This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by O'Connor, J. C. Articles by Carson, D. D. Search for Related Content PubMed PubMed Citation Articles by O'Connor, J. C. Articles by Carson, D. D.

---

Angiogenesis, Metastasis, and the Cellular Microenvironment

Coculture with Prostate Cancer Cells Alters Endoglin Expression and Attenuates Transforming Growth Factor-ß Signaling in Reactive Bone Marrow Stromal Cells

¹ Department of Biological Sciences, University of Delaware; ² Medical Oncology Hematology Consultants, PA, Helen F. Graham Cancer Center, Christiana Care Health Services; ³ Center for Translational Cancer Research, Newark, Delaware; and ⁴ Department of Medicine, Jefferson Medical College, Philadelphia, Pennsylvania

Requests for reprints: Daniel D. Carson, Department of Biological Sciences, University of Delaware, Newark, DE 19716. Phone: 302-831-6977; Fax: 302-831-1033. E-mail: dcarson{at}udel.edu

	Abstract

Top Notes Abstract Introduction Results Discussion Materials and Methods Acknowledgements References

 
A dynamic interplay between prostate cancer cells and reactive bone stroma modulates growth of metastases within bone. We used microarray analysis to screen for changes in gene expression in bone marrow stromal cells cocultured with prostate cancer cells and found reduced expression of endoglin, a transmembrane glycoprotein that functions as an auxiliary coreceptor for members of the transforming growth factor ß (TGF-ß) family of cytokines. The downstream TGF-ß/bone morphogenetic protein signaling pathway including Smad1 and Smad2/3 also was attenuated, as was Smad-dependent gene transcription. Smad1/5/8-dependent inhibitor of DNA binding 1 expression and Smad2/3-dependent plasminogen activator inhibitor I expression both were decreased and were accompanied by decreased cell proliferation. Small interfering RNA–mediated knockdown of endoglin in HS-5 cells verified that the effects on signaling were a direct result of the attenuation of endoglin. These data illustrate that endoglin acts as a positive regulator of both activin receptor–like kinase 1–induced Smad1/5/8 activation and activin receptor–like kinase 5–induced Smad2/3 activation in bone marrow stromal cells. In addition, the data illustrate that one early event of metastasis upon the arrival of prostate cancer cells into the bone stroma is attenuated endoglin expression in the stromal cells, which subsequently alters Smad signaling and cell proliferation. We hypothesize that coculture of bone marrow stromal cells with prostate cancer cells alters TGF-ß signaling in the stromal cells, ultimately facilitating growth of the cancer cells in the bone compartment. Collectively, these studies suggest that prostate cancer cells modulate TGF-ß responsiveness of bone marrow stroma as one means of facilitating their own growth in bone. (Mol Cancer Res 2007;5(6):585–603)

	Introduction

Top Notes Abstract Introduction Results Discussion Materials and Methods Acknowledgements References

 
Prostate cancer is the most common cancer and is the third leading cause of cancer-related deaths among American males (1). It is characterized by a very predictive progression from an early hormone-sensitive form that can be treated by androgen deprivation therapy to a hormone-refractory form that occurs within the axial skeleton (2-4). Once it becomes hormone refractory, prostate cancer is essentially incurable (5, 6) and, ultimately, the secondary metastases (e.g., to bone) and their sequelae are lethal (2, 3).

Whereas the mechanism behind the preferential metastasis and growth of certain cancers (e.g., breast and prostate) in bone is poorly understood, it is generally accepted that cancer is not simply a disease involving a single foreign tumor cell that grows out of control, but is a dynamic exchange between the tumor cells and the surrounding host cells that disrupts normal homeostasis (7, 8). Factors present within the bone microenvironment, whether insoluble components of the extracellular matrix or soluble factors (cytokines and mitogens) secreted by host cells, and direct cell-to-cell interactions are believed to contribute to the preferential metastasis and growth of prostate cancer cells in bone (4, 8-12).

It is clear that bone stromal cells affect prostate cancer cells and are themselves altered by their interactions with the cancer cells (4, 7, 8). In a series of experiments, Chung and coworkers (13-25) showed that growth of both benign and malignant prostate tumors in vivo is dependent on the cells of the host environment (i.e., stromal cells). For example, when bone marrow stromal cells are coinoculated with the poorly tumorigenic prostate cancer cell line LNCaP, the prostate cancer cells progress to a more tumorigenic phenotype (20, 21, 24, 25). The enhanced tumorigenicity is the result of changes in genotype and phenotype of both the prostate cancer cells and the host stromal cells (19, 22, 23, 26, 27). Interestingly, extracted stromal cells that have been in contact with tumor cells retain the ability to transform nontumorigenic cells, indicating that the phenotypic and genotypic alterations of the stromal cells are permanent (28). Taken together, these results show that many components of the bone microenvironment contribute to the growth and proliferation of prostate cancer metastases within bone.

We used pathway-specific microarray analysis to characterize changes in gene expression in HS-27a bone marrow stromal cells in response to coculture with C4-2B prostate cancer cells, a human prostate cancer cell line that readily metastasizes to bone (24, 29). The coculture experiment was designed to replicate early events that occur when prostate cancer cells first arrive in the bone microenvironment and to provide an opportunity to examine gene changes in the bone stroma that are induced by the prostate cancer cells. We identified 13 genes that were differentially expressed between HS-27a cells that were cultured alone and HS-27a cells that were cocultured for 3 days with C4-2B prostate cancer cells. One of the altered genes was endoglin.

Endoglin is a homodimeric transmembrane glycoprotein that binds with high affinity to transforming growth factor ß1 (TGF-ß1) and TGF-ß3, as well as activin A and several bone morphogenic proteins (BMP), after association with one of two transmembrane serine-threonine kinases known as TGF-ß receptor I and TGF-ß receptor II (30). Although it is constitutively phosphorylated, endoglin is not an active signaling molecule itself but functions as a regulatory component of the TGF-ß receptor heteromeric complex whereby it modulates signaling of distinct TGF-ß receptor I isotypes known as activin receptor–like kinase (ALK)-1 and ALK5 (31, 32). Formation of the activated heteromeric complex initiates an intracellular signaling cascade involving activation of specific Smad (small mothers against decapentaplegic) proteins that transduce signals from the receptor complex into the nucleus where they regulate transcription of a variety of genes involved in maintaining normal physiologic processes such as cell proliferation, apoptosis, cell adhesion, and motility (33). It is the role that endoglin plays in regulating signaling of the TGF-ß superfamily and the possible implications that it may have on the ability of prostate cancer cells to survive and proliferate in the bone microenvironment, and ultimately give rise to osteoblastic/osteosclerotic lesions, that led us to focus on the alterations in endoglin. Our results indicate that prostate cancer cells secrete a soluble factor that attenuates endoglin expression in the bone marrow stromal cells and alters TGF-ß signaling to decrease cell proliferation in the bone stroma, creating a more permissive microenvironment for growth of bony metastases.

