Activated 3',5'-cyclic AMP-dependent protein kinase is sufficient to induce neuroendocrine-like differentiation of the LNCaP prostate tumor cell line.

Neuroendocrine (NE) differentiation within prostate tumors is proposed to be a contributing factor in disease progression. However, the cellular origin and molecular mechanism controlling differentiation of prostatic NE cells are unresolved. The prostate tumor cell line, LNCaP, can reversibly acquire many NE characteristics in response to treatment with beta-adrenergic receptor agonists and activators of adenylate cyclase. In this study, we demonstrate that these treatments induce protein kinase A (PKA) activation in LNCaP cells and that ectopic expression of a constitutively activated form of the PKA catalytic subunit, CIalpha, results in acquisition of NE characteristics, including the extension of neuritic processes, cessation of mitotic activity, and production of neuron-specific enolase. Forskolin-, epinephrine-, and isoproterenol-dependent NE differentiation of LNCaP cells was significantly inhibited by expressing a dominant negative mutant of the PKA regulatory subunit, RIalpha. These results demonstrate that prostatic NE differentiation in response to these agents depends on PKA activation, and this signaling pathway may provide a therapeutic target for treating advanced forms of prostate cancer.

Efforts to develop novel therapeutic strategies for treating advanced prostate cancer have resulted in a considerable interest in the potential role of prostatic neuroendocrine (NE) 1 cells in disease progression (reviewed in Refs. 1 and 2). It has become evident that NE cells are a normal component of both the developing and mature prostatic epithelium and are postulated to be derived from stem cells common to both the exocrine and basal cell populations (3,4). NE cells produce a variety of neurosecretory products that exhibit growth-promoting activities, including parathyroid hormone-related peptides, neurotensin, serotonin, calcitonin, and bombesin-related pep-tides (5)(6)(7)(8)(9), suggesting that these cells function through endocrine/paracrine mechanisms to regulate the normal development and secretory activity of the prostate.
Virtually all prostatic adenocarcinomas contain foci of cells with NE characteristics (10). The presence of extensive multifocal NE features in tumors is an indication of increased aggressiveness and androgen independence (11)(12)(13). While the prognostic value of tumor NE status remains controversial, a strong link between NE status and long term, disease-specific survival has been reported (14). Tumor cell populations become enriched for NE cells following long term anti-androgen therapy (15), and although the NE cells appear to be nonmitotic, the carcinoma cells adjacent to these NE foci have been noted to exhibit increased proliferative activity (16,17). These studies suggest that NE cells that develop within prostate tumors may produce neurosecretory factors that contribute to increased malignancy and decreased responsiveness to androgen ablation therapy.
The androgen-responsive prostate tumor cell line, LNCaP, has emerged as a useful model for testing the development of a NE phenotype in adenocarcinoma cells with an exocrine phenotype. LNCaP cells have been shown to acquire NE characteristics in response to increased intracellular cAMP levels (18,19), long term androgen ablation (20), and stimulation with the cytokines interleukin-1␤ and -6 (21,22). Treating LNCaP cells with the ␤-adrenergic receptor agonists epinepherine (Epi) and isoproterenol (Isop) or with the adenylate cyclase activator, forskolin (Fsk), induces the rapid accumulation of intracellular cAMP and acquisition of numerous NE characteristics, including development of neuritic processes; cessation of mitotic activity; increased expression of neuron-specific enolase (NSE) and the biogenic amine serotonin; and the release of parathyroid hormone-related peptide and neurotensin into the culture medium (19). However, maintenance of the NE phenotype is lost upon withdrawal of differentiation-inducing agents, resulting in retraction of neuritic processes, reinitiation of mitotic activity, and loss of neuronal marker expression (19). These findings suggest that the phenotype of tumor cells may be dynamic and determined in part by the balance of differentiative and mitogenic factors in the local environment.
