STEAP1 Is Associated with the Invasive and Oxidative Stress Phenotype of Ewing Tumors

Ewing tumors are highly metastatic bone-associated cancers of enigmatic histogenesis mostly affecting children. Established therapies still have limited success in advanced stages of the disease despite high toxicity (1). Thus, selective and less toxic drugs are the prerequisites to reduce the toxic burden of cure.

Ewing tumors express chimeric EWS/ETS (Ewing sarcoma breakpoint region 1/E-twenty-six) fusion proteins derived from chromosomal translocations with EWS/FLI1 (Ewing sarcoma breakpoint region 1/friend leukemia virus integration 1) being the predominant one (85%). EWS/FLI1 encodes an oncogenic transcription factor that determines the complex and highly malignant phenotype of Ewing tumors (2). Hence, the detailed functional characterization of the EWS/FLI1-induced transcriptome may be key to understand the underlying mechanisms of the disease and ultimately to halt its progression (3, 4).

Previously, we identified a specific expression signature of approximately 40 genes that are highly upregulated in Ewing tumors compared with benign tissues (3). Part of this signature is the six-transmembrane epithelial antigen of the prostate 1 (STEAP1)—a membrane-bound protein possibly contributing to transmembrane electron transfer (5, 6).

Among STEAP proteins, only STEAP1 is overexpressed in many carcinomas including prostate and bladder cancer, where it locates to plasma and endosomal membranes (6, 7), but its precise cellular function remains elusive. Recently, STEAP1 was validated as a bona fide marker for mesenchymal stem cells (MSC; ref. 8) supporting the relationship of Ewing tumors with MSCs (9). Moreover, STEAP1 mRNA circulates in peripheral blood (6, 10), and its detection in bone marrow of patients with Ewing tumors is indicative for occult Ewing tumor cells (11).

In contrast, STEAP1 mRNA is not detectable in blood of healthy donors and is hardly expressed in benign tissues, except for low amounts in urothelium and prostate (6, 10). Because of its high tumor specificity and membrane-bound localization, STEAP1 might serve as a promising candidate for targeted therapy (6, 10, 11). In accordance, monoclonal antibodies against STEAP1 inhibit growth of xenografted prostate and bladder cancer (12).

STEAP proteins are homologues of NADPH oxidases (NOX; refs. 13, 14), which are involved in cellular reactive oxygen species (ROS) metabolism and frequently overexpressed in cancer (15). Like NOX, all STEAP proteins, except for STEAP1, are equipped with an N-terminal NADP⁺ oxidoreductase (7, 13). In addition, all STEAP members contain a C-terminal ferric oxidoreductase. Thus, a role in cellular iron homeostasis is assumed for these proteins (16). However, in contrast to other STEAP proteins, STEAP1 does not facilitate iron uptake and reduction, suggesting another distinct function (7).

ROS play a key role in oncogenic signaling and elevated ROS levels are a salient feature of many highly invasive cancers (17, 18), possibly including Ewing tumors (19). Sound evidence suggests that many malignancies take advantage of a permanently active “oxidative stress phenotype” leading to enhanced invasiveness, which has been recognized as an additional hallmark of cancer (20). On the basis of its homology to NOX and its overexpression in highly metastatic cancers, we hypothesized that STEAP1 is involved in the invasive behavior and oxidative stress phenotype of Ewing tumors.

In the present study, we investigate the putative oncogenic function of STEAP1. We prove that STEAP1 is induced by EWS/FLI1 and that its expression promotes proliferation, invasiveness, anchorage-independent colony formation, tumorigenicity, and metastasis of Ewing tumors. Moreover, transcriptome and proteome analyses as well as functional studies reveal that STEAP1 expression is associated with elevated ROS levels that regulate ROS-sensitive signaling molecules and proinvasive genes via STAT1.

