RETRACTED: Polycomb repressive complex 2 binds and stabilizes NANOG to suppress differentiation-related genes to promote self-renewal

•NANOG-PRC2 inhibited OXPHOS to generate TICs and refractory to therapy•ARID1A gene loss promoted sorafenib-resistant and self-renewal through PRC2 complex•ARID1A mutation promoted HCC in humanized liver of FRG mice•A therapy approach was established to overcome the chemoresistance of TICs The synergistic effect of alcohol and HCV mediated through TLR4 signaling transactivates NANOG, a pluripotency transcription factor important for the stemness of tumor-initiating stem-like cells (TICs). NANOG together with the PRC2 complex suppresses expression of oxidative phosphorylation (OXPHOS) genes to generate TICs. The phosphodegron sequence PEST domain of NANOG binds EED to stabilize NANOG protein by blocking E3 ligase recruitment and proteasome-dependent degradation, while the tryptophan-rich domain of NANOG binds EZH2 and SUZ12. Human ARID1A gene loss results in the resistance to combined FAO and PRC2 inhibition therapies due to reduction of mitochondrial ROS levels. CRISPR-Cas9-mediated ARID1A knockout and/or constitutively active CTNNB1 driver mutations promoted tumor development in humanized FRG HCC mouse models, in which use of an interface inhibitor antagonizing PRC2-NANOG binding and/or FAO inhibitor blocked tumor growth. Together, the PRC2-NANOG interaction becomes a new drug target for HCC via inducing differentiation-related genes, destabilizing NANOG protein, and suppressing NANOG activity. The synergistic effect of alcohol and HCV mediated through TLR4 signaling transactivates NANOG, a pluripotency transcription factor important for the stemness of tumor-initiating stem-like cells (TICs). NANOG together with the PRC2 complex suppresses expression of oxidative phosphorylation (OXPHOS) genes to generate TICs. The phosphodegron sequence PEST domain of NANOG binds EED to stabilize NANOG protein by blocking E3 ligase recruitment and proteasome-dependent degradation, while the tryptophan-rich domain of NANOG binds EZH2 and SUZ12. Human ARID1A gene loss results in the resistance to combined FAO and PRC2 inhibition therapies due to reduction of mitochondrial ROS levels. CRISPR-Cas9-mediated ARID1A knockout and/or constitutively active CTNNB1 driver mutations promoted tumor development in humanized FRG HCC mouse models, in which use of an interface inhibitor antagonizing PRC2-NANOG binding and/or FAO inhibitor blocked tumor growth. Together, the PRC2-NANOG interaction becomes a new drug target for HCC via inducing differentiation-related genes, destabilizing NANOG protein, and suppressing NANOG activity. Alcoholism and hepatitis C virus (HCV) infection are major risk factors for hepatocellular carcinoma (HCC) which lacks any available mechanism-based therapy.1Goessling W. Deciphering hepatocellular carcinoma: from bench to bedside and back.Gastroenterology. 2009; 137: 786-788https://doi.org/10.1053/j.gastro.2009.07.033Abstract Full Text Full Text PDF PubMed Scopus (3) Google Scholar It is the second most deadly cancer in the world (5-year survival rate is 2%–21%)2Barbara L. Benzi G. Gaiani S. Fusconi F. Zironi G. Siringo S. Rigamonti A. Barbara C. Grigioni W. Mazziotti A. et al.Natural history of small untreated hepatocellular carcinoma in cirrhosis: a multivariate analysis of prognostic factors of tumor growth rate and patient survival.Hepatology. 1992; 16: 132-137Crossref PubMed Scopus (380) Google Scholar and the most rapidly escalating cause of cancer mortality (24,550 deaths and 33,660 new HCC cases for 2014) in the US.3Kanwal F. Hoang T. Kramer J.R. Asch S.M. Goetz M.B. Zeringue A. Richardson P. El-Serag H.B. Increasing prevalence of HCC and cirrhosis in patients with chronic hepatitis C virus infection.Gastroenterology. 