Revisiting the consequences of deleting the X inactivation center

Significance

Mammalian cells equalize X-linked dosages between the male (XY) and female (XX) sexes by silencing one X chromosome in the female sex. X chromosome inactivation (XCI) requires the X inactivation center (Xic), which spans several hundred kilobases in the mouse and includes several required noncoding genes. Over three decades, transgenic and deletional analyses have demonstrated that the Xic is both necessary and sufficient to induce X chromosome counting, choice, and the initiation of whole-chromosome silencing. However, one recent study reported that deleting the noncoding Xic surprisingly had no effect on counting or initiation of XCI. Here, we revisited the question, created independent Xic deletion lines, and reaffirmed the essential nature of the Xic for counting and X chromosome silencing.

Abstract

Mammalian cells equalize X-linked dosages between the male (XY) and female (XX) sexes by silencing one X chromosome in the female sex. This process, known as “X chromosome inactivation” (XCI), requires a master switch within the X inactivation center (Xic). The Xic spans several hundred kilobases in the mouse and includes a number of regulatory noncoding genes that produce functional transcripts. Over three decades, transgenic and deletional analyses have demonstrated both the necessity and sufficiency of the Xic to induce XCI, including the steps of X chromosome counting, choice, and initiation of whole-chromosome silencing. One recent study, however, reported that deleting the noncoding sequences of the Xic surprisingly had no effect for XCI and attributed a sufficiency to drive counting to the coding gene, Rnf12/Rlim. Here, we revisit the question by creating independent Xic deletion cell lines. Multiple independent clones carrying heterozygous deletions of the Xic display an inability to up-regulate Xist expression, consistent with a counting defect. This defect is rescued by a second site mutation in Tsix occurring in trans, bypassing the defect in counting. These findings reaffirm the essential nature of noncoding Xic elements for the initiation of XCI.

In mammals, sex is determined by a pair of unequal X and Y chromosomes, with males carrying one each (XY) and females carrying two X chromosomes (XX). Because the X chromosome is gene-rich while the Y is gene-poor, this scheme results in a dosage imbalance between the two sexes that requires a systematic method of “dosage compensation” known as X chromosome inactivation (XCI) (1–7). XCI transcriptionally silences one X chromosome in female cells. The choice of which X chromosome to inactivate is random, such that the maternal and paternal X’s have a roughly equal chance of being inactivated. Recent work has brought to light a number of epigenetic mechanisms that initiate and establish XCI (8–13). Once inactivated, the same X chromosome is maintained as the inactive X (Xi) through embryonic and postnatal development. While the field is converging on mechanisms by which silencing is propagated, established, and maintained on the Xi, there remains a debate regarding how cells count and choose X chromosomes (2, 6, 14–17). Counting and choosing are key steps that every cell—regardless of whether it is XX or XY—must traverse in order to determine whether XCI will be initiated in the cell. The decision is made in a cell-autonomous fashion.

XCI requires a master locus called the X inactivation center (Xic) that spans several hundred kilobases (Fig. 1A) (18). Early transgenic studies in the mouse system have demonstrated that a region of 100–500 kb may be both necessary and sufficient to drive most if not all steps of XCI, including counting, choice, and the initiation of silencing (19, 20). Most genes within the Xic do not encode proteins, but instead produce long transcripts associated with critical activities of XCI, including Xist, RepA, Tsix, Xite, Jpx, Ftx, and Tsx (14, 21–35). These factors together regulate expression of the key silencer, Xist, whose RNA spreads along the X chromosome to initiate the silencing cascade. Xist is regulated both positively and negatively. In the positive pathway, Jpx RNA acts as a numerator element to control counting by titrating away the architectural factor CTCF, which otherwise sits on the Xist promoter to repress Xist’s transcriptional activation (29, 36). Also proposed to be in the Xist induction pathway are the short RepA RNA (10), and the Ftx locus (31, 37). In the negative pathway are Tsix and Xite, which together control X chromosome pairing and the allelic choice of which X will become the active X (Xa) (24, 27, 38). Studies support the idea that pairing drives formation of epigenetic asymmetry between two initially active X’s and thereby mediates the mutually exclusive choice of Xa and Xi (17, 39–41). In addition to the noncoding regulators, the protein-coding gene, Rnf12/Rlim (42–44), has also been implicated in XCI regulation. The RNF12 enzyme is an E3 ubiquitin ligase that degrades REX1, a transcription factor that stimulates Tsix and represses Xist (43). The study proposed RNF12 as a counting factor whose dose-dependent degradation of REX1 triggers Xist expression.

