The corpus callosum (CC) represents the major forebrain commissure connecting the 2 cerebral hemispheres. Midline crossing of callosal axons is controlled by several glial and neuronal guideposts specifically located along the callosal path, but it remains unknown how these cells acquire their position. Here, we show that the Gli3 hypomorphic mouse mutant Polydactyly Nagoya (Pdn) displays agenesis of the CC and mislocation of the glial and neuronal guidepost cells. Using transplantation experiments, we demonstrate that agenesis of the CC is primarily caused by midline defects. These defects originate during telencephalic patterning and involve an up-regulation of Slit2 expression and altered Fgf and Wnt/β-catenin signaling. Mutations in sprouty1/2 which mimic the changes in these signaling pathways cause a disorganization of midline guideposts and CC agenesis. Moreover, a partial recovery of midline abnormalities in Pdn/Pdn;Slit2−/− embryos mutants confirms the functional importance of correct Slit2 expression levels for callosal development. Hence, Gli3 controlled restriction of Fgf and Wnt/β-catenin signaling and of Slit2 expression is crucial for positioning midline guideposts and callosal development.
The corpus callosum (CC) connects neurons of the 2 cerebral hemispheres and coordinates information between the left and right cortex. CC malformations have been associated with mental retardation involving a wide range of cognitive, behavioral, and neurological consequences (Richards et al. 2004; Paul et al. 2007) and have been identified in over 50 human congenital syndromes (Richards et al. 2004). During CC formation, several guidance events control midline crossing of callosal axons. The midline zipper glia (MZG) have been suggested to initiate the fusion of the dorsal midline producing the substrate on which callosal axons navigate (Silver et al. 1993). Moreover, several guidepost cells are located along the path of callosal axons including the midline glial cell populations composed of the indusium griseum glia (IGG) and the glial wedge (GW) (Richards et al. 2004), and GABAergic and glutamatergic neurons that transiently populate the CC (Niquille et al. 2009). Finally, axons from the cingulate cortex pioneer the CC and function as scaffolds for neocortical axons (Koester and O'Leary 1994; Rash and Richards 2001; Piper, Plachez et al. 2009). Several axon-guidance molecules, including Slit2, that are produced by midline glial cells and by the glutamatergic neurons have essential roles in callosal development (Bagri et al. 2002; Niquille et al. 2009). While these studies reveal complex interactions between callosal axons and their environment, it remains largely unknown how guidepost cells acquire their correct positions and how the expression of essential guidance molecules is regulated.
Gli3 encodes a zinc-finger transcription factor with crucial roles in early patterning of the dorsal telencephalon (Theil et al. 1999; Tole et al. 2000; Kuschel et al. 2003; Fotaki et al. 2006) acting both cell autonomously (Quinn et al. 2009) and cell nonautonomously by controlling the expression of signaling molecules essential for telencephalic development (Grove et al. 1998; Theil et al. 1999; Tole et al. 2000; Aoto et al. 2002). Moreover, Gli3 functions in axon pathfinding in the forebrain. The Gli3 hypomorphic mouse mutant Polydactyly Nagoya (Pdn) shows defects in the corticothalamic and thalamocortical tracts (Magnani et al. 2010) and lacks the CC (Naruse et al. 1990) though for unknown reasons. Using transplantation experiments, we here demonstrate that midline abnormalities are primarily responsible for agenesis of the corpus callosum (ACC). We show that Pdn mutants display mislocated glial and neuronal guidepost cells. The Pdn cingulate cortex contains ectopic glial cells transecting the path of callosal axons. These midline abnormalities are associated with an up-regulation and down-regulation of Fgf and Wnt/β-catenin signaling, respectively. These changes in these signaling pathways are mimicked in Sprouty1/2 double mutants, which display a mislocation of midline guideposts and ACC. Pdn mutants also show an up-regulation of Slit2 expression, and positioning of the neuronal guideposts is largely rescued in Pdn/Pdn;Slit2−/− double mutants suggesting that maintaining correct Slit2 expression levels is crucial for callosal development. Collectively, these analyses reveal a novel role for Gli3 in controlling the positioning of midline guideposts by regulating Fgf and Wnt/β-catenin signaling and Slit2 expression levels and provide new insights into the mechanisms underlying CC pathogenesis.
Materials and Methods
The mutant mouse lines Pdn, τGFP, Slit2, Sprouty1, and Sprouty2 and mating strategies have been described previously (Naruse et al. 1990; Pratt et al. 2000; Plump et al. 2002; Basson et al. 2005; Shim et al. 2005; Simrick et al. 2011). All experimental procedures involving mice were performed in accordance with local guidelines. In analyses of Pdn mutant phenotypes, heterozygous and wild-type embryos did not show qualitative differences and both were used as control embryos. For quantitative analyses, wildt-ype and Pdn/Pdn embryos were compared to avoid the possible risk of Pdn/+ embryos having subtle defects. For each marker and each stage, 3–5 embryos were analyzed.
