Balanced Shh signaling is required for proper formation and maintenance of dorsal telencephalic midline structures

Background The rostral telencephalic dorsal midline is an organizing center critical for the formation of the future cortex and hippocampus. While the intersection of WNTs, BMPs, and FGFs establishes boundaries within this critical center, a direct role of Shh signaling in this region remains controversial. In this paper we show that both increased and decreased Shh signaling directly affects boundary formation within the telencephalic dorsal midline. Results Viral over-expression of Shh in the embryonic telencephalon prevents formation of the cortical hem and choroid plexus, while expanding the roof plate. In a transgenic model where cholesterol-lacking ShhN is expressed from one allele (ShhN/+), genes expressed in all three domains, cortical hem, choroid plexus and roof plate expand. In Gli1/2 -/- mutant brains, where Shh signaling is reduced, the roof plate expands, again at the expense of cortical hem and plexus. Cell autonomous activation of Shh signaling in the dorsal midline through Gdf7-driven activated Smoothened expression results in expansion of the Wnt3a-expressing cortical hem into the plexus domain. In addition, developmental stage determines dorsal midline responsiveness to Shh. Conclusions Together, these data demonstrate that balanced Shh signaling is critical for maintaining regional boundaries within the dorsal midline telencephalic organizing center.


Background
The telencephalic dorsal midline contains two organizing centers: the roof plate and cortical hem [1,2]. The roof plate is initially induced by signals from the overlying epidermal ectoderm, and once established, provides a secondary source of secreted TGFβ-family members along the entire dorsal midline of the developing neural tube [3]. Fate-mapping experiments show that the roof plate is derived from Wnt1-expressing cells in the overlying neuroectoderm, and develops from lineagerestricted cellular compartments [4]. Evidence that the roof plate may be an organizer stems from genetic ablation experiments demonstrating roof plate-dependent dorsal interneuron specification in the spinal cord, and cortical and choroid plexus development in the telencephalon [5][6][7]. In addition, it has been shown that the roof plate directs choroid plexus formation through a cell non-autonomous mechanism [5]. The cortical hem, originally identified as an embryonic structure marked by Wnt expression [8], exhibits hippocampal organizer activity [9]. Targeted inactivation of Wnt3a and the signaling co-factor Lef-1 confirm the hem's role in hippocampal growth and development [10,11].
The intersection of multiple secreted factors expressed in the telencephalic midline are known to contribute to the patterning of this region. These include BMPs, WNTs and FGFs [12][13][14]. FGF signals from the rostral forebrain regulate anterior-posterior regionalization in the telencephalon [15]. The current model suggests that a number of FGFs, specifically FGF8, function to coordinate the competing morphogenic signals expressed from the dorsal and ventral midlines [16].
However, recent questions have been raised as to a direct role of the secreted signaling protein Shh in patterning the dorsal telencephalic midline. While Shh is expressed in the ventral telencephalic midline as early as E10.5, a time when the first invagination of the dorsal midline region occurs, it is not detectable in the dorsal telencephalic midline. However, disruption of Shh signaling through mutation of Shh or its downstream targets affects dorsal midline patterning in both humans and mice [17][18][19]. Genetic interactions between Shh and Gli3 suggested a mechanism [20] whereby Shh expression in the telencephalic ventral midline represses the transcriptional repressor Gli3 through an activity gradient [21]. However, loss of both Shh copies in Gli3 mutants fails to rescue telencephalic dorsal midline defects, challenging the idea that Shh acts solely through a Gli3-dependent mechanism [22]. Together these data suggest that the role of Shh in dorsal telencephalic midline patterning is still not well understood.
In this paper we show that Shh expression from the dorsal extent of the zona limitans intrathalamica (ZLI, [23]) is positioned correctly to directly influence gene expression boundaries critical for telencephalic dorsal midline formation, specifically the choroid plexus and cortical hem. Using mouse models that contain increased or decreased Shh activity in the developing forebrain, we show that balanced Shh signaling is required for proper telencephalic dorsal midline development. Further, we show that disruption of FGF signaling does not result in dorsal midline patterning defects resembling the Shh loss-or gain-of-function mouse models, suggesting that Shh signaling in the dorsal midline is not mediated by FGFs. Together, these data support a direct role of Shh in patterning the dorsal telencephalic midline.

