Background: Targeting cancer stem cells is critical for suppressing cancer progression and recurrence. Finding novel markers or related pathways could help eradicate or diagnose cancer in clinic. Methods: By constructing STARD13-correlated ceRNA 3′UTR stable overexpression or knockdown breast cancer cells, we aimed to explore the effects of STARD13-correlated ceRNA network on breast cancer stemness in vitro and in vivo. Further RNA-sequencing was used to analyze transcriptome change in combination with functional studies on candidate signaling. Clinical samples obtained from The Cancer Genome Atlas data were used to validate the correlation between STARD13 and related pathways. Finally, in vitro and in vivo experiments were used to examine the effects of STARD13-correlated ceRNA network on chemotherapy sensitivity/resistance. Results: Here, we revealed that this ceRNA network inhibited stemness of breast cancer. Mechanistically, we found that activation of STARD13-correlated ceRNA network was negatively correlated with YAP/TAZ activity in breast cancer. Specifically, this ceRNA network attenuated YAP/TAZ nuclear accumulation and transcriptional activity via collectively modulating Hippo and Rho-GTPase/F-actin signaling. Finally, we demonstrated that YAP/TAZ transcriptional activity regulated by this ceRNA network was involved in chemoresistance. Conclusions: Our results uncover a novel mechanism of YAP/TAZ activation in breast cancer and propose the possibility to drive STARD13-correlated ceRNA network to inhibit breast cancer stem cell traits. Keywords: STARD13 ceRNA YAP/TAZ Hippo F-actin CSC breast cancer * Correspondence: email@example.com; firstname.lastname@example.org Lufeng Zheng and Chenxi Xiang contributed equally to this work. Jiangsu Key Laboratory of Drug Discovery for Metabolic Diseases, Department of Pharmacology of Chinese Materia Medica, China Pharmaceutical University, 24 Tong Jia Xiang, Nanjing 210009, China School of Life Science and Technology, Jiangsu Key Laboratory of Carcinogenesis and Intervention, China Pharmaceutical University, Nanjing 210009, China Full list of author information is available at the end of the article © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Zheng et al. Journal of Hematology & Oncology (2018) 11:72 Page 2 of 18 Background could phosphorylate and inactivate the downstream ef- Breast cancer is the most common cancer in women fectors YAP/TAZ that mediate transcriptional output of worldwide, and its incidence is increasing yearly . the Hippo pathway. A previous study has indicated that At present, chemotherapy and surgery are the main TAZ could confer CSC-related traits on breast cancer methods for breast cancer treatment; especially, cells . YAP/TAZ are considered as “stemness factors” chemotherapy is the only option for triple negative in several types of stem cells . Inhibition of YAP/ breast cancer . However, tumor relapse and che- TAZ activity could attenuate breast cancer EMT and in- moresistance constitute a major detriment to patients’ vasion . Notably, LATS1/2 could suppress breast treatment and survival, and the mechanisms under- cancer EMT and metastasis via inactivating YAP/TAZ pinning these phenomena remain elusive, which is an activity . In line with this, aberrant YAP/TAZ tran- urgent need to solve. Since tumor relapse and che- scriptional activity is frequent in numerous tumors, moresistance are a complex network that integrates making YAP/TAZ as a potential target for cancer treat- multiple growth control signals through an expanding ment. However, efforts in this direction are frustrated by setofcoreelements, a singlefactorcould notmakea the fact that the Hippo cascade is largely undruggable good assessment of chemotherapy or tumor recur- . While the core components of the Hippo signaling rence. In fact, we have only a scattered understanding are well established, numerous additional upstream reg- of the molecular mechanisms that are responsible for ulators of YAP/TAZ activity are emerging, such as the tumor relapse and chemoresistance. transcriptional activity of YAP/TAZ is influenced by cell Cancer stem cells (CSCs) are proposed to drive tumor mechanics, and this process is driven through Rho- relapse and chemoresistance . CSCs are a small group GTPase/F-actin signaling, a manner largely independent of cells within the tumor bulk, which remain dominant on LATS1/2 . In addition, actin remodeling factors under the rapid growth of tumor cells, thus circumvent- control ciliogenesis by regulating YAP/TAZ activity . ing chemical agents, and then spawn non-CSC progeny, Rho-GTPase/F-actin signaling could promote the long- leading to tumor recurrence or chemoresistance . It term survival and expansion of human embryonic stem was proposed that in the course of malignant progres- cells . The fact that LATS1/2 and F-actin sion, the increased proportion of the CSCs within the organization act independently to regulate YAP/TAZ is tumor is inextricably twined with activation of the also supported by genetic evidence in Drosophila . epithelial-to-mesenchymal transition (EMT) . EMT is Despite the clear association of Rho-GTPase/F-actin and a complex transdifferentiation program that is instru- Hippo-YAP signaling in various cancers, targeted therap- mental for the acquisition of stemness by non- ies aiming at these two pathways remain limited. Col- transformed tumor cells [6, 7]. Several genes involved in lectively, these findings speak to that coordinately maintaining self-renewal capacity were also demon- activating Hippo signaling and inactivating Rho-GTPase/ strated to enhance EMT features . Over the past sev- F-actin pathway might be an ideal way to suppress YAP/ eral years, competing endogenous RNAs (ceRNAs) have TAZ activity, and thus CSC formation. emerged as an important class of post-transcriptional Here, we found that STARD13-correlated ceRNA net- regulator that alters gene expression through an work suppressed breast CSC formation in vitro and in miRNA-mediated mechanism [9, 10]. We and others vivo. To characterize the mechanisms and roles of have revealed that ceRNAs have significant roles in can- STARD13-correlated ceRNA network, we performed a cer pathogenesis by altering the expression of key candidate functional screen and identified LATS1/2 and tumorigenic or tumor-suppressive genes in both solid RhoA/F-actin signaling as essential for STARD13- tumors and hematopoietic malignancies [11–14]. Our correlated ceRNA network-mediated inhibition on breast recent work has confirmed that STARD13, CDH5, CSC formation. We further found that YAP/TAZ were HOXD1, and HOXD10 could co-regulate each other the major downstream factors in this process. Finally, we through competing for several shared miRNA binding indicated that STARD13-correlated ceRNA network en- sites, and thus formed a ceRNA network to coordinately hanced doxorubicin sensitivity in breast cancer cells. inhibit breast cancer EMT and metastasis, which is de- noted as STARD13-correlated ceRNA network . Methods Here, we further seek to explore the mechanisms by Cell culture which STARD13-correlated ceRNA network regulates HEK293T cells and human breast cancer cells MCF-7 EMT and its roles in breast CSC formation. and MDA-MDB-231 were stored in our laboratory. Cell The mammalian Hippo pathway is a kinase cascade in- line authentication was assessed using short tandem re- volving mammalian STE20-like protein kinase 1 (MST1) peat (STR) DNA profiling method every year. HEK293T and the large tumor suppressor 1 (LATS1) and LATS2 and MCF-7 cells were cultured in DMEM medium . LATS1/2, as the core elements of Hippo pathway, (Gibco, Grand Island, NY, USA), and MDA-MB-231 Zheng et al. Journal of Hematology & Oncology (2018) 11:72 Page 3 of 18 cells were cultured in L-15 medium (Gibco) at 37 °C Shanghai, China) following the manufacturer’s protocol. under a humidified atmosphere with 5% CO . Both of Briefly, cells were lysed by NP-40 lysis buffer (Beyotime, the media were supplemented with 10% FBS (Gibco), China). Then, 100 μl cell lysates were incubated with 80 U/ml penicillin, and 0.08 mg/ml streptomycin. NP-40 buffer containing Protein A/G Agarose Resin 4F conjugated with human anti-Ago2 antibody (Cell Signal- Cell transfection ing Technology) at 4 °C overnight. After that, agarose Transfection of plasmids was performed using Lipofecta- beads were isolated by centrifugation and incubated with mine 2000 (Invitrogen, Carlsbad, CA) on MCF-7 cells, protease K to dissociate Ago2-RNA complex from the TransIT-BrCa Transfection Reagent (Mirus, USA) on beads. The RNA fraction precipitated by RIP was ana- MDA-MB-231 cells, and Lentifection (ABM, Vancouver, lyzed by qRT-PCR. Canada) on HEK293T cells. A final concentration of siRNA (GenePharma, China) was 50 nM. Sequences of In vivo tumor initiation