Axial Hox9 activity establishes the posterior field in the developing forelimb

Abstract

Vertebrate limb buds emerge from the lateral plate mesenchyme, and early polarity in the limb bud is established through the interaction of a series of signaling pathways. The antagonism between Hand2 in the posterior field and Gli3 in the anterior field is a key determinant in the establishment of anteroposterior (AP) polarity before Shh signaling in the zone of polarizing activity (ZPA) in both the forelimb and the hindlimb (1). Previous work has shown that Hand2 and posterior HoxA/D genes can activate Shh transcription during limb AP patterning (2–7). Once initiated, Shh maintains the expression of Hand2 and posterior HoxD genes, resulting in a Gli3 repressor/activator gradient that further controls AP patterning in the limb (3, 8, 9). The mechanism by which AP patterning is initiated before the expression of Hand2 and Gli3 remains elusive, but likely is related to differential gene expression along the body axis. Because of their collinear expression along the AP axis, Hox genes are excellent candidates for this activity.

Posterior HoxA and HoxD genes play important roles in patterning along the proximodistal axis of the forelimb (10–13). They also are required for activating and maintaining Shh expression in the developing forelimb (6, 7). Exactly how the collinear expression of the HoxA/D AbdB group genes are established and maintained in the limb, and the details regarding the feedback loop with Shh, are areas of active investigation, but these mechanisms engage only after the initial establishment of AP polarity, because Hand2 expression is established unperturbed in compound HoxA/D mutant forelimbs (7). Furthermore, only HoxA and HoxD genes exhibit forelimb-specific collinear expression and no loss-of-function mutations in HoxB or HoxC group genes have resulted in forelimb phenotypes (14, 15). Here we report an analysis of Hox9 quadruple mutant mice (Hoxa9^−/−; Hoxb9^−/−; Hoxc9^−/−; Hoxd9^−/−) and demonstrate that these genes are critical for early forelimb AP patterning, revealing an unexpected role for Hox genes in limb development.

Results

Hox9 single mutants show no obvious limb deformities (11, 16); however, loss of both Hoxa9 and Hoxd9 (Hoxa9^−/−; Hoxd9^−/−) results in mild defects in stylopod patterning compared with controls, including a shorter humerus and loss of the deltoid crest, as reported previously (Fig. 1 A and C) (11). The loss of three additional alleles of Hoxb9 and Hoxc9 does not enhance the forelimb phenotype, and loss of function of other possible seven-allele combinations also results in relatively normal forelimb phenotypes (Fig. 1D). However, Hox9 quadruple mutants exhibit severe forelimb defects, with complete loss of all posterior skeletal elements (Fig. 1B), remarkably similar to phenotypes previously reported in Shh mutants and Hand2 conditional loss-of-function mice (5, 8, 17). This phenotype is 100% penetrant in Hox9 quadruple mutant forelimbs; however, no defects in hindlimb pattern are observed (Fig. 1 E and F). These unexpected forelimb defects demonstrate that Hox9 paralogous genes play important roles in forelimb AP patterning.

Open in a separate window

Fig. 1.

Skeletal defects in Hox9 quadruple mutant forelimbs. Compared with WT forelimbs (A), cartilage and skeletal patterns of Hoxa9/d9 double mutants at E18.5 show a shorter humerus and loss of the deltoid crest (C, red arrowhead), identical to those reported previously (11). The severity of this phenotype is not affected with the addition of up to three mutant alleles of Hoxb9 and Hoxc9 (seven-alelle mutant shown in D). Severe forelimb skeletal defects are observed in Hox9 quadruple mutants, showing loss of all posterior elements with 100% penetrance (B). However, these phenotypes are restricted to Hox9 mutant forelimbs. The hindlimbs in Hox9 quadruple mutants do not exhibit any skeletal malformations compared with controls (E and F).

Examination of the expression of all four Hox9 genes in the forelimb demonstrates that Hoxa9 and Hoxd9 are expressed throughout the forelimb bud at early stages (Fig. 2 A, B, G, and H) (11, 18, 19). Hoxb9 expression is detected in posterior regions of forelimb buds at E9.5 (Fig. 2C), but is no longer detected by E10.5 (Fig. 2D) (16). Hoxc9 is expressed throughout the E9.5 forelimb bud, but also is not detected by E10.5 (Fig. 2 E and F). These experiments suggest that redundant Hox9 function in forelimb patterning is likely to occur early in the establishment of the forelimb field, before the role for collinear HoxA/D function is established. Consistent with the lack of disruption of hindlimb development, Hox9 expression in the hindlimb differs markedly from expression in the forelimb, with no obvious Hoxb9 or Hoxc9 expression observed during hindlimb emergence stages (Fig. S1).

Open in a separate window

Fig. 2.

