YY1 DNA binding and interaction with YAF2 is essential for Polycomb recruitment

Abstract

Polycomb Group (PcG) proteins are crucial for epigenetic inheritance of cell identity and are functionally conserved from Drosophila to humans. PcG proteins regulate expression of homeotic genes and are essential for axial body patterning during development. Earlier we showed that transcription factor YY1 functions as a PcG protein. YY1 also physically interacts with YAF2, a homolog of RYBP. Here we characterize the mechanism and physiologic relevance of this interaction. We found phenotypic and biochemical correction of dRYBP mutant flies by mouse YAF2 demonstrating functional conservation across species. Further biochemical analysis revealed that YAF2 bridges interaction between YY1 and the PRC1 complex. ChIP assays in HeLa cells showed that YAF2 is responsible for PcG recruitment to DNA, which is mediated by YY1 DNA binding. Knock-down of YY1 abrogated PcG recruitment, which was not compensated by exogenous YAF2 demonstrating that YY1 DNA binding is a priori necessary for Polycomb assembly on chromatin. Finally, we found that although YAF2 and RYBP regulate a similar number of Polycomb target genes, there are very few genes that are regulated by both implying functional distinction between the two proteins. We present a model of YAF2-dependent and independent PcG DNA recruitment by YY1.

Figure 1.

REPO domain silencing of BGUZ, a PcG-dependent reporter, requires a wild-type dRYBP background. (A) Maps of transgenic constructs and expected results. The bxd enhancer, GAL4 UAS-binding sites, and Ubx promoter sequences are shown driving the lacZ gene. The hunchback promoter driven GAL4DBD-REPO expressing transgene is shown below. Expected fly embryo LacZ staining patterns controlled by hbGAL4DBD-REPO in a wild-type or dRYBP¹ mutant background are indicated on the right. (B) dRYBP is required for PcG repression in Drosophila. The panels depict representative Drosophila embryos stained for expression of LacZ using X-Gal. The crossing strategy for each embryo is depicted on the left with male and female genotypes indicated. The top panel represents BGUZ, the transgenic reporter gene where there is uniform staining of the embryos. The second panel represents the cross between hbGAL4DBD-REPO flies on the BGUZ background and shows anterior repression by the absence of LacZ stain. The third panel represents embryos that are crossed between hbGAL4DBD-REPO and the dRYBP¹ mutant flies where the absence of dRYBP results in loss of anterior repression. Additional controls crossed are shown in panels 4–6. hbGAL4DBD-REPO indicates hunchback-driven expression of GAL4DBD-REPO (GAL4DBD with YY1 201–226), dRYBP¹ designates the mutant dRYBP allele. Anterior is oriented toward the left, ventral is oriented down.

Figure 2.

Mouse YAF2 can substitute for dRYBP. (A) CLUSTALW comparison of dRYBP and mYAF2 proteins highlighting the conserved residues suggestive of functional conservation of the two proteins across species. The residues are numbered according to the mouse YAF2 sequence. (B) Genotyping PCR for the wild-type dRYBP allele (WT), dRYBP¹ allele (m) and transgenic hsp-FlagYAF2 expression construct in Drosophila strains. The marker lanes are reproduced for each panel for size reference of PCR products. Genotypes and phenotypes are located to the right of each panel. The fertility/viability and the observed wing phenotype of the flies are indicated. (C) ChIP assays were performed at the endogenous ultrabithorax PRE_D locus using the antibodies shown on the x-axis. The PRE_D sequence lies between Drosophila melanogaster genomic sites 12 589 500 to 12 590 299 (version 3.1). Enrichment in wild-type rosy⁵⁰⁶ flies is indicated by white bars, whereas the dotted bar shows a reduction in the enrichment of Pc and H3K27me3 in the absence of dRYBP in the mutant flies. This loss of PcG binding is recovered upon overexpression of mYAF2 by heatshock induction of the hspFlagYAF2 transgene in the mutant flies as shown by the black bars. The error bars indicate the standard deviation of three independent replicates of quantitative PCR analysis. Fly strains: Rosy (ry⁵⁰⁶, wild-type), white; dRYBP¹/CyO (dRYBP¹ mutant), gray; dRYBP¹/CyO; hsp-FlagYAF2 (corrected mutant), black. Single and double asterisks denote P < 0.05 and P < 0.001, respectively.

Figure 3.

