Early Cellular Responses of Prostate Carcinoma Cells to Sepantronium Bromide (YM155) Involve Suppression of mTORC1 by AMPK

The imidazolium compound YM155, first discovered as a potent inhibitor of Survivin, effectively kills many carcinomas in preclinical models. However, the upstream signaling mechanism triggered by YM155 remains unclear. Here we studied early signaling responses in vitro in prostate and renal cancer cell lines in a dose-dependent manner. We found that YM155 rapidly activates the retinoblastoma protein, correlating with the loss of expression of all three Cyclin Ds. Using Western blot, various selective chemical inhibitors and q-PCR, we show that YM155-mediated decrease in protein levels of Cyclin Ds, Survivin and Mcl-1 is independent of transcription or proteasomal control mechanisms. Moreover, we provide the first evidence that YM155 changes the phosphorylation status of known mTOR-target proteins involved in translational control, namely ribosomal protein S6 (rS6) and 4E-BP1. Our data support that YM155 achieves this by blocking mTORC1 via the phosphorylation of Raptor at S792 through activated AMPKα (T172). Furthermore, we also used a polysome profile, supporting that YM155 markedly suppresses cap-dependent translation of mRNAs which include Survivin, Cyclin D1 and Mcl-1. We provide the first evidence that YM155 functions as a potent activator of AMPKα, a robust suppressor of mTORC1 and an attenuator of global protein synthesis.

www.nature.com/scientificreports www.nature.com/scientificreports/ YM155 concentrates in human tumor xenografts 7 . Owing to the highly positive outcome of these preclinical studies, YM155 quickly entered phase I and II clinical trials. In phase I studies, YM155 was well tolerated when administered as a continuous intravenous infusion 12,13 . In Phase II trials, YM155 showed moderate activity as a single agent on CRPC, NSCLC, and stage III and IV melanoma [14][15][16] .
Some preclinical tumor studies more recently demonstrate markedly improved efficacy (or synergy) of YM155 in killing tumors when combined with other cancer therapeutics [17][18][19][20][21] . What remains a bottleneck in exploring its full clinical potential is an understanding of YM155's early biological responses and direct molecular target(s) mediating them. The study of the early and late drug responses will allow the rational design of drugs, using YM155 as a lead compound. Here we provide new insight into the roles of mTOR and AMPK on early signals triggered by YM155.

YM155 kills various prostate and kidney cancer cells.
We examined the relative potency of YM155 (72 h treatment) in killing various prostate genitourinary cancer cell lines (Table 1, Supplementary Fig. S1A,B), which included the PC-3 AR-negative bone metastatic human prostate cancer (PCa), the AR-negative DU-145 brain metastatic human PCa cell line, the LNCaP androgen-dependent PCa lymph node metastatic cell line, the castration-refractory PCa variant of LNCaP (C4-2), bone metastatic castrate-resistant variant of C4-2 (C4-2B), castrate-resistant PCa cell line derived from a clinically localized tumor (CWR22Rv1), normal human prostate epithelial cells (HPEpiC), immortalized prostate epithelial lines (RWPE-1, RWPE-2), and renal cell carcinoma cell lines (RCC4, 786-O) ( Table 1). Among the CRPC cell lines tested, the PC-3 line was the most sensitive to YM155, with an IC 50 of 1.84 nM, while the androgen-dependent prostate cancer cell line, LNCaP, was the least sensitive (IC 50 = 44 nM). The CRPC (C4-2) and metastatic (C4-2B) variant of the LNCaP line had intermediate levels of sensitivity to YM155 ( Table 1). The immortalized and ras-transformed human prostate epithelial cell lines RWPE-1 and RWPE-2 were half-maximally growth inhibited at about 25 nM and 29 nM YM155, respectively. YM155 also had potent growth inhibitory activity on renal clear cell carcinoma line RCC4 (IC 50 = 1.5 nM) and was less effective on renal clear cell carcinoma line 786-O (IC 50 = 16 nM). YM155 also inhibited the growth of prostate epithelial cells (HPEpiC) but with an IC 50 of 99.6 nM (Fig. S1B). Overall, YM155 effectively kills or inhibits the growth of aggressive PCa cells in the low nM range, and the IC 50 for growth inhibition is generally lower for the more aggressive PCa lines.

YM155 suppresses Survivin expression and activates Rb in prostate and kidney cancer cells.
