Sonic hedgehog (Shh) is a signaling molecule critically involved in many aspects of embryonic development (McMahon et al.,2003). Shh is a morphogen, providing positional information to cells in a concentration-dependent manner, and is crucial for the patterning of neural cell fates in the developing central nervous system, as well as the control of digit pattern in the developing limbs. Signaling through the hedgehog (Hh) pathway is initiated when a Hh ligand binds to the transmembrane receptor Patched-1 (Ptc1), which relieves repression of another transmembrane protein, Smoothened (Smo). Activated Smo then allows the Gli transcription factors (Gli2 and Gli3) to activate transcription of target genes, such as Ptc1 and Gli1. In the absence of ligand, the full length Gli3 transcriptional activator (Gli3A) is proteolytically processed to form a transcriptional repressor (Gli3R). During neural tube patterning Gli3 functions as a repressor required to restrict cell fate in the intermediate region along the dorsoventral (D-V) axis of the developing neural tube (Persson et al.,2002), and as an activator, which is required along with Gli2 to induce cell fates in the ventral neural tube (Bai et al.,2004). Gli2 acts predominantly as a transcriptional activator, required for the specification of the most ventrally located neural cells (Ding et al.,1998; Matise et al.,1998).
The morphogenetic role of Shh has been established through studies of limb and neural tube morphogenesis. In limb buds, Shh is produced in cells located within the zone of polarizing activity located at the posterior margin, which results in posterior-most cells being exposed to the highest concentrations for longer periods of time than cells in more anterior positions (Riddle et al.,1993; Harfe et al.,2004; Scherz et al.,2007). This localized Shh production antagonizes the formation of Gli3R in the posterior limb bud creating an anterior–posterior (A-P) gradient of Gli3 activity with Gli3R levels highest in the anterior limb bud (Wang et al.,2000). It has been demonstrated that this antagonism between Shh and Gli3 is essential for proper limb patterning (Litingtung et al.,2002; te Welscher et al.,2002). During neural tube development Shh is produced at the ventral midline in the floor plate and notochord, creating a ventral-to-dorsal gradient such that ventral neural cells are exposed to higher levels of Shh ligand than dorsal cells (Briscoe and Ericson,1999). It is proposed that this gradient results in the establishment of several domains of progenitor cells along the D-V axis that express different combinations of homeodomain transcription factors, giving rise to different neural cell types (Briscoe et al.,2000). Class I genes which are defined as being repressed by high levels of Shh signaling, such as Pax6 and Pax7, are expressed laterally and dorsally respectively, while Class II genes, which are defined as being induced by Shh, such as Nkx2.2, and Nkx6.1, are expressed ventrally. Deregulation of components of the Shh signaling network can have drastic effects on these patterning processes.
Mutations leading to inappropriate pathway activation, such as targeted disruptions of Ptc1, SuFu, and Fkbp8 or mutations in Rab23 or Thm1 genes have been shown to cause polydactyly (in cases where embryos survive long enough to develop digits), and severe neural tube defects characterized by a dorsal expansion of ventral neural cell identity (Goodrich et al.,1997; Eggenschwiler et al.,2001; Bulgakov et al.,2004; Svard et al.,2006; Tran et al.,2008). In contrast, disruption of Shh results in loss of all digits in the forelimb and all but the anterior-most digitin the hindlimb of mouse and chick (Chiang et al.,2001; Wijgerde et al.,2002; Ros et al.,2003). Furthermore, loss of Shh or Smo results in a lack of ventral neural tube cell types (Chiang et al.,1996; Wijgerde et al.,2002).
