Open Full Page

Published OnlineFirst July 3, 2015; DOI: 10.1158/0008-5472.CAN-15-0930 Cancer Tumor and Stem Cell Biology Research

Epigenetic Activation of TWIST1 by MTDH Promotes Cancer Stem–like Cell Traits in Breast Cancer Yajun Liang1,2, Jing Hu1, Jiatao Li1, Yingjie Liu1, Jingyi Yu1, Xueqian Zhuang1, Lili Mu3, Xiangyin Kong1,2, Dengli Hong3, Qifeng Yang4, and Guohong Hu1,2

Abstract

Cancer stem–like cells (CSC) are a cell subpopulation that can mediated by the histone acetyltransferase CBP. Mechanistic inves- reinitiate tumors, resist chemotherapy, and give rise to metastases. tigations showed that MTDH interacts with CBP and prevents its Metadherin (MTDH) contributes widely to tumor growth, drug ubiquitin-mediated degradation, licensing its transcriptional acti- resistance, relapse, and metastasis, but its molecular mechanisms vation of TWIST1. In clinical specimens of breast cancer, MTDH of action are not well understood. Here, we report that MTDH expression correlates positively with TWIST1 expression and CSC drives CSC expansion by promoting the expression of TWIST1,a abundance. Overall, our work revealed that MTDH promotes transcription factor critical for cancer cell stemness and metastasis. CSC accumulation and breast tumorigenicity by regulating MTDH activates TWIST1 expression indirectly by facilitating TWIST1, deepening the understanding of MTDH function in histone H3 acetylation on the TWIST1 promoter, a process cancer. Cancer Res; 75(17); 1–9. 2015 AACR.

Introduction was predominantly localized in the cytoplasm and nucleus rather than cell surface, thus not likely to directly mediate the interaction Tumor cells are highly heterogeneous and cancer stem–like of cancer cell and lung endothelium (7), its function in cancer cells (CSC) arise as the subpopulation that can perpetuate the metastasis, as well as chemoresistance, was revealed again by an growth of malignant cells indefinitely (1). Previous studies dem- independent approach combining bioinformatic screening and onstrated that breast cancer contains CSCs that are responsible for experimental validation (5). Since then, rapidly accumulated data tumor initiation, chemoresistance, and metastasis (2, 3). It is of have documented its multifaceted roles to help tumors grow, biologic and clinical importance to elucidate the mechanisms of evade apoptosis, resist chemo- or radiotherapy, and metastasize CSC maintenance and expansion in breast cancer to identify to distant organs (5, 6, 8–11). Gene amplification of MTDH is effective strategies for cancer treatment. often observed in breast cancer, prostate cancer, colorectal carci- MTDH (also called AEG1 or LYRIC/3D3) is a single-pass noma, and hepatocellular carcinoma, especially in metastatic transmembrane protein with the encoding gene located at chro- tumors (5, 12–14). The expression of MTDH is also aberrantly mosome 8q22, a genomic region that is frequently amplified in elevated in a broad range of cancers (5, 15, 16), and correlates with cancers (4, 5). The link of MTDH with cancer was first uncovered clinical stage, tumor size, lymph node spread, distant metastasis, by a phage cDNA screening for cancer cell proteins that bound to and poor survival (17). However, so far the functional mechanism lung vasculature (6). Although later studies showed that MTDH of MTDH in cancer is not clear. Although studies have reported the regulation of a number of genes and pathways by MTDH (9, – 1Key Laboratory of Stem Cell Biology, Institute of Health Sciences, 18 20), little is known about how MTDH delivers oncogenic Shanghai Institutes for Biological Sciences, Chinese Academy of signals to the downstream molecules and performs the pluor- Sciences and Shanghai Jiao Tong University School of Medicine, ipotent functions in cancer. 2 Shanghai, China. China Collaborative Innovation Center of Systems TWIST1 is a basic helix–loop–helix transcription factor that Biomedicine, Shanghai Jiao Tong University School of Medicine, Shanghai, China. 3Key Laboratory of Cell Differentiation and Apopto- potently drives the EMT process in development and cancer (21). sis of National Ministry of Education, Department of Pathophysiology, Activated TWIST1 upregulates N-cadherin (CDH2), downregu- Shanghai Jiao Tong University School of Medicine, Shanghai, China. lates E-cadherin (CDH1), and leads to morphologic change, 4Department of Breast Surgery, Qilu Hospital of Shandong University, Ji'nan, China. migration, and metastasis of cancer cells. In addition, the acquisition of mesenchymal traits resulted from TWIST1 activation Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/). confers CSC characteristics, including tumorsphere formation, and in vivo tumor initiation, on cancer cells (22, 23). Transcrip- Y. Liang and J. Hu contributed equally to this article. tionally, TWIST1 can be regulated by miRNAs and transcription Corresponding Author: Guohong Hu, Institute of Health Sciences, 320 Yueyang factors such as HMGA2, KLF17, and STAT3 (24–27). Epigenetic Road, Shanghai 200031, China. Phone: 86-21-54923296; Fax: 86-21-54923297; modifications, including DNA and histone methylation, also play E-mail: [email protected] roles in the activation of TWIST1 expression (28, 29). However, doi: 10.1158/0008-5472.CAN-15-0930 the regulation of the histone acetylation status of TWIST1 has not 2015 American Association for Cancer Research. been reported. In addition, although recently MTDH has been

www.aacrjournals.org OF1

Liang et al.

