Expression of ptc and gli genes in talpid³ suggests bifurcation in Shh pathway*

The developing chick limb bud is a well-established model for investigating cellular and molecular basis of embryonic pattern formation. Extensive experimental analysis has elucidated aspects of patterning of all three limb axes, anteroposterior, proximodistal and dorsoventral, and shown that they are interdependent and act coordinately. In addition, limb signalling pathways are also used elsewhere and therefore investigations of limb development may illuminate understanding of patterning of other regions of the embryo (reviewed Johnson and Tabin, 1997; Cohn and Tickle, 1996). Anteroposterior limb patterning is controlled by a group of mesenchymal cells in the posterior limb bud known as the polarising region. Polarising region grafts to the anterior of another limb bud produce mirror-image digit duplications (Saunders and Gasseling, 1968; Tickle et al., 1975). The polarising region expresses shh, a signalling molecule that has been implicated in many patterning processes in developing embryos (reviewed Hammerschmidt et al., 1997). Ectopic expression of shh at the anterior of a developing limb bud can also induce mirror-image duplications suggesting that Shh mediates patterning activity of the polarising region (Riddle et al., 1993). Several other potential components of anteroposterior patterning have been identified, either because of their asymmetrical expression across the anteroposterior axis of the limb bud or because of their involvement in the Shh signalling pathway. These include members of bmp and hox gene families (Izpisua-Belmonte et al., 1991; Francis et al., 1994; Francis-West et al., 1995). For example, bmp2 and 5′ members of the hoxD complex (hoxD-11, hoxD-12 and hoxD- 13), normally expressed in posterior limb bud mesoderm, are ectopically expressed in anterior limb buds after shh mis- expression or polarising region grafts. However, this induction of bmp and hox-D gene expression by Shh also requires the apical ectodermal ridge (AER), which runs along the distal limb bud rim. This apical ridge requirement can be substituted for by FGF4 protein, which is encoded by a gene that is normally expressed in posterior ridge and is itself induced in response to shh (Laufer et al., 1994; Niswander et al., 1994). Therefore expression of these bmp and hox genes probably requires convergence of Shh and FGF signalling pathways.

Several components of the Hh pathway have been identified in both vertebrates and Drosophila. Gli proteins have been implicated as possible transcription factors for Hh pathways and Ptc proteins as receptors (reviewed Ingham, 1998a; Johnson and Scott, 1998). In chick, two gli genes (gli1 and gli3) and one ptc gene have been described (Marigo et al., 1996a,b). Note gli3 may be more closely related to gli2/4 of other vertebrates (Borycki et al., 1998); gli1 was originally referred to as gli (Marigo et al., 1996b). Ptc is an unusual receptor as it is both a target and a repressor of the Hh pathway. Basal levels of ptc are expressed throughout domains competent to respond to Hh signals, where they repress transcription of Hh target genes. Hh signals antagonise this repression by binding to Ptc, one result of which is increased ptc expression (reviewed Ingham, 1998b). The Drosophila homologue of gli genes, ci, has been postulated to have both repressor and activator functions. In the absence of Hh signals, it represses expression of dpp, a Drosophila bmp homologue, and hh itself. However, in presence of Hh signals, Ci activates transcription of at least some targets of the Hh pathway including ptc and dpp. These different functions are mediated by different forms of Ci protein (Aza-Blanc et al., 1997; Dominguez et al., 1996; Alexandre et al., 1996). In vertebrates, different Gli proteins may fulfil these different functions. Analysis of mouse mutations suggests that Gli3 represses shh expression (Masuya et al., 1995; Buscher et al., 1997). Gli1 has been implicated as a transcriptional activator for the Shh pathway (Hynes et al., 1997; Marigo et al., 1996b; Lee et al., 1997; Sasaki et al., 1997). In addition to being components of the Shh pathway, all three genes (gli1, gli3 and ptc) in vertebrates are targets for transcriptional control by Shh: ptc and gli1 expression increases in response to Shh; gli3 expression decreases (Marigo et al., 1996a,b). However, in contrast to induction of bmp2 and 5′ hoxD genes, induction of ptc and gli1 expression in the chick limb bud does not require apical ridge.

talpid³ (ta³) is an embryonic lethal chicken mutation in a molecularly uncharacterised autosomal gene. Homozygous embryos have a pleiotropic phenotype, but ta³ limbs buds are initially shorter and broader than normal and eventually about 7 digits form, all of which are fused and look morphologically similar (Ede and Kelly, 1964b; Hinchliffe and Ede, 1967). In ta³ the relationship between shh expression and expression of 5′ hoxD genes, fgf-4, bmp2 and bmp7 is uncoupled: normally posteriorly restricted fgf-4, bmp and hoxD genes are expressed symmetrically across the entire anteroposterior axis but shh expression is still restricted to posterior limb (Izpisua-Belmonte et al., 1992; Francis-West et al., 1995). bmp4 expression, which is normally strong anteriorly, is reduced in ta³ anterior limb buds and symmetrical around the bud rim (Francis-West et al., 1995). This suggests that ta³ may act between shh transcription and expression of bmp and hoxD genes. This is consistent with the ta³ phenotype, which includes abnormalities in many other patterning processes where Shh signalling has been implicated, such as dorsal/ventral patterning of neural tube and somites, correct spacing of eyes and development of face and feathers (Ede and Kelly, 1964a,b). However, since it is not known whether bmp and 5′ hoxD genes are direct targets of Shh signalling or further downstream, it is unclear at which point in these developmental signalling cascades ta³ normally acts.

