Prostate cancer cell growth is modulated by adipocyte-cancer cell interaction


Y. Tokuda, MD, PhD, Department of Urology, Saga Medical School, Nabeshima 5-1-1, Saga, 849–8501, Japan.



To assess whether adipocytes (mesenchymal stromal cells thought to affect the proliferation and differentiation of epithelial cells) affect the behaviour of prostate cancer cells in vitro, as prostate cancer metastasizes to the bone, which is an adipocyte-rich environment.


The human bone-metastatic prostate carcinoma cell line PC3 was cultured with or without adipocytes in a three-dimensional collagen gel matrix. Histological and immunohistochemical assays were used to evaluate the proliferation and differentiation of PC3 cells. The cytokine expression of this culture assembly was also examined by reverse transcription-polymerase chain reaction (RT-PCR).


The proliferation and differentiation of cancer cells were clearly changed on co-culture with adipocytes compared with the control cultures. The mean (range) bromodeoxyuridine (BrdU) indices estimated (according to uptake) to evaluate the growth of the cultured cells were 36.3 (8.55)% in the co-culture and 26.95 (10.50) in the control (P < 0.02). PC3 cells in co-culture formed larger clusters than in the control, at 16.0 (11.0) vs 14.0 (10.0), respectively (P < 0.01). Cancer cells also showed pleomorphism, varying from cuboidal to spindle-shaped. The expressions of vascular endothelial and platelet-derived growth factor were greater in co-culture than in the control.


These findings suggest that adipocytes modulate the growth, morphology and cytokine expression of prostate cancer cells. This specific mesenchymal stromal cell type is important in the biological behaviour of prostate cancer cells.


It is known that fibroblasts are important in mesenchymal stromal cell–cancer cell interactions. Many studies have assessed adipocyte-human normal mammary epithelial cell and adipocyte-mammary cancer cell interactions [1,2]. In these studies, mammary adipocytes significantly enhanced casein and lipid accumulation within the mammary epithelial organoids. Furthermore, adipocytes induced mammary epithelial organoids to undergo alveolar morphogenesis, inhibiting squamous outgrowth, and increasing lumen size [1]. Rahimi et al.[2] showed that 3T3-L1 adipocytes stimulated murine mammary carcinoma SP1 cell growth by secreting hepatocyte growth factor (HGF).

We previously reported interactions between some epithelial cells and adipocytes, using the three-dimensional collagen gel matrix culture system. Subcutaneous adipocytes promoted the differentiation of both reconstructed skin and cutaneous carcinoma cells (DJM-1) [3,4]. Adipocytes clearly enhanced the invasion activity of fibroblasts in laryngeal squamous cell carcinoma on collagen gel matrix culture [5]. We also studied the effect of epididymal adipocytes on rat prostatic tissue in vitro. The co-culture with adipocytes induced higher proliferation and greater differentiation of prostatic epithelia than in control culture [6]. These results suggest that adipocytes participate in both epithelial cell–stromal cell interaction and in stromal cell–stromal cell interaction.

The aim of the present study was to evaluate how adipocytes influence prostate cancer cell growth. With age, adipose tissue becomes the predominant component within bone marrow. Bone metastasis occurs frequently in clinical prostate cancer. Thus, studies of the effect of adipose tissue on prostate cancer cell growth would seem to be useful for assessing the biological behaviour of prostate cancer within bone.

Epidermal growth factor (EGF) and platelet-derived growth factor (PdGF) promote the proliferation of prostatic cells by activating the specific membranous receptors of both epithelial and stromal cells. The discovery of angiogenesis-dependent tumour growth led to the identification of several endothelial cell growth factors, including acidic and basic fibroblast growth factors, vascular endothelial growth factor (VEGF), TGF-α and -β, and PdGF [7]. Ravindranath et al.[8] showed that EGF greatly enhanced the expression of VEGF mRNA in DU145 and PC3 cell lines in a dose-dependent manner. Recently, it has also become apparent that adipocytes produce so-called ‘adipocytokines’, e.g. leptin, adipsin, plasminogen activator inhibitor 1, TNF-α, EGF, IGF-II and adiponectin [9]. In the present study, the expression of the angiogenesis-dependent tumour growth factors, VEGF and PdGF, was assessed using RT-PCR.