Although cAMP-dependent protein kinase (PKA) is often considered essential for mediating the wide range of physiological effects initiated by cAMP, direct evidence for its role is often limited. The recent identification of cAMP-responsive guanine nucleotide exchange factors for Rap1 (23,24) highlights the existence of PKA-independent cAMP signaling pathways. Additionally, treatment of LNCaP cells with a butyrate analogue was shown to be sufficient to induce cell cycle arrest (25), suggesting the possibility that dibutyryl cyclic AMP (Bt 2 cAMP) may have antiproliferative effects due to the butyrate component of the reagent and not due to activation of cAMP-dependent effectors. In this report, we establish that PKA activation is sufficient to initiate acquisition of a NE phenotype by LNCaP cells and that PKA activation is required for acquisition of these characteristics by ␤-adrenergic receptor agonists or Fsk. These results indicate that PKA activation plays a central role in regulating the phenotype of LNCaP cells and suggest that PKA activation may be a novel therapeutic target for controlling NE differentiation in prostatic adenocarcinomas.
Kinase Assays-PKA kinase activity was assessed by immune complex kinase assay derived from previously published protocols (28,30). Immunoprecipitations were performed from 500 g of lysate protein using 1 g of rabbit polyclonal anti-PKA CI␣ catalytic subunit antibody (Santa Cruz Biotechnology) collected on Protein A-Sepharose (Sigma) or 10 l of M2 anti-FLAG epitope antibody conjugated to agarose beads (Kodak). Immune complexes were washed 3 times with HO lysis buffer and once with kinase buffer (50 mM HEPES, pH 7.4, 10 mM MgCl 2 , 1 mM EGTA, 0.014% Tween 20) and resuspended in 40 l of kinase buffer. Kinase reactions were started by adding 10 l of kinase buffer containing 28 M synthetic PKA peptide substrate, malantide (Sigma), and 1 mM (1 Ci/nmol) [␥-32 P]ATP (NEN Life Science Products) and incubated at 30°C for 10 min. Reactions were terminated by adding 10 l 1 M HCl, and 35 l of the reaction was spotted onto 1-cm 2 strips of phosphocellulose (P81; Whatman). The P81 strips were washed four times for 10 min each in 75 mM H 3 PO 4 , washed once in methanol, dried, and counted by Cherenkov radiation. Malantide is a synthetic dodecapeptide corresponding to the PKA phosphorylation site of phosphorylase kinase ␤-subunit and is efficiently phosphorylated by PKA in vitro (K m ϭ 15 M; V max ϭ 23 mol of phosphate transferred min Ϫ1 mg Ϫ1 (28)). Relative PKA-specific activity was determined as pmol of phosphate incorporated into malantide per min per unit of immunoprecipitated PKA CI␣ ( Fig. 1) or FLAG-tagged kinase (Fig. 3) as measured by immunoblotting as described above.
Immunocytochemistry-For immunofluorescence analysis, LNCaP cells were plated onto glass coverslips and processed as described previously (31). FLAG-tagged CI␣ or Cqr expression was detected using anti-FLAG antibody, M5, and Texas Red-conjugated goat anti-mouse IgG (Jackson Immunoresearch Laboratories, West Grove, PA). Cells undergoing DNA synthesis were detected by bromodeoxyuridine (Br-dUrd; Sigma) labeling, performed by adding 100 M BrdUrd to the culture medium of cells during the last 20 h of incubation. BrdUrd incorporation into DNA was assessed by immunofluorescence using anti-BrdUrd-fluorescein isothiocyanate antibody (Roche Molecular Biochemicals) as described by the manufacturer.
