The Hippo pathway regulates cell proliferation and organ size through control of the transcriptional regulators YAP (yes-associated protein) and TAZ. Upon extracellular stimuli such as cell–cell contact, the pathway negatively regulates YAP through cytoplasmic sequestration. Under conditions of low cell density, YAP is nuclear and associates with enhancer regions and gene promoters. YAP is mainly described as a transcriptional activator of genes involved in cell proliferation and survival. Using a genome-wide approach, we show here that, in addition to its known function as a transcriptional activator, YAP functions as a transcriptional repressor by interacting with the multifunctional transcription factor Yin Yang 1 (YY1) and Polycomb repressive complex member enhancer of zeste homologue 2 (EZH2). YAP colocalized with YY1 and EZH2 on the genome to transcriptionally repress a broad network of genes mediating a host of cellular functions, including repression of the cell-cycle kinase inhibitor p27, whose role is to functionally promote contact inhibition. This work unveils a broad and underappreciated aspect of YAP nuclear function as a transcriptional repressor and highlights how loss of contact inhibition in cancer is mediated in part through YAP repressive function.
This study provides new insights into YAP as a broad transcriptional repressor of key regulators of the cell cycle, in turn influencing contact inhibition and tumorigenesis.
The Hippo–YAP pathway is a central regulator of cell fate and proliferation and is tightly regulated by mechanical cues such as tension, pressure, and contact with the extracellular matrix and other cells (1). At the core of the pathway are the transcriptional coregulators yes-associated protein (YAP) and TAZ, which bind to gene promoters and enhancers through interaction with transcription factors such as the TEA-domain proteins (TEAD) and others (2, 3). YAP localization depends on cellular density, where under low cell density conditions YAP localizes to the nucleus and modulates the transcription of genes involved in cell growth and survival (4). Increased YAP activity and nuclear localization is commonly observed in a multitude of cancers including schwannoma and cancers of the liver, colon, ovarian, lung, and prostate (5, 6).
YAP has previously been shown to repress the expression of mesendoderm lineage–specific genes in human embryonic stem cells (7). In addition, YAP facilitates the recruitment of the NuRD complex to deacetylate histones and repress the expression of target genes (8). To explore the role of YAP as a transcriptional regulator, we investigated the genomic localization of YAP at low cell density in human Schwann cells. These were chosen due to the critical role YAP plays in promotion of cellular transformation and tumorigenesis, subsequent to loss of the NF2 tumor suppressor gene, which is an upstream effector of the Hippo pathway (9–12). These efforts led to the identification of a transcriptional repressor function for YAP, through interaction with the multifunctional transcription factor Yin Yang 1 (YY1) and enhancer of zeste homologue 2 (EZH2), a member of the Polycomb repressive complex 2 (PRC2). This work unveils a broad and underappreciated aspect of YAP nuclear function and highlights how loss of contact inhibition in cancer is partly mediated through YAP's repressive function.
Materials and Methods
Human Schwann cells
Human Schwann cells (hSC2λ) cells were obtained from the laboratory of Dr. Margaret Wallace (Department of Molecular Genetics & Microbiology, University of Florida, Gainesville, FL; ref. 13). The cells were authenticated by short tandem repeat DNA Profiling (DDC Medical). Cells were maintained in low glucose DMEM (Gibco) supplemented with 10% FBS (Atlas Biologicals) and antibiotics (100 U/mL penicillin and 100 μg/mL streptomycin; Gibco), at 37°C in a humidified atmosphere of 5% CO2 (v/v). Cells were tested every 3 months for Mycoplasma and confirmed free of contamination.
Transfections were performed using an Amaxa Nucleofector with the Amaxa Cell Line Nucleofector Kit V. Lentiviral infection of hSC2λ was performed according to standard protocols. Briefly, lentivirus was prepared in HEK293T cells that were cotransfected with packaging plasmids VSVG, Δ8.2, and GIPZ YY1 short hairpin RNA (shRNA) gene set. Supernatant was collected 48 and 72 hours after transfection, and cells were infected with 6 mL of viral supernatant containing polybrene (8 μg/mL). After 48 hours, transduced cells were selected with puromycin (0.25 μg/mL) and this selection was maintained for 72 hours.