Here we investigate ptc expression in ta³. ptc is a direct target of Hh signalling in Drosophila (Ingham, 1993; Forbes et al., 1993; Alexandre et al., 1996; Struhl et al., 1997; Strigini and Cohen, 1997) and recent analysis suggests that this is also the case in vertebrates (Goodrich et al., 1996, 1997; Marigo et al., 1996a; Concordet et al., 1996; Lewis et al., 1998). ptc expression in ta³ should show whether ta³ acts in the Shh pathway, or whether it acts downstream of, or in parallel to, Shh signalling. ptc is also a possible candidate for ta³: since Ptc represses Hh pathways in absence of Hh signals, mutations in ptc usually result in ectopic activation of Hh pathways throughout embryonic fields that are competent to respond to Hh (Ingham et al., 1991; Ingham and Hidalgo, 1993; Capdevila et al., 1994; Goodrich et al., 1997). Therefore, a ptc mutation in chicks would be expected to have a similar phenotype to ta³: normal shh expression but expanded expression of Shh targets, such as bmp and 5′ hoxD genes.

Our examination of ptc expression in ta³ suggested that ta³ acts between shh and transcription of ptc. Therefore, we examined expression of other components of the Shh pathway in ta³ embryos. We also investigated whether the ta³ defect was in signalling cells or responding cells by determining whether ta³ limbs could respond to Shh protein and whether ta³ mesoderm was capable of signalling to normal mesoderm. Finally, we genetically mapped several components of the Shh pathway and investigated whether ta³ corresponded to any of these.

Embryos were obtained from fertilised Needle Farm White Leghorn chicken eggs or from talpid³ (ta³) stock maintained at Roslin Institute, UK. Eggs were incubated at 38°C: wild-type chick embryos were staged according to Hamburger and Hamilton (1951); and ta³ embryos according to Hinchliffe and Ede (1967). Homozygous ta³ embryos were identified by characteristic limb morphology and/or by reduced spacing of eyes. If there was any ambiguity, then identification was confirmed by grafting a wing onto the wing stump of a wild-type host to allow it to develop further so that the ta³ shape became more obvious, or by in situ hybridisation on one limb with a gene that has an altered pattern of expression in ta³.

Stock solutions of bacterially expressed aminoterminal Shh protein (N-Shh), a kind gift from A. McMahon (Yang et al., 1997), were stored at ?70°C and dilutions made in Tris chloride/sodium chloride buffer. To prepare Shh beads, 2 μl of N-Shh solution (14 mg/ml) was placed on a bacteriological grade Petri dish. 15-20 Affigel CM beads (200-250 μm in diameter) were rinsed in Tris chloride/sodium chloride buffer, then transferred to the drop of N-Shh solution and allowed to soak for between 1 and 2 hours at room temperature.

Wing buds of stage 20 wild-type chicken hosts were removed using tungsten needles and replaced by manipulated or non- manipulated ta³ limb buds, which were kept in place with platinum staples. In the case of manipulated ta³ limb buds, a Shh bead was placed under a loop of apical ectodermal ridge at the anterior margin of each bud before grafting to its new host. For control experiments, the same procedures were repeated with wild-type chick limb buds. In another series of experiments, posterior, apical and anterior mesenchyme was removed from ta³ limb buds and grafted to a host wing under the anterior apical ridge (Fig. 1). In all these cases, limb bud tissue was first soaked for 1 hour in trypsin at 4°C and the ectoderm removed. Host embryos were fixed in 4% paraformaldehyde (PFA) at different time intervals after operations for whole-mount in situ hybridisation.

Embryos were processed according to standard methods (Riddle et al., 1993; Francis-West et al., 1995); precise protocol available on request. Probes were synthesised using the following templates: ptc (Marigo et al., 1996a), gli1 (Marigo et al., 1996b), gli3 (Marigo et al., 1996b), coupTFII (Lutz et al., 1994), bmp-2 (Francis et al., 1994) and shh (Cohn et al., 1995).

Genetic loci were mapped in the East Lansing reference back cross (Burt and Cheng, 1998; Jones et al., 1997). Mapping information has been submitted to chicken genome database at Roslin Institute, UK (http://www.ri.bbsrc.ac.uk/).

