Loss of the Tg737 protein results in skeletal patterning defects



Tg737 mutant mice exhibit pathologic conditions in numerous tissues along with skeletal patterning defects. Herein, we characterize the skeletal pathologic conditions and confirm a role for Tg737 in skeletal patterning through transgenic rescue. Analyses were conducted in both the hypomorphic Tg737orpk allele that results in duplication of digit one and in the null Tg737Δ2-3βGal allele that is an embryonic lethal mutation exhibiting eight digits per limb. In early limb buds, Tg737 expression is detected throughout the mesenchyme becoming concentrated in precartilage condensations at later stages. In situ analyses indicate that the Tg737orpk mutant limb defects are not associated with changes in expression of Shh, Ihh, HoxD11–13, Patched, BMPs, or Glis. Likewise, in Tg737Δ2-3βGal mutant embryos, there was no change in Shh expression. However, in both alleles, Fgf4 was ectopically expressed on the anterior apical ectodermal ridge. Collectively, the data argue for a dosage effect of Tg737 on the limb phenotypes and that the polydactyly is independent of Shh misexpression. Developmental Dynamics 227:78–90, 2003. © 2003 Wiley-Liss, Inc.


During limb development, the presumptive limb bud mesenchyme induces overlaying ectoderm along the anterior–posterior (AP) axis to form the apical ectodermal ridge (AER) required for limb outgrowth. The proteins that mediate AER signaling are the fibroblast growth factors (Fgf). Fgf4 and Fgf8 are expressed in the AER, and transplanted beads impregnated with these factors can replace the AER after its removal (Niswander et al., 1993; Cohn et al., 1995; Makarenkova and Patel, 1999). The Fgfs influence limb patterning by inhibiting apoptosis, regulating proliferation, directing chemotaxis, and by regulating gap junction activity (Niswander et al., 1993; Cohn et al., 1995; Li and Muneoka, 1999; Makarenkova and Patel, 1999). Two models have been proposed on how the AER regulates limb morphogenesis. According to one model, the AER maintains the subjacent mesenchyme in a region called the progress zone (PZ) in a mitotically active state, whereas the cells more distal to the AER begin to differentiate (Wolpert, 1999). Thus, the AER may direct proximal–distal patterning by the amount of time a cell spends in the PZ. In contrast, recent data suggest that most limb segments are established early as groups of progenitor cells that then undergo selective AER-mediated expansion (Dudley et al., 2002; Sun et al., 2002).

The limb bud mesenchyme also plays a role in establishing limb patterning. A region of the posterior bud called the zone of polarizing activity (ZPA) influences the AP axis, as evidenced by digit duplications resulting from transplantation of ZPA cells to the anterior side of the limb bud. The mediator of the ZPA activity is likely the protein sonic hedgehog (Shh), because its domain of expression colocalizes with the ZPA and Shh ectopic expression on the anterior margin of the limb bud correlates with induction of additional digits (Riddle et al., 1993; Chan et al., 1995; Pearse and Tabin, 1998; Sharpe et al., 1999).

The affect of the AER and ZPA on limb patterning is mediated through regulation of gene expression in the limb bud mesenchyme (reviewed by Ng et al., 1999). Potential targets are genes in the HoxD cluster, because their expression is induced by Shh and they are aberrantly expressed in mice and chick exhibiting limb defects (Riddle et al., 1993; Chan et al., 1995; Qu et al., 1997; Buscher and Ruther, 1998; Sharpe et al., 1999). HoxD genes are normally expressed in an AP pattern such that the genes located most 3′ are expressed more anteriorly than those located 5′ in the cluster (Nelson et al., 1996). In general, mutations in the HoxD genes (D11, D12, and D13) have not caused homeotic transformation of digits but rather localized malformation in the length, shape, and number of limb bones (Davis and Capecchi, 1994, 1996; Davis et al., 1995; Fromental-Ramain et al., 1996).

In addition to Hox genes, the AER and ZPA also regulate members of the Gli family (Laufer et al., 1994; Masuya et al., 1995; Buscher and Ruther, 1998; Theil et al., 1999). Gli1 and Gli3 expression is regulated by Shh, and these genes are often ectopically expressed in polydactyly mutants (Buscher and Ruther, 1998). The semidominant extra toes (xt) mutation is an intragenic deletion in Gli3. xt heterozygotes have preaxial duplication of a single digit, whereas homozygous mutants develop as many as eight digits per limb. The excessive digit formation in homozygotes is associated with expansion of HoxD13 expression and ectopic anterior Shh expression.

