Human Osteocalcin and Bone Sialoprotein Mediating Osteomimicry of Prostate Cancer Cells: Role of Camp-Dependent Protein Kinase a Signaling Pathway

Research Article

Human Osteocalcin and Bone Sialoprotein Mediating Osteomimicry of Prostate Cancer Cells: Role of cAMP-Dependent Protein Kinase A Signaling Pathway

Wen-Chin Huang,1 Zhihui Xie,1 Hiroyuki Konaka,1 Jaro Sodek,2 Haiyen E. Zhau,1 and Leland W.K. Chung1

1Molecular Urology and Therapeutics Program, Department of Urology and Winship Cancer Institute, Emory University School of Medicine, Atlanta, Georgia and 2CIHR Group in Matrix Dynamics and Department of Biochemistry, University of Toronto, Toronto, Ontario, Canada

Abstract bone and visceral organ metastasis (1). Despite the prevalence Osteocalcin and bone sialoprotein are the most abundant of prostate cancer metastasis to the skeleton, the molecular noncollagenous bone matrix proteins expressed by osteo- mechanisms of bone tropism are poorly understood. Previous blasts. Surprisingly, osteocalcin and bone sialoprotein are studies suggest that prostate cancer cell adhesion, extravasation, also expressed by malignant but not normal prostate migration, and interaction with bone cells are critical determinants epithelial cells. The purpose of this study is to investigate that govern prostate cancer bone colonization (2–4). Reports using how osteocalcin and bone sialoprotein expression is clinical prostate cancer metastasis specimens (5–7) and experi- regulated in prostate cancer cells. Our investigation mental cell and animal models (8–10) found the bone-specific revealed that (a) human osteocalcin and bone sialoprotein proteins osteocalcin and bone sialoprotein to be expressed in a promoter activities in an androgen-independent prostate heterogeneous manner by human prostate cancer specimens. We cancer cell line of LNCaP lineage, C4-2B, were markedly proposed that prostate cancer cells acquire osteomimetic or bone- enhanced 7- to 12-fold in a concentration-dependent like properties to improve their adhesion, proliferation, and manner by conditioned medium collected from prostate survival in bone (11). This communication delineates the molecular cancer and bone stromal cells. (b) Deletion analysis of mechanisms underlying the induction of human osteocalcin and human osteocalcin and bone sialoprotein promoter regions bone sialoprotein promoter activities and their endogenous gene identified cyclic AMP (cAMP)–responsive elements (CRE) as expression in the bone microenvironment by factors secreted from the critical determinants for conditioned medium–mediated prostate cancer and bone stromal cells. osteocalcin and bone sialoprotein gene expression in Osteocalcin (5-6 kDa) and bone sialoprotein (72-80 kDa) are prostate cancer cells. Consistent with these results, the synthesized and secreted by normal maturing osteoblasts. They are protein kinase A (PKA) pathway activators forskolin and major noncollagenous bone matrix proteins, with osteocalcin dibutyryl cAMP and the PKA pathway inhibitor H-89, comprising 1% to 2% of the total proteins in the skeleton (12). respectively, increased or repressed human osteocalcin and Osteocalcin binds with high affinity to hydroxyapatite crystals, bone sialoprotein promoter activities. (c) Electrophoretic the key mineral component of bone, and regulates bone crystal mobility shift assay showed that conditioned medium– growth (13). Osteocalcin can also act as a chemoattractant in the mediated stimulation of human osteocalcin and bone recruitment of osteoblasts and osteoclasts, contributing to the sialoprotein promoter activities occurs through increased dynamics of new bone formation and bone resorption (14). Ducy et al. interaction between CRE and CRE-binding protein. (d) (15) reported that osteocalcin-null mice exhibit increased bone Conditioned medium was found to induce human osteocal- formation without impaired bone resorption, suggesting a more cin and bone sialoprotein promoter activities via increased complex interaction between recruited osteoblasts and osteoclasts CRE/CRE-binding protein interaction in a cell background– and the participation of bone sialoprotein in this process. Bone dependent manner, with marked stimulation in selected sialoprotein is a highly sulfated, phosphorylated, and glycosylated prostate cancer but not bone stromal cells. Collectively, protein that mediates cell attachment through a RGD motif to these results suggest that osteocalcin and bone sialoprotein extracellular matrices (16). Due to its highly negatively charged expression is coordinated and regulated through cAMP- characteristics, bone sialoprotein can sequester calcium ions while dependent PKA signaling, which may define the molecular conserving polyglutamate regions, which have hydroxyapatite basis of the osteomimicry exhibited by prostate cancer cells. crystal nucleation potential (17). Through the RGD motif, bone (Cancer Res 2005; 65(6): 2303-13) sialoprotein mediates the attachment and activation of osteoclasts (18) and can facilitate attachment of normal bone or cancer cells to Introduction mineralized tissue surfaces (19, 20). Through its binding to factor H, bone sialoprotein can protect cells from complement-mediated The progression of prostate cancer to androgen independence cell lysis, which may be important for cancer cell survival (21). In and bone metastasis is generally lethal. Death often results from the absence of osteocalcin, bone sialoprotein could contribute to an overall metabolic shift toward new bone formation (22–24). The Requests for reprints: Leland W.K. Chung, Molecular Urology and Therapeutics published data suggest that osteocalcin and bone sialoprotein Program, Department of Urology and Winship Cancer Institute, Emory University could complement each other by regulating the homeostasis of School of Medicine, 1365-B Clifton Road, Room B5101, Atlanta, GA 30322. Phone: bone formation and bone resorption via controlled osteosclerotic 404-778-3672; Fax: 404-778-3675; E-mail: [email protected]. I2005 American Association for Cancer Research. and osteoclastogenic reactions. www.aacrjournals.org 2303 Cancer Res 2005; 65: (6). March 15, 2005

