Differential Expression of Osteonectin/SPARC during Human Prostate Cancer Progression¹

The ability of cancer cells to invade and metastasize from the primary site is a hallmark of malignancy. Such is the intolerant nature of this process that patients with distant metastatic disease are almost invariably incurable. Thus, it is imperative to understand the cellular and molecular mechanisms of tumor dissemination to develop novel therapies for intervention. The acquisition of invasive and metastatic potential involves complex cell-cell and cell-ECM³ interactions. Cell surface glycoproteins with their matrix ligand have been implicated in this process. ON also known as BM-40 or SPARC, is a secreted glycoprotein that modulates the interaction of cells with the ECM. Other members of the ON family that have a similar basic structure to ON include hevin (1), SC1 (2), QR-1 (3), follistatin-like protein (TSC-36; Ref. 4) and testican (5). Of these proteins, the 232 COOH-terminal amino acids of hevin (human SC1) show a 62% identity to the homologous portion of human SPARC. Although the specific role of ON is elusive,the high degree of evolutionary conservation of this protein among different species suggests that ON has an important physiological role (6, 7, 8, 9). Although initial studies associated this protein with bone mineralization, recent studies indicate that this protein has pleiotropic effects on biological functions as diverse as proliferation (10, 11, 12), morphogenesis (7, 13), tissue remodeling (14, 15), migration (16), and angiogenesis (10, 17, 18). The phenotype of SPARC-null mice is such that they develop age-related cataracts due to aberrant differentiation of lenticular epithelial cells and incomplete fiber cell differentiation (19).

A variety of in vitro studies have shown that secretion of ON affects cell morphology by reducing the number of focal contacts and blocking the adhesion of cells to their substratum or to neighboring cells (9, 20). ON expression is up regulated in transformed cells (7). Increased expression is also associated with the neoplastic progression of several human tumor types (21, 22, 23, 24, 25, 26, 27, 28). Suppression of ON expression using antisense RNA significantly decreased the tumorigenicity of human melanoma cells by reducing their invasive and adhesive properties (26). To date, however, there are no detailed studies on ON expression in prostatic tissue. The aim of the current study was to analyze ON mRNA and protein expression in malignant as well as normal prostatic tissues, CaP cell lines, and immortalized prostate epithelial cell lines.

Unless otherwise specified, all of the human CaP cell lines were grown in RPMI 1640 (Life Technologies, Inc., Gaithersburg, MD) supplemented with 10% fetal bovine serum. Cultures were maintained at 37°C in a 5% CO₂ humidified incubator. The cell types used in this study were human CaP cell lines including DU 145 (29),parental PC-3 (30), parental LNCaP (31),VeCaPs⁴ (a gift from Dr. K. Pienta, University of Michigan, MI), and prostate epithelial cell line PZ-HPV-7 (32). Other cell lines include sublines of LNCaP (32), the C4, C4-2, and C4-2b, and sublines of PC-3, N-PC3 (noninvasive), and B-PC3 (invasive; both are gifts from Dr. M. Stearns, Medical College of Pennsylvania, Philadelphia, PA). C4 is an androgen-dependent cell line derived from mice that were coinoculated with LNCaP and the human osteosarcoma cell line MS. C4-2 is a second-generation cell line derived from C4, and it is androgen independent. C4-2b is another subline derived from the bone metastasis of a mouse that was inoculated with C4-2. VeCaps is a CaP cell line derived from a spine metastasis and grown in DMEM supplemented with 10% fetal bovine serum. PZ-HPV-7 is a human papilloma virus immortalized cell line derived from normal human prostate epithelial cells and is grown in keratinocyte-serum-free medium supplemented with 5 ng/ml epidermal growth factor and 50 μg/ml bovine pituitary extract(Life Technologies, Inc.).

FITC-labeled anti-CD44 (Clone G44-26) and PE-labeled CD57 (clone NK-1)were purchased from PharMingen (San Diego, CA). A polyclonal serum(LF37) raised against human ON was a kind gift from Dr. L. W. Fisher (NIH; Ref. 33). A mAb (ON-mAb; lot L0122) raised against human platelet ON was purchased from Hematological Technologies, Inc. (Essex Junction, VT). For PSA staining, a polyclonal antibody (α-PSA) raised against human PSA was purchased from DAKO (Carpinteria, CA). Mouse IgG1, κ-chain (MOPC-21) control was purchased from Sigma (St. Louis,MO). Anti-DIG, a mAb to DIG from mouse-mouse hybrid cells (clone 1.71.256), was purchased from Roche Molecular Biochemicals(Indianapolis, IN). Chinese hamster ovary cells expressing rHevin were a gift from Dr. P. S. Nelson (University of Washington, Seattle,WA). Recombinant ON used in this study was expressed in Escherichia coli (34).

