CDK8/19 inhibition induces premature G1/S transition and ATR-dependent cell death in prostate cancer cells

The CDK8/19 kinase module comprises a subcomplex that interacts with the Mediator complex and regulates gene expression through phosphorylation of transcription factors and Mediator subunits. Mediator complex subunits have been increasingly implicated in cancer and other diseases. Although high expression of CDK8/19 has been demonstrated in prostate cancer, its function has not been thoroughly examined. Here we report that CDK8/19 modulates the gene expression of cell cycle regulators and thereby maintains the proper G1/S transition in prostate cancer cells. We show that highly selective CDK8/19 inhibitors exerted anti-proliferative activity in prostate cancer cells both in vitro and in vivo. In CDK8/19 inhibitor-sensitive prostate cancer cells, the compounds reduced the population of G1 phase cells and elevated that of S phase cells through the modulation of G1/S transition regulators at the level of mRNA expression. Furthermore, the premature G1/S transition induced a DNA damage response that was followed by ATR-dependent and caspase-independent cell death. These findings suggest a novel role of CDK8/19 in transcription-mediated cell cycle control, albeit with possible contribution of other proteins inhibited by the compounds. Our data provide a rationale for further investigation of CDK8/19 inhibitors as a new therapeutic approach to prostate cancer.


INTRODUCTION
The cell cycle is controlled by CDKs and their binding partners known as cyclins. Proper progression of the cell cycle is monitored by checkpoints that recognize defects during multiple stages in the cell cycle such as in DNA synthesis and chromosome segregation. Activation of these checkpoints induces cell cycle arrest through the modulation of CDK activity [1]. Although CDKs were originally characterized by their roles in cell cycle regulation, several members of this family have direct functions in the regulation of RNA polymerase II (Pol II) activity, of which the best known representatives are CDK7, CDK8, and CDK9 [2]. Among these transcriptional CDKs, CDK8 and its closely related paralog CDK19 constitute alternative subunits of the regulatory CDK module of the Mediator complex, which links transcription factors with Pol II [3]. The Mediator complex also has a role in transcription elongation and pausing as well as in chromatin remodeling, facilitating the formation of enhancer-promoter gene loops [4,5]. This complex can influence cell identity by regulating genes associated with super-enhancers, which are composed of a large cluster of enhancers [4,5]. By functioning as a hub for input from numerous signaling pathways, the Mediator complex including CDK8/19 activates or represses transcription cycles [3,4]. Notably, the role of the CDK module in transcription is thought to be context-dependent, such that its biological function may vary among different cell types or in response to distinct stimuli.
A certain subset of CDKs (CDK1, CDK2, CDK4, and CDK6) and the corresponding cyclins is directly involved in the control of the cell cycle. In contrast to Research Paper these members of the CDK family, the cyclin subunits of transcriptional CDKs do not show oscillating patterns in their protein levels during the cell cycle. However, several studies have recently demonstrated the direct and indirect role of transcriptional CDKs in cell cycle control [6,7]. Different stages (e.g., G1/S and G2/M transition) in the cell cycle and the related pathways (e.g., p53/p21, Wnt/ β-catenin, and Rb/E2F) have been suggested to require CDK8 for proper function [8][9][10][11][12][13]. Thus, these recent findings begin to clarify a plausible role for the CDK module in Mediator as a hub for integrating transcription regulation with cell cycle control.
Notably, several subunits of the Mediator complex have been implicated in the pathogenesis of human diseases including cancer [4,14,15], for which most evidence has been obtained regarding the involvement of CDK8. In particular, amplification and overexpression of CDK8 have been found in colon, breast, and prostate cancers [14,[16][17][18], which underlies the marked attention that CDK8 has attracted as a potential target for cancer therapy by small molecule inhibitors [19][20][21]. In comparison, although information on CDK19 has been limited, the CDK19 gene has been recently implicated in prostate cancer [14,22]. However, despite their importance as potential therapeutic targets for prostate cancer, the function and importance of CDK8/19 in prostate cancer remain poorly understood.
To address this deficit, in this preclinical study, we used both small molecule inhibitors of CDK8/19 and genetic approaches to investigate the dependence of prostate cancer cells on CDK8/19 activity. Furthermore, we explored the biological roles of CDK8/19 in prostate cancer cells as well.
We then investigated the expression of CDK8 and CDK19 in several commercially available prostate cancer cell lines. In accordance with previous reports [14], CDK19 was highly expressed in some prostate cancer cells at the mRNA and protein levels ( Figure 1C, 1D, and Supplementary Figure 2). We observed that CDK8 protein levels were moderately elevated in CDK19-depleted cells ( Figure 1D and Supplementary Figure 2). Notably, similar compensatory effects in paralogs have been reported previously [24]. CDK8/19 inhibition did not obviously impact proliferation of LNCaP, 22Rv1, PC-3, or DU 145 cells ( Figure 1E and 1F), whereas we observed that treatment with T-474 or T-418 substantially inhibited the proliferation of VCaP cells ( Figure 1G). Furthermore, in VCaP cells, knockdown of CDK8 or CDK19 by siRNA did not obviously impact the cell proliferation (Supplementary Figure 3A). Specifically, only one of four CDK19 siRNAs substantially suppressed cell proliferation; however, the effects appeared to be off-target considering the limited knockdown efficiency (Supplementary Figure 3A). Importantly, the simultaneous knockdown of CDK8 and CDK19 suppressed the proliferation of VCaP cells (Supplementary Figure 3B). These results suggest that inhibition of both CDK8 and CDK19 is essential for suppression of VCaP cell proliferation.

