The formation of a functional cortical circuitry requires the coordinated growth of cortical axons to their target areas. While the mechanisms guiding cortical axons to their targets have extensively been studied, very little is known about the processes which promote their growth in vivo. Gli3 encodes a zinc finger transcription factor which is expressed in cortical progenitor cells and has crucial roles in cortical development. Here, we characterize the Gli3 compound mutant Gli3Xt/Pdn, which largely lacks Neurofilament+ fibers in the rostral and intermediate neocortex. DiI labeling and Golli-τGFP immunofluorescence indicate that Gli3Xt/Pdn cortical neurons form short and stunted axons. Using transplantation experiments we demonstrate that this axon growth defect is primarily caused by a nonpermissive cortical environment. Furthermore, in Emx1Cre;Gli3Pdn/fl conditional mutants, which mimic the reduction of Gli3 expression in the dorsal telencephalon of Gli3Xt/Pdn embryos, the growth of cortical axons is not impaired, suggesting that Gli3 controls this process early in telencephalic development. In contrast to cortical plate neurons, Gli3Xt/Pdn embryos largely lack subplate (SP) neurons which normally pioneer cortical projections. Collectively, these findings show that Gli3 specifies a cortical environment permissive to the growth of cortical axons at the progenitor level by controlling the formation of SP neurons.
Correct functioning of the cerebral cortex requires that cortical neurons make appropriate contacts with neurons either within or outside the cortex. Our understanding of the mechanisms generating the circuitry that confers functional properties on cortical neurons and networks, although poor, has been advanced significantly by recent research (Price et al. 2006; Paul et al. 2007; Grant et al. 2012; Molnar et al. 2012). These analyses have largely focused on mechanisms regulating axon guidance to targets outside of the cortex via the corticothalamic, corticopontine, and cortiospinal tract or to the opposite cortical hemisphere via the corpus callosum. In contrast, little is known about the mechanisms which control the initial formation of cortical axons and their outgrowth toward and within the white matter in vivo. Subplate (SP) neurons are important regulators of these processes as they establish cortical circuitry by pioneering the first subcortical and contralateral projections (McConnell et al. 1989; Jacobs et al. 2007). In addition, several mouse mutants with axonal growth defects in the cortex have been identified (Polleux et al. 1998; Polleux et al. 2000; Wang et al. 2002; Tissir et al. 2005; Armentano et al. 2006; Wang et al. 2006; Dent et al. 2007; Kwiatkowski et al. 2007; Shima et al. 2007). The corresponding mutations affect genes which are active in cortical neurons and which control axon growth cell autonomously. In contrast, genes which act at the level of cortical progenitors and which control the formation of an environment supportive of cortical axon growth within the cortex remain largely unknown.
The zinc finger transcription factor Gli3 has key roles in telencephalic development. Analysis of the Gli3 loss-of-function mouse mutant extra-toes (Gli3Xt) revealed essential roles for Gli3 in the patterning of the dorsal telencephalon (Johnson 1967; Theil et al. 1999; Tole et al. 2000; Kuschel et al. 2003; Fotaki et al. 2006; Quinn et al. 2009). Moreover, neurogenesis is delayed in the Gli3Xt/Xt cortex affecting the formation of the preplate (Theil 2005) which will later split into the SP and marginal zone (MZ). These severe early patterning defects cause dysmorphogenesis of the Gli3Xt/Xt cortex, thereby significantly complicating the analyses of axon tract formation. In contrast, the Gli3 hypomorphic mutant Polydactyly Nagoya (Gli3Pdn) shows weaker regionalization defects (Kuschel et al. 2003) which do not affect axon outgrowth but lead to severe axon pathfinding defects (Magnani et al. 2010). We therefore became interested to analyze cortical circuitry in the Gli3 compound mutant Gli3Xt/Pdn which shows patterning defects intermediate to Gli3Xt/Xt and Gli3Pdn/Pdn embryos (Kuschel et al. 2003). Here, we show that the Gli3Xt/Pdn neocortex nearly completely lacks Neurofilament expression and only forms short and highly fasciculated axons. Transplantation experiments and analyses of a Gli3 conditional mouse mutant demonstrate that environmental defects in the cortex which arise early in development, before the birth of cortical projection neurons, underlie this axon growth defect. Consistent with an early role of Gli3 in controlling axon growth, we further show that the formation of SP neurons is severely compromised in Gli3Xt/Pdn embryos.