The following PCR primer pairs were designed for chicken shh, ptc, gli1 and smo genes using the MIT PRIMER program.

553 bp product (GenBank L28099, nucleotides 879-1431); digested with PvuII (at position 1171) to produce 293 bp and 260 bp products, which were verified by DNA sequencing.

#338 5′-TGAAGGACCTGAGCCCTG-3′ (879-906)

#355 5′-AGGAGCCGTGAGTACCAATG-3′ (1412-1431)

Position of intron-exon boundaries predicted from structure of Drosophila hh gene (L02793).

218 bp (GenBank U40074, exon 22, nucleotides 4096-4313) product verified by digestion with internal AvaII site (at position 4233) to produce 138 bp and 80 bp DNA fragments.

#385 5′-TGCCAGCCTATCACTACTGTG-3′ (4096-4116)

#386 5′-CATTCGACATCCTGAAGCTC-3′ (4294-4313)

Position of intron-exon boundaries predicted from structure of human ptc (U43148) gene.

144 bp product (GenBank U60762, nucleotides 670-813) verified by digestion with internal PvuII site (at position 708) to produce 104 bp and 40 bp DNA fragments.

#457 5′-GTACCAGAACCCTCCGGG -3′ (670-687)

#458 5′-TGTTTGATGGTCCCACGAG-3′ (795-813)

223 bp product (GenBank AF019977, 3′-untranslated region, nucleotides 2817-3039) verified by digestion with internal HindIII site (at position 2867) to produce 51 bp and 172 bp DNA fragments.

#467 5′-ATCCCAAAGTGCCTTCCAG-3′ (2817-2835)

#468 5′-GAGGAAGCCGTGAGCCTAC-3′ (3021-3039)

50 ng of chicken genomic DNA was amplified in 50 μl containing 200 mM dNTPs, 500 nM primers, 2 units Taq Gold DNA polymerase,

20 mM Tris-HCl (pH 8.3), 50 mM KCl and 0.75 mM MgCl₂. Thermal cycling conditions: initial denaturation step at 94°C for 1 minute, then 30 cycles of 94°C for 30 seconds, 65°C for 1 minute, 72°C for 3 minutes, followed by 72°C for 10 minutes. Aliquots (15 μl) of PCR products were mixed with equal volumes of formamide loading dye (95% formamide, 20 mM EDTA, 0.05% bromophenol blue and 0.05% xylene cyanolFF), heated to 99°C for 3 minutes and placed on ice prior to electrophoresis. Samples were run on 16 cm native 15% polyacrylamide gels (37.5:1) for 16 hours at 12 mA at 15°C and gels were silver stained as described previously (Jones et al., 1997).

As previously described (Francis-West et al., 1995), shh is expressed normally in ta³ limbs (confirmed here for 7 legs, 3 wings; Fig. 2B). In contrast, ptc expression is clearly altered in ta³ limb buds: in wild-type chick limbs, ptc is expressed at high levels posteriorly with expression gradually decreasing anteriorly (Fig. 2C; strongly stained wing), whereas in ta³ limbs, high levels of ptc expression are never detected (0/19, 11 wings; 8 legs). At stages 19-21, ptc is expressed uniformly at very low levels throughout ta³ limb mesenchyme except immediately under the AER (Fig. 2D). This resembles the level of expression in the very anterior of wild-type limbs (compare Fig. 2C and D). In ta³ limbs at stage 27, ptc expression is still very weak but has resolved into a wide distal band and more medial proximal band that resemble the fused mesenchymal condensations that form in ta³ limbs (Hinchliffe and Ede, 1967; 2 wings; 2 legs; Fig. 2F). In contrast, wild-type limbs still have strong posterior expression of ptc at this stage, but are also beginning to express ptc more anteriorly, around developing skeletal elements where Ihh is beginning to be expressed (Marigo et al., 1996a; Fig. 2E).

In wild-type chick embryos, high levels of ptc expression are also clearly visible in branchial arches and mouth. Branchial arches are misshapen in ta³ embryos (Fig. 2H) and there is no high-level expression of ptc in either branchial arches or mouth (5 embryos; Fig. 2G,H). Thus, high-level ptc expression, normally associated with shh expression, is absent in several different regions in ta³ embryos.

shh RNA is expressed normally in ta³ but it is possible that processing of Shh protein is defective. Therefore lack of high- level ptc expression in ta³ embryos could be due to either reduction in activity of the Shh signal itself, or a defect in ability of ta³ tissue to respond to Shh signals.