Numerous secreted signaling molecules whose function is required for normal patterning of the limb have been identified (Crossley and Martin, 1995; Marti et al., 1995; Pearse and Tabin, 1998; Zuniga et al., 1999). Often these molecules are released from signaling centers and are thought to form a gradient across the limb bud dictating changes in gene expression and establishing limb pattern. What remains uncertain is how the cells within the limb bud are able to sense their position and differentiate into the appropriate structures. Elucidating the process by which the limb bud cells determine their fate will require molecular and cellular analysis of novel limb bud mutants that result from defects in receptors or other nonsecreted proteins that act in a cell autonomous manner (Wada et al., 1999a). One such mutant is likely the Oak Ridge Polycystic Kidney (Tg737orpk) mouse. Surprisingly with regard to skeletal patterning, the Tg737 gene encodes a protein called polaris that localizes to basal bodies and cilia and is required for assembly of cilia (Taulman et al., 2001; Yoder et al., 2002). Homozygous Tg737orpk mice develop a complex series of pathologies involving the kidney, liver, brain, testis, eye, pancreas, and skeleton (Moyer et al., 1994; Yoder et al., 1995). Herein, we characterize the skeletal defects that include craniofacial abnormalities, cleft palate, supernumerary teeth, and preaxial duplication of digit one. Previous analysis of the Tg737orpk mouse indicates that it is a hypomorphic allele of Tg737 (Moyer et al., 1994; Yoder et al., 1995; Taulman et al., 2001). In contrast to the hypomorph, the Tg737Δ2-3βGal mutation represents a null allele. Homozygous Tg737Δ2-3βGal mutants die in early to midgestation and exhibit random left–right body axis specification, failure of neural tube closure and as demonstrated here, a pronounced expansion of the AP limb bud axis with as many as eight digits on each limb (Murcia et al., 2000).

Although the kidney and liver pathologic states of Tg737orpk mutant mice and the left–right patterning defects in Tg737Δ2-3βGal mutants have been described in detail, the skeletal anomalies have not been characterized (Moyer et al., 1994; Yoder et al., 1995, 1996, 1997, 2002; Richards et al., 1996, 1997, 1998). Here, we characterize the skeletal patterning defects associated with both Tg737 mutants, demonstrate a role for Tg737 in the skeletal patterning by transgenic rescue, and investigate the possible mechanism behind the skeletal pathologic conditions by analyzing the temporal–spatial expression pattern of Tg737 and other genes known to be involved in limb morphogenesis. Unexpectedly, we could detect no overt changes in Shh expression in either of the Tg737 alleles, even though eight digits are present on the limbs of the null mutants. However, in both mutant lines, Fgf4 was ectopically expressed on the anterior margin of the limb bud. Together, the data suggest that misexpression of Fgf4 leads to the limb pathologic state independent of ectopic Shh activation, that the dramatically different limb phenotypes seen in the null and hypomorphic allele is due to a dosage effect of Tg737, and finally that cilia may play a role in normal limb patterning.


Tg737orpk Mutants Exhibit Defects in the Patterning of the Skull and Teeth

To characterize the skeletal phenotype, we compared 12 alizarin red–stained skeletons from Tg737orpk mutant and wild-type mice. Whereas no overt defects were evident in the vertebral column or ribs, there were numerous skull abnormalities. The frontal and parietal sutures were disorganized and failed to fuse (Fig. 1A). In addition, 10 of the 12 mutant skeletons revealed minor clefting of the palatine process and all mutants lacked an opening in the periotic capsule surrounding the inner ear (Fig. 1B,C). Finally, mutants had supernumerary teeth (Fig. 1D). Normally, mice have three molars, the most rostral molar is the largest and has three cuspids, and the most caudal molar is the smallest and has a single cuspid. In contrast, the mutants had four molars on both upper and lower jaws. The extra tooth is located rostral to what is normally the initial molar and is composed of a single cuspid.

Figure 1.

Comparison of alizarin red–stained skulls in Tg737orpk (left column), wild-type (center column), and rescue (right column) mice. A: In the Tg737orpk mutants, the frontal and parietal sutures were disorganized and failed to fuse (arrows). In contrast, the sutures in rescue animals always fused but remained slightly disorganized (arrow in right column). B: Clefting of the palatine process was detected in the Tg737orpk mutants (arrow). This abnormality was corrected in the rescue animals. C:Tg737orpk mutants lack a large foramen in the periotic capsule (arrow, left column). This foramen was present in all rescue and wild-type mice (arrows, center and right columns). D:Tg737orpk mutants have an extra molar tooth (labeled as 0) in both upper and lower jaws (shown is the lower right jaw). In contrast to Tg737orpk mutants, all rescue animals had the normal number of molars (labeled 1–3, right column).

Abnormalities in the Fore and Hindlimbs of Tg737orpk Mutant Mice

In addition to the skull defects, Tg737orpk mutants have abnormalities in the appendicular skeleton. In the autopod, the abnormality involves an extra preaxial digit on both the fore- and hindlimbs. In the forelimbs, the polydactyly is often seen as a fusion with digit I (Figs. 2A, 4). In contrast, the hindlimb polydactyly is a complete duplication containing two phalanges and their corresponding metatarsals, thus resembling digit I (Figs. 2B, 4). The hindlimb abnormality is evident early in development as an increased AP axis, with an expansion on the anterior side of the limb bud (Fig. 3).

Figure 2.