Normal prostate was minced and digested with collagenase (3.0 mg/ml),and three cell populations were separated as described previously (35). The epithelial fraction was resuspended in 50 μl of 0.1% BSA-PBS solution and 5.0 μl of FITC-labeled anti-CD44 (50μg/ml) and PE-labeled CD57 (50 μg/ml) were added to 10⁶ cells. After 30 min of incubation at room temperature,the cells were washed with 0.1% BSA-PBS and resuspended in 0.5 ml of 0.1% BSA-PBS. The cells were analyzed by flow cytometry(FACStar^plus; Becton Dickinson, San Jose, CA) within 60 min. Gene expression analyses on the sorted cells were done using RT-PCR.

Total RNA from sorted and cultured cells were isolated using STAT-60(Tel-Test Inc., Friendswood, TX) according to the manufacturer’s suggested protocol. The quality of the RNA samples was determined by electrophoresis through an agarose gel and staining with ethidium bromide, and the 18S and 28S RNA bands were visualized under UV light. RT of 5 μg of RNA was done in a final volume of 20 μl containing 1× first strand buffer [250 mm Tris-HCl (pH 8.3), 375 mm KCl, and 15 mm MgCl₂], 5 mm DTT, 200 units of Moloney murine leukemia virus reverse transcriptase (Life Technologies, Inc.), 10 μm random hexamers, and 10 units of RNase inhibitor (Roche Molecular Biochemicals). A negative RT reaction control tube (lacking-RNA sample)was also included to check for contamination. The samples were incubated at room temperature for 10 min, followed by incubation at 37°C for 1 h. Finally, the RT reaction was terminated by heating at 95°C for 5 min.

Relative quantitation of ON gene expression among different prostate cell lines was done by a method described previously (36). In the present study, each sample was normalized on the basis of its β2-MIC content. The MIC gene was selected as an endogenous external control because the prevalence of its transcripts is similar to that of the ON target gene. Briefly, for each primer set (ON and MIC), the optimum amount of input RNA and a linear range of amplification were determined under optimum PCR conditions. Dilution assays were performed to determine the optimum amount of input RNA. The optimum cycle number was determined as the midpoint of the linear range of amplification. All subsequent amplifications were performed by using the optimum cycle number for each primer set. For determining linearity of RNA input versus RT-PCR product output, different dilutions of input RNA ranging from 2.0–8.0 μg were reverse-transcribed and subjected to separate amplification for ON and MIC under optimum cycle number. cDNA was amplified in 50-μl volumes using a thermocycler (Omni Gené, National Labnet Co., Woodbridge,NJ). Reaction mixtures contained 2 mm deoxynucleotide triphosphates, 2 mm MgCl₂, 5 pmol of each upstream primer (hON-Us2.0) and downstream primer (hON-Ds2.0), and 2μl of RT reaction products. PCR cycling conditions consisted of denaturation at 80°C for 3 min, followed by a three-step cycle of denaturation at 95°C (5 s), annealing at 65°C (30 s), and extension at 72°C (30 s). Thirty cycles were carried out before a final 7-min extension period at 72°C. These ON-specific primers yielded a 374-bp PCR-DNA product. For MIC, PCR conditions were 80°C for (3 min), 1 cycle; 95°C (5 s); 69°C (60 s), 28 cycles; and 72°C (7 min), 1 cycle. PCR with MIC-specific primers yields a 550-bp fragment. Under the above conditions, PCR amplification was within the linear range. Aliquots (10.0 μl) of PCR products were resolved by agarose gel electrophoresis. The fluorescent image was captured using an IS-1000 digital imaging system (Alpha Innotech, San Leandro, CA) and analyzed with AlphaEase (Alpha Innotech) software. Briefly, the digital image was analyzed to determine the pixel intensity of each band. Relative quantities of ON mRNA among different cell lines were calculated as the ratio of the ON:MIC pixel intensities from three replicate RT-PCR experiments. Positive results were based on the presence of DNA bands of the expected size. All the PCR-DNA products were validated by dideoxy sequencing of both strands.