Statistical Analysis-Assessment of differences in response was performed using analysis of variance models fit using maximum likelihood techniques using single factor analysis of variance models for Figs. 5A and 6 and multifactor analysis of variance models for Figs. 1, 5B, and 8. Variance-stabilizing transformations were used when required to meet assumptions of a linear model (constant variance and normally distributed error terms). A natural logarithmic transformation was used for Figs. 1 and 6 that allowed the use of the antilogarithm of pairwise differences to compare the ratio of two quantities (-fold change). The data in Fig. 8 required an arc sine square root (sin Ϫ1 (͌x)) transformation, for which there is no convenient interpretation of a pairwise difference. Therefore, only the relative magnitude and statistical significance of these results were of interest. Type III F-tests were used to determine the importance of effects and interactions. Pairwise comparisons were made at the highest level of complexity that was significant for this test at the 5% level. Confidence intervals for pairwise differences were generated by Fisher's least significant difference method with a comparison type I error rate of 0.05 (32). A 95% confidence interval that does not contain 0 (or 1 in the case of -fold change) implies statistical significance at the 5% level. Representative photomicrographs and immunoblots are provided to demonstrate the primary data. , or 100 ng/ml EGF for 10 min with anti-PKA CI␣ antibody. Immune complexes were subjected to in vitro kinase assays using malantide as a substrate and immunoblotting using anti-PKA CI␣ antibodies. For untreated lysates, kinase assays were performed in the presence or absence of 2 nM H89 to determine the percentage of basal activity attributable to PKA. The relative PKA-specific activity (pmol of phosphate transferred to malantide min Ϫ1 arbitrary unit of PKA catalytic subunit immunoprecipitated Ϫ1 ) was normalized for the activity of untreated cells and expressed as the mean Ϯ S.E. of five independent experiments. The activity of the immunoprecipitates from untreated cells ranged from ϳ1 to ϳ50 pmol of phosphate transferred to malantide under the described assay conditions. Statistical significance of pairwise differences for PKA activity from untreated cells (No Trt.) as compared with activity from cells treated with Fsk, Isop, IL-6, and EGF or from untreated cells to which 2 nM H89 was added during the immune complex kinase assay (No Trt. ϩ H89), was performed using errors generated by the log transform pairwise comparisons method described under "Experimental Procedures" ( †, p Ͻ 0.05). † is used here only to indicate those treatments for which PKA activity was significantly different from that of no treatment.

PKA Is Activated in LNCaP Cells
CaP cells treated with Epi, Isop, or Fsk induced an increase in cytosolic cAMP levels and the reversible acquisition of a number of NE characteristics. We measured PKA-specific kinase activity from LNCaP cells acutely stimulated with these agents to assess their ability to mediate activation of PKA. PKAspecific kinase activity was significantly increased (ϳ4-fold) in lysates from LNCaP cells treated with Fsk or Isop as compared with that of lysates from unstimulated cells or cells stimulated with IL-6 or EGF (Fig. 1). Additionally, the kinase activity immunoprecipitated from untreated cells could be significantly inhibited (ϳ80%) by addition of the PKA inhibitor, H89, to the kinase assay. These results establish that PKA is activated in LNCaP cells under treatment conditions that induce the acquisition of NE characteristics.
Transfection of PKA Catalytic Subunits into LNCaP Cells-To directly assess whether activated PKA was sufficient to induce acquisition of NE characteristics by LNCaP cells, we employed cDNAs encoding amino-terminal FLAG-tagged wild type PKA catalytic subunit, CI␣, or constitutively activated mutant, Cqr, which is unable to bind PKA regulatory subunit in the absence of cAMP due to the point mutations, H87Q and W196R (27) (Fig. 2). LNCaP cells were transiently transfected with equal amounts of plasmids encoding FLAG-tagged CI␣, Cqr, or extracellular signal-regulated kinase 2 and assessed for expression and PKA-specific kinase activity (Fig. 3). The CI␣ and Cqr subunits were readily detected in anti-CI␣ immunoblots of anti-FLAG immunoprecipitations from the appropriately transfected populations (Fig. 3B, upper panel), while similar levels of all three kinase constructs were detected in anti-FLAG epitope immunoblots of anti-FLAG immunoprecipitates (Fig. 3B, lower panel).