The cationic charge of YM155 resulting in a low cLogP (−2.008; partition coefficient) contributes to YM155's poor cell permeability [22][23][24][25] . Cationic transporters are required for cellular uptake of YM155 21,22 . This transport delays the timing of intracellular drug uptake relative to membrane-permeable drugs. Minematsu et al. 22 studied the cellular uptake of [ 14 C]YM155 and used the Michaelis-Menten Equation to derive the V max (14.93 pmol/min/ mg protein) and K m (0.243 μM) for uptake of YM155 by PC-3 cells. Based on those values we calculated that an intracellular concentration of 3 nM YM155 (IC 50 of YM155 for killing PC-3 cells in 3 days) is achieved in ~6 h at the K m of YM155. Our data revealed that 100 nM YM155 (41% K m ) rapidly (1-2 h) activated the Rb protein (as demonstrated by dephosphorylation of Rb at S807/811) in PC-3 cells, which temporally preceded the loss of Survivin protein at 4-8 h (Fig. 1A, left panel). We observed similar effects in YM155-treated DU-145 cells (Fig. 1A, right panel).
We thus studied those alterations in PC-3 and RCC4 cells at various doses of YM155 (1-100 nM) after 8 h and 24 h of YM155 treatment (Fig. 1B). Previous studies in our laboratory showed that Survivin expression is functionally linked to the inactivation retinoblastoma proteins 26,27 . We first focused on this potential link in the mechanism of YM155 action. We found that activation of Rb and loss of Survivin had the same or similar dose-response curves at both times, and such changes occurred within their IC 50  www.nature.com/scientificreports www.nature.com/scientificreports/ (Table 1). These data are consistent with the possibility that YM155 drives the loss of Survivin expression by activating Rb.
Thus, to study early responses of YM155 we used 100 nM YM155 for our subsequent initial studies. This higher dose decreased the lag time of the cationic pump-dependent YM155 uptake.
YM155 promotes rapid loss of Cyclin Ds. We next explored the mechanism by which YM155 may dephosphorylate Rb. The Cyclin D-CDK complexes are key enzymes that inactivate Rb through directly phosphorylating Rb and other pocket proteins, thereby releasing the E2F factors that initiate the transcription of several genes required for the entry of cells into the S phase of the cell cycle 28 . A common mechanism of such regulation is through changes in cyclin D levels. Thus, we measured changes in levels of Cyclin Ds in prostate and kidney cancer cells following YM155 treatment. We observed that YM155 (100 nM) caused loss of Cyclin D1 as early as 1 h, and Cyclin D2 as early as 30 min of treatment in the PC-3 and DU-145 cells ( Fig. 2A). In the PC-3 and RCC4 cells, Cyclin D1 and D2 expression were decreased by 8 to 24 h with as low as 3 nM YM155, a concentration within its IC 50 for cell death after 3 days (Fig. 2B). We performed a time course of those changes with 3 nM YM155 on PC-3 cells and found YM155 decreased Cyclin D1 and D2 expression following 2 h of treatment (Fig. 2C).
We subsequently examined whether loss of Cyclin Ds by YM155 was driven by a transcriptional versus a post-transcriptional mechanism. We assessed mRNA levels of Cyclin Ds by both semi-quantitative (Fig. 2D) and real-time quantitative RT-PCR (Fig. 2E) following treatment of PC-3 cells with YM155. YM155 did not alter Cyclin D mRNA levels under conditions that caused robust loss of Cyclin D proteins (examined side-by-side), indicating that YM155 suppresses the expression of Cyclin Ds mainly at the protein level. We tested this further by treating PC-3 cells with sufficient concentrations of either the transcriptional inhibitor actinomycin D (Act D) or the translational inhibitor cycloheximide (CHX) that entirely block transcription versus translation at the elongation step before YM155 treatment. Such abrogation for 4 h of YM155 treatment did not appreciably reduce the levels of Cyclin D2 or affect the ability of YM155 to suppress Cyclin D2 (Fig. 2F), supporting transcriptional responses on Cyclin D2 cannot explain the loss of Cyclin D2 protein. In contrast, CHX effectively suppressed levels of Cyclin D2 and dampened the response to YM155. These data lend further support that YM155 does not suppress the expression of Cyclin D2 through a transcriptional mechanism or message stability. www.nature.com/scientificreports www.nature.com/scientificreports/ www.nature.com/scientificreports www.nature.com/scientificreports/ We next tested whether YM155 promoted loss of Cyclin Ds through activating its proteasomal degradation. Pretreatment of PC-3 cells with 5 μM proteasomal inhibitor MG132 did not rescue cells from the suppressive effects of YM155 (Fig. 2G), although it substantially enhanced the basal levels of those Cyclins (evidence that MG132 was effective in blocking proteasomal degradation). Doubling the dose of MG132 in the above experiment failed to rescue the loss of Cyclins by YM155 (Fig. 2G), suggesting proteasomal degradation does not play a major role in the ability of YM155 to drive their downregulation. In another experiment, we re-evaluated the effect of CHX side-by-side with MG132 on the expression of Cyclin Ds (Fig. 3H). Under those conditions, our data support that YM155 suppresses the expression of Cyclin Ds in the absence of MG132, but not when combined with CHX. These data suggest that YM155 may promote loss of Cyclin Ds, particularly Cyclin D1, mainly by inhibiting their translation.