Plasmids and siRNA/shRNA
The pCMV-Flag-YAP-5SA (#27371), pCMV-Flag-YAP-S127A (#27370), and pCellFree_G03 YY1 (#67082) expression plasmids were purchased from Addgene. The siGENOME Human YY1 (7528) siRNA set (MU-011796-02-0002) was purchased from GE Healthcare Dharmacon and its siGENOME nontargeting control (D-001210-01-05). The siYAP Flexitube siRNA (1027418) was purchased from Qiagen. The “AllStars” negative control siRNA (SI03650318) was purchased from Qiagen. Expression plasmids for YY1 shRNA (V2LHS_219592, V2LHS_172065, and V3LHS_412955) were purchased from GE Healthcare Dharmacon. All siRNAs were used at a final concentration of 20 nmol/L.
The following antibodies were used in this study: Anti-YAP antibody ChIP (Abcam, ab52771) and Anti-YY1 antibody-ChIP Grade (Abcam, ab38422); Anti-EZH2 ChIP (Active motif, #39901); Anti-Lamin A/C [sc-6215 (N-18), 1:500]; Anti-tubulin (T5168) anti-actin (A4700) and anti-vinculin (V4505) were purchased from Sigma; Anti-YAP Cell Signaling (#4912); Anti-p27 Kip1 antibody (Cell Signaling Technology, 2552S); Anti-Flag (Sigma, F1804-200UG); Anti-His (Abcam, #18184); and Anti-TAZ (D3I6D) (Cell Signaling Technology, 70148).
This was carried out as described previously (14). Briefly, cells were incubated on ice for 20 minutes in ice-cold cytoplasmic buffer (20 mmol/L Tris-HCl pH 7.4, 150 mmol/L KCl, 1.5 mmol/L MgCl2, 1 mmol/L PMSF, 1 mmol/L DTT, 0.5% Nonidet P-40, and protease inhibitor mixture), centrifuged at 4,000 × g for 5 minutes at 4°C. Supernatant was kept for the cytosolic and plasma membrane fractions. The pellet was washed 5× with nuclear washing buffer (10 mmol/L HEPES pH 7.9, 10 mmol/L KCl, 1.5 mmol/L MgCl2, 0.34 mol/L sucrose, and complete protease inhibitor mixture), lysed on ice for 20 minutes in 400 μL of RIPA (200 mmol/L NaCl, 50 mmol/L TRIS HCl pH = 7.4, 1 mmol/L EDTA, 1% NP40, and 0.25% DOC) buffer, and centrifuged at 13,000 × g for 10 minutes at 4°C. Supernatant was kept as nuclear fraction. To separate the cytosolic and plasma membrane fractions, the supernatant was spun at 200,000 × g for 30 minutes at 4°C and the resultant supernatant was centrifuged again at 13,000 × g for 5 minutes at 4°C and kept as the cytosolic fraction. The pellet was washed in 500 μL lysis buffer (50 mmol/L Tris pH 7.4, 1 mmol/L EDTA, 2.5 mmol/L MgCl2, 150 mmol/L NaCl, and complete protease inhibitor mixture) and resedimented at 200,000 × g for 30 minutes at 4°C. The pellet was then resuspended in ice-cold IP buffer [1 mol/L Tris-HCl pH 7.4, 4 mol/L NaCl, 10% (w/v) Triton X-100, and complete protease inhibitor mixture], incubated on a rocker for 30 minutes at 4°C, and cleared by centrifugation at 13,000 × g for 15 minutes at 4°C. The supernatant was kept as the plasma membrane fraction.
