Identification of a novel 2-oxindole fluorinated derivative as in vivo antitumor agent for prostate cancer acting via AMPK activation

The key metabolic sensor adenosine monophosphate-dependent kinase (AMPK) has emerged as a promising therapeutic target for cancer prevention and treatment. Besides its role in energy homeostasis, AMPK blocks cell cycle, regulates autophagy and suppresses the anabolic processes required for rapid cell growth. AMPK is especially relevant in prostate cancer in which activation of lipogenic pathways correlate with tumor progression and aggressiveness. This study reports the discovery of a new series of 2-oxindole derivatives whose AMPK modulatory ability, as well as the antitumoral profile in prostate cancer cells, was evaluated. One of the assayed compounds, compound 8c, notably activated AMPK in cultured PC-3, DU145 and LNCaP prostate cancer cells. Likewise, compound 8c caused PC-3, DU145 and LNCaP cells viability inhibition. Selective knocking down of α1 or α2 isoforms as well as in vitro assays using human recombinant α1β1γ1 or α2β1γ1 AMPK isoforms revealed that compound 8c exhibit preference for AMPKα1. Consistent with efficacy at the cellular level, compound 8c was potent in suppressing the growth of PC-3 xenograft tumors. In conclusion, our results show that a new 2-oxindole fluorinated derivative exerts potent in vivo antitumor actions against prostate cancer cells, indicating a promising clinical therapeutic strategy for the treatment of androgen-independent prostate cancer.


Results
Chemistry. The 2-oxindole derivatives were prepared following the synthetic route depicted in Fig. 1B.
Compounds 6a-6c were synthesized by condensation of commercially available 2-oxindole derivatives (4 and 5) with different aromatic aldehydes 30 . Reduction of 6a-6c was achieved with sodium borohydride in methanol to result 7a-7c in 37-64% yield. The subsequent fluorination reaction placed a fluorine substituent at C-3 position on the 2-oxindole skeleton, which is easily performed using compounds 7 a-c with Selectfluor, as electrophilic fluorine source, to yield the novel compounds 8 a-c 31,32 . Finally, N-alkylation using commercially available alkylation agents was achieved in anhydrous DMF using NaH as a base to led the final derivatives 9a-9b with moderated yields 33 .

Evaluation on AMPK activity in prostate cancer cells.
To investigate the effect of the compounds on AMPK activity at cellular level, compounds 8a-c and 9a-b were selected to perform AMPK phosphorylation assays in human prostate cancer cell lines. We use the PC-3 cell line as it is androgen-resistant and represents the advanced and aggressive stage of prostate cancer. We found that compounds 8a, 9a and 8b slightly, although significantly, inhibited AMPK phosphorylation compared to controls and compound 9b exerted a weak AMPK activation. In contrast, compound 8c caused a significant increase in Thr172 phosphorylation of AMPK compared to control cells ( Fig. 2A). AMPK inhibits lipogenic pathway through phosphorylation of its well-known substrate acetyl-CoA carboxylase (ACC). To confirm the modulation of AMPK activity in prostate cells by the compounds, we determined levels of phosphorylated ACC in 2-oxindole fluorinated derivatives-treated cells. As shown in Fig. 2, levels of pACC correlated to the levels of pAMPK. Whereas compounds 8a, 9a and 8b inhibited ACC phosphorylation, compounds 8c and 9b, according to their increased AMPK activation, augmented pACC. It is worthy to note that the highest ACC phosphorylation was obtained with compound 8c denoting a powerful activity of this compound in PC-3 cells.
To further confirm the effect of the compounds in prostate cells, we used the androgen-insensitive cell line DU145 in which LKB is inactive and the androgen-sensitive prostate cancer LNCaP cell line. Similar results were obtained although AMPK stimulation by compound 8c was not so strong ( Fig. 2A). The fact that compound 8c stimulate AMPK in DU145 cells suggest that LKB1 is not necessary for 8c effect and points to a direct effect of 8c on AMPK.