Effects of CDK8/19 inhibition on cell cycle progression
Given that CDK8/19 forms a subcomplex of Mediator, it was plausible that inhibition of CDK8/19 might affect the gene expression pattern. To understand the mechanism of action, we performed a microarray analysis in CDK8/19 inhibitor-sensitive VCaP cells. A comprehensive evaluation of transcriptional changes using a parametric analysis of gene set enrichment (PAGE) revealed that down-regulated genes following T-474 treatment were enriched for multiple processes related to cell migration (Table 1), in accordance with a previous report [14]. Notably, the down-regulated processes included regulation of androgen receptor (AR) activity, which is important for pathophysiology of prostate cancer. However, we also observed that treatment with T-474 or T-418 reduced the mRNA expression of KLK3 (a well-known AR target gene) in LNCaP and 22Rv1 cells (Supplementary Figure 4). Considering that treatment with CDK8/19 inhibitors suppressed the proliferation of VCaP but not LNCaP and 22Rv1 cells, the effects on AR signaling are considered unlikely to be involved in the anti-proliferative activity. In comparison, the up-regulated processes following CDK8/19 inhibitor treatment were primarily related to cell cycle progression and DNA replication (Table 1). Flow cytometric analysis revealed www.impactjournals.com/oncotarget that the population of G1 phase cells was decreased and that of S phase cells was increased in VCaP cells treated with T-474 or T-418 (Figure 2A and 2B). In contrast, in CDK8/19 inhibitor-insensitive LNCaP and 22Rv1 cells, changes in the cell cycle profile following CDK8/19 inhibitor treatment were less obvious (Figure 2A and 2B). Consistent with these findings, treatment with the CDK8/19 inhibitors induced Cdc6 expression and MCM2 phosphorylation in VCaP cells ( Figure 2C), both of which are implicated in DNA replication [25]. Conversely, the expression of Cdc6 and the phosphorylation of MCM2 were not affected by treatment with either compound in LNCaP and 22Rv1 cells ( Figure 2D), although a reduction of STAT1 phosphorylation was detected ( Figure 2D) as also observed in VCaP cells ( Figure 1B). These results indicate that CDK8/19 inhibition decreases the population of G1 cells and increases that of S phase cells in CDK8/19 inhibitor-sensitive VCaP cells.