Expression of Hox9 paralogous genes in early-stage WT forelimb buds. All four Hox9 paralogous genes are expressed in WT forelimb buds. At E9.5, Hoxa9 (A), Hoxc9 (E), and Hoxd9 (G) transcripts can be detected throughout the forelimb bud, whereas Hoxb9 expression (C) is restricted to posterior domains. By E10.5, Hoxa9 (B) and Hoxd9 (H) expression are still observed in forelimb buds, but Hoxb9 (D) and Hoxc9 (F) are no longer detected.

Based on the striking phenotypic similarities between Hox9 quadruple mutants and Shh mutant forelimbs, along with the importance of Shh in limb AP patterning (8, 17), we examined Shh expression in Hox9 mutant forelimbs. Consistent with the phenotypic defects, Shh expression is not initiated in Hox9 quadruple mutant forelimbs, although expression is normal in mutant hindlimbs (Fig. 3 A, B, E, and F and Fig. S2 A and B). Lack of maintenance of Fgf8 expression is consistent with loss of Shh expression in the Hox9 mutants. Although normal expression of Fgf8 is observed in Hox9 mutants at early stages (Fig. 3 C and D), expression is reduced and restricted by E10.5 (Fig. 3 G and H).

Open in a separate window

Fig. 3.

Hox9 paralogous genes are required for the initiation of Shh expression and to maintain collinear Hoxd10–13 expression. (A–H) Shh and Fgf8 expression in Hox9 quadruple mutant forelimb buds. Shh expression is not initiated in Hox9 quadruple mutant forelimb buds at any stage (control and Hox9 mutants at E9.5 in A and B and at E10.5 in E and F). Fgf8 expression is unaltered at early stages in Hox9 mutant forelimbs compared with controls (C and D); however, by E10.5, Fgf8 expression is reduced and restricted to the posterior AER in Hox9 mutant forelimb buds (G and H). (I–T) Expression of Hoxd10–13 in Hox9 mutant limbs. Expression of Hoxd10 and Hoxd11 is detected at significantly lower levels in Hox9 mutant forelimb buds compared with controls at E10.5 (I–L), and is no longer detected in mutants by E11.5 (O–R). Expression of Hoxd13 is not detected at any stages examined in Hox9 mutants (M, N, S, and T).

Hoxd10–13 genes have been proposed to participate in activating and maintaining Shh expression in forelimb AP patterning (2, 6); thus, we examined the epistasis between Hox9 and posterior HoxD genes. In Hox9 mutants, Hoxd10 and Hoxd11 expression is initiated, although at reduced levels compared with controls (Fig. 3 I–L). The expression of Hoxd10 and Hoxd11 is not maintained and is no longer detected by E11.5 (Fig. 3 O–R). Hoxd13 expression is not detected at any limb stage in Hox9 mutant forelimbs (Fig. 3 M, N, S, and T). Consistent with the lack of hindlimb phenotype, Hoxd10–13 expression is unaltered in mutant hindlimbs compared with controls (Fig. S2 C–H). These data demonstrate that Shh and posterior HoxD genes are downstream of the Hox9 genes, and that posterior Hox expression is not maintained in the absence of Shh expression, as has been shown in the absence of Shh expression (10, 17).

Given that axial Hox9 expression can be detected at the earliest stages of lateral plate mesoderm formation (20), and that all four Hox9 genes are expressed in the emerging forelimb bud (Fig. 2) (16, 19), we examined initial outgrowth and patterning of the forelimb field in Hox9 mutants. Tbx5 is critical for forelimb bud initiation (21, 22). Tbx5 is expressed and maintained indistinguishably from controls in Hox9 quadruple mutant forelimb buds (Fig. 4 A and B), consistent with no general disruptions in forelimb bud initiation.

Open in a separate window

Fig. 4.

The Gli3-Hand2 pathway is disrupted in Hox9 mutant forelimbs. Comparable Tbx5 expression is observed in control and Hox9 mutant forelimb buds at E10.5 (A and B), consistent with no disruption in forelimb bud emergence in Hox9 mutants. Hand2 expression is dramatically down-regulated in Hox9 mutant forelimbs compared with controls (E–H; red arrows denote the expected anterior expression boundary); however, relatively normal expression of Hand2 is observed in other regions of E9.5 Hox9 quadruple mutant embryos, including the branchial arches (b) and posterior lateral plate mesoderm (C and D). As expected with loss of posterior Hand2 expression in the limb bud, anterior Gli3 expression is expanded in Hox9 mutant forelimbs compared with controls (I–L; yellow arrows denote the expected posterior expression boundary).

