A Dual Role for Hox Genes in Limb Anterior-Posterior Asymmetry

See allHide authors and affiliations

Science  11 Jun 2004:
Vol. 304, Issue 5677, pp. 1669-1672
DOI: 10.1126/science.1096049


Anterior-to-posterior patterning, the process whereby our digits are differently shaped, is a key aspect of limb development. It depends on the localized expression in posterior limb bud of Sonic hedgehog (Shh) and the morphogenetic potential of its diffusing product. By using an inversion of and a large deficiency in the mouse HoxD cluster, we found that a perturbation in the early collinear expression of Hoxd11, Hoxd12, and Hoxd13 in limb buds led to a loss of asymmetry. Ectopic Hox gene expression triggered abnormal Shh transcription, which in turn induced symmetrical expression of Hox genes in digits, thereby generating double posterior limbs. We conclude that early posterior restriction of Hox gene products sets up an anterior-posterior prepattern, which determines the localized activation of Shh. This signal is subsequently translated into digit morphological asymmetry by promoting the late expression of Hoxd genes, two collinear processes relying on opposite genomic topographies, upstream and downstream Shh signaling.

Anterior-posterior (AP) asymmetry in tetrapod limbs is reflected by the anatomy of lower arms and hands. In humans, the thumb is shorter and more mobile than other digits. These differences result from the presence in the developing posterior limb bud of a zone of polarizing activity (ZPA) (1), defined by its potential both to induce supernumerary digits and to modify digit identity when transplanted anteriorly. Cells within the ZPA express the Shh gene (2), whose product propagates posterior identity in the growing bud, likely through a graded, long-range intercellular signaling mechanism (3, 4).

The effects of Shh signaling in limbs are mediated, at least in part, by posterior Hoxd genes (2, 59) because of the potential of SHH to prevent the production of the repressor form of GLI3 protein, which negatively regulates Hox gene transcription (1012), likely through a global digit enhancer located near the HoxD cluster (13, 14). Models for the restriction of Shh expression in posterior limb bud cells have been proposed whereby the antagonism between the Gli3 and dHand transcription factors would initially divide the bud into anterior and posterior domains (15). Although this model is supported by genetics and experimental data (11, 12, 1618), it falls short in explaining the spatial restriction of Shh expression.

A similar limb bud posterior specificity was observed for both Hoxa and Hoxd genes in their earliest phases of expression (1921). Hoxd genes are activated in a collinear fashion, with Hoxd1 and Hoxd3 expressed throughout the early bud, whereas Hoxd12 and Hoxd13 are expressed posteriorly (Fig. 1A) in a domain containing future SHH-positive cells. This restriction occurs before Shh expression (5, 6, 9, 22), which suggested a role for Hox genes in AP polarity (19). In addition, ectopic expression of Hoxb8 and Hoxd12 revealed the potential of some HOX products to trigger Shh expression (2325). Here, we use two novel genomic rearrangements to show that posterior Hoxd genes are key determinants in the early organization of limb AP asymmetry.

Fig. 1.

Targeted inversion of the HoxD cluster. (A) Collinear expression patterns (in gray) of Hoxd genes in limb bud. The HoxD cluster is shown with blue arrows for a Hoxd11lac reporter gene (27) and white block arrows for Hoxd genes. Red arrowheads are loxP sites. The presence of an ELCR is shown in blue. The loxP/Cre conditional inversion allele is also shown. Exposure to the Cre recombinase in vitro (red arrows) generated both flox and Inv alleles. (B) Control Hoxd13 expression (one copy) in an E9.5 embryo. (C) Hoxd13 expression in a de novo isolated (28) (Materials and Methods) Inv/+ embryo, showing an anterior shift in expression including the forelimb field (dotted lines), similar to Hoxd1 (D). Hoxd11 expression in E9 embryo with one copy (E) or Inv/+ embryo (F). Hoxd13 staining in forelimb buds of normal (G) and Inv chimeric (H) embryos. The expected posterior pattern is seen in the larger two specimens [(G), bottom], whereas chimeras show premature expression in the entire bud (H).

