Open Full Page

Published OnlineFirst March 22, 2012; DOI: 10.1158/2159-8290.CD-11-0234

RESEARCH ARTICLE

Functional Metabolic Screen Identifies 6-Phosphofructo-2-Kinase/Fructose-2, 6-Biphosphatase 4 as an Important Regulator of Prostate Cancer Cell Survival

Susana Ros1, Claudio R. Santos1, Sofia Moco4, Franziska Baenke1, Gavin Kelly2, Michael Howell3, Nicola Zamboni4, and Almut Schulze1

The BATTLE Trial: Personalizing Therapy for Lung Cancer RESEARCH ARTICLE

ABSTRACT Alterations in metabolic activity contribute to the proliferation and survival of cancer cells. We investigated the effect of siRNA-mediated gene silencing of 222 metabolic enzymes, transporters, and regulators on the survival of 3 metastatic prostate cancer cell lines and a nonmalignant prostate epithelial cell line. This approach revealed significant complexity in the metabolic requirements of prostate cancer cells and identified several genes selectively required for their survival. Among these genes was 6-phosphofructo-2-kinase/fructose- 2,6-biphosphatase 4 (PFKFB4), an isoform of phosphofructokinase 2 (PFK2). We show that PFKFB4 is required to balance glycolytic activity and antioxidant production to maintain cellular redox balance in prostate cancer cells. Depletion of PFKFB4 inhibited tumor growth in a xenograft model, indicating that it is required under physiologic nutrient levels. PFKFB4 mRNA expression was also found to be greater in metastatic prostate cancer compared with primary tumors. Taken together, these results indicate that PFKFB4 is a potential target for the development of antineoplastic agents.

SIGNIFICANCE: Cancer cells undergo several changes in their metabolism that promote growth and survival. Using an unbiased functional screen, we found that the glycolytic enzyme PFKFB4 is essential for prostate cancer cell survival by maintaining the balance between the use of glucose for energy generation and the synthesis of antioxidants. Targeting PFKFB4 may therefore present new therapeutic opportunities. Cancer Discov; 2(4); 328–43. ©2012 AACR.

INTRODUCTION In this study we used an unbiased functional approach to identify metabolic enzymes required for the survival of It is now widely accepted that cancer cells alter their prostate cancer cells. Expression of glycolytic and lipogenic metabolic activity to satisfy their increased needs for energy enzymes is induced in response to androgens in prostate can- and biosynthetic precursors. Many cancers show increased gly- cer (12). Prostate cancers also show a high rate of de novo lipo- colysis, a feature generally known as the Warburg effect, and genesis, and inhibition of this process decreases the viability the importance of biosynthetic processes such as lipid syn- of prostate cancer cells (13, 14). The prostate epithelium se- thesis for cancer cell growth is increasingly recognized (1). In cretes large amounts of citrate into the seminal fluid, which numerous studies, authors have shown that oncogenic signal- is achieved by inhibition of aconitase, resulting in a truncated ing pathways regulate the expression and activity of metabolic tricarboxylic acid cycle. Interestingly, during transformation, enzymes to support macromolecule synthesis in cancer cells, aconitase is reactivated, and prostate cancer cells oxidize ci- which is required for their rapid proliferation (2). Alterations trate to generate energy (15). Because many cancer cells ex- in metabolic activity can also protect cells from apoptosis (3). hibit attenuated mitochondrial oxidative metabolism (16), Furthermore, in cancer genome-sequencing studies, research- prostate cancer may represent a unique metabolic situation. ers have revealed that metabolic genes themselves can also be The characterization of several metabolic features of non- subjected to mutational alterations (4). Enzymes within the malignant and prostate cancer cell lines presented here shows core metabolic pathways, including glycolysis, glutaminolysis, that the latter are glucose dependent when cultured both in and fatty acid biosynthesis, have been shown to be essential full medium (FM) or in lipid-depleted medium (LDM) and for the growth and survival of cancer cells (5–9). More re- sensitive to lipid synthesis inhibition when cultured in LDM. cently, serine biosynthesis emerged as an important metabolic Using these cells in a siRNA screen targeting basic glucose me- pathway in the development of breast cancer (10, 11). It is tabolism, lipogenesis, and amino acid biosynthesis genes, in therefore evident that metabolic enzymes represent interest- both full medium (FM) and in LDM, we were able to identify ing targets for cancer therapy. several genes selectively required for the survival of the prostate cancer cell lines. Among these was 6-phosphofructo-2-kinase/ Authors’ Affiliations: 1Gene Expression Analysis Laboratory, 2Bio- PFKFB4 3 fructose-2,6-biphosphatase 4 ( ), an isoform of the informatics and Biostatistics Service, High Throughput Screening PFK2 Facility, Cancer Research UK London Research Institute, London, United glycolytic enzyme phosphofructokinase 2 ( ). A detailed Kingdom; 4Institute of Molecular Systems Biology, ETH Zurich, Zurich, analysis of the role of this gene in prostate cancer cell survival Switzerland revealed that PFKFB4 is required to maintain redox balance S. Ros and C. R. Santos contributed equally to this work. and to support tumor growth in vivo, highlighting the impor- Note: Supplementary data for this article are available at Cancer Discovery tance of the metabolic regulation of bioenergetics and antioxi- Online (http://www.cancerdiscovery.aacrjournals.org). dant production in cancer cells. Corresponding author: Almut Schulze, Cancer Research UK London Research Institute, 44 Lincoln’s Inn Fields, London WC2A 3LY, United RESULTS Kingdom. Phone: 44-207-269-3663; Fax: 44-207-269-3479; E-mail: [email protected] Metabolic Characterization of Prostate Cell Lines doi: 10.1158/2159-8290.CD-11-0234 To identify metabolic weaknesses of prostate cancer ©2012 American Association for Cancer Research. cells, we first characterized the metabolic requirements of 2