The coupling of G proteins with a cyclic AMP (cAMP)– vector has been described previously (36). The single deletion constructs dependent protein kinase A (PKA) signaling pathway is triggered DCRE1 (79 to 72) and DCRE2 (674 to 667) and a double deletion by a large number of ligand and receptor systems (25). Activation construct DCRE2/CRE1 in the human bone sialoprotein promoter/pGL3- of this signaling pathway is associated with the control of cell luciferase plasmid were generated by recombinant PCR technique. All plasmid constructs were confirmed by DNA sequencing. growth and differentiation (26, 27), ion channel conductivity (28), Transfection and Luciferase Activity Assay. Cells were trypsinized and and gene transcription (29–31). The cAMP-responsive element seeded at a density of 1.5 105 cells per well (LNCaP, C4-2B, DU145 and (CRE), a cis-element, interacts with a basic domain/leucine zipper ARCaP), and 1.0 105 cells per well (PC3 and MG63) in 12-well plates 24 motif contained within a transcription factor termed CRE-binding hours before transfection. Plasmid DNAs were introduced into cells by protein (CREB) through the cAMP-dependent PKA pathway (32). complexing with a commercial reagent N-[1-(2,3-dioleoyloxyl)propyl]- Studies in which both osteocalcin (30) and bone sialoprotein N,N,N-trimethylammonium methylsulfate (Roche Molecular Biochemicals) (33, 34) promoter activities were stimulated by the cAMP- according to the manufacturer’s protocol. Each transfection reaction dependent PKA pathway suggest the presence of the putative contained 1.25 Ag of tested DNA constructs and 0.25 Ag of the transfection h CRE site within these promoter regions that may interact with efficiency control cytomegalovirus promoter-driven -galactosidase plas- CREB to initiate osteocalcin and bone sialoprotein expression. mid DNA. After 6 hours of transfection, DNA-liposome mixtures were replaced by fresh T-medium or conditioned medium. Transfected cells were In this communication, we show that conditioned media harvested and lysed in 1 reporter lysis buffer (Promega) after 36 hours of collected from various human prostate cancer and bone stromal additional incubation. Cell lysates were vortexed for 15 seconds and spun cell lines induce human osteocalcin and bone sialoprotein for 10 minutes. For luciferase activity assay, 20 AL of the lysate supernatant promoter activities and increase their steady-state levels of were mixed with 100 AL of the luciferase substrate (Promega) and detected endogenous mRNA through a cAMP-dependent PKA pathway in by a luminometer (Monolight 3010 luminometer, PharMingen, San human prostate cancer cells that targets CRE cis-elements within Diego, CA). For h-galactosidase activity assay, 100 AL of the supernatant the human osteocalcin and bone sialoprotein promoter regions. were mixed with 100 ALof2 h-galactosidase substrate [200 mmol/L sodium phosphate buffer (pH 7.3), 2 mmol/L MgCl2, 100 mmol/L h- mercaptoethanol, 1.33 mg/mL ONPG] and incubated at 37jC for 30 Materials and Methods minutes. h-galactosidase activity was detected by Microplate spectropho- Reagents. Tissue culture medium and fetal bovine serum were obtained tometer (Molecular Devices Corp., Sunnyvale, CA) at 405 nm wavelength. from Life Technologies, Inc. (Rockville, MD) and Sigma (St. Louis, MO), Data were presented as normalized luciferase activity (means F SD) defined respectively. Reagents used for the study of cAMP-dependent signaling as luciferase activity normalized to internal control h-galactosidase activity pathway, forskolin, dibutyryl cAMP (db cAMP), phorbol 12-myristate 13-acetate, for transfection efficiency. All studies were done in three independent and N-[2-(p-bromocinnamylamino)ethyl]-5-isoquinolinesulfonamide (H-89), experiments with duplicate assays. were purchased from Alexis Biochemicals (San Diego, CA). Synthetic Reverse Transcription-PCR Analysis. Total RNA was isolated from oligonucleotides were ordered from Invitrogen (Carlsbad, CA). Restriction confluent monolayers of cells using RNAZol B (Teltest, Inc., Friendswood, enzymes, T4 DNA ligase and T4 polynucleotide kinase, were obtained from TX). Total RNA (5 Ag) was used as template and random hexanucleotide New England Biolabs (Beverly, MA). Radioactive nucleotides were purchased primers (0.15 Ag) were added for reverse transcription and amplification in from Amersham Biosciences Corp. (Piscataway, NJ). Taq DNA polymerase a reaction volume of 20 AL according to the manufacturer’s instruction was obtained from Roche Molecular Biochemicals (Indianapolis, IN). (Invitrogen). After reverse transcription reaction, first-strand cDNA (3-5 AL) QuikChange site-directed mutagenesis kit was obtained from Stratagene was used for PCR with a PTC-100 programmable thermal controller (MJ (La Jolla, CA). Research, Inc., Waltham, MA). The oligonucleotide primer sets used for Cells, Cell Culture, and Conditioned Medium Collection. Human PCR analysis of cDNA are human osteocalcin 5V-ACACTCCTCGCCCTATTG- prostate cancer cell lines LNCaP, C4-2B, DU145, PC3, and ARCaP and 3V (forward) and 5V-GATGTGGTCAGCCAACTC-3V (reverse), human bone- human bone stromal cell lines MG63 (a human osteosarcoma cell line) and sialoprotein 5V-GCATCGAAGAGTCAAAATAG-3V (forward) and 5V- KeesII (a human normal osteoblast cell line) were cultured in T-medium TTCTTCTCCATTGTCTTCTC-3V (reverse), and glyceraldehyde-3-phosphate (Life Technologies) supplemented with 5% fetal bovine serum and 1% dehydrogenase (GAPDH) 5V-ACCACAGTCCATGCCATCA-3V (forward) and penicillin/streptomycin as described previously (35). The cells were 5V-TCCACCACCCTGTTGCTGT-3V (reverse). The thermal profile for human maintained at 37jCin5%CO2. For conditioned medium collection, cells osteocalcin amplification is 30 cycles starting with denaturation of 1 minute were cultured in T-medium with serum until 80% confluent. The cells were at 94jC followed by 1 minute of annealing at 55jC and 1 minute for washed subsequently twice in PBS and incubated in T-medium without extension at 72jC. For human bone sialoprotein amplification, the thermal serum. After 2 days of additional incubation, conditioned media were profile is 35 cycles starting with denaturation for 1 minute at 94jC followed collected, centrifuged, and stored at 20jC until use. The concentration of by 30 seconds of annealing at 48jC and 40 seconds of extension at 72jC. total proteins in conditioned medium was determined by the Bradford The program for GAPDH amplification is 25 cycles starting with method using Coomassie plus protein reagent (Pierce, Rockford, IL). denaturation for 30 seconds at 94jC followed by 30 seconds of annealing Plasmid Construction. Genomic DNA was used in the PCR of the 0.8-kb at 60jC and 1 minute of extension at 72jC. The reverse transcription-PCR human osteocalcin promoter inserting KpnI and XhoI sites, subsequently products were analyzed by 1.2% agarose gel electrophoresis. Quantity one- cloned into pGL3-Basic-luciferase reporter vector (Promega, Madison, WI) 4.1.1 Gel Doc gel documentation software (Bio-Rad, Hercules, CA) or NIH as described previously (9). The deletion constructs, DA [374 bp upstream of image were used for quantification of human osteocalcin and bone AP1/VDRE (AV) element], DTst-1 (POU-factor Tst-1/Oct-6 binding site), sialoprotein mRNA expression normalized by GAPDH mRNA expression. DCRE, and DIRF-1 (IFN regulatory factor-1 binding site) in human osteo- Immunohistochemical Staining. Human primary and bone metastatic calcin promoter/pGL3 were generated by recombinant PCR technique. prostate cancer tissue specimens were deparaffinized, treated with 3% The point mutation plasmids in CRE (643 to 636 in human osteocalcin H2O2, blocked with SuperBlock (Scytek Laboratories, Logan, UT), and promoter, 5V-TGACCTCA-3V) were constructed with a QuikChange site- reacted with antibodies against osteocalcin (OC 4-30, PanVera Corp., directed mutagenesis kit. The mutation constructs (see below) are the one- Madison, WI) or bone sialoprotein (kindly provided by Dr. J. Sodek, point substitution mutants, Mut1 (642 G!T), Mut2 (641 A!C), Mut3 University of Toronto, Toronto, Ontario, Canada). The staining signals were (640 C!A), Mut4 (639 C!A), and Mut5 (638 T!G); a two-point amplified by biotinylated peroxidase-conjugated streptavidin system (Bio- substitution mutant, Mut6 (640 and 639 CC!AA); and a two-point Genex Laboratories, San Ramon, CA). Osteocalcin and bone sialoprotein deletion construct, Mut7 (640 and 639 CC!XX). The 1.5-kb human bone were visualized in selective areas of the clinical tissue specimens where the sialoprotein promoter construct prepared in the pGL3-luciferase reporter immunohistochemical staining was revealed by the conjugated peroxidase

