Oxidative Stress Induces ADAM9 Protein Expression in Human Prostate Cancer Cells

Research Article

Shian-Ying Sung,1,2 Hiroyuki Kubo,1,6 Katsumi Shigemura,1 Rebecca S. Arnold,4 Sanjay Logani,4 Ruoxiang Wang,1 Hiroyuki Konaka,1 Masayuki Nakagawa,6 Spiro Mousses,7 Mahul Amin,4 Cynthia Anderson,2 Peter Johnstone,2 John A. Petros,1,4 Fray F. Marshall,1 Haiyen E. Zhau,1 and Leland W.K. Chung1,3,5

1Molecular Urology and Therapeutics Program, Department of Urology, 2Department of Radiation Oncology, 3Winship Cancer Institute; Departments of 4Pathology, 5Biochemistry, and Hematology/Oncology, Emory University School of Medicine, Atlanta, Georgia; 6Department of Urology, Faculty of Medicine, Kagoshima University, Kagoshima, Japan; and 7Translational Genomics Research Institute, Gaithersburg, Maryland

Abstract and survival of cancer cells (1). Recent data suggest that chronic The ADAM (a disintegrin and metalloprotease) family is a inflammation of the prostate and production of reactive oxygen group of transmembrane proteins containing cell adhesive species (ROS) could contribute to DNA damage and genomic and proteolytic functional domains. Microarray analysis instability, which may facilitate subsequent progression of cancer cells (2). It also has become apparent that oxidative stress such as detected elevated ADAM9 during the transition of human LNCaPprostate cancer cells from an androgen-dependent to toxins, dietary fat consumption, or high level of androgen may be an androgen-independent and metastatic state. Using a important etiologic factors in the development and progression of prostate tissue array (N = 200), the levels of ADAM9 protein prostate cancer (3). Fisher et al (4) showed that oxidative and expression were also elevated in malignant as compared with osmotic stresses on tumor cells increase the shedding of pro- benign prostate tissues. ADAM9 protein expression was found heparin-binding epidermal growth factor (EGF), which is processed in 43% of benign glands with light staining and 87% of by proteins of the ADAM family, specifically ADAM9, ADAM10, and malignant glands with increasing intensity of staining. We ADAM17. There are concomitant inductions of ADAM9 (5) and found that ADAM9 mRNA and protein expressions were ROS (6) in prostate carcinogenesis. It seems that ROS can induce elevated on exposure of human prostate cancer cells to stress expression of the ADAMs via p38 mitogen-activated protein kinase activation (4). The induced ADAM proteins are responsible for conditions such as cell crowding, hypoxia, and hydrogen peroxide. We uncovered an ADAM9-like protein, which is release of processed heparin-binding EGF, which further promotes predominantly induced together with the ADAM9 protein by a cancer cell growth and survival through an EGFR-dependent brief exposure of prostate cancer cells to hydrogen peroxide. mechanism. Induction of ADAM9 protein in LNCaPor C4-2 cells can be Recent studies showed a relationship between the elevation of completely abrogated by the administration of an antioxidant, ADAMs and the progression and metastasis of cancer cells (7–11). ebselen, or genetic transfer of a hydrogen peroxide degrada- Among them, ADAM9 has been shown to possess potent biological tive enzyme, catalase, suggesting that reactive oxygen species activities. The metalloprotease domain of ADAM9 could cleave h a h (ROS) are a common mediator. The induction of ADAM9 by insulin -chain, tumor necrosis factor , gelatin, -casein, and stress can be inhibited by both actinomycin D and cyclohex- fibronectin (12, 13), and induce shedding of EGF, fibroblast growth imide through increased gene transcription and protein factor receptor 2IIIB (14), or heparin-binding EGF-like growth synthesis. In conclusion, intracellular ROS and/or hydrogen factor (4, 15). The disintegrin domain of ADAM9 contains an ECD peroxide, generated by cell stress, regulate ADAM9 expression. (Glu-Cys-Asp) motif, which may mediate cellular adhesion through a h (16) and a h integrins (17). The cytoplasmic tail of ADAM9 ADAM9 could be responsible for supporting prostate cancer 6 1 v 5 cell survival and progression. By decreasing ADAM9 expres- has potential SH3 binding motifs, which have been reported to interact with potentially important regulatory proteins, endophilin sion, we observed apoptotic cell death in prostate cancer cells. h (Cancer Res 2006; 66(19): 9519-26) 1 (SH3GL2), SH3PX1 (18), and mitotic arrest deficient 2 (19). The abilities of ADAM9 to degrade specific extracellular matrix substrates, release active growth factors, interact with key Introduction regulatory factors, and appear in the invasion front of several Prostate cancer progression, invasion, and metastasis are tumor metastases suggest that ADAM9 provokes cancer cell complex processes involving the interplay between cancer cells growth, invasion, and metastasis. This conclusion is further and microenvironment with tight regulations to ensure the growth supported by a recent report where ADAM9 knockdown mice, when crossed with transgenic mice bearing prostate carcinoma, exhibited reduced tumor growth and local invasion (14). Note: S-Y. Sung and H. Kubo contributed equally to this work. In this communication, we show increased ADAM9 expression in Requests for reprints: Shian-Ying Sung, Department of Urology, Emory University prostate cancer cells on androgen-independent progression and by School of Medicine, Suite 5100, Room B5101, 1365B Clifton Road, Atlanta, GA 30322. Phone: 404-778-3663; Fax: 404-778-3675; E-mail: [email protected] or Leland W.K. exposing prostate cancer cells to stress conditions. Our findings on Chung, Molecular Urology and Therapeutics Program, Department of Urology, Emory the roles of ROS in the induction of ADAM9 protein expression University School of Medicine, Atlanta, GA 30322. Phone: 404-778-3672; E-mail: provide a framework for the development of future antioxidant [email protected]. I2006 American Association for Cancer Research. therapies to arrest the growth of clinical human prostate cancer doi:10.1158/0008-5472.CAN-05-4375 and its distant metastasis. www.aacrjournals.org 9519 Cancer Res 2006; 66: (19). October 1, 2006