APRIL 2012 CANCER DISCOVERY | 329

RESEARCH ARTICLE Ros et al. androgen-independent prostate cancer cell lines, DU145 and amino acids did not show significant differences between cell PC3, and an androgen-dependent prostate cancer cell line, lines or conditions. However, the rate of glycine and isoleucine LNCaP. These were compared with the nontransformed pros- uptake and glutamate secretion was markedly greater in the 3 tate epithelial cell line RWPE1 (17). We first investigated glu- prostate cancer cells compared with RWPE1. Increased gluta- cose uptake, lactate secretion, and oxygen consumption in the mate secretion has been previously reported in cancer cells (20). 4 cell lines. All cell lines show high rates of glucose consump- Taken together, the metabolic characterization of these cell tion and lactate secretion, indicating high glycolytic activity lines revealed that despite some heterogeneity in glucose and (Fig. 1A). DU145 and RWPE1 cells show the greatest rates of oxygen usage, the high dependence on glucose is a unifying both glucose uptake and lactate secretion, which is consistent feature of the 3 prostate cancer cell lines, whereas the non- with their high proliferation rate (Supplementary Fig. S1A). malignant RWPE1 cells may have the ability to use alternate Basal oxygen consumption rate (OCR) was greatest in substrates. Furthermore, reliance on lipid synthesis, which RWPE1 and LNCaP cells. PC3 cells showed a low basal OCR has been previously described for prostate cancer cells in a but increased oxygen consumption after treatment with the xenograft model (19), is only observed when they are grown uncoupling agent FCCP. Only DU145 cells showed a com- in lipid-depleted conditions. plete loss of mitochondrial respiration (Fig. 1B), which is a common feature of cancer cells. We next investigated Functional Metabolic Screen potential differences in the rates of lipid synthesis in the dif- Using a custom library of siRNA oligonucleotides targeting ferent cell lines. Pyruvate-dependent lipid synthesis in FM 222 metabolic enzymes, transporters, and regulators located was similar in the 4 cell lines (Fig. 1C). The removal of lipo- within the core glucose metabolism (see Supplementary Fig. proteins from the growth medium (LDM) strongly increased S2A and Supplementary Discussion), we performed a func- the rate of lipogenesis (Fig. 1C) and reduced growth rates of tional screen to identify metabolic genes required for pro- the 3 prostate cancer cells (Supplementary Fig. S1A and S1B). liferation and survival of prostate cancer cell lines. Because Many cancer cell lines depend on glucose for survival. In these cells only showed sensitivity to inhibition of lipogenesis addition, glutamine has been shown to be a major anabolic in LDM (Fig. 1E), we used both FM and LDM growth condi- substrate in some cancer cells (18). We therefore investigated tions for the screen. Induction of apoptosis (caspase activity) the sensitivity of these cell lines to the depletion of glucose and reduction in total protein amount (cell mass) in response or glutamine. Remarkably, the withdrawal of glucose for to siRNA transfection was measured (Supplementary Fig. 72 hours induced apoptosis in all prostate cancer cell lines S2B and Supplementary Table S2). We used the apoptosis in both FM and LDM (Fig. 1D). Only DU145 cells were af- z-scores to perform a 2-way cluster analysis (Fig. 2A), which fected by glutamine depletion (Fig. 1D). In contrast, RWPE1 revealed considerable overall differences between the cell cells were not affected by the withdrawal of either glucose or lines and conditions whereas the different serum conditions glutamine. This observation confirms that the prostate for each cell line clustered together. Importantly, the results cancer cell lines are highly dependent on the availability of from the 3 prostate cancer cell lines were clearly separated glucose for their survival. from the results of the nonmalignant RWPE1 cell line. A de- The inhibition of fatty acid biosynthesis blocks the growth tailed discussion of the primary screen results is provided in of prostate cancer cells in a xenograft model (19). We there- the Supplementary Discussion. fore investigated the dependency of these lines on de novo siRNA sequences that caused apoptosis (z-score $2.5) or lipogenesis. Inhibition of lipid synthesis by treatment with a loss of cell mass (cell mass #0.6 relative to control) in at inhibitors of fatty acid synthase or ATP-citrate lyase (C75 and least 2 of the cancer cell lines but had no effect in RWPE1 SB204990, respectively) reduced viability of the prostate can- cells (Table 1) were selected for validation by the use of 4 indi- cer cells only in the absence of lipoproteins (Fig. 1E). RWPE1 vidual siRNA sequences (Supplementary Table S3). The same cells were sensitive to inhibition of lipid synthesis, indicating 18 genes were also used to perform a gene set-enrichment that their recommended growth medium (KGM) does not analysis with 2 public expression datasets from prostate can- provide sufficient exogenous lipids for growth when de novo cer (GSE3325 and GSE6752). This analysis showed that these lipogenesis is blocked. genes tended to be overexpressed in samples from metastatic Next, we established differences in metabolite levels be- prostate cancer compared with samples from primary sites tween the 3 prostate cancer cell lines and RWPE1 cells under (Fig. 2B). This observation is particularly interesting because the different aforementioned growth conditions described the 3 prostate cancer cell lines used for this study were all by using mass spectrometry (Supplementary Fig. S1C and originally isolated from metastases. Supplementary Table S1). The 3 prostate cancer cell lines Only 2 genes compromised the viability of all 3 prostate showed greater intracellular levels of several glycolytic inter- cancer cell lines (Fig. 2C and D). Silencing of PRKAB1 induced mediates, including fructose-6-phosphate, phosphoenolpyr- apoptosis in both FM and LDM conditions whereas silenc- uvate, and pyruvate. Interestingly, levels of adenine, a purine ing of PFKFB4 caused apoptosis in LDM only. Importantly, derivative required for nucleotide synthesis, were reduced in in several independent studies, investigators show that ex- the prostate cancer cell lines, whereas the concentration of pression of these 2 genes is increased in prostate cancer phosphoribosyl pyrophosphate, an intermediate in purine metastasis when compared with primary tumors (Fig. 2E). biosynthesis, was increased. PRKAB1 codes for the β1 subunit of AMP-activated protein We also measured the uptake and secretion rates of me- kinase (AMPK), a kinase that regulates metabolic processes tabolites that could be readily detected by mass spectrometry in response to low energy levels (21). PFKFB4 is an isoform of (Supplementary Fig. S1D). Overall, the uptake rate for most PFK2 and has been originally identified in testes (22). PFK2