When normalized to the amount of FLAG-tagged kinase immunoprecipitated, the PKA kinase activity from Cqr-transfected cells was ϳ2-fold greater than the anti-FLAG immunoprecipitated activity detected in CI␣-transfected cells stimulated with Fsk and ϳ5-fold greater than the activity of unstimulated CI␣-transfected cells (Fig. 3A). Treatment of Cqrtransfected cells with Fsk had no discernible effect on FLAGprecipitated PKA kinase activity. FLAG-extracellular signalregulated kinase 2-transfected LNCaP cells exhibited a basal level of malantide phosphorylating activity equivalent to onetenth the activity detected from unstimulated, CI␣-transfected, cells. This basal activity was independent of Fsk stimulation. Each construct has been assessed three times with similar results, indicating that tagging and subcloning these constructs did not affect the previously reported differences in enzymatic activity attributed to these CI␣ isoforms (27).
PKA Activation Is Sufficient to Induce Morphological Differentiation of LNCaP Cells-We have established that treatment of LNCaP cells with ␤-adrenergic receptor agonists or Fsk increases intracellular cAMP levels (19) and PKA activity (Fig.   1). To determine whether expression of activated PKA catalytic subunit was sufficient to induce acquisition of NE characteristics in LNCaP cells, we compared the morphology of cells cotransfected with empty vector or CI␣-or Cqr-expressing plasmids along with the GFP reporter plasmid. Three days after transfection, cells were fixed and stained for the FLAG epitope. When either the Cqr-or CI␣-containing plasmid was transfected into LNCaP cells along with the GFP reporter plasmid at a 5:1 molar ratio, greater than 95% of the GFP-positive cells were also positive for the FLAG epitope. Transient transfection of LNCaP cells routinely resulted in ϳ30% of the cells in the  ; lanes 2, 4, and 6) for 10 min. The FLAG-tagged kinases were immunoprecipitated with M2 anti-FLAG antibody and subjected to in vitro kinase assays using malantide as a substrate. Relative PKA kinase activity (pmol of phosphate transferred to malantide min Ϫ1 arbitrary unit of immunoprecipitated FLAG-tagged kinase Ϫ1 ) was normalized to the activity from untreated, CI␣-transfected cells (1.2 pmol of phosphate transferred to malantide min Ϫ1) . The corresponding immunoblots using anti-PKA CI␣ antibody (upper panel) or anti-FLAG antibody (lower panel) were performed as described under "Experimental Procedures." population being transfected. These observations allowed us to compare the morphology of PKA subunit-transfected cells with both the untransfected cells and cells cotransfected with GFP along with empty pRSV1 vector.
Cqr-transfected cells exhibited a distinct NE morphology, including the elaboration of long branched processes (Fig. 4, D and E) terminated with well defined growth cones (Fig. 4, G  and H), while the surrounding untransfected cells and cells transfected with CI␣ (Fig. 4, A-C, F, and I) exhibited morphologies typical of untreated LNCaP cells. Greater than 80% of the Cqr-transfected cells exhibited a NE morphology (cells with processes greater than two cell bodies (40 m) in length). This was significantly greater than the less than 20% of the CI␣-or empty vector-transfected cells determined to exhibit such morphological characteristics (Fig. 5A). These results indicate that activation of PKA alone is sufficient to induce the NE differentiation morphology in LNCaP cells.
PKA Activation Is Sufficient to Induce Mitotic Arrest in LN-CaP Cells-We previously showed that Fsk-and Epi-induced acquisition of NE characteristics included the cessation of mitotic activity. We therefore assessed the mitotic activity of the PKA catalytic subunit-transfected cells cultured for 3 days in T-Media containing 5% FBS, serum-free and phenol red-free RPMI 1640 (S-F), or phenol red-free RPMI 1640 containing 5% charcoal-stripped FBS (CS-FBS) to characterize the extent to which the different transfections affected the LNCaP phenotype. Under all conditions tested, the rate of BrdUrd incorporation observed in the Cqr-transfected cells was significantly reduced as compared with that of the CI␣-transfected cells (Fig.  5B) with the exception of the case discussed below. Furthermore, the rate of BrdUrd incorporation observed in CI␣-transfected cells was indistinguishable from that of the untransfected cells (data not shown). Approximately 48% of the CI␣transfected cells grown in FBS were positive for BrdUrd at the end of the 20-h labeling indicating a doubling time of about 2 days. This is in excellent agreement with the previously reported doubling time for LNCaP cells (26,33) and indicates that the doubling time of transfected cells had not been influenced by expression of the wild type CI␣ subunit.