Ewing tumor cell lines (MHH-ES1, SK-ES1, RDES, SK-N-MC, TC-71), neuroblastoma lines (CHP126, MHH-NB11, SHSY5Y, SIMA), rhabdomyosarcoma cell line RH-30, and B-cell leukemia lines (Nalm6, 697, cALL2) were obtained from the German Collection of Microorganisms and Cell Cultures (DSMZ). A673 was purchased from American Type Culture Collection (LGC Standards). SB-KMS-KS1, previously described as SBSR-AKS, is an Ewing tumor cell line with a type 1 EWS/FLI1 translocation established in our laboratory (4). Human MSCs L87 and V54.2 were immortalized with SV40 large T-antigen (4). Retrovirus packaging cell line PT67 was obtained from Takara Bio Europe/Clontech. Cells were grown at 37°C in 5% CO₂ in a humidified atmosphere in RPMI-1640 (Invitrogen) containing 10% FBS (Biochrom), 1% glutamine, and 100 μg/mL gentamycin (Invitrogen). Cell lines were checked routinely for purity (EWS/FLI1 translocation product, surface antigen, or HLA-phenotype) and Mycoplasma contamination. Reagents were purchased from Sigma, if not otherwise specified.

Gene expression was analyzed using TaqMan Universal PCR Master Mix, TaqMan Gene Expression Assays, and fluorescence detection with an AB 7300 Real-Time PCR System (Applied Biosystems). Results were normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and quantified by the ΔΔC_t method. Primers are listed in Supplementary Methods.

Transfection was described previously (4). For siRNA, see Supplementary Methods.

Chromatin immunoprecipitation was essentially carried out as described (4). For primers, see Supplementary Methods.

Procedures were described previously (4). For antibodies, see Supplementary Methods. Specificity of the STEAP1 antibody was assessed previously (21, 22).

The Institutional Review Board of the Technische Universität München (Munich, Germany) approved the current study. Archival tumor samples were obtained from the Department of Pathology of the Technische Universität München.

Procedures were described previously (4). See Supplementary Methods.

Experiments were essentially carried out as described previously (4). See Supplementary Methods.

Two-dimensional (2D) isoelectric focusing/SDS-PAGE were essentially carried out as described previously (23). For spot selection and mass spectrometry, see Supplementary Methods.

Cells were stained 48 to 72 hours after transfection as described (24). Samples were analyzed on a FACScalibur flow cytometer (Becton Dickinson). At least 30,000 events per sample were recorded. See Supplementary Methods.

Cell numbers were counted in real-time with a bioelectric xCELLigence instrument (Roche/ACEA Biosciences) monitoring impedance across gold micro-electrodes on the bottom of E-plates. Immediately after transfection with siRNA, 1.6 × 10⁴ cells were seeded in wells in 200 μL media containing 10% FBS. Cellular impedance was measured periodically. Transfection efficacy was controlled by Western blot analysis and/or quantitative real-time PCR (qRT-PCR).

A total of 5 × 10⁵ transiently transfected cells in 500 μL serum-free media was seeded into the upper chambers of Matrigel-covered Transwell plates (Becton Dickinson). Bottom chambers contained 500 μL media with 10% FBS. After 48 hours, invading cells were stained with 4 μg/mL CalceinAM (Becton Dickinson) in Hank's balanced salt solution and photographed with a Zeiss AxioCam MRm camera on a Zeiss Axiovert 100 microscope (Zeiss). The number of invading cells was normalized to proliferation as assessed with xCELLigence (Roche/ACEA Biosciences). N-Acetylcysteine (NAC) or H₂O₂ pretreatment of Matrigel plates did not affect invasiveness of untreated Ewing tumor cells plated subsequently.

See Supplementary Methods.

Procedures were described previously (4).

Immunodeficient Rag2^−/−γ_c^−/− mice on a BALB/c background were obtained from the Central Institute for Experimental Animals (Kawasaki) and maintained under pathogen-free conditions in accordance with the institutional guidelines and approval by local authorities. Experiments were carried out in 6- to 16-week-old mice. For in vivo tumor growth, 3 × 10⁶ Ewing tumor cells in 0.2 mL PBS were subcutaneously injected in groins. The amount of 2 × 10⁶ to 5 × 10⁶ cells has been previously reported to be optimal for assessment of local growth of Ewing tumor xenografts (4, 25). Mice bearing tumors greater than 10 mm in diameter (determined with a caliper) were considered positive (event). To analyze metastatic potential, tumor cells were injected intravenously. Five weeks later, mice were euthanized and metastasis was monitored in individual organs. All macroscopically visible metastases within an organ were counted. Tumors and affected tissues were excised for histology and gene expression analysis.

