An essential role for the DXPas34 tandem repeat and Tsix transcription in the counting process of X chromosome inactivation
-
Communicated by Mary F. Lyon, Medical Research Council, Oxon, United Kingdom, March 23, 2006 (received for review February 7, 2006)
Abstract
A counting process senses the X chromosome/autosome ratio and ensures that X chromosome inactivation (XCI) initiates in the early female (XX) embryo and in differentiating female ES cells but not in their male (XY) counterparts. Counting depends on the X inactivation center (Xic), which contains the Xist gene encoding a nuclear RNA, which coats the inactive X chromosome and induces gene silencing. A 37-kb sequence lying 3′ to the Xist gene is known to prevent initiation of XCI in male differentiating ES cells. This region contains the major and minor promoters of the Tsix gene, which runs antisense to Xist, and the DXPas34 tandem repeat lying close to the Tsix major promoter. We have addressed the role of these elements in counting by using male ES cells. Targeted deletion of DXPas34 impaired recruitment of RNA-polymerase II and TFIIB at the Tsix major promoter, resulting in low levels of Tsix expression in ES cells and moderate ectopic initiation of XCI upon differentiation. A deletion extending 3′ to Xist and including the Tsix major promoter resulted in almost complete impairment of Tsix transcription and in efficient ectopic XCI upon differentiation of male ES cells. Internal deletions within the Tsix gene did not affect significantly the level of antisense transcription within Xist and had only minor effects upon differentiation. Our results identify a function for DXPas34 in murine XCI and demonstrate the critical role of Tsix transcription in preventing XCI in differentiating male ES cells and in normal functioning of the counting pathway.
In mammals, X chromosome inactivation (XCI) ensures dosage compensation of X-linked genes between females (XX) and males (XY) through silencing of one of the X chromosomes in female cells (see ref. 1 for review; see also refs. 2–4). The parentally imprinted form of XCI results in the paternal X being first silenced at the four-cell stage but reactivated in the inner cell mass (ICM) before de novo random XCI in the differentiating epiblast from which the somatic tissues originate (5–7). ES cells, which are derived from the ICM, recapitulate counting, choice, and silencing upon in vitro differentiation (8–12). Consistently, XCI is inducible in XX but not in XY ES cells upon differentiation.
Proper regulation of random XCI depends on the implementation of the counting and choice mechanisms, both poorly understood. Counting evaluates the X/autosome ratio and controls the decision to initiate or not the silencing process. It is thought to ensure that XCI does not occur in XY cells, and that a single X is silenced in diploid XX cells. Choice controls the mutually exclusive designation of one active X (Xa) and one inactive X (Xi) in each female cell. Clearly, counting and choice must involve in trans X–X and X–autosome interactions. Highly relevant to the X–X interactions may be the observation that a transient pairing between specific X-linked loci occurs at the onset of XCI (13, 14).
Silencing of X-linked genes is induced by the accumulation on the Xi elect of the Xist nuclear RNA (see ref. 4 for review), which recruits enzymatic complexes able to ensure the heterochromatinization of the Xi (reviewed in ref. 15). The Xist gene maps to within the Xic (X inactivation center), an X-linked locus spanning several hundreds of kilobases, which includes, in addition to Xist, genomic elements involved in counting and choice. In female ES cells, a heterozygous deletion spanning the 65-kb region located immediately 3′ to Xist has been shown to result upon differentiation in every cell choosing the mutated X as the Xi and the unmutated X as the active Xa (8). Confirmation that choice had indeed been affected by this mutation came subsequently from the detailed identification of genomic elements within this 65-kb span (16–19). The 65-kb deletion, however, is also associated with a counting defect, because ectopic initiation of XCI occurs in XY differentiated ES cells carrying the deletion (20). Reestablishment of normal counting through site-specific cre-loxP reinsertion of DNA from within the 65-kb span into the deleted locus refined the candidate region for the counting function to a 37-kb sequence lying immediately 3′ to Xist (20).