Materials and Methods confluence were treated with different concentrations of H2O2 for 0, 0.5, 1, 2, 4, and 8 hours. In some experiments, ebselen was added to the cells for Cell lines, cell culture, and chemical reagents. The androgen- 30 minutes before exposure to the H O . Cells were cultured and harvested dependent and androgen receptor–positive parental LNCaP human 2 2 4 hours later. To determine whether ebselen could affect ADAM9 expression prostate cancer cell line and its lineage-derived androgen-independent, induced by cell crowding, cells were first cultured under the crowding androgen receptor–positive, metastatic C4-2 and C4-2B sublines were used condition for 48 hours and then treated with ebselen (0.02-20 Amol/L) for for this study. Unless otherwise stated, these cell lines were cultured in 2 hours. T-medium (Invitrogen, Carlsbad, CA) supplemented with 5% fetal bovine ADAM9 expression in clinical prostate cancer specimens. Clinical serum (FBS; Sigma, St. Louis, MO), 100 IU/mL penicillin G, and 100 Ag/mL prostate cancer tissue specimens were obtained following an institution- streptomycin at 37jC under 5% CO . Ebselen (2-phenyl-1, 2-benzisoselena- 2 approved protocol. Paraformaldehyde-fixed, paraffin-embedded tissues zol-3[2H]-one), actinomycin D, or cycloheximide (Sigma) was freshly were deparaffinized and rehydrated. Antigen retrieval was done in citrate reconstituted with DMSO before use. buffer (pH 6) using an electric pressure cooker at 120jC for 5 minutes. The Generation of human prostate cancer cell lines that stably express section was then exposed to 3% H O for 5 minutes to quench endogenous catalase. Construction and overexpression of catalase was done as 2 2 tissue peroxidase Iand then incubated for 30 minutes with the monoclonal previously described (20). In brief, human catalase cDNA was recovered anti-ADAM9 antibody (1:50 dilution). Immunohistochemical staining was from the pCI-neo vector and ligated into the HindIII site of the pZeoSV visualized with the Envision system (DAKO Corporation, Carpinteria, CA) vector. A clone containing the catalase in the sense orientation (pZeoSV/ and the results were independently evaluated by two pathologists (S.L and catalase) was verified by DNA sequencing. Transfection of the cDNA to M.A.). The anti-ADAM9 antibody from R&D Systems has been validated by prostate cancer cells was carried out with Fugene 6 (Roche Applied Science, several other groups to be specific for the detection of ADAM9 protein Indianapolis, IN) according to the instructions of the manufacturer. After expression in pancreatic cancer (8, 22), small-cell lung carcinoma (23), and 2 days, the cells were subcultured into 10-cm plates and treated with hepatic cancer (24). 0.4 mg/mL zeocin (Invitrogen) until individual colonies formed. Short hairpin RNA mediated gene silencing of ADAM9. To determine Quantitative reverse transcription-PCR. Quantitative reverse tran- the potential functions of ADAM9 protein in C4-2 cells, we transfected these scription-PCR (RT-PCR) was done with the iCycler thermal cycler system cells with a predesigned short hairpin RNA (shRNA; Invitrogen), which was (Bio-Rad, Hercules, CA). Total RNA (5 Ag) from cells at 80% to 95% shown to knock down specifically ADAM9 mRNA and ADAM9 and ADAM9- confluence was used as the starting material for reverse transcription like protein. C4-2 cells were seeded in six-well plates and cultured until using the Moloney murine leukemia virus reverse transcriptase system 90% confluence. The cells were then transfected with 50 nmol/L ADAM9 (Invitrogen). In brief, quantitative RT-PCR was done with a CYBR green shRNA using Lipofectamine 2000 transfection reagent (Invitrogen) according RT-PCR kit (Applied Biosystems, Foster City, CA). A 25-AL reaction to the recommendations of the manufacturer and cultured for an additional contained 1Â CYBR green reaction mix with 50 ng of primers and 0.4 AL 6 hours in serum-free RPMI. Cells were washed and cultured in T-medium for of cDNA from the reverse transcription. Forty cycles of PCR reactions were an additional 18 hours. The cells were evaluated by Western blot analysis done with 30 seconds of denaturing at 95jC, 30 seconds of annealing at for expression of ADAM9 proteins with h-actin as an internal control. 60jC, and 1 minute of PCR reaction at 72jC. Fold of induction of expression Flow cytometry analysis of cell apoptosis. C4-2 cells were seeded at was determined according to a published method (21). Average Ct value 2 Â 105 per well in six-well dishes overnight before ADAM9 gene (n = 3) was used to measure the fold change. The formula used to calculate knockdown. shRNA knockdown of ADAM9 was done as described above fold changes was DCt = Ctsample À Ct18SrRNA; ÀDDCt = À(unknown DCT À À parental DCt) and the fold change was 2 DDCt (21). Primers used were with scramble shRNA or liposome as controls. Cells were cultured for an 18S-F1, 5¶-GTTCCGACCATAAACGATGCC-3¶; 18S-R1, 5¶-TGGTGGTGCCCT- additional 48 hours before harvesting for Annexin V flow cytometry analysis TCCGTCAAT-3¶; ADAM9-1088(+), 5¶-GTTCCTGTGGAGCAAAGAGC-3¶; and according to the recommendations of the manufacturer (Invitrogen). In ADAM9-1274(À), 5¶-CCAGCGTCCACCAACTTATT-3¶. brief, cells were collected and washed with cold PBS and diluted in Annexin- Western blot. Prostate cancer cell lines cultured at 80% to 95% con- binding buffer [10 mmol/L HEPES, 140 mmol/L NaCl, and 2.5 mmol/L Â 6 fluence were lysed in a lysis buffer [50 mmol/L Tris (pH 8.0), 150 mmol/L CaCl2 (pH 7.4)] with 1 10 cells/mL. Cells were incubated with Annexin V at room temperature for 15 minutes before analysis by flow cytometry NaCl, 0.02% NaN3, 0.1% SDS, 1% NP40, 0.5% sodium deoxycholate] containing 100 Amol/L phenylmethanesulfonyl fluoride and 1Â Protease Inhibitor (BD FACS) to determine the population of cells with or without Annexin V Cocktail Complete (Roche Diagnostics, Indianapolis, IN). Protein concen- by the use of anti-Annexin V antibody (Invitrogen) with a green filter and Â 4 trations of the lysates were determined with the BCA Protein Assay Kit counting 1 10 C4-2 cells. (Pierce, Rockford, IL). Unless otherwise stated, 20 Ag of total protein were Hypoxia culture conditions. C4-2 cells were seeded at 75% confluence fractionated on an SDS-PAGE. Protein was transferred onto a nitrocellulose overnight and then exposed to low oxygen tensions in humidified airtight membrane, which was incubated with anti-human ADAM9 monoclonal chambers with inflow and outflow valves. The chambers were flushed with antibody (1 Ag/mL; R&D Systems, Inc., Minneapolis, MN), mixed with preanalyzed gas (1% O2,5%CO2, balance N2; SGS, Atlanta, GA) mixtures and j antihuman EF1-a or actin antibody as internal control, and visualized by continuously maintained at 37 C for 24 hours. Normal cell culture enhanced chemiluminescence (Amersham Biosciences, Piscataway, NJ). conditions were defined by placing cells in an incubator with 5% CO2 Intensity of the detected signals by Western blot was quantified with and 95% air for 24 hours. Cells were harvested and total cellular protein was Image J8 after scanning and inverting the images into TIFF files and con- extracted and subjected to Western blot analysis for ADAM9 protein verting into grayscale mode. Relative ADAM9 expression was semiquantified expression. after normalization with either endogenous h-actin or EF1-a as indicated. Assessment of endogenous ROS concentration. Intracellular ROS Effects of cell crowding on ADAM9 expression. Effects of the crowding was determined according to a published protocol (25, 26). Confluent cells on ADAM9 expression were evaluated after growing cells to 70% or 100% (85-90%) were rinsed with HBSS and detached by trypsin treatment. confluence. For low-density culture, 8 Â 105 cells were seeded onto 10-cm Cells were pelletted, resuspended in HBSS, and incubated in 2.5% FBS A dishes to avoid overlapping and clumping of the cells. For crowding and 2 mol/L dichlorofluorescein diacetate in the dark at room temperature conditions, 6 Â 106 cells were seeded. Cells were harvested 48 hours after for 45 minutes. Fluorescence was determined with FACSCalibur from seeding and Western blot analysis of ADAM9 expression was done. Becton Dickinson (San Jose, CA) (excitation wavelength, 488 nm; emission Effects of hydrogen peroxide on ADAM9 protein expression. wavelength, 545 nm). Effects of actinomycin D and cycloheximide on ADAM9 expression. Hydrogen peroxide (H2O2) and ebselen, an antioxidant, were used to evaluate the contribution of ROS to ADAM9 expression. Cells at 90% To examine if ADAM9 expression was regulated at the level of transcription or translation, we employed an RNA synthesis inhibitor, actinomycin D, and

