Implications for Tsix Function

Human Molecular Genetics, 2003, Vol. 12, No. 2 125–136 DOI: 10.1093/hmg/ddg010 Characterization and quantitation of differential Tsix transcripts: implications for Tsix function

Shinwa Shibata and Jeannie T. Lee*

Howard Hughes Medical Institute, Department of Molecular Biology, Massachusetts General Hospital, Department of Genetics, Harvard Medical School, Boston, MA 02114, USA

Received September 7, 2002; Revised and Accepted November 2, 2002 GenBank accession number: AF541962 Downloaded from https://academic.oup.com/hmg/article/12/2/125/608526 by guest on 27 September 2021

In dosage compensation of female mammals, the accumulation of Xist RNA initiates silencing of one X-chromosome. Xist action is repressed by the antisense gene, Tsix, whose full-length RNA product is complementary to Xist RNA in mice. While previous work showed that Tsix transcription blocks the accumulation of Xist RNA, it is still unclear whether this repression requires the antisense RNA product or whether the antisense transcriptional movement is sufficient. A better understanding of potential mechanisms requires elucidation of Tsix RNA structure and determination of Tsix RNA copy number relative to that of Xist RNA. Previous work indicated that at least some of murine Tsix is spliced and that human TSIX truncates within the 30 end of XIST. Here, further characterization and quantitation of murine Tsix RNA reveal three new findings: first, in undifferentiated embryonic stem cells, Tsix RNA is present at 10–100- fold molar excess over Xist RNA. Second, only 30–60% of Tsix RNA is spliced at known exon–intron junctions. The nearly equal abundance of spliced and unspliced species leaves open possible roles for both isoforms. Finally, Tsix is spliced heterogeneously at the 50 end and most detectable splice variants exhibit only a 1.9 kb region of complementarity between sense and antisense RNAs. Implications for Tsix’s possible mechanisms of action are discussed.

INTRODUCTION transcription blocks Xist RNA accumulation on the same chromosome (16,17). Current models propose that, prior to the In mammals, either of the two X-chromosomes in females is onset of XCI (undifferentiated cellular state), the co-expression inactivated to compensate for dosage (1), a phenomenon of Tsix along with Xist in cis prevents high level Xist expression referred to as X-chromosome inactivation (XCI). The master and the initiation of XCI. At the onset of XCI, silencing of the switch for this long-range chromosomal silencing is termed the future Xi can only proceed with the downregulation of Tsix ‘X-inactivation center’ (Xic) (2,3) and is sufficient for in cis (12,14) and, conversely, the maintenance of the active chromosome counting, choice, initiation of XCI, and establish- state on the Xa depends on the persistent expression of Tsix on ment of heterochromatin (4). Two unusual noncoding tran- that chromosome (16,17). scripts have been identified within the Xic. The 17 kb What is the molecular basis of Tsix’s action on Xist? Three untranslated product of the Xist (X-inactive specific transcript) classes of mechanisms have been proposed (8,9,12). First, locus is unique in that it is transcribed only from inactive X Tsix’s action may be independent of its transcription and RNA chromosome (Xi) in female somatic cells and accumulates in product. ‘Enhancer competition’ (18) between Xist and Tsix is cis along the Xi (5–7, for Xist reviews, refer to 8,9). Xist is one model that has been proposed for Tsix’s role in determining indispensable for the silencing step of XCI (10,11). Its X-chromosome choice (19). Two other classes of mechanisms antisense counterpart, Tsix, originates 12 kb downstream of include ‘transcription-dependent’ and ‘RNA-dependent’ mod- Xist and transverses the entire Xist locus, thus encompassing els. In the former, antisense transcriptional action in itself over 40 kb of the Xic (12). Prior to the onset of XCI, Tsix and provides the repressive force. For example, the opposing Xist RNAs are co-localized to the Xic. Unlike Xist RNA, movement of the Tsix RNA polymerase complex could either however, Tsix RNA has not been observed to paint the X (12). result in topological constraints on Xist RNA production or lead Indeed, its action is strictly limited to the Xic, where its to ‘counter-current collision’ of the converging Xist and Tsix expression represses the upregulation of Xist and designates the polymerase machinery. Alternatively, antisense transcription future active X (Xa). When Tsix expression is eliminated on across the Xist promoter (15,20) could impair the recruitment one X in female cells, XCI occurs predominantly on the of transcriptional initiation machinery to Xist, perhaps by mutated X (13–15). In contrast, the augmentation of Tsix ‘promoter occlusion’. In contrast, the ‘RNA-dependent’ class

*To whom correspondence should be addressed. Email: [email protected] Human Molecular Genetics, Vol. 12, No. 2 # Oxford University Press 2003; all rights reserved 126 Human Molecular Genetics, 2003, Vol. 12, No. 2 of mechanisms postulates that Tsix works as a functional RNA. In one scenario, the Tsix transcript may anneal to and mask the functional domain of Xist RNA, thereby preventing Xist RNA from making a complex with silencing protein partners. The hybridization of the complementary RNAs may also enhance the degradation of sense and antisense RNAs. In this type of RNA-dependent action, the repressive mechanism is stoichiometric rather than catalytic. Thus, the regulatory antisense RNA would be expected to occur at a molar excess over Xist RNA. All of these models have yet to be evaluated experimentally. Ultimately, the testing of each hypothesis will first require detailed characterization of Tsix RNA structure and quantita- Downloaded from https://academic.oup.com/hmg/article/12/2/125/608526 by guest on 27 September 2021 tion of sense/antisense RNA abundance. Although the full- length antisense transcript spans >40 kb (12), several spliced isoforms have been described (15). One additional isoform (15) initiates approximately 16 kb upstream of the previously reported Tsix start site (12). An interesting feature of all these isoforms is that splicing eliminates virtually all of the complementarity to Xist, with the notable exception of Xist’s 50 end. The relative abundance of spliced versus full-length transcript has not been determined. Furthermore, the functional relevance of spliced isoforms remains unclear. Some understanding of the antisense mechanism may be shed by cross-species comparison. By strand-specificRT–PCR analysis, human TSIX transcription reportedly terminates within the 30 end of XIST (within XIST intron 4) so that there is only partial overlap between the two genes (21). This observation has one of three implications: first, it may imply that the functional domains of Tsix/TSIX lie in its 50 half. Second, it may instead suggest that the mechanism of antisense regulation differs completely at the mouse and human loci. Finally, the critical domain of murine Tsix may actually reside in its 30 terminus, a possibility consistent with one interpretation that human TSIX has lost function (22). A 30 functional end is appealing in light of the fact that Xist’s silencing domain lies in a repeat sequence complementary to this 30 terminus of Tsix (23,24). Thus, to gain a better understanding of potential Tsix mechanisms, a more detailed structural and quantitative analysis of Tsix RNA is warranted and is achieved below.

