Paracrine interactions between cells have traditionally been studied in vitro by co-culturing cells in separate chambers while sharing the same culture medium. We have developed a non-conventional mixed culture system where different cell populations are cultured within the same chamber, thus, allowing them to establish physical associations that most likely occur in vivo and influence cell-cell signalling.

Shh-expressing LNCaP human prostate cancer cells obtained via stable transfection with hShh cDNA cloned into a pIRES2-EGFP vector, designated as LNShh cells, or vector-transfected control LNCaP cells were cultured within the same culture chamber with MC3T3 mouse pre-osteoblasts. The cells in this mixed culture system established, in a time-dependent manner, a distinctive morphologic pattern. There were no apparent differences in the spatial organization of the mixed cultures of MC3T3 cells and either control LNCaP or LNShh cells, and only images of mixed cultures of MC3T3 and LNShh cells are shown in Figure 1. Beginning about two weeks of mixed culture, clusters of LNShh cells, identified by their GFP fluorescence, were surrounded by a compact "stroma" of MC3T3 cells which exhibited more intense phalloidin staining of their F-actin filaments (Figure 1, panels a, b and c). Identification of the epithelial-derived LNShh cells in mixed cultures was further confirmed by their positive immunostaining for human cytokeratin 8 (Figure 1, panel d). When cultured alone, LNShh cells also formed colonies which became interconnected over time forming a meshwork instead of remaining discrete as shown in mixed cultures. On the other hand, single cultures of MC3T3 cells which reached confluence at about two weeks formed a sheet of cells instead of the reticulate network evident in mixed cultures (data not shown).

To demonstrate activation of Shh signalling in pre-osteoblasts by Shh-expressing prostate cancer cells in mixed cultures, the expression of known Shh target genes Gli1 and Ptc1 were determined by quantitative real time RT-PCR analysis using species specific primer sequences (Table 1) which amplified these genes in mouse MC3T3 cells but not in the human prostate cancer cells. Thus, amplification of Gli1 and Ptc1 in mixed cultures is highly, if not solely, attributable to their expression in MC3T3 mouse pre-osteoblasts. Conversely, using primers that amplified genes in human but not mouse cells, we determined that expression of human GLI1 did not differ significantly between control LNCaP and LNShh cells cultured with MC3T3 cells suggesting the absence of significant autocrine or feedback paracrine regulation of the hedgehog pathway, at least under conditions used in these studies (data not shown). We have previously used species-specific primers in RT-PCR analysis to demonstrate the upregulation of mouse Gli1 in mouse xenografts of human prostate carcinoma 9.

MC3T3 pre-osteoblasts in mixed cultures with LNShh cells expressed increased levels of Gli1 and Ptc1 compared to those cultured with control LNCaP cells (Figure 2A). In a traditional co-culture system where cells were grown in separate chambers but shared the same culture medium, MC3T3 cells co-cultured with LNShh cells also expressed higher levels of Gli1 and Ptc1 (Figure 2B). Serum-free conditioned media from single confluent cultures of LNShh cells also increased Gli1 and Ptc1 expression in MC3T3 cells cultured alone which is consistent with the soluble nature of the secreted Shh ligand (data not shown).

Cyclopamine, a specific chemical inhibitor of hedgehog signalling through functional inhibition of the Shh receptor complex protein Smo, dramatically abrogated the increase in both Gli1 and Ptc1 expression in MC3T3 cells in mixed culture with LNShh cells (Figure 2C). Activation of the hedgehog pathway in MC3T3 cells was a direct and immediate effect of Shh action since exogenous Shh peptide (Shh-N) increased Gli1 and Ptc1 message as early as 24 h following treatment (Figure 2D).

These data demonstrate specific and direct paracrine activation of Shh signalling in pre-osteoblasts by Shh-expressing prostate cancer cells.

The ability of LNShh cells to induce osteoblast differentiation was examined. The initial phase of osteoblast differentiation is characterized by cell proliferation followed by growth arrest 15. As shown in Figure 3A, proliferation of MC3T3 cells co-cultured with LNShh cells was significantly decreased by days 5 and 7 of co-culture compared to that of MC3T3 cells co-cultured with control LNCaP cells or cultured alone. These data suggest that Shh-expressing prostate cancer cells can inhibit cell proliferation in pre-osteoblasts.

