Differentiation-dependent requirement of Tsix long non-coding RNA in imprinted X-chromosome inactivation

Abstract

Imprinted X-inactivation is a paradigm of mammalian transgenerational epigenetic regulation resulting in silencing of genes on the paternally inherited X-chromosome. The preprogrammed fate of the X-chromosomes is thought to be controlled in cis by the parent-of-origin-specific expression of two opposing long non-coding RNAs, Tsix and Xist, in mice. Exclusive expression of Tsix from the maternal-X has implicated it as the instrument through which the maternal germline prevents inactivation of the maternal-X in the offspring. Here, we show that Tsix is dispensable for inhibiting Xist and X-inactivation in the early embryo and in cultured stem cells of extra-embryonic lineages. Tsix is instead required to prevent Xist expression as trophectodermal progenitor cells differentiate. Despite induction of wild-type Xist RNA and accumulation of histone H3-K27me3, many Tsix-mutant X-chromosomes fail to undergo ectopic X-inactivation. We propose a novel model of lncRNA function in imprinted X-inactivation that may also apply to other genomically imprinted loci.

Figure 1. Absence of ectopic Xist induction from the X^ΔTsix maternal X-chromosome in embryonic day (E) 3.5 blastocyst embryos

(a) Schematic representation of the genomic structure of Xist, Tsix, and the Tsix RNA truncation mutant X^ΔTsix(b) RNA FISH detection of Xist (white), Tsix (green) and Atrx (red) RNAs in representative E3.5 embryos. Nuclei are stained blue with DAPI. Insets show representative nuclei. Scale bar, 25 µm. (c) Quantification of Xist, Tsix, and Atrx RNA expression patterns in blastocyst nuclei. The X-axis of each graph represents the average % nuclei observed in each class for each genotype. n=4 embryos per genotype. Diagrams along the Y-axis depict all observed expression patterns. +, RNA expression detected from a single X-chromosome; + +, RNA expression detected from both X-chromosomes; -, absence of RNA detection. Gene expression pattern does not differ significantly between wild-type and Tsix mutant blastocysts (Fisher’s exact test). Error bars, S.D. (d) RT-PCR detection of Xist, Tsix, and control b-actin RNAs. Three individual embryos are shown for each genotype. M, marker; NTC, no template control; +, reaction with reverse transcriptase (RT); -, no RT control lane. (e) Sanger sequencing chromatograms of representative Xist RT-PCR products. Highlights mark a single nucleotide polymorphism that differs between the maternal X^Lab / X^ΔTsix alleles and the paternal X^JF1 allele (see Methods). Both X^LabX^JF1 and X^ΔTsixX^JF1 females express Xist only from the paternally-inherited X-chromosome (Xp). The X^LabX^JF1 epiblast is a control sample displaying expression from both parental alleles.

Figure 2. Xist induction from the X^ΔTsix maternal X-chromosome in E6.5 extra-embryonic cells

(a) RNA FISH detection of Xist (white), Tsix (green), and Pgk1 (red) RNAs in E6.5 extra-embryonic cells. Nuclei are stained blue with DAPI. Dashed boxes mark representative nuclei. Scale bar, 10 µm. (b) Quantification of Xist, Tsix, and Pgk1 RNA expression patterns. The X-axis of each graph represents the % nuclei in each class out of 100 total nuclei counted per genotype (from n > 3 embryos per genotype). Diagrams along the Y-axis depict all observed expression patterns. +, RNA expression detected from a single X-chromosome; + +, RNA expression detected from both X-chromosomes; -, absence of RNA detection. Pairwise comparisons of the frequency of individual gene expression patterns between wild-type and X^ΔTsix mutant embryos were performed using Fisher’s exact test. *, 0.001<p<0.01; **, p ≤ 0.001. Extra-embryonic cells show significantly increased level of inactivation of the X^ΔTsix X-chromosome (p=0.0003 for males; p=3.5×10⁻⁵ for females). (c) RT-PCR detection of Xist and Tsix RNAs in extra-embryonic tissues from individual E6.5 embryos. Results from three individual embryos of each genotype are shown. M, marker; NTC, no template control; +, RT; -, no RT control lane. (d) Sanger sequencing chromatograms of Xist RT-PCR products. Highlights mark a single nucleotide polymorphism that differs between the maternal X^Lab / X^ΔTsix alleles and the paternal X^JF1 allele. X^LabX^JF1 females express Xist only from the paternally-inherited X-chromosome, while X^ΔTsixX^JF1 females express Xist biallelically in extra-embryonic tissues. X^ΔTsixY embryos variably express Xist from the maternally-inherited X-chromosome.