Induction of DNA damage followed by ATRdependent cell death by CDK8/19 inhibition
Next, to investigate the cell fate of CDK8/19 inhibitor-treated VCaP cells, we examined the long-term effects of CDK8/19 inhibition. Flow cytometric analysis revealed a sustained decrease in G1 phase and an increase in S phase populations following T-474 treatment ( Figure 3A). Consistent with this phenomenon, treatment with T-474 or T-418 not only increased Cdc6 expression and MCM2 phosphorylation, but also decreased the expression of p21 and p27 ( Figure 3B), which are inhibitors of the CDK2/4/6-mediated G1/S transition [26]. Notably, western blot analysis revealed that treatment with the CDK8/19 inhibitors induced markers of the DNA damage response (DDR) [27] such as Chk1 phosphorylation, mono-ubiquitination of FANCD2 (as indicated by a reduced mobility of bands), and γH2AX, at later time points relative to the changes in cell cycle regulators ( Figure 3B). We therefore next examined whether the activation of DDR is involved in the antiproliferative activity of CDK8/19 inhibitors. As DDR is known to be regulated by sensor kinases such as ATR, ATM, and DNA-PK [27], we utilized inhibitors of these kinases [28][29][30]. Treatment with the ATR inhibitor VE-821 significantly reversed the reduced cell viability mediated by CDK8/19 inhibition in VCaP cells, whereas treatment with the ATM inhibitor KU-55933 and DNA-PK inhibitor NU7026 did not ( Figure 3C). In flow cytometric analysis, we noted that T-474 treatment elevated the sub-G1 fraction ( Figure 3A), which is indicative of apoptotic cell death. Treatment with CDK8/19 inhibitors also induced cleavage of PARP ( Figure 3B); however, caspase inhibitor Z-VAD treatment did not reverse the reduced cell viability caused by CDK8/19 inhibition ( Figure 3D), suggesting that CDK8/19 inhibition causes caspase activation but leads to caspase-independent cell death. In addition, despite its rescue effects on cell viability, the ATR inhibitor treatment did not prevent the caspase activation mediated by CDK8/19 inhibition ( Figure 3E), further supporting the notion that the caspase activation by CDK8/19 inhibitors does not contribute to their cell-killing effects. Previous findings on topoisomerase inhibitor-mediated replication stress and DNA damage [31,32] prompted us to test the combination treatment of CDK8/19 inhibitors with topoisomerase inhibitors. CDK8/19 inhibitor-treated VCaP cells were more sensitive to the topoisomerase inhibitors SN-38 (an active metabolite of irinotecan), etoposide, and doxorubicin ( Figure 3F and Supplementary Figure 5), probably owing to the accumulation of DNA damage caused by the effects of the drug combination.

