|Home | About | Journals | Submit | Contact Us | Français|
X chromosome inactivation (XCI) is a dynamically-regulated developmental process with inactivation and reactivation accompanying the loss and gain of pluripotency, respectively. A functional relationship between pluripotency and lack of XCI has been suggested, whereby pluripotency transcription factors repress the master regulator of XCI, the noncoding transcript Xist, by binding to its first intron (intron1). To test this model, we have generated intron1-mutant embryonic stem cells (ESCs) and two independent mouse models. We found that Xist’s repression in ESCs, its transcriptional upregulation upon differentiation, and its silencing upon reprogramming to pluripotency are not dependent on intron1. Although we observed subtle effects of intron1-deletion on the randomness of XCI and in the absence of the antisense transcript Tsix in differentiating ESCs, these have little relevance in vivo as mutant mice do not deviate from Mendelian ratios of allele transmission. Together, our findings demonstrate that intron1 is dispensable for the developmental dynamism of Xist expression.
To balance the expression of X-linked genes between males and females, female mammals silence one of the two X chromosomes in a developmentally regulated process called X chromosome inactivation (XCI). XCI occurs in two waves in the course of mouse embryogenesis. The earliest form of XCI is imprinted as it is selective for the paternally-inherited X chromosome (Xp) and starts at the 2–4 cell stage in pre-implantation embryo (Huynh and Lee, 2003; Kalantry et al., 2009; Namekawa et al., 2010; Patrat et al., 2009). At the pre-implantation blastocyst stage, imprinted XCI is retained in the trophectoderm and primitive endoderm lineages, but reversed in arising pluripotent epiblast cells yielding a state with two active X chromosomes (XaXa) (Mak et al., 2004; Okamoto et al., 2004; Silva et al., 2009; Williams et al., 2011). Upon implantation, these epiblast cells establish a random form of XCI that stochastically initiates on the maternal or paternal X chromosome and is retained through the lifetime of mitotic divisions (Kay et al., 1993; Rastan and Robertson, 1985). Similarly, mouse embryonic stem cells (ESCs), which are derived from epiblast cells of the pre-implantation blastocyst, undergo random XCI when induced to differentiate ex vivo. The only exception to somatic maintenance of random XCI is inactive X (Xi) reactivation in the germline, which is assumed to be essential for female fertility and occurs in primordial germ cells as they traverse the hindgut to seed the genital ridges (Chuva de Sousa Lopes et al., 2008; de Napoles et al., 2007; Sugimoto and Abe, 2007). Xi reactivation is also a feature of experimentally induced acquisition of pluripotency via transcription factor-mediated reprogramming to induced pluripotent stem cells (iPSCs), fusion of somatic cells with ESCs, or somatic cell nuclear transfer (Eggan et al., 2000; Maherali et al., 2007; Tada et al., 2001).
The cycles of X chromosome inactivation and reactivation are associated with changes in Xist RNA coating, where cells with a Xi display coating by the non-coding Xist RNA on the inactive X chromosome, and those with two active X chromosomes lack Xist RNA expression (Brockdorff et al., 1991; Brown et al., 1991). Xist’s function has been most studied in the random form of XCI in the mouse system, where it is shown to be the critical trigger of XCI. The upregulation of Xist RNA and coating of the X at the onset of random XCI immediately lead to transcriptional silencing of X-linked genes and result in the exclusion of RNA polymerase II and the recruitment of repressive chromatin-modifying protein complexes such as the Polycomb complex PRC2 which establishes an accumulation of H3K27me3 (Chaumeil et al., 2006; Chow and Heard, 2009; Plath et al., 2003; Silva et al., 2003). A stereotypic order of changes in chromatin structure culminates in heritable silencing of either the maternally or paternally transmitted X chromosome in each cell of the female adult mammal. Xist is essential for XCI to occur in cis as its deletion leads to silencing of the other X chromosome carrying an intact Xist allele, regardless of parent-of-origin (Marahrens et al., 1997; Penny et al., 1996). Moreover, the importance of Xist regulation for the developmental and sex-specific context of XCI is demonstrated by its sufficiency: overexpression of a X-linked Xist cDNA transgene in male mESCs (XY:tetOP-Xist) initiates XCI and cell death due to silencing of the single X chromosome (Wutz and Jaenisch, 2000).
Xist is transcribed from a larger locus on the X chromosome that has been defined as the minimal critical region for XCI and besides housing Xist, contains other protein-coding and noncoding activators and repressors of Xist, some of which act in cis and others in trans (Rastan and Robertson, 1985; reviewed in Minkovsky et al., 2012). The best characterized repressor of Xist is its antisense transcript, Tsix, that is highly transcribed in epiblast cells of the pre-implantation blastocyst and in undifferentiated mouse ESCs/iPSCs, where Xist is repressed (Lee et al., 1999; Sado et al., 2001; Maherali et al., 2007). Deletion of Tsix leads to only slight Xist upregulation without causing precocious XCI or Xist RNA coating in self-renewing, undifferentiated ESCs. However, upon differentiation, XCI is skewed to the Tsix-deleted X in female cells heterozygous for the mutant Tsix allele (Lee et al., 1999; Lee, 2000; Luikenhuis et al., 2001; Sado et al., 2001). The effect of Tsix deletion on Xist indicates that it participates in parallel pathways with other regulators of Xist repression or activation.
