Prostatic carcinogenesis evoked by cellular interaction.

Inoculation of tumorigenic prostatic stroma and nontumorigenic prostatic epithelia into the subcutaneous space of syngeneic rats induced the development of carcinosarcoma. The induced tumors, which were composed of a mixture of adenocarcinoma and fibrosarcoma, were androgen responsive. This model offers a novel mechanism for prostatic carcinogenesis in which prostatic fibroblasts determine epithelial growth, androgen responsiveness, and tumorigenicity. Our results emphasize the potential importance of an epigenetic pathway in prostatic carcinogenesis.


Introduction
Cellular interaction between stroma and epithelium are the key determinants during normal development for morphogenesis, cytodifferentiation, and hormonal responsiveness in most glandular tissues in vertebrate species (1)(2)(3)(4)(5). Classical approaches to deciphering the underlying mechanisms of cellular interactions often rely on successfully dissecting and dissociating embryonic tissues and recombining them either homotypically or heterotypically. This laboratory and others have employed this technology for the study of the regulation and growth of the prostate gland (6,7) or regression of the mammary gland (8) induced by androgenic steroids. These earlier studies have led to the general conclusion that androgen receptors in the stroma, rather than in the epithelium, are essential to androgen action during normal development.
To understand the cellular and biochemical basis of stromal-epithelial interactions and the possible involvement of chemical signals between these tissues, we have taken the approach of establishing cell lines from rat ventral prostate gland. We have studied the androgen responsiveness of prostatic cells in culture and in vivo. We report here our observation that an androgen-responsive tumorigenic fibroblast cell line induced a non-*Department of Urology, The University of Texas M. D. Anderson Hospital and Tumor Institute at Houston, 1515  tumorigenic epithelial cell line to form an adenocarcinoma. The tumors that resulted from the SC injections of mixtures of fibroblasts and epithelia were histologically characterized as carcinosarcoma and were composed of mixtures of adenocarcinoma and fibrosarcoma. These tumors were found to be androgen responsive in vivo. This observation emphasized the role of the stromal component in determining androgen responsiveness and its involvement in initiating epithelial cancer development.

Establishment of Rat Prostatic Epithelial and Fibroblast Cell Lines
Rat prostatic cell lines were established by a slight modification of a procedure described by Phillips and Rice (9). In brief, ventral prostate glands obtained from intact adult male rats (inbred Nb-strain) were cut into 1-mm cubes with scissors. The tissue fragments were placed on Falcon plastic dishes (Becton Dickinson Labware, Lincoln Park, NJ) and submerged in tissue culture T-medium [Dulbecco's modified Eagle medium: F-12K (4:1), fetal bovine serum (5%), biotin (0.244 ,ug/mL), adenine (25 ,ug/mL), insulin (5 ,ug/mL), transferrin (5 [ug/mL), choleratoxin (10 ng/mL), epidermal growth factor (10 ng/mL), and hydrocortisone (0.4 ,ug/mL)]. Prostatic epithelia were separated from their adjacent fibroblasts by removing differentially the loosely attached fibroblasts with phosphate-buffered saline containing ethylene diamine tetracetic acid disodium salt (EDTA). The epithelial cells attached to the plastic dishes were removed by digestion with trypsin (2 mg/ mL, Gibco, Grand Island, NY) and replated in T-me-dium. Morphologically distinct epithelial colonies, uncontaminated by fibroblast, were obtained after the fifth passage. Pure fibroblast culture without epithelial contamination was obtained by a procedure described previously (10). This procedure relied on the differential enzymatic digestion of fibroblast from minced prostatic tissues with 0.1% each of collagenase and hyaluronidase. In the present study, the morphology of fibroblasts was further confirmed by electron microscopic observation. The tumor-inducing spindle-shaped fibroblasts were stained blue by Masson's trichrome stain, indicating the lack of smooth muscle cells in the fibroblast preparation. Because the initial growth rate of the fibroblasts far exceeded that of the epithelia, a fibroblast cell line devoid of epithelial contamination was obtained before the third passage in 5% calf serum in DMEM (F-medium). Unless otherwise specified, the established fibroblast cell line, maintained for 11 to 36 passages, and the epithelial cell line, maintained for 30 to 34 passages, were used for this study. Figure 1 represents the morphologic structure of the established rat prostatic epithelium and fibroblast. Only the cultured epithelium stained positively with antikeratin antibody. These cell lines were also characterized biochemically for the presence of testosterone 5areductase [NADPH: 4A-3-ketosteriod 5a-oxido reductase (11)] and acid phosphatase activity (12). Both of these enzyme activities were found to be associated primarily with the epithelium. Activity in the cultured epithelial and fibroblast cells, respectively, was: testosterone 5a-reductase, 8.87 and 0.43 pmole/pLg DNA; acid phosphatase, 17.8 and 1.06 unit/10 cells.

