Background
Prostate cancer is one of the leading causes of cancer in men worldwide. Although the growth of both normal prostate epithelial cells and cancer cells is dependent on the presence of androgens, chemical or surgical androgen depletion therapy is the mainstay of treatment for metastatic prostate disease. However, in many patients an androgen-independent (castration-resistant) form of prostate cancer develops within 18–24 months. Castration-resistant prostate cancer (CRPC) is currently an incurable stage with poor prognosis [
1]. During attempts to find new treatment modalities for CRPC it has been hypothesized that one of the events contributing to the development of anti-androgen resistance is neuroendocrine transdifferentiation (NED) of prostate cancer cells (summarized in [
2]). NED thus serves as an example of one of the many levels of tumor heterogeneity and cancer cell plasticity that in general represent a challenging issue for effective clinical diagnosis and therapy [
3].
In prostate carcinoma there is an increase in the number of cells with neuroendocrine-like properties over time. Because normal NE cells are thought to be post-mitotic [
4], it is proposed that new cells with NE-like properties originate through the process of NED from pre-existing epithelial cancer cells [
5]. Such cancer cells acquire a NE-like phenotype—they are able to secrete several neuropeptides and are androgen-independent. It was shown that NED can be induced
in vitro by various stimuli, such as androgen depletion [
6,
7], increased levels of interleukin-6 (IL-6) [
8], activation of Wnt [
9] and EGF [
10] signaling pathways, activation of the cyclic adenosine 3′, 5′-monophosphate (cAMP) signaling pathway [
11‐
13], or ionizing radiation [
14,
15]. In addition, several genes and transcription factors were shown to be involved in NED, for example protocadherin-PC and the transcription factors Foxa2 and NeuroD1 (summarized in [
2]).
Androgen depletion, which induces NED, is associated with cell cycle arrest in G1 phase [
16,
17]. This cell cycle arrest is linked to modulation of well-known cell cycle regulators involved in G1 phase progression and the G1 to S phase transition [
16,
18]. Another mechanism that contributes to cell cycle arrest is the phenomenon of contact inhibition. High-density cultivation is associated with arrest in G1 phase that is accompanied by decreased Cdk2 and Cdk4 activity, even in cancer cells that are refractory to the typical contact inhibition exhibited by normal cells. Furthermore, cell density can also influence intracellular signaling, as shown by density-dependent changes in intra- and extra-cellular distribution of cAMP [
19].
In the present study, we focused on the role of cell cycle modulation in the regulation of NED in prostate cancer cells. We showed that androgen depletion and cell cycle modulation mediated by high cell density both promoted NED, which was demonstrated by increased expression of characteristic markers both in AR-positive and AR-negative prostate epithelial cell lines of different origin. We identified an important role of Cdk1 and Cdk2 activity in promoting NED by cell cycle attenuation. Finally, our results suggest a role of cAMP signaling activation in NED promotion by high cell density in AR-positive prostate cancer cell lines. Taken together, our data identify a novel condition leading to the promotion of NED in prostate cancer cells and define specific molecular mechanisms that determine this process.
Discussion
To our knowledge, this is the first demonstration that the plasticity of prostate cancer cells enables promotion of NED by modulation of the cell cycle in conditions of high cell density. It has been shown that NED can be induced predominantly by androgen depletion [
6,
7], but also by a wide variety of other stimuli, including IL-6 [
8], Wnt(s) [
9], EGF [
10], or cAMP signaling [
11,
12,
27]. Androgen deprivation therapy (ADT) represents a standard treatment for advanced prostate cancer [
36]. The action of androgens is predominantly mediated through AR and its co-activators, which have been shown to be critical regulators of the G1 to S transition in prostate cancer cells [
16,
17]. The important role of cyclin D1 and cyclin D3 in this process was demonstrated by Xu et al. [
37]. However, the link between ADT-induced NED and mechanisms of cell cycle regulation remains unclear. Here, we showed that androgen depletion and high cell density both led to the promotion of NED and modulation of the cell cycle machinery in a very similar manner, although inhibition of AR activity was confirmed only in androgen-depleted conditions and not in the high cell density condition. We demonstrated that Cdk1 and/or Cdk2 inhibition, but not cyclin D1 or D3 down-regulation, is sufficient for NED promotion in the presence of androgens. More importantly, we identified a key role of the cAMP/PKA signaling pathway in NED promoted by high cell density. Our study defines a novel mechanism highlighting how high tumor density could modulate the plasticity of prostate cancer cells and influence disease progression.
