Comparison of Pax6 and Cdca7 expression patterns in wild-type and Pax6
−/− embryos
Pax6 is expressed at high levels in the ventricular zone of the developing cerebral cortex and at lower levels in the ventricular zone of the lateral ganglionic eminence (LGE). At the earliest stages of murine cortical neurogenesis, around embryonic days 12–13 (E12-13), Pax6 expression levels are graded from rostro-lateral [high] to caudo-medial [low]. As corticogenesis continues, this gradient of Pax6 expression becomes progressively less steep [11]. At all embryonic ages, Pax6 expression levels drop sharply at the boundary between pallium and subpallium. We tested whether the expression pattern of Cdca7 normally correlates with that of Pax6 and whether the expression of Cdca7 is affected in mutants that are null for Pax6.
In situ hybridization was used to reveal the expression patterns of Pax6 and Cdca7 in adjacent sections from wild-type (WT; n = 4) and Pax6
−/− (n = 4) embryos at E12.5 (Fig. 1). Note that while the Pax6 null mutants used here do not produce functional Pax6 protein (insets in Fig. 1A, C), they do still express mutant Pax6 mRNA [21, 22]. The expression pattern of Pax6 described above can be seen in Fig. 1A, A′, B, B′. There were much higher levels of Pax6 in lateral cortex than in LGE, with a sharp transition at the pallial-subpallial boundary (PSPB). In contrast, Cdca7 expression levels were lower in the lateral cortex than in the LGE (Fig. 1E, E′, F, F′). In E12.5 Pax6
−/− embryos, the decline in Pax6 expression across the PSPB was more gradual than normal (Fig. 1C, C′, D, D′) and the dip in Cdca7 expression levels in the region of high Pax6 expression in the lateral cortex was no longer obvious (Fig. 1G, G′, H, H′). By E14.5, wild-type Pax6 expression levels were consistently high across the pallium and, as at E12.5, they showed a sharp decline at the PSPB (Fig. 2A, A′; n = 4). Wild-type Cdca7 expression levels again showed an opposing pattern, being lower in the lateral cortex than in the LGE with a sharp change in levels at the PSPB (Fig. 2C, C′). In E14.5 Pax6
−/− embryos (n = 4), Pax6 expression declined more gradually than normal across the PSPB, as at E12.5 (Fig. 2B, B′), but the high levels of Cdca7 expression observed in subpallium continued across the PSPB deep into the lateral cortex where they now overlapped high Pax6 expression (Fig. 2D, D′). These findings suggest that Pax6 is a negative regulator of Cdca7 in the lateral cortex.
We used a quantitative approach to confirm the breakdown in the normal expression of Cdca7 when Pax6 is mutated. In situ hybridizations for Pax6 and Cdca7 were carried out on comparable sections from wild-type and Pax6
−/− littermate embryos aged E12.5, E13.5 and E14.5, mounted and processed together on the same slide to minimise batch effects. Images were analysed with Fiji software. Each image was converted to 8-bit (giving values on a greyscale of 0–255) and a ribbon whose width corresponded to that of the gene expression domains was drawn through the telencephalon from dorsal to ventral as shown in Fig. 3A, B. Average pixel intensities were calculated in 10 µm bins along the ribbon. Background staining was measured in areas expressing neither gene and average pixel intensities were corrected by subtracting BG values to obtain intensity profiles such as those shown in Fig. 3C, D. The profiles were divided into a pallial and a subpallial part based on the expression pattern of the Pax6 gene. Regression analysis gave values for the slopes on the two sides (Fig. 3C, D). Average slopes from several animals were combined to give the graphs in Fig. 3E, F (data are from 4 wild-type and 4 Pax6
−/− embryos at E12.5 and at E14.5 and 3 embryos of each genotype at E13.5).
In the pallium (i.e. dorsal to the PSPB; Fig. 3E), two way ANOVA showed significant effects of both genotype and age on the gradients of Pax6 and Cdca7, with significant interaction effects for both genes (for Pax6: p < 0.001 for age, p < 0.001 for genotype and p = 0.017 for interaction; for Cdca7: p < 0.001 for age, p < 0.001 for genotype and p = 0.026 for interaction). In wild-types, the gradients of Pax6 and Cdca7 expression were opposite at E12.5 (Fig. 3E). Their magnitudes declined significantly over the following days, approaching (in the case of Pax6) or reaching (in the case of Cdca7) zero by E14.5 (dark blue and red lines in Fig. 3E) (E12.5 vs E14.5: for Pax6, p = 0.011; for Cdca7, p = 0.018; Tukey tests). In Pax6
−/− pallium, the gradient of Pax6 expression was much smaller than normal at E12.5 (p < 0.05, Student’s t test) and then reversed (light blue line in Fig. 3E). There was no gradient of Cdca7 expression at E12.5 (a significant difference to the situation in wild-types; p < 0.05, Student’s t-test) but one appeared, with a slope opposite to normal, over the subsequent 2 days (pink line in Fig. 3E; gradients were significantly different at E14.5, p < 0.05, Student’s t-test) (E12.5 vs E14.5: for Pax6, p < 0.001; for Cdca7, p < 0.001; Tukey tests).