indicated to promote cancer cell EMT (30–33), a direct link of one million cells were incubated with 5 mg/mL Hoechst 33342 MTDH to prominent EMT regulators, such as TWIST1, is still (Sigma) for 90 minutes at 37C. Control samples were treated unknown. In this study, we demonstrate a MTDH–TWIST1 signal with 50 mmol/L verapamil (Sigma) to block the efflux of axis that drives CSC expansion and tumorigenesis of cancer cells. the dye. Mechanistically, the interaction between MTDH and the histone acetyltransferase CBP prevents CBP from degradation, thus facil- Chemoresistance clonogenic assay itating the epigenetic activation of TWIST1 and CSC expansion in For chemoresistance analysis, 2 104 cancer cells were seeded breast cancer as well as prostate cancer. in each well of a 48-well plate. Two days later when the culture grew to approximately 80% confluence, the cells were treated with Materials and Methods 25 nmol/L paclitaxel for 24 hours, and reseeded to a 10-cm dish at Plasmids and reagents various dilutions. Two weeks later, the colonies were counted with For MTDH knockdown, the sense and antisense shRNA crystal violet staining. oligonucleotides were annealed and cloned into the BglII and HindIII site of pSUPER.retro.puro (OligoEngine) as previously Luciferase reporter assay described (5). The same method was used for TWIST1 and CBP 293T cells were cotransfected with the control or MTDH over- knockdown. For MTDH overexpression, the human MTDH expression/knockdown plasmids, the indicated firefly luciferase gene was cloned into the pcDNA3.1 vector with BamHI and reporter plasmid and a Renilla luciferase plasmid with a ratio of EcoRI digestion. For TWIST1 promoter reporter assays, a 2.3-Kb 2:2:1. Lysates were collected at 48 hours after transfection, fol- DNA fragment upstream of TWIST1 transcription start site was lowed by analyses of firefly and Renilla luciferase activities with a cloned into pGL3-basic (Promega) with EcoRI and HindIII Dual-Luciferase Reporter System (Promega). digestion. All of these constructs were confirmed by sequenc- ing. The sequences of primers and shRNA constructs were available in Supplementary Table S1. The mouse anti-human RNA stability assay fl m MTDH (Abnova H00092140-B01P), the mouse anti-human Cancer cells of 80% con uence were treated with 5 g/mL MTDH (Cell Signaling Technology 9596s), the rabbit anti- actinomycin D (Dalian Meilun Biology), and mRNA was collected human CBP (Santa Cruz Biotechnology sc-369), the rabbit at the indicated time and analyzed with real-time PCR. anti-human CD24 (Santa Cruz Biotechnology sc-11406), the rat anti-human CD44 (Santa Cruz Biotechnology sc-18849), Microarray hybridization and data analysis the mouse anti-human TWIST1 (Santa Cruz Biotechnology Total RNAs of MCF7 with or without MTDH overexpression sc-814417) antibodies, the rabbit anti-human p300 (Santa were subjected to hybridization to Affymetrix U133 plus 2.0 Cruz Biotechnology sc-584), and DAPI (Roche 10236276001) arrays, performed by Shanghai Biotechnology Corporation. The were used in this study for Western blot analysis or immuno- microarraydatawerenormalizedaccordingtothemedian fluorescence analyses. The rabbit anti-acetyl-histone H3 anti- intensity of each sample. The genes with expression ratios body (Upstate 06-599) was used for chromatin immunopre- between the overexpression and control cells greater than 4 cipitation (ChIP) assays. C646 (Selleckchem S7152, 25 mmol/L were defined as gene candidates regulated by MTDH. The for 24 hours), cycloheximide (Beyotime s1560, 25 mg/mL for microarray data were available in the Gene Expression Omni- 4–24 hours), and MG-132 (Beyotime s1748, 100 mmol/L for bus database (GSE67249). 12 hours) were used to inhibit CBP, protein translation, and proteasome activity, respectively. The reagents used to inhibit Chromatin immunoprecipitation DNA methylation and histone deacetylation were 5-Aza-2- For ChIP assays, cell lysates were cross-linked with 1% m – deoxycytidine (2.5 or 5 mol/L, 48 72 hours) and Trichostain formaldehyde, and 125 mmol/L glycine was used to quench – A (300 or 600 nmol/L, 12 24 hours). the formaldehyde. The nuclear extracts were sonicated and incubated with control IgG or anti-acetyl-histone H3 antibody Tumorsphere culture for immunoprecipitation. The precipitated complexes were Cells were cultured as tumorspheres in DMEM containing eluted and reverse cross-linked. The captured genomic DNA 20 ng/mL rhEGF (R&D Systems), 10 ng/mL rhbFGF (R&D Sys- was purified with the silica membrane purification kit (TIAN- tems), 4 mg/mL heparin sulfate (Sigma), B27 supplement, and 1% GEN) and was used for PCR analyses. One percent of total penicillin G-streptomycin. Five thousand cells were seeded in genomic DNA from nuclear extracts was used as input. each well of a 6-well ultra-low attachment plate. After 2 weeks of culture, spheres with diameters larger than 50 mm were counted. When tumorspheres were passaged, sphere pellets were centri- Animal studies fuged and digested with trypsin-EDTA, and 5,000 cells were All animal experiments were performed according to the reseeded in a new plate. guidelines for the care and use of laboratory animals and were approved by the institutional biomedical research ethics com- FACS analyses mittee of Shanghai Institutes for Biological Sciences (Shanghai, FACS analyses were performed using a MoFlo Astrios Flow China). NOD/SCID or Balb/c nude mice at the age of 4 to 6 Cytometer (Beckman). Nonviable cells were excluded from weeks were used in all studies. Orthotopic injection was further analyses. The ALDEFLUOR Kit (Stem Cell Technologies) performed to study primary tumor growth as previously was used to analyze ALDH-positive subpopulation. The specific described (11). Starting from one week before cancer cells ALDH inhibitor diethylaminobenzaldehyde (DEAB) was used injection, the mice were treated with subcutaneous injection to assess the background signal. For side population analyses, of estradiol cypionate (5 mg/kg in 100 mL sesame oil, Dalian