In our previous study, we demonstrated that both androgen depletion and high cell density cultivation led to increased expression of cytokeratins, general markers of epithelial differentiation [
38]. Because androgen deprivation is a well-known promoter of NED [
39], we compared the expression of NED markers between prostate cancer cells cultured at high density and cells cultured in the absence of androgens. Surprisingly, androgen depletion and high cell density both promoted increase in expression of several NED markers (γ-enolase, tubulin β-III, and aromatic L-amino acid decarboxylase) in prostate cancer cells. The effect of high density on NED promotion is androgen-independent, since similar regulation of NED markers was observed in C4-2 cells, an androgen-independent sub-line of LNCaP. Moreover, a similar high density-induced increase in the level of NED markers was detected using the AR-negative prostate cell line BPH-1 and its tumorigenic derivative CAFTD03, as well as the cancer cell lines DU-145 and PC3. Furthermore, re-introduction of AR into AR-negative PC3 cells did not have any effect on high cell density-induced NED. This phenotype was also confirmed in a 3D culture system in all above mentioned cell lines. Moreover, we observed a slight trend towards the correlation between decreased proliferation and NED promotion in a small cohort of patients with advanced prostate cancer. This is in contrast to previously published data showing that a high Ki-67 labeling index is weakly associated with high chromogranin A expression (Spearman’s correlation 0.164) [
40]. However, the authors of that study showed no correlation between Ki-67 labeling index and expression of AR or NeuroD1, another neuroendocrine marker. NeuroD1 was previously shown to be expressed in aggressive prostate cancer cell lines and prostate cancer samples, although co-expression with chromogranin A was found only rarely [
41]. Bubendorf et al. [
42] did not found significant association between Ki-67 and neuroendocrine differentiation while Grobholz et al. [
43] observed higher proliferation index (assessed by Ki-67) in tumors with large clusters of NE differentiation in comparison to negative tumors or with solitary NE cells. With respect to our observations (both in-vitro and in our small patient cohort) and varying results in the literature, we may state that the neuroendocrine transdifferentiation may be present in tumors with slower proliferation in some patients.
Because the correlation of NED with cell cycle arrest had been described in several experimental models [
35,
44], we next focused on modulation of the cell cycle machinery. We showed that the induction of cell cycle arrest by androgen depletion is associated with down-regulation of cyclin D1 protein in LNCaP cells but not in LAPC-4 cells, and with down-regulation of cyclin D3 in both cell lines. Similar effects were observed when NED was promoted by high cell density. These results are in accordance with previously published data showing that androgen-induced proliferation of prostate cancer cells is accompanied by increased levels of D-type cyclins [
37] and, conversely, androgen deprivation causes down-regulation of protein levels of cyclin D1 and cyclin D3 in LNCaP cells [
16,
37]. RNAi-mediated down-regulation of cyclin D1 in LNCaP cells and cyclin D3 in LAPC-4 cells led to modulation of the cell cycle and an increased percentage of cells in the G0/G1 phase; however, this was not associated with the promotion of NED. Thus, we conclude that although the down-regulation of D-type cyclins leads to cell cycle arrest in the G0/G1 phase, it is not sufficient for promoting NED in AR-positive prostate cancer cells.