Despite G1 growth arrest signals, cell cycle analysis of PC-3 cells at 24 h with 3 nM and 10 nM YM155 revealed an increased sub-G1 and S phase ( Supplementary Fig. S2), consistent with bypassing G1, arresting in S phase and undergoing cell death.

YM155 suppresses the expression of Survivin through a transcriptional mechanism. Cyclin Ds
have been shown to regulate Survivin through a transcriptional mechanism 26,27,29 . However, our semi-quantitative RT-PCR (Fig. 3A,B) and real-time Q-PCR (Fig. 3C), each of which was done side-by-side, showed that YM155 did not alter Survivin mRNA levels. In all of these experiments, we found that doses and treatment times with YM155 that suppressed Survivin protein levels did not suppress Survivin mRNA levels, as confirmed by biological replicates and two different sets of primers (see Material and Methods). Consistent with the above results, we www.nature.com/scientificreports www.nature.com/scientificreports/ show YM155 inhibited the expression of Survivin in either control or Act D-treated cells. However, YM155 did not further suppress Survivin levels over that inhibited by CHX alone (Fig. 3D). These results suggest that YM155 blocks the expression of Survivin protein independent of the Survivin promoter.
We next tested whether YM155 promoted loss of Survivin expression through activating its proteasomal degradation. PC-3 cells were pre-treated with MG132 for 30 min before a standard 4 h treatment with YM155 and changes in Survivin expression were assessed by Western blot (Fig. 3E). MG132 afforded no significant reversal of Survivin loss by YM155, although MG132 robustly elevated basal levels of Survivin as we expected.

YM155 suppresses the expression of the anti-apoptotic protein Mcl-1 through a non-transcriptional mechanism.
Mcl-1, an anti-apoptotic member of the Bcl-2 family, has been reported to be robustly suppressed by YM155 8,9,30 . While a transcriptional mechanism of Mcl-1 control by YM155 has been previously proposed 8,9 , the mechanism of such control remains poorly understood. In our time course experiments, we found that treatment of PC-3 cells with 100 nM YM155 can suppress the expression of Mcl-1 by 4 h, with significant suppression at 3 nM and 10 nM YM155 by 8 h and 24 h, respectively (Fig. 4A). YM155 also suppressed the expression of Mcl-1 in DU-145, RCC4 (Fig. 4B), with dose-response patterns comparable to their respective IC 50 s for cell death (Table 1).
We next examined the mechanism by which YM155 suppresses Mcl-1 using PC-3 cells. We assessed levels of Mcl-1 mRNA by semi-quantitative RT-PCR and real-time Q-PCR in PC-3 cells at various times and doses of YM155 www.nature.com/scientificreports www.nature.com/scientificreports/ treatment that promoted Mcl-1 protein loss, as conducted side-by-side. YM155 did not suppress levels of Mcl-1 mRNA (Fig. 4C,D), as confirmed with biological replicates and two different sets of primers (in Material and Methods), while parallel cultures treated identically responded to YM155 with loss of Mcl-1 protein. These data suggest that YM155 suppresses the expression of Mcl-1 through a non-transcriptional mechanism at those early treatment times.
We next examined whether YM155 promoted loss of Mcl-1 expression through activating its proteasomal degradation. PC-3 cells were pre-treated for 30 min with MG132 before a standard 4 h treatment with YM155, and changes in Mcl-1 expression were assessed by Western blot (Fig. 4E). MG132 (~ 4-fold) stabilized levels of Mcl-1 and appeared to at best minimally block (<30%) its suppression by YM155. To further test the relative roles of the proteasomal and translational controls in the mechanism by which YM155 suppresses Mcl-1, we examined the effect of YM155 on translational control by suppressing protein synthesis at the elongation step with CHX in the absence of proteasomal degradation. Under those conditions, the effect of YM155 on the suppression of Mcl-1 was efficiently voided by CHX in the presence of MG132 (Fig. 4F). These data suggest that YM155 also promotes loss of Mcl-1 by inhibiting its translation. Collectively, these data are consistent with the possibility that YM155 drives loss of Mcl-1 through translational control.