Material and Methods
Animals were handled in accordance with local guidelines for the ethical use of animals in research at the University of Edinburgh. Animal husbandry was in accordance with the UK Animals (Scientific Procedures) Act 1986 regulations. Gli3Xt/+ and Gli3Pdn/+ animals were kept in a mixed C57Bl6/C3H and C3H/He background, respectively, and were interbred. In qualitative analyses of mutant phenotypes, Gli3Xt/+, Gli3Pdn/+, and wild-type embryos did not show differences and were used as control embryos. Gli3Xt/Pdn embryos were distinguished from control embryos by forebrain and/or limb morphology. The Emx1Cre, Gli3flox/flox, τGFP, and Golli-τGFP mouse lines and their genotyping have been described previously (Naruse et al. 1990; Pratt et al. 2000; Gorski et al. 2002; Jacobs et al. 2007; Blaess et al. 2008). Emx1Cre mice were bred into the Pdn line to obtain Emx1Cre;Gli3Pdn/+ which were mated with Gli3flox/flox;Golli-τGFP mice. For Emx1Cre;Gli3Pdn/flox conditional embryos, Gli3flox/flox, Gli3flox/+, Emx1Cre, and Gli3flox/+ embryos were used as controls in qualitative analyses, whereas Gli3flox/+ embryos served as controls in quantitative reverse transcription–polymerase chain reaction (RT–PCR) analysis. For each marker and each stage, 3 to 5 nonexencephalic embryos were analyzed at rostral, medial, and caudal levels of the developing forebrain.
Immunofluorescence analysis on 15-µm cryosections and immunohistochemistry were performed as described previously. The following antibodies were used: Ctip2 (1:1000, Abcam), green fluorescent protein (GFP) (1:1000, Abcam), hippocalcin (1:3000; Abcam), neural cell adhesion molecule L1 (1:1000, Chemicon), microtubule associated protein 2 (Sigma; 1:1000), NF (2H3, Developmental Studies Hybridoma Bank, 1:5), Satb2 (1:50; Abcam), and Tbr1 (Chemicon; 1:2500). Primary antibodies for immunohistochemistry were detected with Alexa- or Cy2/3-conjugated fluorescent secondary antibodies. For nonfluorescent detection, we used biotinylated goat anti-mouse antibodies (Dako) followed by avidin-horseradish peroxidase and 3, 3'-diaminobenzidine detection (Vector Labs).
In Situ Hybridization
In situ hybridization on 12-µm paraffin sections was performed as described recently (Theil 2005). The following in situ probes were used: Ctgf (Hoerder-Suabedissen et al. 2009), Gli3 (Hui et al. 1994), Nurr1 (Quina et al. 2009), Pcp4 (Osheroff and Hatten 2009), and Pls3 (Oeschger et al. 2012).
DiI crystals (D282, Molecular Probes) were placed in target areas of paraformaldehyde (PFA) fixed, E14.5, and P0 brains using glass capillaries. Brains were incubated in the dark at room temperature in 4% PFA for 4 to 6 weeks. Vibratome sections of 100 µm size were prepared from the brains embedded in 4% agarose. After mounting in vectashield sections were examined under epifluorescence microscopy with a rhodamine filter using a Leica confocal microscope.