a protein synthesis inhibitor, cycloheximide, to inhibit the H2O2-induced 8 http://rsb.info.nih.gov/ij/. ADAM9 expression. Cells at 90% confluence were exposed to H2O2 in

Cancer Res 2006; 66: (19). October 1, 2006 9520 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on October 1, 2021. © 2006 American Association for Cancer Research. ADAM9 Expression and Reactive Oxygen Species the presence of actinomycin D (1 Ag/mL) or cycloheximide (10 Ag/mL) for found in androgen-independent, androgen receptor–negative PC3 2 hours. The cells were harvested and ADAM9 protein expression was and DU-145 cells in which cell crowding increased ADAM9 protein assessed by Western blots. expression by 2.0- and 2.6-fold, respectively. Likewise, when cells were cultured under hypoxic conditions, a 1.3-fold increase in ADAM9 Results protein expression was noted in C4-2 cells (Fig. 1D). While evaluating Increased ADAM9 protein expression in the lineage-related ADAM9 protein overexpression under stress conditions, such as LNCaPhuman prostate cancer progression model. We used the crowding and hypoxia, we noted that, in addition to the increase lineage related prostate cancer cell lines, androgen-dependent of ADAM9 protein expression (f80 kDa protein), a smaller size of LNCaP and androgen-independent C4-2 (27), as models to identify immunoreactive protein, tentatively defined as an ADAM9-like genes of which the expression is differentially activated in prostate protein in the molecular weight range of 60 to 65 kDa, reacted cancer progression. Microarray analysis was used to compare gene positively with ADAM9 antibody and was induced under stress expression profiles between lineage-related androgen-dependent conditions and by hydrogen peroxide (Figs. 1C and 3B). LNCaP and androgen-independent C4-2 and C4-2B cells (27, 28). Increased ADAM9 protein expression in clinical specimens. These analyses suggested a differentially elevated expression of To confirm that ADAM9 and/or ADAM9-like expression in ADAM9 in the androgen-independent derivative sublines of the response to stress conditions in cultured human prostate cancer LNCaP. We confirmed differential ADAM9 mRNA and protein cells is relevant to clinical conditions, tissue microarray of ADAM9/ expression in LNCaP lineage cells with quantitative RT-PCR and ADAM9-like expression was done in clinical prostate cancer Western blot, respectively. As shown in Fig. 1A, a 2.8- and 3.3-fold specimens. Tissue microarray was conducted in 200 prostate increase in ADAM9 mRNA expression was noted in C4-2 and specimens (80 benign hyperplastic prostate and 120 prostate C4-2B cells, respectively, over that of the parental LNCaP cells. cancer; Table 1). Results of these studies revealed that 57.5% Western blot analyses corroborated that ADAM9 mRNA and ADAM9 (46 of 80) of benign foci and 12.5% (15 of 120) of cancer were protein expression in C4-2 and C4-2B cells were also 1.5- to 1.4-fold negative for ADAM9/ADAM9-like protein expression. In the higher than parental LNCaP cells (Fig. 1B). Under stress conditions, cancerous foci, the distribution of intensity of the 87.5% of the such as cell crowding from 70% to 100% confluence, an additional positively stained samples was as follows: 19.2% (23 of 120) showed increase of ADAM9 protein expression by 1.6- and 2.0-fold over 1+, 20.8% (25 of 120) showed 2+, and 47.5% (57 of 120) showed 3+. those of the basal levels was observed in C4-2 and C4-2B cells, Location of the positive ADAM9/ADAM9-like stain was found to respectively (Fig. 1C). This stress induction of ADAM9 also can be be associated with the cell membrane or apical surface of cancer