Results
Shh is expressed adjacent to the Lhx5-expressing roof plate domain in the embryonic telencephalon At E12.5, the telencephalic dorsal midline is delineated by three regions based on morphology and gene expression [8]: the Wnt3a-expressing cortical hem, the rTTRexpressing choroid plexus, and the Lhx5-expressing roof plate (Figure 1). At E10.5, prior to the appearance of rTTR in the choroid plexus, Wnt3a and Lhx5 overlap in the dorsal midline ( Figure 1A, C). Between E11.5 and E12.5, the appearance of the choroid plexus divides the roof and hem boundaries from each other ( Figure 1E-G, I-K). In E10.5-E12.5 coronal sections, Shh expression is restricted to telencephalic ventral midline, and therefore unlikely to directly influence formation of dorsal telencephalic midline structures ( Figure 1D, H, L). However, sagittal analyses at these stages show that dorsal diencephalic extension of Shh in the ZLI is adjacent to the Lhx5-expressing telencephalic roof plate, the region known to be a telencephalic dorsal midline organizer ( Figure 1O, P, S, T). The latter raises the possibility that Shh signaling directly influences dorsal midline structures/boundaries during the time that the choroid plexus becomes juxtaposed between the cortical hem and roof plate.

Increased Shh signaling changes telencephalic dorsal midline morphology and boundaries
In order to directly test the influence of Shh on dorsal midline development, we used ultrasound image-guided injections to introduce viruses expressing cholesterollacking Shh (ShhN) and full length Shh (wtShh) into the E9.5 embryonic forebrain ventricle. This technique provides a powerful approach to disentangle the spatial and temporal roles of Shh in brain development. At E12.5, three days after viral misexpression, wtShh and ShhN virally infected brains are characterized by expanded lateral ventricles and a thinned cortex. The dorsal midline fails to properly invaginate and cortical hem and choroids plexus differentiation is disrupted ( Figure 2F, K, 3E, I, 4K, P, 5F, K).
Since dorsal midline morphology is significantly altered by Shh activation, it is likely that the dorsal midline is still influenced by Shh signaling. Therefore, we next addressed how genes distinguishing the cortical hem, choroid plexus and roof plate may be altered by Shh activation. Figure 4 shows that Shh activation results in the loss of hem (Wnt3a, Figure 4M, R) and plexus (rTTR, Figure 4N, S) structures. Msx-1, which is normally expressed by both hem and plexus, is also lost upon Shh activation ( Figure 4O, T). Therefore, Shh activation results in the loss of specific midline structures, but does not transform the dorsal midline to a ventral fate.
We next asked if Shh activates downstream targets in the dorsal midline, or whether the remaining dorsal midline tissue is refractory to Shh signaling. Figure 5 shows that the dorsal midline responds to Shh activation by ectopic activation of its transcriptional target Gli-1 [29] ( Figure 5G, L), its co-receptor Ptc-1 [30][31][32]  The experiments in Figure 2, 3, 4, and 5 show the effects of ectopic Shh activation by viral delivery to E9.5 telencephalic dorsal midline. In order to address if increased Shh signaling from its normal source (ventral midline or ZLI) would affect telencephalic dorsal midline formation, we utilized an activated Shh transgenic model where one allele of Shh is replaced by ShhN (ShhN/+) [33]. In this model, ShhN is proposed to travel further, resulting in Shh activation in the dorsal telencephalon. Figure 6 shows that all three telencephalic dorsal midline boundaries are expanded in ShhN/+ mice ( Figure 6D-F). While viral activation and genetic modification both cause roof plate Lhx5 expansion, rTTR and Wnt3a expand in ShhN/+ mice ( Figure 6) and are absent in brains injected with Shh virus (Figure 4). Together, these data not only support the idea that increased Shh signaling alters telencephalic dorsal midline patterning, but that timing and dose are critical factors in determining the outcome.