and doxorubicin sensitivity assays siRNA against a specific target in this study were listed Four- to six-week male athymic BALB/c nude mice were in Additional file 1: Table S1. purchased from Model Animal Research Center of Nan- jing University and were housed and fed in standard RNA isolation and quantitative real-time PCR analysis pathogen-free conditions. For tumor-limiting dilution Total RNA was extracted by TRIZOL reagent (Invitro- assays, tumor cells were mixed 1:1 with Matrigel matrix gen, USA) according to the manufacturer’s instructions. (BD Biosciences) and orthotopically implanted in the in- qRT-PCR was performed on triplicate samples in a reac- guinal mammary gland of mice. On day 8, all mice were tion mix of SYBR Green (Vazyme, China) with Roche killed, and tumor tissues were collected, weighed, and Real-Time PCR system (Roche, USA). mRNA and fixed in 10% formalin at room temperature and embed- miRNA levels were normalized to GAPDH or U6 sRNA, ded in paraffin for immunohistochemistry or immuno- respectively. The relative expression levels of indicated histofluorescence assay. For doxorubicin sensitivity -ΔΔCt genes were calculated using 2 method. Sequences assay, MCF-7 cells with STARD13 or its ceRNA stable of primers used for qRT-PCR in this study were listed in knockdown or not and MDA-MB-231 cells with Additional file 2: Table S2. STARD13 3′UTR overexpression or not were injected subcutaneously into each flank of 5-week-old BALB/c fe- Immunohistochemistry and immunohistofluorescence male mice. When the tumors reached the volume of ~ assays 100 mm , we randomly allocated the mice to groups in Paraffin-embedded sections were deparaffinized and which they received doxorubicin (0.5 mg/kg, Shenzhen rehydrated, followed by antigen retrieval. After pri- Main Luck Pharmaceuticals Inc., China) or saline. mary and secondary antibody incubation, the slide Tumor growth was monitored by caliper measurements. was finally incubated with diaminobenzidine (DAB) Tumor volume was calculated by the following formula: (Dako, USA) for IHC analysis and observed with the Volume (cubic millimeters) = L (length) × W (width) × confocal microscopy. 1/2. Mice were euthanized 18 days after the inoculation. The weight of each tumor was measured. Immunofluorescence and F-actin visualization The detailed procedure was referred to our previous Plasmid and stable expression cell line constructions study . For stable expression of STARD13-3′UTR, CDH5-3′ UTR, HOXD1-3′UTR, and HOXD10-3′UTR, sequences RhoA GTPase assay of 3′UTRs were subcloned into pLVX-ZsGreen and re- The detailed procedure was referred to our previous ferred as pLVX-ceRNAs-3′UTR. shRNA oligos were study . synthesized by Sangon Co., Ltd. After annealing, double- strand oligos were inserted to lentiviral pLKO.1-Puro Western blot analysis vector (Addgene). To package lentivirus, HEK-293T cells Protein lysates were obtained from cells grown for 48 h were co-transfected with the lentiviral vector described at high density. The Western blot procedure was carried above and packaging vectors psPAX2 and pMDG.G out as described in our previous work . The informa- using Lentifectin (ABM, USA). Cells were infected with tion of primary antibodies were listed in Additional file 3: the virus in the presence of 2 μg/ml polybrene. The in- Table S3. fected cells were selected with puromycin (Sigma, 2 μg/ ml) for 2 weeks. After two rounds of infection, qRT-PCR RNA immunoprecipitation and Western blot analyses were used for verification. RNA immunoprecipitation (RIP) assays were conducted Meanwhile, cells infected with pLVX-ceRNAs-3′UTR using the Protein A/G Agarose Resin 4FF (YEASEN, were selected by fluorescent cell sorting. Zheng et al. Journal of Hematology & Oncology (2018) 11:72 Page 4 of 18 The 3′UTRs of LATS1 and LATS were cloned into Qvalue is smaller, and the degree of enrichment is the luciferase reporter vector (pMIR-Report, more significant. Ambion, Carlsbad, CA, USA), and the corresponding plasmids were denoted as pMIR-LATS1-3′UTR and Patient samples pMIR-LATS2-3′UTR. Sequences of primers used for Paired mRNA profiling data were downloaded from The plasmid construction in this study were listed in Cancer Genome Atlas (TCGA) data portal (http://can- Additional file 4: Table S4. All constructs were con- cergenome.nih.gov). The dataset from the Mixed Tumor firmed by DNA sequencing. Breast – Clynes – 121 – MAS5.0 – u133p2 public (http://hgserver1.amc.nl/cgi-bin/r2/main.cgi), which in- Luciferase reporter assay cludes 121 breast cancer samples, was obtained as a val- Cells were co-transfected with pMIR luciferase reporter idation set. The microarray dataset was deposited in the or β-gal and siRNA using Lipofectamine 2000. Each Gene Expression Omnibus (GEO) (accession number group was run in triplicate in 96-well plates. The lucifer- GSE42568) according to “minimum information about a ase activity was detected by Luciferase Reporter Assay microarray experiment” (MIAME) guidelines. The R2 System (Promega, USA) after 48 h of transfection. Lucif- platform was used to analyze the microarray (http://r2. erase activity was normalized to β-gal activity. amc.nl). Flow cytometric assay Statistical analysis Cells were detached from plates with Accutase (Invi- All data were obtained from at least three independent trogen), resuspended (1 × 10 cells/ml), incubated in experiments (n ≥ 3) and presented as the mean ± SD running buffer (PBS 1×, BSA 0.5%, and EDTA 5 mM) (standard deviation). Statistical analyses were performed with anti-human CD44 (APC-conjugated, BD Biosci- using Student’s t test except for qRT-PCR. Data from the ences) and anti-human CD24 (PE-conjugated, BD Bio- qRT-PCR test were analyzed using one-way analysis of sciences), and finally analyzed on a C6 flow cytometer variance (ANOVA). The differences between the groups (BD Biosciences). Flow cytometry values have been were analyzed using ANOVA with the Tukey-Kramer normalized by subtracting the appropriate isotype post-test. *p < 0.05 was considered statistically signifi- control value. cant. **p < 0.01, ***p < 0.001, and ns indicate no signifi- cant differences from control. Mammosphere formation assay Cells were grown in the MammoCult medium (Stem Results Cell Technologies, Vancouver, Canada) supplemented STARD13-correlated ceRNA network inhibits the with MammoCult Proliferation Supplements (Stem Cell stemness-related traits in breast cancer cells Technologies, Vancouver, Canada) and plated in 24-wells To investigate whether STARD13-correlated ceRNA net- plate with ultra-low attachment at a density of 10,000 vi- work is involved in breast CSC formation, we first de- able cells/ml and grown for 10 days. Mammospheres tected the expression levels of STARD13, CDH5, were counted and photographed. HOXD1, and HOXD10 in a pool of CSCs naturally aris- ing within breast cancer cells. Fluorescence-activated cell RNA sequencing and data analysis sorting (FACS) was used to sort MCF-7 cells based on RNA from MDA-MB-231 and STRARD13 3′UTR- the expressions of cell surface antigen markers CD44 overexpressed MDA-MB-231 cells (MDA-MB-231 (S- +/CD24−, which have been identified as the breast CSC UTR)) was extracted using TRIzol. RNA-seq libraries markers . As shown in Fig. 1a, the qRT-PCR analysis were acquired using TruSeq PE Cluster Kit V4 and revealed that a CD44+/CD24− subpopulation displayed sequenced by HiSeq 2000 sequencer. We used Cuff- lower levels of STARD13-correlated ceRNAs. As previ- diff to estimate fragments per kilobase of transcript ous studies have demonstrated that non-adherent per million (FPKM) values for known transcripts and spheres are highly enriched for CSCs [27, 28], the ex- analyze differentially expressed transcripts. p <0.05 pression levels of STARD13-correlated ceRNAs were ex- was considered as significant. Heatmap of gene ex- amined in non-adherent spheres and parental cells, and pression was generated based on log2 (FPKM) using an identical result was acquired (Fig. 1b). We next tested HemI 18.104.22.168 (http://hemi.biocuckoo.org/down.php). whether STARD13-correlated ceRNAs-3′UTR overex- GO analysis of gene expression changes was per- pression could impair the capacity of tumorsphere for- formed using GO-TermFinder. KEGG enrichment mation. MDA-MB-231 cells stably transfected with analysis was performed according to Rich factor, Qva- STARD13-correlated ceRNAs-3′UTRs or an empty vec- lue, and numbers of enriched genes. The richer factor tor were subjected to non-adherent sphere formation is bigger, and the degree of enrichment is bigger; the assay. The infection efficiency of lentivirus was verified Zheng et al. Journal of Hematology & Oncology (2018) 11:72 Page 5 of 18 Fig. 1 STARD13-correlated ceRNA network inhibits tumor sphere formation capacity of breast cancer in vitro. a, b Detection of