Cancer Res 2005; 65: (6). March 15, 2005 2304 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2005 American Association for Cancer Research. Osteocalcin and Bone Sialoprotein Expression by PKA reacted with AEC or 3,3V-diaminobenzidine as the substrate. Positive genetic variations among the prostate cancer cells or be an osteocalcin and bone sialoprotein are defined as >15% of the cell epiphenomenon of prostate cancer cell interaction with the populations reacted positively with either anti-osteocalcin or anti–bone microenvironment. We sought to evaluate the regulation of sialoprotein antibody. osteocalcin and bone sialoprotein expression in human prostate Electrophoretic Mobility Shift Assay. A LNCaP-lineaged metastatic cancer and bone cells to define whether extrinsic factor(s) secreted human prostate cancer cell line, C4-2B, and the human osteosarcoma cell line MG63 were plated in 15 cm diameter tissue culture dishes in T-medium (with by prostate cancer or bone cells could mediate osteocalcin and 5% serum) until 80% confluent. Cells were then switched to 1-day complete bone sialoprotein promoter activities and their respective steady- serum-free condition and then treated with or without ARCaP conditioned state levels of endogenous mRNA expression. medium (15 Ag/mL) or forskolin (105 mol/L) for an additional 16 hours. To test the hypothesis that bone-specific protein expression by Nuclear extracts were prepared from C4-2B and MG63 for electrophoretic prostate cancer cells may be induced by factor(s) secreted from mobility shift assay as described by Ausubel et al. (37). Briefly, cells were prostate cancer and bone stromal cells, we compared the effects of washed with cold PBS twice, harvested, and homogenized in hypotonic buffer conditioned medium harvested from either prostate cancer or [10 mmol/L HEPES (pH 7.9), 1.5 mmol/L MgCl2, 10 mmol/L KCl, 0.5 mmol/L bone stromal cells on the expression of osteocalcin and bone DTT, 0.2 mmol/L phenylmethylsulfonyl fluoride]. After centrifugation, the sialoprotein in an androgen-independent C4-2B prostate cancer nuclear pellets were stirred and incubated for 30 minutes in high-salt buffer cell line of LNCaP lineage. A luciferase reporter construct with a [20 mmol/L HEPES (pH 7.9), 25% glycerol, 1.5 mmol/L MgCl2, 1.2 mol/L KCl, 0.2 mmol/L EDTA, 0.5 mmol/L DTT, 0.2 mmol/L phenylmethylsulfonyl 0.8-kb human osteocalcin promoter or a 1.5-kb human bone fluoride] on ice. The nuclear pellets were centrifuged and the nuclear extract sialoprotein promoter was transfected into C4-2B cells, and the supernatants were dialyzed twice against dialysis buffer [20 mmol/L HEPES transfected cells were exposed to serum-free conditioned medium (pH 7.9), 20% glycerol, 100 mmol/L KCl, 0.2 mmol/L EDTA, 0.5 mmol/L DTT, collected from various human prostate cancer cell lines with a 0.2 mmol/L phenylmethylsulfonyl fluoride] overnight. The concentration of gradient of malignant potentials [from LNCaP (the least malig- nuclear proteins was determined by the Bradford method using Coomassie nant), C4-2B, DU145, and PC3 (cells with intermediate levels of plus protein reagent. Synthetic oligonucleotides were purified by PAGE. aggressiveness) to ARCaP (the most malignant)], a normal j Appropriate pairs were annealed by heating up to 95 C for 10 minutes and nonmalignant osteoblast KeesII cell line or a malignant osteosar- naturally cooled down to room temperature. The oligo sequences used as coma MG63 cell line. As shown in Fig. 1B, conditioned medium probes or competitors were wild-type CRE 5V-ACCAACCGGCTGACCT- stimulated human osteocalcin promoter activity in a concentration- CATCTCCTGCC-3VandMut65V-ACCAACCGGCTGAAATCATCTCCTGCC-3V. A The double-stranded probes were end labeled with [g-32P]ATP (3,000 Ci/ dependent manner (0-15 g/mL of total proteins in conditioned mmol at 10 mCi/mL) using T4 polynucleotide kinase. Forty thousand medium). Conditioned medium from LNCaP cells maximally counts per minute of the labeled probe and 10 Ag of nuclear extracts were stimulated human osteocalcin promoter activity by only 1.2 F incubated with binding buffer containing 10 mmol/L Tris-HCl (pH 7.5), 0.1-fold, whereas conditioned medium collected from the most 50 mmol/L NaCl, 0.5 mmol/L EDTA, 0.5 mmol/L DTT, 4% glycerol, 1 mmol/ aggressive ARCaP prostate cancer cell line maximally enhanced the L KCl, and 1 Ag poly(deoxyinosinic-deoxycytidylic acid) (Amersham highest human osteocalcin promoter activity at 7.1 F 0.3-fold. Pharmacia Biotech, Piscataway, NJ) at 30jC for 30 minutes. The samples Conditioned medium from other cell lines, C4-2B, DU145, PC3, were subjected to 6% nondenaturing PAGE in 1 TGE buffer [25 mmol/L KeesII, and MG63, induced human osteocalcin promoter activity at Tris-HCl (pH 8.5), 188 mmol/L glycine, 1 mmol/L EDTA] at 35 mA and 200 intermediate levels (2.0 F 0.2- to 5.1 F 0.4-fold). These data suggest V for 2 hours at room temperature. In competition experiments, unlabeled that the extent of stimulation of human osteocalcin promoter competitor oligos were preincubated with nuclear extracts for 30 minutes at room temperature before the addition of the probe. For the supershift activity by conditioned medium correlated positively with the experiment, 2 Ag of anti-CREB antibody (Santa Cruz Biotechnology, Inc., aggressiveness of the prostate cancer. In parallel with the induction Santa Cruz, CA) or anti-AML3/Runx2 antibody (Active Motif, Inc., Carlsbad, of human osteocalcin promoter activity, ARCaP conditioned CA) were added to the nuclear extract reaction mixtures at room medium also up-regulated human bone sialoprotein promoter temperature for 30 minutes before the incubation of the probe. After activity as much as 12-fold in a concentration-dependent manner electrophoresis, gels were dried by Gel Dryer (model 583, Bio-Rad) and in C4-2B cells (Fig. 1C). exposed to BioMax film (Kodak, Rochester, NY). To determine further whether ARCaP conditioned medium is Statistical Analysis. All statistical analyses were done using Microsoft capable of stimulating human osteocalcin and bone sialoprotein Excel software. Significant differences were analyzed using Student’s t test promoter activities in a series of other human prostate cancer and and two-tailed distribution. bone stromal cell lines, we tested these promoter activities in LNCaP, DU145, PC3, ARCaP, and MG63 cells. As shown in Fig. 1D, Results both human osteocalcin and bone sialoprotein promoter activities Human Osteocalcin and Bone Sialoprotein Expression in were elevated by ARCaP conditioned medium in LNCaP and C4-2B Tissue Specimens, Promoter Regulation in Human Prostate but not in DU145, PC3, ARCaP, and MG63 cell lines. Cancer Cell Lines. Osteocalcin protein is prevalently expressed by We compared the effect of ARCaP conditioned medium in both primary (85% positively stained) and metastatic (both lymph inducing human osteocalcin and bone sialoprotein promoter- node and bone were 100% positively stained) human prostate reporter activities in LNCaP and C4-2B cells, and its effects on cancer specimens (8). Likewise, bone sialoprotein protein is also the steady-state levels of osteocalcin and bone sialoprotein expressed preferentially by malignant (89-100%) primary prostate mRNA, in several human prostate cancer cell lines by cancer tissues (7). One common feature of osteocalcin and bone semiquantitative reverse transcription-PCR. Figure 1E shows sialoprotein immunostaining in human prostate cancer tissues is that adding ARCaP conditioned medium (15 Ag/mL) to LNCaP the marked heterogeneities among prostate cancer cells in primary and C4-2B cells for a 12-hour period enhanced the steady-state and bone metastatic specimens. Some cells stain strongly for levels of endogenous osteocalcin and bone sialoprotein mRNA osteocalcin and bone sialoprotein proteins (Fig. 1A, bold arrows) expression by 4.8- and 5.9-fold and 4.5- and 7.8-fold (GAPDH as and others seem to be lightly stained or not stained at all (Fig. 1A, an internal control), respectively. In cells that already had high arrowheads). Differential staining could either reflect intrinsic basal levels of osteocalcin and bone sialoprotein mRNA, such as www.aacrjournals.org 2305 Cancer Res 2005; 65: (6). March 15, 2005

Figure 1. Immunohistochemical staining of osteocalcin (OC) and bone sialoprotein (BSP) in human primary and bone metastatic prostate cancer tissue specimens, and conditioned medium stimulates human osteocalcin and bone sialoprotein promoter activities and the steady-state levels of osteocalcin and bone sialoprotein mRNA expression in human prostate cancer cell lines. A, positive and strong osteocalcin and bone sialoprotein stains were detected in both primary and bone metastatic clinical prostate cancer specimens (bold arrows). Some areas of the prostate cancer cells were found only lightly or not stained at all (arrowheads). Osteomimicry was found to exist in prostate cancer cells when present in primary. Magnification, 75. B, conditioned media (CM) were collected from human prostate cancer cell lines (LNCaP, C4-2B, DU145, PC3, and ARCaP), a normal human osteoblastic cell line (KeesII), and a human osteosarcoma cell line (MG63). The human osteocalcin promoter-reporter construct was cotransfected with cytomegalovirus promoter-driven h-galactosidase plasmid (for the correction of transfection efficiency as an internal control) into an androgen-independent and metastatic LNCaP cell subline, C4-2B. Conditioned medium induced human osteocalcin promoter activity in a dose-dependent manner (total protein concentration 0-15 Ag/mL). C, ARCaP conditioned medium also stimulated human bone sialoprotein promoter activity in a dose-dependent manner (total protein concentration 0-15 Ag/mL). D, human osteocalcin and bone sialoprotein promoter-reporter activities were determined in LNCaP, C4-2B, DU145, PC3, ARCaP, and MG63 cell lines in the presence or absence of ARCaP conditioned medium (15 Ag/mL). Human osteocalcin and bone sialoprotein promoter activities were dramatically elevated by ARCaP conditioned medium in LNCaP and C4-2B cells. Fold induction was calculated from the promoter activities assayed in the presence or absence of conditioned medium. Columns, mean of three independent experiments with duplicate assays in each experiment; bars, SD. Significant differences of the fold inductions of human osteocalcin or human bone sialoprotein reporter activity were observed by the addition of ARCaP conditioned medium: **, P < 0.005. E, reverse transcription-PCR was done using total RNA isolated from LNCaP, C4-2B, PC3, and MG63 cells in the absence (-) or presence (+) of ARCaP conditioned medium (15 Ag/mL of total protein) for a 12-hour incubation period. Expression of the housekeeping gene GAPDH was used as a loading control. Relative expression values of osteocalcin and bone sialoprotein mRNA, normalized by the amounts of GAPDH mRNA expression, were measured by Gel Doc gel documentation software. Fold induction represents the ratios of ARCaP conditioned medium–treated versus vehicle-treated control of each cell line.

