Report

Impeding Xist Expression from the Active X Chromosome Improves Mouse Somatic Cell Nuclear Transfer

See allHide authors and affiliations

Science  22 Oct 2010:
Vol. 330, Issue 6003, pp. 496-499
DOI: 10.1126/science.1194174

Abstract

Cloning mammals by means of somatic cell nuclear transfer (SCNT) is highly inefficient because of erroneous reprogramming of the donor genome. Reprogramming errors appear to arise randomly, but the nature of nonrandom, SCNT-specific errors remains elusive. We found that Xist, a noncoding RNA that inactivates one of the two X chromosomes in females, was ectopically expressed from the active X (Xa) chromosome in cloned mouse embryos of both sexes. Deletion of Xist on Xa showed normal global gene expression and resulted in about an eight- to ninefold increase in cloning efficiency. We also identified an Xist-independent mechanism that specifically down-regulated a subset of X-linked genes through somatic-type repressive histone blocks. Thus, we have identified nonrandom reprogramming errors in mouse cloning that can be altered to improve the efficiency of SCNT methods.

Cloned animals have been generated from embryonic cells (blastomeres) or somatic cells by nuclear transfer. The latter type of cloning, somatic cell nuclear transfer (SCNT), has more practical applications and has been applied successfully to more than 20 animal species (1). However, despite extensive efforts to improve the technique, the efficiency in terms of normal birth remains low. For example, screening for tissue-specific stem cells that might provide a more efficient donor cell has shown limited success (2, 3). The observation of many SCNT-specific phenotypes in cloned animals, such as placental abnormalities (4) and immunodeficiency (5), led us to hypothesize that SCNT might be associated with some definable, nonrandom epigenetic errors. By combining genetics, functional genomics, and cloning technologies, we now identify nonrandom reprogramming errors in cloned embryos that provide promising clues for improving SCNT cloning.

To define the gene expression patterns specific for SCNT, we generated mouse embryos from cumulus cells and immature Sertoli cells under standardized SCNT conditions (6). Single cloned blastocysts were analyzed for their global gene expression patterns by comparing them with genotype-matched controls produced by means of in vitro fertilization (IVF) at the same time (7). When the relative expression levels of filtered genes in cloned embryos taken from a 44,000 oligo DNA microarray were plotted on the 20 chromosomes, genes on the X chromosome were specifically down-regulated (Fig. 1A). This phenomenon was sex- and genotype-independent because the average X:autosome (X:A) expression ratio in the three types of cloned embryos (cumulus and Sertoli clones with different genotypes) was consistently lower than in the corresponding control embryos (Fig. 1B). Detailed observations on the entire X chromosome revealed that although there seemed to be some gene-specific variations, the X-linked genes were largely down-regulated in most regions (Fig. 1C). We then performed a statistical analysis using Student’s t test to identify the number of affected genes in cloned embryos. In each clone group, 2560 to 5540 out of 39,448 gene probes were expressed differentially as compared with that of the genotype-matched IVF controls (fig. S1A). However, the affected genes common to all the clone groups represented only 129 genes (145 probes), with 90 being up-regulated and 39 down-regulated (fig. S1B). Thus, SCNT caused dysregulation of a large subset of genes, but most followed a pattern specific to each donor cell type. Twenty-one out of 39 (54%) of the commonly down-regulated genes (“CDGs”) were mapped to the X chromosome (P < 1.0 × 10−72 versus the expected number from the X-linked gene population with Pearson’s χ2test) (table S1 and fig. S1C), compared with a nonbiased population of genes in the up-regulated genes (P > 0.05) (table S2 and fig. S1C). For some CDGs, their clone-associated down-regulation was confirmed by means of quantitative real-time polymerase chain reaction (RT-PCR) experiments (fig. S2). We also analyzed embryos cloned from fibroblasts and from blastomeres of four-cell embryos and confirmed that the X-linked down-regulation largely could be attributed to SCNT cloning and not generally to nuclear transfer cloning (fig. S3 and table S1). Next, we tested whether the X-linked down-regulation of cloned embryos could be ameliorated through treatment with trichostatin A (TSA), which is a histone deacetylase inhibitor (HDACi) known to improve mouse cloning efficiency (8). However, this treatment produced no significant improvement in the X:A expression ratio (P > 0.05) (Fig. 1B) or in the expression levels of X-linked genes (Fig. 1C) as compared with that of untreated cloned embryos.

