Dual-Specificity Phosphatase DUSP6 Has Tumor-Promoting Properties In

Oncogene (2011) 30, 3813–3820 & 2011 Macmillan Publishers Limited All rights reserved 0950-9232/11 www.nature.com/onc SHORT COMMUNICATION Dual-speciﬁcity phosphatase DUSP6 has tumor-promoting properties in human glioblastomas

S Messina1,2, L Frati1, C Leonetti3, C Zuchegna4,5, E Di Zazzo4, A Calogero6 and A Porcellini5

1Department of Pathology, IRCCS Neuromed, Pozzilli, Italy; 2Department of Health and Motor Sciences, University of Cassino, Cassino, Italy; 3Department of Experimental Chemotherapy, Regina Elena Cancer Institute, Rome, Italy; 4Department of Health Sciences, University of Campobasso, Campobasso, Italy; 5Department of Structural and Functional Biology, Federico II University of Naples, Naples, Italy and 6Department of Medical and Surgical Sciences and Biotecnology, Sapienza University of Rome, Rome, Italy

Dual-specificity phosphatase 6 (DUSP6, mitogen-acti- signaling by dephosphorylating components of the vated protein kinase (MAPK) phosphatase 3 or PYST1) MAPK cascade. DUSPs are typically negative feedback dephosphorylates phosphotyrosine and phosphothreonine regulators of MAPK activity that become transcription- residues on extracellular signal-regulated kinase (ERK1/ ally induced after stress or mitogenic stimulation; this 2; MAPK1/2) to inactivate the ERK1/2 kinase. DUSP6 is leads to early production of proteins that inactivate a critical regulator of the ERK signaling cascade and has MAPKs within 20–40 min (Owens and Keyse, 2007). been implicated as a tumor suppressor. We report here Importantly, some DUSPs may provide biomarkers of experimental evidences that DUSP6 is transcriptionally therapeutic responsiveness and patient outcome in upregulated in primary and long-term cultures of human specific cancers, as well as DUSP1 and DUSP2 in glioblastoma, as assayed by northern hybridization and different type of cancers (reviewed in Patterson et al., real-time quantitative PCR, producing constitutive high 2009). level of protein expression. Functional assays were performed DUSP6 is a cytoplasmic member of this family that with adenovirus-mediated expression of DUSP6 in glio- has been implicated in the development of cancer. blastoma cultures. Protein overexpression inhibits growth by Pancreatic cancer cells progress towards a more inducing G1-phase delay and increased mitogenic/anchorage aggressive and invasive phenotype, following loss of dependence and clonogenic potential in vitro. Changes in cell DUSP6 expression (Furukawa et al., 2003, 2005). morphology were associated with an increased tumor growth Overexpression of DUSP6 in estrogen receptor-positive in vivo. Chemoresistance is a major cause of treatment failure breast cancer cells confers resistance to the growth and poor outcome in human glioblastomas. Importantly, inhibitory effects of tamoxifen (Cui et al., 2006) and its DUSP6 overexpression increased resistance to cisplatin- expression is strongly upregulated in HER2-positive mediated cell death in vitro and in vivo.Antisense-mediated breast tumors (Lucci et al., 2010). Moreover, DUSP6 depletion of DUSP6 acted in lowering the threshold to is part of a high-risk signature gene for non-small cell anticancer DNA-damaging drugs. We conclude that upregu- lung cancer (Chen et al., 2007), and its expression is lation of DUSP6 exerts a tumor-promoting role in human significantly correlated with high extracellular signal- glioblastomas exacerbating the malignant phenotype. regulated kinase (ERK) 1/2 activity in primary human Oncogene (2011) 30, 3813–3820; doi:10.1038/onc.2011.99; ovarian cancer cells (Chan et al.,2008).Remarkably,in published online 18 April 2011 human glioblastoma cells DUSP6 gene was induced upon platelet-derived growth factor (PDGF) stimulation Keywords: glioblastoma; dual-specificity phosphatase; (Tullai et al., 2004). Moreover, DUSP6 can be upregu- tumor promoting lated in specific cancers exhibiting aberrant receptor tyrosine kinase and Ras/Raf signaling, such as non-small cell lung cancer, potentially as a negative-feedback regulator of mitogenic signaling (Sato et al.,2006). Introduction Although the DUSP6 gene is a potential tumor- suppressor in some cancers (Furukawa et al., 2003, Dual-specificity phosphatases (DUSPs) negatively 2005; Okudela et al., 2009; Zhang et al., 2010), here, we regulate mitogen-activated protein kinase (MAPK) show that DUSP6 exerts tumor-promoting properties in glioblastoma cancers. Recently, some papers report that Correspondence: Dr S Messina, Department of Pathology, IRCCS upregulation of tumor suppressor genes might promote Neuromed, via Atinense, 18, Pozzilli, Isernia 86077, Italy. the malignant phenotype of cancer cells (Hong et al., E-mail: [email protected] or Professor A Porcellini, 2007). This split of personality was recently documented Department of Structural Biology, School of Biology, Federico II for PTEN and IDH1 genes in glioblastoma cancers University of Naples, Naples, Italy. E-mail: [email protected] (Li et al., 2008; Yan et al., 2009). Indeed, particular Received 8 September 2010; revised 3 February 2011; accepted 21 DUSPs may have contradicting roles in different tumor February 2011; published online 18 April 2011 types, such as DUSP1, that increased or decreased its DUSP6 regulation and function in glioblastoma cancer cells S Messina et al 3814 expression in different type of cancers, depending on the cultures of human glioblastoma, producing constitutive stage of carcinogenesis (reviewed in Patterson et al., high level of protein expression. Functional assays were 2009). performed with adenovirus-mediated expression of Glioblastoma multiforme is the most aggressive and DUSP6 in glioblastoma cultures. Protein overexpression frequent subtype of glioma, and shows very little inhibits growth by inducing G1-phase delay and response to chemotherapy (Maher et al., 2001). Re- inducing cell flattening that was associated with an cently, members of the DUSP family, such as DUSP26 increased tumor growth in vivo and inhibition of and DUSP4, have been involved in determining the epidermal growth factor (EGF)-detachment in vitro. malignant phenotype of glioblastomas (Tanuma et al., Finally, DUSP6 overexpression decreases tumor cell 2009; Waha et al., 2010). sensitivity to anticancer DNA-damaging cisplatin, We report here experimental evidence that DUSP6 is indicating that DUSP6 may function as a tumor transcriptionally upregulated in primary and long-term promoter in these cancers.