Tissue specimens were obtained from (a) glands resected by radical prostatectomy, (b) organ donors, or (c)“rapid” autopsies conducted within 2–3 h of death. Excess prostatic tissues not needed for pathological diagnosis were obtained from the operating room and used for cell isolation. The histological compositions of these samples were assessed by a pathologist(L. D. T.). Adjacent sections were used in these studies. All the metastatic foci were tested with antibody against PSA. Immunohistochemical labeling for ON is known to vary under different conditions of tissue processing and fixation.⁵ In this study, we used frozen, formalin-fixed-, or Histochoice-fixed tissues. For frozen sections, tissue samples were snap-frozen in Tissue-Tek OCT (Sakura Finetek USA, Inc, Torrance, CA) by immersing in isopentane (−70°C). The frozen blocks were stored at −80°C. Five-μm sections were cut on a cryostat and mounted on a glass slide. The sections were briefly fixed in acetone and air-dried for 1 h before staining. For paraffin-embedded sections, tissues samples were fixed in either formaldehyde (10% v/v; Mallinckrodt Baker Inc., Paris, KY) or in Histochoice (Bio World Laboratories, Dublin, OH) before embedding in paraffin. Metastatic foci from bone were decalcified in 10% formic acid before being embedded in paraffin. Serial sections of 5.0 μm were cut and mounted on Superfrost Plus slides (VWR Scientific, West Chester, PA). The slides were air-dried at 60°C for 1 h. After deparaffinization in three changes of xylene and clearing in 100%ethanol, slides were hydrated and kept in APK solution (Ventana Medical System, Tucson, AZ) until used for ISH or IHC staining. For IHC using formalin-fixed tissues, antigen retrieval was performed for all slides. Slides were placed in 0.01 m sodium citrate (pH 6.0) and microwaved at full power for 16 min. These slides were allowed to cool for 10 min, washed three times in deionized water, and placed in Tris buffer (pH 7.4) until they were stained. These slides were immediately used for IHC staining.

DIG-labeled RNA probes for ON were created as follows. A part of the ON cDNA fragment (454 bp) was PCR-amplified using hOSN.Us3(5′-ACTGAGGTATCTGTGGGAGCTAATCC-3′) and hOSN.Ds2(5′-CAGTCAGAAGGTTGTTGTCCTCATCC-3′) primers. The PCR fragment was cloned into pGEM-T vector (Promega, Madison, WI) and analyzed by sequencing. These plasmids were linearized with Nco1 (for sense probe) and Not1(for antisense probe). These linearized plasmids were in vitro transcribed with DIG-11-uridine-5′-triphosphate along with 1 unit/μl T7 polymerase and SP6 polymerase. The labeled riboprobes were analyzed and quantitated before use. All enzymes and DIG RNA labeling mix were purchased from Roche Molecular Biochemicals. ISH was performed according to the manufacturer’s protocol on the Ventana gen^II automated ISH system (Ventana Medical System). ISH for ON was run using preprogrammed recipe files and consisted of a series of steps of buffer rinse, protease digestion, hybridization,detection reactions, and counterstains. DIG-labeled RNA probe was added manually. Anti-DIG was used as a primary antibody. Denaturation was at 80°C, and the hybridization was done at 41°C for 4 h. Washes were performed at 37°C with 2×, 1×, and 0.5× saline sodium citrate. The system uses a mixture of antirabbit and antimouse secondary IgG biotinylated antibody with an indirect biotin avidin diaminobenzidine detection system (Ventana Medical System). Finally,the sections were counterstained with hematoxylin. Appropriate ON-positive (skin tissue) and -negative (with sense probe) controls were processed in parallel.

Proteins were separated under reducing conditions on 4–15% PAGE(Bio-Rad Laboratories, Hercules, CA) with 0.1% SDS. Gels were stained with SYPRO Orange protein gel stain (Bio-Rad Laboratories) and photographed. The proteins were then transferred to nitrocellulose membrane (Schleicher and Schuell, Keene, NH) for immunoblotting. The blots were blocked for 1 h with NAP-SureBlocker (Geno Technology,Inc., St. Louis, MO) and then incubated with primary antibody (either 1:500 LF37 or 2.0 μg/ml ON-mAb) in 0.5% gelatin Tween-Tris-buffered saline solution. As a negative controls, identical blots were incubated with nonimmune rabbit serum (1:500) or mouse IgG1, κ-chain (2.0 μg/ml). The nitrocellulose membrane was then incubated with 0.5% gelatin Tween-Tris buffered saline containing 1.5 × 10⁶ cpm/ml ¹²⁵I-labeled rProtein A(NEN Life Science Products, Inc., Boston, MA) for 1 h at 37°C. Bands were visualized by exposing the immunoblot to Kodak XAR5 film in an X-ray cassette with an intensifying screen (Eastman Kodak,Rochester, NY). Images were acquired using a charged-coupled device camera (Alpha Innotech) and analyzed with PhotoEditor (Microsoft Inc.,Redmond, WA).