Interestingly, the pluripotency factors Oct4, Sox2, and Nanog, have been implicated in the control of Xist expression in pluripotent cells. Navarro and colleagues found that in ESCs, Oct4, Sox2, and Nanog bind the first intron of the Xist gene (intron1) (Navarro et al., 2008), a finding that has been recapitulated in many genomic datasets and extends to additional pluripotency regulators such as Tcf3 and Prdm14, and early developmental regulators such as Cdx2 (Fig S1A, Loh et al., 2006; Marson et al., 2008; Ma et al., 2011; Erwin et al., 2012). Such genomic regions of extensive pluripotency transcription factor co-occupancy in the ESC genome occur more commonly than would be expected by chance (Chen et al., 2008). It is thought that these co-bound genomic regions represent functionally important sites and often represent enhancer elements (Chen et al., 2008). Further support for a gene regulatory role of intron1 is that, in ESCs, the intron1 region has a propensity to be in the three-dimensional proximity to the promoter of Xist and adopts a DNAse hypersensitive state (Tsai et al., 2008). Additionally, pluripotency factors appear directly linked to Xist regulation. Upon Nanog deletion or inducible repression of Oct4, Xist is upregulated and binding of the pluripotency factors to intron1 is lost (Navarro et al., 2008). In males ESCs, which normally do not upregulate Xist, experimentally forced Oct4 repression can even induce Xist RNA coating in up to 10% of the cells (Navarro et al., 2008). Another study could not replicate Xist RNA coating upon Oct4 knockdown in male ESCs, but observed biallelic XCI in differentiating female ESCs upon Oct4 knockdown (Donohoe et al., 2009). A role for Nanog in Xist suppression is also supported by its expression pattern with regard to domains of Xi reactivation in the pre-implantation blastocyst, where the restriction of Nanog expression demarcates the fraction of cells undergoing reactivation of the imprinted Xi (Silva et al., 2009). Furthermore, pre-implantation embryos lacking Nanog are unable to specify epiblast cells and to lose Xist RNA, whereas forced expression of Nanog induces a more rapid loss of Xist RNA coating in developing pre-implantation embryos (Silva et al., 2009; Williams et al., 2011).
Together these findings led to the model that pluripotency factor binding to intron1 is critical for repression of Xist in undifferentiated XaXa ESCs. However, in the experiments leading to this conclusion, cell identity and therefore likely the expression of many genes were modulated by experimental changes in pluripotency factor expression, which could confound the interpretation that Oct4, Nanog and other pluripotency factors act directly on intron1 of Xist to regulate XCI. It has also been suggested that the pluripotency transcription factors control the levels of positive and negative regulators of Xist, as they are binding to Tsix and the trans-acting activator of XCI, Rnf12 (Donohoe et al., 2009; Gontan et al., 2012; Navarro et al., 2010, 2011) Accordingly, an experiment directly addressing the functional importance of binding to intron1 showed only subtle dysregulation of XCI: in female ESCs carrying a heterozygous deletion of intron1 of Xist, XCI remained suppressed in the undifferentiated state. However, upon differentiation, Xist appeared more highly expressed from the chromosome carrying the mutation supporting a role for intron1 in suppressing Xist during differentiation (Barakat et al., 2011). Furthermore, deletion of intron1 in the context of a transgene carrying the extended Xist locus moderately increased expression of Xist in undifferentiated ESCs, which was amplified by simultaneous deletion of the antisense transcript Tsix (Nesterova et al., 2011). Notably, these results were variable between clones potentially reflecting the effect of transgene copy number and variations (Nesterova et al., 2011). Binding to Xist intron1 has also been proposed to govern the switch from imprinted to random XCI in pre-implantation development (Erwin et al., 2012). In vitro, gel shift assays suggest that the binding events between Xist’s intron1 and the pluripotency regulator Oct4 and the trophectoderm regulator Cdx2 are direct but mutually exclusive (Erwin et al., 2012).
Collectively, these findings motivated us to examine the role of Xist intron1 further to test the model wherein pluripotency factor binding silences Xist to prevent XCI in pluripotent cells, and to determine the role of the intronic region in X chromosome reactivation events, both in vivo and in vitro.
To further define the role of Xist intron1, we used gene targeting to generate a conditional allele in male and female mouse ESCs. We tested the requirement of intron1 in both sexes since male ESCs are able to undergo XCI upon forced expression of Xist, providing a sensitive background for monitoring Xist regulation independently of other X chromosomes present in a cell (Wutz and Jaenisch, 2000). By contrast, heterozygous female ESCs permit investigation of kinetics of XCI upon induction of differentiation and insight into potential effects on skewing of XCI between the targeted and wildtype chromosome.
To delineate the region of intron1 involved in Xist repression, we inspected where pluripotency transcription factors bind within the intron1 region as detected by published ChIP-seq data sets (Marson et al., 2008). We also determined the localization of pluripotency factor DNA binding motifs, and considered sequence conservation across mammals (Fig S1). We found that co-occupancy of pluripotency factors occurs in a 600bp region within the full 2.8kb sequence of intron1. Most of the intron1 sequence is not conserved in placental mammals, however, two highly conserved composite Oct4-Sox2 DNA binding motifs, which are found to stabilize a ternary Oct4-Sox2-DNA complex in the expression of many ESC-specific genes, underlie the ChIP-Seq binding peaks of Oct4 and Sox2 (Fig S1, Reményi et al., 2003; Marson et al., 2008; Mason et al., 2010, UCSC phastCons). Based on these data, we decided to delete 800 bp of intron1, and subsequently refer to this mutation as ‘intron1’ (Minkovsky/Plath allele, Fig S1).