Recently, novel and unexpected regulators of mammalian hedgehog signaling have been identified, such as proteins involved in primary cilia formation and function, as well as vesicle transport, illustrating the complexity of this signaling network and divergence of the mammalian pathway from that of Drosophila (Huangfu and Anderson,2006). Tubby-like protein 3 (Tulp3) is a member of small gene family consisting of Tub and Tulps1–3 generally implicated in the function and maintenance of neural cells (Ikeda et al.,2001,2002a; Carroll et al.,2004). Although a definitive function for these proteins has not yet been established, studies have suggested potential roles as a transcription factor, and in vesicle trafficking (Boggon et al.,1999; Mukhopadhyay et al.,2007). Mutations in the parent Tub gene result in adult-onset obesity and retinal and cochlear degeneration in mice (Kleyn et al.,1996; Noben-Trauth et al.,1996), and mutations in Tulp1 have been associated with retinal degeneration in mice and humans (Hagstrom et al.,1998; Ikeda et al.,2000). Tulp3, however, is the only member of this family that has been shown to play a critical role during embryonic development, as Tulp3−/− embryos die during gestation by embryonic day (E) 14.5 and develop various abnormalities such as exencephaly, spina bifida, and facial clefting (Ikeda et al.,2001).
In this study, we present evidence for a novel and critical functional role of Tulp3 in the regulation of the mouse Hh pathway. Tulp3−/− mice develop abnormalities characteristic of deregulated Shh activity such as polydactyly, exencephaly, and ventralization of neural progenitors in the developing neural tube. We further show that signaling through the Shh pathway takes place in the absence of Shh ligand in Tulp3 mutants. Tulp3/Shh compound-null embryos express Shh target genes and develop multiple digits. Taken together, these findings indicate that Tulp3 functions as a key suppressor of the Hh pathway.
RESULTS AND DISCUSSION
Increased Shh Signaling in Tulp3−/− Embryos
It has been previously reported that loss of Tulp3 caused severe neural tube defects in both cranial and caudal regions (Ikeda et al.,2001). We have found that these mice also develop preaxial polydactyly (Fig. 1A,B), and a phenotype which strongly resembles that resulting from mutations in negative regulatory factors of the Shh pathway, such as Rab23, Fkbp8, and Thm1 (Eggenschwiler et al.,2001; Bulgakov et al.,2004; Tran et al.,2008). To assess the effect of loss of Tulp3 on Shh signaling, we used whole-mount in situ hybridization (WMISH) to examine the expression of Ptc1, a direct transcriptional target of the Shh pathway. At E9.5 and E10.5 increased Ptc1 expression was detected in the caudal neural tube (Fig. 1C,D), craniofacial region (Fig. 1E,F), and forelimbs (Fig. 1G,H) of mutants as compared to wild-type, indicating that activation of the Shh pathway is increased in Tulp3−/− mice. Expression of Ptc1 in the Tulp3−/− hindlimbs did not appear to be affected at this stage (Fig. 1I,J).
Previous data gathered by following expression of a knock-in Tulp3-green fluorescent protein (GFP) fusion protein, along with in situ hybridization, showed that Tulp3 is expressed throughout the neuroepithelium, and is found ubiquitously between E8.5 and E14.5 (Ikeda et al.,2001). Consistent with these results, we have detected Tulp3 transcripts ubiquitously and throughout the A-P axis of the developing limb buds (Fig. 2). Thus, Tulp3 is expressed in tissues recognized as targets of Hh signaling during development.
Loss of Tulp3 Causes Ventralization of the Caudal Neural Tube
Shh plays a critical role in specifying neural cell fate within the ventral neural tube. During neural tube patterning, Shh is normally expressed in the notochord and floor plate where it diffuses dorsally, activating transcription of target genes in a concentration dependent manner (reviewed in Dessaud et al.,2008). On the other hand, the roof plate, which is a source of bone morphogenetic proteins (BMPs) and Wnts, is required for the generation of neural subtypes found in the dorsal neural tube (Lee et al.,2000). Genetic mutations that either disrupt or elicit inappropriate signaling from either the floor plate or roof plate can alter the pattern of neuronal cell fates along the D-V axis. The expression of floor plate and roof plate markers was examined in Tulp3 mutant embryos. At E9.5, the expression of the transcription factor Hnf3β, a marker of floor plate cells, was expanded dorsally in the caudal neural tube in Tulp3 mutants as compared to wild-type (Fig. 3A,B). Wnt1, which is normally expressed in the roof plate, was lost in the corresponding region of the Tulp3−/− embryo at this stage (Fig. 3C,D), suggesting that ventral cell types are inappropriately expanded dorsally. This phenotype is also consistent with other mutations that result in activation of Hh signaling, such as in the open brain mutant which disrupts the function of the Rab23 protein (Eggenschwiler and Anderson,2000; Eggenschwiler et al.,2001). Shh expression in E9.5 Tulp3 mutants resembled that of wild-type embryos (Fig. 3E,F), indicating that the expansion of ventral fates may not be dependent on Shh. At E10.5, Hnf3β remained expressed in more dorsal cells in Tulp3 mutant embryos, and Wnt1 expression was severely reduced or lost in the caudal neural tube (not shown). Shh expression at E10.5 was expanded dorsally in Tulp3 mutants as compared to wild-type (Fig. 3G,H). Because high levels of Shh from the notochord induce Shh expression in the floor plate (Marti et al.,1995; Roelink et al.,1995), this expansion of the Shh expression domain likely does not reflect a role for Tulp3 in the regulation of Shh transcription, but is instead the consequence of cells adopting a more ventral fate than their position would normally dictate. Taken together, these results suggest that Tulp3 plays a role in suppressing signaling through the Shh pathway in neural progenitor cells in the absence of ligand.