STARD13, CDH5, HOXD1, and HOXD10 expressions in sorted CD44+/CD24− cell subpopulation (a) and mammospheres (b) of MCF-7 cells enriched in 3D semi- solid culture medium by qRT-PCR analysis. (c) Phase contract images of mammospheres formed by MDA-MB-231 derivatives (MDA-MB-231 stably overexpressing with STARD13 3′UTR, CDH5 3′UTR, HOXD1 3′UTR, and HOXD10 3′UTR, and their negative control, hereafter indicated as pLVX- STARD13-3′UTR, pLVX-CDH5-3′UTR, pLVX-HOXD1-3′UTR, pLVX-HOXD10-3′UTR, and pLVX-Ctrl, respectively). d Quantification of mammospheres in c. e Representative FACS profile of MDA-MB-231 derivatives described in c with CD24− and CD44+ markers. f, g Examination of stemness-related genes expression (ALDA1, OCT4, and Nanog) by qRT-PCR (f) and Western blot analysis (g) in cells described in c. Data were presented as the mean ± SD, n =3, *p < 0.05, **p < 0.01, ***p < 0.001 vs. vector or control by qRT-PCR assay (Additional file 5: Figure S1A). Both efficiency of shRNAs against STARD13-correlated ceRNAs the sphere size and number were significantly decreased (Additional file 5: Figure S1B). Functionally, the knock- after STARD13-correlated ceRNAs-3′UTR overexpres- down of STARD13-correlated ceRNAs formed more pri- sion (Fig. 1c, d). Additionally, overexpression of mary mammospheres than control cells (Additional file 6: STARD13-correlated ceRNAs-3′UTRs decreased the Figure S2A and S2B) and induced an enlargement of the CD44+/CD24− population by flow cytometry analysis CD44+/CD24− population (Additional file 6:FigureS2C). (Fig. 1e). Moreover, the expression of several pluripotent As expected, the expression of stemness markers was up- transcription factors, namely, Oct3/4, ALDH1, Nanog, regulated in cells with STARD13-correlated ceRNA knock- and Sox2, was decreased in cells with STARD13- or its down (Additional file 6:FigureS2D). Since wefound that ceRNAs-3′UTR overexpression (Fig. 1f, g). We further the CD44+/CD24− subpopulation is more than 90% in tested whether the knockdown of STARD13-correlated MDA-MB-231 cells, and less than 20% in MCF-7 cells, this ceRNAs could confer stemness properties in MCF-7 phenomenon indicated that MDA-MB-231 and MCF-7 cells. Western blot assay confirmed the knockdown cells held high CSC-related traits and low CSC-related Zheng et al. Journal of Hematology & Oncology (2018) 11:72 Page 6 of 18 traits, respectively. Thus, MDA-MB-231 and MCF-7 cells Ki67, which is required for the maintenance of CSCs were used for overexpression and knockdown experiments, , were decreased in tumors derived from STARD13- respectively. correlated ceRNAs-3′UTR-overexpressed cells (Fig. 2b). Additionally, we further performed in vivo tumorigenic STARD13-correlated ceRNA network inhibits CSC traits of assay with MCF-7 cells after STARD13-correlated breast cancer cells in vivo ceRNA knockdown and indicated that the knockdown of We further evaluated whether STARD13-correlated STARD13-correlated ceRNAs held much stronger ceRNA network regulates tumor-initiating potential of tumor-initiating potentials as compared with control breast cancer cells in vivo. We compared the capacity of cells (Additional file 7: Figure S3D and S3E and Fig. 2c) STARD13-correlated ceRNAs-3′UTR-overexpressed and increased staining intensity and number of Ki67 MDA-MB-231 cells to seed tumors at limiting dilutions. (Fig. 2d). Taken together, these results indicate that Although all cell lines could form tumors at a density of STARD13-correlated ceRNA network inhibits the tumor 1×10 cells, STARD13-correlated ceRNAs-3′UTR-over- initiation ability of breast cancer. expressed cells showed a decrease of tumor size and weight (Additional file 7: Figure S3A and Fig. 2a). Not- STARD13-correlated ceRNA network activates Hippo ably, the tumor-seeding ability of STARD13-correlated signaling ceRNAs-3′UTR-overexpressed cells was significantly de- To explore the related mechanisms contributing to 5 4 creased at a density of 1 × 10 and 1 × 10 cells STARD13-correlated ceRNA network in prohibiting (Additional file 7: Figure S3B and S3C and Fig. 2a). breast cancer CSC formation, we tried to characterize Concordantly, the staining intensity and number of the pathways regulated by STARD13-correlated ceRNA Fig. 2 STARD13-correlated ceRNA network inhibits CSC traits of breast cancer cells in vivo. a TIC (tumor-initiating cell) frequencies (left), χ values, and associated probabilities (right) of MDA-MB-231 with STARD13-correlated ceRNAs-3′UTR overexpression. Limiting dilution analysis was performed using ELDA software. b Represented images of Ki67 staining of tumors harvested when 1,000,000 cells were injected in a. c TIC frequencies (left), χ values, and associated probabilities (right) of MCF-7 derivatives stably depleted of STARD13 and its ceRNAs (hereafter indicated as pLKO.Ctrl, pLKO.STARD13, pLKO.CDH5, pLKO.HOXD1, and pLKO.HOXD10). Limiting dilution analysis was performed using ELDA software. d Represented images of Ki67 staining of tumors harvested when 1,000,000 cells were injected in c Zheng et al. Journal of Hematology & Oncology (2018) 11:72 Page 7 of 18 network. Firstly, we analyzed the transcriptome of and the target of YAP (Fig. 4a, c and Additional file 9: MDA-MB-231 cells with STARD13-3′UTR stable ex- Figure S5). Importantly, among 52 YAP-regulated genes pression or not based on RNA sequencing data. Hippo examined, 25 genes were commonly reduced when over- signaling was shown on the top of the most upregulated expressing STARD13-3′UTR, the core member of pathways by STARD13-3′UTR overexpression (Fig. 3a), STARD13-correlated ceRNA network (Fig. 4d). Con- and LATS1/2 expression was markedly increased in versely, STARD13-correlated ceRNA knockdown re- STARD13-3′UTR-overexpressed cells (Fig. 3b). By ana- sulted in YAP/TAZ dephosphorylation and upregulation lyzing the mRNA microarrays from TCGA, we deter- of CTGF expression in MCF-7 cells (Fig. 4b). Collect- mined that LATS1/2 level was positively correlated with ively, our results indicate that STARD13-correlated STARD13, CDH5, HOXD1, and HOXD10 levels in ceRNA network could activate Hippo signaling. breast cancer tissues, respectively (Additional file 8: Fig- ure S4A). And the ceRNA sequences and the genomic STARD13-correlated ceRNA network acts as a sub-ceRNA locus were denoted in Additional file 8: Figure S4B. network to activate Hippo signaling LATS1/2 is the key component of Hippo signaling path- We sought to investigate the mechanisms by which way, consisting of a core kinase cascade of MST1/2, STARD13-correlated ceRNA network regulates Hippo LATS1/2, and downstream effectors YAP/TAZ. Consist- signaling. By computational prediction, LATS1/2-3′UTR ently, ectopic expression of STARD13-correlated was shown to share 16 miRNA binding sites with ceRNAs-3′UTRs in breast cancer cells elevated LATS1/ STARD13-3′UTR (Additional file 10: Table S5). This re- 2 levels, while the knockdown of STARD13-correlated sult was provocative because the previous studies have ceRNAs decreased LATS1/2 levels (Figs. 3c, d and 4a, indicated that the more miRNAs are shared, the more b). We also examined the phosphorylation level of YAP/ probability of ceRNA network exists [32–34], which led TAZ on Ser-127/Ser-66, which is phosphorylated by us to examine whether STARD13-3′UTR could activate LATS1/2 and associated to YAP/TAZ cytoplasmic reten- Hippo signaling by acting as a ceRNA for LATS1/2 and tion , thus being sequestered in the cytoplasm by 14- thus regulate LATS1/2 expression. To validate this hy- 3-3 proteins . Enforcing the expression of STARD13- pothesis, we examined several predicted shared miRNAs correlated ceRNAs-3′UTRs in MDA-MB-231 cells in- using 3′UTR-luciferase reporter assays. Five miRNAs creased the phosphorylation of YAP/TAZ and their cyto- (miR-424, miR-374a, miR-590-3p, miR-448, and miR- plasm retention and dampened the expression of CTGF 15a) significantly repressed the luciferase activity of Fig. 3 STARD13-correlated ceRNA network activates Hippo signaling via increasing LATS1/2 expression . a, b Enrichment of signaling signatures differentially expressed (a) and identification of LATS1/2 expression (b) between pLVX-Ctrl and pLVX-STARD13-3′UTR based on RNA-sequencing analysis. c, d LATS1/2 was downregulated when STARD13-correlated ceRNAs were depleted and upregulated when overexpressing STARD13- correlated ceRNAs-3′UTRs in MCF-7 (c) and MDA-MB-231 cells (d). Data were presented as the mean ± SD, n =3, **p < 0.01 vs. vector or control Zheng et al. Journal of Hematology & Oncology (2018) 11:72 Page 8 of 18 Fig. 4 STARD13-correlated ceRNA network activates Hippo signaling via increasing LATS1/2 expression . a, b Western blot analysis of lysates from MDA-MB-231 cells overexpressing STARD13-correlated ceRNAs-3′UTRs (a) and MCF-7 cells depleted of STARD13-correlated ceRNAs (b). c Confocal images showing