We flanked the 800 bp intron1 region with loxP sites, simultaneously inserting a hygromycin resistance cassette (yielding a targeted allele with 3loxP sites), and subsequently generated experimental (1lox) and control (2lox) alleles by transient expression of Cre recombinase in hemizygously targeted male and heterozygous female ESCs (Fig 1, Fig S2). To be able to monitor the effects of the deletion of intron1 on Xist in cis in female cells, we employed genetically polymorphic F1 2–1 female ESCs (129/Cas) carrying a MS2 RNA tag in exon 7 of Xist on the 129 allele (Jonkers et al., 2008). Southern blotting and PCR analysis confirmed that intron1 was targeted in cis to the MS2 RNA tag in females ESCs (Fig S2). Male and female targeted ESC lines showed normal chromosome complement upon karyotyping (Fig S2 and data not shown).
To confirm that deletion of 800 nucleotides from intron1 sufficiently removes pluripotency factor binding, we performed chromatin immunoprecipitation against Oct4 and Sox2 coupled to quantitative PCR for the targeted region of intron 1, neighboring intronic regions, the Xist promoter, and previously validated control regions (Navarro et al., 2008). Importantly, we did not observe an increase in Oct4 or Sox2 binding in these regions upon deletion of intron1 (Fig 1G–I). Thus, compensatory binding at cryptic binding sites upon intron1 deletion appears unlikely.
To understand the role of intron1 in the regulation of XCI, we first performed fluorescence in situ hybridization (FISH) to analyze the expression and localization of Xist and Tsix RNA at the single cell level using strand-specific RNA probes. Undifferentiated male and female ESC lines displayed no significant Xist RNA cloud or pinpoint signal in the presence or absence of intron1 (Fig 2A/B). The absence of Xist RNA coating in the undifferentiated ESC state was confirmed by the lack of a Xi-like enrichment of H3K27me3, which is known to occur on the Xi when Xist RNA coats (Plath et al., 2003; Silva et al., 2003) in Nanog-positive cells (Fig S3A/B). In agreement with this finding, the signal for Tsix was present in the majority of cells in each case and indistinguishable among all tested genotypes (Fig 2A).
Upon induction of differentiation by embryoid body (EB) formation, the lack of intron1 did not induce Xist RNA in male ESCs to a level detectable by FISH (data not shown), and yielded no Xi-like enrichment of H3K27me3 (Fig S3C/D), indicating that intron1 is not an essential regulator of Xist suppression in differentiating male ESCs when all other regulators of XCI are intact. Heterozygous 1lox/wt female ESCs formed Xist RNA clouds and H3K27me3 Xi foci at comparable rates to 2lox/wt control ESCs (Fig 2C/D, Fig S3C/D). Xist RNA levels were also similar between undifferentiated and differentiating male and female ESCs, with or without intron1, in RT-PCR experiments (Fig 2E). Proper differentiation was confirmed by decrease in Nanog transcript levels (Fig 2F). Furthermore, the use of Xist intron1-spanning PCR primer pairs ruled out dramatic secondary effects of intron1 deletion on Xist splicing (data not shown).
Next, we assessed whether XCI is skewed upon intron1 deletion in differentiating female ESCs. The polymorphic 129/cas F1 2–1 female ESC line is known to have a baseline skewing of XCI towards the 129 allele such that approximately 70% of the cells will silence the 129 allele, due to strain-specific haplotypes (Cattanach and Isaacson, 1967). Due to the integration of the MS2 RNA tag on the intron targeted 129 X chromosome, combined RNA-FISH for MS2 and Xist sequences can distinguish between Xist being expressed from the targeted chromosome (positive for both Xist and MS2 signals) and the untargeted X (only marked by the Xist probe) (Fig 2C, Jonkers et al., 2008). We found that, at the single cell level, female 1lox intron/wt ESCs consistently had ~15% more cells expressing the MS2-tagged Xist than their 2lox/wt counterparts, in three of four ex vivo differentiation methods (Fig 2G, Fig S4). This mild skewing effect in differentiating female ESCs is consistent with published results (Barakat et al., 2011).
Next, we investigated the possibility that the intron1-dependent skewing of XCI in differentiating female ESCs represents a mild effect on the intron1-deleted X chromosome at the transition to the differentiated state. We reasoned that such an effect may be more strongly revealed in the absence of other regulators of Xist and sought to assay such an effect on a ‘sensitized’ background for Xist transcription. Tsix represents the prime candidate for a redundant Xist repressor that could compensate to repress Xist in the absence of intron1. One study supports the view that a functional role for the intron can be uncovered in the absence of Tsix, as male ESCs with randomly integrated genomic Xist transgenes lacking intron1 and a functional Tsix allele dysregulated the expression of the transgenic Xist (Nesterova et al., 2011). We therefore performed the above analyses in male ESCs lacking intron1 in the endogenous Xist allele on the background of a previously characterized Tsix loss of function mutation at the endogenous locus (Fig 3, Lee et al., 1999; Luikenhuis et al., 2001; Sado et al., 2001). We targeted the disruption of Tsix to both 2lox and 1lox intron male ESCs using a construct from the Sado lab that inserts a splice acceptor-IRESβGeo cassette in exon 2 of Tsix resulting in an early transcriptional stop (Fig 3A). Correct targeting and loss of the Tsix transcript were confirmed by Southern blotting and absence of the FISH signal for Tsix after targeting (Fig 3B/C).