YM155 rapidly suppresses the phosphorylation of the ribosomal protein S6. To examine whether YM155 can suppress the basic, translational machinery common to Survivin, Cyclin Ds, and Mcl-1, we tested the impact of YM155 on the phosphorylation of ribosomal protein S6 (rS6), which plays a vital role in translational control 31 . We found that YM155 rapidly (1-2 h) suppressed the phosphorylation of rS6 at S235/236 in PC-3 cells, with an almost complete loss at 2 to 4 h (Fig. 5A), without suppressing total rS6. Although p70-rS6-kinase (S6K1) phosphorylates rS6 at S235/236, S240 and S244, the ERK and MAPK mitogen-activated kinases can also phosphorylate rS6 at S235/236 [32][33][34] . Thus, we assessed phosphorylation of S240 with a mouse monoclonal antibody as an extra measure of specificity for activation by S6K1. Thus, our data support that YM155 suppresses activation of S6K1. Basal and YM155-suppressed levels of P-rS6 were not affected by Act D or CHX (Fig. 5B), consistent with a protein kinase or phosphatase response. Consistent with this hypothesis, the serine/threonine phosphatase inhibitor calyculin A, which blocks PP1, blocked the dephosphorylation of P-S6 by YM155, but not the ability of YM155 to suppress the expression of Cyclin D2 and Mcl-1 (Fig. 5C). Intriguingly, calyculin A robustly reversed the ability of YM155 to inhibit the expression of Survivin. This phosphatase inhibitor effectively promoted loss of Cyclin D2 and Mcl-1 protein. These results suggest that calyculin A's ability to reverse YM155's action on Survivin likely requires a serine/threonine phosphatase(s).
YM155 suppresses mTORC1 activity. Although our results in Fig. 5 suggested that YM155 inhibited phosphorylation of rS6 through a potential PP1 phosphatase-dependent mechanism 35 , it left open the likelihood that YM155 suppresses the activity of S6K1 (the primary kinase that phosphorylates rS6 at S240 33 ). Western blot analysis confirmed that brief treatment of PC-3 cells with YM155 blocked phosphorylated S6K1 at T389 (Fig. 6A), supporting that YM155 suppresses the phosphorylation of S6K1, suggesting it is related to S6K1's activity 36 .
The major kinase shown to phosphorylate S6K1 at T389 is mTOR in complex 1 (mTORC1) 37 , which is comprised of mTOR and Raptor 38 . We, therefore, examined whether YM155 suppressed mTORC1 kinase by targetting mTOR or Raptor, by Western analysis of phosphorylation sites on mTOR and Raptor (reported to be critical to mTORC1 function) 39 . Treatment of PC-3 cells with YM155 did not alter the levels of total mTOR, although it slightly suppressed P-mTOR (S2448) (Fig. 6A). Of note, studies support that S6K1 rather than Akt is the kinase that phosphorylates mTOR at S2448 [40][41][42] . This suppressive effect was relatively small compared to the robust suppression of S6K1 phosphorylation reported in the literature, suggesting a cell line discrepancy. To test the basis for this discrepancy, we examined the effect of the classical mTOR inhibitor rapamycin side-by-side with that of YM155 in PC-3 cells. Rapamycin, which effectively blocked phosphorylation of rS6 at S240, had a similar effect relative to YM155 on suppressing phosphorylation of mTOR at S2448 (Fig. 6B). This suggests that the limited loss of mTOR phosphorylation at S2448 is perhaps a cell type difference, likely limited by access of the phosphatase that normally removes the phosphorylated residues on mTOR at S2448.
Treatment of PC-3 cells with YM155 induced the phosphorylation of Raptor at S792 (Fig. 6C), while it did not alter the levels of total Raptor. Phosphorylation of Raptor at S792 monitors AMPK activity and AMPK-mediated suppression of mTORC1 functions specifically during energetic stress 39 . In contrast to suppressing mTORC1, our data support that YM155 may possibly activate the mTORC2 complex, as demonstrated by loss of the putative inhibitory phosphorylation site (T1135) on Rictor (Fig. 6C) 43 , an obligate component of the mTORC2 complex 44 . However, the suppression of Rictor activity by its phosphorylation at T1135 generally seems modest and not entirely reproducible 45,46 . Consistent with a potential activation of mTORC2, we found YM155 also slightly increased the phosphorylation of its substrate, Akt (S473) (Fig. 6D), a reported target of mTORC2 44 . YM155 does not seem to significantly alter the levels of total Akt or P-Akt (T308). The putative effect of YM155 on mTORC2 activation is consistent with the potential inactivation of S6K1, as this kinase is known to phosphorylate Rictor 43 .
Consistent with YM155's inhibition of mTORC1, we found that YM155 also caused loss of the phosphorylation of 4E-BP1 (Fig. 6E), a suppressor of the eukaryotic translation initiation factor eIF4E 47 . Probing the blot for total 4E-BP1, we identified that YM155 affects all four of 4E-BP1's phospho-variants, most prominently it suppresses phosphorylation of S65. This corresponds to the most massive migrating form, the γ-phosphorylation one 48 When mTOR is active, phosphorylation of eIF4E by active mTOR kinase causes dissociation of 4E-BP1 from eIF4F, thus enabling the formation of a translationally competent eIF4F complex 49 .

Impact of YM155 on mRNA translation.