Cell proliferation arrest is associated with a time-dependent increase in alkaline phosphatase (ALP) activity, a well-established indicator of osteoblast differentiation 15. ALP activity in mixed cultures of MC3T3 and LNShh cells increased with time and was significantly higher relative to that in mixed cultures of MC3T3 and control LNCaP cells (Figure 3B). Staining for ALP activity was performed to determine the spatial localization of activity in mixed cultures. Consistent with data shown in Figure 3B, staining for ALP activity was greater in mixed cultures of LNShh and MC3T3 cells and intensified progressively through time (Figure 3C). Staining for ALP activity was localized almost exclusively in MC3T3 cells which surrounded clusters of LNShh cells (Figure 3C). Single cultures of MC3T3 cells also exhibited ALP staining when grown to confluence (data not shown).

Correspondingly, expression of the alkaline phosphatase gene Akp2, an early osteoblast differentiation marker gene, was increased in MC3T3 cells in mixed culture with LNShh cells, and this effect was mimicked by direct treatment of MC3T3 cells with exogenous Shh-N (Figures 3D and 3E, respectively).

MC3T3 cells in mixed culture with either control LNCaP or LNShh cells in ascorbic acid (AA)-free α-MEM media for 21 days were negative for both Alizarin red and Von Kossa staining indicating the absence of significant calcium deposition in and mineralization of the extracellular matrix (ECM), respectively (data not shown). These results are consistent with the known requirement for AA and inorganic phosphates for ECM mineralization during late stage osteoblast differentiation.

To establish that LNShh cell-induced osteoblast differentiation occurred via the Shh signalling pathway, MC3T3 cells were stably transfected with human GLI1 cDNA which has a deletion in the region containing the

t

a

d

Both parental MC3T3 and the M-TAD cells expressed endogenous mouse Gli1 mRNA (Figure 4A, lanes 1 and 2). However, only M-TAD cells expressed the message for the dominant negative human GLI(-)TAD transgene (Figure 4A, lanes 3 and 4). Surprisingly, the phenotype of M-TAD cells appeared different from that of parental MC3T3 cells. When grown to confluence, MC3T3 cells changed from a spindle-shaped phenotype to a cuboidal morphology with generally round nuclei indicative of the differentiated state (Figure 4B, panel a). M-TAD cells, on the other hand, retained their fusiform shape with nuclei elongated along the long cell axis consistent with an immature osteoblast state (Figure 4B, panel b). As with MC3T3 cells, M-TAD cells in mixed culture with LNShh cells formed a stroma surrounding human prostate cancer cells although their stroma appeared to be more compact (Figure 4B, panels c and d).

Expression of the dominant negative GLI1 effectively blocked the activation of the Shh pathway and inhibited the upregulation of Gli1 and Ptc1 in M-TAD cells cultured with LNShh cells (Figure 4C). Consequently, M-TAD cells cultured with LNShh cells for 14 days exhibited markedly decreased Akp2 expression (data not shown) and ALP activity (Figure 4D, panels a, b and c). Hematoxylin staining of mixed cultures following staining for ALP activity showed confluent populations of both MC3T3 and M-TAD cells in mixed cultures with LNShh cells indicating that the decrease in ALP activity in M-TAD cells was not a function of differences in cell confluence in the co-cultures (Figure 4D, panels d, e and f).

The transcription factor Runx2 plays a pivotal role in osteoblast differentiation and function 1718. The effect of LNShh cells on Runx2 expression in MC3T3 cells in mixed cultures was determined. Endogenous levels of Runx2 were not significantly changed in MC3T3 cells cultured with LNShh cells (Figure 5A, b relative to a). In addition, the expression of Runx2 target genes osteocalcin (Ocn) and osteopontin (Opn) were not significantly increased (Figure 5A, c and d, respectively, relative to a).