On the ventral side of the PSPB (Fig. 3F), two way ANOVA showed a significant effect of genotype but not age on the gradients of Pax6 and Cdca7, with no significant interaction effect for either gene (for Pax6: p < 0.001 for age; for Cdca7: p = 0.015 for age). Pax6 and Cdca7 were expressed in opposing gradients from E12.5-14.5 (dark blue and red lines in Fig. 3F), while in Pax6
−/− embryos the gradients of Pax6 were less steep than normal (light blue line in Fig. 3F; differences were significant at p < 0.05 at all three ages, Student’s t-tests) and the gradients of Cdca7 were abolished (pink line in Fig. 3F; differences with wild-type were significant at p < 0.05 at E12.5 and E14.5, Student’s t-tests). These quantifications support our conclusion that the Cdca7 expression gradient is disrupted by mutation of Pax6.
Double-immunohistochemistry for Pax6 and Cdca7 in wild-type E12.5 embryos was used to test Pax6 and Cdca7 co-expression by cells in the ventricular zone. This confirmed the complementarity of Pax6 and Cdca7 expression patterns around the PSPB (Fig. 4A, B, C, A′, B′, C′). Cells on the subpallial side of the PSPB showed low Pax6 levels and high Cdca7 levels (Fig. 4i). Interestingly, Cdca7 levels in the subpallium were particularly high in the subventricular zone (Fig. 4B′), in line with the relatively strong staining for Cdca7 mRNA in this layer (Fig. 1E′). This strong subventricular zone staining came to an abrupt end at the PSPB (Fig. 4B′). On the pallial side of the PSPB, cells with higher levels of Pax6 staining showed lower levels of Cdca7 (Fig. 4ii). In the region further away from the PSPB, cells with higher levels of Cdca7 showed lower Pax6 staining (Fig. 4iii). Immunohistochemistry on Pax6
−/− embryos showed the elevation of Cdca7 protein levels in lateral cortex and the loss of the medial to lateral gradient in its expression (Fig. 4D).
Our new findings, together with previous microarray and quantitative PCR data showing that levels of Cdca7 increase in the pallium of Pax6
−/− mutant embryos [11], indicate that high levels of Pax6 normally present in the lateral cortex repress Cdca7 expression in this region. We next assessed the functional importance of this repression by asking what would happen if Cdca7 levels were to increase in the lateral cortex of normal embryos.
The consequences of elevating Cdca7 expression in lateral cortex
As cells exit the E12.5 Pax6+ cortical ventricular zone, many of them upregulate Tbr2 and become intermediate progenitors in the subventricular zone before dividing again to generate Tbr1+ neurons that enter the cortical plate (Fig. 5A) [4, 10]. Interestingly, this process is most advanced in the lateral cortex, where Cdca7 levels are most strongly suppressed by Pax6. To test whether low levels of Cdca7 in the lateral cortex are important for cortical development to proceed normally in this region, we elevated Cdca7 levels in progenitors in the lateral cortex of normal embryos and examined the effects on cell proliferation and differentiation into Tbr2+ and Tbr1+ cells.