To determine the effect of YM155 on protein synthesis, we treated PC-3 cells with 100 nM of YM155 for 4 h and analyzed global translation by running polysome profiles. The ratio of polysomes to monosomes (5.2 in control and 1.4 in YM155 treatment, respectively) was decreased dramatically by YM155 treatment (Fig. 7A,B, upper panels), which indicated a repressive effect of YM155 on www.nature.com/scientificreports www.nature.com/scientificreports/ mRNA translation. This is also shown by the observation that YM155 treatment shifted the accumulation of the 28S and 18S ribosomal subunits from the polyribosomes to the free ribosomal pool (Fig. 7A,B, lower panels). The association of Survivin, Cyclin D1 and Mcl-1 mRNAs with the free 80S pool and ribosomes were measured. YM155 treatment caused a severe shift of the Cyclin D1 mRNA from the heavy polyribosomes to the light polyribosomes and the free ribosomal pool (71% decrease in the heavy polysomes, a 2.2-fold increase in the light polysomes, and 6.1-fold increase in the free pool) (Fig. 7C). Translation of the Mcl-1 mRNA was similarly inhibited (39% decrease in the heavy polysomes, a 3.2-fold increase in the light polysomes, and 2.3-fold increase in the free pool (Fig. 7C), although to a lesser extent in the 4 h treatment. The survivin mRNA shifted from both types of polysomes to the free pool (70% decrease in the heavy polysomes, 32% decrease in the light polysomes, and a 4.6-fold increase in the free pool) (Fig. 7C). These data are in agreement with the decrease of the protein levels for these mRNAs (Figs 1, 2, 4). Specifically, Mcl-1 which has a half-life of >30 min 50 , decreased >75% at the levels at 4 h of YM155 treatment (Fig. 4). We conclude that in agreement with the negative regulation of mTORC1 by YM155, this drug has inhibitory effects on protein synthesis and translation of mRNAs that encode for proteins promoting the survival of the prostate carcinoma cells. www.nature.com/scientificreports www.nature.com/scientificreports/ Role of AMPK in the suppression of mTOR by YM155. We next examined the mechanism by which YM155 phosphorylates Raptor. Because AMPK is well-established to phosphorylate Raptor at S792 39 , we tested if YM155 can activate AMPK. AMPK is a heterotrimer comprised of a catalytic subunit of AMPKα and two regulatory subunits, AMPKβ and AMPKγ 51 . An ideal marker of activated AMPK is the phosphorylated AMPKα at T172 52 , which we detected using a phospho-specific antibody by Western blot. YM155 (100 nM)-treated cells better separation of high MW proteins. Analysis of P-4E-BP1(S65) in (E) was both through a 4-12% gradient SDS-PAGE (gel 1) and a 15% non-gradient SDS-PAGE (gel 2 Each group of blots represents data obtained from a single gel and experiment, except for P-mTOR and mTOR in (A,C,E) and P-Raptor, Raptor, P-Rictor, Rictor, P-4E-BP1, 4E-BP1 (same sample, protein concentrations, Ponceau S staining, but used different blots to avoid loss of immunoreactivity following blot stripping. β-Actin or GAPDH of the lower blots were used as loading controls). Unless indicated on the figure, each group of blots represents data obtained from a single gel and experiment. Results shown are representative of 2 to 3 independent experiments. www.nature.com/scientificreports www.nature.com/scientificreports/ caused rapid (within 10 to 30 min) phosphorylation of AMPKα at T172 (Fig. 8A). This suggested that YM155 activates AMPK. To confirm that AMPK is active, we examined its ability to phosphorylate its known target, ULK1 at S555 53 , which upon activation is a key inducer of the pathway leading to autophagy. As expected, we found that YM155 promoted the phosphorylation of ULK1 at S555 (Fig. 8A), and the time course of such phosphorylation matched that of the activation of AMPKα (Fig. 8A). Moreover, we showed the other AMPK target protein, Raptor, gets phosphorylated at S792 with similar rapid kinetics as the activation of ULK1 (Fig. 8A). These data, therefore, further support that YM155 rapidly activates AMPK, but such activation may not be direct. However, to the best of our knowledge, the only kinase known to phosphorylate Raptor at S792 is AMPK. Thus, the near temporal coincidence of the phosphorylation of Raptor at S792 within 10 to 30 min of YM155 addition suggests that AMPK directly phosphorylates Raptor, thereby inhibiting mTORC1. To test whether YM155's ability to active AMPK was limited to a high dose of this drug, we conducted a dose-response study of YM155 on the activation AMPK and suppression of its downstream targets (Fig. 8B). Our results support that 3 nM of YM155 was sufficient to effectively activate AMPK within 4 h treatment, which is within the killing/cytostatic activity of this agent.