Consistent with these data, activation of the Shh pathway in MC3T3 cells by exogenous Shh-N, as evidenced by increased expression in Gli1 and Ptc1, did not upregulate Runx2 expression (Figure 5B). On the other hand, exogenous AA significantly increased Runx2 levels in MC3T3 cells without activating the Shh pathway (Figure 5B). Combined stimulation with Shh-N and AA did not significantly increase the expression of Runx2 relative to treatment with AA alone, or the expression of Shh target genes Gli1 and Ptc1 relative to treatment with Shh-N alone indicating the absence of synergistic interaction between these two factors. AA upregulates the expression of osteoblast differentiation marker genes including Runx2 and promotes osteoblast differentiation 1920. These data, therefore, indicate that early phase osteoblast differentiation in pre-osteoblasts induced by paracrine Shh signalling might proceed in a mechanism that does not require the transcriptional regulation of Runx2.

LNCaP human prostate cancer cells (ATCC, Rockville, MD) have been previously stably transfected with a 1.44 kb human Shh cDNA cloned into a pIRES2-EGFP mammalian cell expression vector (designated as LNShh cells) or with pIRES2-EGFP vector alone 9. LNShh and control LNCaP cells were maintained at 37C, 5% CO₂ in complete culture medium consisting of RPMI-1640 supplemented with 10% fetal bovine serum (FBS), 100 U/ml penicillin and 100 μg/ml streptomycin (Gibco Invitrogen). Shh gene and protein expression were routinely determined by quantitative real time RT-PCR and western blot analysis, respectively, and GFP expression was monitored by fluorescence microscopy.

Cells from the mouse calvaria-derived non-transformed pre-osteoblast cell line MC3T3-E1 (subclone 4; ATCC, Rockville, MD: designated as MC3T3 cells), were transfected with pCMV-GLI(-)TAD: a human GLI1 cDNA lacking a

t

a

d

Control LNCaP or LNShh cells (5 × 10⁴) and MC3T3 cells (0.5 × 10⁴) were mixed in AA-free α-MEM complete culture medium and seeded per well of 6-well tissue culture plates. Cultures were maintained for the length of time specified in the experiments with media changes every 2–3 days.

Control LNCaP or LNShh cells (5 × 10⁴) were seeded per cell culture insert (transparent PET membrane, 0.4 μm pore size: BD Biosciences) in RPMI-1640 complete culture medium. MC3T3 cells (1 × 10⁴) were seeded per well of 6-well tissue culture plates in AA-free α-MEM complete culture medium. After two days, culture inserts grown with prostate cancer cells were transferred to wells grown with MC3T3 cells, and co-cultures were maintained in AA-free α-MEM complete culture medium for the length of time specified in the experiments with media changes every 2–3 days.

Total RNA was extracted using Trizol (Invitrogen), purified using the RNeasy Mini Kit (Qiagen) and subjected to DNase treatment with RQ1 RNase-free DNase (Promega) to remove contaminating genomic DNA. The TaqMan^® Gold PCR Core Reagent Kit along with MuLV Reverse Transcriptase and RNase Inhibitor (Applied Biosystems) were used for cDNA synthesis. PCR primers (Invitrogen) and FAM-QSY7 probes (MegaBases, Inc.) for genes of interest and the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (Gapdh), whose sequences are shown in Table 1, were designed using the Primer Express 3.0 software program. mRNA expression was measured in duplicate or triplicate per sample using 40 cycles of amplification in the 7500 Fast Real-Time PCR System (Applied Biosystems). Reactions were routinely performed without Reverse Transcriptase to demonstrate RNA dependence of the reaction products. Results were analyzed using the comparative C_t method. C_t is the threshold cycle at which the reporter fluorescence signal for each amplicon passes above a fixed baseline. The average C_t of the gene of interest was normalized to that of the housekeeping gene Gapdh to obtain the ΔC_t for that gene. ΔC_t values between experimental and control samples were then compared (ΔΔC_t) to reflect the relative fold change in gene expression.