Figure 1. Identification of a novel Tsix splice variant. (A) The strategy of Tsix RESULTS cDNA library construction. Tsix-specific RT primer is shown as an arrow. The DNA region used as a probe for colony hybridization is indicated as a Identification of a novel Tsix splice variant from an bi-directional arrow. Open tall rectangles and smaller gray ones represent Xist and Tsix exons, respectively. Symbols xE1, xE2-6, and xE7 represent Xist embryonic stem (ES) cell cDNA library exons and tE2, tE3, and tE4 are those of Tsix.B,Bam HI; X, Xho I restriction enzyme sites. (B) Resultant Tsix clones from library screening, shown as filled A striking feature of previously identified spliced Tsix RNAs is rectangles. Note the asterisk in clone B7, which represents newly identified their relatively short lengths and their minimal overlap with splice donor at the 30 end of Tsix exon 2. The 50 end of these cDNA clones Xist RNA (Fig. 1A). Notably, the region of complementarity are: base pair 1938 (clone 8B), 2362 (12B), 2876 (10B) of Genbank sequence lies exclusively at the 50 end of Xist (termed Tsix exon 4). accession no. L04961; base pair 77697 (9B), 77687 (D37), 77674 (B7), 79118 (C7 and B43) of Genbank no. X99946. The intron of C7 and B43 starts from Because the structural variations in Tsix RNA could have base pair 79839 of X99946. (C) PCR amplification of Tsix transcripts spanning implications for the antisense mechanism, here we endeavored DXPas34 locus and exon 4. M, 100 bp ladder marker. to determine if additional spliced variants existed, particularly longer isoforms which might have been missed by PCR-based methods such as rapid amplification of cDNA ends (RACE). may potentially interfere with first-strand synthesis. To select We constructed a Tsix-specific cDNA library from wild-type for higher molecular weight cDNAs, the inserts were size- male ES cells by placing a primer specific for the antisense fractionated prior to cloning. Positives were subsequently transcript near the Xho I site within Xist exon 1 (Fig. 1A, identified by colony hybridization using an Xist exon 1 probe arrow), a position that circumvented upstream repeats which (Fig. 1A, bi-directional arrow). Human Molecular Genetics, 2003, Vol. 12, No. 2 127 Downloaded from https://academic.oup.com/hmg/article/12/2/125/608526 by guest on 27 September 2021

Figure 2. 50 and 30 RACE to determine the structure of the novel splice variant. A detailed map of the Tsix clone C7 at DXPas34 locus and the 30 and 50-RACE products depicted as filled rectangles. The gene-specific primers used for RACE are shown as arrowheads. The 30 ends of 30-RACE clones are: base pair 649 (clone no. 5), 1503 (no. 7) of Genbank sequence accession no. L04961; base pair 79991 (no. 14) of Genbank no. X99946. The hexanucleotide polyadenylation sequence (ATTAAA) starts from base pair 79959 of X99946. The 50 ends of 50-RACE clones are: base pair 79428 (no. 19), 77777 (no. 101), 77675 (no. 105) of X99946. The single and double asterisks in the figure represent gaps of 34 bases (79311–79344) and 62 bases (78460–78521) each, which were found between the RACE clones and Genbank sequence X99946. The upstream intron is from base pair 78569 to 78942 of X99946. A, AgeI; B, BamHI; M, MluI; P, PstI; Sm, SmaI; Sl, SalI; X, XhoI restriction enzyme sites. Bars with dots on their top represents HpaII sites. Symbols xE1, xE2–6, and xE7 represent Xist exons. Symbols tE2, tE3, tE4, tE1a, and tE2a are Tsix exons.

After multiple rounds of screening, we isolated eight Tsix of the Xist gene body save 1.9 kb at Xist’s50-most terminus clones (Fig. 1B). Clones 8B, 12B and 10B are likely to be (exon 4). The intron–exon structure of C7 and B43 obeyed the prematually terminated cDNAs because no promoter activity GT–AG rule. Second, the 50 ends of clones C7 and B43 lay in has been described in this region (No transcriptional initiation DXPas34, suggesting either premature termination of ﬁrst- was observed in TsixDCpG ES cells; see below). The clones 12B strand synthesis (possibly due to the repetitive nature of and 10B contained an additional 0.4 and 0.9 kb of proximal DXPas34) or potential promoter activity within DXPas34.To sequence and could therefore either represent prematurely determine if this pattern of splicing occurred in vivo, we carried truncated full-length Tsix or longer spliced isoforms. The out RT–PCR of wild-type male ES cells using PCR primers in spliced variants 9B, D37 and B7 appeared to be similar to DXPas34 and Tsix exon 4. A product of the expected size was cDNA clones previously described (15), each originating in the obtained (Fig. 1C), suggesting that the novel splice junctions vicinity of the reported major transcription start site of Tsix are indeed utilized in vivo. Sequencing of the PCR product (12). However, there also appeared to be some variation in showed that it is identical to the corresponding region of clones splice junctions, as the 30 junction of exon 2 in clone B7 C7 and B43 (data not shown). Thus, our attempts to isolate included an additional four bases (asterisk in Fig. 1B). This novel Tsix splice variants produced one new species. variability is consistent with the previous report that the 30 end 0 of exon 2 could occur 17 bases upstream (exon 2 ) (15). We 50- and 30-RACE of the novel Tsix splice variant refer to this new exon 2 variant as exon 200. Finally, clones C7 and B43 were novel and were identical to In order to obtain the complete 50 structure of the novel species, each other. They differed from all previously described variants we carried out 50-RACE using a primer positioned downstream in two ways: ﬁrst, they included the DXPas34 repeat (13), a of DXPas34 (Fig. 2, right-directed arrowhead). We obtained region in which CTCF binding sites have been described (19). three distinct clones (nos 19, 101 and 105). Clone no. 19 was As in all prior Tsix isoforms, C7 and B43 spliced out almost all truncated at the 50 end. The truncation may be due to an artifact 128 Human Molecular Genetics, 2003, Vol. 12, No. 2 Downloaded from https://academic.oup.com/hmg/article/12/2/125/608526 by guest on 27 September 2021 Human Molecular Genetics, 2003, Vol. 12, No. 2 129