localization of YAP in MDA-MB-231 cells overexpressing STARD13-correlated ceRNAs-3′UTRs. d Expression of YAP targeting genes in pLVX-Ctrl and pLVX-STARD13-3′UTR cells based on RNA-sequencing analysis. Data were presented as the mean ± SD, n =3, **p < 0.01 vs. vector or control STARD13- or LATS1/2-3′UTR-luciferase reporters (ex- STARD13-correlated ceRNAs reduced the luciferase ac- cept for miR-448 on LATS1) (Fig. 5a), suggesting that tivity of pMIR-LATS1/2-3′UTR; this effect was attenu- the crosstalk between STARD13 and LATS1/2 may be at ated by knocking down Dicer enzyme (Fig. 5c). Those least in part by these five miRNAs. In the light of the data further supported the idea that LATS1/2 could be crucial role of miRNAs in ceRNA networks, we sought regulated by STARD13-correlated ceRNA 3′UTRs to knockdown Dicer, an important enzyme for miRNA through competing for shared miRNAs. To further con- biogenesis . The knockdown of STARD13-correlated firm the direct interaction between these five miRNAs ceRNAs failed to regulate LATS1/2 protein level when and STARD13 or LATS1/2 at endogenous levels, we per- Dicer was knocked down (Fig. 5b). Notably, YAP/TAZ formed RIP analysis to pull down endogenous miRNAs dephosphorylation induced by STARD13-correlated associated with Ago2 in STARD13- and LATS1/2-3′ ceRNA knockdown was abolished by knocking down UTR-overexpressed cells. The precipitated miRNAs were Dicer enzyme in MCF-7 cells (Fig. 5b), suggesting that subjected to qRT-PCR analysis, and the results showed the regulation is miRNA dependent. To further validate that these five miRNAs were enriched in RNAs retrieved that STARD13-correlated ceRNAs share miRNA-binding from STARD13- and LATS1/2-3′UTR-overexpressed sites at 3′UTR of LATS1/2 transcripts, we conducted lu- cells (Fig. 5d), supporting that STARD13 and LATS1/2 ciferase reporter assays to examine whether STARD13- were the bona fide targets of these five miRNAs. The correlated ceRNAs could compete for the miRNAs tar- qRT-PCR analysis combined with an internal standard geting at LATS1/2 3′UTR. The knockdown of curve revealed that LATS1 and LATS2 were Zheng et al. Journal of Hematology & Oncology (2018) 11:72 Page 9 of 18 Fig. 5 STARD13-correlated ceRNA network acts as a sub-ceRNA network to increase LATS1/2 expression. a Luciferase activity of pMIR reporter containing LATS1- or LATS2- or STARD13-3′UTR was detected upon co-transfecting with indicated miRNA mimics in 293T cells. pMIR-reporter empty vector was used as positive control. Data are presented as the ratio of luciferase activity to β-gal activity. b Western blot analysis of lysates from MCF-7 cells with STARD13-correlated ceRNA knockdown plus a control or anti-Dicer siRNA co-transfection. Depletion of Dicer rescued the expressions of LATS1/2 and downstream YAP/TAZ phosphorylation in STARD13-correlated ceRNA knocked down MCF-7 cells. c Luciferase activity of pMIR reporters which contained LATS1 (left panel) or LATS2 (right panel) 3′UTR with indicated treatment in MCF-7 cells. d qRT-PCR was used to measure the level of enrichment of miRNAs in the anti-Ago2-immunoprecipitated complexes in MCF-7 cells with empty vector or pMIR-LATS1- 3′UTR or pMIR-LATS2-3′UTR or pLVX-STARD13-3′UTR transfection. miR-17 which did not hold binding sites on LATS1/2 and STARD13 was served as a negative control. e Copy number of LATS1 (left) and LATS2 (right) per cell was measured in MCF-7 cells calibrated with an internal standard curve of synthetic constructs containing LATS1 3′UTR or LATS2 3′UTR. f RIP assay of enrichment of STARD13, LATS1, and LATS2 transcripts on Ago2 relative to input in MDA-MB-231 cells with STARD13-correlated ceRNAs-3′UTR overexpression. Data were presented as the mean ± SD, n =3, *p < 0.05, **p < 0.01 vs. control/vector 9 10 expressed at 8.82 × 10 and 3.02 × 10 in MCF-7 cells complex. Overexpression of STARD13-correlated 8 9 and 6.02 × 10 and 3.34 × 10 in MDA-MB-231 cells, ceRNAs-3′UTRs resulted in a substantial decrease of respectively (Fig. 5e). As our previous study has LATS1/2 recruitment to Ago2 (Fig. 5f). Notably, the above shown that STARD13 was expressed at 1.22 × 10 five miRNA mimics (miR-424, miR-374a, miR-590- and 8.06 × 10 in MCF-7 and MDA-MB-231 cells, re- 3p,miR-448, and miR-15a) were transfected into MDA- spectively . These results are consistent with the MB-231 cells with STARD13-correlated ceRNAs-3′UTR previous reports that ceRNA interaction is optimal overexpression, and we found that the overexpression when the transcript abundance of ceRNAs within a of these target miRNAs abolished the effects of network is near equimolarity [33, 35, 36]. Next, to as- STARD13-correlated ceRNA network on LATS1/2 sess directly whether STARD13-3′UTR induces and its downstream effectors p-YAP/p-TAZ expression LATS1/2 expression through competing for miRNAs, (Additional file 11:FigureS6).Thus, theseresults we performed RIP assays against Ago2 to address suggest that STARD13-correlated ceRNAs release whether enforced expression of STARD13-3′UTR LATS1/2 transcripts from the post-transcriptional re- coulddisplaceLATS1/2 transcriptsawayfromRISC pression mediated by RISC complexes. Zheng et al. Journal of Hematology & Oncology (2018) 11:72 Page 10 of 18 Furthermore, we sought to confirm the central role of metastasis by suppressing EMT process . It led STARD13 in this ceRNA network, namely, STARD13 us to explore that whether the inhibition of EMT ceRNAs (CDH5, HOXD1, and HOXD10)-3′UTRs regu- caused by STARD13-correlated ceRNA network is late LATS1/2 expression through STARD13. As ex- mediated through its effects on LATS1/2 expression. pected, the upregulation of LATS1/2 and p-YAP/TAZ Overexpression of STARD13-correlated ceRNAs im- levels and downregulation of YAP target and CTGF ex- paired mesenchymal marker expression such as N- pression level, both were induced by STARD13 cadherin, MMP-9, α-SMA,and vimentin,and these ceRNAs-3′UTR overexpression, were attenuated by effects were attenuated or reversed by knocking STARD13 knockdown, while the overexpression of down LATS1 or LATS2 separately or simultaneously STARD13-3′UTR decreased or even reversed the (Additional file 14:FigureS9).These data suggest knockdown of STARD13 ceRNA-caused downregula- that STARD13-correlated ceRNA network suppresses tion of LATS1/2 and p-YAP/TAZ levels and upregula- breast cancer EMT and CSC traits at least partly tion of CTGF level (Additional file 12:Figure S7A and through Hippo signaling pathway. S7B). Consistently, the impaired mammary sphere for- mation capacity resulted from CDH5-, HOXD1-, and STARD13-correlated ceRNA network regulates YAP HOXD10-3′UTR overexpression was attenuated or re- nuclear abundance by inhibiting RhoA-ROCK signaling versed by STARD13 knockdown (Additional file 12: As the previous studies have shown that knockdown of Figure S7C). Additionally, the shrinking of CD44 LATS had marginal effects on YAP/TAZ signaling in +/CD24− subpopulation caused by CDH5-, HOXD1-, dense cultures [21, 25] which seems to contradict with and HOXD10-3′UTR overexpression was attenuated or our results that knockdown of STARD13-correlated ceR- reversed after STARD13 depletion (Additional file 12: NAs could rescue YAP/TAZ nuclear localization in Figure S7D). We further tested whether this interaction dense monolayers (Fig. 7a and Additional file 15: Figure occurred in vivo as well. Expectedly, the protein levels S10A). Importantly, a mutant form of YAP (YAP-5SA) of LATS1/2 were remarkably upregulated in the tumors was transiently transfected in MDA-MB-231 cells to es- derived from the STARD13-correlated ceRNAs-3′UTR- cape LATS1/2-mediated phosphorylation. YAP-5SA was overexpressing cells (Additional file 12: Figure S7E) via still sensitive to with STARD13-correlated ceRNAs-3′ immunohistofluorescence stain assay, and the overex- UTR overexpression (Fig. 7b), suggesting that pression of STARD13-correlated ceRNAs-3′UTRs ele- STARD13-correlated ceRNA network could regulate vated the phosphorylation level of YAP through YAP subcellular localization in a LATS1/2-independent immunohistochemistry analysis (Additional file 12: manner. These results promote us to explore whether Figure S7E). All those data testify our hypothesis that STARD13-correlated ceRNA network regulates YAP nu- STARD13-correlated ceRNA network regulates LATS1/ clear localization via another signaling pathway inde- 2 expression through acting as a sub-ceRNA network pendent of LATS1/2. STARD13 was demonstrated by for LATS1/2. our groups and others to block RhoA-ROCK signaling axis by acting as a Rho GTPase-activating proteins STARD13-correlated ceRNA network regulates breast CSC (GAP), thus disorganizing F-actin structures . More- traits and EMT through LATS1/2 over, the fact that LATS and F-actin organization act in- We then asked whether STARD13-correlated ceRNA dependently