Skeletal analysis of the forelimb (top) and hindlimb (bottom) of Tg737orpk mutant (left column), wild-type (middle column), and rescue (right column) mice. A: The preaxial duplicated digit in the forelimb of Tg737orpk mutants is observed as a fusion with digit I into a single wider bone (digit I, arrow, left column). This abnormality was corrected in rescue mice (arrow, right column). The abnormality in digit II is localized to the second phalange (digit II, P2, arrow, left column). The size of this phalange was reduced in Tg737orpk mutants compared with wild-type controls (arrow, center column). In rescue animals, the size of P2 was generally corrected; however, in a few rescue animals, a small degree of reduction was detected (arrow, right column). In digit V, P2 was also affected where it was reduced in size or fused with P1 (digit V). In many cases, P2 was completely absent. Unlike digit II, there was no correction in the size of P2 of digit V in rescue animals. Many Tg737orpk mice (5 of 12) exhibited an abnormal fusion between the lesser multangular and the centrale (carpals, arrowhead, left column) along with morphologic changes in the size and shape of the hamate (not shown). These defects were corrected in rescue mice. B: Hindlimb. The ectopic preaxial digit (arrows) in the hindlimb of Tg737orpk mutants resembles digit I as evidenced by the presence of two phalanges and corresponding metatarsals (digit I, arrow). No preaxial duplications were observed in the hindlimb of rescue animals; however, the size of digit I was reduced compared with wild-type controls (arrow, center column). In digit II, the size of P2 was slightly reduced in mutants and was fully corrected in rescue animals (digit II, arrows). The decrease in the size of digit V in Tg737orpk mutants was also due to defects in P2 (digit V, arrow, left column). The defect in P2 was essentially corrected in rescue animals (digit V, arrow, right column); however, minor reduction was seen in a few rescue mice. All Tg737orpk mutants analyzed had abnormalities in tarsal bone patterning (tarsals). Four or five aberrant bones (tarsals, arrows, left and right columns) were located at the distal ends of the tarsals. In addition, there were fusions between the third cuneiform and the navicular (tarsal, arrowhead, left column) or cuboid and navicular (not shown). In contrast to Tg737orpk mutants, no fusions were detected in the navicular, cuboid, or third cuneiform bones; however, two of the aberrant bones (Tg737orpk rescue, arrows, right column) at the distal end of the tarsals were still present. Tg737orpk mutant mice exhibit defects in the zeugopod, i.e., tibia and fibula completely failed to fuse (Tibia/Fibula, arrow, left column). This defect was corrected in the rescue mice (Tibia/Fibula, arrow, right column).

Figure 4.

Alizarin red–stained skeletons from 4-day-old Tg737orpk mutant (left) and wild-type (right) mice analyzed for the formation and size of the ossification centers. Tg737orpk mutants had a significant delay and reduction in size of the ossification centers. The anterior side is denoted by arrows. The ossification center that corresponds to phalange 2 (P2) on digit II is marked by asterisks, and P2 on digit V is marked by pound signs. Note complete loss of the ossification center for P2 on the forelimb of digit V in the mutant and significant reduction in the hindlimb.

Figure 3.

Comparison of the wild-type (left), Tg737orpk (middle), and Tg737Δ2-3βGal (right) limb buds from embryonic day 12.5 embryos. In the Tg737orpk mutants, the expansion is mainly on the anterior side, whereas in the Tg737Δ2-3βGal mutants the expansion appears to be both anterior and posterior. The anterior side is toward the left (labeled A).

Patterning abnormalities were also evident in other structures of the distal autopod. To evaluate these defects, we measured the bone length of each phalange along with its metacarpal or metatarsal. To correct for the growth retardation of Tg737orpk mutants, we standardized the length of each phalange to its corresponding metacarpal or metatarsal (P:M ratio). The most dramatically affected structure on both the fore- and hindlimb was phalange 2 (P2) of digits II and V. On the mutant forelimb, the length of P2 was severely reduced or completely absent (digits II/V, Fig. 2A; Table 1). On the hindlimb, P2 on digit V was also affected, but the length of P2 on digit II was only mildly reduced (digits II/V, Fig. 2B; Table 2). No significant differences in the length of the other phalanges relative to their metacarpals or metatarsals were evident. These abnormalities could be detected in newborn mutant mice as a reduction in size and delay in formation of ossification centers (Fig. 4).

Table 1. Phalange to Metacarpal Ratio for Digits II and V of Tg737orpk, Wild-Type, and Rescue Micea
ForelimbMutantWild typeRescue
  • a

    Due to the growth inhibition associated with the Tg737orpk mutation, phalange (P) length was standardized to the corresponding metatarsal (M) or metacarpal. The P:M ratio was calculated for each digit on both right and left fore- and hindlimbs from a minimum of 12 adult animals in each class. Data are presented as the mean ± 1 standard deviation.

Digit II Phalange II0.31 ± 0.050.55 ± 0.030.53 ± 0.14
Digit V Phalange II0.10 ± 0.050.70 ± 0.070.11 ± 0.12
Table 2. Phalange to Metatarsal Ratio for Digits II and V of Tg737orpk, Wild-Type, and Rescue Micea
HindlimbMutantWild typeRescue
  • a

    Due to the growth inhibition associated with the Tg737orpk mutation, phalange (P) length was standardized to the corresponding metatarsal (M) or metacarpal. The P:M ratio was calculated for each digit on both right and left fore- and hindlimbs from a minimum of 12 adult animals in each class. Data are presented as the mean ± 1 standard deviation.