Fig. 1.

Generation of Xic deletion in ES cell lines. (A) Map showing parts of the Xic and location of the primers used to genotype the ESC clones. (Top) Four deletions created by Barakat et al. (15) and one 209-kb deletion from our study. Scissors indicate ΔXic (Tsx - Ftx) (209 kb) using CRISPR/Cas9. The region around the Xist (red) gene harbors noncoding RNA genes (orange), as well as protein-coding genes (purple or gray). Pairing region shown in blue. Positive (+) or negative (–) effect on Xist/Tsix regulation is indicated. (B) ∆Xic deletions made using CRISPR/Cas9 on the Cast (Top), Mus/129 (Middle), or both (Bottom) alleles. (C, Top) PCR analysis on genomic DNA of the parental cell line WT (16.7) and ΔXic ES clones using primers (P1–P2). (Bottom) PCR products showed that 1A1, 3C9, and 3G10 clones are heterozygous; 1C4 and 4H10 clones are homozygous. Xist primers P3–P4 were used. M, 100-bp markers. (D) Time course analyses of Oc4, Nanog, Xist, and Tsix expression by qRT-PCR in differentiating cells showed appropriate down-regulation of pluripotency markers and Tsix RNA. Averages ± SE from two independent differentiation experiments are plotted. All values are normalized to beta-actin RNA and WT d0 levels are set to 1.0.

Altogether, these events lead to an asymmetry in Tsix—the key to Xist induction. Persistent expression of Tsix on the future Xa then blocks Xist up-regulation in cis, while repression of Tsix enables Xist up-regulation on the future Xi (24). Once expressed, Xist RNA binds to a single nucleation site located within the Xist gene body (45) and spreads in three dimensions (3D) to the rest of the X chromosome to establish chromosome-wide silencing (46–48). Xist RNA serves as platform to recruit a large proteome (49–51) and collaborates with that proteome to form heterochromatin on the X chromosome. Additionally, Xist RNA also evicts transcriptional activators from the X chromosome (51, 52) and reshapes its conformation during its transformation to the Xi (51, 53–56). Thus, the initiation of XCI hinges on a complex interplay among noncoding elements and the linked coding gene, Rnf12/Rlim.

Critically, all of the aforementioned factors reside within the core region of the Xic or are closely linked to it. The Xic is therefore associated with all critical regulatory steps involved in counting and choosing of X chromosomes. One exception is PAR-TERRA RNA, the X-linked form of TERRA that originates in the distal end of the X chromosome, within the pseudoautosomal region (PAR) and extending to the telomeric repeats, (TTAGGG)n (57). PAR-TERRA is implicated in promoting the homology search between two X chromosomes for inter Xic allelic pairing. This is currently the earliest known step of XCI. Interestingly, when the pairing region of the Xic is moved to an autosome, the Xic transgene directs an ectopic pairing with the X chromosome, despite the Xic transgene becoming dissociated from the X-linked telomere (40). The pairing function is likely retained in the transgenic context because PAR-TERRA is diffusible and also binds to autosomal telomeres, where we suggest that it would direct homology searching between the autosomal Xic and the native Xic (57). Other exceptions may include two noncoding regions, Firre (58, 59) and Dxz4 (60, 61). However, while Firre and Dxz4 may influence the 3D structure of the Xi, neither locus is required for initiation or maintenance of XCI in mouse cells (62–65).