In Situ Hybridization and Immunohistochemistry
Antisense RNA probes for Bmp7 (Furuta et al. 1997), Msx1 (Hill et al. 1989), Sema3C (Bagnard et al. 2000), Slit2 (Erskine et al. 2000), Fabp7 (Genepaint. RNA probe 653), Fgf8 (Crossley and Martin 1995), Sprouty2 (Minowada et al. 1999), Axin2 (Lustig et al. 2002), Wnt7b (Parr et al. 1993), Wnt8b (Richardson et al. 1999), Nf1b (IMAGE: 4038233), Nf1x (IMAGE: 3491917), Emx1 (Simeone et al. 1992), and Six3 (Oliver et al. 1995) were labeled with digoxigenin. In situ hybridization on 12-μm serial paraffin sections of mouse brains were performed as described (Theil 2005).
Immunohistochemical analysis was performed as described previously (Theil 2005) using antibodies against the following molecules: β-III-tubulin (Tuj1 antibody; 1:1000, Sigma); brain lipid-binding protein (Blbp; 1:500, CHEMICON); calbindin (CB; 1:1000, Swant); calretinin (CR; 1:1000, CHEMICON); Glast (1:5000, CHEMICON); glia fibrillary acidic protein (GFAP; 1:1000, DakoCytomation); green fluorescent protein (GFP; 1:1000, Abcam); Nf1a (1:1000, Active Motif); neural cell adhesion molecule L1 (1:1000, CHEMICON); Neurofilament (2H3; 1:5, DSHB); Neuropilin-1 (Npn-1; 1:1000, R&D Systems); Satb2 (1:50, Abcam); Tbr1 (1:2500, CHEMICON). Primary antibodies for immunohistochemistry were detected with Alexa- or Cy2/3-conjugated fluorescent secondary antibodies. For non-fluorescent detection, we used biotinylated goat antimouse antibodies (Dako) followed by avidin-HRP and DAB detection (Vector Labs).
CB+ neurons in the indusium griseum of E16.5 and E18.5 embryos were quantified by determining total CB+ cell numbers in this region. For quantifying CR+ neurons, a box with constant area (170 μm2 for E16.5 and 297 μm2 for E18.5 embryos) was placed in the cingulate cortex immediately dorsal to the CC, and the numbers of CR+ neurons were counted within this box. Numerical values are given as a proportion of CR+ cells per μm2. For statistical analyses, an analysis of variance test was used followed by a Bonferroni's multiple comparison test.
Organotypic slice cultures of rostral levels of the embryonic mouse telencephalon were prepared as previously described (Magnani et al. 2010). Brain slices were cultured on polycarbonate culture membranes (8-μm pore size; Corning Costar) in organ tissue dishes containing 1 mL of medium (Neurobasal/B-27 [Gibco] supplemented with glutamine, glucose, penicillin, and streptomycin). For transplantation experiments, slices were cultured for 72 h, fixed with 4% PFA, and processed for antiGFP immunofluorescence as described above. For Fgf blocking experiments, slices were cultured in the presence of either DMSO or of 25 or 100 μM SU5402 (Calbiochem) for 48 h , fixed with 4% PFA, and processed for in situ hybridization or Blbp immunofluorescence as described above.
Quantitative Reverse Transcription PCR
Total RNA was prepared from the E14.5 rostromedial telencephalon of wild-type or Pdn/Pdn embryos. Quantitative reverse transcription polymerase chain reaction (qRT-PCR) was performed using a TaqMan® Gene Expression Assay (Applied Biosystems) for Slit2 (Mm00662153.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 reaction kinetics using sequence detection software (SDS) version 1.2.3 running an absolute quantification protocol including background calibrations.