Organotypic slice cultures of E14.5 and E17.5 mouse telencephalon were prepared as previously described (Magnani et al. 2010). Brain slices were cultured on polycarbonate culture membranes (8μm pore size; Corning Costar) in organ tissue dishes containing 1 mL of medium (Neurobasal/B-27 (1:50), [Gibco] supplemented with glutamine (1×), glucose (0.5%), penicillin and streptomycin(1×)). Slices were cultured for 72h, fixed with 4% PFA and either processed for anti-GFP immunofluorescence as described above (E17.5 sections) or DiA crystals were placed in the transplanted tissue (E14.5 sections). After 10 days at room temperature to allow for the diffusion of the fluorescent dye, sections were transferred to 1:1 glycerol/phosphate buffered saline (PBS), counterstained with TO-PRO 3 in 9:1 glycerol/PBS and examined using a confocal microscope.
Quantitative Reverse Transcription–Polymerase Chain Reaction
Total RNA was isolated from the E11.5 and E12.5 dorsal telencephalon of wild-type and Gli3Xt/Pdn littermate embryos and of Gli3fl/+ and Emx1Cre;Gli3Pdn/flox littermates. Quantitative reverse transcription–PCR (qRT–PCR) was performed using a TaqMan Gene Expression Assay (Applied Biosystems) for Gli3 (Mm00492345_m1, probe dye FAM-MGB) with ACTB (#4352933, probe dye FAM-MGB) and GAPDH (#4352932, probe dye FAM-MGB) as endogenous controls and a 7000 Sequence Detection System. The abundance of each transcript in the original RNA sample was extrapolated from PCR kinetics using Sequence Detection software SDS Version1.2.3 running an absolute quantification protocol including background calibrations.
Axon Outgrowth Defects in the Gli3Xt/Pdn Mutant Cortex
To study the role of Gli3 in cortical development, we previously reported characterizations of the Gli3 loss-of-function mutant extra-toes (Gli3Xt) and the Gli3 hypomorphic mutant Polydactyly Nagoya (Gli3Pdn). While Gli3Xt/Xt mutants show severe patterning defects which preclude the analysis of fiber tract formation, Gli3Pdn/Pdn embryos show moderate regionalization, but severe axon pathfinding defects in the thalamocortical tract (Magnani et al. 2010). Therefore, we became interested whether axon forebrain tracts are affected in the Gli3 compound mutant Gli3Xt/Pdn which displays intermediate patterning defects relative to Gli3Xt/Xt and Gli3Pdn/Pdn embryos (Kuschel et al. 2003; Friedrichs et al. 2008). We therefore performed Neurofilament immunohistochemistry on sectioned P0 control and Gli3Xt/Pdn brains. In control embryos, this analysis revealed the corpus callosum which is absent in Gli3Xt/Pdn animals (Fig. 1A,B) consistent with the highly dysmorphic rostral dorsomedial telencephalon (Friedrichs et al. 2008). In control embryos, thalamocortical axons (TCAs) form dense axon bundles in the internal capsule, defasciculate in the striatum, and enter the cortex where they form dense Neurofilament+ axon fibers in the cortical intermediate zone (IZ) together with corticothalamic axons (CTAs; Fig. 1A,C). In contrast, the internal capsule is broadened in the mutant and Neurofilament+ fibers fail to project into the cortex but turn ventrally at the pallial/subpallial boundary (PSPB) (Fig. 1B,D). Most strikingly, however, the Gli3Xt/Pdn neocortex nearly completely lacks neurofilament staining at rostral and intermediate levels. Neurofilament+ fibers only formed in the IZ at caudal levels but did not project into the striatum (Fig. 1F). However, at this level, ectopic fibers were found in the MZ. Taken together, these findings suggest severe defects in TCA pathfinding and in the formation and/or growth of CTAs in Gli3Xt/Pdn mutants.
