SATB1 defines the developmental context for gene silencing by Xist in lymphoma and embryonic cells

Abstract

The noncoding Xist RNA triggers silencing of one of the two female X chromosomes during X inactivation in mammals. Gene silencing by Xist is restricted to a special developmental context in early embryos and specific hematopoietic precursors. Here, we show that Xist can initiate silencing in a lymphoma model. We identify the special AT-rich binding protein SATB1 as an essential silencing factor. Loss of SATB1 in tumor cells abrogates the silencing function of Xist. In lymphocytes Xist localizes along SATB1-organized chromatin and SATB1 and Xist influence each other's pattern of localization. SATB1 and its homolog SATB2 are expressed during the initiation window for X inactivation in ES cells. Importantly, viral expression of SATB1 or SATB2 enables gene silencing by Xist in embryonic fibroblasts, which normally do not provide an initiation context. Thus, our data establish SATB1 as a crucial silencing factor contributing to the initiation of X inactivation.

Figure 1. Xist Initiates Ectopic X Inactivation in Lymphoma Cells and Blocks Tumorigenesis

(A) Tumor growth was monitored in Doxycycline (Dox)-treated (red) and untreated (black) nude mice after subcutaneous injection of 2 × 10⁶ NPM-ALK TX/Y R26^rtTA/rtTA lymphoma cells (n = 10). Error bars represent standard deviation. (B) Tumor weight on Day 30 (n = 10). (C and D) Tumor morphology (C) and TUNEL analysis ([D], green) of histological tumor sections showing apoptotic cells in the Dox-treated mice. Nuclei are counterstained with DAPI (blue). Scale bar = 10 μm. (E) Xist induction on day 12 after tumor graft (blue arrow) causes remission of established tumors (blue). Untreated tumors (black) and tumors treated with Dox from day 2 after tumor graft (red) are controls (n = 10). Error bars represent standard deviation. (F) NPM-ALK TX/Y R26^rtTA/rtTA lymphoma cell cultures on feeders on days 1 and 5 with and without Dox. Scale bar = 50 μm. (G) RT-PCR showing silencing of the X-linked Pgk1 and Hprt genes after 48 hr of ectopic Xist induction in NPM-ALK TX/Y R26^rtTA/rtTA lymphoma cells. (H and I) Combined RNA FISH immunofluorescence analysis showing colocalization of H3K27me3 (H) and H2AK119ub1 (I) with Xist RNA. Percentage of total cells is given in the panels (n = 500). Scale bar = 5 μm.

Figure 2. Derivation of an Xist-Resistant Tumor for Identification of Silencing Factors

(A) Tumor growth after transplantation of 2 × 10⁶ cultured NPM-ALK TX/Y R26^rtTA/rtTA lymphoma cells into nude mice. After 22 days (red arrow), Xist-resistant tumors appear in the Dox-treated group (red; n = 5). Error bars represent standard deviation. (B) Xist RNA FISH (red) and H3K27me3 (green) in histological sections of the resistant tumor. Percentage of colocalization is given. (C) Quantitative RT- PCR analysis showing that the X-linked Pgk1 and Hprt genes are still expressed in the resistant tumors when Xist is induced with Dox at a comparable level to normal thymus or parental tumor without Xist induction. Results were normalized to Gapdh. Error bars represent standard deviation (n = 3). (D) Heatmap showing upregulated (red) and downregulated (green) genes after hierarchical clustering of gene expression profiles of Xist-resistant and Xist-responsive NPM-ALK TX/Y R26^rtTA/rtTA lymphoma and normal male thymus. Probes for Xist and SATB1 are indicated (arrows) and gene groups are annotated (right).

Figure 3. SATB1 Is a Critical Silencing Factor for Xist in Lymphoma Cells

(A) Western analysis of SATB1 expression in Xist-resistant and Xist-responsive tumor and normal thymus. Actin serves as loading control. (B) Immunohistochemistry in sections reveals nuclear ring-like SATB1 structures in normal thymus and Xist-responsive tumor, whereas nuclear SATB1 staining is lost in resistant tumors. (C) SATB1 RNAi in cultured NPM-ALK TX/Y R26^rtTA/rtTA lymphoma cells significantly increases cell survival after Xist induction. Cell survival after 5 days of culture in presence of Dox relative to uninduced cultures is plotted. Scrambled siRNA treatment serves as control (n = 3). Error bars represent standard deviation. (D) Western analysis confirming loss of SATB1 protein in SATB1 siRNA treated but not scrambled control siRNA treated cells. Lamin B serves as loading control. (E) Virally expressed SATB1 restores Xist silencing function in Xist-resistant tumor cells. Cell survival after 5 days of cultures in presence of Dox relative to uninduced cultures is plotted. Empty vector-infected cells serve as control (n = 3). (F) Western analysis confirming SATB1 protein expression in cells infected with SATB1 but not control virus. Actin serves as loading control. (G) Xist RNA FISH (red) combined with SATB1 immunofluorescence (green) analysis showing Xist spreading along the SATB1 ring in TX/Y R26^rtTA/rtTA thymic cells after 48 hr of Xist induction with Dox. (H) Delocalization of Xist into a ring-like pattern is observed in thymocytes of wild-type (WT) but not SATB1 deficient female mice. Representative Xist RNA FISH images are shown. A statistical analysis of Xist patterns in SATB1 deficient and control thymocytes is given (n = 150).

Figure 4. SATB1 Expression Defines the Silencing Function of Xist in Embryonic Cells

(A) Combined RNA FISH immunofluorescence analysis of TX/Y R26^rtTA/rtTA ES cells shows SATB1 (green) surrounding the characteristic Xist focus (red) after 24 hr of induction with Doxycycline. Scale bar = 5 μm. (B and C) Western analysis of SATB1 (B) and SATB2 (C) protein during ES cell differentiation shows that SATB1 is lost and SATB2 shifts to a lower molecular weight on day 3, precisely when Xist loses its silencing capability. Asterisk indicates the high molecular weight form of SATB2. Actin serves as loading control. (D) Repression of the puromycin (puro) marker, which is cointegrated with an Xist transgene on chromosome 11 in clone 36 ES cells, upon Xist induction is reduced by SATB1 or SATB2 RNAi compared to scrambled control. Quantification of puro expression after Xist induction relative to uninduced cells is given. (E) Western analysis demonstrates strongly reduced SATB1 protein in ES cells treated with SATB1 siRNA but not control siRNA. Western analysis of SATB2 in ES cells treated with scrambles control, SATB2, and SATB1 siRNAs. The higher molecular weight form of SATB2 (asterisk) observed in ES cells is depleted by SATB2 RNAi. The lower molecular weight form of SATB2 is observed in brain. Lamin serves as loading control. (F) Survival of immortalized fibroblasts established from TX/Y R26^rtTA/rtTA embryos and infected with SATB1 and SATB2 expression vector or control empty vector (vector). The percent survival is calculated from the cell number with Xist induction relative to untreated cells (n = 6). Error bars represent standard deviation. (G) Quantitative RT-PCR showing repression of the X-linked Hprt and Pgk1 genes upon Xist induction in MEFs infected with SATB1 and SATB2 but not empty vector controls. Expression relative to Gapdh is plotted (n = 6). Error bars represent standard deviation.