Variability of Sequence Surrounding the Xist Gene in Rodents Suggests Taxon-Specific Regulation of X Chromosome Inactivation

One of the two X chromosomes in female mammalian cells is subject to inactivation (XCI) initiated by the Xist gene. In this study, we examined in rodents (voles and rat) the conservation of the microsatellite region DXPas34, the Tsix gene (antisense counterpart of Xist), and enhancer Xite that have been shown to flank Xist and regulate XCI in mouse. We have found that mouse regions of the Tsix gene major promoter and minisatellite repeat DXPas34 are conserved among rodents. We have also shown that in voles and rat the region homologous to the mouse Tsix major promoter, initiates antisense to Xist transcription and terminates around the Xist gene start site as is observed with mouse Tsix. A conservation of Tsix expression pattern in voles, rat and mice suggests a crucial role of the antisense transcription in regulation of Xist and XIC in rodents. Most surprisingly, we have found that voles lack the regions homologous to the regulatory element Xite, which is instead replaced with the Slc7a3 gene that is unassociated with the X-inactivation centre in any other eutherians studied. Furthermore, we have not identified any transcription that could have the same functions as murine Xite in voles. Overall, our data show that not all the functional elements surrounding Xist in mice are well conserved even within rodents, thereby suggesting that the regulation of XCI may be at least partially taxon-specific.


Introduction
X chromosome inactivation (XCI) is a developmentally regulated process, which results in heterochromatization and transcriptional silencing of one of the two X chromosomes in eutherian females [1]. Imprinted XCI occurs on the paternal X chromosome (Xp) in the preimplantation embryo of some eutherians (for example, rodents) and is further maintained in the placenta [2,3]. Random XCI takes place on either the maternal or paternal X chromosomes after Xp is reactivated in the cells giving rise to the embryo proper.
A complex X-linked locus termed the X-inactivation centre (Xic) governs both imprinted and random XCI (reviewed in [4]). It has been shown that the initiation of XCI and propagation of silencing are mainly provided by the Xist gene which produces a 17 kb nuclear RNA associated with the inactive X chromosome [5][6][7][8]. This is the only functional element of the Xic that has been identified in all eutherians studied [9][10][11]. The studies of the Xic in mice have detected multiple elements in surrounding Xist with roles at different stages of XCI. Two non-coding nuclear RNA genes Enox (Jpx) and Ftx are localized 59 to Xist [12,13]. Both genes are positive regulators of Xist [14,15]. The microsatellite region DXPas34, the Tsix gene (the antisense counterpart to Xist), regulatory element Xite, and a 37 kb bipartite counting element have been mapped 39 to Xist (reviewed in [16,17]). As is demonstrated, these elements in mice regulate Xist expression during imprinted and random XCIs and are involved in the mechanisms underlying the counting of X chromosome number per diploid set of autosomes and the choice of the X chromosome to be inactivated during random XCI. However, these regulatory elements have not been definitely identified in other eutherians.
In this study, we intended to find any conserved elements surrounding Xist in rodents (mouse, rat, and common voles). Mouse and rat represent Muridae rodents. These two species diverged from a common ancestor 2 million years ago [18,19]. Voles are Arvicolidae rodents, which diverged from Muridae lineage 15-25 million years ago. We have earlier identified and described the nucleotide sequences of Xist in four common vole species [20]. In this work, we have extended our analysis of vole Xic further downstream of Xist and compared the region 39 to Xist in vole, mouse, and rat. We have also determined all transcription upstream, downstream, and across Xist in vole, as such transcription has been suggested to regulate Xist expression and be involved in counting and choice function of Xic [21].
We have found that the vole sequences downstream of Xist have homology to the minisatellite DXPas34 region and the Tsix major promoter of mouse Xic. We have demonstrated that this putative Tsix promoter identified in voles is a site of origin for transcription antisense to Xist, displaying a very similar expression pattern to mouse Tsix. Conservation of the Tsix major promoter sequences and the pattern of Tsix expression between vole, mouse and rat suggest a crucial role for antisense transcription in the regulation of Xist expression and XIC in rodents. However, we have found that the region containing the Tsix minor promoter and the regulatory element Xite in mice is replaced with the Slc7a3 gene and its surrounding sequence in voles. This allows us to suggest that both the nucleotide sequences of these elements and the transcription associated with them are not absolutely necessary for XCI even in rodents.

