In addition, we noted a gain of Hoxd1 expression in the paraxial mesoderm, neural tube and tail bud. These patterns suggest either that a repressive regulatory activity was contained within the deleted region or that Hoxd1 was brought closer to mesodermal enhancers located further in the telomeric gene desert.
Figure 17 - Hoxd1 expression is absent in the somites in Del(AttP-SB2) context.
Whole-‐mount in situ hybridizations of Hoxd1 in E9.5 embryos, the line names are to the below the pictures.
2.7 Screening for enhancer candidates driving Hoxd1 expression
Two hypotheses can be proposed in the light of our results. Regulatory elements involved in transient Hoxd gene expression during the formation of the whisker pad may have been recruited from more ancestral mechanisms at work during somitogenesis, so that both developmental processes involve overlapping sets of enhancers. Alternatively, they may be regulated by distinct groups of enhancers that would nevertheless be situated in close proximity compared to the wide regulatory landscape known to modulate Hoxd gene expression in other tissues.
In order to address these hypotheses, we searched for enhancer candidates. We performed qPCRs for each of the whisker pad tissues collected by different dissection methods (Figure 18A). Consistent with the results of the previous in situ hybridizations (Figure 14), we observed high levels of Hoxd1 transcription. Hoxd3 was lower than Hoxd1 but higher than other Hoxd genes. Hoxd4 expression was low but it can be explained by the stage that is early for the onset of this particular gene.
On the other hand, Hoxd8 was surprisingly high in every sample (Figure 18A), although in situ hybridization in the whisker pad showed only weak Hoxd8 expression (Figure 14D;Hoxd8). It may be explained by transcriptional activity in adjacent tissues, deeper under the skin, that could not be detected.
Given that the qPCRs show a clear Hoxd1 signal, we pursued our experiment and did an H3K27ac ChIP on whisker pads dissected according to the most suited method, i.e.
with the highest Hoxd1 expression and the best ratio between Hoxd8 and Hoxd1 (Figure 18A, purple asterisk). This histone modification is known to mark active promoters and enhancers. As a negative control, we used an H3K27ac profile of E12.5 embryonic brain, where Hox genes are known not be expressed (Figure 18B,C;
upper track). We wanted to compare the profiles in order to discriminate whisker-pad specific peaks from sequences that are constitutively acetylated.
To select enhancer candidates that would be active both in the whisker pad and during somitogenesis, we compared the E12.5 profiles with an H3K27ac ChIPseq of E8 posterior trunks that comprise somites (Figure 18B,C; blue track), using an E6 whole embryo ChIPseq as control, a stage where Hox genes did not start yet to be transcribed (Figure 18B,C; upper track). Analysing the ChIP-seq profiles revealed a marked increase of acetylation in a portion of telomeric desert that spans 400 kb starting from the HoxD cluster (Figure 18B; grey and green top tracks, global_ac).
This area was specifically acetylated in whisker pads and posterior trunks, but not in brains and in the E6 embryos. It comprises the 140-kb region that we previously highlighted (Figure 18B, green rectangles and blue highlight). This first analysis indicates that our selected 140-kb region contains specific enhancer activity to both the whisker pad and the somites.
We then looked more in detail at the 400 kb-long acetylated area to try to select enhancer candidates. We observed specific peaks common to the whisker pad and the posterior trunk, reinforcing the idea that the same genomic region contains regulatory elements active in both tissues (Figure 6C;green and grey rectangles). Interestingly, two sequences were clearly enriched in H3K27ac marks in the 140 kb-long highlighted region (Figure 6C;green rectangles), corroborating our transgenic results that pointed towards regulatory activity in this location.
Additional peaks were found further in the desert (Figure 18C;grey rectangles), but earlier BAC and in situ hybridization experiments showed that this portion of the acetylated region is not sufficient to drive reporter expression (Figure 16B;
BAC(TDproximal). However, our genetic approaches indicate that most of the regulatory elements active in the whisker pad are contained in the 140kb-long highlighted region (Figure 18B,C; blue highlight). For this reason, we chose to continue our analysis focusing mainly on this domain.
In summary, the detailed analysis of ChIP-seq acetylation profiles revealed two wide but specific peaks, located in our 140kb-long region and common between the somites and the whisker pad.
Figure 18 - Promoter and enhancer activities of the close telomeric desert
A Validation of the E12.5 WP dissection method: quantification of Hoxd1, Hoxd3, Hoxd4, Hoxd8, Hoxd9 and Hoxd13 transcript levels by RT-qPCR in E12.5 embryos (top) with corresponding schemes of the dissection technics (bottom). The purple asterisk indicates the method used to produce the sample of the ChIPseq. C is a zoom of B H3K27ac ChIPseq tracks, control tracks (brain E12.5 and whole embryo E6) are in grey, E12.5 whisker pad track is in green and E8 posterior trunk track in blue.
Blue highlight corresponds to the 140kb-long region of interest selected in Figure 16A. Selected acetylated regions are indicated by green rectangles under the green tracks, other peaks are in grey.
Schemes of the genomic landscape are depicted under the tracks, genes are represented by black rectangles, the CS39 conserved region is in red and the BACs in salmon color.
2.8 Genomic contacts between Hoxd1 and the telomeric gene