As expected, in the presence of intron1 (2lox intron1), Tsix deletion in male ESCs induced a mild transcriptional upregulation of Xist RNA compared to XY:2lox/Tsix-wt ESCs in RT-PCR experiments, reaching a level found in female ESCs (Fig 3D). Upon differentiation XY:2lox intron/Tsix-Stop ESCs further upregulated Xist transcript levels ~5 fold (Fig 3D). However, this induction was rarely correlated with an Xist RNA cloud signal detectable by RNA FISH in or an Xi-like H3K27me3 accumulation (Fig 3E-H) before and after induction of differentiation, in agreement with previous reports (Luikenhuis et al., 2001; Sado et al., 2002). Combined deletion of intron1 and Tsix did not alter the Xist status in undifferentiated ESCs, but upon induction of differentiation resulted in a Xist RNA cloud-like signal in FISH experiments in 3–6% of cells compared to 0.2–0.8% in differentiating XY:2lox intron/Tsix-Stop cells (Fig 3E/G). We did not, however, see any significant intron1-dependent effect on Xist RNA levels by RT-PCR comparing XY:2lox intron/Tsix-Stop and XY:1lox intron/Tsix-Stop cells (Fig 3D) or an increase in the number of H3K27me3 Xi-like accumulations (Fig 3F/H). Thus, even though Xist RNA was induced in a slightly larger proportion of differentiating cells in the absence of both Tsix and intron1 than in the absence of either Tsix or intron1, this upregulation does not appear to be sufficient to mediate H3K27me3 enrichment on the targeted X chromosome, suggesting that the RNA does not efficiently coat the chromosome in these cells or that the recruitment of Polycomb proteins is affected. We conclude that these experiments reveal a subtle role of intron1 in the control in Xist expression, which may be related to the weak skewing phenotype of XCI described above for differentiating intron1-mutant heterozygous female ESCs (Fig 2, Fig S4).
In a second assay, we tested the consequence of intron1 deletion upon modulation of global Oct4 transcript levels. We first confirmed the previously reported relationship between the decrease of Oct4 levels and Xist RNA induction (Navarro et al., 2008; Donohoe et al., 2009). Specifically, upon Oct4 depletion in the male ZHBTc4 ESC line, in which Oct4 expression can be silenced acutely by the addition of doxycycline, Xist RNA levels were induced almost 100-fold 96 hours post induction of Oct4 repression (Fig S5A), and Xist RNA could be detected by FISH in a small number of cells (Fig S5 B/C). Notably, we observed that Oct4 transcript levels drop with faster kinetics than Xist RNA levels increase, suggesting that the effect of Oct4 on intron1 is indirect and may require efficient differentiation, which occurs at 96 hours post Oct4 repression as indicated by the loss of the pluripotency factor Nanog (Fig S5D). In agreement with this conclusion, siRNA-mediated knockdown of Oct4 in ESCs did not increase Xist RNA levels more than two-fold after 72 hours confirming a previous report (Fig S5E, Donohoe et al., 2009). Furthermore, the absence of intron1 did not significantly alter Xist RNA levels in female ESCs or male ESCs lacking Tsix in Oct4 knockdown conditions (Fig S5E). These data indicate that the slight increase in Xist levels immediately upon Oct4 depletion are independent of intron1.
The model of pluripotency-factor binding to intron1 to repress Xist motivated us to directly assess whether intron1 behaved as a silencer in ESCs in a reporter assay. We transfected constructs with intron1 or control sequences upstream of a minimal promoter driving luciferase and did not see intron1-dependent decreases in reporter activity (data not shown). The small effect of intron1 deletion on Xist RNA levels detected in differentiating ESCs in the absence of Tsix motivated us to revisit these experiments and instead investigate whether Xist intron1 represents a developmentally-regulated enhancer that becomes active upon induction of differentiation. We therefore tested transactivation activity of intron1 in undifferentiated and differentiating ESCs using stably integrated luciferase reporter constructs (Fig 4). Male ESCs were electroporated with hygromycin resistance-bearing constructs containing either the part of intron1 that we deleted in our experimental cell lines, or two control sequences representing the upstream and downstream flanking regions of the intron1 region (Fig 4A). The experimental intron1 region (B in Fig 4B) was cloned in triple copy to amplify any putative enhancer activity of this region. Pooled clones were subjected to monolayer differentiation by LIF withdrawal with and without retinoic acid treatment. Only cells bearing the intron1 construct covering the pluripotency factor binding site showed a robust increase in luciferase activity upon differentiation (Fig 4B). In agreement with the notion that intron1 does not act as an active enhancer in undifferentiated ESCs, we did not find a histone acetylation mark characteristic of active enhancers, namely H3K27ac, examining our own and published Chip-Seq data sets from ESCs, despite binding of intron1 by a battery of pluripotency factors and p300 in undifferentiated ESCs (mouse ENCODE, Creyghton et al., 2010, data not shown).
