HOX transcription factors determine the identity of body regions along the rostro-caudal axis during bilaterian embryogenesis. In vertebrates Hox genes distinctively lie organized in dense clusters, each typically composed of a dozen paralogous transcription units spread over 100 kb of genomic DNA. In every Hox cluster the Hox genes are arranged following a complete 5'-3' transcriptional polarization. Even though Hox are found across the whole animal kingdom, this ordered physical arrangement distinctly characterizes vertebrate Hox clusters. This organization, and the attendant collinearity phenomena have been proposed to be a crucial feature to Hox genes' co-option in several structures organized along secondary body axes, as exemplified by the patterning of various tetrapod-specific structures (e.g. limbs, dig-its and external genitalia). The aim of this thesis is to investigate to what extent Hox transcriptional polarization results from constraints imposed by their close linear proximity when arranged in clusters. Therefore, we challenged the murine HoxD cluster by engineering inversions inside the cluster using CRISPR/Cas9 mutagenesis approaches in vivo. We produced two novel alleles bearing inversions of either Hoxd11 or Hoxd12. Complementarily, we also reanalyzed extensively a larger targeted inversion encompassing Hoxd11-d12 loci in a variety of embryonic organs and tissues. Together, the comparative analysis of these alleles by a combination of approaches (WISH, RNAseq, ChIPseq and 4Cseq) illuminated different types of outcomes resulting from the disruption of HoxD transcriptional polarity and the inversion of non-coding regulatory elements. From our work, we developed the view that engineered inversions of particular Hox genes are tolerated by this genetic system and does not result systematically in major functional disruption of the affected cluster. This result therefore demonstrates that Hox transcriptional polarity is not a feature arising from an absolute developmental constraint. This conclusion is nevertheless nuanced by two instances where intra-HoxD inversion disrupts the regulation and function of the HoxD cluster. In one case, the inversion of Hoxd11 affects the transcriptional output of its posterior neighbor Hoxd12. This perturbation is likely implying interference mechanism related to the transcriptional leakage emitted onto Hoxd12 by the inverted Hoxd11 locus. In the second case, the inversion of the Hoxd11-d12 loci in a Hoxd13lacZ background results in the specific gain-of-expression of this reporter gene in a variety of embryonic structures. Of interest, we present here diverse pieces of evidence that this dysregulation may not be directly related to the disruption of Hox gene transcriptional polarity but instead caused by the translocation and inversion of a CTCF binding site. This observation supports the hypothesis that CTCF dependent topological organization of the chromatin plays a role in the regulatory insulation of Hoxd13. It appears therefore that spontaneous inversions within polarized Hox clusters might affect the function and regulation of the Hox patterning mechanism in different functional contexts. This conclusion leads us to conjecture that once disposed in dense clusters, the developmental expression of Hox genes is extremely sensitive to genetic perturbations, due notably to the presence of several conserved polarized regulatory elements interspersed between the individual Hox genes i