The chick oligozeugodactyly (ozd) mutant lacks sonic hedgehog function in the limb

Patterning of the amniote limb is organized by three well-defined signaling centers (reviewed by Capdevila and Izpisua Belmonte, 2001; Schaller et al., 2001). The apical ectodermal ridge (AER) permits limb bud elongation and the determination of skeletal elements along the proximal-distal (PD) axis, through the action of fibroblast growth factor(FGF) family members (Martin,1998; Moon et al.,2000; Sun et al.,2002). The non-ridge limb bud ectoderm controls dorsal-ventral(DV) polarity through the action of Wnt7a and engrailed1 respectively(reviewed by Chen and Johnson,1999). The anterior-posterior (AP) limb axis is controlled by a small group of mesodermal cells along the posterior limb bud border called the zone of polarizing activity (ZPA) through the activity of Sonic hedgehog(Shh), which is synthesized by ZPA cells (reviewed byPearse and Tabin, 1998).

Evidence from a variety of sources points to an interdependence of the limb bud signaling centers for continued synthesis of effector molecules and signaling function. The AER is necessary for the induction(Crossley et al., 1996;Grieshammer et al., 1996;Noramly et al., 1996;Ros et al., 1996) and maintenance of Shh expression(Riddle et al., 1993) by ZPA cells. AER induction and maintenance has long been known to be dependent on the limb bud mesoderm (e.g. Saunders,1977). Recent data suggest the Bone morphogenetic protein (BMP)inhibitor gremlin (Gre) is downstream of Shh and required for AER maintenance(Capdevila et al., 1999;Merino et al., 1999;Zúñiga et al.,1999). Thus, the molecular framework of a possible AER-to-ZPA-to-AER feedback loop is emerging. The possibility that FGF10 is the effector growth factor of AER induction has been suggested(Ohuchi et al., 1997). The complexity of limb signaling center interaction is further demonstrated by the observation that Wnt7a^-/- mice showed reduced Shhexpression and posterior limb deficiencies(Parr and McMahon, 1995).

The mechanisms that precisely define the location and subsequent maintenance of the limb bud signaling centers are poorly understood. This is especially true of the ZPA (Tanaka et al.,2000). There is evidence of a role for retinoids in Shhinduction and maintenance from studies using retinoid inhibitors and retinoid deficiency models (e.g. Lu et al.,1997; Power et al.,1999; Stratford et al.,1997; Stratford et al.,1999), and from Shh induction after treatment with retinoids (reviewed by Tickle and Eichele,1994). Misexpression of Hoxb8 in anterior limb mesoderm results in ectopic Shh expression but only in the proximity of the AER (Charité et al.,1994). Interestingly, Hoxb8 expression precedes retinoid-induced Shh expression in the anterior limb bud mesoderm(Lu et al., 1997;Stratford et al., 1999). Similarly, ectopic expression of the transcription factor dHAND results in ectopic Shh expression(Charité et al., 2000;Fernandez-Teran et al., 2000;McFadden et al., 2002) anddHAND^-/- mice fail to express Shh in the limb(Charité et al., 2000). At present, it is not clear how these observations can be integrated to explain how the ZPA is spatially delineated or how Shh expression is maintained.

While it would appear that a cohort of cells in the emerging limb bud has the competence to express Shh when exposed to FGFs(Ros et al., 1996), analyses of mouse mutants with anterior polydactyly, and other studies, indicate the existence of negative regulators that restrict Shh expression to the posterior bud. The transcription factors Alx4 and Gli3 have domains of expression in the limb bud complementary to that of Shh and have been proposed to repress Shh expression in the anterior limb mesoderm(Büscher et al., 1997;Marigo et al., 1996b;Masuya et al., 1997;Masuya et al., 1995;Qu et al., 1997;Qu et al., 1998;Takahashi et al., 1998). A gradient of the repressor form of Gli3 has been described in the limbs of mice and chickens with the highest concentration in the anterior portion of the limb (Litingtung et al., 2002;Wang et al., 2000). It has also been proposed recently that Gli3 is an obligate component of ZPA function, required in responding cells for Shh mediated polarizing activity(Litingtung et al., 2002).

