An Xist-activating antisense RNA required for X-chromosome inactivation

Abstract

The transcriptional imbalance due to the difference in the number of X chromosomes between male and female mammals is remedied through X-chromosome inactivation, the epigenetic transcriptional silencing of one of the two X chromosomes in females. The X-linked Xist long non-coding RNA functions as an X inactivation master regulator; Xist is selectively upregulated from the prospective inactive X chromosome and is required in cis for X inactivation. Here we discover an Xist antisense long non-coding RNA, XistAR (Xist Activating RNA), which is encoded within exon 1 of the mouse Xist gene and is transcribed only from the inactive X chromosome. Selective truncation of XistAR, while sparing the overlapping Xist RNA, leads to a deficiency in Xist RNA expression in cis during the initiation of X inactivation. Thus, the Xist gene carries within its coding sequence an antisense RNA that drives Xist expression.

Figure 1. Expression of a novel Xist antisense transcript from the inactive X chromosome.

(a) Schematic representation of the Xist locus and the location of the primers and RNA FISH probes used. P1 and P2 are start sites of two distinct Xist isoforms. (b) Detection of an Xist antisense transcript from the inactive paternal X chromosome (X_p) in F1 hybrid TS and XEN cell lines lacking expression of the Xist antisense transcript Tsix (X^ΔTsixX^JF1) from the maternal X chromosome (X_m) by strand-specific RT–PCR. Sanger sequencing of cDNAs from females reveals an X_p-specific SNP at bp 804 in Xist exon 1. Genomic DNAs from the same cell lines displays nucleotides from both Xs. (c,d) Strand-specific RT–PCR amplification of Xist antisense RNAs in wild-type (WT) F1 hybrid (F1_i, initial cross; F1_r, reciprocal cross) TS (c) and XEN (d) cell lines. Sanger sequencing of cDNAs detects SNPs from both Xs in females. The active X_m expresses Tsix; the inactive X_p expresses the novel Xist antisense transcript XistAR. (e) Strand-specific amplification of XistAR in an X^ΔTsixX^JF1 EpiSC line with biased inactivation of the Tsix-mutant X_p. Sanger sequencing of the cDNAs detects Xist antisense expression from both Xs. As in c,d while Tsix is expressed from the active X_m, XistAR is expressed from the inactive X_p. M, marker; NTC, no template control; +, reaction with reverse transcriptase (RT); −, reaction without RT; NP, ‘no primer' control RT reaction without added primer in the RT step but with primers included in the PCR step to exclude reverse transcription of the sense Xist RNA via cell intrinsic primers. Strand-specific RNA FISH detection of XistAR (red), Xist (green) and Atrx (white) RNAs in TS, XEN and EpiSC cell lines. Atrx RNA marks the active X chromosome. Nuclei are stained blue with 4,6-diamidino-2-phenylindole. Three different TS and XEN cell lines and one EpiSC line were stained and >100 nuclei counted in each cell line. Numerical values in the images indicate the percentage of nuclei that display an antisense signal adjacent or overlapping with Xist RNA coat; ±, s.d.

Figure 2. XistAR expression in embryos.

Detection of XistAR in female embryonic day (E) 3.5 blastocyst embryos (a), E6.5 extra-embyonic ectoderm (b) and epiblast cells (c) by strand-specific RNA FISH. Numerical values in the images indicate the percentage of nuclei that display an antisense signal adjacent to or overlapping with Xist RNA coat. n=4 E3.5 embryos; n=3 E6.5 embryo-dissected extra-embryonic ectoderm and epiblast; ±, s.d.

Figure 3. Mapping of XistAR.

(a) Schematic representation of 5′ RACE and RT–PCR strategy to delineate the structure of XistAR in WT F1_i hybrid TS cells. SNPs that distinguish the maternal and paternal X-chromosome alleles (X_m and X_p, respectively) are shown under the amplicons. Green amplicons, detection of XistAR expression from the inactive X. Red amplicon, lack of XistAR detection. (b) Mapping of 5′ end of XistAR to bp 2,802 of Xist by 5′ RLM-RACE. Sanger sequencing of the major amplified cDNA detects X_p-specific SNPs. Amplified genomic DNA displays SNPs from both alleles. (c–f) Mapping of XistAR by overlapping RT–PCRs. All amplicons except XF7/XR883 detect XistAR. The 3′ end of XistAR therefore maps to ∼Xist bp 13. In c Sanger sequencing demonstrates primary amplification of XistAR, but not the Xist antisense transcript Tsix. The PCR primers XR2640 and XR2530 in this amplicon map to intron 3 of Tsix, and thus do not amplify spliced Tsix RNA. The low-level amplification in males is presumptive unspliced Tsix RNA. In d,e Sanger sequencing detects SNPs from both Xs, since both XistAR and Tsix are reverse transcribed and amplified. In f, Sanger sequencing detects amplification of Tsix but not XistAR. M, marker; NTC, no template control; +, reaction with reverse transcriptase (RT); −, reaction without RT; NP, ‘no primer' control RT reaction without added primers.