We next examined the expression of genes involved in the patterning of neural progenitor cell domains. Expression of Shh was first analyzed by in situ hybridization on sections through the caudal neural tube of E10.5 wild-type and Tulp3−/− embryos, and we again detected a dorsally expanded region of Shh expressing cells (Fig. 4A,B). Expression of the V3 interneuron marker Nkx2.2 is induced by high levels of Shh and is normally expressed in the ventral-most cells flanking the floor plate. In the neural tube of Tulp3 mutants Nkx2.2 was present in a dorsally expanded domain compared with wild-type (Fig. 4C,D). Nkx6.1, also induced ventrally by Shh, was similarly expressed in more dorsal cells and was expressed extensively along the D-V axis (Fig. 4E,F). HB9 is a homeodomain transcription factor which requires Shh for its expression and is specifically expressed in motor neuron progenitor cells found in a domain immediately dorsal to the V3 domain. We found that the region of HB9 expressing cells was also dorsally expanded in Tulp3−/− embryos as compared to wild-type (Fig. 4G,H). The expression domain of Pax6, which is inhibited by high levels of Shh, was shifted dorsally and was detected in cells at the dorsal midline, where Pax6 is normally excluded from the roof plate (Fig. 4I,J). Likewise, the expression domain of Pax7 showed a dorsal shift, being restricted to the dorsal-most cells (Fig. 4K,L).
Altered A-P Patterning of Developing Limb Buds in Tulp3 Mutants
Shh expressed in the polarizing region of the limb bud plays a critical role in the specification and development of digits (Riddle et al.,1993; Harfe et al.,2004; Scherz et al.,2007). In Tulp3 mutants at E10.5, the domain of Shh expressing cells was not tightly restricted to the posterior margin in forelimb buds and appeared disorganized and expanded into more anterior regions (Fig. 5A,B). Expression of Shh in the hindlimbs, however, did not appear to be affected at this stage (Fig. 5C,D). At E11.5, ectopic Shh expressing cells are still detected in the forelimbs (Fig. 5E,F), while a small domain of Shh-expressing cells was present at the anterior margin of the hindlimbs (Fig. 5G,H). Ptc1 expression was detected throughout the A-P axis of the forelimb at E10.5 and E11.5 indicating an increase in Shh pathway activation (Figs. 1G,H, 5I,J), while in the hindlimb an ectopic anterior expression domain was detected in Tulp3 mutants, consistent with the ectopic Shh expression in this area (Fig. 5K,L). Expression of the autopod marker Hoxd13, which is normally repressed by Gli3R in the anterior limb bud, was extended anteriorly in both fore- and hindlimbs at E11.5 (Fig. 5M–P), and was present along the entire A-P axis of the limb bud. Expression of Pax9 is normally restricted to the anterior limb bud and is dependent on the activity of Gli3R, such that its expression is lost in the Gli3Xt/Xt mutant and is expressed throughout the limb bud in Shh−/− mutants (McGlinn et al.,2005). In the Tulp3−/− limb bud, Pax9 was not detected in either fore- or hindlimbs at E11.5 (Fig. 5Q–T). The altered expression of Gli3-regulated genes could be the result of a loss of Gli3 function. However, we found Gli3 to be expressed similarly in wild-type and Tulp3 mutant embryos (Fig. 5U–X), indicating that the polydactyly phenotype is not due to a loss of Gli3 expression, but rather that Gli3 processing to the repressor form may be compromised. Taken together, these expression patterns suggest that there is an increase in activation of the Shh signaling pathway in cells along the A-P axis