Figure 1. ADAM9 protein expression in human prostate cancer cells increases on progression from androgen-dependent to androgen-independent state. A, real-time quantitative RT-PCR analysis of ADAM9 transcripts showed significant (*, P V 0.05) elevation of ADAM9 mRNA on progression of human prostate cancer cells from androgen-dependent (LNCaP) to androgen-independent state (C4-2 and C4-2B). B, Western blot analysis of ADAM9 expression in prostate cancer cell lines. ADAM9 protein was elevated on progression of LNCaP (1.0Â) to C4-2 (1.5Â) and C4-2B (1.4Â). C, ADAM9 protein was elevated by 1.1-, 1.6-, and 2.0-fold, respectively, in LNCaP, C4-2, and C4-2B cells, and 2.0- and 2.6-fold, respectively, in PC-3 and DU-145 cells on cell culture in crowding conditions. Note that ADAM9-like protein also increases in crowding conditions. D, ADAM9 protein was also shown to be elevated by 1.3-fold under hypoxic cell culture conditions. www.aacrjournals.org 9521 Cancer Res 2006; 66: (19). October 1, 2006

Table 1. Tissue array analysis of ADAM9 expression in prostate samples

Samples ADAM9 positive stain 1+ 2+ 3+ Negative

n Total no. n % n % n % n % n %

Benign 80 34 42.5 34 42.5 0 0 0 0 0 0

Gleason 6 113 101 89 19 25 57 12 Cancer 120 19.2 20 47.5 12.5 Gleason 7 7 4 57 4 0 0 3

Stroma 200 0 0 0 0 0 0 0 0 200 100 cells (Fig. 2A). In contrast, in benign hyperplastic prostate tissues, relatively small sample size, we cannot draw a direct correlation only 1+ stain (42.5%; 34 of 80) was detected. None of the samples between increased ADAM9/ADAM9-like protein expression and scored 2+ or 3+. In this study, protein expression was found to the Gleason score of prostate cancer tissues (113 of 120 in Gleason be absent from prostate stromal cells (0 of 200). Because of a V6 as opposed to 7 of 120 in Gleason z7). The specificity of this

Figure 2. Immunohistochemical study of ADAM9 expression in clinical prostate cancer specimens. Human prostate cancer tissues were stained with ADAM9 antibody. A, stronger ADAM9 protein staining was observed at a magnification of Â20 in malignant (closed arrows) than in benign (open arrows, marked N) prostate tissues. B, ADAM9 knockdown by shRNA resulted in decreased expression of both 80-kDa and 65-kDa protein in C4-2 cells under crowding culture condition (95% confluence at the time of cell harvest). We observed a down-regulation of ADAM9 protein (80 kDa) in noncrowding condtions (75% confluence) at the time of cell harvest. C, knockdown of ADAM9 expression using shRNA strategy also resulted in an apoptosis of C4-2 cells as revealed by Annexin V flowcytometry analyses.