Premature G1/S transition by CDK8/19 inhibition
The reduction of G1 phase cells and accumulation of S phase cells can be caused by the delay of S phase progression or the acceleration of G1/S transition. To address whether CDK8/19 inhibition directly modulates S phase progression, we performed EdU pulse chase analysis in which cells were pre-treated with CDK8/19 inhibitors and then pulse-labeled with EdU in the presence of CDK8/19 inhibitors. T-474 treatment did not affect the population of early S phase cells in VCaP cells 0-10 hours after the release of EdU (Supplementary Figure 6), indicating that CDK8/19 inhibition does not directly impact S phase progression.
The transition from G1 phase to S phase is tightly regulated by various regulators [1]. As CDK8/19 can impact gene expression patterns as Mediator-associated kinases, we examined whether treatment with CDK8/19 inhibitors might affect the gene expression of the G1/S transition regulators. To analyze the gene expression changes that occurred prior to the changes in cell cycle profile, we performed microarray analysis using VCaP cells treated with CDK8/19 inhibitors for 6 hours. Treatment with CDK8/19 inhibitors not only increased the mRNA expression of positive regulators (e.g., CDC25A, cyclin E1, or c-Myc) but also decreased that of negative regulators (e.g., p19, p21, or p27) of the G1/S transition ( Figure 4A). Western blot analysis revealed that T-474 treatment elevated c-Myc protein expression at least from 2 hours after compound treatment, whereas an increase in MCM2 phosphorylation was induced by the 24-hour treatment but not by the 2-or 6-hour treatment ( Figure 4B). The results indicate that the gene expression changes in G1/S transition regulators would be earlier events. Notably, c-Myc upregulation following CDK8/19 inhibitor treatment was not observed in CDK8/19 inhibitor-insensitive LNCaP and 22Rv1 cells (Supplementary Figure 7). As the post-translational regulation of c-Myc expression has been well investigated [33], we also examined the turnover of c-Myc protein. The normalized c-Myc protein level in T-474-treated cells was maintained somewhat longer than in control cells, in the presence of the protein translation inhibitor cycloheximide (Supplementary Figure 8), suggesting that CDK8/19 inhibition delays the degradation of c-Myc protein, albeit with moderate effects. Given that T-474 treatment clearly increased MYC mRNA levels ( Figure 4A), CDK8/19 may regulate c-Myc expression mainly transcriptionally but also post-transcriptionally to a certain degree. Consistent with the upregulation of c-Myc expression, we observed CDK8/19 inhibitor-mediated gene expression changes in previously identified c-Myc downstream effectors [34] ( Figure 4C).
A shortened G1 phase and the fast transition through G1 phase enables cells to enter the S phase in the presence of unrepaired DNA damage followed by ATR activation [35]. We therefore investigated whether aberrant G1/S transition contributes to the induction of DDR in CDK8/19 inhibitor-treated cells by performing siRNA experiments. Western blot analysis revealed that c-Myc depletion reversed the changes in cell cycle regulator expression and the activation of DDR following T-474 treatment ( Figure 4D). We confirmed that c-Myc knockdown did not obviously impact cell viability at that time point (Supplementary Figure 9), precluding the possibility that the effects of c-Myc knockdown on cell proliferation alleviated the effects of CDK8/19 inhibitors on the cell cycle. Furthermore, ATR inhibitor treatment did not affect the population of G1 or S phase cells when combined with CDK8/19 inhibition (Supplementary Figure 10), indicating that the DDR activation is not responsible for the aberrant cell cycle profile. These results suggest that CDK8/19 inhibitor-mediated gene expression changes accelerate the G1/S transition and thereby induce DDR.

In vivo antitumor activity of CDK8/19 inhibition
Next, we examined the in vivo activity of a CDK8/19 inhibitor in VCaP xenograft models. T-474 exhibited potent antitumor activity when orally administered at 5 mg/kg once daily [Treatment over control (T/C) = 23%, P < 0.01] for 21 days without severe body weight reduction ( Figure 5A and 5B). Western blot analysis in parallel studies of mice treated for 3 days revealed significant reductions of STAT1 phosphorylation in VCaP tumors ( Figure 5C). Consistent with in vitro observation, T-474 administration up-regulated c-Myc, MCM2 phosphorylation, Chk1 phosphorylation, and γ-H2AX, whereas it down-regulated p21 ( Figure 5C). These results indicate that T-474 administration causes aberrant G1/S transition and activates the DDR machinery in VCaP xenografts, thereby exhibiting significant antitumor activity.

