One Tumour Surpessive

THE JOURNAL OF BIOLOGICAL CHEMISTRY VOL. 280, NO. 41, pp. 34985–34996, October 14, 2005 © 2005 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in the U.S.A.

Tumor-suppressive Maspin Regulates Cell Response to Oxidative Stress by Direct Interaction with Glutathione S-Transferase* Received for publication, March 31, 2005, and in revised form, June 23, 2005 Published, JBC Papers in Press, July 26, 2005, DOI 10.1074/jbc.M503522200

Shuping Yin‡, Xiaohua Li‡, Yonghong Meng‡, Russell L. Finley, Jr.§, Wael Sakr‡, Heng Yang‡, Neelima Reddy‡, and Shijie Sheng‡¶1 From the ‡Department of Pathology, §Center of Molecular Medicine and Genetics, and ¶Protease Program of the Karmanos Cancer Institute, Wayne State University School of Medicine, Detroit, Michigan 48201 Maspin, a novel serine protease inhibitor, suppresses tumor progression in several cancer models, including an in vivo model for prostate cancer bone metastasis. However, the molecular mechanism of maspin remains illusive, primarily because its molecular targets are unknown. To this end, we used a full-length maspin cDNA bait to screen against both a primary prostate tumor cDNA prey library and a HeLa cDNA prey library by the yeast two-hybrid method. We found that heat shock protein 90, glutathione S-transferase (GST), and heat shock protein 70 interacted with maspin with the highest frequencies. We confirmed the maspin/GST interaction using purified proteins, human epithelial cell lines, and human prostate tissues. A maspin variant that has a point mutation of Arg340 to Ala (MasR340A) showed a significantly decreased affinity for GST. Although purified maspin had no effect on the activity of purified GST in vitro, intracellular interaction between endogenous maspin and GST correlated with an elevated total GST activity in both MDA-MB-435- and DU145-derived stably transfected cells. Consistently, tumor cells treated with purified wild type maspin, but not MasR340A, enhanced cellular GST activity. Maspin expression in cancer cell lines also correlated with decreased basal levels of reactive oxygen species (ROS). Furthermore, H2O2 treatment not only induced GST expression but also increased intracellular maspin/ GST interaction, which was inversely correlated with the level of ROS generation. Conversely, maspin knockdown by small interfering RNA increased the basal, as well as H2O2-induced, ROS generation. Furthermore, the maspin effect on ROS generation was completely abolished by a GST inhibitor, indicating an essential role of GST in maspin-mediated cellular response to oxidative stress. Consistently, oxidative stress-induced vascular endothelial growth factor A expression was significantly inhibited in maspin-expressing cells. Together, our data suggest a new mechanism by which maspin, through its direct interaction with GST, may inhibit oxidative stress-induced ROS generation and vascular endothelial growth factor A induction, thus preventing further adverse effects on tumor genetics and stromal reactivity.

Maspin, a novel serine protease inhibitor (serpin),2 remains a promising potential therapeutic agent for a tumor suppressive effect in sev-

* This work was supported by National Institutes of Health Grant CA84176, the Ruth Sager Memorial Fund, the Fund for Cancer Research (to S. S.), and Department of Defense DAMD17-03-1-0038 (to S. Y.). 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. 1 To whom correspondence should be addressed: Dept. of Pathology, Wayne State University School of Medicine, 540 East Canfield Ave., Detroit, MI 48201. Tel.: 313-9938197; Fax: 313-993-4112; E-mail: [email protected]. 2 The abbreviations used are: serpin, serine protease inhibitor; DCFH-DA, 2⬘,7⬘-dichlorofluorescein diacetate; ELISA, enzyme-linked immunosorbent assay; EA, ethacrynic

OCTOBER 14, 2005 • VOLUME 280 • NUMBER 41

eral cancer models (1– 4). More recently, we showed that maspin expression inhibited prostate carcinoma bone tumor growth, osteolysis, and angiogenesis in a clinically relevant in vivo model (5). Consistently, differential expression of maspin has been linked to the progression of many types of human cancers. For example, in breast, colon, thyroid, lung, oral squamous, and prostate cancer, maspin expression predicts a better prognosis (6). Considering the therapeutic potential of maspin, it is critical to understand its molecular mechanisms. Three functional features of maspin are particularly noteworthy. First, an inhibitory effect of maspin on cell surface-associated urokinase-type plasminogen activator/ urokinase-type plasminogen activator receptor complex is consistent with its effect on tumor cell motility and invasion (7, 8). In contrast to known inhibitory serpins that bind to active serine protease targets, we found that maspin specifically bound to pro-urokinase-type plasminogen activator and inhibited its proteolytic activation.3 Second, although maspin is expressed by epithelial cells, documented evidence demonstrates that it also exerts an inhibitory effect on tumor-induced stromal reactivities such as angiogenesis and bone remodeling in vivo (5). Third, intracellular, but not extracellular, maspin sensitizes breast and prostate cancer cells to drug-induced apoptosis (10 –13). In fact, maspin is among only a few pro-apoptotic serpins. These observations led to the question whether the multifaceted biological effects of maspin derive from one central tumor-suppressive mechanism. The key to this question is to identify the maspin-associated molecules. In particular, despite the long-standing observation that maspin is also a cytoplasmic protein, its intracellular interacting proteins have not been identified. An earlier yeast two-hybrid screening using a C-terminal-truncated maspin bait against a fibroblast cDNA library led to the identification of collagen I and IV as the candidate maspin targets (14). However, this result has not been reproduced in epithelial cells and is not supported by the crystallographic analyses of maspin structure (15). We conducted a yeast two-hybrid screening study using the fulllength maspin as the bait and prey cDNA libraries derived from a human primary prostate tumor and HeLa cells. We report here that maspin specifically interacted with three stress-related proteins, heat shock protein 90 (Hsp90), glutathione S-transferase (GST), and heat shock protein 70 (Hsp70). While further studies of the maspin interaction with other candidates are underway, we further characterized the acid; FACS, fluorescence-activated cell sorting; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; GSH, reduced glutathione; GST, glutathione S-transferase; PMA, phorbol-12-myristate-13-acetate; ROS, reactive oxygen species; VEGF, vascular endothelial growth factor; Hsp, heat shock protein; rMaspin, recombinant wild type maspin; McAb, monoclonal antibody; siRNA, small interfering RNA; BSA, bovine serum albumin; PBS, phosphate-buffered saline; RT, room temperature; IP, immunoprecipitation; RIPA, radioimmune precipitation assay; Ct, critical threshold. 3 H. Biliran, submitted for publication.

JOURNAL OF BIOLOGICAL CHEMISTRY

34985

Maspin and GST in Oxidative Stress maspin interaction with GST. Our cellular, molecular, and biochemical evidence suggests that maspin may exert a tumor-suppressive effect on tumor genetic/epigenetic instability and tumor-induced VEGFdependent angiogenesis by regulating the GST-based cellular defense against oxidative stress. This novel mechanism could not have been predicted had GST not been identified as a maspin interactor. Nonetheless, it not only helps explain the biological activities of maspin but also suggests exciting new avenues for maspin-based drug developments.

