Abstract
X chromosome inactivation (XCI) is an essential epigenetic process which involves several non-coding RNAs (ncRNAs), including Xist, the master regulator of X-inactivation initiation. Xist is flanked in its 5′ region by a large heterochromatic hotspot, which contains several transcription units including a gene of unknown function, Ftx (five prime to Xist). In this article, we describe the characterization and functional analysis of murine Ftx. We present evidence that Ftx produces a conserved functional long ncRNA, and additionally hosts microRNAs (miR) in its introns. Strikingly, Ftx partially escapes X-inactivation and is upregulated specifically in female ES cells at the onset of X-inactivation, an expression profile which closely follows that of Xist. We generated Ftx null ES cells to address the function of this gene. In these cells, only local changes in chromatin marks are detected within the hotspot, indicating that Ftx is not involved in the global maintenance of the heterochromatic structure of this region. The Ftx mutation, however, results in widespread alteration of transcript levels within the X-inactivation center (Xic) and particularly important decreases in Xist RNA levels, which were correlated with increased DNA methylation at the Xist CpG island. Altogether our results indicate that Ftx is a positive regulator of Xist and lead us to propose that Ftx is a novel ncRNA involved in XCI.
INTRODUCTION
Non-coding RNAs (ncRNAs) are increasingly recognized as playing important functions in many cellular processes, including the regulation of gene expression and chromatin conformation. The exact number of ncRNAs encoded within the human genome is currently unknown, yet transcriptomic and bioinformatic studies suggest the existence of thousands of ncRNAs. Indeed, a recent study suggested that only one-fifth of transcription across the human genome is associated with protein-coding genes (1). In addition to abundant and functionally important small structural and regulatory RNAs, ncRNAs also include a category of long transcripts that bear many signatures of mRNAs, including 5′ capping, splicing and polyadenylation.
X chromosome inactivation (XCI), the transcriptional silencing of one X chromosome which takes place in female mammals to compensate for X-linked gene dosage imbalance, is controlled and regulated by a cis-acting region on the X chromosome termed the X-inactivation center (Xic). Defined on the basis of chromosomal rearrangements, the Xic has the intriguing characteristic of harboring several such long ncRNAs genes (2), including Xist, the master control locus in this process. The Xist gene, which encompasses >30 kilobases (kb) of genomic DNA, is transcribed by RNA polymerase II (3), spliced and polyadenylated to produce a 17 kb-long transcript with no coding potential that remains in the nucleus and coats the chromosome from which it is expressed. Although Xist expression and coating of the X chromosome are absolutely essential for gene silencing, very little is known concerning the underlying mechanisms. Xist RNA appears to be able to recruit chromatin-modifying complexes that are involved in the heterochromatinization of the inactive X, but these events are believed to be involved in the maintenance of silencing rather that in its establishment. More importantly, Xist is responsible for creating a repressive nuclear compartment from which the transcription machinery is excluded (4). A specific region of the Xist transcript, the A repeat, plays a prominent role in the silencing function of Xist (5) and it was recently shown that this conserved region is able to adopt a stem-loop 2D structure and to interact with several proteins, including members of the polycomb repressive complex 2 (6).
Xist is essentially required for the initiation of the X-inactivation process, and its expression has to be tightly controlled to ensure the developmental regulation of XCI during embryogenesis. Mouse ES cells have proven to be a powerful tool to study random X-inactivation in general and to identify the molecular actors involved in the control of Xist. Using this system, Tsix, a non-coding transcript antisense to Xist, was identified as a negative regulator which prevents Xist upregulation. We and others have shown that Tsix is not a direct repressor of Xist transcription; it is rather involved in the programming of Xist expression through the establishment of a repressive conformation at the Xist promoter (7,8). More recently, the pluripotency factors NANOG, OCT4 and SOX2 were identified as direct repressors of Xist, thus establishing the first molecular link between X-inactivation, development and pluripotency (9). The repression of Xist, therefore, likely results from the combined action of trans-acting transcriptional repressors and a cis-acting chromatin remodeler.
Although Xist upregulation represents a crucial step in the X-inactivation process, very little is known so far concerning the underlying network. The E3 ubiquitin ligase Rnf12 was recently identified as an X-encoded activator of X-inactivation which may act, directly or indirectly, through Xist activation (10). Another region of the X chromosome located 200 kb upstream to Xist was also shown to be involved in promoting X-inactivation. This Xpr region is involved in the pairing of both X chromosomes that occurs very early during X-inactivation and is proposed to trigger Xist upregulation (11). The Xpr region is part of a larger domain that was found to display, prior to inactivation, several chromatin features of the inactive X, including an enrichment for H3K9me2 and H3K27me3 (12,13). Aside from the Xpr region, the regulation and the function of this heterochromatic hotspot remain unknown. Strikingly, two ncRNAs are embedded within the hotspot, Jpx and Ftx. Jpx/Enox has been described previously as partially escaping X-inactivation (14).
Here, we describe the molecular and functional characterization of Ftx (five prime to Xist). Ftx is well conserved during evolution and contains several microRNAs (miRs) which are also conserved. It produces a transcript with no apparent coding potential that predominantly remains in the nucleus. Strikingly, Ftx is specifically upregulated during female ES cell differentiation at the onset of XCI. Loss of Ftx expression through the deletion of the promoter region leads to local alteration in the chromatin structure. More importantly, Ftx impacts on gene expression within the Xic and in particular on Xist, as Xist is significantly downregulated in the absence of Ftx. These results identify Ftx as an activator of Xist in ES cells and suggest that Ftx could be a novel ncRNA involved in XCI.
RESULTS
Genomic organization and conservation of the Ftx gene
We have identified previously a large hotspot of H3K9 and H3K27 methylation upstream of Xist (12,13). Despite its heterochromatic nature, several genes were found to lie within this region (Fig. 1A), including the protein-coding genes Cnbp2 and Xpct, the latter mapping to within the Xpr region involved in the pairing of the two X chromosomes at the onset of X-inactivation (11). Strikingly, the hotspot also harbors in its most centromeric part (close to Xist) two non-coding genes, Jpx/Enox and Ftx. The function of Jpx/Enox is unknown, but the gene is expressed independently of Xist/Tsix and has been shown to partially escape X-inactivation (14). Ftx was initially identified as containing seven exons lacking coding potential (2). To gain insight into the functional relevance of Ftx, we analyzed sequence conservation by comparing the mouse gene with human and bovine orthologs for which several ESTs are available. The genomic organization of the Ftx gene was deduced in the three species, using genomic ESTs alignment.