To corroborate the activation of AMPK by compound 8c, it was compared with the two well-recognized AMPK activators AMP and A-769662 which bind to the regulatory subunits and allosterically activate the kinase activity of the AMPKα 34 . As shown in Fig. 2B, compound 8c induced AMPK and ACC phosphorylation comparable to AMP or A-769662 (Fig. 2B).

Effect of 2-oxindole fluorinated derivatives on prostate cancer cells proliferation.
We then study the effect of the compounds on the proliferative capacity of PC-3, DU145 and LNCaP prostate cells. We also used the prostate cell lines PTN2 and RWPE-1 which are derived from a histologically normal adult and therefore are considered as normal prostate cells, for comparison. All the compounds were tested at concentrations ranging from 1 to 100 μM. We found that compounds 9a and 8c effectively inhibited cell proliferation in the range of 1-100 µM in a dose-dependent manner in the five cell lines (Fig. 3A), although compound 8c exhibited a clear antiproliferative effect. Notably, compound 8c exhibited a greater efficacy and potency against the cancer cell lines LNCaP, DU145 and PC-3 compared to normal cells (Fig. 3A).
In fact, the highest potency of this compound is observed in the androgen-resistant PC-3 cells, which represent the most aggressive form of prostate cancer. IC50 values for compound 8c were 18 μM, 19 μM, 30 μM, 65 μM, and >1000 μM in PC-3, DU145, LNCaP, PNT2 andRWPE-1, cells, respectively. As shown in Table 1, at 50 μM compound 8c was the most efficacious in all cell lines displaying a higher antiproliferative effect against the cancer cell lines PC-3, DU145 and LNCaP. Nevertheless, at 100 µM, compound 8c almost totally inhibited cell viability in the five cell lines probably due to a high cytotoxic effect (Fig. 3A). These data indicate that the highest inhibitory effect correlates with the greater activation of AMPK in prostate cancer cells and prompted us to continue the study with the compound 8c, (structure in Fig. 1C) being selected for further studies.
To confirm the antiproliferative effect of compound 8c, cell viability was evaluated by cell counting at different times in PC-3, DU145 and LNCaP cells. As observed in Fig. 3B, 8c induced a time-dependent strong cytotoxic effect in the three cell lines although it was more pronounced in PC-3 cells. β-tubulin (β-Tub) serves as a loading control. (B) PC-3, DU145 and LNCaP cells were treated with 25 µM of compound 8c, 25 µM of AMP or 25 µM of A-769662 for 1 hour and processed as above. Lower panels, densitometric analyses of bands represented as the mean of the ratio pAMPK/AMPK or pACC/ACC ± SD of three different experiments. *p < 0.05 compared with the controls by one-way ANOVA followed by Fisher's LSD test; # p < 0.05 compared with 2-oxindole fluorinated derivates or with the AMPK activators AMP or A769662 by one-way ANOVA followed by Tukey test.
The inhibitory efficacy of 8c in prostate cells was contrasted with that of AMP and A-769662 by cell counting and results indicate that compound 8c is very potent and effective especially in the PC-3 cell line (Fig. 3B).
Upon demonstration that compound 8c strongly inhibited prostate cancer cells growth, we investigated its effect on cell cycle. For that, we treated PC-3 and LNCaP cells with 25 μM and 50 μM of compound 8c for 24 h and performed cell cycle analysis by flow cytometry using Propidium Iodide (PI). As shown in Fig. 4 an increase in Values were normalized with the control and represent the mean ± SD of at least three independent experiments. *p < 0.05 and **p < 0.01 compared to control by the Student's -test.
subG0 cells was detected in cells treated with the compound 8c, fact that indicates apoptosis. It is worthy to note that when PC-3 cells were treated with compound 8c, 80% of cells were in subG0 phase (Fig. 4A).