To determine whether the endogenous Shh signal is active in ta³ limbs, mesodermal tissue from the posterior of ta³ limbs was grafted to anterior of wild-type limbs and ptc expression assayed after 18 hours. Posterior ta³ mesenchyme can induce high-level expression of ptc in normal limb tissue (7/8: 4 grafts showed strong induction – Fig. 3A; 3 showed weaker induction). However, ta³ tissue itself still did not express high levels of ptc (confirmed with sections: data not shown). Grafts of wild-type posterior tissue also induced high-level ptc expression in normal host tissue (3 grafts showed strong induction; 1 weaker; Fig. 3E).

ta³ limb buds grafted to wild-type chicken embryos also induced ptc expression in host flank near the posterior of the grafted ta³ limb, despite there still being no high-level expression of ptc in the grafted ta³ limbs themselves (2/2, assayed 25–26 hours after grafting; Fig. 3I). However, ptc was never induced in host flank when wild-type limbs were grafted to wild-type hosts (0/4; assayed 19-24 hours after grafting; Fig. 3H). These results show that ta³ posterior limb mesenchyme can induce ptc expression in normal tissue, suggesting that the ta³ defect is not in Shh itself. In addition, the second result suggests that ta³ posterior limb mesenchyme can induce ptc at a greater distance than its wild-type counterpart.

At stages 20-24, polarising activity in ta³ limbs is not as posteriorly restricted as in wild-type limbs (Francis-West et al., 1995), so we tested whether apical or anterior ta³ tissue could also increase levels of ptc expression in wild-type tissue. Some, but not all, of these grafts induced ptc expression in wild-type host tissue around the graft (2/4 anterior grafts; 1/1 apical graft; Fig. 3B-D). In contrast, grafts of anterior and of apical wild- type tissue never induced ptc expression (0/3 apical; 0/1 anterior; Fig. 3F,G).

To determine whether ability of mesodermal cells to respond to Shh was affected in ta³ embryos, we investigated whether high-level expression of ptc could be induced in ta³ limbs with purified Shh protein. In one case, an Shh-bead was implanted to the wing of a ta³ embryo, which was fixed 24 hours later. Contralateral wing and wing with implanted Shh bead had identical low-level expression of ptc. In other cases, Shh beads were implanted anteriorly in ta³ limbs and then each manipulated ta³ limb grafted to a wild-type chicken embryo. ptc expression was not induced in any of these ta³ limbs (0/7 – 3 legs; 4 wings). However, there was clear induction of ptc in normal host flank, suggesting that Shh protein from the bead was active and had diffused across the ta³ limb (Fig. 2J). Beads were also inserted into wild-type wings and these wings were grafted to different wild-type hosts to check that grafting did not interfere with the ability of limb tissue to respond to ectopic Shh protein. In these cases, ptc was strongly induced in wild- type limbs (2/2). Weak expression of ptc was also induced in host flank (Fig. 2I). These experiments show that wild-type and ta³ limbs respond differently to Shh beads: unlike in wild-type limbs, purified Shh protein does not induce high-level ptc expression in ta³ limbs, suggesting a defect in response to Shh signals in ta³.

Drosophila Ci and vertebrate Gli-1 proteins are implicated as transcription factors for induction of high-level ptc expression (Alexandre et al., 1996; Marigo et al., 1996b; Hynes et al., 1997). In addition, expression of gli genes in vertebrates is regulated by Shh: expression of gli1 increases in response to Shh and expression of gli3 decreases (Marigo et al., 1996b; Lee et al., 1997).

In wild-type limbs, gli1 is expressed posteriorly in a domain similar to ptc (Marigo et al., 1996b; Fig. 4A). However, in ta³ limbs, there is no high-level gli1 expression: rather gli1 is expressed very weakly throughout the limb except distally under the AER (14/14 – 10 legs; 4 wings, stages 20-24; Fig. 4B). Again, as for ptc, the level of gli1 expression in ta³ limbs resembles the most anterior expression in wild-type limbs (compare Fig. 4B to 4A).

In contrast to ptc and gli1, gli3 expression is expanded in ta³ limbs (7/7; 3 wings, 4 legs; stages 20-23; Fig. 4D,F). In wild- type limbs, there is a clear area posteriorly where gli3 is not expressed (Fig. 4C,E). In some ta³ limbs (4/7), gli3 expression is weaker posteriorly (Fig. 4F) but the area of weaker expression is always very small and much less distinct than the clear region in wild-type limbs. However, gli3 expression extends to the very posterior of other ta³ limbs (3/7; Fig. 4D). These changes in gli1 and gli3 expression suggest that transcription of these genes, like that of ptc, is unable to respond to endogenous Shh signals in ta³ limbs.