	Results

Top Notes Abstract Introduction Results Discussion Materials and Methods Acknowledgements References

 
Coculture of HS-5/HS-27a Cells with C4-2 and C4-2B Cells Suppresses Endoglin mRNA and Protein Expression in HS-5 and HS-27a Cells
Microarray analysis was used to characterize changes in gene expression in HS-27a bone marrow stromal cells in response to coculture conditions with prostate cancer cells. Using a 2-fold change as the criterion for array analysis, we identified 13 genes that were differentially expressed between HS-27a cells that were cultured alone and HS-27a cells that were cocultured with C4-2B prostate cancer cells (Table 1 ). Of the 13 genes that were differentially expressed, 10 genes were down-regulated and 3 genes were up-regulated. We chose to validate changes in endoglin gene expression due to its role in regulating signaling of the TGF-ß superfamily and the possible implications that it may have on the ability of prostate cancer cells to survive and proliferate in the bone microenvironment, and ultimately give rise to osteoblastic/osteosclerotic lesions; however, during the course of our studies, we validated several other changes observed in the microarray (see below). In addition, we also verified the decrease in endoglin mRNA levels using conventional PCR (data not shown).

The soluble factor(s) released from DU-145, C4-2, and C4-2B cells that induced the decreased endoglin expression is heat stable. Boiling of the conditioned media before application to the bone marrow stromal cell cultures did not reduce the ability of the conditioned media to attenuate endoglin expression (data not shown). Identification of the active factor(s) that is secreted by the prostate cancer cells is currently under way.

Exposure to Conditioned Media from DU-145, C4-2, and C4-2B Cells Suppresses Endoglin Protein Expression in HS-5 and HS-27a Cells
Conditioned media from several cell lines were evaluated for the ability to alter cellular endoglin protein levels in HS-5 and HS-27a cells, both cell lines being surrogates for normal bone marrow stromal cells. Exposure of HS-5 (Fig. 1A and C ) and HS-27a (Fig. 1B and D) cells to conditioned media from DU-145, C4-2, and C4-2B cells significantly decreased endoglin protein levels to 25% of control levels, results that confirmed the decreases in endoglin mRNA levels observed in the microarray experiment. The LNCaP cell line, the nonmetastatic parental line of the C4-2 and C4-2B cell lines, did not alter endoglin levels, suggesting that during the progression of the LNCaP cells to the more metastatic phenotype of the LNCaP sublines, there is also a concurrent change in the ability of the LNCaP sublines to attenuate endoglin protein levels. The breast carcinoma cell line ZR-75-1 did not alter endoglin protein levels in HS-5 cells but decreased endoglin in HS-27a cells. The DU-145 cell line, an osteolytic cell line derived from a prostate cancer brain metastasis, decreased endoglin levels to a similar magnitude as the C4-2 and C4-2B cells. However, the osteolytic prostate cancer cell line PC-3 did not alter endoglin protein levels, illustrating that not all osteolytic prostate cancer cell lines have the ability to decrease endoglin protein expression. Foreskin fibroblasts, used as a nonepithelial cell line control, did not alter endoglin levels in either HS-5 or HS-27a cells. Overall, DU-145, C4-2, and C4-2B cells showed a similar capacity to attenuate endoglin protein levels in both HS-5 and HS-27a cells, suggesting that all three cell lines may secrete a common soluble factor that causes the effect.

View larger version (32K): [in this window] [in a new window] [Download PPT slide]   FIGURE 1. Attenuation of cellular endoglin levels after conditioned medium exposure. After 24 h of serum deprivation in DMEM/1% FBS, confluent cultures of HS-5 (A and C) and HS-27a (B and D) bone marrow stromal cells plated into six-well culture plates were exposed to DMEM/1% FBS (control) or conditioned media from foreskin fibroblasts (For. Fib.), PC-3, DU-145, ZR-75-1, LNCaP, C4-2, or C4-2B cells for 3 d. Total cell protein was extracted and 10 µg of protein per lane were analyzed by SDS-PAGE and Western blotting for endoglin (95-kDa band) as described in Materials and Methods. ß-Actin (42-kDa band) served as a loading and transfer control and all data were normalized to ß-actin in (C) and (D). Columns, mean ratio of endoglin signal to ß-actin signal from duplicate determinations from a representative experiment; bars, SD. *, P 0.05; #, P 0.001.

View larger version (31K): [in this window] [in a new window] [Download PPT slide]   FIGURE 2. Dose- and time-dependent attenuation of cellular endoglin levels after conditioned medium exposure. After 24 h of serum deprivation in DMEM/1% FBS, confluent cultures of HS-5 bone marrow stromal cells plated into six-well culture plates were exposed to DMEM/1% FBS (0%; control) or increasing concentrations of conditioned medium from C4-2B cells for 3 d (A and C) or exposed to 100% C4-2B conditioned medium for 0, 1, 2, or 3 d (B and D). Total cell protein was extracted and 10 µg of protein per lane were analyzed by SDS-PAGE and Western blotting for endoglin (95-kDa band) as described in Materials and Methods. ß-Actin (42-kDa band) served as a loading control and all data were normalized to ß-actin in (C) and (D). Columns, mean ratio of endoglin signal to ß-actin signal from duplicate determinations from a representative experiment; bars, SD. *, P 0.05; #, P 0.001.

View larger version (9K): [in this window] [in a new window] [Download PPT slide]   FIGURE 3. TGF-ß secretion from HS-5 and HS-27a cells after conditioned medium exposure. After 24 h of serum deprivation in DMEM/1% FBS, confluent cultures of HS-5 (A) and HS-27a (B) bone marrow stromal cell lines were exposed to DMEM/1% FBS (control) or conditioned media from foreskin fibroblasts, PC-3, DU-145, ZR-75-1, LNCaP, C4-2, or C4-2B cells for 3 d. On test day 3, conditioned media (1 mL) were collected from all wells for analysis of secreted TGF-ß concentration by ELISA (Materials and Methods). The conditioned medium feeder cultures (foreskin fibroblast, PC-3, DU-145, ZR-75-1, LNCaP, C4-2, and C4-2B) were analyzed for secreted TGF-ß concentration to correct for the TGF-ß secreted from the feeder cultures when calculating TGF-ß secreted from the test cell lines. The adherent cells in each well were lysed and the protein concentration was measured for normalization of secreted TGF-ß levels to total cell protein per well. Columns, mean of duplicate determinations; bars, SD. *, P 0.05; #, P 0.001.