2011; 140: 1182-1188.e1https://doi.org/10.1053/j.gastro.2010.12.032Abstract Full Text Full Text PDF PubMed Scopus (333) Google Scholar Deaths due to metastatic HCC continue to mount worldwide (660,000 deaths/year); thus, HCC remains a largely incurable malignancy. Alcoholism increases gut permeability, leading to endotoxemia and the activation of Toll-like receptors (TLRs),4Hritz I. Mandrekar P. Velayudham A. Catalano D. Dolganiuc A. Kodys K. Kurt-Jones E. Szabo G. The critical role of toll-like receptor (TLR) 4 in alcoholic liver disease is independent of the common TLR adapter MyD88.Hepatology. 2008; 48: 1224-1231Crossref PubMed Scopus (334) Google Scholar which induce the inflammatory response resulting in development of alcohol-related liver disease.5Kharbanda K.K. Todero S.L. King A.L. Osna N.A. McVicker B.L. Tuma D.J. Wisecarver J.L. Bailey S.M. Betaine treatment attenuates chronic ethanol-induced hepatic steatosis and alterations to the mitochondrial respiratory chain proteome.Int. J. Hepatol. 2012; 2012962183https://doi.org/10.1155/2012/962183Crossref Google Scholar,6Filiano A.N. Millender-Swain T. Johnson Jr., R. Young M.E. Gamble K.L. Bailey S.M. Chronic ethanol consumption disrupts the core molecular clock and diurnal rhythms of metabolic genes in the liver without affecting the suprachiasmatic nucleus.PLoS One. 2013; 8e71684https://doi.org/10.1371/journal.pone.0071684Crossref PubMed Scopus (66) Google Scholar Alcoholism and HCV synergistically increase the risk of HCC.7Yuan J.M. Govindarajan S. Arakawa K. Yu M.C. Synergism of alcohol, diabetes, and viral hepatitis on the risk of hepatocellular carcinoma in blacks and whites in the.Cancer. 2004; 101: 1009-1017Crossref PubMed Scopus (250) Google Scholar Thus, understanding the mechanisms of HCV- and alcohol-induced hepatocarcinogenesis is crucial for the improvement of therapeutic modalities.8Crippin J.S. McCashland T. Terrault N. Sheiner P. Charlton M.R. A pilot study of the tolerability and efficacy of antiviral therapy in hepatitis C virus-infected patients awaiting liver transplantation.Liver Transplant. 2002; 8: 350-355Crossref PubMed Scopus (311) Google Scholar Sorafenib is a tyrosine kinase inhibitor used to treat patients with HCC. More specifically, sorafenib has been shown to be more effective for patients with HCV (+) than patients with HCV (−).9Llovet J.M. Di Bisceglie A.M. Bruix J. Kramer B.S. Lencioni R. Zhu A.X. Sherman M. Schwartz M. Lotze M. Talwalkar J. et al.Design and endpoints of clinical trials in hepatocellular carcinoma.J. Natl. Cancer Inst. 2008; 100: 698-711https://doi.org/10.1093/jnci/djn134Crossref PubMed Scopus (1453) Google Scholar Therefore, it is crucial to identify patient subpopulations unresponsive to sorafenib so as to justify development of new therapies which solely target tumor-initiating cells (TICs). These therapies should potentially help to reduce morbidity, mortality, and treatment costs since over 40% of HCCs are clonal.10Yao Z. Mishra L. Cancer stem cells and hepatocellular carcinoma.Cancer Biol. Ther. 2009; 8: 1691-1698Crossref PubMed Google Scholar In this study, we developed a mouse model to study human hepatocarcinogenesis. These mice, which exhibit liver-specific expression of the HCV NS5A protein,11Majumder M. Ghosh A.K. Steele R. Zhou X.Y. Phillips N.J. Ray R. Ray R.B. Hepatitis C virus NS5A protein impairs TNF-mediated hepatic apoptosis, but not by an anti-FAS antibody, in transgenic mice.Virology. 2002; 294: 94-105https://doi.org/10.1006/viro.2001.1309Crossref PubMed Scopus (105) Google Scholar when fed alcohol and high-cholesterol high-fat diet for 12 months, develop HCC.12Chen C.L. Tsukamoto H. Liu J.C. Kashiwabara C. Feldman D. Sher L. Dooley S. French S.W. Mishra L. Petrovic L. et al.Reciprocal regulation by TLR4 and TGF-beta in tumor-initiating stem-like cells.J. Clin. Invest. 