Among the elements that lie within this 37-kb region is the 5′ end of the Tsix gene. The Tsix gene is transcribed in antisense orientation to Xist and appears to be an important regulator of XCI. Tsix encodes a spliced nuclear transcript (16, 18), which, however, seems of lesser functional importance than the primary transcript or the act of Tsix transcription itself (21, 22). Tsix transcripts initiate from both a proximal major promoter and a distal minor promoter (16, 18, 19). DXPas34, a 1.2-kb CG-rich tandem repeat, lies 750 bp downstream from the Tsix major promoter. This element has a unique DNA methylation profile marking specifically the active X after the onset of XCI (23, 24). DXPas34 also contains numerous binding sites for the nuclear factor CTCF of putative functional signification (25). Despite the interesting features associated with this element, its function has yet to be addressed by a deletion specifically targeted to it.
Tsix transcription has been shown in ES cells to antagonize Xist expression (26–28) and is primarily involved in regulating Xist at the onset of XCI, because these two transcripts show inverse regulation on the elect Xi (16). Tsix appears also to mark the Xist/Tsix locus for deposition of histone H3-dimethylK4 (29). Several reports based on targeted mutagenesis have shown that Tsix plays a role in choice (17, 18, 26, 30), with the X chromosome impaired in Tsix transcription being majorly or exclusively chosen as the Xi elect. However, a deletion targeting the Tsix major promoter in male ES cells was reported not to result in ectopic XCI and not to affect counting (17). This latter result sits uncomfortably with others, suggesting a function in counting for Tsix DNA sequences based on transgenic experiments in differentiating female ES cells (31). Crucial to the interpretation of these experiments, but remaining to be determined, is whether the inhibition of XCI in females and the induction of ectopic XCI in males reveal identical or distinct mechanisms within the counting pathway.
In addition to Tsix, several other genomic features are also found in the region lying 3′ to Xist. A 9-kb bipartite region centered on the Tsix major promoter has been shown in episomal luciferase reporter assays to contain DNA elements capable of enhancing transcription from the Tsix promoter (32). Xite, located between the major and minor Tsix promoters, contains both a cluster of DNase hypersensitive sites and a bidirectional promoter of intergenic transcription (19, 32). Xite has also been suggested to be involved in choice by targeted deletion experiments (19) and in counting through transgenesis experiments in female cells (31).
One of the first models of counting postulated the existence of a blocking factor present in quantity just sufficient to prevent the inactivation of a single X per diploid cell (33). This model was supported by the phenotype associated with the 65-kb deletion, which similarly resulted in inactivation of the deleted X in XX, XO, and XY ES cells (8, 20). More recently, an additional competence factor specific for XX cells, required for inactivation to occur, has been postulated on the basis of the different in cis effects of Tsix transcription impairment on XCI reported in male and female differentiated ES cells (17).
Here we report on the functional characterization of elements from within the Tsix gene with respect to counting based on the transcriptional analysis of the Xist/Tsix locus in a series of targeted deletions in male ES cells. We demonstrate that, in male ES cells, the 1.2-kb tandem repeat DXPas34 is essential for the initiation of transcription from the major Tsix promoter and for the repression of XCI upon differentiation. Through the analysis of a series of nested targeted deletions within the Tsix transcription unit in male ES cells, we show that the Tsix genomic elements located 3′ to Xist do not play a significant role in Xist regulation at the onset of XCI unless they impair Tsix transcription. Reanalysis of the previously described Ma2L ES cell line (26), in parallel with our deletions, confirms that Tsix transcription is necessary for preventing ectopic XCI in differentiated male ES cells. These results show that Tsix transcription is an essential regulator of the counting pathway. Our observation that impairment of Tsix transcription has a similar in cis effect in both male and female cells makes it unnecessary to postulate the existence of a female-specific competence factor in XCI.
Results
Targeted Deletion of DXPas34 Affects Tsix Expression at the Transcriptional Level.
We precisely targeted the 1.2-kb DXPas34 tandem repeat for deletion in ES cells (Fig. 1 A), because its function has until now been analyzed only in the context of larger deletions (17, 18, 30). DXPas34 was first replaced by a floxed pgk-Neo cassette, which was subsequently removed by cre expression. Southern blot analysis (Fig. 5, which is published as supporting information on the PNAS web site) confirmed the genomic structure of the targeted locus.