Figure 3. The X^ΔTsix maternal X-chromosome displays ectopic Xist induction only upon differentiation in trophoblast stem (TS) cells

RNA FISH detection of Xist (white), Tsix (green), and Atrx (red) RNAs in representative TS cell lines. Nuclei are stained blue with DAPI. Scale bar, 10 µm. Three cell lines of each genotype were analyzed. (b) RT-PCR detection of Xist, Tsix (two different amplicons), and control b-actin RNAs in wild-type (WT) and Tsix-mutant TS cells. Three TS cell lines of each genotype were analyzed. M, marker; NTC, no template control; +, RT; -, no RT control lane. (c) RNA FISH detection of Xist (white), Tsix (green), and Atrx (red) RNAs in 6-day (d6) differentiated TS cell lines. Three cell lines of each genotype were analyzed. Scale bar, 10 µm. (d) Quantification of Xist induction from the X^ΔTsix X-chromosome in females. Aberrant Xist RNA coating (defined as two Xist RNA coats in a diploid cell or Xist coating of all chromosomes in polyploid giant cells) is observed in mutant but not WT d6 differentiated TS cells. For giant cells, the number of X-chromosomes was identified based on distinct Xist and Atrx RNA FISH signals. n=100 nuclei counted for each cell line per day of differentiation. (e) Quantification of Xist induction from the X^ΔTsix X-chromosome in males. Aberrant Xist RNA coating is observed in mutant but not WT d6 differentiated TS cells. n=100 nuclei counted for each cell line per day of differentiation. (f) RT-PCR detection of Xist and control b-actin RNA in undifferentiated (d0) and d6 differentiated wild-type (WT) and Tsix-mutant TS cells. A single representative TS cell line from each genotype is shown. M, marker; NTC, no template control; +, RT; -, no RT control lane. (g) Sanger sequencing chromatograms of representative X^LabX^JF1 and X^ΔTsixX^JF1 Xist RT-PCR products (RNA), and an Xist genomic DNA amplicon (gDNA) within exon 1. Highlights mark a single nucleotide polymorphism that differs between the maternal X^Lab / X^ΔTsix alleles and the paternal X^JF1 Xist allele.

Figure 4. Characterization of differentiation-dependent Xist RNA induction from the X^ΔTsix maternal X-chromosome

(a) RNA FISH detection of Xist, Tsix, and the X-linked gene Atrx in undifferentiated and 6-day (d6) differentiated trophoblast stem (TS) cells. Immunofluorescence (IF) staining of the same cells detects CDX2, a marker of undifferentiated trohoectodermal cells. Scale bar, 10 µm. (b) Quantification of Xist induction in CDX2 positive and negative in undifferentiated and differentiated female TS cells. 100 nuclei were counted per cell line at each time point (n = 3 cell lines per genotype). No aberrant Xist induction is observed from the X^ΔTsix in undifferentiated cells. In d6 differentiated X^ΔTsixX^JF1 TS cells, ectopic Xist induction is restricted to cells that lack CDX2 staining. A subset of differentiated nuclei show both multiple Xist-coated inactive X-chromosomes and multiple active X-chromosomes, due to endoreduplication. (c) IF/RNA FISH analysis of male TS cells, as in (a). Scale bar, 10 µm. (d) Quantification of Xist induction in undifferentiated and d6 differentiated male TS cells. 100 nuclei were counted per cell line at each time point (n = 3 cell lines per genotype). (e) IF/RNA FISH detection of Xist, Tsix, and the X-linked gene Atrx, in differentiated TS cells. p57^Kip2, a marker of trophoblast giant cells, is detected in the same cells by IF. Scale bar, 10 µm. (f) Quantification p57^Kip2 positive cells and aberrant Xist induction in the TS cells. 100 nuclei were counted per cell line (n = 3 cell lines per genotype). X^ΔTsixX^JF1 and X^ΔTsixY TS cells show significantly reduced levels of p57^Kip2 staining, suggesting failure of these genotypes to terminally differentiate. Error bars, S.D.

