Mice Carrying Autosomal Xist Transgenes Are Viable but Show Specific Phenotypes.

To investigate the requirements for imprinting, we created transgenic mice by microinjecting the 200-kb BAC8 Tg carrying Xist, Tsix, and surrounding sequences (TgBAC8; Fig. 1A) into the pronuclei of fertilized mouse embryos. We obtained six independent Tg lines. In contrast to the lack of phenotype observed previously with YAC Tgs (39⇓–41), intriguing phenotypes were observed in five of the six hemizygous transgenic lines, here named Tg2080, Tg2082, Tg2086, Tg2087, Tg2091, and Tg2092 (Fig. 1B). Two of the lines, Tg2080 and Tg2092, were unstable and were lost after a few generations.

Tg2080/+ had a male founder that carried two to seven Tg copies at a single integration site. The male founder was fertile. Crossing to WT females, however, resulted in F1 progeny that showed a female-specific Tg lethality. Tg2080/+ females died during the neonatal period. On the other hand, Tg2080/+ males were found at reduced frequency. These viability differences were not observed in prenatal embryos, which were recovered at expected Mendelian frequencies regardless of sex (Fig. 1C and Fig. S1). Tg2082/+ also had a male founder that showed a single integration site with four to seven Tg copies. Although both male and female F1 hemizygotes were viable, homozygosity of the Tg resulted in zygotic lethality of both sexes. By contrast, Tg2086 had a female founder with a single integration site and a Tg copy number of 2. Tg2086/+ mice had no gross phenotype. Both female and male offspring produced litters of expected size and at normal Mendelian frequencies (Fig. 1C).

• Download figure
• Open in new tab
• Download powerpoint

Fig. S1.

No embryonic defect is associated with Xist Tgs in Tg2080. (A) Embryos at 12.5 d post conception (dpc). Shown are seven deciduas dissected from a WT female mated with a Tg2080 male. (B) Embryos at 18.5 dpc. Shown are nine pups isolated by Caesarian delivery from a WT female mated with a Tg2080 male.

On the other hand, for several hemizygous Tg lines (Tg2087, Tg2091, and Tg2092), sterility was a major issue. Intriguingly, in each case, the sterility was observed only in the male. Notably, all three lines had female founders. In the case of Tg2087/+, male sterility occurred when the autosomal Tg was homozygosed. In the case of Tg2091/Y and Tg2092/+, male sterility was observed in the hemizygous state. Tg2091 is an X-linked integration with 7–16 copies of the Tg. Tg2092 had two integration sites for a total of 30 Tg copies.

To understand the basis of these phenotypes, we first examined Xist expression in somatic cells of the progeny. We isolated tail-tip fibroblasts from male and female progeny of each transgenic line and performed Xist RNA FISH with accompanying DNA FISH to distinguish the transgenic locus from the endogenous Xic. Robust expression of Xist RNA was observed from the hemizygous transgenic locus in both male and female cells from all transgenic lines (Fig. 1D). The Xist “cloud” produced by the Tg was qualitatively similar to the endogenous Xist cloud in the female cell. In the case of hemizygous Tg2087/+, for example, an Xist cloud was observed over the transgenic locus in 74.6% of male cells (n = 114). Among female cells, 59% showed two Xist clouds, one over the Xi and one over the autosomal insertion site; 41% showed an Xist cloud only over the Xi (n = 61). These experiments showed, somewhat surprisingly, that Xist remained robustly expressed in somatic cells well into adulthood despite potentially detrimental effects. Indeed, transgenic Xist was capable of recruiting PRC2 to trimethylate histone H3 at lysine 27 (H3K27me3) (Fig. 1E), consistent with formation of repressive chromatin.

Male-Specific Sterility and Germ-Cell Defects.