We reported previously that bmp-2 expression is expanded in ta³ limbs and that the apical ridge shows very strong anterior expansion of bmp-2 expression (Francis-West et al., 1995; here, 8/8, 4 legs, 4 wings, stages 21-24; Fig. 4H-K.) Our more recent analysis revealed further subtleties, in that expanded bmp-2 expression in ta³ limb mesoderm is weaker than apical ridge expression and there is always a thin band where bmp2 is not expressed just under the ridge. Despite expansion of its domain, bmp-2 is sometimes still expressed at slightly higher levels in posterior ta³ limb (6/8 limbs): in one leg, bmp-2 was expressed in a wide, anteroposterior stripe that decreased anteriorly (stage 21/22; Fig. 4H) and, in 5 limbs, slightly higher posterior expression could be distinguished (Fig. 4J).

As bmp-2 is expressed at high levels in ta³ limbs, unlike ptc and gli1, this suggests that even though transcription of all of these genes can be increased by Shh signals in wild-type limbs, these genes are regulated independently in ta³ limbs. We therefore tested whether bmp-2 expression in ta³ limbs is altered by inserting a Shh bead into the anterior of the limb. However, we were unable to detect any changes in bmp-2 expression (5/5; 22-23 hours after bead insertion; Fig 4I), suggesting that anterior bmp2 expression may have already been maximal in these ta³ limbs.

We also tested whether ectopic expression of bmp-2 in ta³ tissue was maintained in a wild-type environment. We grafted mesenchyme from apex and anterior of ta³ limbs (Fig. 1) into anterior wild-type wings and examined expression of bmp-2, 18-20 hours later. In all cases, bmp-2 was ectopically expressed in host AER near the ta³ graft and there was some persistent mesodermal expression of bmp-2 in the graft edges. This residual expression was weaker than ectopic AER expression and wild-type bmp-2 expression in posterior host limb mesoderm (3 apical grafts; 1 anterior graft; Fig. 5). In contrast, grafts of posterior wild-type tissue induced bmp-2 expression (2/2), while a graft of apical wild-type tissue did not (data not shown).

In ta³ limbs, expression of some target genes of Shh signalling is expanded (Francis-West et al., 1995; this paper), but transcription of gli1, gli3 and ptc no longer seems to respond to Shh signalling. coupTFII is a direct target gene of Shh signalling in vitro but, unlike ptc, its transcription is probably not mediated by Gli-1 (Krishnan et al., 1997). We therefore determined whether expression of coupTFII in ta³ limbs followed either of these two patterns. It has been reported that coupTFII is expressed in chick limbs but no details were given (Lutz et al., 1994). At stages 21-24, coupTFII is expressed in the centre of normal limbs. Wing expression is broader proximally and extends slightly more distally than leg expression (Fig. 6A,B).

In ta³ wings, coupTFII is expressed throughout, except in a thin band of tissue just beneath the AER (3/3, stage 20-26; Fig. 6C,E). In younger ta³ legs, expression of coupTFII is similar (2/2, stage 20; Fig. 6D) but, in contrast, in older ta³ legs, coupTFII is not expressed in posterior mesoderm (3/3 stage 25- 26; Fig. 6F) and the band of distal cells that does not express coupTFII is wider than at younger stages, or in ta³ wings of the same stage (compare Fig. 6F to E and D).

Expansion of coupTFII expression in ta³ limbs suggests that this gene is regulated in a similar way to normally posteriorly expressed bmp and hox-D genes (Izpisua-Belmonte et al., 1992; Francis-West et al., 1995). However, expression of coupTFII in wild-type limbs is more anterior than these other Shh targets and the experiments that suggest that coupTFII expression is induced by Shh have been conducted either in vitro or in neural tube (Lutz et al., 1994; Krishnan et al., 1997). Therefore, to establish if, and how, coupTFII can be regulated by Shh in limbs, we inserted Shh beads into anterior wild-type wings and examined coupTFII expression after 18-19.5 hours. In most wings, coupTFII expression increased (4/5; Fig. 6G), although in a few cases (2/5) there was less high-level expression adjacent to the bead than elsewhere (Fig. 6H). One wing (Fig. 6G) had an expression pattern very reminiscent of coupTFII expression in ta³ wings. Therefore coupTFII expression can respond to Shh signalling in the limb and expression of coupTFII in ta³ limbs is consistent with ectopic Shh signalling.

Our analysis of gene expression in ta³ limb buds is consistent with ta³ encoding a component of the Shh pathway. We therefore investigated several genes in this pathway to see if they co-segregated with the ta³ phenotype. Polymorphisms were detected in ptc (Marigo et al., 1996a), shh (Riddle et al., 1993), gli1 (Marigo et al., 1996b) and smoothened (smo) (Quirk et al., 1997) by SSCP analysis in the East Lansing reference mapping population (Burt and Cheng, 1998).The segregation patterns of parental alleles in the back cross progeny for all loci were compared with the inheritance of previously mapped markers.

