Polydactylous limbs in Strong’s Luxoid mice result from ectopic polarizing activity

Classical developmental studies have shown that the anteroposterior axis of the chick limb is determined as early as stage 12, well before the limb primordia are morphologically distinguishable from the surrounding flank (Hamburger, 1938). Furthermore, when prospective limb mesoderm from stage 12 chicks is implanted under the flank ectoderm of a host chick, a supernumerary limb develops, which displays the anteroposterior characteristics of the donor tissue, regardless of its orientation with respect to the host (Saunders and Reuss, 1974). These results indicate that the anteroposterior axis is determined early in limb development and that this axial information resides in the mesoderm. Little, however, is known about the molecules that establish and maintain the polarity of the anteroposterior axis.

One important consequence of the determination of the anteroposterior limb axis is the induction of a zone of polarizing activity (ZPA, or polarizing region) in the posterior mesoderm of the limb (Tabin, 1991; Tickle and Eichele, 1994). When this polarizing region is grafted into the anterior mesoderm of a host chick limb bud, the resulting wing, instead of having the normal IV-III-II (posterior to anterior) digit pattern, has the symmetric pattern of IV-III-II-II ′-III ′-IV ′ (′ indicates duplicated digits; Saunders and Gasseling, 1968). The polarizing region, perhaps by secreting a diffusible morphogen, is thought to provide positional information for mesodermal limb bud cells along the anteroposterior axis (Tickle et al., 1975). Cells close to the polarizing regionacquire posterior identities, while cells far from the polarizing region acquire anterior identities.

Sonic hedgehog (shh), a vertebrate homolog of the Drosophila segment polarity gene hedgehog, is likely to be a key player in mediating the polarizing activity of posterior limb mesoderm. Shh is expressed in posterior limb mesoderm, in a domain coincident with the polarizing region (Riddle et al., 1993). Shh is also expressed in other embryonic tissues that display polarizing activity when implanted into chick limb buds, including the notochord, floor plate and Hensen’s node (Echelard et al., 1993; Krauss et al., 1993; Riddle et al., 1993; Roelink et al., 1994). Tissue culture cells infected or transfected with shh-expressing constructs acquire polarizing activity, as demonstrated by their ability to induce mirror-image digit duplications when implanted into the anterior mesoderm of host chick limb buds (Riddle et al., 1993; Chang et al., 1994).

The apical ectodermal ridge (AER), a specialized rim of limb ectoderm that is required for distal outgrowth of the underlying mesoderm (Saunders, 1948), is likely to play an indirect role in patterning the anteroposterior axis through its interactions with the polarizing region. Removal of the AER results in loss of polarizing activity (Vogel and Tickle, 1993) and shh expression (Vogel and Tickle, 1993; Laufer et al., 1994) in the posterior mesoderm. In the absence of the AER, maintenance of polarizing activity and shh expression can be provided by beads releasing fibroblast growth factor 4 (FGF- 4; Vogel and Tickle, 1993; Laufer et al., 1994; Niswander et al., 1994), which is normally expressed in the posterior portion of the AER (Niswander and Martin, 1992).

In spite of these recent advances in our understanding of the patterning of the anteroposterior limb axis, very little is known about the earlier molecular mechanisms that establish and maintain anteroposterior asymmetry. Hoxb-8 encodes a candidate molecule for playing such a role. This homeodomain-containing gene is expressed in the posterior mesoderm of day 9.5 mouse forelimbs, before the onset of shh or fgf-4 expression, and disappears shortly thereafter (Charité et al., 1994). In transgenic mice, misexpression of Hoxb-8 in the anterior mesoderm of forelimb buds results in mirror-image duplication of digits (Charité et al., 1994). Such digit patterns are similar to those observed in chick limbs containing a polarizing graft (ZPA graft) in the anterior mesoderm (Saunders and Gasseling, 1968). These ZPA-like duplications are preceded by ectopic expression of shh in the anterior limb mesoderm and fgf-4 in the anterior portion of the AER. These results raise the possibility that Hoxb-8 may normally be involved in activating the expression of shh and fgf-4, and thereby establish the polarizing region. However, Hoxb-8 is clearly not required for maintenance of the polarizing region because its expression in the forelimb disappears shortly after day 9.5, and therefore is absent at the time of maximal polarizing activity. Furthermore, Hoxb-8 does not appear to have an analogous role in the hindlimbs since its expression there is throughout the mesoderm (Charité et al., 1994). In addition to Hoxb-8, fgf-4, in cooperation with retinoic acid, may play a role in inducing the expression of shh (Laufer et al., 1994; Niswander et al., 1994).