Epididymal adipose tissues were dissected from adult male Wistar rats (8–12 weeks old), housed in environmentally controlled rooms, with food and water freely available. After the rats were killed with pentobarbital sodium, epididymal adipose tissues were removed using small scissors. To obtain unilocular fat cells, the adipose tissue was digested with a collagenase solution as reported previously [6]. Briefly, adipose tissues were chopped into pieces and digested with 0.2% collagenase solution at 37°C. After filtration and centrifugation of the digested cell suspension, unilocular adipocytes were obtained in a thin, white, floating layer. The adipocytes were washed three times to eliminate collagenase.


Co-cultures with adipocytes were examined using the bone metastatic prostate cancer cell line, PC3 (Dainippon Seiyaku, Japan). Briefly, eight volumes of acid-soluble type I collagen solution (Nitta Gelatin, Osaka, Japan) were mixed with one volume of 10-fold concentrated Ham's F-12 growth medium (with no NaHCO3) and one volume of reconstruction buffer (2.2 g of NaHCO3 and 4.77 g of HEPES in 100 mL of 0.05 mol/L NaOH). Cancer cells were embedded in 100 × 104/2 mL of collagen solution and immediately poured into a flask. All procedures were conducted on ice to delay gelation of the collagen solution. The number of adipocytes in the culture assemblies was ≈ 15 × 104/2 mL. After 0.5 h, when the gel became strong enough, it was overlaid with 15 mL/dish of Ham's F-12 growth medium supplemented with 10% neonatal calf serum and antibiotics. The cultures were incubated at 37°C in 5% CO2/95% air in a water-saturated atmosphere. Then 10 mL of the growth medium was routinely changed every 48 h. A cancer cell-only medium was cultured as a control.


Bromodeoxyuridine (BrdU; 30 µL, 2 µg/mL) was added 24 h before fixing the culture assemblies with 30% acetic ethanol, and embedding in paraffin wax. Deparaffinized sections were immunostained by the procedures of the BrdU Kit (Amersham, Arlington Heights, IL). Groups of ≥ 500 cancer cells were counted manually to assess the BrdU-labelling index. The number of cells in cancer clusters was also counted to estimate the size of aggregations. Data were expressed as the median (interquartile range), with differences between the groups evaluated using the Mann–Whitney U-test.


The collagen gels fixed with 10% formalin were stained with haematoxylin-eosin. To identify lipid droplets within the cytoplasm of culture cells, the frozen sections of the collagen gels were stained by Oil red-O [10]. The ultrastructure of the culture cells was examined by transmission electron microscopy. To observe the fine structure, the collagen gels were fixed with 2.5% glutaraldehyde and processed by standard methods.


Real-time quantitative RT-PCR was performed starting with 50 ng of reverse-transcribed total RNA, with 10 µmol/L of both sense and antisense primers in a final volume of 20 µL using the Syber Green PCR core reagents. Relative quantification for a given gene, expressed as ‘-fold’ was calculated after normalization to α-actin. The mRNA expression of cytokines was assessed by the ratio between the culture assemblies of cancer cells and those of adipocytes only; experiments were repeated twice. For Light Cycler reaction, a mastermix of the following reaction components was prepared to the indicated end-concentration: 6.8 µL water, 1.2 µL MgCl2 (3 mmol/L), 0.5 µL forward primer (0.5 µmol/L), 0.5 µL reverse primer (0.5 µmol/L) and 1.0 µL Light Cyler (Fast Start DNA Master SYBR Green I; Roche Diagnostics, USA). Light Cycler glass capillaries were filled with Light Cycler mastermix (9 µL), and 1 µL cDNA (reverse-transcribed total RNA) was added as the PCR template. Capillaries were closed, centrifuged and placed into the Light Cycler rotor. The following Light Cycler experimental protocol was used: denaturation programme (96°C for 10 min), amplification and quantification programme repeated 60 times (60°C for 5 s and 72°C for 10 s with a single fluorescence measurement), melting curve programme (70–96°C with a heating rate of 0.1°C/s continuous fluorescence measurement) and finally a cooling step to 40°C. Each of the primers is described in Table 1.