The mechanisms involved in skeletal patterning downstream of Shh are being actively investigated. Bmp2 has been considered a candidate Shh effector gene because it is expressed in a domain that overlaps Shhexpression and because it is induced in the anterior limb mesoderm by ectopic Shh expression (Duprez et al.,1996; Yang et al.,1997). Recently, it was proposed that Shh acts to specify digit formation, while concurrently setting up a gradient of Bmp2 that subsequently specifies digit identity in a dose-dependent manner(Drossopoulou et al., 2000). It has been demonstrated that BMP activity in the interdigital mesoderm at autopod stages is required for the interdigits to specify digit identity(Dahn and Fallon, 2000).

We have analyzed a new limb mutant in the chicken first described by Smyth et al. (Smyth et al., 2000)and previously named Ametapodia 2. These chickens develop limbs that lack ulna and fibula and all digits except digit 1 (d1) of the foot. Digit identity was proposed on the basis of genetic evidence. Here we rename this mutation as oligozeugodactyly (ozd) meaning reduced zeugopod and digits and report data consistent with the complete absence ofShh expression and activity specifically in the developing limb buds. We report that the stylopod is normal in ozd limbs and the zeugopod develops with only radius or tibia. While the wing lacks digits, the leg develops a clearly identifiable d1. Consistent with the absence of Shh signaling, neither Ptc1 nor Gli1 are detectable in mutant limb buds, and we observe that the expression of Bmp2, dHAND and 5′ Hoxd genes in posterior wing and leg bud mesoderm is comparable to that observed in the limbs of Shh^-/- mice. We conclude that Shh becomes necessary for limb skeletal patterning distal to the elbow and knee joints, similar to Shh^-/- mice(Chiang et al., 2001;Kraus et al., 2001). The data presented are consistent with a developmental model proposing the PD axis is specified in the limb field, and that the radius/tibia and d1 are Shh independent, while the ulna/fibula and other digits are Shh dependent.

Oligozeugodactyly (ozd) mutant and normal chick embryos were obtained from a heterozygous mating flock maintained at the University of Wisconsin Poultry Science Department (Madison, WI). Normal chick embryos were also obtained from Granja Santa Isabel (Cordoba, Spain) and from a white leghorn flock supplied by the S&R egg farm (Whitewater, WI). Eggs were incubated, opened, and staged as described previously(Hamburger and Hamilton, 1951;Ros et al., 2000). Visualization of cartilage patterns was achieved by routine Victoria Blue or Alcian Green staining.

To analyze gene expression in ozd embryos by whole-mount in situ hybridization, we used two methods: the batch method and the hemisection method. By the batch method we analyzed groups of appropriately staged embryos from the ozd flock, of which approximately one quarter should be homozygous for the ozd mutation. For each gene expression analyzed we used a minimum of 16 embryos of the ozd flock, giving a probability of 0.99 that at least one of them is homozygous. The hemisection method was based on the fact that limb buds of ozd homozygous embryos do not express Shh at any stage. Embryos were hemisected along their midline and one half was hybridized for the gene of interest and the other half forShh. Embryos in which Shh was not detected were confirmedozd mutants. In order to analyze gene expression before theozd mutant phenotype was discernible, we surgically removed the right wing buds from embryos in ovo and allowed the embryo to develop to show the phenotype (Carrington and Fallon,1988). Similar results were obtained by all three methods.

Right wing buds of stage (st.) 19-21 embryos from the ozd flock were removed in ovo, and embryos allowed to develop to confirm the phenotype. The isolated buds were incubated in 0.5% trypsin for 1 hour at 4°C to separate the ectoderm from the mesoderm. The isolated ectoderm was recombined with wild-type mesoderm. Using the same approach, isolated limb bud mesoderm from the ozd mutant flock was recombined with ectoderm from wild-type embryos. The recombinant limbs were allowed to heal for 1 hour and then grafted to the flank level somites of host embryos as described previously(Fernandez-Teran et al.,1999).

The ZPA was removed in ovo from st. 19-20 embryos of the ozdflock, and donor embryos were allowed to develop to confirm the phenotype. ZPA grafts were performed as described previously(Tickle, 1981). Heparin acrylic beads (Sigma, H5263) were soaked in recombinant mouse Shh (4 mg/ml). The beads were implanted into the posterior wing bud mesoderm of st. 20 embryos from the ozd flock.

For application of retinoic acid (RA; all-trans-retinoic acid,Sigma), beads (AG1X2, Bio-Rad) were soaked for 20 minutes at room temperature in 0.1 mg/ml, 0.6 mg/ml or 1 mg/ml RA suspended in DMSO and rinsed several times in saline before use. RA-soaked beads were implanted under the AER at the anterior or posterior border of the developing wing and leg buds(Tickle et al., 1985).