in the developing Tulp3−/− limbs, with cells in the anterior limb bud taking on a more posterior identity. These results are reminiscent of gene expression patterns observed in the polydactylous chicken mutant talpid2 in which an anterior expansion of normally posteriorly restricted genes is observed, but without a detectable change in Shh expression (Caruccio et al.,1999). However, a partially rescued Ptc1 mutant mouse that survives until E14, develops severe polydactyly and shows an ectopic anterior domain of Shh expression, similar to the phenotype observed here (Milenkovic et al.,1999). Therefore, the ectopic Shh expression seen in the Tulp3 mutant limb is also likely due to deregulated Shh signaling. Altered gene expression in the Tulp3−/− limbs is also distinct from the polydactyly phenotypes of both Gli3−/− limbs in which the Ptc1 expression domain is not expanded (Litingtung et al.,2002), as well as Ift88/polaris mutants which show a reduced level of Ptc1 expression resulting from an impairment of all Gli function (Liu et al.,2005).
Ligand-Independent Shh Signaling in Tulp3 Mutants
In order to determine whether there is a genetic interaction between Tulp3 and Shh, we examined the phenotype of Tulp3/Shh compound-null embryos. At E10.5, the double mutants resembled Tulp3 single mutants (Fig. 6A–D), indicating that Tulp3 is epistatic to Shh. Moreover, Tulp3/Shh compound mutants were found to express Ptc1, which is not detected in Shh mutants (Fig. 6E–H). Analysis of the neural patterning phenotype in Tulp3/Shh double mutants showed specification of ventral neural subtypes in the absence of Shh. Expression of Nkx2.2, which is lost in Shh mutants, is expressed in an expanded domain in Tulp3/Shh mutants, similarly to that observed in Tulp3 mutants (Fig. 6I–L). Similarly, HB9-positive motor neuron progenitors, which are not specified in Shh mutants, were detected in Tulp3/Shh null embryos (Fig. 6M–P). This epistasis analysis confirms that Tulp3 plays a key suppressive role in the hedgehog signal transduction pathway, and is required to silence signaling in the absence of ligand.
Loss of Shh causes severe defects in A-P patterning of the limbs, where null embryos display a single zeugopod element, and loss of all digits in forelimbs and all but one digit in the hindlimbs (Chiang et al.,2001). When Tulp3 was lost in a Shh mutant background, double mutants resembled Tulp3 mutants at E14 and formed multiple digits (Fig. 7A–D). Surprisingly, compound mutants were not found to be polydactylous, but instead developed five digits in both fore- and hindlimbs (Fig. 7E–H, forelimbs are shown). Ptc1 is expressed in Tulp3/Shh double mutant limbs at low levels compared with wild-type or Tulp3−/− embryos (Fig. 7I–L), unlike Gli3/Shh compound mutants which do not express Ptc1 (Litingtung et al.,2002). Embryos containing two copies of an engineered form of Gli3 that cannot be processed into its repressor form (Gli3P1-4/P1-4), express low levels of Gli1 in a Shh mutant background (Shh−/−Gli3P1-4/P1-4) (Wang et al.,2007). The low level Ptc1 expression in Tulp3/Shh double mutants, therefore, may also be the result of impaired Gli3 processing and an altered balance between Gli3R and Gli3A. The observation that Tulp3/Shh mutant limbs develop multiple digits as well as two zeugopod elements (Fig. 7P) clearly shows that Tulp3 plays a repressive role independently of Shh. However, the observation that these mutants are not polydactylous indicates that Shh-dependent pathway activation must also contribute to the Tulp3−/− limb phenotype.