Figure 3. Determination of splicing efficiency by RNAse protection assays. (A) RPA at the 30 portion of Tsix exon 2 and exon 3. The probe TxE2 yields 245 nt signal for unspliced transcripts and 166, 149 or 170 nt signal for spliced ones corresponding to exon 2, 20 or 200, respectively (arrowheads). The original transcript size of the probe is 332 nt. Probably, the probe E3SD yields 501 and 110 nt signals for unspliced and spliced transcripts, each (arrows). The latter corresponds to exon 3. The original size of E3SD riboprobe is 518 nt. Gray rectangles in the map represent Tsix exon 2 (tE2) and exon 3 (tE3). (B) RPA at the 30 part of DXPas34 locus. Gray rectangles in the map show newly identified Tsix exons 1a (tE1a) and 2a (tE2a). The C7PM probe yields 293 and 176 nt signals for unspliced and spliced transcripts, respectively. The original size of C7PM riboprobe is 343 nt. The position of uncharacterized third signal was shown as an arrowhead. (C) RPA at the 50 boundary of Tsix exon 4 (tE4). The probe TxE4 yields 186 nt signal for unspliced transcripts and 110 nt signal for spliced ones, while its original transcript size including multi cloning site of the vector is 273 nt. Various ES cell lines either undifferentiated (at day 0, d0) or partially differentiated (for 4 days, d4), and adult mouse liver RNA were examined. Yeast RNA with or without RNase digestion was used as controls (Yþor Y). The open larger rectangle and gray smaller one represent Xist exon 1 and Tsix exon 4, respectively. The faint band about 273 nt in size observed in a lane of TsixEF-1a-female ES/d0 is likely to be a remnant of undigested probe. A, AgeI; B, BamHI; D, DraII; M, MluI; P, PstI; Sm, SmaI; Sl, SalI; X, XhoI restriction enzyme sites. Bars with dots on their top represents HpaII sites. DCpG, TsixDCpG; EF-1a, TsixEF-1a; WT, wild-type; M, 100 nt marker. (D) The estimation of the amount of unspliced (unspl) versus spliced (spl) Tsix transcripts at each exon–intron boundary. The signal volume of the bands marked as *1 to *4 (see C and A) was measured and the relative copy number was calculated and shown. Downloaded from https://academic.oup.com/hmg/article/12/2/125/608526 by guest on 27 September 2021 of premature termination by reverse transcriptase or may reflect placing probes at known Tsix exon–intron boundaries, the possible transcription initiation within DXPas34. The 50 ends of amount of splicing at each exon–intron junction could be two other clones, nos 101 and 105, closely coincided with the estimated by comparing the intensity of the bands corres- known major Tsix start sites (12). Aside from the 50 termini, the ponding to fully protected (unspliced) versus partially protected sequences of these two clones were identical to each other and (spliced) probes. In each assay, we examined the extent spanned all of the DXPas34 repeat. The repeat sequence of splicing in ES cells of four genetic backgrounds: wild- differed slightly from the published genomic sequence type male (129 strain), wild-type female (hybrid 129 Â (Genbank accession no. X99946) in having two gaps (asterisks M.castaneus), mutant male carrying the TsixDCpG null geno- in Fig. 2), most likely due to genomic sequence polymorphism type (14), and mutant female carrying the TsixEF1a gain- rather than to RNA processing, because these gaps did not fit of-function genotype (17). The specificity of RPA bands for the GT–AG rule and coincided precisely with a single unit of Tsix was inferred from their presence in wild-type ES cells and the DXPas34 repeat. Nonetheless, some splicing activity was absence from the TsixDCpG mutant, somatic cells and yeast. evident within DXPas34 (Fig. 2) and occurred in accordance First, we examined the highly heterogeneous 50 end using with the GT–AG rule. Thus, in addition to the splicing patterns RPA probes, E3SD and TxE2, which spanned sequence just previously reported (15), our data suggested that at least one downstream of the Tsix major starts and overlapped exons 2 new splice variant includes the DXPas34 sequence. We name and 3 (15) (Fig. 3A). When the TxE2 probe was used, we found the novel DXPas34 exons of this new species ‘exon 1a’ and that the majority of RNA was not spliced at this junction, as the ‘exon 2a’ (Fig. 2). fully protected band of 245 nucleotides (nt) was more To determine the 30 structure of the novel Tsix isoform, we prominent than any band corresponding to the spliced variants performed 30-RACE using two gene-specific primers posi- (170, 166 and 149 nt) (Fig. 3A, arrowheads). When the E3SD tioned downstream of DXPas34 (Fig. 2, left-directed arrow- probe was used, we found that the unspliced form (501 nt) was heads). Three products were of the approximate sizes, 1.5 again most abundant (Fig. 3A), while the spliced form was less (no. 5), 0.7 (no. 7), and 0.4 kb (no. 14) (Fig. 2). Sequence detectable (110 nt). The minimal overlap between the probe and analysis revealed that the splicing pattern of clone nos 5 and 7 exon 2 probably made signals from exon 2 variants hard to was identical to that of C7 in that Tsix was spliced from detect. Additional bands between 200 and 400 nt might be DXPas34 to exon 4. Notably, the 30 ends of both clones attributed to RNA secondary structure or to sequence occurred within blocks of A-rich sequence (bp 648–629 and differences between 129 and M.castaneus (since some were bp 1502–1493 of Genbank sequence accession no. L04961), observed only in the hybrid female line). However, they may perhaps indicating mis-priming of the oligo-dT primer also reflect as yet unidentified splice variants. during first-strand synthesis (25). (This sequence is unlikely We next asked how much splicing takes place at the to be a true poly-A site because it does not conform to the DXPas34 boundary (Fig. 3B). At this junction, the unspliced poly-A consensus.) Clone no. 14 extended further 30 for form (293 nt) was once again more abundant than the spliced 300 bp. A polyadenylation signal (ATTAAA) (26) occurs in variant (176 nt). The RPA also revealed a third band of 120 nt this region, so the 30 end of clone no. 14 may actually (arrowhead), suggesting yet more possible splicing variation at represent one poly-A site for Tsix RNA. We termed this exon the boundary of DXPas34. To quantitate the extent of splicing, 2a variant exon 2a0. we determined the relative signal intensities of fully protected versus spliced bands by phosphorimaging and normalization How much of Tsix RNA is spliced? for number of adenine bases within the target sequence. This analysis showed that there is at least two-fold more unspliced As a first step towards understanding the significance of Tsix RNA (Fig. 3D). Based on relative band intensities by RPA (data splicing, we next determined what fraction of Tsix is actually not shown) and the results of cDNA cloning (Fig. 1B), splicing spliced by adapting the RNase protection assay (RPA). In the at this junction appeared to take place at a comparable RPA technique, hybridization of a strand-specific RNA probe frequency relative to exons 2 and 3, suggesting that the novel to target cellular RNA protects the annealed strands from splice variant may be as abundant as the previously reported RNase digestion, thereby providing information regarding both Tsix isoforms. These findings further argue for heterogeneity of the length and the relative quantity of target RNA (27). By splicing at the 50 end of Tsix. 130 Human Molecular Genetics, 2003, Vol. 12, No. 2

It should be noted that even greater splicing variation may Table 1. Quantitation of Tsix and Xist RNA at various genomic positions exist at the 50 end. Because our cloning strategy involved a size-selection step, splice variants of both lower and higher Amplicona DCpG male WT male WT female molecular weights may have been missed. Indeed, RNase 1 4.0 1.4 102b 8.9 8.9 101 5.2 2.5 102 protection analysis occasionally yielded specific bands that 2NAc 1.9 0.5 103 2.6 0.6 103 c 3 3 were not of predicted molecular weights (Fig. 3A and B), 3NA 1.5 0.5 10 1.9 0.6 10 4NDd 1.5 3.1 101 0.5 1.4 102 suggesting that additional splice junctions may have been 0 2 2 0 5 2.9 0.9 10 1.5 0.5 10 2.7 0.6 10 utilized at the 5 end. Regardless, the results of RPA indicate 6 7.7 5.6 101 8.1 1.2 101 1.2 1.4 102 that a large fraction of the antisense transcript remains 7 1.3 2.5 101 6.1 2.8 102 7.1 3.2 102 8 4.5 2.3 100 2.2 0.3 102 7.9 2.9 102 unspliced at each known exon–intron boundary. 0 2 2 We finally examined the 30 end of Tsix using probes which lie 9 1.7 1.8 10 1.1 0.3 10 3.3 4.3 10 10 2.5 1.9 100 2.7 0.8 102 3.8 0.5 102 at the upstream junction of exon 4 (Fig. 3C). Two specific 0 2 2

