Published OnlineFirst March 31, 2014; DOI: 10.1158/0008-5472.CAN-13-2321 Cancer Molecular and Cellular Pathobiology Research LEO1 Is Regulated by PRL-3 and Mediates Its Oncogenic Properties in Acute Myelogenous Leukemia Phyllis S.Y. Chong1, Jianbiao Zhou1, Lip-Lee Cheong2, Shaw-Cheng Liu1, Jingru Qian3, Tiannan Guo3, Siu Kwan Sze3, Qi Zeng4, and Wee Joo Chng1,2,5 Abstract PRL-3, an oncogenic dual-specificity phosphatase, is overexpressed in 50% of acute myelogenous leukemia (AML) and associated with poor survival. We found that stable expression of PRL-3 confers cytokine independence and growth advantage of AML cells. However, how PRL-3 mediates these functions in AML is not known. To comprehensively screen for PRL3-regulated proteins in AML, we performed SILAC-based quantitative proteomics analysis and discovered 398 significantly perturbed proteins after PRL-3 over- expression. We show that Leo1, a component of RNA polymerase II–associated factor (PAF) complex, is a novel and important mediator of PRL-3 oncogenic activities in AML. We described a novel mechanism where elevated PRL-3 protein increases JMJD2C histone demethylase occupancy on Leo1 promoter, thereby reducing the H3K9me3 repressive signals and promoting Leo1 gene expression. Furthermore, PRL-3 and Leo1 levels were positively associated in AML patient samples (N ¼ 24; P < 0.01). On the other hand, inhibition of Leo1 reverses PRL-3 oncogenic phenotypes in AML. Loss of Leo1 leads to destabilization of the PAF complex and downregulation of SOX2 and SOX4, potent oncogenes in myeloid transformation. In conclusion, we identify an important and novel mechanism by which PRL-3 mediates its oncogenic function in AML. Cancer Res; 74(11); 1–11. Ó2014 AACR. Introduction and breast, and PRL-3 protein was overexpressed in an PRL-3, encoded by the PTP4A3 gene, is a small dual- average of 22.3% of 1,008 human carcinoma samples exam- specificity phosphatase characterized by the conserved ined using immunohistochemistry (9, 10). Together with the fact that it has a highly restricted basal pattern of expression C(X5)R catalytic domain, and a unique C-terminal prenyla- tion domain essential for its proper subcellular localization in adult tissues (11), PRL-3 is deemed as an attractive (1, 2). PRL-3 has been shown to promote cellular processes, therapeutic target that spares normal tissues. The potential – fi such as cell motility, invasion, cell growth, and survival, of this target has been demonstrated using PRL-3 speci c in vivo through various mechanisms (2–7). PRL-3 was first linked to antibodies in an model (12). cancer when it was consistently found at elevated levels in In recent years, accumulating evidence suggests that PRL-3 colorectal cancer metastases, but at much lower levels in is also a novel therapeutic target and biomarker in leukemia fi matched early-staged tumor and normal colorectal epithe- (13, 14). We were the rst to report that elevated PRL-3 protein lium (8). Since then, elevated expression of PRL-3 has been expression occurs in about 47% of human acute myelogenous implicated in the progression and metastasis of an array of leukemia (AML) cases while absent from normal myeloid cells cancer types, including gastric, ovarian, cervical, lung, liver, in bone marrow (13). In addition, a large-scale gene expression profiling study of 454 primary AML samples demonstrates that high PRL-3 levels is an independent negative prognostic Authors' Affiliations: 1Cancer Science Institute of Singapore; 2Depart- factor in AML, both for overall survival and event-free survival ment of Medicine, Yong Loo Lin School of Medicine, National University of (14). These reports collectively suggest that PRL-3 may be of Singapore; 3Department of Biological Sciences, Nanyang Technological University; 4Institute of Molecular and Cell Biology (AÃSTAR); and 5Depart- biologic and clinical relevance in AML and warrants further ment of Haematology-Oncology, National University Cancer Institute of investigation. Singapore, National University Health System, Singapore In the present study, we created a TF-1 AML cell line stably Note: Supplementary data for this article are available at Cancer Research overexpressing PRL-3 (TF1-hPRL3) as a model to study the Online (http://cancerres.aacrjournals.org/). biologic relevance of PRL-3 in AML, and we use a quantitative P.S.Y. Chong and J. Zhou contributed equally to this work. proteomic strategy to profile on a global scale changes in Corresponding Author: Wee-Joo Chng, National University Health Sys- protein expression induced by PRL-3 in AML. We found that tem, 1E Kent Ridge Road, NUHS Tower Block, Level 10, Singapore 119228. PRL-3 has pro-oncogenic properties in AML, and Leo1, a Phone: 65-6772-4612; Fax: 65-6777-5545; Email: [email protected] component of the human RNA polymerase II–associated factor doi: 10.1158/0008-5472.CAN-13-2321 (PAF) complex, is one of the most differentially expressed Ó2014 American Association for Cancer Research. proteins induced by PRL-3. Further, we showed