to regulate YAP/TAZ is also supported by network modulates breast cancer cells stemness genetic evidence in Drosophila . Thus, we conclude through LATS1/2. LATS1/2 was knocked down in that STARD13-correlated ceRNA network might also MDA-MB-231 cells with STARD13-correlated regulate YAP/TAZ activity through RhoA-ROCK path- ceRNAs-3′UTR overexpressing by lentivirus shRNAs way. As expected, Glisa assay indicated that enforcing infection (Fig. 6a, b). Additionally, both the upregu- STARD13-correlated ceRNAs-3′UTR expressions in lation of p-YAP/TAZ level and the downregulation MDA-MB-231 cells reduced Rho-GTPase activity (Fig. of its target (CTGF) induced by STARD13- 7c). In line with this, MLC phosphorylation was damp- correlated ceRNAs-3′UTR overexpression were at- ened, but RhoA expression was unaffected (Fig. 7d), and tenuated by LATS1/2 knockdown (Fig. 6a, b). impressive disorganization of F-actin occurred in LATS1/2 knockdown attenuated the shrinking of the STARD13- or its ceRNAs-3′UTR-overexpressed MDA- mammary sphere formation and CD44+/CD24− sub- MB-231 cells (Fig. 7e). Notably, cytoplasm abundance of population, which were both resulted from YAP/TAZ was recovered by ROCK inhibitor in MCF-7 STARD13-correlated ceRNAs-3′UTR overexpression cells (Fig. 7f and Additional file 15: Figure S10B), and (Fig. 6c, d and Additional file 13:FigureS8).Our ROCK inhibitor could also facilitate the cytoplasmic re- previous study has shown that STARD13-correlated tention of YAP-5SA in MDA-MB-231 cells (Fig. 7g), in- ceRNA network could inhibit breast cancer dicating STARD13-correlated ceRNA network regulates Zheng et al. Journal of Hematology & Oncology (2018) 11:72 Page 11 of 18 Fig. 6 STARD13-correlated ceRNA network regulates breast CSC traits through LATS1/2. a and b Western blot analysis of lysates from MDA-MB- 231 with STARD13- or its ceRNAs-3′UTR overexpression plus LATS1 or LATS2 or LATS1/2 knockdown by lentiviral infection. c Phase contract images of mammospheres formed by MDA-MB-231 cells described in a. Depletion of LATS1 or LATS2 or LATS1/2 rescued the mammosphere formation impaired by enforced STARD13-correlated ceRNAs-3′UTR expression in MDA-MB-231 cells. d Quantification of mammosphere formation in c. Data were presented as the mean ± SD, n =3, *p <0.05, **p < 0.01 vs. pLVX-Ctrl YAP/TAZ nuclear translocation at least partly through cancer cells to doxorubicin. Therefore, we sought to RhoA/ROCK/F-actin axis. Moreover, ROCK inhibitor examine the doxorubicin response in MDA-MB-231 decreased the CD44+/CD24− subpopulation and the cells with STARD13- or ceRNAs-3′UTR overexpres- promotion of STARD13 knockdown on CD44+/CD24− sion. We noticed that the overexpression of subpopulation (Additional file 15: Figure S10C). In con- STARD13- or its ceRNAs-3′UTR decreased the ac- clusion, our data provide evidence that STARD13- tivity of multidrug resistance (MDR) proteins, as re- correlated ceRNA network could regulate YAP/TAZ nu- vealed by the enlargement of SP (Fig. 8a), and clear abundance in a LATS-independent manner, but enhanced the cell sensitivity to doxorubicin charac- through disorganizing F-actin formation by inhibiting terized as the decrease of IC value (Fig. 8b). Add- RhoA-ROCK signaling. itionally, Pgp, an MDR protein, was also downregulated in MDA-MB-231 cells with STARD13-correlated ceRNA network sensitizes breast STARD13- or ceRNAs-3′UTR overexpression (Fig. cancer cells to doxorubicin 8c). Notably, the overexpression of STARD13- or We have established the inhibitory role of STARD13- ceRNAs-3′UTR enhanced the intake of doxorubicin correlated ceRNA network in conferring CSC traits in MDA-MB-231 cells (Fig. 8d). On the contrary, to breast cancer cells. Conferring CSC traits was well STARD13orits ceRNA knockdowndecreased the confirmed to endow chemoresistance to tumor cells cell sensitivity to doxorubicin characterized as the . It led us to speculate whether STARD13- increase of IC value (Additional file 16:Figure correlated ceRNA network could also sensitize breast S11A) and upregulated Pgp expression levels Zheng et al. Journal of Hematology & Oncology (2018) 11:72 Page 12 of 18 Fig. 7 STARD13-correlated ceRNA network regulates YAP nuclear abundance by inhibiting RhoA-ROCK signaling. a Confocal images of YAP distribution in MCF-7 cells with STARD13-correlated ceRNAs knockdown. b Confocal images of YAP-5SA distribution in MDA-MB-231 cells with STARD13-correlated ceRNAs-3′UTR overexpression or not. Overexpression of STARD13-correlated ceRNAs-3′UTRs could promote YAP- 5SA cytoplasmic retention even when it could not be phosphorylated by LATS1/2. c Glisa assay of MDA-MB-231 cells with STARD13- correlated ceRNAs-3′UTR overexpression. Overexpression of STARD13-correlated ceRNAs-3′UTRs impaired RhoA activity in MDA-MB-231 cells. d Western blot assay of lysates from MDA-MB-231 cells described in c. Overexpression of STARD13-correlated ceRNAs-3′UTRs in MDA-MB- 231 cells dephosphorylated MLC. e Confocal images of filamentous actin (F-actin) stained with rhodamine-labeled phalloidin in MDA-MB- 231 cells described in c. f Confocal images of YAP distribution in MCF-7 cells with STARD13-correlated ceRNA knockdown plus ROCK inhibitor (Y-27632) treatment or not. g Confocal images of YAP-5SA distribution in MDA-MB-231 cells with ROCK inhibitor (Y-27632) treatment or not. Data were presented as the mean ± SD, n =3, **p < 0.01 vs. pLVX-Ctrl (Additional file 16: Figure S11B) in MCF-7 cells. Im- knockdown (Additional file 16: Figure S11F). Thus, portantly, STARD13 or its ceRNA knockdown de- we confirm that STARD13-correlated ceRNA net- creased doxorubicin sensitivity (Fig. 8e–h), while the work could increase the sensitivity of breast cancer overexpression of STARD13 3′UTR enhanced doxo- cells to doxorubicin in ceRNA-dependent. rubicin sensitivity in vivo (Additional file 16:Figure To sum up, our results demonstrate that STARD13- S11C–E). Consistently, the intake of doxorubicin in correlated ceRNA network acts as a potential inhibi- MCF-7 cells was dampened under the absence of tor of YAP/TAZ transcriptional activity at least STARD13-correlated ceRNAs (Additional file 16: through two concurring pathways: one was by facili- Figure S11G). However, the increase of Pgp expres- tating LATS1/2 expression and the other by blocking sion induced by STARD13 or its ceRNA knockdown RhoA-F-actin signaling cascade. This inhibition results was attenuated or even reversed by Dicer in EMT prohibition and thus inhibits breast CSC Zheng et al. Journal of Hematology & Oncology (2018) 11:72 Page 13 of 18 Fig. 8 STARD13-correlated ceRNA network enhances the sensitivity of breast cancer cells to doxorubicin. a Representative profile of side population (SP) fractions of MDA-MB-231 cells with STARD13-correlated ceRNAs-3′UTR overexpression stained with Hoesch 33342. b IC curves of MDA-MB-231 cells with STARD13-correlated ceRNAs-3′UTR overexpression and were fitted with a nonlinear regression model and were presented as log (doxorubicin) vs. cell viability. c Pgp protein expression was detected in cells described in a. d Confocal images of MDA-MB-231 cells described in a with doxorubicin treatment. Overexpression of STARD13-correlated ceRNAs-3′UTRs promoted the cellular retention of doxorubicin. e Images of tumors harvested when STARD13- correlated ceRNA stable knockdown cells were planted and followed by doxorubicin treatment or not. f Thevolumeof tumorsharvested in e was monitored. g The weight of tumors harvested in e was monitored. h The weight of mice depicted in e was monitored. Data were presented as the mean ±SD, n =3, **p <0.01 vs. pLVX-Ctrl formation, and then enhances doxorubicin sensitivity downregulated in both breast CSC subpopulation sorted (Fig. 9). by cell surface stemness markers and breast cancer spheres; (2) STARD13-correlated ceRNA network in- Discussion hibits CSC traits of breast cancer in vitro and in vivo; STARD13-correlated ceRNA network is a molecular and (3) our previous study has shown that STARD13- determinant of CSC traits in breast cancer correlated ceRNA network activity is a clinically relevant Here, we revealed that STARD13-correlated ceRNA net- tool to predict the proclivity to develop metastasis and work attenuated breast CSC formation. This conclusion EMT, itself another hallmark of CSC activity . is supported by the following evidence: (1) the mRNA Functionally, we indicated that activating STRAD13- levels of STARD13-correlated ceRNAs are significantly correlated ceRNA network inhibited CSC-related traits, Zheng et al. Journal of Hematology & Oncology (2018) 11:72 Page 14 of 18 Fig. 9 Proposed model that STARD13-correlated ceRNA network inhibits breast cancer EMT and CSC formation, and thus drug resistance. STARD13-correalted ceRNA network acts as a potential inhibitor of YAP transcriptional activity at least