Ftx genomic organization and conservation. (A) Schematic representation of the mouse X-inactivation center (Xic). Gray arrows represent protein-coding genes and colored ones non-coding transcription units. The light green box represents the so-called region B, a complex transcriptional unit that produces both sense and antisense transcripts (2). The localization of the H3K9/K27 methylation hotspot (12,13) is indicated. (B) Genomic organization of the Ftx gene in the mouse and description of the corresponding ESTs. Exons are indicated by rectangles; in color are alternatively spliced exons (gray: skipped exons; green: exons with alternative 3′ splicing site; orange: intron retention). CpG islands are indicated in red. The position of the two miRs is shown. (C) Conservation of Ftx in the mouse, human and cow. The size of the genomic region covered by the Ftx gene in the corresponding species is indicated in bracket. The similarity between exons is indicated by dotted lines. Note the presence of an additional miR cluster in the 5′ part of the gene in the human and cow.
In the mouse, the analysis of 6 ESTs identified 8 additional Ftx exons, indicating that the Ftx gene is composed of 15 exons spanning ∼63 kb (Fig. 1B). Ftx gene gives rise to several RNA isoforms that originate through varied combinations of alternative promoters usage, splicing and termination. The transcription is initiated from at least three promoters, as suggested by the usage of alternative first exons (exons 1, 2 and 3; Fig.1B). The UCSC genome database predicts CpG islands associated with exon 1, 3 and 4, supporting the conclusion that the use of alternative promoters regulates transcription. Ftx transcripts are also characterized by the alternative splicing of internal exons 5 and 10, and by the usage of alternative termination, with exon 12 spliced to either exon 13 or exons 14 and 15 (see legend to Fig. 1B for details). Ftx gene complexity is increased by the localization, within the largest intron (intron 12), of a cluster of two miRs (miR-374 and miR-421). These miRs have been identified experimentally (15,16), and miR-421 seems to target ATM (ataxia-telangiectasia mutated), a serine/threonine kinase that plays a central role in genome integrity by activating cell-cycle checkpoint and promoting double-strand break DNA repair (17).
Genomic alignment of six human ESTs allowed the identification of 12 exons spanning ∼330 kb in the syntenic region on the X chromosome (Supplementary Material, Fig. S1A). In the cow genome, 15 exons spanning ∼82 kb were similarly identified (Supplementary Material, Fig. S1B). As in the mouse, the human and bovine Ftx gene gives rise to various transcripts that originate from different combinations of alternative promoter usage and splicing (Supplementary Material, Fig. S1A and Supplementary Data).
Comparative analysis revealed that the 5′ region of Ftx is particularly well conserved, with human and cow exons a and b corresponding to mouse exons 1 and 2, respectively (Fig. 1C). In addition, a CpG island is predicted to be associated with exon a in the cow and 1 in the mouse, supporting the homology between the two exons. In the human, the CpG density around exon a is high but below the threshold to be defined as a CpG island. A CpG island is, however, predicted to be associated with exon b. Four other exons in the 5′ of the gene show similarity between the human and cow (corresponding to human exons d, e, f and g), and three of them are homologous to mouse exons 8, 9 and 12 (Fig. 1C). Interestingly, these conserved exons are present in all the alternative transcripts identified so far by EST analysis (except human exon g, which is alternatively spliced in the EST CN412172; Fig. 1C; Supplementary Material, Fig. S1A and Supplementary Data). Thus, among the first seven human exons, six are conserved in the cow, and five are conserved in the mouse. This conservation of exons contrasts with the overall poor conservation of introns (Supplementary Material, Fig. S2). Taken together, our results suggest that the 5′ part of Ftx transcripts may be involved in functional domains and structures essential for all the Ftx isoforms. However, mouse exons 3–5 and 11 are not found in the human and cow, which suggests that they correspond to new exons, gained in the rodent lineage. This indicates that the function of Ftx is tolerant to gain or loss of some exons.
The miRNA cluster (miR-374 and miR-421) is also conserved in all three species (and called miR-374b and 421 in the human and cow, Fig. 1C; Supplementary Material, Fig. S1A and Supplementary Data), whereas a second cluster, miR-374a and miR-545, is localized in intron b of both the human and the cow. Interestingly, miR-374a and miR-545 show strong sequence similarities to miR-374b and miR-421, which could indicate that they arose by duplication. This second cluster is absent from mouse and rat Ftx gene, suggesting its pseudo-genization during the evolution of the rodent lineage. The conserved localization of these two miRNA clusters within introns of Ftx may suggest that these miRNAs are matured from the Ftx pre-RNA.
The Ftx gene produces nuclear non-coding transcripts
The Ftx transcript, when initially characterized, was suggested to be non-coding, based on the absence of a conserved open-reading frame (2). Since additional exons were identified in our analysis, we re-investigated the coding potential of Ftx by analyzing the alternative ESTs, looking for ORFs >300nt (100 codons). This arbitrary threshold was used by the FANTOM consortium to identify putative mRNA (18) and is consistent with the observation that >95% of proteins in public databases are >100 amino acids in length (19). The sequence analysis revealed multiple stop codons and multiple short-predicted ORFs (<300nt) in all three frames of all the ESTs analyzed. The absence of an ORF >100 codons suggests that Ftx is a non-coding gene. A BLASTX analysis of ORF conservation further showed that the putative short proteins encoded by the alternative ESTs do not show significant sequence similarity with any of the proteins in public databases. In addition, the comparison of mouse and human Ftx transcripts using the NCBI TBLASTX tool did not reveal any conserved ORF. These data suggest that the short ORFs contained in mouse Ftx transcripts are likely random, non-functional occurrences.
To assess the intracellular localization of Ftx transcripts, we measured their enrichment in nuclear and cytoplasmic fractions relative to a cytoplasmic transcript (Gapdh). Male (TC1) and female (PGK) undifferentiated ES cells were fractionated to extract nuclear, cytoplasmic and total cellular RNAs. An identical amount of RNA from each fraction (1 μg) was reverse transcribed and analyzed by quantitative real-time PCR (qPCR). In this assay, the values are normalized to Gapdh in each fraction and expressed as enrichment relative to this mRNA. The distribution of Gapdh mRNA is shown in Supplementary Material, Fig. S3A. For each histogram, the value of the total cellular RNA (after normalization to Gapdh) is set to 1. Thus, a cytoplasmic mRNA such as β-actin shows no enrichment in the nucleus relative to Gapdh, and cytoplasmic and nuclear levels are approximately equal to 1 as reported in Figure 2. In contrast, nuclear-localized transcripts such as Xist show enrichment in the nucleus relative to Gapdh in both male and female ES cells (Fig. 2). We then evaluated the distribution of the Ftx RNAs and showed enrichment of the transcripts in the nucleus compared with Gapdh, with a similar nucleo/cytoplasmic ratio in both male and female ES cells (Fig. 2). Because the enrichment reflects the distribution of a transcript within a cell independently of transcript levels, this indicates that Ftx transcripts have similar subcellular localization in male and female cells. Predominant nuclear localization was found for all the Ftx transcript variants (data not shown). However, the ratio observed is lower compared with the known nuclear transcripts analyzed (Xist, Tsix and β-act pre-mRNA, Fig. 2B), suggesting that Ftx transcripts are prevalently localized into the nucleus but not exclusively. A similar distribution of Ftx transcripts was found in differentiating male and female ES cells (Supplementary Material, Fig. S3). In conclusion, the absence of a conserved open-reading frame in Ftx transcripts and their prevalent nuclear localization indicate that Ftx produces ncRNAs.
