INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Mammalian X-chromosome inactivation is the process by which one of the two X chromosomes in female cells is transcriptionally silenced, resulting in dosage compensation for X-linked gene products between the sexes (31). This process takes place during early female development and is closely associated with cellular differentiation during early embryogenesis. Similarly, in vitro differentiation of female embryonic stem (ES) cells results in inactivation of one of the two X chromosomes (43).

Classical genetic studies have defined a critical region of the X chromosome, the X-inactivation center (Xic), which controls the initiation of X inactivation (see reference 21 for a review). The Xic not only is required for the spread in cis of inactivation but also appears to be involved in a counting process that senses the number of X chromosomes in a cell and ensures that only a single X chromosome remains active per diploid cell, all supplementary X chromosomes being inactivated (see reference 32 for a review).

The Xist gene, which lies within the critical Xic region, is expressed exclusively from the inactive X chromosome (4, 6, 9), producing a long, processed transcript which does not appear to be protein coding (7, 10). The Xist RNA appears to associate with or "coat" the inactive X chromosome (10, 14), leading to suggestions that it may play a chromatin remodeling role in the initiation and propagation of X inactivation (53). A role for Xist in the initiation of X inactivation is supported by its developmental expression pattern (26). The onset of X inactivation is preceded by a marked increase in Xist RNA levels in early embryos and differentiating ES cells, and recent studies have shown that this up-regulation is achieved via stabilization of Xist transcripts, rather than higher transcription rates (39, 48). Formal proof of Xist's involvement in X inactivation has been provided by two independent targeted deletions of Xist sequences (35, 40). Both studies demonstrated that Xist expression is essential for inactivation in cis, thus supporting the idea that the Xist RNA is the propagator of X inactivation. In both cases, however, an undeleted X chromosome could still be inactivated in Xist mutant cells, implying that counting had not been disrupted. Thus, counting and cis inactivation appear to be controlled by genetically separable elements.

The presence of sequences involved in counting outside the Xist gene has recently been suggested by the deletion of a 65-kb region lying 3' to Xist exon 6, where the X chromosome bearing the deletion undergoes inactivation even in the absence of a second X chromosome (15). Furthermore, the genetically defined X-chromosome controlling element (Xce), which affects the random nature of X inactivation in the mouse and thus appears to be involved in the choice of X chromosome that is inactivated (13), has been found to be closely linked to but genetically distinct from Xist, lying 3' to it (49). Elements potentially affecting choice have also been reported to lie within the Xist gene (34) and 5' to it (41).

Taken together, the above genetic and knockout data support the idea that several distinct elements, distributed across the region containing Xist, may be required to regulate the initiation of X inactivation. Although targeted deletions can establish whether a particular sequence is necessary for X inactivation, such an approach cannot easily address the question of what the extent of the sequences necessary and sufficient to carry out X inactivation might be. A stringent test for determining whether a given region is sufficient to bring about correctly regulated X inactivation is to examine its behavior at ectopic loci in transgenic experiments. A number of such experiments have now been reported (see reference 19 for a review). In one case, ES cell lines carrying multiple copies of a 450-kb yeast artificial chromosome (YAC) including the Xist gene were found to show a number of functions attributed to the Xic upon differentiation: the transgenic Xist RNA could coat the autosome carrying the transgene; and inactivation in cis was observed, as was inactivation of the single X chromosome (counting) (29, 30). In another study, Matsuura et al. (37) found ectopic Xist expression in transgenic mice containing four copies of a 240-kb Xist YAC integrated onto the Y chromosome, although inactivation was not demonstrated. More recently, multiple copies of a 35-kb cosmid transgene encompassing Xist were also found to be capable of some ectopic Xic function, although the variability in Xist expression levels observed between cell lines suggested that these short transgenes were subject to position effects and may therefore be missing elements for autonomous function (24).