We also considered recently published spatial organization data which demonstrated that the Xist gene lies in a topologically associating domain (TAD) with genes encoding the non-coding RNAs Ftx and Jpx/Enox, and the protein-coding genes Rnf12/Rlim, Zcchc13, and Slc16a2 (Nora et al., 2012). It has been proposed that promoters and enhancers predominantly interact (loop) within TADs (Dixon et al., 2012; Nora et al., 2012). Notably, significant intra-TAD contacts originating from within intron1 of Xist, indicative of putative enhancer/promoter looping, were only found in differentiated and not in undifferentiated ESCs (Nora et al., 2012) (Fig S6A), consistent with our finding of reporter activity upon differentiation. However, similar to our result that Xist levels in female and males ESCs did not significantly change in the absence of intron1, we also did not see intron1-dependent transcriptional differences in the three genes that come in contact with intron1 within the Xist-containing TAD, before and during differentiation (Fig S6B). Thus, even though intron1 is pluripotency factor- bound in ESCs, it may only gain significant enhancer activity upon differentiation though still not to an extent where deletion affects transcription of Xist or of neighboring protein-coding genes.
Together these ex vivo studies in undifferentiated and differentiating male and female ESCs point to a minor role for intron1 in the regulation of Xist expression, uncovered only when another Xist repressor is deleted, and some aspect of X chromosome choice (potentially also through slight modulation of Xist RNA levels). These data do not support intron1 as a main aspect of the mechanism of transcriptional repression of Xist, at least in this tissue culture model.
Next, we assayed the significance of intron1 in vivo. Our male ESCs deleted for intron1 (1lox) were injected into C57BL/6 blastocysts. Chimeras were obtained at high efficiency and bred with C57BL/6 females to obtain germline transmission of the mutant allele. Importantly, the 1lox intron1 allele showed normal propagation through the maternal or paternal germline and mice completely lacking intron1 (crossing 1lox/1lox females with 1lox males) could be efficiently bred without any female-specific defect (Fig 5A). Since X chromosome reactivation occurs in the female germline and is likely essential for female fertility, we assessed litter size of the F2 generation of female homozygous knockout mice, finding their litter sizes unaffected (data not shown).
To strengthen these observations of normal transmission of the intron1 mutation and rule out that genetic background obscured a potential intron1 phenotype in vivo, we generated a second mouse model carrying an independent intron1 mutation. We generated mice using previously published females 129/cas F1 female ESCs in which a larger (1.815 kb) region was deleted on the 129 X chromosome (Barakat/Gribnau allele) (Fig S1A, Barakat et al., 2011). We previously observed a slight upregulation of Xist RNA levels on the deleted chromosome in differentiating female ESCs, in agreement with our results indicating skewing of X-inactivation towards the deleted chromosome. Importantly, this second mouse model also displayed normal mendelian transmission of the intron1 lox allele (Fig 5B).
To assay whether random XCI has occurred in female mice carrying a paternally inherited X chromosome lacking intron1 and a maternally inherited wildtype X chromosome, and whether the lack of the intron leads to any skewing of XCI in vivo, we analyzed the allele-specific expression of Xist and two X-linked genes, Mecp2 and G6pdx, in polymorphic heterozygous females (1loxC57BL/6/wtCAST/Ei) and a wildtype control (wtC57BL/6/wtCAST/Ei), respectively, where the C57BL6 X chromosome was transmitted from the father and the CAST/Ei wildtype X from the mother, by semi-quantitative RT-PCR on RNA isolated from various tissues (Fig 5C–E). In these experiments, we used the Barakat/Gribnau mouse model described in Figure 5B (Fig S1A). Normally, the paternal X chromosome initially undergoes imprinted XCI, which is reversed in the epiblast cells of the pre-implantation blastocyst to allow subsequent random XCI. The intron1 region has been implicated to be important for Xi-reactivation in the ICM, and thus, if the absence of intron1 prevents reactivation of imprinted XCI, we may observe non-random XCI in the adult mouse (Navarro et al., 2008).
However, we did not find differences in allele-specific expression pattern in the presence and absence of intron1 in heterozygous female mice (Fig 5C–E). As expected, the C57BL/6 Xist allele is more often expressed than the CAST/Ei X, consistent with a modifier effect, likely resulting in more cells with an inactivated C57BL/6 X (Cattanach and Isaacson, 1967). Because of the stochastic and clonal nature of XCI patterns in the adult mouse, variations in skewing towards Xist RNA from the C57BL/6 allele ranged from 50–90% (Fig 5C, E). Notably, we did not see a preference of Xist upregulation on the intron1-deleted X chromosome in tissues of the adult mouse in vivo, albeit we observed slightly skewed Xist RNA levels in heterozygous differentiating female ESCs carrying the same mutant intron1 allele (Barakat et al., 2011). In agreement with this notion, the X-linked genes Mecp2 and G6pdx, both subject to silencing on the Xi, showed reciprocal and intron1-independent levels of expression from the C57BL/6 chromosome compared to Xist, as would be expected from the fact that the Xist-expressing chromosome is more likely to be silent (Fig 5D/E). These data suggest that the paternal transmission of the intron1 mutation does not interfere with reactivation of imprinted XCI and subsequent random XCI. A reverse cross in which the maternal allele lacked intron1 also resulted in random XCI (data not shown). In summary, the intron1 genomic region is dispensable in the mouse and does not critically control Xist expression and skewing of XCI in vivo.