View larger version (35K): [in this window] [in a new window] [Download PPT slide]   FIGURE 4. Apoptosis and cell proliferation of HS-5 and HS-27a cells after conditioned medium exposure. After 24 h of serum deprivation in DMEM/1% FBS, confluent cultures of HS-5 or HS-27a bone marrow stromal cell lines were exposed to DMEM/1% FBS (control) or conditioned media from foreskin fibroblasts, PC-3, DU-145, ZR-75-1, LNCaP, C4-2, C4-2B, or PrEC cells for 3 d. For apoptosis evaluation, HS-5 (A) and HS-27a (B) cells were lysed directly in the wells and apoptosis was evaluated by measuring histone-associated DNA fragments (Materials and Methods). For cell proliferation analysis, growing cultures of HS-5 (C) and HS-27a (D) cells were labeled with 100 µmol/L 5-bromo-2-deoxyuridine for 4 h. After labeling, cell proliferation was measured using a commercially available assay kit as described in Materials and Methods. Three additional wells of HS-5 and HS-27a cells were plated and treated concurrently for quantification of total cell protein per well for normalization of apoptosis and cell proliferation data. For cell cycle analysis, growing cultures of HS-5 cells were cultured as above, and confluent cultures of HS-5 cells were exposed to DMEM/1% FBS (control; E and G) or conditioned medium from C4-2B cells (F and G) for 3 d. Cells were trypsinized and pelleted, fixed for 30 min in 70% ethanol, and stained with propidium iodine as described in Materials and Methods. Twenty thousand cells were evaluated by fluorescence-activated cell sorting. Columns, mean of duplicate (fluorescence-activated cell sorting), triplicate (apoptosis), or sextuplicate (cell proliferation) determinations from a representative experiment; bars, SD. *, P 0.05; #, P 0.001.

With the decreased cell proliferation and decreased inhibitor of DNA-binding 1 (Id-1) protein expression, we evaluated whether the bone marrow stromal cells were undergoing differentiation into osteoblast-like cells (data not shown). Under the correct conditions, bone marrow stromal cells can be induced to differentiate into osteoblast-like cells by treatment with dexamethasone or BMP-2 in vitro. The earliest marker of this differentiation is expression of alkaline phosphatase. Under normal growth conditions [DMEM/10% fetal bovine serum (FBS)], HS-27a cells showed some alkaline phosphatase staining after 12 days of culture at postconfluence, suggesting that they possess some intrinsic potential to differentiate along an osteoblast-like pathway. Consistent with this observation, HS-27a cells that were exposed to dexamethasone show increased alkaline phosphatase staining when compared with the untreated controls. In contrast, BMP-2 treatment actually decreased alkaline phosphatase staining in HS-27a cells compared with the untreated controls, suggesting that these transformed bone marrow stromal cells do not possess the normal physiologic potential for osteoblastic differentiation. HS-5 cells did not show alkaline phosphatase staining after treatment with either dexamethasone or BMP-2. When HS-27a cells were exposed to C4-2B conditioned medium, alkaline phosphatase staining was decreased in the untreated controls as well as the cultures treated with dexamethasone or BMP-2.

Endoglin Attenuation in HS-5 and HS-27a Cells Alters Smad Signaling
Due to the role of endoglin in TGF-ß signaling, we looked at cellular targets in the TGF-ß/BMP signaling pathway that are downstream of endoglin to determine if attenuated endoglin expression altered TGF-ß/BMP signaling in the bone marrow stromal cells. Endoglin interacts with both ALK1 and ALK5, subsequently leading to phosphorylation of Smad1/5/8 and Smad2/3, respectively. Both ALK1 and ALK5 mRNA were detected in bone marrow stromal cells by reverse transcription-PCR (data not shown).

Smad1 was decreased in HS-5 cells (Fig. 5A and C ) after exposure to conditioned media from both C4-2 and C4-2B cells. In HS-27a cells (Fig. 5B and D), Smad1 was decreased only after exposure to conditioned medium from C4-2B cells; in contrast, Smad1 was increased after exposure to conditioned medium from foreskin fibroblasts. Smad2/3 also was decreased in both HS-5 (Fig. 5A and E) and HS-27a cells (Fig. 5B and F) after exposure to conditioned medium from C4-2B cells, and also was decreased in HS-5 cells after exposure to conditioned media from DU-145 and C4-2 cells. These results illustrate that attenuated endoglin protein expression results in down-regulation of both Smad1 and Smad2/3 in HS-5 and HS-27a cells, although the effects were less pronounced in the cells exposed to conditioned media from DU-145 and C4-2 cells.

View larger version (45K): [in this window] [in a new window] [Download PPT slide]   FIGURE 5. Western blot analysis of Smad proteins. After 24 h of serum deprivation in DMEM/1% FBS, confluent cultures of HS-5 (A, C, and E) or HS-27a (B, D, and F) bone marrow stromal cells plated into six-well culture plates were exposed to DMEM/1% FBS (control) or conditioned media from foreskin fibroblasts, PC-3, DU-145, ZR-75-1, LNCaP, C4-2, or C4-2B cells for 3 d. Total cell protein was extracted and 10 µg of protein per lane were analyzed by SDS-PAGE and Western blotting for total Smad1 (60-kDa band; C and E) or total Smad2/3 (58-kDa band; D and F) as described in Materials and Methods. ß-Actin (42-kDa band) served as a loading and transfer control and all data were normalized to ß-actin in (C) to (F). Columns, mean ratio of Smad signal to ß-actin signal from duplicate determinations from a representative experiment; bars, SD. *, P 0.05; #, P 0.001.

View larger version (26K): [in this window] [in a new window] [Download PPT slide]   FIGURE 6. Attenuation of Smad1/5/8 phosphorylation in HS-5 cells after conditioned medium exposure. After 24 h of serum deprivation in DMEM/1% FBS, confluent cultures of HS-5 bone marrow stromal cells plated into six-well culture plates were exposed to DMEM/1% FBS (control; A) or conditioned medium from C4-2B cells (B) for 3 d. Cells were then stimulated with 500 ng/mL recombinant human BMP-2 for the indicated times from 5 to 90 min. Total cell protein was extracted and 25 µg of protein per lane were analyzed by SDS-PAGE and Western blotting for pSmad1/5/8 (60-kDa band) as described in Materials and Methods. ß-Actin (42-kDa band) served as a loading and transfer control. pSmad1/5/8 was normalized to either ß-actin (C) or total Smad1 (D) levels. *, P 0.05.

View larger version (33K): [in this window] [in a new window] [Download PPT slide]   FIGURE 7. Western blot analysis of cellular PAI-I after conditioned medium exposure. After 24 h of serum deprivation in DMEM/1% FBS, confluent cultures of HS-5 (A and C) or HS-27a (B and D) bone marrow stromal cells plated into six-well culture plates were exposed to DMEM/1% FBS (control) or conditioned media from foreskin fibroblasts, PC-3, DU-145, ZR-75-1, LNCaP, C4-2, or C4-2B cells for 3 d. Total cell protein was extracted and 10 µg of protein per lane were analyzed by SDS-PAGE and Western blotting for PAI-I as described in Materials and Methods. ß-Actin (42-kDa bands) served as a loading and transfer control and all data were normalized to ß-actin in (C) and (D). Columns, mean ratio of PAI-I signal to ß-actin signal from duplicate determinations from a representative experiment; bars, SD. *, P 0.05; #, P 0.001.