330 | CANCER DISCOVERY APRIL 2012 www.aacrjournals.org

Functional Metabolic Screen in Prostate Cancer RESEARCH ARTICLE 0.71 0.69 0.67 0.55 0.77 PC3 0.56 0.69 0.63 0.82 0.65 0.80 1.000.76 1.00 0.65 0.62 0.66 0.74 1.02 0.73 0.52 0.51 0.41 0.59 0.57 0.57 1.16 0.85 0.91 0.53 0.31 0.54 0.46 LNCaP 0.40 1.00 0.73 0.72 0.57 0.66 0.48 0.72 0.57 0.44 0.31 0.52 0.55 0.53 0.59 0.56 0.64 1.09 0.74 0.70 0.98 DU145 Cell mass (fold of negative control) Cell mass (fold of negative 0.56 0.510.34 0.60 0.40 0.56 0.58 0.46 0.30 0.43 0.58 0.94 0.74 0.850.79 0.71 0.67 0.65 0.79 0.80 0.66 0.84 1.02 0.69 0.88 0.88 0.930.86 0.65 0.94 0.67 0.86 0.91 0.63 0.96 1.10 0.79 0.66 0.73 0.890.89 0.66 0.78 1.32 0.91 1.12 0.97 1.33 0.94 RWPE1 2.02 0.92 0.73 0.80 0.62 1.49 0.71 0.64 2.63 1.37 0.63 2.11 0.78 0.94 0.71 0.73 1.411.44 0.71 0.69 4.27 7.09 PC3 1.26 1.14 0.87 0.84 0.90 2.65 7.05 1.77 1.51 0.66 0.68 2.75 2.73 0.483.96 0.85 3.43 0.78 0.97 0.65 0.28 2.72 4.99 0.83 –0.64 2.92 2.67 3.02 2.99 4.65 4.35 6.99 7.31 1.52 7.61 LNCaP 3.47 3.79 4.79 3.14 1.63 1.09 0.16 2.09 1.48 0.67 2.05 0.19 0.70 0.88 3.64 4.15 2.79 2.76 3.69 –0.86 –0.02 –0.19 13.44 4.61 Apoptosis (z-score) 1.35 2.87 6.04 2.091.56 –0.51 1.70 2.11 4.19 6.81 3.38 6.36 DU145 2.75 7.10 5.49 7.66 6.80 8.34 4.46 12.98 6.57 16.89 3.96 12.19 6.54 0.33 1.31 2.06 1.60 2.12 0.31 –0.23 –0.13 1.57 0.74 1.67 RWPE1 PDK3 0.55PRKAA2 0.16PRKAB1 –0.60 0.87 SLC16A3 0.65 SLC25A10 2.06 WBSCR14 0.83 1.51 –0.43 0.39 1.55 –0.17 1.27 MVD 0.56 PFKFB4 1.36 1.48 ME1 MDH1 0.90 0.25 –0.63 ACSS2 + 0.6) are marked in bold. 0.6) are marked 2.5 or cell mass < 1. List of genes targeted by siRNA pools that induce apoptosis or reduce cell mass in at least two cancer cell lines but not in RWPE1 siRNA pools that induce apoptosis or reduce cell mass in at least two 1. List of genes targeted by Table Name Symbol KGM FM LDM FM LDM FM LDM KGM FM LDM FM LDM FM LDM Pyruvate dehydrogenase Pyruvate dehydrogenase kinase, isozyme 3 AMP-activated kinase, a -2 catalytic subunit AMP-activated kinase, b -1 noncatalytic subunit acid Monocarboxylic 4 transporter Mitochondrial carrier; transporter dicarboxylate response Carbohydrate element binding protein Diphosphomevalonate Diphosphomevalonate decarboxylase 2 CoA transferase 3-Oxoacid OXCT26-Phosphofructo-2-kinase/ fructose-2,6-biphosphatase 4 0.19 –0.35 Aconitate hydratase 1, solubleAconitate hydratase ACO1 dependent, cytosolic control) is indicated. Results that passed the filtering criteria used for hit identification (apoptosis z-scores each gene, the effect of gene silencing on apoptosis (z-score) and cell mass (fold negative For NOTE: Glutathione peroxidase 4Glutathione peroxidase Similar to aconitase 2 GPX4 LOC441996Malic enzyme 1, NADP –0.23 1.22 Choline phosphotransferase 1Choline phosphotransferase CHPT1 synthaseCitrate 0.22 1, NAD Malate dehydrogenase (soluble) –0.03 CS 0.99 –0.12 –0.39 Acyl-CoA synthetase short- member 2 chain family 1 inhibitory factor ATPase ATPIF1 –0.58 0.09 –0.07 –0.46 >

APRIL 2012 CANCER DISCOVERY | 331

RESEARCH ARTICLE Ros et al.

A BC

300 500 600 Basal 1.6

) FM 4 400 FCCP LDM Oligomycin 1.2 200 400 300 0.8 200 100 200 100 0.4 ( D RFU/min/ m g protein) Glucose uptake rate Lipid synthesis rate Lactate secretion rate (nmoles/day/cell mass unit) (nmoles/day/cell mass unit) (cpm/cell mass unit 3 10

0 0 Oxygen consumption rate 0 0

PC3 PC3 PC3 PC3 RWPE1DU145LNCaP RWPE1DU145LNCaP DU145 LNCaP RWPE1 RWPE1 DU145 LNCaP D 1.5 RWPE1 (FM) 6 DU145 (FM) 2.0 LNCaP (FM) 2.0 PC3 (FM)