that PRL-3 www.aacrjournals.org OF1 Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2014 American Association for Cancer Research. Published OnlineFirst March 31, 2014; DOI: 10.1158/0008-5472.CAN-13-2321 Chong et al. upregulates Leo1 through a novel epigenetic mechanism. On RNA extraction and quantitative reverse transcription the other hand, knockdown of Leo1 significantly diminishes PCR the oncogenic effects of PRL-3. Our current work implicates a Detailed protocol is available in Supplementary Methods. pro-oncogenic role of PRL-3 in AML, and reveals Leo1 as a novel downstream molecule required for PRL-3 oncogenic Western blotting function in leukemia. Cells were counted and lysed in radioimmunoprecipita- tion assay buffer. Anti-PRL3 (clone 318) was a kind gift from Materials and Methods Dr. Zeng Qi [Institute of Molecular and Cell Biology, Agency for Science, Technology, and Research (A*STAR), Singa- Cell culture pore]; anti-Leo1, anti-Paf1, and anti-Ctr9 antibodies were fi HEK293T cells were cultured in Dulbecco's Modi ed Eagle from Bethyl Laboratories; anti-actin, anti-hnRNPE1, anti- Medium with 10% fetal calf serum (FCS). TF1-derived cell lines HSP90, anti-JMJD2C, anti-GFP, and anti-c-Myc (9E10) anti- were cultured in RPMI 1640 medium containing 10% FCS (R10) bodies were from Santa Cruz Biotechnology; anti- supplemented with 5 ng/mL human interleukin (IL)-3 (Milte- Stathmin, anti-HDAC2 and anti-H3, anti-H3K9me2, anti- nyi Biotec). Molm-14, HEL, and HL-60 were cultured in R10. þ H3K27me2, anti-H3K27me3, anti-H3K4me1, anti-H3K4me2, Human CD34 cells were grown in StemSpan SFEM II medium anti-H3K4me3, and anti-H3K79me2 antibodies were from supplemented with StemSpan CC100 cytokine cocktail (Stem- Cell Signaling Technology; anti-H3K9me3 antibody was Cell Technologies). Primary AML cells were grown in same from Active Motif. conditions, with the addition of granulocyte macrophage colony-stimulating factor (20 ng/mL). Cell lines were obtained Chromatin immunoprecipitation from American Type Culture Collection and authenticated. Chromatin immunoprecipitation (ChIP) assays were per- Plasmid details are available in Supplementary Methods. formed using the respective antibodies according to the man- ufacturer's instructions (Thermo Scientific). Quantitative PCR SILAC-based mass spectrometry (qPCR) was performed using the eluted DNA (sample) and 1% TF-1 was cultured in "light" stable isotope labeling by amino input, with primers spanning Leo1 promoter region with acids in cell culture (SILAC) medium containing normal lysine respect to Leo1 transcriptional start site. The percentage of and arginine amino acids, whereas the TF1-hPRL3 was grown input was calculated as described previously (15). Primer 13 in "heavy" SILAC medium with stable isotope-labeled C6 sequences used for ChIP–qPCR are available in Supplementary þ 13 15 þ lysine ( 6-Da shift) and C6 N4 arginine ( 10-Da shift; Table S3. Thermo Scientific). The cellular lysates were combined and proteolytically digested by trypsin followed by tandem mass fi Luciferase assay spectrometry (MS) identi cation. Peptides were subjected to Detailed protocol is available in Supplementary Methods. target-decoy database search strategy with a false discovery rate (FDR) of less than 1% (<1%). Differential protein expres- Cell proliferation, apoptosis, and CFA sion was quantified from the relative intensity ratios in the MS Detailed protocol is available in Supplementary Methods. spectra between the "heavy" and "light" states. Xenograft mode models Results Six-week-old female nonobese diabetic/severe combined PRL-3 promotes cytokine-independent growth, colony- immunodeficient (NOD/SCID) mice were provided by Dr. forming capacity of AML cells in vitro and Chan Shing-Leng (CSI Singapore, NUS). Exponentially growing tumorigenecity in vivo TF1-pEGFP and TF1-hPRL3 cells (5 Â 106) were subcutane- To assess the roles of PRL-3 in pathogenesis of AML, we ously injected into loose skin between the shoulder blades and developed a pair of stable, isogenic cell lines, TF1-pEGFP and the left and right front leg of NOD/SCID-recipient mice (3 mice TF1-hPRL3 by transfecting pEGFP (vector control) and total), respectively. The length (L) and width (W) of the tumor pEGFP-hPRL-3 vectors into TF-1 cells, respectively, followed were measured with callipers, and tumor volume (TV) was by G418 selection and fluorescence-activated cell sorting calculated as TV ¼ (L Â W2)/2. The protocol is reviewed and (Fig. 1A). TF-1 is a cytokine-dependent AML cell line. Quan- approved by Institutional Animal Care and Use Committee in titative reverse transcription PCR (qRT-PCR) and Western compliance to the guidelines on the care and use of animals for blot validated the overexpression of PRL-3 on both mRNA and scientific purpose. protein levels in the TF1-hPRL3 cells relative to TF1-pEGFP cells (Fig. 1B). In the absence of cytokine (human IL-3), the Transfection and lentiviral-mediated shRNA delivery majority of TF1-pEGFP cells became apoptotic after 72 hours.