Digit II Phalange II0.30 ± 0.030.31 ± 0.020.32 ± 0.02
Digit V Phalange II0.21 ± 0.050.31 ± 0.030.28 ± 0.03

Another defect associated with the autopod involved the tarsals and carpals, which were malformed or fused with neighboring bones. In the forelimb, 5 of 12 mutants had a fusion between the lesser multangular and centrale and 2 of 12 had morphologic changes in the hamate (carpals, Fig. 2A). In the hindlimb, the second cuneiform was malformed (5 of 12), and the third cuneiform and the navicular (8 of 12) or the cuboid and the navicular (4 of 12) were fused (tarsals, Fig. 2B). In addition, mutants had four or five aberrant bones located at the distal end of the tarsals along with numerous ectopic sesamoid bones (tarsals, Fig. 2B). Finally, in Tg737orpk mutants, the tibia and fibula failed to fuse (tibia/fibula, Fig. 2B). Normally, these bones fuse one third of the way up the tibia.

Rescue of the Skeletal Defects

To confirm that the Tg737orpk skeletal pathologic condition is due to disruption of Tg737, we crossed transgenic lines that express the wild-type Tg737 cDNA under control of the human β-actin promoter (BAP-737) or the endogenous Tg737 promoter fragment (Tg737RSQ; Yoder et al., 1996, 1997) and crossed them with animals carrying the Tg737orpk mutation. Mice homozygous for Tg737orpk that also carry one of the transgenes expressing wild-type Tg737 cDNA (rescue transgenes) are referred to as “rescue” animals. Thirteen skeletons were analyzed from each class of rescue mice. Identical results were obtained for both the Tg737RSQ and the BAP-Tg737 lines, and data are shown for the analysis of the BAP-Tg737 mice only.

Analysis of the rescue mice indicates that the majority of skeletal defects are corrected or strongly influenced by the rescue transgene. In the skull, 7 of 13 rescues had complete correction of the frontal and parietal sutures (Fig. 1A). In the remaining mice, the sutures showed a minor degree of disorganization. Furthermore, all rescue mice had proper dentition, they had no clefting of the palatine process, and they contained the foramen in the periotic capsule (Fig. 1B–D). In the appendicular skeleton, none of the rescue mice exhibited the preaxial polydactyly or fusion of tarsal or carpal bones (carpals, Fig. 2A and tarsals, Fig. 2B). Although aberrant bones were still detected distal to the tarsals in 11 of 13 rescue mice analyzed, instead of the four or five bones seen in the mutants, only two bones were detected in rescues. Similarly, 7 of 13 rescue animals showed normal distal fusion between the tibia and fibula, and in the remaining mice, the fusion occurred at a slightly more distal position than normal (tibia/fibula Fig. 2B). There was no significant difference between the lengths of P2 on digit V of the hindlimb in rescue and wild-type mice (digit V, Fig. 2B; Table 2). The only mutant limb traits that were not substantially influenced were the truncation of P2 on digit V and P2 on digit II (Table 1) of the forelimb.

Limb Phenotype in Tg737Δ2-3βGal Null Embryos

Data from several studies suggest that Tg737orpk is a hypomorphic allele as evidenced by the continued expression of alternatively spliced transcripts and by the presence of low levels of the polaris protein in Tg737orpk mice (Murcia et al., 2000; Taulman et al., 2001; Pazour et al., 2002). Additionally, mice homozygous for the Tg737orpk mutation are viable, but mice homozygous for the targeted Tg737Δ2-3βGal null allele die in early to midgestation (Murcia et al., 2000). To determine whether the severity of the limb phenotype differs in the hypomorphic and null alleles, we compared the limb morphology of Tg737orpk and Tg737Δ2-3βGal embryos. The limbs in Tg737orpk mutants show minor expansion that appears mainly on the anterior side of the limb (Fig. 3). In contrast to the minor phenotype in Tg737orpk mice, in Tg737Δ2-3βGal mutants, the limbs appear paddle-like with a dramatic expansion on both the anterior and posterior sides of the bud (Fig. 3). In the few Tg737Δ2-3βGal embryos surviving past E11.5, we normally detect eight digits on each limb (Fig. 3). Due to the early lethality of the Tg737Δ2-3βGal mutants, it was not possible to assess the identity of the digits. Similar to the results with the Tg737orpk mice, expression of the rescue transgene in the Tg737Δ2-3βGal mutants corrected the embryonic lethality, the left–right axis defects, and limb pathologic conditions (Murcia, unpublished observations).

Expression of Tg737

To analyze Tg737 expression, we used the β-galactosidase reporter gene incorporated in the Tg737Δ2-3βGal allele (Murcia et al., 2000). Previous polaris immunofluorescence and in situ hybridization analyses verified that the β-galactosidase activity paralleled endogenous polaris expression in multiple tissues (Taulman et al., 2001). At E8.5, Tg737 appears to be predominantly located the late primitive streak/presomitic mesoderm and in the condensing somites (Fig. 5A). At E9.5 through E12.5, there was a general increase in β-galactosidase activity throughout the embryo (Fig. 5B). The β-galactosidase staining appears to be indicative of Tg737 expression in that identical results were obtained by in situ hybridization analysis (data not shown) and because there is no β-galactosidase activity detected in wild-type littermate controls (Fig. 5C).

Figure 5.