In this light, one recent study that deleted a 400- to 600-kb region from Xite to Slc16a2 was especially surprising (Fig. 1A) (15). Despite this large deletion that included established noncoding factors, Xite, Tsix, Jpx, and Xist, female embryonic stem (ES) cells carrying only one Xic (∆Xic/+) reportedly still possessed the ability to register two X chromosomes and undergo XCI, as evidenced by up-regulation of the remaining Xist allele. This outcome argued that essential counting factors do not reside within the noncoding Xic. The study proposed that the linked coding gene, Rnf12/Rlim, which was not deleted (Fig. 1A), is instead the required counting factor (43, 66). While RNF12 protein clearly plays a role in XCI, subsequent studies have shown that it is dispensable for initiation of random XCI in vivo, although it is important for imprinted XCI (42, 44). Notably, imprinted XCI occurs exclusively on the paternal X chromosome in preimplantation mouse embryos and placental tissues, and does not involve a counting mechanism. Furthermore, Rnf12^+/− female ES cells and embryos can still up-regulate Xist expression (15, 44, 66). Therefore, the deletional study (15) is difficult to reconcile with other existing studies, raising the possibility that creation of the large ∆Xic deletion might have perturbed the X chromosome in unanticipated ways and induced a nonphysiological response in ES cells. Because the conclusions have important implications for the current understanding of XCI mechanism, we revisited the question here by creating independent ∆Xic cell lines to test the function of the Xic during XCI initiation.

Results and Discussion

Barakat et al. (15) previously created various deletions from 400 to 600 kb within the Xic spanning loci from Xite to Slc16a2 (Fig. 1A). Here, using CRISPR/Cas9 technology, we constructed a similar deletion that spans Tsx to Ftx (Fig. 1A). We targeted the deletion to the X chromosome in wild-type hybrid female ES cell line, 16.7, which carries X chromosomes of different strain origin, one from M. musculus (129/Sv or “Mus”) and one from M. castaneus (Cast/EiJ or “Cast”) (67). To obtain a comprehensive view, we generated clones where the Xic was deleted on the Mus (∆Xic_mus/+) or Cast (+/∆Xic_cast) or both (∆Xic_mus/∆Xic_cast) alleles (Fig. 1B). Critically, our ∆Xic mutation deleted all noncoding elements involved in the positive and negative regulation of Xist, including Tsx, Xite, Tsix, Xist, Jpx, and Ftx, and spanned ∼209-kb region around the Xist locus. We isolated five independent female clones, verified their genotypes by Sanger sequencing, and took advantage of interspecies single-nucleotide polymorphisms (SNPs) located within the Xist region to determine which X chromosome harbored the deletion (Fig. 1C and SI Appendix, Fig. S1). Genotyping revealed that the Cast allele was targeted in clone 1A1 (+/∆Xic_cast), the Mus allele was targeted in 3C9 and 3G10 (∆Xic_mus/+), and both alleles were targeted in 1C4 and 4H10 (∆Xic_mus/∆Xic_cast).

We differentiated the female ES cells to induce initiation of XCI, a process tightly coupled to cell differentiation (68). Quantitative RT-PCR (RT-qPCR) analysis of Tsix RNA and pluripotency markers, Oct4 and Nanog, showed appropriate down-regulation between days 4 and 7 (d4–d7) (Fig. 1D), indicating that all clones could initiate the cell differentiation pathway. Time course analysis showed that the parental wild-type (WT, 16.7) female cells appropriately up-regulated Xist expression between d4 and d7. RNA fluorescence in situ hybridization (FISH) showed 6.73% ± 0.72% (mean ± SE) of female cells displaying Xist cloud by d4 and 41.51% ± 1.39% by d7 (Fig. 2 A and B). In contrast, the heterozygous deletion clones showed severely compromised Xist up-regulation. RT-qPCR showed a nearly complete absence of Xist RNA between d4 and d7 (Fig. 1D). Consistent with this, RNA-FISH revealed only 0.20% ± 0.14% of mutant cells displaying Xist clouds on d4 and equally poor expression on d7 (1.05 ± 0.23%). This was the case regardless of which X chromosome carried the deletion (∆Xic_mus/+ versus +/∆Xic_cast) and was true of all clones (Figs. 1D and 2 A and B). The severely compromised XCI was similar to that observed in homozygous ∆Xic_mus/∆Xic_cast female clones on d4 (P = 0.00072) and d7 (P = 0.00064), where Xist clouds were not observed at all.

Fig. 2.