Chromatin immunoprecipitation and qRT-PCR analysis
A total of 20 × 106 cells were fixed with 1% formaldehyde for 10 minutes at room temperature. Fixation was halted with 125 mmol/L glycine for 5 minutes at room temperature. Fixed cells were washed 2× with cold PBS. Cell pellets were then resuspended in chromatin immunoprecipitation (ChIP) lysis buffer and chromatin was sheared with Misonix S-3000 bath sonicator for 15-minute sonication at 280 W, 30″ ON and 30″ OFF to obtain 0.3–0.5 kb DNA fragments. Antibody (5 μg) and Dynabeads Protein A were added to the cell lysate and incubated overnight at 4°C. Beads were washed with buffer 1 (150 mmol/L NaCL, 20 mmol/L TrisCl pH 8.0, 5 mmol/L EDTA, 65% w/v sucrose, 10% Triton-X-100, and 20% SDS) and then washed with TE buffer. DNA was eluted by resuspending the beads in TE/1% SDS. ChIP DNA and Input were treated with RNase A (5 μg) for 1 hour at 37°C. Proteinase K (0.5 mg/mL) was added and incubated overnight at 65°C to reverse cross-linking. DNA was then purified in phenol:chloroform and resuspended in a 30 μL of elution buffer. DNA was used for real time-PCR using SYBR Green PCR kit. A standard dilution curve was obtained for each Input and 1 μL of ChIP DNA was used in each PCR reaction. Melt curves were analyzed to confirm specificity of the amplified target.
Fly RNA extraction
RNA from cells was extracted using the Qiagen RNeasy Kit. RNA from Drosophila heads was extracted using the RNAaqueous-micro total RNA isolation kit. cDNA was made using the SuperScript III Kit (Life Technologies). qRT-PCR was performed with SYBR Green (Applied Biosystems). Relative gene expression was calculated with the 2−ΔΔCt method. Primer sequences' detail is provided in Supplementary Table S1.
ChIP-seq library preparation and sequencing
The ChIP-seq libraries were prepared using the NEBNext Ultra II DNA Library Prep Kit for Illumina (catalog no. E7645S) from New England Biolabs Inc. Fragmented DNA (1 ng) was used as input. The following steps were followed for library preparation: end repair, 5′ phosphorylation, dA-tailing, adapter ligation, U excision, cleanup of adaptor-ligated DNA without size selection, and library amplification. Twelve to 13 cycles followed by clean-up with 1 X Agencourt AMPure XP. The quality of the libraries was assessed using the Agilent High Sensitivity DNA Kit (catalog no. 5067-4626) on the Agilent 2100 Bioanalyzer (Agilent Technologies) and were quantified using the NEBNext Library Quant Kit for Illumina (catalog no. E7630S) by New England Biolabs Inc. The libraries were sequenced using the NextSeq 500 High Output v2 Kit (75 cycles; catalog no. FC-404-2005) on the NextSeq 500 Platform from Illumina. Integrative Genomics Viewer was used to generate browser tracks (15).
CDKN1B promoter region (from −80 to 960 relative to the transcription start site) was cloned in place of the PGK promoter into the pmirGLO- Dual-Luciferase plasmid (E1330 Promega). hSC2λ cells were transfected and seeded in 96-well plates. For the TEAD and YY1 mutations, the binding sites were, respectively, mutated from CCAT to TGTA for YY1 and from CTTC to AGGA for TEAD. On the following day, luciferase activity was measured with Dual Luciferase Assay System (Promega; #E1910) according to the manufacturer's instructions. Firefly luciferase signal was normalized with the Renilla luciferase signal.
Hydrodynamic tail vein injection
DNA (10 μg/mL) was mixed in saline solution (10% volume/body weight; ref. 2). The tail was placed under a heat lamp for 20 seconds to increase the vascular volume and using 27-gauge needles, the DNA solution was injected at approximately 20 mL/minute.
The Pho RNAi strain (#110466) and its background control (#60100) were obtained from the Vienna Drosophila RNAi Center. The RNAi line is predicted to have no off targets (RNAi-phiC31 construct and insertion data submitted by the Vienna Drosophila RNAi Center). The UAS driven yorkieS168A (BDSC 28818 UAS-yki-S168A) was obtained from the Bloomington Drosophila Stock Center. The lines were recombined with GMR-Gal4 to mediate expression in the eye (16).
CRISPR cell line
The PX459 plasmid was purchased from Addgene (#62988). Guide (TCCGGACCCGGGCAACCG) targeting the first exon of YAP was cloned into PX459. hSC2λ cells were transfected with the plasmid and treated with puromycin (0.25 μg/mL) for 48 hours. Single clones were selected, expanded, and subjected to Western blot analysis. Cleavage was assessed by DNA sequencing.