Subcellular localization of Ftx transcripts. RNA was extracted from whole-cell, nuclear and cytoplasmic fractions of undifferentiated male (TC1) and female (PGK1) ES cells. (A) Subcellular distribution of Ftx and control transcripts. All values are normalized to Gapdh and expressed as enrichment relative to this mRNA, and for each histogram the value of the total cellular RNA is set to 1. β-Actin, a cytoplasmic mRNA, shows no enrichment relative to Gapdh. In contrast, nuclear-localized transcripts such as Xist show enrichment in the nucleus relative to Gapdh. Ftx similarly show predominant nuclear localization. (B) Quantification of the nuclear/cytoplasmic ratio of several transcripts including those shown in (A) relative to Gapdh.
Ftx is ubiquitously expressed but specifically upregulated at the onset of X-inactivation
The expression profile of Ftx was first investigated in adult mouse tissues by northern and RT-PCR analysis (Fig. 3A and B). Two major bands can be detected on northern blot in most of the tissues tested, and each likely corresponds to multiple transcripts as suggested by the analysis of ESTs and cDNAs (see the size of the different ESTs in Fig. 1B). The ubiquitous expression of Ftx was confirmed by RT-PCR analysis (Fig. 3B). No significant difference in tissue distribution could be detected between the different spliced forms tested (data not shown).
Analysis of Ftx expression profile. Ftx expression in the adult spleen, lung, liver, muscle, kidney and testis/ovary analyzed by (A) northern and (B) RT-PCR using Ftx primers spanning exons 8, 9 and 10. Since several commonly used reporters show strong variation between tissues (see Supplementary Material, Fig. S6), the results are presented as raw data without normalization. Ftx transcript levels were similarly analyzed in male and female ES cells. (C) RT-PCR analysis of Ftx expression during male and female ES cell differentiation after normalization to Arpo PO transcript levels. Primers used are the same as in (B). The graph shows the average of seven independent experiments using two male and three female ES cell lines. Note that several Ftx primers have been used and all display the same profile (data not shown). Xist expression during ES cell differentiation is shown as a comparison. (D) RNA-FISH in female ES cells, both undifferentiated and after 4 days of differentiation using a Xist probe (p510, in green) and an Ftx probe (in red). (E) Allele-specific analysis of Xist, Nap1l2 and Ftx in XEN cells carrying an inactive, paternal X chromosome of Pgk origin (in light gray) and an active, maternal X chromosome of 129 origin (in dark gray) (n= 2). Xist and Nap1l2 polymorphic primers were described previously (20,35). A polymorphism in Ftx exon 15 was used to design allele-specific primers (see Supplementary Material, Fig. S5).
We next investigated the pattern of Ftx expression in ES cells and during their differentiation. ES cells are derived from the inner cell mass of the blastocyst when X chromosomes are active. Differentiation of female ES cells rapidly leads to the random inactivation of one of the two X chromosomes, whereas the single X of male cells remains active. Female ES cells, therefore, constitute a powerful tool for the study of random X-inactivation. Strikingly, whereas Ftx expression remains stable during the course of male ES cell differentiation, a specific and reproducible upregulation of Ftx is observed early during female ES cell differentiation at the onset of Xist upregulation and X-inactivation (Fig. 3C). This expression profile, which resembles that of Xist, is specific for Ftx and not a general feature of the hotspot region, as Jpx expression was shown to be elevated during the onset of both male and female ES cell differentiation (Supplementary Material, Fig. S4). Since Ftx levels in differentiated tissues are similar to that of ES cells, this upregulation appears to be transient and to be lost in later stages of differentiation.
To study the status of Ftx regarding XCI, we analyzed its expression by RNA-FISH in the course of female ES cell differentiation (Fig. 3D). In undifferentiated cells, two pinpoints of Ftx are detected in the vicinity of the Xist/Tsix signal, indicating that Ftx is expressed from both X chromosomes. During differentiation, we observed in about half of the cells a loss of the Ftx signal from the Xist RNA-coated chromosome, revealing correct silencing of this gene on the inactive X. However, in the other half (62%, n= 93), two Ftx signals continue to be detected, one of which is located at the periphery of the Xist domain. In all these cases, the Xist-associated pinpoint was always smaller than the pinpoint on the active X. These data, which suggest that Ftx partially escapes X-inactivation, were confirmed in other independent cellular models. TS and XEN cells are stem cells derived respectively from the trophectoderm and the extraembryonic endoderm. In both cell types, imprinted X-inactivation has already been established, with the paternally derived X being the inactive one (20,21). We took advantage of TS and XEN cell lines derived from interspecific crosses, where X-linked polymorphisms allowed us to distinguish between the paternal and maternal X chromosomes. We used specific primer pairs within polymorphic regions of several X-linked genes, including Ftx, so that only the paternal or the maternal allele can be amplified. Allele-specific analysis of Xist and Nap1l2 (a gene subjected to X-inactivation) in XEN cells (Fig. 3E) confirmed the complete bias in X-inactivation, with Xist being exclusively expressed from the paternal inactive X, while Nap1l2 is transcribed only from the active, maternal chromosome. In contrast, Ftx displayed some levels of biallelic expression, showing that while Ftx is predominantly expressed from the maternal allele, significant expression can be detected from the inactive X. Allelic Ftx primers were verified for their efficiency and specificity on genomic DNA (Supplementary Material, Fig. S5A). This result, which was confirmed in TS cells (Supplementary Material, Fig. S5B), further demonstrates that Ftx partially escapes X-inactivation. This property has also been reported for the neighboring ncRNA Jpx/Enox (14), which may indicate that a subregion of the hotspot is protected, in a partial but coordinate manner, from X-inactivation. Partial escape from X-inactivation might explain the slight increase in Ftx RNA levels in most female somatic tissues compared with males (Fig. 3B). Importantly, it cannot solely account for the female-specific upregulation during ES cell differentiation, which can reach up to a 5-fold increase.
Targeted invalidation of the Ftx promoter region
Several characteristics of Ftx, including its conservation across species and its peculiar expression profile, strongly suggest that Ftx is a functional RNA. One important question is the involvement of Ftx in the heterochromatic structure of the region and the subsequent consequences on gene expression. To test this, we generated Ftx mutant male ES cells, which provide an appropriate cellular context in which the above-mentioned aspects can be analyzed in detail without contamination from a wild-type X chromosome.