In all of the above studies, however, the transgenic lines contained Xist transgenes present in multiple linked copies (3 to 24 copies [30]; 4 copies [37]; 2 to 8 copies [24]). The ability of a single-copy transgene encompassing Xist to show X-inactivation function has thus not been demonstrated so far, and it could not be ruled out in the above studies that complex interactions among linked YAC copies may have somehow enabled ectopic Xic function to be observed (52).

In a previous report we described several mouse lines carrying single-copy Xist YAC transgenes in which Xist was not expressed and there was no evidence of inactivation (22). In the study presented here, we wished to establish definitively whether single copies of Xist-containing YAC transgenes are capable of inducing inactivation in cis and/or counting in ES cells. ES cells and their in vitro-differentiated derivatives were used as the selective pressure against transgene function is likely to be less stringent in vitro than in vivo, where cell selection might occur as a result of autosomal cis inactivation (leading to monosomy) or counting (leading to X-chromosome nullisomy). We have generated an extensive series of transgenic cell lines by transferring two different YACs into male and female ES cells. We also derived ES cells from one of our previously described transgenic mouse lines (22), which carries a single intact copy of a 460-kb YAC, to see whether the absence of Xic function that we had previously observed was due to the single-copy nature of the transgene or to in vivo counterselection. Transgene activity was followed at the single-cell level using RNA fluorescent in situ hybridization (FISH) to detect expression both of Xist and of Brx, a gene susceptible to inactivation which is located 75 kb 3' to Xist (51) and which is expressed from the active X chromosome in both differentiated and undifferentiated ES cells. The degree of mosaicism associated with transgene function in differentiating ES cell populations could thus be assessed.

We report that, unlike multicopy arrays, large, single-copy Xist YAC transgenes lead to neither inactivation in cis nor counting, despite the presence of all of the sequences defined to date by deletion as being essential for the correct regulation of X inactivation. Our results indicate that certain elements necessary for the initiation of X inactivation at ectopic sites must be lacking from the Xist YAC transgenes tested, and that their lack is compensated for by the presence of multiple copies. A number of hypotheses to explain the differences in behavior between single and multicopy Xist transgenes are presented, and the implications of our findings for autonomous Xic function are discussed.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

YAC manipulation. The two vector arms of YAC PA-2 were retrofitted with I-PpoI sites, and the right YAC arm was retrofitted with the pgk1 promoter-neomycin resistance gene (pgk-neo) cassette as described elsewhere (18). YAC PA-3 F1n (20) was retrofitted by using the pLNA-1 vector, which carries the pgk1 promoter-neo cassette (45). This vector could integrate into the ampicillin resistance gene present on either of the two YAC vector arms.

Generation of transgenic ES cell lines. High-molecular-weight yeast DNA was prepared in agarose plugs and separated by pulsed-field gel electrophoresis, and the YAC DNA was excised by previously described procedures (22). YAC DNA was lipofected into CK35 ES cells (11) by using the protocol described in reference 28. Transfer of YAC PA-2 by spheroplast fusion into the XX HP310 ES cell line (2a) was performed as previously described (18). In both cases, Neo^r clones were selected 24 h after YAC transfer by treatment with G418 (0.425 mg/ml for HP310-derived F1 clones and 0.25 mg/ml for CK35-derived L1 and L4 clones). ES cell lines were derived from transgenic mouse line 53 (22) by using standard procedures (46) that involved crosses between transgenic females and transgenic males, both of which were on an essentially (>98%) 129 genetic background.

Characterization of transgenic ES cell lines. Southern analysis of EcoRI- or HindIII-digested DNA was carried out by using standard procedures. Transgene copy number was quantitated by PhosphorImager (Molecular Dynamics) analysis of blots hybridized with Xist probes generated by PCR (e.g., nucleotides 5379 to 5547) and a 1.1-kb fragment in exon 1 (22), as well as the X-chromosome probe Xpct (128E2 probe [16]), which does not lie within the YACs. Other probes used for mapping (e.g., IVA5, DXPas34, DXPas19, and DXPas27) have been described previously (20, 23). The Tsx probe (50) was generated by PCR using primers GAG ATG TTC ACT GAT CAG AA and CTC AGT GGC TCA TCT GC. Pulsed-field gel analysis of ES cell DNA in agarose blocks digested with rare-cutter enzymes such as SalI was performed by using standard procedures (23). I-PpoI digestion was performed as specified by the manufacturer (Promega).