The level of BrdUrd incorporation observed in Cqr-transfected cells cultured in FBS was comparable with the level of inhibition of [ 3 H]thymidine incorporation observed in LNCaP cells treated with Bt 2 cAMP, Fsk, or Epi (19) and indistinguishable from that observed in CI␣-transfected cells cultured under S-F conditions (Fig. 5B). This latter condition has been reported to induce growth arrest and eventual acquisition of NE characteristics in LNCaP cells (20). These results indicate that, even under optimal growth conditions, expression of Cqr is sufficient to reduce the mitotic activity level of transfected cells to that observed for untransfected or CI␣-transfected cells under S-F conditions. Additionally, while the rate of BrdUrd incorporation observed in CI␣-expressing cells under charcoalstripped FBS conditions was significantly greater than that of cells under S-F conditions, the BrdUrd incorporation rate of the Cqr-expressing cells under the same conditions were indistinguishable (Fig. 5B). These results indicate that while androgen withdrawal had a modest impact on the mitotic activity of LNCaP cells, it significantly increased the susceptibility of LNCaP cells to PKA activation by initiating cell cycle arrest.
PKA Activation Is Sufficient to Induce NSE Expression in LNCaP Cells-To further characterize the extent to which activation of PKA affected the LNCaP phenotype, we assessed the expression of NSE in cells transfected with the FLAGtagged PKA catalytic subunit vectors (Fig. 6). For comparison, NSE levels were measured from lysates of LNCaP cells transfected with a GFP expression plasmid and cultured for 2 days in the presence or absence of Bt 2 cAMP and from lyastes of bovine adrenomedullary chromaffin cells. As compared with untreated GFP-transfected cells, NSE expression was ϳ5-fold higher in CI␣-transfected cells and ϳ60-fold higher in Cqrtransfected cells. NSE expression in Cqr-transfected cells was ϳ35% of that detected in Bt 2 cAMP-treated GFP-transfected cells. To compare the NSE levels in cells transfected with CI␣ or Cqr, the NSE/FLAG ratio of CI␣-transfected cells was nor- malized and compared with the NSE/FLAG ratio of Cqr-transfected cells for five independent experiments using the pairwise difference described under "Experimental Procedures." By this analysis, Cqr-transfected cells expressed significantly more NSE than CI␣-transfected cells (-fold change (95% confidence interval) ϭ 8.5 (4.8, 15.2)). These results are in agreement with the morphologic and mitotic activity data and indicate that expression of PKA is sufficient to induce acquisition of a NE phenotype by LNCaP cells.
PKA Activation Is Required for ␤-Adrenergic Receptor-and Fsk-mediated Acquisition of NE Characteristics-To determine if PKA activation is required for Fsk or ␤-adrenergic receptor agonists to mediate acquisition of NE characteristics, LNCaP cells were transfected with pMTdnRI␣ in which expression of a dominant-negative mutant of the PKA regulatory subunit, RI␣, is induced in response to treatment with Cd/Zn (Fig. 2). This construct contains mutations at site A (G200E) and site B (G324D, R332H), which render it unresponsive to cAMP and maintain the catalytic subunit in an inactive conformation even in the presence of increased intracellular cAMP levels (28). LNCaP cells were cotransfected with pMTdnRI␣ or pcDNA3 as a control plasmid along with the GFP reporter plasmid. Transfected cells were cultured in medium with or without Cd/Zn, in the presence or absence of Fsk for 3 days before assessing the morphology of the transfected and untransfected cells (Fig. 7). In the absence of Cd/Zn, pMTdnRI␣transfected, Fsk-stimulated LNCaP cells acquired a NE morphology indistinguishable from that of the untransfected cells (Fig. 7, A and B). However, in the presence of Cd/Zn, pMTd-nRI␣-transfected, Fsk-stimulated cells retained the typically fusiform LNCaP morphology, while the surrounding cells acquired a NE morphology (Fig. 7, C-F).
