SUMO-specific protease 2 is essential for suppression of polycomb group protein-mediated gene silencing during embryonic development

Abstract

SUMO-specific protease 2 (SENP2) has a broad de-SUMOylation activity in vitro. However, the biological function of SENP2 is largely unknown. Here, we show that deletion of SENP2 gene in mouse causes defects in the embryonic heart and reduces the expression of Gata4 and Gata6, which are essential for cardiac development. SENP2 regulates transcription of Gata4 and Gata6 mainly through alteration of occupancy of Pc2/CBX4, a polycomb repressive complex 1 (PRC1) subunit, on its promoters. We demonstrate that Pc2/CBX4 is a target of SENP2 in vivo and that SUMOylation is essential for Pc2/CBX4-mediated PRC1 recruitment to methylated histone 3 at K27 (H3K27me3). In SENP2 null embryos, SUMOylated Pc2/CBX4 accumulates and Pc2/CBX4 occupancy on the promoters of PcG target genes is markedly increased, leading to repression of Gata4 and Gata6 transcription. Our results reveal a critical role for de-SUMOylation in the regulation of PcG target gene expression.

Figure 1. SENP2−/− embryos have a defect in myocardial development

(A) Appearance of SENP2+/+ and SENP2−/− embryos at E10.5. The SENP2−/− embryo has a smaller heart with pericardial effusion. (B) H& E-stained sections of heart from E10.5 wildtype (a, c) and SENP2−/− (b, d) embryos. The SENP2−/− embryos showed hypocellular endocardial cushions and myocardial hypoplasia with thinner myocardium. (C) TUNEL staining of sections of hearts from SENP2 +/+ or −/− embryos at E10.5 showed no significantly increased in apoptosis in the mutant embryos (P>0.01). “n” indicates the number of sections examined. (D) Phosphohistone H3 (PH3) staining of sections of hearts from SENP2 +/+ or −/− embryos at E10.5 revealed reduced proliferation in the mutant embryos (P<0.01). “n” indicates the number of sections examined.

Figure 2. Reduced expression of Gata4, Gata6, and Mef2c in SENP2−/− embryos

(A) RT–PCR analysis of genes that are involved in cardiac development in E10.5 SENP2+/+ or −/− embryos. Samples were amplified for 32, 34, and 36 cycles for Hand1; 30, 32, and 34 cycles for Hand2 and Isl1; 28, 30, and 32 cycles for Gata4, Gata6, Nkx2.5, and Srf; 26, 28, and 30 cycles for Mef2c. β-Actin mRNA levels (amplified for 18, 21, and 24 cycles) were measured as control. (B) SENP2+/+ or −/− embryos at E10.5 days were stained by whole mount in situ hybridization for the indicated transcripts.

Figure 3. SENP2 is essential for expression of PcG target genes

(A) Quantitative expression level of PcG target genes in wildtype and SENP2−/− embryos. Transcript levels were measured by real-time PCR, normalized to a wildtype control and depicted as a fold change between SENP2−/− and wild-type embryos (set as 1, red dotted line). Error bars are based on the standard deviation derived from triplicate PCR reactions in three independent experiments. (B) Quantitative expression level of PcG target genes in SENP2+/+ or SENP2−/− MEF cells, or SENP2−/− MEF cells infected with retroviral vector containing SENP2 or SENP2 catalytic mutant (SENP2m). Transcripts of indicated genes were measured by real-time PCR, normalized to a SENP2+/+ control (set as 1). Error bars are based on the standard deviation derived from triplicate PCR reactions in three independent experiments. (C) Silencing of SENP2 down-regulates expression of PcG target genes. The transcripts of PcG target gene were analyzed by real-time PCR in 293 cells transfected with si-NS or si-SENP2. The mRNA level is shown in means±s.d. of three independent transfection experiments.

