Several clinically important classes of neurons (e.g., the oculomotor complex and midbrain dopaminergic neurons) reside within the ventral midbrain of vertebrates (Agarwala and Ragsdale,2002). Recent advances in stem cell technology and breakthroughs in identifying the transcriptional cascades involved in specifying midbrain dopaminergic neurons have renewed interest in understanding how overall midbrain patterning is accomplished (Bayly et al.,2007; Agarwala et al.,2001; Simeone,2005; Blaess et al.,2006). Two major signaling centers, the midbrain–hindbrain boundary (MHB), and the floor plate have been implicated in establishing midbrain pattern (Marti et al.,1995; Joyner,1996; Blaess et al.,2006). The midbrain/rostral floor plate (rFP) is a source of the signaling molecule Sonic Hedgehog (SHH), whose role in specifying ventral neural cell-fates is well established (Chiang et al.,1996; Jessell,2000; Agarwala et al.,2001; Fedtsova and Turner,2001; Blaess et al.,2006).
The core Hedgehog (HH) signaling cascade is conserved across species and is relatively well understood (Ingham and McMahon,2001; Lum et al.,2003; Hooper and Scott,2005; Huangfu and Anderson,2006). Of the three known HH ligands (SHH, Indian [IHH], and Desert [DHH] Hedgehog), only SHH has been shown to play a role in neural tube patterning among higher vertebrates (Bitgood and McMahon,1995; Zhang et al.,2001). The HH ligand binds to its receptor PTC, a 12-pass transmembrane protein (Marigo et al.,1996; Stone et al.,1996). PTC acts both as a receptor and a negative regulator of HH signaling by maintaining a constitutive block on the transmembrane protein SMO (Akiyama et al.,1997; Alcedo and Noll,1997). In the presence of the ligand, the blockade on SMO is lifted and downstream HH signaling occurs by means of the action of the GLI/Ci family of transcription factors (Ohlmeyer and Kalderon,1998; Methot and Basler,2001).
In addition to PTC1, the identity of several other HH-binding proteins, which can positively (Dally-like proteins, CDO, BOC) or negatively (HHIP, PTC2, BOC) regulate the HH signaling cascade are known (Chuang and McMahon,1999; Schweitzer et al.,2000; McCarthy et al.,2002; Tenzen et al.,2006; Yao et al.,2006). Further regulation of the HH signaling cascade occurs at the level of GLI transcription factors (Jiang,2002). Several genes (e.g., SUFU, DZIP) complex with the GLI/Ci genes and are involved in the nucleocytoplasmic shuttling and cytoplasmic sequestration of the latter (Kogerman et al.,1999; Murone et al.,2000; Sekimizu et al.,2004; Vokes and McMahon,2004; Wolff et al.,2004; Huangfu and Anderson,2006; Svard et al.,2006). In addition, some genes (e.g., BTRC/SLIMB) that are not solely dedicated to HH signaling also regulate this signaling cascade by targeting GLI genes for hyperphosphorylation and eventual degradation (Jiang,2002).
The midbrain is an ideal system in which to study the HH signaling cascade. During development, the ventral midbrain is organized into a longitudinal array of cell-fates (midbrain arcs) specified at specific distances by a graded SHH signal emanating from the rFP (Agarwala et al.,2001). Each midbrain arc expresses a unique set of homeobox transcription factors, thus providing a precise readout for understanding the role of HH signaling in the midbrain (Agarwala et al.,2001; Sanders et al.,2002). We have also shown that midbrain nuclei (e.g., dopaminergic neurons, red nucleus, oculomotor complex) can develop within the context of midbrain arcs (Agarwala and Ragsdale,2002). Finally, overexpression and HH blockade experiments suggest a requirement for HH signaling, not only in specifying ventral midbrain cell fates, but also in establishing a spatially coherent pattern of arcs, in midbrain size regulation and in maintaining midbrain boundaries, for example the MHB (Bayly et al.,2007; Ishibashi and McMahon,2002; Blaess et al.,2006).
