**1. Introduction**

The number of men diagnosed with prostate cancer (PCa) is increasing worldwide [1]. Most patients with early-stage PCa can be treated with therapies such as radical prostatectomy or irradiation. In contrast, patients with advanced PCa are treated with androgen deprivation therapy (ADT), the standard systemic therapy. Although ADT induces temporary remission, the majority of patients (approximately 60%) eventually develop and progress to castration-resistant prostate cancer (CRPC), which is associated with a high mortality rate [2,3]. In the development and progression of CRPC, a decrease or loss of androgen sensitivity in PCa cells often occurs. Low androgen sensitivity of PCa cells is associated with a more malignant phenotype and is difficult to cure. Changes in the androgen sensitivity of PCa cells are often caused artificially as negative effects of ADT and by spontaneously arising variants of androgen receptor (AR) even before ADT is started [4,5].

ADT for patients with advanced PCa aims to decrease the concentration of circulating androgen and block AR signaling in PCa cells [6]. Well-di fferentiated PCa cells are generally androgen and AR dependent, i.e., AR signaling regulates cell cycle progression and di fferentiation. Loss of AR signaling after ADT triggers AR-independent outgrowth, generating poorly di fferentiated uncontrollable PCa cells [7]. To prevent the development and progression of CRPC, we hypothesize that preservation of AR signaling after ADT is essential. Nelson et al. described four molecular-state frameworks for activation of AR signaling after ADT as follows: state 1, endocrine androgen and AR dependent; state 2, intracrine androgen and AR dependent; state 3, androgen independent and AR dependent; and state 4, androgen and AR independent [8]. Several molecular mechanisms responsible for changes in the AR dependency of PCa cells have been suggested, e.g., androgen-independent activation of AR signaling by mutations in the AR or altered levels of coactivators, and activation of alternative growth factor pathways [4,5]. In patients with CRPC, a number of growth factors and cytokines contribute to malignancy of PCa cells through activation of AR signaling in an androgen-independent manner, which is often called the "outlaw pathway" [8]. Previous studies have suggested that epidermal growth factor (EGF), fibroblast growth factor (FGF)-7 (also known as keratinocyte growth factor (KGF)), insulin-like growth factor (IGF)-1, and interleukin (IL)-6 can activate AR signaling in the absence of androgen [9–11] via various signaling pathways, including Akt, signal transducer and activator of transcription (STAT) 3, and p44/42 mitogen-activated protein kinase (MAPK) pathways [12]. In the tumor microenvironment, tumor stroma surrounding PCa cells is enriched in fibroblasts that secrete AR-activating factors, such as EGF, FGF-7/KGF, IGF-1, and IL-6 [13,14].

Most fibroblasts do not express AR and can survive in the absence of androgen [15–17]. Several studies have reported that androgen-independent interactions between PCa cells and fibroblasts determine how PCa cells respond to ADT [18,19]. In androgen-insensitive PCa cells, we demonstrated that fibroblast-derived FGF-7/KGF may bypass the functionally inactive AR and may promote cell proliferation after ADT [19]. In androgen-sensitive PCa cells, however, we have recently reported that fibroblast-derived EGF, IGF-1, and IL-6 can activate AR signaling, leading to preservation of AR signaling after ADT [17]. Thus, we hypothesize that fibroblast-dependent AR activation may preserve AR signaling after ADT and may play a critical role in the prevention of CRPC development and progression. Clinically, PCa is a heterogeneous disease with various biological behaviors, such as androgen sensitivity and response to ADT. To investigate the relationship between fibroblast-dependent AR activation and androgen sensitivity in PCa cells, well-established PCa cell lines with a variety of androgen sensitivities are strongly required.

LNCaP human PCa cell lines are androgen-sensitive PCa cell lines that are useful for investigating the molecular mechanisms responsible for changes in the androgen sensitivity and AR dependency of PCa cells. Notably, LNCaP cells are a heterogeneous cell population containing various clones with naturally occurring di fferences in androgen sensitivity caused by spontaneously arising changes [20,21]. Accordingly, we generated two sublines of LNCaP cells (E9 and F10) showing low androgen sensitivity by using a limiting dilution method with regular culture conditions [22,23]. In addition, we established androgen-insensitive AIDL cells from parental LNCaP cells by continuous passaging under hormone-depleted conditions [24]. The parental LNCaP cell line and its derivative sublines (E9, F10, and AIDL) expressed similar levels of AR protein, and AR-dependent secretion of prostate-specific antigen (PSA) was detected in LNCaP and E9 cells [14]. As compared with parental LNCaP cells, we have previously reported that combination of E9 or AIDL cells with embryonic rat urogenital sinus mesenchyme promoted tumor progression in vivo even under androgen ablation [19]. Comparing the characteristic features of paternal LNCaP cells and its sublines allows us to investigate the molecular mechanisms responsible for changes in the androgen sensitivity and AR dependency of PCa cells.

In this study, we defined "androgen sensitivity" in PCa cells as the degree of androgen-dependent AR activation in vitro. In contrast, we defined "AR dependency" in PCa cells as the degree of androgen- and growth factor/cytokine-dependent AR activation. The objective of this study was to

investigate the role of fibroblast-dependent AR activation in the tumorigenesis of three LNCaP sublines di ffering in androgen sensitivity under androgen deprivation conditions.

### **2. Materials and Methods**