The genetic locations of the genes are:

ptc, Chr. Z (137 cM) (LEI0254 – 0.0 cM, LOD 14.4 – PTCH – 0.0 cM, LOD 15.7 -MSU0352).

shh, Chr. 2 (61 cM) (MSU308 – 6.2 cM, LOD 9.8 – SHH – 10.4 cM, LOD 7.7 – ADL0270).

gli1, Chr. E22C19W28 (42 cM) (LEI0003 – 1.9, LOD 13.5 – GLI-1 – 20.3, LOD 4.6 -LEI0019).

smo, Chr. 1 (16 cM) (MSU0343 – 0.0 cM, LOD 15.7 – SMOH – 0.0 cM, LOD 12.0 -MSU0314).

The mutant allele in all ta³ carriers is derived from a single grand sire and is inherited as a single autosomal, recessive lethal mutation. Therefore a simple co-segregation test of smo, shh, gli1 and ta³ in the ta³ carrier parents (n=13) and ta³ progeny (n=19) will either support or exclude the role of these genes in the ta³ gene defect. If the ta³ gene defect were in one of these genes, then all ta³ homozygous progeny should also be homozygous for one of the candidate gene polymorphisms. Conversely, if there is no association then all genotypes for the candidate gene will be found within ta³ homozygotes. We were unable to detect any polymorphisms in the ptc gene of any ta³ carriers, but genetic mapping (described above, Fig. 7A) showed that this gene maps to the Z sex chromosome. Since ta³ is inherited as an autosomal mutation, we can clearly exclude ptc as a potential candidate for the ta³ defective gene. SSCP alleles were detected in the other candidate genes (Fig. 7B). All classes of genotypes were found in ta³/ta³ mutants, thus excluding smo, shh and gli1 as the defective gene in ta³. Unforunately, we were unable to extend this analysis to gli3, as we were unable to detect any polymorphisms in this gene in either East Lansing or ta³ crosses.

coupTFII is normally strongly expressed in two domains in ventral neural tube (Lutz et al., 1994; Fig. 8A), and has been implicated in induction of motoneurons by Shh. Therefore we investigated whether coupTFII expression was altered in ta³ neural tubes. We found that, unlike ta³ limbs, in which coupTFII expression is expanded, coupTFII is not expressed in neural tubes of ta³ embryos though it is still expressed at high levels in lateral somites (compare Fig. 8B and A). However, the effect of the ta³ mutation on shh expression also differs between limb and neural tube. In ta³ limbs, shh is expressed normally, but shh expression in ventral neural tube is discontinuous and reduced in both trunk and head (compare Fig. 8C,E to D,F,G). However, shh is still expressed normally in ta³ notochords (Fig. 8G).

We also examined expression of ptc, gli1 and gli3 in trunks of ta³ embryos. ptc is still expressed in medial neural tube, though at lower levels than in wild-type embryos. However, compared to wild-type embryos, there is hardly any expression of ptc around the notochord in ventral somites (compare Fig. 8H to I). gli1 is also still expressed in neural tubes of ta³ embryos but like ptc there is less expression of gli1 in ta³ somites (compare Fig. 8J to K). In contrast gli3 is strongly expressed in ta³ neural tube and somites. Neural tube expression of gli3 is expanded ventrally and appears stronger than in wild-type embryos (compare Fig. 8M to L).

We also observed characteristic morphological abnormalities in ta³ trunks. ta³ neural tubes are usually ovoid and have larger lumens than in wild-type embryos (Ede and Kelly, 1964b; Fig. 8I,M). In addition, in two ta³ embryos, the neural tube bifurcated caudally.

shh is expressed normally in ta³ limbs, but we show here that the high-level expression of ptc and gli1 that is normally induced in and around shh-expressing cells is absent in ta³ limbs, and that normal posterior repression of gli3 expression does not occur. This suggests that the ta³ gene product acts between shh expression and Shh regulation of ptc, gli1 and gli3 transcription. It is unlikely that ta³ is required for correct processing of Shh protein as posterior ta³ tissue can induce ptc expression in wild-type tissue. Instead the ta³ gene product appears to be required for response to Shh signals because purified recombinant Shh protein does not induce ectopic ptc expression in ta³ limbs as it does in wild-type limbs. Furthermore, ta³ tissue grafted into wild-type limbs continues to express low levels of ptc, even when surrounding normal tissue is expressing higher levels. The pleiotropic morphological phenotype of ta³ embryos is also consistent with ta³ being required for correct Shh signalling as abnormalities have been observed in many embryonic regions in which Shh signalling has been implicated. In two of these regions (branchial arches and mouth), high-level ptc expression is also absent. However, gene mapping and co-segregation analyses of genetic markers for ptc, smo, shh and gli1 genes exclude them as candidate genes for ta³.