Expression of Tg737 was analyzed in embryos by using the β-galactosidase reporter gene in Tg737Δ2-3βGal heterozygous mice at embryonic day 8.5 (E8.5) through E12.5. A: At E8.5, β-galactosidase activity was concentrated around the areas of the late primitive streak and presomitic somatic (arrow) during late neurulation, in the somites (arrowhead), and at lower levels throughout the embryo. B: At E11.5, β-galactosidase activity increased throughout much of the embryo. C: No β-galactosidase activity was detected in wild-type control littermates (E11.5). The analysis of Tg737 promoter driven β-galactosidase expression in whole-mount hindlimb buds from E10.5 (D), E11.5 (E), and E12.5 (F) heterozygous embryos. Initially, β-galactosidase expression was detected evenly throughout the limb bud mesenchyme with little expression detected in the apical ectodermal ridge (D). E,F: At later stages, Tg737 expression became concentrated in the condensing mesenchyme of the forming digits. The anterior side of the limb bud or embryo is marked with asterisks.G: Increased staining in the condensing mesenchyme is also detectable in sections through E12.0 limb buds. In the top panel, the section was cut parallel to the proximal–distal axis. In the bottom panel, the section was cut perpendicular to the proximal–distal axis.

In the E9.5–E10.5 limb, Tg737 expression is present throughout the limb bud with very low or no expression in the AER (Fig. 5D, data not shown). At E11.5 through E12.5, Tg737 expression a appears to increase in the condensing mesenchyme of forming digits (Fig. 5E,F). Low levels were still present in the mesenchyme surrounding the digits. The increased digit expression is not a consequence of tissue thickness in the whole embryos, because sectioning of the limb bud before staining gives identical results; however, we cannot eliminate the possibility that the staining is due to increased cell density rather than increased expression per cell (Fig. 5G).

Expression of Limb Bud Markers in Tg737 Embryos

To evaluate further the Tg737 limb phenotype, we compared the expression pattern of several genes involved in limb morphogenesis in Tg737orpk and wild-type embryos. Numerous mutants with preaxial digit duplication exist and in most, the polydactyly is associated with ectopic ZPA activity as defined by Shh, or Ihh in the case of the doublefoot (dbf) mutant (Chan et al., 1995; Masuya et al., 1995; Yang et al., 1998). This ectopic ZPA induces aberrant expression of HoxD genes, patched (ptch), Fgf genes, Gli genes, and BMP genes that are involved in limb patterning. Thus, the Tg737 limb phenotype suggested that Shh or its responsive genes would be ectopically expressed. However, our analysis in Tg737orpk mutants failed to show any overt expression changes in Shh, Ihh, Ptch, Bmp2, Bmp4, Fgf2, Fgf8, HoxD11–13s, HoxA11, HoxA13, Gli1, or Gli3 (Fig. 6A–J and data not shown). Most significantly, Shh expression was restricted to the ZPA in the posterior limb bud even with prolonged staining to detect possible weak hybridization on the anterior margin of limb (Fig. 6A–C). Ptch expression was identical to that in wild-type controls (Fig. 6D,E; Wada et al., 1999b). Similarly, the 5′ HoxD and HoxA clusters showed no anterior expansion as seen in most other polydactyly mutants, and Bmp2 and Bmp4 were normally expressed in the AER and posterior mesenchyme (Fig. 6F–J; Chan et al., 1995; Qu et al., 1997; Rijli and Chambon, 1997; Yang et al., 1998). In all of the Tg737orpk embryos, Fgf8 expression was normal throughout the entire AER (Fig. 6I) and the morphology of the cells in the AER was identical to that in wild-type controls (data not shown). Likewise, Fgf4 expression was normal at E10.5, being restricted to the posterior two thirds of the AER (Fig. 6K; Niswander and Martin, 1992; Heikinheimo et al., 1994). However, in three of the five E11.5 Tg737orpk mutants, a small anterior domain of Fgf4 expression was seen in the AER corresponding to the site where the extra digit will emerge (Fig. 6L). This aberrant Fgf4 expression was not observed in mutant embryos at other ages nor was it detected in wild-type embryos (data not shown).

Figure 6.

Expression of limb bud markers in Tg737orpk embryos. A–C: No overt changes in Shh expression were detected in Tg737orpk mutants. Shown are whole-mount embryonic day 10.5 (E10.5) mutant (A, left) and wild-type (A, right) embryos and magnified images of E10.5 mutant (B, left) and wild-type (B, right) hindlimbs (anterior is to the left). C: Shh expression in E11.5 mutant hindlimb from a side view (left), anterior view (middle), and posterior view (right). D,E: There was no overt change in patched expression in mutant (D, left) and wild-type (D, right) E11.0 and higher magnification of E11.0 forelimbs (E). Anterior is toward the left. Note the lack of an ectopic domain of patched expression on the anterior side of the limb, which is in agreement with the data for Shh. There were no changes in expression pattern detected between the mutant (left) and wild-type (right) embryos for HoxD11 (E11.0; F), HoxD12 (E11.0; G), HoxD13 (E11.0; H), or for HoxA11 or HoxA13 (data not shown) at any age analyzed between E9.0 and E12.0. In addition, there were no changes in the expression pattern of Ihh, Gli1, or Gli3 in any mutant embryos analyzed compared with wild-type controls at any age (data not shown). I: The apical ectodermal ridge (AER) appeared normal in E11.0 embryos as revealed by Fgf8 expression in mutant (left) and wild-type (right). J: There were no overt changes in Bmp4 expression detected in mutant (left) embryos compared with wild-type (right) at any age analyzed (E11.0). K: Similarly, Fgf4 expression was normal in the posterior AER at E10.5 in both mutant (left) embryos and wild-type (right) controls. L: However, in several of the E11.5 Tg737orpk mutants (left), a small zone of Fgf4 expression was present on the anterior side of both hindlimbs (arrows).