OF2 Cancer Res; 75(17) September 1, 2015 Cancer Research

MTDH Activates TWIST1 and Promotes Cancer Cell Stemness

Meilun Biology) every week for 4 weeks in MCF7 tumorigen- Results esis assays. MTDH knockdown inhibits CSC expansion and tumorigenicity of breast cancer cells Clinical analyses To study the functional role of MTDH in CSCs of breast cancer, fi Frozen and paraf n-embedded breast tumor specimens were MTDH was knocked down by two different shRNA in LM2 obtained from Qilu Hospital of Shandong University (Ji'nan, (Fig. 1A), a lung-metastatic subline derived from MDA-MB-231 China) with informed patient consent and approval from the cells (34), and the in vitro stemness of the cells were assessed. As Institutional Review Board. Frozen tissues were used for RNA showed in Fig. 1B, MTDH knockdown significantly inhibited extraction, followed by qPCR analyses of MTDH and TWIST1 tumorsphere formation of cancer cells. Through FACS analyses, expression levels. For Cox survival analysis, the patients were we also found that the ALDH-positive CSC subpopulation was fi classi ed according to MTDH median expression level. significantly reduced in the knockdown cells (Fig. 1C and D). fl fi For immuno uorescence analysis, paraf n-embedded speci- To examine the effects of MTDH knockdown on tumorigenic- fi fi mens were rst deparaf nized. Then the antigen was unmasked ity, we performed serial transplantations of limiting dilutions of with 10 mmol/L sodium citrate at a sub-boiling temperature for the cells into the mammary fat pads of NOD/SCID mice and 10 minutes. After blocking, the specimens were incubated with assessed the tumor growth. Interestingly, when 10,000 cells were aprimaryantibodyofCD44,CD24,orMTDHovernightat4 C. injected, no difference in tumor growth was observed between the The samples were washed with PBS for three times and then a control and MTDH knockdown groups. However, tumor inhibi- fl uorochrome-conjugated secondary antibody was incubated tion by MTDH knockdown was revealed with 1,000 or 200 cells with samples. After washing, the samples were mounted with injected (Fig. 1E). We further analyzed the difference in tumor fl coverslips and the immuno uorescence signal was collected incidence and growth by injecting only 40 cells. MTDH knock- with confocal microscopy. We distinguished the tumor epithe- down significantly delayed the appearance of tumors (Fig. 1F). At lial and stromal cells with H/E staining. For statistic analyses, the sixth week, 5 of 6 injected mice developed tumors in the fi fl ten random view elds of each sample were assessed for uo- control group, whereas only one tumor formed in the 12 mice rescence intensities with the ZEN blue edition software. Each injected with the MTDH knockdown cells (Fig. 1G). The tumor samplewasscoredaslowandhighaccordingtoMTDHstaining fi þ sizes were also signi cantly reduced in the knockdown groups intensities and CD44 CD24 cells were counted in the tumor when some tumors eventually appeared after 8 weeks (Fig. 1E and region. H). Hence, MTDH inhibition significantly impaired CSC expansion and tumorigenicity of breast cancer cells. Statistical analyses Unless stated otherwise, data are presented as mean SD in the MTDH overexpression promotes CSC properties of breast and figures. A Student t test was performed to compare the in vitro data. prostate cancer cells A log-rank test was performed to compare tumor-free survival. Next, we analyzed the function of MTDH by overexpressing it in P values less than 0.05 were considered statistically significant. MCF7 cells (Fig. 2A). Compared with the cells expressing the

Figure 1. MTDH knockdown inhibits CSC properties of breast cancer cells. A, MTDH knockdown in LM2 cells. B, quantitation and representative images of tumorsphere formation in LM2 cells. Scale bars, 200 mm. C, ALDEFLUOR analyses of LM2. Quadrangles indicate ALDH-positive population and the ALDH inhibitor DEAB is used for negative control. D, quantitation of ALDH-positive population in LM2. E, tumor volumes of NOD/SCID mice injected with the indicated numbers of LM2 cells at week 4 (left) and week 6 (right). F, Kaplan–Meier analysis of tumor-free survival of the mice injected with 40 LM2 cells. G, tumor incidence of the mice injected with 40 LM2 cells by week 6 (n 6 in each group). H, tumor images of LM2 at week 8. , P < 0.05; , P < 0.01; NS, nonsigniﬁcant. Ctrl, control.

www.aacrjournals.org Cancer Res; 75(17) September 1, 2015 OF3

Liang et al.

Figure 2. MTDH overexpression promotes CSC properties of breast cancer cells. A, MTDH overexpression in MCF7 cells. B, quantitation and representative images of sphere formation in MCF7 cells. Scale bars, 200 mm. C, relative sphere numbers of MCF7 cells in successive passage experiments. D, side population analyses of MCF7. Verapamil was used for negative control. E, side population quantitation of MCF7. F, tumor incidence of the mice injected with MCF7 cells at week 7 (n 6 in each group). G, tumor images of the mice injected with MCF7 cells at the sixth month. , P < 0.001.

