LIM homeobox transcription factors integrate signaling events that control three-dimensional limb patterning and growth

Abstract

Vertebrate limb development is controlled by three signaling centers that regulate limb patterning and growth along the proximodistal (PD), anteroposterior (AP) and dorsoventral (DV) limb axes. Coordination of limb development along these three axes is achieved by interactions and feedback loops involving the secreted signaling molecules that mediate the activities of these signaling centers. However, it is unknown how these signaling interactions are processed in the responding cells. We have found that distinct LIM homeodomain transcription factors, encoded by the LIM homeobox (LIM-HD) genes Lhx2, Lhx9 and Lmx1b integrate the signaling events that link limb patterning and outgrowth along all three axes. Simultaneous loss of Lhx2 and Lhx9 function resulted in patterning and growth defects along the AP and the PD limb axes. Similar, but more severe, phenotypes were observed when the activities of all three factors, Lmx1b, Lhx2 and Lhx9, were significantly reduced by removing their obligatory co-factor Ldb1. This reveals that the dorsal limb-specific factor Lmx1b can partially compensate for the function of Lhx2 and Lhx9 in regulating AP and PD limb patterning and outgrowth. We further showed that Lhx2 and Lhx9 can fully substitute for each other, and that Lmx1b is partially redundant, in controlling the production of output signals in mesenchymal cells in response to Fgf8 and Shh signaling. Our results indicate that several distinct LIM-HD transcription factors in conjunction with their Ldb1 co-factor serve as common central integrators of distinct signaling interactions and feedback loops to coordinate limb patterning and outgrowth along the PD, AP and DV axes after limb bud formation.

Fig. 1.

Expression of Lhx2, Lhx9, Lmx1b and Ldb1 in wild-type limb buds. (A) Whole-mount in situ hybridization of Lhx2, Lhx9 and Lmx1b. Only forelimbs are shown. Hindlimbs had the same expression patterns. (B) Expression of the Ldb1 protein in the hindlimb, as shown by immunohistochemistry on cryosections (9.5 dpc and 10.5 dpc) and paraffin sections (11.5 dpc and 12.5 dpc). Ldb1 was ubiquitously expressed in the limb bud. See Fig. S1 in the supplementary material for controls of the Ldb1 antibodies.

Fig. 2.

Limb defects in the Lhx2^-/-;Lhx9^-/- and Ldb1^c/-;T-Cre embryos. Limb skeletal preparations of mouse embryos are shown. (A) At 15.5 dpc, only Lhx2^-/-;Lhx9^-/- mutant limbs showed severely shortened zeugopod and autopod (arrows), and fewer digits in both the forelimb and the hindlimb. (B) At 15.5 dpc, Ldb1^c/-;T-Cre mutant hindlimb showed severely shortened zeugopod and autopod (arrows) and fewer digits. The phenotype in the hindlimb was more severe than that in the forelimb. FL, forelimb; HL, hindlimb; H, humerus; R, radius; U, ulna; Fe, femur; T, tibia; Fi, fibula; S, stylopod; Z, zeugopod; A, autopod.

Fig. 3.

Shh expression and response were impaired in the Lhx2^-/-;Lhx9^-/- and Ldb1^c/-;T-Cre limbs. Whole-mount in situ hybridization was performed to examine gene expression. In situ hybridization with the ³⁵S-labelled probes of Lmx1b and Grem1 was performed on limb sections at 10.5 dpc. (A) Expression of indicated genes in the 10.5 dpc Lhx2^-/-;Lhx9^-/- limbs. Reduced gene expression in the mutant limbs is indicated by arrows. FL, forelimb; HL, hindlimb. (B) Expression of indicated genes in the 10.75 dpc Ldb1^c/-;T-Cre hindlimb. (C) Significantly reduced Grem1 expression in the 10.5 dpc hindlimb of the Ldb1^c/-;T-Cre embryo and the 10.5 dpc forelimb of the Lhx2^-/-;Lhx9^-/- embryo. In A-C, reduced gene expression in the mutant limbs is indicated by arrows. (D) Loss of Grem1 expression was more severe in the ventral limb bud of the Lhx2^-/-;Lhx9^-/- embryo (arrow). Lmx1b expression in the dorsal limb was not altered in the Lhx2^-/-;Lhx9^-/- limb. (E) Grem1 expression was similarly lost in both dorsal and ventral Ldb1^c/-;T-Cre hindlimb bud at 11.0 dpc (arrow). D, dorsal; V, ventral. AER, apical ectodermal ridge.

Fig. 4.

Regulation of Grem1 expression by Shh was impaired in the Lhx2^-/-;Lhx9^-/- and Ldb1^c/-;T-Cre limbs. (A) Shh-coated beads were implanted into the hindlimb buds at 10.5 dpc and Grem1 expression was examined. (B) Grem1 expression was reduced in the 10.5 dpc Lhx2^-/-;Lhx9^-/- hindlimb bud, most notably in the posterior part. Ectopic Grem1 expression in the anterior limb bud was induced by Shh beads in the control Lhx2^+/-;Lhx9^+/- limb, but not in the Lhx2^-/-;Lhx9^-/- limb (arrows). Control beads were inserted to the contralateral limb buds. D, dorsal; V, ventral. (C) Shh-coated beads were implanted into 10.25 dpc forelimb buds and Fgf4 expression was not induced in the Lhx2^-/-;Lhx9^-/- limb (arrow). (D) Shh-coated beads were implanted into the hindlimb buds at 11.0 dpc. Ptch1 expression was upregulated in the Ldb1^c/-;T-Cre and wild-type limb buds.