PC3 and MG63 (38) cells, ARCaP conditioned medium did not CRE Is Responsible for Regulation of Conditioned Medium– enhance further osteocalcin and bone sialoprotein mRNA Mediated Human Osteocalcin and Bone Sialoprotein Promoter expression (0.9- to 1.3-fold induction, respectively). As with the Activities. Our previous data (9) showed that three cis-elements are promoter activity assay, the steady-state levels of endogenous critical for the regulation of human osteocalcin promoter activity: osteocalcin and bone sialoprotein mRNA also did not show an OSE1, OSE2, and AV. To determine whether these elements are increase in DU145 and ARCaP cell lines on exposure to ARCaP important for ARCaP conditioned medium–activated human conditioned medium (data not shown). osteocalcin promoter activity, we assessed the activities of several

Cancer Res 2005; 65: (6). March 15, 2005 2306 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2005 American Association for Cancer Research. Osteocalcin and Bone Sialoprotein Expression by PKA human osteocalcin promoter deletion constructs, including single, double-base mutations at 640 and 639 CC!AA (Mut6) and double, or triple deletion constructs of OSE1, OSE2, and AV gen- deletion at this same region, CC!XX (Mut7), dramatically reduced erated by the recombinant PCR method as described previously (9). ARCaP conditioned medium–induced human osteocalcin promoter Figure 2A shows that among the single deletion constructs, DAV did activity. Although we did not perform similar mutational analysis of not seem to affect ARCaP conditioned medium–induced human human bone sialoprotein promoter because of the structural osteocalcin promoter luciferase activity. In comparison, a slight drop identity between CREs among human osteocalcin and bone of human osteocalcin promoter activity was observed on the de- sialoprotein promoters (with the exception of a single base letion of OSE1 or OSE2. No further decrease in human osteocalcin difference found in CRE1), it is highly likely that similar mutations promoter-luciferase activity induced by ARCaP conditioned medi- in CRE1 and CRE2 will result in disruption of human bone um was noted by deleting additional cis-elements, including the sialoprotein promoter activity when assayed in prostate cancer complete deletion of all three critical human osteocalcin regulatory cells. Taken together, the results show that two nucleotides, 640 (C) elements, DAV, DOSE2, and DOSE1. These data suggest that regions and 639 (C) within the CRE cis-element of human osteocalcin outside of OSE1, OSE2, and AV must be responsible for human promoter, are cooperatively responsible for the ARCaP conditioned osteocalcin promoter activation by ARCaP conditioned medium. medium–mediated human osteocalcin promoter activation. To address this question, we generated three new deletion cAMP-Dependent PKA Signaling Pathway Is Essential for constructs with regions outside of the OSE1, OSE2, or AV element Mediating ARCaP Conditioned Medium–Activated Osteocalcin systematically deleted. These are designated DA (upstream of the and Bone Sialoprotein Gene Expression in Human Prostate AV element, 374 bp; Fig. 2B), DB (between AV and OSE2 site, 327 Cancer Cells. The cAMP-dependent PKA pathway has long bp), and DC (between OSE2 and OSE1 site, 99 bp). As shown in Fig. been shown to mediate specific intracellular signaling events, 2B, a dramatic decrease in human osteocalcin promoter activity including the transcription of specific genes via the CRE cis-element was observed only when region A was deleted. Minimal loss of (25, 30, 31). To determine whether ARCaP conditioned medium ARCaP conditioned medium–induced human osteocalcin pro- stimulation of human osteocalcin and bone sialoprotein promoter moter luciferase activity was detected with deletion of region B or activities may be mediated through an activation of the PKA C (data not shown). signaling pathway, we assessed the effects of PKA pathway To find the specific cis-DNA element within region A that may activators, db cAMP and forskolin, on human osteocalcin and bone be responsible for ARCaP conditioned medium–induced human sialoprotein promoter activities in C4-2B cells and correlated this osteocalcin promoter activity, we site-specifically deleted selected with endogenous mRNA expression in various human prostate regions of A, DTst-1 (POU-factor Tst-1/Oct-6, 848 to 834), cancer cell lines. The PKA pathway activators, db cAMP (106-103 DCRE (643 to 636), and DIRF-1 (609 to 597), based on a mol/L) and forskolin (108-105 mol/L), stimulated human computer database search, and tested the activity of these osteocalcin (Fig. 3A) and human bone sialoprotein (Fig. 3B) constructs in C4-2B cells exposed to either ARCaP conditioned promoter activities in a ligand concentration–dependent manner medium or control medium. Figure 2B shows that only the in C4-2B cells. These results were confirmed by an assessment of DCRE construct exhibited a marked decreased in ARCaP endogenous osteocalcin and bone sialoprotein mRNA expression conditioned medium–induced human osteocalcin promoter after induction by a PKA activator, forskolin. Following forskolin luciferase activity, suggesting that cAMP may mediate down- treatment (106 mol/L), osteocalcin and bone sialoprotein mRNA stream signaling through CRE, regulating ARCaP conditioned expression in LNCaP and C4-2B but not in PC3 and MG63 were medium–induced human osteocalcin promoter activity. elevated (Fig. 3C). In this study, the steady-state levels of osteocalcin Then, we generated CRE deletion constructs of human bone mRNA were elevated by 5.2- and 7.8-fold, whereas the levels of bone sialoprotein promoter. Based on computer search, there are two sialoprotein mRNA were increased by 3.2- and 5.4-fold after forskolin putative CRE sites, CRE1 (79 to 72) and CRE2 (674 to 667), stimulation in LNCaP and C4-2B cells, respectively. This result was within human bone sialoprotein promoter. As shown in Fig. 2C, in general agreement with the effect of ARCaP conditioned medium human bone sialoprotein promoter luciferase activity decreased on endogenous osteocalcin and bone sialoprotein mRNA expres- partially in the two single deletion constructs of either CRE1 or sion (Fig. 1E), further supporting the involvement of PKA as the CRE2 (designated as DCRE1 and DCRE2). However, human bone key downstream signaling pathway regulating soluble factor– sialoprotein promoter activation was markedly reduced in the mediated osteocalcin and bone sialoprotein gene expression in double-deleted DCRE2/CRE1 construct when exposed to ARCaP LNCaP and C4-2B human prostate cancer cells. conditioned medium. This study shows that CREs are also The involvement of the PKA pathway in mediating ARCaP important for the regulation of human bone sialoprotein promoter conditioned medium activation of osteocalcin and bone sialo- reporter activity enhanced by ARCaP conditioned medium. protein expression was further confirmed using a selective To delineate the specific nucleotide(s) within the CRE of human inhibitor of PKA, H-89 (39). Forskolin-stimulated human osteo- osteocalcin promoter that may be responsible for ARCaP condi- calcin and bone sialoprotein promoter activities were both tioned medium–regulated promoter activity, we introduced either inhibited by H-89 (108-106 mol/L) in a concentration- one or two point mutations in CRE and examined ARCaP dependent manner (Fig. 4A). Consistent with this observation, conditioned medium–induced human osteocalcin promoter lucif- H-89 also inhibited ARCaP conditioned medium– and db cAMP- erase activity in C4-2B cells. Only Mut3 (640 C!A) and Mut4 mediated activation of human osteocalcin promoter activity in (639 C!A) greatly diminished the ARCaP conditioned medium– prostate cancer cells (Fig. 4B). Interestingly, phorbol 12-myristate activated human osteocalcin promoter activity (Fig. 2D). Other 13-acetate (PMA), an activator of the protein kinase C pathway, single base mutations, 642 G!T (Mut1), 641 A!C (Mut2), and also induced human osteocalcin promoter activity to a lesser 638 T!G (Mut5), failed to exert much influence on conditioned extent and such activation, as expected, was not blocked by H-89 medium–mediated human osteocalcin promoter activity when (Fig. 4B). In agreement with the above results from the human assayed under the same conditions. Consistent with these results, osteocalcin and bone sialoprotein promoter study, we confirmed www.aacrjournals.org 2307 Cancer Res 2005; 65: (6). March 15, 2005