Oncogene DUSP6 regulation and function in glioblastoma cancer cells S Messina et al 3815 Results and discussion major isoform in peripheral blood lymphocytes and, to a lesser extent, in normal human astrocytes. In human In order to establish differences in the expression of glioblastomas, in contrast, the shorter transcript was DUSP6 between normal and tumor cells, we first only induced by mitogenic stimulation. To determine analyzed the mRNA levels of DUSP6 by northern blot whether DUSP6 protein levels correlate with transcript and quantitative real-time PCR analysis in short- levels in human glioblastoma cultures, we performed passage cultures (FCN, MTR and MZC) (Calogero biochemical analysis by western blot. Immunoblots et al., 2004) and long-term human glioblastomas on cytosolic fraction showed negligible immuno- (U87MG and variants U87MGvEGFRIII, and kinase- reactivity of endogenous DUSP6 using commercial deficient U87MGDK) (Figure 1a). Comparing the antibodies, as reported previously (Mark et al., 2008). neoplastic cells with the normal counterpart (normal Next, we probed immunoblot membranes with a rabbit human astrocytes), we observed an increased expression polyclonal antibody (generated by Dr Lennartson) of DUSP6 (three- to sixfold) at baseline that increased against DUSP6, which recognizes a protein of 45 kDa upon mitogenic stimulation (Figure 1a, lower panel). (Figure 1c). Importantly, this antibody recognized Consistently, gene expression studies revealed DUSP6 constitutive high levels of endogenous DUSP6 as upregulation (mean fold changes: 3.14) in human multiple bands, probably due to various phosphoryla- astrocytic brain tumors compared with normal brain tion states (Dr Lennartson’s personal communication tissue (Petalidis et al., 2008). All cells were serum- and Figure 2d). starved overnight and stimulated with 100 ng/ml of EGF Functional assays were performed to focus on the for 3 h. Despite some fluctuation monitored, northern implication of DUSP6 overexpression in glioblastoma hybridization of non-neoplatic cells (unstimulated cul- cancers. In vitro assays were performed with replication- tures of murine pheochromocytoma PC12 cells, mouse deficient adenoviral vector harboring murine sense, fibroblasts NIH-3T3 and normal human astrocytes) antisense and C293S mutant DUSP6 (Muda et al., showed basal negligible mRNA levels and powerful 1996; Kim et al., 1996). The plasmid coding for murine induction of DUSP6 mRNA upon mitogenic stimula- DUSP6 includes a COOH-terminal myc-tag cloned in tion (Figure 1a, upper panel). As observed by Marchetti pAdTRK-CMV adenoviral shuttle in BglII-EcoRV et al. (2005), half-lives of DUSP6 mRNA was less than sites, which contains green fluorescent protein as a 1 h in heterologous expression systems; however, this is tracer (Porcellini and De Blasi, 2004). not the case for native DUSP6 in our system. Following Growth experiments (proliferation, thymidine incor- this line of evidence, we quantified DUSP6 mRNA by poration and cell-cycle analysis) were performed using real-time quantitative PCR in several cultures of as controls the empty vector, the antisense-DUSP6 glioblastomas (Figure 1b). We compared DUSP6 (DUSP6-AS) vector and the mutant C293S (DUSP6- expression in primary astrocytoma FCN, MTR, DME, CS)-expressing vector to ensure that DUSP6 is acting on CNT, MZC and GB58 (WHO grade IV) and glioblas- its cellular target. Antisense expression inhibited effec- toma cell lines U251, T98G and U87MG, with human tively concomitant expression of sense-DUSP6, as shown breast carcinoma MCF-7 cells as other neoplastic tissue in Supplemetary Figure S2. Two different glioblastoma (Nunes-Xavier et al., 2010). As expected, we observed a cell lines, U87MG (p53 wild type) and U251MG (p53 significant increase in the percentage of major transcript mutant), were infected with empty vector and sense- at 543 bp both in unstimulated and mitogen-stimulated or antisense-DUSP6 and assayed for cell growth and cells (Figure 1b, lower panel). Interestingly, we detected (methyl-3H) thymidine incorporation. Comparable results two variants transcritpt encoding for different DUSP6 were obtained with two lines that showed decreased isoforms by PCR analysis, as described in Supplemen- growth (41.4-fold) and thymidine incorporation (1.5-fold) tary Figure S1. The shorter transcript was detected as compared with antisense- and control empty vector