While a more detailed analysis of the Gli3Xt/Pdn TCA phenotype will be presented elsewhere, we focus here on a further analysis of the CTA phenotype. Neurofilament is a panneuronal marker that labels all axon tracts in the forebrain. To label cortical axons specifically, we made use of Golli-τGFP transgenic mice (Jacobs et al. 2007). This transgene is expressed in a subset of SP cells and of layer VI neurons and labels their axonal projections toward the corpus callosum and along the corticothalamic tract (Jacobs et al. 2007), thereby revealing a broad IZ populated by numerous, parallel organized axons (Fig. 2A,B). In Gli3Xt/Pdn embryos, we observed very few GFP+ fibers forming thick fascicles that appear in cross section, strongly suggesting that the growth of these axons is impeded (Fig. 2C,D). Moreover, the dendritic palisade appears less dense and much less organized than in control embryos. To further evaluate the axon growth in Gli3Xt/Pdn embryos, we performed DiI tracing experiments. Crystal placement in the P0 neocortex resulted in labeling of callosal and corticofugal projections in wild-type embryos while in the mutant neocortex very few and short fibers were filled (Fig. 2E–H). Slightly longer fibers were detected only at caudal levels, but these axons failed to enter the striatum and projected along the PSPB. Taken together, these analyses suggest a specific and severe defect in axon growth in the neocortex of Gli3Xt/Pdn animals.
These severe axon defects could be due to a general failure of cortical neurons to outgrow axons or due to a degeneration of earlier-formed axons. To distinguish between these possibilities, we analyzed the expression of axonal markers at E14.5. At this stage, we detected Neurofilament+ fibers in the lateral neocortex of control embryos originating from corticofugal axons which have just started to penetrate the lateral ganglionic eminence (LGE) and from TCAs which have already entered the neocortex (Fig. 3A). In Gli3Xt/Pdn embryos, we did not observe any neurofilament immunoreactivity in the neocortex but thalamocortical fibers were detected in the LGE although they did not project into the cortex (Fig. 3B). This defective development of cortical projections was confirmed by DiI labeling. In wild-type embryos, crystal placement in the neocortex resulted in anterograde and retrograde labeling of CTAs and TCAs, respectively, revealing the complete corticothalamical tract. In Gli3Xt/Pdn brains, however, this staining detected only a very short fiber bundle that did not enter the LGE but projected along the PSPB (Fig. 3C,D). Furthermore, staining for the Golli-τGFP transgene in control embryos revealed many early corticofugal axons, organized in parallel to the ventricular surface of control embryos. In contrast, cortical axons are less dense and follow more irregular pathways in Golli-τGFP;Gli3Xt/Pdn brains (Fig. 3E–H). Taken together with the P0 analyses, these findings suggest that Gli3Xt/Pdn cortical neurons start to form axons but their growth is severely retarded.
Defects in the Cortical Environment Primarily Underlie the Axonal Growth Defects of Gli3Xt/Pdn Cortical Neurons
Next, we started to address the cellular basis for the axonal growth defects of Gli3Xt/Pdn cortical neurons. By in situ hybridization, Gli3 mRNA can only be detected in cortical progenitors but not in differentiated neurons (Hui et al. 1994). Based on this expression pattern, Gli3 could confer cortical neurons with the ability to grow their axon at progenitor stages. Alternatively, it could control the development of the environment through which these axons grow. Our previous analyses showed that corticofugal (Tbr1+; Sox5+) and upper layer callosal neurons (Cux2+) are formed in Gli3Xt/Pdn mutants, but these neurons occupy abnormal positions in