ABSTRACT Topologically associating domains (TADs) have been proposed to both guide and constrain enhancer activity. Shh is located within a TAD known to contain all its enhancers. To investigate the importance of chromatin conformation and TAD integrity on developmental gene regulation, we have manipulated the Shh TAD – creating internal deletions, deleting CTCF sites, and deleting and inverting sequences at TAD boundaries. Chromosome conformation capture and fluorescence in situ hybridisation assays were used to investigate the changes in chromatin conformation that result from these manipulations. Our data suggest that these substantial alterations in TAD structure have no readily detectable effect on Shh expression patterns or levels of Shh expression during development – except where enhancers are deleted – and result in no detectable phenotypes. Only in the case of a larger deletion at one TAD boundary could ectopic influence of the Shh limb enhancer be detected on a gene ( Mnx1 ) in the neighbouring TAD. Our data suggests that, contrary to expectations, the developmental regulation of Shh expression is remarkably robust to TAD perturbations.

Limb-specific Shh expression is regulated by the (∼1 Mb distant) ZRS enhancer. In the mouse, limb bud-restricted spatiotemporal Shh expression occurs from ∼E10 to E11.5 at the distal posterior margin and is essential for correct autopod formation. Here, we have analysed the higher-order chromatin conformation of Shh in expressing and non-expressing tissues, both by fluorescence in situ hybridisation (FISH) and by chromosome conformation capture (5C). Conventional and super-resolution light microscopy identified significantly elevated frequencies of Shh /ZRS colocalisation only in the Shh -expressing regions of the limb bud, in a conformation consistent with enhancer-promoter loop formation. However, in all tissues and at all developmental stages analysed, Shh -ZRS spatial distances were still consistently shorter than those to a neural enhancer located between Shh and ZRS in the genome. 5C identified a topologically associating domain (TAD) over the Shh /ZRS genomic region and enriched interactions between Shh and ZRS throughout E11.5 embryos. Shh /ZRS colocalisation, therefore, correlates with the spatiotemporal domain of limb bud-specific Shh expression, but close Shh and ZRS proximity in the nucleus occurs regardless of whether the gene or enhancer is active. We suggest that this constrained chromatin configuration optimises the opportunity for the active enhancer to locate and instigate the expression of Shh .

Coordinated gene expression controlled by long-distance enhancers is orchestrated by DNA regulatory sequences involving transcription factors and layers of control mechanisms. The Shh gene and well-established regulators are an example of genomic composition in which enhancers reside in a large desert extending into neighbouring genes to control the spatiotemporal pattern of expression. Exploiting the local hopping activity of the Sleeping Beauty transposon, the lacZ reporter gene was dispersed throughout the Shh region to systematically map the genomic features responsible for expression activity. We found that enhancer activities are retained inside a genomic region that corresponds to the topological associated domain (TAD) defined by Hi-C. This domain of approximately 900 kb is in an open conformation over its length and is generally susceptible to all Shh enhancers. Similar to the distal enhancers, an enhancer residing within the Shh second intron activates the reporter gene located at distances of hundreds of kilobases away, suggesting that both proximal and distal enhancers have the capacity to survey the Shh topological domain to recognise potential promoters. The widely expressed Rnf32 gene lying within the Shh domain evades enhancer activities by a process that may be common among other housekeeping genes that reside in large regulatory domains. Finally, the boundaries of the Shh TAD do not represent the absolute expression limits of enhancer activity, as expression activity is lost stepwise at a number of genomic positions at the verges of these domains.

Conservation within intergenic DNA often highlights regulatory elements that control gene expression from a long range. How conservation within a single element relates to regulatory information and how internal composition relates to function is unknown. Here, we examine the structural features of the highly conserved ZRS (also called MFCS1) cis -regulator responsible for the spatiotemporal control of Shh in the limb bud. By systematically dissecting the ZRS, both in transgenic assays and within in the endogenous locus, we show that the ZRS is, in effect, composed of two distinct domains of activity: one domain directs spatiotemporal activity but functions predominantly from a short range, whereas a second domain is required to promote long-range activity. We show further that these two domains encode activities that are highly integrated and that the second domain is crucial in promoting the chromosomal conformational changes correlated with gene activity. During limb bud development, these activities encoded by the ZRS are interpreted differently by the fore limbs and the hind limbs; in the absence of the second domain there is no Shh activity in the fore limb, and in the hind limb low levels of Shh lead to a variant digit pattern ranging from two to four digits. Hence, in the embryo, the second domain stabilises the developmental programme providing a buffer for SHH morphogen activity and this ensures that five digits form in both sets of limbs.