Culture and differentiation of ES cells. CK35 and HP310 ES cells and transgenic derivatives were maintained in the undifferentiated state by culturing in ES medium (high-glucose Dulbecco modified Eagle medium [DMEM], 2 mM glutaMAX I [GIBCO] 0.1 mM 2-mercaptoethanol, 15% fetal calf serum and 10³ U of recombinant leukemia inhibitory factor [GIBCO] per ml, 0.05 mg of streptomycin per ml, 50 U of penicillin per ml) on mitomycin C-treated mouse fibroblast feeder cells and appropriate G418 selection for transgenic cells. For differentiation into embryoid bodies (EBs), feeders were first removed by successive adsorptions on gelatinized dishes and then cultivated for 3 days under adherent conditions in ES cell medium (46). Aggregates were formed following mild trypsinization and transferred to suspension culture (day 0) in EB medium (DMEM, 10% newborn calf serum, 10⁴ M 2-mercaptoethanol), without G418 selection. Half of the EB medium was changed every day. Four day EBs were attached onto LabTek chamber slides (Nunc) for monolayer outgrowth for 3 to 10 days in DMEM-10% fetal calf serum.

DNA and RNA FISH analysis. Metaphase spreads were obtained from ES cells, and FISH was performed as described elsewhere (14). Interphase nuclei were prepared as described previously (30). Briefly, cells grown on chamber slides or cytospun (4 min, 400 rpm) onto baked glass slides were permeabilized with Triton X-100 in ice-cold cytoskeletal buffer for 7 min, fixed with 4% paraformaldehyde for 10 min on ice, and then stored in 70% ethanol. EBs were taken from culture at days 1, 2, 3, 4, and 6 of differentiation and disaggregated by pipetting in phosphate-buffered saline prior to cytospinning. For RNA FISH, slides were dehydrated and used directly for hybridization. For DNA FISH, nuclei were denatured for 2 min in 70% formamide, 2× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate) at 75°C; for simultaneous RNA-DNA FISH, nuclei were denatured for just 1.5 min. Hybridization and washing conditions were as described previously (14). Nuclei were mounted in Vectashield (Vector) and counterstained with 4',6-diamidino-2-phenylindole (DAPI). A Quantix change-coupled device camera and IPLab and Photoshop software were used for image acquisition and treatment. All probes were labeled by nick translation (Vysis) with spectrum green-dUTP or spectrum red-dUTP (Vysis). The Xist probe was either plasmid pXist3K (30) or lambda 510 (47), which covers most of the Xist gene. The Brx probe consisted of the partially overlapping lambda clones IIIC2 and IG10 (47). The YAC vector probe was pYAC4 vector DNA, and the bacterial artificial chromosome (BAC) X probe was a BAC isolated by using the microsatellite marker DXMit158, which lies outside the X-chromosome region covered by YACs PA-2 and PA-3 F1n.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Generation of transgenic ES cell lines with YACs PA-2 (460 kb) and PA-3 F1n (320 kb). To study the capacity of different regions containing Xist to induce the X-inactivation process, we used two YACs, PA-2 (460 kb) and PA-3 F1n (320 kb) (Fig. 1), both of which have previously been mapped in detail (20, 23). Both YACs carry inserts derived from the C3H/He mouse strain. YAC PA-2 contains sequences 130 kb 5' and 310 kb 3' to Xist, while YAC PA-3 F1n contains sequences 170 kb 5' and 100 kb 3' to Xist. Using homologous recombination in yeast, both YACs were retrofitted with the pgk1 promoter-neo selectable marker in one vector arm (see Materials and Methods).