View larger version (31K): [in this window] [in a new window] [Download PPT slide]   FIGURE 8. Western blot analysis of Id-1 after conditioned medium exposure. After 24 h of serum deprivation in DMEM/1% FBS, confluent cultures of HS-5 (A and C) or HS-27a (B and D) bone marrow stromal cells plated into six-well culture plates were exposed to DMEM/1% FBS (control) or conditioned media from foreskin fibroblasts, PC-3, DU-145, ZR-75-1, LNCaP, C4-2, or C4-2B cells for 3 d. Total cell protein was extracted and 25 µg of protein per lane were analyzed by SDS-PAGE and Western blotting for Id-1 (15-kDa band) as described in Materials and Methods. Twenty-five micrograms of protein per lane were analyzed. ß-Actin (42-kDa band) served as a loading and transfer control and all data were normalized to ß-actin content in (C) and (D). Columns, mean ratio of Id-1 signal to ß-actin signal from duplicate determinations from a representative experiment; bars, SD. *, P 0.05; #, P 0.001.

View larger version (35K): [in this window] [in a new window] [Download PPT slide]   FIGURE 9. Conditioned medium effects on protein expression in primary human bone marrow stromal cells. After 24 h of serum deprivation in DMEM/1% FBS, confluent cultures of primary human bone marrow stromal cells plated into six-well culture plates were exposed to DMEM/1% FBS (control) or conditioned media from foreskin fibroblasts, PC-3, DU-145, LNCaP, C4-2, C4-2B, or PrEC cells for 3 d. Total cell protein was extracted and analyzed by SDS-PAGE and Western blotting for endoglin (B), total Smad1 (C), total Smad2/3 (D), PAI-I (E), or Id-1 (F) as described in Materials and Methods. Ten micrograms of protein per lane were analyzed for endoglin (95-kDa band), Smad1 (60-kDa band), Smad2/3 (58-kDa band), and PAI-I (44-kDa bands); 25 µg of protein per lane were analyzed for Id-1 (15-kDa band). ß-Actin (42-kDa band) served as a loading and transfer control and all data were normalized to ß-actin in (B) to (F). Columns, mean from duplicate determinations from a representative experiment; bars, SD. *, P 0.05.

PAI-I (Fig. 9E) and Id-1 (Fig. 9F) protein levels were both also decreased in primary bone marrow stromal cells exposed to conditioned media from C4-2 and C4-2B cells. DU-145 cell conditioned medium and LNCaP conditioned medium caused an increase and a decrease in PAI-I concentrations, respectively, in the primary bone marrow stromal cells. Consistent with the increased endoglin protein levels observed in the primary bone marrow stromal cells after exposure to conditioned medium from PrEC cells, both PAI-I and Id-1 protein levels were increased. Thus, increased endoglin protein levels were accompanied by increases in downstream gene products from both the ALK1 (Id-1 via pSmad1/5/8) and ALK5 (PAI-I via pSmad2/3) signaling pathways.

Increased Endoglin Increases Smad Signaling in HS-5 and HS-27a Cells
The increased Smad signaling in primary bone marrow stromal cells exposed to conditioned medium from PrEC cells prompted us to evaluate the effects of PrEC conditioned medium on endoglin protein levels and Smad signaling in HS-5 and HS-27a cells (Fig. 10A-F ). Similar to the observation in the primary bone marrow stromal cells, exposure to conditioned medium from PrEC cells increased endoglin expression in HS-5 cells; however, endoglin protein levels in HS-27a cells were unaffected (Fig. 10A and B). Both Smad1 (Fig. 10A and C) and Smad2/3 (Fig. 10A and D) protein levels were increased in HS-5 cells exposed to PrEC conditioned medium, although the increase in Smad1 was not statistically significant. Smad protein levels in HS-27a cells were not altered. Consistent with the effect observed in primary bone marrow stromal cells, PAI-I was increased in HS-5 as well as HS-27a cells (Fig. 10A and E). Id-1 protein levels were not altered in either HS-5 or HS-27a cells after PrEC conditioned medium exposure (Fig. 10F). Overall, these data suggest that the HS-5 cell line is more sensitive to changes in endoglin expression than HS-27a cells. In addition, these data are consistent with increased endoglin expression causing increased Smad signaling.

View larger version (32K): [in this window] [in a new window] [Download PPT slide]   FIGURE 10. Effect of PrEC conditioned medium on protein expression in HS-5 and HS-27a bone marrow stromal cells. After 24 h of serum deprivation in DMEM/1% FBS, confluent cultures of HS-5 or HS-27a bone marrow stromal cells plated into six-well culture plates were exposed to DMEM/1% FBS (control) or conditioned medium from PrEC cells for 3 d. Total cell protein was extracted and analyzed by SDS-PAGE and Western blotting for endoglin (B), total Smad1 (C), total Smad2/3 (D), PAI-I (E), or Id-1 (F) as described in Materials and Methods. Ten micrograms of protein per lane were analyzed for endoglin (95-kDa band), Smad1 (60-kDa band), Smad2/3 (58-kDa band), and PAI-I (44-kDa bands); 25 µg of protein per lane were analyzed for Id-1 (15-kDa band). ß-Actin (42-kDa band) served as a loading and transfer control and all data were normalized to ß-actin in (B) to (F). Columns, mean from duplicate determinations from a representative experiment; bars, SD. *, P 0.05; #, P 0.001.

View larger version (32K): [in this window] [in a new window] [Download PPT slide]   FIGURE 11. siRNA-mediated knockdown of endoglin alters Smad signaling via ALK1 and ALK5. After electroporation with dsRNA specific for human endoglin or a scrambled sequence (control), HS-5 bone marrow stromal cells were plated into six-well culture plates and cultured in DMEM/10% FBS (control) for 3 d for evaluation of total cellular protein levels by Western blotting (A-F) or plated into 96-well culture plates and cultured in DMEM/10% FBS (control) for 3 d for evaluation of apoptosis (G) and cell proliferation (H). Total cell protein was extracted and analyzed by SDS-PAGE and Western blotting (A) for endoglin (B), total Smad1 (C), total Smad2/3 (D), PAI-I (E), or Id-1 (F) as described in Materials and Methods. Ten micrograms of protein per lane were analyzed for endoglin (95-kDa band), Smad1 (60-kDa band), Smad2/3 (58-kDa band), and PAI-I (44-kDa bands); 25 µg of protein per lane were analyzed for Id-1 (15-kDa band). ß-Actin (42-kDa band) served as a loading and transfer control and all data were normalized to ß-actin in (B) to (F). HS-5 (G) cells were lysed directly in the wells, and apoptosis was evaluated by measuring histone-associated DNA fragments using a commercially available apoptosis assay as described in Materials and Methods. For cell proliferation analysis, growing cultures of HS-5 (H) cells were labeled with 100 µmol/L 5-bromo-2-deoxyuridine for 4 h. After labeling, cell proliferation was measured using a commercially available assay kit as described in Materials and Methods. Three additional wells of HS-5 cells were plated and treated concurrently for quantification of total cell protein per well for normalization of apoptosis and cell proliferation data. Columns, mean of triplicate (apoptosis) or sextuplicate (cell proliferation) determinations from a representative experiment; bars, SD. *, P 0.05; #, P 0.001.