Results
Comparative Study of the Sequences 39 to Xist in Voles,

Mouse and Rat
We have earlier identified and described the nucleotide sequences of Xist in four common vole species [20]. In this study, we have extended our analysis further downstream of Xist, as it is known that this region in mouse comprises the DXPas34, Tsix, and Xite regulatory elements, which act at different steps of XCI (reviewed in [16,17]). Several clones were isolated from phage genomic libraries for each vole species, and the sequence contig extending over 24 kb downstream of the vole Xist built (Fig. 1). Comparative sequence analysis shows that sequences 39 to Xist of the four vole species display an overall similarity.
Then we compared the mouse and rat Xic sequences 39 to Xist up to the Tsx gene from the UCSC Genome Bioinformatics database and found that this region of the two rodent species displayed a high similarity. In the rat sequence, we have identified all the regulatory elements mapped in the mouse Xic region (Fig. S2).
Comparison of the vole sequence 39 to Xist with the corresponding mouse and rat regions detected homology to the mouse minisatellite DXPas34 region, the major promoter of mouse Tsix, and some sequences surrounding these elements (Fig. 1, Fig. S3). We have not found any significant homology to the mouse minor Tsix promoter and the associated exon or Xite within the vole region studied. Surprisingly, we located the vole ortholog of the mouse Slc7a3 gene in close proximity (5 kb) to the putative vole Tsix major promoter (Fig. 1, Fig. S3). Then we examined in more detail all the similarities and differences found in the region 39 to Xist between vole and two other rodent species.

DXPas34
DXPas34 in mice has been described as a block of minisatellite repeats with a monomer of 34 bp [22]. The repeats are CpG-rich, and each monomer contains a binding site for the CTCF protein, known to be involved in XCI regulation [23][24][25]. However, our more comprehensive analysis of mouse DXPas34 using the Tandem Repeat Finder program revealed three blocks of repeats composed of monomers of 34, 31, and 30 bp (Mus-34, Mus-31, and Mus-30) respectively ( Fig. 2A). Two blocks of repeats composed of the 31-and 32-bp monomers (Rn-31 and Rn-32) have been found in rat ( Fig. 2A). Three blocks of tandem repeats were detected in the four studied vole species. The first block comprises the monomers with an average length of 70 bp (Mc-70); the second, of 48 bp (Mc-48); and the third, of 34 bp (Mc-34) ( Fig. 2A). The vole species differ in the copy number of monomers in the blocks. The similarity of vole monomers within blocks varies from 53 to 89 %. We managed to identify a motif that retained a high degree of conservation in all monomer types of mouse, rat, and voles (Fig. 2C). This motif may be necessary for binding of a protein factor involved in XCI regulation.
The CpG dinucleotide content and the number of binding sites for CTCF factor in the DXPas34 region in rodents displays considerable variation ( Fig. 2A). The corresponding region of rat appeared the richest in CpG content; however, it contained considerably smaller number of CTCF binding sites compared with mouse. The least CpG content and number of CTCF binding sites was observed in voles ( Fig. 2A). Note that this region in M. arvalis completely lacked CTCF binding sites and almost lacked CpG dinucleotides. Note also that the M. arvalis X chromosome is predominantly active in the cells of interspecific hybrids obtained by reciprocal crosses of this species with the three remaining vole species [26].