One approach towards uncovering early patterning mecha-nisms in the limb is to study the molecular embryology of mouse mutants that show abnormalities in limb patterning. We have been analyzing the semidominant mouse mutant Strong’s Luxoid (lst^D) (Forsthoefel, 1962, 1963). Mice heterozygous for lst^D have preaxial polydactyly (extra digits are anterior) in the hindlimbs. Mice homozygous for lst^D have preaxial polydactyly in all four limbs, as well as abnormalities in the preaxial long bones (radius and tibia) of the limb. We present evidence that polydactyly in lst^D mice results from the induction of additional AER containing posterior characteristics and a new polarizing region in the anterior region of lst^D limb buds. These results suggest that the lst gene product plays an important role in early anteroposterior patterning.

Heterozygous mating pairs of lst^D mice were checked daily for vaginal plugs, with noon of the day of plug appearance designated as day 0.5. Embryos were dissected from extraembryonic membranes, fixed overnight in 4% paraformaldehyde, and stained with Alcian Green as described previously (Riddle et al., 1993), except that staining was done overnight. After clearing in methylsalicylate, limbs were dissected from the body to facilitate photography.

Embryos were obtained from matings between heterozygous lst^D parents and staged as described above. Whole-mount in situ hybridizations with digoxigenin-labeled riboprobes were performed as previously described (Riddle et al., 1993), except that 0.1% Tween 20 was used in the post-antibody washes. The in situ riboprobes have also been described previously: fgf-4 (Hébert et al., 1990), shh (Echelard et al., 1993), Hoxd-12 (Dollé et al., 1989), and Hoxb-8 (Charité et al., 1994). Embryos were genotyped using a CA dinucleotide-repeat assay (Chan and Leder, unpublished results) on DNA extracted from extraembryonic membranes. The CA-repeat resides in the limb deformity locus and shows <10% recombination with lst^D. Control experiments (Chan and Leder, unpublished results) showed that genotypes could be reliably determined by morphological examination (Forsthoefel, 1963) of the limbs of day 11.5 and older embryos. Younger embryos show no morphological distinctions and must be genotyped molecularly. For the Hoxb-8 studies, at least 5 embryos identified as homozygotes, using the CA-repeat assay, were examined for each stage (days 9.5, 9.75, 10.0, and 10.5). Since the CA marker shows <10% recombination with the lst^D locus, there is >99.9% probability that a homozygote was examined for each stage.

Anterior mesoderm from day 11.5 mouse limb buds was transplanted into a slit made in the anterior region of stage 18-20 host chicken wing buds. Homozygous lst^D embryos were identified morphologically in litters from heterozygous parents. Homozygous forelimbs at this stage have a distinct anterior bulge (Forsthoefel, 1963), and mesoderm from the anterior margin of this bulge was used for the transplantations, since in situ hybridization (Fig. 2) showed shh expression there. As a positive control, grafts were performed using posterior mesoderm from these limbs. As a negative control, grafts were also performed with anterior limb mesoderm from embryos resulting from wild-type matings. For in situ hybridization, the chick embryos were harvested 36 hours after grafting and processed as described above. For skeletal analysis, embryos were harvested 7 days after grafting and stained with Alcian Green.