Proximity ligation assay
Proximity ligation assay (PLA) was carried out using the Duolink In Situ Red Start Kit Mouse/Rabbit (Sigma, DUO92101) per the manufacturer's instructions. Cells were fixed in 4% paraformaldehyde for 20′ at room temperature, washed 3 × 5′ in PBS, room temperature. Cells were then permeabilized in 0.3% triton-X for 3′ and blocked in blocking buffer for 2 hours at 37°C. Primary antibodies (YAP = CST-12395, YY1 = CST-63227, and EZH2 = CST-5246) were used at 1:600. Images were taken on an Olympus FV3000RS confocal scanning microscope. DAPI was visualized at 405 nm, and Duolink red visualized at 594 nm.
ChIP-seq data analysis
Adapter sequences and low quality read ends were trimmed using cutadapt v1.8.1 (17). For processing the ChIP-seq samples, the reads were aligned to the hg19 genome build using bowtie2 v2.2.9 (18). Post-alignment filtering was performed according to the AQUAS pipeline (github.com/kundajelab/chipseq_pipeline). Peaks were called using macs2 v18.104.22.16860309 using the shift size values calculated from the run_spp.R script from the SPP peak caller (19, 20). High-quality peaks were then identified using the idr1 pipeline, selecting peaks with an idr score of less than or equal to 0.02. HOMER v4.9 was used for peak annotation, gene ontology (GO) analysis, motif identification, and binding heatmaps (21).
All animal experiments complied with NIH guidelines and were approved by The Scripps Research Institutional Animal Care and Use Committee (IACUC). NOD/SCID mice (6–8 weeks old) were used for the hydrodynamic tail vein injection. Luciferase-labeled SC4 cells were resuspended in PBS at a concentration of 5 × 105 cells per mL. Six-week-old NSG mice (Jackson Laboratory Stock #005557) were anesthetized with isoflurane. A total of 2.5 × 105 cells were injected subcutaneously in the right hind flank. Tumor growth was monitored using IVIS 200, drug treatment was started when tumors reached a threshold value. Mice were treated with EPZ005687 at dose of 10 mg/kg/day. EPZ005687 was resuspended in DMSO at 2 mg/mL. Control animals received DMSO without drug. EPZ005687 was injected daily via intraperitoneal injection. Mice were sacrificed after 17 days of treatment (once control animals had tumor size that required euthanasia according to IACUC-approved protocol). Tumors were isolated, weighed at the conclusion, as described previously (22).
Statistical analysis of data was performed using GraphPad Prism (version 6). Individual statistical methods are described in the respective figure legends. Unpaired Student t test was calculated to determine the significance of the results and indicate the two-tailed P values. Unless otherwise noted, mean and SD was used to assess the significance.
Data and software availability
ChIP-seq data are stored in the Gene Expression Omnibus under accession number GSE112932.
YAP binds at the promoter of p27
To understand the transcriptional network regulated by YAP, we employed human Schwann cells (hSC2λ), grown at low density, when YAP is nuclear (Supplementary Fig. S1A). We performed ChIP followed by next-generation sequencing (ChIP-seq) using an anti-YAP antibody and identified a total of 7,019 peaks, including peaks at the promoters of previously identified YAP targets genes (CTGF, TEAD1, AMOTL2, and ANKRD1; Fig. 1A; Supplementary Fig. S1B; Supplementary Table S2). YAP peaks were located mainly in intron and intergenic regions and, to a lesser extent, at gene promoters (7.72%), which is in accordance with previously reported ChIP-seq data (23). Motif analysis at YAP peaks reveals that TEAD-binding consensus sequences are enriched in 57.91% of YAP peaks (Fig. 1B; Supplementary Table S3), in proximity to the summit of peaks (Fig. 1C). GO analysis indicated that YAP is involved in structure morphogenesis, cell communication, and signaling (Supplementary Table S4). Among the peaks, we identified YAP binding to the promoter of the CDKN1B gene (Fig. 1D), which codes for the cyclin-dependent kinase inhibitor (CDKI) p27 (p27Kip1). Analysis of YAP ChIP-seq datasets in a variety of cell lines shows enrichment of YAP within the promoter or enhancer region of p27 (23–25). We confirmed YAP binding by ChIP-RT-PCR and observed a significant enrichment of YAP on the CDKN1B promoter compared with an IgG control (Fig. 1E; Supplementary Fig. S1C). We also observed enrichment of TAZ at the promoter of CDKN1B (Supplementary Fig. S1D). Previous studies indicate p27 mediates contact inhibition of cell proliferation by binding to cyclin–CDK complexes and preventing cells from progressing into S-phase (26, 27). Therefore, we set out to further investigate the mechanism by which YAP regulates p27.