ΔXic causes loss of XCI and massive cell death in female ES cells. (A) Xist RNA-FISH to examine the differentiation time course (d0, d4, and d7). Genotypes and clone IDs are indicated. Xist probe, Cy3-labeled. DNA is stained with DAPI (blue). (B) RNA-FISH for Xist quantified at different time points (d0, d4, and d7). Averages ± SE from three independent differentiation experiments are shown. Sample sizes (n): d0, 299–375; d4, 432–559; d7, 327–569 (t test; ns, nonsignificant, *P < 0.05; **P < 0.01; ***P < 0.001). (C) Bright-field photographs of WT, heterozygous, and homozygous ES cells from d0 to d7 of differentiation. The arrows point to disintegrating, necrotic EBs present in heterozygous and homozygous cultures. (D) Cell proliferation analysis during differentiation (d0, d3, d5, and d7) of WT and ΔXic ES lines. Error bars represent SD from two independent replicates (t test; ns, nonsignificant; *P < 0.05; **P < 0.01; ***P < 0.001). (E) Massive cell death caused by deleting the Xic in female ES Cells. The trypan blue staining results of two independent differentiation experiments were averaged and plotted with SE. d0, n = 300–1,000 cells; d4, n = 300–600 cells; d7, n = 500–1,300 cells.

Because dosage compensation is required for proper cell differentiation, the absence of XCI would be expected to compromise formation of embryoid bodies (EBs) in culture. To test this, we followed differentiation of ES cells into EBs over time and examined EB morphology and cell proliferation. WT, ∆Xic_mus/+, +/∆Xic_cast, and ∆Xic_mus/∆Xic_cast clones were indistinguishable on d0 (Fig. 2C), consistent with the lack of need for Xist expression in pre-XCI cells. Differentiation uncovered dramatically different effects, however; WT cells showed the typical EB globularity, with smooth and radiant borders between d2 and d4 when grown in suspension (Fig. 2C). ΔXic heterozygous and homozygous clones exhibited necrotic centers, irregular edges, and disaggregation (Fig. 2C, arrows). All clones showed a similar phenotype. The difference became even more obvious during the adherent phase of growth (d4–d7). Whereas WT EBs adhered to plates and displayed exuberant cellular outgrowth, all ΔXic EBs—both heterozygous and homozygous—attached poorly and showed scant outgrowth, aberrant cellular disaggregation, and death. Cell proliferation assays demonstrated that, whereas undifferentiated cells grew normally, differentiating ∆Xic mutants exhibited significantly reduced proliferation, especially evident between d5 (57.06–74.88% reduction) and d7 (55.89–74.68% reduction) (Fig. 2D). Cell death analysis confirmed a significant increase in cell death among all ∆Xic clones (Fig. 2E). The poor EB growth and massive cell death therefore mirrored the failure of Xist induction over the time course. Thus, deleting one Xic copy in females—either on the Cast or Mus X chromosome—caused severe XCI abnormalities and aberrant cell growth during differentiation. These results sharply contrast with the previous findings of Barakat et al. (15), who reported that none of their ∆Xic/+ female cells were XCI-impaired.

We interpret our findings to mean that two copies of an element(s) within the Xic are required for a cell to register the presence of two X chromosomes. It is well established that counting involves determining the X chromosome number relative to autosomal content, or the so-called X-to-autosome ratio (X:A). In diploid cells, one X chromosome remains active (Xa), and all others are inactivated. In tetraploid cells, two X chromosomes remain as Xa and all others are inactivated. Initially, the counting mechanism was proposed to involve only a “blocking factor” comprising a complex of autosomal factors that binds to one Xic and blocks its function (69–72). The blocking factor has been proposed to bind and transactivate Tsix (38), the antisense repressor of Xist that persists on the future Xa to block the up-regulation of Xist RNA in cis.

The blocking factor does not act alone in the counting process, however, as deleting Tsix in male cells does not cause Xist up-regulation (67, 73). It was therefore proposed that a “competence factor” must also exist (38, 67, 74). This factor comprises a complex of X-linked and autosomal factors that would titrate each other and effectuate the X:A ratio, ensuring that XCI is initiated only when X:A ≥ 1. Jpx RNA (75–77) has been implicated as a numerator of the X:A ratio and as a subunit of the competence factor (14, 29). Indeed, deleting one allele of Jpx renders female ES cells able to up-regulate Xist RNA (29). By contrast, extra copies of Jpx in male ES cells causes Xist to turn on (14). The current findings in our ∆Xic heterozygotes are consistent with Jpx RNA being a numerator and the Tsix/Xite locus harboring binding sites for an autosomal blocking factor.