empty vector, the overexpression cells formed more and larger population was significantly reduced in the knockdown cells tumorspheres (Fig. 2B). This enhancement was stable for at least (Supplementary Fig. S3B), together with the inhibition of ABC three passages of the spheres (Fig. 2C). By Hoechest efflux anal- transporter genes (Supplementary Fig. S1B), resistance to pacli- ysis, we found that MTDH overexpression expanded the side taxel and doxorubicin (Supplementary Fig. S1C), as well as cancer population of the cells for more than two folds (Fig. 2D and cell EMT (Supplementary Fig. S2). We also tested the in vivo þ E). The CD44 CD24 population was also upregulated by tumorigenicity of DU145 cells by transplanting the cells subcu- MTDH (Supplementary Fig. S1A). In addition, ectopic expression taneously into NOD/SCID mice. Similar to the observations of of MTDH correlated with increased expression of ABC transporter LM2 cells, there was no difference in tumor growth between the genes, including ABCB1 and ABCG2 (Supplementary Fig. S1B), control and MTDH knockdown groups when 1 106 cells were and enhanced cell resistance to chemotherapeutic drugs such as injected (Supplementary Fig. S3C). However, when fewer cells, paclitaxel, which are also hallmark properties of CSCs (Supple- including 1 105,1 104,2 103,4 102 cells, were injected, mentary Fig. S1C). MTDH knockdown significantly reduced tumorigenicity and sup- To assess the in vivo tumorigenicity of the cells, we implanted pressed tumor growth compared with the control group (Sup- 1 106 MCF7 cancer cells into the mammary fat pads of nude plementary Fig. S3C–S3E). Overall, these data demonstrated that mice with estrogen supplements, and found that MTDH over- MTDH was a pivotal factor for CSC regulation. expression significantly promoted the tumor incidence. By week 6, all mice with MTDH overexpression cells developed palpable TWIST1 mediates the role of MTDH in CSCs tumors, whereas only 1 out of 10 mice formed a tumor in the Given the fundamental role of MTDH in CSCs, we searched for control group (Fig. 2F). When we repeated the experiments by the downstream genes mediating its function by performing a injecting 5 105 cancer cells and waited for 6 months to observe microarray analysis of MCF7 cells with or without MTDH over- the tumor formation, MTDH overexpression still markedly expression (Fig. 3A). The microarray data showed that TWIST1 increased tumor incidence and tumor sizes (Fig. 2G). Therefore, ranked at the top of the candidate genes regulated by MTDH, MTDH is indispensable and sufficient to promote CSC propaga- which was confirmed in the mRNA and protein levels by qPCR tion and tumorigenicity of breast cancer cells. and Western blot analyses in MCF7 (Fig. 3B). We also found that We also found obvious changes in the expression of EMT MTDH knockdown led to the downregulation of TWIST1 in LM2 marker proteins, including CDH1 and CDH2, following MTDH (Fig. 3B). Because TWIST1 is an important factor to promote EMT overexpression and knockdown in MCF7 and LM2 cells (Supple- and CSC in breast cancer (22, 34–36), we examined whether mentary Fig. S2), an observation consistent to previous reports on TWIST1 mediated the function of MTDH in driving CSC expan- the role of MTDH in EMT (30–33). sion and tumorigenicity. We stably silenced TWIST1 in the To substantiate the functional study of MTDH in CSCs, we MTDH-overexpressing MCF7 cells (Fig. 3C). TWIST1 knockdown silenced the expression of MTDH in DU145 prostate cancer cells significantly impaired side population abundance, tumorsphere þ (Supplementary Fig. S3A) and found that the CD44 CD24 formation, and in vivo tumorigenicity that were enhanced by

OF4 Cancer Res; 75(17) September 1, 2015 Cancer Research

MTDH Activates TWIST1 and Promotes Cancer Cell Stemness

MTDH overexpression. Notably, the cells with both MTDH over- showed that each of the chemicals was able to activate the expression and TWIST1 inhibition displayed a tumor-initiating expression of TWIST1 in the control cells, and the upregulation ability that was comparable with that of the control cells without was more obvious when both chemicals were used together. MTDH overexpression (Fig. 3D and E). However, in the MTDH-expressing cells, TSA was less potent for We further overexpressed TWIST1 in LM2 with or without TWIST1 induction than Aza (Fig. 4C). Further analyses demon- MTDH knockdown (Fig. 3C), and found that TWIST1 rescued strated that when histone deacetylation was inhibited by TSA, the the tumor-suppressing phenotypes of MTDH knockdown. More expression of TWIST1 was almost the same in control and MTDH- importantly, MTDH knockdown led to no obvious difference in expressing cells (Fig. 4D). In contrast, Aza was not able to tumor incidence or animal survival when TWIST1 was constitu- diminish the upregulation of TWIST1 by MTDH. The differential tively expressed in the cancer cells (Fig. 3F and G). Therefore, effects of these two inhibitors on TWIST1 expression was not TWIST1 acts downstream of MTDH to regulate CSC expansion related to the treatment duration or doses, as shown by repeating and tumorigenicity of breast cancer cells. the experiments with different time courses and compound con- centrations of the inhibitor treatment (Supplementary Fig. S4A MTDH activates TWIST1 expression by upregulating histone and S4B). These data suggested that TWIST1 activation by MTDH acetylation might be due to regulation of histone acetylation. To verify this, Next, we examined how TWIST1 was regulated by MTDH. we performed a ChIP assay to analyze the histone acetylation Because the mRNA level of TWIST1 was extensively regulated status around TWIST1 promoter and found that MTDH over- by MTDH, we cloned the promoter of TWIST1 and performed expression indeed enhanced acetylation of histone H3 on TWIST1 luciferase reporter assays in 293T cells. The results showed that promoter (Fig. 4E). Moreover, the overall level of H3 acetylation MTDH knockdown or overexpression failed to signiﬁcantly was also increased after MTDH overexpression, as revealed by affect the reporter activity (Fig. 4A), suggesting that MTDH Western blotting (Fig. 4F). Notably, there was no obvious binding cannot directly regulate the promoter activity of TWIST1. Next, of acetylated histone on the TWIST1 reporter promoter ectopically we analyzed the mRNA stability of TWIST1 in breast cancer cell introduced into the cells (Supplementary Fig. S4C), providing an lines. Following actinomycin D treatment to inhibit de novo explanation why the reporter construct was not responsive to transcription, the mRNA level of TWIST1 declined within MTDH overexpression or knockdown (Fig. 4A). hours. However, MTDH knockdown or overexpression led to no signiﬁcant changes in the speed of TWIST1 mRNA degra- MTDH interacts with CBP to prevent ubiquitin-mediated CBP dation(Fig.4B).Therefore,MTDH was not likely to stabilize degradation TWIST1 mRNA. Because MTDH regulated histone H3 acetylation on the pro- Then we analyzed whether epigenetic regulation was involved moter of TWIST1, we further searched for the executor of histone in TWIST1 activation by MTDH. We treated MCF7 cells with acetylation. Previous studies showed that MTDH was able to MTDH overexpression and control cells using the histone deace- interact with the histone acetyltransferase CBP (9). In addition, tylase (HDAC) inhibitor trichostatin A (TSA) and the DNA CBP and its homolog p300 were connected to TWIST1 expression methylation inhibitor 5-Aza-2-deoxycytidine (Aza). The data (37, 38). Therefore, we tested the possibility that CBP was

Figure 3. TWIST1 mediates the role of MTDH in CSCs. A, expression heatmap of the genes regulated by MTDH in MCF7. B, veriﬁcation of TWIST1 regulation by MTDH in MCF7 and LM2. C, TWIST1 knockdown in MTDH overexpression cells, and TWIST1 overexpression in MTDH knockdown cells. D, side population and tumorsphere formation of TWIST1 knockdown cells. Scale bars, 200 mm. E, tumor incidence of the mice injected with TWIST1 knockdown cells by day 80 (n 6in each group). F, tumor incidence of the mice injected with TWIST1 overexpression cells by day 30 (n 6 in each group). G, tumor-free survival analysis of the mice injected with TWIST1 overexpression cells. , P < 0.05; , P < 0.01. NS, nonsigniﬁcant.

www.aacrjournals.org Cancer Res; 75(17) September 1, 2015 OF5

Liang et al.