Regulation of p27 expression by cell density is YAP dependent
The mRNA and protein levels of p27 are reported to be elevated with increased cell density and cell–cell contact (28, 29), which we confirmed by qRT-PCR and Western blot analysis in hSC2λ cells (Fig. 2A). We also confirmed that the YAP target gene CTGF is upregulated at low cell density (Fig. 2A). To assess the role of YAP in regulating p27, we employed hSC2λ cells in which YAP was inactivated by CRISPR-mediated editing (hSC2λYAP−/−) or knocked down by siRNA (hSC2λ-siYAP) with an efficiency of more than 80% (Supplementary Fig. S2A and S2B). In hSC2λYAP−/− cells grown at low density, we observed a 2.5-fold upregulation of p27 mRNA (Fig. 2B). Similarly, knockdown of YAP by siRNA led to an upregulation of p27 at both the protein and mRNA levels (Fig. 2C). Conversely, to assess whether forced expression at high cell density of a constitutively active YAP would downregulate endogenous p27, we transfected hSC2λ cells with vector expressing a constitutively active YAP (YAP5SA) cDNA (30). As expected, we observed upregulation of the known YAP target gene CTGF and downregulation of p27 at high cell density, both at the mRNA and protein levels (Fig. 2D). We also assessed the effect of TAZ overexpression by transfecting hSC2λ cells with vector expressing a constitutively active TAZ (TAZS89A) and similarly observed downregulation of p27 at the mRNA and protein levels (Supplementary Fig. S2C).
To assess whether YAP represses p27 expression in vivo, we introduced expression vectors for wild-type (WT) YAP, constitutively active YAP (YAP5SA and YAPS127A), or control plasmid into the liver of mice by hydrodynamic tail vein injection. The levels of p27 were assessed 2 weeks later by qRT-PCR and a significant downregulation of p27 mRNA was observed in the presence of WT or activated YAP (Supplementary Fig. S2D). We confirmed the expression of the exogenous YAP qRT-PCR using primers against the FLAG-tag (Supplementary Fig. S2E). In another approach, we implanted hSC2λ cells that stably expressed either a YAP-5SA or WT YAP into the sciatic nerve of NOD/SCID mice and allowed tumors to develop. Once tumors were detected they were harvested and analyzed for the expression of p27. As expected, levels of p27 proteins were significantly reduced in the YAP-5SA tumors compared with the YAP WT control (Fig. 2E). Finally, as another in vivo validation, we used the fruit fly, Drosophila melanogaster (D. melanogaster), in which the core of the Hippo pathway is highly conserved (31–33). The expression of a constitutively active form of yorkie (ykiS168A), the fly homologue of YAP, in the eye using a GMR eye-specific driver, led to a significant reduction in dacapo (fly homologue of p27) mRNA levels (Fig. 2F). Overall, these results indicate that YAP regulates p27 at the transcriptional level, in cultured cells and in vivo.