The morphological limb phenotype of ta³, especially polydactyly, resembles mutant phenotypes that arise because of increased Shh signalling (or a reduction in Ptc activity). For example, mice and humans heterozygous for a mutation in ptc have phenotypes that occasionally include polydactyly (Hahn et al., 1996; Johnson et al., 1996; Goodrich et al., 1997); transgenic mice that ectopically express shh specifically in their skin show polydactyly of both forelimbs and hindlimbs, with each limb forming about 8 morphologically similar digits (Oro et al., 1997), and most polydactylous mutations in mouse investigated so far (Strong’s luxoid, hemimelic extra toes, extra toes, recombinant induced mutant 4, luxate, X-linked polydactyly, Alx-4) are associated with ectopic expression of shh in anterior of the limb bud (Buscher and Ruther, 1998; Buscher et al., 1997; Chan et al., 1995; Masuya et al., 1995, 1997; Qu et al., 1997). One exception is the mouse mutation doublefoot, which has normal shh expression, and, unlike the other mouse mutants, does not have mirror-image digit duplications but identical digits like ta³. However, unlike ta³, doublefoot has a dominant phenotype and ectopic expression of Shh target genes including ptc and gli1 (Hayes et al., 1998; Yang et al., 1998). This suggests that its phenotype is also due to ectopic activation of the Shh pathway, either because of a mutation in an, as yet unknown, component of the pathway (Hayes et al., 1998) or due to ectopic expression of another hh gene, Ihh (Yang et al., 1998). ta³ is the only polydactylous mutation examined so far in which ptc and gli1 are not ectopically expressed at high levels.

Other aspects of ta³ resemble mutations that have reduced Shh activity. For example, in ta³ embryos, the eyes are closer together and, in extreme cases, fuse, which is reminiscent of holoprosencephaly and cyclopia seen in humans and mice with shh mutations (Chiang et al., 1996; Roessler et al., 1996; Belloni et al., 1996). Our analysis of gene expression in ta³ neural tubes is also consistent with reduced Shh signalling: expression of shh is reduced in ventral neural tube, the motoneuron domain of coupTFII expression is lost and there is lower expression of ptc and gli1, and expanded expression of gli3. This is reminiscent of mouse embryos lacking Shh function, which do not form a morphologically distinct floor plate, but it contrasts with mice lacking Ptc function in which floor plate characteristics, including expression of shh, are expanded dorsally (Chiang et al., 1996; Goodrich et al., 1997). This paradox – that some aspects of the ta³ phenotype resemble a loss of Shh signalling, and some resemble a gain of Shh signalling – suggests a more complex basis for the ta³ phenotype than for any of the other polydactylous mutations.

The ta³ gene product is required for high-level expression of ptc and gli1 and for posterior repression of gli3 expression in the limb, but is not required for expression of fgf-4 or the normally posteriorly expressed bmp or 5′ hoxD genes (Izpisua-Belmonte et al., 1992; Francis-West et al., 1995). ta³ limb morphology most resembles what we would expect from gain of Shh signalling, but expression of ptc and gli genes suggests loss of Shh signalling. Recent experiments in cell culture and Drosophila have suggested that the Hh pathway is not necessarily linear and that Gli proteins are not the only transcription factors for Shh signalling (Ohlmeyer and Kalderon, 1997; Lessing and Nusse, 1998; Krishnan et al., 1997). This is also suggested by recent analysis of mice lacking Gli1 and Gli2 function (Matise et al., 1998). There is considerable evidence that Gli proteins are transcription factors that mediate Shh upregulation of ptc transcription (Marigo et al., 1996b; Hynes et al., 1997), but expression of coupTFII in vitro does not seem to be mediated by Gli, even though it is directly induced by Shh (Krishnan et al., 1997). The ta³ limb phenotype could thus be explained if Gli proteins regulate expression of gli1, gli3 and ptc, but another transcription factor regulates expression of bmp2 and 5′ hoxD genes, fgf4 and coupTFII, either directly or indirectly, and ta³ is only required for the Gli branch of the pathway (Fig. 9A). A bifurcation in the Shh pathway could also explain why some aspects of ta³ more resemble loss of Shh signalling and others gain of Shh signalling. This would be predicted if different branches are required to different extents in different tissues. However, lack of coupTFII expression in ta³ neural tubes is puzzling. Based on our observations in the limb, we would have expected loss of high-level gli1 and ptc expression and an expansion of gli3 and coupTFII expression in ta³ trunks but, while the other expression patterns are as expected, coupTFII expression is lost in ta³ neural tubes. This could be a secondary effect of reduced floor plate in ta³ embryos. The vertebrate trunk is an unusual site of Shh activity in that very high levels are required to induce floor plate, which then itself expresses shh (Roelink et al., 1995; Marti et al., 1995). Floor plate development and associated shh expression is disturbed in ta³ embryos, probably because the Gli pathway branch is explicitly needed for floor plate development (Sasaki et al., 1997; Ruiz i Altaba, 1998; Ding et al., 1998; Matise et al., 1998). This could result in an overall level of Shh signal too low for coupTFII expression to be induced in ventral neural tube. However, in gli2 mouse mutants, motoneuron development can occur even in absence of floorplate, although, in these embryos, notochord remains in close association with neural tube for longer than in normal mouse development (Ding et al., 1998; Matise et al., 1998).