Tsix Major Promoter and Associated Exon
The 100-bp region located upstream of the main Tsix transcription start site, identified in mouse by RACE experiments [27] is the most conserved between rodent species (Fig. S3). Presumably, this region represents a basal Tsix gene promoter, which contains the sites necessary for initiation and regulation of its expression. The sequence immediately adjacent to the vole region similar to the Tsix promoter has a homology to mouse Tsix exon and contains a CpG island in three vole species (except M. arvalis). As CpG islands in this region of the mouse female inactive X chromosome are hypermethylated [28][29][30], we decided to clarify the methylation status of CpG dinucleotides in the corresponding vole region. We digested the genomic DNA isolated from M. arvalis and M. rossiaemeridionalis with the restriction endonuclease HpaII, sensitive to methylation, and assayed it by Southern blot hybridization (Fig. 3). The genomic region was not methylated on the only active male X and, consequently, its DNA was completely digested with HpaII. In females, carrying one active and one inactive X chromosome, in addition to the HpaII-digested fraction, a fraction inaccessible for digestion with HpaII due to the presence of methylated DNA on the inactive X, was also detected. Thus, at least individual CpG dinucleotides in this region of the vole female inactive X chromosome are hypermethylated.

Pseudo NIF3L1BP1
Immediately adjacent to the major promoter of Tsix, we have found several regions homologous to the NIF3L1BP1 gene in voles, mouse, and rat ( Fig. 1, Fig. S3). They are likely to represent the remains of an ancient pseudogene the parts of which have been separated by an intensive integration of mobile elements. Pseudo NIF3L1BP1 is the last element common to the vole, mouse, and rat regions 39 to Xist. In voles, pseudo NIF3L1BP1 is located at the boundary of the rearrangement that led to Slc7a3 embedding.

Slc7a3
Unlike mouse and rat, Slc7a3 was found in voles at distance 5 kb from the region similar to the pseudogene NIF3L1BP1 (Fig. 1, Fig.  S3). Slc7a3 is not located within the Xic in any other eutherians studied. In all other eutherians, the Xic region downstream of Xist is flanked with the protein-coding genes Tsx, Chic1, and Cdx4. In mice, Slc7a3 is located 2.4 Mb upstream of the 59 boundary of Xist. We have found that vole Slc7a3 has the same exon-intronic structure as mouse Slc7a3 and consists of 13 exons. The presence of the sequences homologous to promoter and the surrounding of this gene in mouse, similarity of the exon-intron structure, and absence of stop codons and frameshifts in the coding region allowed us to assume that we had identified functional Slc7a3, brought close to vole Xic as a result of a chromosome rearrangement.
Having discovered the rearrangement ';in the vole Xic region, we decided to determine the localizations of Xist, Slc7a3, Chic1, and Cdx4 on the M. rossiaemeridionalis metaphase chromosomes. Despite Chic1 and Cdx4 having been obviously removed from the close proximity of Xist in voles, they were nevertheless located within the same cytogenetic band on the vole X chromosome as Xist and Slc7a3 (Fig. S4).

Bipartite Counting Element
The bipartite counting element is defined as a 37 kb region 39 to Xist independent of DXPas34 and Tsix. Deletion of the region in XY and XO embryonic stem (ES) cells results in aberrant inactivation of the only X chromosome [31][32][33], and insertion of certain sequences from the region into autosomes in XX ES cells interferes with normal counting process and blocks XCI [34]. In voles, the distance between 39 of Xist and the nearest proteincoding gene Slc7a3 is only 15 kb, of which only 4 kb displays similarity to the sequences of the mouse 37-kb bipartite counting element. This similarity is detectable across a region of about 3 kb adjacent to DXPas34 and over 1 kb upstream of the major Tsix transcription start site (Fig. S3).

Comparison of 39 Xist Region in Rodents and Other Mammalian Species
We have compared the region 39 to Xist in rodents and other mammals and found no sequences homologous to Tsix, DXPas34, and Xite in primates (human and chimpanzee), ungulates (bovine), and carnivores (dog) (data not shown). Thus, the sequences homologous to Xic functional elements found in the mouse 39 to Xist region are detectable only in the order Rodentia. However, we have identified four areas of homology between mouse and human 39 to Xist regions (R1 -R4) (Fig. 4), three of which (R1-R3) have been previously reported [35]. We found that human R3 and R4 are separated by species-specific transposable elements, whereas an insertion of DXPas34 with the 59 region of Tsix occurred between mouse R3 and R4.
Previously, within a transposable element 39 to Xist in human, several short dispersed regions of similarity to DXPas34 consensus monomers were identified [36]. Nevertheless, the results of our comparative analysis show that the homology in human is not between R3 and R4 (Fig.4). Taken together these facts suggest that microsatellite repeats and adjacent sequences could have emerged from a mobile element, but that a transposon with homology to DXPas34 found in human is probably not the genuine ancestor of the region.