2013; 123: 2832-2849https://doi.org/10.1172/JCI65859Crossref PubMed Scopus (132) Google Scholar By studying this mouse model, we reported several important findings in prior publications. (i) HCV NS5A induces the expression of TLR4 in hepatocytes, leading to Tlr4-mediated expression of Nanog (a transcription factor that is crucial for stemness)12Chen C.L. Tsukamoto H. Liu J.C. Kashiwabara C. Feldman D. Sher L. Dooley S. French S.W. Mishra L. Petrovic L. et al.Reciprocal regulation by TLR4 and TGF-beta in tumor-initiating stem-like cells.J. Clin. Invest. 2013; 123: 2832-2849https://doi.org/10.1172/JCI65859Crossref PubMed Scopus (132) Google Scholar (ii) The HCV-TLR4-Nanog axis enhances the genesis of TICs and hepatocarcinogenesis with chemoresistance. (iii) Nanog is upregulated in TICs as shown in three different HCC mouse models. (iv) NANOG reduces mitochondrial oxidative phosphorylation (OXPHOS), reactive oxygen species (ROS) production, and activated fatty acid oxidation (FAO) to support the self-renewal and drug resistance properties of TICs, leading to HCC relapse and metastasis. Previous ChIP-seq analysis of TICs using anti-Nanog antibody revealed its enrichment on genes associated with OXPHOS (i.e., Cox6a2 and Cox15) and FAO (i.e., Acadvl).13Jiang J. Chan Y.S. Loh Y.H. Cai J. Tong G.Q. Lim C.A. Robson P. Zhong S. Ng H.H. A core Klf circuitry regulates self-renewal of embryonic stem cells.Nat. Cell Biol. 2008; 10: 353-360Crossref PubMed Scopus (607) Google Scholar The enhanced production of mitochondrial ROS by Nanog silencing supported the functionality of Nanog on these genes. Approximately 39% of HCV-alcohol-associated HCCs are associated with mutations in ARID1A, a component of the chromatin remodeling complex SWI/SNF. Our genome-wide meta-analysis of HCCs showed increases in components of polycomb repressive complex 2 (PRC2), together with existing ARID1A mutations were implicated in hepatomas. We compared alcohol-associated human HCCs and alcohol-fed, HCV NS5A transgenic (Tg) mice for HCC development and identified NANOG as a core stem cell factor in both systems. The appearance of NANOG followed TLR4 activation as a crucial component needed for genesis and maintenance of TICs. These findings support our hypothesis that alcohol/HCV-mediated NANOG and PRC2 inductions and ARID1A mutations cooperatively generate chemoresistant TICs in alcohol/HCV-associated HCC via inhibition of OXPHOS and activation of FAO. In this study, we tested how NANOG activated by HCV and alcohol interacts with PRC2 to suppress OXPHOS genes for the generation of TICs in HCV-associated human HCCs. To identify candidate genes required for tumor growth, we screened human genes in liver progenitor cells (LPCs) using a genome-wide CRISPR-Cas9 knockout (GeCKO) library to identify genes that are needed for tumor growth. This approach used a GeCKO library of 64,751 single-guide RNAs (sgRNAs) combined in a lentiviral Cas9-vector library. Human LPCs used in this study were transduced with the GeCKO library followed by HCV infection and engrafted into immunodeficient FRG mice. These mice were chronically fed ethanol for 6 months. Tumors were then isolated, and we performed RNA and exome-sequencing for both RNA and DNA, respectively (Figure 1A). The analysis of resulting sgRNA distribution revealed differences between alcohol-treated and vehicle-treated mouse tumors. This screen revealed loss-of-function mutations in ARID1A and COX6A2, which are targets of NANOG (Figure 1B), are essential for HCC development. The most frequently mutated oncogenic driver genes in human alcohol-associated HCCs (not in dysplastic macro-nodules) are alleles of ARID1A, a component of the chromatin remodeling complex that regulates gene transcription (8%–38% of HCC), CTNNB1 (involved in β-Catenin/Wnt pathway), and TP53 (30%–65% of HCC) (Figure 1C).14Guichard C. Amaddeo G. Imbeaud S. Ladeiro Y. Pelletier L. Maad I.B. Calderaro J. Bioulac-Sage P. Letexier M. Degos F. et al.Integrated analysis of somatic mutations and focal copy-number changes identifies key genes and pathways in hepatocellular carcinoma.Nat. Genet. 