MATERIALS AND METHODS Cell Cultures and Reagents—Human prostatic carcinoma cell lines DU145 and PC3, breast carcinoma cell line MDA-MB-435, normal prostate cell line CRL2220N, and normal breast cell line MCF10A were obtained from American Type Culture Collection (Rockville, MD). Maspin stable transfectant clones (M3, M7, and M10) and the mock transfectant clone (Neo) derived from DU145 cells were generated and cultured as described previously (8). Maspin stable transfectant clones (Tn15 and Tn16) and the mock transfectant clone (M-Neo) derived from MDA-MB-435 cells were generated and cultured as described (16). Recombinant wild type maspin (rMaspin) was overexpressed in baculovirus-infected insect cells Sf9 and purified as previously described (17). A maspin variant that has a point mutation of Arg340 to Ala (rMasR340A) was expressed and purified as described.3 GST protein (a hybrid of isoforms ␣ and ␮) was expressed in Escherichia coli XL-1 Blue cells that were transformed by commercial pGEX-2T vector and purified as described (17). Anti-maspin monoclonal antibody (McAb) and polyclonal antibody were purchased from BD Biosciences. AntiGST monoclonal antibody was purchased from Zymed Laboratories (San Francisco, CA). Anti-GST ␮ polyclonal antibody and protease inhibitor mixture were purchased from Calbiochem. The GST Detection Module kit and horseradish peroxidase-linked anti-mouse and anti-rabbit antibodies were purchased from Amersham Biosciences. Oregon Green-conjugated goat anti-rabbit secondary antibody, Texas Red-conjugated goat anti-mouse secondary antibody, 2⬘,7⬘-dichlorodihydrofluorescein diacetate (H2DCFDA), and Prolong Antifade kit were purchased from Molecular Probes (Eugene, OR). The iScriptTM cDNA synthesis kit, siLentFect lipid kit, and reagents for protein concentration analysis and protein gel electrophoresis were obtained from Bio-Rad. Protein G plus Protein A (Protein G/A)-agarose beads were from Oncogene Research Products (Boston, MA). GST inhibitor ethacrynic acid (EA) was purchased from BioVision (Mountain View, CA). Four pooled SMART-selected siRNA duplexes were purchased from Dharmacon, Inc. (Lafayette, CO). Fluorescein isothiocyanate-labeled, nonspecific siRNA with scrambled sequence, RNeasy Mini kit, and Effectene transfection reagent kit were purchased from Qiagen (Cambridge, MA). Brilliant SYBR Green QPCR master mix and pCMV-Tag2-luciferase (expressing FLAG-luciferase) were purchased from Stratagene (La Jolla, CA). All other chemicals, unless otherwise specified, were obtained from Sigma in the highest suitable purities. Transient Transfection—pCMV-Tag2-maspin plasmid was constructed to express FLAG-maspin. The full-length maspin cDNA was inserted downstream in-frame to the FLAG sequence into the pCMVTag2 vector (Stratagene). DU145 cells grown on 6-well culture plates were transiently transfected with 0.4 ␮g of pCINeo-maspin (8), FLAGmaspin, and FLAG-luciferase plasmid DNA, using the Effectene transfection reagent kit (Qiagen). Transient transfection efficiency was monitored based on the score of red fluorescent cells resulting from a parallel control transfection experiment using the mRFP1 plasmid (a generous gift from Dr. Roger Y. Tsien, Howard Hughes Medical Institute Labo-

34986 JOURNAL OF BIOLOGICAL CHEMISTRY

ratories at the University of California, San Diego). Western blotting and GST activity assay were performed after 48 h of transfection. To knock down maspin expression, PC3 cells grown on 6-well culture plates were transiently transfected with 10 nM siRNA SMARTpool maspin (Dharmacon) using the siLentFect lipid kit. In parallel, cells were transfected with a fluorescein isothiocyanate-labeled, nonspecific siRNA with scrambled sequence. Transfection efficiency was monitored by the fluorescein isothiocyanate label under a Leica fluorescence microscope (Model DM IRM). For further molecular analyses, cells were harvested 40 h after the transfection. Yeast Two-hybrid Screening—The yeast expression vector pEG202 (18), which contains the yeast HIS3 gene and the coding sequences for the LexA DNA-binding domain, was used to express the maspin bait. The full-length maspin coding sequence was excised from the previously constructed plasmid pGEX-2T-maspin (17) and fused in-frame N-terminal to the LexA DNA-binding domain of pEG202. The correct orientation and in-frame fusion were confirmed by DNA sequencing. The bait plasmid containing the LexA-maspin fusion protein was introduced into yeast strain RFY206, which has the lacZ reporter plasmid pSH18 –34 (19). The expression of the fusion protein was confirmed by Western blot analysis using both anti-maspin and anti-LexA antibodies. The human primary prostate tumor cDNA library and HeLa cDNA library cloned into pJG4 –5 plasmid were obtained from OriGene Technology Inc. (Rockville, MD) and maintained in yeast strain RFY231, which contains LEU2 reporter gene (20). Yeast two-hybrid screening was performed as described (19, 20). To check whether maspin bait alone would activate the reporter gene LEU2, the bait strain was mated with the RFY231 strain containing the empty pJG4 –5 vector (18). The number of total colony-forming units and Leu⫹ colonies were counted. The ratio of Leu⫹ colonies against total colony-forming units was 2 ⫻ 10⫺7, demonstrating that the background was low and the maspin bait plasmid was applicable for yeast two-hybrid screening. Maspin bait strain was mated with the prey strain. With each prey library, 250 Leu⫹ colonies were selected from 5 ⫻ 107 diploid colony-forming units. Colonies were screened for their galactose-dependent Leu⫹ and LacZ⫹ phenotype. Prey plasmids were rescued from the galactose-dependent Leu⫹ and LacZ⫹ colonies by a yeast mini-prep method as described (21). PCR amplification was performed using primers BCO1 and BCO2 to amplify the insert (21). The PCR products were digested with restriction enzymes AluI and HaeIII to avoid sequencing identical clones (22). The resulting DNA was used to transform E. coli KC8 for amplification and purification. The purified prey plasmid DNAs were introduced back into yeast strain RFY231, and transformants were mated with RFY206/pSH18 –34 containing maspin bait or other randomly chosen baits (LexA fusion bait plasmids pEG202-DmcycJ (23), pEG202-DmRux (24), and pJG21–1 (18)). The prey clones that gave rise to a specific interaction with the maspin bait, but not with random baits, were sequenced. The sequences were analyzed using the NCBI BLAST search program. ELISA Assay—Triplicate of 0.5 ␮g of rMaspin, rMasR340A, or BSA in 100 ␮l of coating buffer (40 mM Na2CO3䡠H2O and 60 mM NaHCO3, pH 8.9) were added to a 96-well Nunc-ImmunoTM plate (MaxiSorpTM Surface; Rochester, NY), incubated at 4 °C for overnight, and blocked by 5% skim milk/PBS solution for 1 h at room temperature (RT). 100 ␮l of purified GST at 5 ␮g/ml was added to each well and incubated for 2 h at RT. The plate was then incubated sequentially with 100 ␮l/well of antiGST McAb (4,000-fold diluted in 5% milk/PBS solution) for 1 h at RT and with 100 ␮l/well of 5,000-fold diluted horseradish peroxidaselinked anti-mouse antibody for 1 h at RT. Between steps, the plate was washed three times with PBS containing 0.1% Tween 20. Triplicate