SUMMARY AND PERSPECTIVES
In this study, we have shown that Tulp3 plays a critical role in the regulation of Shh signaling. Tulp3 mutant embryos develop defects associated with increased Shh activity and show increased expression of Ptc1 in tissues where Shh signaling normally takes place. The caudal neural tube of Tulp3 mutants is ventralized, and ventral markers that are lost in Shh−/− embryos are expressed in Tulp3/Shh compound mutants. This clearly demonstrates that the enhanced pathway activation in the Tulp3 mutant neural tube is independent of Shh, and that Tulp3 plays an important negative regulatory role in the pathway. Cells in the ventral neural tube need to be able to sense and appropriately respond to different levels of extracellular Shh, dictated by their relative position along the D-V axis. How these cells are able to translate a particular Shh signal into a response, mediated by the Gli proteins, is not entirely understood but involves the activity of several positive and negative regulatory factors downstream of Smo activation. We have shown here that Tulp3 is one factor required for cells in the ventral neural tube to generate the appropriate response to a particular level of Shh, because these cells in Tulp3−/− embryos adopt an incorrect positional identity. We also demonstrated that Tulp3 represses the Hh pathway in the limbs through both Shh-dependent and -independent mechanisms. The altered expression of genes normally regulated by Gli3R in Tulp3−/− limb buds suggests that Tulp3 is involved in regulating Gli3 processing. Further epistasis analyses with mutants that exhibit altered Hh responses will help to define the precise location of Tulp3 in the Hh pathway.
Several other key regulatory factors of Hh signaling have been identified in mice, which lead to a loss of or enhanced activation of the pathway. Studies using forward genetic screens of mutations affecting early embryogenesis in mice have revealed that the ability of cells to respond to Hh ligands requires intraflagellar transport (IFT) proteins, which are involved in intracellular transport and in the formation of primary cilia (Huangfu et al.,2003). Since this discovery, several further studies have shown that components of the Hh signal transduction pathway such as Ptc1 and Smo, as well as overexpressed full-length GFP-tagged Gli proteins, localize to primary cilia (Corbit et al.,2005; Haycraft et al.,2005; Rohatgi et al.,2007), and IFT proteins have been shown to be required for both activator and repressor functions of Gli proteins (Haycraft et al.,2005; Huangfu and Anderson,2005; Liu et al.,2005). Tulp3 mutants show a gain of Hh signaling phenotype in both neural tube and limbs, which is consistent with mutations in factors that have been shown to act at a step in the pathway that is dependent on cilia function, such as Rab23 and Fkbp8 (Huangfu et al.,2003; Liu et al.,2005; Cho et al.,2008), suggesting that Tulp3 function may also be dependent on cilia function. In fact, several studies have implied a role for the Tubby proteins in the regulation of intracellular transport and cilia-related functions. The tubby protein ortholog in C. elegans, Tub-1 has been shown to influence fat storage by regulating the activity of the vesicle transport protein Rab-7 through the interaction with a Rab GTPase activating protein, RGB-3 (Mukhopadhyay et al.,2007). Tub-1 also undergoes anterograde and retrograde ciliary transport in ciliated neurons (Mukhopadhyay et al.,2005,2007), and functions in the same genetic pathway as bbs-1 (Mak et al.,2006), a member of the BBS family of proteins associated with the basal body of primary cilia (Ansley et al.,2003). In mice, inactivation of the Tulp1 gene causes retinal degeneration and impaired rhodopsin transport in photoreceptors (Ikeda et al.,2000; Hagstrom et al.,2001), and mutations in the human TULP1 gene cause retinitis pigmentosa, a human retinal degenerative disease (Hagstrom et al.,1998). Moreover, the microtubule associated protein 1a (Mtp1a) gene, which encodes a protein involved in intracellular trafficking, was found be a genetic modifier of the cochlear degeneration phenotype leading to hearing loss in tub mutant mice (Ikeda et al.,2002b). Although the precise role that Tulp3 plays in the Hh signaling pathway is currently unclear, the evidence from other Tubby family members suggests that it may be involved in intracellular transport event(s) which, when deregulated, lead to ligand-independent pathway activation.