Overall Transcription Upstream, Downstream, and Across the Xist Gene in Voles
We analyzed the transcriptional activity at Xist and adjacent regions using strand-specific RT-PCR.
First, we decided to confirm the transcriptional activity of the Slc7a3 sequence detected in voles. It has been found that this gene, as expected, is transcribed antisense to Xist and ubiquitously expressed in fetal tissues of both males and females (Fig. 1V,  amplicon 14).
We have not identified any transcription in the intergenic region between 39 end of Slc7a3 and the putative transcription start site of vole Tsix (Fig. 1B). Thus, we have found in voles neither the sequences similar to Xite nor the intergenic antisense to Xist transcription that could have the same functions as Xite in mouse.
Antisense to Xist transcription was detected between the putative vole Tsix promoter and Xist promoter in the embryo and placenta of both males and females. Transcription in vole is identical to the transcription of the mouse Tsix (Fig. 1B).
Xist expression was revealed only in female tissues. It fits the vole Xist transcription unit reported previously (Fig. 1B). Transcription neither sense nor antisense to Xist was detected in the intergenic region starting 2 kb upstream of the Xist promoter and continuing up to the GpG island near the Enox (Jpx) gene regions (Fig. 1V, amplicons 6 to 4). We did however identify transcription both sense and antisense to Xist transcription in the region of the vole Enox (Jpx) putative promoter. In mouse, similar sense-antisense transcription in this region occurs because of the bidirectional activity of Enox (Jpx) promoter [21].

Allele-Specific Profile of Sense and Antisense Transcription in the Vole Xic
Differences in the Xic nucleotide sequences between the vole species allow us to determine the allele-specific profile for Slc7a3, Tsix, Xist, and Enox (Jpx) expressions. For this purpose, we used 12.5 dpc vole XX placentas obtained by interspecies crosses with imprinted inactivation of the paternal X chromosome.
The electrophoretic patterns of sequencing reactions for PCR products of these genes obtained from the genomic DNA of hybrid placentas displayed double peaks at the polymorphic positions, thereby confirming the presence of both parental alleles in the hybrid placentas (Fig. 1C). Sequencing of the PCR products obtained from the cDNA of hybrid XX placentas demonstrated that only Xist was expressed from the inactive X chromosome, whereas the remaining genes Slc7a3, Tsix, and Enox (Jpx) were expressed from the active X chromosome.

Exon-Intronic Structure and Boundaries of the Vole Tsix and Enox Genes
To determine the exon-intron structure and transcription boundaries for Tsix and Enox in voles, we used 59 and 39 RACE and strand-specific RT-PCR. The RNA for 59 and 39 RACE was isolated from 12.5 dpc XX and XY embryos and placentas of M. rossiaemeridionalis and M. arvalis.