the cortex (Kuschel et al. 2003; Friedrichs et al. 2008). Moreover, the P0 Gli3Xt/Pdn neocortex contains Ctip2+ corticospinal motor neurons and Satb2+ upper layer neurons though their lamina position is affected (Fig. 4), lending further support to the notion that the Gli3Xt/Pdn mutation does not interfere with fate specification of cortical neurons. To further distinguish between neuronal and environmental defects, we performed intercortical transplantation experiments using E17.5 embryos and labeling cortical axons with a ubiquitously expressed τGFP fusion protein (Pratt et al. 2000). To this end, we used sections through rostral and intermediate cortex where the Neurofilament staining and DiI labeling revealed the strongest axon growth defects in Gli3Xt/Pdn embryos. The transplanted tissue contained ventricular, IZ, SP and cortical plate, and MZ to adjust for the different thickness of the mutant and control cortices. When we transplanted control;GFP+ cortical tissue homotopically into control host cortex, GFP+ cortical axons grew medially and laterally from the transplant toward the contralateral cortical hemisphere and toward the striatum, respectively (n = 7/7) (Fig. 5A,D,G). Transplantation of Gli3Xt/Pdn;GFP+ cortical tissue into control cortex led to a similar outgrowth of axons into the host cortex (n = 8/8) (Fig. 5B,E,G) while control;GFP+ tissue formed only very few and short axons in a Gli3Xt/Pdn cortical environment (n = 0/3) (Fig. 5C,F,G). Similarly, transplanting cortical tissue from E14.5 embryos also led to the outgrowth of axons from Gli3Xt/Pdn cortical tissue in a control environment (n = 5/6) while control cortical tissue produced only very short axons in a Gli3Xt/Pdn cortical environment (n = 0/6) (Supplementary Fig. 1). Taken together, these data indicate that Gli3Xt/Pdn cortical neurons are capable of growing axons but the Gli3Xt/Pdn environment through which these axons would grow is not permissive to axon growth.
The Axon Growth Defect in Gli3Xt/Pdn Embryos Arises during Early Telencephalic Development
During cortical development, Gli3 is expressed in progenitor cells from early forebrain stages (E8.5) till late cortical maturation stages (E18.5) (Hui et al. 1994). We therefore investigated when the cortical growth defects start to arise in Gli3Xt/Pdn embryos. To this end, we used a Gli3 conditional mouse mutant (Blaess et al. 2008) and an Emx1Cre driver line which directs Cre expression specifically in the dorsal telencephalon from E9.5 (Gorski et al. 2002). To mimic cortical development in Gli3Xt/Pdn mutants, we mated Emx1Cre; Gli3flox/+ embryos animals with Gli3Pdn/+ mutants to generate Emx1Cre;Gli3Pdn/flox conditional mutants. Since these embryos carry the Gli3 floxed allele over the Pdn allele, they are predicted to have reduced Gli3 transcript levels in the dorsal telencephalon similar to that of Gli3Xt/Pdn mutants once recombination in the dorsal telencephalon is completed. Indeed, Gli3 in situ hybridization on sections of E12.5 Emx1Cre;Gli3Pdn/flox brains show reduced Gli3 expression in the dorsal telencephalon while expression in the ventral telencephalon is not affected (Fig. 6A,B). Moreover, qRT–PCR revealed that Gli3 expression levels in the E11.5 cortex of Emx1Cre;Gli3Pdn/flox conditional mutants are reduced compared with Gli3flox/+ embryos but are at a higher level than in Gli3Xt/Pdn mutants (0.19 ± 0.05-fold for Gli3Xt/Pdn mutants compared with 0.35 ± 0.06-fold for Emx1Cre;Gli3Pdn/flox embryos; n = 3 for each genotype) while they have reached that level in E12.5 Emx1Cre;Gli3Pdn/flox mutants (0.22 ± 0.01-fold for Gli3Xt/Pdn mutants compared with 0.24 ± 0.02-fold for Emx1Cre;Gli3Pdn/flox embryos; n = 3 for each genotype) (Fig. 6C). Taken together, these findings suggest that Gli3 recombination is complete by E12.5.