Figure 4. MTDH activates TWIST1 expression by upregulating histone acetylation. A, luciferase activities of TWIST1 promoter in 293T cells transfected with MTDH overexpression or knockdown plasmids. B, TWIST1 mRNA levels in MTDH overexpression or knockdown cells after treatment with actinomycin D. C, TWIST1 mRNA levels in control and MTDH overexpression cells after treatment with AZA (2.5 mmol/L for 72 hours) or TSA (300 nmol/L for 25 hours). D, TWIST1 mRNA levels in MTDH overexpression cells normalized to that of control cells after treatment with AZA or TSA. E, ChIP-qPCR analyses of acetylated histone H3 binding on TWIST1 promoter. F, histone H3 acetylation levels in MCF7 cells. , P < 0.01. NS, nonsigniﬁcant.

involved in MTDH-regulated histone acetylation. First, we MTDH to regulate TWIST1 expression and CSC population in confirmed that MTDH interacted with the CBP protein with cancer cells. reciprocal coimmunoprecipitation assays in cancer cells (Fig. 5A). Then we found that CBP transcription was unaffected by MTDH correlates positively with CSC abundance and TWIST1 MTDH in MCF7, but its protein level was obviously increased expression in breast cancer samples upon MTDH overexpression (Fig. 5B and C). CBP mRNA Next, we studied the clinical significance of our findings. To stability analysis showed that MTDH could not protect CBP investigate the correlation of MTDH with CSC abundance in transcript from degradation (Fig. 5D). However, when de novo human breast tumors, we assessed the expression of MTDH, protein synthesis was inhibited by cycloheximide (CHX), the CD44, and CD24 in 16 invasive ductal breast carcinomas by degradation of CBP protein was pronouncedly slower in MTDH immunofluorescence staining (2). The analysis showed that overexpression cells than the control cells (Fig. 5E and F). In higher expression of MTDH was significantly linked to increased further analyses, we blocked ubiquitin-mediated proteasome þ presence of CD44 CD24 CSC population in the tumors (Fig. 6A activity by treating cells with MG132 and analyzed the ubiqui- and B). To unravel the clinical relevance of MTDH and TWIST1 tination status of CBP. Accordantly, less CBP ubiquitination expression, we collected a set of fresh-frozen breast tumors from was observed when MTDH was overexpressed (Fig. 5G), argu- Qilu Hospital and assessed the mRNA levels of the two genes. ing for the conclusion that MTDH protects the CBP protein MTDH expression correlated positively with that of TWIST1 in from ubiquitin-mediated degradation. these samples (Fig. 6C, Pearson r ¼ 0.38). IHC analysis also To further verify that CBP was responsible for MTDH regulation showed that both MTDH and TWIST1 proteins were upregulated of TWIST1 expression and CSC properties, we used a p300/CBP- in breast tumor edges as compared with intratumor areas (Sup- specific inhibitor C646 to treat MCF7 and found that CBP inhi- plementary Fig. S6), suggesting the enrichment of the molecules bition blocked TWIST1 expression and sphere formation that in invasive front of tumors. Moreover, MTDH expression was also were induced by MTDH overexpression (Fig. 5H and I). We also a prognostic factor of distant metastasis in this cohort (Fig. 6D). stably silenced CBP with a shRNA construct in MTDH overexpres- Thus, MTDH is clinically linked to TWIST1 expression, CSC sion cells (Supplementary Fig. S5A) and observed the same abundance, and prognosis of breast cancer. changes (Fig. 5J). In addition, side population analyses in these cells showed that although MTDH increased CSC fraction in MCF7, CBP silencing reversed the phenotype induced by MTDH Discussion (Fig. 5K). Notably, the CBP homolog p300 was not affected by Accumulated data of functional studies and clinical analyses MTDH overexpression or CBP shRNA. In addition, inhibition of have firmly established MTDH as an important regulator of p300 with a siRNA construct did not influence the expression of neurodegeneration, cancer metastasis, and chemoresistance TWIST1 (Supplementary Fig. S5B and S5C). Collectively, these (5, 11, 17, 39, 40). However, compared with the unanimous data demonstrated that CBP, but not p300, acted downstream of functional roles of MTDH in these processes, the underlying

OF6 Cancer Res; 75(17) September 1, 2015 Cancer Research

MTDH Activates TWIST1 and Promotes Cancer Cell Stemness

Figure 5. MTDH interacts with CBP and prevents CBP from ubiquitin- mediated degradation. A, coimmunoprecipitation assays in HeLa cells transfected with CBP and MTDH plasmids. B, CBP mRNA levels in MCF7. C, CBP protein levels in MCF7. D, CBP mRNA levels in MCF7 after treatment with actinomycin D. E, CBP protein levels in MCF7 after treatment with cycloheximide (CHX). F, quantitation of CBP protein levels in MCF7 after treatment with cycloheximide. G, CBP protein was immunoprecipitated and detected with anti-uibiquitin antibody. H, TWIST1 protein levels in MCF7 after C646 treatment. I, tumorsphere formation of MCF7 after C646 treatment. J, tumorsphere formation of MCF7 with CBP knockdown. K, side population of MCF7 with CBP knockdown. , P < 0.05. NS, nonsigniﬁcant.