11 2.8 2.2 10 1.5 0.3 10 2.6 0.7 10 Downloaded from https://academic.oup.com/hmg/article/12/2/125/608526 by guest on 27 September 2021 signals were detectable in male and female ES cells (Fig. 3C), a 12 2.8 0.8 102 1.0 0.4 101 8.5 6.6 101 186 nt band corresponding to the unspliced RNA and a 110 nt band for the spliced variant. As their intensities were not so aAmplicons are defined in Fig. 4. different, it was likely that there were comparable amounts of bValues are expressed as copy number/50 ng of total RNA. spliced and unspliced transcripts at this exon boundary. cNot applicable for corresponding genomic loci are deleted in TsixDCpG cells. d Quantitation by phosphorimaging demonstrated that there Not detected. was either equal amounts of (wild-type female) or more (wild-type male) spliced RNA as compared with unspliced (Fig. 3D). Thus, Tsix RNA is more efficiently spliced at exon 4 cells was slightly higher; we believe that much of this may be than its 50 exons. Nevertheless, a significant fraction at the 30 due to the inevitable presence of differentiating cells which end is also not processed. To determine whether the extent of would exhibit high-level Xist expression.) Thus, with respect to splicing changes at the onset of XCI, we compared the RNA Tsix’s50 portion, there was a 100-fold molar excess of Tsix ratio between undifferentiated (day 0) and partially differen- RNA over Xist RNA. With respect to Tsix’s30 end, there was a tiated (day 4). We could not see any significant difference in the 10-fold excess. These results indicated that there is indeed a ratio of unspliced versus spliced signal intensity (Fig. 3C). significant excess of Tsix RNA over Xist RNA. These results suggest that splicing at exon 4 of Tsix RNA Finally, we examined how Xist levels are altered in the occurs more efficiently than at the 50 portion and that the extent absence of Tsix RNA. Previous studies have shown that, even of splicing does not change during differentiation. in undifferentiated ES cells, the loss of Tsix expression resulted in increased steady-state Xist levels (14,28). Here, we first used The stoichiometry of Tsix and Xist RNA real-time RT–PCR to more accurately quantitate the degree of upregulation in the mutant TsixDCpG cell line and found a At the onset of XCI on the future Xi, the downregulation of 10-fold increase in Xist RNA levels (Table 1, wild-type male Tsix enables Xist to transition from a low to high expression versus TsixDCpG male ES cells). However, despite this order of state in cis. Current models of Tsix function invoke either magnitude increase, Xist RNA neither spread along the X nor an RNA-based mechanism or a mechanism based rather on initiated silencing (14). We next asked if Xist levels further transcriptional action. If Tsix functioned as an antisense RNA increased upon cell differentiation and the onset of XCI. We which titrates out sense RNA, its stoichiometry would be found that somatic female cells (liver) showed a further crucial to its function. upregulation (about 100-fold, data not shown). This suggests Here, we used real-time RT–PCR analysis to estimate Tsix the transition from low to high level state involves increase in RNA copy numbers in undifferentiated wild-type male, wild- steady-state Xist RNA amount, a result roughly consistent with type female, and TsixDCpG male ES cells using primer pairs the previous estimate (17,29). Taken together, the results of placed at 11 positions across the locus (Fig. 4A, Table 2). In quantitative real-time RT–PCR demonstrated that, prior to the this method, the starting template amount is quantitated against onset of XCI, Tsix RNA is indeed in great molar excess over a synthetic control template whose copy number is determined Xist RNA and that its loss of expression correlates with Xist’s by spectrophotometry. The results revealed a gradient of RNA transition to a high level state on the Xi. abundance, with 10 times more transcript at the 50 portion as compared with the 30 end (Fig. 4B). This suggested that a significant portion of Tsix transcription terminated early before crossing the Xist gene body. Relevant to this, the transcript DISCUSSION levels at positions 4–6 were low at steady state. Aside from A synthesis of Tsix genetic studies has led to the ‘transcrip- transcription termination in the middle, the low copy number tion-dependent’ and ‘RNA-dependent’ models of antisense could reflect intronic RNA degradation since these positions regulation (8,9,12). Although not mutually exclusive, the two occur in a Tsix ‘intronic’ region. models differ in significant ways and make clear predictions. To determine if Tsix RNA exists at a molar excess relative to The transcription-dependent models invoke topological and Xist, we estimated Xist RNA abundance in undifferentiated ES steric constraints on Xist imposed by antisense transcriptional cells, cells which have not undergone XCI. We found that activity and do not require the antisense RNA product. In undifferentiated ES cells contained less Xist RNA than Tsix contrast, the RNA-dependent models postulate a repressive (Fig. 4B, Table 1). (Note: the Xist copy number in female ES role for Tsix RNA and posit a stoichiometric titration of Xist Human Molecular Genetics, 2003, Vol. 12, No. 2 131

Table 2. Strand-speciﬁc primers, PCR primers, and template plasmid constructs used in real-time PCR analyses for Tsixa

Amplicon Position Strand-specific primer PCR primers Template plasmid construct 1 Tsix exon 1 — TCA GTT TGA GTA CAG ACA CCA GGC pB1/C7.E14.10Bb GAC AGA GTG AAA ATC CGG AAG TTG 2 Tsix exon 2 — AAA GTA CCT GCA AGC GCT ACA CAC pB1/C7.E14.10Bb TGG CTA TCA CGC TCT TCT TCC ATC 3 DXPas34 — AAA GCG TTC AAT AAG CCT GGC GTG pB1/C7.E14.10Bb TTC TAG ACC CTG CTA CAA GTC ACG 4 Xist exon 7 AAC GGA TGC GGA ATA CAG CAT GTC ACA ATG AGT CCG TGT GTG pCC1 (4219-13672 of Genbank U41394) CTA CAC GGG TAT AAC ACC CAA ACC

5 Xist intron 5–6 — CTT TCC TGG CAT TCA TAC AGC TTC pCC1 (4219-13672 of Downloaded from https://academic.oup.com/hmg/article/12/2/125/608526 by guest on 27 September 2021 Genbank U41394) TCA GAA GAC TGC AAT GCT AGT GAC 630 of Xist exon 1 CCC TCC TTT CCA CTC TTT GTA TCT CTT GCA CTC CTG CAA GGG ppXB8 (1877-6948 of Genbank L04961) AGG TGG CAG TGC ATA CGC ATA CAT 750 of Tsix exon TTT TTC GAA GTG CCT GCC GGA GAG CGC ATG CTT GCA ATT CTA pB1/10B (803-2876 of 4–intron Genbank L04961) GTA AGA GAC TAT GAA CGC AAG CGG 8 Tsix exon 4 GTG TGA GTG AAC CTA TGG TTA AAC TGA GTG GGT GTT CAG GGC ppXB1 (680-4210 of Genbank AF138745) CTA CCA CAA ATC AAG GCG AAT CCC 9 Tsix exon 4 TCT TCC GTG GTT TCT CTC TTT TAG TTC CCC TAC CAC CAA GCC ppXB1 (680-4210 of Genbank AF138745) GGT AAG TAT CCA AAA CCC CGT TGG 10 Tsix exon 4 — GTA CGT AAG CTC AGT GAC ATG ACG pGEM/4/4 (2718-4260 of Genbank AF138745) TCT ACC CTT TCC TCT CCT CAT CTG 11 Tsix exon 4 — ACC GTG TAC ATC AAG GTA TGT CAG pGEM/4/4 (2718-4260 of Genbank AF138745) CAC TGT CGG TCA CTG TTC AGA TAC 12 (Xist) Xist exon 1–3 — CCC GCT GCT GAG TGT TTG ATA TG pXPC2 (5763-11297 of Genbank L04961) CAG AGT AGC GAG GAC TTG AAG AG b-actin — ACA CGC AGC TCA TTG TAG ATG GAT GAC GAT ATC GCT GC pGEM/b-actin (81-1208 of X03672) AGA TCT TCT CCA TGT CGT CC aAll the sequences are presented from 50 to 30 direction. bFor details, see Materials and Methods.