We engineered a loxP/Cre-dependent inversion of the HoxD cluster (Fig. 1A) and asked whether gene expression would be concomitantly modified. Because embryos carrying the inversion (Inv) did not survive, we induced the inversion in vivo, along with a wild-type chromosome, or analyzed mosaic fetuses containing both the Inv and the non-inverted alleles after recombination in vitro. When the Inv allele was present, expression of both Hoxd13 (Fig. 1, C, G, and H, and fig. S1) and Hoxd11 (Fig. 1F) was seen in the anterior domain. Expression of Hoxd1 is shown in Fig. 1D as a control. Expression of a single copy of Hoxd13 (Fig. 1B) and Hoxd11 (Fig. 1E) using the floxed configuration over a deletion of the HoxD cluster (flox/Del1-13) showed no anterior ectopic expression. We concluded that, on inversion, Hoxd13 and Hoxd11 had been brought to the vicinity of an early limb control region (ELCR) (Fig. 1), located telomeric to HoxD, that normally drives expression throughout limb bud cells.

To ascertain both the existence and the location of an ELCR, we analyzed a deletion of Hoxd1 to Hoxd10 obtained by targeted meiotic recombination (fig. S2A) [Del(1-10)]. This deficiency generated a minicluster composed of Hoxd11, Hoxd12, and Hoxd13, now located as close to a potential ELCR as in the inversion but in their native orientation and genomic order (Fig. 2A). Homozygous Del1-10 mice died soon after birth; hence specimens were analyzed either as fetuses or newborns. Homozygous newborns showed a markedly elongated thumb in forelimbs (Fig. 2, B and E, asterisks), making the AP asymmetry hardly recognizable. Furthermore, homozygotes displayed a perfectly symmetrical carpus, with bones organized in a mirror image (Fig. 2E), and the distal radius acquired the morphology of distal ulna. This bilateral symmetry, seen in the right limbs of eight animals, occurred less frequently in the corresponding left limbs. In addition, three homozygous animals had only four digits in their forelimbs (Fig. 2C), and hindlimb polydactyly was scored in both homozygous and heterozygous individuals. In all cases, digits appeared organized following a mirror symmetry, as exemplified by an additional large palmar cushion at the base of the most anterior digit (Fig. 2G).

Fig. 2.

A mini HoxD cluster. (A) Derivation of the Del1-10 allele by meiotic recombination (2831) (Materials and Methods). Symbols are as in Fig. 1A. Forelimb paws of homozygous (hom) mutant [(B) and (C)] and wild-type (wt) (D) newborns. Bone (red) and cartilage (blue) pattern of newborn homozygous mutant (E) and wild-type (F) autopods. r, radius; u, ulna. Palm clay prints of a heterozygous (het) mutant (G) and a normal (H) adult animal. Asterisks point to the elongated digit I. The supernumerary anterior cushion is shown with arrowhead. The red dotted lines indicate the plane of AP mirror symmetry.

The loss of AP asymmetry suggested that the deletion had modified the expression of the remaining Hoxd genes. Hoxd13 expression (Fig. 3A) was detected at high amounts in both heterozygous and homozygous animals. However, contrary to what is seen in wild-type mice, Hoxd13 was scored in anterior cells of homozygous mice at day 11 of the embryo (E11) within the digits as well as along the anterior circumference of the limb, including both lower and upper arm primordia (Fig. 3B). Consequently, Hoxd13 transcript pattern in homozygous limbs was no longer skewed posteriorly. Therefore, symmetrical expression of Hox genes coincided with double-posterior hand plates. At E10, Hoxd13-expressing cells are normally located distal and posterior (Fig. 3C). In heterozygotes, two additional domains located at the anterior and posterior base of the bud expressed Hoxd13. At E9.5, homozygous limbs displayed an even more homogenous signal, because we detected Hoxd13 and Hoxd11 transcripts throughout the limb buds (Fig. 3D), supporting the presence of an ELCR 3′ of Hoxd1. All three posterior genes showed similar misregulation (fig. S3).

Fig. 3.

Hoxd13 expression in Del1-10 embryos. Homozygous, heterozygous, and wild-type littermates are compared. Expression of Hoxd13 at E13 (A), E11 (B), E10 (C), and E9 (D) shows loss of asymmetry (arrowhead indicates presumptive digit I). Anterior regions displayed ectopic Hoxd13 signal, and a mirror image pattern was visible in all homozygous and some heterozygous specimens. Whereas at E9, both Hoxd11 and Hoxd13 were expressed throughout the bud (D), this ectopic domain was not maintained and appeared as two weak patches located proximally at E10 (C). At this latter stage, the wild-type domain appeared distally, though with an unusual central position [(C), middle and right]. Red and black dotted lines indicate the plane of AP symmetry and limb bud field, respectively.