2012; 44: 694-698https://doi.org/10.1038/ng.2256Crossref PubMed Scopus (1060) Google Scholar Decreased expression of ARID1A or loss-of-function mutations in ARID1A are associated with chemoresistance, tumor progression, metastasis, and reduced overall survival in mice and humans.15Nhieu J.T. Renard C.A. Wei Y. Cherqui D. Zafrani E.S. Buendia M.A. Nuclear accumulation of mutated beta-catenin in hepatocellular carcinoma is associated with increased cell proliferation.Am. J. Pathol. 1999; 155: 703-710Abstract Full Text Full Text PDF PubMed Google Scholar,16He F. Li J. Xu J. Zhang S. Xu Y. Zhao W. Yin Z. Wang X. Decreased expression of ARID1A associates with poor prognosis and promotes metastases of hepatocellular carcinoma.J. Exp. Clin. Cancer Res. 2015; 34: 47https://doi.org/10.1186/s13046-015-0164-3Crossref PubMed Scopus (69) Google Scholar Bioinformatics analyses demonstrated that ARID1A mutations were recurrent mutations in HCV/ethanol-associated HCCs; by contrast, in HCV-associated HCC, ARID1A was less frequently mutated and instead CTNNB1 was more frequently mutated (Figure 1C). Integrated comparisons of GEO datasets for HepG2 cells, a human hepatoblastoma cell line with wild-type (wt) ARID1A against NANOG ChIP-seq data, showed OXPHOS genes were co-identified in NANOG-bound regions as well as in ARID1A and in SNF (components of SWI/SNF complex) ChIP-seq datasets (Figure 1D). OXPHOS genes were targets of both NANOG and PRC2 as per ChIP-seq analysis of TICs (mutant ARID1A); wt ARID1A was present as a component SNF of SWI/SNIF complexes, but was also associated with NANOG targets in ChIP-seq analyses of HCC: The reference genome was compared for possible co-enrichment of NANOG and ARID1A on the same genes of HepG2 cells that express wt ARID1A. Peak-calling analyses demonstrated that the SWI/SNF complex (including ARID1A) and NANOG were together co-enriched on OXPHOS genes in this cell line (Figure 1D). We previously demonstrated that HCV and alcohol synergistically activate TLR4 to induce the expression of Nanog.17Chen C.L. Uthaya Kumar D.B. Punj V. Xu J. Sher L. Tahara S.M. Hess S. Machida K. NANOG metabolically reprograms tumor-initiating stem-like cells through tumorigenic changes in oxidative phosphorylation and fatty acid metabolism.Cell Metabol. 2016; 23: 206-219https://doi.org/10.1016/j.cmet.2015.12.004Abstract Full Text Full Text PDF PubMed Scopus (240) Google Scholar Additionally, EED, EZH2, and SUZ12 genes (PRC2 component) were induced under these conditions (Figure 1E) allowing for NANOG and PRC2 components (i.e., EED, EZH2, and SUZ12) to be present on the promoters of OXPHOS genes. Similarly, wt ARID1A expression showed SWI/SNF-NANOG co-enrichment on OXPHOS genes. Notably, a gene set enrichment analysis of our NANOG ChIP-seq and RNA profiling data of mouse liver TICs isolated from alcohol-associated HCC with ARID1A mutations displayed increased NANOG signals with components of PRC, viz., EZH2 (enzymatic subunit responsible for histone H3K27 methylation),18Simon J.A. Kingston R.E. Mechanisms of polycomb gene silencing: knowns and unknowns.Nat. Rev. Mol. Cell Biol. 2009; 10: 697-708https://doi.org/10.1038/nrm2763Crossref PubMed Scopus (1104) Google