CC Midline GuidePost Cells are Severely Disorganized in Pdn/Pdn Brains
Neurofilament, Tuj1 and L1 immunhistochemical stainings, and cortical DiI labeling confirmed a previous description of CC malformation in Pdn mutants (Naruse et al. 1990), showing that the path of callosal axons is disrupted at several positions in the cingulate cortex and that those axons which approach the midline fail to cross it, forming Probst bundles instead (Fig. 1 and Supplementary Fig. 1). To gain insights into the origins of these defects, we analyzed the navigation of the cingulate pioneer axons and the formation of glial and neuronal guideposts that are essential for callosal development (Paul et al. 2007). In P0 control animals, the cingulate pioneer axons are immunopositive for Npn-1 occupying the dorsal-most part of the CC (Fig. 1A,B). In Pdn mutants, Npn-1+ axons fail to project to the contralateral hemisphere, but form dense bundles ipsilaterally (Fig. 1C,D). Glutamatergic guidepost neurons express Tbr1, CR, or CB (Niquille et al. 2009). In control embryos, CR+ and CB+ neurons are both located in the IG region, and CR+ neurons are also found within the CC where they delineate its ventral and dorsal parts (Niquille et al. 2009; Fig. 1A,B,E,F). In Pdn mutants, CR+ neurons are dramatically disorganized, but maintain their spatial association with callosal axons, with clusters of CR+ neurons surrounding the Probst bundles (Fig. 1C,D). CB+ neurons remain concentrated in the medial cortex although they are more diffusely distributed and clusters of CB+ neurons intermingle abnormally with callosal axons (Fig. 1G,H). Finally, GFAP immunostaining labels the GW, the IGG, and the MZG in control embryos (Fig. 1I,J). In Pdn brains, several GFAP+ fascicles are formed ectopically in the cingulate cortex (Fig. 1K,L). Some fascicles span the whole cortical width and transect the path of callosal axons. The IGG could not be identified and the MZG expands into more ventral regions of the septum. Taken together, these data show a dramatic disorganization of glial and neuronal guidepost cells.
Given the severity of this disorganization, we started to study the origins of these midline defects by investigating the formation of the midline guideposts and of the cingulate pioneer neurons at earlier stages. In E16.5 control embryos, cingulate pioneer axons approach the midline and start to cross it (Supplementary Fig. 2A,B). In Pdn mutants, few axons have reached the corticoseptal boundary (CSB) and many have abnormally formed clusters in the cingulate cortex (Supplementary Fig. 2C,D). Tbr1+, CR+, and CB+ glutamatergic guidepost neurons form a well organized band of neurons at the CSB of control embryos, but their organization is severely disturbed in Pdn embryos with less CB+ neurons in the IG region (Supplementary Figs 2A–L and 7P). In the cingulate cortex, the cortical plate is disrupted in several positions where callosal axons stop their navigation (Supplementary Fig. 2G,H). Finally, radial glial cells (RGCs) at the CSB, which co-express GFAP and the RGC marker Glast, have started to differentiate into GW cells, to translocate to the pial surface, and to form the IGG in control embryos (Supplementary Fig. 2M,N). In Pdn mutants, GFAP+;Glast+ cells are present ectopically in the cingulate cortex and extend projections from the ventricular to the pial surface (Supplementary Fig. 2O,P). Taken together, these findings suggest that midline guidance cues are already disorganized in Pdn mutants when callosal axons approach the CSB.
Agenesis of the CC in Pdn Mutants is Caused by CC Midline Defects
Since Gli3 is widely expressed in progenitor cells that give rise to callosal neurons and to guidepost cells, agenesis of the CC in Pdn mutants could either be caused by the disorganization of midline guideposts or by a primary failure of callosal axons to navigate in the midline region leading to the formation of Probst bundles and to a secondary redistribution of guideposts. Previous marker and BrdU birthdating analyses in Pdn/Pdn mutants failed to find major defects in cortical lamination (Magnani et al. 2010). Moreover, Satb2 upper layer callosal projection neurons are borne at E15.5 (Supplementary Fig. 3), suggesting that these neurons are specified correctly. To test directly whether Pdn mutant callosal axons are capable of following midline guidance cues we performed in vitro transplantation experiments using mice ubiquitously expressing a τGFP fusion protein (Pratt et al. 2000). Homotopical transplantation of frontal cortex of E17.5 GFP+ embryos into cortical sections of age-matched GFP− embryos resulted in growth of axons into the host tissue and in midline crossing of callosal axons (n = 8 of 9; Fig. 2A). After transplantation of Pdn/Pdn;GFP+ cortex into control cortex, Pdn/Pdn axons also migrated across the midline (n = 7 of 8; Fig. 2B). However, control;GFP+ callosal axons did not grow into Pdn/Pdn;GFP− dorsomedial cortex (n = 0 of 7; Fig. 2C) and only a few axons projected along the surface of the mutant host tissue (n = 4 of 7; Fig. 2C). In contrast, corticofugal axons project into the lateral cortex and striatum under these conditions (Magnani et al. 2010). These results show that normal levels of Gli3 are not required to generate callosal neurons with the ability to project their axons across the midline, but indicate a requirement for Gli3 in the generation of the midline guideposts.