through two concurring pathways: one is by facilitating LATS1/2 expression via ceRNA interaction and the other is by blocking RhoA-F-actin signaling cascade through Rho GTPase activity of STARD13. This inhibition results in EMT prohibition and inhibits breast CSC formation, and thus enhances drug sensitivity i.e., tumor sphere formation and tumor initiation prop- larger ceRNA network, and sub-networks contained erties of breast cancer cells while depleting STARD13 or within larger networks could explain the pronounced its ceRNAs promoted these CSC-related traits. Thus, the crosstalk alterations observed when distal potent ceR- activity of STARD13-correlated ceRNA network has NAs are silenced , such as we have previously dem- been linked to breast CSCs, i.e., tumor seeding. Interest- onstrated that the ceRNA network between pseudogene ingly, the previous study has shown that normal and CYP4Z2P and CYP4Z1 could enhance ERα transcrip- non-stem cells could spontaneously revert to a stem-like tional activity, thus conferring tamoxifen resistance in state under appropriate intrinsic or extrinsic cues . MCF-7 cells via acting as a sub-network for CDK3 in The unsettled issue is whether the effects of STARD13- ER-positive breast cancer . We further showed that correlated ceRNA network on breast CSCs also exist in this ceRNA network exerts an anti-apoptotic function non-stem cells, such as mammary epithelial cells. via acting as a sub-network for hTERT in breast cancer cells . In this work, we showed that STARD13- A mechanism that links STARD13-correlated ceRNA correlated ceRNA network acted as a sub-network for network to Hippo signaling LATS1/2 and thus promoted LATS1/2 activity, demon- In addition, we have revealed the STARD13-correlated strating a role of this ceRNA network in modulating ceRNA network as a deterministic upstream controller Hippo pathway. These molecular and cellular findings of YAP/TAZ transcriptional activity. Recent studies have are corroborated by the results that CDH5-, HOXD1-, uncovered ceRNA network as a crucial modulator in hu- and HOXD10-3′UTR could regulate LATS1/2 activity man cancer [38, 39]. Characterizing the identity, func- and downstream Hippo signaling in an miRNA- tion, and mechanism of the ceRNAs will not only dependent manner. However, the luciferase reporter further our fundamental understanding of RNA- assay showed that miR-448 did not decrease the lucifer- mediated cancer pathogenesis but may also shed light ase activity of LATS1 (Fig. 5a), which is in accordance on the development of new RNA-based therapeutic with the result of RIP assay with Ago2 (Fig. 5d). These strategies for treating cancer . Notably, extensive results indicate that the ceRNA interaction between ceRNA-ceRNA interaction networks could regulate de- STARD13-correlated ceRNA network and LATS1 is at velopment in multiple rhesus tissues . Such indirect least through the other four miRNAs. Of note, CDH5-, interactions critically result in ceRNA crosstalk in a HOXD1-, and HOXD10-3′UTRs could not work in this Zheng et al. Journal of Hematology & Oncology (2018) 11:72 Page 15 of 18 way without STARD13, suggesting a key and bridge role It is worth noting that most studies to date have of STARD13 in this larger ceRNA network, which com- pointed to YAP/TAZ in the regulation of the Hippo municated its ceRNAs (CDH5, HOXD1, and HOXD10) pathway or Rho-GTPase/F-actin pathway alone. Among with LATS1/2, the crucial member of Hippo signaling the numerous upstream signals, mechanical inputs re- cascade. In line with this, the absence of LATS1/2 atten- main a central pillar in the control of YAP/TAZ activity uated the inhibitory roles of STARD13-correlated  and represent a prerequisite for it. And this is testi- ceRNA network on CSC and EMT traits, highlighting fied by the fact that the loss of LATS1/2 fails to revive that STARD13-correlated ceRNA network exerts its YAP/TAZ activity in the cells that are mechanically tumor-suppressive effects at least partly through regulat- inhibited [25, 46]. By identifying LATS1/2 and Rho- ing LATS1/2. Notably, genetically depleting LATS1/2 in GTPase as two synergistic instruments of STARD13- MCF-7 cells did not additionally attenuate the inhibitory correlated ceRNA network in regulating breast CSC effects of STARD13-correlated ceRNA network com- traits, our findings forge a novel and unanticipated col- pared to depleting LATS1 or LATS2 alone, demonstrat- laboration between oncogenic Rho-GTPase/F-actin ing that alternative activation of LATS1 or LATS2 is function and the tumor-suppressive Hippo pathway in sufficient to inhibit YAP activity and downstream effects. breast cancer via ceRNA crosstalking. To sum up, since STARD13-correlated ceRNA net- A mechanism that links STARD13-correlated ceRNA work could simultaneously promote LATS1/2 (the sup- network to Rho-GTPase/F-actin signaling pressor of YAP/TAZ transcriptional activity) expression, Data from multiple groups, including our own, have and inhibit Rho-GTPase/F-actin signaling cascade (the shown that STARD13 could serve as a Rho GTPase- derepressor of YAP/TAZ nuclear translocation), it might activating protein (GAP) to inhibit Rho-GTPases and exert a more profound regulation on YAP/TAZ activity thus RhoA activity, resulting in the reorganization of the than either of those factors alone. Accordingly, one may cytoskeleton [37, 43]. Previous studies have indicated propose a therapeutic strategy driving STARD13- that Rho-GTPase and F-actin rearrangements are re- correlated ceRNA network considering its comprehen- quired for YAP/TAZ activity [19, 23]. Galphaq stimu- sive effects on YAP transcriptional activity. lates YAP/TAZ activity through a Trio-Rho/Rac signaling circuitry promoting actin polymerization, inde- STARD13-correlated ceRNA network modulates pendently of the canonical Hippo pathway . Add- doxorubicin sensitivity via regulating YAP/TAZ activity itionally, the regulators of Rho-GTPase could modulate Increasing evidence has demonstrated that CSCs play cell shape and YAP/TAZ localization in triple negative vital roles in breast cancer metastasis, relapse, and che- breast cancer . YAP/TAZ responds to mechanical moresistance. Although traditional therapies work in the signals from the surrounding ECM, which informs cells early treatment of breast cancers, they fail to target and the need to preserve stem cell attributes depending on wipe out the CSCs, which contribute to chemoresistance tissue needs, and mechanotransduction is intimately and tumor recurrence . Notably, salinomycin has interwoven with cytoskeletal dynamics . Therefore, been identified to specifically target breast CSCs and we speculated that STARD13-mediated inhibition of thus effectively prohibit breast cancer progression . Rho-GTPase activity could also regulate YAP/TAZ activ- Additionally, mifepristone, a drug regularly used for ity. Since STARD13-correlated ceRNAs-3′UTRs could abortion, has been shown to inhibit triple negative breast promote STARD13 expression via their ceRNA activities, cancer growth via reducing breast CSC population . we strongly believe that STARD13-correlated ceRNA Therefore, it is an urgent need to develop more therap- network could also regulate YAP/TAZ activity via inhi- ies specifically targeting CSCs for breast cancer treat- biting Rho-GTPase/F-actin signaling. Consistently, Glisa ment, especially for triple negative breast cancer which assay confirmed that STARD13-correlated ceRNA net- innately holds a poorly differentiated “stem/progenitor” work reduced Rho-GTPase activity, and impressive cell phenotype . Attributing responses to CSCs is disorganization of F-actin occurred in STARD13- or its also recommended as the key to accelerating an under- ceRNAs-3′UTR-overexpressed MDA-MB-231 cells; this standing of their biology and developing more effective result is also consistent with our recent work . Im- methods for their eradication in patients . portantly, cytoplasm abundance of YAP/TAZ and YAP- In this work, we showed that STARD13-correlated S5A was recovered by activating STARD13-correlated ceRNA network could regulate CSC traits of breast can- ceRNA network or Rock inhibitor. These results sub- cer cells through two independent pathways, which col- stantiate our speculation that STARD13-correlated laboratively led to the nucleus-cytoplasm translocation ceRNA network regulates YAP/TAZ activity at least of YAP/TAZ. Despite the more details that need to be partly through Rho-GTPase/F-actin signaling and be- elucidated, we proposed that verteporfin, an inhibitor of yond through Hippo pathway. YAP-TEAD binding, might target breast CSCs and could Zheng et al. Journal of Hematology & Oncology (2018) 11:72 Page 16 of 18 be used as a combinative treatment with doxorubicin in Additional file 12: Figure S7. CDH5, HOXD1, HOXD10-3′UTRs regulate