Scholar as well as the other core subunits of PRC (EED and SUZ12) in NANOG-bound regions (Figure 1E). To test if NANOG together with the PRC2 complex inhibits OXPHOS to generate TICs in the presence of ARID1A mutations, the candidate genes EZH2, SUZ12, and EED were silenced in patient-derived TICs (with or without ARID1A mutations) by Piggyback lentivirus carrying both Cas9 and sgRNA. Surviving cells were subjected to the spheroid formation assay on ultra-low attachment plates for analysis of their self-renewal ability. ARID1A-mutant HCC cell lines (PLC/PRF/5 and Hep3B and HCC-LM6) and wt ARID1A HCC cell lines (as representatives of lower sorafenib IC50: HepG2, Huh7, and HCC-97L) were compared for the expression of NANOG target genes (i.e., COX6A2 and COX15) (Figure 1F). These were compared to ARID1A-mutant HCC cells (Hep3B, PLC/PRF/5, and HCC-LM6) which are more resistant to sorafenib than wt ARID1A HCC cells (Table S1). Accordingly, we tested if NANOG together with the PRC2 complex inhibits OXPHOS to generate TICs in the presence of ARID1A mutations. As ARID1A-mutant HCC cells (Hep3B, PLC/PRF/5, and HCC-LM6) are more resistant to sorafenib than wt ARID1A HCC cells, the outcome of such a comparison is germane to our model. To corroborate if NANOG together with the PRC2 complex inhibits OXPHOS to generate TICs in the presence of ARID1A mutations, candidate genes, EZH2, SUZ12, and EED (identified by the GeCKO screen), were individually silenced by the Piggyback lentivirus approach in patient-derived TICs that were isolated in our previous studies (Figure 1G).17Chen C.L. Uthaya Kumar D.B. Punj V. Xu J. Sher L. Tahara S.M. Hess S. Machida K. NANOG metabolically reprograms tumor-initiating stem-like cells through tumorigenic changes in oxidative phosphorylation and fatty acid metabolism.Cell Metabol. 2016; 23: 206-219https://doi.org/10.1016/j.cmet.2015.12.004Abstract Full Text Full Text PDF PubMed Scopus (240) Google Scholar As a mimic for previously described endotoxin effects, RT-qPCR analysis of LPS-treated TICs resulted in higher NANOG mRNA expression compared to primary hepatocytes. Knockdown of ARID1A further amplified the LPS-induced NANOG expression in TICs (Figure 1H). These results indicated that loss of ARID1A function might have inactivated SWI/SNF complex leading to augmented NANOG expression in TICs. Thus, the role of endotoxin in vivo is expected to increase NANOG expression via ARID1A-mediated augmentation of SWI/SNF. Based on these findings, we hypothesized that cancer-promoting mutations (i.e., ARID1A) increase PRC2 complex activity (including EZH2) and induce an HCV/alcohol-mediated stem cell program (slow growing cells), leading to TIC-initiated HCC development (Figure 1I). In The Cancer Genome Atlas (TCGA) data analyses, expression of HNF4A was correlated with ARID1A but inversely correlated with PRC2 components, EZH2, EED, and AEBP2 in HCC patients (Figure 1J). From these results, we postulate that the well-balanced chromatin remodeling activities between SWI/SNF and PRC2 complexes are disrupted upon mutation of ARID1A, leading to functional loss of SWI/SNF activity, thus enhancing EZH2 (PRC2) activity. The enhanced PRC2 activity promotes alcohol and HCV-induced stem cell reprogramming in liver TICs by elevation of NANOG and repression of OXPHOS gene expression (Figure 1K). We next aimed to determine how alcohol/HCV-induced NANOG interacts with PRC2 to suppress OXPHOS genes for the generation of slow-cycling, chemoresistant TICs. We evaluated this further by examining NANOG for direct binding to PRC2, thus leading to inhibition of OXPHOS genes in the context of ARID1A deficiency. Second, we examined if NANOG cooperated with specific genetic alterations to generate TICs for