High-level ptc expression, at least in Drosophila, sequesters Hh and prevents its diffusion (Chen and Struhl, 1996; Taylor et al., 1993; Fig. 9B). Therefore, in absence of high-level ptc expression in ta³ limbs, it is possible that Shh protein is diffusing farther and can act over a longer distance than normal. This possibility is suggested by induction of ptc expression in wild-type hosts when we grafted ta³ limbs, but not when we grafted wild-type limbs, and by induction of ptc in wild-type tissue by grafts of anterior and apical ta³, but not wild-type tissue. As ptc expression can be induced by Shh, these results suggest that there may be more widespread distribution of Shh protein in ta³ limbs. This would also explain anterior expansion of normally posterior gene expression and polarising activity in ta³ limbs (Francis-West et al., 1995). Consistent with this hypothesis, reaggregates of disaggregated mesoderm from a whole leg or just the posterior third, which have polarising region cells, and hence shh expression and protein distributed randomly throughout the limb, show many similarities to ta³ limbs: identical digits and hox-D, bmp2 and fgf-4 genes expressed across the entire anteroposterior axis (Hardy et al., 1995).

Sensitivity of cells to Hh signals may also increase in the absence of high-level ptc expression. Ptc normally represses Hh signalling pathways in the absence of Hh, but high levels of ptc, normally induced by Shh, may also dampen the effect of Shh signals. Evidence that a precise balance between Hh and Ptc is required for correct development comes from the fact that overexpression of ptc can reduce Hh signalling (Johnson et al., 1995; unpublished results K. E. L., P. Currie and P. W. I.), and from analyses of Gorlin’s syndrome and human holoprosencephaly, which are caused by heterozygosity for loss-of-function PTCH or SHH mutations, respectively (Johnson et al., 1996; Hahn et al., 1996; Belloni et al., 1996; Roessler et al., 1996). If this ‘dampening’ effect is lacking, i.e. there is no ptc upregulation, a lower exogenous Shh signal may be sufficient to activate posterior genes, and therefore enough Shh may be diffusing even to anterior regions of the limb to produce uniform expression of genes such as bmp-2.

The combination of Ptc sequestering Shh protein and/or dampening Shh signals could also explain the difference between ta³ and the phenotype that is obtained from placing beads soaked in aminoterminal Shh (N-Shh) in the anterior of the limb. N-Shh diffuses a long way from the bead in these experiments, yet this results in symmetrical digit duplications rather than posterior identity across the limb bud (Yang et al., 1997). However, N-Shh induces ptc (Yang et al., 1997; Porter et al., 1996) and it seems likely that Ptc protein in cells around the bead begins to sequester N-Shh. Therefore, cells far away from the bead initially receive high levels of Shh activity but then receive lower levels. In contrast, Shh activity in ta³ limbs is not localised or dampened by ptc expression.

Interaction between Hh and Ptc may also explain why, in contrast to the ta³ polydactylous phenotype where all digits are morphologically similar, most mouse polydactylous mutations specify a sequence of different ectopic digits (Buscher and Ruther, 1998; Buscher et al., 1997; Chan et al., 1995; Masuya et al., 1997; Qu et al., 1997). One prediction would be that these mutants all ectopically express ptc, which localises any ectopic Shh signalling. Indeed, ptc is expressed ectopically in extra-toes and Strong’s luxoid (Goodrich et al., 1996; Platt et al., 1997; Buscher et al., 1997).

In conclusion, genes that are regulated by Shh in chicken limbs behave in one of two abnormal ways in ta³: transcription of ptc, gli1 and gli3 no longer responds to Shh signalling; in contrast expression of bmp, hox-D and coupTFII genes expands. Since the ta³ defect is in responding cells and not production of Shh signal, we propose a bifurcation in the Shh pathway and suggest that the ta³ gene product is a component of the Shh signal transduction pathway, required after the branch point, for regulation of the first group of genes. In addition, we propose that expanded expression of the second group of genes, and expansion of polarising activity in ta³ limbs, is due to a wider distribution of Shh protein, caused by lack of high-level ptc expression. This hypothesis can also account for different aspects of the pleiotropic ta³ phenotype.

K. E. L was funded by a postgraduate student bursary from ICRF; G. D. by MRC and an Anatomical Society Studentship; K. E. R. by a MRC studentship. We thank A. McMahon for Shh protein, C. Tabin for probes kindly provided before publication; P. Francis-West and M. Tsai for probes. We also thank colleagues at UCL, Dundee, ICRF and Sheffield for their interest and advice and particularly David Stark, Juan-José Sanz Ezquerro and Isabelle Le Roux for helpful comments on early drafts of the manuscript.