How do the effectors of the HH signaling cascade establish these patterning parameters? To understand how the HH signal is regulated, it is critical to determine the spatial and temporal expression patterns of the key modulators of HH signaling. In this study, we have cloned out novel HH pathway genes and analyzed the expression pattern of 14 genes involved in the regulation of the HH signaling cascade between Hamburger and Hamilton (H&H) stage 3+ and embryonic day (E) 7. The expression patterns of HH pathway genes corroborate our previous findings that HH signaling is likely to be required very early and briefly in the ventral midline, followed by a sustained requirement in the intermediate regions of the ventral midbrain flanking the rFP SHH source (Bayly et al.,2007). Of interest, several HH-binding factors (MEGALIN, BOC) and HH effectors (SMO, GLI1, 2, 3, DZIP) are expressed in the dorsal midbrain and the MHB. These expression patterns thus provide a basis for understanding recent reports that implicate HH signaling in the regulation of dorsal midbrain structures (e.g., inferior colliculus) and the MHB (Bayly et al., unpublished observations; Aoto et al.,2002; Ishibashi and McMahon,2002; Blaess et al.,2006). Finally, we report for the first time the expression of IHH in the intermediate regions of the spinal cord, but not the ventral midbrain between H&H stage 18 and E4.
RESULTS AND DISCUSSION
The chick midbrain is organized into a transient series of longitudinal territories termed the midbrain arcs (Supplementary Figure S1, which can be viewed at http://www.interscience.wiley.com/jpages/1058-8388/suppmat) (Sanders et al.,2002). They appear in mediolateral (ML) sequence between H&H stage 12 and E5 in response to HH signaling (Bayly et al., unpublished observations; Agarwala et al.,2001; Sanders,2001; Agarwala and Ragsdale,2002). To understand how the HH signal is interpreted by its downstream effectors to produce the midbrain arc pattern, we have examined the expression of 14 HH pathway genes between H&H stage 3+ and E7. Their spatiotemporal expression patterns are reported below.
Midbrain/rFP Expression of SHH Begins at E1 and Continues to E7
SHH expression is seen within axial mesendoderm (Hensen's node, anterior primitive streak, prechordal plate) and ectoderm by H&H stages 3–5 (Fig. 1A–C; Hamburger and Hamilton,1951; Lawson et al.,2001; Patten et al.,2003). Between H&H stages 7 and 10 (3–10 somites), SHH is seen throughout the ventral midline of the neural tube and within axial mesoderm (Fig. 1D–G; data not shown; Lawson et al.,2001). Although the SHH source in the midbrain is linear at these stages (Fig. 1D,E), it broadens rostrally by H&H stage 13 (e.g., Figs. 1H, 3E). By H&H stage 15, the rFP displays a distinct medial region expressing high levels of SHH and a lateral region where lower levels of SHH expression can be found (Figs. 1H, 5F). SHH continues to be expressed in the rFP at least until E7 (Fig. 1I,J). Thus, SHH expression is dynamic temporally as well as spatially, with different levels of expression along the dorsoventral (DV) as well as the anteroposterior (AP) axes (Fig. 1H–J). Our previous HH blockade experiments suggest that the requirement for HH in ventral midbrain cell-fate specification is extinguished in almost all regions of ventral midbrain by H&H stage 13 (Bayly et al.,2007). The continued expression of HH until E7 may, therefore, have a role in cell survival, midbrain size regulation, dorsal patterning, or axon guidance (Ishibashi and McMahon,2002; Blaess et al.,2006; Cayuso et al.,2006; Okada et al.,2006; Stoeckli,2006).
Dynamic Modulation of SMO Expression in Ventral Midbrain Arcs and in Tectum
SMO is an obligate component of the HH signaling cascade (Stone et al.,1996; Alcedo and Noll,1997). In agreement with the early expression of SHH, broad SMO expression is seen throughout the presumptive neural plate, head process, Hensen's node, and the primitive streak by H&H stage 5 (Fig. 2A; data not shown; Meier and Jacobson,1982). At H&H stage 10, strong SMO expression is seen in the dorsal and ventral midbrain as well as the notochord (Fig. 2B,C). Strong expression of SMO continues in the midbrain through H&H stage 16 (Fig. 2D). By E5, SMO expression displays an arcuate pattern within ventral midbrain and strong expression within dorsal midbrain/tectum (Fig. 2E).
