Enhancers were first discovered in simian virus 40 (SV40), where it was found that they function in an orientation-independent manner to stimulate transcription on heterologous genes [50]. Since then, a variety of analyses have revealed that animal genomes contain a large number of putative enhancers, out numbering coding genes [51,52]. It is challenging to identify cis-regulatory elements, such as enhancers, encoded in the genome through sequence analyses and computational methods alone [53]. Major efforts have been made to find ways of identifying and characterizing enhancers and their properties on a genome-wide and individual basis, which is important to facilitate our ability to decode regulatory information embedded in the genome [54,55,56,57,58,59,60]. While many development specific enhancers, including some of those discovered in the Hox clusters, are evolutionarily conserved [35,61,62,63], many adult or tissue-specific enhancers can be highly variable across species [64,65]. Even when enhancers are highly conserved, it can be challenging to understand the information content and the critical arrangements of the cis-elements that govern their ability to regulate expression [54,59,60]. Furthermore, highly conserved patterns of gene expression can arise through enhancers that display divergence [65]. Enhancers serve to stimulate transcription by integrating a variety of different regulatory inputs and binding sites for TFs to confer precise temporal, spatial and cell-type specific gene expression programs. Precise regulatory outputs from enhancers do not require that upstream factors have highly restricted domains of expression and can arise through the cumulative integration of weak, imprecise or wide-spread inputs by TFs [66,67]. The convergence of inputs can result in the integration of disparate and very broad patterns of regulatory signals into robust and tightly controlled specific outputs. Similarly, clusters of weak enhancers can synergize to serve as super enhancers to robustly regulate gene expression [68].

Enhancers can be located directly upstream of a gene or up to over a megabase away from its target gene promoter [69,70]. They frequently reside within introns of genes, even in ones they do not regulate, [70], and there is evidence for enhancers and cis-regulatory elements embedded in coding exons, including those of Hox genes [71,72,73]. Studies have shown that enhancer regions are themselves transcriptionally active. Several groups have demonstrated that non-coding enhancer RNAs are more than just transcriptional noise or byproducts of the transcriptional machinery, but are useful indicators in predicting active enhancers [74].

A challenge in identifying the targets of enhancer activity is that they can function independently of their orientation with respect to target genes and can make long-range enhancer–promoter contacts to more than the near adjacent genes [75]. Typically, an average vertebrate enhancer can be ~5 to 50 kb away from target promoters and ~1 to 10 kb away in the more compact Drosophila genome [39]. Intriguingly, the proximity of an enhancer to its target promoter required for functional activity is variable. Hence, there are no clear rules on how close enhancers should be positioned relative to the promoters they activate. Some studies have shown through proximity-dependent ligation techniques, such as 3C (Chromatin conformation capture), that enhancers physically come into contact with promoters, resulting in the activation of gene transcription [76]. Imaging approaches have shown that following the activation of genes, the distance between the enhancers and their target promoter tends to increase, suggesting a change in their interactions dependent upon their activity state [77]. This raises fundamental questions, such as, how do enhancers locate and distinguish between the target genes they activate; what confers enhancer–promoter specificity; and what degree of proximity is essential for the enhancer interactions required for gene regulation [39,55,69,75,78,79,80]?

These questions are relevant to understanding the regulation of the Hox clusters because of the high gene density and compact nature of the clusters. The enhancers embedded within and flanking an individual Hox cluster can display selective preferences, competition between genes and can regulate both near adjacent genes or act more globally on other genes in the complex. For example, in the mouse Hoxb complex, there are three RAREs in the middle of the cluster, two upstream and one downstream of Hoxb4 (Figure 2A), which participate in mediating its response to RA by regulating multiple coding and long non-coding (lncRNAs) transcripts [33,35,37,81]. One of these RAREs (DE-RARE) is an essential cis element of an RA-dependent enhancer, which undergoes epigenetic modifications, and is required to coordinate the global regulation of Hoxb genes in hematopoietic stems cells [34]. This functional role for an enhancer raises many questions regarding the mechanisms through which the DE-RARE participates in regulating so many transcripts, how targets are selected and the dynamics of the process. Why, in contrast, do other enhancers embedded in the Hoxb cluster only appear to work on a single near adjacent gene [37,62,82,83,84]?