Fig. 1

Large-scale down-regulation of X-linked genes in SCNT embryos. (A) A representative pattern of relative gene expression levels of a B6D2F1 IVF embryo, a cumulus cell cloned embryo, and a Sertoli cloned embryo, plotted on the genomic positions from chromosomes 1 to X (except for Y). The red bar indicates down-regulated X-linked genes in a cloned embryo. (B) The ratio of the expression levels of X-linked genes to autosomal genes. Wild-type cloned embryos, including those treated with TSA, showed lower X:A ratios as compared with the corresponding IVF controls. The data are represented as the mean ± SEM. a, a′P < 0.01, b, b′P < 0.0001, c, c′P < 0.05 (one-way analysis of variance and Student’s t test). (C) Relative gene expression levels of (red) cumulus cell cloned embryos, (blue) cumulus cell cloned embryos treated with TSA (n = 3 embryos), and (gray) IVF embryos plotted on the positions of the X chromosome. Dotted lines represent a single embryo, and solid lines indicate their mean values. Arrowheads 1 and 2 indicate the position of the Xlr and Magea clusters, respectively (Fig. 3B).

The chromosome-wide gene down-regulation on the X chromosome in cloned embryos was reminiscent of X chromosome inactivation (XCI). This process normally triggers inactivation of one of the two X chromosomes in female embryos so that the gene dosage is comparable with that in males. XCI is initiated by Xist RNA coating in cis, although it is completed and maintained by many other molecules (Fig. 2D) (9). We then examined whether Xist was expressed excessively in our cloned embryos, as has been reported for embryos cloned from cumulus cell nuclei (10, 11). In both female and male cloned embryos, the Xist expression levels were significantly higher than in control IVF embryos (P < 0.05), as confirmed with microarray (Fig. 2A) and quantitative RT-PCR analyses (fig. S4A). We postulate from these findings that Xist was expressed ectopically from the active X chromosome (Xa) in cloned embryos. We then observed the number of Xist domains within each blastomere nucleus at the morula or early blastocyst stage by use of RNA fluorescent in situ hybridization (FISH). As expected, about half of the IVF embryos consistently showed a single domain in each blastomere, and the remaining half showed no domain, probably representing female and male embryos, respectively (Fig. 2, B and C). In female clones, all four embryos contained blastomeres with unusual biallelic Xist domains with a variable frequency from 20.0 to 51.7% (Fig. 2, B and C). In male clones, all seven embryos analyzed showed one strong Xist RNA domain in the majority of blastomeres (Fig. 2, B and C), whereas their donors had no Xist expression (12). We could exclude the possibility of involvement of tetraploidy in this excessive number of Xist RNA domains because there were very few blastomeres with duplicated X chromosomes in the cloned embryos (fig. S5). We further confirmed the localization of trimethylated histone H3 at lysine 27 (H3K27me3) and of Eed, which are responsible for the repressive chromatin state in the inactive X (9). In male and female cloned embryos, they colocalized exclusively in one and two domains in the nucleus, respectively, suggesting that the ectopic Xist expression indeed leads to XCI (Fig. 2D). The ectopic expression of Xist first appeared at the four-cell stage and increased up to the blastocyst stage, as revealed through quantitative RT-PCR and RNA FISH analysis by use of male cloned embryos (fig. S6). These findings support our hypothesis that Xist is ectopically expressed and aberrantly inactivate Xa in both male and female clones. At present, we do not know the causes of the ectopic expression of Xist in cloned embryos. However, because it is assumed that the major mechanisms of genomic memory for Xi (or conversely, Xa) in preimplantation embryos and somatic cells are different (9, 13, 14), reestablishment of the Xi (or Xa) memory in the somatically derived genome in reconstructed embryos might have been incomplete.

Fig. 2

Xist is ectopically expressed on the active X chromosome in female and male cloned embryos. (A) The expression levels of Xist in female and male embryos. The expression levels are significantly higher in cloned embryos of both sexes (P < 0.05, Student’s t test). (B) Morula or early blastocyst stage embryos with localizations for (red) Xist RNA and (blue) nucleus. Ectopic expression of Xist is evident from the existence of two domains in females (white arrowheads) and one domain in males for Xist RNA (black arrowheads). (C) The ratio of blastomeres classified by the number of Xist RNA domains within single embryos (0 to 2). Each bar represents one embryo. (D) Immunostaining for H3K27me3 and Eed in IVF and cloned blastocysts. The signals of H3K27me3 and Eed are colocalized in single or double domains within blastomere nuclei. There are two localizations in embryos cloned from cumulus cells (females) and one in embryos cloned from Sertoli cells (male), suggesting that the Xa chromosome is inactivated aberrantly in cloned embryos of both sexes.