We next analyzed the formation of major axon tracts in Emx1Cre;Gli3Pdn/flox mutants using L1 immunofluorescence and the Golli-τGFP transgene. In control E14.5 embryos, L1 stains cortical axons in the IZ, the anterior commissure, and TCAs in the striatum (Fig. 6D,H) while the Golli-τGFP transgene more specifically labels early cortical axons and the anterior commissure (Fig. 6F,J). At this stage, the L1 and Golli-τGFP staining patterns in Emx1Cre;Gli3Pdn/flox mutants are nearly identical to that of control embryos, only fewer CTAs have entered the LGE in conditional mutants (Fig. 6E,G,I,K). In E18.5 control embryos, L1 staining revealed the corpus callosum and dense fiber bundles in the cortical IZ (Fig. 6L,P). Golli-τGFP also stains the corpus callosum and corticofugal axons projecting through the striatum and labels a dense palisade of dendrites emanating from SP and layer VI neurons (Fig. 6N,R). In Emx1Cre;Gli3Pdn/flox mutants, these axon tract are also present (Fig. 6M,O,Q,S) although we detected several axon pathfinding defects. Some callosal axons do not cross the midline (Fig. 6O) and fewer corticofugal axons enter the striatum (Fig. 6M,O). Moreover, GFP+ axons appear slightly disorganized in the IZ (Fig. 6S). Despite these pathfinding defects, these data show that, unlike in Gli3Xt/Pdn embryos, axon growth of cortical neurons is not affected in Emx1Cre;Gli3Pdn/flox mutants, suggesting that the axonal defect in Gli3Xt/Pdn embryos arises as a result of defective processes prior to E12.5.
Development of the SP is Severely Affected in Gli3Xt/Pdn Embryos
The previous analyses showed a requirement of Gli3 before E12.5 in controlling the growth of cortical axons. This finding provided a basis for a further investigation of the cellular and molecular basis for this defect. The dorsal telencephalon only starts to give rise to cortical projection neurons from E12.5 onwards while preplate neurons, which later populate the MZ and the SP, have already been generated at this stage. Moreover, ablation of preplate neurons results in a similar axon growth defect as in Gli3Xt/Pdn embryos (Xie et al. 2002) and SP neurons have a crucial role in pioneering the corticothalamic tract and the corpus callosum (McConnell et al. 1989; Jacobs et al. 2007). We previously showed that the E12.5 Gli3Xt/Pdn neocortex has mild regionalization defects but our birth dating and initial marker analyses failed to identify SP neurons in their normal location underneath the cortical plate (Friedrichs et al. 2008). This finding raised that the possibilities that SP neurons are either not formed or have defects in cell migration, differentiation, and/or survival. To further explore these possibilities, we performed an expression analysis of various SP-specific markers throughout cortical development. Initially, we analyzed the expression of Ctgf and Nurr1 in newborn brains. mRNAs of both genes are detected in distinct subsets of the SP and Nurr1 expression spreads across most layers in lateral cortex (Hoerder-Suabedissen et al. 2009 and Figure 6A,B). In the P0 Gli3Xt/Pdn neocortex, the SP expression of both markers is lost but the cortical plate expression of Nurr1 expands into neocortical areas (Fig. 7G,H) confirming the results of our previous birth dating analysis (Friedrichs et al. 2008). Next, we determined the expression of Pls3 and Pcp4 which label the SP in E15.5 embryos (Osheroff and Hatten 2009; Oeschger et al. 2012 and Figure 6C,D) when cortical plate neurons start to project their axons in the IZ. In contrast, the expression of both genes was detected only in the lateral most neocortex of Gli3Xt/Pdn embryos but was not detected more medially (Fig. 7I,J), suggesting that SP development is already severely affected in E15.5 mutants and that Gli3Xt/Pdn embryos might not form SP neurons. To test for this possibility, we analyzed the expression of the earliest known SP marker hippocalcin in E12.5 embryos (Osheroff and Hatten, 2009). This analysis revealed extensive hippocalcin+ neurons in the preplate of E12.5 control embryos (Fig. 7E,F), but no such cells in the Gli3Xt/Pdn neocortex (Fig. 7K,L). Since Caspase3 staining did not reveal a significant increase in cell death in the neocortex at this stage (data not shown), we conclude that the Gli3Xt/Pdn neocortex generates only few SP neurons at lateral most levels. In contrast, the expression of these SP markers is not affected in Emx1Cre;Gli3Pdn/fl mutants (Fig. 7M–R) suggesting that defective SP development underlies the axon outgrowth defects in Gli3Xt/Pdn embryos.