HH Binding Proteins in the Midbrain
PTC1 is a receptor and a direct transcriptional target of HH and its expression is considered to be a reliable readout of HH signaling (Marigo and Tabin,1996; Akiyama et al.,1997; Alcedo and Noll,1997; Goodrich et al.,1997). Although the expression of PTC1 in axial mesendoderm begins after the onset of SHH expression, low levels of PTC1 expression coincide with SHH expression within ventral neural tube by H&H stage 7 (Figs. 1A, 3A; data not shown; Lawson et al.,2001; Pearse et al.,2001; Dathe et al.,2002; Patten et al.,2003). By H&H stage 8, strong PTC1 expression is seen along the ventral midline overlapping with the expression of SHH (Figs. 1D, 3A; data not shown). By H&H stage 10, PTC1 expression is down-regulated at the ventral midline of the prospective midbrain and is instead expressed strongly in stripes flanking the SHH+ rFP (Fig. 3B,B′,C). This pattern of PTC1 and its relationship to SHH remains essentially unchanged between H&H stage 10 and E5 (Fig. 3B–G). Of interest, from H&H stage 10 to E5, PTC1 expression is also seen in the MHB where it overlaps with FGF8 expression medially (arrowheads, Fig. 3B″, D′). Detectable levels of PTC1 are not noted in the dorsal midbrain at any stage examined (data not shown).
Gene expression of PTC2 precedes PTC1 expression within Hensen's node and the head process/notochord at H&H stage 5 (Fig. 3H; data not shown; Pearse et al.,2001). The expression of PTC2 is stronger than that of PTC1 between H&H stages 5–10 and overlaps with that of SHH more extensively (compare Figs. 3B and 3I). Thereafter, the expression of PTC2 resembles that of PTC1 and circumscribes the SHH expression domain until E5 (Fig. 3J,K).
The HH-binding protein HHIP is typically expressed in cells near the HH source and has a role in attenuating the HH signal (Chuang and McMahon,1999). The onset of HHIP expression occurs relatively late in midbrain development (H&H stage 14+) and overlaps with the band of strong PTC1 and PTC2 expression flanking the SHH+ rFP (Fig. 3L,M; data not shown).
MEGALIN/LRP-2 is a HH-binding protein found on the apical surface of neurepithelial cells (McCarthy et al.,2002). It is involved in the endocytosis and/or transcytosis of multiple ligands, including SHH, Retinol, and bone morphogenetic protein (BMP; McCarthy et al.,2002; Spoelgen et al.,2005). In the midbrain, low levels of MEGALIN expression are seen by H&H stage 10+ (11 somites; data not shown). By H&H stage 17, strong MEGALIN expression is seen in the ventral midbrain and weak expression is noted in the tectum (Fig. 4A). By E5, MEGALIN expression is mainly seen within the medial region of the rFP, the MHB, and the roof plate (Fig. 4B,C).
BOC and a related protein, and CDO are a novel class of HH-binding proteins belonging to the Immunoglobulin/fibronectin superfamily (Mulieri et al.,2002; Tenzen et al.,2006; Yao et al.,2006). In the mouse, boc expression is present within dorsal spinal cord and in commissural axons (Mulieri et al.,2002; Okada et al.,2006; Tenzen et al.,2006). BOC expression is similarly confined to the tectum in the chick when it is first seen at H&H stage 13 (Fig. 4D). Strong BOC expression continues to be seen in dorsal midbrain through E5 (Fig. 4E,F).
Taken together, multiple HH-binding proteins are expressed in the chick midbrain, including PTC1, PTC2, HHIP, MEGALIN, and BOC. However, the onset of expression of all of these genes occurs after H&H stage 3, when SHH expression is already present within the embryo (Lawson et al.,2001). Thus, whether and how HH signaling is mediated at this early time point is not yet known.
Of interest, the genes for several HH-binding proteins (BOC, GAS1, MEGALIN) are expressed in the dorsal midbrain (Fig. 4; data not shown). Of these, MEGALIN has been implicated in the endocytosis/transcytosis of multiple ligands, including HH, Retinol, and the BMPs, which are expressed in midbrain roof plate (RP; Spoelgen et al.,2005; data not shown). In addition, BOC and MEGALIN may both have a role in the sequestration of the HH ligand in dorsal midbrain (McCarthy et al.,2002; Tenzen et al.,2006).