Procedures were described previously (26). See Supplementary Methods.

Cellular glutathione was assessed with a colorimetric assay kit following the manufacturer's instructions (Cayman Europe).

Procedures were described previously (27, 28).

Unpaired two-tailed Student t test or independent one-sample t test were used; P < 0.05 was considered significant.

Previously, we identified STEAP1 in an Ewing tumor expression signature (3). To substantiate this observation, we matched our microarray data against those of neuroblastomas and a normal body map composed of benign fetal and adult tissues. As seen in Fig. 1A and B, STEAP1 is highly expressed in Ewing tumors but virtually not expressed in neuroblastomas and benign tissues. High STEAP1 expression in Ewing tumors was confirmed by qRT-PCR and Western blot analysis (Fig. 1C). To test whether EWS/ETS transcription factors can induce STEAP1, we transfected MSCs with EWS/FLI1-containing vectors and observed a 5- to 6-fold increase of STEAP1 expression in 2 MSC lines (L87 and V54.2; Fig. 1D). Vice versa, RNA interference–mediated EWS/FLI1 silencing in Ewing tumor cells reduced STEAP1 expression (Fig. 1E). Moreover, the STEAP1 promoter contains 2 evolutionarily conserved ETS-binding sites (−1,465 and −250 bp upstream of the transcriptional start site), which proved to be enriched for FLI1 in chromatin immunoprecipitation (Fig. 1F). We next confirmed STEAP1 overexpression in primary Ewing tumors on protein level. As positive controls, we chose prostate cancer because of its known overexpression of STEAP1 (6). Figure 1G shows that among a series of tumors, which are usually included in the histologic differential diagnosis of Ewing tumors (29, 30), only Ewing tumors display very high STEAP1 levels. Consistently, the analysis of a sarcoma gene expression library (137 sarcomas; 14 entities) revealed that STEAP1 discriminates Ewing tumors from other sarcomas (sensitivity, 89.5%; specificity, 82.2%; ref. 31).

Because STEAP1 is overexpressed in Ewing tumors, we tested whether RNA interference–mediated STEAP1 silencing impacts the Ewing tumor phenotype. Using an xCELLigence instrument, we observed that STEAP1 knockdown reduced proliferation of Ewing tumor cells (Fig. 2A) without affecting rates of apoptosis and/or necrosis or inducing cell-cycle arrest (Supplementary Fig. S1). Moreover, STEAP1 silencing inhibited cellular invasiveness through Matrigel (Fig. 2B), whereas STEAP1 overexpression in Ewing tumor cells with low STEAP1 expression increased invasiveness and proliferation (Supplementary Fig. S1). Interestingly, STEAP1 levels correlated with invasiveness of different Ewing tumor cell lines (Supplementary Fig. S1). To evaluate the effect of long-term knockdown of STEAP1, we generated STEAP1 short hairpin RNA–expressing infectants of 2 Ewing tumor cell lines (SK-N-MC and SB-KMS-KS1). Constitutive STEAP1 silencing reduced colony formation of Ewing tumor cells in methylcellulose in a dose-dependent manner (Fig. 2C and Supplementary Table S1). Consistently, these infectants exhibited delayed tumor growth in Rag2^−/−γ_c^−/− mice (Fig. 2D and E). Persistence of STEAP1 knockdown was confirmed ex vivo in each xenograft (Fig. 2D). Comparing the xenografts with or without STEAP1 silencing, we did not find changes in immunohistochemistry for the apoptosis marker caspase-3 or for tumor-infiltrating macrophages (tested by MAC3; not shown). In addition, no differences in vascularization, quantified by CD31 staining (not shown), and in intratumoral necrosis were observed (Fig. 2F). Similarly to local tumor growth, experimental metastasis into the liver was diminished after STEAP1 knockdown (Fig. 2G). Although our cell lines showed a high propensity to metastasize into livers, we only noted kidney metastases in mice injected with control cells (pSIControl; not shown). Taken together, these results suggest that STEAP1 supports Ewing tumor growth and invasiveness.