Digoxigenin-labeled antisense riboprobes were prepared, and wholemount in situ hybridization analysis performed according to standard procedures(Nieto et al., 1996). S³⁵-labeled riboprobes were prepared and hybridized to tissue sections as described previously(Wilkinson and Nieto, 1993). The probes used were Shh, Fgf4, Fgf8, Bmp2, Hoxd11, Hoxd12, Hoxd13, Gli1,Gli3, Ptc1, Hoxb8 and dHAND (kindly provided by C. Tabin, T. Jessel, J-C Izpisua-Belmonte, P. Beachy and D. Srivastava).

In situ detection of DNA fragmentation was performed using terminal deoxynucleotidyl transferase (TdT) mediated deoxyuridine-triphosphate (dUTP)nick end-labeling (TUNEL) with the In Situ Cell Death Detection Kit,Fluorescein (Boehringer-Mannheim).

Recently, a new mutation was reported in the chicken, characterized by a limb phenotype resembling the Ametapodia mutation(Cole, 1967) and was namedAmetapodia-2 (Smyth et al.,2000). We have performed a detailed analysis of theAmetapodia-2 limbs. Because Ametapodia refers to a dominant mutation resulting in reduced or absent metapodial bones (metacarpal and metatarsal bones) (Cole, 1967;Ede, 1969) we renamed the mutant oligozeugodactyly (ozd) indicating fewer than normal elements at zeugopod and autopod levels. This mutation is inherited as a simple Mendelian recessive trait, and the gross overall morphology ofozd embryos appeared normal except for the limb. ozd mutants hatch normally and are alert, but have impaired mobility; death occurs for unknown reasons within the first days of hatched life [personal observations and (Smyth et al., 2000)].

Limb development in ozd embryos proceeds normally until st. 23/24 when the limb buds become abnormally narrow across the AP axis. The narrowing becomes more evident during subsequent stages of development; by st. 26 the mutant limb buds acquire a pointed and hooked shape; eventually, the mutant limbs adopt a spiked shape (Fig. 1).

Skeletal preparations at 10 days of incubation(Fig. 1A,B) showed ozdmutant wings composed of humerus, radius and a hypoplastic carpal element while the ulna, metacarpals and digits were absent(Fig. 1B). ozd mutant legs displayed femur, tibia, tibiale and first toe with a total absence of fibula, and digits 2, 3 and 4 (Fig. 1B). It is important to emphasize that the skeletal elements present in the mutant limb were of normal morphology and easily recognizable except for the rudimentary carpal. The single element present in the leg autopod showed the characteristic morphologies of the first metatarsal and proximal phalanx of d1, making the identification unequivocal(Fig. 1C). According to the current classification of limb mutations, ozd can be considered a longitudinal postaxial defect (Stoll et al., 1998).

Alcian Green staining of st. 25 and 27 mutant and wild-type limbs failed to detect evidence of cartilage condensations corresponding to the absent skeletal elements in the day-10 mutant limb, indicating the development of these elements was never initiated (Fig. 1D).

To determine whether ozd limb bud narrowing resulted from abnormal cell death we performed TUNEL analysis in wild-type and ozd limbs at st. 24, when the mutant phenotype became discernible(Fig. 2). Wild-type wing buds showed two areas of well-defined mesodermal apoptosis, one in the center of the wing bud, known as the opaque patch (OP), and another along the posterior border called the posterior necrotic zone (PNZ;Fig. 2A)(Fell and Canti, 1934;Hinchliffe, 1982;Hurlé et al., 1995;Saunders and Fallon, 1967). In contrast, comparably staged ozd wing buds showed extensive abnormal apoptosis in the anterior border mesoderm that extended into the distal mesoderm (Fig. 2B) as well as increased apoptosis in the OP (Fig. 2A,B). However, cell death was not detected in the posterior border mesoderm in mutant wing buds (arrow inFig. 2B, compared with the control in Fig. 2A). TUNEL analysis of leg buds gave similar results. st. 24 wild-type leg buds show apoptosis in a fairly extensive anterior zone, called the anterior necrotic zone (ANZ) as well as in the OP and a small PNZ(Fig. 2C). The ozd leg buds showed massive apoptosis along the anterior and distal borders of the limb and increased central cell death (Fig. 2D). No evidence of cell death in the posterior mesoderm was found(arrow in Fig. 2D). During subsequent development of the mutant limb, the anterior-distal area of cell death persisted, but posterior apoptosis was not observed (not shown). The absence of posterior cell death was a surprising result since the shape of the mutant buds gives the appearance of a less substantial posterior border that eventually formed a concavity. Our results indicate that the increased apoptosis in the mutant contributes to the progressive narrowing of the bud to a pointed shape over the course of development. However, the predominantly anterior pattern of apoptosis in the mutant cannot account for the loss of posterior structures characteristic of ozd wings and legs.