The open reading frame of a full length Cdca7 cDNA was tagged with influenza hemagglutinin (HA) sequence to generate an HA epitope-tagged version of Cdca7. This was cloned into the expression vector pCAGGS_GFP, which expresses green fluorescent protein (GFP) tagged with a nuclear localization signal (NLS) (Fig. 5B). The Cdca7 expression plasmid was validated by transient transfection of HEK293 cells. At 48 h post transfection, cell lysates were analysed by western blot with an antibody against the HA epitope (Fig. 5C). The level of expression of Cdca7 correlated with the concentration of vector used in the transfection. The lateral cortex of wild-type E12.5 or E14.5 embryos was electroporated in utero with either the Cdca7 expression plasmid pCAGGS_Cdca7 or empty vector (pCAGGS_GFP) as a control. Embryos were injected with bromodeoxyuridine (BrdU) 15.5 h later and culled after a further 0.5 h (Fig. 5D). GFP+ cells were detected through the depth of the lateral cortex 16 h post-electroporation (Fig. 5E). Immunohistochemistry for Cdca7 showed that they expressed higher levels of Cdca7 than their neighbours (Fig. 5F, G). Since the Cdca7 antibody gave clear evidence of increased expression, confirmation using the HA tag in vivo was not considered necessary. We examined 625 GFP+ cells randomly selected from six different sections through the electroporated areas of these embryos and found that 98.5% of them showed Cdca7 expression increased above endogenous levels (Fig. 5H–J; these images illustrate a rare example of a GFP+ cell that did not appear to have elevated Cdca7, white arrows). In general, staining for Cdca7 was stronger in cells with stronger GFP signals, but there were frequent exceptions (e.g. red arrows in Fig. 5H–J show an example of a cell with high Cdca7 but low GFP). Most likely this was due to the use of an internal ribosomal entry site (IRES) in the construct. It is known that IRES activities can vary greatly between cells due to intercellular variation in the levels of positive or negative regulatory IRES trans-acting factors that influence IRES function but not cap-mediated initiation [35].
We estimated the level of Cdca7 expression from the electroporated cells relative to surrounding non-electroporated cells by randomly selecting 30 GFP-expressing cells and 30 intermingled GFP-non-expressing cells within the lateral cortex, outlining each cell and measuring the intensity of the Cdca7 signal from each using Fiji software. We found that the average intensity of the Cdca7 signal was 4.4-fold higher in the GFP+ than in the GFP− cells. To put this difference into context, we then carried out the same analysis with 30 GFP-non-expressing cells from the ventral telencephalon, where Cdca7 is highly expressed by some cells; in this case we selected cells with high levels of expression, located in the ventral telencephalic subventricular zone (Fig. 4B′). We found that the average intensity of the Cdca7 signal from these high-expressing ventral cells was 3.9-fold higher than that from the non-electroporated lateral cortical cells. These values suggest that the electroporated cells in the lateral cortex achieved levels of Cdca7 that were similar to the endogenous levels present in the most highly expressing cells in the ventral telencephalon. This approximates to the situation in Pax6
−/− mutants (Fig. 4D, D′), where the intensity of Cdca7 signals from lateral telencephalic cells becomes closer to that from ventral telencephalic cells (measurements done as above showed an average intensity of Cdca7 signal in ventral telencephalic cells only 1.8-fold above that of lateral cortical cells).
We examined all GFP+ electroporated cells 16 h after electroporation at either E12.5 or E14.5 in a series of sections stained for: GFP and Tbr2; GFP and Tbr1; GFP and BrdU. We found no evidence that the elevation of Cdca7 levels altered the distribution of affected cells through the cortical depth 16 h after electroporation (Fig. 6). Average proportions of GFP+ cells that were double-labelled for Tbr2, Tbr1 or BrdU were obtained from an analysis of multiple electroporated embryos (Figs. 7, 8).
There was a significant reduction by ~20–25% in the proportion of GFP+ cells that were Tbr2+ in embryos electroporated with Cdca7 at E12.5 (Fig. 7A–E), although not in those electroporated at E14.5 (Fig. 7F–J). In both cases, the GFP+ Tbr2+ cells were in the same layer as the non-electroporated (GFP−) Tbr2+ cells. These data suggest that raised levels of Cdca7 reduce the generation of Tbr2+ intermediate progenitors in the early stages of corticogenesis. Early-generated Tbr2+ progenitors are known to produce Tbr1+ neurons [4] and, 16 h after electroporation on E12.5, small numbers of GFP+ Tbr1+ cells had reached the deep edge of the cortical plate. The proportions of GFP+ cells that were Tbr1+ were significantly lower in cells overexpressing Cdca7 than in control cells Fig. 7K–O, which is likely to be a consequence of the underproduction of Tbr2+ intermediate progenitors.
Analysis of the proportion of cells labelled by BrdU shortly before animals were sacrificed allowed us to assess whether elevated Cdca7 levels in the lateral cortex had a drastic effect on proliferation, for example bringing effected cells out of the cell cycle. We found that 20–30% of GFP+ cells took up BrdU in both control and experimental groups following electroporation at E12.5 or E14.5 (Fig. 8). At neither age did electroporation with the Cdca7-expressing plasmid have a significant effect on the proportions of GFP+ cells that were in S-phase in the lateral cortex 16 h later (Fig. 8).
Our findings indicate that the main effect of elevated Cdca7 expression in early embryonic lateral cortical progenitors is to reduce the production of cells expressing Tbr2, which is the hallmark of intermediate progenitors.