This latter ectopic expression nevertheless rapidly subsided and thus could not account for the subsequent symmetrical Hox profiles in digits. Because Hox expression in digits is likely under the control of Shh signaling, we looked at Shh transcripts in these mutant buds. In limbs of early and late E9.5 wild-type embryos, Shh transcripts were restricted to a posterior spot (Fig. 4, A and B). By E10, this positive domain had extended while remaining posterior and shifting distally (Fig. 4C), and by E10.5 transcript accumulated within the posterior half (Fig. 4D). Del1-10 homozygous embryos showed clusters of ectopic Shh-expressing cells, either as an extension of the normal domain anteriorly and distally (Fig. 4, E and F) or along the anterior edge (Fig. 4G), leading in some cases to mirror-image domains (Fig. 4H). In these buds, the expression profiles of the Gli3 and dHand genes were modified concomitantly (fig. S3). Therefore, ectopic posterior Hox products in early anterior buds were sufficient to trigger Shh transcription in cells that eventually established an ectopic Shh anterior domain, suggesting that Shh expression is normally controlled by posterior Hox genes.

Fig. 4.

Shh expression in mutant limb buds. Wild-type embryos at E9.5 [(A) and (B)], E10 (C), and E10.5 (D). Shh expression in Del1-10 mutants at E9.5 [(E) and (F)], E10 (G), and E10.5 (H). Ectopic Shh expression in early buds was systematically scored [compare (A) to (E) and (B) to (F)]. The normal Shh profile (B) was lost, and a larger, more distally located domain was observed (F). Subsequently, this domain was preferentially reinforced and maintained at both margins of the growing limb [(G) and (H)]. Note the scattered anterior signal, which could be observed (G), and the robust ectopic domain seen only in homozygotes (H).

In these experiments, we did not observe mirror-image duplications with polydactyly, as seen with ZPA grafts where a similar amount of ectopic SHH was present. We think this is because of the absence of many Hoxd genes and the observed down-regulation of Hoxd11 in digits (fig. S2, B and C). Hox genes are required for proper growth of the appendages, and their absence prevented mutant buds from fully developing a duplicated pattern because of an initial size reduction. Full mirror-image duplications obtained with SHH, ZPA grafts, or retinoid treatment are thus likely Hox-dependent. Supernumerary digits in forelimbs were only scored in heterozygotes containing a complete set of Hoxd genes, which supports this view.

The mechanism progressively restricting Hox gene transcription to posterior limb bud cells is crucial for the limb AP asymmetry by triggering localized Shh transcription. Although the underlying process is elusive, both the inverted and deleted configurations reported here suggest that a regulatory sequence localized telomeric to the cluster (ELCR) (Fig. 5) is needed to implement this collinear regulation. Because Hoxd promoters equally respond to this ELCR when located nearby, a distance effect may progressively prevent anterior cells from transcribing genes located far from the ELCR. Once activated posteriorly, Shh will subsequently control Hoxd gene transcription in presumptive digits with the use of another collinear strategy (14), likely mediated by an enhancer located centromeric to the cluster [global center region (GCR)] (13). In early limb development, 5′-located Hoxd genes are repressed in anterior cells, probably by the GLI3 repressor protein (26). By contrast, in the late phase, the same genes are heavily, but unequally, transcribed in most distal cells because of an opposite distance effect whereby posterior Hoxd genes preferentially respond to the GCR sequence (14). In this view, Shh signaling acts as a relay mechanism between two “opposite” collinear strategies to translate an initial molecular asymmetry into its morphological readout, digit identities.

Fig. 5.

Model for the onset of limb AP asymmetry. Red bars and green arrows indicate negative and positive effects, respectively. Anterior is on top. In the early bud phase (top), Gli3 is expressed without AP asymmetry. The repressor form (GLI3R) (10) is present throughout the bud, suppressing transcription of posterior Hoxd genes Shh and dHand (11, 12). Hox gene collinear activation leads to specific accumulation of group 10 to 13 transcripts posteriorly, perhaps as a distance effect from the ELCR (top, purple domain). These latter products trigger expression of Shh and dHand, which activate each other and limit the level of GLI3R accumulation posteriorly, either by inhibiting GLI3-to-GLI3R protein conversion (12) or by directly suppressing Gli transcription (15). Subsequently (bottom), in the posterior part, positive feedback loops between 5Hox genes, Shh, and dHand trigger the progressive expansion of this posterior identity (2, 16, 17), mostly through the graded impact of the SHH product on Hox gene expression in the distal bud (32), a process presumably under the control of the GCR (13). I to V indicate presumptive digits, and the graded pink zones represent the SHH gradient.

Supporting Online Material

Materials and Methods

Figs. S1 to S3

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article