To investigate which tissue layer is affected by the ozd mutation we performed recombination experiments interchanging mesoderm and ectoderm between mutant and normal donors(Fernandez-Teran et al.,1999). Control experiments exchanging mesoderm and ectoderm from normal limb buds resulted in completely normal skeletal patterns(Fig. 2E). Recombinant limbs constructed with mutant ectoderm and wild-type mesoderm also developed into limbs with a normal skeletal pattern (Fig. 2F), indicating that the ozd ectoderm is capable of supporting normal development. However, mutant mesoderm recombined with normal ectoderm resulted in limbs exhibiting the mutant phenotype(Fig. 2G). These experiments demonstrate that the mesoderm is defective in the mutant while the ectoderm is capable of normal function.

The lack of posterior elements in both the zeugopod and autopod of theozd mutants indicated a defect along the AP axis, so we began our molecular analysis by looking at Shh expression.

Batch analysis of st. 19 and older embryos revealed that approximately one quarter (14/50) lacked normal posterior Shh expression(Fig. 3E-H). This correlated with the expected percentage of homozygous embryos, suggesting ozdmutants did not express detectable levels of Shh in the limb.

We detected Shh transcripts in st. 17/18 wild-type embryos (cf.Riddle et al., 1993) by whole-mount in situ hybridization (n=10;Fig. 3A). But, since there is some variability in the developmental time at which Shh expression is initiated in the limb bud (cf. Riddle et al., 1993), the batch method was not completely satisfactory for the study of these stages. In order to determine if mutant embryos expressed transient levels of detectable Shh prior to st. 19, we analyzedShh expression in st. 17/18 wing buds of confirmed ozdembryos. For this specific experiment, we removed the right limb buds in ovo and allowed the embryo to develop to determine the phenotype. Confirmedozd limbs were embedded, sectioned and hybridized with³⁵S-labeled Shh riboprobe. We found that Shhexpression was undetectable in all confirmed ozd buds (n=4,Fig. 3D) while control buds,acquired in the same way, expressed Shh (n=11,Fig. 3C). Thus, ozdembryos do not express detectable levels of Shh in the posterior limb bud at any stage. We stress at this point that the defect in Shhexpression is specific for the limb bud, since expression at other embryonic sites, e.g. the floor plate of the neural tube, appeared normal and these structures had no morphological defects(Fig. 1A-B andFig. 3).

We also analyzed the posterior ozd mesoderm for polarizing activity. ZPA grafts from confirmed ozd limbs gave no duplications(n=3, not shown) while ZPA tissue from non-ozd siblings gave the expected digital duplications (n=8, not shown); polarizing activity of 71.8%, calculated according to the method of Drossopoulou et al.(Drossopoulou et al.,2000).

To confirm that Shh was not expressed in ozd limbs, we analyzed the expression of Patched1 (Ptc1) and Gli1, genes directly regulated by Shh and considered to be highly sensitive indicators of Shh signaling (Ingham and McMahon,2001).

During normal limb development Ptc1, the receptor for Shh, andGli1, a target of Shh signaling, are expressed in domains overlapping the expression domain of Shh but extend more anteriorly(Fig. 4A). Using the batch method, we found that roughly 25% of embryos from the ozd flock did not express detectable levels of Ptc1 in the wing bud (5/22;Fig. 4A). Utilizing the hemisection technique, expression of Ptc1 was never detected in st. 18/19 ozd mutant limb buds (st. 18-19, 36-38 somites: n=5;Fig. 5B). Similar toPtc1, we found that roughly 25% of hybridized embryos (3/18) did not express detectable Gli1 in the limb (compareFig. 4C with 4D). These data confirm that detectable Shh activity is not present in ozd limb buds.