Telencephalic dorsal midline defects result from decreased Shh signaling
Increased Shh signaling, both indirectly through loss of a repressor (Gli3, [8]) and directly through transgenic or viral expression (this paper) can affect telencephalic dorsal midline boundaries. However, it is unknown whether Shh normally plays a direct role in patterning the telencephalic dorsal midline. The dramatic loss of both dorsal and ventral telencephalic midline structures in Shh -/-mutants supports the idea that Shh is critical for proper formation of both regions [34]. Since Shh has been shown to activate transcriptional targets through the action of Gli1 and Gli2, combined loss of these two transcription factors reduces, but does not eliminate  direct Shh signaling [35,36]. Figure 7 shows that reduced Shh signaling in Gli1/2 -/-mutants results in the loss of Wnt3a ( Figure 7D) and rTTR ( Figure 7E) in the hem and plexus, respectively. In place of the hem and plexus, the Gli1/2 -/-mutant telencephalic dorsal midline now expresses Lhx5 throughout ( Figure 7F), without proper thinning to a single epithelial layer characteristic of the roof plate [1]. This suggests that Shh signaling is required for inducing proper morphology and gene expression boundaries that establish the hem, plexus and roof relationships. The current model predicts that Shh indirectly influences dorsal midline development through the action of FGF8 [18,33]. Shh expression from the ventral midline is necessary for the maintenance of FGF8 expression [37], and FGF8 is thought to restrict the spread of BMPs and WNTs in the dorsal telencephalon [12]. Given the hypothesis that Shh acts through FGF8 to influence the dorsal midline, we investigated the effects of FGF inhibition during dorsal telencephalic midline patterning. An FGF receptor tyrosine kinase inhibitor (PD173074, Calbiochem) was injected into mouse E9.5 forebrain, and gene expression changes were examined at E12.5. While the FGF inhibitor interferes with the morphological development of the dorsal midline, Wnt3A and rTTR are still expressed in the cortical hem and choroid plexus structures, respectively ( Figure  7G, H). In addition, Lhx5 expansion and thickening of the roof plate is not observed ( Figure 7I). Therefore, the dorsal midline defects resulting from reduced Shh signaling (Gli1/2 -/-) are distinct from reduced FGF signaling, raising the possibility that Shh signaling in the telencephalic dorsal midline may be independent of FGF. However, we cannot rule out the possibility that FGF inhibition is incomplete, and then fails to generate the complete roof plate phenotype. Future experiments that include both gain and loss of function FGF action in the telencephalon combined with Shh activation will be necessary to determine whether Shh acts through FGF in this region. Cell autonomous increase in telencephalic dorsal midline Shh signaling alters cortical hem/plexus boundary In order to distinguish between cell-autonomous and non-autonomous Shh activity, we next tested the effects of Gdf7-promoted activated Smoothened on telencephalic dorsal midline gene expression boundaries and morphology. Using a Gdf7cre transgenic mouse [6] crossed to a conditionally activated Smoothened (ActSmo) [38], it is possible to limit Shh activation to roof plate cells, and to a small group of hem and plexus cells. Figure 8 shows that cell autonomous activation of Shh through act-Smo expression in the Gdf7 domain increases roof plate thickness, and prevents formation of the single epithelial layer characteristic of the roof plate [1]. Wnt3a, which is normally limited to the cortical hem, expands into the choroid plexus region (arrows, Figure 8D). In addition, rTTR expression expands slightly dorsally into the cortical hem region (arrows, Figure 8E), and the choroid plexus appears somewhat reduced in size. Together these data suggest that Shhdependent cell autonomous changes in the roof plate shifts boundaries between hem and plexus.

Time-dependent effects of Shh ectopic activation on dorsal midline boundaries
The appearance of the telencephalic choroid plexus (between E11.5 and E12.5) follows the formation of the roof plate and cortical hem (between E9.5 and E11.5).
The appearance of Shh in the ZLI occurs between E10.5 and E12.5 [39]. If balanced Shh signaling were required to maintain relationships between the telencephalic roof plate, hem, and plexus, then it would be expected that the roof plate remains competent to respond to Shh signaling later than E9.5. Therefore, we next asked whether increased Shh signaling affects dorsal midline patterning after roof plate and hem formation have already begun. Figure 9 shows E12.5 dorsal telencephalic midline analysis of brains infected with wtShh virus at E10.5. Roof plate expansion, as indicated by Lhx5 expression, is still observed ( Figure 9B). In addition, the choroid plexus gene rTTR is not detected rostrally in the telencephalon ( Figure 9E), but is detected in more caudal sections at the level of the diencephalon ( Figure 9H). The morphological invagination of the choroid plexus epithelium is apparent without induction of rTTR. The cortical hem can be distinguished both morphologically and by expression of Wnt3A ( Figure 9C, F, I). Expansion of the hem into the plexus domain is also observed ( Figure  9F, I), similar to that found in GDF7cre-ActSmo brains ( Figure 8D). Therefore, hem and plexus structures change in Shh responsiveness between E9.5 and E10.5, while the roof plate retains its earlier sensitivity. The latter is consistent with the idea that Shh is important for maintaining the roof plate after its formation.