mechanisms are far from clear. Indeed, the downstream events of (42). Here, we corroborated this conclusion with various CSC MTDH remain fragmentary despite previous efforts to explain the assays in cancer cell lines. In addition, our data uncovered the link potency of MTDH in cancer regulation. On one hand, although of MTDH with TWIST1, the pivotal transcription factor in CSC numerous studies reported that MTDH activated oncogenic path- expansion and cancer cell EMT (21, 22). Therefore, these results ways such as PI3K/Akt, NF-kB, and Wnt/b-catenin (9, 18–20), it is suggest that the pluripotency of MTDH in cancer can be explained unclear how MTDH is involved in these signaling pathways. On by a fundamental role in CSC regulation, and support the ratio- the other hand, recent data demonstrated that the physical inter- nale of MTDH-targeting strategies to prevent cancer initiation, action of MTDH with the nuclease protein SND1 was critical for recurrence, and metastasis. cancer initiation (41–43), the downstream signaling of this inter- The challenge in studying the functional mechanisms of MTDH action is unknown. Here, we reported a different and detailed is largely due to the lack of homology of MTDH to any known pathway that can connect MTDH to the malignant behaviors of protein structures. However, among the misty data regarding the cancer cells. First, we provided evidence of the role of MTDH in molecular nature of MTDH, the observations that MTDH plays its CSC regulation by showing that MTDH elevated the abundance of roles by protein–protein interactions seem to be increasingly ALDH-positive fraction and side population of cancer cell lines, convincing. In addition to CBP, a number of proteins, such as þ and correlated with the CD44 CD24 subpopulation in clinical CEACAM1, BCCIPa, PLZF, p65, AGO2, and SND1, have been samples. It also promoted sphere formation, drug resistance as documented to interact with MTDH, and these interactions are well as in vivo tumorigenesis of cancer cells. Then we demonstrated indispensible for MTDH regulation of cancer cell features that the regulation of CSC traits by MTDH was dependent on the (7, 9, 19, 43–46). Notably, MTDH stabilizes some of these partner downstream molecules CBP and TWIST1. MTDH interacted with proteins, including BCCIPa and SND1. Here, we not only show CBP to prevent ubiquitin-mediated degradation of the histone that MTDH prevents the degradation of CBP through physical acetyltransferase and enhanced histone H3 acetylation on interaction, but also demonstrate the involvement of ubiquitin- TWIST1 promoter, thus leading to the transcriptional activation mediated proteasome activity in this process. Therefore, disrupt- of TWIST1 and CSC promotion. Thus our study delineated a ing the interaction of MTDH with its partner proteins may be molecular pathway of MTDH in driving cancer progression (Fig. useful to target MTDH in cancer therapeutics. However, further 6E) and will profoundly expand our understanding for this master studies are required to illustrate how MTDH selects its interacting regulator of cancer. partners and interrupts protein degradation in order to rationalize Metastasis and chemoresistance are the two most malignant such approaches. features of tumor cells that account for disease progression and Histone modiﬁcations are important epigenetic mechanisms to eventually therapeutic failure of cancer. As we and many others regulate a broad range of developmental and disease processes. In have shown the dual roles of MTDH to promote metastasis and this study, we unravel a new role of MTDH in epigenetic regula- chemoresistance (11, 40), it would be interesting to test whether tion to interact and stabilize the CBP protein. As a core member of MTDH also regulates cancer initiation and recurrence. Previously the p300/CBP histone acetyltransferase (HAT) family, CBP reg- Kang and colleagues demonstrated that MTDH promoted tumor ulates gene expression as a transcriptional coactivator by acety- initiation using oncogene-driven breast cancer mouse models lating histones and other transcription factors. CBP has been

www.aacrjournals.org Cancer Res; 75(17) September 1, 2015 OF7

Liang et al.

Figure 6. MTDH correlates positively with the CSC population and TWIST1 expression in breast cancer samples. A, representative images of MTDH, CD44, and CD24 þ immunoﬂuorescence in invasive ductal carcinomas (n ¼ 16). Scale bars, 20 mm. B, quantitation of CD44 CD24 population in invasive ductal carcinomas with different MTDH protein levels. C, correlation of MTDH and TWIST1 expression in breast cancer samples. D, survival analysis of the Qilu patients stratiﬁed by MTDH expression. E, a schematic model of the role of MTDH to regulate CSC traits. , P < 0.05.

widely known to be involved in normal stem cell regulation and Acquisition of data (provided animals, acquired and managed patients, cancer progression (47–49), but its role in CSC is seldom provided facilities, etc.): Y. Liang, J. Hu, J. Li, Y. Liu, J. Yu, X. Zhuang, D. Hong, reported. Our data show that CBP directly regulates the expression Q. Yang Analysis and interpretation of data (e.g., statistical analysis, biostatistics, of TWIST1 and promotes CSC traits, thus expanding the role of computational analysis): Y. Liang, J. Hu, J. Li, Y. Liu, J. Yu, X. Zhuang CBP in cancer. Our data, together with previous studies (37, 38), Writing, review, and/or revision of the manuscript: Y. Liang, J. Hu, J. Li, Y. Liu, suggested that CBP acted upstream of TWIST1 and its homolog J. Yu, X. Zhuang, G. Hu p300 was a downstream target of TWIST1, highlighting the Administrative, technical, or material support (i.e., reporting or organizing connection of the MTDH–TWIST1 axis to histone modification. data, constructing databases): L. Mu, D. Hong More importantly, as many histone modification inhibitors are Study supervision: G. Hu being developed in therapeutics, our findings that link MTDH to Grant Support epigenetic regulation may provide new opportunities to target This work was supported by National 973 programs (2011CB510105 and CSC and MTDH signaling in metastatic cancer. 2013CB910904) and grants from the National Natural Science Foundation of China (81430070, 81222032, 31371409) and the Ministry of Science and Disclosure of Potential Conflicts of Interest Technology of China (2012ZX09506-001-005). The costs of publication of this article were defrayed in part by the payment of No potential conflicts of interest were disclosed. page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. Authors' Contributions Conception and design: X. Kong, D. Hong, G. Hu Received April 8, 2015; revised June 16, 2015; accepted June 29, 2015; Development of methodology: D. Hong, Q. Yang published OnlineFirst July 3, 2015.