A 12kb region including the five most 5′ exons and all the putative Ftx promoters was deleted from an Ftx containing BAC (BAC 561P13 from the 129/Sv mouse strain, BAC Library, Research Genetics, Inc.) by recombineering in Escherichia coli. The modified BAC D NeoDTA-ΔFtx was then transfected into TC1 male ES cells (Fig. 4A). Seven positive clones were obtained and verified by Southern analysis (Fig. 4B). The genomic integrity of the region in these clones was extensively analyzed by restriction digestion (data not shown). Three clones were used for further analysis following the removal of the selection cassette by transient transfection with a plasmid expressing the Cre recombinase. As expected, Ftx expression is completely abrogated in all the clones tested, even when primers corresponding to the most 3′ exons, located >40 kb from the deletion, are used (Fig. 4C). This was shown to be the case in both undifferentiated and differentiated cells, which confirms that all the putative Ftx promoters had been deleted. Notably, expression of both miR-374 and miR-421 is also lost in mutant clones (Fig. 4D), which, together with the localization of the miRNA cluster within Ftx intron 12, may suggest that these miRNAs are processed from the Ftx pre-messenger RNA. It cannot be completely excluded, however, that the deletion we have generated impacted the activity of an miR-specific promoter.
Generation of Ftx mutant ES cells. (A) Schematic representation of the targeting strategy. A 12.36 kb region including Ftx exons 1 to 5 was deleted from TC1 male ES cells by homologous recombination using a modified BAC (BAC D NeoDTA-▵Ftx). The selection cassette (shown by green and blue arrows and boxes) was subsequently removed by transfecting recombinant ES clones with a plasmid expressing the Cre recombinase, leaving in place a LoxP site (blue arrow). (B) Southern validation of recombinant clones. Genomic DNAs from control (gDNA) and recombinant clones (1–9) were digested using NcoI and hybridized with a genomic probe located outside of the homology arms, which detects an 8.5 kb band from WT DNA and a 6.6 kb restriction fragment from the deleted allele. (C) RT-PCR analysis of Ftx expression in undifferentiated and differentiated WT and Ftx mutant clones using primers as in Figure. 3. Note that the same profile was obtained for all Ftx primers tested. The average of n > 3 experiment is shown. For ΔFtx, the results obtained with the three clones were pooled. (D) RT-PCR analysis of miR expression in undifferentiated WT and Ftx mutant clones (n= 2).
Ftx mutation induces local changes in chromatin structure
In order to study the impact of Ftx invalidation on the chromatin structure of the hotspot and of the Xic in general, we used a PCR-based custom-made array of a 300 kb region in chromatin immunoprecipitation (ChIP) assays (22) (Fig. 5A). We first analyzed H3K9 di- and H3K27 tri-methylation, two marks that had previously been shown by local PCR amplification to characterize the hotspot (12,13). Enrichment for both marks was confirmed within the hotspot region compared with other parts of the Xic in WT cells (Fig. 5B and C). However, the Ftx genomic region itself displays reduced levels of H3K27me3 and, to a lesser extent, H3K9me2. Strikingly, this appears to be linked to Ftx expression as in the mutant cell lines—the Ftx region accumulates both marks (from position 250 000 up to the deletion). Ftx transcription, therefore, appears to prevent accumulation of heterochromatic histone marks.
ChIP analysis of Ftx mutant clones. (A) Schematic diagram of the 300 kb region analyzed in our ChIP experiments showing the location of the different transcription units of the Xic. Arrows indicate the direction of each transcription unit. Coding genes are indicated in black, non-coding genes in colored arrows, and CpG islands are indicated by stars. The position of the miR cluster is also indicated. Orange box represents the Xite locus, an enhancer region that displays transcriptional activity on the major Tsix promoter. We used in each ChIP experiment a set of 383 primers pairs that were designed automatically to cover the 300 kb region (except repeat regions, see Materials and Methods). Each primer pair is represented by a vertical red line. (B–D) ChIP analysis of H3K9me2, (B) H3K27me3 and (D) H3K4me2 in WT (black line) and Ftx mutant (red line) male ES cells (n= 2; for Ftx mutants, results obtained for the three clones were pooled). Graphs show the percentage of immunoprecipitation (%IP) obtained after normalization to the input. All values are mean ± SD. The average %IP calculated for each position was plotted against the genomic location (pb). The +1 coordinate corresponds to position 1 005 322 247 in NCBI Build 37. The vertical dotted lines indicate the boundaries of the deletion.
With the exception of the Ftx genomic region, the profiles of H3K9me2 and H3K27me3 are mostly unaffected by the absence of Ftx, both within and outside the hotspot region. This result indicates that the expression of Ftx is not involved in the maintenance of heterochromatic marks characterizing the hotspot. We cannot rule out, however, that Ftx plays a role in the establishment of such a structure.
We next investigated H3K4 di-methylation distribution in WT and Ftx mutant cells. As shown previously (22) and in Fig. 5D, the hotspot region is globally deprived of this histone mark, with the exceptions of the promoters of Jpx and Ftx and of a few interspaced regions. This profile is extremely well conserved in Ftx mutant cells, where the only detectable change concerns a 25 kb region that includes the miR cluster (position 225 000 to 250 000), with the strongest effect restricted to a 5 kb subregion 5′ of the cluster (the miR cluster is around position 236 200). Accumulation of substantial levels of H3K4me2 in this subdomain is lost in the Ftx mutant and raises the intriguing possibility that the activity of an internal miR-specific promoter or regulatory element is altered by the Ftx mutation.
Ftx controls gene expression within the Xic
In addition to the Jpx and Ftx ncRNAs, the hotspot region is characterized by complex sense and antisense transcription units located in between these two genes (called B region, Fig. 6A) (2). The impact of Ftx deletion on transcription within the B region and in the Xic in general was investigated by RT-PCR. As shown in Fig. 6B, the loss of Ftx has major and complex consequences on expression levels within the Xic. Although gene expression levels are stable (Cdx4 and Nap1l2) or slightly increased (Chic1 and Tsx) in the proximal Xic (Fig. 6A), a significant decrease is observed for all positions tested from Tsix to within the hotspot, whether intra- or intergenic. In particular, a gradient of downregulation was observed from Ftx to Tsix, with the strongest effect detected for positions closest to Ftx. Because this region is characterized by sense and antisense transcription, we next asked whether transcription on both strands was affected using strand-specific RT-PCR. Analysis of two positions revealed that only sense transcription (i.e. in Ftx orientation) is lost or strongly reduced in Ftx mutant cells (Fig. 6C). This may suggest that unspliced transcription is also initiated from the Ftx promoter region and extends beyond the known Ftx exons. Alternatively, one cannot exclude the possibility of an indirect effect of Ftx promoter deletion on remote transcription. This is likely to be the case for other genes of the Xic, such as Tsix, Jpx and Cnbp2, which are in the opposite orientation compared with Ftx. In the case of Cnbp2, whose 5′ end is located in close proximity (7 kb) to Ftx, one might hypothesize that a regulatory element necessary for Cnbp2 expression has been altered by the Ftx mutation.