VOLUME 280 • NUMBER 41 • OCTOBER 14, 2005

Maspin and GST in Oxidative Stress reactions without purified GST were performed to calibrate the background. The bound horseradish peroxidase was detected by its reaction with o-phenylenediamine substrate as described by the manufacturer’s instructions (Sigma) and was quantified by photometric absorbance at 492 nm using a Benchmark microplate reader (Bio-Rad). Immunoprecipitation (IP)—Total lysate of MCF10A, CRL2220N, or PC3 cells was obtained by lysing the cells in RIPA buffer. Protein G/Aagarose beads, at a final concentration of 5% (v/v), were incubated with 400 ␮g of total lysate to block nonspecific immunoglobulins. The resulting mixture was centrifuged at 10,000 ⫻ g for 30 s, and the harvested supernatant was used for subsequent IP. Briefly, cell lysate supernatant was incubated with 5 ␮g of maspin McAb or 5 ␮g of normal mouse IgG with gentle agitation at 4 °C overnight. Protein G/A beads were then added to a final concentration of 2.5% (v/v). The mixture was further incubated for 1 h at RT with gentle agitation and centrifuged at 10,000 ⫻ g for 30 s. The harvested beads were washed four times with RIPA buffer, each with a decreasing concentration of Nonidet P-40 (1, 0.5, 0.25, and 0.1%). The beads were then washed with PBS and resuspended with SDS sample buffer, heat denatured, and subjected to Western blot analyses. To test maspin/GST interaction in fresh human tissues, histologically confirmed prostate carcinoma tissues, along with the adjacent normal tissues of the respective patients, were briefly minced on ice. Tissue granules were then transferred into precooled 1.5-ml Eppendorf tubes with 300 ␮l of RIPA buffer and homogenized 20 strokes with a motordriven mini-homogenizer in an ice bath. After homogenization, samples were centrifuged at 10,000 ⫻ g for 20 min at 4 °C. The supernatants were harvested and 800 ␮g of total protein were used for each IP. Antimaspin McAb and anti-GST ␮ polyclonal antibody were used for IP. GSH Affinity Pulldown Assay—Binding of purified maspin to GST was examined by the affinity pulldown assay. Briefly, 1 ␮g of purified GST (in 500 ␮l of PBS) was incubated with 50 ␮l of glutathione (GSH)Sepharose 4B beads (50% slurry) for 1 h. The beads were washed three times with PBS containing 0.5% Tween 20 and incubated with 1 ␮g of rMaspin in 500 ␮l of PBS at RT for 2 h in the presence or absence of 20 ␮g of BSA. After washing the beads six times with PBS containing 0.5% Tween 20, the bound proteins were eluted with reduced glutathione buffer, denatured in SDS sample buffer, and subjected to Western blot assay. In a negative control experiment, 1 ␮g of maspin was heat denatured by boiling for 10 min prior to incubation with the GST/GSH beads. Binding of endogenously expressed maspin to GST was examined in a similar manner except 200 ␮g of total extracts of cells or freshly harvested frozen human prostate tissues were used in the place of purified GST. Cell extracts were made by lysing the cells in RIPA buffer (20 mM Hepes, 100 mM NaCl, 0.1% SDS, 1% Nonidet P-40, 1% deoxycholate, 1 mM Na3VO4, 1 mM EGTA, 50 mM NaF, 10% glycerol, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, and 1⫻ protease inhibitor mixture). FLAG-Maspin/GST Interaction under H2O2 Treatment—To express FLAG-Maspin, cells were transiently transfected with pCMV-Tag2maspin plasmid DNA using the Effectene transfection reagent kit (Qiagen). Twenty-four hours after the transfection, cells were treated with either H2O2 at various concentrations for another 24 h or TRAIL (50 ng/ml) for 1 h as a non-oxidative stress control. Cells were washed with Tris-buffered saline and lysed by lysis buffer (50 mM Tris-HCL, pH 7.4, 150 mM NaCl, 1 mM EDTA, 1% TRITONTM X-100 buffer, 1⫻ protease inhibitor mixture). Total cell lysate (1 mg) from each sample was subjected to IP using FLAG M2 antibody immobilized to agarose beads as instructed by the manufacturer (Sigma). After the resin was collected and washed with Tris-buffered saline, the bound proteins were eluted with Tris-buffered saline buffer containing 3⫻ FLAG peptide (150 ␮g/ml) and subjected to Western blotting.

OCTOBER 14, 2005 • VOLUME 280 • NUMBER 41

FIGURE 1. Yeast two-hybrid mating assay. A representative phenotypical profile for the mating of the prey strain with bait strain expressing pJG21–1 (lane 1), pEG202-DmcycJ (lane 2), pEG202-DmRux (lane 3), or pEG202-maspin (lane 4). The mating medium contained either glucose or galactose. The specific interaction between maspin and prey was identified by galactose-dependent growth on the Leu⫺ plate and by galactose-dependent ␤-galactosidase activity as indicated by the blue colony on the 5-bromo-4chloro-3-indolyl-␤-D-galactopyranoside (X-Gal) plate.

GST Activity Assay—To test the effect of maspin on cellular GST activity, cells, with or without maspin overexpression, were lysed by three cycles of sonication on ice (10 s/cycle with a 1-min interval) and centrifuged at 10,000 ⫻ g for 20 min at 4 °C. To test the specificity of the maspin effect on GST activity, MDA-MB-435 cells were cultured in a 6-well plate (2 ⫻ 105 cells/well) and treated with 5 ␮g/ml of rMaspin, rMasR340A, or BSA for 24 h. The resulting supernatants were analyzed by the same GST activity assay. The total GST activity in the resulting supernatants was determined using a GST detection module kit (Amersham Biosciences). The changes in absorbance at 340 nm over a 5-min period were used to calculate GST activities (25). Quantification of Reactive Oxygen Species (ROS)—Fluorogenic substrate 2⬘,7⬘- dichlorodihydrofluorescein diacetate (DCFH-DA) was used to react with intracellular ROS for fluorescence-activated cell sorting (FACS) as described (26). Briefly, 1 ⫻ 107 maspin or mock transfectant cells were incubated with 20 ␮M DCFH-DA in a 5-ml suspension at 37 °C for 10 min. The cells were divided into 250-␮l aliquots and added into FACS sample tubes. Triplicate aliquots of cells that were either untreated or treated with H2O2 at the indicated final concentrations at 37 °C for 30 min, or PMA at 1 ␮g/ml at 37 °C for 30 min, were analyzed using a FACScalibur flow cytometer (BD Biosciences) with an excitation wavelength of 488 nm and an emission wavelength of 530 nm. The FACS measurements made at the moment of H2O2 addition were used as background. Data analysis was based on 10,000 detected events using the Cell Quest software. Alternatively, to detect ROS in cell lysates, cells pretreated with 20 ␮M H2DCFDA for 30 min at 37 °C were cultured either with or without the H2O2 (1 or 10 mM) insult for 30 min, washed with PBS two times, and lysed by RIPA buffer. The resulting total cell lysate was centrifuged at 10,000 ⫻ g for 20 min at 4 °C. Changes of fluorescence at 488/583 nm in the supernatants were quantified by a SPECTRAmax Gemini plate reader (Molecular Devices). To study the effect of GST inhibition on the cellular capacity of ROS generation, cells harvested from adherent monolayer cultures were suspended to a final density of 2 ⫻ 106 cells/ml in a mitochondria isolation buffer (180 mM KCl, 10 mM EGTA, 5 mM HEPES, and 0.5% fatty acidfree BSA, pH 7.2) containing 20 ␮M H2DCFDA. The cells were lysed by incubation with digitonin to the final concentration of 0.02% for 4 min at RT and subsequently diluted with the same volume of the mitochondria isolation buffer. The GST activity in the resulting total cell lysates was neutralized with 200 ␮M GST inhibitor EA (RT for 3 min). The mixtures thus obtained were treated with 1 mM H2O2 for 10 min prior to the fluorescence measurements at 488/538 nm using a SPECTRAmax Gemini plate reader. Cells processed in parallel treated with only EA or H2O2 were used as background controls. Confocal Immunofluorescent Staining—The intracellular colocalization of maspin with GST was examined using an immunofluorescent staining procedure as described previously (12). M7 cells cultured on chamber slides were either untreated or treated with 100 ␮M H2O2 for 24 h. Cells were washed two times with PBS, fixed in 3.8% formaldehyde

JOURNAL OF BIOLOGICAL CHEMISTRY

34987

Maspin and GST in Oxidative Stress TABLE ONE

Candidate maspin interactors identified by yeast two-hybrid screening Identification frequency Prostate tumor HeLa library library 5 3 1 2 0 0 0 0 0 0 1 1 1 1 1