The effect of RI␣ expression on acquisition of a NE phenotype in response to induction by Fsk, Epi, and Isop was quantitatively assessed for three independent sets of transfected and treated cells (Fig. 8). Because of the relatively high endogenous levels of PKA regulatory subunit in LNCaP cells, we were unable to determine the fold induction of RI␣ expression in the transfected cells. However, treatment of RI␣-transfected cells with Cd/Zn significantly inhibited the ability of Isop-, Epi-or Fsk-stimulated cells to acquire a NE morphology. These results suggest that PKA signaling is required for LNCaP cells to acquire NE characteristics in response to ␤-adrenergic agonists and Fsk, further supporting the idea that activated PKA is involved in controlling cAMP-mediated NE differentiation in prostatic adenocarcinomas. DISCUSSION NE cells are found in essentially all prostatic adenocarcinomas, and since these cells produce factors capable of affecting the proliferative and metastatic state of prostate tumor cells, the potential role of NE differentiation in prostatic cancer progression is of interest. However, the lack of a consensus concerning the prognostic value of tumor NE status clearly indicates that a more accurate assessment of the potential significance of NE differentiation in the progression of prostatic . Bovine adrenomedullary chromaffin cell lysate (Chrom., lane 5; 100 g) was utilized as a positive control for NSE expression. Statistical significance of the pairwise difference for the ratio of NSE expressed per unit of FLAG-tagged PKA catalytic subunit expressed from CI␣ to Cqr was averaged from five independent experiments and is described under "Results." disease is needed.
In this report, we used the LNCaP cell line to demonstrate directly that activated PKA plays an essential role in mediating the acquisition of NE characteristics induced by ␤-adrener-gic receptor agonists and activators of adenylate cyclase. This conclusion is based on our demonstration 1) that these agents induce activation of PKA kinase activity, 2) that ectopic expression of an activated isoform of PKA catalytic subunit is sufficient to initiate acquisition of NE characteristics by these cells, and 3) that inhibition of PKA signaling by expressing a dominant-negative isoform of the PKA regulatory subunit is capable of inhibiting the ability of ␤-adrenergic receptor agonists and activators of adenylate cyclase to induce NE differentiation. We have measured acquisition of a neuritic morphology and inhibition of mitotic activity because these were the most readily quantifiable measures of NE differentiation applicable to the single-cell assays utilized here, and because these changes in morphology and mitotic activity were completely coincident with the expression of the NE markers, NSE (Fig. 6), serotonin, parathyroid hormone-related peptide, and neurotensin (19). These results, coupled with our previous report (19) indicating that cAMP-mediated acquisition of NE characteristics in LN-CaP cells is completely reversible, such that maintenance of the NE phenotype requires the continual presence of differentiating agents, implicate PKA as a central player in mediating NE differentiation in adenocarcinomas.
The prostate has been demonstrated to be a rich source of ␣ 1and ␤ 2 -adrenergic receptors with both adrenergic and cholinergic innervation into the glandular epithelia (34). Moreover, ␤-adrenergic receptor agonists have been implicated in the maintenance of prostate homeostasis and as mitogens for prostatic stromal cells (35). Perhaps the requirement for continual cAMP signaling for the maintenance of a NE phenotype in LNCaP cells reflects a normal requirement for ␤-adrenergic signaling for proper prostatic function.
A potentially strong selective advantage is provided to tumors that, in response to factors either present (such as catecholamines) or absent (such as androgens) in their environment, can reversibly generate cells that serve as endocrine/ paracrine sources and are themselves nonproliferative and androgen-independent. Our observations indicating that PKA activation initiates acquisition of NE characteristics and that these effects are accentuated under conditions of androgen withdrawal (Fig. 5B) suggest that the LNCaP cell line provides just such a model.
The use of LNCaP cell lines expressing wild type and mutant isoforms of components of the PKA signaling pathway should provide systems to critically evaluate the outstanding issues concerning control of NE differentiation and the possible influence NE cells have on prostate cancer progression.