DISCUSSION
In this study, we used two chemically distinct CDK8/19 inhibitors, both of which showed high selectivity against CDK8 and CDK19. These compounds provided us a means to precisely understand the biological phenomena dependent on CDK8/19 kinase activity. The data presented here show that inhibition of CDK8/19 by the compounds and dual depletion of Recently, several studies have explored the role of CDK8/19 in cancer by using small molecules that selectively inhibit CDK8/19 [19,21,36]. As CDK8 is amplified and functions as an oncogene in colorectal cancer through the regulation of Wnt/β-catenin signaling [17,18,37], the therapeutic potential of the small molecules has been actively tested in preclinical models of colorectal and other cancers [19,36]. Additionally, cortistatin A, a natural product inhibitor of CDK8/19, has been shown to demonstrate antitumor activity in acute myelogenous leukemia (AML) cells [21]. Specifically, Pelish et al. reported that the increased activation of super-enhancer genes in AML cells by cortistatin A results in antileukemic activity [21]. In the current study, to characterize the mechanism by which CDK8/19 contributes to the proliferation of prostate cancer cells, we primarily utilized prostate cancer VCaP cells, which displayed the highest sensitivity to CDK8/19 inhibition among the prostate cancer cells tested. Given that VCaP cells show the highest expression of CDK19 among a large number of cell lines (approximately 1000 cell lines) in the Cancer Cell Line Encyclopedia (CCLE) database (https:// portals.broadinstitute.org/ccle/home), it is plausible that overexpression of CDK19 plays a predominant role in prostate cancer cell proliferation. However, the CDK19 protein level in VCaP cells was high although comparable to that of LNCaP and 22Rv1 cells, which were insensitive to T-474 treatment. Factors other than CDK19 expression level are therefore likely involved in determining the sensitivity to CDK8/19 inhibitors.
We expect that the VCaP sensitivity is not due to an off-target effect of our compounds because the antiproliferative effects were observed for two structurally distinct compounds, each of which has high kinase selectivity. Furthermore, the simultaneous depletion of CDK8 and CDK19 suppressed the proliferation of VCaP cells. However, we cannot completely deny the possibility of off-target effects in the absence of a rescue experiment using a CDK8/19 mutant refractory to T-474 or T-418 binding. Thus, our data suggest that inhibition of CDK8/19 by our compounds contributes to anti-proliferative activity in prostate cancer cells both in vitro and in vivo, although it remains possible that the effects may be influenced by the inhibition of a protein other than CDK8/19. Although only a limited number of prostate cancer cell lines are currently available, further investigation will be required to extend the generality of our findings.
Our results demonstrate that CDK8/19 prevents premature G1/S transition by modulating the gene expression of G1/S transition regulators in VCaP cells. Consistent with this finding, it has been reported that CDK8 positively regulates the p53-p21 pathway activation and thereby inhibits the G1/S transition [8]. However, this mechanism would not be expected to play an important role in VCaP cells, because these cells harbor an R248W mutation in the TP53 gene, which is a gain-of-function mutation abrogating wild-type p53 activity [38]. In addition, a genetic screen in Caenorhabditis elegans also demonstrated that CDK8 module components are required for the control of entry into G0 during development [39]. Conversely, it has been suggested that CDK8 functions as a positive regulator of the G1/S transition through Wnt/βcatenin and/or E2F signaling [7,13,17]. These findings indicate that the CDK8/19 module contributes to regulation of the G1/S transition in a context-dependent manner.
Phosphoproteome and transcriptome analyses have revealed that CDK8/19 predominantly modulates the regulators of transcription [40]. Among the G1/S transition regulators tested in the current study, the expression of c-Myc was substantially affected by CDK8/19 inhibition. Notably, a previous report showed that CDK8 binds to the MYC promoter as a component of the Mediator complex, albeit with a positive impact on c-Myc transcription [17]. The increased c-Myc expression mediated by our CDK8/19 inhibitors may be linked to super-enhancer activation, as MYC is known to be a super-enhancer-associated gene [41]. However, one caveat for this interpretation is that MYC has not been confirmed to be SE-associated in VCaP cells; therefore, further study is needed to clarify this issue. Given that leukemic cells are sensitive to the dosage of super-enhancer-associated genes affected by CDK8/19 inhibition [21], the possibility exists that certain prostate cancer cells such as VCaP cells may also show vulnerability to the activation of super-enhancer-associated genes. In CDK8/19 inhibitor-sensitive prostate cancer cells, we found that c-Myc depletion prevents premature G1/S transition and the subsequent DDR resulting from CDK8/19 inhibition. Furthermore, we also showed that c-Myc upregulation and the defects in cell cycle control did not occur in CDK8/19 inhibitor-insensitive cells. The findings support the upregulation of the MYC gene as a critical molecular response to CDK8/19 inhibition in CDK8/19 inhibitor-sensitive cells. However, we cannot exclude the contribution of other factors to this process.
It has been reported that overexpression of c-Myc not only accelerates the G1/S transition but also induces DNA damage and a hypersensitivity to DNA-damaging agents [42][43][44]. The premature G1/S transition might enable cells to enter the S phase in the presence of unrepaired damage. In support of this mechanism, we found that CDK8/19 inhibitor-treated VCaP cells acquire DNA damage and are sensitive to multiple topoisomerase inhibitors. Furthermore, a previous report showed that cells expressing c-Myc at high levels undergo p53-dependent G2 arrest [44]. It has been also shown that the G2/M checkpoint against DNA damage is impaired in cells with p53 R248W [45]. Consistent with these reports, VCaP cells (which carry p53 R248W ) [38] did not undergo G2 arrest when c-Myc expression and DDR were induced by CDK8/19 inhibitors. The G2/M checkpoint impairment in c-Myc overexpressing cells is expected to cause uncontrolled proliferation without appropriate DNA repair. Notably, CDK8/19 inhibitors did not promote the proliferation of VCaP cells but rather led to cell death. Thus, the possibility exists that VCaP cells would be particularly vulnerable to the stress resulting from c-Myc-driven premature G1/S transition. Further investigation will allow us to better understand the mechanisms underlying their observed sensitivity to CDK8/19 inhibition.
DDR induces cell cycle arrest for the repair of errors in DNA or to trigger cell death, depending on the type of cell and the level of DNA damage. Although many studies have examined the cytoprotective role of ATR, a pro-apoptotic role of ATR has been also reported [46][47][48]. In our study, an ATR inhibitor clearly reversed the reduced cell viability of CDK8/19 inhibitor-treated cells. Further investigation will provide a better understanding of the means by which CDK8/19 inhibition triggers cell death via ATR and its downstream effectors.
In summary, our preclinical study demonstrates that CDK8/19 inhibition induces cell death in VCaP prostate cancer cells. Inhibition of CDK8/19 activity modulates the mRNA expression of G1/S transition regulators, leading to premature G1/S transition followed by DDR and cell death in an ATR-dependent manner. These results shed light on the importance of CDK8/19 kinase as a hub for integrating transcription regulation with cell cycle control in prostate cancer cells.  [49,50]. KU-55933, NU7026, and VE-821 were obtained from Selleck Chemicals. IFNγ and Z-VAD were obtained from R&D Systems and Sigma-Aldrich, respectively. www.impactjournals.com/oncotarget