Hippo breast cancer therapy. However, we must admit that signaling and CSC traits of breast cancer cells through STARD13. (A) Western blot analysis of lysates from MCF-7 cells with its ceRNA knockdown plus STARD13 3′ other first-line drugs in breast cancer treatment were UTR co-transfection or not. (B) Western blot analysis of lysates from MDA-MB-231 not examined in this work, which could be performed in cells with STARD13 ceRNAs-3′UTR overexpression plus STARD13 knockdown or our future work. not. (C and D) Phase contract images of mammospheres (C) formed by MDA- MB-231 cells described in (A). Representative FACS profile (D) of them with CD24 − and CD44+ markers by flow cytometry analysis. (E) Represented images of p- Conclusions YAP, LATS1, and LATS2 staining of tumors harvested when 1,000,000 cells were Our work reveals an unpredicted layer of YAP regulation injected in Fig. 2a. (TIF 5010 kb) and put the activation of STARD13-correlated ceRNA Additional file 13: Figure S8. Representative FACS profile of MDA-MB- 231 cells with STARD13- or its ceRNAs-3′UTR overexpression plus LATS1 network as a potential novel therapeutic strategy to spe- or LATS2 or LATS1/2 knockdown by lentiviral infection. (TIF 1568 kb) cifically target breast CSCs. Additional file 14: Figure S9. STARD13-correlated ceRNA network regulates breast cancer EMT through LATS1/2. EMT marker (see in main text) expressions were measured in MDA-MB-231 cells with STARD13- or its Additional files ceRNAs-3′UTR overexpression plus LATS1 or LATS2 or LATS1/2 knockdown by lentiviral infection. (TIF 2846 kb) Additional file 1: Table S1. Sequences of siRNA against specific target Additional file 15: Figure S10. STARD13-correlated ceRNA network in this study. (DOC 34 kb) regulate TAZ nuclear abundance by inhibiting RhoA-ROCK signaling. (A) Additional file 2: Table S2. Sequences of primers used for qRT-PCR in Confocal images of TAZ distribution in MCF-7 cells with STARD13- this study (DOC 44 kb) correlated ceRNA knockdown. (B) Confocal images of TAZ distribution in MCF-7 cells with STARD13-correlated ceRNA knockdown plus ROCK Additional file 3: Table S3. Primary antibodies used in this study. (DOC 36 kb) inhibitor (Y-27632) treatment or not. (C) Representative FACS profile of MCF-7 cells with STARD13 knockdown plus ROCK inhibitor treatment. Additional file 4: Table S4. Sequences of primers used for plasmid Data were presented as the mean ± SD, n =3, *p < 0.05, **p <0.01 VS. constructions. (DOC 41 kb) Control or Rock inhibitor. (TIF 5007 kb) Additional file 5: Figure S1. The infection efficiency of lentivirus. (A) Additional file 16: Figure S11. Depletion of STARD13-correlated ceR- Lentiviral infection efficiency of MDA-MB-231 cells stably expressing NAs dampens the response of breast cancer cells to doxorubicin. STARD13-3′UTR, CDH5-3′UTR, HOXD1-3′UTR, and HOXD10-3′UTR was ex- (A) IC curves of MCF-7 cells with STARD13-correlated ceRNA amined by qRT-PCR. (B) Lentiviral infection efficiency of MCF-7 cells stably knockdown and were fitted with a nonlinear regression model and depleted of STARD13, CDH5, HOXD1, and HOXD10 was verified by West- were presented as log (Doxorubicin) vs cell viability. (B) Western ern blot analysis. Data were presented as the mean ± SD, n = 3, ***p < blot assay of lysates from MCF-7 cells with STARD13-correlated 0.001 vs. Ctrl. (TIF 1373 kb) ceRNA knockdown. (C) Images of tumors harvested when Additional file 6: Figure S2. MCF-7 cells depleted of STARD13- STARD13 3′UTR stable overexpression cells were planted and correlated ceRNAs gain CSC traits. (A) Phase contract images of mammo- followed by doxorubicin treatment or not. The weight of tumors spheres formed by MCF-7 cells with STARD13-correlated ceRNA knock- harvested in (C) was monitored. (D) The weight of mice depicted down. (B) Quantification of mammospheres formed in (A). (C) in (C) was monitored. (E) The volume of tumors harvested in (C) Represented FACS profile of MCF-7 cells described in (A). (D) Identifica- was monitored. (G) Confocal images of MCF-7 cells described in tion of stemness-related genes expression (ALDH1, OCT4, and Nanog) by (B) with doxorubicin treatment. Depletion of STARD13-correlated Western blot analysis in MCF-7 cells described in (A). Data were pre- ceRNAs impaired the cellular retention of doxorubicin. (F) Western sented as the mean ± SD, n =3, *p < 0.05, **p < 0.01 vs. pLVX-Ctrl. (TIF blot assay of lysates from MCF-7 cells with STARD13-correlated 3016 kb) ceRNA knockdown plus si-Dicer or not. (TIF 3999 kb) Additional file 7: Figure S3. STARD13-correlated ceRNA network in- hibits CSC traits of breast cancer cells in vivo. (A, B, and C) Images (left) Abbreviations and weight (right) of tumors harvested when serially diluted MDA-MB- ceRNA: Competing endogenous RNAs; CSCs: Cancer stem cells; 231 cells with STARD13-correlated ceRNAs-3′UTR overexpression were ECM: Extracellular matrix; EMT: Epithelial-to-mesenchymal transition; planted. (D and E) Images (left) and weight (right) of tumors harvested FACS: Fluorescence-activated cell sorting; FBS: Fetal bovine serum; when serially diluted MCF-7 cells with STARD13 or its ceRNA knockdown GEO: Gene Expression Omnibus; IHC: Immunohistochemistry; LATS: Large were planted. (TIF 4230 kb) tumor suppressor; MDR: Multidrug resistance; qRT-PCR: Quantitative real-time Additional file 8: Figure S4. (A) Correlation analysis between LATS1/2 PCR; RIP: RNA immunoprecipitation; STR: Short tandem repeat; TCGA: The and STARD13, CDH5, HOXD1, and HOXD10, based on the microarray data Cancer Genome Atlas; UTR: Untranslated region downloaded from the TCGA data portal. (B) The ceRNA sequence and the genomic locus. (TIF 1525 kb) Funding Additional file 9: Figure S5. STARD13-correlated ceRNAs-3′UTRs regulate This work was supported by the National Nature Science Foundation of TAZ nuclear abundance. Confocal images of TAZ distribution in MDA-MB- China, No. 81702957; China Postdoctoral Science Foundation, No. 231 cells with STARD13-correlated ceRNAs-3′UTR overexpression or not. 2017M620230; Postdoctoral Research Funding Scheme of Jiangsu Province Data were presented as the mean ± SD, n =3, *p <0.05, **p < 0.01 vs. Vector. (2017), No.1701197B; and the Priority Academic Program Development (TIF 4364 kb) (PAPD) of Jiangsu Higher Education Institutions. Additional file 10: Table S5. The number of common miRNA binding Availability of data and materials sites on STARD13 3′UTR and LATS1/2 3′UTR is predicted using Targetscan All data generated or analyzed during this study are included in this 6.2 and microRNA.org. (DOC 33 kb) published article (and its Additional files). Additional file 11: Figure S6. Target miRNAs attenuated the promotive effects of STARD13-correlated ceRNA network on Hippo signaling. Target Authors’ contributions miRNAs (miR-424, miR-374a, miR-590-3p, miR-448, and miR-15a) mimics mix LZ and TX designed the research. LZ and CX analyzed the data. LZ, CX, XL, was co-transfected with STARD13-correlated ceRNAs-3′UTR overexpression QG, LG, and HN performed the research. LZ, CX, and XL wrote the paper. YX constructs or not, the protein level of LATS1/2 and downstream effectors contributed new reagents or analytic tools. All authors read and approved (p-YAP/p-TAZ, YAP/TAZ, and CTGF) was examined. (TIF 1437 kb) the final manuscript. Zheng et al. Journal of Hematology & Oncology (2018) 11:72 Page 17 of 18 Ethics approval and consent to participate 16. Moroishi T, Hayashi T, Pan WW, Fujita Y, Holt MV, Qin J, et al. The All animal experiments were performed with the approval of Ethics Hippo pathway kinases LATS1/2 suppress cancer immunity. Cell. 2016; Committee for Animal Experimentation of China Pharmaceutical University. 167:1525–39. e17 17. Cordenonsi M, Zanconato F, Azzolin L, Forcato M, Rosato A, Frasson C, et al. Competing interests The Hippo transducer TAZ confers cancer stem cell-related traits on breast The authors declare that they have no competing interests. cancer cells. Cell. 2011;147:759–72. 18. Tremblay AM, Camargo FD. Hippo signaling in mammalian stem cells. Semin Cell Dev Biol. 2012;23:818–26. Publisher’sNote 19. Wang Z, Wu Y, Wang H, Zhang Y, Mei L, Fang X, et al. Interplay of Springer Nature remains neutral with regard to jurisdictional claims in mevalonate and Hippo pathways regulates RHAMM transcription via YAP to published maps and institutional affiliations. modulate breast cancer cell motility. Proc Natl Acad Sci U S A. 2014;111: E89–98. Author details 20. Zhao B, Li L, Lei Q, Guan KL. The Hippo-YAP pathway in organ size School of Life Science and Technology, Jiangsu Key Laboratory of control and tumorigenesis: an updated version. Genes Dev. 2010;24: Carcinogenesis and Intervention, China Pharmaceutical University, Nanjing 862–74. 