Autosomal Dominant Mutations Affecting X Inactivation Choice in the Mouse

See allHide authors and affiliations

Science  10 May 2002:
Vol. 296, Issue 5570, pp. 1136-1139
DOI: 10.1126/science.1070087


X chromosome inactivation is the silencing mechanism eutherian mammals use to equalize the expression of X-linked genes between males and females early in embryonic development. In the mouse, genetic control of inactivation requires elements within the X inactivation center (Xic) on the X chromosome that influence the choice of which X chromosome is to be inactivated in individual cells. It has long been posited that unidentified autosomal factors are essential to the process. We have used chemical mutagenesis in the mouse to identify specific factors involved in X inactivation and report two genetically distinct autosomal mutations with dominant effects on X chromosome choice early in embryogenesis.

During early preimplantation development, female cells have two active X chromosomes. As these cells begin to differentiate, they undergo X chromosome inactivation, the epigenetic process that results in the stable silencing of a majority of genes on one X chromosome (1–4). Historically, control of inactivation has been attributed to a genetically defined region on the X chromosome known as the X inactivation center (Xic); three elements have been mapped to Xic that play a role in the mechanism of inactivation. A key component is the Xist gene and its product (5), a large untranslated RNA that coats the inactive X chromosome (Xi). It is transcribed from the Xi, whereas it is repressed on the active X (Xa). Xist is required in cis for initiation and propagation of silencing (2, 3,6). Its antisense counterpart, Tsix(7), is transcribed exclusively from Xa and appears to repress Xist before initiation of X inactivation, thereby maintaining two active X chromosomes (7, 8). Once silenced, the inactive X maintains its state by a variety of epigenetic regulatory mechanisms, including modification of both classic and variant histones, methylation of Xi DNA, and late replication timing (2–4).

The most elusive component of the X inactivation pathway remains the initial choice between the two X chromosomes, the earliest step in the process. In the mouse, choice is under the control of the X-controlling element (Xce), a third locus within the Xic(9–12). Three Xce alleles (Xcea , Xceb , andXcec ) have been described (13,14), and mutation studies of the Xist andTsix genes suggest that they also play a role in choice (2, 3). Further, DXPas34, a region downstream from the Tsix promoter, is hypermethylated onXce alleles that are more likely to remain active (15). It has long been hypothesized that multiple trans-acting autosomal factors interact with cis-acting elements in theXic to determine X chromosome choice (16–19). Recent identification of CTCF-binding sites within Tsix suggests that the general transcriptional regulator CTCF may behave as one such factor (20), although there is no genetic evidence that CTCF influences X inactivation (21). To identify specific trans-acting elements that act on the choice process, we conducted a phenotype-driven genetic screen involving chemical mutagenesis in the mouse.

The screen took advantage of a quantitative mouse model of choice in which the effect of different Xce alleles is used as a baseline measure of the X inactivation pattern (22). Nonrandom choice occurs when a “strong” Xce allele (one that is more likely to be on the active X, such as theXcec allele derived from Mus musculus castaneus mice) is inherited in combination with a “weak” allele (one that is less likely to be on the active X, such asXcea or Xceb derived fromMus musculus domesticus mice). In these heterozygous animals, 25 to 30% of cells choose the X bearing the weakerXce allele as the Xa, instead of the 50% expected for a completely random process (10–14,22). The use of Xcea/c andXceb/c heterozygotes in the genetic screen facilitated recovery of mutations disrupting general aspects of X chromosome choice as well as those specifically involved inXce allelic discrimination.

The X inactivation phenotype was first established in two unmutagenized control populations of Xcea/c andXceb/c heterozygotes to provide a basis for screening the mutagenized populations (22, 23). We then screened 84 Xcea/c and 252Xceb/c G1 female progeny of males treated with N-ethyl-N-nitrosourea (ENU) (Fig. 1, A and D) (23,24). We considered any G1 animals with X inactivation patterns greater than two standard deviations from the mean of the control populations candidates for X inactivation mutants. To avoid recovering animals in which deleterious X-linked mutations unrelated to the X inactivation choice pathway caused highly skewed X inactivation patterns secondary to differential cell survival (25), we concentrated on animals that preferentially maintained the mutagenized X chromosome as the Xa. To determine the heritability of the putative mutations, we progeny tested G1 females fulfilling the candidate mutant criteria.