Mice heterozygous for the semidominant mutation Strong’s Luxoid (lst^D) have preaxial polydactyly usually limited to the hindlimbs, while homozygotes have a more severe preaxial polydactyly in both the fore- and hindlimbs (Forsthoefel, 1962) (Fig. 1). To examine their digit patterns in more detail, we used Alcian Green to stain the cartilage of day 15.5 embryos obtained from matings between heterozygous lst^D parents. Our results are consistent with previous skeletal analyses of lst^D mice (Forsthoefel, 1962). The normal mouse foot has five digits, numbered V, IV, III, II, and I (posterior to anterior direction). Digits V to II each have three phalanges, whereas digit I has two phalanges. In heterozygous lst^D embryos, the most common hindlimb digit pattern is V, IV, III, II, I, II ′ (Fig. 1B,E; ′ denotes extra digit). The forelimbs are generally normal (not shown). The exact identity of the extra digits in homozygotes often could not be determined, but forelimbs of homozygotes usually contain one or more triphalangeal digits positioned anterior to the biphalangeal pollex, or digit I (Fig. 1C,F). Homozygous hindlimbs usually lack digit I and contain six or more triphalangeal digits (Fig. 1D). These digit patterns, in which posterior digits are duplicated in the anterior portion of the limb, resemble those obtained after grafts of polarizing tissue are transplanted into the anterior mesoderm of chick limb buds (Saunders and Gasseling, 1968).

To explore the possibility that lst^D limb buds might contain an ectopic polarizing region, we used whole-mount in situ hybrid-ization on mutant and wild-type embryos to examine the expression of fgf-4 (Niswander and Martin, 1992), shh (Echelard et al., 1993; Krauss et al., 1993; Riddle et al., 1993; Roelink et al., 1994), and Hoxd-12 (Dolle et al., 1989), which are molecular markers for, respectively, the posterior apical ectodermal ridge (AER), the zone of polarizing activity (ZPA), and posterior limb mesoderm. With these probes, we found three in situ staining patterns resulting from the three genotypes generated from matings between heterozygous lst^D parents. Wild-type embryos showed staining patterns consistent with published reports. Fgf-4 is expressed in the central and posterior portion of the AER, most intensely at embryonic day 10.5 (not shown) and fading by day 11.5 (Niswander and Martin, 1992) (Fig. 2A). Shh is expressed in posterior limb mesoderm coincidently with the ZPA (Echelard et al., 1993; Riddle et al., 1993; Fig. 2A); Hoxd-12 is restricted to posterior and distal mesoderm (Dollé et al., 1989; Fig. 2H).

Heterozygous lst^D embryos show normal staining patterns for these genes in the forelimbs and in the posterior half of the hindlimbs. However, they show ectopic expression of fgf-4, shh, and Hoxd-12 in the anterior region of the hindlimbs, starting at day 11.5 (Fig. 2B,I). Fgf-4 is ectopically expressed in a strip of anterior AER in day 11.5 hindlimbs, by which time the normal expression of fgf-4 in the posterior AER has started to disappear (Fig. 2B and data not shown). At the same time, shh is ectopically expressed in a small patch of anterior mesoderm that underlies the strip of fgf-4 ectopic expression (Fig. 2B and data not shown). Hoxd-12 is ectopically expressed in a broader region of the anterior mesoderm (Fig. 2I).

Homozygotes show ectopic expression of limb markers in the anterior portion of all limbs (Fig. 2C,J), whereas expression patterns in the posterior portion of the limbs are normal. Ectopic expression of fgf-4 is first observed at day 10.5 in an anterior strip of AER in the forelimb, resulting in two discontinuous strips of fgf-4 production (Fig. 2D). As in heterozygotes, these homozygotes do not show ectopic expression of fgf-4 in the hindlimb until day 11.5 (Fig. 2C,F), probably reflecting the normally delayed development of murine hindlimbs relative to forelimbs. By late day 11.5, the homozygous lst^D limb buds show a distinct anterior outgrowth from which the extra digit(s) will form (Forsthoefel, 1963). At this time, fgf-4 is expressed in the AER at the anterior border of this bulge (Fig. 2E,F), whereas the wild-type expression in the posterior AER has largely faded. Mutant fore- and hindlimbs also express shh ectopically in a small patch of anterior mesoderm that directly underlies and extends anterior to the ectoderm expressing fgf-4 (Fig. 2E,G), such that this juxtaposition of fgf-4 and shh is in mirror-image relationship to their normal day 10.5 expression domains in the posterior region (Laufer et al., 1994). The onset of shh expression is delayed relative to fgf-4 (data not shown), raising the possibility that fgf-4 may play a role in inducing shh expression (Laufer et al., 1994; Niswander et al., 1994). Interestingly, the ectopic staining patterns in homozygous hindlimbs were significantly more intense than in heterozygous hindlimbs (compare Fig. 2B with 2C, and Fig. 2I with 2J), consistent with the more severe polydactyly in homozygotes. Hoxd-11 (not shown) and Hoxd- 12 are ectopically expressed in a broad region of anterior mesoderm in day 11.5 limbs (not shown), and in later embryos are expressed throughout the anterior bulge (Fig. 2J).