View larger version (40K): [in this window] [in a new window]   FIG. 1.   (A and B) Structures of the two YACs used in this study, PA-2 (460 kb) and PA-3 F1n (320 kb). More detailed physical maps of these YACs have been published previously (20, 23). The DNA probes used in characterization of the YACs and transgenic clones are shown as black boxes below each YAC map along with the I-PpoI, SalI (Sa), and SfiI (Sf) sites that were informative in the analysis of the transgenes. (A) Transgene structures of the L4 (lipofection-derived) and F1 (spheroplast fusion-derived) lines created with the 460-kb YAC PA-2. I-PpoI meganuclease sites present at each extremity of this YAC enabled YAC integrity of the transgenic clones to be assessed (see panel C). SfiI mapping (data not shown) provided further resolution concerning the extent of the region covered by the transgenes (particularly in line F1 A). Multicopy transgenic line F1 C showed, in addition to the major YAC fragment illustrated, a minority of smaller transgenic fragments. (B) The transgene structures of L1 (lipofection-derived) lines created with the 320-kb YAC PA-3 F1n. Both SalI and SfiI restriction sites were used to analyze the structures of these transgenes (see panel D for an example of the SfiI analysis). A SalI site within the YAC arm of PA-3 F1n which leads to the generation of a SalI band detected by the Tsx and DXPas34 probes of approximately the same size as that at the endogenous locus is indicated as Sa*. It should be noted that there are other SalI sites present in this region (23), but these are methylated and thus uncut by SalI in mouse genomic DNA. Multicopy transgenic line L1 17 showed, in addition to the major YAC fragment illustrated, a minority of smaller transgenic fragments. (C) Example of pulsed-field gel analysis of I-PpoI-digested DNA of cell lines carrying YAC PA-2 transgenes and the nontransgenic control lines CK35 and HP310. In lines L4 8, L4 13, and F1 E (single-copy transgenes), a 460-kb fragment detected by all probes tested lying within the YAC (Xist probe shown) following I-PpoI digestion suggested that the YAC was intact. In lines F1 A (single copy) and L4 12 (two copy), larger (>750-kb) transgene fragments were detected, suggesting that at least one I-PpoI site had been lost. In line F1 C (three to four copies), detection of a 360-kb band as well as a >750-kb band suggested that at least two copies of the truncated YAC, as shown in panel A, were present, with additional end fragment(s) (i.e., minus a second I-PpoI site) being responsible for the higher-molecular-weight band. (D) Example of pulsed-field gel analysis of SfiI-digested DNA of lines carrying YAC PA-3 F1n transgenes. The internal SfiI bands detected by various probes (B) appear to be intact in lines L1 25 (single copy), L1 6 (two copies) and L1 12 (six to seven copies) and are detected as an increase in the intensity of the endogenous bands compared with the control cell line CK35 (170 and 280 kb with the Xist probe shown here). In L1 17 (five to eight copies), the majority of the SfiI fragments appear to be intact, although a minority of smaller and larger fragments, corresponding to truncated fragments and partial digests of neighboring transgene copies, respectively, can also be seen.

View larger version (112K): [in this window] [in a new window]   FIG. 2.   Xist RNA expression in single-copy and multicopy transgenic lines in undifferentiated ES cells. (A) DNA FISH on metaphase chromosomes from line L1 17 showing a single integration site for the multicopy transgene in a mouse autosome. A spectrum red-labeled YAC PA-3 F1n probe hybridized with both the transgenic locus (T) and its endogenous site on the X chromosome (X). The larger signal at the transgene locus reflected its higher copy number (seven in this case). This was confirmed in other hybridizations involving a YAC-specific probe. (B to F) Under conditions that do not denature chromosomal DNA, undifferentiated ES cells were hybridized with a spectrum green labeled Xist probe to detect Xist RNA. In lines L4 8 (single copy, XY) (B), F1 A (single copy, XX) (C), and 53.2 (two unlinked copies, XO) (D), Xist RNA pinpoint signals over the X chromosome and the single-copy transgene loci were indistinguishable. In lines L4 12 (multicopy, XY) (E), and L1 17 (multicopy, XY) (F), the transgenic Xist RNA pinpoint tended to be slightly larger than the endogenous signal. (D and F) The Xist RNA FISH signals in positioned nuclei were photographed, and the cells were then denatured and DNA FISH was performed. (D) A probe specific for the YAC vector (pYAC4, spectrum red labeled) detected the two single-copy transgenes (T) in line 53.2; (F) a probe specific for the X chromosome (BAC X, spectrum red labeled) demonstrated that the larger Xist pinpoint signal corresponded to the seven-copy transgene (T) present in line L1 17. (G to I) Two-color RNA FISH for simultaneous detection of Xist (spectrum green) and Brx (spectrum red) transcripts under nondenaturing conditions (see above). (G) L4 8 (single copy, XY); (H) F1 A (single copy, XX); (I) L4 12 (two copy, XY).

