Maternal Rnf12/RLIM is required for imprinted X-chromosome inactivation in mice

Abstract

Two forms of X-chromosome inactivation (XCI) ensure the selective silencing of female sex chromosomes during mouse embryogenesis. Imprinted XCI begins with the detection of Xist RNA expression on the paternal X chromosome (Xp) at about the four-cell stage of embryonic development. In the embryonic tissues of the inner cell mass, a random form of XCI occurs in blastocysts that inactivates either Xp or the maternal X chromosome (Xm). Both forms of XCI require the non-coding Xist RNA that coats the inactive X chromosome from which it is expressed. Xist has crucial functions in the silencing of X-linked genes, including Rnf12 (refs 3, 4) encoding the ubiquitin ligase RLIM (RING finger LIM-domain-interacting protein). Here we show, by targeting a conditional knockout of Rnf12 to oocytes where RLIM accumulates to high levels, that the maternal transmission of the mutant X chromosome (Δm) leads to lethality in female embryos as a result of defective imprinted XCI. We provide evidence that in Δm female embryos the initial formation of Xist clouds and Xp silencing are inhibited. In contrast, embryonic stem cells lacking RLIM are able to form Xist clouds and silence at least some X-linked genes during random XCI. These results assign crucial functions to the maternal deposit of Rnf12/RLIM for the initiation of imprinted XCI.

Fig. 1

RLIM accumulates during oocyte maturation. Oocytes are indicated by red arrows in A–D. A) Immunohistochemical staining of ovary sections from a fl/fl female 10 weeks of age using RLIM antibodies. Right panel: Higher magnification of boxed region. B) RLIM staining of ovaries from 5 week old fl/fl animals. C) MMTV-LTR-induced Cre expression in oocytes. The Rosa26 loxP-Stop-LoxP –lacZ reporter strain mouse (R26R) was crossed to MMTV-Cre (line F) mice. Paraffin-embedded sections of ovaries of 5 weeks old females were stained with β-Gal antibodies. D) Ovarian section of a 10 weeks old Rnf12^fl/fl x MMTV-Cre female stained with RLIM antibodies. Note the loss of RLIM detection in oocytes but not surrounding granulosa cells. E) Knockout of the Rnf12 gene leads to a loss of RLIM protein in mice. Upper panel: Western blot analysis of protein extracts prepared from brain and breast tissues of wt and Rnf12 KO male mice (wt/Y and Δ/Y, respectively) stained with RLIM antibodies. Lower panel: As control, the same blot was probed with GAPDH antibodies. Scale bars = 80 μm.

Fig. 2

A maternally transmitted Rnf12 deletion allele leads to early embryonic lethality specifically in females. Embryos were first photographed and then processed for genotyping in B–H. A) MMTV-Cre mediated loss of Δm females. Schematic diagram of born pups of indicated mating schemes (1–5). Parental genotypes of female (upper) and male (lower) mice with respect to Rnf12 and MMTV-Cre is shown and the total number (n) of F1 offsprings and the mean litter size is indicated. Number of offsprings (grouped in female and male) and their genotypes with respect to Rnf12 are indicated in the abscissa and ordinate, respectively. m (maternal) and p (paternal) indicate the origin of the KO (Δ) allele. In mating scheme 4 and 5, maternally transmitted wt, floxed and Δ alleles are indicated in grey, blue and red, respectively. Three asterisks indicate P values <1×10⁻⁷. B–E) Heterozygote Δm/fl female and homozygote Δm/Y littermates from a fl/Δp × fl/Y cross at E7.5 (A); E8.5 (B); E9.5 (C) and E10.5 (D). F) Heterozygote fl/Δp female at E10.5 from a fl/fl × Δ/Y cross. G) Best developed Δm/fl embryo (n=28) detected at E10.5 (magnification of the Δm/fl shown in F). H) Representative Δm female embryo at E10.5. Scale bars = 0.15 mm (B); 0.4 mm (C); 0.6 mm (D); 1 mm (E, F); 0.5 (G); 0,25 mm (H).

Fig. 3

Rnf12 is not required for initiation of random XCI. A) MEFs of E12.5 of fl/Δp and fl/Y males were co-stained with RLIM and H3K27me3 antibodies or with a Xist probe using RNA FISH. Upper panel: Summary graph representing 3 independent experiments. Error bars represent SD. Cells were scored for RLIM expression, the presence of H3K27me3 staining or Xist clouds. Lower left panel: Representative image with RLIM (red) and H3K27me3 (green) staining. Lower right panel: Representative image of Xist staining (red). B) Left panel: Summary graph of RNA FISH experiments on E12.5 fl/Δp MEFs co-stained with probes against Xist and Rnf12. Left panel: Representative images showing monoallelic expression. C, D) Time course of initiation of XCI (C) and silencing of Rnf12 (D) in ES cell lines. ES cells were EB differentiated for the indicated time before co-staining with RNA FISH using Xist and Rnf12 probes. Percentage of Xist-positive ES cells (C). Error bars represent SD. Asterisks indicate significant differences between wt and each of the three Δ cell lines (P < 0.05). Xist positive cells were scored for monoallelic or biallelic Rnf12 expression, or no signal (D). Upper panel: Summary graph; lower panel: representative images of various ES cells with biallelic and monoallelic Rnf12 expression. Asterisks indicate significant differences between wt and each of the three Δ cell lines (P < 0.05).

Fig. 4

Regulation of Xist cloud formation and X silencing by Rnf12 during imprinted XCI. Probes for RNA FISH were Xist (green) and Rnf12 (red). A) Representative E4.5 blastocyst outgrowths stained with Xist. ICM = inner cell mass; troph = trophoblasts. B) Representative trophoblasts of fl/wt, Δm/wt, and Δ/Δ stained with Xist and Rnf12 are shown. Right panel: For quantification, Xist-positive trophoblasts (clouds or pinpoints) were scored for co-localization with Rnf12, no co-localization with Rnf12, and no Rnf12 signal. Numbers of blastocysts and trophoblasts examined are indicated. C) Representative 8-cell stage embryos. Boxed area is magnified on the right. Right panel: Quantifications. D) Representative 4-cell stage embryos hybridized with Xist and Rnf12 probes. Right table: Summary of Xist signals detected in embryos at the 4 and 8 cell stages.