2 2 2 0 1 1 1 1 1 1 0 0 0 0 0

Candidate maspin interactors Heat shock 90-kDa protein 1, ␣ Glutathione S-transferase ␮3 Heat shock 70-kDa protein 5 Chromosome 2 open reading frame 2 Glutathione S-transferase ␻1 Heat shock 90-kDa protein 1, ␤ Histone deacetylase 1 Transcription factor IID Peptidylprolyl isomerase F (Cyclophilin F) Caenorhabditis briggsae cosmid Malate dehydrogenase 1, NAD (soluble) Transcriptional repressor (GCF2) 5-aminoimidazole-4-carboxamide ribonucleotide formyltransferase/IMP cyclohydrolase (ATIC) Brain-abundant, membrane-attached signal protein 1 Golgi autoantigen, golgin subfamily a, 4

for 10 min, and permeabilized with 0.1% Triton X-100 for 5 min. Cells were washed three times with PBS, blocked with 2% BSA for 1 h, and subsequently incubated with maspin polyclonal antibody and GST McAb at 4 °C overnight. Cells were washed five times with PBS and incubated with Oregon Green-conjugated goat anti-rabbit antibody and Texas Red-conjugated goat anti-mouse antibody at RT for 1 h. Finally, the cells were mounted using Prolong Antifade solution. Confocal microscopy examination was performed using a Zeiss LSM 310 model (The Imaging Core Facility of the Karmanos Cancer Institute, Detroit, MI). Real-time PCR—Total RNA was extracted using RNeasy Mini kit (Qiagen) following the manufacturer’s instructions. The quality of the RNA preparations was verified by RNA-agarose gel electrophoresis showing intact 18 and 28 S RNA and by UV spectrophotometry showing an optimal A260:A280 ratio (1.8 –2.0). One microgram of each total RNA sample was reverse transcribed. For real-time PCR, Brilliant SYBR Green QPCR master mix (Stratagene) was used together with 1 ␮l of cDNA and 200 nM primers for maspin, GST ␮3, VEGF-A, or GAPDH. The primer sequences for VEGF-A are 5⬘-TGCCTTGCTGCTCTACCTCC-3⬘ and 5⬘-TCACCGCCTCGGCTTGTCAC-3⬘. The primer sequences for maspin are 5⬘-CTACTTTGTTGGCAAGTGGATGAA-3⬘ and 5⬘-ACTGGTTTGGTGTCTGTCTTGTTG-3⬘. The primer sequences for GST ␮3 are 5⬘-AGTTGTGTGCGGAAATCCAT-3⬘ and 5⬘-GTGCGAGTCGTCTATGGTTC-3⬘. The primer sequences for GAPDH are as described (8). The real-time PCR thermal profile is: 1 cycle of 10 min at 95 °C and then 40 cycles of 30 s at 95 °C, 1 min at 55 °C and 1 min at 72 °C followed by 1 cycle of 1 min at 95 °C, and finally 41 cycles of increasing the temperature one degree (°C) every 30 s starting at 55 °C. Critical threshold cycle numbers (Ct) were obtained using the built-in software of the Mx4000TM Multiplex Quantitative PCR system (Stratagene). Miscellaneous Experiments—Protein concentration was determined by using the Bio-Rad protein assay reagent. SDS-PAGE, Western blotting, DNA-agarose gel electrophoresis, and ethedium bromide staining were performed as described (27).

RESULTS Screening for Candidate Targets of Maspin by Yeast Two-hybrid— To identify the molecular targets of endogenously expressed full-length

34988 JOURNAL OF BIOLOGICAL CHEMISTRY

FIGURE 2. The biophysical interaction between maspin and GST. A, determination of rMaspin and rMasR340A binding affinity for GST by ELISA. The bound GST was quantified by absorbance of o-phenylenediamine substrate at 492 nm, which was further normalized by the binding of BSA in control experiments. The bars represent the S.E. B, affinity pulldown assay. a, Western blotting of rMaspin eluted from GST affinity beads. rMaspin (1 ␮g) was either directly loaded onto the GST beads (lane 2) or preincubated with 20 ␮g of BSA prior to the affinity pulldown procedure (lane 3). In parallel, 1 ␮g of heat-denatured rMaspin was loaded (lane 4). rMaspin standard was used as a reference for Western blotting (lane 1). b, Western blotting of maspin eluted from the GST affinity beads using Neo and M7 cell extracts. Western blotting of co-eluted GST was used for loading control.

maspin, we screened a human primary prostate tumor cDNA library and a HeLa cDNA library using yeast two-hybrid screening methods. A total of 1 ⫻ 108 diploid colony-forming units was screened, among which 500 Leu⫹ colonies were further screened for their galactose-dependent Leu⫹ and LacZ⫹ phenotype. cDNA plasmids from 45 positive colonies were purified and introduced back into yeast cells for specific

VOLUME 280 • NUMBER 41 • OCTOBER 14, 2005

Maspin and GST in Oxidative Stress

FIGURE 3. Intracellular interaction between maspin and GST. A, Western blotting of maspin and GST in either total tissue extracts or in the GSH affinity pulldown fractions. The loading of tissue extracts was monitored by Western blotting of GAPDH on the same membrane. N, normal tissue; T, tumor tissue. Each pair of tissue samples is numerically labeled (1, 2, and 3). The pulldown results with matched pairs of human prostate tissues are shown in panel a, whereas the results with DU145-transfected cell lines are shown in panel b. B, Western blotting of co-immunoprecipitated maspin and GST from a matched pair of normal (N4) and cancerous (T4) human prostate tissues (a) and from the indicated epithelial cell lines (b). IP was performed using either maspin antibody or preimmune IgG. Western blotting of IgG was used as a loading control.

interaction mating assay. Twenty-eight candidates specifically interacted with full-length maspin, but not with three randomly chosen baits. Fig. 1 shows the representative specific interaction between the maspin bait and a prey clone. DNA sequencing analysis revealed that these 28 cDNA tags encoded 15 different genes (TABLE ONE). Among these genes Hsp90, GST, and Hsp70 were recovered with the highest frequencies using either cDNA library. All other candidate maspin interactors were also pulled out with lower frequencies. Our earlier studies showed that maspin sensitized both breast and prostate cancer cells to drug-induced apoptosis (12, 13). Considering the documented involvement of heat shock proteins, especially Hsp90 and Hsp70, in apoptosis (28 –31), it is intriguing to speculate that maspin interaction with heat shock proteins may play a role in the regulation of cell sensitivity to apoptosis. This exciting prospect is currently under investigation in our laboratory. In the meantime, the high frequency of maspin/GST interaction came as a surprise because no serpin/GST interaction has been reported. Characterization of the Maspin/GST Interaction—Specific binding of maspin to GST was first confirmed by ELISA. As shown in Fig. 2A, as compared with BSA purified GST specifically bound to rMaspin that was precoated on an ELISA plate. The interaction of maspin and GST was not affected by GSH (data not shown). Interestingly, rMasR340A, a maspin variant that has a point mutation of Arg340 to Ala at the putative P1 site, showed a significantly reduced affinity for GST, suggesting an important role of the maspin reactive site loop sequence in this interaction. In an affinity pulldown assay, rMaspin was shown to directly bind to GST that had been immobilized to GSH beads (Fig. 2Ba, lane 2). The binding of maspin to GST was insensitive to the competition by BSA (lane 3) but was abolished by heat denaturation of maspin prior to the column procedure (lane 4). These data indicate that maspin binding to GST is specific. To test whether endogenously expressed maspin specifically interacts with GST, proteins extracted from maspin and mock-