Tsix
The 39 RACE primers for Tsix were designed for the regions 2100 and +1500 bp relatively of Xist transcription start site (Table  S2). The PCR products obtained by RACE were subcloned and sequenced. Analysis of the sequences of 39 RACE products demonstrated that Tsix transcription in voles terminated at multiple sites encompassing the Xist transcription start site (Fig. 5). The most distant Tsix transcription termination site was detected at position -1279 bp from the Xist transcription start site.
The Tsix gene-specific primers for 59 RACE were designed for the region +1200 bp relatively to Xist transcription start site. For M. arvalis and M. rossiaemeridionalis, 59 RACE clones with a length of 600 bp were obtained (Fig. 5). The first 134 bp in the clones correspond to the sequence located immediately after the primer. The next 110 bp correspond to the sequence of Xist exon 1, located at a distance of 6 kb from the first homologous region. The remaining 450 bp are identical to the newly determined genomic sequence localized 12 kb downstream of Xist exon 8. This fragment is homologous to the mouse exon located downstream of the major Tsix start site. In addition, we found clones with regions of continuous 600-800-bp homology to Xist exon 1 detected after the 134-bp region; these homology regions started 1.5 kb downstream of the 59 boundary of the 110-bp fragment. Presumably, these clones were truncated at the 59 end due to a premature termination of reverse transcription. To obtain the complete structure of the Tsix transcript isoform, we carried out  strand-specific RT-PCR using the primer set positioned within Xist exon 1 sequence identically to the 600-800-bp region and in the vicinity of the reported Tsix major transcription start site. As a result, we amplified a 2 kb cDNA fragment. Part of this fragment with a length of 1.4 kb was identical in its sequence to Xist exon 1, which corresponded to the 600-800-bp fragments from its 39 end and to 110-bp fragment from its 59 end; the next 75 bp was homologous to Xist intron 3; and 400 bp at the 59 end corresponded to the sequence similar to the mouse exon located after the major Tsix transcription start site.
Thus, the RACE experiments allowed us to detect four exons (designated A, B, C, and D) in the vole Tsix gene (Fig. 5). Exon A (419 bp) is localized 12 kb downstream of the end of Xist exon 8. It corresponds to the exon found in mouse after the major Tsix start site reported in [27]. Exon B has a length of 121 bp and is localized to Xist intron 3. Its sequence is a fragment of B1 element. Exon C (1.4 kb) is located near the 39 boundary of Xist exon 1. Several variants of alternative splicing were detected for this exon. Exon D (3.2 kb) terminates 1279 bp upstream of the Xist transcription start site and spans 2.9 kb of Xist exon 1. It corresponds to the terminal exon of mouse Tsix reported in [27,37].
To detect Tsix transcript in vole tissues, Northern blot hybridization with strand-specific probe corresponding to exon A was performed. Three major hybridization signals were observed (Fig. 6). The first high molecular band appears to represent unspliced Tsix transcript and to correspond in size to the whole transcription unit. The second major band is about 5 kb and presumably represents spliced transcripts containing all the four vole Tsix exons, total size of which is 5,14 kb. The third band detected is about 2,2 kb. In summary, the results show that the Tsix transcript is only partially spliced and its main portion is present in an unspliced form.

Enox
We performed both 59 and 39 RACE to identify boundaries of Enox (Jpx) and its antisense counterpart. However, only 39 RACE for Enox (Jpx) and 59 RACE for its antisense transcript was successful. Gene-specific primers were designed within region 2 for 39 RACE of Enox (Jpx) and within region 3 for 59 RACE of transcript antisense to Enox (Jpx) (Fig. 1A, Table S2). All clones obtained in 39 and 59 RACE were identical to the genomic DNA sequences located immediately after the primers. The nucleotide sequences of the 59 and 39 RACE clones were deposited with Gene Bank (accession numbers [GenBank:JF519003], [GenBank: JF519004]). We found that vole Enox (Jpx) terminated 840 bp downstream from the region homologous to the mouse Enox (Jpx) promoter, while the antisense transcription start site was mapped within the Enox (Jpx) transcription unit 342 bp before its 39 end. Thus, in vole, both Enox (Jpx) and its antisense counterpart represent a single exon transcription unit, which taking into account RT-PCR data (Fig. 1A,B) spans about 1500-2000 kb.