The DXPas34 tandem repeat enhances Tsix expression in male ES cells. (A) Map of the DXPas34 locus. Previous and new deletions of DXPas34 are shown with the nomenclature of the deleted cell lines on the right. Blue marks, CTCF-like sites. (B) Map of the Xist/Tsix locus. RNA-FISH probes and PCR primers used are indicated. (C) Normalized qRT-PCR analysis of Xist and Tsix (2−CT Xist or Tsix/2−CT Arpo, n = 3). Xist levels are increased ≈7-fold, whereas Tsix transcripts at positions downstream from the Tsix major promoter are reduced 10-fold in the Δ34#1 and #2 cell lines. No effect was detected upstream of the Tsix major promoter (Tpg aud Upt1). CTs for RT− reactions were at least nine cycles greater than RT+. Two other culture samples for each line were in the same range. (D) A Tsix RNA-FISH signal of reduced intensity is detected in the nuclei of the Δ34#1 line. White arrowheads identify the position of the signals. (E) ChIP analysis showing reduced levels of H3dimethyl K4 within the Tsix gene in the Δ34#1 and #2 lines. (F) ChIP analysis showing reduced levels of TFIIB within the Tsix major promoter and of RNA PolII within the major promoter and the body of the Tsix gene in the Δ34#1 and #2 lines. In E and F, IPs and inputs were analyzed in triplicate by qPCR. The Hprt and Arpo promoters and X3 give ChIP controls. Duplicate experiments on separate culture samples gave similar results.
Quantitative RT-PCR (qRT-PCR) analysis showed that Tsix RNA steady-state levels were uniformly reduced 9- to 10-fold at all positions tested downstream of the major Tsix promoter in both Δ34#1 and #2 cell lines, as compared with the parental cell line CK35 (Fig. 1 C). In contrast, no significant effect was observed on the steady-state levels of transcripts upstream from the major promoter (Fig. 1 C). RNA-FISH analysis using a probe specific for Tsix showed a pinpoint signal of reduced intensity in most nuclei of the Δ34#1 cell line (Fig. 1 D) and Δ34#2 cell line (not shown). We thus conclude that DXPas34 exerts a positive effect on Tsix expression derived from the major promoter of this gene.
We observed both a 6- to 7-fold increase in spliced Xist RNA expression levels (Fig. 1 C) and a significant reduction of the levels of H3-dimethylK4 within the Tsix/Xist locus (Fig. 1 E) in the Δ34#1 and #2 ES cell lines as compared with the CK35 ES cell line. It is known that Tsix antagonizes Xist expression in ES cells (26–28) and mediates the deposition of H3-dimethylK4 within the Tsix/Xist locus (29). In this context, we believe that the impact of deleting DXPas34 on Xist expression levels and chromatin structure of the Xist/Tsix locus in ES cells are mostly, if not completely, mediated by a reduction in the steady-state levels of expression of Tsix. Chromatin immunoprecipitation (ChIP) analysis was performed to measure the recruitment of the general transcription factor TFIIB and of RNA–polymerase II (PolII) at the Tsix major promoter. A significant reduction in the presence of these factors was found at the Tsix major promoter in the Δ34 line as compared with the CK35 parental line (Fig. 1 F). Similarly, the amount of PolII associated with the body of the Tsix gene was reduced 3- to 5-fold (Fig. 1 F), a level compatible with the lowered steady-state level of Tsix RNA. We conclude that the deletion of the DXPas34 tandem repeat in ES cells affects Tsix at the transcriptional level.
Ectopic Xist Up-Regulation in Differentiated Male ES Cells Deleted for DXPas34.