Next, we focused on the role of the more general cell cycle regulator Cdk2. Both the activity and expression of Cdk2 were down-regulated during NED promotion by androgen depletion and high cell density, and this down-regulation correlated with up-regulation of the Cdk inhibitor p27
Kip1 in both of our models. It was previously shown that Cdk2 expression and activity together with Rb protein phosphorylation, are regulated by androgens [
16,
45]. Generally, inhibition of Cdk activity causes cell cycle arrest and inhibits proliferation of prostate cancer cells. Our results showed that inhibition of Cdk2 causes cell cycle attenuation, in particular accumulation in the G2/M phase in LAPC-4 cells; this arrest is associated with significant promotion of NED. These results, which indicate a functional role of the inhibition of Cdk2 activity in the regulation of NED, are supported by the findings of other investigators. For example, it has been reported that silibinin-induced NED in LNCaP cells is also associated with cell cycle arrest and decreased Cdk2 levels [
46]. We did not observe a significant change in the expression of NED markers when Cdk2 expression was reduced using a siRNA-mediated approach (data not shown). However, transfection of Cdk2 siRNA in combination with Cdk1 siRNA resulted in a slight trend towards NED promotion. This observation is in agreement with results obtained using CVT-313 inhibitor, which at the applied dose preferentially inhibits Cdk2, but might also partially inhibit Cdk1 [
34]. To reveal the role of Cdk1 in NED promotion we used selective inhibitors. Treatment with subtoxic concentrations of the Cdk1-specific inhibitor CGP 74514 only slightly modulated the cell cycle, whereas RO-3306 caused accumulation of cells in G2/M phase, in accordance with previously published data [
47]. This was accompanied by increased expression of NED markers at the protein and mRNA level in LNCaP cells but not in LAPC-4 cells, which only showed up-regulation of γ-enolase expression at the protein level. Further studies of the mechanisms by which Cdk1 and Cdk2 are involved in the plasticity of prostate cancer cells are necessary based on the fact that different approaches to the modulation of their expression and activity were not uniformly reflected in terms of NED promotion. Since experiments for elucidating the involvement of Cdk1 and Cdk2 in promoting NED were performed only in AR-positive prostate cancer cell lines, investigating the AR-negative cell lines might shed more light in proposed role of Cdk1 and Cdk2 deregulation in promoting NED. Our observations suggest that the association between NED and the cell cycle, and the role of particular regulators of cell cycle machinery is more complex and also cell type-dependent. The clinical potential of pharmacological inhibition of Cdk activity in cancer therapy has been demonstrated in several studies (for review see [
48]). However, based on our observations, it is important to consider the possible effects of Cdk1 and Cdk2 inhibition in NED promotion in prostate cancer cells.
It has previously been shown that NED can be induced by physiological and pharmacological agents that elevate intracellular cAMP levels [
27]. Treatment of prostate cancer cells with cAMP leads to changes in the expression of Hox genes located at the HOXD locus [
13] including the Neuro D1 transcription factor, which is expressed in malignant NE cells [
41]. Moreover, other downstream targets of cAMP, PKA [
11] and CREB [
14], are directly involved in NED. Furthermore, promotion of NED by cAMP-inducing agents is a reversible process. Interestingly, it has been shown that the cAMP level can be modulated by cell density [
19]. Based on these facts, we hypothesized that the promotion of NED by high cell density can be mediated by the activation of cAMP signaling. Our results demonstrated that cAMP signaling is indeed activated in response to high cell density, as demonstrated by increased levels of downstream target molecules of cAMP, such as phosphorylated PKA regulatory subunit II and phosphorylated CREB. Interestingly, cAMP inhibits proliferation of breast cancer cells via increased expression of p27
Kip1 and decreased activity of Cdk2 [
49]. These observations are indirectly in accordance with our observation, since in response to high cell density we observed increased activation of cAMP-mediated signaling, increased p27
Kip1, and decreased Cdk2 expression. Moreover, it was shown that cAMP inhibits Cdk2 activity and Rb phosphorylation in adipose stem cells [
50]. More importantly, the functional involvement of cAMP was confirmed by the demonstration that treatment of prostate cancer cells with MDL-12330A, a potent inhibitor of adenylate cyclase, abolished the promotion of NED by high cell density. In summary, these experiment support our hypothesis that activation of cAMP signaling mediates NED promotion by high cell density in AR-positive prostate cancer cell lines (Figure
5F). Based on our results we conclude that modulation of the cell cycle by high cell density can promote reversible NED in prostate epithelial cancer cells. Our study also suggests that prostate cancer tissue remodeling, in association with disease progression or therapy, might contribute to tumor progression by modulating the plasticity of cancer cells and by promoting NED.