YY1 is required for YAP-mediated repression of p27
To identify the mechanisms mediating p27 repression by YAP, we analyzed motif distribution at the promoter of p27 and identified a YY1-binding motif adjacent to the TEAD-binding motif (Supplementary Fig. S3A). YY1 is a multifunctional zinc-finger transcription factor of the Polycomb Group protein (PcG) family that bind promoters and enhancers of various cellular and viral genes (34, 35). YY1 plays a key role in growth and differentiation and can act both as an activator and as a repressor, depending on its binding partners (34, 35). To assess whether YAP repression of p27 requires YY1, we generated a stable knockdown of YY1 in hSC2λ cells using shRNA (hSC2λshYY1). We confirmed the knockdown of YY1 by Western blot analysis (Supplementary Fig. S3B). In hSC2λshYY1 cells, the overexpression of activated YAP failed to repress p27 both at the transcriptional and protein levels (Fig. 3A; Supplementary S3C). Similarly, transient knockdown of YY1 using small interfering RNA (siRNA) showed similar results (Supplementary Fig. S3D). This holds true for TAZ, where overexpression in cells transiently knockdown for YY1 using siRNA showed similar results (Supplementary Fig. S3E). To test whether YY1 mediates the repressive function of YAP on p27 in vivo, we employed the GMR-YkiS168A flies and crossed them to flies that carry a dsRNA construct KK110466 targeting pho (the fly homologue of YY1; ref. 36) under conditional UAS control. This cross results in the expression of YorkieS168A in combination with downregulation of pho, specifically in the eye. As expected, the GMR-YkiS168A line shows downregulation of dacapo. While the knockdown of pho alone does not affect dacapo expression and result in a normal eye phenotype, dacapo levels were significantly rescued in the YkiS168A/pho RNAi line (Fig. 3B; Supplementary Fig. S3F). This provides in vivo support for YY1 being required, at least in part, for YAP-mediated repression.
YY1 was shown to be required for the recruitment of EZH2 to chromatin in the vicinity of muscle-specific genes (37). EZH2 is the catalytic component of the PRC2 complex and silences gene expression through trimethylation of histone H3 lysine 27 (H3K27me3; ref. 38). We performed ChIP-RT-PCR for EZH2 and H3K27me3 in hSC2λ cells and observed enrichment at the p27 promoter at low cell densities (Fig. 3C). This enrichment was significantly reduced in hSC2λ YAP−/− cells grown at low density, suggesting that EZH2 recruitment requires YAP (Fig. 3C). To test whether YAP and YY1 can cooccupy the p27 promoter, we carried out a sequential ChIP-RT-PCR in 293T cells in which YAP is knocked out. Cells were transfected with either FLAG-YAP5SA or a combination of 6X(HIS)-YY1 and FLAG-YAP5SA and immunoprecipitated for YAP5SA (FLAG) alone or YAP5SA (FLAG) followed by YY1 (HIS). We observed enrichment for YY1 specifically in the presence of YAP5SA at p27 promoter (Fig. 3D). To test whether YY1 is required for YAP-mediated repression of p27, we cloned the promoter and first exon of p27 into a reporter plasmid upstream of a firefly luciferase cassette. Upon YAP5SA expression, we observed a decrease in luciferase signal compared with the control. However, when mutating the YY1 DNA–binding site (CCAT→TGTA) in the cloned promoter sequence, we no longer observed this decrease in signal (Fig. 3E). Similarly, mutation of the TEAD-binding site significantly diminished the ability of YAP5SA to repress luciferase activity (Fig. 3E). Because YAP knockdown leads to upregulation of p27, we sought to assess how downregulating YY1 and EZH2 would affect p27 levels. We performed a knockdown of YY1 and EZH2 using siRNA and observed upregulation of p27 at the transcriptional and protein levels, suggesting that in addition to YAP, YY1 and EZH2 are both required for p27-mediated repression (Fig. 3F). Finally, to determine whether YY1 is required for YAP recruitment to the p27 promoter, we assessed the enrichment for YAP at the promoter by ChIP-RT-PCR in the presence or absence of YY1. This analysis indicates that YY1 is required for the binding of YAP to the p27 promoter (Fig. 3G).
To assess whether cells overexpressing YAP exhibit loss of contact inhibition, we performed cell-counting assays of hSC2λ cells stably expressing the activated YAP-5SA allele and observed that the YAP-5SA–expressing cells have a growth advantage and continue to proliferate after the control-expressing cells have reached a plateau (Supplementary Fig. S3G). Moreover, cell-cycle analysis of the cells at high densities demonstrates a higher fraction of YAP5SA-expressing cells in S-phase, compared with the control cells (Supplementary Fig. S3H). In addition, we observed enrichment of YY1, YAP, and EZH2 at the promoter of CDKN1B in the YAP-5SA cell line, at high cell density (Supplementary Fig. S3I). These findings suggest that forced expression of YAP at high cell density can lead to the recruitment of YY1 and EZH2 to the CDKN1B promoter and override contact inhibition due to high cell density.