Gli3 and Shh have mutually exclusive expression domains in the developing limb and are believed to repress one another's expression(Büscher et al., 1997;Marigo et al., 1996a;Masuya et al., 1997;Schweitzer et al., 2000). By the batch method, at st. 18/19, no differences in Gli3 expression were detected among embryos of the ozd flock(Fig. 4E). This was confirmed in mutant limb buds (n=4) as early as late st. 18/19 (37-40 somites)by hemisection analysis. However, at st. 21, ozd embryos failed to down-regulate Gli3 expression at the posterior border of the limb(25% of batch, Fig. 4F).

We next compared expression patterns of the bHLH transcription factor dHAND which has been proposed to act upstream of Shh and establish a positive feedback loop with Shh later in development(Charité et al., 2000;Fernandez-Teran et al., 2000). Expression of dHAND in the ozd limbs started normally (batch method), but then was reduced to a weak domain of expression restricted to the posterior border of the limb, in a very similar pattern to that observed in the limbs of the Shh^-/- mice(Fig. 4G,H)(Charité et al., 2000;Fernandez-Teran et al.,2000).

We also analyzed the expression patterns of other genes considered to be major downstream targets of Shh. Bmp2, previously thought to be a downstream target of Shh, was expressed in the mesoderm and AER of both mutant wing and leg buds as early as st. 18/19(Fig. 4I,J). It was expressed in a reduced area and at a slightly lower level than normal as determined by both batch (n=2/4, st. 20-23) and hemisection methods (st. 19, 37-39 somites; n=3).

Expression of Hoxd11-13 was also analyzed in ozd limbs. Using the batch method, it was determined that Hoxd11-13 expression was initiated in a temporally and spatially normal pattern (not shown), but progressively declined with time (Fig. 5). Hoxd11 pattern of expression was virtually normal inozd wings up to st. 25, although its level of expression was slightly reduced compared to wild-type (Fig. 5A-B). During subsequent stages Hoxd11 expression in the mutant wing was restricted to the posterior border (Fig. C,D). In theozd leg bud Hoxd11 expression was very reduced compared to wild type at st. 21/22 (Fig. 5A), becoming undetectable at st. 24/25(Fig. 5B-D). Hoxd12expression was reduced in the ozd wing buds as early as st. 21/22(Fig. 5E) and its expression continued restricted to the posterior border(Fig. 5F-H). Expression ofHoxd12 was much more affected in the mutant leg where it became undetectable at st. 24 (Fig. 5E-H). In the mutant wing and leg, Hoxd13 expression was very reduced and became undetectable by st. 23/24(Fig. 5I-L). Interestingly,Hoxd13 was re-expressed in the distal mutant leg mesoderm at st. 27(Fig. 5K) and persisted in the distal leg mesoderm (Fig. 5L). Re-expression of Hoxd11 or 12 was never observed.

Genes involved in PD and DV patterning were normally expressed inozd limbs. For PD specification we analyzed the expression ofMeis1 and 2 and Hoxa11 and Hoxa13 genes. We found that expression of Meis1 and 2 was not modified inozd limbs (not shown). While the expression of Hoxa11 was normal in ozd wing buds, the expression of Hoxa13,considered a marker for the autopod, was dramatically diminished to a thin low-level stripe of distal expression in the mutant wing mesoderm. In the mutant leg Hoxa13 expression was similar to normal (not shown). For DV specification we analyzed the expression of Wnt7a andLmx1; both showed a normal pattern of expression in ozdlimbs (not shown).

Although our molecular characterization and experimental study of theozd mutant limb indicates that the defect is in the mesoderm,reciprocal interactions between the mesoderm and the AER are well documented(Deng et al., 1997;Ohuchi et al., 1997). Therefore, we analyzed the expression of Fgf8 and Fgf4 in the AER of ozd limbs. The mutant AER always expressed high levels ofFgf8 throughout development of both the wing and leg(Fig. 6A-F). Coincident with the progressive narrowing of the mutant limb, the posterior extent of the AER was reduced, showing an abrupt end at the posterior border at the point of the posterior concavity in the mutant limb shape(Fig. 6B). Fgf8expression persisted in the mutant AER up to st. 27 in the wing and st. 28 in the leg. At later stages, Fgf8 was dramatically reduced throughout the anterior AER (Fig. 6C). The anterior loss of Fgf8 together with its reduced posterior extension resulted in a discrete point of Fgf8 expression at the very tip of the mutant limbs at st. 28 (not shown). The expression of Fgf4appeared reduced except in the most posterior of the mutant AER(Fig. 6D), where a spot of elevated expression became apparent by st. 22/23(Fig. 6E). Fgf4expression was not maintained in the mutant AER and declined with time, so that by st. 25 it was undetectable except for residual levels of expression in the posterior spot of high-level expression seen at st. 22/23(Fig. 6F, compare withFig. 6E).