There are many intracellular mechanisms whereby AMPK activation inhibits cell growth. Recent findings reveal that one possible mechanism is the downregulation of Cyclin D1 and cell cycle arrest [35][36][37][38] . Cyclin D1 senses and integrates extracellular signals and intracellular changes of cells in G1 phase and therefore its modulation reflects intracellular status. To further investigate the mechanism underpinning the cell viability inhibitory effect of compound 8c, levels of Cyclin D1 were determined by Western blot in PC-3 and LNCaP cells. As shown in Fig. 4B, a marked decrease in Cyclin D1 level was observed in 8c-treated cells. The decrease was prevented by incubation with the AMPK inhibitor Dorsomorphin (also known as Compound C), which has been shown to inhibit AMPK on prostate cells 39 , pointing to an AMPK-dependent effect (Fig. 4B).
Cyclin D1 activity is modulated by several inhibitors such as p21 which negatively regulate cell cycle progression by interacting with cyclin dependent kinases (cdks). P21 protein plays a crucial role in the signaling pathway of the tumor suppressor gene p53, which induces growth arrest and apoptosis when DNA is damaged. To further study the role of compound 8c in cell cycle, we determined the expression of p21, p53 and the levels of cleaved caspase 3 in the prostate cells. P53 was only determined in LNCaP cells since PC-3 cells are p53-negative 40 . When cells were treated with 8c, a remarkably increase in p21 protein was observed both in PC-3 and in LNCaP cells (Fig. 4B). In LNCaP cells, and increase in the total and phosphorylated forms of p53 was also observed. These changes correlated with an increase in cleaved caspase 3 (17 kDa fragment) indicating a cell cycle arrest and apoptosis induction by compound 8c in prostate cells (Fig. 4B).

Characterization of AMPK activation by compound 8c.
To further confirm the compound 8c-induced activation of AMPK, PC-3 and LNCaP cells were incubated with the compound for 1 h and 24 hours in the presence of the AMPK inhibitor Dorsomorphin. As shown in Fig. 5A, activation of AMPK by compound 8c was also seen at 24 hours, being even more robust that at 1 hour of treatment. Dorsomorphin blocked AMPK activation by compound 8c both at 1 h and 24 h. Accordingly, ACC phosphorylation was increased at 1 h and 24 h and prevented by Dorsomorphin in both cell lines. AMPK inhibits mTORC1 through phosphorylation of Raptor on two well-conserved serine residues (ser722 and ser792) inducing cell-cycle arrest in response to an energy stress. To further support the 8c-induced AMPK activation, we determined Raptor phosphorylation on Ser792 in prostate cells. An increase in p-Raptor was observed in PC-3 and LNCaP cells treated with 8c which was partially prevented by Dorsomorphin (Fig. 5A). It has been demonstrated that AMPK activation suppresses the hypoxia-inducible factor 1 A (HIF-1A) accumulation induced by hypoxic stress, counteracting the inflammatory response and the HIF-1A-promoted tumorigenesis. As shown in Fig. 5A, the expression of HIF-1A was decreased in cells treated with compound 8c. These results reinforce the notion of the 8c antitumoral role through AMPK activation in prostate cells.
To analyze whether AMPK was involved in the antiproliferative effect of compound 8c, AMPK pharmacological inhibition as well as AMPK knocking down were carried out and cell viability was assayed. Figure 5B shows that the decrease in cell viability induced by compound 8c in PC-3 and LNCaP cells was reduced in cells treated with the AMPK inhibitor dorsomorphin. Likewise, AMPK silencing by specific siRNA, prevented 8c-induced cell death, further supporting the involvement of AMPK in the antiproliferative effect of compound 8c.