1.5 1.5 1.0 4

1.0 1.0

(fold of FM) 0.5 2 0.5 0.5 Caspase assay / CM 0 0 0 0

FM FM FM FM

FM -glucFM -glut FM -glucFM -glut FM -glucFM -glut FM -glucFM -glut FM low gluc FM low gluc FM low gluc FM low gluc 4 DU145 (LDM) 2.5 LNCaP (LDM) 3 PC3 (LDM)

2.0 3 2 1.5 2 1.0

(fold of LDM) 1 1 0.5 Caspase assay / CM

0 0 0

LDM LDM LDM

LDM -glucLDM -glut LDM -glucLDM -glut LDM -glucLDM -glut E LDM low gluc LDM low gluc LDM low gluc

Cell mass Cell mass

C75 (mg/mL) SB204990 (mg/mL)

Figure 1. Metabolic profiling and characterization of reliance on de novo lipid synthesis of prostate cell lines. A, glucose uptake (left) and lactate (right) secretion in 3 prostate cancer cell lines (DU145, LNCaP, and PC3) and a nonmalignant prostate epithelial cell line (RWPE1). B, OCR of cells in the presence or absence of either 10 μM FCCP (gold bars) or 0.2 μg/mL oligomycin (purple bars). Total protein concentration was used for normalization. C, de novo pyruvate-dependent lipid synthesis rates of cells incubated in FM or LDM. D, glucose and glutamine dependency of cells incubated in FM or LDM. Induction of apoptosis was determined 72 hours after incubation in the indicated medium (CM, cell mass). E, cell mass assay after 72 hours of treatment with increasing concentrations of the fatty acid synthase inhibitor C75 (left) or the ATP-citrate lyase inhibitor SB204990 (right) in FM or LDM. The significance of the difference between FM and LDM was tested for the 2 greatest concentrations of each inhibitor. All graphs show mean and SD of 3 independent replicates (*P # 0.05; **P # 0.005, Student t test).

332 | CANCER DISCOVERY APRIL 2012 www.aacrjournals.org

Functional Metabolic Screen in Prostate Cancer RESEARCH ARTICLE

C Apoptosis D Cell mass FM LDM FM LDM DU145 DU145 DU145 DU145 7 10 3 7 3 5 3 4 2 2 3 1 1 2 1 1 143 10 162 10 121 13 111 9

LNCaP PC3 LNCaP PC3 LNCaP PC3 LNCaP PC3 PRKAB1 PRKAB1 PFKFB4

Figure 2. Identification of metabolic enzymes required for prostate cancer cell survival. A, functional viability profiling of prostate epithelial cells. Prostate cell lines were transfected with Dharmacon SMARTpool siRNAs targeting 222 different metabolic genes in either FM (KGM for RWPE1, FM or LDM for cancer cell lines). Loss of viability was determined by measuring caspase activity and cell mass. The dendrogram shows a 2-way clustering of apoptosis z-scores for the different cell lines and medium conditions. B, genes for which silencing caused a significant induction in apoptosis or a loss of cell mass in at least 2 prostate cancer cell lines but has no effect in RWPE1 were used to perform a gene set–enrichment analysis (i.e., GSEA) using published datasets from human prostate cancer (GSE3325 and GSE6752). P-values represent the enrichment score of all probe sets representing these genes and indicate increased expression in samples from metastatic disease compared to primary site tumors. C, Venn diagrams showing the number of siRNA sequences that induce apoptosis in each cancer cell line in FM or LDM (z-score $ 2.5) but have no effect in RWPE1. D, Venn diagram showing the number of siRNA sequences that caused a reduction in cell mass in FM or LDM (cell mass #0.6 relative to siCtrl) but have no effect in RWPE1. E, P values and fold-change showing increased expression of PRKAB1 and PFKFB4 in metastatic prostate cancer compared with primary site tumors. Data are publicly available in Oncomine, and references are provided in the Supplementary Methods.