Surprisingly, several HH-binding proteins (MEGALIN, PTC1), GLI 1, 2, and 3,and the GLI regulator DZIP are expressed in the MHB (Figs. 3D,D′, 3G, 4B, 6G). A role for HH signaling in the maintenance of fgf8 expression in the MHB has recently been established in the Shh−/− mouse (Fogel and Agarwala, unpublished observations; Aoto et al.,2002; Blaess et al.,2006). In the mouse limb, the maintenance of fibroblast growth factor-8 signaling in the apical ectodermal ridge by SHH (derived from the zone of polarizing activity) is indirect, by means of the action of the BMP antagonist Gremlin (Khokha et al.,2003). However, the presence of multiple HH effectors makes it possible that the maintenance of the MHB could in part be directly mediated by SHH, particularly around E5–E6 when multiple HH pathway genes are expressed within the MHB.
GLI Gene Expression in the Midbrain
GLI1 expression is seen in the neural plate by H&H stage 7 (Fig. 5A). Between H&H stages 7 and 10, graded expression of GLI1 is noted in the ventral midbrain, with low levels of GLI1 within the ventral midline and high levels in the flanking regions (Fig. 5A–C). Between H&H stage 9 and 10+, GLI1 expression is cleared from the ventral midline, although it continues to be expressed in regions flanking the SHH domain until E3 (Fig. 5B,E,F). Surprisingly, GLI1 is present within the dorsal midbrain at H&H stage 10, where it partially overlaps with the dorsal marker PAX7 (Fig. 5C; Supplementary Figures S2, 3). Dorsal expression of GLI1 at the DV boundary and caudal tectum can also be seen at E5 (Supplementary Figure S4).
Of interest, expression of GLI1 is noted within the MHB continuously between H&H stages 10 and E5 (Fig. 5D,G,H). By late E5, GLI1 expression is entirely confined to the MHB where it coincides with FGF8 expression (Fig. 5H; Supplementary Figure S5). Thus, a very complex and dynamic modulation of GLI1 expression is noted within the ventral midbrain, where, consistent with the expression of PTC1, PTC2, and HHIP (Fig. 3), the highest levels of GLI1 are seen in the region flanking the SHH+ rFP and not the ventral midline. Furthermore, the expression of GLI1 in tectum and the MHB supports a role for GLI1 in the patterning of these regions, especially around E5.
The expression of GLI2 is first seen in the embryo at H&H stage 4 (Granata and Quaderi,2005). By H&H stage 6, strong GLI2 expression is seen in the ectoderm and weak expression is seen within axial mesendoderm (Fig. 5I,J). Of interest, only scattered GLI2+ cells are seen within the midline overlapping with the mesendodermal expression of SHH (Fig. 5I,J). Low GLI2 expression is seen throughout the ventral midbrain at H&H stage 9 (eight somites), but it is almost entirely cleared from the ventral midline by nine somites (data not shown). At H&H stage10, GLI2 is expressed in tectum, although it is confined to low levels of expression in the dorsal midline (roof plate), where it is only present in the most ventricular cells (Fig. 5K,L). At H&H stage 15, GLI2 expression resembles that of GLI1 and flanks the SHH+ rFP (Fig. 5M). This complementary relationship with SHH is maintained through E5 (Fig. 5N,O). As with GLI1, GLI2 is also present within the MHB between H&H stage 10 and E5 (Fig. 5N,W; Supplementary Figure S6). Taken together, these results suggest an overlapping role for GLI1 and GLI2 in early patterning of the ventral midline (before H&H stage 10) and a continued role in patterning intermediate and dorsal regions of the midbrain and the MHB at least until E5.