We next used Compound C, a selective chemical inhibitor of AMPK 54 , to test our hypothesis that YM155 inhibits mTORC1 through activating AMPK. Focusing on a direct downstream target of mTORC1, we found that pre-treatment of cells with Compound C effectively reversed the dephosphorylation of S6K1 by YM155 (Fig. 8C). These data support our model that YM155 suppresses mTORC1 activity by activating AMPK. www.nature.com/scientificreports www.nature.com/scientificreports/ www.nature.com/scientificreports www.nature.com/scientificreports/ YM155 differentially affects the activities of AMPKα1 and AMPKα2. There are two isoforms of AMPKα: AMPKα1 and AMPKα2. Each is expressed in normal prostate and prostate cancer tissues, and have unique functions in prostate cancer, either tumor suppressive or oncogenic [55][56][57][58] . We, therefore, assessed the role of each of these isoforms in the activity of YM155 by transducing PC-3 cells with lentiviral particles to deliver shRNA for silencing each of those subunits. We achieved >2-fold loss of each isoform, as confirmed by Western blot (Fig. 8D,E). Although silencing AMPKα1 suppressed levels of AMPKα1, it did not effectively block P-AMPKα (T172) activated by treatment with 100 nM YM155 (at 4 h). In contrast, silencing AMPKα2 somewhat enhanced basal and YM155-induced levels of P-AMPKα (Fig. 8D, Supplementary Fig. S3), suggesting that AMPKα1 is the primary AMPKα activated by YM155, and that AMPKα2 functions as a negative regulator of P-AMPKα (T172). This supports our model that YM155 differentially affects each of those two isoforms in PC-3 cells. Intriguingly, although YM155 promoted the phosphorylation of AMPKα, YM155 suppressed the levels of total AMPKα1 and total AMPKα2 (Fig. 8D,E). Moreover, YM155 retarded the migration (through SDS-PAGE) of each isoform, and this change in migration was not observed by another activator of AMPK such as curcumin (Supplementary Fig. S6A). The latter result suggests that YM155 induces an additional modification in AMPKα1 over that by another activator of AMPK. We are not clear what causes loss of AMPKα1 or AMPKα2 levels, as detected by Western blotting. However, such loss is unlikely to be through proteasomal degradation, as MG132 pre-treatment failed to reverse the loss of immunodetectable levels of AMPKα1 (Supplementary Fig. S6B).
We next assessed the effect of silencing AMPKα1 and AMPKα2 on the expression of Survivin and Mcl-1 in response to YM155. Silencing AMPKα1 and AMPKα2 suppressed and enhanced the expression of Survivin, respectively, and silencing each dampened the overall suppression of Survivin by YM155 (Fig. 8E, Supplementary  Fig. S5). However, basal levels of Mcl-1 were modestly suppressed and elevated by silencing AMPKα1 and AMPKα2, respectively (Fig. 8E, Supplementary Fig. S5). We also found that knockdown of AMPKα1 or AMPKα2 in PC-3 cells suppressed the relative levels of growth/cell viability compared to control (Fig. 8F). In fact, the AMPKα1-silenced cells were completely growth arrested and could not be expanded under optimal growth. Figure 8F shows growth of those cells by 5 days of culture normalized to their plating density at day 1. These results support that each AMPKα isoform impacts PC-3 cell growth, although AMPKα1 appears to be more critical. Two other constructs we developed to knockout AMPKα1 in PC-3 cells caused the transduced cells to become non-viable, in contrast to the high viability of the respective controls (data not shown). Thus, in PC-3 cells, activation followed by loss of AMPKα1 likely promotes a strong cytotoxic response to YM155.
Importantly and consistent with our data, Sh1-AMPKα1 did not effectively repress the levels of P-AMPKα1(T172) activated by YM155, and both Sh2-AMPKα2 and Sh3-AMPKα2 enhanced P-AMPKα1(T172) (Supplementary Fig. S3). These results support that effective silencing of AMPKα1 and AMPKα2 required to block YM155's response will repress PCa growth. Thus, a different experimental approach beyond the scope of the current study will likely be required to elucidate mechanistic insight into the role of AMPKs on the action of YM155.

Discussion
The mechanism by which YM155 triggers early signals that mediate cell death remains poorly understood. In this study, we provide new mechanistic insight on the early signaling responses of prostate and renal cancer cells to YM155. Figure 9 illustrates a summary of our model. We present the first evidence that YM155 robustly and rapidly inhibits mTORC1 kinase, a critical regulator of cell growth and metabolism. We provide data consistent with a model in which YM155 suppresses mTORC1 through phosphorylation of Raptor by AMPK. YM155 promotes such activation by promoting the phosphorylation of the catalytic subunit of AMPK complexes, AMPKα1 or AMPKα2 at T172. How YM155 promotes such activation is not clear, but it is likely to involve enhanced targeting of LKB1, the key kinase that phosphorylates AMPKα at T172, particularly in PC-3 cells that lack CAMKK2 (88). Although a previous study reported that YM155 induces autophagy in PC-3 cells 59 , consistent with our data on the activation of AMPK and concomitant phosphorylation of ULK1 (Fig. 9A), our study here provides the first underlying mechanism on this problem.