Cancer Res 2006; 66: (19). October 1, 2006 9522 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on October 1, 2021. © 2006 American Association for Cancer Research. ADAM9 Expression and Reactive Oxygen Species monoclonal antibody has been extensively validated by several ADAM9 protein expression up-regulated by hydrogen other groups of investigators in which they showed that ADAM9 peroxide requires active gene transcription and translation: immunoreactivity was blocked by a sequence-specific ADAM9 effects of actinomycin D and cycloheximide. To determine if the adsorbing peptide in pancreatic carcinoma and small-cell lung up-regulation of ADAM9 protein by H2O2 may require active mRNA carcinoma (8, 22, 23). We have also confirmed the specificity of this and protein syntheses, we monitored the steady state of ADAM9 antibody by conducting the ADAM9 knockdown study where we protein in C4-2 cells after exposing them to actinomycin D, an showed that on ADAM9 shRNA transfection, both ADAM9 and inhibitor of RNA synthesis, or to cycloheximide, an inhibitor of ADAM9-like proteins were decreased as analyzed by Western blot protein synthesis. Figure 5 shows that induction of ADAM9 protein analysis (see Fig. 2B). In addition, we routinely did isotype-specific expression in C4-2 cells by H2O2 has a rapid onset and this immunoglobulin G control staining of comparable tissue speci- induction can be blocked within 1 to 2 hours by the addition of mens to ensure that there is no nonspecific staining of antigenic either actinomycin D or cycloheximide to the cell culture medium epitope in clinical prostate cancer specimens (29–31). In our study, (Fig. 5). We showed that this short-term (1-2 hours) exposure to we observed ADAM9 shRNA treatment of C4-2 cells induced actinomycin D and cycloheximide did not alter the basal ADAM9 apoptotic cell death of prostate cancer cells in culture as assessed protein accumulation in prostate cancer cells. by Annexin V FACS analysis (Fig. 2C). Induction of ADAM9 expression under stress is mediated through ROS. When cells are subjected to crowding and hypoxia Discussion stress conditions, they produce high levels of ROS, primarily H2O2 Using human prostate cancer androgen-dependent LNCaP cells (32). Figure 3A (left) shows that when C4-2 cells were exposed to and their lineage-derived androgen-independent C4-2 and C4-2B exogenously added H2O2 at 0.01 to 10 mmol/L, a concentration- sublines as models for cDNA microarray analysis, we detected an dependent elevation of ADAM9 protein was observed. Addition of elevation of ADAM9 mRNA expression on androgen-independent 10 mmol/L H2O2 resulted in rapid induction of ADAM9 protein progression. These results were confirmed by quantitative RT-PCR accumulation in C4-2 cells at 4 hours (Fig. 3A, right). The H2O2 and Western blot for ADAM9 mRNA and protein, respectively induction of ADAM9 protein expression by H2O2 was more marked (Fig. 1). On further evaluation of ADAM9 protein expression in when cells were exposed to relatively higher concentrations of cultured human prostate cancer cell lines, we observed that the H2O2 (>5 mmol/L). The doublet bands, detected by exposing the steady-state level of this protein was determined by certain X-ray films for shorter time, observed on H2O2 addition were pathophysiologic stress conditions, such as cell crowding, hypoxia, ADAM9 and ADAM9-like protein, and both proteins were shown to and the addition of H2O2 to cell culture medium (34). We also be elevated (Fig. 3B). The role of H2O2 in the induction of ADAM9 confirmed the earlier study by McCulloch et al. (9) that androgen protein expression is further supported by experiments where stimulated ADAM9 expression and such stimulation can be blocked exogenously added H2O2 failed to induce ADAM9 protein by the simultaneous administration of an antiandrogen (data not expression in C4-2 cells that were transfected with catalase (an shown). Both the native ADAM9 and its smaller ADAM9-like protein enzyme that converts H2O2 to water; ref. 33; Fig. 3C, C4-2/catalase). were elevated by placing prostate cancer cells under stress con- In addition, recombinant catalase added to C4-2 cells exposed to ditions or exposing them to H2O2 (Figs. 1C and 3B), and the induc- H2O2 also abolished ADAM9 induction (Fig. 3C). tion of both proteins was blocked by catalase (Fig. 3C). Recently, To evaluate the potential pathophysiologic role of H2O2 in the Fischer et al. (4) showed that stress can induce ROS production, expression or accumulation of ADAM9 protein in prostate cancer which can cause the release of heparin-binding EGF through cells and tissues, we measured ROS and ADAM9 protein in several enzymatic shedding by ADAM9, ADAM10, and ADAM17. H2O2 human prostate cancer cell lines. Figure 3D shows a linear increased ADAM9 but failed in ADAM10, ADAM15, and ADAM17 correlation between ROS and the respective steady-state levels of (data not shown) in LNCaP, suggesting a special role of ADAM9 in ADAM9 protein in cells. These results are in agreement with a prostate cancer progression through increased growth factor recently published report by Lim et al. (6) who showed that a key signaling in response to stress conditions. The roles of H2O2 in ROS-producing enzyme, Nox1, is predominately expressed in ADAM9 induction may be cell context dependent because ebselen malignant rather than benign human prostate tissues. abolished crowding-induced ADAM9 protein accumulation in PC-3 Increased ADAM9 in prostate cancer cells under crowding and DU-145, but not C4-2, cells (Fig. 4). It was noted, however, that conditions is associated with ROS in a cell context-dependent destruction of endogenous H2O2 by catalase transfection also manner. Because cell crowding induced ADAM9 protein accumu- eliminated ADAM9 protein accumulation in C4-2 cells (Fig. 3C). lation in prostate cancer cells (Fig. 1C), we assessed the potential of The increased basal and cell crowding–induced ADAM9 protein H2O2 as a mediator. C4-2, DU-145, and PC3 cells were grown to expression in metastatic androgen-independent human prostate 100% confluence, then added with ebselen, an antioxidant, and cancer cell lines was confirmed in tissue array using human prostate ADAM9 protein accumulation in cells was determined. ADAM9 cancer specimens. For example, higher and more intense staining of protein induction in C4-2 cells by H2O2 can be effectively ADAM9/ADAM9-like proteins was associated with malignant antagonized by the antioxidant ebselen in a concentration- compared with benign prostate tissues (Fig. 2A). Recent studies dependent manner with effective blockade shown at 0.2 Amol/L showed that the secreted form of ADAM9-S promotes cancer cell ebselen (Fig. 4A). As shown in Fig. 4B, the addition of ebselen invasion (24). Further, increased ROS may be associated with decreased ADAM9 protein expression in 100% confluent cells of increased stress responses that result in increased ADAM9 protein PC-3 from a 2.0-fold to a 1.0-fold induction and DU-145 from a expression in prostate cancer cells. In our study, it seems that 2.6-fold to a 1.95-fold induction as compared with a 75% confluent ADAM9 protein may be responsible for cell survival on prostate culture (see Fig. 1C). Similar effects were not detected in C4-2 cells cancer progression in which we noted that decreased ADAM9 despite the evidence suggesting that H2O2 is a potent inducer for protein promoted apoptotic death of prostate cancer cells (Fig. 2C). ADAM9 protein expression in these cells. This suggestion is also supported by the observations that (a) www.aacrjournals.org 9523 Cancer Res 2006; 66: (19). October 1, 2006

Figure 3. The induction of ADAM9 expression by H2O2. A, Western blot. Left, ADAM9 expression in C4-2 cells induced by H2O2 in a dose-dependent manner. Right, induction of ADAM9 is rapid and effective, occurring within 4 hours of H2O2 exposure at 10 mmol/L. B, increased expression of ADAM9-like protein (arrow, with shorter time exposure) was observed on induction by either stress or addition of H2O2. Note that this protein was not expressed in LNCaP, C4-2, and C4-2B cells at 0 and 0.2 mmol/L H2O2 treatments; marked induction of ADAM9-like protein was observed when elevated H2O2 was added (5 or 10 mmol/L H2O2). C, the role of H2O2 in ADAM9 protein induction was further confirmed by abrogating H2O2 in C4-2 cells stably transfected with catalase (C4-2/catalase) and by the addition of catalase in culture medium (catalase+). D, a positive correlation between increased endogenous ROS levels, as measured by dichlorofluorescein diacetate assay, and increased ADAM9 protein expression, as assessed by Western blot, was observed.