	Discussion

Top Notes Abstract Introduction Results Discussion Materials and Methods Acknowledgements References

We used microarray analysis to characterize changes in gene expression in HS-27a bone marrow stromal cells in response to coculture conditions with C4-2B prostate cancer cells, a human prostate cancer cell line with a predilection for bone metastasis (24, 29). The coculture experiments were designed to replicate early events that occur when prostate cancer cells first arrive in the bone microenvironment and to provide an opportunity to examine gene changes in the bone stroma that are induced by the prostate cancer cells. TGF-ß is the archetypical member of a superfamily of secreted cytokines that includes the TGF-ßs, activins, BMPs, and Müllerian-inhibiting substance, which are involved in a number of physiologic processes including regulation of cell growth/proliferation, apoptosis, cell adhesion, and motility (32, 33). Within the bone microenvironment, in which TGF-ß is in high abundance, TGF-ß family members regulate bone mass and architecture, as well as the mechanical properties and composition of the bone matrix (42). Dysregulation of TGF-ß signaling has been implicated in numerous bone diseases, vascular disorders, and in many advanced-stage cancers such as metastatic prostate cancer (43-45). Our microarray data prompted us to focus on the alterations in TGF-ß/BMP signaling that occur as a result of the decrease in endoglin in the bone stroma after coculture with prostate cancer cells and the implications this may have on the establishment and growth of the invading prostate cancer cells. It also provided an opportunity to evaluate the role of endoglin in TGF-ß/BMP signaling in bone marrow stromal cells, in contrast to previous research focusing on the role of endoglin in endothelial cells.

Our data showed that soluble factors released by prostate cancer cells have effects on cells within the bone microenvironment. Because the factor(s) was not inactivated by boiling the conditioned medium, it is likely to be either a small peptide capable of refolding after it is denatured by heating or a heat-stable molecule such as a lipid or oligosaccharide. Identification of the active factor(s) that is secreted by the prostate cancer cells is currently under way. Koutsilieris et al. (46-48) observed that conditioned media from benign prostatic hypertrophic cells, malignant prostatic adenocarcinoma, and PC-3 cells all contain a factor that stimulates rat osteoblasts in vitro. Similarly, Perkel et al. (49) showed that PC-3 conditioned medium stimulates human osteoblasts in vitro. In the current experiments, DU-145, C4-2, and C4-2B prostate cancer cells released a soluble factor(s) that significantly decreased endoglin protein expression and subsequently altered TGF-ß/BMP signaling in the bone marrow stromal cells. ZR-75-1 conditioned medium also decreased endoglin protein levels, although to a much lower extent than the DU-145, C4-2, and C4-2B conditioned media, suggesting that there is a common mechanism of action of osteoblastic/osteosclerotic cell lines to attenuate endoglin expression because all three cell lines evaluated decreased endoglin expression. The difference in the magnitude of the effect between ZR-75-1 cells and the other cell lines may represent cell line variability and illustrate the complexity of the early cellular events that occur upon prostate cancer cell arrival in bone.

Accompanying the effects on endoglin protein levels, TGF-ß secretion was altered in HS-5 and HS-27a cells exposed to conditioned media from several of the test cell lines. PC-3, DU-145, and C4-2B conditioned media caused significant decreases in TGF-ß secretion, with almost complete ablation of TGF-ß secretion caused by DU-145 and C4-2B cells. Overall, LNCaP and ZR-75-1 cell conditioned media caused increases in TGF-ß secretion. Whereas the patterns of effects between endoglin and TGF-ß secretion were not identical for all cell lines evaluated, the data clearly illustrate that TGF-ß signaling is altered in HS-5 cells after exposure to conditioned media from several of the cell lines that also caused the decreased endoglin protein levels.

The bone marrow stromal cells expressed both TGF-ß receptor isoforms (ALK1 and ALK5) and showed active signaling through both ALK1-induced Smad1/5/8 activation and ALK5-induced Smad2 activation. However, in contrast to endothelial cells, bone marrow stromal cells with decreased endoglin expression had attenuated signaling through both TGF-ß receptor I isoforms (Fig. 12 ). In the bone marrow stromal cells, conditioned medium from C4-2 or C4-2B cells decreased Smad1 and Smad2/3 protein levels and reduced levels of activated (e.g., phosphorylated) Smad1/5/8 after BMP-2 stimulation. The effects in HS-27a cells were only induced by the C4-2B conditioned medium, most likely reflecting sensitivity differences between HS-5 and HS-27a cells. The decrease in pSmad1/5/8 decreased the expression of Id-1 in both HS-5 and HS-27a cells exposed to conditioned medium from C4-2 or C4-2B cells. Id-1 is an inhibitor of basic helix-loop-helix transcription factors and contains a BMP-responsive element within its promoter. Id-1 is often used as a marker of ALK1-induced Smad1/5/8 activation (32, 50, 51). Similarly, the decrease in Smad2/3 resulted in a decrease in PAI-I, a downstream target of pSmad2/3 (52), supporting the hypothesis that activated Smad2 levels were also decreased although pSmad2 levels were below the limit of detection by Western blotting. Overall, the effects on Smad2/3 signaling seemed to be more prevalent after conditioned medium exposure. Id-1 protein levels were variable in both HS-5 and HS-27a cells but also indicated a decrease in both cell lines after exposure to conditioned medium from DU-145 cells, a finding that would be consistent with the decreased endoglin protein levels but inconsistent with the lack of an affect on Smad1 protein levels. Overall, this illustrates that in bone marrow stromal cells, attenuation of endoglin protein levels decreases Smad1/5/8 and Smad2/3 levels and subsequently attenuates gene transcription via both ALK1-induced Smad1/5/8-dependent and ALK5-induced Smad2/3-dependent signaling pathways, respectively. This is in contrast to endothelial cells where siRNA-mediated knockdown of endoglin does not affect BMP signaling (32). Therefore, in the bone marrow stromal cells, and in contrast to endothelial cells, endoglin acts as a positive regulator of both ALK1-induced Smad1/5/8 activation and ALK5-induced Smad2 activation and directly participates in BMP signaling via ALK1.

View larger version (16K): [in this window] [in a new window] [Download PPT slide]   FIGURE 12. Model of regulation of TGF-ß/BMP signaling in bone marrow stromal cells by the auxiliary coreceptor endoglin. Under conditions of normal endoglin expression, signaling of TGF-ß signaling occurs via ALK5-induced Smad2/3 activation and BMP signaling occurs via ALK1-induced Smad1/5/8 activation. Under conditions of attenuated endoglin expression, as was the case after coculture of the bone marrow stromal cells with osteoblastic/osteosclerotic prostate cancer cells, signaling via both signaling pathways is attenuated, leading to decreased cell proliferation of the bone marrow stromal cells.