Figure 4. SENP2 plays a crucial role in the regulation of PRC1 occupancy on the promoter of PcG target genes

(A) Mutation of SENP2 gene enhances binding of PRC1 to the promoter of Gata4 and Gata6. Eight antibodies (anti-IgG, -H3K27me3, -CBX2, -Pc2, -Ring1b, -Bmi1, -Ezh2, and -Suz12) were used in the qChIP assays in SENP2+/+ or −/− MEF cells. Data are shown in means±s.d. of three independent experiments. (B) Over-expression of SENP2, but not SENP2 catalytic mutant, reduces Pc2 recruitment in SENP2−/− MEF cells. Two antibodies (anti-Pc2, and -H3K27me3) were used in the qChIP assays in SENP2−/− MEF cells infected with empty retroviral vector, retroviral vector containing Flag-SENP2, or Flag-SENP2 catalytic mutant (SENP2m). Data are shown in means±s.d. of three independent transfection experiments. (C) Pc2 and histone 3 (H3) occupancy at the Hoxa2 promoter were analyzed by a qChIP assay in the 293 cells transfected with siRNA against SENPs (si-SENP) or non-specific control (si-NS) as indicated. Data are shown in means±s.d. of three independent transfection experiments. (D) Silencing of SENP2 reduces the expression of Hoxa2. The expression of Hoxa2 was analyzed by real-time PCR in the cells described in (C). The mRNA level is shown in means±s.d. of three independent transfection experiments. (E) SENP2 binds to PRC1. Flag-SENP2 was immunoprecipitated with anti-Flag antibody or IgG from chromatin fraction of Flag-SENP2-transfected 293 cells. The precipitates were detected with anti-Flag, -Pc2, -Bmi1, or -Ring1b antibodies.

Figure 5. SENP2 de-conjugates SUMOylated Pc2

(A) SENP2 de-SUMOylates Pc2 in vivo. COS-1 cells were transfected with indicated plasmids. The immunoprecipitates with anti-Flag (IP) from transfected cell lysates were detected by immunoblotting with anti-HA and anti-Flag (IB). Whole-cell lysates (WCL) were immunoblotted (IB) with anti-RGS antibody. (B) SUMOylated Pc2 accumulated in SENP2−/−, not SENP2+/+ nor SENP1−/− MEF cells. Chromatin fractions from 5×10⁷ MEF cells were treated with micrococcal nuclease, and then immunoprecipitated with Pc2 and SUMO1 antibodies. Bound proteins were detected by immunoblotting with anti-Pc2 or anti-SUMO1 (IB).

Figure 6. SUMOylation increases binding affinity of Pc2 to H3K27me3

(A) Mutation of SUMOylation site decreases the ability of Pc2 to occupy the promoter of Gata4. Two antibodies (anti-Flag and -H3) were used in the qChIP assays using SENP2+/+ or −/− MEF cells transfected with indicated plasmids. Data are shown in means±s.d. of three independent transfection experiments. (B) The immunoprecipitates with Pc2 (IP) from chromatin fraction of SENP2+/+ (WT), SENP2−/−, or SENP1−/− MEF cells were detected by immunoblotting with anti-Pc2 and anti-H3K27me3, H3K27me2, H3K27me1, and H3K9me3 (IB). Chromatin fractions were immunoblotted (IB) with anti-H3 as input. (C) SUMOylated Pc2 binds to tri-methylated H3K27, but not H3K9, with higher affinity than un-conjugated Pc2. Biotinylated histone H3 peptides that were either unmodified or tri-methylated on K27 (upper panel) or on K9 (bottom panel), were incubated with the chromatin fractions from SENP2+/+ (WT) or SENP2−/− (Mut) MEFs in the presence of streptavidin-conjugated sepharose beads. The precipitates were detected by immunoblotting (IB) with anti-Pc2 or anti-SUMO1. (D) SUMOylation facilitates binding of Pc2 chromodomain to tri-methylated H3K27. Biotinylated histone H3 peptides that were tri-methylated on K27 (mK27), were incubated with GST-SUMO1, GST-Pc2(1-100), GST-SUMO1-fused Pc2(1-100), GST-Pc2(409-531), and SUMOylated GST-Pc2(409-531) recombinant proteins in the presence of streptavidin-conjugated sepharose beads. Inputs and precipitates by mK27 peptides as detected with anti-GST were shown in the left panel. The precipitates by mK27peptidea as detected with anti-SUMO1 were shown in the middle panel. The relative binding affinity was shown as relative fold change of signal density of precipitates detected by anti-GST and standardized with input (Right panel). “*” indicates a non-specific band. “” indicates a degraded band. (E) A model depicting the role of SENP2 in the regulation of PRC1 recruitment to H3K27me3 through controlling the SUMOylation status of Pc2. E3 ligase for Pc2 may be Pc2 itself or an undefined E3 ligase.