The Pdn Mutation Affects the Patterning in the Rostromedial Telencephalon
Next, we became interested in identifying causes underlying these midline defects. Our previous analyses showed that Pdn mutants display patterning defects during early telencephalic development (Kuschel et al. 2003). We therefore hypothesized that these defects might cause the defective positioning of the midline guidance cues. To test this idea, we started to analyze the development of the E12.5 corticoseptal region where callosal axons later cross the midline. We showed previously that expression of the Emx1 homobox gene is lost in Pdn mutants (Kuschel et al. 2003). Moreover, Emx1 mutants display ACC (Qiu et al. 1996; Yoshida et al. 1997) and Emx1 has recently been shown to belong to a group of transcription factors including Six3 and Nfia whose expression domains delineate the regions where the CC, the hippocampal, and anterior commissures cross the midline at E16.5 (Moldrich et al. 2010). As these genes have important roles in forebrain and/or callosal development (Qiu et al. 1996; das Neves et al. 1999; Lagutin et al. 2003; Shu et al. 2003; Campbell et al. 2008; Plachez et al. 2008; Piper, Moldrich et al. 2009), we investigated their expression at the E12.5 CSB where callosal axons later cross the midline. In control embryos, Six3 is expressed in the septum, but Six3 expression expands dorsally in Pdn mutants (Supplementary Fig. 4A,E). Nfia, Nfib, and Nfix, are expressed at high levels in the cortex and at lower levels in the dorsalmost septum (Shu et al. 2003; Plachez et al. 2008; Supplementary Fig. 4B–D). In Pdn mutants, their cortical expression domains are lost, while low-level septal expression remains except for Nfia which is strongly expressed in the septum (Supplementary Fig. 4F–H). Taken together, these data indicate that the expression of several transcription factors with important roles in callosal development is altered in Pdn mutants, suggesting that the CSB is poorly defined.
Previous analyses had also shown a requirement of Gli3 for the correct expression of several signaling molecules in the telencephalon, including Shh, Bmp/Wnt genes, and Fgf8 (Grove et al. 1998; Theil et al. 1999; Tole et al. 2000; Aoto et al. 2002; Kuschel et al. 2003; Magnani et al. 2010). We therefore analyzed the expression of these signaling molecules specifically at the E12.5 CSB. This analysis revealed a slight extension of Shh expression into the ventral-most part of the septum in Pdn/Pdn embryos, but Shh signaling as judged by Ptc1 expression remains confined to the septum and does not reach the CSB (Supplementary Fig. 5). Moreover, Bmp7, which is essential for callosal development (Sanchez-Camacho et al. 2011), and its target gene Msx1 are expressed on the cortical side of the CSB though only at caudal levels with no obvious difference between control and Pdn/Pdn embryos (Fig. 3A,B,F,G). In contrast, we observed severe changes in the Wnt7b/8b expression patterns. In control embryos, Wnt7b and Wnt8b expression are confined to the dorsomedial telencephalon with a sharp expression boundary at the CSB (Fig. 3C,D), while Wnt7b and Wnt8b expression is nearly absent from the Pdn dorsomedial telencephalon, and Wnt7b transcription is increased in the septum (Fig. 3H,I). Consistent with reduced Wnt/β-catenin signaling, expression of the Wnt target gene Axin2 is severely reduced in Pdn mutants (Fig. 3E,J).
Since telencephalic patterning is controlled by a balance between Bmp/ Wnt/β-catenin and Fgf signaling (Theil et al. 1999; Kuschel et al. 2003; Shimogori et al. 2004) and since Fgf8 is required for callosal development (Huffman et al. 2004; Moldrich et al. 2010), we also investigated Fgf8 expression in Pdn mutants. In control embryos, Fgf8 transcripts are confined to the commissural plate, but expand further dorsally in the E12.5 Pdn corticoseptal region (Fig. 3K,O) consistent with our previous whole-mount expression analysis (Kuschel et al. 2003). Expression of and phospho-Erk (pErk), targets of Fgf signaling, also extends dorsally into the cortex (Fig. 3L,M,P,Q). A similar expansion of Fgf8 and sprouty2 expression were already observed in E11.5 Pdn embryos (data not shown), indicating that Fgf signaling is ectopically activated during patterning. We also analyzed Fabp7 expression which in control embryos marks neurogenic RGC on the cortical side of the CSB (Fig. 3N) and which is increased upon up-regulation of Fgf signaling in the rostromedial telencephalon (Faedo et al. 2010). Interestingly, the Pdn dorsomedial cortex lacks this high-level Fabp7 expression domain, but shows clusters of RGCs with high levels of Fabp7 expression next to cells having little Fabp7 transcripts (Fig. 3R) reminiscent of the ectopic Glast+ fibers which we observed at E16.5. Taken together, these analyses indicate severe changes in Fgf and Wnt/β-catenin signaling in the rostromedial telencephalon of Pdn mutants.