APRIL 2012 CANCER DISCOVERY | 333

RESEARCH ARTICLE Ros et al. is a bifunctional enzyme that regulates the levels of fructose- studied whether depletion of PRKAB1 or PFKFB4 affects the 2,6-bisphosphate (Fru-2,6-BP), a strong allosteric activator of ability of prostate cancer cells to form tumors in nude mice. the most important regulatory enzyme of glycolysis phos- For this purpose, we generated lentiviral inducible expres- phofructokinase 1 (PFK1). sion constructs encoding short-hairpin RNAs (shRNA) tar- We confirmed that the efficiency of mRNA depletion cor- geting the expression of PRKAB1 and PFKFB4 (TetOnPLKO). related with the biological effect for each individual siRNA Transduction of PC3 cells with these constructs resulted in sequence targeting these 2 genes (Supplementary Fig. S2C efficient depletion of mRNA and protein for both genes after and S2D) and that silencing of either gene had no effect in treatment with doxycycline (Fig. 4A and B) and led to a signifi- nonmalignant RWPE1 cells despite efficient depletion of the cant reduction in cell viability (Fig. 4C and D). respective mRNA (Supplementary Fig. S2E). Because RWPE1 To detect live tumor cells in vivo, PC3 cells expressing cells are grown in the presence of supplements, including epi- luciferase (PC3luc) were generated and transduced with len- dermal growth factor, we investigated whether the ability to tiviruses coding for inducible shRNA sequences. These cells tolerate depletion of PRKAB1 or PFKFB4 was determined by were then injected subcutaneously into the dorsal flank of the growth medium. DU145 cells did not tolerate culture in immunocompromised mice (nu/nu). At 12 days after injec- KGM (data not shown). However, both LNCaP and PC3 cells tion, animals with similar tumor burden (as detected by were still sensitive to PRKAB1 and PFKFB4 depletion un- luminescence analysis) were divided into 2 groups, and one der the medium conditions used to grow the nonmalignant group was treated with doxycycline. Results show a mod- RWPE1 cell line (Supplementary Fig. S2F). erate reduction in tumor volume in doxycycline-treated

animals injected with PC3luc-TetOnPLKO-shPRKAB1 Dependency on PRKAB1 and PFKFB4 for cells compared with the untreated control group (PC3luc- Cancer Cell Survival Is Isoform Specific and TetOnPLKO-shB1 #70; Fig. 4E). However, we did not ob- Not Restricted to Prostate Cancer Cells serve significant differences in bioluminescence intensity in We analyzed whether the requirement for PRKAB1 in the doxycycline-treated mice compared with untreated con- prostate cancer cell survival was specific to this isoform. trols (Supplementary Fig. S4A). Transfection with siRNAs targeting PRKAB2, the second iso- Histologic analysis of the tumors showed a small increase form of the AMPK β subunit, did not induce apoptosis in in cells with apoptotic morphology compared with the un- any of the prostate cancer cell lines (Fig. 3A), although both treated controls (Fig. 4J). When we investigated the efficiency isoforms PRKAB1 and PRKAB2 were efficiently depleted (Fig. of gene depletion, we observed a 50% to 60% reduction in 3B). In contrast, silencing of PRKAB1 was detrimental to the PRKAB1 mRNA levels after treatment with doxycycline 3 prostate cancer cells but had no effect in RWPE1 cells. (Fig. 4F) but failed to find a correlation between tumor size There are 4 genes that