Discussion
In this paper, we show that increased or decreased Shh signaling disrupts dorsal telencephalic midline development, suggesting that balanced Shh signaling is required for the proper formation and maintenance of the telencephalic dorsal midline. We show that Shh activity can directly influence dorsal midline patterning through ectopic viral expression at E9.5 (Shh E9.5vir ). The Shh E9.5vir telencephalon lacks both cortical hem and choroid plexus. In addition, the Shh E9.5vir telencephalic dorsal midline is refractory to ventralization, with expanded expression of the roof plate gene Lhx5. Recent characterization of the ShhN/+ telencephalon also shows Lhx5 roof plate expansion accompanied by a loss of cortical hem and choroid plexus [33]. These results are consistent with the finding that ShhN, the cholesterol lacking form of Shh, is more potent in telencephalic explant assays [25], and exhibits an extended range of signaling [33,40,41]. Also consistent with this idea is that ectopic expression of ShhN in the dorsal midline results in a more extreme phenotype than wtShh ( Figure 2F vs 2K, Figure 3E vs 3I, Figure 4K vs 4P, Figure 5F vs 5K). While our analysis of Shh E9.5vir using wtShh or ShhN supports that ShhN/+ mutant phenotypes in the telencephalon result from increased Shh activity, our analysis of ShhN/+ dorsal midline defects differs from that of Huang et al. [33]. Both analyses show that ShhN/+ mutants exhibit roof plate expansion. However, unlike Huang et al. [33], our analysis shows that ShhN/+ mutants display expanded cortical hem and plexus structures along with the expanded roof plate, generating a larger dorsal midline. Phenotypic differences in ShhN/+ may result from differential cre-induced removal of the ShhN conditional allele (EIIAcre compared with Sox2Cre), or possibly from mouse strain differences.

Wnt3a rTTR Lhx5
A B C The source of Shh in ShhN/+ mutants derives from adjacent or ventral domains. The dorsal midline expansion in ShhN/+ embryos may be attributed to increased Shh signaling from the ventral telencephalon, increased signaling from the diencephalic ZLI, or both. In contrast, viral delivery results in direct expression of Shh in the dorsal midline. Thus, the level of exposure at the dorsal midline is likely to be higher in Shh E9.5vir compared with ShhN/+. The use of viral delivery methods can help elucidate direct effects of Shh activity on dorsal midline boundary formation. Another important difference between these two methods of ectopic Shh expression is the timing of delivery. In Shh E9.5vir , ectopic Shh activity begins at E9.5, after the telencephalon has begun to form. In ShhN/+, increased Shh is co-expressed with endogenous Shh concurrent with formation of the telencephalon. These data suggest that the ability of Shh to prevent or expand cortical hem and plexus may be determined by changes in source, level, or timing of exposure to Shh.

ShhN/+
In a direct test of the effect of timing of Shh exposure on dorsal midline patterning, Shh viral infections were performed at E10.5 (Shh virE10.5 ). Several interesting differences are detected when dorsal midline patterning is compared in Shh virE9.5 and Shh virE10.5 brains. While the roof plate expands in response to Shh at both delivery time points, opposite effects occur on the hem, which expands after E10.5 Shh exposure rather than diminishing (E9.5). In the Shh virE10.5 telencephalon, morphological appearance of the choroid plexus occurs without activation of the plexus-specific gene rTTR, suggesting that choroid plexus development is still sensitive to Shh activity at this time. Differential alteration of dorsal midline patterning as a result of different methods of increased Shh activity supports the idea that the dorsal

Wnt3a rTTR Lhx5
F midline, and particularly the roof plate, is a sensitive responder of Shh signaling. It has previously been proposed that Shh influences dorsal telencephalic midline patterning through a Gli3 gradient or FGF signaling rather than through direct signaling [21,22,33]. However, the sensitivity of the dorsal telencephalic midline to ectopic Shh signaling suggests that Shh can directly influence patterning of this region. In addition, decreased Shh signaling in mice lacking both Gli1 and Gli2 results in dorsal telencephalic midline defects similar to that found when Shh is increased (Shh E9.5vir ): loss of cortical hem and choroid plexus, with a concomitant expansion of the roof plate gene Lhx5. Therefore, both increased and decreased Shh signaling can have similar effects on telencephalic dorsal midline patterning.