Figure 5. Lack of Xist induction from the X^ΔTsix maternal X-chromosome in cultured extra-embryonic endoderm (XEN) cells

(a) RNA FISH detection of Xist (white), Tsix (green) and Atrx (red) RNAs in representative XEN cell lines. Nuclei are stained blue with DAPI. Three cell lines of each genotype were analyzed. Scale bar, 10 µm. (b) Quantification of Xist RNA coating and X-linked gene expression in the XEN cells. The X-axis of each graph represents average % nuclei in each class from 100 cells counted per cell line (n = 3 cell lines per genotype). Diagrams along the Y-axis depict all observed expression patterns. +, RNA expression detected from a single X-chromosome; + +, RNA expression detected from both X-chromosomes; -, absence of RNA detection. A subset of tetraploid XEN nuclei show two Xist-coated inactive X-chromosomes and two active X-chromosomes, due to endoreduplication. Gene expression patterns do not differ significantly between wild-type and Tsix mutant XEN cells (Fisher’s exact test). Error bars, S.D. (c) RT-PCR detection of Xist, Tsix, and control b-actin RNAs in three individual XEN cell lines of each genotype. M, marker; NTC, no template control; +, reaction with reverse transcriptase (RT); -, no RT control lane. (d) Sanger sequencing chromatograms of representative X^LabX^JF1 and X^ΔTsixX^JF1 RT-PCR products spanning Xist exons 1–4 (RNA), and an Xist genomic DNA amplicon (gDNA) within exon 1. Highlights mark a single nucleotide polymorphism that differs between the maternal X^Lab / X^ΔTsix alleles and the paternal X^JF1 allele.

Figure 6. Disassociation of Xist induction, H3-K27me3 enrichment, and inactivation of the X^ΔTsix maternal X-chromosome in E6.5 extra-embryonic cells

(a) RNA FISH detection of Xist, Tsix, and Pgk1 RNAs coupled with immunofluorescence (IF) detection of H3-K27me3 in extra-embryonic cells of E6.5 embryos. Dashed boxes mark representative nuclei. Scale bar, 10 µm. (b) Quantification of H3-K27me3 enrichment and Pgk1 expression in nuclei displaying Xist RNA coating of both X-chromosomes in X^ΔTsixX extra-embryonic cells (50 nuclei with Xist RNA coating of both X-chromosomes were analyzed [n=5 X^ΔTsixX embryos]). Wild-type (WT) XX embryos show Xist RNA coating and enrichment of H3-K27me3 on a single X-chromosome (n=5 embryos). (c) Quantification of H3-K27me3 enrichment and Pgk1 expression in nuclei displaying Xist RNA coating of the X-chromosome in X^ΔTsixY extra-embryonic cells (50 nuclei with Xist RNA coating of the single X-chromosome [n = 4 X^ΔTsixY embryos] were analyzed). WT XY cells show neither Xist RNA coating nor H3-K27me3 enrichment (n = 4 embryos).

Figure 7. Reactivation of the inactive X^ΔTsix paternal X-chromosome in the inner cell mass (ICM)

RNA FISH detection of Xist (white), Tsix (green), and Atrx (red) RNAs in E4.0 ICMs. Nuclei are stained blue with DAPI. Insets show representative reactivated nuclei. Scale bar, 20 µm. (c) Quantification of the number of reactivated nuclei, as characterized by loss of Xist RNA coating and biallelic Atrx expression, in individual ICMs (n=6 ICMs per genotype). The mean number of reactivated cells per ICM does not differ significantly between XX and XX^ΔTsix ICMs (p = 0.23, two-tailed T-test). Error bars, S.D. (d) Allele-specific X-linked gene expression analysis in E5.0 epiblast cells. Representative chromatograms of sequenced cDNAs show biallelic expression of the X-linked genes Pdha1, Rnf12, and Utx, regardless of genotype.

Figure 8. A model for the role of Tsix in imprinted X-Inactivation

The maternal X-chromosome, but not the paternal-X, is marked by histone modifications during gametogenesis that are transmitted to the offspring upon fertilization. In the pre-implantation embryo, these histone modifications prevent inactivation of the maternal-X, while the paternal X-chromosome is subject to inactivation. Xist is induced from the paternally-derived X-chromosome in the pre-implantation embryo, and helps recruit protein complexes that catalyze histone marks characteristic of facultative heterochromatin on the paternal-X. The oocyte-configured chromatin of the maternal-X, conversely, prevents Xist induction from the maternal X-chromosome during the initiation phase of X-inactivation (pre-implantation) and does not require Tsix. The maternal-X then remains active during the maintenance phase of imprinted X-inactivation in undifferentiated extra-embryonic nuclei (post-implantation, undifferentiated cells), independently of Tsix expression. Tsix is induced from the maternal-X due to the absence of Xist expression from this X-chromosome. Upon differentiation, Tsix transcription across the Xist promoter region is required to induce heterochromatinization of the Xist promoter to keep Xist silenced in the extra-embryonic trophectodermal lineage. In XX differentiated extra-embryonic cells, the wild-type maternal-X remains transcriptionally competent while the paternal-X is maintained as transcriptionally inactive in imprinted X-inactivated cells. Upon differentiation of X^ΔTsixX trophectoderm cells, the Tsix-mutant maternal X-chromosome induces Xist.