Figure 1.
Strategy to characterise the transcriptional networks regulated by Pax6 in neocortical stem cells.
(A) Location analysis was performed for Pax6 binding in the genome of neocortical stem and progenitor cells at embryonic day 12.5 (E12.5) to define those promoters bound by Pax6 in vivo. (B) To define the transcriptional responses downstream of Pax6, both direct and indirect, global gene expression was compared at E12.5 between wild-type and Pax6 loss-of-function (LOF) cortices (Sey/Sey) and also between wild-type and Pax6 gain-of-function (GOF) cortices (D6-Pax6). (C) Phenotypic analyses of the Pax6 gain- and loss-of-function cortices were used to both interpret the gene expression changes observed upon altering Pax6 dosage and also to test the hypotheses generated from the combination of the Pax6 binding and regulation data (D).
Figure 2.
In vivo location analysis identifies the set of genes whose promoters are bound by Pax6 in neocortical stem and progenitor cells.
(A–D) Identification of Pax6-bound promoters. ChIP array data for four typical Pax6-bound promoters, all of which were calculated as significantly bound above background levels (see Materials and Methods for details). Each graph plots fold-enrichment (in log base 2) on the y-axis, against relative chromosomal position on the x-axis. (E,F) Validation of the specificity and accuracy of the binding of the Pax6 promoter by Pax6, confirmed by qPCR in (G). Scanning PCR primers were designed to amplify DNA from a series of genomic regions in the Pax6 promoter (red bars represent amplified regions in the plot of fold-enrichment, y-axis, against relative chromosomal position, x-axis). The peak of Pax6 binding from the ChIP array analysis is shown in E. Pax6 binding was confirmed in this assay only in the region around the oligonucleotide, as shown in the gel images in F. Pax6, Pax6 ChIP; WCE, whole cell extract. (G) Validation of Pax6-binding to a set of promoters by quantitative PCR in independent ChIPs (not used for the array hybridisations) using two different Pax6 polyclonal antibodies. Enrichment of a region is calculated relative to that observed in control ChIPs without the primary antibody and normalised to a non-bound control region (NBCR). (H) Three independent ChIPs and promoter array hybridisations were carried out for Pax6, in which Pax6-bound material was compared with total genomic DNA (whole cell extract, WCE) to identify regions enriched in the Pax6-ChIP material by raw ratio measurements. Pax6 is found on the promoters of 1560 genes, 1172 (75%) of which are annotated genes (Table S1).
Figure 3.
Pax6-bound genes are enriched in genes required for nervous system development.
(A) Gene Ontology analysis for enrichment of categories of genes in the Pax6-bound set of genes. The set of Pax6-bound genes was compared to the whole genome to identify functional or biological sets over-represented in the Pax6-bound set, as described in the Materials and Methods. (B) Examples of genes in selected functional categories shown in A. (C,D) Cellular localisation of genes representative of some of the biological (C) and molecular (D) categories shown in (A), at E14.5 from the public in situ hybridisation database GenePaint [38]. The majority of genes are expressed in neocortical stem and progenitor cells in the ventricular zone (VZ) and basal progenitor cells in the subventricular zone (SVZ), with a significant minority also expressed in differentiating neurons in the cortical plate (CP). Scale bars, 500 µm.
Figure 4.
Expression analysis of Pax6 gain and loss-of-function cortices identifies sets of genes dependent on Pax6 for expression.