The androgen receptor (AR) plays a key role in prostate cancer progression and, as a result, targeting of AR signaling is a major therapeutic strategy. Recent findings reveal that AR is a potential downstream mediator of AMPK in LNCaP cells and that AMPK activation inhibited AR activity and translocation to the nucleus 41 Therefore we analyzed the expression of the androgen receptor as well as of its two well-recognized targets prostate specific antigen (PSA) and prostate-specific membrane antigen (PSMA) in the androgen-sensitive LNCaP cell line. Results in Fig. 5C indicate that compound 8c markedly reduced the expression of AR and to a lesser extent those of PSA and PSMA. These data all together indicate that compound 8c inhibits prostate cell growth by an AMPK activation mechanism.
AMPK is a heterotrimeric protein composed by catalytic (α) and regulatory (β and γ) subunits. In mammals, different isoforms of α, β and γ subunits can associate to produce up to 12 AMPK different combinations 42 . It is worthy to note that the most abundant form of AMPK α subunit present in PC-3 cells is α1 (AMPKα1-containig complexes are at least 10-fold more abundant than AMPKα2) 39 . To gain insight into the mechanism underlying AMPK activation by compound 8c in prostate cells, we selectively silenced AMPKα1 or AMPKα2 in prostate cancer cells. We observed that when AMPKα1 subunit was knocked down, 8c-induced AMPK phosphorylation was almost totally blocked, and ACC phosphorylation notably decreased (Fig. 6). In contrast, when the expression of AMPKα2 subunit was silenced compound 8c still was able to phosphorylate AMPK and ACC at a similar level than in non-silenced (siC) cells (Fig. 6). These findings suggest that compound 8c needs α1 subunit to exert its effect and therefore preferably activates α1 subunit-containing AMPK isoforms. As a further proof of AMPK activation by compound 8c, we performed in vitro AMPK activity assays using human recombinant α 1 β 1 γ 1 and α 2 β 1 γ 1 AMPK isoform. This assay evaluated the ability of the compound to activate AMPK and in turn phosphorylate the synthetic substrate peptide AMARAASAAALARRR. Figure 7A shows that compound 8c increases AMPK α 1 β 1 γ 1 activity in 75%, p < 0.001 whereas no activation was observed when AMPK α 2 β 1 γ 1 isoform was used. This is an interesting finding that indicates a preference of compound 8c for α 1 isoform. However, it should be noted that this is a partial AMPK activation compared with AMP or the Abbott  AMPK activator A-769662 which induce a higher response (Fig. 7A). We also assayed the effect of compound 8c on a fragment of the α 1 isoform (aa 1-550) that contains the catalytic domain. Results show that in this case there is no change in AMPK activity (Fig. 7B), inferring that compound 8c does not bind to the catalytic site.
AMPK may be allosterically activated by AMP, which binds at up to three sites on γ subunit, or by synthetic drugs like A-769662, PF-06409577 and salicylate that directly bind and induce conformation changes in the AMPK complex enhancing protection of the activation loop against dephosphorylation 43,44 . The A-769662-binding site is located at the interface of the α-kinase domain and the β subunit and is different from the AMP binding sites 42,43,45 . Then, we analyzed AMPK α 1 β 1 γ 1 activity when A-769662 and 8c compounds were added together. As shown in Fig. 7C, an additive effect of 8c and A-769662 on AMPKα 1 β 1 γ 1 activation was only observed at a submaximal A-769662 dose (0.3 µM) when the system is not saturated. At 30 μM or 3 μM A-769662 a maximal effect on AMPKα 1 β 1 γ 1 activation was reached and the addition of 8c did not increase this effect. No synergy was observed at the doses used (Fig. 7C).
We then tested the selectivity of compound 8c to a panel of 9 kinases. Among the 9 kinases analyzed, compound 8c only showed a slight inhibitory activity on PDK1 (K I = 3.1 × 10 −5 M) and PKA K I = 3.5 × 10 −5 M) at the low micromolar range (Supplementary Table 1).
These results indicate that compound 8c selectively activates AMPK and inhibits prostate cells proliferation and hence it was a good candidate to prove his effectiveness in vivo.