To pressure test the system, we took the genetic analysis one step further. We reasoned that, given Tsix’s role in silencing Xist, deleting Tsix on the opposite X chromosome should at least partially rescue the ∆Xic phenotype. Here, we took advantage of the existing TST (Tsix^TST/+) cell line, which produces a truncated version of Tsix RNA on Mus allele and thereby skews XCI to occur almost exclusively on the Mus chromosome (67, 78). We targeted the ∼209-kb Xic deletion and created a compound heterozygote, Tsix^TST/∆Xic, in which ∆Xic occurred on the Cast X chromosome and Tsix is truncated on the Mus X chromosome (Fig. 3 A and B). CRISPR-Cas9 targeting yielded multiple Tsix^TST/∆Xic_cast cell lines. The targeting scheme also created ∆Xic_mus/+ and ∆Xic_mus/∆Xic_cast (Fig. 3 A and B). We genotyped five independent female clones by Sanger sequencing of genomic DNA and determined which X chromosome was targeted using allelic SNPs (Fig. 3B and SI Appendix, Fig. S2). The Cast X chromosome was targeted in clones B5 and B9 (Tsix^TST/∆Xic_cast), the Mus X chromosome was targeted in clones G4 and D2 (∆Xic_mus/+), and both X’s were targeted in F11 (∆Xic_mus/∆Xic_cast). None of the five clones showed any growth defects when propagated as undifferentiated ES cells.

Fig. 3.

Generation of Xic deletion in TST cell lines. (A) In TST (Tsix^TST/+) cell lines, the 209-kb ∆Xic deletion was constructed on the Cast, Mus/129, or both alleles. The red bars represent truncation of Tsix in Mus allele. (B, Top) PCR analysis on genomic DNA of the parental Tsix^TST/+ cell line or ΔXic ES derivatives using primers (P1–P2; Fig. 1A). (Bottom) PCR products of Xist showed that B5, B9, G4, and D2 clones are heterozygous; F11 clone is homozygous using primers (P3–P4; Fig. 1A). M, 100-bp markers. (C) Appropriate down-regulation of stem cell markers during differentiation of ∆Xic cell lines in the Tsix^TST/+ background. Time course analyses of Oc4, Nanog, Xist, and Tsix expression by qRT-PCR in differentiating cells. Averages ± SE from two independent differentiation experiments are plotted. All values are normalized to beta-actin RNA and WT d0 levels are set to 1.0.

We performed a time course analysis to examine cell differentiation and XCI. RT-qPCR of Tsix RNA, Oct4, and Nanog showed appropriate down-regulation between d4 and d7 (Fig. 3C), indicating that the cell differentiation pathway per se was not compromised in any of the clones. In the parental Tsix^TST/+ cells (“TST”), Xist induction occurred between d4 and d7, as expected (Fig. 3C). Formation of an RNA cloud was observed in the same time frame, with 45.11 ± 2.06% (mean ± SE) of cells displaying Xist clouds at d4 and 85.40% ± 0.27% at d7. The ∆Xic_mus/∆Xic_cast showed no Xist up-regulation at all, in keeping with the homozygous deletion (Fig. 4 A and B). By contrast, the ∆Xic_mus/+ clones in the Tsix^TST/+ background showed severely compromised Xist expression—similar to the ∆Xic_mus/+ clones in the 16.7 background above. RNA-FISH demonstrated that only 9.64% ± 0.60% displayed Xist clouds on d4 and 16.09 ± 0.98% on d7 for ∆Xic_mus/+ clones created in the Tsix^TST/+ background. This was significantly different from the 45–85% positivity for the parental cell line (d4, P = 0.00009; d7, P = 0.000004).

Fig. 4.