HCC development. Third, we investigated if ARID1A mutations with NANOG induction promoted tumorigenesis in vivo. To examine if NANOG interacts with PRC2 components EZH2 and SUZ12 in vitro, we performed co-immunoprecipitation. We first generated various expression constructs including full-length protein coding sequences and deletion mutants of human NANOG containing N-terminal Flag epitope tag as well as versions for full-length hEZH2 and hSUZ12 containing N-terminal Myc-epitope tags (Figure 2A left panel). Cell lysates prepared after co-transfections of HEK293T with Nanog deletion mutants and either EZH2 (Figure 2B) or SUZ12 (Figure 2C) plasmids were immunoprecipitated (IP) with anti-Myc antibody. The eluted immunocomplexes were analyzed by SDS-PAGE and immunoblotted (IB) with either anti-Flag or anti-Myc antibodies. The analysis of IP Myc and IB Flag experiments showed Nanog interacted with EZH2 and SUZ12 through carboxyl terminal domains (Figure 2A middle and right panel, respectively). Similarly, cell lysates from HEK293T cotransfections with full-length NANOG and either EZH2 (Figure 2B) or SUZ12 (Figure 2C) deletion mutants were immunoprecipitated with anti-Flag antibody. The immunocomplex eluates were analyzed by SDS-PAGE and immunoblotted with anti-Flag or anti-Myc antibodies. The result of IP Flag and IB Myc experiments showed EZH2 interacted with NANOG through all domains (Figure 2B) and SUZ12 interacted with NANOG through its amino terminal domain (Figure 2C). To further examine if NANOG interacts with PRC2 components EZH2 and SUZ12 in a manner independent of DNA binding of PRC2 complexes, portions of HEK293T cell lysates were further treated with DNase I, sonicated, and immunoprecipitated with anti-Myc antibody after cotransfection with Nanog deletion mutants and full-length SUZ12. IP-western blots showed that DNase I treatment and sonication did not abrogate the binding of NANOG to SUZ12, indicating the interaction between NANOG and or other PRC2 components (i.e., EZH2 and SUZ12) was not mediated by their DNA binding (Figure 2D). Reciprocal IP-Western blot analysis further confirmed the interactions by NANOG-EZH2 and NANOG-SUZ12 (Figures 2E–2G). Domain mapping studies using deletion mutants of NANOG were performed for binding to PRC subunits. HEK 293T cells were cotransfected with the indicated Myc-tagged NANOG deletion constructs and Flag-tagged EED full-length constructs (Figure 3A). Immunoprecipitation of protein complexes in cell lysates was followed by immunoblotting to examine possible protein interactions. These results revealed that EED interacted with NANOG through N-terminal domains (aa1-aa94) and the C1-W region (aa155-aa240) (Figures 3A and 3B). By contrast, other PRC2 components EZH2 and SUZ12 only bound NANOG through C1-W domains. This is surprising difference in the interacting domain, suggests EED may bind NANOG independently of the PRC2 complex formation. Reciprocal IP-Western blot analyses confirmed interactions between NANOG and EED (Figure 3C). To test the functional relationship of the NANOG pathway to EED, we knocked down EED. TICs expressed abundant NANOG (Figure 3D) while primary hepatocytes do not express NANOG (data not shown). EED knockdown (KD) reduced NANOG protein levels (Figure 3D). The PRC2 component EED was knocked down in TICs and their RNA was examined by RT-qPCR analyses. Silencing of EED did not alter NANOG mRNA levels while silencing of EED induced COX6A2 expression (Figure 3E). EED silencing reduced SOX2 and OCT4 mRNA levels, indicating that NANOG destabilization reduced stemness genes, including OCT4 and SOX2 whose promoters have NANOG binding sites for transactivation. Spheroid colony formation assay demonstrated that EED expression transformed