OCTOBER 14, 2005 • VOLUME 280 • NUMBER 41

transfected DU145 cells were tested in parallel GST affinity pulldown assays. As shown in Fig. 2Bb, maspin was only detected in the elution fractions derived from the M7 cell lysate. To confirm the specific interaction between endogenously expressed maspin and GST, GSH affinity pulldown experiments were performed with three matched pairs of normal and cancerous human prostate tissues, freshly harvested from radical prostatectomies. As shown in Fig. 3Aa, maspin protein was detected in all the tissue samples. In noninvasive prostate carcinoma tissue (T1), maspin expression was significantly higher than in the corresponding normal tissues (N1). In contrast, the expression of maspin in invasive prostate cancer specimens was significantly reduced (T2 and T3 versus N2 and N3, respectively). GST was pulled down by GSH beads from all those tissue specimens. Maspin was also pulled down, mostly from the normal tissues, irrespective of its various differential expression patterns in each pair of tissues. In parallel, total cell lysates of DU145-derived stably transfected Neo (with an empty vector) and M7 (with full-length maspin cDNA) cells were tested by the GSH affinity pulldown assay. GST was pulled down from both cell lines. Not surprisingly, maspin was only pulled down from M7 cell lysates (Fig. 3Ab). The molecular interaction of endogenous maspin and GST was further tested in the complementary IP experiments using maspin McAb. As shown in Fig. 3Ba, GST was co-precipitated with maspin from another matched pair of normal (N4) and noninvasive tumor (T4) tissues of human prostate. Comparable amounts of maspin were detected in N4 and T4 (data not shown). Consistently, GST also co-precipitated with maspin from three maspin-expressing human epithelial cell lines, MCF10A, CRL2220N, and PC3 (Fig. 3Bb). In control experiments using purified preimmune IgG, neither maspin nor GST was precipitated. Consistent with this evidence, maspin and GST appeared to be colocalized in maspin-transfected DU145 cells by confocal dual immunostaining analysis under a permeabilizing condition (Fig. 6A).

JOURNAL OF BIOLOGICAL CHEMISTRY

34989

Maspin and GST in Oxidative Stress

FIGURE 4. The effect of maspin on GST activity. The total GST activity in the DU145-derived Neo and maspin-transfected cell lines (A) as well as the MDA-MB-435-derived Neo and maspin-transfected cell lines (B) were quantified. The transient transfection of DU145 cells with PCINeo-maspin, FLAG-maspin, and FLAG-luciferase plasmids was monitored by Western blotting of maspin and FLAG tag (Ca). Western blotting of GAPDH on the same membrane was used to monitor the loading. The GST activities of the transiently transfected cells were quantified by a chromogenic assay (Cb). GST activities in the lysates of parental MDA-MB-435 cells that had been treated with 5 ␮g/ml of BSA, rMaspin, or rMasR340A were quantified (D). Specific GST activities in the unit of ⌬A340/min/mg are presented as an average of five repeats. The bars represent the S.E. The p values were obtained by one-tailed matched pair Student’s t tests.

The Effect of Maspin/GST Interaction on GST Activity and ROS Generation—Maspin expression in both breast and prostate tumor cells led to increased total GST activity as measured by a chromogenic substrate (Fig. 4, A and B). To test whether the maspin-induced cellular GST activity was a nonspecific stress response due to the presence of any overexpressed protein, DU145 cells were transiently transfected in parallel with three different vectors encoding FLAG-luciferase, FLAGmaspin, and maspin, respectively. The transient transfection efficiency was estimated to be ⬃30% based on the score of red fluorescent cells that resulted from a control transfection experiment using the mRFP1

34990 JOURNAL OF BIOLOGICAL CHEMISTRY

plasmid. As shown in Fig. 4Ca, the transiently transfected cells produced the corresponding transgenes as expected at comparable levels. The expression of maspin or FLAG-maspin significantly enhanced cellular GST activity, whereas the effect of FLAG-luciferase did not reach statistical significance (Fig. 4Cb). Furthermore, MDA-MB-435 cells that were treated with rMaspin exhibited increased GST activity. rMasR340A, which has a significantly reduced affinity for GST in vitro (Fig. 2A), did not have any significant effect on the total cellular GST activity in parallel (Fig. 4D), indicating the effect of maspin on GST activity may be dependent on a direct

VOLUME 280 • NUMBER 41 • OCTOBER 14, 2005

Maspin and GST in Oxidative Stress

FIGURE 5. The effect of maspin on basal cellular ROS level. FACS profiles of fluorescently labeled ROS in DU145-derived transfectant clones (Aa) and MDA-MB-435-derived transfectant clones (Ab) were quantified by FACS with an excitation wavelength of 488 nm and an emission wavelength of 530 nm. B, down-regulation of maspin expression in PC3 cells by siRNA. a, Western blotting of maspin in control PC3 cells is shown as well as PC3 cells transiently transfected with maspin siRNA or scramble siRNA. The Western blotting of GAPDH on the same membrane was performed to monitor the loading. b, the real-time quantitative PCR profile of maspin is shown in control PC3 cells (Œ), PC3 transfected with maspin siRNA (▫), and PC3 cells transfected with scramble siRNA (f). c, the basal level of ROS, as labeled by fluorescent dichlorofluorescein, is shown in the controls and siRNAtransfected PC3 cells. Data in panels A, Ba, and Bb are representative results of three repeats of the experiments. Data in panel Bc represent an average of three repeats. The bars represent S.D. The p values were obtained from one-tailed matched pair Student’s t tests.

protein/protein interaction. However, maspin may not directly regulate the catalytic activity of GST, because rMaspin did not have any significant effect on the activity of purified GST, whether or not GSH was present (data not shown). We have previously shown that maspin does not act as an inhibitory serpin against any serine proteases tested in solution. In an attempt to test whether the molecular interaction between maspin and GST alters the biochemical presentation of maspin, thus rendering it proteolytically inhibitory, an in vitro chymo-

OCTOBER 14, 2005 • VOLUME 280 • NUMBER 41

trypsin activity assay was performed. However, the proteolytic activity of chymotrypsin was not affected in the presence of various combinations of rMaspin and GST (data not shown). GST, by sulfhydrolizing electrophilic metabolic intermediates or drugs, acts as a housekeeping regulator of ROS homeostasis (32–34). Because maspin confers a higher cellular GST activity, it is possible that maspin expression may also affect ROS generation. To test this possibility, ROS in M3 and M7 cells were labeled with DCFH-DA and quan-

JOURNAL OF BIOLOGICAL CHEMISTRY

34991

Maspin and GST in Oxidative Stress

FIGURE 6. Maspin/GST interaction in response to H2O2 treatment. A, confocal imaging of immunofluorescently stained GST (red) and maspin (green) in M7 cells that were either untreated (a– c) or treated with H2O2 (100 ␮M for 24 h) (d–f). Co-localization of maspin and GST are shown in the superimposed images in panels c (a⫹b) and f (d⫹e); ⫻620. B, Western blotting detection of GST associated with FLAG-maspin in transiently transfected DU145 cells that had undergone treatment with H2O2 at the indicated concentrations for 24 h or with TRAIL for 1 h. Western blotting of maspin was used to control the loading. C, the fold differences of GST ␮3 mRNA in FLAG-maspin-transfected DU145 cells that were treated with the indicated concentrations of H2O2. GST ␮3 mRNA was determined by real-time quantitative PCR. Each reaction was performed in triplicate and was internally normalized first by the detection of GAPDH mRNA. The fold differences were calculated as a ratio to the untreated control.

tified by FACS. As shown in Fig. 5Aa, the basal levels of ROS in M3 and M7 clones were significantly lower than that in the Neo control cells. Similar results were obtained when the ROS levels in MDA-MB-435derived stable transfectant Tn15 and Tn16 clones were compared with that of the respective mock control cells (Fig. 5Ab). In contrast, in maspin-expressing PC3 cells, maspin knockdown by a pooled siRNA significantly increased the basal ROS level as judged by a fluorometric assay (Fig. 5B). A parallel experiment with a scrambled siRNA had no effect. The Effect of Maspin/GST Interaction on Oxidative Stress-induced ROS Generation and VEGF-A Expression—A gain of cellular capacity to detoxify increased ROS generation in response to oxidative stress may protect cells from direct cell injury and, consequently, from further genetic and epigenetic instabilities. To test whether the maspin/GST interaction plays a role in the cellular response to oxidative stress, M7 cells were challenged with oxidative stressor H2O2, which is also commonly used as a ROS inducer. The resulting cells were analyzed for protein/protein interaction by dual immunofluorescence staining under a permeabilizing condition. Confocal imaging revealed increased maspin/GST interaction in H2O2-treated cells (Fig. 6, Ad–f ) as compared with the untreated cells (Aa– c). Independently, the H2O2-induced maspin/GST interaction was confirmed in DU145 cells that had been transiently transfected with a pCMV-Tag2-maspin construct. The transfected cells were treated with 100 ␮M H2O2 for 24 h before the cells were lysed and analyzed by IP using FLAG M2 antibody. As shown in Fig. 6B, the amount of GST co-precipitated with maspin increased dose dependently by H2O2 treatment. This maspin/GST co-precipitation was not increased by TRAIL treatment in a parallel experiment, suggesting that the maspin interaction with GST could be a cellular response specifically to oxidative stress. The detectable H2O2-dependent increase of maspin/GST interaction may result from increased protein/protein interaction affinity and/or additional increase of GST expression in the presence of maspin. While the former possibility is