Figure 2. CRE is responsible for the human osteocalcin and bone sialoprotein promoter activation induced by ARCaP conditioned medium. A, deletion analysis of human osteocalcin promoter. Three cis-elements, AV, OSE2, and OSE1 (9), are not critical for the human osteocalcin promoter activation regulated by ARCaP conditioned medium (basal luciferase activities, expressed as relative luciferase activity, in control human osteocalcin/Luc and DAV/OSE2/OSE1 were 1440 F 58 and 1160 F 200, respectively). ARCaP conditioned medium–mediated human osteocalcin promoter reporter activity was not affected by the elimination of these three cis-elements. B, deletion of CRE element abrogated the ARCaP conditioned medium–mediated activation of human osteocalcin promoter reporter activity. Region A (374 bp), upstream from AV element, contains three cis-elements, Tst-1 (848 to 834), CRE (643 to 636), and IRF-1 (609 to 597). Deletion of region A (DA) mutant in human osteocalcin promoter dramatically decreased the conditioned medium–mediated activation of the promoter activity. Subsequently, DTst-1, DCRE, and DIRF-1 mutant constructs were generated from the human osteocalcin promoter using the recombinant PCR method. Only the DCRE construct abolished ARCaP conditioned medium–induced human osteocalcin promoter activity. Relative activities of various human osteocalcin mutation reporter constructs were determined in the presence or absence of ARCaP conditioned medium (minus ARCaP conditioned medium of the human osteocalcin/Luc promoter activity was designed as 1.0).*, P < 0.05; **, P < 0.005, significant difference of the relative luciferase activity. C, two putative CRE sites were cooperatively regulated in the human bone sialoprotein promoter activity by ARCaP conditioned medium. Single deletion of CRE1 (DCRE1, 79 to 72) or CRE2 (DCRE2, 674 to 667) in human bone sialoprotein promoter reduced partially the promoter activation; the double deletion DCRE2/CRE1 construct markedly decreased the ARCaP conditioned medium–induced human bone sialoprotein promoter activity (the human bone sialoprotein/Luc promoter activity was assigned as 1.0 in the absence of ARCaP conditioned medium). *, P < 0.05; **, P < 0.005. Columns, mean of three independent studies with duplicate assays in each experiment; bars, SD. D, point mutation constructs of the CRE site within human osteocalcin promoter were constructed using the QuikChange site-directed mutagenesis kit (see Materials and Methods). Relative activities of human osteocalcin mutation reporter constructs were determined and compared with the human osteocalcin/Luc construct (assigned as 1.0 without adding ARCaP conditioned medium). Columns, mean of three independent studies with duplicate assays in each experiment; bars, SD. *, P < 0.05; **, P < 0.005, significant difference from the relative luciferase activity of human osteocalcin/Luc. that H-89 also inhibited the induction of endogenous osteocalcin specific osteocalcin and bone sialoprotein gene expression in and bone sialoprotein mRNA expression by ARCaP conditioned human prostate cancer cell lines by soluble factor(s) in prostate medium or forskolin in LNCaP and C4-2B cells (data not shown). cancer or bone conditioned medium occurs primarily through We evaluated further whether the cAMP/PKA signaling pathway the PKA signaling pathway involving cAMP as a mediator. may be operative under induction by conditioned medium Evidence in Support of Nuclear CREB and Cis-Element, CRE, harvested from C4-2B, DU145, PC3, and MG63 cell lines. Figure in the Regulation of Bone-Specific Gene Expression in Human 4C shows that H-89 (106 mol/L) nearly abolished conditioned Prostate Cancer Cells: Electrophoretic Mobility Shift Assay. medium–induced human osteocalcin promoter activity in all of To further establish a downstream link between the cAMP- conditioned medium harvested from prostate cancer and bone dependent PKA signaling pathway and human osteocalcin and stromal cell lines. These data suggest that stimulation of bone- bone sialoprotein promoter activation in prostate cancer cells, we

Cancer Res 2005; 65: (6). March 15, 2005 2308 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2005 American Association for Cancer Research. Osteocalcin and Bone Sialoprotein Expression by PKA conducted electrophoretic mobility shift assay to compare the formed a low but detectable basal level of complexes with 32P- binding of a 32P-labeled oligonucleotide CRE probe and nuclear labeled-CRE probe before and after treatment with ARCaP factors extracted from C4-2B cells (an ARCaP conditioned conditioned medium (Fig. 5B, lanes 4 and 5). These complexes medium–positive responder) and MG63 cells (an ARCaP condi- could be competed off by unlabeled-CRE probe (lane 6) but failed tioned medium–negative responder) that had been exposed to be supershifted by anti-CREB-specific antibody (lane 7). previously to ARCaP conditioned medium (15 Ag/mL) or forskolin As expected, nuclear extracts from C4-2B cells exposed to ARCaP (105 mol/L) for 16 hours in culture. Cells exposed to vehicle were conditioned medium did bind to CRE and these CRE-nuclear used as controls. Nuclear factors extracted from either ARCaP protein complexes can be supershifted by anti-CREB antibody conditioned medium– or forskolin-treated C4-2B cells strongly (lane 2) but not by anti-Runx2 antibody in both cases (lanes 3 enhanced the specific CRE-nuclear protein complex formation and 8). These data show that the trans-acting factor CREB may (Fig. 5A, lanes 3 and 5) in comparison with cells exposed to control play a critical role in regulating bone-specific gene transcription medium (Fig. 5A, lane 2). These DNA-protein complexes could be through the cAMP/PKA pathway by ARCaP conditioned medium in competed off by unlabeled specific CRE-oligo probe (lanes 4 and 6). human prostate cancer but not in bone stromal cells. However, no competition was observed with a mutant form of CRE-oligo probe, the Mut6-oligo (lane 9, two-point substitution; see Fig. 2D). Consistent with the biochemical data, H-89 was shown to Discussion abolish both ARCaP conditioned medium– and forskolin-induced Prostate cancer expresses proteins normally restricted to bone, CRE binding to the nuclear proteins extracted from C4-2B cells such as osteocalcin, bone sialoprotein, osteopontin, and osteo- (lanes 7 and 8). Nuclear extracts from MG63 cells, in contrast, nectin (6, 7, 11, 40–43). We hypothesized that the osteomimetic

Figure 3. Human osteocalcin and bone sialoprotein promoter activities were stimulated by db cAMP and forskolin (FSK) in a dose-dependent manner. C4-2B cells were cotransfected with human (A) osteocalcin or (B) bone sialoprotein promoter plus cytomegalovirus/h-galactosidase plasmid (an internal control plasmid for transfection efficiency). Transient transfected cells were treated with different concentrations of db cAMP (106-103 mol/L, M), forskolin (108-105 mol/L, M), or ARCaP conditioned medium (15 Ag/mL) for 16 hours. Human osteocalcin and bone sialoprotein promoter activities were induced by these pharmacologic reagents in a concentration-dependent manner through the activation of cAMP-dependent PKA pathway. Columns, mean fold induction of three separate studies with duplicate assays in each experiment; bars, SD. *, P < 0.05; **, P < 0.005, significant difference from control. C, reverse transcription-PCR was done using total RNA (5 Ag) isolated from LNCaP, C4-2B, PC3, and MG63 cell lines in the absence (-) or presence (+) of forskolin (106 mol/L, M) exposure for 12 hours. Relative expression values of osteocalcin and bone sialoprotein mRNA, normalized by the amounts of GAPDH mRNA, were measured by Gel Doc gel documentation software. Fold induction represents the ratios of forskolin-treated versus untreated specimens from each cell line. www.aacrjournals.org 2309 Cancer Res 2005; 65: (6). March 15, 2005