Genomic colocalization of YAP, YY1, and EZH2
To further understand YY1/YAP/EZH2 genome-wide distribution in human Schwann cells, we performed additional ChIP-seq experiments for YY1 and EZH2. We observed that all three factors colocalize at the p27 promoter (Fig. 4A). To assess whether YY1, YAP, and EZH2 can form a complex, we performed immunoprecipitation studies looking at endogenous levels of expression or in cells transfected with expression plasmids for all three proteins. These studies demonstrate that YY1/YAP/EZH2 can be coimmunoprecipitated using antibodies against any of the three, when overexpressed (Fig. 4B and C) and at endogenous levels of expression (Fig. 4D; Supplementary Fig. S4A). To further validate the interaction and close proximity between YAP, YY1, and EZH2, we performed a PLA in nontransfected hSC2λ cells. This analysis demonstrates the colocalization and close proximity of YAP-YY1 and YAP-EZH2 in the nuclei of these cells (Fig. 4E).
Across the genome, we observed a significant colocalization of YAP, YY1, and EZH2 (Fig. 5A). The overall number of peaks identified for each mark were YAP = 12,480, YY1 = 9,455, and EZH2 = 6,928. Pairwise affinity analysis of the common peaks shared between YAP and YY1 reveal a total of 8,560 (64%). Of these shared YAP-YY1 peaks, 6,813 (80%) were shared with EZH2 (Fig. 5B). Motif analysis for triple-bound peaks revealed enrichment for TEAD1 and YY1 motif at bound sites across datasets (Fig. 5C). When comparing the predicted motifs enrichment in YAP peaks versus YAP/YY1/EZH2 peaks, there is no clear enrichment for binding motifs for known transcriptional activators in the YAP peaks alone (Supplementary Tables S3 and S5). In addition, read densities for YAP and YY1 chromatin–binding show a strong overlap (Fig. 5D). From the triple-bound peaks, we identified several cyclin-dependent kinase inhibitors (p57, p15, and p21) as potential targets for YAP-mediated repression (Supplementary Table S6). All of these targets have a YY1 motif adjacent to a TEAD motif at their respective promoters (Supplementary Table S7). Overexpression of YAP in hSC2λ cells led to a significant repression of these targets (Fig. 5E, top). The overexpression of TAZ in hSC2λ cells led similar results (Supplementary Fig. S5A). Significantly, in the context of downregulated YY1, the overexpression of YAP failed to repress mRNAs levels, suggesting that YY1 is required for YAP-mediated repression of these targets (Fig. 5E, bottom). In addition, we observed a loss of H3K27me3 marks at the promoter of these genes upon YAP knockout (Fig. 5F). We analyzed H3K27me3 marks at the promoters of additional genes bound by YAP/YY1/EZH2 and three of nine genes analyzed displayed a YAP-dependent H3K27me3 expression profile, suggesting that additional mechanisms can mediate the repression at other loci (Supplementary Fig. S5B). One possibility, given the similar repressive role for TAZ, is that TAZ may mediate the repressive function at these loci. These results suggest that YAP-mediated transcriptional repression, in conjunction with YY1, plays a broad and significant role in regulation of the cell cycle. A Kyoto Encyclopedia of Genes and Genomes pathway analysis on YAP-bound genes further supports this by identifying cell-cycle–associated genes as significantly enriched in the list of differentially bound promoters (Supplementary Fig. S5C).