Sprouty1/2 Double Mutants Show Agenesis of the CC
To investigate the importance of these changes in Fgf and Wnt/β-catenin signaling for callosal development, we made use of Sprouty1/2 double mutants. Sprouty1 and Sprouty2 encode negative feedback regulators of Fgf signaling (Kim and Bar-Sagi 2004). In the E12.5 rostromedial telencephalon of Sprouty1/2 double mutants, Fgf signaling is up-regulated which in turn leads to a down-regulation of Wnt/β-catenin signaling (Faedo et al. 2010) similar to the situation in Pdn mutants. We first determined the effects of these alterations to the signaling pathways on the development of guidepost neurons. At E14.5, prior to the arrival of callosal axons, the CR+ guidepost neurons accumulate at the CSB forming a well organized band of neurons which, however, is largely missing in Sprouty1/2 double mutants (Fig. 4A,B). The mutants also lack CB+ neurons that can already be detected in the midline region of control embryos (Fig. 4C,D), suggesting that the development of guidepost neurons is disturbed in these mutants before callosal axons approach the CSB. Next, we analyzed CC formation in E18.5 embryos. While the formation of Satb2+ callosal projection neurons and their positioning in the upper cortical layers is not affected (Fig. 4E,F). Neurofilament and Tuj1 staining revealed agenesis of the CC in Sprouty1/2 mutants (Fig. 4G–L,O,P). Callosal fibers project toward the midline, but fail to cross and form ectopic axon bundles. The analysis of the midline guideposts showed no dramatic differences in the distribution of CB+ neurons, but CR+ neurons formed abnormal fibers in the ectopic axon bundles (Fig. 4I–N). Several GFAP+ glia fibers abnormally cluster at the CSB, transecting the path of callosal axons, while the IGG could not be identified (Fig. 4O,P). Taken together, these data show that up-regulation of Fgf signaling is sufficient to induce callosal malformation.
Fgf Signaling is Reduced in the E16.5 Pdn Cingulate Cortex
A recent analysis had shown that Fgf signaling is required between E15.5 and E17.5 for the translocation of glial cells toward the indusium griseum (Smith et al. 2006). Since interfering with Fgf signaling at this stage leads to glial translocation defects very similar to those in E18.5 Pdn mutants, we investigated Fgf8 expression and that of its target gene sprouty2 in E16.5 Pdn embryos. In the rostral cortex of control embryos, both genes are expressed in the IGG and in the GW and sprouty2 expression expands into the cingulate cortex (Supplementary Fig. 6A,B). At more caudal levels, Fgf8 and sprouty2 transcripts were detected in the septum and in the stria medullaris thalami (Supplementary Fig. 6C,D). In contrast, Fgf8 expression is absent from the IG region and from the GW of Pdn embryos and is confined to the caudal septum (Supplementary Fig. 6E,G). At this caudal level, septum and cingulate cortex are only connected by a thin bridge of tissue. This abnormal morphology and the absence of Fgf8 expression in the GW and IG region suggests that Fgf8 might not signal to the cingulate cortex. Consistent with this idea, sprouty2 is only expressed in the septum but not in the cingulate cortex of Pdn mutants (Supplementary Fig. 6F,H). Taken together with the results of our E12.5 analysis, these data strongly suggest an early phase when Fgf signaling is up-regulated in the E12.5 rostromedial Pdn telencephalon causing patterning defects and a clustering of RGCs followed by a later phase with a down-regulation of Fgf signaling in the E16.5 cingulate cortex due to an abnormal morphology of the Pdn rostral midline tissue. This down-regulation coincides with the glial translocation defect in Pdn mutants.
Positioning of Midline Guidance Cues is Rescued in Pdn/Pdn;Slit2−/− Embryos
The findings described above indicate that altered Fgf signaling plays an important part in the development of the Pdn callosal phenotype. However, the callosal phenotype of sprouty1/2 embryos appears relatively mild compared with that of Pdn mutants, suggesting additional abnormalities in Gli3 mutants. We therefore started to analyze the expression of axon guidance molecules in Pdn mutants. In E16.5 control embryos, Sema3c is expressed in glutamatergic guidepost and cingulate neurons thereby attracting callosal axons toward the midline (Niquille et al. 2009; Piper, Plachez et al. 2009), but its expression is only slightly reduced in Pdn mutants (Supplementary Fig. 7). Slit2 normally prevents callosal axons from projecting into the septum (Bagri et al. 2002) and is already expressed in the commissural plate of E9.5 embryos (Yuan et al. 1999) and in the septum of E12.5 control embryos (Fig. 5A). Interestingly, our in situ hybridization showed a slight expansion of Slit2 expression into the cortical region of E12.5 Pdn embryos (Fig. 5E). This expansion became more prominent by E14.5 when strong Slit2 expression is confined to the septum of control embryos with a graded but weaker expression in cortical midline progenitors. In contrast, Slit2 expression is up-regulated in the rostromedial Pdn cortex and Slit2 transcripts were ectopically detected in the septal midline (Fig. 5B,C,F,G). To confirm this potential increase in Slit2 expression, we used qRT-PCR on rostromedial telencephic tissue to show a significant increase in Slit2 mRNA expression levels (Fig. 5I). Moreover, expanded Slit2 expression is maintained in the E16.5 cingulate cortex (Fig. 5D,H). Thus, Pdn mutants show an expansion of Slit2 expression in the rostromedial cortex from patterning stages until time points when callosal axons approach the CSB.