One of the important roles of mTORC1 is to control the translational of a large fraction of cellular proteins, particularly those involved in growth and survival of cells 49 . It is now well recognized that mRNA translation and energy metabolism are dysregulated in cancer and that such dysregulation is a driver of malignancy 60 . mTORC1 controls the translation initiation of those transcripts mainly by phosphorylating S6K1 and 4E-BP1, which activates and inhibits those proteins, respectively. Sabatini's laboratory conducted global analyses of the rates of mRNA translation in WT and 4E-BP1-knockout MEFs following 2 h of inhibiting mTORC1 with Torin 1 versus vehicle. They identified significant changes in translational control of 4,840 transcripts using transcriptome-scale ribosome profiling 49 . Survivin and Cyclin Ds 1, 2 and 3 were among the transcripts that were translationally suppressed by Torin1 through a 4E-BP1-dependent mechanism (Supplementary Table 1 of 49 ). Other studies, using various cell lines, have also demonstrated that mTORC1 controls the levels of Survivin 61 , Cyclin Ds and Mcl-1 through translational regulation 62,63 . These studies are consistent with our model that YM155 suppresses the expression of those proteins partly or entirely through suppressing their translation. Indeed our data in Fig. 7 supports the translation of Survivin, Cyclin D1 and Mcl-1 are significantly suppressed as early as 4 h of YM155 treatment, which is before any noticeable changes in the levels of their transcripts.
Here we established a new paradigm to explain the mechanism by which YM155 causes robust loss of Survivin and Mcl-1. Previous studies claimed that YM155 was a transcriptional suppressor of the Survivin gene. However, our extensive effort (unpublished work) of analyzing the Survivin promoter led us to question that premise and compelled us to postulate an alternative hypothesis that YM155 is not the immediate target of the Survivin promoter. Our hypothesis is clearly supported by the absence of changes in the levels of Survivin mRNA under conditions in which Survivin protein is robustly lost. We showed that YM155 also rapidly suppresses the levels www.nature.com/scientificreports www.nature.com/scientificreports/ of Mcl-1 through a transcription-independent mechanism since YM155 can suppress Mcl-1 protein without changing its mRNA level.
Although YM155 has been shown to suppress the expression of Survivin mRNA in other systems or conditions, such suppression was relatively small (~50%) compared to a rather robust loss of Survivin protein and appears to require 24 h treatment with high concentrations of YM155 7,11,64 . Though Sp1, ILF3, and p54 nrb have been implicated in YM155's ability to suppress exogenous Survivin promoter constructs 11,65 , their role in suppression of the endogenous Survivin promoter by YM155 is not clear. Plainly, the endogenous promoter functions in the context of chromosomal DNA embedded with chromatin, which is different from exogenous promoters in being chromatin-free.
Even though the potent pro-apoptotic effect of YM155 on cancer cells was ascribed to loss of Survivin 7,66 , YM155 was shown to induce a number of other cytostatic and cytotoxic signals, including inhibition of the phosphorylation of PI3K, ERK1/2, Akt, and STAT3, degradation of EGFR 64 , DNA damage response 67,68 , DNA damage, and oxidative stress 69 . AMPK can be activated by oxidative stress generating peroxides that directly activate ATM, which is the first sensor of the DNA damage response 70 . These data suggest that YM155 may potentially activate AMPK through ATM activated by DNA damage and/or oxidative stress. If so, ATM would need to be activated within 10 min of YM155. Alternatively, due to a potentially unique protein modification of AMPK by YM155 (gel mobility shift), it is possible that the YM155-altered AMPK plays a role in oxidative stress, DNA damage or/and downstream responses.
YM155 has been reported to inactivate Akt through PI3K by blocking upstream receptor tyrosine kinases 64 . However, we do not show any inactivation of Akt after 4 h of YM155 treatment under the conditions where rS6, S6K1, and mTORC1 are suppressed, supporting that YM155 does not block mTORC1 by inactivating Akt or a kinase upstream of Akt. The latter data is consistent with no change in phosphorylation of mTOR at the Akt target site (Fig. 6D).
Our  www.nature.com/scientificreports www.nature.com/scientificreports/ Our data in Table 1 suggests that YM155 is more effective on aggressive prostate cancers compared to less aggressive ones. A good comparison of such response can be achieved with the androgen-dependent LNCaP cell line versus its castrate resistant variant C4-2 along with its castrate-resistant bone metastatic variant C4-2B. Consistent with their IC 50 s for cell death, YM155 more effectively suppressed the expression of Cyclin Ds 1 & 2, Survivin, Mcl-1, and P-S6 levels in C4-2 and C4-2B cells compared with LNCaP cells (unpublished study). Consistent with these results, a recent report demonstrated that androgens enhance the ability of YM155 to kill AR-positive PCa cell lines presumably by inducing the expression of a cationic transporter 21 . Because AR activation promotes the expression of CAMKK2 71 , an AMPKα1 activator, it is possible that the synergistic effect of androgen on the killing of prostate cancer cells by YM155 also or alternatively occurs through enhanced activation of AMPKα by elevated levels of CAMKK2 induced by androgen. Further work is required to explore the role of CAMKK2 in the mechanism by which androgens enhance the ability of YM155 to kill cells and to explore the clinical potential of YM155 and its analogs.