The proper functioning of the cerebral cortex requires the establishment of cortical connections and therefore the growth of cortical axons to their target areas. Here, we present a severe axon outgrowth phenotype in the Gli3 compound mutant Gli3Xt/Pdn and a detailed analysis of the cellular mechanisms underlying this defect. In the rostral and intermediate cortex of these mutants, cortical neurons fail to express the pan-axonal marker Neurofilament. DiI-labeling experiments and expression analysis of the Golli-tGFP transgene further indicated that Gli3Xt/Pdn cortical neurons are able to form axons but these processes are stunted. Many axons also form highly fasciculated bundles which appear in cross section, strongly suggesting that the loss of Neurofilament staining in the neocortex not only indicates a differentiation defect of cortical axons but also indicates that their growth is severely impaired. Moreover, this axon outgrowth defect represents a more severe CTA phenotype than in Gli3Pdn/Pdn embryos where CTAs leave the cortex and project toward the diencephalon though their migration is delayed (Magnani et al. 2010). Taken together, these findings suggest that correct Gli3 expression levels are crucial for determining the growth of cortical axons and their pathfinding.
Recently, a number of mouse mutants have been identified in which the growth of cortical axons is affected. The corresponding genes encode a variety of functions including axon guidance molecules (Polleux et al. 1998; Polleux et al. 2000), components of the Wnt planar cell polarity pathway (Wang et al. 2002; Tissir et al. 2005; Wang et al. 2006; Shima et al. 2007), actin-binding proteins (Dent et al. 2007; Kwiatkowski et al. 2007), and transcription factors (Armentano et al. 2006) that are thought to act within or on cortical neurons to regulate axon growth. In contrast to these genes, Gli3 acts in progenitor cells and the Gli3Xt/Pdn mutant provided a unique possibility to analyze how the growth of cortical axons is regulated at the progenitor level. On the one hand, Gli3 could specify the ability of cortical neurons to grow their axons at the progenitor level but the expression of lamina-specific markers and of callosal and corticofugal fate determinants are not affected (Friedrichs et al. 2008). Moreover, the cortical lamination defects in Gli3Xt/Pdn mutants are unlikely to cause axon outgrowth defects since a more severe disruption of cortical lamination in reeler mutant mice does not interfere with the growth of cortical axons (Caviness and Yorke 1976; Molnar, Adams, Goffinet et al. 1998). Alternatively, Gli3 could direct the development of an environment permissive for axon growth which is strongly supported by our transplantation experiments. While a control host environment allowed the growth of cortical axons derived from a Gli3Xt/Pdn tissue transplant, control axons fail to grow in a Gli3Xt/Pdn mutant environment. Moreover, analyses of Emx1Cre;Gli3Pdn/fl conditional mutants indicated that growth defects already arise before E12.5. Taken together, these findings suggest that Gli3 specifies an environment in the cortex supportive of axon growth at early stages in telencephalic development well before axons of cortical projection neurons start their migration.