Figure 1

Mutagenesis screen and progeny tests of two mutants: 24.21 and 1.19. Distribution of the X inactivation pattern in 84 G1 Xcea/c (closed circles) (A) and 252 G1 Xceb/c (open circles) (D) female progeny of ENU-treated males. Females 24.21 and 1.19 are designated by diagonal arrows in (A) and (D), respectively. The X inactivation pattern reflects the proportion of M. musculus domesticus allele transcription relative to total RNA levels of both M. musculus domesticus and M. musculus castaneus alleles (22, 24). Progeny tests of the Xcea/c G1 female 24.21 (B) and the Xceb/c G1female 1.19 (E) demonstrate dominant transmission of each mutant phenotype through the seventh generation. Unaffected females from each pedigree generated only unaffected female progeny [right (B) and (E)]. In (A), (B), (D), and (E), the horizontal arrowhead represents the mean X inactivation pattern, and the dashed lines represent the 2 SD demarcation of matched control (unmutagenized) populations (22, 26). Representative branches of the 24.21 (C) and 1.19 (F) pedigrees demonstrate dominant segregation of each mutation through affected females (half-filled circles) and through carrier males with a wild-type X chromosome (squares with a dot) but not through unaffected females (open circles). Affected females are defined by phenotypes that are at least 2 SD greater than the mean of controls. Carrier status was confirmed by transmission of the affected phenotype to at least two female offspring (26).

Two G1 females, 24.21 and 1.19 (26), transmitted the altered X inactivation pattern to their female offspring and were investigated further. Both pedigrees segregated the mutant X inactivation pattern through at least the seventh generation and demonstrated dominant inheritance patterns (Fig. 1, B, C, E, and F). In both pedigrees, carrier females (26) displayed X inactivation patterns significantly different from those predicted by their respective Xce genotypes, as the X chromosome bearing the strong Xce allele was the Xi in a significantly higher proportion of cells than expected (22). Phenotyping several generations of outcrossed control animals confirmed that the aberrant X inactivation patterns did not result from genetic background variability (27), consistent with previous studies (22). As further validation of the mutant phenotypes, we progeny tested unaffected female siblings in both pedigrees; as expected, their offspring demonstrated X inactivation patterns not significantly different from controls of the same cross (P > 0.1) (Fig. 1, B, C, E, and F).

Neither carrier males nor females exhibited obvious developmental defects, impaired fertility, or reduced segregation of the mutant phenotypes in either pedigree, which indicates specificity of the mutant phenotype to X inactivation. We characterized carrier mice for each mutation at three different X-linked loci: Xist,Pctk, and Pgk-1 (24). The results of the assays were highly concordant for females tested in both pedigrees (Fig. 2, A and D), which indicates that the phenotypes reflect the chromosomal pattern of X inactivation and do not reflect, for example, simple locus-specific effects. We also characterized the X inactivation pattern in a variety of tissues representing the three embryonic lineages and determined that all tissues are similarly affected in carrier females from both pedigrees (Fig. 2, B and E).

Figure 2

Mutant phenotype analysis.Xcea/c control females and noncarrier and carrier females from pedigree 24.21 (A) andXceb/c control females and noncarrier and carrier females from pedigree 1.19 (D) were phenotyped at the Xist (circles), Pgk-1 (diamonds), and Pctk (squares) loci. Open symbols represent progeny of unmutagenized control animals; filled symbols represent progeny of carriers of the mutation. Animals phenotyped from both pedigrees were informative at all three loci tested and represent a subset of those depicted in Fig. 1, B and E. Carrier status was determined by progeny testing (26). The X inactivation phenotype at the Xist locus was characterized in five tissues [toe/ear (TE), heart (H), kidney (K), liver (L), and skeletal muscle (SM)] in control females and carrier females from pedigree 24.21 (B) and in control females and carrier females from pedigree 1.19 (E). To minimize variability due to clonal cell populations, we sampled whole tissues. (C) X inactivation phenotype at the Xist locus in 6.5 days postcoitus (dpc) and 7.5 dpc control female embryos (open circles) and in 6.5 dpc 24.21 and 7.5 dpc 1.19 carrier embryos (closed circles). Carrier status of 24.21 embryos was established by genotyping at loci on chromosome 15 (34). Carrier status of 1.19 embryos was assigned on the basis of X inactivation patterns > 2 SD above the control mean of postnatal tissues (dashed line).