Figure 1 DUSP6 mRNA is overexpressed in human glioblastomas. (a) Representative northern blot for DUSP6 mRNA in human glioblastomas. mRNA extraction was performed using TriPure isolation reagent (Roche Diagnostics, Indianapolis, IN, USA), according to manufacturer’s instructions. Membranes probed with the 32P-labeled using a 2.7-kb human DUSP6 HindIII/XhoI cDNA fragment and 18S ribosomal probe as normalized loading. All experiments were conducted on cells serum-starved overnight and stimulated with EGF, 100 ng/ml for 3 h. Northern blot (lower panel) signals were quantified by scanning densitometry using ImageJ 1.43d software (Wayne Rasband, NIH, Bethesda, MD, USA; http://rsb.info.nih.gov/ij). All experiments were carried out in triplicates and statistical significance obtained by Student’s t-test. *Po0.01 as compared with the untreated normal astrocytes (Student’s unpaired test); **Po0.02 as compared with the untreated sample (Student’s matched pairs test). (b) Real-time quantitative PCR for DUSP6 mRNA in glioblastomas. Total mRNA (500 ng) was reverse transcribed using random primers and Superscript III (Invitrogen, Carlsbad, CA, USA), according to the manufacturer’s instruction. For primers and methods, see Supplementary Materials. Human glioblastoma cultures U-251MG, U-87MG and T98G were obtained from the American Type Culture Collection (ATTC, Rochville, MD, USA) and grown according to manufacturer’s instructions. U87MGDEGFR and U87MGDK were a generous gift from Dr Cavenee and Dr Furnari. Astrocytoma primary FCN, MTR, DME, CNT, MZC and GB58 cell lines (WHO grade IV) were established from tumor specimens of patients and cultured as described (Calogero et al., 2004; Ricci-Vitiani et al., 2010). Normal human astrocytes were purchased from Cambrex Charles City, Inc. (Charles City, IA, USA) and grown according to the manufacturer’s instructions. Rat PC12, mouse NHI-3T3, human MCF-7 and normal human astrocytes were used as control cells. (c) Western blot analysis of native DUSP6 protein. Constitutive high protein levels were detected in U87MG, U251 and T98G, and were stimulated with EGF for 30 min. Endogenous DUSP6 was detected on the same membrane with a ‘home-made’ specific antibody against DUSP6 (Lennartson’s lab). Western blot analysis of phospho-ERK and total ERK was performed in the same experiment. Monoclonal anti-b-tubulin as loading control was purchased by Sigma–Aldrich (St Louis, MO, USA).

Oncogene DUSP6 regulation and function in glioblastoma cancer cells S Messina et al 3816 (Figures 2a and b). Phosphorylation status of ERK1/2 DUSP6 overexpression in human glioblastomas are in these functional experiments was assayed by western shown in Supplementary Figure S3. blot analysis of U87MG overexpressing sense- and Taking in consideration phosphatase-mediated effects antisense-DUSP6. Mitogenic stimulation by EGF in- on cellular morphology and adhesiveness in human duced a strong phosphorylation in P-ERK1/2 that was glioblastomas (Tanuma et al., 2009), we next measured blunted by overexpression of sense-DUSP6 adenovirus adhesive properties of DUSP6-expressing cells in these (Figure 2d). Overexpression of wild-type DUSP6 delays cancers by in vitro analysis using clonogenic and EGF- progression through the cell cycle as assayed by induced detachment assays (Figure 3). We carried out ﬂuorescence-activated cell sorting analysis of propidium colony-formation assay with method by Porcellini et al. iodide staining. U87MG cells expressing sense-DUSP6 (1997); three human glioblastoma cultures, U251-MG, showed an increased G1-phase compared with control U87MG and the mutant U87MGDEGFR were infected and were insensitive to EGF stimulation (Figure 2c). and seeded at appropriate dilutions to form colonies In vitro growth properties mediated by mutant C293S- (consist of at least 50 cells) in 2–3 weeks; then stained