Three of the six independent Tg/+ lines showed male-specific sterility. The Tg2092/+ line carried the highest number of Tg copies (∼30), originally carried on two autosomes. Consistent with germ-cell defects, the testis mass of male mice was <20% that of WT littermates (Fig. 2A). Examination of male germ cells revealed that spermatogenesis was arrested at the pachytene stage of meiosis I (Fig. 2B). There were few, if any, germ cells of a more mature stage, including spermatozoa. Intriguingly, although pachytene spermatocytes of WT mice were characterized by a single large γ-H2AX domain localizing to the partially synapsed “sex body” or “XY body” (42), Tg2092 spermatocytes showed multiple γ-H2AX domains beyond the XY body (Fig. 2C). Such aberrant γ-H2AX distributions were prevalent and were observed in 78% (n = 50) of pachytene meiotic cells from Tg2092/+ males. Because γ-H2AX is induced by meiotic asynapsis and is associated with meiotic silencing by unpaired chromatin (43), the occurrence of multiple diffuse γ-H2AX domains suggested multiple aberrant synapses and chromatin changes in the presence of the Xist Tg.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Male germline defects in Xist transgenic mice. (A) Male hemizygous Tg2092 mice are sterile and have small testes. Shown are representative testes from a transgenic male and a WT littermate control male. (B) Pachytene arrest in the male germ line of hemizygous Tg2092 and Tg2091 mice. PS, pachytene spermatocytes. RS, round spermatids, SC, Sertoli cells; Sp, spermatozoa. (C) γH2AX and SCP1 immunostaining in pachytene spermatocytes reveal the XY body in WT mice and aberrant foci formation in Tg2092 mice. (D) Normal XY-body formation in the pachytene spermatocytes of hemizygous Tg2091 mice. Immunostaining fluorescence for γH2AX and DNA FISH for Tg BAC8 integration are shown. (E) Normal XY-body formation in the pachytene spermatocytes of homozygous (Tg/Tg) Tg2087 males. Immunostaining fluorescence for γH2AX and DNA FISH for Tg BAC8 are integration shown. (F) Defective spermatozoa of hemizygous Tg2091 male: Morphological abnormality in sperm heads of Tg2091 is visualized with DAPI in the spermatocytes spread. (G) Defective sperm cells of homozygous (Tg/Tg) Tg2087 males. Mature sperm isolated from epididymis of the homozygous Tg/Tg male (sterile) and the hemizygous +/Tg male (fertile). Sperm morphology was visualized by differential interference contrast.

Although the multiple γ-H2AX domains were consistent with there being multiple Tg integration sites, the size of the domains indicated that γ-H2AX must have spread along the autosomes beyond the several hundred kilobases or several megabases associated with the tandem Tgs. These autosomal abnormalities likely accounted for the sterility of Tg2092 males. The Tg2092 genotype therefore could be transmitted only maternally. Unfortunately, because the Tgs were inserted into two distinct autosomes, the Tgs were unstable and also segregated after one generation of passage, and the original Tg2092 genotype could not be maintained for further analysis.

The hemizygous Tg2091 and homozygous Tg2087 male mice were sterile also, despite having normal γ-H2AX localization on the XY body (Fig. 2 D and E). On the XY body, a defined γ-H2AX domain was present in 76% (n = 51) of 2091 Tg/Y pachytene spermatocytes and in 69% (n = 96) of 2087 Tg/Tg pachytene spermatocytes. The presence of microγ-H2AX signals on the unpaired Tgs could not be ruled out, because small regions of H2AX phosphorylation could be present at domains of several hundred kilobases without being detectable at the level of immunostaining. This asynapsis and any resulting recruitment of the double-strand break repair mechanism could lead to localized silencing (43, 44). Interestingly, 2091 Tg/Y and 2087 Tg/Tg males showed morphological anomalies in the spermatozoa. In the case of Tg2091, dysmorphic head shapes and failed separation of meiotic tetrads were readily apparent (Fig. 2F). There was an absence of mature spermatozoa in the epididymis. In the case of Tg2087, the homozygous males carried mature sperm in the epididymis but failed to sire offspring. Examination of mature spermatozoa revealed heterogeneous morphological abnormalities ranging from a kinked tail to a missing head (Fig. 2G). These anomalies were present only in the homozygous state and provided logical explanations for the male-specific sterility. The female counterparts were fully fertile.