Transcription Antisense to Xist in Rat
We searched for the mRNA and EST antisense to rat Xist at the Blat server (http://genome.ucsc.edu/cgi-bin/hgBlat). Two spliced RNAs were detected (Fig. S5). EST [GenBank:CF978550] has three exons. The first exon is within the block of tandem repeats DXPas34 and does not coincide with any known mouse or vole M. rossiaemeridionalis, were isolated by 59RACE. Rt1 represents a strand-specific RT-PCR product obtained from M. rossiaemeridionalis placenta using primers Bt11 -SNTR (Table S2). doi:10.1371/journal.pone.0022771.g005 Tsix exons. The second exon coincides with vole exon C and the third, with the terminal Tsix exon of vole and mouse. This RNA seems to be a spliced variant of the rat Tsix transcript.
The other RNA, [GenBank:AY539944], comprises ten exons and contains a native open reading frame encoding a hypothetic protein LRRGT00193. This cDNA is referred as the gene Lrrrn [38,39,40], the transcription of which has been detected in the rat liver. The exons of this gene are localized in the region homologous to the Xite regulatory element, in Tsix exon downstream of the major promoter and in Xist introns (Fig.  S5A). Neither vole nor mouse retains the native open reading frame of the rat Lrrrn gene.
Using strand-specific mRNA from 14 dpc rat placentas and embryos, we confirmed the exon-intronic structure of Tsix and the transcription corresponding to Enox (Jpx) identified by EST analysis (Fig. S5A) We also confirmed by RT-PCR that in rat Tsix, transcription passed through the Xist gene promoter. Northern blot hybridization with a strand-specific probe representing the exon which follows the major Tsix start site revealed in rat a high molecular weight transcript, presumably corresponding to unspliced Tsix and two spliced isoforms of ,6 kb and ,1,8 kb (Fig. 6).

Variability of Transcription of Xist Surrounding in Rodents
In this study we identified in vole and rat 59 to Xist transcription corresponding to the mouse Enox (Jpx) gene. We also found in vole transcription antisense to Enox (Jpx) which was previously described in mouse [21]. However, the transcription boundary and its start site in mouse and vole differ both for Enox (Jpx) and its antisense transcript. Vole Enox (Jpx) transcript is not spliced and the transcription unit spans about 2 kb from the initiation site, while mouse Enox (Jpx) is transcribed through tens of kb and gives spliced mRNA containing up to 5 exons [12,13]. It should be noted that vole Enox (Jpx) transcription initiates from the speciesspecific CpG-rich region and obviously transcriptional regulation of the gene differs in vole and mouse. Thus, we can propose that the regulation of XCI by transcription associated with Enox (Jpx) is not the same in vole and mouse. It is intriguing that Enox (Jpx) transcription is more similar in mouse and human [12,13] than in mouse and vole. Finally, it should be also noted that Enox (Jpx) is not completely silenced on the inactive X chromosome during random inactivation and is biallelically expressed in female mice [13,15]. However, Enox (Jpx) appears to be expressed only on the active X chromosome in vole during imprinted inactivation. The difference in Enox (Jpx) expression between vole and mouse could be ascribed in equal degree both to species-specific features and different function of the gene during imprinted and random inactivation. Further studies are needed to clarify these issues.
Our results demonstrate that in both rat and vole there is expression antisense to Xist which corresponds to mouse Tsix. It starts from the region homologous to the mouse Tsix major promoter and ends around Xist transcription start site. The vole Tsix gene has an expression pattern similar to that of mouse Tsix. Both vole and rat Tsix transcripts cover the Xist promoter, which, as has been demonstrated for mouse, is obligatory for Tsix functioning in XCI [41][42][43]. Thus, it is most likely that similar to the mouse Tsix, the vole and rat Tsix is able to regulate Xist expression. However, this requires additional confirmation. Similar to mouse the vole and rat Tsix RNA undergoes, at least in part, alternative splicing. The terminal exons of the Tsix genes in mouse, rat and vole are identical, whereas the remaining exons differ between rodent species. It has been shown that many exons in the noncoding RNA genes of Xic, such as Xist and Enox (Jpx), originated from mobile elements of various classes [9]. In this work, we have found another example of how a part of a speciesspecific SINE is present in spliced mature RNA of vole Tsix and represents one of its exons. This example illustrates the idea that integration of mobile elements into noncoding RNA genes of Xic continues in contemporary eutherian species and also supports the assumption that the emergence of exons from mobile elements is a general way of evolution and rearrangement of genes encoding large nuclear regulatory RNAs. However, the differences in the exon-intronic structures of the rodent Tsix genes together with the data on the absence of the differences in the ratio of spliced and unspliced Tsix transcripts [37] and the absence of XCI abnormalities caused by mutations of mouse Tsix splicing sites [44] confirms the earlier assumption that Tsix splicing is not necessary for its normal function.
It should be noted that transcription in the region of the Tsix major promoter is bidirectional in mouse [36], however, we only identified transcription antisense to Xist in vole. Xite is present and expressed in mouse [35], but not in vole. Tissue-specific proteincoding gene Lrrrn is transcribed through Xite, Tsix, and Xist in rat [38][39][40], but is not even found in mouse and vole. Thus, comparative analysis data has shown that many transcripts that surround the Xist gene in rodents are taxon-specific. The taxonspecific transcripts either are able to influence XCI (as Xite), or are not absolutely linked with it (as Lrrrn).