code for different PFK2 isoforms and PRKAB1 mRNA expression (Supplementary Fig. S4D). (PFKFB1-4), and these isoforms differ in their tissue-specific ex- Although we do not rule out the possibility that the level of pression as well as their kinetic properties (23). One of the splice PRKAB1 gene depletion was not sufficient to prevent tumor variants of the PFKFB3 isoform is induced by mitogens and growth, it is also possible that PC3 cells can tolerate the de- hypoxia and is overexpressed in tumors (24, 25) and is required pletion of this gene under the physiological metabolite levels for Ras-induced tumorigenesis (26). Expression of PFKFB4 is provided by the in vivo situation. induced in response to hypoxia (27) and in some types of cancer In contrast, treatment with doxycycline prevented the (28–30). We compared the effects of silencing of PFKFB3 and growth of PC3 cells expressing an inducible shRNA target- PFKFB4 In vivo PFKFB4 on prostate cancer cell survival in LDM. Depletion of ing (PC3luc-TetOnPLKO-shB4 #68; Fig. 4G). PFKFB4 selectively induced apoptosis in the 3 cancer cell lines detection of bioluminescence also confirmed a dramatic but not in RWPE1 (Fig. 3C). In contrast, silencing of PFKFB3 loss of viable tumor cells after treatment with doxycycline did not impair survival of any of the prostate cell lines, although (Supplementary Fig. S4B), suggesting that ablation of PFKFB4 both isoforms were efficiently depleted (Fig. 3D–F). mRNA not only prevents tumor formation but also induces We next investigated whether the requirement of PRKAB1 tumor regression. Notably, we were only able to recover one and PFKFB4 for cancer cell survival is specific to prostate small tumor from the treated cohort. Histologic analysis of cancer cells. PRKAB1 silencing induced apoptosis in a variety this tumor showed an increase in cells with apoptotic mor- of cancer cell lines from different tissues in both FM and phology compared with the untreated controls (Fig. 4J). LDM (Fig. 3G). Silencing of PFKFB4 reduced the viability of Experiments in which we used PC3 cells that expressed a the majority of the cell lines cultured in LDM, whereas the second, less-efficient, shRNA sequence targeting PFKFB4 effect in FM was observed in a smaller number of cell lines. (PC3luc-TetOnPLKO-shB4 #64; Fig. 4H) also showed a signifi- Sensitivity to silencing did not correlate with the level of ex- cant reduction in tumor volume and in viable tumor cells pression across the different cell lines (Supplementary Fig. upon treatment with doxycycline (Supplementary Fig. S4C). S3). Depletion of PRKAB2 or PFKFB3 had only minor effects, When we investigated the efficiency of gene depletion, we indicating that the specific dependency on PRKAB1 and observed a 50% to 60% reduction in PFKFB4 mRNA levels PFKFB4 extends to cancer cell lines from a variety of tissues. after treatment with doxycycline (Fig. 4I). Remarkably, we observed a positive correlation between tumor growth and PFKFB4 Is Required for Tumor Growth PFKFB4 mRNA expression (Supplementary Fig. S4E). Taken An important aspect of the metabolic adaptations of cancer together, these results demonstrate that PFKFB4 is required cells is to support the growth and survival under the condi- to support the survival of prostate cancer cells both in vitro tions encountered by a growing tumor in vivo. We therefore and in vivo.