Androgen Responsiveness of Rat Prostatic Cell Lines
To determine if these cell lines were androgen responsive, we exposed the cultured epithellal or fibroblast cells to various concentrations of dihydrotestosterone (DHT) (0, 0.1, 1.0, 10, 100, 1000, and 10,000 ng/ mL) in a downshift medium for 24 hr. The cells were then pulse-labeled for 2 hr with [methyl-3H]-thymidine (4 Ci/mmole), and the rate of DNA synthesis activities was determined by collecting and counting the perchloric acid precipitates on Millipore ifiters. Figure 2 shows that DHT at a concentration of < 1000 ng/mL stimulated DNA synthesis and increased cellular proliferation only in cultured prostatic fibroblasts. In contrast, DHT slightly inhibited [3H]-thymidine incorporation and cellular proliferation in cultured epithelium. DHT at 10 ,ug/mL markedly inhibited the growth of both prostatic epithelial and fibroblast cells. The stimulatory action of DHT on cultured prostatic fibroblasts is androgen-receptor mediated because it can be inhibited by the antiandrogens cyproterone acetate and 4hydroxy-flutamide (unpublished results).

Role of Tumorigenic Prostatic Fibroblasts on Prostatic Epithelial Cancer Development
Because the cultured prostatic epithelial and fibroblast cells proliferated quite well in vitro-their generation times were 13.4 and 17.3 hr, respectively-we  decided to investigate whether these cells were tumorigenic in vivo. Figure 3 demonstrates that prostatic fibroblasts at later passages (e.g., the 24th to 36th passages) were tumorigenic; however, prostatic epithelial lines and/or early passages of prostatic fibroblasts (e.g., 6.0 v4.0-3 3.0-020 the 11th through 14th passages) were nontumorigenic (Fig. 3A). To determine if the tumorigenic fibroblasts could induce the nontumorigenic epithelia to become tumorigenic through cellular interaction, we injected cultured fibroblasts and epithelia together into the syngeneic male hosts. Figure 3B shows that the nontumorigenic epithelia could be induced to form adenocarcinoma only by the tumorigenic fibroblasts; nontumorigenic prostatic fibroblasts were completely ineffective. If the total number of cells inoculated into adult male hosts was kept constant (2 x 106 cells), the optimal ratios of fibroblasts/epithelia to form the largest size of tumor was 1:10. As few as 0.1% of the fibroblasts (2000 fibroblasts) among epithelia was sufficient to induce tumor formation (Fig. 3C). In these studies, no linear relationship was apparent between the size of the tumor formed and either the absolute number or the percent of either cell type present at the time of tumor inoculation (DNA content per milligram of tumor tissue was equivalent among all tumors examined). These results suggest that bidirectional cell-cell interaction may be the most likely mechanism to account for the increase in tumor volume. We have examined the histologic composition of the tumors at the end of a 9-day growth period. Figure 4A shows that tumorigenic fibroblasts alone developed in to pure fibrosarcoma, whereas Figure 4B shows that the mixtures of fibroblasts and epithelia developed into carcinosarcoma. In the latter case, nontumorigenic epithelia, which were stained positively by the anticytokeratin antibody, were induced by the tumorigenic fibroblasts to divide and to form typical acini composed of adenocarcinoma mixed with sarcoma (Fig. 4C). In all ofthe tumors formed, we noted that the adenocarcinoma component is only a small fraction (2-5%) of the total 0.001 0.01 0.10 0.20 0.33 1.0 Ratio of Fibroblast/Epithelium tumor. The precise volume of the epithelial component in the tumor is difficult to assess because of the extensive intermixing between fibroblasts and epithelia. Nests of epithelial colonies, which were demonstrated by antikeratin immunofluorescence, were found to be randomly scattered among the sarcomatoid component. Numerous mitotic activities (Fig. 4B) were detected in the epithelia of the induced glandular structures, sug- gesting that the underlying tumorigenic fibroblasts have induced epithelial proliferation and organization to form adenocarcinoma through cell-cell interaction. We have subsequently recovered epithelial cells from the carcinosarcoma through cell culture method and found that the epithelial component remained nontumorigenic when inoculated into adult male hosts (data not shown). These results emphasize the role of fibroblast in epi-thelial tumorigenesis. The tumorigenic fibroblasts effectively promoted the nontumorigenic epithelia to undergo adenocarcinoma development.
Corroborating the in vitro DHT stimulation of cultured prostatic fibroblast growth, we found that both the fibrosarcoma induced by tumorigenic fibroblasts alone and the carcinosarcoma induced by the mixture of fibroblasts and epithelia were androgen responsive. Figure 5 shows that the fibrosarcoma grew somewhat faster in intact than in castrated adult male hosts and that castration greatly reduced the growth rate of the experimentally induced carcinosarcoma. No palpable growth of the transplantable prostatic tumors was detected in females during a 9-day observation period (data not shown).