Figure 4. Mapping of XistAR by RNA FISH.

(a) Schematic representation of the Xist locus and the position of strand-specific RNA FISH probes used to test XistAR expression in TS cells. Probes depicted by green arrows detect XistAR expression, whereas probes represented by red arrows do not. (b) Representative nuclei with or without XistAR expression. XistAR RNA (red); Xist RNA (green); blue, 4,6-diamidino-2-phenylindole (DAPI) staining of nuclei.

Figure 5. Identification of putative XistAR promoter and enhancer element.

(a) Relative luciferase activity in cells transfected with constructs containing the Xist P1 promoter, RepA promoter, putative XistAR promoter and enhancer sequences. In the enhancer assay, the fragment was cloned upstream of the putative XistAR promoter sequence (see Methods for full details). Each construct was tested in triplicate in each of three separate transfections. a.u., arbitrary units. (b) Comparisons of relative expression of Xist, XistAR, Tsix RNAs by RT-qPCR in three XEN and TS cell lines (XistAR was profiled in X^ΔTsixX cell lines). Each cell line was tested in three technical replicates. P values were calculated using Welch's two-sample t-test. Error bars represent s.d.

Figure 6. XistAR is required for robust Xist induction.

(a) Schematic representation of Tsix^pA (X^pA) and Xist^IVS (X^IVS) mutations and the locations of the RT–PCR primers used. (b) RT–PCR detection of XistAR proximally and distally to the insertion cassette in X^JF1X^IVS and X^JF1X^pA blastocysts. M, marker; NTC, no template control; +, reaction with reverse transcriptase (RT); −, reaction without RT; NP, ‘no primer' control RT reaction without added primer in the RT step but with primers included in the PCR step. (c) RT–PCR detection of Xist in X^JF1X^Lab, X^JF1X^IVS and X^JF1X^pA blastocysts. Xist levels were quantified via pyrosequencing (see Methods). Three blastocysts of each genotype were analysed. (d) Defective silencing of the X-linked genes Utx, Rnf12, Atrx and Pdha1 in X^JF1X^pA compared with X^JF1X^Lab and X^JF1X^IVS blastocysts. X_p, paternal allele; X_m, maternal allele. Allele-specific expression levels in c,d were quantified by pyrosequencing and compared using Welch's two-sample t-test. In d P values were adjusted using the Bonferroni correction to account for multiple testing. Five to fifteen embryos were analysed for each gene. Error bars represent s.d. (e) RT–qPCR comparisons of Xist expression proximally, across, distally and at the junctions of the inserted cassette in XX, XX^IVS and XX^pA blastocysts. Data were normalized to Gapdh expression. Three different embryos were analysed from each genotype in triplicate. a.u., arbitrary units. (f) Relative quantification of XistAR expression by RT–qPCR upstream of the inserted cassette (that is, towards the 5′ end of XistAR) in XX^Lab, XX^IVS and XX^pA blastocysts. The amplicon is unique to XistAR and not present in mature Tsix (location of primers shown in Fig. 6a). Two different embryos of each genotype were analysed in triplicate. In e,f P values were calculated using Welch's two-sample t-tests. Error bars represent s.d. (g) Allele-specific analysis of Tsix expression in X^JF1X^Lab, X^JF1X^IVS and X^JF1X^pA blastocysts. The SNP position, at 110 bp of Tsix, is shaded in blue. Representative genomic DNA displays both alleles.

Figure 7. Profiling of Xist RNA coating in XX^Lab, XX^IVS and XX^pA embryos.

(a) RNA FISH detection of Xist RNA coating in XX^Lab, XX^IVS and XX^pA blastocysts. Red, XistAR RNA; green, Xist RNA; white, Atrx; blue, nucleus. Atrx RNA marks the active X chromosome. Numerical values in the images indicate the percentage of nuclei that display Xist RNA coats; ±, s.d. n=3 embryos per genotype. XX^pA embryos display significantly fewer nuclei with Xist RNA coating compared with both XX^IVS (P<10⁻³; Welch's two-sample t-test) and XX^Lab (P<10⁻⁴). (b) Sanger sequencing chromatograms of Xist cDNAs in F1 hybrid X^JF1X^Lab, X^JF1X^IVS and X^JF1X^pA E6.5 epiblasts (top). Similar analysis was performed with embryonic epiblasts from the reciprocal cross of these genotypes (bottom). In X^JF1X^pA and X^pAX^JF1 epiblasts, Xist is expressed exclusively from the X^JF1X chromosome.