A chromosomal memory triggered by Xist regulates histone methylation in X inactivation

Abstract

We have elucidated the kinetics of histone methylation during X inactivation using an inducible Xist expression system in mouse embryonic stem (ES) cells. Previous reports showed that the ability of Xist to trigger silencing is restricted to an early window in ES cell differentiation. Here we show that this window is also important for establishing methylation patterns on the potential inactive X chromosome. By immunofluorescence and chromatin immunoprecipitation experiments we show that histone H3 lysine 27 trimethylation (H3K27m3) and H4 lysine 20 monomethylation (H4K20m1) are associated with Xist expression in undifferentiated ES cells and mark the initiation of X inactivation. Both marks depend on Xist RNA localisation but are independent of silencing. Induction of Xist expression after the initiation window leads to a markedly reduced ability to induce H3K27m3, whereas expression before the restrictive time point allows efficient H3K27m3 establishment. Our data show that Xist expression early in ES cell differentiation establishes a chromosomal memory, which is maintained in the absence of silencing. One consequence of this memory is the ability to introduce H3K27m3 efficiently after the restrictive time point on the chromosome that has expressed Xist early. Our results suggest that this silencing-independent chromosomal memory has important implications for the maintenance of X inactivation, where previously self-perpetuating heterochromatin structures were viewed as the principal form of memory.

Figure 1. Epigenetic Imprints at the Initiation of X Inactivation

(A–H) Indirect immunofluorescence and subsequent DNA FISH analysis on mitotic chromosomes prepared from undifferentiated clone 36 ES cells after 3 d of Xist induction. H3K27m3 (A), H4K20m1 (B), and Ezh2 (D) are enriched on the arms of Chromosome 11 upon ectopic Xist expression. H3K9m2 (C) is not enhanced upon Xist expression. H3K4m2 (E) is reduced on Chromosome 11 upon Xist induction (green box) and absent from pericentric heterochromatin and the Y chromosome (orange arrow). (F) Histone H4 multiple-lysine acetylation is partially reduced (green box, left panel). Hypoacetylation (red) is restricted to chromosomal regions which show high levels of H3-K27 trimethylation (green, right panel). H3K9m3 (G) and H3K27m1 (H) are enriched at constitutive heterochromatin of pericentric regions and the Y (orange arrows). (I–K) Indirect immunofluorescence (upper panels) and subsequent Xist RNA FISH (red, Xist RNA; blue, DAPI) analysis of H3K27m3 (I), H4K20m1 (J), and Ezh2 (K) in interphase nuclei of undifferentiated clone 36 ES cells expressing Xist for 3 d.

Figure 2. ChIP Mapping of H3K27m3, H4K20m1, H3K9m2, H3K4m3, and H3K4m2 on the Xist-Expressing Chromosome 11 during Differentiation of Clone 36 ES Cells

A genetic map of Chromosome 11 indicating the loci analysed is given on the left (Xist-TG, approximate integration site of Xist transgene; puro, PGKpuromycin marker). (A to F) Chromatin was prepared from undifferentiated clone 36 ES cells grown for 3 d in the presence (light bars) or absence (dark bars) of doxycycline. H3K27m3 and H4K20m1 were enriched at three intergenic microsatellite sequences at 18.0 (A), 45.5 (C), and 75.2 (D) cM. (B) H3K27m3 was established over the coding sequence of PGKpuromycin in doxycycline-induced cells, which was accompanied by a loss of H3K4m2 and H3K4m3. (E) Tubulin control. (F) Control microsatellite located on Chromosome 15. (G–L) Analysis of H3K27m3, H4K20m1, and H3K9m2 in clone 36 ES cells differentiated for 9 d with (light bars) or without (dark bars) doxycycline. Histone methylation marks were monitored. Experiments were performed in duplicate, and the standard error is indicated in the graphs.

Figure 3. Sequences of Xist RNA Required for H3K27m3 Establishment

(A) Schematic representation of the Xist cDNA (top) indicating repeats A to E, restriction sites, and the locations of deletions (coloured bars) relative to the location of sequences required for localisation (black and hatched boxes; Wutz et al. 2002). (B) Analysis of H3K27m3 on metaphase chromosome spreads from undifferentiated ES cells after 3 d of Xist induction (see text). The staining patterns (n > 100) were scored as chromosome-wide dense methylation (black), reduced methylation (grey), and a single band (open). (C) Pattern of H3K27m3 triggered by different Xist mutants on metaphase chromosomes after 3 d of induction. Enlarged view of Chromosome 11 (clone 36) or the X chromosome (T20 lines, J1 knock-in line). (D) Focal H3K27m3 staining in interphase nuclei (percentage given; n > 100) of undifferentiated ES cells expressing Xist constructs.