General Comments and Conclusion
Results of this study demonstrate that both androgen responsiveness and tumorigenicity of the experimentally induced prostatic tumors are determined by the fibroblasts. Our data also suggest that androgen responsiveness alone is not sufficient to activate tumor development in the fibroblasts (e.g., after the 11th passage, fibroblasts are responsive to DHT-induced DNA synthesis but are completely nontumorigenic). The roles of fibroblasts in prostatic cancer development and androgen sensitivity are consistent with our previous finding that fetal urogenital sinus mesenchyme (fibroblast) is the target for DHT that determines the normal growth and development of fetal urogenital sinus epithelia in vivo in tissue recombinants (13).
As Days After Tumor Inoculation mechanism for prostatic carcinogenesis. The tumorigenic fibroblasts were found to induce the nontumorigenic epithelia to form a tumor that contained a mixture of sarcoma and adenocarcinoma. Although the condition of carcinosarcoma in the prostate is a rare disease in man (14), the concept that a directive influence from the stromal component induces patterns of adenocarcinoma formation deserves greater attention. In this regard, it is interesting to note that the epithelial cells in culture, isolated from normal or neoplastic human prostate glands, exhibited no difference in their growth responses to various hormones and growth factors (15). This novel mechanism of inducing prostatic tumor development via cell-cell interaction does not preclude the influence of genetic factor(s) that may predispose the epithelia to proliferate and to develop tumors in response to neighboring tumorigenic fibroblasts. Furthermore, it has not been determined if the changes introduced into the responding epithelium are genetically stable. However, electron microscopy provided no evidence for the presence of viral particles in the experimentally induced prostatic tumors. Finally, it is unlikely that the induced epithelial cancers were the result of cell fusion between tumorigenic fibroblast and nontumorigenic epithelium, because tumorigenicity, in general, was found to be a recessive trait (16). Work is currently in progress to determine DNA content profiles in cells through flow cytometry and karotyping of epithelial cells harvested before and after cellular interaction.
This study provides, for the first time, strong evidence that tumorigenic events can be activated across the cellular membranous barrier through cell-cell inter-  Adult male rats were either sham-operated or castrated 1 week prior to inoculation with tumorigenic prostatic fibroblasts (6.7 x 106 cells). Tumor volumes were measured by a microcaliper and calculated according to a standard procedure (24). Data represent the average sizes of tumors from three specimens each. In another experiment, two rats were inoculated with a total of 2 x 106 cells (10% fibroblasts), and the tumors were allowed to grow for 4 days. At this time, one animal was castrated and the other was sham-operated. Individual tumor volumes were measured daily or every other day for an additional 7 days. Data were expressed as average ± SEM of four determinations. In some cases, the SEs were not shown because they were small and fell within the average bars. Tumor volumes are expressed relative to day 4, which is designated as 1.0.
action. It is possible that certain chemical signaling, in the form of soluble growth factors (17), angiogenesis factors (18), or an extracellular matrix (19)(20)(21), is involved in maintaining cell-cell communication and phenotypic expression. Certain protein growth factors have been isolated from both whole normal rat (22) and hyperplastic human prostate glands (23). The functional significance of these chemical signaling systems between epithelium and stroma and their role in normal and abnormal prostate growth and development remain obscure. In this model system, we have observed that prostatic fibroblasts determine epithelial growth, androgen responsiveness, and tumorigenicity. The concept of the role of cell-cell interaction in prostatic carcinogenesis may, in the future, yield alternate therapeutic approaches to the regulation of cancer growth.