In silico drug profile type evaluation of compound 8c. Previous to the in vivo study, we evaluated the drug-like profile of compound 8c. One of the most common tools in the early stages of drug development is to determine the drug-like properties, because it allows making decisions based on fundamental criteria, such as permeability, oral absorption or passage of blood-brain barrier. In order to evaluate the drug-like profile of compound 8c and to determine its potential as future drug candidate, we employed computational predictive methods. Pharmacokinetic properties of compound 8c were predicted using the QikProp 3.5 program, integrated in Maestro (Schrödinger LLC, New York, USA). As presented in Table 2, the predicted properties of 8c are within the ranges predicted by QikProp for 95% of known oral drugs. These data indicate that fluorinated oxindole 8c presents an adequate druggability profile and therefore is suitable to use in vivo.  with 5 mg/Kg of compound 8c. In vehicle-treated mice, an increase in tumor growth volume was observed during the course of the study and at the end of experimentation (Fig. 8A). Nevertheless, treatment with compound 8c significantly reduced tumor growth over the 15 days experimental period (Fig. 8A). The volume of the tumor at the end of the treatment was notably smaller in 8c-treated animals than in controls (Fig. 8B). No effect on mice weight or animal welfare was observed in compound 8c-treated mice. Serum biochemical parameters were similar between control and 8c-treated animals and were within the reference range (Supplementary Figure 1). Liver morphology was normal either in control than in mice treated with 8c (not shown). Consistent with our in vitro   findings, compound 8c promoted AMPK activation, increased ACC phosphorylation, decreased Cyclin D1 and decreased procaspase-3 levels in xenograft tumors (Fig. 8C). These data show the novel finding that the 2-oxindole fluorinated derivative compound 8c is effective at inhibiting the growth of prostate tumor xenografts and activating AMPK in vivo and accordingly has a great potential as a candidate to treat prostate cancer.

Discussion
The cellular metabolic sensor AMPK has emerged as a key therapeutic target for many cancer types as it controls the metabolic pathways that are usually reprogrammed in cancer cells. This is especially significant in prostate cancer in which increased lipid synthesis allows the survival of cancer cells and is associated with tumor progression 5 .
In the present study we have synthesized new AMPK modulators and have selected one that exhibits potent antitumoral properties against prostate cancer cells. The small collection of fluorinated derivatives of 2-oxindole were prepared with the aim of contributing to the knowledge of some of the main structural requirements for AMPK modulation by this family of compounds. Phosphorylation assays in cellular cultures show that compounds 8a, 8b and 9a are AMPK inhibitors. Nevertheless, compound 8c shows a cellular profile suitable with an AMPK activator whereas compound 9b slightly increases AMPK phosphorylation in PC-3 cells but not in LNCaP cells. Notably, compound 8c is the only non-substituted compound in which a furan moiety has been introduced as substituent at C-3 position of the 2-oxindole system and is a powerful in vivo activator of AMPK and the best inhibitor of prostate cells growth. This suggests that this chemical feature together with the disposition of the substituent may be playing an important role in the interaction with the enzyme being important to determine the activator profile of the compound. We have identified and pharmacological characterized a new AMPK activator, compound 8c, which exhibits potent antitumoral properties against prostate cancer cells.
The antiproliferative effect of 8c is associated with cell cycle arrest and apoptosis. Notably, the cell cycle assays state that the compound 8c increases subG0 cells and decreases Cyclin D1 levels. These results are in line with recent findings that underscore the importance of AMPK in the regulation of cell cycle 12,35,46 . Moreover, recent data indicate that the AMPK activator metformin inhibited proliferation of prostate cells by promoting cell cycle arrest in the G0-G1 phase, inhibiting cyclin D1 and inducing cell apoptosis 38 , which is in good agreement with our observations According with previous research showing that AMPK induces p53 activation 47 , compound 8c increased the phosphorylation and expression of p53 in prostate cells. In addition, 8c modified the expression of other AMPK-dependent proteins such as p21, HIF-1A, or the androgen receptor and its targets. Our results demonstrate that compound 8c activates AMPK in prostate cells and regulate genes involved in cell survival and growth.