increased ROS in prostate cancer cells (Fig. 3D) and tissues (6) tasis, there is evidence supporting this hypothesis. For example, C4-2 correlated with increased malignant potential of the prostate cancer and C4-2B cells expressing higher levels of H2O2 and ADAM9 also cells and tissues; and (b) the antioxidant ebselen (Fig. 4A) and seemed to be more tumorigenic and metastatic than the lineage- H2O2-degrading enzyme catalase (Fig. 3C) effectively blocked the related parental LNCaP cells (28). Shintani et al. (11) showed a close ability of prostate cancer cells to respond to stress- or H2O2-induced correlation between ADAM9 mRNA expression and brain metastasis increases in ADAM9 protein expression by prostate cancer cells. from lung cancer, and recently, Petudo et al. (14) showed that À À Although we have not established a direct correlation between stress ADAM9 / W10 mice developed well-differentiated prostates response–induced ADAM9 expression and prostate cancer metas- whereas littermate controls (ADAM9+/+ W10) had predominantly

Figure 4. Differential blockade of ADAM9 expression by ebselen in human prostate cancer cells subjected to cell crowding conditions. A, antioxidant ebselen blocked H2O2-induced ADAM9 and ADAM9-like protein expression in a concentration-dependent manner in C4-2 cells. C4-2 cells were cultured in the presence of 10 mmol/L H2O2 for 2 hours and in the presence of different concentrations of ebselen (0.02-20 Amol/L) for an additional 2 hours. B, to assess the possible relation between induction of ADAM9 and ADAM9-like protein expression in crowding conditions, C4-2, DU-145, or PC3 cells were grown to 100% confluence. Ebselen (20 Amol/L) effectively blocked cell crowding–induced ADAM9 and ADAM9-like protein expression in PC3 and DU-145, but not C4-2, cells.

Cancer Res 2006; 66: (19). October 1, 2006 9524 www.aacrjournals.org

Figure 5. Effects of actinomycin D and cycloheximide on ADAM9 and ADAM9-like protein expression up-regulated by hydrogen peroxide. To determine if stress-induced ADAM9 expression involved active gene transcription and translation, we assessed ADAM9 and ADAM9-like protein expression in C4-2 cells subjected to treatment with a classic RNA synthesis inhibitor, actinomycin D (Act D), or a protein synthesis inhibitor, cycloheximide (CHX). C4-2 cultured to 90% confluence was exposed to hydrogen peroxide (10 mmol/L) for 2 hours and then exposed to actinomycin D (1 Ag/mL), cycloheximide (10 Ag/mL), or vehicle for 0, 1, or 2 hours. Induction of ADAM9 protein expression in C4-2 cells by H2O2 has a rapid onset and this induction can be blocked within 1 to 2 hours by the addition of either actinomycin D or cycloheximide to the cell culture medium. poorly differentiated and significantly larger tumors. They also found to enhance cell signaling through specific growth factors, showed that overexpression of ADAM9 in mouse prostate epithe- extracellular matrices, and matrix metalloproteinases that facilitate lium resulted in significant abnormalities, including epithelial tumor growth and invasion (20, 37, 38). Likewise, cells subjected to hyperplasia at 4 to 6 months of age, and prostatic intraepithelial oxidative stresses such as cell crowding and hypoxic conditions neoplasia after 1 year. A report by Mozzocca et al. (24) indicated that were also found to have an altered repertoire of growth stimulatory the soluble form of ADAM9 could help tumor migration by cleavage and survival factors (39, 40). It is likely that ADAM9 up-regulation of the laminin in basement membrane. Taken together, our results is not only associated with androgen-independent progression but suggest that ADAM9 is a stress-inducible protein involved in is also required for cells to survive and to overcome harsh oxidative prostate cancer progression. ADAM9 may play a pivotal role stress conditions (41–46). ROS, represented by a host of oxygen supporting the growth, survival, and metastasis of human prostate radicals such as superoxide anion, hydrogen peroxide, hydroxyl cancer cells under pathophysiologic stress conditions (11, 14, 24). By radical, and peroxynitrate, can cause macromolecular damage and inactivating ADAM9 in human prostate cancer cell lines, we may play important roles in tumor development and aging (47–49). obtained preliminary evidence that these cells become highly The possibility exists, though unproven, that ADAM9 protein sensitive to radiation and chemotherapy (35). expression may be intimately linked with the critical balance Whereas our study suggests that H2O2 could play a key role in between the formation and destruction of ROS. Several human cell directly regulating ADAM9 protein expression in prostate cancer lines derived from cancers have significantly elevated hydrogen cells, its pathophysiologic role warrants further investigation. The peroxide (33). Lim et al. (6) showed that the increased capability of reasons for this conclusion are as follows: (a) In addition to H2O2, invasive androgen-independent human prostate cancer cell lines several other ROS, including nitric oxide, reactive nitrogen species, for ROS production and the higher level of ROS-producing enzyme, and lipid peroxides, have been shown to exert regulatory roles Nox1, were associated with malignant but not benign clinical controlling cell proliferation, survival, and differentiation (36). (b)A prostate tumors. Higher levels of ROS could be generated from cells more pronounced induction of ADAM9 protein expression was that have mitochondrial DNA mutations during tumor progression, observed at higher concentrations of H2O2, which could cause tumor development, and aging (50, 51). apoptosis albeit ADAM9 protein induction could be effectively In summary, intracellular ROS, generated by cell stress, controls prevented by exposing the treated cells to ebselen. The possible ADAM9 expression during androgen-independent and malignant local ROS gradients and the heterogeneity of ADAM9 protein progression of human prostate cancer cells and tissues. Up- induction, some with live cells and others with apoptotic cells, regulation of ADAM9 could facilitate cancer development (14) and warrant further investigation. (c) ADAM9 protein expression could promote cancer cell invasion (24), migration (11), and resistance to be the consequence of stress conditions induced by cell cycle stress-induced injuries (4). Ebselen, a lipid-soluble selenoorganic arrest. We have addressed this concern by using low concen- antioxidant, was shown to block ADAM9 through inhibition of ROS trations of H2O2 that do not cause cell cycle arrest and yet still production. A further understanding of the downstream pathways stimulate ADAM9 protein accumulation in prostate cancer cells. linking ROS production and the oxidative-induced stress response Additionally, the use of metabolic poisons, such as actinomycin D could potentially be the basis for a new clinical therapeutic develop- and cycloheximide, inhibited cell cycle progression but did not ment for advanced and androgen-refractory prostate cancer. induce ADAM9 protein expression (Fig. 5). Our results raise an interesting possibility about the develop- Acknowledgments ment of androgen-independent progression and the possible role Received 12/8/2005; revised 6/19/2006; accepted 7/25/2006. of ROS as mediators for stress and androgen-independent pro- Grant support: NIH grants CA76620, CA82739, and CA098912 and the Georgia state cancer progression in prostate cancer cells. Evidence indi- Cancer Coalition. The costs of publication of this article were defrayed in part by the payment of page cates the critical role of ROS in the regulation of cell growth charges. This article must therefore be hereby marked advertisement in accordance and differentiation. Cells engineered to overexpress ROS were with 18 U.S.C. Section 1734 solely to indicate this fact. www.aacrjournals.org 9525 Cancer Res 2006; 66: (19). October 1, 2006