Expression analysis of Ftx mutant clones. (A) Schematic representation of the Xic with the position of the primers (represented as asterisks) used in our RT-PCR analysis. Enlargement of the region from Jpx to Ftx, including the B region, is shown below with the corresponding positions of primers. (B) Random RT-PCR analysis of the Xic in WT (black) and Ftx mutant clones (red). For each position analyzed, data were normalized to Arpo PO transcript levels, and value of the WT was set to 1 (n= 2–5; for Ftx mutants, results obtained for the three clones were pooled). (C) Strand-specific RT-PCR analysis of positions 20 and 17 within region B in WT (black) and Ftx mutant clones (red) shows that only sense transcript (in Ftx orientation) is affected by Ftx mutation at these two positions (n= 2).
Ftx deletion leads to decreased expression and increased DNA methylation at Xist
We showed that the effect of Ftx promoter deletion on transcription is gradual, with the strongest effect observed for positions within or close to the Ftx gene. One notable exception is Xist, which displays strongly reduced RNA levels (70%) as a result of Ftx mutation despite its location some 150 kb from Ftx 5′ end. This contrasts with the two flanking genes Jpx and Tsix, for which a reduction of only 30% is observed (Fig. 6B). Importantly, this is, to our knowledge, the first example of a context in which Tsix and Xist are affected in the same direction. Since Tsix downregulation itself is known to lead to the elevation of Xist RNA levels, it is, in addition, likely that the effect we observe at Xist in Ftx mutant cells is being underestimated. These results identify Ftx as a novel element participating in the control of Xist RNA levels in ES cells.
In order to decipher the underlying molecular mechanisms and to tease apart direct from indirect effects, we investigated whether the Ftx mutation causes any modification in the Xist promoter chromatin structure. Re-analysis of the Xic-wide histone modification profiles focusing on the 4 kb region spanning the Xist transcription start site did not reveal any significant changes at the levels of H3K9 di- and H3K27 tri-methylation. However, we observed a slight but significant decrease in H3K4me2 within the first 2 kb of Xist exon 1 (Fig. 7A), a region that corresponds to a CpG island (CpG-22 in Fig. 7C, position 145 183 to 146 442). H3K4 methylation has been shown to be anti-correlated to DNA methylation and to prevent the action of the DNA methyltransferase Dnmt3L (23). This prompted us to analyze DNA methylation at Xist and at the level of the entire Xic in WT and Ftx mutant cells by MeDIP (methylated DNA immunoprecipitation).
Ftx impacts on Xist chromatin structure. (A) Enlargement of the Xist 5′ region for the H3K4me2 data shown in Figure 5. The green box below represents the start of Xist exon 1 and the black asterisks symbolizes CpG island. Note that only the first 3.2 kb of Xist exon 1 is shown. (B) DNA methylation analysis by MeDIP of the Xist 5′ region as in (A). (C) Xic-wide MeDIP analysis of WT (black) and Ftx mutant (red) ES cells, from which data presented in (B) were extracted (n = 2). Results were analyzed as in Figure 5. The map of the region is shown below the graph, with CpG islands represented as asterisks.
As reported previously (24), the Xist CpG island displays substantial levels of methylation in WT ES cells which are controlled, at least in part, by Tsix (7). Importantly, these levels are increased significantly in Ftx mutant cells (Fig. 7B), whereas no changes could be detected for all the other CpG islands that were included in our analysis as well as for the rest of the Xic (Fig. 7C). Two ncRNAs, therefore, appear to have antagonist effect on Xist promoter methylation: Tsix promotes DNA methylation (7), whereas Ftx prevents hypermethylation. Since Ftx mutation also leads to Tsix downregulation, the DNA hypermethylation we observe at Xist in the absence of Ftx is again likely to be underestimated.
DISCUSSION
XCI is a complex, developmentally regulated epigenetic process in which untranslated transcripts play central role. Here, we have characterized Ftx, a non-coding gene of the Xic which hosts several miRs. Our comparative analysis revealed that the Ftx gene and the embedded miRNAs are conserved among eutherian mammals. It has been suggested that the emergence of the Xic is a relatively recent event in the evolution of sex chromosomes and that genes of the Xic evolved, at least in part, as a result of pseudo-genization of protein-coding genes (25). Indeed, Xist derives from Lnx3, a protein-coding gene whose function is still unknown. Other non-coding transcripts of the Xic also appear to have protein-coding gene ancestors, and Ftx was shown to have significant homology to the Wave4 (Wasf3) gene (26). It is interesting to note that the most striking homology is found for exons 8 and 9 of the murine Ftx (where both sequence and exon-intron junctions are conserved), which are among the most conserved exons in eutherian mammals.
The function of the Ftx miRs is currently unknown; however, ATM was recently identified as a putative target for miR-421. Interestingly, disruption of ATM function, by RNAi or through the use of a specific inhibitor, leads to heterochromatin and silencing defects on the inactive X chromosome (27), thus providing the first molecular connection between Ftx and XCI. Other long non-coding genes have been described as encoding miRs (28), and strikingly, several of them are located within clusters of imprinted genes (29). This raises the intriguing question as to whether these long non-coding genes have a function beyond serving as a primary transcript of miR. In the case of Ftx however, the conservation of Ftx exons across evolution suggests that Ftx is a functional RNA and has a role beyond the production of the embedded miRs.
Ftx is located within a larger region of the Xic which harbors both peculiar chromatin features reminiscent of facultative heterochromatin and complex non-coding transcription in both sense and antisense orientations. It has been shown in fission yeast and plants that forward and reverse transcription participates in heterochromatin formation at centromeres and other loci through a small interfering (si)RNA-dependent pathway (30). So far, no such direct connection between RNAi and chromatin modification has been made in vertebrates, but it was tempting to speculate that a related process could be involved in the establishment of the hotspot of methylation that we have identified 5′ to Xist. Although this hypothesis cannot be ruled out, we have clearly shown that Ftx transcription is not necessary for the maintenance of the heterochromatic marks of the hotspot. It remains possible, however, that Ftx participates to the hotspot formation during embryonic development, a step that cannot be addressed in our ES cell model system.
Although the overall chromatin profile of the hotspot was unaffected by Ftx deletion, some local modifications were observed. In particular, repressive histone marks were accumulating within the Ftx locus in mutant cells. This is reminiscent of what we have recently described for Tsix mutants, in which H3K27me3 was shown to spread from the hotspot to the Xist/Tsix region (22). In this case however, the spreading was shown to concern H3K27me3, but not H3K9me2. Whether the mechanisms by which Ftx and Tsix block the accumulation and/or spreading of repressive histone marks are the same remains to be investigated. It is, however, interesting to note that this is not a general property of ncRNA, as Jpx, which is located in between Tsix and Ftx, does not appear to display such activity; indeed, H3K9 and K27 methylation levels within the Jpx genomic region are similar to other parts of the hotspot.