Recent transgenic mouse experiments have shown that ectopic expression of Hoxb-8 in early stage anterior limb mesoderm causes anterior ectopic expression of fgf-4 and shh and results in ZPA-like transformations in the forelimb (Charité et al.,1994). This observation raises the possibility that the ectopic expression of fgf-4 and shh observed in lst^D limbs may result from prior ectopic expression of Hoxb-8. Since ectopic fgf-4 is detectable in homozygous lst^D forelimbs by day 10.5, we examined mutant embryos at days 9.5, 9.75 (not shown), 10.0, and 10.5 (not shown) for ectopic Hoxb-8 expression (Fig. 3). Homozygous mutant embryos show no misexpression of Hoxb-8 in the forelimb at any stage examined, suggesting that the lst^D mutation causes digit duplications either by acting downstream of Hoxb-8 or by functioning through a Hoxb-8- independent pathway.

To test directly the idea that lst^D limb buds contain an ectopic ZPA, we transplanted anterior mesoderm from day 11.5 lst^D limb buds into host chick limb buds. At day 11.5, homozygous lst^D embryos can be identified in litters from heterozygous parents because their forelimbs contain a distinct bulge in the anterior region. We decided to transplant anterior mesoderm from homozygous limbs since homozygotes contain considerably higher levels of ectopic shh expression than heterozygotes (compare Fig. 2B with 2C). As a negative control, we used anterior limb mesoderm from day 11.5 embryos generated from wild-type parents. Transplants of lst^D anterior mesoderm harvested 36 hours later showed clear induction of ectopic chick Hoxd-11 expression at the site of the graft, as is typical for grafts containing polarizing activity (Izpisúa-Belmonte et al., 1991; Nohno et al., 1991) (3 of 6 grafts; Fig. 4B). Grafts of anterior mesoderm from wild-type embryos produced no induction of Hoxd-11 (0 of 6 grafts; Fig. 4A). Furthermore, ten percent of lst^D transplants that were allowed to develop for a week showed an extra digit II or other skeletal abnormality (4 of 39 grafts; Fig. 4D), whereas transplants of wild-type anterior mesoderm showed no abnormality (0 of 20 grafts; Fig. 2C). Since we were able to obtain more extensive duplications (up to digit III) by grafting mouse posterior limb mesoderm (not shown), lst^D limb buds appear to contain lower levels of polarizing activity in the anterior limb mesoderm relative to the normal polarizing activity in the posterior mesoderm. The extent of digit duplication obtained with lst^D anterior limb mesoderm indicates weak ZPA activity (Tickle, 1981), consistent with the relatively lower levels of shh expression in the anterior mesoderm of lst^D limb buds (Fig. 2C,G) and with the limited number of digits duplicated in lst^D mice (Fig. 1).

FGF-4 and SHH have been shown to cooperate in patterning the anteroposterior and proximodistal axes of the limb. Each factor may play a role in inducing and maintaining the expression of the other (Laufer et al., 1994; Niswander et al., 1994). Maintenance of polarizing activity (Niswander et al., 1993; Vogel and Tickle, 1993) and shh expression (Laufer et al., 1994; Niswander et al., 1994) requires the presence of an intact AER, and this requirement can be largely replaced by FGF-4 The anterior margins of lst^D mutant limb buds ectopically express both fgf-4 and shh in adjacent cell layers and would therefore appear to have the necessary conditions to support polarizing activity. Consistent with this idea, lst^D embryos show expression of Hoxd cluster genes in the outgrowth of mesoderm that gives rise to the extra digits (Fig. 2I,J); Hoxd genes are induced in response to grafts of polarizing tissue into chick anterior mesoderm (Izpisua-Belmonte et al., 1991; Nohno et al., 1991). Finally, anterior mesoderm from lst^D homozygous limb buds show polarizing activity when transplanted into host chick wing buds.