GLI3 expression is seen in the embryo by H&H stage 5 and in the neural tube by H&H stage 8 (data not shown; Schweitzer et al.,2000; Granata and Quaderi,2005). At H&H stage 9, GLI3 expression is confined to intermediate and dorsal regions of the neural tube (Fig. 5P–R). By H&H stage 15, graded expression of GLI3 is seen with low levels of expression ventrally and strong expression is seen in the tectum (Fig. 5S). By E5, GLI3 is largely confined to dorsal midbrain (Fig. 5T). Of interest, GLI3 expression is prominently seen in the MHB (Fig. 5U–W) and in the tectum, where it partially overlaps with GLI2 and GLI1 (Fig. 5H,T–V; Supplementary Figs. S4–6). Finally, at late E5, a domain of GLI3 expression is seen in ventral midbrain, which is more pial and ventral to GLI2 expression (Fig. 5W, arrowhead).
These results corroborate our findings of an early requirement (before H&H stage 11) for HH signaling at the ventral midline (Bayly et al.,2007). Based on PTC and GLI expression and our HH blockade studies, these results also suggest that the highest and the most prolonged HH signaling within ventral midbrain occurs at intermediate regions of the ventral midbrain flanking the ventral midline. Of interest, our results implicate GLI1 in dorsal patterning both early (∼H&H stage 10) and ∼E5 (Supplementary Figures S2–S4). Finally, a role for HH signaling at the MHB is indicated by the expression of all GLI genes, PTC1, MEGALIN, and DZIP (Figs. 3G, 4B, 5H,U–W, 6J) within this region.
Gli Modulators in the Midbrain: SUFU, DZIP/IGUANA, and BTRC
SUFU complexes with GLI proteins in the cytoplasm and participates in their cytoplasmic sequestration and nucleocytoplasmic shuttling (Lum et al.,2003). SUFU is ubiquitously expressed within the presumptive neural tissue by H&H stage 5 (Fig. 6A). At H&H stage 8, SUFU is present throughout the anterior neural tube, including the prospective midbrain (Fig. 6B). By H&H stage 10, SUFU is present throughout the AP and DV extent of the neural tube (Fig. 6C). This pattern of SUFU expression continues until E5, when the SUFU expression becomes arcuate, with strong expression in Arcs 1 and 2 and the midbrain tectum (Fig. 6D,E).
DZIP expression is present throughout the neural tube by H&H stage 13 (Fig. 6F; Supplementary Figure S7). By E5, DZIP expression is down-regulated and is present at low levels within the medial regions of rFP and the MHB (Fig. 6G).
BTRC is involved in the hyperphosphorylation of the GLIs targeting the latter for further phosphorylation and degradation in the proteasome (Jiang,2002). Low levels of BTRC is first seen in the midbrain at H&H stage 10+, when it is expressed everywhere except the ventral midline (Fig. 6H). At E4, a similar pattern is seen, although the expression of BTRC is much stronger (Fig. 6I). As with SUFU, BTRC expression is prominent in Arc 1 and Arc 2 at E5 (Fig. 6J). Compared with early embryonic stages (<H&H stage 10), the strong expression of BTRC and SUFU, with lowered levels of PTC1, GLI1, and SMO (Figs. 2E, 3G, 5G), suggest diminished HH signaling within medial rFP and Arc 1 and 2 around E5, although further investigations are required to confirm our gene expression data.
HH Effectors in the Spinal Cord
We examined the gene expression of the principal effectors of HH signaling in the spinal cord for comparison with the midbrain. At H&H stage 10, when patterning is under way in the ventral spinal cord, SHH is strongly expressed in the floor plate (FP) and in the notochord (Fig. 7A; Ericson et al.,1996). Graded expression of PTC1 is noted in the spinal cord at this stage with the strongest and most uniform expression occurring in the region immediately dorsal to the FP (Fig. 7A,B). Interestingly, strong, but punctate expression of PTC1 is noted in select cells of the FP and the notochord at this time (Fig. 7A,B). The expression of HHIP is similar to that of PTC1, although it is more circumscribed (Fig. 7C).