To test for a role of this expanded Slit2 expression, we analyzed CC development in Pdn/Slit2 double mutants. Initially, we determined the positioning of guidepost cells in E16.5 embryos. This analysis showed that the organization of the cortical midline is much improved in Pdn/Pdn;Slit2+/− and in Pdn/Pdn;Slit2−/− embryos. The positioning of the CB+ and CR+ guidepost neurons is largely rescued (Supplementary Fig. 8B–D,G–I). The numbers of CB+ neurons are increased in double mutants, though not to wild-type levels, while CR+ neurons are present in normal numbers in the double mutants (Supplementary Fig. 8P,Q). The formation of GFAP+ GW cells is restricted to the CSB, although the GFAP staining appears more irregular with a few isolated GFAP+ fascicles (Supplementary Fig. 8L–N). Moreover, in contrast to Pdn/Pdn embryos, L1+ callosal axons progress through the cingulate cortex without disruption in double-mutant embryos (Supplementary Fig. 8B–D;G–I;L–N). We also analyzed the positioning of guidepost cells in Slit2−/− embryos (Supplementary Fig. 8E,J,O). While the CB+ and many CR+ guidepost neurons acquire their correct position in the prospective IG region of Slit2−/− mutants, some CB+ and CR+ neurons intermingle ectopically with callosal axons in the septum, where callosal axons are misdirected. In addition, there is a dramatic increase in the number of callosal axons reaching the midline region in Slit2−/− embryos as reported previously (Bagri et al. 2002).
Finally, we analyzed CC formation in E18.5 Pdn/Slit2 double mutants. This analysis confirmed our findings on the much improved organization of midline guidepost neurons, but callosal axons do not cross the midline (Fig. 6C,D,H,I,M,N). In the Pdn cingulate cortex, the intermediate zone is disrupted by several, large Probst bundles (Fig. 6B,G,L). In Pdn/Pdn;Slit2+/− and in Pdn/Pdn;Slit2−/− embryos, callosal axons migrate uninterrupted through the cingulate cortex without forming Probst bundles (Fig. 6C,D,H,I,M,N). CB+ neurons are located in the IG region similar to control embryos, but are scattered in the Pdn cortex (Fig. 6A–D). CR+ neurons occupy positions in the dorsomedial cortex of the double mutants, while these cells are mostly associated with the Probst bundles in Pdn mutants (Fig. 6F–I). In addition to their correct position, normal numbers of CB+ and CR+ neurons are present in the midline region of double mutants (Fig. 6P,Q). In contrast, the midline glia develops abnormally in Pdn/Slit2 double mutants (Fig. 6K–N). The IGG is missing and ectopic glial fascicles are still formed at the CSB but only in the ventralmost part of the cortex (Fig. 6M,N). Interestingly, the guidepost neurons are also severely affected in Slit2−/− mutants. Few CR+ neurons occupy their normal position in the IG, while large clusters of CR+ neurons were detected ventrally to the callosal axons crossing the midline (Fig. 6E,J,O). In addition, 2 large ectopic bundles of fibers were also found at either side of the CC as described previously (Bagri et al. 2002). Taken together, these analyses show a remarkable recovery of midline morphology in Pdn/Slit2 double mutants.
Up-Regulation of Fgf Signaling Controls Slit2 Expression and is Required for RGC Clustering in Pdn Mutants
Taken together, our analyses demonstrate roles for Fgf signaling and Slit2 in positioning callosal guidance cues raising the possibility that both pathways are interconnected. To test for this, we employed an ex vivo explant assay in which we prepared coronal sections of E13.5 control and Pdn/Pdn rostral telencephalon, including the commissural plate as the Fgf8 signaling centre, and maintained these sections in culture for 48 h in the presence of DMSO or various concentrations of SU5402, which selectively inhibits Fgf signaling. We first determined the effects of these treatments on the expression of sprouty2. Under control conditions, sprouty2 expression is detected in the septum on sections of control and Pdn/Pdn embryos (Fig. 7A,B). While the addition of 100 μM SU5402 severely disrupted tissue morphology (data not shown), sprouty2 expression was abolished in the presence of 25 μM SU5402 (Fig. 7C), indicating that this concentration is sufficient to block Fgf signaling in this ex vivo explant culture assay. Next, we analyzed the expression of Slit2 after SU5402 treatment. In the presence of DMSO, Slit2 transcripts are confined to the septum of control embryos (Fig. 7D), but Slit2 expression expands into the cortex and into the ventral-most septum on Pdn/Pdn sections (Fig. 7E). SU5402 treatment of Pdn mutant sections resulted in a loss of Slit2 expression in this latter tissue and in reduced expression in the cortex (Fig. 7F), suggesting that up-regulated Fgf signaling in Pdn mutants plays at least a partial role in controlling Slit2 expression. Finally, we used the same assay to determine a role for Fgf signaling in the formation of the ectopic RGC clusters. Immunofluorescence for the Blbp antigen which is encoded by Fabp7 revealed RGCs in the cortex dorsally to the CSB on control sections and widespread RGC clusters on Pdn mutant sections (Fig. 7G,H) similar to our in vivo findings (compare with Fig. 3N,R). In contrast, addition of 25 μM SU5402 nearly completely abolished the formation of RGC clusters on Pdn/Pdn sections (Fig. 7I) strongly suggesting that their formation depends on up-regulated Fgf signaling.