Recently, it was proposed that Fgf4 upregulation by Shh in the posterior AER is mediated by the BMP antagonist Gre and expression ofGre in the limb mesoderm is considered necessary for AER maintenance(Capdevila et al., 1999;Zúñiga et al.,1999). During development of the ozd limb budsGre expression appeared reduced and restricted to the posterior border (Fig. 6G-I) as confirmed by the hemisection technique (st. 20/21, 40-44 somites; n=5). This pattern of Gre expression is similar to that reported in theShh mutant mice(Zúñiga et al.,1999) and is consistent with the reduced Fgf4 expression observed in ozd limb buds.

Since Shh expression and signaling is undetectable in mutant limbs, we tried to rescue the mutant phenotype by grafting a normal ZPA or applying exogenous SHH-N to the posterior border of st. 20 mutant limb buds. ZPA fragments from st. 20 wild-type limb buds were grafted under the posterior AER of either the wing or leg of embryos from the mutant flock. For wings,pieces of leg ZPA were used and for legs, pieces of wing ZPA were used. When the ZPA was grafted to an ozd limb, the mutant phenotype was restored to normal (n=2; Fig. 7B). In the specimen showed inFig. 7B, the piece of ZPA of leg origin has also formed a digit characteristic of the leg (asterisk inFig. 7B). The appearance of a digit of graft (leg) origin may occur if the grafted ZPA is large. ZPA grafts into the ozd leg buds gave equivalent results (not shown).

Next, heparin acrylic beads loaded with SHH-N protein (4 mg/ml) were applied to the posterior border, attempting to mimic a normal ZPA. In these cases, a total pattern restoration of the AP axis was observed at the zeugopod level with formation of a normal ulna, and improved development of carpals,although the limbs were truncated at wrist level(Fig. 7C). The sequential application of a second SHH-N loaded bead 24 hours after the first restored wing patterning at both zeugopod and autopod levels(Fig. 7D).

Retinoic acid (RA) induces Shh expression when applied to the anterior wing bud mesoderm (Helms et al.,1994; Riddle et al.,1993) and is implicated in the normal induction of the ZPA(Lu et al., 1997;Stratford et al., 1997). We applied RA to either the anterior or posterior mesoderm of st. 20/21ozd limb buds to determine if Shh could be induced and the mutant phenotype rescued. We first applied beads soaked in RA (0.1 and 1 mg/ml) under the posterior AER. In wild-type limb buds (n=10), the level of normal Shh expression was reduced when analyzed 24 hours after the operation (Fig. 7E)and resulted in a range of skeletal alterations varying from a loss of digits(n=9) to the complete inhibition of outgrowth (n=1). These data are consistent with previous reports(Tickle et al., 1985). Application of an RA bead to the posterior mesoderm of ozd wings did not induce Shh expression after 24 hours (n=3;Fig. 7F) and resulted in the total absence of the right wing (n=2; not shown).

RA-soaked beads (0.1 and 1 mg/ml) applied to anterior ozd limb mesoderm did not induce Shh at 24 hours (n=2; compareFig. 7H withFig. 7G) or 48 hours(n=1; not shown) after the operation and the mutant phenotype was not modified (n=2). It has been shown in the wing that the induction ofShh by RA may be mediated by the early, transient activation ofHoxb8 (Lu et al.,1997; Stratford et al.,1997), and that it is also preceded by the activation ofdHAND expression (Fernandez-Teran et al., 2000). Thus, we analyzed at what point RA induction ofShh failed in the mutant. RA-soaked beads (1 mg/ml and 0.6 mg/ml)were placed in the anterior border of wing buds, and embryos were fixed after 5 hours to analyze Hoxb8 expression, and after 12 or 20 hours to analyze dHAND expression. Hoxb8 was normally expressed byozd limb mesoderm in response to RA signaling (n=5;confirmed by hemisection technique; not shown). RA applications also induceddHAND expression in the anterior mutant limb mesoderm, similar to the normal limb (n=5; Fig. 7I-L). These observations indicate that the ozd mutation lies downstream of Hoxb8 and dHAND activation by RA.