YAC transgene function in undifferentiated ES cells. In undifferentiated ES cells, Xist is expressed as an unstable transcript from every X chromosome present (whether XX or XY) (39, 48). This unstable transcript can be detected by RNA FISH as a punctate signal (pinpoint) at the site of the Xist gene. We wished to determine whether single and multicopy transgenes were capable of expressing Xist in its unstable form in undifferentiated ES cells. In single-copy transgenic lines (referred to as single-copy lines hereafter), a Xist RNA pinpoint was observed from the transgene locus as well as from the X chromosome. In lines L4 8 and 53.1 (single copy, XY) two pinpoints were observed; in 53.2 (two single copies, XO) and F1 A (single copy, XX), three pinpoints were observed (Fig. 2B to D). Confirmation that the transgene was indeed able to express Xist was provided by performing YAC-specific DNA FISH on nuclei following Xist RNA FISH detection (as shown, for example, in line 53.2 [Fig. 2D]). No mosaicism in transgenic Xist expression was observed in these undifferentiated cells, with the expected number of Xist pinpoints, derived from the transgene(s) and the X chromosome(s), being detected in almost all cells. Similar results were obtained in experiments using a probe to detect Brx transcripts in lines L4 8 (single copy, XY) and F1 A (single copy, XX) (Fig. 2G and H), where in 80% of cells, two and three Brx pinpoints, respectively, were detectable. Furthermore, Xist (and Brx) pinpoints from the X chromosome or the transgene could not be distinguished, suggesting that the genes were being expressed at similar levels at both loci.

Single-copy transgene function in differentiated cells. When induced to differentiate in vitro, female ES cell lines undergo X inactivation, the first sign of this being the coating of the prospective inactive X chromosome with stabilized Xist RNA (detected by RNA FISH as a Xist RNA domain) (39, 48). The unstable Xist transcript (Xist pinpoint) on the active X chromosome is still expressed when the Xist RNA domain appears but gradually disappears as differentiation progresses. In male or XO ES cells where X inactivation does not normally occur, no Xist RNA domain is ever observed upon differentiation and Xist expression on the single active X chromosome eventually disappears, as it does on the active X in female cells.

View larger version (126K): [in this window] [in a new window]   FIG. 3.   Xist and Brx expression in single-copy and multicopy transgenic lines in fully differentiated ES cells. (A to D) Under conditions that do not denature chromosomal DNA, ES cells differentiated by EB attachment and outgrowth on chamber slides for several days, were hybridized with a spectrum green-labeled Xist probe to detect Xist RNA as well as a spectrum red-labeled Brx probe to detect Brx RNA in panel D. (A) Several nuclei in the control female cell line (HP310, XX) showing Xist RNA domains (inactive X chromosomes) in the majority of cells; in some cells, a single Xist RNA pinpoint (on the active X chromosome) is also detected. (B) In differentiated cells of the control male cell line (CK35, XY), either a single Xist RNA pinpoint or no Xist signal at all is detected. (C) In the single-copy transgenic line L4 8, either one, two, or no Xist RNA pinpoints are detected. (D) Two representative differentiated nuclei of line 53.2 (two unlinked single-copy transgenes, X0) are shown, one with no Xist RNA signals (green) but three Brx signals (red) and the other with three Brx signals and a single Xist RNA signal (yellow as it overlaps with Brx). (E and F) Line F1 A (single-copy transgene, XX) simultaneous RNA-DNA two-color FISH on mildly denatured nuclei (see Materials and Methods), using a spectrum green-labeled Xist probe and spectrum red-labeled YAC vector (E) or BAC X (F) probe. The Xist RNA domain is associated with the X chromosome (X) and not the single-copy transgene (T). (G to I) RNA FISH on undenatured nuclei of differentiated L4 12 ES cells (multicopy, XY) using a Xist probe (spectrum green) was followed by denaturation and DNA FISH using a spectrum red-labeled YAC vector probe. In the majority of nuclei (G), a single Xist RNA domain associated with the transgene (T) was observed. Occasionally the Xist RNA domain was associated with the X chromosome (H) or with both the X chromosome and the transgene (I).