One of the most striking characteristic of Ftx is its specific upregulation during female ES cell differentiation, which strongly correlates with XCI. Moreover, we have shown that Ftx impacts on the chromatin structure of the Xist promoter (by preventing DNA hypermethylation and promoting accumulation of H3K4 di-methylation) and regulates, directly or indirectly, Xist expression. We propose that Ftx acts as a novel positive regulator of Xist and an activator of XCI, a hypothesis that ultimately needs to be validated through the analysis of X-inactivation in Ftx mutant female embryos or ES cells. Ftx is, to our knowledge, the first element of the Xic shown to favor Xist expression independently of X-inactivation. Xpr and Rnf12 are two other loci recently reported as playing activating function on X-chromosome inactivation (10,11), but their effect on Xist has not been directly addressed. Interestingly, both the Xpr/Xpct region and Rnf12 are located on the X chromosome, telomeric to Xist and Ftx, and all four genes are transcribed in the same orientation. Whether this specific organization is random or not is unknown, but this raises the intriguing possibility that Ftx is an intermediate in the cascade of events leading from Rnf12 and/or Xpr to Xist activation (Fig. 8A). Alternatively, Ftx may exert its function independently of Rnf12 and Xpr (Fig. 8B). This is in agreement with the observation that Rnf12+/− female ES cells are still able to initiate XCI, indicating the existence of additional XCI activators (10).
A model for Ftx action on Xist. Xist is negatively controlled by the combined action of Tsix and the pluripotency factors Oct3/4, Nanog and Sox2 (schematized by red arrows). So far, very little is known concerning the positive regulation of Xist. The Xpr region and Rnf12, both located upstream of Xist, were described as positive regulators of X-inactivation, which may exert their function through direct Xist activation. Here, we have shown that Ftx is a positive regulator of Xist, which is located in between Rnf12/Xpr and Xist. This peculiar genomic organization together with the positive effect that these three loci exert on Xist expression/X-inactivation raises the possibility that Ftx is an intermediate factor in the cascade of event leading from Rnf12/Xpr to XCI activation (A). Alternatively, Rnf12, Xpr and Ftx may represent three independent pathways acting in concert to promote X-inactivation (B).
In conclusion, Xist appears to be controlled at least in part by two cis-acting ncRNAs exerting antagonistic actions. On the one hand, Tsix acts as a repressor and promotes accumulation of repressive histone marks (including DNA methylation) at the Xist promoter. On the other hand, Ftx has a positive effect on Xist expression and inhibits DNA methylation at its promoter. Understanding the orchestrated action of this triumvirate of ncRNAs will undoubtedly bring new light on the early regulatory steps of the X-chromosome inactivation process.
MATERIALS AND METHODS
In silico analysis of Ftx ESTs and conservation
To identify the exon–intron structure of the Ftx gene in the mouse, human and cow, alternative spliced ESTs overlapping with the Ftx locus were extracted from UCSC Genomic Database (http://genome.ucsc.edu/). The genomic EST alignment was performed using the following assemblies: mouse, July 2007 (NCBI/mm9); human, February 2009 (GRCh37/hg19) and cow, October 2007 (Baylor 4.0/BosTau4). Six ESTs were analyzed in the mouse (AK035525, BC027389, AK085125, BC076616, BB619145, AK020989), six in the human (AK057701, BM546361, AK311345, BF678669, BG718066, CN412172) and seven in the cow (DY165977, EE361536, EE373422, DY455742, DN281600, DN276300 and EH183043).
The comparative analysis of the Ftx gene to identify homologous exons was performed using the bl2seq software at NCBI (http://www.ncbi.nlm. nih.gov/). The alignment was performed using the blastn algorithm and +1/−1for match/mismatch, 1 for Gap existence and 2 for Gap extension as parameters.
To compare the mouse and human Ftx genes, genomic sequences were first analyzed with RepeatMasker to identify and mask repeated elements (A.F. Smit, unpublished; http://www.repeatmasker.org/). Pairwise local alignments of genomic sequences were performed with SIM (31), using the following scoring scheme: match = 1, mismatch = −1, gap opening penalty = 6 and gap extension penalty = 0.2. Only local alignments with a score >20 were retained. Pictures of pairwise local alignments were obtained with LALNVIEW (32). CpG-rich regions were identified with CpGproD (33).
Cell culture
ES cell lines were grown in DMEM, 15% FCS and 1000U/ml LIF (Chemicon). Female PGK1 ES cells were cultured on gelatin-coated plates in the absence of feeder cells. Male TC1 and TC1-derived Ftx mutant ES cells were cultured on mitomycin C-treated male embryonic fibroblast feeder cells, which were removed by adsorption before chromatin and RNA extractions. To induce ES cell differentiation, cell lines were plated on gelatin-coated flasks and cultured in DMEM, 10% FCS, supplemented with retinoic acid (RA) at a final concentration of 10−7m. The medium was changed daily throughout differentiation. RA-treated ES cell lines exhibited morphological features of differentiated cells. Differentiation was also evaluated by analysis by real-time PCR of Oct3/4 and Nanog expression (data not shown).
Subcellular fractionation
Cell fractionation of undifferentiated and differentiated TC1 (XY) and PGK1 (XX) ES cells was performed using the PARIS kit (Ambion) as described in the manufacturer's protocol. RNA (1 μg) from each fraction was reverse transcribed using a QuantiTect reverse transcription kit (Qiagen). Q-PCR was performed with the 7900 HT Fast-Real Time PCR System (Applied Biosystem) under the following cycling conditions: 2 min 50°C, 10 min 95°C, 40 cycles of 15 s 95°C and 1 min 60°C. The Q-PCR was carried out using the following primers: Gapdh (Up: 5′-GTATGACTCCACTCACGGCAAA-3′; Lo: 5′-TTCCCATTCTCGGCCTTG-3′), β-actin (Up: 5′-ACGTTGACATCCGTAAAGACCT-3′; Lo: 5′-GCAGTAATCTCCTTCTGCATCC-3′), β-actin pre-mRNA (Up: 5′-CAGCTTCTTTGCAGCTCCTT-3′; Lo; 5′-CATGGTGTCCGTTCTGAGTG-3′), Xist (Up: 5′-GGATCCTGCTTGAACTACTGC-3′; Lo: 5′-CAGGCAATCCTTCTTCTTGAG-3′), Tsix (Up: 5′-GTGATGGAAGAAGAGCGTGA-3′; Lo: 5′-GCTGCTTGGCAATCACTTTA-3′), Ftx7-8 (Up: 5′-TATGCCACCTAGCCTTTCTACA-3′; Lo: 5′-ATCTCTTCAAAAGCGGCATAAT-3′), Ftx8-9 (Up: 5′-GCCGCTTTTGAAGAGATAAACT-3′; Lo: 5′-TGATTATCCTCATGTGTTGCTG-3′).