In accordance with data in HEK-293T cells (7), we did not find evidence for STEAP1 to impact iron uptake and reduction in Ewing tumors (Supplementary Fig. S2). However, to gain functional insight into how STEAP1 influences Ewing tumor malignancy, we conducted whole transcriptome microarrays (GSE26422) to identify concordantly regulated genes in A673 and SK-N-MC cells after STEAP1 silencing (minimum mean log₂ fold change ±0.32; maximum variation of 40% across siRNAs). STEAP1 knockdown differentially regulates 87 genes (41 upregulated and 46 downregulated; Fig. 3A and Supplementary Table S2). STEAP1-dependent gene regulation was confirmed on selected genes after STEAP1 knockdown (Fig. 3B) and STEAP1 overexpression (Supplementary Fig. S2).

Among the 40 top-regulated genes, 20% were assigned to the ubiquitin–proteasome system (UPS), suggesting that STEAP1 might play a role in protein modification that requires enhanced proteasomal decay. Consistently, gene set enrichment analysis (GSEA) revealed that STEAP1 silencing regulates gene sets involved in oxidative stress responses, type II conjugation, and proteolysis (Supplementary Table S3), which are part of the oxidative stress phenotype seen in cancer (32, 33).

In support of the prediction that changes in oxidative stress responses influence overall proteome composition, STEAP1 knockdown altered the protein levels of 121 spots (81 spots upregulated and 40 spots downregulated) of 845 spots detected in 2D gel electrophoresis of SK-N-MC cells (minimum linear fold change ±2 in 3 independent experiments as assessed by densitometric analysis with PDquest Advanced; BioRad; P < 0.05).

The 24 most significantly regulated spots and 1 nonregulated control spot were excised for proteomic analysis. By mass spectrometry, 132 different proteins were identified, of which 17% were assigned to protein transport and folding, 13% to invasion, and 11% to the redox system according to their gene ontology (GO) annotations (Fig. 3C). Among these proteins are several redox enzymes such as peroxiredoxins and superoxide dismutases, which are dysregulated in a large cohort of cancers (15, 34). Moreover, endosomal redox stress response proteins (35, 36) were identified including T-complex chaperones, heat shock proteins, protein-disulfide isomerases as well as co-chaperones. Furthermore, we identified mediators of posttranscriptional mRNA processing such as heterogeneous nuclear ribonucleoproteins, THO complex proteins, and eukaryotic translation initiation factors (Supplementary Table S4). Notably, we found dysregulation of 2 of 6 key modules of the hexameric proteasomal ATPase (PSMC5 and PSMC6) and identified 2 further subunits (PSMC3 and PSMC4) in regulated spots by proteomic means as well as key modules of the proteolytic cavities of the proteasome and immunoproteasome (PSMA3, PSMB5, PSMB8, and PSMB9), which are critical for protein quality control upon oxidative stress (ref. 37; Supplementary Fig. S2). These data were confirmed by GSEA pathway analysis (Table 1). In summary, these analyses suggest that STEAP1 silencing leads to adaptations in oxidative stress response systems, supporting the hypothesis that STEAP1 is associated with the oxidative stress phenotype of Ewing tumors.

We next investigated whether the long-term knockdown of STEAP1 can lead to altered ROS levels. Indeed, constitutive STEAP1 silencing decreased ROS-levels (Fig. 4A), whereas STEAP1 overexpression increased ROS levels (Fig. 4B) as quantified by dihydroethidium fluorescence. Moreover, STEAP1 knockdown reduced mitochondrial ROS without changing mitochondrial mass as quantified by MitoSOX Red and MitoTracker Green staining, respectively (Fig. 4C and D). Consistently, STEAP1 knockdown decreased the cellular pool of oxidized glutathione (glutathione disulfide) but increased the amount of reduced glutathione (GSH; Fig. 4E). Interestingly, reassessment of our microarray data revealed that STEAP1 is most prominently expressed in Ewing tumors among other NOX and STEAP proteins (Supplementary Fig. S3). It should be noted that we did not detect obvious morphologic changes of Ewing tumor cells and their mitochondria upon STEAP1 knockdown (Fig. 4F).