(A) Generation of transgenic mice over-expressing Pax6 specifically in the developing cerebral cortex, using the D6 promoter [35]. (B) Fluorescent immunohistochemistry for Pax6 in control and D6-Pax6 E12.5 cortices demonstrates the marked increase in Pax6 protein in the transgenic cortex. Scale bar, 100 µm. (C) Quantification of the intensity of anti-Pax6 immunofluorescence in WT and D6-Pax6 cortical ventricular zones, relative to signal in the basal ganglia in which the D6 transgene is not expressed. Error bars, s.e.m. (D) Microarray analysis of gene expression changes in the cortex of D6-Pax6 mice at E12.5. Gene expression in cortices from single E12.5 embryos was compared between mice carrying the D6-Pax6 transgene and control littermates in a set of 6 paired, dye-swapped array hybridisations. Paired hybridisations were averaged and genes showing reproducible, statistically significant (FDR<0.1) changes in expression were clustered to identify 1567 up-regulated and 2238 down-regulated transcripts (see Table S2 for details). (E) Microarray analysis of gene expression changes in the cortex of Pax6 mutant (Sey/Sey) mice at E12.5. Gene expression in cortices from single E12.5 embryos was compared between Sey/Sey mice and control littermates in a set of 6 dye-swapped array hybridisations. A set of 938 genes showing reproducible, statistically significant (FDR<0.1) changes in expression were clustered to identify 338 down-regulated and 600 up-regulated transcripts (see Table S3 for details). (F) A set of 298 genes was found to have altered expression in both the Sey/Sey and the D6-Pax6 E12.5 cortex to a statistically significant level. (G) From the total set of 4404 genes showing altered expression in one of more of the Sey/Sey and D6-Pax6 E12.5 cortices, the union of the sets of genes upregulated in the D6-Pax6 cortex and genes down-regulated in the Sey/Sey cortex defines a set of genes that are positively regulated by Pax6 (both directly and indirectly), denoted by the green shading. In contrast, the union of the sets of genes downregulated in the D6-Pax6 cortex and genes up-regulated in the Sey/Sey cortex defines a set of genes that are negatively regulated by Pax6 (both directly and indirectly), denoted by the red shading. (H) Examples of the genes found in each of the four categories defined in (C).
Figure 5.
The set of genes bound and regulated by Pax6 in the embryonic cerebral cortex.
(A) Identification of a set of 343 genes both bound and regulated by Pax6 by intersecting the three genomics datasets reported here: Pax6 ChIP-on-chip and the array analyses of Pax6 gain- and loss-of-function cortices. The intersection is highlighted in blue (see Table S4 for details). (B) Definition of the sets of genes bound and positively regulated by Pax6 (green shading) and negatively regulated by Pax6 (red shading). (C) Breakdown of the evidence for positive or negative regulation of genes by Pax6: 23 genes show regulation following both gain and loss of Pax6 function, 42 show regulation only upon loss of function and 278 show regulation only upon gain of function. (D–F) In situ hybridisation data for cellular expression of Pax6 bound and regulated genes in whole embryos at E14.5 was available for 279 genes (GenePaint), 56 of which were not detected at this age. Blind scoring of cellular distribution of expression patterns of the bound and regulated genes (D, E) found that the majority of genes were expressed by neocortical stem/progenitor cells (VZ), including substantial numbers of genes expressed in both the VZ and in differentiating neurons, as well as genes exclusively expressed in the VZ (E). Many Pax6 target genes were also expressed by basal progenitor cells in the SVZ. Examples of genes expressed in all three cell types (VZ, SVZ and CP) are shown in F. Scale bar, 500 µm.
Figure 6.
Pax6-regulated transcriptional networks in neocortical stem cells.
Pax6 directly regulates sets of genes that control four major processes in cortical neurogenesis: stem cell maintenance/self-renewal, neurogenesis, production of basal progenitor cells and cortical identity (black lines). These direct programmes are underpinned by indirect control of larger numbers of genes involved in each of these processes, as shown by the grey lines. Hes1 suppresses neurogenesis when continuously over-expressed in cortical stem/progenitor cells [51] and that is associated with repression (blue lines) of many of the same genes positively regulated by Pax6, including those genes required for genesis of basal progenitor cells.
Figure 7.
Increased expression of Pax6 specifically in the developing cortex alters neocortical stem cell self-renewal and cell fate determination.