Materials and Methods
Cell culture and treatment
LNCaP cells (DSMZ) [
51] (androgen-sensitive cell line carrying mutation in the gene encoding AR [
52]), and LAPC-4 cells (androgen-dependent cell line carrying WT gene encoding AR [
53]) were cultivated as described previously [
38]. Culture conditions for the androgen-independent subline C4-2 [
54] were similar to those for the parental LNCaP cell line. Under experimental conditions (high-density NED promotion, siRNA transfection, treatment with inhibitors), LNCaP cells were cultivated with 5% FBS or 5% dextran/charcoal-stripped FBS (CS, for androgen depletion), and LAPC-4 cells were cultivated with 10% FBS and 1 nM R1881, or with 10% CS. BPH-1 cells [
55] and the BPH-1 tumorigenic clone CAFTD03 [
56] were cultivated as described previously [
57]. PC3 (ATCC) and PC3 cells stably expressing AR [
30] were cultivated in F12 with 10% FBS and penicillin and streptomycin. DU-145 cells (ATCC) were cultivated in RPMI 1640 with 10% FBS and penicillin and streptomycin. AmpFLSTR® Identifiler® PCR Amplification Kit (Life Technologies) was used to verify the origin of cell lines.
To evoke NED, LNCaP and LAPC-4 cells were cultivated as follows: cells were seeded at a density of 20,000/cm2 in the appropriate complete medium with FBS (day -1). After 24 hours, the medium was exchanged for medium with FBS or CS (day 0). Cells were continuously cultivated for 2 to 16 days without splitting, but with exchange of the medium for fresh medium twice a week. Cells were collected for further analysis on days 2, 4, 8, and 16 after the change of medium on day 0.
For cultivation in 3D conditions, we used Alvetex® polystyrene scaffold inserts in 6-well plates (AVP004), 12-well plates (AVP002), or 24-well plates (AVP006) containing 200 μm thick Alvetex polystyrene scaffold (Reinnervate). Cells were seeded at a density of 0.5 × 106, 1.0 × 106, and 1.5 × 106 cells per insert and cultivated for 72 to 96 hours with regular media exchanges. Cells that were seeded on standard Petri dishes in standard media and at standard seeding densities and cultivated for either 1 day or 4 days were used as a 2D control. Experiments in 3D were performed with two independent repetitions.
For the inhibition of Cdk2 activity we used a selective ATP-competitive Cdk2 inhibitor III [CVT-313, 2 (bis-(Hydroxyethyl) amino)-6-(4-methoxybenzylamino)-9-isopropyl-purine)] (#238803 Merck). For inhibition of Cdk1 activity we used the Cdk1 inhibitor CGP 74514A [N-(cis-2-Aminocyclohexyl) -N-(3-chlorophenyl)-9-ethyl-9H-purine-2,6-diamine, #217696, Calbiochem] and ATP-competitive Cdk1 Inhibitor IV RO-3306 [(5Z)-2-((Thiophen-2-yl)methylamino)-5-((quinolin-6-yl) methylene) thiazol-4(5H)-one, #217699, Calbiochem], both dissolved in DMSO. For inhibition of cAMP signaling we used an adenylate cyclase-specific inhibitor MDL-12330A hydrochloride [N- (cis- 2-phenyl-cyclopentyl) azacyclotridecan-2-imine-hydrochloride, M-182, Sigma-Aldrich]. For all treatments, LNCaP and LAPC-4 cells were seeded at a density of 20,000 or 30,000 cells/cm2 in appropriate media (IMDM + 10% FBS + 1 nM R1881 + antibiotics for LAPC-4, RPMI + 5% FBS + antibiotics for LNCaP cells). After 48 hours, cells were treated with the indicated concentrations of selected inhibitors and control cells were treated with the equivalent concentration of dMSO (not exceeding 0.1%). Cells were collected for further analysis 48 hours after treatment. All experiments were performed at least twice with technical duplicates.