34992 JOURNAL OF BIOLOGICAL CHEMISTRY

being investigated, we detected the mRNA of GST ␮3. As shown in Fig. 6C, under the same condition, DU145 cells transiently expressing FLAG-maspin exhibited a H2O2-dependent increase of GST ␮3 mRNA. The increased maspin/GST interaction in H2O2-treated cells correlated with attenuated ROS induction. When H2O2 was used as the oxidative stressor, ROS generation in the Neo control cells increased rapidly and dose responsively following the treatments (Fig. 7A). However, ROS generation in M3 cells that had a lower basal level was only moderately induced by the same treatments. To test whether the maspin/ GST interaction may also confer resistance to ROS generated by other intracellular mechanism, cells were also treated with PMA at 1 ␮g/ml. As shown in Fig. 7B, treatment with PMA for 30 min led to a 28% increase of ROS in the Neo control cells. In parallel, the same treatment induced only a 16% increase of ROS in M3 cells. In contrast, specific maspin knockdown by siRNA not only led to increased basal level of ROS in PC3 cells (Fig. 5Bc) but also resulted in increased ROS generation in cellular response to H2O2 (Fig. 7C). Because multiple candidate maspin-associated proteins were identified by the yeast-two-hybrid system (TABLE ONE), it is important to find out whether GST is the essential mediator of the maspin effect on ROS generation. To test this possibility, cell lysates were treated with EA, a synthetic GST inhibitor. As shown in Fig. 8A, EA used at 200 ␮M completely inhibited the cellular GST activity in DU145-derived stably transfected cells, despite their different basal levels of GST activities. Furthermore, when treated with a combination of H2O2 and EA, the maspin-transfected cells and the Neo control cells exhibited the same ROS generation capacities (Fig. 8B). When these cells were treated with EA alone, there was no significant change in the basal level of ROS (data not shown). Together, these data suggest that the maspin effect on ROS generation is predominantly mediated by GST. ROS is linked to hypoxia-induced VEGF-A expression and angiogenesis in tumor progression (34, 35). Maspin, on the other hand, has been shown to suppress tumor-induced angiogenesis in several in vivo mod-

VOLUME 280 • NUMBER 41 • OCTOBER 14, 2005

Maspin and GST in Oxidative Stress

FIGURE 7. Maspin/GST interaction and ROS generation under oxidative stress. Aa, FACS profiles of ROS levels in untreated or H2O2-treated Neo (red) and M3 (green) cells. Ab, the mean ROS fluorescence intensities of Neo (●) and M3 (E) in panel Aa were plotted against the H2O2 concentrations. Ba, FACS profiles of ROS levels in untreated and PMA-treated Neo (red) and M3 (green) cells. Ab, the mean ROS fluorescence intensities of Neo and M3 in panel Ba are summarized in panel Bb. Open columns, untreated; filled columns, PMA-treated. In both panels Ab and Bb, data represent an average of triplicate analyses (⫹ S.E.). C, ROS detection by dichlorofluorescein fluorescence in H2O2-treated PC3 cells that had been transiently transfected with a scramble siRNA (open columns) or the pooled maspin siRNA (filled columns). Data represent the average of three repeats, and the bars represent S.D. The p values were obtained from one-tailed matched pair Student’s t-tests.

els (2, 5, 36). To test whether the maspin effect on ROS generation correlates with less oxidative stress-induced VEGF-A expression, DU145-derived stably transfected cells that have been treated with H2O2 were further analyzed by real-time PCR to detect VEGF-A. Par-

OCTOBER 14, 2005 • VOLUME 280 • NUMBER 41

allel real-time PCR detection of GAPDH was performed to normalize the total RNA in each reaction. As summarized in TABLE TWO, the Ct of VEGF-A dose dependently decreased in H2O2-treated Neo cells, whereas the Ct of VEGF-A in M3 cells remained largely unchanged.

JOURNAL OF BIOLOGICAL CHEMISTRY

34993

Maspin and GST in Oxidative Stress DISCUSSION

FIGURE 8. The effect of GST inhibition on ROS generation. A, the specific GST activities in DU145-derived transfected cells in the presence (open column) and absence (filled columns) of 200 ␮M EA. B, H2O2 induction of ROS in DU145-derived transfected cells in the presence of 200 ␮M EA. In both panels A and B, data represent an average result of three repeats. The bars represent S.D. The p values were obtained from one-tailed matched pair Student’s t-tests. TABLE TWO

Summary of real-time PCR Ct numbers Cell line

H2O2

GAPDHa

VEGF-Aa

17.10 ⫾ 0.08 17.05 ⫾ 0.17 17.09 ⫾ 0.7 17.30 ⫾ 0.09 17.22 ⫾ 0.03 17.21 ⫾ 0.14

24.28 ⫾ 0.16 23.15 ⫾ 0.21 22.45 ⫾ 0.17 25.59 ⫾ 0.14 25.56 ⫾ 0.04 25.89 ⫾ 0.19

p valuesb

␮M

Neo

M3

a b

0 20 100 0 20 100

⬍0.001 ⬍0.001 ⬎0.1 0.049

Data represent an average of triplicate samples (⫹ S.E.). The p values were obtained from one-tailed matched pair Student’s t-tests.

FIGURE 9. Determination of VEGF-A expression by real-time PCR. Neo and M3 cells were treated with H2O2 at the indicated concentrations for 24 h. Total RNA was isolated followed by quantitative real-time PCR detection of VEGF-A and GAPDH. PCR products were visualized by gel electrophoresis (1% agarose).

These data indicate an induction of VEGF-A in H2O2-treated Neo cells, but not in H2O2-treated M3 cells. The magnitude of these differences was also visualized by agarose gel electrophoresis of the resulting amplified DNA (Fig. 9).