ON was identified using either LF37 or ON-mAb. IHC staining for PSA was performed using the polyclonal antibody α-PSA. All of the IHC stainings were performed according to the manufacturer’s protocol for the Ventana Nexus automated IHC system (Ventana Medical System). The system uses an indirect biotin-avidin diaminobenzidine detection system. Positive controls were sections of human skin (37, 38); negative controls included replacement of the primary antibody with equal volumes of nonimmune rabbit serum (for LF37) and with nonimmune rabbit IgG1 (for α-PSA) or equal volumes of mouse IgG1, κ-chain (for ON-mAb). Analysis of ON staining extent for epithelial layers of normal and tumor glands was performed on the entire section. For stroma, at least 10 high-powered areas were counted and averaged. Percentage staining was calculated based on the number of positive cells, irrespective of the staining intensities.

Expression of ON mRNA steady-state levels was studied in various prostate cell lines using semiquantitative RT-PCR. Ratios of ON:MIC signals from three independent replicates were used for quantitation. As shown in Fig. 1, all cell lines, with the exception of DU 145, showed expression of ON mRNA, although the intensity of expression ranged from low to high. Whereas the noninvasive N-PC3 cell line showed moderate expression, as did its parental cell line PC-3, the invasive subline B-PC3 showed high levels of this transcript. ON mRNA was also detected in LNCaP and its sublines, C4, C4-2, and C4-2b, with C4-2b giving the most intense band in this group. In addition, PZ-HPV-7, a cell line derived from normal prostate epithelial cells, shows high levels of ON mRNA. High levels of the transcript were also observed in VeCaPs, the cell line derived from a spine metastasis. These data demonstrate ON expression by eight of nine CaP cell lines, with the most intense expression associated with bone metastatic cell lines.

To clarify the specificity of the antibodies used in this study,Western blot analyses were done using rHevin. No cross-reaction was observed between LF37/ON-mAb and the ON-like protein rHevin (Fig. 2).

To investigate the expression and location of ON in tissues, ISH and IHC were performed on normal prostate and CaP specimens. Because CaP is histologically heterogeneous and often multifocal within the prostate (39, 40) we compared ON expression in normal areas as well in the neoplastic areas of the same prostate (matched pairs). The results of ON immunoreactivity are summarized in Table 1. In normal prostate, glandular epithelium shows low to moderate levels of ON immunoreactivity in both luminal and basal cells. In addition, cell-specific RT-PCR showed the presence of the ON transcript in both luminal cells and basal cells(Fig. 3). Immunoreactivity was observed in tumor cells from three CaP cases (3 of 10). Weak staining was also observed in fibromuscular stromal cells of both normal and tumor glands.

To assess the possible association between ON expression and tumor progression, we analyzed primary CaP with its specific lymph node metastatic site. The results of ON immunopositivity are summarized in Table 2. In general, low to moderate ON immunoreactivity is seen in the primary cancer cells. The corresponding lymph node metastatic samples showed a moderate to strong ON signal within the tumor mass.

We also analyzed CaP metastases from different tissues to assess the possible association of ON expression with tumor progression. Metastases that we analyzed included sites from the bone, liver, lymph,lungs, colon, and urinary bladder (Table 3). In metastatic foci, two histological patterns of tumor were observed: (a) sheets of tumor cells that did not form glands (denoted in Table 3 as “sheets”); and(b) aggregates of tumor cells that formed glandular structures (denoted in Table 3 as “glandular”). Serial sections of metastatic foci were analyzed for PSA and ON expression. ISH analysis showed high levels of ON mRNA in the cytoplasm of the tumor cells (Fig. 4). As shown in Fig. 5, PSA immunoreactivity was intense and homogenous in the tumor cells from different metastatic sites. Most metastatic foci showed elevated levels of ON expression, although areas of intense and weak immunostaining were frequently observed in the same foci. Generally, intense ON staining was observed in metastatic foci from bone (Fig. 5-2c), whereas a diffuse pattern was observed in soft tissue metastasis. In soft tissue metastasis,increased immunoreactivity was observed in tumor cells adjacent to normal tissue (arrow, Fig. 5-4c).

