins. The paraformaldehyde was then removed and cells were kept in PBS until further processing.

In Utero Electroporation
In utero electroporation was carried out essentially as previously described [100,102]. Briefly, tamoxifen-treated pregnant dams carrying E13.5 embryos were anesthesized using isofluorane. Embryos were injected intraventricularly either with 0.5-3 mg ml -1 control or Pax6-expressing plasmid in PBS containing 0.25% Fast Green (Sigma) using a glass micropipette followed by electroporation (30 V, six 50-msec pulses with 1 sec intervals). Electroporated brains were dissected at the indicated developmental stages and fixed for 20-70 h at 4°C in 4% paraformaldehyde in phosphate buffer for further analysis.

Thymidine Analog Labeling
Single EdU pulses were administered by injecting 0.1 ml of 1 mg ml -1 EdU intraperitoneally into pregnant dams carrying embryos of the indicated developmental stages.
For the cell cycle re-entry experiments, we injected such a single pulse of EdU at E14.5 and sacrificed the animals 24 h later (Fig 6A). At this developmental stage, the length of S+G2+Mphase of cortical progenitors is 11 h [52], and a single EdU pulse is unlikely to be effective for >5 h [103]. Hence, the 24 h period between the EdU administration and analysis should be more than sufficient for essentially all cortical progenitors that incorporated EdU and that had been derived from electroporated aRG (i.e., that were RFP+) to go through M-phase, and thus for determining by Ki67 immunofluorescence whether or not the resulting daughter cells had re-entered the cell cycle.
For the dual pulse chase experiments, 0.1 ml of 1 mg ml -1 of IdU and BrdU were sequentially injected intraperitoneally into pregnant dams carrying embryos of the indicated developmental stages (S6A Fig). The length of S-phase was calculated as described previously [61].

Live Imaging
It has previously been shown that electroporation does not randomly target APs irrespective of the phase of the cell cycle, but preferentially targets APs in late S-, G2-and M-phase [104]. Conditional Pax6 expression upon electroporation would thus be confined to a synchronized cohort of progenitors, which precludes the use of cumulative labeling with a thymidine analog to determine the length of the cell cycle and its various phases. We therefore used live imaging to measure the cell cycle length of electroporated Tis21-positive aRG. In these analyses, we have exploited the fact that the RFP+ cells still contain residual membrane-GFP fluorescence (either by inheritance, or because not all plasmid copies electroporated into a given aRG underwent Cre recombination, or both).
AP divisions were defined as those occurring at the ventricular surface. The time period between two successive mitoses of the neurogenic aRG is taken to be the length of the cell cycle, Tc.
In addition, we used live imaging to track the fate of the bRG progeny and for the reconstruction of the RFP-positive bRG lineage tree. We defined bRG divisions as those occurring away from the ventricular surface (with no apical contact) and as BPs exhibiting a basally and/ or apically directed process just prior to, and often persisting through, mitosis. We included only RFP-positive bRG that had undergone division and tracked their progeny for at least an additional 20 h (i.e., for a time period longer than the average Tc of self-renewing bRG).

Immunofluorescence
For immunofluorescence of transfected cells [65], fixed cells were permeabilised with 0.3% Triton X-100 in PBS for 30 min and then quenched with 0.1 M glycine in PBS for 30 min. Cells were sequentially incubated with primary antibodies for 3 h followed by secondary antibodies for 1 h at room temperature. Coverslips were mounted onto glass slides using Mowiol.
For vibratome sectioning [105], fixed brains were embedded in 3% low-melting agarose. Sections (50-70 μm) were cut using a vibratome (Leica 1000) and were stored in PBS (maximally for 2 wk) until further processing. For cryosectioning [105], fixed brains were equilibrated in 30% (wt/vol) sucrose in PBS overnight at 4°C. Brains were embedded with Tissue-TEK (O.C.T, Sakura Finetek) and stored at −20°C. Brains were cryosectioned at 10-12 μm. Cryosections were rehydrated with PBS before further processing. Both vibratome and cryosections were subjected to an antigen retrieval protocol as follows. Unless indicated otherwise, sections were heated in 0.01 M citrate buffer pH 6.0 at 70°C for 1 h. For comparative quantification of Pax6 and phosphohistone H3 immunofluorescence levels in mouse and human mitotic APs, cryosections of paraformaldehyde-fixed embryonic mouse and fetal human neocortex were heated in the citrate buffer using a microwave oven at 800 W for 1 min followed by 140 W for 10 min. Sections were permeabilized using 0.3% Triton X-100 in PBS for 30 min and quenched with 0.1 M glycine for 30 min. Sections were then incubated with primary antibody overnight at 4°C, followed by secondary antibody for 1 h at room temperature in a solution of 0.2% gelatin, 300 mM NaCl, and 0.3% Triton X-100 in PBS. Floating sections were mounted to Superfrost Plus microscope slides (Thermo Scientific) using Mowiol (Merck Biosciences). For BrdU and IdU detection, slices were processed after RFP immunofluorescence as follows. An additional antigen retrieval step was performed by using HCl (2 N HCL, 30 min incubation at 37°C). Slices were then blocked with 10% goat serum and incubated for 3 h at room temperature followed by 1 h of secondary antibody incubation.
The following primary antibodies were used; ßIII-tubulin (Sigma, T8578 1:1,000). In case of thymidine analog-labeled samples (i.e., BrdU, EdU, and IdU), sections were postfixed with 4% paraformaldehyde for 20 min after the secondary antibody incubation. Incorporated EdU was detected using the Click-iT EdU kit with Alexa Fluor 647 (Invitrogen) as described previously [52].

