Trunk  mesodermal  enhancers  at  the  HoxD  locus

Correct Hox gene activation in the main body axis in time and space is instrumental for the correct patterning of the axial skeleton (Deschamps and van Nes, 2005). The

regulatory mechanisms behind spatial and temporal collinearity have been reported to mostly be located within the clusters. However, previous work has uncovered a role for sequences located outside of the cluster in the activation of Hoxd gene expression along the main body axis of the developing embryo (Tschopp and Duboule, 2011; Tschopp et al., 2009).

Transgenic-based experiments have identified several local cis-regulatory elements that can reproduce, for the most part, a Hox-like expression pattern in the main body axis (Bel-Vialar et al., 2002; Brend et al., 2003; Charité et al., 1995; Kwan et al., 2001;

Oosterveen et al., 2003; Sharpe et al., 1998). However, our transgenic mouse lines containing Hoxd clusters from different vertebrate species clearly show that, in most vertebrates, regulatory sequences located in the gene deserts that flank the Hoxd cluster are necessary for normal expression in the embryonic trunk mesoderm. This is further explored in the mouse context where we find that this regulatory potential is mostly located in the telomeric desert.

The genes that are mostly under the control of this regulatory region, Hoxd9-Hoxd3, are interestingly the subset of Hoxd genes that is expressed in the upper trunk mesoderm.

Accordingly, 5’ Hoxd genes Hoxd12-d10 that mostly contact the centromeric desert have been shown to have local regulatory elements that correctly reproduce their expression in the trunk (Gérard et al., 1993; Herault et al., 1998). However, although these small construct transgenic approaches are useful to assess the regulatory capabilities of a given sequence, erroneous conclusions can be extrapolated in what concerns the endogenous activity of this sequence. Indeed, previous work has shown that a Hoxd4 mouse transgene was able to recapitulate the domain of expression of the endogenous gene in the mouse trunk (Zhang et al., 1997). However, the use of a single-copy version of a similar Hoxd4 construct, elicited reporter gene expression exclusively in the neural tube (Tschopp et al., 2012). The same study also showed that both Hoxd9 and Hoxd3 single-copy transgenic construct were able to drive gene expression in the upper trunk mesoderm (Tschopp et al., 2012). Therefore, it would seem that the enhancers responsible for trunk Hoxd gene expression are also located inside of the cluster but that they are not sufficient to drive reporter gene expression in the main body axis of the embryo either in single copy number or within a larger genomic context.

3.2.1 Reallocation of snake mesodermal enhancers

Although most vertebrate BAC lines generated showed dorsally restricted expression in the trunk of the embryo, the snake BAC revealed a strikingly different regulatory potential.

Indeed, the snake HoxD cluster appears to contain most of the necessary regulatory elements that allow for normal expression in the trunk mesoderm. This finding indicates a major change in long-range regulatory allocation at the snake HoxD locus. Indeed, an H3K27ac

ChIP of the snake upper trunk only revealed two putative active enhancer regions in the 3’

gene regulatory desert. In addition, a validated mouse trunk mesodermal enhancer contains a sequence that is highly conserved all the way to amphibians but absent in snakes. However, a mouse construct lacking this region would be necessary to confirm the functional relevance of the sequence conservation. Alternatively, transgenic constructs of different species, such as the frog, could be generated in order to assess if the ability to drive reporter gene expression in the mesoderm is maintained.

3.2.2 Mechanisms of Hox gene repression in the posterior trunk

Although the corn snake has suffered extensive regulatory changes in what concerns Hoxd gene regulation in the upper trunk, similar regulatory mechanisms to those that operate in mouse and the human appear to be employed in posterior trunk Hox gene downregulation.

While the shutting down of Hox gene transcription in this context has not been addressed before, previous work has shown that a Polycomb silencing mechanism is associated with the transition between the two phases of limb development. Therefore, in distal limb tissue, H3K27me3 is detected both over the most 3’-located Hoxd genes as well as on the telomeric gene desert that contains enhancers for activating gene expression in the proximal limb (Andrey et al., 2013).

In the main body axis of the embryo, we find that the telomeric gene desert has a role in the downregulation of Hoxd gene regulation. However, the regulatory mechanism underlying this repression does not seem to require H3K27me3 coverage either in the cluster or in the surrounding gene deserts. It is therefore possible that negative cis-regulatory sequences lie in the telomeric gene desert and that, upon their deletion, Hoxd genes are free to be expressed in the embryonic tail. Further experiments would be necessary to locate more precisely these regulatory sequences. We also cannot exclude that this histone mark will be detected later in development to reinforce the silenced state since only a single stage has been assayed.

3.2.3 Towards a disorganized cluster?

Hox clusters of jawed vertebrates are known for their high level of compaction and organization, a very distinctive structure that has been hypothesized to arise due to the appearance of long-range global enhancers (Duboule, 2007). However, squamate Hox clusters seem to stray form this trend by having accumulated a big amount of transposable elements.

We find that snakes (and perhaps squamates in general) have acquired a larger number of

trunk mesodermal enhancers within the HoxD cluster. This change in regulatory sequence allocation could be a result of the failure of long-range enhancers to correctly activate Hox gene expression in the main body axis tissues of the snake. In this scenario, proximal enhancers would be positively selected in detriment of more distant regulatory sequences (Figure 24).

Another possibility can arise if we consider that long-range regulation at the HoxD locus is normally of a global nature, acting on groups of genes rather than individually.

Therefore, the snake might have lost the need for the existence of a global regulatory modality at the HoxD locus. Alternatively, a more gene-specific regulation in the developing trunk might have been required to pattern the snake trunk.

While the change in regulatory strategies operating at the HoxD cluster could have been caused by the presence of transposable elements at this locus, we cannot exclude the possibility that TEs have appeared as a consequence of global regulation becoming obsolete.

In order to test this hypothesis it would be interesting to use another TE rich cluster, such as the lizard’s in a transgenic mouse approach. Such an experiment would indicate if a short-range type of regulation is found in other TE-rich clusters or if it is restricted to snakes in particular.

If indeed a short-range regulation is sufficient for shaping the snake body plan, we can envision a scenario whereby an organized cluster would no longer be essential. If this is the case, evolutionary time and perhaps the sequencing of other squamate species might reveal the emergence of an atavistic disorganized cluster (Figure 24).

Figure 25 – Towards a disorganized cluster?

Proposed model for the re-allocation of mesodermal regulatory potential in the snake.

While initially most regulatory activity would be achieved by local enhancers (blue) acting only on neighbouring genes, the appearance of long-range global enhancers (red) would have promoted the consolidation of the cluster (Duboule, 2007). Perhaps due to the unusual amount of repeat elements within the cluster of snakes, and other squamates, local enhancers might have been selected in detriment of long-range global regulators.

Eventually such a scenario could result in an even larger and perhaps disorganized cluster in the squamate clade.