We evaluated alkaline phosphatase expression in the bone marrow stromal cells after exposure to the conditioned medium from C4-2B cells as an early marker of osteoblast differentiation (54-57). Under the test conditions, HS-27a cells, but not HS-5 cells, can be induced to differentiate under conditions of high cell density and/or after exposure to dexamethasone. However, the normal induction of alkaline phosphatase by BMP-2 was not observed. BMP-2 treatment actually decreased alkaline phosphatase staining in HS-27a cells, suggesting that HS-27a cells do not differentiate via the classic osteoblastic differentiation pathway that can normally be induced by BMP-2 (55-57). Whereas the data suggest that the conditioned medium from C4-2B cells decreases differentiation of HS-27a cells, differentiation of osteoprogenitor cells is a very complex process that is difficult to control, especially in vitro, and the lack of differentiation may not reflect what would occur under normal physiologic (i.e., in vivo) conditions. For example, Walsh et al. (58) observed similar effects when they administered TGF-ß to human bone marrow stromal cells and saw a decrease in cell proliferation and decreased alkaline phosphatase staining. Whether alterations in endoglin may affect differentiation of osteoprogenitor cells is unknown and should be investigated further under in vivo conditions in which endoglin levels can be manipulated by either knockdown or overexpression methods to examine the effect on prostate cancer bone metastases.

Overall, primary bone marrow stromal cells displayed activities similar to those observed in the bone marrow stromal cell lines. In primary bone marrow stromal cells exposed to conditioned medium from C4-2 or C4-2B cells, endoglin protein levels were decreased. Smad1 and Smad2/3 were both decreased by C4-2 and C4-2B conditioned media, leading to decreases in both PAI-I and Id-1. Interestingly, in the primary bone marrow stromal cells, DU-145 conditioned medium did not induce effects similar to the C4-2 and C4-2B cells. DU-145 conditioned medium actually caused a slight increase in endoglin protein levels, whereas Smad1, Smad2/3, and Id-1 protein levels were unchanged and PAI-I levels were slightly increased. This suggests that the osteoblastic/osteosclerotic cell lines may in fact have a common mechanism of action for the decrease in endoglin protein levels and subsequent alterations in Smad signaling. Overall, the data in the primary bone marrow stromal cells validated the results of the experiments using the immortalized bone marrow stromal cell lines, confirming that both signaling pathways are attenuated by a soluble factor(s) that is secreted by C4-2 and C4-2B cells.

The primary bone marrow stromal cells also displayed a doublet band for PAI-I. Conditioned media from both osteolytic cell lines, PC-3 and DU-145, caused a striking shift in the band intensity for the PAI-I doublet bands that was not observed for the other cell lines. Whereas the overall intensity of the doublet band was not changed by the conditioned media from PC-3 and DU-145 cells, there was a shift in band intensity in which the upper band was less intense and the lower band was more intense in the primary bone marrow stromal cells exposed to conditioned media from PC-3 and DU-145 cells, when compared with the control or foreskin fibroblast control. The importance of this observation is unclear, but may reflect the increased proteolytic activity associated with the PC-3 and DU-145 cell lines, resulting in an increase in the lower band, representing the proteolytically cleaved form of PAI-I.

As another negative control, we used normal PrEC conditioned medium to look at what might mimic the earliest stage of metastasis. The results were intriguing and provided further insight into the role of endoglin in the bone marrow stromal cells. In contrast to the effects observed in the cell lines exposed to conditioned media from a variety of cell types in which endoglin levels were either unchanged or decreased, there was a significant increase in endoglin expression in the primary bone marrow stromal cells and HS-5 cells that were exposed to PrEC conditioned medium. The increase in endoglin allowed us to compare the effects of both attenuation and "overexpression" of endoglin on Smad signaling. The increase in Smad protein levels and Smad signaling in HS-5 cells correlated with an increase in cell proliferation (data not shown), effects that are in opposition to the HS-5 and HS-27a cells that had attenuated endoglin protein levels. In effect, this provided a situation of endoglin overexpression. The responses observed in the primary bone marrow stromal and HS-5 cells were consistent with a role for endoglin as a positive regulator of both ALK1-induced Smad1/5/8 activation and ALK5-induced Smad2/3 activation.

To determine if the effects were the result of a primary affect on endoglin attenuation or the result of other genotypic or phenotypic changes in the bone marrow stromal cells, endoglin expression was selectively attenuated in HS-5 cells using siRNA. Similar to what was observed in the conditioned medium experiments, endoglin protein levels were decreased and Smad signaling was altered in the HS-5 cells transfected with endoglin siRNA. Smad1 and Smad2/3 protein levels were significantly decreased. PAI-I protein levels were also significantly decreased, whereas Id-1 protein levels were unchanged. Whereas this may suggest that the effects on Id-1 that were observed in the conditioned medium exposures were not due to the attenuation of endoglin, it is likely that the 50% decrease in endoglin in the siRNA-transfected HS-5 cells was not sufficient to attenuate Id-1 protein levels. Overall, the data from the siRNA experiment support the hypothesis that a direct effect of the coculture experiment was attenuated endoglin expression in the bone marrow stromal cells, which subsequently alters Smad signaling and cell proliferation.

Together, these data indicate that endoglin is a positive regulator of both ALK1-induced Smad1/5/8 activation and ALK5-induced Smad2/3 activation. This is in contrast to the role endoglin plays in endothelial cells, where endoglin acts as a positive regulator of ALK1 signaling and a negative regulator of ALK5 signaling. Our data also suggest that highly metastatic and osteoblastic/osteosclerotic cell lines such as C4-2 and C4-2B release soluble factor(s) that decreases endoglin expression in the bone stroma, resulting in altered Smad signaling and, ultimately, decreased cell proliferation. These experiments add to our understanding of early events that occur during prostate cancer metastasis to bone and the alterations that can occur in the host bone marrow stromal cell microenvironment.

	Materials and Methods

Top Notes Abstract Introduction Results Discussion Materials and Methods Acknowledgements References

 
Cell Culture
Human foreskin fibroblasts (SCRC-1042), human breast carcinoma cell line ZR-75-1 (CRL-1500), human bone marrow stromal cell lines HS-5 (CRL-11882) and HS-27a (CRL-2496), and human prostate carcinoma cell lines LNCaP (CRL-1740), PC-3 (CRL-1435), and DU-145 (HTB-81) were obtained from American Type Culture Collection. Human prostate carcinoma cell lines C4-2 and C4-2B were obtained from Dr. Leland Chung (Emory University School of Medicine, Atlanta, GA). Normal human PrEC were obtained from Clonetics. Foreskin fibroblasts, ZR-75-1 (ZR-75), HS-5, HS-27a, PC-3, and DU-145 cells were maintained in low-glucose DMEM (Gibco Life Technologies) supplemented with 10% (v/v) heat-inactivated FBS (DMEM/10% FBS). LNCaP, C4-2, and C4-2B cells were maintained in T-medium (Invitrogen Corp.) supplemented with 5% (v/v) heat-inactivated FBS. PrEC were maintained in the manufacturer's recommended growth medium. All cultures were maintained at 37°C in a humidified air/CO₂ (95:5, v/v) atmosphere.