334 | CANCER DISCOVERY APRIL 2012 www.aacrjournals.org

Functional Metabolic Screen in Prostate Cancer RESEARCH ARTICLE

A B

C D E

Figure 3. Dependency on PRKAB1 and PFKFB4 for cancer cell survival is isoform specific and not restricted to prostate cancer. A, cells were transfected with pooled siRNAs targeting PRKAB1 or PRKAB2, and induction of apoptosis was measured after 96 hours of culture in either KGM (RWPE1) or LDM. PLK1 silencing was used as positive control. B, lysates from cells transfected as in A were analyzed for expression of PRKAB1 and PRKAB2. C, cells were transfected with pooled siRNAs targeting PFKFB3 or PFKFB4, and induction of apoptosis was measured after 96 hours culture in either KGM (RWPE1) or LDM. PLK1 silencing was used as positive control. D, efficiency of PFKFB3 mRNA depletion by siRNA transfection. E, efficiency of PFKFB4 mRNA depletion by siRNA transfection. F, lysates from cells transfected as in C were analyzed for expression of PFKFB4. G, cancer cell lines from different tissues were transfected with siRNAs targeting PRKAB1, PRKAB2, PFKFB3, or PFKFB4. Caspase activity normalized to cell mass is displayed as fold change over siCtrl in FM or LDM. PLK1 and UBB1 were used as positive controls. All graphs show mean and SD of 3 independent replicates (*P # 0.05; **P # 0.005, Student t test).

APRIL 2012 CANCER DISCOVERY | 335

RESEARCH ARTICLE Ros et al.

A B

C D E F

Figure 4. In vivo silencing of PFKFB4 prevents tumor growth. A, PC3 cells were transduced with TetOnPLKO plasmids expressing inducible shRNAs targeting PRKAB1 (sequences #70 and #71) or PFKFB4 (#64 and #68). Expression of the shRNA was induced with 1 μg/mL doxycycline in FM for 96 hours and PRKAB1 and PFKFB4 mRNA levels were determined. Values represent the means and SEM of three independent experiments. B, lysates from cells treated as in A were analyzed for expression of PRKAB1 and PFKFB4 proteins. C, the same cells were grown in the presence or absence of doxycycline for 10 days, fixed, and stained with crystal violet. D, colorimetric quantitation of crystal violet staining shown in panel C. Values represent the mean and SEM of 4 independent experiments each performed in duplicate. E, nude mice (nu/nu) were injected subcutaneously × 6 with 1 10 PC3luc-TetOnPLKO-shPRKAB1 #70 cells. Silencing of PRKAB1 was induced in the treatment group by addition of (continued)