References 17. Zhou M, Graham R, Russell G, Croucher PI. MDC-9 amounts of hydrogen peroxide by human tumor cells. (ADAM-9/meltrin g) functions as an adhesion molecule Cancer Res 1991;51:794–8. 1. Liotta LA, Kohn EC. The microenvironment of the by binding the a(v)h(5) integrin. Biochem Biophys Res 34. Murray JI, Whitfield ML, Trinklein ND, Myers RM, tumour-host interface. Nature 2001;411:375–9. Commun 2001;280:574–80. Brown PO, Botstein D. Diverse and specific gene 2. Nelson WG, De Marzo AM, DeWeese TL, Isaacs WB. 18. Howard L, Nelson KK, Maciewicz RA, Blobel CP. expression responses to stresses in cultured human The role of inflammation in the pathogenesis of prostate Interaction of the metalloprotease disintegrins MDC9 cells. Mol Biol Cell 2004;15:2361–74. cancer. J Urol 2004;172:S6–11; discussion S-2. and MDC15 with two SH3 domain-containing proteins, 35. Anderson CS, Sung SY, Kubo H, Hsieh CL, Johnstone 3. Pathak SK, Sharma RA, Steward WP, Mellon JK, endophilin Iand SH3PX1. J Biol Chem 1999;274:31693–9. P, Chung LW. Radiation induction of ADAM9 expression Griffiths TR, Gescher AJ. Oxidative stress and cyclo- 19. Nelson KK, Schlondorff J, Blobel CP. Evidence for an promoted prostate cancer cell survival: possible mech- oxygenase activity in prostate carcinogenesis: targets interaction of the metalloprotease-disintegrin tumour anism and potential therapeutic application. J Urol 2005; for chemopreventive strategies. Eur J Cancer 2005;41: necrosis factor a convertase (TACE) with mitotic 173S:108 (#396). 61–70. arrest deficient 2 (MAD2), and of the metalloprotease- 36. Payne CM, Bernstein C, Bernstein H, Gerner EW, 4. Fischer OM, Hart S, Gschwind A, Prenzel N, Ullrich A. disintegrin MDC9 with a novel MAD2-related protein, Garewal H. Reactive nitrogen species in colon carcino- Oxidative and osmotic stress signaling in tumor cells is MAD2h. Biochem J 1999;343 Pt 3:673–80. genesis. Antioxid Redox Signal 1999;1:449–67. mediated by ADAM proteases and heparin-binding 20. Arnold RS, Shi J, Murad E, et al. Hydrogen peroxide 37. Brar SS, Corbin Z, Kennedy TP, et al. NOX5 NAD(P)H epidermal growth factor. Mol Cell Biol 2004;24:5172–83. mediates the cell growth and transformation caused by oxidase regulates growth and apoptosis in DU 145 5. Karan D, Lin FC, Bryan M, et al. Expression of ADAMs the mitogenic oxidase Nox1. Proc Natl Acad Sci U S A prostate cancer cells. Am J Physiol Cell Physiol 2003;285: (a disintegrin and metalloproteases) and TIMP-3 (tissue 2001;98:5550–5. C353–69. inhibitor of metalloproteinase-3) in human prostatic 21. Livak KJ, Schmittgen TD. Analysis of relative gene 38. Hsu TC, Young MR, Cmarik J, Colburn NH. Activator adenocarcinomas. Int J Oncol 2003;23:1365–71. expression data using real-time quantitative PCR and protein 1 (AP-1)- and nuclear factor nB(NF-nB)- 6. Lim SD, Sun C, Lambeth JD, et al. Increased Nox1 and the 2(ÀDDC(T)) method. Methods 2001;25:402–8. dependent transcriptional events in carcinogenesis. Free hydrogen peroxide in prostate cancer. Prostate 2005;62: 22. Asayesh A, Alanentalo T, Khoo NK, Ahlgren U. Radic Biol Med 2000;28:1338–48. 200–7. Developmental expression of metalloproteases ADAM 39. Zafarullah M, Li WQ, Sylvester J, Ahmad M. 7. Carl-McGrath S, Lendeckel U, Ebert M, Roessner A, 9, 10, and 17 becomes restricted to divergent pancreatic Molecular mechanisms of N-acetylcysteine actions. Cell Rocken C. The disintegrin-metalloproteinases ADAM9, compartments. Dev Dyn 2005;232:1105–14. Mol Life Sci 2003;60:6–20. ADAM12, and ADAM15 are up-regulated in gastric 23. Kakiuchi S, Daigo Y, Ishikawa N, et al. Prediction of 40. Calvo A, Xiao N, Kang J, et al. Alterations in gene cancer. Int J Oncol 2005;26:17–24. sensitivity of advanced non-small cell lung cancers to expression profiles during prostate cancer progression: 8. Grutzmann R, Luttges J, Sipos B, et al. ADAM9 gefitinib (Iressa, ZD1839). Hum Mol Genet 2004;13:3029–43. functional correlations to tumorigenicity and down- expression in pancreatic cancer is associated with 24. Mazzocca A, Coppari R, De Franco R, et al. A regulation of selenoprotein-P in mouse and human tumour type and is a prognostic factor in ductal secreted form of ADAM9 promotes carcinoma invasion tumors. Cancer Res 2002;62:5325–35. adenocarcinoma. Br J Cancer 2004;90:1053–8. through tumor-stromal interactions. Cancer Res 2005; 41. Halliwell B, Murcia MA, Chirico S, Aruoma OI. Free 9. McCulloch DR, Harvey M, Herington AC. The 65:4728–38. radicals and antioxidants in food and in vivo: what they expression of the ADAMs proteases in prostate cancer 25. Ubezio P, Civoli F. Flow cytometric detection of do and how they work. Crit Rev Food Sci Nutr 1995;35: cell lines and their regulation by dihydrotestosterone. hydrogen peroxide production induced by doxorubicin 7–20. Mol Cell Endocrinol 2000;167:11–21. in cancer cells. Free Radic Biol Med 1994;16:509–16. 