These data suggest that the Tg737orpk polydactyly is independent of Shh or, alternatively, that the aberrant Shh expression is below the levels of detection in this assay. This finding would be similar to the Gli3 extra toes (xt) mutant. xt heterozygotes exhibit an extra preaxial digit without detectable changes in Shh; however, in the xt homozygotes, eight digits form and Shh is ectopically expressed on the anterior limb bud (Hui and Joyner, 1993). Because similar phenotypic results are observed in the limb between the two Tg737 alleles, we questioned whether there might a dosage effect of Tg737 that results in detectable changes in Shh. Even though the Tg737Δ2-3βGal embryos develop eight digits per limb, there was still no change in Shh expression detected at any age (Fig. 7A–D). Thus, the polydactyly in Tg737 mutants is not associated with ectopic Shh expression. In addition, there are no overt defects in AER of Tg737Δ2-3βGal embryos and in general, Fgf8 expression appeared to be normal; however, in a small number of the null embryos, expression of Fgf8 was slightly elevated on the lateral sides of the AER (Fig. 7C). Intriguingly, Fgf-4 was misexpressed in all of the null mutants analyzed at E10.5 through E12.5 (Fig. 7D). This result is similar to that in Tg737orpk mutants except that the level of expression was significantly elevated in the nulls mutants. Thus, the data suggest there is a correlation between the level of Fgf-4 expression and the severity of the limb phenotype.

Figure 7.

Expression of limb bud markers in Tg737Δ2-3βGal embryos. A: No overt changes in expression Shh were detected at embryonic day 11.5 (E11.5) Tg737Δ2-3βGal mutants (left) and wild-type (right) embryos. B: Magnified image of an E11.5 mutant (left), and wild-type (right) hindlimb (anterior is to the left). Note the increased anterior–posterior axis in the limb of the mutants. C: In most embryos, there were no detectable changes in expression of Fgf8; however, in a few of the Tg737Δ2-3βGal mutants (left), we could detect slight up-regulation at the lateral regions of the AER relative to wild-type (right) controls. D: Analysis of Fgf4 expression in the fore- and hindlimb of Tg737Δ2-3βGal mutants (left; shown is E11.5). Higher magnification (right) image of a forelimb showing ectopic anterior expression of Fgf4 in the anterior AER. No anterior expression of Fgf4 was seen in the wild-type controls. The anterior side is toward the left.


Analysis of Tg737orpk mutant skeletons reveals a complex array of limb and skull patterning defects. In general, the defects were corrected in rescue animals, conclusively demonstrating Tg737 involvement in skeletal patterning. Our analysis indicates that the polydactyly in Tg737 mutants is independent of Shh misexpression, although we cannot eliminate a possible role for downstream genes in light of the limited in situ hybridization experiments conducted on the null embryos. Furthermore, our data suggest that there is a correlation between Fgf4 signaling and the limb defect severity and that there is a dosage affect of Tg737 on different limb pathologic conditions seen in the two Tg737 alleles.

During limb development, mesenchymal cells proliferate under the control of the AER and form prechondrogenic condensations that lay down a primary pattern for the development of the digits. Perturbations in the number or distribution of the cells that contribute to a digit can cause duplication or truncation of the developing digit (Shubin and Alberch, 1986; Dolle et al., 1993; Davis and Capecchi, 1996; Zakany and Duboule, 1996). It is intriguing that the only limb structures not dramatically improved by the rescue transgene are terminal structures formed during limb development, namely P2 on digits II and V. This finding raises the possibility that the Tg737 skeletal defects are caused by an increase in cell proliferation in specific regions or subpopulations of cells. This localized expansion could then result in too many cells on the anterior limb bud leading to the polydactyly; however, in preliminary analysis we can detect no overt changes in proliferation in the limb of the Tg737orpk mice relative to wild-type controls. An alternative mechanism might be that disruption of Tg737 results in changes in cell migration or recruitment. Thus, excess cells recruited to the anterior side of the limb could result in and extra digit while reducing the number of cells available to form P2 on digits II and V.

The expression of Tg737 throughout the limb in early development suggests that it is unlikely to function as part of the ZPA or the AER, or in positioning of these organizing centers as proposed for Gli3, dHand, and Alx4 (Qu et al., 1997). The pattern of Tg737 expression also suggests that it does not act directly downstream of Shh, because it is not restricted to or eliminated from the vicinity of the ZPA as seen for most Shh targets. Rather, Tg737 expression appears to be progressively more concentrated in presumptive digits, although this finding could reflect an increased cell density rather than an actual increase in expression.