References 1. Valent P, Bonnet D, De Maria R, Lapidot T, Copland M, Melo JV, et al. 3. Clevers H. The cancer stem cell: premises, promises and challenges. Nat Cancer stem cell definitions and terminology: the devil is in the details. Nat Med 2011;17:313–9. Rev Cancer 2012;12:767–75. 4. Garnis C, Coe BP, Zhang LW, Rosin MP, Lam WL. Overexpression of LRP12, 2. Al-Hajj M, Wicha MS, Benito-Hernandez A, Morrison SJ, Clarke MF. a gene contained within an 8q22 amplicon identified by high-resolution Prospective identification of tumorigenic breast cancer cells. Proc Natl array CGH analysis of oral squamous cell carcinomas. Oncogene Acad Sci U S A 2003;100:3983–8. 2004;23:2582–6.

OF8 Cancer Res; 75(17) September 1, 2015 Cancer Research

MTDH Activates TWIST1 and Promotes Cancer Cell Stemness

5. Hu G, Chong RA, Yang Q, Wei Y, Blanco MA, Li F, et al. MTDH activation by 28. Gort EH, Suijkerbuijk KPM, Roothaan SM, Raman V, Vooijs M, van der 8q22 genomic gain promotes chemoresistance and metastasis of poor- Wall E, et al. Methylation of the TWIST1 promoter, TWIST1 mRNA levels, prognosis breast cancer. Cancer Cell 2009;15:9–20. and immunohistochemical expression of TWIST1 in breast cancer. Cancer 6. Brown DM, Ruoslahti E. Metadherin, a cell surface protein in breast tumors Epidemiol Biomarkers Prev 2008;17:3325–30. that mediates lung metastasis. Cancer Cell 2004;5:365–74. 29. Ezponda T, Popovic R, Shah MY, Martinez-Garcia E, Zheng Y, Min DJ, et al. 7. Blanco MA, Aleckovic M, Hua Y, Li T, Wei Y, Xu Z, et al. Identification of The histone methyltransferase MMSET/WHSC1 activates TWIST1 to pro- Staphylococcal Nuclease Domain-containing 1 (SND1) as a Metadherin- mote an epithelial-mesenchymal transition and invasive properties of interacting Protein with Metastasis-promoting Functions. J Biol Chem prostate cancer. Oncogene 2013;32:2882–90. 2011;286:19982–92. 30. Zhu K, Dai Z, Pan Q, Wang Z, Yang GH, Yu L, et al. Metadherin promotes 8. Wei Y, Hu G, Kang Y. Metadherin as a link between metastasis and hepatocellular carcinoma metastasis through induction of epithelial-mes- chemoresistance. Cell Cycle 2009;8:2132–3. enchymal transition. Clin Cancer Res 2011;17:7294–302. 9. Sarkar D, Park ES, Emdad L, Lee SG, Su ZZ, Fisher PB. Molecular basis of 31. Liu XW, Wang DG, Liu HL, Feng Y, Zhu TY, Zhang L, et al. Knockdown of nuclear factor-kappaB activation by astrocyte elevated gene-1. Cancer Res astrocyte elevated gene-1 (AEG-1) in cervical cancer cells decreases their 2008;68:1478–84. invasiveness, epithelial to mesenchymal transition, and chemoresistance. 10. Yoo BK, Chen D, Su ZZ, Gredler R, Yoo J, Shah K, et al. Molecular Cell Cycle 2014;13. mechanism of chemoresistance by astrocyte elevated gene-1. Cancer Res 32. Suh SS, Yoo JY, Cui R, Kaur B, Huebner K, Lee TK, et al. FHIT suppresses 2010;70:3249–58. epithelial-mesenchymal transition (EMT) and metastasis in lung cancer 11. Liang Y, Fu D, Hu G. MTDH: an emerging key regulator of malignant through modulation of MicroRNAs. PLoS Genet 2014;10. progression in multiple cancers. Thorac Cancer 2011;2:143–8. 33. Ward A, Balwierz A, Zhang JD, Kublbeck M, Pawitan Y, Hielscher T, et al. Re- 12. Wang K, Lim HY, Shi S, Lee J, Deng S, Xie T, et al. Genomic landscape of expression of microRNA-375 reverses both tamoxifen resistance and copy number aberrations enables the identification of oncogenic drivers in accompanying EMT-like properties in breast cancer. Oncogene 2013;32: hepatocellular carcinoma. Hepatology 2013;58:706–17. 1173–82. 13. Wan L, Hu G, Wei Y, Yuan M, Bronson RT, Yang Q, et al. Genetic ablation of 34. Minn AJ, Gupta GP, Siegel PM, Bos PD, Shu WP, Giri DD, et al. Genes that metadherin inhibits autochthonous prostate cancer progression and mediate breast cancer metastasis to lung. Nature 2005;436:518–24. metastasis. Cancer Res 2014;74:5336–47. 35. Hollier BG, Tinnirello AA, Werden SJ, Evans KW, Taube JH, Sarkar TR, et al. 14. Casimiro S, Fernandes A, Oliveira AG, Franco M, Pires R, Peres M, et al. FOXC2 expression links epithelial-mesenchymal transition and stem cell Metadherin expression and lung relapse in patients with colorectal carci- properties in breast cancer. Cancer Res 2013;73:1981–92. noma. Clin Exp Metastasis 2014;31:689–96. 36. Beck B, Blanpain C. Unravelling cancer stem cell potential. Nat Rev Cancer 15. Liu LP, Wu JH, Ying Z, Chen BX, Han AJ, Liang YJ, et al. Astrocyte elevated 2013;13:727–38. gene-1 upregulates matrix metalloproteinase-9 and induces human glioma 37. Hamamori Y, Sartorelli V, Ogryzko V, Puri PL, Wu HY, Wang JY, et al. invasion. Cancer Res 2010;70:3750–9. Regulation of histone acetyltransferases p300 and PCAF by the bHLH 16. Lee SG, Jeon HY, Su ZZ, Richards JE, Vozhilla N, Sarkar D, et al. Astrocyte protein twist and adenoviral oncoprotein E1A. Cell 1999;96:405–13. elevated gene-1 contributes to the pathogenesis of neuroblastoma. Onco- 38. Akimaru H, Hou DX, Ishii S. Drosophila CBP is required for dorsal- gene 2009;28:2476–84. dependent twist gene expression. Nat Genet 1997;17:211–4. 