The strong antiproliferative effect of compound 8c was dependent on AMPK as suggested by the Dorsomorphin prevention and AMPK knocking down. Despite Dorsomorphin may exert unspecific AMPK-independent effects 48-50 , in our model, it inhibits AMPK activation by compound 8c (Fig. 4A) and thus indicates that the 8c-induced cell death was mediated, at least in part, by AMPK activation. This notion was confirmed by AMPK silencing, which blocked 8c-induced cell death.
At cellular level, compound 8c shows a profile suitable with an AMPK activator. This fact prompted us to investigate whether 8c was acting on specific AMPK isoform since prostate cells primarily express α1 isoform. Knocking down of selective α isoforms and in vitro experiments show that compound 8c is acting through α1-containig complexes.
There are however, several unanswered questions regarding the exact mechanism whereby 8c activates AMPK in prostate cells. Assays with recombinant AMPK showed that compound 8c had additive activating effect with the Abbot AMPK activator A-769662. At dose analyzed, no synergistic effect was observed, suggesting that both compounds might bind to the same site. A-769662-binding site, termed the allosteric drug and metabolite (ADaM) binding site, is located at the interface of the α-kinase domain small lobe and a surface of the carbohydrate-binding module (β-CBM) opposite to the glycogen-binding site, as recently revealed by crystal structures of AMPK 42,43,45 . Further studies are necessary to better characterize the binding site of compound 8c.
Finally, the antitumor activity of compound 8c was further reinforced by the repression of xenograft tumors growth in vivo at a lower dose than that previously used for AMPK activators in prostate cells 12 . Moreover, in 8c-treated tumors an increase in AMPK phosphorylation, ACC phosphorylation and a decrease in Cyclin D1 and procaspase-3was observed showing that this compound can activate AMPK also in vivo. In line with or results, Zadra et al. 12 showed that the novel AMPK direct activator MT63-78 inhibited LNCaP prostate tumor growth in vivo and promoted AMPK activation in xenograft tumors.
Overall, these findings suggest that the AMPK act as a tumor suppressor in prostate cancer and therefore its activation might be a quite promising therapeutic strategy to fight against castration-resistant prostate cancer. The results shown here indicated that compound 8c is very powerful in activating AMPK and inhibiting prostate tumor growth and accordingly is a promising candidate with therapeutic value for prostate cancer. NMR spectra were recorded on a Bruker 300 (300 and 75 MHz) at 25 °C. Samples were prepared as solutions in deuterated solvent and referenced to an internal non-deuterated solvent peak. Chemical shifts are expressed in parts per million (δ) downfield of tetramethylsilane. Melting points were determined on an MP70 Mettler Toledo apparatus. (6a-c). To a solution of 2-oxindole (1 eq.) in MeOH (25 mg/mL), the corresponding aldehyde (1 eq.) was added. The reaction mixture was heated at reflux for 3 h. After cooling at 0 °C, the product was isolated by filtration.

General procedures for 3-(arylmethyl)-1-benzyl-3-fluoro-2-oxindoles (9a-9b).