Because X inactivation occurs early in embryonic development (1, 28–31), these observations suggested that the mutations likely act on an early step in the X inactivation pathway. To address this directly, we assayed the X inactivation pattern in early embryos, after choice has occurred (2,3, 32) but before any secondary cell selection might have taken place (25). Indeed, we detected the mutant phenotype in early embryos of both families (Fig. 2C), confirming that the mutations have a primary effect on X inactivation and clearly ruling out secondary cell selection as the mechanism responsible for the mutant phenotypes. Further, the mutation segregating in pedigree 24.21 induced the mutant phenotype when transferred to a differentXce heterozygous genotype, which confirms the effect of this mutation on X chromosome choice and rules out the possibility of a strain-specific oddity (27). Carrier females from both pedigrees exhibited appropriate imprinting of the well-characterized H19 gene (27, 33). Together, these data strongly indicate that the mutations detected in pedigrees 24.21 and 1.19 specifically affect an early step of the X inactivation pathway to result in primary X inactivation patterns that are significantly different from those predicted by theirXce genotypes.

To determine whether the mutations were X-linked, we examined segregation of the mutant phenotype through two categories of male siblings: those carrying either a wild-type or an ENU-treated X chromosome. If the mutation were X-linked, none of the males in the first category and all in the second should transmit the mutation to each of their daughters. On the other hand, if the mutation were autosomal, half the males in each category should transmit the phenotype to half their daughters. Indeed, about half the males in both categories in each family clearly transmitted the abnormal X inactivation pattern to a subset of their daughters (27), which indicates autosomal dominant modes of inheritance. Analysis of X-linked microsatellite markers in affected and unaffected family members in both pedigrees confirmed these observations (27). These data rule out X linkage in both families and are consistent with expectations of autosomal inheritance.

To localize the mutation in pedigree 24.21, we conducted a genome scan that took advantage of the genetic cross used to generate the pedigree. In this cross, we outcrossed the mutagenized G0BALB/cByJ founder male (23, 26) and all subsequent generations to a tester strain carrying a M. musculus castaneus X chromosome (22, 23). For the mapping analysis, we genotyped five affected and five unaffected distantly related females from the fourth (G4) and fifth (G5) generations with microsatellite markers evenly spaced across the genome with a maximum spacing of 30 centimorgans (cM) [Fig. 3 (27, 34); fig. S1 (35)]. We observed alleles from the mutagenized BALB/cByJ founder male at the expected frequency, about 6.25 and 3.13% in G4 and G5 animals, respectively. The proximal half of chromosome 15 was the only region of the genome with significant linkage to the mutant phenotype (P < 0.004) [Fig. 3 (34); fig. S1 (35)]. This region does not contain any genes for factors known to be involved in any aspect of X inactivation or epigenetic silencing (including CTCF), which indicates that the mutation affects a specific factor in the X inactivation pathway. To determine whether the mutation in pedigree 1.19 involves the same locus, we genotyped chromosome 15 in affected animals from this pedigree (27); we observed no association between the mutant phenotype and chromosome 15. Although mapping of the 1.19 mutation to a specific location has been hindered by the lack of suitable informative markers that distinguish the strains used in this cross (26, 27), these data indicate that the mutation segregating in the 1.19 pedigree involves an autosomal locus distinct from that mutated in the 24.21 pedigree.

Figure 3

Genome scan of pedigree 24.21 demonstrates linkage of the mutant phenotype to chromosome 15. Five affected G4 and G5 females were genotyped with microsatellite markers spaced evenly throughout the genome with a maximum spacing of 30 cM (34). The location of the markers is designated by arrowheads on the left of each chromosome. Genotypes for the five individual females are represented as vertical bars 1 through 5 for each chromosome. Unfilled regions of the bars represent wild-type genotypes (from the nonmutagenized parent); filled regions represent BALB/cByJ alleles inherited from the ENU-treated founder male. The probability that the ENU-treated alleles were inherited in the proximal region of chromosome 15 by chance isP < 0.004. Alleles from the ENU-treated male were not observed on chromosomes 4, 6, 7, 8, 11, 16, 17, and 18 (Other Chs.) (27) [fig. S1 (35)].

These mutations, which we designate X-inactivation autosomal factors 1 and 2 (Xiaf1 and Xiaf2), represent direct genetic evidence of autosomal factors in X inactivation. Functional and genetic studies of possible interactions between Xiaf1 andXiaf2 and elements of the Xic(16–19), as well as further genetic mapping and positional cloning, will be required to elucidate the role of these autosomal factors in the X inactivation pathway.

  • * Present address: Department of Medicine, University of California, San Francisco, CA 94143, USA.

  • To whom correspondence should be addressed. E-mail: bartolom{at} (M.S.B.) and willard{at} (H.F.W.).


Stay Connected to Science

Navigate This Article