Figure 2 Overexpression of DUSP6 reduces growth properties in human glioblastoma. (a) Proliferation 3-(4,5-dimethyl-thiazol-2-yl)- 2,5-diphenyl tetrazolium bromide assay was performed in 96-well flat-bottomed tissue cultures as follows. In all,1500 cells were seeded in eight replicates and infected as described in Porcellini and De Blasi (2004) and Messina et al. (2004). All values are means±s.d. of at least three experiments in triplicate. Differences between groups were tested for statistical significance using Student’s t-test. *Po0.01 as compared with the control vector-expressing cells. (b)(methyl-3H) thymidine incorporation in U87MG-expressing sense- and antisense- DUSP6 adenoviruses. For thymidine method, see Messina et al. (2004); briefly, U87MG cells were infected and starved overnight; incubated with 3–5 mCi/well of (methyl-3H) thymidine for 16 h with and without EGF (100 ng/ml). All experiments were carried out in triplicates and statistical significance obtained by Student’s t-test. *Po0.01 as compared with the control vector-expressing cells; **Po0.01 as compared with DUSP6-espressing cells (Student’s matched pairs test). (c) Fluorescence-activated cell sorting (FACScan) cell-cycle analysis was carried out as follows: 5 Â 105 of infected cells were plated in 60 mm dishes and grown in low-serum medium (0.5% fetal bovine serum) for 18 h and then stimulated with EGF 100 ng/ml in low serum. Fluorescence was determined by using the FACScan (Becton Dickinson, Franklin Lakes, NJ, USA) measuring propidium iodide staining on 40 000 events. All experiments were carried out in triplicates and statistical significance obtained by ModFit LT software (Verity Software House, Topsham, MA, USA) module. (d) Western blot analysis of recombinant sense- and antisense-DUSP6 protein and phospho-ERK status in U87MG. Ectopical overexpression of DUSP6 was detected on the same membrane with antisense-DUPS6 and control (Lennartson’s antibody). Western blot analysis of phospho-ERK and total ERK was performed in the same functional experiment. Monoclonal anti-b-tubulin was used as loading control.

Oncogene DUSP6 regulation and function in glioblastoma cancer cells S Messina et al 3817 and counted using a stereomicroscope. Unexpectedly, detachment that is normally observed after stimulation cells overexpressing sense-DUSP6 had twofold more with EGF. In the mutant U87MGDEGFR glioblastoma colonies than control with both high (10% fetal calf cultures, this efffect was abrogated due to its desensi- serum) and low (0.2% fetal calf serum) serum stimula- tization to mitogenic stimulus. Nevertheless, DUSP6 tion (Figure 3a). Moreover, phosphatase-overexpressing overexpression enhanced tumor adhesiveness compared cells exhibited a more flattened morphology by fluores- with wild-type U87MG cultures (Figure 3c). cence microscopy of green fluorescent protein-positive Among human astrocytic gliomas, glioblastoma cells (Figure 3b). Interestingly, we obtained similar multiforme is the most common and aggressive kind results with U87MGDEGFR, the glioblastoma multi- of tumor characterized by high invasivity and shows formes expressing the characteristic EGFR mutant very little response to chemotherapy (Maher et al., (EGFRvIII, de 2–7) that does not bind ligand, signals 2001). Thus, we examined the chemosensitivity of constitutively, and is more tumorigenic than the wild- glioblastoma-overexpressing DUSP6 by survival analy- type receptor (Furnari et al., 2007). We also examined sis. Cisplatin (anticancer DNA-damaging cisplatin) is a anchorage-dependent regulation by adhesion assay. DNA-adduct-inducing anticancer frequently used to Overexpression of DUSP6 strongly inhibited cellular treat brain tumors (Nagane et al., 1996; Silvani et al., 2009). We measured chemoresistance to increasing doses of drug by colorimetric 3-(4,5-dimethyl-thiazol-2-yl)- 2,5-diphenyl tetrazolium bromide assay. Mitochondrial permeability was measured in cisplatin-treated U87MG cells at 48 and 72 h, showing enhanced chemoresistance compared with controls. Indeed, depletion by antisense DUSP6-AS reversed the effect in lowering the threshold sensitivity to drug treatment (Figure 4a). The fraction of propidium iodide-positive cells, measured by bivariate analysis on green fluorescent protein-positive cells with