RNA extraction
To analyze Ftx expression in the adult mouse, organs were dissected from 9-week-old C57Bl6/J mice (two males and two females) and homogenized in TRIzol (Invitrogen) using a cell homogenizer. RNA was subsequently extracted according to the manufacturer's protocol and treated with TURBO DNAse (0.2 U/μg, Ambion) for 30 min at 37°C. Following two steps of phenol extraction, RNA was precipitated and resuspended in DEPC-treated water. RNAs were quantified using a NanoDrop 1000 (Thermo Scientific).
To extract RNA from cell culture, cell pellets were rinsed in PBS and directly resuspended in TRIzol.
Expression analysis by qPCR and northern blot
Random-primed RT was performed at 42°C using Vilo reverse transcriptase (Invitrogen) with 0.5 to 1 µg of total RNA isolated from organs or cell cultures. Control reactions lacking enzyme were verified negative unless specified otherwise. qPCR measurements using SYBR Green Universal Mix were performed in duplicate. The Ftx primers used span exons 8, 9 and 10 (FtxE8.9Up: 5′-CTTGATTCAGCAACACATGAGGA-3′; FtxE10.9Lo: 5′-TCCAGGCAAGAGGGACCAG-3′). Xist primers were described previously (34).
The other primers used are:
Nap1l2: 5′-CAGACCGTCCAAAAGGACTTA-3′/5′-AGTAAGGGTTGGTACATTTCAG-3′
Cdx4: 5′-AGCTCCAGCCGTTGATGATCT-3′/5′-CATGGATGCGCAAAACTGTG-3′
Chic1: 5′-GATATGCCCCAGACCCTGTGT-3′/5′-GCCACCTTCCCTGTCAGAACT-3′
Tsx: 5′-GATCAGGCAGATGAGCCACTG-3′/5′-GCAAGTGTCCTCATCATCGGT-3′
Tsix: 5′-AGAGCTTTGCGAGTTTGAGG-3′/5′-CTTTGGTTTTGATGCGGATT-3′
Jpx: 5′-ATAAAATGGCGGCGTCCAC-3′/5′-GGCCAGTTTCTCCACTCTCCT-3′
24: 5′-CGAAAACAACTATGCTGGCATG-3′/5′-GGAAGACCTCCATAGCTGGCAT-3′
21: 5′-ACAGGCTGTGAACCAGAGTACC-3′/5′-ACAGGAATGTAGGATTCACCAA-3′
19: 5′-GGGGCAGAGCCTAGACCAAA-3′/5′-CACTGGCATCCCCATTAGGG-3′
18: 5′-CAGAATGTTTCAATTCCAAGGA-3′/5′-TTGGGAAATTCATGGAAGTATG-3′
17: 5′-ACTTAACACCCACCGCCTCTG-3′/5′-AGTGACTGCAGGATGCCAACC-3′
16: 5′-TCTTTCTCCCTTGTGGGTGATG-3′/5′-ATGTGGAGGTCTGTGGACCATA-3′
15: 5′-CCCTCAGGCCCAAATATCAG-3′/5′-AAATCCACTGGCCACCCTCT-3′
9: 5′-GGCGGAATTTAAGGGCTGTAA-3′/5′-CACGCTACCCCATTGTTGGT-3′
Cnbp2: 5′-TCTGGATGTGCCCAAATTCAC-3′/5′-ATCTGCAGTCAGCCTGGTCAT-3′
Xpct: 5′-GCCCTTGGTTACTTCGTCCC-3′/5′-ACACCAAGAGCACCCAGGTC-3′
Arpo P0 transcript levels were used to normalize between samples. Rrm primers used in Supplementary Material, Fig. S6 were the following: Rrm1F: 5′-CCGAGCTGGAAAGTAAAGCG-3′; Rrm1R: 5′-ATGGGAAAGACAACGAAGCG-3′.
For miR expression analysis, 1 μg of total RNA was reverse transcribed using the miScript Reverse Transcription Kit (Qiagen), according to the manufacturer's protocol. cDNAs were diluted five times, and 2 μl of this dilution were used for real-time PCR, using the miScript SYBR Green PCR Kit (Qiagen). Mature miRNAs were amplified using specific primers for the detection of mmu-miR-421 (Mm_miR-421_1: AUCAACAGACAUUAAUUGGGCGC) and mmu-miR-374 (Mm_miR-374-5p_1: AUAUAAUACAACCUGCUAAGAG), together with a universal primer provided with the kit. PCR reactions were conducted at 95°C for 15 min, followed by 40 cycles comprising three steps of denaturation (95°C, 15 s), annealing (55°C, 30 s) and extension (70°C, 30 s), on a Light Cycler 7900 HT (Applied Biosystems).
Mouse multiple northern blot was obtained from Clontech Laboratories and hybridized according to manufacturer recommendations with an Ftx probe corresponding to exons 8, 9 and 10.
Strand-specific RT
Strand-specific RT was performed at 50°C using SuperScript reverse transcriptase (Invitrogen) with 0.5 μg of total RNA using specific primers (position 17: 5′-GCTGCCCCAGAAACTTAACA-3′ for sense orientation and 5′-GATGGGGAGGGACTTCATTT-3′ for antisense orientation; position 20: 5′-GGAGGAACCTTTCCCCTGT-3′ for sense orientation and 5′-CAACCAATTGGAGCAGGACT-3′ for antisense orientation) and the Arpo primer (5′-TGCGGACACCCTCCAGAA-3′) for normalization. Control reactions lacking enzyme or primer were verified negative. Reactions of controls, using the specific primer without the Arpo primer and vice versa, were also performed to verify the absence of cross-reaction. qPCR measurements using SYBR Green PCR Master Mix (Applied Biosystems) were performed in triplicates and standardized against Arpo cDNA (7).
Allele-specific RT-PCR
Allelic Xist and Nap1L2 cDNA quantitations in XEN and TS cells carrying a 129 allele and a Pgk allele were carried out as described previously (20,35). Two informative SNPs identified in exon 15 of Ftx between the 129 and Pgk alleles were used to design allele-specific primers: 5′-AGATGGCACAGTGTCAGAAACG-3′/5′-TGCAGATGGTGTGAAGCTGC-3′, which are specific for the 129 allele; and 5′- CAGATGGCACAGTGTCAGAAACA-3′/5′-TGCAGATGGTGTGAAACTGC-3′, which are specific for the Pgk allele. All cDNA quantifications using SYBR Green PCR Master Mix were standardized against Arpo cDNA.
RNA-FISH
The p510 Xist probe used was a 19 kb genomic fragment that covers most of the Xist gene (34). The Ftx probe was a mix of the BAC D NeoDTA-ΔFtx used to target ES cells and a plasmid covering the deleted region generated by gap-repair cloning from the BAC 561P13. Visualization was performed with a Zeiss Axioplan epi-fluorescence microscope and a Hamamatsu CCD camera. Image acquisition was done with Volocity software.