Daughter Cell Pair Analysis
Vibratome sections (50-μm) and 12-μm cryosections were used for Tbr2 and PCNA analysis, respectively. We used similar criteria as used previously [65], with some modification. In summary, we examined only sparsely electroporated areas and defined two closely located RFPpositive cells as a pair of daughter cells derived from a single electroporated AP if (i) no other RFP-positive cells were observed within the distance of one cell body around the two cells in the z-stack; (ii) both cells exhibited the same RFP fluorescence intensity; and (iii) the two RFPpositive cells were aligned in the same radial axis and were located above one another. In the case of the Tbr2-/Tbr2-daughter cell pairs, we measured the distance of the center of the nucleus of the ventricular-most daughter cell from the ventricular surface in Fiji.

Image Acquisition
Fluorescence images were acquired using a Zeiss 700 confocal microscope using 25x and 63x objectives. Images were taken as either 2.1 μm (25x) or 0.9 μm (63x) single optical sections. All images used for scoring of parameters in control versus conditional Pax6 expression had comparable RFP fluorescence intensities. All images showing these parameters for control versus conditional Pax6 expression were acquired with the same settings during each microscope session. Images taken as tile scans were stitched together using the ZEN software (Zeiss). Quantifications were performed using Fiji. Whole-brain images were acquired with an Olympus SZX12 stereomicroscope.

Determination of Cleavage Plane Orientation
Cleavage plane orientation of electroporated mitotic APs and BPs was measured in 2-D based on the position of the DAPI-stained sister chromatids during late anaphase and was expressed relative to the ventricular surface. A cleavage plane parallel to the ventricular surface (i.e., horizontal cleavage plane) is defined as 0°.

Identification of Germinal Zones, Cell Counting, Quantification of Immunofluorescence Intensity, and Statistical Analysis
Germinal zones were identified based on their different histological characteristics. The VZ was identified as the ventricular-most layer of densely packed, radially aligned, elongated nuclei. The SVZ was identified as the layer basal to the VZ containing less densely packed, randomly orientated, rounded nuclei.
Unless specified otherwise, cells were counted in a rectangular area, 200-μm wide at the ventricular surface, within the electroporated region of the dorsolateral telencephalon. For quantifications using double-transgenic animals (Tis21-CreER T2 : JoP6), cells were counted in a rectangular area, 100-μm wide at the ventricular surface. Cells were counted without using pseudocolour. All quantifications were confined to RFP-positive cells only, with the exception of (i) the determination of the total nuclei present in the VZ (Fig 4J), and (ii) the analyses of the neocortex of the double-transgenic animals, (Tis21-CreER T2 : JoP6); in both cases, all DAPI-stained nuclei were quantified.
For quantification of immunofluorescence intensity levels, the area of the nucleus of interphase cells in VZ and SVZ was selected using the DAPI staining as a guide, and the area of the cell body of mitotic APs was selected using the phosphohistone H3 immunofluorescence as a guide. Selected areas were quantified using Fiji [106].
Data was further processed using the Prism software (GraphPad software). Student's t test was used to determine statistical significance. Steps carried out with ES cells. Using the targeting construct shown (i), the coding sequence in Exon1 of Tis21 (shaded box, ii) was replaced by a CreER T2 cassette (yellow box, i) followed by a neomycin cassette (neo, red box, i) flanked by FRT sites (green triangles, i), yielding the Tis21-CreER T2 -neo allele (iii). (iii-iv) Steps carried out with mice. Removal of the neomycin cassette by crossing mice carrying the Tis21-CreER T2 -neo allele (iii) with transgenic hACTB-FLPe mice expressing FLPe recombinase, yielding mice carrying the Tis21-CreER T2 allele (iv). (i-iv) Shaded boxes, Tis21 ORF; blue and red triangles, 5' and 3' ends, respectively, of homology between the targeting construct (i) and the Tis21 wildtype allele (ii); P1-P6, primers used for genotyping PCR; blue, yellow, and red bars, location of probes used for Southern blot analyses. Summary of the 13 bRG-derived lineage trees observed upon live time-lapse imaging of E14.5 organotypic slices prepared from dorsolateral telencephalon of tamoxifen-treated Tis21-CreER T2 heterozygous mice electroporated with control or Pax6-expressing plasmid (see Fig 7). Control, 7 bRG divisions; Pax6, 6 bRG divisions. Tc, total cell cycle length; red circles, bRG; yellow circles, bIPs; grey circles, no further mitosis detected for progeny until 20 h; white circles, unknown cell type. 00:00 (hh:mm) denotes the start of mitosis. Yellow arrowheads, mother bRG; white arrowheads, basal process; green and red arrowheads, bRG daughter and neuron daughter, respectively, of mother bRG; green arrows, daughter cells of bRG daughter. (TIF)