AP patterning is first established in the developing limb field by expression of Hand2 in the posterior emerging limb bud. Hand2 is initially expressed broadly in the flank lateral plate mesoderm, but becomes restricted to the posterior limb bud during outgrowth stages (3, 4). Consequently, Gli3 expression is restricted to the anterior limb bud, and the mutual antagonism between these two genes establishes limb field polarity, which is followed by restriction of Shh to the posterior ZPA (1). In Hox9 quadruple mutants, Hand2 expression in the branchial arches and the hindlimb bud is not affected, and Hand2 expression can be observed in the early flank lateral plate mesoderm (Fig. 4 C and D and Fig. S3 A and B). However, expression of Hand2 is severely affected in Hox9 mutant forelimb buds (Fig. 4 E–H). Hand2 expression is dramatically reduced at E9.5 and is not detected at later stages in Hox9 mutant posterior forelimb buds (Fig. 4 E–H). As expected given the antagonistic relationship between Hand2 and Gli3 (23), Gli3 expression in Hox9 quadruple mutants is not restricted anteriorly as in controls, and expression is observed throughout the mutant forelimb bud (Fig. 4 I–L).

Discussion

Our data identify the Hox9 paralogous genes as the earliest factor promoting AP polarity in the developing forelimb, acting upstream of Hand2 to establish the posterior domain of the forelimb bud. The defects in Hox9 quadruple mutant forelimbs result from the loss of Hand2 expression in the posterior limb bud, which is critical for the initiation of Shh signaling and expansion of the posterior limb skeleton (3, 4, 8). Although deletion of the HoxA/HoxD complex leads to loss of Shh signaling, these compound mutants do not result in altered Hand2 expression (6). Therefore, the early expression of Hand2 specifically requires the Hox9 paralogous gene function and defines these genes as the earliest known factors for initiating this polarity in the forelimb.

We integrate these data into a genetic model (Fig. 5) in which axial Hox9 paralog activity results in the establishment of Hand2 expression in the emerging forelimb bud. Gli3 expression in the anterior domain restricts Hand2 expression posteriorly, leading to activation of Shh in the ZPA. Shh is also activated by HoxA/D10–13 genes. Based on our genetic evidence, it is also possible that Hox9 participates in the activation or maintenance of Shh and/or HoxA/D10–13 as well.

Open in a separate window

Fig. 5.

Model for Hox9 gene function in forelimb AP polarity initiation. Our findings demonstrate that Hox9 paralogous genes are required to trigger posteriorly restricted Hand2 expression to initiate the AP field in the forelimb bud. Hand2 and HoxA/D10–13 activate Shh expression in the ZPA, and the forelimbs transition from the initiation stage to the propagation stage, where Shh is maintained by HoxA/D10–13 and Hand2. Although the same antagonism between Hand2 and Gli3 and initiation of posterior Shh expression is used in hindlimb AP patterning, no function for Hox9 genes is observed in the hindlimb. The factors acting upstream of Hand2/Gli3 in the hindlimb remain to be identified.

Even though the same genetic mechanism using the antagonism of Hand2 and Gli3 exists in the hindlimb, loss of Hox9 function plays no role in establishing these domains in the hindlimb. Furthermore, no combinations of more posterior Hox paralogous mutants to date have revealed any similar contribution to hindlimb AP patterning (13), suggesting that whereas forelimb AP patterning is established by the activity of paralogous Hox genes along the body axis, other non-Hox genes and/or more complex combinations of posterior Hox genes might have been co-opted for this activity in the hindlimbs.

Previous studies have demonstrated that AbdB-related HoxA and HoxD genes are important in forelimb development (6, 10–13, 24). However, there has been no evidence that HoxB and HoxC genes are required for forelimb patterning events (14, 15). Our data demonstrate unequivocally that HoxB and HoxC complex genes also make critical contributions to forelimb development. The forelimb posterior domain is established by the concerted activity of all four Hox9 genes (Hoxa9, Hoxb9, Hoxc9, and Hoxd9) to regulate Hand2 expression in this region. Further work is needed to determine how the posterior domain is determined in the hindlimb and what additional factors along the rostrocaudal axis participate in the establishment of limb AP patterning.

Materials and Methods

Mutant mouse strains, early skeletal preparations, and the standard procedure for whole-mount in situ hybridization used in this study have been described previously (25). Hoxa9 and Hoxb9 sequences were subcloned into pBluscript SK plasmids by PCR amplification of their cDNAs using primers Hoxa9-F 5′-gttggtcgctcctgactttc-3′, Hoxa9-R 5′-ggaagctgcaaggactgaag-3′, Hoxb9-F 5′-tacccaagtgagtggggaag-3′, and Hoxb9-R 5′-tcctgcagggaaaatacgag-3′, and antisense probes were transcribed by T7 polymerase (Roche) after the linearization of plasmids by EcoRI digestion. The probes used in this study, including Tbx5, Shh, Fgf8, Hoxd13, Hand2, and Gli3, were kindly provided by Dr. Licia Selleri (Cornell University) and Dr. Xin Sun (University of Wisconsin-Madison) (2, 26).

Supplementary Material

Acknowledgments

Footnotes

References