(A) Genes characteristic of the three main cell types in the embryonic cortex (stem/progenitor cells in the ventricular zone; basal progenitor cells and differentiating neurons) are changed in expression in the D6-Pax6 cortex: there is a striking increase in a set of genes specifically expressed in basal progenitor cells (Eomes/Tbr2, Hes6, Tcfap2c), indicating that there is an increased production of this cell type when Pax6 levels are increased. Genes specifically expressed in layer 5 and 6 neurons are upregulated when Pax6 levels are increased, as are genes associated with cortical neurogenesis (Neurog2, Sox4, Sox21). Conflicting changes in the expression of cell cycle regulators are observed, with increases in expression of positive regulators of G1 progression (Cdk2, Cdk4, Hmga2) and negative regulators of proliferation (Pten, Fzr1), accompanied by decreased expression of genes specifically transcribed in S- and M-phase (Foxm1, Mcm3, 5 & 7). (B) Over-production of basal progenitor cells (Eomes/Tbr2) and layer 6 neurons (Tbr1) observed at E12.5, resulting from increased Pax6 expression that also upregulates Neurog2 expression. Arrowheads indicate the region of Pax6 over-expression within the cortex, which does not extend throughout the entire lateral dimension of the VZ. Scale bars, 100 µm. (C) The increased expression of Pax6 in the D6-Pax6 cortex results in microcephaly at E14.5 that is more pronounced caudally (where the D6 transgene drives the highest increase in Pax6 expression). The cortical plate is reduced in thickness compared to wildtype, with an overall reduction in the number of neurons (as assessed by Tbr1 staining). Increased Neurog2 expression is maintained at this stage, as is increased Eomes/Tbr2 expression. Scale bar, 200 µm. (D,E) Increased Pax6 expression results in a reduction in total cortical size at E12.5. At both E12.5 and E14.5 the microcephaly is not associated with a marked change in mitotic index, as assayed by phopho-histone H3 staining for cells in M-phase at the ventricular surface. Green, Tuj1; red, phospho-histone H3; blue, DAPI. Scale bars, D, 100 µm; F, 200 µm.
Figure 8.
A neurogenesis and self-renewal regulatory circuit operating in neocortical stem cells, regulated by Pax6.
(A) Pax6 binding and regulation data indicate that Pax6 levels are critical in maintaining the balance between stem cell maintenance, neurogenesis and SVZ genesis, as well as enforcing cortical identity. Combining these data with published data on genes downstream of Neurog2, Ascl1 and Hes1 (see text for details) enables construction of a basic network regulating cortical stem cell neurogenesis. Under normal conditions, Pax6 positively regulates Neurog2 and negatively regulates Ascl1 (Mash1), while directly and indirectly repressing the transcription factors Isl1 and Lhx8 respectively, both required for inhibitory interneuron genesis. Overall, this promotes genesis of glutamatergic projection neurons. Pax6 also positively regulates basal progenitor cell genes, including Eomes/Tbr2. Thick lines indicate direct regulation, thin lines indicate evidence from expression studies. (B) Loss of Pax6 function (Sey/Sey) removes the positive regulation of Neurog2 and the negative regulation of Ascl1, Lhx8 and Isl1. This would be predicted to result in the production of GABAergic interneurons in the cortex, as has been previously observed [52]. The loss of Pax6 also removes the positive regulation of the set of basal progenitor cell genes, however this is compensated for many of those genes in large part by the upregulation of Ascl1, with the exception of Eomes/Tbr2. Thus an SVZ is still generated in the absence of Pax6, but without Pax6 it loses its cortical identity and becomes Ascl1-expressing, similar to the SVZ of the ventral forebrain [14]. (C) Increasing the levels of Pax6 (indicated by thick lines) would be predicted to result in an over-production of basal progenitor cells by increasing Neurog2 expression and synergising with Neurog2 to increase expression of basal progenitor cell determinants, including Eomes/Tbr2. This increase in neurogenesis would potentially be at the expense of stem cell maintenance, although both Neurog2 and Pax6 increase expression of the stem cell maintenance factor Hes5.
Figure 9.
Pax6 protein levels are constant in cortical stem and progenitor cells, compared to the variation in Neurog2 and Hes1 levels: implications for network behaviour.
(A) Confocal microscope images of immunofluorescent staining for Pax6, Neurog2 and Hes1 in the E12.5 cortex. Single channel images of staining for each antibody are shown, together with the merged image. Arrowheads highlight Pax6+ nuclei (red), Neurog2+ nuclei (blue) and Hes1+ nuclei (green). Almost all Hes1+ cells are also strongly Pax6+. Both Neurog2+/Pax6+ and Neurog2+/Pax6− cells were observed. Scale bar, 50 µm. (B) Prediction of the output of the self-renewal/neurogenesis network when Hes1 levels are high (active Notch signalling): under these conditions, Hes1 antagonises the positive regulation of Neurog2 by Pax6 to suppress neurogenesis. (C) Prediction of the output of the self-renewal/neurogenesis network when Hes1 levels are low (no active Notch signalling): under these conditions, Pax6 positively regulates Neurog2 expression, unopposed by Hes1. Pax6 and Neurog2 cooperate to promote neurogenesis and basal progenitor cell genesis.