RNA by the antisense transcript. As a step towards Recently, sense and antisense transcripts produced within distinguishing between these potential mechanisms, we have centromeric repeats of Schizosaccharomyces pombe have been characterized Tsix RNA structure and quantitated RNA proposed to give rise to siRNAs which in turn regulate abundance in vivo (Fig. 5). Several conclusions can be drawn heterochromatic silencing at the centromere (34,35). However, from our results. at the Xist/Tsix locus, such siRNAs have not been found despite First, Tsix RNA occurs at a large molar excess over Xist some effort (S. Shibata, unpublished observations). An RNA. Quantitative real-time RT–PCR analysis indicated that, alternative means by which Tsix RNA could titrate Xist RNA in undifferentiated ES cells, Tsix RNA is present at >10–100- could be a masking of functional domains in Xist RNA. Recent fold molar excess over Xist RNA per chromosome. Thus, if work (24) demonstrated that Xist RNA contains a silencing Tsix worked as an RNA entity, the local concentration of Tsix domain located at the 50 terminus and a chromosome-binding would be sufﬁciently high to bind and titrate out Xist RNA. At domain more broadly distributed along the middle stretch of the molecular level, titration could involve enhancement of Xist the 17 kb RNA. As will be further discussed later in this RNA degradation through the Xist : Tsix double-stranded RNA section, both the Tsix splicing patterns and its premature (dsRNA) intermediate. The modulation of Xist RNA stability transcription terminations may provide additional clues to has been proposed to mediate the post-transcriptional regula- the molecular basis of Tsix action. tion of Xist expression (30,31). This has led to much The quantitative analysis demonstrated a strong inverse speculation about the potential involvement of RNA inter- correlation between Tsix and Xist levels. In cells deleted for the ference (RNAi) (32), an anti-viral and anti-transposon cellular antisense gene, Xist RNA levels increased by 10-fold. defense mechanism in which aberrant dsRNAs are cleaved into Interestingly, this increase is not accompanied by XCI (14). 21–25 nt small interfering RNAs (siRNA) that then direct In female somatic cells that have undergone XCI, Tsix becomes sequence-speciﬁc degradation of homologous RNAs (33). undetectable while Xist RNA increases by approximately two 132 Human Molecular Genetics, 2003, Vol. 12, No. 2 Downloaded from https://academic.oup.com/hmg/article/12/2/125/608526 by guest on 27 September 2021

Figure 4. Quantitation of Tsix transcripts by strand-specific real-time RT–PCR analyses. (A) The positions of amplicons examined. The symbol at each amplicon shows the type of template used, i.e. filled circles mean random primed cDNA and filled triangles represent strand-specific cDNA (antisense). The amplicons 1–3 for Tsix exon 1, 2, and the DXPas34 downstream exon (exon 2a), respectively, were placed inside each exon. Therefore, they allow amplification of both spliced and unspliced transcripts. The amplicon 7 spans across the 50 boundary of Tsix exon 4 and represents only unspliced transcript. The Xist copy number was examined using random primed cDNA at amplicon 12 (open circle), whose primers were placed in 30 of Xist exon 1 and in exon 3. pA, polyadenylation signal (AATAAA). The asterisk adjacent to one of the pA means there are three pA signals positioned closely. Large open rectangles, Xist exons (xE1, xE2–6, and xE7); middle gray rectangles, Tsix exons (tE1, tE2, tE3, and tE4); small filled rectangles, newly identified Tsix exons in DXPas34 locus (tE1a and tE2a). (B) The average copy number of Tsix and Xist transcripts in various ES cell lines. The mean number and standard deviations are presented in Table 2. The light gray round columns represents male TsixDCpG ES cells, dark gray round columns male wild-type ES cells, and black square columns female wild-type ES cells. DCpG, TsixDCpG; WT, wild-type. orders of magnitude. These results suggest that the repression 2a. Interestingly, these novel exons span DXPas34, a repeat of Tsix is only sufficient to partially upregulate Xist. Additional sequence that was recently shown to contain binding sites for events at the onset of cell differentiation must occur to fully the transcription factor and chromatin insulator, CTCF (19). stimulate Xist expression and initiate XCI. Quantitative analyses by RPA suggested that this variant is not A second conclusion of this work is that Tsix RNA is so different in abundance from isoforms containing exons 2 or heterogeneously spliced at the 50 end. By employing a cDNA 3 (15). We also showed that the pattern of splicing to exon 4 cloning strategy aimed at identifying larger splice variants, we does not obviously change upon the onset of XCI. The discovered one new species which contains novel exons 1a and possibility of additional splicing variation at the 50 end is not Human Molecular Genetics, 2003, Vol. 12, No. 2 133 Downloaded from https://academic.oup.com/hmg/article/12/2/125/608526 by guest on 27 September 2021