Figure 3 DUSP6 overexpression increases colony-forming efficiency and inhibits EGF-induced detachment in U87MG cells. (a) Clonogenic survival was determined by a colorimetric colony- forming assay at high dilution with cells overexpressing DUSP6 or control adenovirus. For plating efficiency experiments, 200 cells were seeded in six multiwell systems, after 15 days, clones were stained with 0.25% Comassie brilliant blue in methanol and 10% acetic acid. Visible clones were counted and the cellularity was analyzed by phase contrast microscopy. All experiments were carried out in triplicates and statistical significance obtained using Wilcoxon matched test. *Po0.01 as compared with the control vector-infected cells growing in 10% fetal bovine serum (FBS); **Po0.001 to the control vector-infected cells growing in 0.2% FBS. (b) Representative fluorescence images showing an altered morphology in U87MG-expressing sense-DUSP6. Green fluorescent protein-positive cells enhanced cellular density and flattened morphology. Cells were viewed under fluorescence microscopy 48 h post infection at Â 20 magnification (upper panel) and Â 60 magnification (lower panel). (c) Adhesion assay of EGF stimulated U87MG-expressing sense-DUSP6. Adhesion assay was performed as follows: flat-bottom 96-well tissue cultures (Corning Corp., Corning, NY, USA) were seeded with infected cells and subse- quently serum-starved in Dulbecco0s modied Eagle0s medium (DMEM) plus 0.5% bovine serum albumin (BSA) for 16–18 h. Then, treated with EGF (100 ng/ml) in DMEM 0.5% BSA for the indicated times. Non-adhered cells were removed by washing twice with 75 ml of 73% percoll and 0.9% NaCl, and adherent cells fixed with 50 ml of 10% glutaraldehyde in percoll and stained with 100 ml of 0.1% crystal violet, solubilized in 50 ml of 0.5% Triton X-100, and solutions were read at 570 nm in an MRX Revelation absorbance reader (Thermo Labsystems, Franklin, MA, USA). Results are expressed as mean % adhesion±s.d. from three separate experiments in triplicate, and statistical significance obtained using Wilcoxon matched test. *Po0.001 as compared with the untreated U87MG infected with the control vector (matched test); **Po0.001 to the U87MG infected with the control vector treated with EGF (unmatched test); 1Po0.02 as compared with the untreated U87MG infected with the control vector (unmatched test); 11Po0.001 to the untreated U87MG.EGFRvIII infected with the control vector (unmatched test); 111Po0.001 to the U87MG.EGFRvIII infected with the control vector treated with EGF (unmatched test).

Oncogene DUSP6 regulation and function in glioblastoma cancer cells S Messina et al 3818

Figure 4 DUSP6 overexpression enhances U87MG chemoresistance in vitro and in vivo.(a) Cisplatin-induced cytoxicity was measured by colorimetric 3-(4,5-dimethyl-thiazol-2-yl)-2,5-diphenyl tetrazolium bromide assay. Cell viability of DUSP6-expressing U87MG treated with increasing doses (anticancer DNA-damaging cisplatin (CDDP), 2–10 mg/ml) was determined at 48 and 72 h. Cellular depletion of endogenous DUSP6 was obtained by overexpression of antisense-DUSP6 that was able to induce cell death mainly at higher doses of CDDP. Differences between groups were tested for statistical significance using Student’s t-test. *Po0.01 as compared with the control vector treated with the same CDDP dosage; **Po0.02 as compared with the control vector treated with the same CDDP dosage. (b) Propidium iodide staining measured by fluorescence-activated cell sorting scan flow cytometer (Becton Dickinson) in U87MG-expressing sense- or antisense-DUSP6 and treated with increasing doses of cisplatin for 8 h. Differences between groups were tested for statistical significance using Student’s t-test. *Po0.01 as compared with the control vector treated with the same CDDP dosage; **Po0.05 as compared with the control vector treated with the same CDDP dosage. (c) Enhanced tumor volumes measured by in vivo xenograft model. U87MG-expressing sense-DUSP6 were subcutaneously injected in immunosuppressed CD1 mice, as described in Messina et al. (2004). Flank (heterotopic) tumors displayed increased sizes compared with control group mice in the absence/presence of cisplatin treatment. All data are the means±s.d. of three independent experiments. The tumor weight was calculated from caliper measurements, as previously reported. Differences between groups were tested for statistical significance using non-parametric Wilcoxon matched pair Test. *Po0.001 as compared with the control vector; **Po0.001 as compared with the control vector treated with the same CDDP dosage.

fluorescence-activated cell sorting, is shown in tin-treated U87MG tumors failed in lowering tumor Figure 4b. As anticancer DNA-damaging cisplatin progression by direct cytostatic or cytotoxic effect. induces a prolonged activation of ERK1/2, whose Notably, we observed higher tumor size (approxima- outcomes on cell death/cell survival are cell-type tively 30%) in mice inoculated with sense-DUSP6 and dependent (Gozdz et al., 2003; Arany et al., 2004), we treated with cisplatin than in untreated control groups monitored ERK phosphorylation by western blot (Figure 4d). analysis during dose-response treatment of cisplatin on Dysregulated expression of DUSP6 was observed in U87MG-expressing sense- and antisense-DUSP6 several types of human tumor, however, its role in (Figure 4c). Immunoblot analysis showed that forced tumorigenesis has not been fully clarified. The expres- expression of DUSP6 was sufficient to inhibit phos- sion of DUSP6 in glioblastoma cancers was investigated phorylation of ERK, and increasing doses of cisplatin to monitor possible changes in mRNA or protein were not able to change it at all (as reported in U87MG expression levels. Quantification of the DUSP6 tran- DUSP6-infected cells). scripts showed an increased expression in several types Finally, to evaluate tumorigenesis, xenograft injec- of human glioblastoma both primary and long-term tions into nude mice were carried out, showing that cultures (Figure 1). These data were unexpected as in DUSP6-cell tumors grew significantly faster than their many cancers DUSP6 expression is downregulated, counterparts (Figure 4d). Thus, overexpression of according to its putative role as tumor suppressor. DUSP6 was per se sufficient for tumor development Moreover, despite evidence already suggesting the and hence became refractory to chemotherapy. Cispla- existence of two transcript variants (see Gene Bank

