The spontaneous mutation Ironside (Irn). (A, bottom) Mice heterozygous for Irn (HoxD^Irn/+) display severe hindlimb paralysis, as illustrated by footprint analyses. (B) The Irn mutation is a combined deletion/insertion event at the HoxD cluster. Hox genes are indicated in blue, whereas Evx2 is in black. The mutated locus contains two deletions in cis (dashed lines) removing from Hoxd13 to within Hoxd8, and from within Hoxd8 further 3′. A fragment of Hoxd8 was left in place (bottom line; blue box), along with a piece of DNA translocated from a region located centromeric to the locus (top and bottom lines; green flag). This latter, noncoding sequence is still present at its original locus in the mutant configuration, in an inverted orientation. No other genomic rearrangement was detected outside the HoxD cluster.

Scheme of deleted mouse strains used in this work. On the top, the HoxD cluster with the position of the various LoxP sites (red). Each site represents an independent mouse strain. The strains containing the L5, L6, L7, and L8 sites were produced for this study, whereas others were described previously. Below are 11 deleted strains generated by recombinations between pairs of LoxP parental lines using the technique TAMERE (). The Ironside (Irn) and Del(1–13) alleles are also included. Breakpoints are in red, with a triangle for the remaining LoxP site. The deleted fragments are shown in dashed lines. For convenience, the strains are referred to as, for example, Del(10–13) for a deletion removing from Hoxd10 to Hoxd13, inclusively [HoxD^Del(10–13)]. One configuration carries in addition a stop codon engineered within the neighboring Evx2 gene (Evx2^stop). These deleted strains are classified in three groups, with the same color code throughout the paper, based on the severity of the paralysis, documented by footprint analyses of representative homozygous specimen. Group A (red) deletions are dominant and lead to complete paralysis, whereas group B (blue) animals have a recessive and less severe phenotype. Group C (green) animals have no obvious hindleg problem. The inactivation of Evx2 in the Del(10–13) deletion significantly rescues the locomotion defect.

Peroneal nerve defects in HoxD deletion strains. (A) Whole-mount neurofilament immunostainings of E12.5–E13.5 hindlimbs (posterior view, dorsal is right and distal at the bottom) differentiate various peroneal nerves as “close to normal,” “thinner/shorter,” and “truncated” or “absent.” Group C animals show close to normal or thinner/shorter peroneal nerves; group B animals display thinner/shorter or truncated phenotype, whereas severely affected group A animals show no peroneal nerve. (B, right) In addition to the complete loss of the dorsosacral plexus and emanating peroneal nerve (P), group A mutants also show innervation defects in the dorso-lumbar plexus (Ld), where fewer fibers are visible. Pictures are lateral views of a right hindlimb. (C) E14.5 leg cross-sections at shank level stained for neurofilaments revealed the absence of the deep (dP) and superficial (sP) peroneal nerves in group A specimens. (T) Tibial nerve (ventral); (P) peroneal nerve (dorsal); (L) lumbar plexus (anterior); (S) sacral plexus (posterior); (Fi) fibula; (Ti) tibia.

Homeotic transformations in HoxD deletion strains. (A) Neurofilament stainings at E11.5–E12.5 show that roots 21–23 participate in the lumbar plexus, whereas roots 23–25 form the sacral plexus, with characteristic projection patterns. (B) The 21st and 22nd roots are unchanged in Irn mutants, whereas more caudal roots are transformed into anterior identities, with different magnitudes, such that, for example, the 25th and 26th mutant roots appear like normal 24th and 25th. (C) Transformations are more restricted in Del(10–13); Evx2^stop mutants, as only the caudal-most 25th and 26th roots appeared transformed.(D) In contrast, posterior transformations are seen in the Del(10) allele, with all LS roots looking like their immediate posterior neighbor. (E) Crosses between HoxD deletions and a Lim1taulacZ knock-in transgenic line reveal the motor projections to the dorsal limb. Motor fibers in roots 21–24 are labeled, whereas only few (if any) fibers are detected in the 25th root. (F) The Irn mutant shows two motor defects: the absence of the dorso-sacral plexus (S) and anterior homeotic transformations. In this latter case, projections from the 23rd root mostly target the lumbar plexus, instead of going to both lumbar and sacral plexii, and the 25th root is stained, indicating a more anterior identity. (G) In contrast, a dorso-sacral plexus is visible in Del(10–13); Evx2^stop mutants, which have more subtle anterior homeosis, such as some positive fibers in the 25th root. (H) The Del(10) allele also shows a dorso-sacral plexus, whereas posterior homeosis transformed all motor nerves contributing to the hindlimb. Roots are numbered, with plain and dashed arrows indicating complete or partial transformation, respectively. Arrows to the left and right indicate anterior and posterior transformations, respectively. (L) Lumbar plexus; (S) sacral plexus. Asterisks indicate the dorso-sacral plexus and emanating peroneal branch.