Table 1.  Primers for RT-PCR



Numbers of cluster cells varying in size are shown in Fig. 1a. Larger clusters were more frequently formed in the co-culture system than in the control. The box-plot (Fig. 1b) shows that the clusters of the co-culture system were significantly larger than those of the control (P < 0.01). Figure 1c shows the BrdU indices of clusters in the co-culture system and the control; the uptake ratio of the co-culture system was also significantly higher than that of the control (P < 0.02). On histological examination, the co-culture system had a densely packed cellular aggregation under low power (Fig. 2a,b).

Figure 1.

Figure 1.

a, Histograms of PC3 clusters; the number of larger clusters was greater with adipocytes present (a1) than with no adipocytes (a2). The size of each cluster was expressed by the number of component cells. b, The number of PC-3 cells in the clusters of co-cultured cells was significantly greater than in those of the control (* P < 0.01). c, The BrdU uptake ratio of cell clusters on day 14; proliferative activity in the co-culture was significantly more vigorous than that in the control (** P < 0.02). The central line is the median, the box the interquartile range, the bars the sd and the points the outliers.

Figure 1.

Figure 1.

a, Histograms of PC3 clusters; the number of larger clusters was greater with adipocytes present (a1) than with no adipocytes (a2). The size of each cluster was expressed by the number of component cells. b, The number of PC-3 cells in the clusters of co-cultured cells was significantly greater than in those of the control (* P < 0.01). c, The BrdU uptake ratio of cell clusters on day 14; proliferative activity in the co-culture was significantly more vigorous than that in the control (** P < 0.02). The central line is the median, the box the interquartile range, the bars the sd and the points the outliers.

Figure 2.

Haematoxylin and eosin staining. a, PC3-only culture on day 14 (× 20); cancer cells diffusely formed small nests. b, Co-culture system on day 14 (× 20); these cancer cells proliferated more vigorously and formed larger nests than those of the control. c, Control culture on day 14 (× 200); cancer cells grew by forming spherical aggregations. d, Co-culture system on day 14 (× 400); cancer cells showed acinus-like structures with well defined luminae and lacy cytoplasm. Some cancer cells with a spindle shape were scattered; e, Co-culture system on day 14 (× 400); cancer cells formed a cribriform configuration and proliferated by surrounding the adipose tissue. F, Co-culture system on day 14 (× 400); some cancer cells became spindle-shaped. g, Oil red-O lipid stain of PC3-only culture on day 14 (× 800); there were small lipid droplets within the cancer cytoplasm. h, Co-culture system on day 14 (× 800); there were many lipid droplets within the cancer cytoplasm. *Adipocyte; arrowhead, spindle-shaped cancer cell.


Cancer cells of the control culture tended to form round and spherical clusters (Fig. 2c). Cancer cells co-cultured with adipocytes showed a trend of pleomorphism, e.g. acinus-like or spindle-shaped constructions (Fig. 2d–f). The cells in acinus-like structures had larger and more lacy glassy cytoplasm than those of the control, and showed well-defined laminae under high magnification (Fig. 2d). Moreover, some cancer clusters had a cribriform configuration and grew by surrounding the adipocytes (Fig. 2e). Spindle-shaped cells were PSA-positive by immunohistochemistry (data not shown). Oil red-O lipid stain and ultrastructural examination showed larger and richer lipid droplets within the cytoplasm of co-cultured cancer cells (Figs 2g,h and 3a). These findings indicate the differentiation of prostatic epithelium. Spindle-shaped cancer cells also had lipid droplets within the cytoplasm (Fig. 3b).

Figure 3.

Electron microscopy: PC3 with adipocytes on day 14. a, Cancer cells had rich lipid droplets within the cytoplasm (× 2500). b, Some cancer cells grew by forming a spindle shape (× 1500). L, lipid droplets; *adipocyte.


The mRNA expression of cytokines was assessed by RT-PCR. Both VEGF and PdGF expression in co-culture was 20-fold that in the control culture, with values for PC-3-only, co-culture and adipocytes-only of 0.0008, 0.0207 and 0.001, and 0.0045, 0.02 and 0, respectively.