Figure 5. Summary of findings. (A) Deduced levels of antisense transcripts based on real-time PCR data in wild-type male ES cells. Targeting schemes and Xist functional domains in other studies are shown for comparison. þ,0–500; þþ, 500–1000; þþþ, >1000 copies/50 ng of total RNA. The amount at 30 of DXPas34 exon and at 50 of exon 4 was based on both real-time PCR and RPA. The result in wild-type female ES cells is essentially the same. Large open rectangles, Xist exons; middle gray rectangles, Tsix exons; small filled rectangles, newly identified Tsix exons in DXPas34 locus. SA/pA, splice accepter and pA cassette; EF-1a, elongation factor 1a promoter; Sc, ScaI; Sm, SmaI; Sl, SalI; M, MluI; H, HindIII. (B) Summary of Tsix splicing variants revealed by this study and others. (a) Previously reported Tsix exons (15). Note the two transcriptional start sites. Exon skipping is also observed. (b) DXPas34 variant of Tsix including exon 4 or (c) truncated form. (d) Unspliced Tsix. Open rectangles, Xist exons (xE1, xE2–6, and xE7); filled rectangles, Tsix exons. Numbers in the figure represent Tsix exons. excluded by our study. In fact, low levels of specific but target of binding for chromatin-associated proteins (24). By unaccountable bands in RPA analysis very likely suggest virtue of its complementary to the Xist silencing domain, the greater variability in splice site usage than presently measur- spliced Tsix variant may be instrumental in masking this able. The heterogeneity of splicing must be factored into domain. In this model, one facet of Tsix’s repressive action models of Tsix mechanism. would result from direct base-pairing between antisense and Third, although the functional relevance of splicing remains sense RNAs. On the other hand, Tsix transcripts which are not unknown, a most striking feature is that all of the spliced spliced at exon 4 may participate in regulating other aspects of variants identified to date share exon 4, the only exon that Xist action. Indeed, other domains within Xist RNA are occurs within the region of complementarity between Xist and necessary for its cis-localization to the Xi (24,36) and for Tsix (15). Interestingly, this region coincides with a region of targeting of the Xi-enriched protein, macroH2A (37). It is Xist that is required for silencing and has been proposed to be a possible this localization is blocked by Tsix transcripts that 134 Human Molecular Genetics, 2003, Vol. 12, No. 2 remain full-length or even by transcripts that terminate early. complementary between mouse Xist and Tsix RNAs were This repression may be achieved by direct RNA base-pairing or crucial for regulation, a similar mechanism might be utilized by by the indirect action of antisense transcriptional process. human TSIX. Thus, further examination of the 30 terminus of Further work will be required to distinguish between the ‘RNA- human TSIX will be of special interest in the future. As an dependent’ and ‘transcription-dependent’ models. increasing number of antisense genes has come to light in Fourth, a significant fraction of Tsix RNA persists in the recent years (38–43), a clearer understanding of how Tsix/TSIX unspliced state. The results of RNase protection analyses regulates Xist/XISTwill have implications beyond the control of revealed that only 30–60% of Tsix RNA is actually spliced at X-chromosome inactivation. measurable exon–intron boundaries. It is certainly possible that splicing at other junctions was missed by our analysis—splice variants of large size would have been very difficult to clone MATERIALS AND METHODS and their splice site usage would not have been factored into Downloaded from https://academic.oup.com/hmg/article/12/2/125/608526 by guest on 27 September 2021 our calculations. Nonetheless, the data showed clearly that only Construction and screening of the cDNA library for a fraction of total RNA is spliced at the exon 4 junction and novel Tsix variants more than half of Tsix transcripts remain unspliced at the 50 end. The persistence of unspliced Tsix RNA seems surprising A Tsix-specific cDNA library was prepared using the given that other mammalian mRNAs are very efficiently SuperScript Choice System (Life Technologies) according to spliced to the extent that very little unprocessed RNA is the manufacturer’s instruction. Briefly, the first strand was detectable at steady state. Therefore, this persistence may have synthesized using a Tsix-specific primer (50-GTG TGA GTG significant implications for how Tsix works. If Tsix action AAC CTA TGG-30) from 20 mg of total RNA extracted from J1 requires its RNA product, the comparable abundance of spliced male ES cells (44). For efficient first-strand elongation, the and unspliced forms leaves open possible roles for both forms reverse transcriptase reaction was carried out at 50C for 4 h, in regulating Xist expression. Spliced and unspliced forms may adding 200 units of the enzyme every hour. Subsequently, the in fact play different roles. For example, splicing to the second strand was synthesized and the cDNA ligated to Eco RI invariant exon 4 junction may block the silencing activity at the adapters. The cDNA pool was then selected on a size-exclusion 50 end of Xist RNA, while the unspliced variants (many of column, from which the first nine fractions (highest molecular which terminate before reaching the 50 end of Xist RNA) may weight fractions) were chosen for further work. These fractions block Xist RNA localization to the X. We must also consider were dominated by cDNA clones of 1.0 kb or greater in size. the possibility that the operational domain of Tsix lies Eco RI-adapted cDNA molecules were ligated to Eco exclusively in its 50 half, especially given a recent report that RI-digested pBluescript vector and electroporated into E.coli human TSIX terminates in the 30 exons of XIST (21). competent cells (Electromax DH10B cells, Life Technologies). Indeed, our findings in mice have possible implications for Tsix cDNA clones were isolated by colony hybridization human TSIX structure and function. The identification of the screenings using random-primed Xho I-Bam HI 1 kb fragment human TSIX homolog has raised the possibility that human from Xist exon 1 as a probe. Sequencing was carried out by the XIST expression is regulated by a similar antisense mechanism DNA sequencing core facility at the Massachusetts General (21). However, the human gene has been reported to differ from Hospital. murine Tsix in at least one significant way: strand-specificRT– PCR analysis showed that human TSIX RNA is detectable only PCR and RACE as distally as XIST intron 4 (21). This result was interpreted to mean that TSIX terminates in XIST intron 4 and that full-length PCR amplification of the Tsix clone C7 was performed with TSIX and XIST RNAs only partially overlap. This interpretation primers rt50R (50-GGA GAG CGC ATG CTT GCA ATT CTA- could have one of several implications. First, the critical 30) and TRXp (50-AAA GCG TTC AAT AAG CCT GGC domain of TSIX function may lie in the region of overlap with GTG-30) using random-primed cDNA from J1 cells as XIST exons 5–7. Second, TSIX may not work as a functional template. The amplification condition was 94C, 9 min; RNA but may instead exert its repressive action by antisense (94C, 30 s; 66C, 30 s; 72C, 1 min) Â 39 cycles. 50- and 30- transcription per se. Finally, unlike mouse Tsix, human TSIX RACE were carried out using the GeneRacer Kit (Invitrogen, may have no function in regulating XIST (22). CA, USA). Briefly, for 50-RACE total RNA from J1 cells was However, relevant to these interpretations, our current study reverse-transcribed with a random primer or a Tsix-specific demonstrated a diminishing gradient in murine Tsix transcript primer (rt50R) and amplified by nested PCR. In the first round, copy numbers, with the highest level of expression at the 50 end the template was amplified by touch-down PCR with TFXp2 (Fig. 4B). This suggests that a considerable fraction of total primer (50-ACG CCA GGC TTA TTG AAC GCT TTG-30) and transcript terminates early without crossing the Xist gene body. the 50 primer (supplied by kit) using the following conditions: Thus, it must be considered that human TSIX may also exhibit a 94C, 7 min; (94C, 30 s; 72C, 30 s; 72C, 3 min) Â three gradient of expression and that the lack of RT–PCR detection cycles; (94C, 30 s; 70C, 30 s; 72C, 3 min) Â three cycles; beyond XIST intron 4 may reflect a relative decrease in steady- (94C, 30 s; 68C, 30 s; 72C, 3 min) Â three cycles; (94C, state RNA copy number. A quantitative approach to studying 30 s; 66C, 30 s; 72C, 3 min) Â 36 cycles using human TSIX expression will be critical to address whether the AmpliTaqGold Taq DNA polymerase (Roche). In the second human X-inactivation center (XIC) is truly regulated differently round, the first PCR product was diluted 1/100 and amplified from murine Xic. Indeed, human XIST and mouse Xist RNAs with TFXp2 primer and the 50 nested primer (kit) using the share the required silencing domain at the 50 terminus. If the same conditions. For 30-RACE, the oligo-dT primed cDNA was Human Molecular Genetics, 2003, Vol. 12, No. 2 135 amplified with the 30 primer (kit) and TRXp primer using the Table 2 was used for strand-specific Tsix cDNA synthesis. The following conditions: 94C, 9 min; (94C, 30 s; 72C, 30 s; plasmid construct pBl/C7.E14.10B was made by introducing 72C, 3 min) Â three cycles; (94C, 30 s; 70C, 30 s; 72C, the inserts of library clone C7 (base pair 1112–2828 of 3 min) Â three cycles; (94C, 30 s; 68C, 30 s; 72C, 3 min) Â Genbank accession no. AF541962), 10B (base pair 803–2876 three cycles; (94C, 30 s; 66C, 30 s; 72C, 3 min) Â 26 cycles. of Genbank accession no. L04961), and PCR amplified DNA Subsequently the PCR product was diluted 1/100 and amplified fragment including Tsix exon 1, 2, 3 and 50 end of exon 4 (base again with the 30 nested primer (kit) and TRXn primer (50-AGT pair 1–732 of Genbank accession no. AF138745) into TAA GGG CGT GAC TTG TAG CAG-30) as follows: 94C, pBluescript vector. pB1/C7.E14.10B was digested with NotI 9 min; (94C, 30 s; 66C, 30 s; 72C, 3 min) Â 25 cycles. The and SpeI restriction enzyme to separate the three inserts for final PCR products were purified by gel-electrophoresis, ligated standard curve preparation. The amount of sample cDNA to the plasmid vector, and sequenced. template used in each PCR reaction was an equivalent to that converted from 50 ng of total RNA. The PCR amplification was