Our data and recent discussion in the literature (19) indicate that ROS might promote Ewing tumor aggressiveness. In accordance, treatment of Ewing tumor cells with the antioxidant NAC reduced colony formation, proliferation, and invasiveness of Ewing tumor cells in a dose-dependent manner (Fig. 5A–C). Similar results have been obtained with PEGylated-superoxide dismutase and PEGylated-catalase (Supplementary Fig. S4). Reciprocally, treatment of STEAP1 silenced Ewing tumor cells with H₂O₂ rescued the invasive phenotype of STEAP1 knockdown cells (Fig. 5D).

We next tested whether STEAP1-regulated genes contribute to the oxidative stress phenotype observed. We choose MMP-1, ADIPOR1, and DTX3L for follow-up due to their high expression in Ewing tumors and their involvement in oxidative stress responses (34, 38, 39). Notably, expression of MMP-1, ADIPOR1, and DTX3L appeared to be ROS sensitive in Ewing tumor cells as H₂O₂ treatment induced their expression in a time-dependent manner (Fig. 5E). Moreover, H₂O₂ treatment rescued MMP-1, ADIPOR1, and DTX3L, but not STEAP1, expression in a dose-dependent manner implying that STEAP1 is upstream of ROS signaling whereas the other genes are potentially downstream (Fig. 5F). Interestingly, silencing of these genes reduced invasiveness through Matrigel (Fig. 5G), whereas only ADIPOR1 knockdown significantly reduced Ewing tumor proliferation (Fig. 5H). In summary, these data suggest that oxidative stress may support Ewing tumor aggressiveness possibly, in part, via enhanced expression of MMP-1, ADIPOR1, and DTX3L.

Using GSEA to search for common transcription factor motifs within the 87 differentially regulated genes after STEAP1 silencing, we identified STAT1 (P = 0.07) and its downstream cofactors, IFN response factors (IRF) 1, 2, 7, and 8 (P < 0.05; ref. 40) as top-ranked putative STEAP1 targets. Consistently, interrogation of the GSEA Molecular Signatures Database (C2; v3.0) with these 87 genes revealed a strong overrepresentation of gene sets involved in IFN signaling accompanying STEAP1 expression (Fig. 6A and B). Interestingly, STAT1, a downstream effector of ROS and IFNs (41), is predominantly expressed in Ewing tumors (Fig. 6C). Of note, alike STEAP1 silencing, STAT1 silencing reduced the expression of STEAP1 target genes but left STEAP1 expression unaffected. Moreover, the downregulation of these genes could (apart from MMP-1 in 2 of 3 cell lines tested) not be rescued by H₂O₂, suggesting that STAT1 may be downstream of ROS and STEAP1 (compare Figs. 6D and 5F). Indeed, STEAP1 silencing results in less phosphorylated STAT1, which can be rescued by exogenous H₂O₂ and IFN-γ (Fig. 6E). However, Ewing tumor cells virtually do not endogenously produce IFNs and STEAP1 silencing neither alters their expression nor their secretion as seen by microarray (Fig. 6F) and ELISPOT analyses (Supplementary Fig. S5). In summary, these data indicate that STEAP1 and ROS may, in part, mediate their transcriptional effects via IFN-independent activation of STAT1 (Fig. 6G).

The current study assessed the involvement of STEAP1 in the invasive and oxidative stress phenotype of Ewing tumors. We show that STEAP1 is induced by EWS/FLI1 and important for Ewing tumor malignancy. Moreover, our data support a model whereby STEAP1 expression is linked to the maintenance of oxidative stress of Ewing tumors and increased Ewing tumor aggressiveness, probably mediated via STAT1. We show that STEAP1 is highly expressed in Ewing tumors compared with benign tissues and a series of other sarcomas implying that STEAP1 could be used in routine pathology as an additional marker for Ewing tumor diagnosis.

Our data indicate that STEAP1 is important for anchorage-independent colony formation and invasiveness of Ewing tumor cells in vitro and for tumorigenicity and metastasis in vivo. Moreover, we show that STEAP1 expression correlates with increased cellular ROS levels, which in turn induce the redox-sensitive and proinvasive genes MMP-1, DTX3L, and ADIPOR1. Consistently, Pan and colleagues found that STEAP1 overexpression promotes ROS-mediated hyperproliferation of thyroid epithelial cells (42).