Cell lines and culture
Prostate cancer cell lines (22Rv1, DU 145, LnCaP, PC-3, and VCaP), multiple myeloma cell line (RPMI8226), and colorectal cancer cell line (SW480) were obtained from American Type Culture Collection (ATCC) in 2013. The cell lines were cultured at 37° C with 5% CO 2 in the recommended medium supplemented with 10-20% FBS (Thermo Fisher Scientific). The cell lines were stocked after Mycoplasma testing (performed by the Central Institute for Experimental Animals) and used within 2 months after resuscitation. The ATCC uses short tandem repeat profiling for authentication of cell lines. No authentication was performed by the authors.

Recombinant kinome-wide selectivity profiling
Kinase selectivity of T-474 and T-418 was evaluated at 300 nmol/L using a panel of 456 kinases at DiscoveRx (KINOMEscan). The platform employs an active sitedirected competition binding assay to quantitatively measure interactions between test compounds and kinases.

Public gene expression data
The data for CDK8 and CDK19 mRNA expression were collected from public databases of the CCLE.

Transfection and siRNA
Cells were transfected with siRNAs using Lipofectamine RNAi MAX (Invitrogen). CDK8 and CDK19 siRNAs were obtained from GE Dharmacon (OnTARGETplus). c-Myc siRNAs were obtained from Thermo Fisher Scientific (Silencer Select) or GE Dharmacon

Reporter gene assay
The Wnt reporter gene assay was performed as previously described [51,52]. The reporter gene activity was assessed using the Dual-Glo luciferase assay (Promega), according to the manufacturer's protocol.