Several glial and neuronal guidepost cells are organized in strategic positions at the CSB and play crucial roles in the midline crossing of callosal axons, but it remains largely unknown how the guideposts acquire their correct position. The Gli3 hypomorphic mutant Pdn provides an interesting model to address this as the normal distribution of callosal guideposts is severely affected in this mutant. The cortical midline region contains ectopic glial fibers that transect the path of callosal axons and shows an up-regulation of the Slit2 guidance molecule. Several lines of evidence strongly suggest that the ACC in Pdn mutants is caused by these midline defects rather than by defects in callosal axons. Cortical layering, the expression of the callosal determinant Satb2 (Alcamo et al. 2008; Britanova et al. 2008) and the birthdate of upper layer callosal neurons are not affected in Pdn embryos (Magnani et al. 2010). Moreover, Pdn mutant callosal axons are capable of midline crossing in a wild-type environment. Finally, molecular changes in the cortical midline relevant to the callosal malformation occur as early as E12.5. As these alterations occur well before callosal axons arrive at the midline, our findings strongly suggest that Gli3-controlled early patterning events are crucial for setting up the spatial organization of midline guideposts and hence for callosal development.
Pdn mutants showing a very severe callosal phenotype present an interesting tool to identify pathways controlling patterning of the CSB. In fact, our analyses led to the identification of altered activities in key signaling pathways and of changed expression patterns of several transcription factors emphasizing this link between patterning and callosal development. First, several transcription factors with important functions in early forebrain and callosal development have altered expression patterns in the corticoseptal region of E12.5 Pdn embryos. Mutations of the human and mouse SIX3 genes lead to holoprosencephaly (Wallis et al. 1999) and to severe truncations of the prosencephalon (Lagutin et al. 2003), respectively, but the severity of these phenotypes might obscure potential role(s) in callosal formation. In contrast, Emx1 mutants show ACC due to a lack of the indusium griseum (Qiu et al. 1996; Yoshida et al. 1997). Furthermore, Nfia, Nfib, and Nfix have high expression level domains dorsally to the CSB (Shu et al. 2003; Campbell et al. 2008; Plachez et al. 2008) overlapping with the domains of Wnt7b/8b expression, suggesting regulatory relationships between these genes. Mutations in Nfia and Nfib lead to callosal defects due to malformations in the midline glial cell populations and to defective development of the cingulate pioneer neurons (Shu et al. 2003; Steele-Perkins et al. 2005; Piper, Moldrich et al. 2009). Our data suggest that these factors have an earlier patterning role that might be obscured by redundancy between these factors.
Secondly, we identified altered Fgf signaling and Wnt/β-catenin signaling at the CSB in E12.5 Pdn mutants as important regulators of callosal development. In fact, Sprouty1/2 double mutants, in which increased Fgf signaling down-regulates Wnt/β-catenin signaling in the rostromedial telencephalon (Faedo et al. 2010), display agenesis of the CC. Interestingly, these mutants already show defective development of CR+ and CB+ guidepost neurons at the E14.5 CSB. Although we cannot exclude the possibility that elevated Fgf signaling after E14.5 might further disrupt callosal formation, this altered development of guidepost neurons prior to the arrival of callosal axons strongly suggest that increased levels of Fgf signaling at patterning stages already interfere with guidepost and hence callosal development. This idea is supported by recent findings on callosal development in Rfx3 mutant mice in which a mild up-regulation of Fgf signaling underlies a mislocalization of glutaminergic guidepost neurons (Benadiba et al. 2012). Consistent with recent findings on a regulatory role of Fgf signaling in RGC development (Kang et al. 2009; Sahara and O'Leary 2009), we also show here that up-regulating Fgf signaling is required for the formation of RGC clusters in the rostromedial telencephalon of Pdn mutants. Our Blbp/GFAP double staining further indicates that these RGC clusters give rise to the ectopic glial cells in the E16.5 Pdn cingulate cortex, which due to morphological alterations lacks Fgf signaling at this state. This lack is likely to result in a failure of ectopic glial cells to translocate (Smith et al. 2006). Taken together, these findings indicate 2 phases for Fgf signaling in callosal development. During a newly identified, early patterning phase Fgf signaling sets the CSB and positions glial and neuronal guidepost cells. In a second phase, Fgf signaling is required for glial cell translocation (Smith et al. 2006). These data also demonstrate