Figure 4. Restriction of H3K27m3 Establishment and Transcriptional Silencing in Differentiation

(A) Initiation of H3K27m3 during clone 36 ES cell differentiation. Xist expression was induced at the beginning (+) or at various time points (24 to 120 h) after the start of differentiation, or not induced (−). The percentages of interphase cells showing H3K27m3 (black bars; n > 700) and Ezh2 (grey bars; n > 200) staining were determined at day 12 of differentiation. (B) Initiation of transcriptional silencing during differentiation was assessed by Northern blot analysis of PGKpuromycin (puro) and Gapd as a loading control in parallel cultures as described for (A). (C) Western analysis of Ezh2 and Eed protein levels during differentiation of clone 36 ES cells after induction with retinoic acid. Histones H3 and H4 were used as a loading control. (D) Establishment of H3K27m3 during embryonic development. Xist expression was induced from the single X chromosome of male Xist-tetOP embryos (see text) for 3 d (E9.5–12.5 and E13.5–16.5). The percentage of cells with H3K27m3 staining in interphase (left) and clusters of Xist RNA (right, open bars) are given (n > 300). Grey areas indicate the proportion of H3K27m3-positive cells to Xist-positive cells. (E) Xist RNA FISH (top) and H3K27m3 (bottom) staining of histological sections prepared from neck connective tissue of embryos described in (C).

Figure 5. Kinetic Study of H3K27m3 Stability

(A) The percentage of interphase nuclei (n > 100) showing H3K27m3 staining and Xist RNA was analysed for undifferentiated clone 36 ES cells, which expressed Xist for 3 d (+) or were further grown without inducer for 6, 12, 24, or 48 h. (B) Representative images of the time points analysed in (A) are shown. (C) Reversibility of H3K27m3 in differentiating clone 36 ES cells. The percentage of interphase cells showing H3K27m3 staining (n > 100) was determined for cells differentiated for 4 d in the presence of doxycycline (+) or further differentiated for 48, 72, or 96 h in the absence of inducer.

Figure 6. Early Xist Expression Imparts a Chromosomal Memory Independent of Silencing

Transgenic Xist expression was induced from Chromosome 11 in clone 36 ES cells (black bars) or a silencing-deficient Xist RNA from the X in J1:XistΔSX-tetOP ES cells (open bars) at time points during differentiation (see text). The percentage of cells showing H3K27m3 staining is plotted (n > 250). Below, a scheme of Xist induction is given for all cultures, with arrows representing time of analysis.

Figure 7. Establishment of Chromosomal Memory during ES Cell Differentiation

(A) Clone 36 ES cells were differentiated for 13 d in the presence of doxycycline (lane 1) or in the absence of inducer (lane 2) and the percentage of cells with H3K27m3 staining was determined (n > 800). At the beginning of differentiation, parallel cultures received either no Xist induction (lane 3) or a pulse of doxycycline for 24 h (lane 4), 36 h (lane 5), 48 h (lane 6), 60 h (lane7), or 72 h (lane 8) followed by withdrawal of inducer and concerted late induction from day 8 to day 13. A dashed red line indicates the 24-h interval of the transition when the chromosomal memory is recruited. (B and C) Establishment of irreversible transcriptional silencing during differentiation. (B) Ectopic inactivation of Chromosome 11 caused by Xist induction in differentiating clone 36 ES cells was assessed by Northern blot analysis of PGKpuromycin (puro) and Gapd as a loading control. Lanes were aligned electronically for better readability. ES cells were differentiated for 13 d in the presence of doxycycline (lane 1) or in the absence of inducer (lane 2). At the start of differentiation, parallel cultures received a Xist pulse for 24, 36, 48, or 60 h followed by withdrawal of inducer for the rest of the time (lanes 3 to 7) or followed by reinduction of Xist at day 8 of differentiation (lanes 8 to 11). All cells were analysed at day 13 of differentiation. (C) A quantitation of the puro expression relative to Gapd was derived from two independent Northern blots using tnimage software. A dashed red line indicates the 24-h interval in which the transition from reversible to irreversible silencing occurs.

Figure 8. Model for the Transition from Initiation to Maintenance of X Inactivation

Phases of X inactivation are given relative to days of ES cell differentiation (bottom). (A) In undifferentiated ES cells, efficient chromosome-wide H3K27m3 depends on both Xist RNA localisation to the chromosome in cis and initiation of transcriptional silencing via the A repeat (black triangles). (B) Early in differentiation, silencing becomes dispensable for high-level H3K27m3 (dotted arrow). (C) The beginning of the critical window is specified in that Xist loses its potential to trigger H3K27m3 (dotted arrow) and transcriptional silencing. The critical window is negotiated by sustaining high levels of H3K27m3, which is thought to constitute—together with Xist RNA—the signal for the recruitment of the chromosomal memory (black oval). The memory is established on the Xi exactly when silencing becomes irreversible and Xist independent. (D) During the maintenance phase of X inactivation the chromosomal memory allows Xist RNA to establish H3K27m3 efficiently.