Xist Imprinting: Epigenetic Effects of Transgene Hemizygosity.

Unlike the high-copy-number Tg2092/+ line, transgenic lines with low-copy Xist integrations generally showed normal XY bodies in pachytene spermatocytes (Figs. 2 D and E and 3A). In the male germ line, large regions of asynapsed chromatin are known to be subject to meiotic silencing, as observed in cases of chromosomal translocations (43, 45), and can interfere with meiotic sex chromosome inactivation (46). Our Xic Tg was 200 kb, and its insertion at low copy numbers (two to seven times) did not obviously interfere with the meiotic synapsis. Intriguingly, however, the ectopic Xic chromatin was changed by the unpaired state of the Tg, most evidently during the pachytene stage of meiosis (Fig. 3A). DNA FISH using BAC8 as a probe showed the ectopic Xic, when hemizygous (+/Tg) in the male germ line, to be compact chromatin visualized as a small and tight pinpoint signal (Tg2086 is shown as an example in Fig. 3A). This compact signal is typical of observations of BAC8 Tgs in somatic cells (e.g., in mouse embryonic fibroblasts) (Fig. 1D) and therefore was not surprising.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Epigenetic consequences of the +/Tg versus Tg/Tg states in the male germ line. (A) Pachytene spermatocytes from Tg2086 males. Hemizygous (+/Tg) or homozygous (Tg/Tg) transgenic integrations examined by immunostaining/DNA FISH and DAPI. The pachytene stage is identified by the presence of the XY body γH2AX domain. On the XY body, a defined γ-H2AX domain was present in 64% (n = 139) of +/Tg pachytene spermatocytes and in 63% (n = 116) of Tg/Tg pachytene spermatocytes. DNA FISH was carried out using a BAC8 DNA probe, which marks the transgenic Xist outside the XY body γH2AX domain. Arrowheads indicate Tg insertion sites. (B) Quantitation of chromatin decompaction. Z-stack images were captured at 0.2-μm intervals using a Zeiss LSM780 confocal microscope, and the 3D image was projected on a single 2D plane. Using image analysis tools in the Volocity software, we measured the size and spread of the DNA FISH signal by drawing a minimal loop that is inclusive of the full signal. The area within the loop represents the Tg chromatin spread (a). We measured the nucleus size (A) by drawing a loop covering the full DAPI signal. The spread of the DNA FISH signal (a) divided by the size of the corresponding nucleus (A) is the normalized compass (N.C.). N.C. = a/A, which is the condensation index. n, number of cells measured for each genotype. P, statistical significance of difference between the two distribution profiles (+/Tg versus Tg/Tg) calculated by the nonparametric Kolmogorov–Smirnov test. (C) DNA FISH of spermatogonial cells from 2086 Tg/Tg males. The BAC5 DNA probe (probe 2 in Fig. 1A) marks the endogenous Xic; the BAC8 DNA probe marks the transgenic Xist. Arrowheads indicate Tg insertion sites. (D) Xist expression levels in the testes. qRT-PCR of Xist normalized by Gapdh for WT and transgenic testis samples. +/Tg and Tg/Tg males are littermates obtained from the Tg2086 or Tg2087 mouse line. Triplicates were performed for each sample. Averages ± SE are shown. The P values indicated in the pairwise comparisons were determined by the one-tail t test.