View larger version (46K): [in this window] [in a new window]   FIG. 4.   Evolution of Xist RNA signals in differentiating ES cells. (A) One-day EB of control female ES line HP310; two-color RNA FISH detecting Xist (spectrum green) and Brx (spectrum red) transcripts. Xist pinpoint clusters can sometimes be seen at this stage. We considered "clusters" to consist of 2 to 10 distinct pinpoints randomly scattered around the site of transcription, while "domains" were dense, more extensive signals that covered an interphase chromosome-sized area. (B) Two-day EB, control female ES line HP310; Xist (spectrum green) RNA FISH showing a Xist RNA domain. (C) One day EB of line L4 8 (single-copy transgene, XY); two-color RNA FISH detecting Xist (spectrum green) and Brx (spectrum red) transcripts. Xist pinpoint clusters can sometimes be seen at this stage. (D) L4 8 1-day EB. Performance of Xist RNA FISH (spectrum green) was followed by denaturation and DNA FISH using a YAC-specific probe (spectrum red), which shows that Xist RNA pinpoint clusters tend to be associated with the transgene loci. (E and F) L4 12 (two-copy transgene, XY), 2-day EB. Performance of Xist RNA FISH (spectrum green) was followed by denaturation and DNA FISH using a YAC-specific probe (spectrum red). (E) Xist RNA pinpoint clusters are sometimes seen either at the transgenic locus alone or (as shown here) at both the X chromosome and the transgene (T). (F) Xist RNA domains are mostly found to be associated with the transgene. (G) Representative view of L4 12 cells (two-copy transgene, XY) from 1-day EBs. Two-color RNA FISH detecting Xist (spectrum green) and Brx (spectrum red) transcripts shows that the majority of cells contain one or two Xist pinpoint clusters.

Multicopy transgene function in differentiated cells. Five multicopy transgenic ES cell lines were examined for the ability to induce Xist RNA domain formation and inactivation following differentiation via EB attachment and outgrowth (Table 2). In these lines, areas of cells with Xist RNA domains could easily be detected. The majority of transgenic cells contained only a single Xist RNA domain, although two domains were sometimes observed. The results concerning the origins of the Xist RNA domains for lines L4 12 (two copies of YAC PA-2, XY) and L1 17 (five to eight copies of YAC PA-3F1n, XY) are summarized in Table 3. The majority of Xist domains were associated with the transgene locus alone (Fig. 3G), although in a small percentage of cells only the X chromosome or both the transgene and the X chromosome were affected (Fig. 3H and I). The Xist RNA domains associated with the X chromosome were often smaller than those at the transgene. This may be indicative of rapid counterselection against those cells in which the single X chromosome becomes fully inactivated, leading to nullisomy.