To a stirred solution of the corresponding 3-(arylmethyl)-3-fluoro-2-oxindole (1 eq.) in anhydrous DMF (12 mg/mL), NaH (1.1 eq.) was added at rt. After 10 min., benzyl bromide (1 eq.) was added and the reaction mixture was stirred for 15 min at rt. Finally, brine (10 mL) was added under ice-cooling and the resulting precipitate was filtered off and purified by flash column chromatography using mixtures of ethyl acetate:hexane (3:1) to give compounds 9a-9b. In vitro AMPK enzymatic assay. The effect of compounds in the in vitro human recombinant AMPK activity assay was measured by monitoring phosphorylation of the "AMARA" peptide using Delfia technology. Briefly, AMPK enzyme activities were carried out in 96 well microtiter plates (containing 50 mM Hepes buffer, pH 7.4 with 125 µM ATP and 19 mM MgCl2) in the presence of the human recombinant AMPKα 2 β 1 γ 1 or AMPKα 1 β 1 γ 1 proteins (Invitrogen, Carlsbad, CA, USA), 1 mM synthetic peptide substrate (AMARAASAAALARRR, the "AMARA" peptide), and the compounds. Final concentration of DMSO was 1%. Reactions were initiated by the addition of AMPK (0.1U). Following mixing, the plates were incubated for 30 min at room temperature. Enzyme activity was assayed by using an anti-phospho-serine antibody to measure the quantity of phosphate incorporated into the AMARAA peptide.

(±) (R, S)-3-((1H-pyrrol-2-yl)methyl)-1-benzyl-3-fluoroindolin-2-one (9a).
Protein Kinase Assay. A radiometric protein kinase assay (33PanQinase ® Activity Assay) was used for measuring the kinase activity of the 10 protein kinases according to the manufacturer's (Biochemical Screening Systems, ProQinase GmbH, Freiburg, Germany). IC50 values were measured by testing 10 concentrations of the compounds in the range from 1 × 10-04 M to 3 × 10-09 M in duplicate in each kinase assay. The assay for all protein kinases contained 70 mM HEPES-NaOH pH 7.5, 3 mM MgCl2, 3 mM MnCl2, 3 μM Na-orthovanadate, 1.2 mM DTT, ATP (variable amounts, corresponding to the apparent ATP-Km of the respective kinase), [γ-33 P]-ATP (9 × 10 05 cpm per well), the target human recombinant kinase (ProQinase GMBH ® , Freiburg, Germany) and substrate. The reaction cocktails were incubated at 30 °C for 60 minutes. The reaction was stopped with 50 μl of 2% (v/v) H 3 PO 4 , plates were aspirated and washed two times with 200 μl 0.9% (w/v) NaCl. Kinase activity dependent transfer of 33 Pi (counting of "cpm") was determined with a microplate scintillation counter (Microbeta, Wallac). All assays were performed with a BeckmanCoulter Biomek 2000/SL robotic system. Cell culture. Human prostate cancer epithelial PC-3, DU145 and LNCaP and human prostate normal RWPE-1 cells were purchased from American Type Culture Collection (ATCC CRL-1435 CRL-1740 and CRL 11609 respectively) (Rockville, MD, USA). The human prostate normal PNT2 cell line was purchased to ECACC (European Collection of Cell cultures, Salisbury, SP4 0JG, UK). Cells were routinely grown in RPMI 1640 medium supplemented with 100 IU/ml penicillin G sodium, 100 µg/ml streptomycin sulfate, 0.25 µg/ml amphotericin B (Invitrogen, Paisley, UK) and 10% fetal calf serum. Cells were used at passages between 10-20. For treatment experiments, cells were plated and grown over night, the medium was then replaced with serum-free RPMI 1640 for 24 hours and then incubated with different doses of the compounds for the indicated times.