property of prostate cancer cells results from transcription factor switching (9, 44). Because of the potential importance of osteomimicry in enhancing cancer cell adhesion, invasion, and metastasis (6, 7, 11), we designed studies to define how the expression of bone-specific proteins in prostate cancer cells is regulated. We chose to study osteocalcin and bone sialoprotein because of the prevalence of expression of these genes by prostate cancer cell lines (8, 45) and in clinical prostate cancer specimens (7, 8, 46). Our data showed that osteocalcin and bone sialoprotein expression is stimulated in a selective manner in human prostate cancer but not in bone stromal cell lines. We observed that the induction of osteocalcin and bone sialoprotein expression in prostate cancer cell lines is mediated by paracrine/ autocrine factors harvested from conditioned medium of prostate cancer and bone stromal cells (47). The selective nature of soluble factors serving as paracrine mediators in stimulating osteocalcin and bone sialoprotein expression in prostate cancer cell lines agrees with their immunohistochemical staining patterns that are generally heterogeneous (see Fig. 1A; refs. 7, 8, 46). Using the stimulatory response of human osteocalcin/ human bone sialoprotein promoter reporter activities and the endogenous steady-state levels of osteocalcin/bone sialoprotein mRNA in C4-2B as the assay end points, we showed that the extent of human osteocalcin promoter activation by conditioned medium collected from prostate cancer and bone stromal cell lines correlated directly with the malignant potential of these cells in laboratory immune-compromised mice. We also observed that conditioned medium stimulated human osteocalcin and bone sialoprotein promoter activities in a dose-dependent manner and that these increased promoter reporter activities corresponded with the enhancement of the steady-state levels of endogenous osteocalcin and bone sialoprotein mRNA expressed in LNCaP and C4-2B cells. These results support the idea that osteocalcin and bone sialoprotein, once induced in prostate cancer cells, could facilitate the formation of hydroxyapatite complexes, leading to altered biological functions of prostate cancer cells (i.e., increased cell adhesion, migration and invasion, and recruitment of bone cells). Enhanced bone turnover could contribute to prostate cancer ‘‘seeding’’ in the skeleton (4, 11). A CRE cis-element, located upstream of the AV region, is responsible for mediating ARCaP conditioned medium activated human osteocalcin promoter activity and osteocalcin mRNA expression in human prostate cell lines (Fig. 2B). This cis-element seems to be functional in LNCaP and C4-2B cells but remains silent Figure 4. Effects of a selective inhibitor of PKA pathway H-89 on ARCaP in several other human prostate cancer (DU145, PC3, and ARCaP) conditioned medium–, db cAMP-, or forskolin-induced elevation of and osteosarcoma (MG63) cell lines. This is supported by the lack human osteocalcin (hOC) and bone sialoprotein (hBSP) promoter activities. A, C4-2B cells transfected with human osteocalcin or human bone of increased osteocalcin expression in these prostate cancer and sialoprotein promoter-reporter constructs were treated with various osteosarcoma cell lines after exposure to ARCaP conditioned concentrations of H-89 (108-106 mol/L, M) for 2 hours and subsequently exposed to forskolin (106 mol/L, M) for an additional 16 hours. H-89 medium (Fig. 1D and E). Further, nuclear extracts from prostate exerted a concentration-dependent inhibition of human osteocalcin and cancer and bone cells that failed to respond to PKA pathway bone sialoprotein promoter-reporter activities induced by forskolin. Fold activators showed a lack of CRE binding and supershift activity. In induction represents the folds of forskolin-treated reporter activities assayed in the presence or absence of H-89. *, P < 0.05; **, P < 0.005. B, H-89 sharp contrast, however, an increased complex formation between inhibited ARCaP conditioned medium–, db cAMP-, or forskolin-induced cis-element CRE and trans-acting factor CREB and an expected human osteocalcin promoter activity but did not inhibit the promoter-reporter supershift of this complex were observed in the responsive C4-2B activity assayed under stimulation by phorbol 12-myristate 13-acetate (PMA). H-89 (106 mol/L, M) was added to human osteocalcin promoter-reporter cells on the addition of anti-CREB antibody (Fig. 5A and B). The transiently transfected C4-2B cells for 2 hours and then exposed to complex formation is likely to account for increased human ARCaP conditioned medium (15 Ag/mL), db cAMP (103 mol/L, M), forskolin (106 mol/L, M), or the protein kinase C pathway activator phorbol osteocalcin promoter activity and enhanced endogenous osteocal- 12-myristate 13-acetate (PMA, 106 mol/L, M) for 16 hours. C, H-89 also cin mRNA expression in human prostate cancer cells in response abolished the human osteocalcin promoter activation induced by C4-2B, to factor(s) from the tumor cell microenvironment. A similar DU145, PC3, or MG63 conditioned medium. Columns, mean of three independent experiments with duplicate assays; bars, SD. *, P < 0.05; **, mechanism regulating human bone sialoprotein promoter activity P < 0.005. and mRNA expression was also observed in LNCaP and C4-2B cells

Cancer Res 2005; 65: (6). March 15, 2005 2310 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2005 American Association for Cancer Research. Osteocalcin and Bone Sialoprotein Expression by PKA after exposure to ARCaP conditioned medium, a result supported and 4A). Our results show for the first time that conditioned by additional studies using interfering pharmacologic agents that medium harvested from prostate cancer and bone stromal cells either stimulated (db cAMP and forskolin) or repressed (H-89) stimulated osteocalcin and bone sialoprotein expression primarily cAMP accumulation in target cells (Figs. 1D and E,2C,3B and C, through a cAMP-dependent PKA signaling pathway in LNCaP and C4-2B human prostate cancer cells. This stimulation by conditioned medium lends further support to the observation that osteocalcin and bone sialoprotein are prevalently expressed by clinically localized and bone and lymph node metastatic human prostate cancer tissue specimens and can be the molecular basis of osteomimicry during disease progression. Several growth factors (fibroblast growth factor-2 or basic fibroblast growth factor, insulin-like growth factor-I, and trans- forming growth factor-h) and hormones (glucocorticoids, estro- gens, parathyroid hormone, and parathyroid hormone–related peptide) have been shown to regulate osteocalcin and bone sialoprotein expression in rodent, chick, or human cell lines (48–54). Boudreaux and Towler (55) showed the induction of osteocalcin promoter activity by a growth factor, fibroblast growth factor-2, in MC3T3-E1 cells. Boguslawski et al. (30) observed that rat and human osteoblast-like cell lines stably transfected with osteocalcin promoter reporter construct responded to parathyroid hormone and growth factors (fibroblast growth factor-2 and insulin-like growth factor-I) via enhanced osteocalcin transcription mediated by a PKA-dependent pathway. A similar activation of bone sialoprotein promoter activity was documented for fibroblast growth factor-2 (54, 56), parathyroid hormone (33, 53), and prostaglandin E2 (34) in which bone sialoprotein transcription was stimulated through a PKA-dependent pathway, although a protein kinase C–mediated pathway may play a minor role (33). The present study differs from previous reports in two important aspects. First, none of the earlier observations showed conclusively at the molecular level that the involvement of CRE elements within osteocalcin and bone sialoprotein promoters, as defined in the present study, conferred soluble factor–induced osteocalcin and bone sialoprotein expression by cancer or normal cells. Our results emphasized the roles of CRE cis- elements in osteomimicry by prostate cancer cells. Second, none of the previously reported factors tested in our assay system activated human osteocalcin and bone sialoprotein promoter activities in C4-2B cells (data not shown). These important distinctions suggest the following possibilities. (a) There are fundamental differences in terms of the soluble factors, downstream signaling network, and cis-elements within human osteocalcin and bone sialoprotein promoters that are responsible for Figure 5. ARCaP conditioned medium and forskolin enhanced CREB and CRE mediating the osteomimetic properties of prostate cancer and binding through cAMP-dependent PKA signaling pathway in selective human bone cells. (b) At the molecular level, prostate cancer and bone prostate cancer but not in bone cells. A, C4-2B cells were exposed to ARCaP conditioned medium (CM, 15 Ag/mL) or forskolin (F, 105 mol/L) for 16 hours; cells may differ in their cell surface receptors and downstream control cells were exposed to vehicles. Cells were harvested and nuclear cell signaling pathways responding to soluble factor(s) secreted extracts (NE) prepared. Electrophoretic mobility shift assay was done by incubating nuclear extracts and the 32P-labeled CRE probe. Lanes 3 and 5, by prostate cancer and bone stromal cells and their elicited ARCaP conditioned medium and forskolin enhanced the formation of activation of osteocalcin and bone sialoprotein expression. (c) CRE-nuclear protein complexes. Lanes 4 and 6, complexes were competed off With respect to the osteomimetic response of prostate cancer by unlabeled specific CRE-oligo probe. Lane 9, Mut6-oligo (the CRE mutant of two-point substitution; see Fig. 2D) did not compete with the nuclear proteins cells to soluble factors, prostate cancer cell lines clearly show and 32P-CRE-oligo complexes. Lanes 7 and 8, H-89 (106 mol/L) blocked heterogeneity. Because osteocalcin and bone sialoprotein expres- both ARCaP conditioned medium– and forskolin-induced CRE binding to the sion in LNCaP, C4-2, and C4-2B cells are responsive to soluble nuclear factors extracted from C4-2B cells. B, C4-2B (lanes 1-3) and MG63 (lanes 4-8) cells were treated with ARCaP conditioned medium (15 Ag/mL) or factors and osteocalcin and bone sialoprotein are prevalently vehicle for 16 hours and nuclear extracts were prepared. Lanes 4 and 5, no or expressed in both primary and metastatic human prostate minimum changes of the binding complex formation when experiments were conducted using nuclear extracts from MG63 cells either exposed to ARCaP cancer tissues, we suggest that LNCaP and its derivative cell conditioned medium or untreated. Arrow, CRE and CREB complexes, which lines are superior models for the study of human prostate cancer were supershifted by adding anti-CREB antibody to the nuclear extracts from progression. This conclusion is also supported by a large body of C4-2B cells (lane 2) but not from MG63 cells (lane 7). Specificity of the supershift complex was confirmed by the lack of a supershift band when naive Runx2 literature devoted to the study of the molecular mechanisms antibody was used as a reagent (lanes 3 and 8). of androgen receptor and androgen-independent metastatic www.aacrjournals.org 2311 Cancer Res 2005; 65: (6). March 15, 2005