While there was no dramatic effect on XCI state in vivo, we sought to understand the requirement for intron1 in Xist silencing associated with reprogramming to iPSCs. We have shown previously that female iPSCs derived from mouse embryonic fibroblasts (MEFs) carry two active X chromosomes, where Xist is efficiently repressed and Tsix upregulated, as seen in mouse ESCs (Maherali et al., 2007). Another study suggested that Xi-reactivation occurs late in reprogramming at around the time pluripotency genes become expressed, again suggesting that pluripotency transcription factors could contribute to Xi-reactivation and the silencing of Xist, potentially via binding to intron1 (Stadtfeld et al., 2008). To test the role of intron1 in the Xist silencing process during reprogramming, we bred male mice carrying the 2lox intron1 allele (obtained upon blastocyst injection of our male 2lox ESCs described in Figure 1, Minkovsky/Plath allele) with female mice heterozygous for an Xist knockout allele (Marahrens et al., 1997), yielding female XX:2lox intron/ΔXist MEFs. Due to the presence of the Xist knockout allele, the X chromosome bearing the intron1 allele is exclusively inactivated in vivo by normal developmental mechanisms (Marahrens et al., 1998). MEFs isolated from d14.5 embryos had uniform Xist coating (Fig 6C) and were transduced with retroviruses encoding the reprogramming factors Oct4, Sox2, and Klf4, and subsequently infected with adenovirus encoding Cre recombinase at day 4 of reprogramming to efficiently delete the intron1 region or with titer-matched empty adenovirus in control samples (Fig 6A). This experimental setup allowed us to test the role of intron1 on reprogramming efficiency for the same infected fibroblast population. Genotyping confirmed that Ad-Cre addition resulted in efficient deletion of the intron1 region (Fig 6B). To test whether intron1 deletion affects the efficiency of reprogramming, we determined the number of Nanog-expressing colonies at day 13 after reprogramming factor introduction as Nanog expression has been shown to mark faithfully reprogrammed cells in retroviral reprogramming experiments (Maherali et al., 2007). We found a comparable number of Nanog-positive colonies in the presence and absence of intron1 (Fig 6D). Normally, at this point of reprogramming, Xist RNA coating is just lost in Nanog- positive cells (Tchieu/Plath et al, manuscript in preparation). In agreement with this notion, an examination of all Nanog-positive cells for the presence or absence of a Xist RNA cloud demonstrated that nearly all Nanog-positive cells carrying the 2lox intron1 allele (Ad-Null reprogramming cultures) lack a Xist RNA cloud at d13 of reprogramming (Fig 6C/E). Importantly, even in the absence of intron1 (Ad-Cre samples), Nanog-positive cells displayed loss of the Xist RNA cloud (Fig 6C/E) and of the Xi-like H3K27me3 focus (data not shown). Furthermore, from the Ad-Cre treated reprogramming cultures, 14 iPSC clones were isolated and clonally propagated and all confirmed to have lost both intron1 and the Xist RNA cloud, demonstrating the efficient deletion of the intronic sequence early in reprogramming (Fig 6F). To ensure that the ability of an intron1-deleted inactive X chromosome to downregulate Xist was not due to intron 1-dependent events occurring within the first four days of reprogramming, i.e. prior to Cre-mediated deletion, we also reprogrammed MEFs carrying a germline transmitted 1lox intron allele. These XX:1lox intron/ΔXist MEFs displayed normal Xist RNA coating before reprogramming (detectable in 95% of the cells) and lost Xist RNA in Nanog-positive colonies (Fig 6H). When comparing to sibling XX:2lox intron/Xist MEFs, MEFs lacking intron1 form Nanog-positive colonies with similar efficiencies (Fig 6I). Together, these studies rule out that Xist intron1 is necessary for the downregulation of Xist in reprogramming to pluripotency.
In summary, our data argue that Xist intron1 does not represent an essential tether coupling repression of both Xist and XCI to the pluripotent state. ESCs lacking intron1 do not dysregulate Xist expression in the undifferentiated state nor upon in vitro differentiation, reprogramming to the iPSC state leads to Xist repression on an Xi lacking intron1, and mice lacking intron1 do not display any of the gross reproductive abnormalities that would be expected if XCI was perturbed.
The deletion of intron1 represents a clean experimental system to probe the functional role of a genomic element that displays very strong pluripotency transcription factor binding, unhampered by the secondary effects on initiation of XCI associated with global modulation of protein factors implicated in the maintenance of the pluripotent state. While correlative binding studies were supported in part by Xist dysregulation in ESC lines with inducible deletions of the pluripotency factors Nanog and Oct4, our study cautions against extrapolating these findings to the behavior of wildtype ESCs and mice. A compromised pluripotency factor network could unmask the mild intron1 contribution to the Xist repressive pathway. In the case of the ZHBTc4 cell line, this altered network may show Rnf12 upregulation followed by downregulation of the pluripotency factor Rex1, sufficient to trigger XCI in male cells independent of intron1 (Barakat et al., 2011a; Gontan et al., 2012). We also noted that ZHBTc4 ESCs lack pinpoint Tsix signal and draw a corollary between their Xist upregulation and our male ESCs deleted for Tsix (that, when differentating, have a significantly greater number of Xist clouds upon deletion of intron1).