The anatomical, molecular and experimental analyses presented here indicate that ozd limbs develop in the absence of Shh signaling. Our data establish that the defect in the Shh signaling pathway lies upstream ofShh transcriptional activation, suggesting the ozd mutation affects a regulatory element that controls limb-specific expression ofShh. Our analysis further demonstrates that the limb buds develop with an AP identity independent of Shh function. The identification of a naturally occurring `targeted knockout' of Shh in the developing limbs of a experimentally tractable model system offers a unique tool to address its role in amniote limb patterning.

Bmp2 and 5′ Hoxd genes are considered to be downstream effectors of Shh signaling since Shh application to the anterior border induces their ectopic expression(Laufer et al., 1994;Riddle et al., 1993;Yang et al., 1997). We report that in ozd limbs these genes are activated in a pattern similar to that in normal limbs. The 5′ Hoxd genes were also shown to be asymmetrically expressed in the limbless mutant limb bud in the absence of detectable Shh expression(Grieshammer et al., 1996;Noramly et al., 1996;Ros et al., 1996) and in theShh^-/- mouse (Chiang et al., 2001; Kraus et al.,2001). Phase II of 5′ Hoxd genes expression,proposed to be Shh dependent (Nelson et al., 1996), starts normally in ozd but is not fully developed and expression declines with time. Phase III of expression, which corresponds to the autopod (Nelson et al.,1996), is dramatically affected. The more 5′ theHoxd gene, the earlier and more severely its pattern of expression is affected. For example, in the st. 25 ozd wing bud, Hoxd11 is expressed in a pattern similar to normal, while Hoxd12 andHoxd13 expression is progressively diminished. This may indicate a progressive differential requirement for Shh among 5′ Hoxdgenes. However, it is of interest that the distal tip of the ozd leg bud re-expresses Hoxd13 at later stages correlating with the formation of d1 and, interestingly, precedes activation of Indian hedgehog (Ihh) in the digital cartilage (data not shown). This late Hoxd13 expression was also reported to occur in theShh^-/- hindlimb(Chiang et al., 2001;Kraus et al., 2001). Also,dHAND is expressed in a reduced but posteriorly polarized domain of expression in ozd limb buds. Thus, activation and polarization ofBmp2, the 5′ Hoxd and dHAND expression in the posterior limb bud does not require Shh and reflects AP patterning asymmetries in the early limb bud that are independent of Shh. However, Shh inputs are required to stabilize and augment initial gene expressions so that the AP polarization of the limb bud is realized.

Because Shh signaling is absent in the limbs of ozd embryos, it is useful to compare the limb phenotype of ozd mutants andShh^-/- mice (Chiang et al., 2001; Kraus et al.,2001). Interestingly, both types of limbs show a very similar phenotype forming a complete PD axis with a normal stylopod. One digit,identified as d1, forms in the Shh^-/- hindlimb(Chiang et al., 2001;Kraus et al., 2001;Lewis et al., 2001) and also d1 forms in the ozd leg. The main differences between ozdand Shh^-/- limbs occur at the zeugopod. The skeletal elements in the zeugopod of the Shh^-/- mice (one in forelimb and two in hindlimbs) are abnormal while the morphology of the single fore and hindlimb zeugopod element in ozd mutants are virtually normal. Despite the differences, both genotypes demonstrate the necessity for Shh distal to the elbow/knee region, since either loss of AP identity and/or posterior deficits are observed without it. Thus, it is possible to classify the skeletal elements of the limb according to their requirement for Shh signaling. The ozd mutation indicates that in the chick the humerus/femur, radius/tibia and d1 are Shh independent, while the ulna/fibula and rest of the digits require Shh inputs for normal development(Fig. 8). However, the Shh-independent potential of the limb varies between chick and mouse at the zeugopod level since the element that forms in chick is well shaped while it is unidentifiable in mouse.

Experimental removal of the posterior wing mesoderm in chick, including the whole ZPA leads to limbs with a phenotype very similar to ozd limbs(Pagan et al., 1996;Todt and Fallon, 1987). The operated wings form a normal radius with or without d2 and since the surgery is performed at st. 20, before the determination of the zeugopod(Summerbell, 1974), it can be concluded, on the basis of various approaches to this issue, that a completely normal radius can develop in the chick in the absence of Shh input.