In most cases of preaxial polydactyly, Shh is ectopically expressed on the anterior side of the limb bud (Chan et al., 1995; Masuya et al., 1995; Buscher et al., 1997; Sharpe et al., 1999). A few exceptions are seen in mutants such as talpid and diplopodia in the chick and Doublefoot in the mouse. However, in the case of Doublefoot, a hedgehog family member is ectopically expressed in place of Shh, and in all of these mutants, the Shh regulated genes Bmp2, Gli, and the 5′ HoxD genes are aberrantly expressed (Rodriguez et al., 1996; Hayes et al., 1998; Yang et al., 1998). In contrast to what is seen in these classic mutants, we detect no changes in Shh or its downstream genes in either the Tg737orpk mutants exhibiting a single ectopic digit or in the Tg737 null mutants where eight digits form. Our results are also distinct from expression studies performed in xt mutants that develop a single digit in the heterozygous condition and eight digits in the homozygous state. Whereas Shh and its downstream genes are not ectopically expressed in the heterozygotes, they are in the homozygous xt mice. This finding is not the case in the Tg737 mutants analyzed here.

Of the limb marker genes analyzed, it is intriguing that we observed changes in expression for Fgf4 in both Tg737 lines. In Tg737orpk mutants, the ectopically expressed Fgf4 domain was very small and transitory. In contrast, in null Tg737Δ2-3βGal mutants, the domain of ectopic Fgf4 expression was consistently observed and elevated relative to that seen in Tg737orpk mutants. Thus, the anterior domain and level of Fgf4 expression appears to correlate with the location and the extent of the duplicated digits that will form. The mechanism by which Fgf4 is activated in Tg737 mutants is not clear, especially given the Shh results. In addition, we currently do not understand how this ectopic expression may lead to the limb pathologic states in these mutants. Previous data implicate Fgf signaling in cell chemotaxis, in gap junction communication, and recently, in proliferative regulation of determined progenitor cells (Li and Muneoka, 1999; Sun et al., 2002). In light of these data and the limb phenotypes in Tg737 mutants, we predict the defects result from a change in the distribution of cells within the limb caused by alterations in cellular chemotactic properties of Fgf4 responsive cells. An aberrant distribution of cells could explain how P2 on digit II and V is lost while an additional digit forms on the anterior side.

Although we cannot eliminate the possibility that polaris, the protein disrupted in the Tg737 mutants, plays multiple roles in the mouse, data suggest that its primary function is in cilia assembly (Taulman et al., 2001; Yoder et al., 2002). Tg737 mutant mice have multiple defects that include random determination of the left–right axis (situs inversus), hydrocephalus, blindness, sterility, and cystic kidney disease, and as shown here, skeletal patterning abnormalities (Murcia et al., 2000; Taulman et al., 2001; Pazour et al., 2002; Yoder et al., 2002). In the case of hydrocephalus, blindness, sterility, situs inversus, and cystic kidney disease, the pathologic conditions have all been associated with defects in cilia. Thus, the data presented here raise an intriguing possibility that the loss of cilia in the Tg737 mutants is a contributing factor to the skeletal patterning defects.

Polaris is a highly conserved tetratricopeptide motif containing protein that is a key component of the intraflagellar transport (IFT) particle (Moyer et al., 1994; Haycraft et al., 2001; Taulman et al., 2001; Yoder et al., 2002). The IFT particle associates with the kinesin-II (KIF3A and KIF3B) and IFT dynein complexes and is thought to mediate the transport of “cargo” proteins along the cilia in the anterograde and retrograde directions, respectively (reviewed in Rosenbaum and Witman, 2002). In agreement with this proposed role, immunolocalization of polaris in mouse and OSM-5 (polaris ortholog) in Caenorhabditis elegans indicate that the proteins are found in the cilia (Haycraft et al., 2001; Taulman et al., 2001). Furthermore, when OSM-5 is fused with GFP and expressed in C. elegans, it migrates along the cilia at a rate expected for an IFT particle protein (Haycraft et al., 2001).

The Tg737 mutants described here represent the only mammalian mutations in genes encoding IFT particle proteins; thus, it is not possible at this point to evaluate whether skeletal defects are commonly associated with the loss of cilia. However, there are mutations in two subunits (KIF3A and KIF3B) of the heterotrimeric kinesin II complex that mediates anterograde movement of the IFT particle as well as vesicular and organelle transport (Nonaka et al., 1998; Marszalek et al., 1999; Takeda et al., 1999, 2000; Hirokawa, 2000). Similar to the Tg737 null mutants, disruption of KIF3A or KIF3B result in random left–right axis specification. Although there were no limb abnormalities reported, both KIF3A and KIF3B mutants die early in development, which may have precluded the detection of a limb phenotype. In addition to the Tg737 and the kinesin II mice, there are mutants such as fused toes (Ft), inversus viscerum or legless (iv/lgl), connexins-43 (Gja1), and sonic hedgehog that exhibit left–right axis patterning abnormalities along with limb defects (Scott et al., 1994; Heymer et al., 1997; Supp et al., 1997, 1999; Makarenkova and Patel, 1999). Intriguingly, the protein responsible for the phenotype in the inversus viscerum (iv) mice is localized to cilia, and there are defects in cilia function (Supp et al., 1997, 1999; Okada et al., 1999). In contrast, there are also numerous left–right axis mutants in mice such as forkhead homologue 4 (HFH-4, FoxJ1), inversin (inv), delta-1 (dll), polycystin-2 (pkd-2), as well as human syndromes such as Kartagener's or primary ciliary dyskinesia that do not have an associated limb defect, although cilia are affected (Morgan et al., 2002; Olbrich et al., 2002; Pennekamp et al., 2002; Przemeck et al., 2003). Additional studies need to be conducted using limb bud–specific conditional IFT mutants to further evaluate the role of cilia in patterning of the limb.