The regulatory effects of DXPas34 in ES cells prompted us to study Xist expression in the Δ34 ES cell lines upon differentiation. In several differentiation experiments, qRT-PCR analysis of mature Xist transcripts consistently showed a >10-fold increase in steady-state levels of Xist transcript between days 2 and 4 of retinoic acid treatment in both cell lines (Fig. 2 A). Such an up-regulation of Xist expression, although highly significant when compared with differentiated CK35 cells (Fig. 2 A), is still 5- to 10-fold lower than what is commonly observed in differentiated female ES cells (28). Our qRT-PCR observations correlate well with results from RNA-FISH experiments, which showed formation of Xist domains in 15% of nuclei of the Δ34#1 (Fig. 2 B) and Δ34#2 (not shown) ES cell lines differentiated for 3 days with retinoic acid. These numbers, although contrasting completely with the total lack of Xist domains observed in the differentiated CK35 ES cell line (not shown), are significantly lower than those observed after differentiation of female ES cells in which 30% to 50% of cells show nuclear Xist accumulation (14, 28, 34). We conclude that deleting DXPas34 diminishes (but does not completely abolish) the strict control that normally prevents up-regulation of Xist expression in differentiated male ES cells. Interestingly, leaving the neomycin resistance gene in place following deletion of the DXPAs34 tandem repeat attenuated the ectopic XCI phenotype (Fig. 6, which is published as supporting information on the PNAS web site). The reason for this is unclear, but it stresses the necessity of removing selection cassettes in targeted mutagenesis approaches.
Targeted deletion of DXPas34 leads to ectopic up-regulation of Xist in differentiated male ES cells. (A) qRT-PCR analysis of Xist normalized to Arpo in ES cells differentiated with retinoic acid. Between days 2 and 4, the levels of Xist RNA increased 30-fold in the Δ34#1 and #2 lines as compared with the CK35 line. qRT-PCR of Oct3/4 normalized to Arpo controls for ES cell differentiation. The absolute levels of Arpo were constant during differentiation. Cycle thresholds for RT− reactions were >39 cycles. Two other differentiation experiments provided essentially similar results. This experiment also included the ΔAJ#1, ΔAS#1, ΔAV#1, and Ma2L ES cells (results presented in Fig. 4 A), allowing for a comparison of all cell lines. (B) Xist RNA-FISH (green) on DAPI-stained (blue) nuclei at day 3 of differentiation. Ectopic Xist nuclear domains were present in 15% of the Δ34#1 cells (n > 200) but were never observed in differentiated CK35 cells. Two other experiments provided similar results. (C) qRT-PCR analysis of spliced Tsix normalized to Arpo in ES cells differentiated with RA. PCR assays location are shown in Fig. 1 C. The kinetics of extinction of Tsix are similar in the CK35 and Δ34#1 lines (Left has two different scales for CK35 and Δ34#1). For positions X9 and Te3, see Fig. 7.
A decrease in the steady-state levels of Xist RNA and in the percentage of cells showing nuclear accumulation of Xist RNA was routinely observed after day 4 of differentiation (Fig. 2 A and data not shown). This suggests that silencing and cell death resulting from functional nullisomy for X-linked genes occurred in differentiated Δ34 ES cells.
We also quantified Tsix expression upon differentiation of wild-type and DXPas34 deleted ES cells. The rate of extinction of Tsix transcription during differentiation was not affected by the DXPas34 deletion (Fig. 2 C and Fig. 7, which is published as supporting information on the PNAS web site). Importantly, at all positions tested, the 9- to 10-fold difference in Tsix expression observed when comparing the Δ34 and CK35 ES cells was conserved during differentiation (Figs. 2C and 7). This suggests that the DXPas34 tandem repeat, although strictly required for normal recruitment of general transcription factors at the Tsix major promoter, is not involved in the regulation of Tsix during the differentiation process.
Construction and Characterization of a Series of Targeted Nested Deletions Within the Counting Region.