Flow cytometric analysis
Cells were harvested and fixed with 70% ice-cold ethanol for 30 minutes at 4° C. After washing with PBS, the cells were incubated with PBS containing 1 mg/mL RNase for 30 minutes at 37° C. After washing with PBS again, the cells were stained with 50 μg/mL propidium iodide for 30 minutes at 4° C. EdU pulse-chase analysis was performed using the CLICK-iT Plus EdU flow cytometry assay kit (Thermo Fisher Scientific). Analyses were performed on a FACS Calibur (BD Biosciences).

Quantitative PCR
RNA was extracted using the miRNeasy kit (Qiagen). cDNA was synthesized using the Verso cDNA kit (Thermo Scientific). Real-time quantitative PCR (qPCR) was performed on an ABI StepOne real-time PCR system (Applied Biosystems) using TaqMan gene expression assays for AR (Hs00171172_m1), CDK19 (Hs01039930_ g1), and KLK3 (Hs02576345_m1) (Applied Biosystems). Relative gene expression was calculated using the ΔΔC T method following the manufacturer's instructions. The expression ratios of the indicated genes were normalized by the GAPDH expression in each cell line.

Microarray analysis
Total RNA was extracted using the RNeasy Mini kit (Qiagen). Preparation of the cDNA and cRNA, hybridization, and microarray scanning were performed according to the manufacturer's protocols (Affymetrix Inc.). The biotinylated cRNA was hybridized to Affymetrix U133 Plus 2 human genome arrays. The captured signals were normalized to the median expression level using the GeneSpring software package (Agilent Technology) and the normalized data were filtered by present/absent calls and the expression level. Pathway analysis was performed with PAGE [54] using the NCBI BioSystems Database [55].

Cell viability assay
Cell viability was assessed using the CellTiter-Glo luminescent cell viability assay (Promega) or CyQuant Direct (Thermo Fisher Scientific), according to the manufacturer's protocol. Sigmoidal dose-response (variable slope) curves were fitted using non-linear regression analysis (GraphPad Prism version 6; GraphPad Software).

Animal study
Suspensions of VCaP prostate cancer cells (1 × 10 6 cells/site) were subcutaneously implanted into the right flank of 6-week-old male SCID mice (C.B17/Icr-scid/ scid Jcl, CLEA Japan). Tumor volumes were calculated as volume = L × l 2 × 1/2, where L was taken to be the longest diameter across the tumor and l was taken to be the corresponding perpendicular distance. Body weight was also measured. For the antitumor activity test, when the tumor mass was approximately 200 mm 3 , mice were sorted into treatment groups (N = 6/group) such that the mean tumor volume and body weight between groups were similar. Tumors were monitored and mice were euthanized when an endpoint was reached, defined as tumor volume larger than 2000 mm 3 , severe ataxia, body weight loss (>20% compared to the body weight on the day of randomization), or study end, whichever came first. From the day of randomization, T-474 dissolved in distilled water containing 0.5% methyl cellulose was orally administered to mice bearing a xenograft for 21 days. Treatment over control (T/C, %), an index of antitumor activity, was calculated by comparison of the mean change in tumor volume over the treatment period for the control and treated groups. For western blot analysis, tumors were homogenized using a Lysing Matrix I tube (MP Biomedicals) in Lysing solution [10% glycerol, 1% sodium dodecyl sulfate, 62.5 mmol/L Tris-HCl (pH 7.5), protease inhibitors, and phosphatase inhibitors (cOmplete mini and PhosSTOP, Sigma-Aldrich)]. The mice were housed and maintained in accordance with institutional guidelines established by the Institutional Animal Care and Use Committee, in a facility accredited by the American Association for Accreditation of Laboratory Animal Care. The animal experimental protocols were approved by the Institutional Animal Care and Use Committee.

Statistical analysis
Groups were compared by Bonferroni's multiple comparison test using GraphPad Prism software (version 6). To assess in vivo antitumor activity, Welch's t test was performed with treatment over control (T/C, %) values using Excel (Microsoft). Differences were considered significant at P < 0.05.