Fig. 5.

Reduced Fgf10 and Fgf8 expression in the Lhx2^-/-;Lhx9^-/- and Ldb1^c/-;T-Cre limbs. Whole- mount in situ hybridization was performed. (A) Fgf10 expression was rapidly lost in the mutant limb bud. Loss of Fgf10 expression was first observed in the distal-most limb bud (arrow). Loss of Fgf10 expression in the Lhx2^-/-;Lhx9^-/- hindlimb bud was less significant (arrow). (B) Spry4 was expressed almost normally in the Ldb1^c/-;T-Cre mutant hindlimb bud at 10.5 dpc. (C) Fgf8 expression was progressively reduced in the AER of the Ldb1^c/-;T-Cre hindlimb bud (black arrows). The AER was completely flattened (yellow arrows) at 11.5 dpc, as shown by the scanning EM. (D) Fgf8 expression was lost in the posterior AER of the Lhx2^-/-;Lhx9^-/- forelimb bud at 10.5 dpc (arrow and broken line). (E) Lhx9 expression in the 11.5 dpc hindlimb. (F) Sox9 expression in the 12.5 dpc hindlimb bud. Small Sox9-expressing domains the Ldb1^c/-;T-Cre mutant are indicated by arrows. Z, zeugopod; A, autopod.

Fig. 6.

Production of output signals in response to Fgf8 signaling is disrupted in the Ldb1^c/-;T-Cre and Lhx2^-/-;Lhx9^-/- mutant limb buds. (A) The Fgf8-soaked bead implanted in the 10.5 dpc hindlimb failed to upregulate Fgf10 expression in the Ldb1^c/-;T-Cre mutant limb (arrow). (B) Fgf10-soaked beads, but not control beads, rescued Fgf8 expression in the AER of the Ldb1^c/-;T-Cre mutant limb (arrow). Beads were implanted into the 10.5 dpc hindlimb bud. Ventral view of the limbs is shown. (C) Fgf10-soaked beads, but not control beads, rescued Fgf8 expression in the AER of the Lhx2^-/-;Lhx9^-/- forelimb bud mutant limb (arrows). Beads were implanted into the 11.0 dpc forelimb bud. Dorsal view of the limbs is shown. Control beads fell off in the Lhx2^-/-;Lhx9^-/- forelimb bud during the in situ hybridization procedure. (D) Fgf8-soaked beads implanted into the 10.5 dpc hindlimb bud induced Spry4 expression in both Ldb1^c/-;T-Cre mutant and wild-type limb buds. (E) At 10.25 dpc, loss of Shh expression in the hindlimb of the Ldb1^c/-;T-Cre mutant embryo (arrow) was not rescued by the implanted Fgf8-soaked bead. Only small patches of Shh expression were detected around the Fgf8-soaked bead and under the AER. Shh expression in the hindgut is indicated by an arrowhead.

Fig. 7.

Reduced proliferation was detected in the Ldb1^c/-;T-Cre mutant limb. (A) Analysis of cell proliferation in wild-type and the Ldb1^c/-;T-Cre hindlimb bud by staining the limb sections with the anti-phosphohistone H3 (anti-PHH3) antibodies to show cells in M phase. (B) Stained and total cell numbers were counted in the boxed regions in A of the section from three independent limbs at the same stages. The average percentage of M-phase cells in different samples is shown in B. Reduction of cell proliferation was more severe in the ventral limb. (C) Cell death was analyzed by immunohistochemistry with cleaved caspase 3 antibodies in transverse sections of hindlimb buds at 11.5 dpc. The boxed proximal limb regions are shown at higher magnification underneath. Cell death was increased in the mutant. (D) Model of signaling interactions in the early limb bud mediated by the LIM-HD factors and Ldb1. Our data provide strong evidence that a transcriptional machinery composed of LIM-HD transcription factors Lhx2, Lhx9, Lmx1b and their common co-factor, Ldb1, mediates at least two distinct signaling feedback loops (orange and blue arrows, respectively) between the AER and the underlying mesenchyme of the limb bud. These feedback loops integrate limb growth and patterning along the PD, AP and DV axes. The LIM-HD transcription factors enable Fgf8 emanating from the AER to control Shh expression, which in turn governs AER maintenance through Grem1 expression in the mesenchyme (orange arrows). Grem1 expression also regulates Fgf4 expression (stripes) in the posterior AER. Strong Fgf signaling from the AER also inhibits Grem1 in the mesenchyme under the AER (Verheyden and Sun, 2008). In addition, these transcription factors regulate the feedback loop between Fgf10 and Fgf8 expression required for PD limb outgrowth (blue arrows).