ΔXic of Mus allele causes loss of XCI and massive cell death in TST cell lines. (A) Xist RNA-FISH to examine the differentiation time course (d0, d4, and d7) in indicated clones of the Tsix^TST/+ background. Xist probe, Cy3-labeled. DAPI is blue. (B) Quantitation of Xist RNA-FISH for A. Averages ± SE from three independent differentiation experiments are shown. Sample sizes (n): d0, 393–688; d4, 289–493; d7, 568–731 (t test; ns, nonsignificant, *P < 0.05; **P < 0.01; ***P < 0.001). (C) Bright-field images show poor EB formation and outgrowth in G4, D2, and F11 clones but not TST, B5, and B9 clones. The arrows point to disintegrating, necrotic EBs present in G4, D2, and F11 cultures. (D) Cell proliferation analysis during differentiation of ES lines of indicated genotype and clone number. Error bars represent SD from two independent replicates (t test; ns, nonsignificant; *P < 0.05; **P < 0.01; ***P < 0.001). (E) Massive cell death caused by deleting the Xic in the Tsix^TST/+ background. The trypan blue staining results of two independent differentiation experiments were averaged and plotted with SE. d0, n = 300–1,000 cells; d4, n = 300–1,000 cells; d7, n = 400–1,000 cells.

Intriguingly, analysis of the reciprocal heterozygous cell lines carrying the Tsix mutation (Tsix^TST/∆Xic_cast) showed full rescue of Xist expression levels (Fig. 3C, clones B5 and B9). Likewise, RNA-FISH showed full restoration of Xist clouds, with 48.58 ± 1.39% at d4 and 88.5 ± 1.33% at d7—levels similar to those in the parental cell line (Fig. 4 A and B). Consistent with these findings, viability, EB formation, and cell proliferation were severely compromised in all mutant clones, except in the two Tsix^TST/∆Xic_cast clones, B5 and B9 (Fig. 4 C and D). Cell death analysis confirmed a significant increase in cell death in all ∆Xic clones except for B5 and B9 (Fig. 4E). Notably, the reciprocal heterozygotes ∆Xic_mus/+ (G4, D2) also showed impaired EB formation, viability, and proliferation over the 7-d time course. Thus, eliminating Tsix, the antisense repressor of Xist, rescues the absence of one Xic in the female cell. We conclude that the Tsix mutation suppresses the Xic deletion in trans.

Given these outcomes, our current work indicates that Rnf12 is not sufficient to induce Xist expression by itself without the presence of two Xic’s. ∆Xic_mus/+ and +/∆Xic_cast heterozygotes with two fully operational Rnf12 alleles are substantially compromised in their ability to up-regulate Xist. How Barakat et al. (15) came to different conclusions may partly lie in one observation made within the cell lines we created in lieu. Interestingly, while the ∆Xic_mus/+ clones made in the Tsix^TST/+ background (G4, D2) showed severe XCI impairment, we noticed that 9 to 16% of cells between d4 and d7 did show some degree of Xist up-regulation (Fig. 4 A and B). This finding contrasted with the reciprocal heterozygous ∆Xic cell lines in which the Cast Xic was deleted. Because the Cast X chromosome is of paternal origin in this hybrid ES cell line (24), we are reminded of an earlier study suggesting that the epiblast merely overwrites an imprint that is not fully erased in the cells that undergo random XCI (79). Using an induced trophoblast stem cell (iTS) model, the study showed that ES cells transdifferentiated into trophoblasts retained partial memory of the XCI imprint and inactivated Xp in a subpopulation of the trophoblast cells (79), a cell type that would ordinarily manifest imprinted XCI in the mouse placenta. Imprinted XCI would bypass the requirement for counting and choice. In the mouse, epiblast cells switch from imprinted to random XCI, but the underlying mechanisms are still not understood, particularly whether the switch necessitates erasure of the original paternal Xist imprint or, alternatively, whether vestiges of the imprint remain in epiblast-derived cells (but are not acted upon). Because ES cells are derived from the epiblast, we propose that the ability of some ∆Xic_mus/+ cells—for example, in G4 and D2 clones—to up-regulate Xist harks back to the paternal imprint that was not erased. Clearly, this is a low-frequency event and not all cells and clones retain this putative imprint. Notably, ∆Xic_mus/+ clones made in the 16.7 background (3C9, 3G10) did not show this type of Xist up-regulation (Fig. 2 A and B). It is possible that independently derived ES clones could carry different epigenetic profiles, rendering some more able to interpret the imprint than others. Thus, Barakat et al. (15) could have relied on ES clones with a more robustly retained paternal Xist imprint, which would have led to Xist up-regulation despite the loss of one Xic and the counting elements that reside in it. In the future, an epigenomic comparison of various ∆Xic clones could elucidate differences at the level of DNA methylation and/or histone modifications, potentially helping to pinpoint putative imprints. Altogether, our current findings reinforce the concept of a single inactivation center and the previous conclusion that required elements for counting and choosing, as well as silencing, reside within the noncoding region spanning the locus, Tsx-Xite-Tsix-Xist-Jpx-Ftx.