Adipocytes produce ‘adipocytokines’, e.g. TNF-α, EGF, IGF-II, adipsin, leptin and other factors [9]. Rahimi et al.[2] reported stromal adipocyte–tumour cell interaction and suggested that the HGF (TNF-α) secreted by adipocytes may be a key regulator of mammary tumour growth. We previously showed that adipocytes promoted differentiation of the squamous cell carcinoma cell line (DJM-1) and the invasion of laryngeal carcinoma (HEp-2) cells in collagen gel matrix culture [4,5]. When prostate cancer invades across the capsule or metastasizes into the skeletal tissue, the cells acquire a new adipocyte-rich environment. Obesity has also been indicated as a risk factor for prostate cancer [11]. Considering these previous findings we assessed for the first time whether adipose tissue influences prostate cancer cell growth in a three-dimensional environment in vitro.

Festuccia et al.[12] showed that the PC3 cell line contains different cell variants; a first variant grows as spherical multicellular aggregates and shows anchorage-independent growth. A second grows as single small round cells and shows anchorage-dependent growth with no spreading of the cells; a third, representing the most abundant population, grows as adherent cells. In the present study the PC3 cells were co-cultured with dissociated adipocytes in a three-dimensional collagen gel matrix. Morphologically, such PC3 cells had larger cancer cell clusters and higher proliferation than in control culture. Some co-cultured PC3 cells showed clear cytoplasm and acinus-like structure formation; there were also cribriform structures. These results indicate that adipocytes affect the differentiation of PC3. Spindle-shaped cells scattered throughout the co-culture assembly were confirmed to be prostate cancer cells, based on their positive staining for PSA antibody (data not shown). These findings suggest that the differentiation of PC3 cells might be enhanced by co-culture with adipocytes.

In the few cultures in the present study the mRNA expression of VEGF and PdGF was induced by co-culture with adipocytes, despite adipocytes having been reported as secreting none of those cytokines. A statistical assessment was not possible in the present study, but a study showing mRNA expression of VEGF in visceral adipocytes was reported by Miyazawa et al. (Annual Meeting of the Japan Diabetes Society, Kyoto, April 2001). Jones et al.[13] showed that serum VEGF was elevated in patients with metastatic or hormone-escaped prostate cancer, and Soker et al.[14] suggested the possibility that tumour cell-derived VEGF might have an autocrine role in the spread of rat prostate cancer, in addition to its known paracrine activity. It has not been reported previously that adipocytes modify the secretion of VEGF and PdGF by tumour cells.

Adipocytes promoted the differentiation of the squamous cell carcinoma cell line (DJM-1), while they enhanced the invasion of laryngeal carcinoma (HEp-2) cells [4,5]. Although these data are not shown, we also co-cultured other prostate cancer cell lines (DU 145 and LNCaP) with adipocytes. In that experiment, VEGF expression was increased in DU 145, whereas it was suppressed in LNCaP. These results indicate that a difference in cell type may contribute to the different responses to ‘adipocytokines’. In the PC3 cell line, which is derived from bone metastatic prostate cancer, the characteristics of invasion might be manifested as a result of cancer cell–adipocyte interaction. Although the results were significantly different according to the type of prostate cancer cell lines, this study suggests that prostate cancer cells grow differently in an adipocyte-rich environment such as bone marrow. The data may also support the view that obesity could be a risk factor of prostate cancer. We conclude that these specific mesenchymal stromal cell type (adipocytes) are important in the growth of prostate cancer cells.


The English in this manuscript was revised by Miss K. Miller (Royal English Language Centre, Fukuoka, Japan). The part of the results were reported in the 25th Congress of the Societe Internationale d’Ulorogie in Singapore. We are grateful to Mr H Ideguchi, Mr S Nakahara, Mr T Tabata, Mr K Tomoda and Mr S Takuma for their technical assistance. This work was supported, in part, by a grant in aid of Fundamental Scientific Research from the Educational Ministry (B) (2) #08457426 and (B) (2) #12470332.


hepatocyte growth factor


epidermal growth factor


platelet-derived growth factor


vascular endothelial growth factor