34994 JOURNAL OF BIOLOGICAL CHEMISTRY

We screened for candidate maspin-interacting molecules by the yeast two-hybrid method using both human prostate tumor and HeLa cDNA libraries. Considering the hereditary bias of the yeast two-hybrid method against those less abundant cDNA sequences in the prey libraries, it is premature to dismiss candidate genes that were identified with lower frequencies. However, the overwhelming high frequency identification of stress-related proteins such as Hsp90, GST, and Hsp70 suggests that intracellular maspin may be primarily involved in cellular response to stress stimuli. In light of our earlier evidence (12, 13) that maspin sensitizes both breast cancer and prostate cancer cells to druginduced apoptosis, it is possible that maspin may regulate the threshold of heat shock protein-mediated cell survival/apoptosis. Investigations are underway in our laboratory to address this possibility. In this study, we focused on the interaction of maspin with GST. Although not predicted, a role of maspin in regulating the GST-based redox activity, in retrospect, is consistent with the biological observations. GST and GSH peroxidases are major GSH-based reductases responsible for reducing cytotoxic electrophilic metabolic intermediates such as ROS (32–34). ROS generation is often up-regulated in rapidly growing tumors due to the exhaustion of local oxygen supply (hypoxia) and nutrients (35). Known consequences of ROS accumulation include (i) cell injury accompanied by damage to DNA, proteins, and lipids, and (ii) inactivation of prolyl hydroxylase that eventually leads to up-regulation of HIF-1␣-induced VEGF-A expression and angiogenesis (37, 38). A down-regulation of maspin expression correlates with more aggressive phenotypes and increased angiogenesis in cancers (39 – 42). Interestingly, data from this study (Fig. 3) and an earlier investigation revealed a biphasic profile of maspin expression during tumor progression, i.e. prior to its progressive down-regulation in invasive tumors, maspin is “transiently” up-regulated in cancer precursor cells or carcinoma in situ (39). The up-regulated maspin in early stages of tumor progression may facilitate GST-mediated detoxification activity, consequently reducing genetic/epigenetic instabilities in tumor cells and curbing further stromal reactivities such as VEGF-dependent angiogenesis. To this end, maspin expression in primary tumors correlates with a better prognosis for oral squamous cancer and prostate cancer. In a recent study, we also showed the maspin expression in prostate tumor cells led to redifferentiation (5). One of the advantages of the yeast two-hybrid system is that it is based on intracellular protein/protein interaction. We also observed the intracellular interaction of maspin with GST in established human epithelial cell lines and in human prostate tissues (Figs. 3 and 6). The specificity of the maspin/GST interaction was further confirmed by using purified rMaspin and GST (Fig. 2). It is noteworthy that although maspin IP pulled down comparable amounts of GST from both normal and cancerous tissues, the reciprocal GSH affinity pulldown recovered more maspin from the normal tissues than from the match-paired tumor tissues. It is likely that the maspin-associated GST became a smaller fraction in the total GST during tumor progression, even though the maspin affinity for GST remained largely the same. A recent x-ray crystallographic analysis of a cysteine-free maspin mutant confirmed the earlier prediction based on computer modeling that the maspin conformation has a general serpin framework with a distinctly rigid reactive site loop and its surrounding areas (15). This evidence suggests a critical role of maspin reactive site loop in its interactions with novel targets. Consistent with this assessment, rMasR340A, which has a point mutation in the reactive site loop sequence, lost the affinity for GST (Fig. 2A). Although other surface areas of maspin may be involved in protein/protein interactions, maspin is thought unlikely to interact with collagen, a possibility suggested based on a previous inde-

VOLUME 280 • NUMBER 41 • OCTOBER 14, 2005

Maspin and GST in Oxidative Stress pendent yeast two-hybrid screening with a C-terminal-truncated maspin bait (14). We showed that endogenously expressed maspin or maspin treatment enhanced cellular GST activity (Fig. 4). The effect of maspin on cellular GST activity appeared to be dependent on direct protein/protein interaction because cells treated with rMasR340A did not have elevated GST activity. Considering the evidence that maspin is epithelial specific and differentially regulated in development as well as in the progression of human cancers (6), maspin may not have a housekeeping function to directly execute stress-reducing reactions either as a chaperone or as a detoxifying enzyme. Rather, maspin may play a regulatory role. Interestingly, the enzymatic activity of purified GST was not affected by rMaspin, despite the apparent affinity. The fact that the GST activity was insensitive to maspin suggests that the folding of the GST catalytic domain was not compromised by this protein/protein interaction. In the meantime, these data raise the question whether maspin alone is sufficient to enhance GST activity. Although it is possible that the maspin/GST interaction may recruit additional associated molecules that regulate the activity of GST, a definitive conclusion depends on further clarifications on the following issues. First, there are five major isoforms of human GST: ␣, ␲, ␮, ␪, and ␻. Although two isoforms (␮ and ␻) were identified as candidate maspin-interacting proteins (TABLE ONE), it is not clear whether maspin interacts with different GST isoforms with different affinities. To this end, a thorough evaluation is needed of the specific contribution of each GST isoform in the total GST activity as well as in the detoxification of tumor-imposed stress stimuli. Second, in case maspin interacts with specific GST isoforms, the GST commercially available for protein tag may not be the appropriate model for detailed enzymatic analysis. Furthermore, it is not clear whether the GST activity assay using an artificial chromogenic substrate was technically sensitive enough to detect all isoform-specific GST activities (43). The maspin/GST interaction was detected in both normal and cancerous prostate tissues (Fig. 3). Oxidative stress dose dependently enhanced this interaction. The increased interaction between maspin and GST correlated with curtailed ROS generation and VEGF-A expression (Figs. 6 –9 and TABLE TWO). Conversely, maspin downregulation by siRNA in PC3 cells increased not only the basal level of ROS but also the H2O2-induced ROS generation (Figs. 5 and 7). Furthermore, the maspin effect on ROS generation was completely abolished in the presence of a GST inhibitor (Fig. 8). Although oxidative stress further induced GST expression in maspin-expressing cells, our data suggest that maspin may protect oxidation-induced cellular damages, at least in part, by a direct interaction with, and protection of, GST. Because oxidative stress is prevalent and causative in tumor progression, a sustained GST-dependent redox homeostasis in the presence of maspin would have a tumor-suppressive effect. The structural basis for the oxidative stress-induced maspin/GST interaction is not clear. However, considering the evidence that the 8 cysteine residues in maspin may form intracellular disulfide bridges (44) and these disulfide bridges may contribute to the poor solubility of maspin at high concentrations (15), the oxidative states of these cysteine sulfhydryl groups may dictate the redox-dependent maspin conformations. Under oxidatively stressed conditions, maspin conformation may undergo changes in favor of a higher affinity for GST. On the other hand, the evidence that the maspin/GST interaction was independent of GSH does not exclude the possibility that GST may even directly regulate the redox-based conformational changes of maspin at the cost of GSH. Theoretically, the detoxifying activity of GST may contribute to drug resistance. However, clinical trials using GSH homologue GST inhibi-

OCTOBER 14, 2005 • VOLUME 280 • NUMBER 41

tors in cancer treatments do not seem to have a significant benefit (45). In fact, maspin that enhances cellular GST activity has been shown to sensitize tumor cells to drug-induced apoptosis. In light of the possibility that maspin may also interact with heat shock proteins, is possible that the apoptosis-sensitizing effect of maspin is independent of its interaction with GST. However, as we are currently investigating this possibility, we also wish to address whether the molecular interaction between maspin and other potential interactors is involved in the oxidative stress-induced maspin/GST interaction and whether GST further regulates the interaction of maspin with other candidate interactors. In this study we relied on breast and prostate epithelial cells and human prostate tissues to characterize the maspin/GST interaction. Both types of cells are relevant because a tumor-suppressive role of maspin has been shown using various breast and prostate cancer models. Additionally, because the three most frequently identified maspin interactors, Hsp90, GST, and Hsp70, were also pulled out from the cDNA library derived from the human cervical carcinoma HeLa cell line (TABLE ONE), our conclusions may be applicable to other types of cancers. A recent report showed that down-regulation of maspin in cervical cancer correlates with increased angiogenesis (9). Because the yeast two-hybrid method is not adequate to identify extracellular protein interactors, it is not surprising that no extracellular proteins were identified by this method as possible maspin interactors. However, it must be borne in mind that maspin is an intracellular, a cell surfaceassociated, and a secreted protein. Our earlier studies suggest that extracellular maspin may specifically interact with cell surface-associated urokinase-type plasminogen activator/urokinase-type plasminogen activator receptor complex and inhibit pericellular plasminogen activation (8). Further studies are needed to address whether the biochemical presentation of maspin that dictates its interaction with distinct molecular targets may be further regulated by its subcellular localization or trafficking. In summary, our data suggest an important role of maspin in cellular response to stress stimuli. In particular, our initial characterization of the maspin interaction with GST suggests that maspin may exert a tumor-suppressive effect on tumor genetic/epigenetic instability and tumor-induced VEGF-dependent angiogenesis by regulating the GSTbased cellular defense against oxidative stress. This novel mechanism, which could not have been predicted had GST not been identified as a maspin interactor, points to an exciting new direction for future mechanistic studies and maspin-based drug development. Acknowledgment—We thank Jaron Lockett for assistance in proofreading this manuscript.

