Divergence  of  Hoxd  gene  regulation  in  the  snake  genital  bud

The bimodal chromatin structure, whereby "distal" Hox genes and "proximal" Hox genes preferentially contact the respective adjacent gene desert, was reported to be particularly relevant in the making of the tetrapod limb. The 5' genomic neighborhood, in particular, has been shown to be essential in driving Hoxd gene expression in mouse digits and genital tubercle (GT). In the absence of digits in snakes, we investigated if the preferential tropism of Hoxd13 to its adjacent gene desert in a whole-embryo context (see

above) was necessary for the development of snake external genitalia, termed hemipenes (HP). Interestingly, it has been proposed that mammalian genitals have a different embryonic origin compared to other amniotes such as squamates (Tschopp et al., 2014) and thus it is unknown if the regulation of Hox genes in genitals of snake and mouse is conserved.

Expression determined by in situ hybridization showed that 3’ genes such as Hoxd4 are expressed in the snake HP (Figure 10, top) while previous expression analysis in the mouse GT revealed that Hoxd3 and Hoxd4 are completely absent from this structure (Lonfat et al., 2014) (Figure 10, top). These differences in Hoxd gene expression in the developing genitalia of mice and snakes were confirmed by the analysis of the different vertebrate BAC transgenic lines. Since these lines contain a cluster isolated from both regulatory gene deserts, gene expression was expected to be restricted to the main body axis and absent from secondary structures. Indeed human, mouse, chicken and zebrafish Hoxd genes were mostly expressed in the trunk (Figure 10A, bottom and Figure 11). Interestingly, however, the snake BAC transgenics were able to express the snake Hoxd11 to Hoxd3 genes not just in the mouse main body axis but also in hindlimbs and genital bud (Figure 10A and Figure 11). This expression pattern is more consistent with a lack of repressive activity rather than with the presence of limb and genital enhancers located inside the clusters. We further investigated this issue by using RT-qPCR in genitals of both mouse and snake and used as a control the region of the trunk that is located at the same anterior-posterior level. We find that the mouse Hoxd9-d3 genes are expressed at much higher level in the trunk than in the genitals, while in the snake the expression in the two tissues is nearly the same (Figure 10B). These results indicate that the snake HoxD cluster does not contain the necessary sequences that restrict Hox gene expression to the trunk mesoderm allowing other long-range regulatory elements to take over to regulate expression in developing secondary structures.

To further explore how Hox genes are differentially regulated in the snake external genitalia, we performed a 4C-seq analysis using GT/HP tissue and control trunk tissue for both snake and mouse embryos. Interestingly, while the mouse shows, an increase of contacts of about 15% in the centromeric gene desert for both Hoxd11 and Hoxd13, the snake Hoxd13 and Hoxd11 interaction profiles only increase about 2% and 7%, respectively. We then looked at specific sequences that were described to be contacting the cluster in genital tissue only: GT1 and GT2 (Lonfat et al., 2014). Even though these sequences are conserved and therefore identifiable in the snake centromeric desert, they failed to significantly increase the interactions with distal Hox genes in the genitals of the snake. In fact, only Prox, a known GT and limb enhancer in the mouse, seems to gain substantial contacts in the snake HP compared to the control tissue.

Figure 10 - Regulation of mouse and corn snake Hoxd genes in developing genitals.

A. Endogenous Hoxd4 expression both in a E12.5 control mouse embryo and in a 8.5 dpo corn snake embryo. Higher magnifications of the cloacal regions are shown on the right, with the positions of the GT and HP delineated in white. Below are Hoxd4 in situ hybridization of either control or E11.5 embryos transgenic for the human and snake BAC clones. B. Quantifications of Hoxd13, Hoxd11, Hoxd10, Hoxd9, Hoxd4, Hoxd3 and Hoxd1 transcript levels either in mouse E12.5 GT or in snake 2.5 to 4.5 dpo HP by RT-qPCR. The log₂ ratios were calculated between genital and control trunk tissue expression values. C. Smoothed 4C-seq mapping using mouse and snake Hoxd13 and Hoxd11 as viewpoints and GT (mouse) and HP (snake) as samples along with a control sample (left).

The BamCompare subtract function was used for each viewpoint to compare sequence coverage in GT or HP versus control tissues. Genes are represented by grey rectangles and previously characterized mouse limb or GT enhancers are represented by red boxes below. The vertical shaded zones in pink represent sequences that displayed increased read coverage in GT versus control tissue, whereas the grey zones point to sequences

showing increased contact in mouse but not in snake genitals. The percentages show the relative amount of interactions over this particular landscape, calculated as in Figure 3.

Figure 11 - Interspecies comparison of the HoxD cluster regulatory potential

Hoxd11, Hoxd10 and Hoxd9 whole mount in situ hybridization of E11.5 transgenic mice containing either a human, a mouse, a snake or a zebrafish HoxD cluster integrated randomly in the genome. The snake HoxD cluster seems to trigger a much wider transcriptional response in mesoderm derivatives.