A late phase of HoxD activation is crucial for the patterning and growth of distal structures across the anterior-posterior (A-P) limb axis of mammals. Polycomb complexes and chromatin compaction have been shown to regulate Hox loci along the main body axis in embryonic development, but the extent to which they have a role in limb-specific HoxD expression, an evolutionary adaptation defined by the activity of distal enhancer elements that drive expression of 5′ Hoxd genes, has yet to be fully elucidated. We reveal two levels of chromatin topology that differentiate distal limb A-P HoxD activity. Using both immortalised cell lines derived from posterior and anterior regions of distal E10.5 mouse limb buds, and analysis in E10.5 dissected limb buds themselves, we show that there is a loss of polycomb-catalysed H3K27me3 histone modification and a chromatin decompaction over HoxD in the distal posterior limb compared with anterior. Moreover, we show that the global control region (GCR) long-range enhancer spatially colocalises with the 5′ HoxD genomic region specifically in the distal posterior limb. This is consistent with the formation of a chromatin loop between 5′ HoxD and the GCR regulatory module at the time and place of distal limb bud development when the GCR participates in initiating Hoxd gene quantitative collinearity and Hoxd13 expression. This is the first example of A-P differences in chromatin compaction and chromatin looping in the development of the mammalian secondary body axis (limb).

The mechanism by which left-right (LR) information is interpreted by organ primordia during asymmetric morphogenesis is largely unknown. We show that spleen and pancreatic laterality is dependent on a specialised, columnar mesodermal-derived cell layer referred to here as the splanchnic mesodermal plate (SMP). At early embryonic stages, the SMP is bilateral, surrounding the midline-located stomach and dorsal pancreatic bud. Under control of the LR asymmetry pathway, the left SMP is maintained and grows laterally. Mice carrying the dominant hemimelia ( Dh ) mutation lack the SMP. Significantly, the mice are asplenic and the pancreas remains positioned along the embryonic midline. In the absence of Fgf10 expression, the spleno-pancreatic mesenchyme and surrounding SMP grow laterally but contain no endodermal component, showing that leftward growth is autonomous and independent of endoderm. In the Bapx1 –/– mutants, the SMP is defective. Normally, the SMP is a source for both Fgf9 and Fgf10; however, in the Bapx1 mutant, Fgf10 expression is downregulated and the dorsal pancreas remains at the midline. We conclude that the SMP is an organiser responsible for the leftward growth of the spleno-pancreatic region and that Bapx1 regulates SMP functions required for pancreatic laterality.

The middle ear apparatus is composed of three endochondrial ossicles (the stapes, incus and malleus) and two membranous bones, the tympanic ring and the gonium, which act as structural components to anchor the ossicles to the skull. Except for the stapes, these skeletal elements are unique to mammals and are derived from the first and second branchial arches. We show that, in combination with goosecoid ( Gsc ), the Bapx1 gene defines the structural components of the murine middle ear. During embryogenesis, Bapx1 is expressed in a discrete domain within the mandibular component of the first branchial arch and later in the primordia of middle ear-associated bones, the gonium and tympanic ring. Consistent with the expression pattern of Bapx1 , mouse embryos deficient for Bapx1 lack a gonium and display hypoplasia of the anterior end of the tympanic ring. At E10.5, expression of Bapx1 partially overlaps that of Gsc and although Gsc is required for development of the entire tympanic ring, the role of Bapx1 is restricted to the specification of the gonium and the anterior tympanic ring. Thus, simple overlapping expression of these two genes appears to account for the patterning of the elements that compose the structural components of the middle ear and suggests that they act in concert. In addition, Bapx1 is expressed both within and surrounding the incus and the malleus. Examination of the malleus shows that the width, but not the length, of this ossicle is decreased in the mutant mice. In non-mammalian jawed vertebrates, the bones homologous to the mammalian middle ear ossicles compose the proximal jaw bones that form the jaw articulation(primary jaw joint). In fish, Bapx1 is responsible for the formation of the joint between the quadrate and articular (homologues of the malleus and incus, respectively) enabling an evolutionary comparison of the role of a regulatory gene in the transition of the proximal jawbones to middle ear ossicles. Contrary to expectations, murine Bapx1 does not affect the articulation of the malleus and incus. We show that this change in role of Bapx1 following the transition to the mammalian ossicle configuration is not due to a change in expression pattern but results from an inability to regulate Gdf5 and Gdf6 , two genes predicted to be essential in joint formation.

We have previously shown that the paired-box transcription factors Pax1 and Pax9 synergistically act in the proper formation of the vertebral column. Nevertheless, downstream events of the Pax1/Pax9 action and their target genes remain to be elucidated. We show, by analyzing Pax1;Pax9 double mutant mice, that expression of Bapx1 in the sclerotome requires the presence of Pax1 and Pax9, in a gene dose-dependent manner. By using a retroviral system to overexpress Pax1 in chick presomitic mesoderm explants, we show that Pax1 can substitute for Shh in inducing Bapx1 expression and in initiating chondrogenic differentiation. Furthermore, we demonstrate that Pax1 and Pax9 can transactivate regulatory sequences in the Bapx1 promoter and that they physically interact with the Bapx1 promoter region. These results strongly suggest that Bapx1 is a direct target of Pax1 and Pax9. Together, we conclude that Pax1 and Pax9 are required and sufficient for the chondrogenic differentiation of sclerotomal cells.