FGF Antagonist
A shared aspect between Gli1/2 -/-mutants, Shh E9.5vir , Shh E10.5vir and ShhN/+ is Lhx5-expressing roof plate expansion. Cell autonomous Shh activation in the roof plate (GDF7cre-ActSmo) also results in roof plate and hem expansion with minimal plexus expansion, supporting a direct role of Shh in maintaining roof plate/hem/ plexus boundaries. These data further support previous data suggesting a cell-non-autonomous role of the roof plate, presumably mediated by BMPs, influencing choroid plexus formation [5]. Together, these data show that the roof plate is particularly sensitive to Shh signaling, requiring a balanced level of Shh activity for proper morphology and gene expression. Interestingly, mutations in Shh or its downstream targets result in the developmental defect holoprosencephaly [19,42], and interference in roof plate function has been linked to a milder form of the disease affecting dorsal hemispheric separation [43][44][45]. Thus, our data suggests that Shhmediated roof plate effects may be responsible for the dorsal midline defects found in holoprosencephaly.
We show that both increases and decreases in Shh signaling ventrally (ShhN/+ and Gli1/2 -/-) and dorsally (wtShh and ShhN injections) significantly alter dorsal midline gene expression. The current model predicts that Shh acts through FGF8 to influence the telencephalic dorsal midline [18,33]. However, the dorsal midline phenotype obtained with reduced FGF signaling and altered Shh signaling may be distinct. In particular, roof plate expansion of Lhx5 is common to both reduced and increased Shh signaling, but absent from reduced FGF signaling. It is possible that the failure to observe common roof plate defects between FGF inhibition and Shh alterations is due to inefficient FGF reduction. However, an argument against this is that the FGF inhibitor increases rather than prevents Wnt3a expression in the hem, as observed with altered Shh signaling. If FGF inhibition is inefficient, the phenotype obtained would be expected to be a subset of the complete phenotype rather than an opposing phenotype. Another argument against inefficient FGF inhibition is the appearance of obvious dorsal midline morphological defects in response to the FGF inhibitor. If FGF inhibition were inefficient, gene expression boundaries may be changed without obvious morphological defects. Although further manipulations are necessary to definitively rule out Shh action through FGFs in the dorsal midline, the present data support the idea that Shh acts directly rather than through FGF signaling to maintain boundary relationships in the dorsal telencephalic midline. Experiments in this paper indicate that the telencephalic dorsal midline, specifically the roof plate, is a direct Shh target. What is the most likely source of Shh that directs telencephalic dorsal midline formation? During the time that mouse telencephalic dorsal midline boundaries are established (E10.5-E12.5), the Lhx5 expression domain in the roof plate becomes juxtaposed to Shh expression in the ZLI. This is most apparent at E12.5 ( Figure 1S, T). The ZLI is a strip of cells that act as a signaling center to induce expression of regionally restricted transcription factors in flanking areas [23,46].
The ZLI demarcates the boundary between the prethalamus and the functionally distinct thalamus and a gradient of Shh expression from the ZLI is required for establishing regional identity in these bordering structures [47]. While DiI labeling revealed that the Shhexpressing ZLI initially forms from cells in the alar plate, both Shh expression in the ZLI and subsequent diencephalic development depend on a source of Shh secreted from the basal plate of the forebrain [46]. Taken together with results in this paper, reduced or increased Shh activity resulting from mutations in Shh or its downstream targets would be predicted to affect ZLI formation, and ultimately influence patterning and/ or maintenance of the telencephalic dorsal midline.

Conclusions
Our data suggests that the telencephalic roof plate is a sensitive responder to changes in Shh activity, and that its expansion occurs in response to increased or decreased Shh signaling. We show that roof plate expansion in response to changes in Shh signaling occurs at the expense of the hem and plexus or can lead to expansion of the hem and plexus. In addition, responsiveness of the telencephalic dorsal midline to Shh signaling changes between E9.5 and E10.5 in mice. The structural relationship of Shh and dorsal midline markers raises the possibility that dorsally expressed Shh from the diencephalic ZLI is a source of Shh signaling to the telencephalic dorsal midline. We propose that imbalanced Shh signaling directly changes dorsal midline boundaries between the roof plate, choroid plexus and cortical hem.

Animals and Surgery
ShhN floxed mice [48] were obtained from Dr. A. McMahon, and crossed to EIIAcre mice (Jackson Lab) to generate ShhN/+ embryos at E12.5. ActSmo floxed mice [38] were crossed to GDF7cre mice [6] to generate GDF7cre-ActSmo E12.5 embryos. Animals used in these studies were maintained according to protocols approved by Institutional Animal Care and Use Committee at Children's Memorial Hospital Research Center. Timedpregnant Swiss Webster mice used for injections were obtained from Taconic breeding laboratories. Embryonic day 0.5 is defined as noon of the day a vaginal plug was found after overnight mating. Detailed animal care, preparation for surgery and the use of the ultrasound scanner (UBM scanner) have been previously described [49].