Owing to the fact that PC-3 cells silenced for AMPKα1 are unable to grow (Fig. 8F) and our AMPKα1 knockout clones were non-viable, the function of AMPKα1 in PC-3 cells is consistent with prostate carcinomas 55,56,58 . Although in normal prostate, AMPKα proteins function as tumor suppressors, they have oncogenic functions in prostate carcinoma [55][56][57][58] . Data mining from cBioportal and Oncomine support overexpression of AMPKα1 and AMPKα2 in a group of neuroendocrine prostate carcinomas (Supplementary Figs S7, S8). Thus it is likely that AMPK kinase and AMPK protein have different functional roles in prostate cancer cells, such that blocking their kinase is not equivalent to silencing their expression (Fig. 9). In particular, our data support that silencing AMPKα1 or AMPKα2 alone did not noticeably reduce the levels of P-AMPKα (T172), although the loss of AMPKα2 enhanced basal and YM155-induced AMPKα (T172) (Figs 8D, 9, Supplementary Fig. S3). This suggests a potential mechanism by which loss of AMPKα2 may be cytostatic. Importantly, YM155 robustly suppresses the expression of both AMPKα1 and AMPKα2 ( Supplementary Fig. S4).
Our model helps predict relevant drug combination of YM155 in prostate and renal carcinomas. As discussed earlier, androgen receptor agonist, would be expected to enhance AMPK activity through overexpression of CAMKK2 55 , and provide the basis for the enhanced killing of AR-positive prostate cancer cells by YM155. Renal cell carcinomas typically lack the tumor suppressor VHL 72 . Functional loss of VHL induces the expression of HIF-1α and subsequent downstream hypoxia response pathway, leading to over vascularization and poor perfusion. We thus recommend moderate inhibition of VEGF to promote vascular normalization and help delivery of YM155.
While our study has addressed many problems, many more remain. What is the direct target of YM155 leading to the activation of AMPK? Does the treatment of cells with YM155 promote them to undergo a unique posttranslational modification of AMPK? How does YM155 cause loss of AMPK expression? Does loss of AMPKα1 and AMPKα2 play a role in the cytostatic/cytotoxic responses of PCa cells to YM155? How can these new attributes be clinically exploited in the design of new potential clinical trials for patients with cancer? As blocking AMPK and mTOR can treat various non-cancer related conditions 73 , can these properties of YM155 raise the potential off-label therapeutic use of this drug and its derivatives?
Cell growth/viability assay. Cells were plated overnight in 12 well plates at a density of 10,000 cells/1 mL/ well, treated with YM155 (0.1-1000 nM), and three days later stained with crystal violet, as previously described 27 . The IC 50 was calculated using GraphPad Prism. MTT assays were also conducted to confirm viability in a 96-well format, using a kit from Sigma-Aldrich as described by their protocol.
Western blotting. Cells (10 5 cells/ml DMEM/F12 + 5% FBS) were plated in either 6-well dishes (2 ml/well) or 10-cm dishes (10 ml/dish) for the indicated treatments, and cells were washed, lysed and clarified lysates were processed for Western blot analysis, as detailed before 27 . Unless specified, all proteins for Western analysis were www.nature.com/scientificreports www.nature.com/scientificreports/ electrophoresed through 4-12% Bis-Tris NuPAGE TM protein gels using SDS-MES running buffer (200 V, 4 °C). Equal loading of samples was based on total protein as determined by a BCA protein assay (96-well format), using an 8-point BSA standard curve, each point in duplicate and samples were the average of 3 replicates determinations. An equal transfer was confirmed by staining membranes with Ponceau S. Chemiluminescent image was captured by autoradiograph on film (Fuji Medical 100 NIF) 27 .

Reverse transcriptase polymerase chain reaction (RT-PCR). PureLink TM RNA mini kit (Invitrogen)
was used for isolating RNA from cells that were treated with 100 nM YM155 for the indicated time periods. One µg of RNA was reverse-transcribed using M-MLV Reverse Transcriptase (Promega). PCR amplification was performed using Taq (TRCN0000000861, TRCN0000002170, TRCN0000001271 were obtained from Sigma to silence human AMPKα1 and α2 genes. Viral supernatants were produced in HEK293T cells transfected with pLKO.1 sh-RNA, pMDG, and pCMV-dR8.78. Viral supernatants were tittered for viral levels by an ELISA for p24 using a kit from Sino Biological. PC-3 cells were transduced overnight with viral supernatant (MOI = 0.5) in the presence of 8 μg/ml polybrene, and 24 h after replacing with fresh growth medium, cells selected with 1.5 μg/ml puromycin for 4 days for 100% death of the non-transduced cells.