DXPas34 lies within a 37-kb genomic span previously reported to play a role in counting by preventing ectopic XCI in differentiated male ES cells (20). We therefore wished to determine whether the ectopic Xist up-regulation observed in Δ34 differentiated ES cells might have resulted from reduced levels of Tsix expression or from a more direct cis-acting effect of DXPas34 on Xist. To address these questions, we constructed a series of nested deletions in ES cells targeted to this region using a protocol (Fig. 8, which is published as supporting information on the PNAS web site), which left in place a single FRT sequence. Southern blot analysis (Fig. 8) confirmed the genomic structure of the targeted loci. The ΔAJ deletion (Fig. 3 A) spans a sequence homologous to a region located upstream of the mouse pre-T cell receptor α gene containing an enhancer for this gene (35) and a tandem repeat based on a 17-bp monomer (36). The ΔAS deletion (Fig. 3 A) additionally spans a 2.5-kb DNA sequence with homology to the upstream region of the mouse pTa gene and extends exactly up to, but does not include, DXPas34. The ΔAV deletion (Fig. 3 A) extends further to include DXPas34 and the major Tsix promoter. Two independently derived ES cell clones were subsequently analyzed for each deletion. We also made use of the previously described Ma2L cell line, which carries a transcriptional stop signal inserted between the 17-mer and DXPas34 tandem repeats and lacks Tsix transcription downstream of this site (26).
A series of nested deletions targeting Tsix. (A) Map of the Xist/Tsix region. 17, 17-mer tandem repeat; HpTa, homology with the region 5′ to the mouse α pre-T cell receptor, 34, DXPas34. The ΔAJ, ΔAS, and ΔAV deletions obtained in this study and the insertional mutation Ma2L are indicated in red. (B) qRT-PCR analysis of Xist and Tsix normalized to Arpo in ES cells. PCR assay location is shown in Fig. 1 C. Xist is increased >50-fold, and Tsix is reduced >40-fold in the ΔAV#1 and Ma2L lines as compared with the CK35 line. In contrast, minor effects are evident in the ΔAS and ΔAJ ES cell lines. Two other culture samples for each cell line were in the same range. (C) RNA-FISH (green) in DAPI-stained (blue) nuclei by using probe 510 showed a signal similar to wild-type in the ΔAS#1 (and ΔAJ#1, Fig. 9A). The Xist RNA-FISH pattern in the ΔAV#1 (and Ma2L, Fig. 9A) line shows dispersed nuclear dots.
Only minor effects on the levels of Tsix were detected in the ΔAJ and ΔAS cell lines (Fig. 3 B). The steady-state levels of the Xist transcript were only slightly affected in the ΔAJ and ΔAS deletions (Fig. 3 B). Using a double-stranded probe located within the Xist gene, we observed in the ΔAS (Fig. 3 C) and ΔAJ (Fig. 9A, which is published as supporting information on the PNAS web site) ES cell lines a pinpoint signal very similar in intensity to wild type.
As expected, very low levels of Tsix RNA were detectable in both the ΔAV and Ma2L ES cell lines (Fig. 3 B). The ΔAV ES cell lines showed a 50-fold increase in mature Xist RNA as compared with the parental cell line CK35 (Fig. 3 B). The levels of expression of Xist transcript in the ΔAV ES cell lines are in the same range as the levels measured in the Ma2L cell line (Fig. 3 B). Consistently, in both ΔAV (Fig. 3 C) and Ma2L (Fig. 9A) the Xist RNA-FISH patterns in undifferentiated ES cells were characterized by numerous faint nuclear dots appearing as if irradiating from the Xist locus. This is reminiscent to the RNA-FISH pattern previously reported for ES cells carrying a 65-kb deletion, including Tsix (28). It is to be noted that such pattern is both much less bright and is made up of more individualized and dispersed dots than the Xist RNA domains, which characterize differentiated ES cells.
We conclude that the strong increase in Xist expression levels in both ΔAV and Ma2L ES cell lines likely resulted from the impairment in Tsix expression but not from a role for cis-acting DNA elements located within the 15-kb span deleted in the ΔAV ES cell line.
Tsix Impairment Results in Ectopic XCI in Differentiated Male ES Cells.