42. Wynder EL. Cancer prevention: optimizing life-styles 10. O’Shea C, McKie N, Buggy Y, et al. Expression of 26. Carter WO, Narayanan PK, Robinson JP. Intracellular with special reference to nutritional carcinogenesis. ADAM-9 mRNA and protein in human breast cancer. Int hydrogen peroxide and superoxide anion detection in J Natl Cancer Inst Monogr 1992;12:87–91. J Cancer 2003;105:754–61. endothelial cells. J Leukoc Biol 1994;55:253–8. 43. Wasylyk C, Wasylyk B. Oncogenic conversion of Ets 11. Shintani Y, Higashiyama S, Ohta M, et al. Over- 27. Thalmann GN, Sikes RA, Wu TT, et al. LNCaP affects redox regulation in vivo and in vitro. Nucleic expression of ADAM9 in non-small cell lung cancer progression model of human prostate cancer: androgen Acids Res 1993;21:523–9. correlates with brain metastasis. Cancer Res 2004;64: independence and osseous metastasis. Prostate 2000;44: 44. Toledano MB, Leonard WJ. Modulation of trans- 4190–6. 91–103 Jul 1;44. cription factor NF-nB binding activity by oxidation- 12. Roghani M, Becherer JD, Moss ML, et al. Metal- 28. Thalmann GN, Anezinis PE, Chang SM, et al. reduction in vitro. Proc Natl Acad Sci U S A 1991;88: loprotease-disintegrin MDC9: intracellular maturation Androgen-independent cancer progression and bone 4328–32. and catalytic activity. J Biol Chem 1999;274:3531–40. metastasis in the LNCaP model of human prostate 45. Wei H. Activation of oncogenes and/or inactivation 13. Schwettmann L, Tschesche H. Cloning and expres- cancer. Cancer Res 1994;54:2577–81. of anti-oncogenes by reactive oxygen species. Med sion in Pichia pastoris of metalloprotease domain of 29. Gleave M, Hsieh JT, Gao CA, von Eschenbach AC, Hypotheses 1992;39:267–70. ADAM 9 catalytically active against fibronectin. Protein Chung LW. Acceleration of human prostate cancer 46. Li W, Nugent MA, Zhao Y, et al. Lysyl oxidase oxidizes Expr Purif 2001;21:65–70. growth in vivo by factors produced by prostate and basic fibroblast growth factor and inactivates its 14. Peduto L, Reuter VE, Shaffer DR, Scher HI, Blobel CP. bone fibroblasts. Cancer Res 1991;51:3753–61. mitogenic potential. J Cell Biochem 2003;88:152–64. Critical function for ADAM9 in mouse prostate cancer. 30. Pisters LL, Troncoso P, Zhau HE, Li W, von 47. Oberley TD, Oberley LW. Oxygen radicals and cnacer. Cancer Res 2005;65:9312–9. Eschenbach AC, Chung LW. c-met proto-oncogene In: Yu BP, editor. Free radicals in aging. Ann Arbor: CRC 15. Izumi Y, Hirata M, Hasuwa H, et al. A metal- expression in benign and malignant human prostate Press; 1993. p. 247–67. loprotease-disintegrin, MDC9/meltrin-g/ADAM9 and tissues. J Urol 1995;154:293–8. 48. Sohal RS, Weindruch R. Oxidative stress, caloric PKCy are involved in TPA-induced ectodomain shed- 31. Zhau HY, Chang SM, Chen BQ, et al. Androgen- restriction, and aging. Science 1996;273:59–63. ding of membrane-anchored heparin-binding EGF-like repressed phenotype in human prostate cancer. Proc 49. Cerutti PA. Prooxidant states and tumor promotion. growth factor. EMBO J 1998;17:7260–72. Natl Acad Sci U S A 1996;93:15152–7. Science 1985;227:375–81. 16. Nath D, Slocombe PM, Webster A, Stephens PE, 32. Oda D, Nguyen MP, Royack GA, Tong DC. H2O2 50. Bianchi NO, Bianchi MS, Richard SM. Mitochondrial Docherty AJ, Murphy G. Meltrin g (ADAM-9) mediates oxidative damage in cultured oral epithelial cells: the genome instability in human cancers. Mutat Res 2001; cellular adhesion through a(6)h(1)integrin, leading to a effect of short-term vitamin C exposure. Anticancer Res 488:9–23. marked induction of fibroblast cell motility. J Cell Sci 2001;21:2719–24. 51. Mandavilli BS, Santos JH, Van Houten B. Mitochon- 2000;113:2319–28. 33. Szatrowski TP, Nathan CF. Production of large drial DNA repair and aging. Mutat Res 2002;509:127–51.

Cancer Res 2006; 66: (19). October 1, 2006 9526 www.aacrjournals.org

Shian-Ying Sung, Hiroyuki Kubo, Katsumi Shigemura, et al.

Cancer Res 2006;66:9519-9526.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/66/19/9519

Cited articles This article cites 50 articles, 19 of which you can access for free at: http://cancerres.aacrjournals.org/content/66/19/9519.full#ref-list-1

Citing articles This article has been cited by 9 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/66/19/9519.full#related-urls

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

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

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/66/19/9519. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.