View larger version (35K): [in this window] [in a new window]   FIG. 5.   Mosaicism in the presence of Xist RNA domains in fully differentiated ES cells. (A) Xist RNA FISH (spectrum green) on differentiated L1 17 cells (five- to eight-copy transgene, XY). Shown is a representative view with Xist RNA domains present in the majority of cells. (B) Xist RNA FISH (spectrum green) on differentiated L4 12 cells (two-copy transgene, XY). Shown is a representative view with Xist RNA domains present in a minority of cells. (C) Two-color RNA FISH detecting Xist (spectrum green) and Brx (spectrum red) on differentiated L412 cells (two-copy transgene, XY) as in panel B. In cells containing a Xist RNA domain, no Brx signal is detected within the domain (indicating Brx inactivation), while a Brx signal is detected on the other (active) Xic in these cells. In cells with no Xist domain, two Brx signals are detected, indicating that despite the lack of Xist expression and domains in these cells, the transgene is still functional (i.e., Brx is expressed). In all cases, ES cells were differentiated for a total of 9 days (4 days of EB growth in suspension followed by 5 days of outgrowth after EB attachment onto chamber slides) and fixed in situ on the slides.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Based on accumulating genetic and knockout data, it is becoming apparent that the Xic may be a complex locus and that elements situated not only within, but also some distance from the Xist gene may be required for the regulation of X-chromosome counting, choice, and cis inactivation (see reference 12 for a review). Given the potential complexity of the Xic region, we have addressed the question of whether large, Xist-containing transgenes can trigger the X-inactivation process in an autonomous fashion when present as single copies as opposed to multicopy arrays.

ES cell lines carrying single or multiple copies of Xist YAC transgenes have been assessed for the capacity to induce inactivation in cis and counting. Both single-copy and multicopy transgenes express the Xist transcript in its unstable form in undifferentiated ES cells. Upon differentiation, however, none of the six single-copy lines investigated showed any ability to induce the X-inactivation process. Only lines with multicopy transgenes displayed mature Xist RNA domain formation and subsequent inactivation at the transgene-carrying autosome and, to a lesser extent, of the X chromosome. Single-copy transgenes did not induce Xist RNA domain formation either at the autosome on which they were integrated or at the endogenous X chromosome. Thus, although the sequences enabling Xist transcription in undifferentiated cells are clearly present in these single-copy transgenes, other elements necessary for the initiation of X inactivation appear to be lacking. Importantly, this incapacity of single-copy transgenes for Xic function was unchanged by the presence of an X chromosome undergoing inactivation in differentiating XX ES cell lines (F1 A and F1 E), indicating that any trans-acting factors that might be required for the initiation of X inactivation were not capable of compensating for this lack of function.

We were also able to address the question of whether any number of Xist YAC transgenes greater than one would enable ectopic Xic function. We found that lines containing just two copies of either YAC PA-2 (line L4 12) or PA-3 F1n (line L1 6) were capable of both cis inactivation and counting. Furthermore, the efficiency of such function appeared to increase with increasing transgene copy number: compared to lines with seven transgene copies (L1 12 and L1 17), the proportion of differentiated cells with Xist RNA domains in the two-copy line, L4 12, was low. We believe that the inefficiency of low-copy-number transgenes is due to an incapacity to establish Xist RNA domains, rather than to selective loss of cells in which the transgene was functional. This conclusion is based on the observation that the proportion of cells in which mature Xist RNA domains were formed was low and did not decrease over time and that a transiently high proportion of Xist pinpoint clusters rather than domains was found at the onset of differentiation.

We have found that multiple transgene copies have to be linked in cis in order to induce cis inactivation and counting. The presence of two single-copy transgenes on homologous chromosomes in the same cell (line 53.2, YAC PA-2) did not lead to Xist RNA domain formation and associated inactivation, unlike the situation in lines L4 12 (YAC PA-2) and L1 6 (YAC PA-3 F1n), where the two transgene copies are cis linked. This cis requirement suggests that critical elements that may be missing in the YACs can be compensated for in some way by the juxtaposition of two or more YAC copies. Similar requirements for multiple transgene copies in order to establish appropriate genetic control have been described for other systems. For example, -globin locus control region 5'HS2 transgenes appear to be functional when at least two copies are present but are unable to direct position-independent expression when present as single copies (17). Similarly, in the case of H19 minitransgenes, correct expression and imprinting was observed only when two or more linked copies were present (3). The use of larger transgenes (130 kb) overcomes the incapacity of single-copy H19 transgenes to function autonomously (1) either because sequences essential for correct regulation of the H19 gene are present in this larger region or because H19 is buffered by the extra sequences from position effects at random integration sites. As discussed below, Xist transgenes larger than those tested here may similarly enable autonomous Xic function.