This finding of an early Gli3 role in controlling cortical axon outgrowth allowed us to pinpoint the relevant cellular processes. First, this early role ruled out the possibility that the failure of TCAs to enter the Gli3Xt/Pdn cortex could result in the growth defect of cortical axons since TCAs start penetrating the cortex only at E14.5. Prior to E12.5, however, cortical development involves 2 major processes, regionalization and the formation of the preplate. Our previous analyses revealed that neocortical regionalization in general is mildly affected in Gli3Xt/Pdn embryos but showed an expansion of the ventral pallial progenitor domain (Friedrichs et al. 2008) resulting in an enlargement of the piriform cortex (Fig. 4). It is unlikely, however, that this change underlies the axon outgrowth defects since Emx1Cre;Gli3Pdn/fl conditional mutants display a similarly expanded piriform cortex but normal growth of cortical axons (Fig. 7N and Supplementary Fig. 2). We therefore focussed our further analyses on preplate development. Interestingly, ablation of preplate neurons results in a dramatic reduction of cortical projections similar to the phenotype of Gli3Xt/Pdn mutants (Xie et al. 2002). Amongst the preplate derivatives, SP neurons appeared to be exciting candidates. SP cells are among the earliest born neurons of the cerebral cortex and play critical roles in cortical lamination and for the establishment of cortical connectivity (Allendoerfer and Shatz 1994; Molnar and Blakemore 1995; Molnar, Adams and Blakemore 1998; Super et al. 1998; Del Rio et al. 2000). SP neurons pioneer corticofugal projections and the corpus callosum (McConnell et al. 1989; De Carlos and O'Leary 1992; Jacobs et al. 2007) and guide TCAs into the cortex (Ghosh et al. 1990). Our previous BrdU birthdating and marker analyses in Gli3Xt/Pdn mutants had already indicated defects in SP development (Friedrichs et al. 2008). Here, we extend these findings by showing that Gli3Xt/Pdn mutants largely lack the expression of several SP-specific markers throughout cortical development, that is, from E12.5 to P0, suggesting that the generation of SP neurons is severely compromised in Gli3Xt/Pdn mutants. Given their significance in establishing cortical connectivity, this extensive lack of SP cells provides a compelling explanation for the failure of axonal growth in the Gli3Xt/Pdn cortex. According to this scenario, cortical plate neurons rely on SP neurons for their axonal projections and in their absence cannot grow axons properly. However, the molecular properties which distinguish SP neurons from cortical plate neurons and which enable them to pioneer cortical projections in an environment nonpermissive for cortical plate axons remain elusive.
Our findings also raise the question how Gli3 regulates the generation of SP neurons at the molecular level. Compared with the importance of SP neurons, surprisingly little is known about genes which control their early development. Few mouse mutants have been described which affect the specification and early differentiation of SP neurons and which could provide hints to Gli3’s role(s). The Tbr1, Sox5, and COUP-TF1 transcription factors interfere with the differentiation, cortical localization, and survival of SP neurons but not with their formation (Zhou et al. 1999; Hevner et al. 2001; Lai et al. 2008). Therefore, we can only speculate how Gli3 controls SP formation but based on previously described Gli3 functions and genetic interactions we envision the following mutually nonexclusive possibilities. Gli3 could directly determine SP fate similar to its involvement in specifying the lamina fate of cortical plate neurons (Wang et al. 2011). Alternatively, it could regulate SP formation indirectly by controlling the expression of other transcription factors and/or signaling molecules. Wnt/β-catenin signaling, for example, inversely correlates with SP formation (Machon et al. 2007) and is severely reduced in the dorsal telencephalon of Gli3Xt/Pdn embryos (Friedrichs et al. 2008). Moreover, transcription of the Emx1 homeobox gene is lost while Emx2 expression is reduced in Gli3Xt/Pdn embryos (Kuschel et al. 2003). While SP neurons are not formed in Emx1/2 double mutants (Shinozaki et al. 2002; Bishop et al. 2003), their development is not affected in Emx1−/−;Emx2+/− embryos (Bishop et al. 2003), suggesting that altered Emx gene expression may contribute to the Gli3 SP phenotype but also implying additional factor(s) in the generation of SP neurons. Detailed molecular analyses are required in the future to identify these factors and Gli3Xt/Pdn embryos will be a valuable tool to characterize the genetic pathways controlling the generation and specification of this crucial cell type.
This work was supported by grants from the Deutsche Forschungsgemeinschaft (TH770/6-1) and from the Medical Research Council (G0801359).
We thank Drs. John Mason and David Price for critical comments on the manuscript. We are grateful to Trudi Gillespie for help with confocal imaging, Alex Joyner and Sandra Blaess for providing the Gli3 conditional mouse line, and Zoltan Molnar for probes for in situ hybridization. We thank Amira Baharin who worked on this project as part of her Honors research. Conflict of Interest: None declared.