17. Hu G, Wei Y, Kang Y. The multifaceted role of MTDH/AEG-1 in cancer 39. Emdad L, Sarkar D, Su ZZ, Lee SG, Kang DC, Bruce JN, et al. Astrocyte progression. Clin Cancer Res 2009;15:5615–20. elevated gene-1: recent insights into a novel gene involved in tumor 18. Lee SG, Su ZZ, Emdad L, Sarkar D, Franke TF, Fisher PB. Astrocyte elevated progression, metastasis and neurodegeneration. Pharmacol Ther gene-1 activates cell survival pathways through PI3K-Akt signaling. Onco- 2007;114:155–70. gene 2008;27:1114–21. 40. Yoo BK, Emdad L, Lee SG, Su ZZ, Santhekadur P, Chen D, et al. Astrocyte 19. Emdad L, Sarkar D, Su ZZ, Randolph A, Boukerche H, Valerie K, et al. elevated gene-1 (AEG-1): A multifunctional regulator of normal and Activation of the nuclear factor kappaB pathway by astrocyte elevated gene- abnormal physiology. Pharmacol Ther 2011;130:1–8. 1: implications for tumor progression and metastasis. Cancer Res 41. Guo F, Wan LL, Zheng AP, Stanevich V, Wei Y, Satyshur KA, et al. Structural 2006;66:1509–16. insights into the tumor-promoting function of the MTDH-SND1 complex. 20. Yoo BK, Emdad L, Su ZZ, Villanueva A, Chiang DY, Mukhopadhyay ND, Cell Rep 2014;8:1704–13. et al. Astrocyte elevated gene-1 regulates hepatocellular carcinoma devel- 42. Wan LL, Lu X, Yuan S, Wei Y, Guo F, Shen MH, et al. MTDH-SND1 opment and progression. J Clin Invest 2009;119:465–77. interaction is crucial for expansion and activity of tumor-initiating cells 21. Yang J, Mani SA, Donaher JL, Ramaswamy S, Itzykson RA, Come C, et al. in diverse oncogene- and carcinogen-induced mammary tumors. Cancer Twist, a master regulator of morphogenesis, plays an essential role in tumor Cell 2014;26:92–105. metastasis. Cell 2004;117:927–39. 43. Yoo BK, Santhekadur PK, Gredler R, Chen D, Emdad L, Bhutia S, et al. 22. Mani SA, Guo W, Liao MJ, Eaton EN, Ayyanan A, Zhou AY, et al. The Increased RNA-induced silencing complex (RISC) activity contributes to epithelial-mesenchymal transition generates cells with properties of stem hepatocellular carcinoma. Hepatology 2011;53:1538–48. cells. Cell 2008;133:704–15. 44. Britt DE, Yang DF, Yang DQ, Flanagan D, Callanan H, Lim YP, et al. 23. Yang MH, Hsu DSS, Wang HW, Wang HJ, Lan HY, Yang WH, et al. Bmi1 is Identification of a novel protein, LYRIC, localized to tight junctions of essential in Twist1-induced epithelial-mesenchymal transition. Nat Cell polarized epithelial cells. Exp Cell Res 2004;300:134–48. Biol 2010;12:982–92. 45. Ash SC, Yang DQ, Britt DE. LYRIC/AEG-1 overexpression modulates 24. Tsukerman P, Yamin R, Seidel E, Khawaled S, Schmiedel D, Bar-Mag T, et al. BCCIPalpha protein levels in prostate tumor cells. Biochem Biophys Res MiR-520d-5p directly targets TWIST1 and downregulates the metastamiR Commun 2008;371:333–8. miR-10b. Oncotarget 2014;5:12141–50. 46. Thirkettle HJ, Mills IG, Whitaker HC, Neal DE. Nuclear LYRIC/AEG-1 25. Tan EJ, Thuault S, Caja L, Carletti T, Heldin CH, Moustakas A. Regulation of interacts with PLZF and relieves PLZF-mediated repression. Oncogene Transcription Factor Twist Expression by the DNA Architectural Protein 2009;28:3663–70. High Mobility Group A2 during Epithelial-to-Mesenchymal Transition. 47. Goodman RH, Smolik S. CBP/p300 in cell growth, transformation, and Journal of Biological Chemistry 2012;287:7134–45. development. Genes Dev 2000;14:1553–77. 26. Cho KH, Jeong KJ, Shin SC, Kang J, Park CG, Lee HY. STAT3 mediates TGF- 48. Chan WI, Hannah RL, Dawson MA, Pridans C, Foster D, Joshi A, et al. beta 1-induced TWIST1 expression and prostate cancer invasion. Cancer The transcriptional coactivator Cbp regulates self-renewal and differ- Lett 2013;336:167–73. entiation in adult hematopoietic stem cells. Mol Cell Biol 2011;31: 27. Dong PX, Kaneuchi M, Xiong Y, Cao LP, Cai MY, Liu XS, et al. Identification 5046–60. of KLF17 as a novel epithelial to mesenchymal transition inducer via direct 49. Fang F, Xu Y, Chew KK, Chen X, Ng HH, Matsudaira P. Coactivators p300 activation of TWIST1 in endometrioid endometrial cancer. Carcinogenesis and CBP maintain the identity of mouse embryonic stem cells by medi- 2014;35:760–8. ating long-range chromatin structure. Stem Cells 2014;32:1805–16.

www.aacrjournals.org Cancer Res; 75(17) September 1, 2015 OF9

Epigenetic Activation of TWIST1 by MTDH Promotes Cancer Stem−like Cell Traits in Breast Cancer

Yajun Liang, Jing Hu, Jiatao Li, et al.

Cancer Res Published OnlineFirst July 3, 2015.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-15-0930

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2015/07/03/0008-5472.CAN-15-0930.DC1

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/early/2015/08/17/0008-5472.CAN-15-0930. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.