that a reduction and an increase in Fgf signaling can cause ACC, strongly suggesting that regulating Fgf8 expression levels is crucial for callosal development. This regulation might involve a positive feedback loop with Shh (Ohkubo et al. 2002) and/or an interaction with Wnt7b and Wnt8b which have complementary expression patterns to Fgf8 at the CSB. Previous analyses have implicated Wnt5a and Ryk-mediated Wnt/Ca2+ signaling in promoting the escape of callosal axons from the midline into the contralateral hemisphere (Keeble et al. 2006; Hutchins et al. 2011). Moreover, the meninges and neurons of the cingulate cortex use a cascade of signals including Wnt3 to regulate midline crossing of cingulate pioneer axons (Choe et al. 2012). In contrast, Wnt8b mutant mice show normal callosal development probably due to redundancy with other Wnt molecules (Fotaki et al. 2010). However, Wnt7b/8b expression is already down-regulated before the onset of ectopic Fgf8 expression in the E9.0 Pdn telencephalon (Ueta et al. 2008). This and the reduced Wnt/β-catenin signaling in the sprouty1/2 double mutants (Faedo et al. 2010) suggest an antagonistic interaction between Fgf and Wnt/β-catenin signaling to control Fgf8 expression levels in the commissural plate, thereby regulating patterning of the CSB and positioning of midline guideposts (Fig. 8).
Finally, the up-regulation of Slit2 expression represents a major cause of the Pdn callosal phenotype. Pdn/Slit2 double mutants show a dramatic improvement in the growth of cortical axons toward the midline and in midline organization suggesting 2, mutually non-exclusive roles for Slit2 in callosal development. First, Slit2 could control the permissiveness of the cingulate cortex for the growth of callosal axons. Indeed, callosal axons approach the CSB without forming Probst bundles in Pdn/Slit2 double mutants, and many callosal axons approach the midline but miss-project into the septum in Slit2−/− mutants (Bagri et al. 2002). This idea is also consistent with the temporal expression profile of Slit2, which becomes down-regulated in the control cingulate cortex after E14.5 (Fig. 5). Alternatively, Slit2 could regulate the migration of guidepost neurons into the cortical midline (Niquille et al., 2009) similar to its effect on the migration of LGE guidepost cells (Bielle et al. 2011). The positioning and the numbers of guidepost neurons are largely rescued in the Pdn/Slit2 double mutants, while CR+ neurons form ectopic clusters in Slit2−/− embryos. Taken together, these findings raise the interesting possibility that a major role of Slit2 in callosal development is to coordinate the migration of callosal axons with that of the guidepost neurons.
Interestingly, Slit2 expression is already expanded in E12.5 Pdn embryos suggesting that early patterning regulates its expression. Slit2 could be a downstream target of Fgf signaling given its coexpression with sprouty2 (Yuan et al. 1999) and its down-regulation in the septum of Fgfr1 mutant mice (Tole et al. 2006) and after blocking Fgf signaling on rostromedial tissue sections (Fig. 7). Alternatively, Gli3 or transcription factors downstream of Gli3, such as Emx1 or the Nfi transcriptional regulators, could repress Slit2 expression in the rostrodorsal telencephalon. Irrespective of the exact mechanism, the up-regulation of Slit2 provides a link between early patterning and the coordination of midline development.
In summary, our analyses provide insights into how early patterning of the cortical midline controls the organization of midline guideposts and the formation of a permissive environment allowing callosal axons to approach the CSB. In this process, Gli3 takes centre stage by controlling Fgf and Wnt/β-catenin signaling at the rostral midline and the expression of several transcription factors and of the Slit2 axon guidance molecule. Interestingly, the human GLI3 gene is mutated in Acrocallosal syndrome patients who lack the CC (Elson et al. 2002). CC malformations are also a frequent hallmark of ciliopathies in which the function of the primary cilium and hence Gli3 processing is affected (Tobin and Beales 2009). Therefore, our findings provide a framework for understanding the defective processes underlying the ACC in Acrocallosal syndrome and in ciliopathies.
Supplementary material can be found at: http://www.cercor.oxfordjournals.org/.
This work was supported by grants from the Deutsche Forschungsgemeinschaft (TH770/6-1) and from the Medical Research Council (G0801359) (TT) and the Wellcome Trust (091475) (A.B.).
We thank Drs Thomas Becker, Christopher D. Conway, Bénédicte Durand, John Mason, and Tom Pratt for critical comments on the manuscript. We are grateful to Trudi Gillespie for help with confocal imaging and Gail Martin for the Sprouty2 mouse line. We thank Neil Campbell who worked on this project as part of his Honours research. Conflict of Interest: None declared.