Invasion and metastasis involve multiple steps, which are not well characterized at the molecular and cellular levels. The initial steps involved in invasion and metastasis by a tumor cell include the breakdown of cell-cell junctions, dissociation of tumor cells, focal proteolysis of the ECM, and dissemination through neovascularization (24). Recent investigations have identified possible participants in the acquisition of metastatic potential by cancer cells. These include adhesive and/or antiadhesive molecules, proteases,and angiogenic factors. Alterations in the expression pattern of several genes involved in cell attachment have been studied in CaP. These include cadherins (41, 42, 43, 44, 45), integrins (46, 47, 48, 49, 50, 51, 52), laminin 5 (53, 54), hevin (55), and CD44 (56, 57, 58). In the current study, we examined the prototypic antiadhesive protein ON. We report here a detailed analysis of ON expression in malignant as well as normal prostate specimens. The role of ON in neoplastic growth and progression is speculative. It may include (a) modulation of cell-cell interaction, (b) remodeling of ECM via the induction of proteolytic enzymes, and (c) the enhancement of tumor-associated angiogenesis. Some of these biological activities may be triggered by the release of distinct domains of ON by prostate-associated serine proteases such as hK2.

At the light microscope level, alterations in cell-cell and cell-matrix adhesion in tumors are reflected in the degree of cohesiveness and the pattern of tumor growth. Loss of normal cell adhesion properties plays a key role in the progression of cancer. This alteration in tumor cell adhesion is important because detachment of tumor cells is an early step in the invasion of surrounding tissues and metastasis to secondary sites. In vitro studies have shown that introduction of exogenous ON to cultured cells inhibits cell spreading and induces cellular rounding, which results in the detachment of cells (59). Overexpression of ON in stably transfected F9 embryonal carcinoma cell lines results in aggregation and rounding (60). ON may alter these cellular morphologies and cell-cell or cell-matrix interactions by disrupting cell-substrate links and promoting the rearrangement of actin cytoskeletal elements (61). Down-regulation of ON by antisense ON suppresses and abolishes the tumorigenic potential of human melanoma cells (26). This down-regulation is accompanied by reduced adhesive and invasive capacities in vitro.

Pericellular degradation of the matrix component is an essential requirement for the motility of the tumor cells from their primary site. Studies have shown that ON, at physiological concentrations, can induce the expression of MMPs (62) including collagenase,stromelysin, and M_r 92,000 gelatinase,suggesting that ON may alter the nature of the ECM presented to the cell. Studies with synthetic peptides representing unique sequence motifs from the different domains of ON have shown that ON peptides can increase the levels of MMPs (63) and plasminogen activator inhibitor 1 (16, 64), as well as decrease ECM synthesis (65). This was further confirmed by a recent finding that there is a significant increase in MMP-2 activity in CaP cell lines exposed to ON (66). Thus, the secretion of this protein by tumor cells and/or surrounding stromal cells may result in the focal proteolysis of the ECM. This may facilitate local invasion and the subsequent distant spread of tumor cells.

Finally, the establishment and growth of tumors at secondary sites require angiogenesis. An “angiogenic switch” has been associated with active tumorigenesis (67). Several studies have implicated the role of ON in the modulation of angiogenesis (10, 17, 18, 27). Expression of ON is induced in endothelial cells during angiogenesis (65). The addition of exogenous ON to endothelial cells in culture results in a 4-fold decrease in thromospondin, an inhibitor of angiogenesis (64). In vivo studies using synthetic Cu²⁺-binding peptides designed from data derived from ON protease digestion stimulate angiogenesis (17). It is conceivable that ON cleavage by serine proteases of prostate cancer cells and the consequent release of functionally distinct fragments could account for the pleotropic actions of ON. Earlier studies documented that this type of unmasking of biological functions distinct from those observed in the native protein after specific cleavage occurs in laminin 5 (68),osteopontin (69), and fibronectin (70).

In summary, we document here the first evidence of ON expression in CaP cell lines and tissues. Our data further reveal an apparent up-regulation of ON mRNA and protein expression in the cell lines and tissues identified with bone metastasis. These observations are in full concurrence with the hypothesis that osseous metastatic CaP cells must be osteomimetic, which allows the cells to thrive in bone (71). Although we did not address the expression of ON in association with Gleason grade, such studies are necessary and are under way. Finally, one of the more interesting aspects under study is whether the proteolysis of ON by hK2 contributes to the unique biological responses seen in CaP bone metastases (e.g., the osteoblastic reaction).

We thank Lisha G. Brown for tissue culture and Kent R. Buhler for the excellent technical assistance in Western blot analyses. We also thank Lan L. Nguyen for initial PCR studies.