YY1 interacts with YAF2. (A–C) Co-IP of full-length YAF2 and YY1 deletion constructs were performed with nuclear extracts from HEK293-T cells transiently co-transfected with the expression plasmids indicated above each lane. Upper panels show direct western blot analysis of nuclear extracts as input samples for both GAL4DBD and Flag-tagged proteins. The lower panels show co-IP proteins with the indicated antibodies and detection of the interacting partner by western blot analysis using the indicated antibody (WB). (D) Diagram of YY1 mutants and summary of interaction with YAF2.

Figure 4.

The C-terminal region of YAF2 interacts with YY1 and RING proteins and can silence a PcG reporter gene in vivo. (A) Co-IP of full-length YY1 and YAF2 deletion constructs was performed with nuclear extracts from HEK293-T cells transiently co-transfected with the expression plasmids indicated above each lane. The upper panels depict direct western blot analysis of nuclear extracts as input samples for GAL4DBD and Flag-tagged proteins. The lower panels shows co-IP proteins with the indicated antibodies followed by western blot analysis using the indicated antibody (WB). (B) Yeast two hybrid detection of YAF2, RYBP and YAF2 deletion proteins with RING1 and RING2 proteins. GAL4 DBD vector expression constructs (BK) and GAL4 AD vector expression constructs (AD) were cotransformed into S. cerevisiae strain AH109 and grown on Trp/Leu/His/Ade drop out medium. (C) Representative Drosophila embryos stained for expression of LacZ using X-Gal. The crossing strategy for each embryo is depicted on the left with male and female genotypes indicated. BGUZ is the transgenic reporter gene. Hb-GAL4YAF2(1–101) and hb-GAL4YAF2(102–179) indicates hunchback-driven expression of GAL4YAF2 deletion proteins YAF2(1–101) and YAF2(102–179), respectively. The middle panel shows the absence of anterior repression by hb-GAL4YAF2(1–101) whereas the hb-GAL4YAF2(102–179) and hb-GAL4YAF2 wild-type embryos show repression demonstrating that the C-terminal region of YAF2 is involved in PcG repression in vivo. Anterior is oriented toward the left, ventral is oriented down.

Figure 5.

YAF2 acts as a bridge protein between YY1 and the RING proteins. (A) Co-IP of full-length YAF2 or YY1 and full-length RING1 and RING2 proteins were performed with nuclear extracts from HEK293-T cells transiently cotransfected with the expression plasmids indicated above each lane. The upper panels depict direct western blot analysis of nuclear extracts as input samples for GAL4DBD and Flag-tagged proteins. The lower panels show that YAF2 co-immunoprecipitates with both RING1 and RING2 but there is no enrichment of the RING proteins with YY1 (Flag IP). (B) Yeast two-hybrid detection of the YY1 REPO domain with YAF2 and RYBP or RING1 and RING2 proteins shows that YY1 REPO domain interacts with RYBP and YAF2 but not with the RING proteins. GAL4 DBD vector expression constructs (BK) and GAL4 AD vector expression constructs (AD) were cotransformed into S. cerevisiae strain AH109 and grown on Trp/Leu/His/Ade drop-out medium. (C) Co-IP of RING1 with YY1 was observed only in the presence of YAF2. Flag-tagged YY1 was co-transfected with Gal4RING1 in the absence or presence of Gal4YAF2. IP was performed with Flag antibody and western blot analysis was performed with GAL4 antibody. The upper two panels represent the input samples and the lower panel represents the IP fraction.

Figure 6.

Drosophila Pc interacts with RING, YAF2 and the YY1 REPO domain. (A) Co-IP assay was performed with nuclear extracts from HEK293-T cells transiently co-transfected with the expression plasmids indicated above each lane. The upper panels depict direct western blot analysis of nuclear extracts as input samples for GAL4DBD and Flag-tagged proteins. The lower panels show that Pc interacts not only with RING1 and RING2 but also YY1 (Flag IP). (B) Co-IP assay was performed with nuclear extracts from HEK293-T cells transiently co-transfected with the expression plasmids indicated above each lane. Both YY1 and YAF2 coimmunoprecipate with Pc, but the interaction of Pc is stronger with YY1 compared with YAF2, as determined by a ternary co-IP. With increasing salt concentration, the interaction between YAF2 and Pc is lost suggesting a lower-affinity interaction. However, the interaction between YY1 and Pc remains intact at high salt. (C) Drosophila Pc interacts with the YY1 REPO domain. pcDNA3–GAL4DBD plasmid was used as the vector control.