Mice Strains and Genotyping
Creation of mice with a targeted disruption of Tulp3 and a floxed allele of Shh was described previously (Ikeda et al.,2001; Lewis et al.,2004) and were obtained from Jackson Laboratories (Bar Harbor, ME). Mice containing the floxed Shh allele were first bred to CMV-Cre–expressing mice, which act as a general deletor strain for floxed alleles (Dupe et al.,1997). To generate null embryos, animals were mated overnight and females were examined the following day for presence of a vaginal plug. Noon on the day a plug was found was considered as E0.5. To obtain Tulp3−/−/Shh−/− progeny, Tulp3+/−/Shh+/− breeding pairs were intercrossed. Genotyping was performed on genomic DNA isolated either from yolk sacs or tail clips by standard PCR using primers as previously described (Ikeda et al.,2001; Lewis et al.,2004). All animal experiments were performed in ethical accordance with protocols approved by the University's Animal Care and Use Committee.
In Situ Hybridization
Embryos at the relevant stages were subjected to whole-mount or frozen section in situ hybridization analysis as described previously (Chotteau-Lelievre et al.,2006). For analysis of Tulp3 expression, a 1,000-bp fragment of the Tulp3 cDNA sequence, from SacII to SmaI, was cloned into the pBluescript II KS vector (Stratagene, La Jolla, CA) and used to generate sense and antisense RNA probes. Shh, Gli3, Hnf3β, Ptc1, and Wnt1 plasmids for generating antisense RNA probes were kindly provided by Drs. A. McMahon, A. Joyner, B. Hogan, M. Scott, and M. Wassef, respectively. Following mRNA detection, embryos were photographed under a dissecting microscope (Leica MZ 9.5, Leica Microsystems (Canada) Inc., Richmond Hill, ON, Canada) using a digital camera (SPOT-RT, Diagnostic Instruments, Sterling Heights, MI). At least 3 embryos of each genotype were used per marker.
For immunohistochemical analysis of neuronal markers, E10.5 embryos were fixed for 90 min in 4% paraformaldehyde and then cryoprotected using a sucrose series (5% for 30 min, 10% for 2 hr, and 20% overnight). Embryos were then embedded in embedding medium (Anatomical Pathology USA, Pittsburgh, PA) frozen, and sectioned using a Reichert-Jung Cryocut 1800 cryostat (Leica Microsystems [Canada] Inc.). All slides represent transverse sections around the level of the hindlimbs.
Sections were blocked in phosphate buffered saline (PBS)/0.2% Triton containing 10% goat serum for 1 hr at room temperature, followed by incubation with primary antibodies (1:100 dilution) overnight in PBS/0.2% Triton with 1% goat serum at 4°C. After three washes in PBS/0.1% Triton, slides were incubated with rabbit anti-mouse IgG Alexa 555 conjugated secondary antibody (1:500 dilution in PBS/0.2% Triton with 1% goat serum; Molecular Probes, Invitrogen Canada Inc., Burlington, Ontario, Canada) in a humidified chamber for 1–2 hr at room temperature. Coverslips were mounted with fluorescence microscopy mounting medium (Dako Canada, Inc., Mississauga, ON, Canada) and confocal microscopy was performed using a Leica TCS SP2 multiphoton confocal microscope using a ×20 oil immersion lens. All images were captured using Leica Confocal Software.
Primary antibodies were developed by Dr. T.M. Jessell (HB9: 81.5C10 and Nkx2.2: 74.5A5) Dr. A. Kawakami (Pax6 and Pax7), and Dr. O. D. Madsen (Nkx6.1: F55A10) and were obtained through the Developmental Studies Hybridoma Bank at the University of Iowa, USA. At least 2 embryos of each genotype were used per marker.
Alcian blue staining of cartilage in was carried out as described (Abu-Abed et al.,2001).
We thank Dr. J. Fallon for comments on this manuscript, and Drs. J. Naggert and A. McMahon for making available the transgenic mouse lines created in their labs. We also thank Dan Wainman, Jeff Mewburn, Matt Gordon, Jalna Meens, and Doug Richardson for technical assistance, and Dr. Glenn MacLean for critical review of the manuscript and discussions regarding this work.