Morphological differences between the wing and the leg reflect differences in the response to common molecular signals that pattern them. Moreover, wing buds and leg buds may respond differently to experimental manipulation (e.g.Todt and Fallon, 1987;Wada and Nohno, 2001). The formation of a properly patterned digit in the leg but not the wing indicates that Shh is required for the most anterior digit to form in the wing. The identity of the three avian wing digits remains controversial(Burke and Feduccia, 1997) (see also Kundrát et al.,2002; Larsson et al., 2002). However, if we assume the conventional nomenclature of d2, d3, d4, our hypothesis that d1 is Shh independent predicts no wing digits will develop in the absence of Shh function. Admittedly, the loss of d1 in the Shh^-/- mouse forelimb is difficult to explain. It is possible that global loss of Shh function has more deleterious effects on limb development than limb-specific loss of Shh function alone. A conditional null of Shh in the mouse limb will permit a direct comparison of the mouse with the ozdlimb.

Removal of posterior mesoderm was shown to cause cell death similar to our findings for ozd and was attributed to the loss of ZPA function(Todt and Fallon, 1987). It is notable that grafting a bead loaded with Shh protein prevents normal anterior cell death in the chick wing(Sanz-Ezquerro and Tickle,2001), suggesting a role for Shh in regulating cell death in the limb.

Abnormal cell death correlates with the progressive narrowing ofozd limbs. Interestingly, while the anterior mesoderm undergoes increased apoptosis, neither the PNZ nor abnormal cell death are detected in the posterior border. However, there is a significant change in the shape of the posterior border, most notably in the leg, where a concavity forms that contributes to the spike shape of the ozd phenotype. Determination of the mechanism of posterior limb bud shape change is made more challenging by the observation that there are no gross differences in BrdU incorporation in posterior cells as compared to wild type at the stages examined (st. 19, 23 and 25; not shown). It is possible that those cells that will later contribute to posterior structures failed to proliferate and were left behind, beginning slightly before the phenotype becomes obvious, around st. 23/24. A slight change in proliferation at st. 17 and 18, or even at the stages analysed with BrdU, but below a detectable level could still account for the loss of posterior structures. Clarification of this point will require further investigation. Also, it is worth mentioning that a mitogenic effect for Shh has been reported in several developing systems(Bellusci et al., 1997;Duprez et al., 1998;Jensen and Wallace, 1997) and that Hh signaling can induce proliferation during development by promoting expression of cyclin D and cyclin E(Duman-Scheel et al., 2002). Thus, in the absence of Shh, stimulus from the AER would not be sufficient to support enough mesoderm to permit the specification of the whole anterior-posterior axis.

Disruptions in AP limb pattern are among the most common human birth defects (Castilla et al., 1996;Castilla et al., 1998), and understanding the affected developmental mechanisms is of significant clinical importance. Interestingly, studies in human and mouse have mapped several mutations and transgene insertions causing limb-specific AP patterning defects to a syntenic locus near or within the Limb region 1 (Lmbr1)gene, located less than 1 Mbp from the Shh coding region[(Clark et al., 2001;Lettice et al., 2002), and references therein]. Recent genetic analyses demonstrate the Lmbr1gene is incidental to the limb phenotypes; rather, evidence suggests these mutations affect long-range cis regulatory elements, embedded within the Lmbr1 locus, that control Shh expression in the limb. While the majority of these mutations cause dominant pre-axial polydactyly,the small deletion responsible for the autosomal recessive human disorderAcheiropodia maps within the Lmbr1 locus(Ianakiev et al., 2001), and causes longitudinal postaxial deficiencies closely resembling the limb phenotypes of ozd chicks and Shh^-/- mice. Here we have shown that ozd limb mesoderm is incapable of expressingShh, clearly indicating that the mutation affects a limb-specific regulatory element of Shh expression. Although the data presented here are compatible with the mutation affecting either a cis- ortrans-acting element, we hypothesize that the ozd mutation disrupts a cis-acting regulatory element directing Shhexpression in the limb, which lies within the Lmbr1 locus such as inAcheiropodia individuals(Ianakiev et al., 2001); this hypothesis is currently being investigated.

This work was supported by NIH Grant No. 32551 to J. F. F.; R. D. D. was supported in part by a Cremer Fellowship from the University of Wisconsin Medical School. Work in M. A. R.'s laboratory is supported by grants DGICYT-PM98-0151 and FIS 01/1219 from the Spanish Ministry of Science. We thank Prof. John Opitz for helpful discussions on naming the mutant, the Florsheim family for their financial support of this research, Allen W. Clark for help with Fig. 8 and S&R Egg Farm (Whitewater, WI) for providing the White Leghorn flock. This paper is dedicated to the memory of Genie Fallon Hall.