Hotair Is Dispensible for Mouse Development

Abstract

Despite the crucial importance of Hox genes functions during animal development, the mechanisms that control their transcription in time and space are not yet fully understood. In this context, it was proposed that Hotair, a lncRNA transcribed from within the HoxC cluster regulates Hoxd gene expression in trans, through the targeting of Polycomb and consecutive transcriptional repression. This activity was recently supported by the skeletal phenotype of mice lacking Hotair function. However, other loss of function alleles at this locus did not elicit the same effects. Here, we re-analyze the molecular and phenotypic consequences of deleting the Hotair locus in vivo. In contrast with previous findings, we show that deleting Hotair has no detectable effect on Hoxd genes expression in vivo. In addition, we were unable to observe any significant morphological alteration in mice lacking the Hotair transcript. However, we find a subtle impact of deleting the Hotair locus upon the expression of the neighboring Hoxc11 and Hoxc12 genes in cis. Our results do not support any substantial role for Hotair during mammalian development in vivo. Instead, they argue in favor of a DNA-dependent effect of the Hotair deletion upon the transcriptional landscape in cis.

Fig 1. Hotair expression in vivo.

(A) Schematic representation of the wild type Hotair locus and the various Hotair deletion alleles. The deleted DNA is in red. The HoxC allele is from [18] and the shorter deletions in the box from [20] and [21]. (B) Whole mount in situ hybridization (WISH) of Hotair RNAs on E12.5 wild type CD1 (left) and CBA/C57/B6 (center) control embryos and of a Del(Hotair)^-/- mouse embryo (right, n = 3). No signal was detected in Del(Hotair)^-/- embryos, demonstrating the specificity of the probe. Hotair is expressed with a posterior restriction (white dashed line), resembling the transcript distribution of either a Hox11 or a Hox12 gene. Black arrowheads indicate expression domain of Hotair in the hindlimbs, hollow arrowheads indicate the limit of Hotair expression in the trunk and in the genital tubercle. The common artifact signal in the cerebral vesicles results from incomplete opening of these vesicles and subsequent probe trapping. (C) Schematic representation of the dissection patterns for RNA-seq. These dissections involved forelimbs (FL, dark green), hindlimbs (HL, green) and the genital tubercle (GT, magenta), as well as three trunk sections corresponding to the lumbar/sacral (T1, light blue); sacro/caudal (T2, blue) and caudal (T3, purple) regions. (D) Quantification of Hotair expression by RNA-seq (normalized RPKM values). (E) Quantification of Hotair expression (normalized RPKM values) in tail tip fibroblasts (TTF), using data from [20].

Fig 2. Hotair deletion has little impact, if any, on skeletal morphology.

Alizarin Red and Alcian Blue skeletal staining of wild type and Del(Hotair)^-/- mice. (A) Lumbar region of wild type (left) and Del(Hotair)^-/- (right). In our (B6xCBA) background, both control and mutant animals have five lumbar vertebrae (L5), with an equally low incidence of L6 (see Table 1). (B) The sacro-caudal region of wild type (left) and Del(Hotair)^-/- (right) animals, with the black arrowhead pointing to a moderate gain of lateral protrusion in mutant caudal vertebra 5 (C5), usually not observed in control animals. (C) Normal wrist and ankle bones in both wild type (left) and Del(Hotair)^-/- (right) animals. The number and organization of mesopodial bones remained unchanged in the mutant condition.

Fig 3. Overview of gene expression patterns in wild type and Del(Hotair)^-/- embryonic tissues.

First factorial map for the principal component analysis (PCA) of gene expression levels. Tissues are color-coded (upper right corner) and the genotypes are indicated by either a circle (wild type) or a triangle (Del(Hotair)^-/-). The numbers in parentheses indicate the proportion of the variance explained by PC1 or by PC2.

Fig 4. Differential expression between wild type and Del(Hotair)^-/- dissected samples.

A-B) Differential gene expression analysis between wild type and Del(Hotair)^-/- trunk tissues (T1, T2, T3). The absolute fold change is > 1.5 and FDR < 0.05. The different columns correspond to sample type and rows correspond to differentially expressed genes. (A) Heat map of centered and scaled gene expression levels (Z-score log2 RPKM). Genes are color coded vertically, according to tissue and expression changes between genotypes. (B) Venn diagram showing the number of down-regulated (top) and up-regulated (bottom) genes. C-D) Differential gene expression analysis between wild type and Del(Hotair)^-/- forelimbs (FL), hindlimbs (HL) and genital tubercle (GT). The absolute fold change is > 1.5 and FDR < 0.05. (C) Heat map of centered and scaled gene expression levels (Z-score log2 RPKM). Genes are color coded vertically according to the tissue and orientation of expression between genotypes. (D) Venn diagram showing the number of down-regulated (top) and up regulated (bottom) genes.

Fig 5. Expression of Hox genes in the various wild type and Del(Hotair)^-/- embryonic tissues.