Cell cycle analysis
Cells were fixed, stained, and analyzed by flow cytometry using FACSCalibur™ or BD FACSVerse (Becton Dickinson) as described previously [
33]. At least two independent repetitions were performed for each experiment.
Cell transfection and RNA interference
LNCaP and LAPC-4 cells were transfected with small inhibitory RNA (siRNA) duplexes (Santa Cruz Biotechnology) directed against non-targeting control (sc-37007), Cdk1 siRNA (sc-29252), and Cdk2 siRNA (sc-156139) using the Neon® Transfection System (Life Technologies). Transfection was performed in a 10-μl tip according to the manufacturer’s recommendations. Cells were harvested 48 hours after transfection for further analysis. Experiments were performed in three independent repetitions.
RNA isolation and real-time reverse transcription polymerase chain reaction (qRT-PCR)
Total RNA was isolated using High Pure RNA Isolation Kit (Roche). PCR was performed using the One Step SYBR® PrimeScript™ RT-PCR Kit II (Perfect Real Time) according to the manufacturer’s recommendations on a RotorGene 6000 (Corbett Research) [
33]. The sequences of the primers used are listed in Additional file
7: Table S1-A. Changes in gene expression were calculated using the comparative threshold cycle method, with POLR2A as a normalizing gene [
58]. Data from at least three experiments were normalized for each gene using the mean C
T value for the control sample (2 days of incubation with FBS, control siRNA, vehicle-treated cells, or cells harvested at day 0). Alternatively, two-step qRT-PCR was performed. Up to 1 μg of isolated RNA was reverse transcribed to cDNA with the High Capacity RNA-to-cDNA Kit (Applied Biosystems). qRT-PCR was performed on a Light Cycler 480 (Roche) using the Light Cycler 480 Master Mix in combination with Human Universal Probe Library (Roche). Primer and probe combinations used in assays are listed in Additional file
7: Table S1B. Results for genes of interest were normalized to the housekeeping gene POLR2a assessed using Light Cycler 480 software and are presented as
. Experiments were performed in at least two independent repetitions or in technical duplicates.
Electrophoresis and western blotting
Collected cell pellets were lysed and the protein extracts were separated and blotted as described previously [
33]. The primary and secondary antibodies used are listed in Additional file
8: Table S2. Detection of α-tubulin and β-actin served as a control of equal loading. All western blots are presented as typical results of at least two independent repetitions. Densitometry analyses were performed using ImageJ software (NIH). Values given below a particular band represent normalized results of densitometry analysis of the given image (integrated density for the particular band was assessed and all values were normalized to the control of equal loading).
Immunofluorescence microscopy
Cells were cultivated, fixed, permeabilized, and stained as described previously [
38]. The primary and secondary antibodies used are listed in Additional file
8: Table S2. Fluorescence images of the cells were obtained using a confocal microscope (TSC SP5X, Leica Microsystems).
Statistical analysis
Statistical analysis was performed using STATISTICA for Windows software (StatSoft). When the data variance was homogenous, one-way analysis of variance followed by the Fisher or Tukey range test was used. If the data variance was non-homogenous, the Mann-Whitney U-test was performed.
Supplementary Material and Methods can be found in Additional file
9.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
ZP carried out experiments, analyzed results and wrote the manuscript. JB and GK provided IHC and data analysis. MK handled selection of patient samples for IHC analysis. ES, RF, TS and ŠŠ carried out expression analysis and analyzed data. JJ performed automatic image analysis of mRNA FISH. AK revised the manuscript. KS carried out particular flow cytometry analyses, supervised the experimental work, participated in data analysis and interpretation of results, and wrote the manuscript. All authors read and approved the manuscript.