On the other hand, when present in the homozygous stage (Tg/Tg), the Xic chromatin became decompacted, with the DNA FISH signals appearing much more diffuse than typical for somatic cells (Fig. 3A, Lower). To quantitate this difference, we performed DNA FISH and compared the sizes of the BAC8 transgenic signals in Tg/Tg versus +/Tg cells (Fig. 3B). We captured z-stack images at 0.2-μm intervals and measured the size and spread of the DNA FISH signals, expressed in terms of a “condensation index,” as defined by the normalized compass (see the legend of Fig. 3B). A significant difference was observed between the sizes of the Tg loci in +/Tg versus Tg/Tg cells, with loci in Tg/Tg cells occupying a greater territory and, hence, being more open (Fig. 3B). On the other hand, when the 2086 Tg/Tg spermatocytes had not yet entered meiosis (spermatogonial cells), the BAC8 transgenic loci were well separated, and both loci appeared as small pinpoints (Fig. 3C). We conclude that pachytene chromatin is treated differently during pachytene and that the hemizygous, unpaired state of BAC8 sequences resulted in a greater degree of chromatin compaction.

The compaction state of chromatin has been well correlated with the gene activity state, with condensed chromatin associated with gene repression and decondensed chromatin associated with gene activation (47⇓–49). Therefore one might predict that the Tg would be better expressed in the more open chromatin of the homozygous state. However, we were surprised to see increased Xist expression in the +/Tg state and lower expression in the Tg/Tg state (Fig. 3D; elevated Xist expression of the +/Tg cells was not associated with a large Xist cloud). This paradoxical finding is consistent with a previous report indicating that Xist expression is favored by the heterochromatic state of the Xic in somatic cells (50) and also is consistent with Xist being expressed only from the silent Xi (8). Thus, hemizygosity affects not only the chromatin compaction of the Tg but also its expression status during male pachytene. We conclude that Xist expression is favored, in general, by condensed chromatin.

Paternal Effects: Xist Imprinting Promoted by Hemizygosity.

The effect of the hemizygous state on Xist expression was arguably small (twofold). However, because Xist normally is expressed only at very low levels in the male germ line (several RNA copies only) (18, 19), the small decrease was likely all that could be observed in the germ line. Xist is not fully up-regulated until the gene is passed on to female progeny. In mice, the paternal Xist allele is induced initially in the two-cell embryo and becomes fully up-regulated in the morula (14, 16, 17, 22, 41). This paternal-specific Xist expression is responsible, at least in part, for imprinted XCI in the early mouse embryo and in extraembryonic tissues of later-stage embryos (5, 7, 16, 41).

To determine whether the hemi- or homozygosity of the Tg could affect its imprinting, we tested transgenic Xist expression in blastocysts resulting from Tg/Tg versus +/Tg fathers using the Tg2086 line, the only line without the sterility or lethality phenotypes that would preclude homozygosis. In agreement with previous reports on up-regulation of the paternal Xist in the zygotes (41, 51), transgenic Xist inherited from a hemizygous Tg father showed significant expression in the trophectoderm, an early extraembryonic tissue first formed in the blastocyst (Fig. 4A). On the other hand, transmission from a hemizygous mother did not result in Xist expression in either male or female embryos (Tg/+; Fig. 4 B and C). Thus, the effect of hemizygosity on imprinting is paternal specific, as is consistent with an activating Xist imprint being applicable only in the male germ line [repressive imprinting being a maternal-specific phenomenon (24⇓–26)]. The paternal-only Tg expression was observed in both sons and daughters of the +/Tg father, indicating a true imprinting phenotype rather than a sex-specific property and arguing that the paternal germline imprints the Xist gene irrespective of the sex of the inheriting progeny. By performing DNA FISH using a combination of a Xist probe and a nonoverlapping Xic probe (BAC5, outside the BAC8 sequence; probe 2 in Fig. 1A), we could distinguish between the transgenic and the endogenous loci (Fig. 4D). The endogenous Xist gene was silent in all cells of the male embryo, but it was up-regulated in one of two female X chromosomes in the female embryo, as expected (Fig. 5 A and B). These observations held true for multiple independent transgenic lines (e.g., Tg2087, Tg2082, and Tg2086) (Fig. 4A). We therefore conclude that the BAC8 Tg carries sufficient genetic information to imprint Xist correctly.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Imprinted Xist expression in the transgenic preimplantation embryo. (A) Transgenic Xist expression was observed in trophectoderm cells of blastocyst-stage (E3.5) embryos in which the Tg was inherited exclusively from the father (+/Tg) by crossing heterozygous fathers to WT mothers. Xist and Tsix RNA FISH were performed using strand- and locus-specific LNA or DNA oligonucleotide probes. Trophectodermal cells were sampled from the mural trophectoderm surrounding the blastocoel to avoid inner cell mass cells located at the polar region of the embryo. Arrowheads indicate Tg insertion sites. (B) No transgenic Xist expression was detected in trophectoderm cells of blastocyst-stage (E3.5) embryos when the Tg was inherited exclusively from the mother (Tg/+) by crossing heterozygous mothers to WT fathers. Xist RNA FISH was performed using locus-specific LNA probes. (C) Transgenic Xist expression in the blastocyst was affected by the Tg zygosity of the father but not of the mother. For each transgenic embryo (E3.5 blastocyst) with Tg derived from the father, the percentage of cells showing transgenic Xist expression was counted. At least five embryos were analyzed for each category. ^†TgXist RNAs in Tg/Tg cells were qualitatively different, ranging from small pinpoints to modest RNA clouds (Fig. 5). Averages ± SE are shown. *P < 0.05 as calculated by Student’s t test comparing the result of Tg/Tg father with that of +/Tg fathers, both from Tg2086. By convention, the maternal genotype is noted first: Tg/+, maternal transmission; +/Tg, paternal transmission. (D) Xist and BAC5 DNA FISH was performed after the RNA FISH of cells shown in A, with the BAC5 probe (probe 2 in Fig. 1A) marking the endogenous Xic. The Tg (arrowheads) is indicated by Xist DNA signal (red) that is not associated with the BAC5 DNA signal (green).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