In light of the two mild phenotypes (skewing effect of deleting intron1 in female ESCs heterozygous for the allele and the increase in Xist clouds in Tsix and intron1-deleted differentiating male ESCs), we hypothesize that intron1 loss leads to mild destabilization of Xist transcriptional repression at the transition to the differentiated state, in the narrow development window of XCI initiation. Unable to capture a transcriptional difference in Xist levels at the onset of in vitro differentiation, we believe that more sensitive methods of transcript quantitation or investigation of chromatin state may address this hypothesis.
We noted a discrepancy between the ex vivo XCI skewing phenotype and the normally occurring in vivo XCI choice in the absence of the intron. This lack of intron1 deletion effect in adult mice and in ESC differentiation induced by bFGF/Activin (Supp Fig 5) which is sensitive to clonogenic skewing of XCI because of serial passage and outgrowth of few cells (unlike monolayer differentiation, Chenoweth and Tesar, 2010), suggests that Xist regulation is more robust in vivo than in vitro in the absence of the intron1. For instance, slightly different cis-acting elements could be used in vivo and in vitro for regulating Xist expression. Thus, the cell culture-observed favoring of the intron-deleted Xist could not be organismally relevant or the stochastic developmental nature of XCI could overshadow the effect.
It seems that the regulation of Xist, at the helm of a chromosome-wide program of gene expression, is genetically ensured by a complex multifactor mechanism. The dispensability of intron1 for repression of Xist may be mouse-specific as mice appear to be unique in the functionality of Tsix and also in the sufficiency of Xist activators such as Rnf12 to elicit Xist upregulation: addition of one copy of Rnf12 is sufficient to drive Xist expression in undifferentiated female ESCs (Jonkers et al., 2009). Other eutherians such as bovines and humans, with truncated and likely non-functional TSIX, may rely more on intron1-dependent mechanisms for Xist repression (Chureau et al., 2002). Therefore, evolution of the overlapping Tsix gene and the network of XCI activators in mice may have become the dominant mechanism in Xist repression.
Xist intron1 transgenic mice analyzed in Figures 5B–5E were generated from polymorphic XXist:2lox intron (129/Sv) XXist: wt (CAST/Ei) ESC line 29, in which a 1.8 kb region of Xist intron1 was replaced by a floxed neomycin cassette (Barakat et al., 2011). Germline transmission was verified by genotyping for the presence of the neomycin cassette integrated in the intron1 region of Xist and XX:2lox/wt females were bred to males expressing pCAGGS-Cre, to loop out the selection cassette. Loopout of the selection cassette was verified by PCR on genomic tail-tip derived DNA. All other intron1- mutant ESC lines and mice carrying the Plath/Minkovsky allele were derived from a targeting construct generated by cloning the respective genomic fragments representing the 5′ and 3′ homology regions into the pCRII plasmid vector upon PCR amplification (see Supp Table 1 for list of primers used). The 800 bp of intron sequence with a 5′ loxP site were ligated between a 2.2 kb 5′ homology arm and 3′ 2.6 kb homology arm by AgeI/NotI subcloning. A positive-negative CMV-HygroTK cassette flanked by loxP sites was inserted into the unique NotI site. A diphtheria toxin gene (PGK-DTA) was inserted into a unique backbone EcoRI site. 40 μg of plasmid were linearized by MluI digestion and electroporated into male ESCs (V6.5 line; F1 between C57BL/6 and 129SV/Jae) and into female F1 2–1 ESCs carrying the MS2 tag in the final large exon of Xist (F1 between C57BL/6 and CAST/Ei) cells co-cultured with drug-resistant DR4 MEFs (Jonkers et al., 2008; Tucker et al., 1997). Hygromycin selection (140 μg/ml) was started one day after and clones were screened by SpeI/KpnI digest and both 5′ and 3′ external probes. BmtI digest and 3′ external probe were used to assess allelism of targeting in F1 2–1 clones. Targeting efficiency was 30% in V6.5 and 1% in F12–1 cells. Two independent V6.5 and one F1 2–1 clones were expanded, electroporated with pPAC-Cre plasmid, and selected with G418 (300 μg/mL) for 8 days. Southern blot screening was performed with a 5′ probe and XbaI digest for 1lox and SpeI/KpnI for 2lox clones. All subsequent intron1 genotyping was performed by PCR. For intron1/Tsix-Stop double transgenic ESC clones, XY:2lox and XY:1lox V6.5 clones were targeted with pAA2Δ1.7 and screened by Southern blot as previously described (Sado et al., 2001). XY:1lox V6.5 ES cells were microinjected into C57BL/6 blastocysts to produce chimeric mice following standard procedures. High agouti coat color-contributing chimeras were bred with C57BL/6 females for germline transmission. All animal experiments were in accordance with the legislation of the Erasmus MC Animal Experimental Commission and the UCLA Animal Research Committee.