Cells were maintained as separate co-cultures as described above. MC3T3 cells were seeded onto 24-well tissue culture plates at 0.2 × 10⁴ cells per well. Control LNCaP or LNShh cells were seeded at 2 × 10⁴ cells per cell culture insert. Following overnight incubation, culture inserts were transferred to wells and co-cultures were maintained in AA-free α-MEM complete culture medium with media changes every 2–3 days. At specified times, cell inserts were removed and proliferation of MC3T3 cells grown on wells was determined using the Cell Counting Kit-8 (Dojindo Laboratories, Japan) which is based on the formation of a water-soluble formazan dye through the activity of dehydrogenases in living cells.

Cells were maintained as mixed cultures in Lab-TekII CC2-treated chamber slides (Nunc) for the length of time specified in the experiments with media changes every 2–3 days. Cells were then fixed in 10% neutral buffered formalin for 10 minutes and permeated with 0.1% Triton X-100. For fluorescence immunocytochemistry, cells were stained with Alexa Flour 635 Phalloidin (Invitrogen: 1:40 dilution) and incubated in the dark for 20 minutes. Cells were counter-stained with 300 nM DAPI (Invitrogen). Slides were mounted using ProLong Gold antifade reagent (Invitrogen). For cytokeratin expression, fixed cells were stained with mouse monoclonal anti-human cytokeratin 8 antibody (DAKO: 1:50 dilution), developed with diaminobenzidine, and counterstained with hematoxylin. Slides were viewed in a Leica DMR-HC Upright Microscope and images were captured with imaging software (Improvision Openlab). Fluorescent images were captured at the following excitation/emission wavelengths (nm): Phalloidin-560/645; DAPI-360/470; GFP-480/527.

Quantitative determination of ALP activity was done using the p-Nitrophenyl Phosphate (pNPP) Liquid Substrate System according to manufacturer's protocol (Sigma Aldrich). Briefly, cells in culture wells were treated with 0.2% Triton X-100, harvested with a cell scraper, and subjected to freeze-thaw cycles. Lysates were centrifuged and supernatants (10 μg protein) were incubated with 150 μl pNPP for 5 h at room temperature in the dark. Absorbance at 405 nm was measured using a microplate reader, and ALP activity was calculated according to manufacturer's instructions. Protein determination was done using the Bio-Rad DC Protein Microplate Assay according to manufacturer's protocol.

For ALP staining, mixed cultures were fixed with 10% neutral buffered formalin for 10 minutes and incubated with alkaline phosphatase substrate solution for at least 30 minutes at room temperature in the dark. The substrate solution was prepared by adding Naphthol AS-MX Phosphate Alkaline Solution to a diazonium salt solution (Fast Violet B Salt capsule dissolved in distilled water) just before use (Sigma-Aldrich).

MC3T3 cells were seeded onto 6-well culture plates at 1 × 10⁵ cells per well in AA-free α-MEM complete culture medium. Following overnight incubation, cells were treated with 1 μg/ml Shh-N, a modified active N-terminal peptide of human Shh (kindly provided by Curis Inc., Cambridge, MA), in serum-free AA-free α-MEM culture media for 24 h or 72 h.

Cultures of mixed cells (as described above) were maintained for 6 days in serum-free AA-free α-MEM culture medium with 1 or 10 μM cyclopamine (vehicle controls were 0.01 or 0.1% ethanol, respectively). Cyclopamine, a generous gift from Dr. W. Gaffield (Western Regional Research Center, USDA, Albany, CA), has been previously used to inhibit Shh signalling in developing mouse prostate 43.

Cultures of mixed cells (as described above) were maintained for 7 days in AA-free α-MEM complete culture medium without or with 50 μg/ml L-ascorbic acid (AA; Sigma). To determine the direct effect of AA, MC3T3 cells were seeded alone onto 6-well culture plates at 1 × 10⁵ cells per well in AA-free α-MEM complete culture medium. Following overnight incubation, cells were treated without or with 50 μg/ml AA for 24 h.

Data were analyzed by ANOVA and pairwise multiple comparisons were done using the Bonferroni t-test at P < 0.05. Data are presented as means ± SD of at least 2 assays from independent experiments, each assay done in duplicate or triplicate.