Cell viability. Cell viability was determined by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) (Sigma-Aldrich) method. In brief, a total of 5000 cells/well were seeded into 12-well plate in a final volume of 1 mL. After treatments, 20 µl MTT solution (5 mg/ml in PBS) was added to the medium and cells were incubated at 37 °C for 4 h. Then, the supernatant was discarded and isopropanol was added to dissolve the formazan crystals. Treatments were carried out in triplicate. The optical density in each well was evaluated by measurement of absorbance at 490 and 650 nm using a microplate reader (ELX 800 Bio-Tek Intruments, INC). IC50 values were calculated using GraphPad Prism vs 6.0 (GraphPad Software, Inc, La Jolla, CA, USA). Cell viability was also determined by counting viable and dead cells by Trypan blue staining. Trypan blue positive and negative cells were counted using a Countess TM automated cell counter (Invitrogen, Carlsbad, CA, USA). Results were expressed in relation to total number of cells counted. Protein concentrations were measured by BioRad ™ protein assay kit (Richmond, CA, USA). Equal amount of protein (20 μg) from each sample was separated on 8% SDS-PAGE gels and transferred to PVDF membranes (BioRad). The membranes were blocked with blocking buffer (1× TBS, 0.1% Tween 20 with 5% w/v fat-free dry milk), washed and incubated with primary antibodies at 4 °C for 12 h. The membranes were subsequently washed with TBS-T (1× TBS and 0.1% Tween 20) and incubated with horseradish peroxidase-conjugated anti-mouse or anti-rabbit IgG antibodies (Sigma, St Louis, MO, USA) at room temperature for 1 h. The immune complex was visualized with an ECL system (Cell Signaling Technology). Films were analyzed by densitometry. The anti-pAMPKα1-thr172, pACC-ser79, pP53, P21, pRaptor and the antibodies against the corresponding total forms were obtained from Cell Signaling Technology (Danvers, MA, USA). The anti-caspase 3 and anti-procaspase 3 were from Santa Cruz Biotechnology (Dallas, Texas, USA). The anti-HIF1A was from Novus Biologicals LLC (Littleton, CO, USA). The anti-Cyclin D1 antibody was obtained from ThermoFisher (Alcobendas, Madrid, Spain). All the other chemicals were obtained from Sigma (St. Louis, MO, USA).

Cell Cycle analysis.
Knockdown of AMPK α1 AMPK α2. For AMPK α1 and α2 subunits knocking down, LNCaP and PC3 cells were transfected with 5 nM of selective small interfering siRNAs for AMPK α1 or α2 subunits or a negative control siRNA (Ambion-Life Technologies, Carlsbad, CA, USA) using lipofectamine RNAi Max (Life Technologies, Spain) and following the manufacturer's instructions. Control of proteins silencing was carried out by Western blot using AMPK α1 and AMPK α2 specific antibodies (#2795 and #2757, respectively, Cell Signaling Technology, Danvers, MA, USA).
In Silico ADME Calculations. A set of 34 physicochemical descriptors were computed using QikProp ver-  Table 2. The 3D structures used in the calculation of QikProp descriptors were generated using Maestro (version 10.4,) and energy minimizations were carried out using Macromodel (version 9.9). Local minimum energy structures of each compound were used as input for ADME studies with QikProp. were assigned randomly to two groups (n = 8) and subcutaneously injected with compound 8c (5 mg/kg/d, which considering an average animal weight of 28 g correspond to 128 μM), or vehicle (5% ethanol in PBS) for 15 days. Tumors were measured with external calipers and volume was calculated by the modified ellipsoidal formula as volume = length × (width) 2 × 0.52 mm 3 52,53 . At the end of the study, the mice were sacrificed by placing them in a CO2 gas-filled chamber, and the excised tumors were recovered and homogenized in lysis buffer for protein quantification and Western blotting.
Serum Clinical parameters determination. The 9 analytes assayed constitute the routine biochemistry profile. The analyses were performed with an ADVIA 1800 Chemistry System (Siemens Healthcare GmbH, Erlangen, Germany).
Statistical analysis. Statistical analysis of the data was performed by one-way ANOVA test followed by Fisher's least significant difference (LSD) or Tukey multiple comparison test, using the 95% confidence interval. When indicated, the Student's t test was performed for comparison of AMPK activation between 8c-treated and control groups. Statistical was performed using Microsoft Office Excel 2013. Significance levels were defined as P < 0.05 (*,#) and P < 0.01 (**,##). All graphs were drawn using Origin software. Data Availability. The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.