210009, China. Jiangsu Key Laboratory of Drug Discovery for Metabolic 21. Sorrentino G, Ruggeri N, Specchia V, Cordenonsi M, Mano M, Dupont S, Diseases, Department of Pharmacology of Chinese Materia Medica, China et al. Metabolic control of YAP and TAZ by the mevalonate pathway. Nat Pharmaceutical University, 24 Tong Jia Xiang, Nanjing 210009, China. Cell Biol. 2014;16:357–66. Jiangsu Key Laboratory for Pharmacology and Safety Evaluation of Chinese 22. Zanconato F, Cordenonsi M, Piccolo S. YAP/TAZ at the roots of cancer. Materia Medica, School of Pharmacy, Nanjing University of Chinese Medicine, Cancer Cell. 2016;29:783–803. Nanjing 210023, China. 23. Kim J, Jo H, Hong H, Kim MH, Kim JM, Lee JK, et al. Actin remodelling factors control ciliogenesis by regulating YAP/TAZ activity and vesicle Received: 1 March 2018 Accepted: 2 May 2018 trafficking. Nat Commun. 2015;6:6781. 24. Ohgushi M, Minaguchi M, Sasai Y. Rho-signaling-directed YAP/TAZ activity underlies the long-term survival and expansion of human embryonic stem References cells. Cell Stem Cell. 2015;17:448–61. 1. Kroenke CH, Michael YL, Poole EM, Kwan ML, Nechuta S, Leas E, et al. 25. Aragona M, Panciera T, Manfrin A, Giulitti S, Michielin F, Elvassore N, et al. A Postdiagnosis social networks and breast cancer mortality in the after breast mechanical checkpoint controls multicellular growth through YAP/TAZ cancer pooling project. Cancer. 2017;123:1228–37. regulation by actin-processing factors. Cell. 2013;154:1047–59. 2. Gradishar WJ, Anderson BO, Balassanian R, Blair SL, Burstein HJ, Cyr A, et al. 26. Hu J, Li X, Guo X, Guo Q, Xiang C, Zhang Z, et al. CCR2 3′UTR functions as a Invasive breast cancer version 1.2016, NCCN Clinical Practice Guidelines in competing endogenous RNA to inhibit breast cancer metastasis. J Cell Sci. Oncology. J Natl Compr Cancer Netw. 2016;14:324–54. 2017;130:3399–413. 3. Li J, Condello S, Thomes-Pepin J, Ma X, Xia Y, Hurley TD, et al. Lipid 27. Dontu G, Abdallah WM, Foley JM, Jackson KW, Clarke MF, Kawamura MJ, desaturation is a metabolic marker and therapeutic target of ovarian cancer et al. In vitro propagation and transcriptional profiling of human mammary stem cells. Cell Stem Cell. 2016;20:303-14.e5. stem/progenitor cells. Genes Dev. 2003;17:1253–70. 4. Yang M, Liu P, Huang P. Cancer stem cells, metabolism, and therapeutic 28. Liu R, Shi P, Nie Z, Liang H, Zhou Z, Chen W, et al. Mifepristone suppresses significance. Tumour Biol. 2016;37:5735–42. basal triple-negative breast cancer stem cells by down-regulating KLF5 5. Chaffer CL, Brueckmann I, Scheel C, Kaestli AJ, Wiggins PA, Rodrigues LO, expression. Theranostics. 2016;6:533–44. et al. Normal and neoplastic nonstem cells can spontaneously convert to a 29. Cidado J, Wong HY, Rosen DM, Cimino-Mathews A, Garay JP, Fessler AG, stem-like state. Proc Natl Acad Sci U S A. 2011;108:7950–5. et al. Ki-67 is required for maintenance of cancer stem cells but not cell 6. Mani SA, Guo W, Liao MJ, Eaton EN, Ayyanan A, Zhou AY, et al. The proliferation. Oncotarget. 2016;7:6281–93. epithelial-mesenchymal transition generates cells with properties of stem 30. Guo PD, Lu XX, Gan WJ, Li XM, He XS, Zhang S, et al. RARgamma cells. Cell. 2008;133:704–15. downregulation contributes to colorectal tumorigenesis and metastasis by 7. Thiery JP, Acloque H, Huang RY, Nieto MA. Epithelial-mesenchymal derepressing the Hippo-Yap pathway. Cancer Res. 2016;76:3813–25. transitions in development and disease. Cell. 2009;139:871–90. 31. Di Benedetto A, Mottolese M, Sperati F, Ercolani C, Di Lauro L, Pizzuti L, 8. Li Z, Wang Y, Zhu Y, Yuan C, Wang D, Zhang W, et al. The Hippo transducer et al. The Hippo transducers TAZ/YAP and their target CTGF in male breast TAZ promotes epithelial to mesenchymal transition and cancer stem cell cancer. Oncotarget. 2016;7:43188–98. maintenance in oral cancer. Mol Oncol. 2015;9:1091–105. 32. Cheng DL, Xiang YY, Ji LJ, Lu XJ. Competing endogenous RNA interplay in 9. Cesana M, Cacchiarelli D, Legnini I, Santini T, Sthandier O, Chinappi M, et al. cancer: mechanism, methodology, and perspectives. Tumour Biol. 2015;36: A long noncoding RNA controls muscle differentiation by functioning as a 479–88. competing endogenous RNA. Cell. 2011;147:358–69. 33. Denzler R, Agarwal V, Stefano J, Bartel DP, Stoffel M. Assessing the ceRNA 10. Ren Y, Ji N, Kang X, Wang R, Ma W, Hu Z, et al. Aberrant ceRNA-mediated hypothesis with quantitative measurements of miRNA and target regulation of KNG1 contributes to glioblastoma-induced angiogenesis. abundance. Mol Cell. 2014;54:766–76. Oncotarget. 2016. http://www.oncotarget.com/index.php?journal= 34. Guo LL, Song CH, Wang P, Dai LP, Zhang JY, Wang KJ. Competing oncotarget&page=article&op=view&path%5B%5D=12659&path%5B%5D= endogenous RNA networks and gastric cancer. World J Gastroenterol. 2015; 21:11680–7. 11. Karreth FA, Reschke M, Ruocco A, Ng C, Chapuy B, Leopold V, et al. The 35. Bosson AD, Zamudio JR, Sharp PA. Endogenous miRNA and target BRAF pseudogene functions as a competitive endogenous RNA and concentrations determine susceptibility to potential ceRNA competition. induces lymphoma in vivo. Cell. 2015;161:319–32. Mol Cell. 2014;56:347–59. 12. Lu MH, Tang B, Zeng S, Hu CJ, Xie R, Wu YY, et al. Long noncoding RNA 36. Tay Y, Rinn J, Pandolfi PP. The multilayered complexity of ceRNA crosstalk BC032469, a novel competing endogenous RNA, upregulates hTERT and competition. Nature. 2014;505:344–52. expression by sponging miR-1207-5p and promotes proliferation in gastric 37. Tang F, Zhang R, He Y, Zou M, Guo L, Xi T. MicroRNA-125b induces cancer. Oncogene. 2015;35:3524–34. metastasis by targeting STARD13 in MCF-7 and MDA-MB-231 breast cancer 13. Wang Y, Hou J, He D, Sun M, Zhang P, Yu Y, et al. The emerging cells. PLoS One. 2012;7:e35435. function and mechanism of ceRNAs in cancer. Trends in genetics: TIG. 2016;32:211–24. 38. Denzler R, McGeary SE, Title AC, Agarwal V, Bartel DP, Stoffel M. Impact of 14. Zheng L, Li X, Gu Y, Lv X, Xi T. The 3′UTR of the pseudogene CYP4Z2P microRNA levels, target-site complementarity, and cooperativity on promotes tumor angiogenesis in breast cancer by acting as a ceRNA for competing endogenous RNA-regulated gene expression. Mol Cell 2016;64: CYP4Z1. Breast Cancer Res Treat. 2015;150:105–18. 565–579. 15. Li X, Zheng L, Zhang F, Hu J, Chou J, Liu Y, et al. STARD13-correlated ceRNA 39. Xu J, Feng L, Han Z, Li Y, Wu A, Shao T, et al. Extensive ceRNA-ceRNA network inhibits EMT and metastasis of breast cancer. Oncotarget. 2016;7: interaction networks mediated by miRNAs regulate development in 23197–211. multiple rhesus tissues. Nucleic Acids Res. 2016;44:9438–51. Zheng et al. Journal of Hematology & Oncology (2018) 11:72 Page 18 of 18 40. Ala U, Karreth FA, Bosia C, Pagnani A, Taulli R, Leopold V, et al. Integrated transcriptional and competitive endogenous RNA networks are cross- regulated in permissive molecular environments. Proc Natl Acad Sci U S A. 2013;110:7154–9. 41. Zheng L, Li X, Meng X, Chou J, Hu J, Zhang F, et al. Competing endogenous RNA networks of CYP4Z1 and pseudogene CYP4Z2P confer tamoxifen resistance in breast cancer. Mol Cell Endocrinol. 2016;427:133–42. 42. Li C, Zheng L, Xin Y, Tan Z, Zhang Y, Meng X, et al. The competing endogenous RNA network of CYP4Z1 and pseudogene CYP4Z2P exerts an anti-apoptotic function in breast cancer. FEBS Lett. 2017;591:991–1000. 43. Nagaraja GM, Kandpal RP. Chromosome 13q12 encoded Rho GTPase activating protein suppresses growth of breast carcinoma cells, and yeast two-hybrid screen shows its interaction with several proteins. Biochem Biophys Res Commun. 2004;313:654–65. 44. Feng X, Degese MS, Iglesias-Bartolome R, Vaque JP, Molinolo AA, Rodrigues M, et al. Hippo-independent activation of YAP by the GNAQ uveal melanoma oncogene through a trio-regulated rho GTPase signaling circuitry. Cancer Cell. 2014;25:831–45. 45. Pascual-Vargas P, Cooper S, Sero J, Bousgouni V, Arias-Garcia M, Bakal C. RNAi screens for Rho GTPase regulators of cell shape and YAP/TAZ localisation in triple negative breast cancer. Scientific data. 2017;4:170018. 46. Dupont S, Morsut L, Aragona M, Enzo E, Giulitti S, Cordenonsi M, et al. Role of YAP/TAZ in mechanotransduction. Nature. 2011;474:179–83. 47. Mukherjee S, Manna A, Bhattacharjee P, Mazumdar M, Saha S, Chakraborty S, et al. Non-migratory tumorigenic intrinsic cancer stem cells ensure breast cancer metastasis by generation of CXCR4 migrating cancer stem cells. Oncogene. 2016;35:4937–48. 48. Gupta PB, Onder TT, Jiang G, Tao K, Kuperwasser C, Weinberg RA, et al. Identification of selective inhibitors of cancer stem cells by high-throughput screening. Cell. 2009;138:645–59. 49. Valent P, Bonnet D, De Maria R, Lapidot T, Copland M, Melo JV, et al. Cancer stem cell definitions and terminology: the devil is in the details. Nat Rev Cancer. 2012;12:767–75.
Journal of Hematology & Oncology
– Springer Journals
Published: May 30, 2018