Heat map of log2-transformed RPKM expression levels for all Hox genes. The columns correspond to sample type (indicated on top) and the rows correspond to Hox genes (indicated on the right). The blue boxes point to down-regulated genes, whereas the black boxes indicate up-regulated genes (FDR < 10%, no minimal fold change threshold).

Fig 6. The deletion of Hotair does not alter Hoxd genes expression in embryo.

(A) RNA-seq expression profiles of Hoxd genes in both the GT and T3 tissues of wild type (green) and Del(Hotair)^-/- (orange) E12.5 embryos. The Y-axis represents the per-base read coverage, normalized by dividing by the total number of million mapped reads in the corresponding samples. The two biological replicates were pooled for this representation and only uniquely mapping reads were used. (B) WISH of Hoxd10 and Hoxd11 on E12.5 wild type (left) and Del(Hotair)^-/- (right) embryos. The dashed lines indicate the rostral limits of the expression domains in the trunk, neural tube (black) and paraxial mesoderm (white). Adult vertebrae derive from the latter tissue. (C) Double WISH for the MyoD RNAs (for somite visualization) and either the Hoxd10 (upper panel) or Hoxd11 (lower panel) on E12.5 wild type (left) and Del(Hotair)^-/- (right) embryos. There was no detectable difference in the anterior limit of expression for any Hoxd gene analyzed.

Fig 7. The deletion of Hotair affects the expression of the neighboring Hoxc11 and Hoxc12 genes.

Hoxc12 (A) and Hoxc11 (B) expression (normalized RPKM values) in the various dissected tissue samples for wild type (green) and Del(Hotair)^-/- (orange) E12.5 embryos. The asterisk* indicates those samples where significant differences in transcript levels between genotypes were scored (FDR < 10%). (C) WISH using the Hoxc12 probe in both wild type (left) and Del(Hotair)^-/- (right) E12.5 embryos. The spatial expression of Hoxc12 remains globally unchanged. D) WISH of Hoxc11 in wild type (left) and Del(Hotair)^-/- (right) E12.5 embryos. The arrows indicate the slight anterior shift in the expression profile and the increase in signal intensity in the mutant genotype.

Fig 8. In-cis effects of the Hotair deletion on the local transcriptional activity.

A) RNA-seq expression profiles of the genomic region neighboring Hotair in both the developing genitalia (GT, four profiles on top) and the most posterior trunk tissue sample (T3, four profiles at the bottom) from either wild type (green) and Del(Hotair)^-/- (orange) E12.5 embryos. In the wild type GT, only the Hoxc11 gene is expressed along with Hotair on the opposite strand, which shows at least three putative start sites (arrows, TSS1 to TSS3). In the mutant GT, a long form of a new lncRNA (AntiHotair) now extends (grey box) on the Hox DNA strand, going over the deleted region up to the Hoxc11 promoter. On the opposite DNA strand, the Hotair TSS2 and TSS3 are still functional and produce Ghost of Hotair (Ghostair), yet another new species of lncRNA, specific for the Del(Hotair)^-/- allele (in orange) and absent from the control allele (in green). A similar situation is observed in the T3 trunk sample, except that Hoxc12 and AHotair are also expressed there. In the native locus, anti-Hotair is produced and meets with the end of the Hotair transcript. In the deleted allele, Ghostair is produced by the remaining Hotair TSS and terminates close to the 3’ end of the Hoxc12 transcript (bottom two profiles). The gray boxes indicate the genomic regions used for the expression quantifications of AHotair, LAHotair and Ghostair. The Y-axis represents the per-base RNA-seq read coverage, normalized by dividing by the total number of million mapped reads in the corresponding samples. The two biological replicates were pooled for this representation and only uniquely mapping reads were used. B-D) Expression values (normalized RPKM) for AHotair (B), LAHotair (C) and Ghost of Hotair (D) in all tissue samples. Genotypes are color-coded with wild type in green and Del(Hotair)^-/- in orange. The asterisk* indicates those samples where significant differences in expression were scored between the two genotypes (FDR < 10%).

Fig 9. Schematic summarizing the data shown in Fig 8.

In a wild type situation and at a body level anterior to Hoxc11 expression (from example in the upper lumbar region such as L3 or L4), Hoxc10 is active whereas the whole posterior part of the HoxC cluster is repressed (top left). Once Hoxc11 and Hoxc12 become activated, in more posterior regions of the body, both the Hotair and the AntiHotair RNAs are produced, from the anti-Hox and Hox DNA strands, respectively (top, right). Upon deletion of the Hotair locus, the posterior HoxC cluster remains closed for transcription in the anterior parts of the body (bottom, left). In contrast, the activation of the Hoxc11 region (bottom right) triggers the transcription of Ghostair on the opposite DNA strand, which meets the 3’ end of the Hoxc12 transcription unit, perhaps causing the light decrease in Hoxc12 mRNAs. On the Hox strand, the anti-Hotair RNA can now cross over the deleted region and contribute to the transcription of Hoxc11, perhaps inducing the observed light gain of expression of the latter gene.