responsible factors and evaluate cell surface or intracellular receptors coupling to cell signaling systems in prostate cancer and bone cells. In summary, our data using human prostate cancer cell lines show dramatic cell background–dependent differences in responsiveness to ARCaP conditioned medium and the involvement of the cAMP-PKA signaling cascade in the induction of osteocalcin and bone sialoprotein gene expression in LNCaP and C4-2B cells. We established at the molecular level that a specific region of cis- element in human osteocalcin promoter, located between 643 and 636 (CRE), must be responsible for conferring cAMP regulation of human osteocalcin promoter activity in human prostate cancer cells. Likewise, other regions of CRE within human bone sialoprotein promoter, 79 to 72 (CRE1) and 674 to 667 (CRE2), must also be activated on exposure of human prostate cancer cells to cAMP mimetics and yet unidentified growth factor(s) in the conditioned medium of prostate cancer and bone stromal cells. We propose that unknown soluble factors secreted by human prostate cancer or bone stromal cells could assume the key regulatory role in osteocalcin and bone sialoprotein expression in human prostate cancer (Fig. 6). An ongoing study to identify and Figure 6. Proposed cAMP-dependent PKA signaling mechanism describing the characterize the unknown soluble factors in our laboratory has regulation of human osteocalcin and bone sialoprotein promoter activities in revealed a small polypeptide (<30 kDa) with thermal sensitivity and human prostate cancer cells. An unknown soluble factor with a molecular mass of <30 kDa was proposed to be secreted by human prostate cancer and bone stromal ammonium sulfate–precipitable characteristics in conditioned cells. This putative factor is believed to interact with a cell surface receptor in medium, which seems exclusively responsible for the activation prostate cancer cells and subsequently to activate adenylate cyclase (AC), of human osteocalcin and bone sialoprotein promoter activities in resulting in a signal cascade through the PKA signaling pathway to enhance human osteocalcin and bone sialoprotein promoter activities. We propose the prostate cancer cells. These results could have significant molecular basis for osteomimicry as follows: cAMP generated by ligand-receptor implications for understanding osteomimicry and targeting it interaction promoted PKA activation and the activated PKA translocated into the therapeutically in human prostate cancer. cell nucleus and induced CREB phosphorylation, which interacted with CRE cis-elements in human osteocalcin (CRE) and human bone sialoprotein (CRE1 and CRE2) promoter regions and triggered marked downstream promoter Acknowledgments activation and endogenous mRNA expression in human prostate cancer cells. Received 9/27/2004; revised 12/17/2004; accepted 12/30/2004. Grant support: CA-76620, CA-098912, and DAMD-17-03-2-0033 Manhattan Project (L.W.K. Chung). progression using this LNCaP prostate cancer progression model The costs of publication of this article were defrayed in part by the payment of page (57–59). To further understand differences in responsiveness charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. among prostate cancer and bone stromal cells to soluble factors We thank our colleagues at the Molecular Urology and Therapeutics Program for in conditioned medium, we must isolate and characterize the helpful suggestions and discussion and Gary Mawyer for editing the article.

References osteopontin and bone sialoprotein in bone metastasis 13. Boskey AL, Gadaleta S, Gundberg C, Doty SB, Ducy P, of breast and prostate carcinoma. Clin Exp Metastasis Karsenty G. Fourier transform infrared microspectroscopic 1. Greenlee RT, Murray T, Bolden S, Wingo PA. Cancer 2003;20:437–44. analysis of bones of osteocalcin-deficient mice provides statistics, 2000. CA Cancer J Clin 2000;50:7–33. 8. Matsubara S, Wada Y, Gardner TA, et al. A insight into the function of osteocalcin. Bone 1998;23: 2. Martinez J, Silva S, Santibanez JF. Prostate-derived conditional replication-competent adenoviral vector, 187–96. soluble factors block osteoblast differentiation in Ad-OC-E1a, to cotarget prostate cancer and bone 14. Hauschka PV. Osteocalcin: the vitamin K-dependent culture. J Cell Biochem 1996;61:18–25. stroma in an experimental model of androgen- Ca2+-binding protein of bone matrix. Haemostasis 1986; 3. Lang SH, Clarke NW, George NJ, Allen TD, Testa NG. independent prostate cancer bone metastasis. Cancer 16:258–72. Interaction of prostate epithelial cells from benign and Res 2001;61:6012–9. 15. Ducy P, Desbois C, Boyce B, et al. Increased bone malignant tumor tissue with bone-marrow stroma. 9. Yeung F, Law WK, Yeh CH, et al. Regulation of formation in osteocalcin-deficient mice. Nature 1996; Prostate 1998;34:203–13. human osteocalcin promoter in hormone-independent 382:448–52. 4. Chung LW. Prostate carcinoma bone-stroma interac- human prostate cancer cells. J Biol Chem 2002;277: 16. Ganss B, Kim RH, Sodek J. Bone sialoprotein. Crit tion and its biologic and therapeutic implications. 2468–76. Rev Oral Biol Med 1999;10:79–98. Cancer 2003;97:772–8. 10. Brown LF, Papadopoulos-Sergiou A, Berse B, et al. 17. Hunter GK, Goldberg HA. Nucleation of hydroxyap- 5. Curatolo C, Ludovico GM, Correale M, et al. Advanced Osteopontin expression and distribution in human atite by bone sialoprotein. Proc Natl Acad Sci U S A prostate cancer follow-up with prostate-specific anti- carcinomas. Am J Pathol 1994;145:610–23. 1993;90:8562–5. gen, prostatic acid phosphatase, osteocalcin and bone 11. Koeneman KS, Yeung F, Chung LW. Osteomimetic 18. Ross FP, Chappel J, Alvarez JI, et al. Interactions isoenzyme of alkaline phosphatase. Eur Urol 1992;21 properties of prostate cancer cells: a hypothesis between the bone matrix proteins osteopontin and bone h Suppl 1:105–7. supporting the predilection of prostate cancer metas- sialoprotein and the osteoclast integrin av 3 poten- 6. Waltregny D, Bellahcene A, Van Riet I, et al. Prognostic tasis and growth in the bone environment. Prostate tiate bone resorption. J Biol Chem 1993;268:9901–7. value of bone sialoprotein expression in clinically 1999;39:246–61. 19. Mintz KP, Grzesik WJ, Midura RJ, Robey PG, Termine localized human prostate cancer. J Natl Cancer Inst 12. Price PA, Otsuka AA, Poser JW, Kristaponis J, JD, Fisher LW. Purification and fragmentation of 1998;90:1000–8. Raman N. Characterization of a g-carboxyglutamic nondenatured bone sialoprotein: evidence for a cryptic, 7. Carlinfante G, Vassilioul D, Svensson O, Wendel M, acid-containing protein from bone. Proc Natl Acad RGD-resistant cell attachment domain. J Bone Miner Heinegard D, Andersson G. Differential expression of Sci U S A 1976;73:1447–51. Res 1993;8:985–95.