1 Downloaded from https://academic.oup.com/hmg/article/12/2/125/608526 by guest on 27 September 2021 RNase protection assay carried out with the SYBR green PCR master mix (Applied Biosystems, Warrington, UK) using the iCycler iQTM real-time Total RNA was extracted using Trizol reagent (Life detection system (BIO-RAD Laboratories, Inc., Hercules, CA, Technologies) from wild-type male ES cells (J1), wild-type USA) using the following conditions: 95C, 9.5 min; (95C, female ES cells (16.6) (12), mutant male ES cells carrying 30 s; 66C, 30 s; 72C, 1 min) Â 40 cycles. Since SYBR green the TsixDCpG allele (CG7) (14), and mutant female ES cells was used to detect amplified products, the specificity of each carrying a gain-of-function TsixEF-1a allele (2A1.H12) (17). PCR reaction was ensured by either melting curve profiles (in Hereafter, the CG7 and 2A1.H12 ES cells are abbreviated as which a single curve demonstrated the presence of only one TsixDCpG and TsixEF-1a ES cells, respectively. In the differ- duplex DNA species) or by agarose gel electrophoresis of PCR entiation condition, ES cells were cultured without feeder cells products in test runs (which showed the amplification of a in the media lacking LIF. The assay was performed using RPA single correctly sized band). The concentration of plasmid III kit (Ambion). A 10 mg aliquot of total RNA was hybridized DNA used as template was determined based on the spectro- with 5 Â 104 cpm of riboprobe in each reaction. The plasmid photometric absorbance at 260 nm. The 1:10 serial dilution constructs utilized for probe preparation were: Nco I-digested from 107 to 102 copies/ml were prepared for each template in pGEM/TxE4 (50 boundary of Tsix exon 4, base pair 1830–2014 order to make the standard curve in which the threshold cycle of Genbank sequence accession no. L04961) (6), Nco I- showing the first cycle number to detect SYBR green digested pGEM/TxE2 (30 boundary of Tsix exon 2, base pair fluorescence was plotted. The copy number of Tsix and Xist 77759–78003 of Genbank accession no. X99946) (45), Bam transcripts in wild-type and mutant samples was estimated by HI-digested pBl/E3SD (Tsix exon 3 and adjacent sequence, comparison to the standards described above and subsequently base pair 77848–78349 of X99946), and Nco I-digested normalized to b-actin. Every PCR reaction was triplicated and pGEM/C7PM (30 boundary of DXPas34 in relation to Tsix repeated at least twice. start site, base pair 79663–79954 of X99946). The former two plasmids were obtained by TA cloning following PCR, and the latter two were generated by subcloning SmaI–EcoO109I ACKNOWLEDGEMENTS 0.5 kb fragment and PstI–MluI 0.3 kb fragment into pBluescript or pGEM5 vector, respectively. The radioactive We thank N. Stavropoulos and Y. Ogawa for plasmid constructs riboprobes were transcribed with either SP6 or T7 RNA and helpful advice. We are also grateful to M. Donohoe, polymerase using [a-32P] UTP and gel-purified thereafter. The K. Huynh and Y. Ogawa for critical reading of the manuscript. signal intensity of protected bands was measured by S.S. is supported by the fellowship program from the Japan Phosphoimager instrument (Molecular Dynamics). The relative Society for Promotion of Science. J.T.L. is an assistant copy number of unspliced versus spliced Tsix variants was investigator of the Howard Hughes Medical Institute. This calculated with the following formula: (volume of unspliced work was supported by NIH grant RO1GM58835. band)/(number of adenine residues in both exon and intron part of the corresponding probe sequence) versus (volume of spliced band)/(number of adenine residues in exon part of the REFERENCES corresponding probe sequence). 1. Lyon, M. (1961) Gene action in the X-chromosome of the mouse (Mus musculus L.). Nature, 190, 372–373. Real-time PCR 2. Rastan, S. and Brown, S.D. (1990) The search for the mouse X-chromosome inactivation centre. Genet. Res., 56,99–106. See Table 2 for strand-specific first-strand primers, PCR 3. Brown, C.J., Lafreniere, R.G., Powers, V.E., Sebastio, G., Ballabio, A., primers and template plasmid constructs for standard curve Pettigrew, A.L., Ledbetter, D.H., Levy, E., Craig, I.W. and Willard, H.F. (1991) Localization of the X inactivation centre on the human preparation. The random-primed cDNA templates for real-time X chromosome in Xq13. Nature, 349,82–84. PCR analyses were prepared from DNase I-treated total RNA 4. Lee, J.T., Strauss, W.M., Dausman, J.A. and Jaenisch, R. (1996) A 450 kb extracted from the ES cell lines using the Supercript first-strand transgene displays properties of the mammalian X-inactivation center. synthesis system for RT-PCR (Life Technologies). The strand- Cell, 86,83–94. specific cDNA templates were prepared in the same way except 5. Brown, C.J., Ballabio, A., Rupert, J.L., Lafreniere, R.G., Grompe, M., Tonlorenzi, R. and Willard, H.F. (1991) A gene from the region of the the reaction was carried out at 50 C in order to avoid human X inactivation centre is expressed exclusively from the inactive nonspecific priming. A mixture of oligonucleotides shown in X chromosome. Nature, 349,38–44. 136 Human Molecular Genetics, 2003, Vol. 12, No. 2

6. Brockdorff, N., Ashworth, A., Kay, G.F., McCabe, V.M., Norris, D.P., 27. Gilman, M. (1987) Ribonuclease protection assay. In Ausubel, F.M., Cooper, P.J., Swift, S. and Rastan, S. (1992) The product of the mouse Xist Brent, R., Kingston, R.E., More, D.D., Seidman, J.G., Smith, J.A. and gene is a 15 kb inactive X-speciﬁc transcript containing no conserved ORF Struhl, K. (eds), Current Protocols in Molecular Biology. John Wiley, and located in the nucleus. Cell, 71, 515–526. Hoboken, NJ, Vol. 1, pp. 4.7.1–4.7.8. 7. Clemson, C.M., McNeil, J.A., Willard, H.F. and Lawrence, J.B. (1996) XIST 28. Morey, C., Arnaud, D., Avner, P. and Clerc, P. (2001) Tsix-mediated RNA paints the inactive X chromosome at interphase: evidence for a novel repression of Xist accumulation is not sufﬁcient for normal random X RNA involved in nuclear/chromosome structure. J. Cell Biol., 132, inactivation. Hum. Mol. Genet., 10, 1403–1411. 259–275. 29. Buzin, C.H., Mann, J.R. and Singer-Sam, J. (1994) Quantitative RT–PCR 8. Boumil, R.M. and Lee, J.T. (2001) Forty years of decoding the silence in assays show Xist RNA levels are low in mouse female adult tissue, embryos X-chromosome inactivation. Hum. Mol. Genet., 10, 2225–2232. and embryoid bodies. Development, 120, 3529–3536. 9. Avner, P. and Heard, E. (2001) X-chromosome inactivation: counting, 30. Panning, B., Dausman, J. and Jaenisch, R. (1997) X chromosome choice and initiation. Nat. Rev. Genet., 2,59–67. inactivation is mediated by Xist RNA stabilization. Cell, 90, 907–916. 10. Penny, G.D., Kay, G.F., Sheardown, S.A., Rastan, S. and Brockdorff, N. 31. Sheardown, S.A., Duthie, S.M., Johnston, C.M., Newall, A.E., (1996) Requirement for Xist in X chromosome inactivation. Nature, 379, Formstone, E.J., Arkell, R.M., Nesterova, T.B., Alghisi, G.C., Rastan, S.