Using qRT-PCR, we detected up to a 15-fold increase in the expression of mature Xist transcripts in both the ΔAV and Ma2L ES cell lines upon differentiation (Fig. 4 A). This result was confirmed at the RNA-FISH level, where >35% of cells showing nuclear accumulations of Xist RNA after 3–4 days of differentiation either with retinoic acid (Fig. 4 B) or through EB formation (data not shown). The efficiency of this ectopic Xist up-regulation appears to overlap the range that we and others have observed for wild-type differentiated female ES cells (14, 28, 34) and is 3- to 4-fold higher than that observed for Δ34 differentiated ES cells (Fig. 2A). In contrast, we observed a complete absence of ectopic Xist up-regulation in differentiated ΔAJ ES cells (data not shown) and a very weak and variable level in the ΔAS differentiated ES cells (Fig. 4 A). We conclude that neither the 17-mer tandem repeat nor the elements homologous to the 5′ region of the mouse pTa gene play any significant role in preventing ectopic Xist up-regulation in male cells. We can also conclude that the level of Tsix transcription is likely to be the main determining factor in the prevention of ectopic Xist up-regulation in differentiated male ES cells, because deletion (ΔAV) and insertion (Ma2L) mutants showed an identical phenotype.
Impairment of Tsix transcription results in ectopic XCI in differentiated male ES cells. (A) qRT-PCR analysis of Xist normalized to Arpo in ES cells differentiated with retinoic acid. Between days 2 and 5, the levels of Xist RNA in the ΔAV#1 and Ma2L are up-regulated 40- to 170-fold as compared with CK35. ΔAJ#1 showed no up-regulation and the ΔAS #1 a weak up-regulation of Xist during differentiation. This experiment also included the Δ34#1 and#2 ES cells (results in Fig. 2 A), allowing for the comparison of all cell lines. Two other experiments provided similar results. (B) A representative view of Xist RNA-FISH (green) on DAPI-stained (blue) nuclei at day 4 of differentiation. The table (Lower) shows the percentage of cells with Xist nuclear domains between days 3 and 5 (n >200). (C) Immuno-RNA-FISH on DAPI-stained (blue) nuclei of the ΔAV#1 ES cells after 4 days of differentiation with retinoic acid. Xist RNA recruits H3methylK27 in differentiated male ES cells. Similar results were obtained by using the Ma2L ES cell line.
We next addressed whether the Xist RNA up-regulation that we observed in Tsix impaired differentiated ES cells is accompanied by other features of XCI. Immuno-RNA-FISH experiments using an H3methylK27 antiserum performed at day 4 of differentiation on the ΔAV ES cell line showed frequent colocalization of the Xist and H3methylK27 nuclear domains (Fig. 4 C). Additionally, a significant subset of cells that showed a Xist nuclear domain could be shown to lack expression of the MeCP2 X-linked gene, as judged by the absence of a signal at the MeCP2 primary transcription site (Fig. 9B). Essentially identical results were obtained by using the Ma2L differentiated ES cells (data not shown). Such features of XCI are likely to cause cell death, explaining why both the Xist RNA levels and the percentage of cells showing a Xist nuclear accumulation decrease after 5 or 6 days of differentiation.
Discussion
We have previously defined a region of the Xic involved in counting. In the present study, we have systematically evaluated the function of elements from within this candidate region. We have shown that the 1.2-kb deletion of DXPas34 impairs recruitment of PolII and TFIIB to the Tsix major promoter and results in a 90% reduction in the levels of Tsix expression in ES cells. Male ES cells deleted for DXPas34 were able to initiate XCI upon differentiation, albeit with low to moderate efficiency. This correlates with the incomplete reduction in Tsix expression observed in DXPas34 deleted cells. On the other hand, the Ma2L and ΔAV ES cell lines, which are both more strongly impaired for Tsix transcription, initiate highly efficient ectopic XCI upon differentiation. The ectopic Xist up-regulation occurring in the differentiated Ma2L and ΔAV ES cells resulted in H3methyl-K27 deposition and gene silencing, suggesting that the ectopic XCI process is completely induced despite its occurrence in XY cells. The critical role played by Tsix in counting is underlined by the observation that ectopic XCI did not occur in ES cells carrying deletions within the Tsix transcription unit that did not significantly alter the levels of Tsix transcription.
DXPas34, a Tandem Repeat Influencing the Expression of Key Players of the Xic.