Conservation of the Xic 39 to Xist Elements in Rodents and Other Eutherians
Comparison of the vole, mouse, and rat sequences has demonstrated that the most conserved of all the known regulatory elements downstream of Xist in rodents are the regions of Tsix's major promoter and minisatellite repeats. Their conservation and necessity in rodents suggests that they are absolutely essential for XCI in this mammalian order. Note that the region of the Tsix major promoter is the most conserved in rodents. However, targeted deletions of the Tsix promoter in mice do not impair Tsix function in random XCI [36].
As the sequences homologous to mouse DXPas34 and Tsix major promoter are only detected in vole and rat, but not in others mammals, we could propose that this region emerged de novo in Rodentia, most likely as a result of the transposition and subsequent amplification of the same sequences which led to the formation of minisatellite repeat arrays. Note that transcription antisense to Xist is also not well conserved in eutherians. The transcript antisense to XIST is revealed in human, but does not overlap the XIST promoter [45][46][47]. Moreover, it is coexpressed with XIST from the inactive X chromosome in fetal and neonatal female cells and does not downregulate XIST, suggesting that this species-specific transcription is not linked with XCI. Thus, regulation of inactivation by Tsix may have arisen in the evolution quite recently and apparently is unique to rodents.
As voles lack the regions homologous to the Tsix minor promoter and Xite, we can assume that they are not absolutely necessary for XCI in eutherians. Moreover, as Xite sequences are only detected in mouse and rat, we cannot exclude that this element has emerged and evolved only in Muridae.
In all eutherians, for which whole genome sequences were obtained, Xic 39 to Xist is flanked by the protein-coding genes Tsx, Chic1 and Cdx4 [9]. There is a speculation that others besides the known regulatory elements of XCI may lie within the region of these protein-coding genes (reviewed in [16]). The rearrangement detected in the vole Xic region demonstrates that the presence of protein-coding genes Tsx, Chic1, and Cdx4 and the sequences linked with them downstream of the 39 Xist boundary are not necessary for XCI. The Xic rearrangement has evolved and, as a result, as little as 15 kb of the sequences downstream of Xist characteristic of this region in mouse appear sufficient for normal imprinted and random XCI regulation in vole.
In this study we found that not all functional elements flanking Xist in mice were well conserved even within rodents. Non-coding RNA transcripts in the region surrounding Xist can appear, disappear, change their promoters, exon-intron structure and borders. No common conserved elements located 39 to Xist have been identified in eutherians, suggesting that the XIC functional elements responsible for 'counting', 'choice', regulation of Xist and XCI may be at least partially taxon-specific.

Ethics statement
The study was carried out according to "The Guidelines for Manipulations with Experimental Animals." The study was approved by the Ethical Committee of the Institute of Cytology and Genetics, Novosibirsk, permit number: (order of the Presidium of the Russian Academy of Sciences of April 02, 1980 no. 12000-496).