Because Xist has a chromosome-wide repressive effect on X-linked genes in cis, next we asked to what extent its ectopic expression might be responsible for the aberrant gene expression observed in cloned embryos. To this end, SCNT was performed by using donor cells containing an Xist-deficient (XΔXist) X chromosome (15) for Xa and analyzed the embryos for their gene expression patterns. In both female (cumulus cell) and male (Sertoli cell) clones, the numbers of down-regulated X-linked genes in wild-type clones were considerably decreased in XΔXist clones by 85% (80 → 12) and 85% (141 → 21) in female and male embryos, respectively (Fig. 3A). This effect is clearly noted in the upper shift of the gene expression levels plotted on the X chromosome (Fig. 3, B and C) and A:X ratios (Fig. 1B). This had a genome-wide effect, and the numbers of down-regulated autosomal genes also decreased by 85% (461 → 71) and 73% (340 → 91) in female and male embryos, respectively (Fig. 3A). These results indicate that the ectopic Xist expression could have adversely affected gene expression in cloned embryos in a genome-wide manner, probably through complex gene networks connecting autosomal genes and X-linked genes that direct embryonic development. However, two discrete groups of genes remained down-regulated (Fig. 3, B and C). These were the Magea and Xlr gene clusters localized on XqF3 and XqA7.2–7.3, respectively (Fig. 1C). Twelve of the 21 X-linked CDGs were classified into one of these two clusters [table S1 and supporting online material (SOM) text].

Fig. 3

Deletion of Xist on the active X chromosome (Xa) in SCNT embryos improves their gene expression patterns and developmental ability in vivo. (A) The numbers of down-regulated genes in SCNT embryos compared with corresponding IVF embryos. With deletion of Xist on the Xa, they are reduced by 73 to 85% for both the X chromosomes or autosomes in both female cumulus cell and male Sertoli cell clones. (B and C) The relative expression levels of X-linked genes plotted on the X chromosome position in (B) cumulus and (C) Sertoli cell cloned embryos. The majority of down-regulated genes are increased in their expression levels, except for genes within the Xlr (arrowheads 1) or Magea (arrowheads 2) clusters, in Xist-knockout clones (n = 5 and 4 cumulus and Sertoli clones, respectively, in green), compared with (red) wild-type cloned embryos (SOM text). (D) The birth rates per embryos transferred. Eight- to ninefold increases were observed in Xist knockout clones. (E) Fetuses born after nuclear transfer by using Sertoli cells (left) with or (right) without the Xist gene on Xa. The birth rates were 1.6 and 15.4% of embryos transferred, respectively (table S2).

In the next series of experiments, we transferred SCNT embryos containing an Xist-deficient Xa into pseudopregnant recipient females. In both cumulus and Sertoli cell–derived clones, their development was greatly improved; the average birth rates reached 12.7 and 14.4% per embryos transferred (up to 19.2%), corresponding to eight- to ninefold higher levels than wild-type controls, respectively (Fig. 3, D and E, and table S3). Mouse cloning from this standard genetic strain background (B6D2F1) has not reached such high efficiencies (1, 16). Most clones grew into normal adults and showed no gross abnormalities (table S3).

In this study, we identified two types of SCNT-associated errors specifically affecting the X chromosome in mice: (i) the ectopic Xist expression from Xa and (ii) persistence of repressive histone modifications (H3K9me2) in the Magea and Xlr regions (SOM text). These errors were resistant to TSA treatment, indicating that they cannot be rescued by simply enhancing the accessibility of the putative ooplasmic reprogramming factors. Thus, we can broadly classify the epigenetic errors in cloned mouse embryos into two categories: One is random and can be overcome to some extent by enhancing genomic reprogramming (such as through HDACi treatment), whereas the other is more specific and probably beyond the ability of the putative ooplasmic factors that are to reprogram the germ cell genome (17). We found that XIST expression was also elevated in female and male bovine SCNT embryos (fig. S4B); therefore, this could have broad implications for improving mammalian SCNT techniques. Indeed, there is a clear association between the death of bovine cloned embryos and aberrant X-linked gene expression in the placenta (18). Because the data presented in this study are still limited, it is necessary to examine whether certain genetic or epigenetic modifications for XIST might improve the survival of SCNT embryos using other mammalian species or mice from different strains.

A major goal of cloning research is to increase the efficiency of mammalian SCNT to a practical level (for example, >20% per embryos transferred) because of the many potential applications in biological drug manufacturing, regenerative medicine, and agriculture (19). To this end, we need to overcome the fundamental differences between somatic and germ cell genomes. We expect that cloning will become more practical by specifically targeting nonrandom epigenetic errors associated with SCNT.

Supporting Online Material

www.sciencemag.org/cgi/content/full/science.1194174/DC1

Materials and Methods

SOM Text

Figs. S1 to S8

Tables S1 to S3

References

References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. The mice in which Xist had been knocked out (RBRC 01260) were provided by the RIKEN BioResource Center. This research was supported by grants from the Ministry of Education, Culture, Sports, Science and Technology and NOVARTIS Foundation (Japan) for the Promotion of Science. The H3K9me2 antibody used for chromatin immunoprecipitation on chip was a kind gift from H. Kimura. We thank M. Tachibana, Y. Shinkai, H. Koseki, T. H. Endo, and S. L. Marjani for their invaluable suggestions. The microarray data have been deposited in the Gene Expression Omnibus and given the series accession number GSE23181.
View Abstract

Navigate This Article