Primary bone marrow stromal cells were obtained from bone marrow aspirates of a consenting male donor with approval from the Christiana Care Health Services and University of Delaware Institutional Review Boards. Bone marrow aspirates (10 mL) were incubated overnight at 37°C in a humidified air/CO₂ (95:5, v/v) atmosphere. The resulting top layer, containing mostly platelets, was discarded. The bottom layer was combined with 20-mL high-glucose DMEM (Gibco Life Technologies) supplemented with 10% (v/v) heat-inactivated FBS, plated onto a 150-cm² culture plate, and incubated at 37°C in a humidified air/CO₂ (95:5, v/v) atmosphere for 5 days. After 5 days, the growth medium was removed, the cells were washed twice with 25-mL PBS to remove the remaining RBC, and the cultures were refed 25-mL growth medium. Primary cell cultures were maintained at 37°C in a humidified air/CO₂ (95:5, v/v) atmosphere and used within 10 passages.

Coculture Experiments
HS-27a and C4-2B cells were plated at a density of 200,000 per well onto six-well culture plates or 25-mm Anopore tissue culture plate inserts (0.02 µm pore size; Nalge Nunc International), respectively. Plating density was chosen so that all cultures were confluent by visual inspection on the day the experiments were initiated (test day 0). Both cell lines were plated using DMEM/10% FBS to maintain consistent experimental conditions. Four wells were plated for each treatment and duplicate wells were combined at study termination, resulting in duplicate samples for each treatment. The cultures were plated 4 days before initiation of the experiment and both cell lines were maintained in separate six-well culture plates before study initiation. Cultures were serum deprived in low-glucose DMEM containing 1% (v/v) heat-inactivated FBS (DMEM/1% FBS) for 24 h before study initiation. On test day 0, the media on all wells were aspirated and replaced with fresh DMEM/1% FBS. The Anopore tissue culture plate inserts containing the C4-2B cells were then transferred into the six-well culture plates containing the HS-27a cells to initiate the experiment. The medium was not changed on subsequent days to minimize cell loss due to washing cells from the Anopore inserts. All cultures were incubated at 37°C in a humidified air/CO₂ (95:5, v/v) atmosphere for the duration of the experiment. On test day 3, total cellular protein was prepared for Western blot analysis or RNA was prepared for microarray analysis. For total cellular protein preparation, the adherent HS-27a cells in each well were lysed with 75-µL sample extraction buffer [SEB; 50 mmol/L Tris-HCl (pH 7.0), 8 mol/L urea, 1% (v/v) SDS, 1% (v/v) 2-mercaptoethanol] containing a 1:100 dilution of protease (Sigma-Aldrich Corp.) and phosphatase (Sigma-Aldrich) inhibitors, duplicate wells were combined, and the lysates stored at –80°C for protein quantification and Western blotting. For RNA preparation, the adherent HS-27a cells in each well were lysed with 300-µL lysis buffer, duplicate wells were combined, and total RNA was prepared using a Qiagen RNeasy kit. RNA was extracted using 30-µL RNase-free water and stored at –80°C for RNA quantification and analysis.

Conditioned Medium Exposures
Initial experiments were done as cocultures as described above. To circumvent variability in cell attachment that was encountered in the microarray experiments and initial experiments designed to validate the attenuation of endoglin RNA levels, subsequent experiments were done using conditioned media from the "coculture" cell lines (Table 2 ) and shown to consistently alter cellular endoglin protein levels. For conditioned medium exposure experiments, cell lines were plated onto six-well culture plates at a density of 125,000 per well (HS-5, HS-27a, and foreskin fibroblasts), 187,500 per well (PC-3, DU-145, C4-2, C4-2B, and primary bone marrow stromal cells), 250,000 per well (LNCaP and ZR-75-1), or 500,000 per well (PrEC). Plating density was chosen so that all cultures were confluent by visual inspection on the day the experiments were initiated (test day 0). For all experiments, cell lines were plated using DMEM/10% FBS to maintain consistent experimental conditions. Four wells were plated for each treatment and duplicate wells were combined at study termination, producing duplicate samples for each treatment. The cultures were plated 4 days before initiation of the experiment. All cultures were serum deprived in DMEM/1% FBS for 24 h before study initiation. On test day 0, the medium for each test cell line (HS-5, HS-27a, or primary bone marrow stromal cells) was aspirated and the cultures were fed the conditioned media from the feeder cell lines (i.e., foreskin fibroblasts, PC-3, DU-145, LNCaP, C4-2, C4-2B, ZR-75-1, or PrEC). On subsequent days (test days 1 and 2), the medium in each of the wells for the test cell lines was aspirated and replaced with fresh conditioned medium from the appropriate feeder cell line. The feeder cultures were then fed fresh DMEM/1% FBS on each test day. Control wells received DMEM/1% FBS on test days 0 to 2. All cultures were incubated at 37°C in a humidified air/CO₂ (95:5, v/v) atmosphere for the duration of the experiment. On test day 3, total cellular protein was prepared for Western blot analysis or RNA was prepared for reverse transcription-PCR. For total cellular protein preparation, the adherent cells in each well were lysed with 75-µL SEB containing a 1:100 dilution of protease (Sigma-Aldrich) and phosphatase (Sigma-Aldrich) inhibitors, duplicate wells were combined, and the lysates stored at –80°C for protein quantification and Western blotting. For RNA preparation, the adherent cells in each well were lysed with 300-µL lysis buffer, duplicate wells were combined, and total RNA was prepared using a Qiagen RNeasy kit. RNA was extracted using 30-µL RNase-free water and stored at –80°C for RNA quantification and analysis.

Measurement of Secreted TGF-ß Levels in Conditioned Medium
Cell lines were maintained and plated as described above onto six-well culture plates. Cultures were treated as described above for the conditioned medium exposure experiments. On test day 3, conditioned media (1 mL) were collected from the wells of the test cell lines (HS-5 and HS-27a) for analysis of secreted TGF-ß concentration. Conditioned media (1 mL) were also collected from the conditioned media feeder cultures (foreskin fibroblast, PC-3, DU-145, ZR-75-1, LNCaP, C4-2, and C4-2B) and analyzed for secreted TGF-ß concentration to correct for the TGF-ß secreted from the feeder cultures when calculating TGF-ß secreted from the test cell lines. Conditioned media (1 mL) were collected, centrifuged at 1,000 x g for 5 min at 4°C to remove cellular debris, and 500-µL aliquots from each well were stored at –20°C for analysis. After medium collection, the adherent cells in each well were lysed with 75-µL SEB and the protein concentration was measured by the method of Lowry et al. (59) for normalization of secreted TGF-ß levels to total cell protein per well. Conditioned media were thawed on ice and secreted TGF-ß concentration was measured using a Human Quantikine ELISA Assay (R&D Systems) according to the manufacturer's directions.