p53-deficient hepatoblasts and this effect was potentiated by NANOG (Figure 3F, Top). HNF4A transcription was unaffected by NANOG but repressed by EED (Figure 3F, Bottom). To elucidate a possible post-transcriptional effect of EED on NANOG protein turnover, we added cycloheximide to block protein synthesis in Huh7 cells treated with either scrambled short hairpin RNA (shRNA) or specific shRNA to silence EED. Cell lysates were obtained at different time points and probed with anti-NANOG antibody (Figure 3G). NANOG protein levels were reduced upon knockdown of EED while this silencing had no effect on overall β-actin stability (Figure 3G upper). EED silencing reduced NANOG t1/2 to 84 min compared to the scrambled control which showed a t1/2 of 152 min. These results indicated that EED is responsible for increasing the turnover rate of NANOG and the interaction between EED and the NANOG pathway supports the oncogenic activity via NANOG stabilization. As hyaluronan-CD44 interactions activate protein kinase C ε to phosphorylate Nanog at T200 for its nuclear translocation,19Bourguignon L.Y.W. Spevak C.C. Wong G. Xia W. Gilad E. Hyaluronan-CD44 interaction with protein kinase C(epsilon) promotes oncogenic signaling by the stem cell marker Nanog and the Production of microRNA-21, leading to down-regulation of the tumor suppressor protein PDCD4, anti-apoptosis, and chemotherapy resistance in breast tumor cells.J. Biol. Chem. 2009; 284: 26533-26546https://doi.org/10.1074/jbc.M109.027466Abstract Full Text Full Text PDF PubMed Scopus (275) Google Scholar a series of IP-western analyses validated that NANOG interacted with PRC2 complexes leading to its increased phosphorylation at T200 of NANOG proteins (Figure 3H). As NANOG N-terminus (aa1-aa94) contains a phosphodegron sequence (PEST domain enriched with P, E, S, or T amino acid) and binds EED, we hypothesized that EED-NANOG interaction stabilizes NANOG in TICs (Figure 3I). We examined whether EED-NANOG association stabilized NANOG by blocking ubiquitination-mediated degradation. For this, EED was knocked down by shRNA lentivirus transduction to eliminate its association with NANOG. To confirm this at the protein level, we transfected Huh7 cells with Flag-tagged NANOG expression vector in EED knockdown Huh7 cells and sh-scrambled (control) Huh7 cells. At 24 h post-transfection, cells were treated with 10 μM MG132 to inhibit proteosome activity. Whole cell lysates were collected 48 h post-transfection and analyzed by western blot. The NANOG protein level was observed to decrease upon EED knockdown compared to sh-scrambled and this effect was reversed by MG132 treatment; however, under this experimental condition, EZH2 level declined in the presence of sh-EED but not after sh-scrambled treatment (Figure 3J). Interestingly in the reciprocal experiment, both EED and EZH2 protein levels were both significantly increased by MG132 treatment in both sh-EED and shscrambled groups (Figure 3K). These results suggested EED knockdown caused NANOG to be degraded at a higher rate via the proteasome degradation pathway and that both NANOG and PRC2 subunits were tightly regulated by the same mechanism. Finally, we examined other effects of EED on NANOG stabilization by analyzing the levels of NANOG in TICs with or without EED KD. EED KD led to depletion of NANOG as expected, suggesting that EED stabilizes NANOG, possibly by blocking both phosphorylation and ubiquitination of NANOG. In fact, E3 ubiquitin ligase FBXW8 is known to bind the PEST domain of NANOG to ubiquitinate and subsequently degrade NANOG. Indeed, FBXW8 proteins bind NANOG (Figure 3K). Furthermore, knockdown of EED in Huh7 cells, enhanced NANOG-target OXPHOS component COX6A2 expression (Figure 3L), suggesting