Taken together, these results show that lst^D limb buds contain a duplicated polarizing region and posterior AER. The duplication of the AER has been observed previously by histological analysis of lst^D embryos (Forsthoefel, 1963), and was suggested to result from duplication of an AER maintenance center. Alternatively, it is equally possible that the lst^D mutation causes the extension of an existing AER and reverses the polarity of the anterior portion. It is likely that the ZPA- like digit duplications seen in lst^D limbs result from the influence of the ectopic polarizing region that we have demonstrated.

The lst locus is closely linked to the mouse limb deformity (ld) gene. Mice homozygous for the recessive ld mutation display reduction and fusion of digits and fusion of the distal long bones (radius/ulna, tibia/fibula; Kleinebrecht et al., 1982; Woychik et al., 1985). Thus, in a general sense, ld mice have a phenotype opposite to that of lst^D mice. This observation has led to the suggestion that ld and lst^D may be mutations in the same gene (Woychik et al., 1985). It has been proposed that lst^D, possibly a gain-of-function mutation, leads to increased gene activity and polydactyly, while ld, being a loss-of- function mutation, leads to reduced gene activity and syndactyly. The ld gene has been cloned (Woychik et al., 1985) and encodes novel protein isoforms termed formins (Woychik et al., 1990). In situ hybridization studies show that lst^D embryos have normal expression of ld transcripts (Chan and Leder, unpublished observations). Furthermore, western blots with anti-formin antibodies show no gross differences between formins in lst^D and wild-type mice (Chan and Leder, unpublished observations). Therefore, we currently favor the view that lst and ld represent two separate genes that function in similar pathways in limb development.

The lst^D mouse belongs to a class of luxoid/hemimelic mouse mutants (characterized by deficiencies in part or all of the long bones of the limbs) that includes Green’s luxoid and Carter’s luxate (Forsthoefel, 1962). These three mutants display polydactyly when heterozygous and hemimelia when homozygous. The genes affected in these mutations are thought to act in independent biochemical pathways, since mice carrying two or more of these mutations show additive effects in limb abnormalities (Forsthoefel, 1962). In future studies, it will be interesting to examine the effect of these other luxoid mutations on the AER and polarizing region.

We propose that the lst gene product acts early in anteroposterior patterning. Two general models for the function of lst can be proposed, depending on whether the lst^D mutation is a gain-of-function or loss-of-function mutation (Fig. 5). In the former case, lst may normally function in specifying the proper regional expression of fgf-4 and shh in the posterior portion of the limb bud (Fig. 5A). Similar to the role proposed for Hoxb-8 (Charité et al., 1994), lst may be involved in the activation of the polarizing region, acting either downstream or independently of Hoxb-8. Alternatively, if lst^D is a loss-of-function mutation, the lst gene product would be required for the formation or maintenance of anteroposterior asymmetry in the limb; in its absence, the limb takes on a default state in which there is mirror-image symmetry along the anteroposterior axis (Fig. 5B). This requirement for the lst product is dosagedependent, since heterozygotes show a milder manifestation of the digit duplications found in homozygotes. In this model, lst may normally repress the expression of molecules such as shh or fgf-4 in the anterior portion of the limb. In both models, lst is likely to play an important role in early anteroposterior patterning. Although other mouse mutants also exhibit preaxial polydactyly (Grüneberg, 1963), lst^D mutants appear unique in their consistent display of ZPA-like duplications, and hence may yield critical information about early anteroposterior patterning mechanisms in the mouse limb bud.

We are grateful to Anne Burke, Andrew McMahon, and Gail Martin for providing probes. We thank members of the Leder and Tabin labs for helpful advice and comments.