REFERENCES 1. Zou, Z., Anisowicz, A., Hendrix, M. J., Thor, A., Neveu, M., Sheng, S., Rafidi, K., Seftor, E., and Sager, R. (1994) Science 263, 526 –529 2. Zhang, M., Shi, Y., Magit, D., Furth, P. A., and Sager, R. (2000) Oncogene 19, 6053– 6058 3. Shi, H. Y., Zhang, W., Liang, R., Abraham, S., Kittrell, F. S., Medina, D., and Zhang, M. (2001) Cancer Res. 61, 6945– 6951 4. Lefter, L. P., Sunamura, M., Furukawa, T., Takeda, K., Kotobuki, N., Oshimura, M., Matsuno, S., and Horii, A. (2003) Clin. Cancer Res. 9, 5044 –5052 5. Cher, M. L., Biliran, H. R., Jr., Bhagat, S., Meng, Y., Che, M., Lockett, J., Abrams, J., Fridman, R., Zachareas, M., and Sheng, S. (2003) Proc. Natl. Acad. Sci. U. S. A. 100, 7847–7852 6. Sheng, S. (2004) Front. Biosci. 9, 2733–2745 7. McGowen, R., Biliran, H., Jr., Sager, R., and Sheng, S. (2000) Cancer Res. 60, 4771– 4778 8. Biliran, H., Jr., and Sheng, S. (2001) Cancer Res. 61, 8676 – 8682 9. Toussaint-Smith, E., Donner, D. B., and Roman, A. (2004) Oncogene 23, 2988 –2995

JOURNAL OF BIOLOGICAL CHEMISTRY

34995

Maspin and GST in Oxidative Stress 10. Latha, K., Zhang, W., Cella, N., Shi, H. Y., and Zhang, M. (2005) Mol. Cell. Biol. 25, 1737–1748 11. Li, Z., Shi, H. Y., and Zhang, M. (2005) Oncogene 24, 2008 –2019 12. Liu, J., Yin, S., Reddy, N., Spencer, C., and Sheng, S. (2004) Cancer Res. 64, 1703–1711 13. Jiang, N., Meng, Y., Zhang, S., Mensah-Osman, E., and Sheng, S. (2002) Oncogene 21, 4089 – 4098 14. Blacque, O. E., and Worrall, D. M. (2002) J. Biol. Chem. 277, 10783–10788 15. Al-Ayyoubi, M., Gettins, P. G., and Volz, K. (2004) J. Biol. Chem. 279, 55540 –55544 16. Sheng, S., Carey, J., Seftor, E. A., Dias, L., Hendrix, M. J., and Sager, R. (1996) Proc. Natl. Acad. Sci. U. S. A. 93, 11669 –11674 17. Sheng, S., Pemberton, P. A., and Sager, R. (1994) J. Biol. Chem. 269, 30988 –30993 18. Gyuris, J., Golemis, E., Chertkov, H., and Brent, R. (1993) Cell 75, 791– 803 19. Finley, R. L., Jr., and Brent, R. (1994) Proc. Natl. Acad. Sci. U. S. A. 91, 12980 –12984 20. Kolonin, M. G., and Finley, R. L., Jr. (1998) Proc. Natl. Acad. Sci. U. S. A. 95, 14266 –14271 21. Finley, R. L., Jr., and Brent, R. (1995), in DNA Cloning Expresion Systems: A Practical Approach (Hames, B. D. G., and Glover, D. M., eds) pp. 169 –203, Oxford University Press, Oxford 22. Kolonin, M. G., Zhong, J., and Finley, R. L. (2000) Methods Enzymol. 328, 26 – 46 23. Kolonin, M. G., and Finley, R. L., Jr. (2000) Dev. Biol. 227, 661– 672 24. Thomas, B. J., Zavitz, K. H., Dong, X., Lane, M. E., Weigmann, K., Finley, R. L., Jr., Brent, R., Lehner, C. F., and Zipursky, S. L. (1997) Genes Dev. 11, 1289 –1298 25. Habig, W. H., Pabst, M. J., and Jakoby, W. B. (1974) J. Biol. Chem. 249, 7130 –7139 26. Robinson, J. P., Darzynkiewicz, Z., Dean, P. N., Hibbs, A. R., Orfao, A., Rabinovitch, P. S., and Wheeless, L. L. (1997) Current Protocols in Cytometry: Oxidative Metabolism, p. 9.7, John Wiley & Sons, Inc., New York 27. Ausubel, Q. M., Brent, B., Kingston, R. E., Moore, D. D., Seidman, J. G., and Struhl, K. (eds) (1998) Current Protocols in Molecular Biology, John Wiley & Sons, Inc., New York 28. Muller, P., Ceskova, P., and Vojtesek, B. (2005) J. Biol. Chem. 280, 6682– 6691 29. Vanden Berghe, T., Kalai, M., van Loo, G., Declercq, W., and Vandenabeele, P. (2003)

34996 JOURNAL OF BIOLOGICAL CHEMISTRY

J. Biol. Chem. 278, 5622–5629 30. Gaikwad, A., Poblenz, A., Haridas, V., Zhang, C., Duvic, M., and Gutterman, J. (2005) Clin. Cancer Res. 11, 1953–1962 31. Jones, E. L., Zhao, M. J., Stevenson, M. A., and Calderwood, S. K. (2004) Int. J. Hyperthermia 20, 835– 849 32. Perquin, M., Oster, T., Maul, A., Froment, N., Untereiner, M., and Bagrel, D. (2000) Cancer Lett. 158, 7–16 33. Maulik, N., and Das, D. K. (2002) Free Radic. Biol. Med. 33, 1047–1060 34. Maulik, N. (2002) Antioxid. Redox. Signal 4, 783–784 35. Ushio-Fukai, M., and Alexander, R. W. (2004) Mol. Cell Biochem. 264, 85–97 36. Shi, H. Y., Zhang, W., Liang, R., Kittrell, F., Templeton, N. S., Medina, D., and Zhang, M. (2003) Histol. Histopathol. 18, 201–206 37. Dachs, G. U., and Tozer, G. M. (2000) Eur. J. Cancer 36, 1649 –1660 38. Choi, K. S., Bae, M. K., Jeong, J. W., Moon, H. E., and Kim, K. W. (2003) J. Biochem. Mol. Biol. 36, 120 –127 39. Pierson, C. R., McGowen, R., Grignon, D., Sakr, W., Dey, J., and Sheng, S. (2002) Prostate 53, 255–262 40. Hojo, T., Akiyama, Y., Nagasaki, K., Maruyama, K., Kikuchi, K., Ikeda, T., Kitajima, M., and Yamaguchi, K. (2001) Cancer Lett. 171, 103–110 41. Song, S. Y., Lee, S. K., Kim, D. H., Son, H. J., Kim, H. J., Lim, Y. J., Lee, W. Y., Chun, H. K., and Rhee, J. C. (2002) Dig. Dis. Sci. 47, 1831–1835 42. Zou, Z., Zhang, W., Young, D., Gleave, M. G., Rennie, P., Connell, T., Connelly, R., Moul, J., Srivastava, S., and Sesterhenn, I. (2002) Clin. Cancer Res. 8, 1172–1177 43. Habdous, M., Vincent-Viry, M., Visvikis, S., and Siest, G. (2002) Clin. Chim. Acta 326, 131–142 44. Pemberton, P. A., Wong, D. T., Gibson, H. L., Kiefer, M. C., Fitzpatrick, P. A., Sager, R., and Barr, P. J. (1995) J. Biol. Chem. 270, 15832–15837 45. Seifried, H. E., McDonald, S. S., Anderson, D. E., Greenwald, P., and Milner, J. A. (2003) Cancer Res. 63, 4295– 4298

VOLUME 280 • NUMBER 41 • OCTOBER 14, 2005