131–137. and Brockdorff, N. (1997) Stabilization of Xist RNA mediates initiation of Downloaded from https://academic.oup.com/hmg/article/12/2/125/608526 by guest on 27 September 2021 11. Marahrens, Y., Panning, B., Dausman, J., Strauss, W. and Jaenisch, R. X chromosome inactivation. Cell, 91,99–107. (1997) Xist-deficient mice are defective in dosage compensation but not 32. Fire, A., Xu, S., Montgomery, M.K., Kostas, S.A., Driver, S.E. and spermatogenesis. Genes Devl., 11, 156–166. Mello, C.C. (1998) Potent and specific genetic interference by double- 12. Lee, J.T., Davidow, L.S. and Warshawsky, D. (1999) Tsix, a gene antisense stranded RNA in Caenorhabditis elegans. Nature, 391, 806–811. to Xist at the X-inactivation centre. Nat. Genet., 21, 400–404. 33. Zamore, P.D., Tuschl, T., Sharp, P.A. and Bartel, D.P. (2000) RNAi: double- 13. Debrand, E., Chureau, C., Arnaud, D., Avner, P. and Heard, E. (1999) stranded RNA directs the ATP-dependent cleavage of mRNA at 21 to 23 Functional analysis of the DXPas34 locus, a 30 regulator of Xist expression. nucleotide intervals. Cell, 101,25–33. Mol. Cell. Biol., 19, 8513–8525. 34. Volpe, T.A., Kidner, C., Hall, I.M., Teng, G., Grewal, S.I. and 14. Lee, J.T. and Lu, N. (1999) Targeted mutagenesis of Tsix leads to Martienssen, R.A. (2002) Regulation of heterochromatic silencing and nonrandom X inactivation. Cell, 99,47–57. histone H3 lysine-9 methylation by RNAi. Science, 297, 1833–1837. 15. Sado, T., Wang, Z., Sasaki, H. and Li, E. (2001) Regulation of imprinted 35. Reinhart, B.J. and Bartel, D.P. (2002) Small RNAs correspond to X-chromosome inactivation in mice by Tsix. Development, 128, centromere heterochromatic repeats. Science, 297, 1831. 1275–1286. 36. Beletskii, A., Hong, Y.K., Pehrson, J., Egholm, M. and Strauss, W.M. 16. Luikenhuis, S., Wutz, A. and Jaenisch, R. (2001) Antisense transcription (2001) PNA interference mapping demonstrates functional domains in the through the Xist locus mediates Tsix function in embryonic stem cells. noncoding RNA Xist. Proc. Natl Acad. Sci. USA, 98, 9215–9220. Mol. Cell. Biol., 21, 8512–8520. 37. Csankovszki, G., Panning, B., Bates, B., Pehrson, J.R. and Jaenisch, R. 17. Stavropoulos, N., Lu, N. and Lee, J.T. (2001) A functional role for Tsix (1999) Conditional deletion of Xist disrupts histone macroH2A localization transcription in blocking Xist RNA accumulation but not in X-chromosome but not maintenance of X inactivation. Nat. Genet., 22, 323–324. choice. Proc. Natl Acad. Sci. USA, 98, 10232–10237. 38. Wutz, A., Smrzka, O.W., Schweifer, N., Schellander, K., Wagner, E.F. and 18. Bartolomei, M.S., Webber, A.L., Brunkow, M.E. and Tilghman, S.M. Barlow, D.P. (1997) Imprinted expression of the Igf2r gene depends on an (1993) Epigenetic mechanisms underlying the imprinting of the mouse H19 intronic CpG island. Nature, 389, 745–749. gene. Genes Devl., 7, 1663–1673. 39. Rougeulle, C., Cardoso, C., Fontes, M., Colleaux, L. and Lalande, M. 19. Chao, W., Huynh, K.D., Spencer, R.J., Davidow, L.S. and Lee, J.T. (2002) (1998) An imprinted antisense RNA overlaps UBE3A and a second CTCF, a candidate trans-acting factor for X-inactivation choice. Science, maternally expressed transcript. Nat. Genet., 19,15–16. 295, 345–347. 40. Wroe, S.F., Kelsey, G., Skinner, J.A., Bodle, D., Ball, S.T., Beechey, C.V., 20. Sheardown, S.A., Newall, A.E., Norris, D.P., Rastan, S. and Brockdorff, N. Peters, J. and Williamson, C.M. (2000) An imprinted transcript, antisense to (1997) Regulatory elements in the minimal promoter region of the mouse Nesp, adds complexity to the cluster of imprinted genes at the mouse Gnas Xist gene. Gene, 203, 159–168. locus. Proc. Natl Acad. Sci. USA, 97, 3342–3346. 21. Migeon, B.R., Chowdhury, A.K., Dunston, J.A. and McIntosh, I. (2001) 41. Lee, Y.J., Park, C.W., Hahn, Y., Park, J., Lee, J., Yun, J.H., Hyun, B. and Identification of TSIX, encoding an RNA antisense to human Xist, reveals Chung, J.H. (2000) Mit1/Lb9 and Copg2, new members of mouse differences from its murine counterpart: implications for X inactivation. imprinted genes closely linked to Peg1/Mest(1). FEBS Lett., 472, 230–234. Am. J. Hum. Genet., 69, 951–960. 42. Smilinich, N.J., Day, C.D., Fitzpatrick, G.V., Caldwell, G.M., Lossie, A.C., 22. Migeon, B.R., Lee, C.H., Chowdhury, A.K. and Carpenter, H. (2002) Cooper, P.R., Smallwood, A.C., Joyce, J.A., Schofield, P.N., Reik, W. et al. Species differences in TSIX/Tsix genes reveal the roles of these genes in (1999) A maternally methylated CpG island in KvLQT1 is associated with X-chromosome inactivation. Am. J. Hum. Genet., 71, 286–293. an antisense paternal transcript and loss of imprinting in Beckwith– 23. Allaman-Pillet, N., Djemai, A., Bonny, C. and Schorderet, D.F. (2000) The Wiedemann syndrome. Proc. Natl Acad. Sci. USA, 96, 8064–8069. 50 repeat elements of the mouse Xist gene inhibit the transcription of 43. Lee, M.P., DeBaun, M.R., Mitsuya, K., Galonek, H.L., Brandenburg, S., X-linked genes. Gene Expr., 9,93–101. Oshimura, M. and Feinberg, A.P. (1999) Loss of imprinting of a paternally 24. Wutz, A., Rasmussen, T.P. and Jaenisch, R. (2002) Chromosomal silencing expressed transcript, with antisense orientation to KVLQT1, occurs and localization are mediated by different domains of Xist RNA. Nat. frequently in Beckwith–Wiedemann syndrome and is independent of Genet., 30, 167–174. insulin-like growth factor II imprinting. Proc. Natl Acad. Sci. USA, 96, 25. Nam, D.K., Lee, S., Zhou, G., Cao, X., Wang, C., Clark, T., Chen, J., 5203–5208. Rowley, J.D. and Wang, S.M. (2002) Oligo(dT) primer generates a high 44. Li, E., Bestor, T.H. and Jaenisch, R. (1992) Targeted mutation of the DNA frequency of truncated cDNAs through internal poly(A) priming during methyltransferase gene results in embryonic lethality. Cell, 69, 915–926. reverse transcription. Proc. Natl Acad. Sci. USA, 99, 6152–6156. 45. Simmler, M.C., Cunningham, D.B., Clerc, P., Vermat, T., Caudron, B., 26. Sheets, M.D., Ogg, S.C. and Wickens, M.P. (1990) Point mutations in Cruaud, C., Pawlak, A., Szpirer, C., Weissenbach, J., Claverie, J.M. et al. AAUAAA and the poly (A) addition site: effects on the accuracy and (1996) A 94 kb genomic sequence 30 to the murine Xist gene reveals an AT efficiency of cleavage and polyadenylation in vitro. Nucl. Acid Res., 18, rich region containing a new testis specific gene Tsx. Hum. Mol. Genet., 5, 5799–5805. 1713–1726.