Figure 7.

YAF2 is required for PcG recruitment in HeLa cells. (A) Knock-down of YAF2. qRT–PCR of YAF2 transcript levels upon 48 h knock-down of YAF2 using siRNA in HeLa cells is shown on the left. Levels of YAF2 transcripts were normalized to actin. The graph shows the fraction of residual transcript in cells treated with Lipofectamine 2000 only and with YAF2 siRNA when compared with control HeLa cells. YAF2 western blot data of untreated, vehicle control, scrambled siRNA, and YAF2 siRNA treated cells are shown in the middle panel with actin and laminB1 as controls. The right panel shows the expression levels of various Polycomb components upon knock-down of YAF2 for 48 h. Nuclear extracts were made from control and YAF2 siRNA-treated HeLa cells and western blot analysis was performed with antibodies against YY1, EZH2, RYBP and RING1 proteins. Nucleolin (sc-8031, Santa Cruz) was used to normalize for loading errors. (B) ChIP–qPCR analysis to monitor enrichment of PRC1 and PRC2 components at ‘PRE-like’ sequences upon knock-down of YAF2 for 48 h. The open bars represent lipid control HeLa cells and the black bars represent cells where YAF2 has been knocked down. The x-axis indicates the antibodies used for IP. The error bars indicate the standard deviation of three independent replicates of quantitative PCR analysis and the data are represented as percentage of input. (C) ChIP–qPCR analysis at the MYT1D site shows that overexpression of YAF2 results in increased recruitment of PcG proteins at ‘PRE-like’ sites (dotted bar). The x-axis indicates the antibodies used for the IP. The error bars indicate the standard deviation of three independent replicates of quantitative PCR analysis and the data are represented as percentage of input. Single and double asterisks denote P < 0.05 and P < 0.001, respectively.

Figure 8.

YY1 DNA binding is essential for PcG recruitment. (A) Western blot analysis to monitor the levels of YY1 upon knock-down with siRNA. Whole-cell extracts were prepared from HeLa cells transfected with YY1-siRNA after 96 h followed by western blot analysis to detect YY1 levels. The membrane was stripped and reprobed with actin antibody (A 1978, Sigma) to normalize for loading errors. (B) ChIP–qPCR analysis to monitor the enrichment of YY1, PRC1 and PRC2 components at ‘PRE-like’ sequences upon loss of YY1. The open bars represent control HeLa cells transfected with vector control and the black bars represent HeLa cells in which YY1 has been knocked down. Overexpression of YAF2 does not compensate for the loss of YY1 in recruiting Polycomb proteins (dotted bar) indicating that YY1 is essential for PcG binding. The x-axis indicates the antibodies used for IP. The error bars indicate the standard deviation of three independent replicates of quantitative PCR analysis and the data are represented as percentage of input. Asterisks denote P < 0.05. (C) YY1 knock-down results in increased hoxA2 and hoxD13 gene expression. qPCR is shown for actin, YY1 hoxA2 and hoxD13 transcripts in cells treated with vehicle control, scrambled siRNA or YY1 siRNA.

Figure 9.

Effect of YAF2 and RYBP knock-down on Polycomb target gene expression. (A) Hierarchical clustering of gene expression of PcG regulated genes on RT² profiler PCR arrays upon knock-down of YAF2 and RYBP for 48 h and 96 h. Results show only partial overlap of genes regulated by YAF2 and RYBP. Red color represents increased expression and blue color represents decreased expression. (B) Venn diagrams show the overlap of genes regulated upon knock-down of RYBP for 48 h and 96 h and those between RYBP and YAF2.

Figure 10.

Model of PcG recruitment by YY1. The diagram represents proposed mechanisms of PcG recruitment by YY1. In all cases YY1 binds to DNA. YY1 may then interact with YAF2 to recruit the PRC1 complex. Alternatively, in the absence of YAF2, YY1 may recruit the PRC1 complex via interaction with CBX/Pc. Either of these recruitment pathways would lead to ubiquitination of H2A on lysine 119. In addition, our published work showed that YY1 interacts with EZH2 and SUZ12 (PRC2 components) (59) thereby recruiting the PRC2 complex resulting in trimethylation of histone H3 at lysine 27.