Methods

ES Culture and Differentiation.

Methods have been described elsewhere (55). Briefly: ES cells were grown on 0.2% gelatin-coated plates in serum-containing N2B27 media (mixture of 20 mL of FBS [Sigma], 500 mL of DMEM/F12 media [11330032; Thermo Fisher Scientific], 500 mL of Neurobasal media [21103-049; Thermo Fisher Scientific], 5 mL of N2 supplement [17502048; Thermo Fisher Scientific], 10 mL of B27 supplement [17504044; Thermo Fisher Scientific], 10 mL of Pen/Strep [15140163; Thermo Fisher Scientific], 5 mL of GlutaMAX [35050061; Thermo Fisher Scientific], and 2 mL of 55 mM β-mercaptoethanol [21985023; Thermo Fisher Scientific], 1,000 U/mL leukemia inhibitory factor [LIF] [Millipore], supplemented with 2i [3 μM Gsk3 inhibitor CHIR99021; 1 μM MEK inhibitor PD0325901; Axon]).

To induce differentiation, ES cells grown in 2i condition were dissociated by Accutase (07920; StemCell Technologies) and 1 million ES cells were grown in suspension in a 10-cm low-attachment plate with 10 mL of differentiation media (DMEM, high glucose, GlutaMAX supplement, pyruvate [10569044; Thermo Fisher Scientific], 15% HyClone FBS, 25 mM Hepes, pH 7.2–7.5 [15630130; Thermo Fisher Scientific], 1× MEM nonessential amino acids [11140076; Thermo Fisher Scientific], 1× Pen/Strep [15140163; Thermo Fisher Scientific], 0.1 mM β-mercaptoethanol [21985023; Thermo Fisher Scientific]) to allow EB formation. Media were changed daily and 25% EBs were settled on 10-cm tissue culture plate coated with 0.2% gelatin on the fourth day of differentiation (d4). To harvest ES cells at different stages of differentiation, cell were treated with either Accutase for 5 min (d0 ES cells and d4 EBs) or 0.25% trypsin (25200056; Thermo Fisher Scientific) for 10 min (d7 EBs) at 37 °C.

Cell Lines.

Female (16.7) ES cell lines have been described (67). TST (Tsix^TST/+) female mouse ES cell lines have been described (78)

Generation of ΔXic Mutants in Wild-Type Female ES Cell Background Using CRISPR-Cas9.

+/∆Xic_cast, ∆Xic_mus/+, and ∆Xic_mus/∆Xic_cast ES cell lines were generated by deleting a 209-kb region around the Xist locus. To generate these cell lines, a total of four guides flanking the region to be deleted (two on each side) were cloned into the pSpCas9(BB)-2A-Puro vectors (80). To create Xic deletion cell lines, we used two pairs of guide RNAs to ensure proper cutting: Two pairs target the same region of Tsx, and two other pairs target the same region of Ftx. For Tsx (pair 1: CAC​CGG​GGT​ATC​AGC​TCC​ACC-​AAC​A, AAA​CTG​TTG​GTG​GAG​CTG​ATA​CCC​C; pair 2: CAC​CGA​GCT​GAT​ACC​CAA-​GTC​TAG​G, AAA​CCC​TAG​ACT​TGG​GTA​TCA​GCT​C). For Ftx, the two different pairs of guide RNAs were (pair 3: CAC​CGA​GGA​GGT​AAA​GGA​ACG​ACG​A, AAA-​CTC​GTC​GTT​CCT​TTA​CCT​CCT​C; pair 4: CAC​CGT​ACG​AAA​CCT​GCT​ATA​CTT​G, AAA​CCA​AGT​ATA​GCA​GGT​TTC​GTA​C). In total, four gRNA/Cas9 plasmids were delivered into ES cells by Lipofectamine LTX and Plus Reagent (15338100; Thermo Fisher Scientific). Starting 24 h after transfection, cells were seeded into 10-cm plates and cultured for 6 d in 2i medium with puromycin (2 μg/mL). Single colonies were picked, cultured in 96-well plates by expanded, and screened by genomic PCR, Sanger sequencing for clones with Xic deletion.