Sequence of progressive homeotic transformations, from a full segment posterior transformation (B) to a full segment anterior transformation (F). (A) Normal hindlimb bud innervation. LMC, nerve roots, plexii, and approximate projections are indicated with a color code used throughout the figure. (B) Posterior homeosis observed in Del(10). LS roots are fully transformed toward a posterior projection pattern. (C) The wild-type organization is retained in some internal deletions [Del(i–9); Del(8–10); Del(9–10)]. (D) Complete anterior transformations, but restricted to the 25th and 26th roots, are scored in group B Del(10–13); Evx2^stop and Del(11–13) deletions. In this case, the LMC is caudally extended by one segment such that an additional root contributes to the sacral plexus. (E) More severe transformations also affect roots 21–24, to various degrees. The latter roots are nevertheless only partially transformed. For instance, partial transformation of the 23rd root involves rerouting of its axonal projection toward the lumbar plexus, at the expense of the sacral plexus, leading to a root with a hybrid profile between the 22nd and 23rd aspect [group A or Del(1–13) and Del(9) alleles]. (F) Full transformation of all lumbar roots following Hoxa10 inactivation (), pointing to a full-segment caudal shift of the LMC. Lateral and medial subsets of the LMC are shown in blue and orange, respectively. Arrows to the right indicate segment and projection transformation toward a posterior fate, whereas arrows to the left indicate the converse transformation.

Innervation defects are spinal cord-autonomous. (A–D) The floxed HoxD allele [Flox(4–13); depicted under I] and the Prx1-Cre deleter transgene were used to conditionally delete Hoxd4–Hoxd13 in early developing limb tissues. (B, asterisk) Heterozygous Del(10–13) animals often showed a severe peroneal nerve reduction. (B,C) Trans-heterozygous specimens carrying both the Del(10–13) and Flox(4–13) alleles along with the Prx1-Cre showed similar peroneal defects, yet with a normal leg posture at birth. (C,D) In contrast, fully deleted trans-heterozygous animals [Del(10–13)/Del(4–13)] showed the expected truncated peroneal nerve and bilateral clubfoot at birth. (G) The conditional deletion was monitored by the selective loss of Hoxd10 expression in E12.5 hindlimb buds.

Spinal shifts in Hox gene expression in various deletion stocks. (A,B) The lumbar homeosis observed in the Del(9) allele () correlates with significant caudal shifts in the expression of both Hoxd10 (A) and Hoxd11 (B) genes. Consequently, while only Hoxd9 is physically absent, the activities of Hoxd10 and Hoxd11 are also missing in the relevant lumbar region, leading to a severe anterior homeosis. (C) Conversely, a rostral shift of Hoxd11 expression is scored in Del(10) animals, accounting for the unexpected posterior root transformation (). (D) A comparable rostral shift of Hoxd11 expression is seen in Del(9–10). In this case, however, the additional loss of Hoxd9 function likely compensates for part of the Hoxd11 gain of function, generating a new functional balance with a normal nerve root organization ().

Complete leg paralysis in group A animals is associated with the gain of expression of Evx2 in lumbar LMC motoneurons. (A) E11.5 half-spinal cords of control (left) and Irn mutants (right). (Left) Expression of Evx2 is found in a column of interneurons. (Right) In the mutant stock, Evx2 is largely gained in a broad Hox-like spinal domain. (B) E12.5 LS spinal cords in an open-book configuration. All group A animals displayed a related Evx2 gain of expression, though limited to the ventral spinal cord at lumbar level [here Del(10–13)]. In contrast, less severely paralyzed (group B) animals sharing the same deletion breakpoint near Evx2 are devoid of ectopic Evx2 transcripts [e.g., Del(8–13)]. (C) In group A deletions, the Evx2 gene responds to neural regulatory sequences normally acting on Hoxd genes (yellow arrows). A modified HoxD BAC clone, with a deletion mimicking Irn outer breakpoints and a LacZ reporter cassette within Evx2 is used to generate transgenic mice (HoxD^{tg[Evx2lacZ;Irn]}). (D, left panel) X-gal staining of transgenic embryos confirms the impact of the deletion on Evx2 regulation (open-book view). Axon fibers innervating the hindlimb, but not the forelimb, are labeled, demonstrating that the spinal gain of expression includes lumbar LMC motoneurons (upper right and lower right panels, E11.5 and E13.5 hindlimbs, respectively). (T) Tibial nerve; (P) peroneal nerve.

Dominant paralysis in group A specimens is associated with a massive loss of hindlimb LMC motoneurons, including Pea3-positive motor pools. (A–E) E12.0–E13.0 LS spinal cords in an open-book configuration. Numbers refer to DRG levels. (A) At lumbar level, the LMC position is revealed by Raldh2 staining and appeared caudally shifted in both Del(10–13); Evx2^stop and Del(10–13) alleles. However, only the latter group A allele is accompanied by a significant loss of signal at levels 21–22, suggesting a loss of MN innervating the hindlimb. (B,C) Group B alleles show Pea3 expression reminiscent of the nerve root homeotic transformations scored after neurofilament stainings, with some motor pools shifted either caudally in anterior root transformations [Del(1–13)] (B) or rostrally in posterior transformations [Del(10)] (C). (D,E) More dramatic alterations of the Pea3 expression domains are scored in group A mutants. Staining in the 21–24 region (arrowhead) is reduced in Del(10–13) homozygous, indicating that the corresponding motor pools are strongly depleted. A decrease in staining intensity is already visible in Irn heterozygotes in E. The Pea3-positive pools at level 25 appear normal (arrow), although slightly shifted caudally in Del(10–13) homozygous. Differences in staining intensities reflect differences in age, with E12.0 samples [Del(1–13), Irn] showing a stronger signal than E12.5–E13.0 samples [Del(10), Del(10–13)].