Imprinted Xist expression in the progeny is promoted by a hemizygous state in the paternal germ line. (A) Representative E3.5 blastocysts with Xist RNA FISH (Cy5, light blue) and DAPI. Two embryos are shown for each genotype. WT and transgenic embryos derived from heterozygous +/Tg fathers crossed with WT mothers (Top and Middle) and transgenic embryos derived from homozygous Tg/Tg fathers crossed with WT mothers (Bottom). Male and female embryos from Tg2086 are shown. Boxed nuclei in A are magnified in B. Arrowheads Indicate transgenic locus. (B) Xist RNA FISH followed by Xist and BAC5 DNA FISH of the nuclei boxed in A. Arrowheads depict the transgenic Xist locus. The endogenous versus transgenic Xist locus is distinguished by hybridization to the BAC5 probe, which detects only the endogenous locus.

A previous study suggested that Xist imprinting was not affected by the hemi- or homozygosity of the Xist Tg in the male germ line (41). However, as discussed above, the Tg53/Tg80 Tgs may not be fully sufficient to recapitulate imprinted XCI. Given the effects of zygosity on chromatin decompaction and pachytene expression in our transgenic system (Fig. 3), we asked whether imprinting also might be affected by homozygosity of the Tg. Most intriguingly, homozygosing the Tg in the paternal germ line significantly reduced transgenic Xist expression in the trophectoderm (Figs. 4C and 5 A and B). In resulting transgenic male embryos, the number of Xist-expressing cells was reduced dramatically in the embryos inheriting the Tgs from a homozygous father, and Xist was absent in most of the cells (example 1 in Fig. 5B). In resulting female embryos, the number of trophectodermal cells with two Xist RNA clouds (one transgenic, one endogenous) was reduced similarly, and the transgenic Xist cloud was mostly absent (example 3 in Fig. 5B). Additionally, among the minority of Tg/Tg cells that did express transgenic Xist, the Xist foci ranged from small pinpoints to modest RNA clouds and therefore were qualitatively different from normal Xist clouds (Fig. 5). Thus, although the Tg carries sufficient genetic information to imprint Xist, imprinting cannot be executed properly when the Tg is present in the homozygous state during male meiosis.