The cascade of effects following the targeted deletion of the 1.2-kb DXPas34 tandem repeat is likely to result primarily from a reduction in the levels of TFIIB and PolII associated with the Tsix major promoter. DXPas34 appears to act as a transcriptional enhancer, and this is consistent with the existence of a DNase hypersensitive site closely associated with this element in ES cells (32). Such a function also fits with the presence of numerous CTCF-binding sites within DXPas34 (25), because CTCF is known to be able to act as a transcriptional activator (for review, see ref. 37). Enhancers, however, by definition act independently of distance and orientation, features that remain to be tested for DXPas34. Because differentiation of ES cells does not seem to be the switch for DXPas34 control of Tsix expression, it will be of interest to address this question at early embryological stages in the context of imprinted XCI. The frequent presence of repetitive DNA sequences in the vicinity of differentially methylated regions associated with imprinting centers further suggests the interest of studying the function of DXPas34 in imprinted XCI (38). We do not exclude that DXPas34 may also operate as a barrier element, again possibly involving CTCF binding that would shield the Tsix major promoter from a silencer located more 5′ (toward Xist). Because DXPas34 is embedded within a larger region containing several elements regulating positively Tsix (32), it could also serve an architectural function in the topological organization of these different elements. Additional work is clearly necessary to elucidate the precise molecular mechanism by which DXPas34 acts on the Tsix major promoter. Similarly to most tandem repeats, DXPas34 shows extensive length variability (24) whose impact on the antisense promoter activity remains to be evaluated.
Tsix Prevents the Occurrence of Ectopic XCI in Differentiated Male ES Cells.
Our results indicate that the level of expression of Tsix is inversely correlated to the efficiency of ectopic XCI in differentiated male ES cells. The idea that Tsix transcription may act to prevent ectopic XCI in differentiated male ES cells finds support in the literature. It is compatible with our previous observation that the 37-kb sequence lying immediately downstream of Xist, which contains the Tsix major promoter, is both necessary and sufficient to prevent ectopic XCI in differentiated male ES cells (20). It is also consistent with previous reports that both the Ma2L ES cell line, which carries a stop for transcription in Tsix (26) and the Tsix AA2 Δ 1.7 ES cell line in which DXPas34 and the Tsix major promoter are replaced by an IRESβgeopA cassette (18, 39), up-regulate Xist upon differentiation (18, 26, 39). The percentage of differentiated Ma2L cells showing Xist RNA nuclear domains was originally reported to be moderate [9.5% after 3 days of retinoic acid differentiation (26)]. Absolute levels of ectopic XCI induction, however, may not be a particularly robust parameter in the absence of internal control cell lines, because ES cell differentiation is highly sensitive to precise experimental conditions. In the present experiments, to circumvent this difficulty, we performed parallel differentiation of the different mutant ES cell lines and monitored carefully the differentiation process by qRT-PCR for the pluripotential stem cell marker Oct3/4. Under our experimental conditions, the Ma2L cell line showed similar levels of highly efficient ectopic XCI to the ΔAV ES cell line, whereas clearly lower levels of XCI characterized the DXPas34 deleted ES cells upon differentiation.
The only study to date that seems incompatible with a role for Tsix transcription in counting is the ΔCpG deletion, which was reported neither to result in Xist up-regulation in undifferentiated male ES cells nor in ectopic XCI upon differentiation (17). It is possible, however, that this phenotype resulted from the presence of a Pgk1-neo cassette in the mutation, as originally reported (17). We have shown in the NeoΔ34 cell lines that this same Pgk1-neo cassette partially rescues Tsix and Xist expression in ES cells and significantly reduces ectopic XCI upon differentiation (Fig. 6).
The mechanisms underlying the repressive effect of Tsix transcription on Xist expression have been shown to act independently of the initiation of transcription (29). One can, therefore, postulate the existence of either a transcriptional interference mechanism acting on transcriptional elongation or a posttranscriptional effect possibly involving double-stranded RNA pathways. The elucidation of these mechanisms will be critical to our understanding of XCI regulation.
Implications for Regulatory Models of XCI and the Role of Tsix in the Counting Pathway.