Nested expression of GLI1, GLI2, and GLI3, with considerable overlap among their expression domains, is seen in the spinal cord at H&H stage 10 (Fig. 7D–F). Furthermore, the ventral limits of the domains of GLI1, 2, and 3 are progressively more dorsal, similar to the pattern of expression in the mouse spinal cord (Fig. 7D–F; Hui and Joyner,1993; Ding et al.,1998; Matise et al.,1998). Interestingly, the expression of GLI1 is punctate like PTC1 and extends over much of ventral and dorsal (PAX7+) spinal cord (Fig. 7D,G). In the spinal cord, the presence of GLI1 expression and the absence of the dorsal marker PAX7 are both reliable readouts of HH signaling (Ericson et al.,1997; Bai et al.,2004). Surprisingly, their expression coincides over a considerable expanse of dorsal spinal cord (Fig. 7D,G). The dorsally located GLI1 may be sequestered within the cytoplasm and, thus, has no activator function (Wang and Holmgren,2000). Alternatively, it could be involved in mediating HH action in the ventral most region of dorsal spinal cord where HH signaling is required for the specification of dorsal (DBX+, d5) interneurons (Wijgerde et al.,2002). In either case, the coexpression of GLI and PAX7 suggests a more complex interaction between the HH signaling cascade and dorsal patterning than previously supposed.
At stage 10, SMO expression extends throughout the neural tube, exhibiting a punctate expression pattern in the FP and the NC and the RP (Supplementary Figure S8). Interestingly, gene expression of IHH is seen in the chick ventral spinal cord at H&H stages 18–19 and persists at least until E4 (Fig. 7H,I). The expression of IHH is dorsal to that of SHH and does not overlap with the latter (Fig. 7H,I). The function of IHH in the chick spinal cord is currently not known and will require further investigation. No IHH expression is observed in the chick midbrain (data not shown).
Fertilized Leghorn eggs (Ideal Poultry, Texas) were incubated at 38°C in a forced-draft humidified chamber. Embryos were staged according to Hamburger and Hamilton (Hamburger and Hamilton,1951). Early anatomical structures were identified according to Schoenwolf, Bellairs and Osmond, and Hamburger and Hamilton (Hamburger and Hamilton,1951; Schoenwolf,1995; Bellairs and Osmond,2005).
In Situ Hybridization
Embryos were harvested between E1 and E7 and immersion-fixed in 4% paraformaldehyde. Digoxigenin- or FL-conjugated antisense riboprobes were prepared from chick cDNAs for BTRC/SLIMB, BOC, DZIP, GLI1, GLI2, GLI3, FGF8, HHIP, IHH, MEGALIN, PTC1, PTC2, SHH, SMO, and Suppressor of Fused (SUFU). One- or two-color whole-mount in situ hybridizations were conducted using NBT (Roche) or T-NBT (Research Organics, Ohio) tetrazolium histochemistry according to previously established protocols (Agarwala et al.,2001; Agarwala and Ragsdale,2002).
Cloning of HH Pathway Genes
The cloning of BTRC, DZIP and HHIP, and SUFU has been described previously (Agarwala et al.,2005). Chick gene fragments for BOC were isolated by polymerase chain reaction from E6 chick ventral midbrain cDNA and subcloned into pCR-TOPO (Invitrogen). Primers were designed with Mac Vector 7.2 (Accelrys) based on sequences obtained from BLAST searches of chick expressed sequence tag and genomic databases (Boardman et al.,2002; Wong et al.,2004; Caldwell et al.,2005). The primer pair successfully used for cloning BOC cDNA was BOC-forward, 5′-GGAGCAGCATTCATCTTCTACAACG-3′; BOC-reverse, 5′-TGGTCACAGGCTTTGGTGGACTAC-3′.
We thank Roy D. Bayly and Lasse Pederson for cloning out the BOC cDNA and Dr. A. Jacobson for help with anatomical identification. We thank P. Beachy, P. Brickell, J. Briscoe, C. Cepko, D. Cleveland, G. Eichele, C. Fan, C. Goridis, M. Goulding, B. Houston, T. Jessell, J. Lahti, A. Leutz, J. Lewis, C. Logan, A. McMahon, G. Martin, C. Ragsdale, J. Rubenstein, G. Struhl, C. Tabin, and M. Wassef for DNA reagents; and Roy Bayly for critical reading of the manuscript. This research was supported by University of Texas at Austin start up funds, and S.A. was funded by the NIH.