ES cell targeting
The targeting vector BAC D NeoDTA-▵Ftx used for homologous recombination was produced through two rounds of recombineering in bacteria DY380. The first one was to exchange a 12 kb region in 5′ part of Ftx (coordinates of the deleted sequence in UCSC Genome Browser on the mouse, July 2007, NCBI/mm9 Assembly: chrX: 100807185–100819547) from the BAC 561P13 (from the 129/Sv mouse strain, BAC Library, Research Genetics, Inc.) with a PGK/Tn5 neomycin-polyA-floxed cassette (36). Briefly, we generated a construct in which short homology arms designed to surround the targeted Ftx sequence have been cloned on either sides of a PGK/Tn5 neomycin cassette. These were obtained by PCR amplification using primers containing appropriate restriction sites for the cloning and the loxP sequences.
For the left arm: HindIII-F AAAAAGCTTTGGTTCTGGTGCTTTGATGCT/ClaI–loxP-R AAAATCGATATAACTTCGTATAATGTATGCTATACGAAGTTATAAATAGCCTCCACCGATCCG; for the right arm: AscI–loxP-F AAAGGCGCGCCATAACTTCGTATAGCATACATTATACGAAGTTATAAGCATGGCTTCTGCTCATTG/HindIII-R AAAAAGCTTTTCAACCTGGGCTTCTCTGTG.
This construct was used to transform electrocompetent DY380 cells carrying BAC 561P13 and made inducible for recombineering [by heat shock induction of the λ Red recombination proteins (37)]. Recombinant cells were selected for their resistance to kanamycin, and the resulting vector was analyzed by restriction digestion.
The second round of recombineering has been done to shave the BAC and to introduce a negative selection cassette by replacing 90 kb from the BAC obtained after the first recombineering with a PGK-DTA (diphtheria toxin) cassette. The two homology arms used for this step were obtained by amplification using the following primers: for the left arm: NotI-F AAAGCGGCCGCTGGCTTAACTATGCGGCATCA/NdeI-R AAACATATGTTACAATTCACTGGCCGTCGT; for the right arm: BglII-F AAAAGATCTGGTTCCAAGTGCTATAATCAATTCC/NotI-R AAAGCGGCCGCGAGTACAAATTACTAACAGCTAGGCTTGC. Recombinant cells were selected for their resistance to zeocyn (Cayla).
The entire structure of the resulting targeted vector BAC D Neo-DTA ▵Ftx, which contains a 3 kb 5′ homology arm and a 50 kb 3′ homology arm, was then extensively analyzed by restriction digestion, and sequences surrounding the replaced regions were also sequenced. The BAC D NeoDTA-▵Ftx was then linearized using SrfI and electroporated in the TC1 ES cell line (Xcell Gene Pulser; BioRad). After selection with 250 mg/ml G418 (Invitrogen), neomycin-resistant clones were screened by PCR. Seven positive clones were obtained out of 135 and were checked for normal karyotype and analyzed by Southern blot. Three of them were lipofected (Lipofectamin 2000; Invitrogen) with the cre-expression plasmid pOG231 to excise the selection cassette. ES cells lines obtained were further analyzed by Southern blot and checked for normal karyotype.
Southern analysis
Southern analysis was performed by digesting 10 µg of genomic DNA from control TC1 and recombinant TC1ΔFtx ES cell lines with NcoI, which generates an 8.7 kb restriction fragment for non-recombinant clones and a 6.6 kb restriction fragment for recombinant clones. Hybridization was performed with a probe designed outside the homology region used for recombination and obtained from PCR products using primers 5′-ATTCATGGCAATCTAACCTCTTGAC-3′ and 5′-TTTTAGTTGGGAAGCTGGTGCT-3′.
Chromatin and MeDIP assays
ChIP and MeDIP assays were performed as described previously (7,38). Antibodies H3K4me2 (Upstate Biotechnology), H3K9me2 and H3K27me3 (Abcam) were used at 1/125 dilution to immunoprecipitate an equivalent 20 µg of DNA in ChIP assays. Monoclonal antibody against 5-methylcytidine (Eurogentec) was used to immunoprecipitate 4 μg of genomic DNA. Each assay was performed two to six times on independent chromatin/DNA preparations to control for sample variation. To standardize between experiments, we calculated the percentage of immunoprecipitation by dividing the value of the IP by the value of the corresponding input, both values first being normalized for dilution factors.
Real-time PCR analysis of ChIP assays
To analyze ChIP and MeDIP experiments, real-time PCR assays were performed in 384-well plates. The primers used for the ChIP assays were designed automatically to produce 90 to 140bp amplicons that cover the Xic region spanning from position 100 532 247 to 100 832 343 on NCBI Build 37 (22). A program was written with the aim of producing high-quality primers and maximizing coverage of the region (available upon request). The program accesses our local mouse genome database in order to extract the sequence designated as a template, and to mask all sequence variations and repeats based on the NCBI Mouse Genome Build Release 37. This template is passed to Primer 3 with the following minimum, optimum and maximum values applied: GC contents of 30, 50 and 80%; and Tm of 58, 60 and 61°C. The resulted primers are further analyzed by sequence homology using BLAST and the mouse genome as the query sequence to eliminate those that can produce more than one PCR product, and filter out those where the log base of 10, of counts of significant hits per pair, exceeds 2. Besides the GC contents and Tm values already mentioned, our program also screens for long runs of identical nucleotides, and G/C stretches at the 3′ ends. For all primer pairs, PCR efficiencies were verified to be similar.
Liquid handling of the 384-well plate was performed with a Baseplate robotic workstation (The Automation Partnership, UK). The composition of the qPCR assay included 2.5 µl of DNA (the IP DNA or the corresponding input DNA), 0.5 µm forward and reverse primers, 1X Power SYBR® Green PCR Master Mix (Applied Biosystems). The amplifications were performed as follows: 2 min at 95°C, 40 cycles at 95°C for 15 s and 60°C for 60 s in the ABI/Prism 7900HT real-time PCR machine (Applied Biosystems). The real-time fluorescent data of qPCR were analyzed with the Sequence Detection System 2.3 (Applied Biosystems).
SUPPLEMENTARY MATERIAL
Supplementary Material is available at HMG online.
FUNDING
This work was supported by the Agence Nationale pour la Recherche (ANR, contract number 05-JCJC-0166-01); INSERM (Avenir program R0721HS) and the European Research Council under the European Community's Seventh Framework Program (FP7/2007-2013)/ERC grant agreement no. 206875 ncRNAx.
ACKNOWLEDGEMENTS
We wish to thank Deborah Bourc'his for providing mouse organs and for stimulating discussion, and Constance Ciaudo for technical and scientific advice. We also thank Jonathan Weitzman and Ute Rogner for critical reading of the manuscript.
Conflict of Interest statement. None declared.