The precise mechanisms involved in counting are as yet unknown but are assumed to involve the evaluation of the Xs/autosomes ratio followed by the establishment of a “signal” permitting or preventing the initiation of XCI. Our observations indicate that counting involves a repressive mechanism ensured by Tsix transcription in male ES cells. It is unclear whether our results can be easily related to those obtained by transgenesis experiments using 5′ Tsix and Xite sequences, which resulted in a near-complete inhibition of initiation of XCI in differentiating female ES cells (31). It is also as yet unresolved whether these multicopy transgenes were transcribed, and to what extent they are perturbing choice rather than counting.
It appears that impairment of Tsix transcription results in cis up-regulation of Xist in both male (this work and refs. 26 and 39) and female (8, 17, 18) ES cells upon differentiation. Postulating a competence factor (17) whose absence in male ES cells would explain their resistance to the initiation of XCI no longer appears necessary. This latter conclusion is supported by numerous other observations describing the capacity of differentiated male ES cells to undergo ectopic XCI after invalidation of Dnmt1 (40), as well as the introduction of Xist transgenes (41, 42).
Describing a novel function for the DXPas34 tandem repeat in eliciting Tsix transcription and for Tsix in ensuring normal counting in male ES cells may be an important step toward the elucidation of the mechanisms that numerate the Xs/autosomes ratio in the counting process. Although more work will be required to define the molecular mechanisms underlying the role of Tsix, it appears possible that regulators of Tsix transcription are direct targets for the numerator system of the counting process.
Methods
Targeted Deletions.
Oligonucleotides used in this work are listed in Table 1, which is published as supporting information on the PNAS web site. Recombination vectors, selections, and screening methods are described in Supporting Text, which is published as supporting information on the PNAS web site.
Cell Culture.
ES cells culture and differentiation using 10−7 M all-trans retinoic acid or embryoid bodies protocol were as described (20).
FISH.
RNA- and DNA-FISH were performed by using nick translation probes (Vysis) (8). ImmunoRNA-FISH was performed by using the H3 di/tri-meK27 antibody 7B11. A Zeiss Axioplan microscope, a Quantix charge-coupled device camera (Photometrics, Tucson, AZ), and smartcapture2 (Digital Scientific, Cambridge, U.K.) were used for image acquisition (6).
ChIP.
ChIP assays were carried out as described (29) by using TFIIB (C-18, Santa Cruz Biotechnology), RNA-PolII (7C2, Euromedex), and H3 dimeK4 (Upstate Biotechnology, Lake Placid, NY) antibodies. IP/input% was calculated from triplicate qPCR on IP and input DNAs.
Real-Time qRT-PCR.
Total RNA was isolated by using RNAble (Eurobio, Paris) treated with DNase (Roche Applied Science, Indianapolis) and verified by using a Bioanalyser (Applied Biosystems). Random primed reverse transcription was carried out with Superscript II (Invitrogen). qRT-PCRs were performed by using SYBR green mix and an ABI Prism 7700 (Applied Biosystems).
Acknowledgments
We thank E. Heard and C. Rougeulle for critical reading of this manuscript and R. Jaenisch (Whitehead Institute for Biomedical Research, Cambridge, MA) for the Ma2L cell line. S.V. is a doctoral fellow supported by the Ministry for Research. This work was supported by the NoE Epigenome and ACI 032526. P.A. was supported by the Centre National de la Recherche Scientifique, and P.C. was supported by the Pasteur Institute.
Footnotes
- †To whom correspondence should be addressed. E-mail: pclerc{at}pasteur.fr
-
↵*Present address: Centre National de la Recherche Scientifique, Unité Mixte de Recherche 218, Curie Institute, 26 Rue d’Ulm, 75248 Paris Cedex 05, France.
-
Author contributions: S.V. and P.C. designed research; S.V., S.A., P.N., and P.C. performed research; S.V., S.A., and P.C. contributed new reagents/analytic tools; S.V. and P.C. analyzed data; and S.V., P.A., and P.C. wrote the paper.
-
Conflict of interest statement: No conflicts declared.
- Abbreviations:
- XCI,
- X chromosome inactivation;
- qRT-PCR,
- quantitative RT-PCR;
- ChIP,
- chromatin immunoprecipitation;
- PolII,
- polymerase II;
- Xi,
- inactive X.
Abbreviations:
- © 2006 by The National Academy of Sciences of the USA