ROS participate in oncogenic signaling and elevated ROS levels are a salient feature of aggressive cancers (17, 18). Disturbances in the delicate ROS balance can lead to protein misfolding, accumulation of dysfunctional proteins, and activation of cellular stress responses (36). In accordance, STEAP1 silencing provokes transcriptional and posttranscriptional adaptations of oxidative stress responses comprising the redox, chaperone, endopeptidase, and UPS. These observations are compatible with the hypothesis that STEAP1 is associated with an enhanced oxidative stress phenotype in Ewing tumors. Consistently, we found that STEAP1 positively correlates with cytoplasmic and mitochondrial ROS levels. As mitochondrial morphology remained unaffected, we suppose that mitochondrial ROS levels change concomitantly with cytoplasmic ROS levels upon STEAP1-silencing, a mechanism also seen in the context of NOX proteins (“ROS-cross-talk”; ref. 43). The role of STEAP1 in ROS modulation is supported by in silico predictions and crystallography of STEAP proteins defining them as heme-containing redox enzymes (5, 13). Less aggressive ROS are well known to interact with heme iron of heme-containing proteins (15). Here, a nonenzymatic 2-electron oxidation of heme generates ferryl-heme and an unstable free radical that may be released as more aggressive ROS in a site-specific manner depending on the localization of the protein (44, 45). As STEAP1 contains a ferric oxidoreductase, it is tempting to speculate that the protein generates ROS by itself (7, 42). However, although STEAP1 target genes appear to be downstream of ROS (Fig. 5F), it remains possible that part of STEAP1-induced ROS merely accompany upstream events involved in the STEAP1 phenotype.

Multiple studies proved that permanently elevated ROS levels activate prometastatic and proproliferative signaling in cancer (17, 19, 20). In agreement, we provide evidence that antioxidants reduce colony formation, proliferation, and invasion of Ewing tumor cells suggesting that Ewing tumors may benefit from an activated oxidative stress phenotype. Among STEAP1-regulated genes, we focused on MMP-1, ADIPOR1, and DTX3L all of which are implicated in ROS signaling. For instance, MMP-1 has been shown to be highly ROS inducible (34, 46) and its overexpression increases invasiveness and metastasis of a variety of cancers (47). ADIPOR1 is the cognate receptor for adiponectin, which stimulates proliferation of hematopoietic stem cells (48). Although the precise function of the ubiquitin ligase DTX3L is not defined, recent work suggests that DTX3L monoubiquitylates histone H4 lysine 91 and thereby protects DNA from ROS (39). Here, we show that these genes are highly inducible by ROS and regulated by STAT1. STAT1 had been traditionally viewed as a mere IFN signal transducer but has recently been linked with aggressiveness, therapy resistance, and oxidative stress responses of several cancers (49, 50). Consistently, we prove that the coordinated expression of MMP-1, ADIPOR1, and DTX3L fosters the invasive and proliferative phenotype of Ewing tumors.

In summary, this work for the first time provides evidence that an activated oxidative stress phenotype enhances Ewing tumor malignancy: we show that STEAP1 overexpression promotes proliferation, invasiveness, anchorage-independent colony formation, tumorigenicity, and metastasis of Ewing tumors. Because STEAP1 is overexpressed in a wide variety of carcinomas, its oncogenic function may have general relevance for tumor progression and targeted therapy.

No potential conflicts of interest were disclosed.

The authors thank Barbara Grunewald and Veit Buchholz for fruitful discussion and Ulrich Welsch for his help with electron microscopy.

This work was supported by the TU München (B08-05; A09-02), the Deutsche Forschungsgemeinschaft (GR3728/1.1, SFB/TR22, and EB740/6.1), the Wilhelm-Sander-Stiftung (2009.901.1), the Ministerium für Innovation, Wissenschaft und Technologie des Landes Nordrhein-Westfalen, the EU 7thFP Metoxia, the Bundesministerium für Bildung und Forschung (BMBF), and the Translational-Sarcoma-Research-Network supported by the BMBF (FK 01GM0870).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.