Cancer Res 2005; 65: (6). March 15, 2005 2312 www.aacrjournals.org

20. Zhang JH, Tang J, Wang J, et al. Over-expression of 2 response elements in the proximal promoter of the rat the osteoblast transcription factor, Cbfa1/Runx2. J Biol bone sialoprotein enhances bone metastasis of human BSP gene. J Biol Chem 2003;278:28659–67. Chem 2002;277:36181–7. breast cancer cells in a mouse model. Int J Oncol 35. Gleave M, Hsieh JT, Gao CA, von Eschenbach AC, 49. Kim HJ, Kim JH, Bae SC, Choi JY, Ryoo HM. The 2003;23:1043–8. Chung LW. Acceleration of human prostate cancer protein kinase C pathway plays a central role in the 21. Fedarko NS, Fohr B, Robey PG, Young MF, Fisher LW. growth in vivo by factors produced by prostate and fibroblast growth factor-stimulated expression and Factor H binding to bone sialoprotein and osteopontin bone fibroblasts. Cancer Res 1991;51:3753–61. transactivation activity of Runx2. J Biol Chem 2003; enables tumor cell evasion of complement-mediated 36. Kiyoshima T, Yamauchi M, Wong C, Jheon A, Ganss B, 278:319–26. attack. J Biol Chem 2000;275:16666–72. Sodek J. An L1 element disrupts human bone sialopro- 50. Alliston T, Choy L, Ducy P, Karsenty G, Derynck R. 22. Roach HI. Why does bone matrix contain non- tein promoter: lack of tissue-specific regulation by TGF-h-induced repression of CBFA1 by Smad3 collagenous proteins? The possible roles of osteocalcin, distalless5 (Dlx5) and runt homeodomain protein2 decreases cbfa1 and osteocalcin expression and osteonectin, osteopontin and bone sialoprotein in bone (Runx2)/core binding factor a1 (Cbfa1) elements. Gene inhibits osteoblast differentiation. EMBO J 2001;20: mineralisation and resorption. Cell Biol Int 1994;18: 2002;299:205–17. 2254–72. 617–28. 37. Ausubel FM, Brent R, Kingston RE, et al. Current 51. Lian JB, Shalhoub V, Aslam F, et al. Species-specific 23. Stein GS, Lian JB, Owen TA. Relationship of cell protocols in molecular biology. New York: Greene Pub. glucocorticoid and 1,25-dihydroxyvitamin D respon- growth to the regulation of tissue-specific gene Associates and Wiley-Interscience; 1999. p. 12-1-1–12-2-10. siveness in mouse MC3T3-E1 osteoblasts: dexametha- expression during osteoblast differentiation. FASEB J 38. Mahonen A, Pirskanen A, Keinanen R, Maenpaa PH. sone inhibits osteoblast differentiation and vitamin D 1990;4:3111–23. Effect of 1,25(OH)2D3 on its receptor mRNA levels and down-regulates osteocalcin gene expression. Endocri- 24. Kondo H, Ohyama T, Ohya K, Kasugai S. Temporal osteocalcin synthesis in human osteosarcoma cells. nology 1997;138:2117–27. changes of mRNA expression of matrix proteins and Biochim Biophys Acta 1990;1048:30–7. 52. Ogata Y, Yamauchi M, Kim RH, Li JJ, Freedman LP, parathyroid hormone and parathyroid hormone-related 39. Chijiwa T, Mishima A, Hagiwara M, et al. Inhibition of Sodek J. Glucocorticoid regulation of bone sialoprotein protein (PTH/PTHrP) receptor in bone development. forskolin-induced neurite outgrowth and protein phos- (BSP) gene expression. Identification of a glucocorticoid J Bone Miner Res 1997;12:2089–97. phorylation by a newly synthesized selective inhibitor response element in the bone sialoprotein gene 25. Richards JS. New signaling pathways for hormones of cyclic AMP-dependent protein kinase, N-[2-(p- promoter. Eur J Biochem 1995;230:183–92. and cyclic adenosine 3V,5V-monophosphate action in bromocinnamylamino)ethyl]-5-isoquinolinesulfonamide 53. Ogata Y, Nakao S, Kim RH, et al. Parathyroid endocrine cells. Mol Endocrinol 2001;15:209–18. (H-89), of PC12D pheochromocytoma cells. J Biol Chem hormone regulation of bone sialoprotein (BSP) gene 26. Liu AY. Differentiation-specific increase of cAMP- 1990;265:5267–72. transcription is mediated through a pituitary-specific dependent protein kinase in the 3T3-L1 cells. J Biol 40. Chung LW, Hsieh CL, Law A, et al. New targets for transcription factor-1 (Pit-1) motif in the rat BSP gene Chem 1982;257:298–306. therapy in prostate cancer: modulation of stromal- promoter. Matrix Biol 2000;19:395–407. 27. Schwartz DA, Rubin CS. Regulation of cAMP- epithelial interactions. Urology 2003;62:44–54. 54. Shimizu-Sasaki E, Yamazaki M, Furuyama S, Sugiya dependent protein kinase subunit levels in Friend 41. Thalmann GN, Sikes RA, Devoll RE, et al. Osteopon- H, Sodek J, Ogata Y. Identification of a novel response erythroleukemic cells. Effects of differentiation and tin: possible role in prostate cancer progression. Clin element in the rat bone sialoprotein (BSP) gene treatment with 8-Br-cAMP and methylisobutyl xan- Cancer Res 1999;5:2271–7. promoter that mediates constitutive and fibroblast thine. J Biol Chem 1983;258:777–84. 42. Jacob K, Webber M, Benayahu D, Kleinman HK. growth factor 2-induced expression of BSP. J Biol Chem 28. Li M, West JW, Numann R, Murphy BJ, Scheuer T, Osteonectin promotes prostate cancer cell migration 2001;276:5459–66. Catterall WA. Convergent regulation of sodium chan- and invasion: a possible mechanism for metastasis to 55. Boudreaux JM, Towler DA. Synergistic induction of nels by protein kinase C and cAMP-dependent protein bone. Cancer Res 1999;59:4453–7. osteocalcin gene expression: identification of a bipartite kinase. Science 1993;261:1439–42. 43. Thomas R, True LD, Bassuk JA, Lange PH, Vessella RL. element conferring fibroblast growth factor 2 and cyclic 29. Roesler WJ, Vandenbark GR, Hanson RW. Cyclic AMP Differential expression of osteonectin/SPARC during AMP responsiveness in the rat osteocalcin promoter. and the induction of eukaryotic gene transcription. human prostate cancer progression. Clin Cancer Res J Biol Chem 1996;271:7508–15. J Biol Chem 1988;263:9063–6. 2000;6:1140–9. 56. Shimizu-Sasaki E, Yamazaki M, Furuyama S, Sugiya 30. Boguslawski G, Hale LV, Yu XP, et al. Activation of 44. Yeung F, Chung LW. Molecular basis of co-targeting H, Sodek J, Ogata Y. Identification of FGF2-response osteocalcin transcription involves interaction of protein prostate tumor and stroma. J Cell Biochem Suppl 2002; element in the rat bone sialoprotein gene promoter. kinase A- and protein kinase C-dependent pathways. 38:65–72. Connect Tissue Res 2003;44 Suppl 1:103–8. J Biol Chem 2000;275:999–1006. 45. Lin DL, Tarnowski CP, Zhang J, et al. Bone metastatic 57. Pfitzenmaier J, Quinn JE, Odman AM, et al. 31. Servillo G, Della Fazia MA, Sassone-Corsi P. Coupling LNCaP-derivative C4-2B prostate cancer cell line Characterization of C4-2 prostate cancer bone metas- cAMP signaling to transcription in the liver: pivotal role mineralizes in vitro. Prostate 2001;47:212–21. tases and their response to castration. J Bone Miner Res of CREB and CREM. Exp Cell Res 2002;275:143–54. 46. Ou YC, Chen JT, Yang CR, Ko JL, Hsieh YS, Kao C. 2003;18:1882–8. 32. Sassone-Corsi P. Coupling gene expression to cAMP Expression of osteocalcin in prostate cancer before and 58. Letsch M, Schally AV, Szepeshazi K, Halmos G, Nagy signalling: role of CREB and CREM. Int J Biochem Cell after hormonal therapy. Anticancer Res 2003;23:3807–11. A. Preclinical evaluation of targeted cytotoxic luteiniz- Biol 1998;30:27–38. 47. Huang WC, Zhau HE, Chung LW. Invasive prostate ing hormone-releasing hormone analogue AN-152 in 33. Yang R, Gerstenfeld LC. Signal transduction path- cancer cells secret soluble protein factors that upregu- androgen-sensitive and insensitive prostate cancers. ways mediating parathyroid hormone stimulation of late osteocalcin (OC) promoter activity: implication to Clin Cancer Res 2003;9:4505–13. bone sialoprotein gene expression in osteoblasts. J Biol stroma desmoplastic response and cancer metastasis. 59. Zayzafoon M, Abdulkadir SA, McDonald JM. Chem 1996;271:29839–46. J Urol 2003;169:154–4; abstract 597. Notch signaling and ERK activation are important 34. Samoto H, Shimizu E, Matsuda-Honjyo Y, et al. 48. Xiao G, Jiang D, Gopalakrishnan R, Franceschi RT. for the osteomimetic properties of prostate cancer Prostaglandin E2 stimulates bone sialoprotein (BSP) Fibroblast growth factor 2 induction of the osteocalcin bone metastatic cell lines. J Biol Chem 2004;279: expression through cAMP and fibroblast growth factor gene requires MAPK activity and phosphorylation of 3662–70.

www.aacrjournals.org 2313 Cancer Res 2005; 65: (6). March 15, 2005

Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2005 American Association for Cancer Research. Human Osteocalcin and Bone Sialoprotein Mediating Osteomimicry of Prostate Cancer Cells: Role of cAMP-Dependent Protein Kinase A Signaling Pathway

Wen-Chin Huang, Zhihui Xie, Hiroyuki Konaka, et al.

Cancer Res 2005;65:2303-2313.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/65/6/2303

Cited articles This article cites 56 articles, 25 of which you can access for free at: http://cancerres.aacrjournals.org/content/65/6/2303.full#ref-list-1

Citing articles This article has been cited by 18 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/65/6/2303.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/65/6/2303. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.