Unlike single-copy transgenes, multicopy transgenes were also found to induce inactivation of the X chromosome (evidence for counting), although the frequency of differentiated cells in which we observed this was lower than that reported by Lee et al. (30) and Herzing et al. (24). This difference in frequency may be due to sequence differences between the transgenes. An Xce effect biasing the choice of allele (transgene or X chromosome) that is inactivated may be involved. Genetic mapping suggests that Xce is closely linked to, but distinct from, Xist (49). Both of the YACs that we used are derived from the C3H/He strain mice, which carry an Xce^a allele, while the YAC and cosmid used by Lee et al. (30) and Herzing et al. (24), respectively, are both derived from strains carrying an Xce^b allele. As the ES cell X chromosome is 129 derived (Xce^a) in all these studies, it would be more likely to be chosen for inactivation in the presence of an Xce^b allele (24, 30) than in the presence of another Xce^a allele (this study) (13), which is consistent with the differences observed. Another possibility that we cannot formally rule out is that the host ES cell lines we used, or our culture conditions, have led to a more severe selection against cells with X-chromosome nullisomy. This seems unlikely, however, since no evidence was found for a higher proportion of X chromosome Xist RNA domains at the earliest stages of differentiation.

Hypotheses of X-inactivation function based on the differences in behavior of single-copy and multicopy Xist transgenes. (i) Capacity of transgenes to inactivate in cis. One hypothesis which could explain the inability of single copy transgenes to lead to cis inactivation is that the YACs we have used lack elements that are necessary for Xist RNA association in cis and/or the spread of inactivation when X inactivation initiates. Thus, although our YACs may actually include all of the sequences necessary to trigger X inactivation, they may lack elements that are, for example, involved in the spread of the inactive state. The existence of X chromosome-specific relay elements that allow the efficient spread of the inactive state from the Xic, via for example the binding of heterochromatin-inducing protein factors, has previously been proposed (38, 44). The potential role of the Xist RNA as an effector of propagation may be mediated by its association with chromatin in cis and might occur via proteins bound to such relay elements along the X chromosome (8). A high density of such relay elements in the region around Xist might be essential to nucleate formation of Xist RNA-chromatin protein complexes and to establish a Xist RNA domain. The inability of single-copy transgenes to function could be due to the absence of a sufficient number of such elements in the YACs tested. Multiple linked YAC copies would increase the number of such relay elements surrounding the Xist genes, allowing more efficient establishment of Xist RNA association in cis. The Xist RNA clusters observed in both single- and two-copy transgenic lines during early differentiation stages may be evidence for the inability of the Xist RNA to nucleate the formation of a full domain from a limited set of such elements. The prediction would be that Xist transgenes flanked with more extensive X-linked sequences should be functional as single copies at ectopic sites. A 750-kb YAC centered around Xist will be analyzed for single-copy function.

(ii) Capacity of transgenes to count. The inability of our single-copy transgenes to induce counting was surprising in the light of currently favored models for this process. It has been proposed that as X inactivation initiates, one Xic per diploid cell is blocked to remain active by a factor present in limiting quantity, and as differentiation proceeds, X inactivation occurs only from remaining unblocked Xics (42). Given the large size of our transgenes, it seemed likely that sequences involved in this hypothetical blocking step would be present, for example, within the 65-kb region 3' to Xist which when deleted eliminates counting (15). It has also recently been suggested that transcription from a newly identified Xist upstream promoter, Po, from which only the unstable form of Xist RNA is transcribed, may be required for a Xist allele to be registered by the counting mechanism (25). Preliminary results obtained using probes that specifically detect Po-derived transcripts (19a) indicate that Po is efficiently used by our single-copy transgenes, thus suggesting that transcription from Po is not in itself sufficient to induce counting in this context.