ESCs were grown on irradiated DR4 mouse embryonic fibroblasts (MEFs) in standard media (DMEM supplemented with 15% FBS, nonessential amino acids, L-glutamine, penicillin-streptomycin, β-mercaptoethanol, and 1000 U/mL LIF). Prior to induction of RA-differentiation, cells were feeder-depleted for 45 minutes on gelatinized plates and plated at a density of 5.0×104 cells/6-well in MEF media (same as ESC media except 10% FBS and excluding LIF). One day later, MEF media was supplemented with 1μM all trans retinoic acid (Sigma) or with DMSO only (LIF withdrawl) and refreshed every two days. For embryoid body differentiation, ESCs were pre-plated on gelatin overnight to feeder deplete, briefly trypsinized, and put in MEF media for suspension culture on bacterial culture plates for four days, then plated on gelatinized coverslips for another 2 or 6 days. For FGF/Activin differentiation, ESCs were feeder-depleted and 2.0×104 cells plated on 6 wells pretreated with fibronectin in DMEMF12/B-27/N-2 (Invitrogen) supplemented with FGF-2 (R&D Systems 40ng/mL) and Activin A (PeproTech, 20 ng/mL). Media was changed daily and colonies were manually passaged onto fibronectin several times then at passage 4 returned to feeder cells. ZHBTc4 ES cells were induced to differentiate with 1μg/mL doxycycline (resulting in acute repression of Oct4) in standard ESC media (Niwa et al., 2000). For reprogramming, primary MEFs were derived at embryonic day 14.5 and 3-factor retroviral reprogramming was performed following previously published methods (Maherali et al., 2007).
ChIP was performed according to previously published methods (Maherali et al., 2007). In summary, formaldehyde-crosslinked chromatin fragments were generated by sonication and 150ug of material were pre-cleared with Protein A sepharose beads. Immunoprecipitation was performed overnight with 5ug antibodies targeting Oct4 (R&D Systems, AF1759) or Sox2 (R&D Systems, AF2018), or with normal goat IgG (Santa Cruz, sc-2028) and subsequent incubation with protein A sepharose beads for 3 hours. Beads were washed and eluted in TE/0.67% SDS. Both IP and input samples were reverse cross-linked overnight at 65 degrees and treated with RNAse A and Proteinase K before DNA phenol-chloroform purification. The proportion of input material immunoprecipitated was calculated using standard curves constructed from input serial dilutions and comparing fractional measurements in IP and input relative to a known region positive for Oct4 and Sox2 binding (Rest, (Berg et al., 2008) ChIP with goat IgG antibody did not find any enrichment (data not shown).
Cells were plated on glass coverslips (and in the case of embryoid body differentiation permeabilized with 5 minute washes of ice-cold CSK buffer, followed by CSK buffer with 0.5%Triton, and another wash in CSK buffer, washed once with PBS, and fixed for 10 minutes in 4% paraformaldehyde (Plath et al., 2003). Immunonostaining with antibodies against Nanog (BD Pharmingen 560259) and H3K27me3 (Active Motif 39155) and combined immunostaining/FISH with double-strand Xist DNA probe labeled with FITC were performed as previously reported and mounted with Prolong Gold reagent with DAPI (Tchieu et al., 2010). Xist and Tsix strand-specific RNA probes were made by in vitro transcription of T3-ligated PCR products of cDNA templates using Riboprobe system T3 (Promega) with Cy3-CTP (VWR) or FITC-UTP (Perklin Elmer) (Maherali et al., 2007).
Cells were harvested from a 6 well format in Trizol (Invitrogen) and RNA purification was performed with the RNeasy kit (Qiagen) according to manufacturer’s instructions with on-column DNAse treatment (Qiagen). cDNA was prepared using SuperScript III (Invitrogen) with random hexamers and qRT-PCR was performed using a Stratagene Mx3000 thermocycler with primers listed in Supp Table 1. Results were normalized to Gapdh by the ΔCt method. To assess XCI skewing in adult mice, parts of organs were collected, snap-frozen and triturated using micropestles in 1 ml of Trizol reagent (Invitrogen). After an additional centrifugation to clear debris, 700 μl was added to 300 μl fresh Trizol, and RNA was purified following manufacturer’s instructions. RNA was reverse-transcribed with SuperScript II (Invitrogen) using random hexamers. Allele-specific Xist expression was analyzed by RT-PCR amplifying a length polymorphism using primers Xist LP 1445 and Xist LP 1446. To determine allele-specific X-linked gene expression of Mecp2 and G6pdx primers MeCP2-DdeI-F and R and G6PD-ScrFI-F and R were used to amplify respective RFLPs. PCR products were gel-purified and digested with the indicated restriction enzymes and analyzed on a 2% agarose gel stained with ethidium bromide. Allele-specific expression was determined by measuring relative band intensities using a Typhoon image scanner and ImageQuant software.
XY:2lox ESCs were transfected by electroporation, as in intron1 targeting, with 40 μg of one of three BamHI-linearized pgl4.27-cloned constructs (Promega, Supp. Table 1) and transferred to hygromycin selection (140μg/mL) one day later. After serial passaging and outgrowth of stable transfectants 1.0×105 or 2.0×104 ESCs were seeded for differentiation with and without (no LIF) retinoic acid for 3 and 5 days and harvested along with 2.0×105 ES cells and measured for luciferase activity with the luciferase assay system (Promega).
KP is supported by the NIH (DP2OD001686 and P01 GM099134), CIRM (RN1-00564 and RB3-05080), and by the Eli and Edythe Broad Center of Regenerative Medicine and Stem Cell Research at UCLA; AM by National Research Service Award AG039179. J.G. was supported by NWO VICI and ERC starting grants. We thank Ying Wang of the UCLA Transgenic Core Facility for blastocyst injections; Takashi Sado for his pAA2Δ2.17 Tsix targeting construct and helpful input; Elyse Rankin-Gee, Dana Case, Matthew Denholtz, Rebecca Rojansky, Konstantinos Chronis, Ritchie Ho, Bernadett Papp, and Sanjeet Patel for advice and experimental support.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.