Oncogene DUSP6 regulation and function in glioblastoma cancer cells S Messina et al 3819 NM 001946.2 and NM 022652.2), data shown in this correlates with the invasive phenotype but, regretfully, study are the first experimental evidence of the existence the DUSP26 adhesive phenotypes were measured in of both mRNA (Supplementary Figure S1). heterologous expression systems. In our experiments, After some experimental troubles regarding antibody anchorage dependence and clonogenic potential, as well specificity, we biochemically confirmed the same results as all tumorigenic parameters, were measured in several obtained at mRNA level. Despite the endogenous lineages of human-malignant glioblastomas. More, in protein scarcely described in cancer cells (Furukawa other cancers, high levels of DUSP6 displayed a et al., 2003; Cui et al., 2006; Chan et al., 2008), DUSP6 migratory phenotype and formed colonies in soft agar protein was monitored, constitutively upregulated and (Nunes-Xavier et al., 2010). insensitive to acute mitogenic stimulation. Specifically, Enhanced tumor growth in xenograft model recapi- we detected a protein with the expected molecular tulates these altered adhesive properties. Interestingly, weight (45 kDa) expressed at comparable levels in DUSP6 is a selected gene for astrocytic classification in neoplastic cells as shown in Figure 1c. The mechanism angiogenic functional class and, recently, a potential leading to DUSP6 upregulation can be only hypothe- transcriptional effect of hypoxia on the DUSP6 gene has sized at this stage of knowledge. As stated by Bermudez been reported (Petalidis et al., 2008; Bermudez et al., et al. (2011), upregulation of DUSP6 expression may be 2011). Moreover, inducible expression of DUSP6 a way for cancer cells to maintain a level of ERK specifically blunts in vivo fibroblasts growth but, in our activity compatible with survival and cycle progression. opinion, this effect probably relies on Ha-Ras oncogene In this respect, it may explain why DUSP6 expression is transformation (Marchetti et al., 2004). Cisplatin treat- upregulated in many cancers harboring Ras mutations ment (both in vitro and in vivo) failed to sensitize (Warmka et al., 2004 and Bloethner et al., 2005). DUSP6-overexpressing glioblastoma cells in our experi- Our studies focused on exacerbation of existing ments. Remarkably, in ovarian cancer cells, resistance to tumorigenic phenotypes by enforced expression of cisplatin is described in DUSP6-depleted cells (Chan DUSP6 that we found upregulated in glioblastoma et al., 2008) and, in contrast, in breast cancers, DUSP6 samples. We employed adenovirus-mediated expression overexpression is reported as a novel mechanism of of sense-, antisense- and catalitically inactive mutant resistance (Cui et al., 2006). C293S in U87MG cells for viability, colony formation, This paper describes a novel unexpected tumor- DNA synthesis and cell-cycle analysis (Figures 2 and 3 promoting function of DUSP6 and has important and Supplementary Figure S3). As expected, enforced implication for therapeutic strategies that aim at expression of DUSP6 reduced ERK1/2 activity by manipulating DUSP6 expression or function for the inhibiting at some extent cell proliferation, as assayed treatment of GBM. by cell-cycle analysis (3H-thymidine incorporation and fluorescence-activated cell sorting) but, unexpectedly, Conflict of interest exacerbated the malignancy of glioblastoma cells. On the other hand, it may seem surprising not to observe a The authors declare no conflict of interest. strong impact of DUSP6 expression on viability. These counterintuitive mild effects probably depend on cytosolic retention of ERK1/2 by DUSP6 (Murphy and Acknowledgements Blenis, 2006; McKay and Morrison, 2007). Furthermore, glioblastoma malignancy relies on We wish to thank Dr Angela Santoni for her continued adhesive properties (infiltrative and radio- and chemore- support. We thank Dr de Franciscis and Dr Colucci D’Amato for providing DUSP6-MKP3 cDNA in the pSG5 vectors; Dr sistance) more than growth rates (Patterson et al., 2009). Furnari and Dr Cavenee for mutant EGFRvIII U87MG cells Recently, members of the DUSPs family (DUSP26 and and Dr Ricci-Vitiani for primary cultures of glioblastomas. We DUSP4) have been involved in determining the malig- especially thank Dr Johan Lennartson for the ‘home-made’ nant phenotype of glioblastomas (Tanuma et al., 2009; antibody anti-DUSP6/MKP3 and Dr Stephen Keyse for Waha et al., 2010). In both papers, downregulation critical reading of the paper.