that EED cooperated with NANOG to suppress OXPHOS transcription. The sum of these observations indicated to us that EED may antagonize NANOG PEST degradation by binding/hiding the same domain of the latter. To further decipher whether EED affects NANOG at the mRNA or post-transcription level, we examined the effect of EED knockdown effect on NANOG and its downstream mRNA targets level by RT-qPCR. sh-EED and sh-scrambled Huh7 cells were cultured for 2 days and treated with 10 μM MG132 for 24 h. The cells were lysed, and total RNA was collected for cDNA synthesis and subsequent quantitative PCR analysis. Changes due to sh-EED or sh-scrambled had no effect on NANOG mRNA levels; however, the expression of cytochrome c oxidase subunit 6A 2 (COX6A2) increased significantly with EED knockdown. Moreover, this effect was reversed upon treatment with MG132 (Figure 3L). Because NANOG normally downregulates expression of COX6A2, our finding corroborated that NANOG was degraded post-transcriptionally in the EED knockdown group by the proteosome degradation pathway. To further confirm the EED-NANOG cooperation in a more physiologic context, PIL420Jellicoe M.M. Nichols S.J. Callus B.A. Baker M.V. Barnard P.J. Berners-Price S.J. Whelan J. Yeoh G.C. Filipovska A. Bioenergetic differences selectively sensitize tumorigenic liver progenitor cells to a new gold(I) compound.Carcinogenesis. 2008; 29: 1124-1133Crossref PubMed Scopus (73) Google Scholar hepatoblasts with or without lentivirus-based NANOG expression and/or EED KD were orthotopically transplanted into the left lobe of NSG mice for tumorigenesis ability (Figure 3M). PIL4 cells with NANOG expression formed large tumors in two months but EED KD significantly but incompletely reduced this growth (Figure 3M). EED KD reduced tumor growth initiated by TICs transplanted subcutaneously in NSG mice; this effect could be partially reversed exogenously by NANOG expression (Figure 3M, orange lines). Conversely, NANOG expression alone showed significant tumor growth, which was attenuated by EED knockout to a level below the growth achieved by control TICs (Figures 3M and 3N). These results indicated that EED KD had additional antitumor activity besides antagonizing the NANOG tumor promoter effect(s) on downstream genes. We sought to identify potential inhibitors of the interaction(s) between NANOG and EED. For screening of small-molecule inhibitors, we employed FITC-tagged NANOG peptides with PEST domains or N-terminus and C-terminus of NANOG tryptophan-rich domains (W domain) (see Figure 3O), which were wild type or Y > A or F > A mutants as binding probes. Binding of these components to recombinant EED was monitored by fluorescence polarization assays. These two proteins interact via tryptophan (W) or phenylalanine (F) residues of binding pockets in NANOG PEST domain which reportedly bind tightly to target proteins based on the amino acid composition of ligand-protein complex: 37.4% F6, 23.5% Y6, 24.7% H5, 5.5% W5, and 8.9% W6 (6- or 5-member rings of ligands) analyzed by use of TOUGH-D1 dataset.21Brylinski M. Aromatic interactions at the ligand-protein interface: implications for the development of docking scoring functions.Chem. Biol. Drug Des. 2018; 91: 380-390https://doi.org/10.1111/cbdd.13084Crossref PubMed Scopus (54) Google Scholar A search for compounds which antagonized binding of NANOG and EED involved testing both an NCI drug library of 1200 compounds and 630 compounds of an FDA-approved small-molecule library in multiple binding reactions using a 384 well plate format between FITC-tagged NANOGpeptides (containing PEST domains) and EED (Figure 3O). High-ranking competitor ligands were identified (50 compounds) from the FDA library; an additional 50