BMP-2 Stimulation
Basal levels of pSmad1/5/8 and pSmad2 in HS-5 and HS-27a cells were below the limit of detection by Western blotting (data not shown). To evaluate pSmad levels in HS-5 and HS-27a cells, conditioned medium exposures were done as described above. HS-5 and HS-27a cells were exposed to DMEM/1% FBS (control) or conditioned medium from C4-2B cells. On test day 3, test cell lines were treated with 500 ng/mL recombinant BMP-2 (355-BM; R&D Systems) or 10 ng/mL TGF-ß1 (240-B; R&D Systems) for evaluating pSmad1/5/8 or pSmad2, respectively. For cytokine treatment, the conditioned medium in each well was aspirated. BMP-2 and TGF-ß treatments were done in serum-free low-glucose DMEM and incubated at 37°C in a humidified air/CO₂ (95:5, v/v) atmosphere for intervals of 0, 5, 10, 20, 30, 40, 50, 60, 75, or 90 min. At each time interval, the medium was aspirated, the adherent cells in each well were lysed with 75-µL SEB containing a 1:100 dilution of protease and phosphatase inhibitors as indicated above, and the lysates frozen for protein quantification and Western blot analysis. pSmad2 levels were still below the limit of detection by Western blotting even after TGF-ß treatment (data not shown).

Apoptosis
For evaluation of basal apoptosis in HS-5 and HS-27a cultures exposed to conditioned medium, cells were plated onto 96-well culture plates at a concentration of 5,000 per well and cultured as indicated above for the conditioned medium exposure experiments. Exposures were done on triplicate wells for each conditioned medium feeder cell line. On test day 3, basal apoptosis was evaluated using a commercially available apoptosis assay (Roche Cell Death Detection ELISA^PLUS, Roche Diagnostics). Briefly, the medium from each well was aspirated, adherent cells were lysed with 200-µL lysis buffer, and the lysate was centrifuged at 200 x g for 10 min at room temperature. Twenty microliters of the resulting supernatant were incubated under gentle agitation with an immunoreagent containing antibodies specific for histone-associated DNA fragments for 3 h at room temperature. The ELISA plates were washed, incubated for 10 min with a colorimetric substrate, and basal apoptosis was determined by measuring the absorbance at 405 nm. Total cell protein per well was quantitated on three wells that were plated concurrently to normalize basal apoptosis for cell number. Wells designated for protein quantification were lysed with 25 µL of RIPA buffer and quantified using the Pierce BCA Protein Assay.

Cell Proliferation
For evaluation of cell proliferation, HS-5 or HS-27a cells were plated into 96-well culture plates at a concentration of 5,000 per well and cultured as indicated above. Exposures were done on six wells for each conditioned medium feeder cell line. On test day 3, growing cultures of HS-5 and HS-27a cells were labeled with 100 µmol/L 5-bromo-2-deoxyuridine for 4 h at 37°C in a humidified air/CO₂ (95:5, v/v) atmosphere. After labeling, cell proliferation was measured using a commercially available cell proliferation assay (Roche Cell Proliferation ELISA, Roche Diagnostics). Briefly, the medium from each well was aspirated, adherent cells were fixed for 30 min with 200 µL of the assay fixative, and the fixed cells were incubated with an immunoreagent containing antibodies specific for 5-bromo-2-deoxyuridine for 90 min at room temperature. The ELISA plates were washed and incubated for 15 min with a colorimetric substrate, the reaction was stopped by the addition of 1 mol/L sulfuric acid, and cell proliferation was determined by measuring the absorbance at 450 nm. Total cell protein per well was quantitated on three wells that were plated concurrently to normalize cell proliferation for cell number. Wells designated for protein quantification were lysed with 25 µL of RIPA buffer and quantified using the Pierce BCA Protein Assay.

Cell Cycle Analysis
Cell cycle distribution of HS-5 cells exposed to conditioned medium from C4-2B cells was evaluated by fluorescence-activated cell sorting. Conditioned medium exposures were done as described above. HS-5 cells were exposed to DMEM/1% FBS (control) or conditioned medium from C4-2B cells. On test day 3, cells were trypsinized and collected by centrifugation at 1,000 x g for 5 min at room temperature. The resulting cell pellet was fixed with 2-mL cold 70% (v/v) ethanol, immediately mixed, and stored at 4°C until analyzed ( 30 min). After two washes with PBS, cells were incubated in 1-mL PBS containing 50 µg/mL propidium iodide and 5 µg/mL RNase. To determine the proportion of cells in each phase of the cell cycle, propidium iodide staining of the DNA in 20,000 nuclei was quantified by fluorescence-activated cell sorting using a BD FACSCalibur flow cytometer (Becton Dickinson Biosciences).

Western Blot Analysis
Total cellular protein was prepared by solubilizing the adherent cells with SEB as indicated above. The protein concentration of each sample was determined by the method of Lowry et al. (59). Ten micrograms of total cell protein were analyzed for endoglin, Smad1, Smad2/3, and PAI-I. Twenty-five micrograms of total cell protein were analyzed for Id-1 and pSmad1/5/8. All membranes were probed with antibodies to ß-actin to correct for loading and transfer differences among samples. Before loading, all cell lysates were added to an equal volume of Laemmli sample buffer [0.1 mol/L Tris-HCl, 4% (w/v) SDS, 0.001% (w/v) bromophenol blue, 20% (v/v) glycerol], heated at 100°C for 5 min, and resolved by SDS-PAGE.

Proteins were separated by SDS-PAGE using Bio-Rad Criterion Tris-HCl gels with a 4% (w/v) polyacrylamide stacking gel and a 10% (w/v) polyacrylamide resolving gel in gel running buffer [150 mmol/L glycine, 50 mmol/L Tris base, 0.1% (v/v) SDS, pH 8.8] under constant voltage. The fractionated proteins were transferred to a nitrocellulose membrane at 4°C for 5 h at 40 V in transfer buffer (100 mmol/L glycine, 100 mmol/L Tris base, pH 8.3). The nitrocellulose membranes were blocked overnight at 4°C with gentle rotary agitation in TBS/0.1% (v/v) Tween 20 (TBS-T) containing 5% (w/v) bovine serum albumin for ß-actin and plasminogen activator inhibitor I (PAI-I), or TBS-T containing 5% (v/v) Blotto A (Santa Cruz Biotechnology) for endoglin, total Smad1 (Smad1), total Smad2/3 (Smad2/3), pSmad1/5/8, or Id-1. The primary antibodies for ß-actin (1:25,000; Abcam, Inc.) and PAI-I (1:1,000; Molecular Innovations, Inc.) were applied in 5% (v/v) bovine serum albumin/TBS-T. The primary antibodies for endoglin (1:250; Santa Cruz Biotechnology) and Id-1 (1:250; Santa Cruz Biotechnology) were applied in 2.5% (v/v) Blotto A. The primary antibodies for Smad1 (1:1,000; Cell Signaling Technology, Inc.), Smad2/3 (1:1,000; Cell Signaling Technology, Inc.), and pSmad1/5/8 (1:1,000; Cell Signaling Technology, Inc.) were applied in 5% (v/v) bovine serum albumin. All primary antibodies