References

Arany I, Megyesi JK, Kameto H, Price PM, Saﬁrstein LR. (2004). Calogero A, Lombari V, De Gregorio G, Porcellini A, Ucci S, Arcella Cisplatin-induced cell death is EGFR/src/ERK signaling dependent A et al. (2004). Inhibition of cell growth by EGR-1 in human in mouse proximal tubule cells. Am J Physiol Renal Physiol 287: primary cultures from malignant glioma. Cancer Cell Int 4: 1–12. F543–F549. Chan DW, Liu VW, Tsao GS, Yao KM, Furukawa T, Chan KK et al. Bermudez O, Jouandin P, Rottier J, Bourcier C, Gilles P, Gimond C. (2008). Loss of MKP3 mediated by oxidative stress enhances (2011). Post-transcriptional regulation of the DUSP6/MKP-3 tumorigenicity and chemoresistance of ovarian cancer cells. phosphatase by MEK/ERK signalling and hypoxia. J Cell Physiol Carcinogenesis 29: 1742–1750. 226: 276–284. Chen HY, Yu SL, Chen CH, Chang GC, Chen CY, Yuan A et al. Bloethner S, Chen B, Hemminki K, Mu¨ ller-Berghaus J, Ugurel S, (2007). A ﬁve-gene signature and clinical outcome in non-small-cell Schadendorf D et al. (2005). Effect of common B-RAF and N-RAS lung cancer. N Engl J Med 356: 11–20. mutations on global gene expression in melanoma cell lines. Cui Y, Parra I, Zhang M, Hilsenbeck SG, Tsimelzon A, Furukawa T Carcinogenesis 26: 1224–1232. et al. (2006). Elevated expression of mitogen-activated protein

