Discovery of genes required for body axis and limb formation by global identification of retinoic acid–regulated epigenetic marks

Abstract

Identification of target genes that mediate required functions downstream of transcription factors is hampered by the large number of genes whose expression changes when the factor is removed from a specific tissue and the numerous binding sites for the factor in the genome. Retinoic acid (RA) regulates transcription via RA receptors bound to RA response elements (RAREs) of which there are thousands in vertebrate genomes. Here, we combined chromatin immunoprecipitation sequencing (ChIP-seq) for epigenetic marks and RNA-seq on trunk tissue from wild-type and Aldh1a2-/-embryos lacking RA synthesis that exhibit body axis and forelimb defects. We identified a relatively small number of genes with altered expression when RA is missing that also have nearby RA-regulated deposition of histone H3 K27 acetylation (H3K27ac) (gene activation mark) or histone H3 K27 trimethylation (H3K27me3) (gene repression mark) associated with conserved RAREs, suggesting these genes function downstream of RA. RA-regulated epigenetic marks were identified near RA target genes already known to be required for body axis and limb formation, thus validating our approach; plus, many other candidate RA target genes were found. Nuclear receptor 2f1 (Nr2f1) and nuclear receptor 2f2 (Nr2f2) in addition to Meis homeobox 1 (Meis1) and Meis homeobox 2 (Meis2) gene family members were identified by our approach, and double knockouts of each family demonstrated previously unknown requirements for body axis and/or limb formation. A similar epigenetic approach can be used to determine the target genes for any transcriptional regulator for which a knockout is available.

Role of Retinoic Acid Signaling, FGF Signaling and Meis Genes in Control of Limb Development

“…Recently, ChIP-seq (H3K27ac and H3K27me3) analysis combined with RNA-seq analysis performed on wild-type vs. Aldh1a2-/- trunk tissue found that the previously known RARE upstream of Fgf8 is normally marked by H3K27me3, but this mark is lost in the RA-deficient mutant [ 36 ]. In this study, two additional RAREs near Fgf8 were also found to be marked by H3K27me3 when RA is present, highlighting the existence of additional potential Fgf8 RARE silencers that can be further studied to pursue the mechanism underlying RA repression of Fgf8 in heart tissues compared to caudal tissues already examined [ 33 , 34 ].…”

“…In addition, no change in Meis2 expression in the interdigital tissues of the mouse was found in Rdh10 mutants lacking limb RA signaling [ 40 ]. Prior to forelimb bud initiation, RA is also not essential for Meis1/2 expression in trunk lateral plate mesoderm that gives rise to limb buds, but loss of RA does reduce Meis1/2 expression in somites [ 14 , 36 ].…”

“…Candidate targets were defined as genes with RA-regulated mRNA abundance that also have nearby RA-regulated H3K27ac (gene activation) or H3K27me3 (gene repression) marks associated with conserved RAREs. This approach identified many previously known RA target genes that control trunk formation, plus additional RA target genes were identified including Meis1 and Meis2 [ 36 ]. Meis1/2 were both found to have nearby RAREs that are highly conserved from mammals to reptiles or frogs [ 36 ].…”

“…This approach identified many previously known RA target genes that control trunk formation, plus additional RA target genes were identified including Meis1 and Meis2 [ 36 ]. Meis1/2 were both found to have nearby RAREs that are highly conserved from mammals to reptiles or frogs [ 36 ]. Together with previous findings, these new findings suggest that Meis1/2 are normally activated by RA in the trunk somitic mesoderm but not in the lateral plate mesoderm or limbs, and that the effects of RA and RAR antagonist treatments on Meis1/2 expression in limb are due to off-target effects in which high concentrations of these reagents target Meis1/2 RAREs in limb tissue to usurp control of these genes and override their normal limb activators and repressors.…”

“…However, redundancy between Meis1 and Meis2 may have masked a defect. CRISPR/Cas9 gene editing was performed to generate Meis1/2 double mutants in mice, and dissection at E10.5 resulted in embryos that exhibit a body axis defect (small somites) with developmental arrest at E9.5, plus a lack of forelimb buds which are normally easily visible at E9.5; arrested development prevented analysis of hindlimb buds which develop after forelimb buds [ 36 ]. Thus, global knockout of Meis1/2 unexpectedly resulted in the loss of forelimb bud initiation, which indicates an important role in limb development, in addition to a role in somitogenesis ( Figure 2 ).…”