The Hippo pathway has been shown to control cell proliferation and organ size through regulation of the transcriptional regulator YAP. Disruption of the Hippo pathway leads to YAP nuclear localization and is associated with loss of contact inhibition and increased proliferation and tumorigenesis (39). However, the precise molecular mechanisms by which YAP mediates contact inhibition remain unknown. YAP's role as a transcriptional coactivator has been widely studied through its ability to activate the transcription of several genes including connective tissue growth factor (CTGF), epidermal growth factor receptor ligand amphiregulin (AREG), and ankyrin repeat domain 1 (ANKRD1; refs. 40–42). YAP's role as an activator may not fully account for the overgrowth phenotype observed in cancer, suggesting that it possesses additional nuclear functions. Indeed, a limited number of studies have shown YAP can associate with the NuRD complex and transcriptionally repress differentiation markers to maintain pluripotency and repress tumor suppressor genes to promote cell growth and survival in MCF10A cells (7, 8). Our study identifies a previously unknown mechanism by which YAP can act as a transcriptional repressor through association with YY1 and the PRC2 complex and the addition of repressive H3K27me3 marks.
Our initial studies focused on CDKN1B, which codes for p27, and is a major regulator of the cell cycle that functions to prevent cells from entering S-phase (43) and is a regulator of Schwann cell differentiation (44, 45). In addition, several studies report that p27 expression is downregulated in vestibular schwannomas (46, 47). We find that YAP represses p27 through the recruitment of YY1 and the PRC2 complex (see model in Fig. 6). In addition, we show that loss of YY1 in vivo can rescue YAP-mediated repression of p27. Loss of contact inhibition of proliferation is one of the main hallmarks of cancer and is associated with decreased level of p27 and increased cell proliferation (48). Our findings that expression of active YAP can overcome contact inhibition in human Schwann cells and repress p27 expression identify a potential molecular link between the disrupted Hippo pathway, amplification of YAP in cancer, and loss of contact inhibition.
In addition to p27, our analysis identified additional targets of YAP/YY1-mediated repression. Because YAP can regulate thousands of targets, as an activator or repressor, we believe it is unlikely that the effects of YAP are mediated through a single target, or perhaps a single gene family. Rather, it is likely that the effects are mediated through hosts of genes, both up- and downregulated. To dissect the role of these different YAP target genes will likely require tools that isolate YAP's effects on individual or groups of selected targets. Further studies are thus required to elucidate the extent of YAP's activating versus repressive functions.
Previous reports indicate that the deacetylation of H3K27 by the NuRD complex facilitates the recruitment of the PRC2 complex for gene repression (49, 50), this suggests that YAP may first associate with the NuRD complex, which can then lead to recruitment of the PRC2 complex and gene repression. Further studies are needed to determine whether such mechanisms are at play in the regulation of CDKN1B and the other CDKIs by YAP and YY1. Undoubtedly, the function of YAP in mediating cell growth is broad and involves multiple mechanisms and downstream targets. Our findings suggest that transcriptional repression represents a major mechanism through which YAP mediates transcriptional regulation of critical cellular behaviors.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.
Conception and design: S. Hoxha, A. Shepard, J.L. Kissil
Development of methodology: S. Hoxha, A. Shepard, M. Janiszewska, R.M. Witwicki, W.W. Ja, J.L. Kissil
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Hoxha, A. Shepard, S. Troutman, J.R. Doherty, M. Janiszewska, R.M. Witwicki, W.W. Ja
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S. Hoxha, A. Shepard, H. Diao, M.E. Pipkin, W.W. Ja, M.S. Kareta, J.L. Kissil
Writing, review, and/or revision of the manuscript: S. Hoxha, A. Shepard, M. Janiszewska, W.W. Ja, J.L. Kissil
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S. Hoxha, A. Shepard, M.E. Pipkin, J.L. Kissil
Study supervision: S. Hoxha, J.L. Kissil
We thank Dr. Ursula Ehmer (Roman-Herzog-Krebszentrum- Comprehensive Cancer Center) for sharing vectors used in the tail vein injections. Library preparation and sequencing were performed at The Scripps Research Institute Florida Genomics Core. The work was supported by grants R01NS077952 (NINDS/NIH) and R01CA124495 (NCI/NIH to J.L. Kissil) and R01AG045036 (to W.W. Ja from the NIA/NIH). M.S. Kareta was supported by an Institutional Development Award (IDeA) from the National Institute of General Medical Sciences of the NIH under grant number P20GM103620. M.E. Pipkin was supported by R01AI095634 (NIAID/NIH) and the Frenchman's Creek Women for Cancer Research.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.