DNA sequencing and analysis
The recombinant DNA sequences from phage clones were subcloned into plasmid vectors and sequenced using both the universal and specific primers. The sequencing reactions were carried out using a Big Dye Terminator v. 3.1 kit. The reaction products were analyzed in ABI Prism automated sequencers. The DNA sequences localized 39 to Xist were determined in both strands for each of the four vole species and joined with the earlier obtained sequences containing the vole genes Enox Comparative sequence analysis was conducted using the following software packages: BLAST [48], http://www.ncbi.nlm. nih.gov/] for searching for homologous sequences; Tandem Repeat Finder 4 [49] for identifying tandem repeats; RepeatMasker [50] (http://www repeatmasker org) for detecting mobile elements; Fasta [51] and CLUSTALX [52] for aligning two and more sequences (the programs and data are available at http:// genome.ucsc.edu/, http://www.ensembl.org/, and http://bio.cse. psu.edu/); and PipMaker [53] for conducting genomic analysis of extended loci.
The following sequences of mouse, rat, bovine, dog, chimpanzee, and human, extracted from the corresponding databases of sequenced genomes at the UCSC Genome Bioinformatics Site (http://genome.ucsc.edu/), were also used in the comparative analysis: mouse  RNA and DNA isolation, reverse transcription, and PCR RNA and DNA were isolated using a Tri Reagent (Sigma) kit from pooled preimplantation vole blastocysts, 12.5 dpc vole embryos and placentas. The cDNA was synthesized using total RNA by SuperScript III (GIBCO-BRL) reverse transcriptase at 50-55uC according to the manufacturer's instructions.
The strand-specific primers used for reverse transcription and cDNA amplification are listed in Table S1. Each reverse transcription reaction mixture was additionally supplemented with the strand-specific primer BAss to beta actin [21]. A negative control reaction, with the reaction mixture lacking reverse transcriptase, was conduced for each reverse transcription reaction.
The sex of 12.5 dpc embryos and placentas was determined using the primers UB1X and UB1Y [21], which in vole giving a PCR products exclusively with male genomic DNA, and the strand-specific RT-PCR, detecting Xist expression exclusively in females (Fig. S1).

and 39 RACE
The exon-intron structure and boundaries of Tsix were analyzed using SMART RACE kit (Clontech). The gene-specific primers for 59 and 39 RACE are listed in Table S2

Northern blot hybridization
About 5 mg of poly(A)+RNA isolated using Oligotex-dT30 (Qiagen) from vole and rat 14 dpc male embryos were run on a 1% agarose gel containing formaldehyde and transferred onto GeneScreen membrane (PerkinElmer). Both vole and rat cDNA amplicons corresponding to the exon located after the major Tsix start site in mouse [27] were obtained. The primer pairs used were gTF TCCTTCTGGCCTCTTCCGTCA -SNTR CTCTCCC-TGCGCTCCCTCACT for vole and Rnt5 TCTAATATGA-CATTGCCGATG -Rnt6 GGCTCGCTTTCCGGACTATC for rat. For probe labeling 50 ng of the amplicons were added in linear PCR with [a 32 P]dCTP and strand-specific primers (gTF or Rnt5) producing DNA strand antisence to Tsix transcript. Hybridization was performed in x5 SSC, 0,5% SDS at 65uC overnight. Washing was carried out according to the manufacturer's instructions. . .

Allelic expression assay
the DNA sequences. The PCR primers (Table S3) were designed to generate the products containing SNPs. RT-PCR and PCR of genomic DNA were performed by standard methods using RNA and DNA samples of 12.5 dpc placentas of interspecific vole hybrids obtained from the crosses M. rossiaemeridionalis 6M. arvalis.
As voles have imprinted XCI in extraembryonic tissues [26], the majority of placental cells have the inactive X chromosome of paternal origin and the active X chromosome of maternal origin. The PCR products containing SNPs were purified by extraction from agarose gel, and their nucleotide sequences were analyzed in an ABI Prism automated sequencer.

DNA fluorescent in situ hybridizations
DNA FISH was carried out as previously described [54,55]. The phage clones containing the sequences of the M. rossiaemeridionalis genes Xist, Slc7a3, Chic1, Cdx4, Slc16a2, and Pgk1 labeled with biotin and digoxygenin were used as probes. The sex of 12.5 dpc embryos and placentas was determined using the primers UB1X and UB1Y [21], which produce a PCR product exclusively from vole male genomic DNA. (B) Additionally, the strand-specific RT-PCR, which detects Xist expression exclusively in females, was performed. Strand-specific primer for Xist cDNA synthesis was SDX3, CCCAGTGCTGGTGAGCTATTCC. Subsequent PCR was performed with primers NSX19, GTGATTAATTCATTC-TATCTGCC and MSX27, TTGCTCAGATTAGCTAG.