These data raise the question of what is the possible connection between cilia and skeletal patterning. Although cilia are known to be present on chondrocytes osteocytes, cartilage, and on certain mesenchymal cells, the function of these cilia remains elusive (Raynaud and Adrian, 1975; Trelstad, 1977; Ede and Wilby, 1981; Poole et al., 1985, 1997, 2001; Vidinov and Vasilev, 1985; Pieper et al., 1995). One possibility that was put forth in the case of chondrocytes is that cilia act as “cellular cybernetic probes that receive and transduce extrinsic signals” (Poole et al., 1985; Wheatley, 1995). This statement was based on the observation that there is direct contact between cilia and the extracellular matrix through electron-dense patches on the cilia membrane and that this attachment results in deflection of the axoneme (Poole et al., 1985). In the case of the mesenchymal cells of the chick somite, the cilium along with the Golgi apparatus were found to be localized in the posterior side of the cell relative to the direction of migration of the cell, thus indicating a possible association between these organelles and mesenchyme polarity (Trelstad et al., 1967). Similarly, analysis of Golgi location in cells of the chick limb bud indicates that the mesenchymal cells orient themselves relative to the forming condensations (Ede and Wilby, 1981). Whether cilia are also present on these cells and if there is a relationship with regard to cell orientation, migration, formation of condensations, or synthesis and secretion of cartilage matrix remains to be elucidated. We predict that, if cilia are present on mesenchymal cells in the limb bud, that they will contain important sensory machinery necessary for the detection of signals or for interactions with the extracellular matrix that allow a cell to respond its environment. In this regard, it is especially intriguing that FGF receptor 1 (FGFR1) has been detected recently in the cilia of monkey tracheal epithelium (Evans et al., 2002). Thus, if ciliary defects are observed in Tg737 mutant limb buds, as seen in other tissues from the Tg737 mutants mice, the sensory machinery might not be appropriately positioned, resulting in alterations in cellular recruitment or migration and consequently to skeletal patterning anomalies.



The generation of the Tg737orpk and Tg737Δ2-3βGal mutants and the construction of the BAP-Tg737 and the Tg737RSQ rescue lines have been described (Moyer et al., 1994; Yoder et al., 1996, 1997). The mice were genotyped as described previously (Yoder et al., 1996; Murcia et al., 2000), and all mice were maintained according to NIH guidelines.

Skeletal Analysis

Skeletons were prepared as described (Selby, 1987). Bone lengths were measured by using an ocular micrometer. Due to the growth inhibition in the Tg737orpk mutants, phalange (P) length was standardized to the corresponding metatarsal (M) or metacarpal. This P:M ratio was calculated for each digit on both right and left fore- and hindlimbs from a minimum of 12 adult animals for each genotype. Data are presented as the mean ± 1 standard deviation.

β-Galactosidase Staining of Limb Buds

The spatial expression pattern of Tg737 was analyzed in heterozygous Tg737Δ2-3βGal embryos by using the β-galactosidase reporter gene incorporated into the Tg737Δ2-3βGal allele as described (Murcia et al., 2000). β-Galactosidase activity recapitulated endogenous Tg737 expression as determined by in situ hybridization analysis and by immunolocalization of the Tg737 protein (Taulman et al., 2001). For cryostat sectioning, limb buds were fixed in 2% paraformaldehyde, impregnated with 30% sucrose/phosphate buffered saline (PBS), and embedded in OCT compound. Eight-micron-thick sections were cut, attached to slides, and incubated overnight at 30°C in X-gal staining solution (Murcia et al., 2000). Sections were counterstained with nuclear fast red, and images were captured by using a Nikon CoolPix950 digital camera attached to a Nikon TE200 inverted microscope or a Nikon SMZ800 stereomicroscope.

In Situ Hybridization

Embryos for in situ analysis were isolated from timed matings with the morning of the vaginal plug considered as E0.5. Embryos were fixed in 4% paraformaldehyde overnight at 4°C, washed twice in PBS + 1.0% Tween 20 (PBT), dehydrated through methanol in PBT (25%, 50%, 75%, 100%), and stored at −20°C. Digoxigenin-labeled riboprobes were generated by SP6, T7, or T3 RNA polymerase in both sense and antisense orientations. Whole-mount in situ hybridization analysis was performed by using digoxigenin-labeled riboprobes at a concentration of 1 μg/ml as described (Wilkinson, 1992).


Whole-mount in situ hybridization probes for HoxD11, HoxD12, HoxD13, and HoxA11 were obtained from Dr. Capecchi (University of Utah, Salt Lake City, UT); the Shh, Fgf4, Fgf8, and Bmp4 probes were obtained from Dr. Underhill (University of Western Ontario, Ontario, Canada); and the Gli1, Gli3a, and Ptch probes from Dr. Hui (Hospital for Sick Children, Ontario, Canada). N.S.M. and B.K.Y. received funding from the National Institute of Diabetes and Digestive and Kidney Diseases.