Oncogene DUSP6 regulation and function in glioblastoma cancer cells S Messina et al 3820 kinase phosphatase 3 in breast tumors: a mechanism of tamoxifen Nunes-Xavier CE, Ta´ rrega C, Cejudo-Marı´ n R, Frijhoff J, Sandin A, resistance. Cancer Res 66: 5950–5959. Ostman A et al. (2010). Differential up-regulation of MAP kinase Furnari FB, Fenton T, Bachoo RM, Mukasa A, Stommel JM, Stegh A phosphatase MKP3/DUSP6 and DUSP5 by Ets2 and c-jun et al. (2007). Malignant astrocytic glioma: genetics, biology, and converge in the control of the growth arrest versus proliferation paths to treatment. Genes Dev 21: 2683–2710. response of MCF-7 breast cancer cells to phorbol ester. J Biol Chem Furukawa T, Fujisaki R, Yoshida Y, Kanai N, Sunamura M, Abe T 285: 26417–26430. et al. (2005). Distinct progression pathways involving the dysfunc- Okudela K, Yazawa T, Woo T, Sakaeda M, Ishii J, Mitsui H et al tion of DUSP6/MKP3 in pancreatic intraepithelial neoplasia and (2009). Down-regulation of DUSP6/MKP3 expression in lung intraductal papillary-mucinous neoplasms of the pancreas. Mod cancer: its mechanism and potential role in carcinogenesis. Am J Pathol 18: 1034–1042. Pathol 175: 867–881. Furukawa T, Sunamura M, Motoi F, Matsuno S, Horii A. (2003). Owens DM, Keyse SM. (2007). Differential regulation of MAP kinase Potential tumor suppressive pathway involving DUSP6/MKP3/ signalling by dual-specificity protein phosphatases. Oncogene 26: MKP-3 in pancreatic cancer. Am J Pathol 162: 1807–1815. 3203–3213. Gozdz A, Habas A, Jaworski J, Zielinska M, Albrecht J, Chlystun M Patterson KI, Brummer T, O’Brien PM, Daly LJ. (2009). Dual- et al. (2003). Role of N-methyl-D-aspartate receptors in the specificity phosphatases: critical regulators with diverse cellular neuroprotective activation of extracellular signal-regulated kinase targets. Biochem J 418: 475–489. ½ by cisplatin. J Biol Chem 278: 43663–43671. Petalidis LP, Oulas A, Backlund M, Wayland MT, Liu L, Plant K Hong L, Li X, Jin H, Yan L, Wu K, Ding J et al. (2007). Up-regulation et al. (2008). Improved grading and survival prediction of of tumor suppressor genes might promote the malignant phenotype human astrocytic brain tumors by artificial neural network of cancer cells. Med Hypotheses 69: 1379. analysis of gene expression microarray data. Mol Cancer Ther 7: Kim HS, Song MC, Kwak IH, Park TJ, Lim IK. (2003). Constitutive OF1–OF9. induction of p-Erk1/2 accompanied by reduced activities of protein Porcellini A, De Blasi A. (2004). Viral infection protocols. Methods phosphatases 1 and 2A and MKP3 due to reactive oxygen species Mol Biol 259: 155–166. during cellular senescence. J Biol Chem 278: 37497–37510. Porcellini A, Ruggiano G, Pannain S, Ciullo I, Amabile G. (1997). Li Y, Guessous F, Kwon S, Kumar M, Ibidapo O, Fuller L et al. Mutations of thyrotropin receptor isolated from thyroid autono- (2008). PTEN has tumor-promoting properties in the setting of mous functioning adenomas confer TSH-independent growth to gain-of-function p53 p53 mutations. Cancer Res 68: 1723–1731. thyroid cells. Oncogene 15: 781–789. Lucci MA, Orlandi R, Triulzi T, Tagliabue E, Balsari A, Villa-Moruzzi Ricci-Vitiani L, Pallini R, Biffoni M, Todaro M, Invernici G, Cenci T E. (2010). Expression profile of tyrosine phosphatases in HER2 et al. (2010). Tumour vascularization via endothelial differentiation breast cancer cells and tumors. Cell Oncol 32: 361–372. of glioblastoma stem-like cells. Nature 468: 824–828. Maher EA, Furnari FB, Bachoo RM, Rowitch DH, Louis DN, Sato M, Vaughan MB, Girard L, Peyton M, Lee W, Shames DS et al. Cavenee WK et al. (2001). Malignant glioma: genetics and biology (2006). Multiple oncogenic changes (K-RASV12, p53 knockdown, of a grave matter. Genes Dev 15: 1311–1333. mutant EGFRs, p16 bypass, telomerase) are not sufficient to confer Marchetti S, Gimond C, Chambard JC, Touboul T, Roux D, a full malignant phenotype on human bronchial epithelial cells. Pouysse´ gur J et al. (2005). Extracellular signal-regulated kinases Cancer Res 66: 2116–2128. phosphorylate mitogen-activated protein kinase phosphatase 3/ Silvani A, Gaviani P, Lamperti EA, Eoli M, Falcone C, Dimeco F DUSP6 at serines 159 and 197, two sites critical for its proteasomal et al. (2009). Cisplatinum and BCNU chemotherapy in primary degradation. Mol Cell Biol 25: 854–864. glioblastoma patients. J Neurooncol 94: 57–62. Marchetti S, Gimond C, Roux D, Gothie´ E, Pouysse´ gur J. (2004). Tanuma N, Nomura M, Ikeda M, Kasugai I, Tsubaki Y, Takagaki K Inducible expression of a MAP kinase phosphatase-3-GFP chimera et al. (2009). Protein phosphatase Dusp26 associates with KIF3 specifically blunts fibroblast growth and ras-dependent tumor motor and promotes N-cadherin-mediated cell-cell adhesion. formation in nude mice. J Cell Physiol 199: 441–450. Oncogene 28: 752–761. Mark JK, Aubin RA, Smith S, Hefford MA. (2008). Inhibition of Tullai JW, Schaffer ME, Mullenbrock S, Kasif S, Cooper GM. mitogen-activated protein kinase phosphatase 3 activity by inter- (2004). Identification of transcription factor binding sites upstream domain binding. J Biol Chem 283: 28574–28583. of human genes regulated by the phosphatidylinositol 3-kinase McKay MM, Morrison DK. (2007). Integrating signals from RTKs to and MEK/ERK signaling pathways. J Biol Chem 279: 20167–20177. ERK/MAPK. Oncogene 26: 3113–3121. Waha A, Felsberg J, Hartmann W, von dem Knesebeck A, Mikeska T, Messina S, Leonetti C, De Gregorio G, Affatigato V, Ragona G, Joos S et al. (2010). Epigenetic downregulation of mitogen-activated Frati L et al. (2004). Ras inhibition amplifies cisplatin sensitivity protein kinase phosphatase MKP-2 relieves growth suppressive of human glioblastoma. Biochem Biophys Res Comm 320: 493–500. activity in glioma cells. Cancer Res 70: 1689–1699. Muda M, Theodosiou A, Rodrigues N, Boschert U, Camps M, Warmka JK, Mauro LJ, Wattenberg EV. (2004). Mitogen-activated Gillieron C et al. (1996). The dual specificity phosphatases M3/6 protein kinase phosphatase-3 is a tumor promoter target in initiated and MKP-3 are highly selective for inactivation of distinct mitogen- cells that express oncogenic Ras. J Biol Chem 279: 33085–33092. activated protein kinases. J Biol Chem 271: 27205–27208. Yan H, Parsons DW, Jin G, McLendon R, Rasheed BA, Yuan W et al. Murphy LO, Blenis J. (2006). MAPK signal specificity: the right place (2009). IDH1 and IDH2 mutations in gliomas. N Engl J Med 360: at the right time. Trends Biochem Sci 31: 268–275. 765–773. Nagane M, Coufal F, Lin H, Bogler O, Cavenee WK, Huang HJ. Zhang Z, Kobayashi S, Borczuk AC, Leidner RS, Laframboise T, (1996). A common mutant epidermal growth factor receptor Levine AD et al. (2010). Dual specificity phosphatase 6 confers enhanced tumorigenicity on human glioblastoma cells by (DUSP6) is an ETS-regulated negative feedback mediator of increasing proliferation and reducing apoptosis. Cancer Res 56: oncogenic ERK signalling in lung cancer cells. Carcinogenesis 31: 5079–5086. 577–586.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)

Oncogene