Generation of ΔXic Mutants in Tsix^TST/+ Background Using CRISPR-Cas9.

Tsix^TST/∆Xic_cast, ∆Xic_mus/+, and ∆Xic_mus/∆Xic_cast cell lines were generated by transfecting TST cells with a mixture of four plasmids (per above) and pST1374-NLS-flag-linker-Cas9 plasmid. Twenty-four hours later, cells were seeded into 10-cm plates and cultured for 6 d in 2i medium with blasticidin (8 μg/mL). Single colonies were picked, cultured in 96-well plates by expanded, and screened by genomic PCR, Sanger sequencing for clones with Xic deletion.

RNA-FISH.

RNA-FISH protocols and probes was performed as previously described (55) and briefly summarized here: 1 × 10⁵ mouse ES cells were affixed by cytospinning onto glass slides at 1,000 rpm for 10 min. After air-drying the slides, cells were rinsed with chilled PBS for 5 min, preextracted with CSKT for 10 min on ice, followed by fixation with 4% paraformaldehyde in PBS at room temperature for 10 min, and then dehydrated in a series of increasing ethanol concentrations and air-dried. For RNA-FISH, cells were incubated with 40-ng labeled probes at 37 °C for >3 h or overnight. After being washed once in 25% formamide/2× SSC at 37 °C for 20 min and three times in 2× SSC at 37 °C for 5 min each, cells were mounted for wide-field fluorescent imaging. Nuclei were counterstained with Vectashield mounting media with DAPI (Fisher Scientific).

Microscopy.

For wide-field fluorescent imaging, cells were observed on a Nikon 90i microscope equipped with 60×/1.4 numerical aperture (N.A.) or 100×/1.4 N.A. VC objective lens, Orca ER CCD camera (Hamamatsu), and Volocity software (Perkin-Elmer).

Cell Proliferation Assays.

For cell proliferation analysis (15), equal amounts of cells (2 × 10⁵) were allowed to differentiate on gelatinized culture dishes. Cells were washed and trypsinized, and viable cells were counted at the different time points indicated. All measurements and countings were performed in duplicate on two independent differentiations.

Cell Death Analysis.

Cell death assays were performed as described (29, 81) and summarized here: On d0, both supernatant and attached ES cells were collected and stained with trypan blue dye (15250061; Thermo Fisher Scientific). On other time points, both supernatant and floating EBs (EBs on d4) or attached EBs (d7 and onward) were collected and stained with trypan blue. The ratios of dead cells (blue) to total cells (i.e., blue dead cells plus clear viable cells) were calculated and plotted as a function of time. Each sample was counted in duplicate.

Quantitative RT-PCR.

Real-time PCR for Xist, Tsix, Oct4, Nanog, and Beta-actin was performed under the following conditions: 95 °C, 3 min; 95 °C, 10 s, 60 °C, 20 s, and 72 °C, 20 s, for a total of 40 cycles, and 72 °C, 5 min. Melting curves for primer pairs were determined by increasing temperatures from 60 to 95 °C at 0.5 °C interval for 5 s. Primers for Xist qRT-PCR were NS33 and NS66, and for Tsix were NS18 and NS19 (81); for Oct4 were octF and octR; for Nanog were nanogF and nanogR; and for Beta-actin were actin1 and actin2 (79).

Data Availability

All study data are included in the article and/or supporting information.