PFKFB4 Regulates the Levels of the Allosteric the phosphatase activity being slightly stronger [32 (Fig. 5B)]. Activator Fru-2,6-BP in Prostate Cancer Cells We investigated whether levels of this metabolite are affected Our results shown so far have identified PFKFB4 as a gene by silencing of PFKFB3 or PFKFB4 in the 4 prostate cell lines. selectively required to support the viability of prostate cancer Silencing of PFKFB3 caused a substantial reduction in Fru- cells. PFK2 isoforms play an important role in controlling the 2,6-BP levels in RWPE1 cells and a decrease in their glycolytic distribution of metabolites between glycolysis and the pen- rate (Supplementary Fig. S5A). However, silencing of PFKFB3 tose phosphate pathway (Fig. 5A). The metabolic properties did not affect the concentration of this metabolite in any of of the different PFK2 isoforms and their effect on prostate the prostate cancer cell lines (Fig. 5D). Depletion of PFKFB4 cell survival are summarized in Fig. 5B. To investigate the caused a significant increase in Fru-2,6-BP levels in each of the role of PFKFB4 in prostate cancer metabolism and survival, 3 prostate cancer cells but only had a minor effect in RWPE1 we first analyzed expression of PFKFB3 and PFKFB4 in the cells (Fig. 5D). The small reduction in Fru-2,6-BP levels after different prostate cell lines. Expression of PFKFB3 was great- PFKFB4 depletion in RWPE1 cells could be caused by com- est in DU145 cells, but LNCaP and PC3 showed similar levels pensation by other PFK2 isoforms or caused by differential of PFKFB3 expression compared with the noncancer cell line regulation of the relative kinase/phosphatase activity in these RWPE1 (Fig. 5C). In contrast, PFKFB4 expression was consis- cells. In addition, the fold induction of apoptosis after silenc- tently greater in the 3 cancer cell lines in FM and was further ing of PFKFB4 was decreased in glucose-depleted medium increased after incubation with LDM (Fig. 5C). (Supplementary Fig. S5B). PFK2 regulates the steady-state concentration of Fru-2,6-BP. These results show differences in the control of Fru-2,6-BP Although the PFKFB3 isoform has a greater kinase activity levels between RWPE1 and the glucose-dependent prostate can- and strongly favors the formation of Fru-2,6-BP, resulting in cer cells and suggest that PFKFB4 supports their viability by pre- enhanced glycolytic rate (31), PFKFB4 is more balanced, with venting the accumulation of this important allosteric regulator.

336 | CANCER DISCOVERY APRIL 2012 www.aacrjournals.org

Functional Metabolic Screen in Prostate Cancer RESEARCH ARTICLE

G H

I J

Figure 4 (continued). doxycycline to the food pellet (day 12). Mean tumor volumes for each group at the indicated time points are shown. Error bars represent SEM. F, in vivo depletion of PRKAB1 mRNA after doxycycline treatment in tumors recovered at day 35. G, nude mice (nu/nu) × 6 were injected subcutaneously with 1 10 PC3luc-TetOnPLKO-shPFKFB4 #68 cells and treated as in panel E. H, nude mice (nu/nu) were injected × 6 in vivo subcutaneously with 1 10 PC3luc-TetOnPLKO-shPFKFB4 #64 cells and treated as in panel E. I, depletion of PFKFB4 mRNA after doxycycline treatment in tumors recovered at day 35. J, histologic analysis of PC3luc-TetOnPLKO-shPRKAB1 #70 tumors or PC3-TetOnPLKO- shPFKFB4 #68 tumors in control mice (No DOX) or after 35 days of doxycycline (DOX) treatment. Arrows indicate cell with apoptotic morphology. In all cases, P values were determined by use of the Student t test; *P # 0.05; **P # 0.005.

PFKFB4 Maintains Cellular Redox Balance in PFKFB4 in LDM caused a significant reduction in lipid syn- Prostate Cancer Cells thesis in the 3 prostate cancer cell lines (Fig. 6C). Accumulation of Fru-2,6-BP could lead to decreased avail- We also investigated the effect of PFKFB4 depletion on ability of metabolic intermediates for the pentose phosphate ROS accumulation. PFKFB4 silencing had no effect on ROS pathway (Fig. 5A). One of the main functions of the oxidative levels in RWPE1 cells (Fig. 6D). In contrast, we observed a arm of the pentose phosphate pathway is the generation of strong increase in ROS levels upon PFKFB4 depletion in the NADPH for biosynthetic processes and for the maintenance cancer cell lines (Fig. 6E). This induction was comparable of cellular redox balance. We found that silencing of PFKFB4 with ROS levels caused by depletion of superoxide dismutase leads to a decrease in the concentration of several pentose 2 (SOD2), a major mitochondrial ROS detoxifying enzyme. phosphate pathway metabolites (Supplementary Fig. S5C) Of importance, the differential sensitivity to PFKFB4 and therefore investigated whether ablation of PFKFB4 af- silencing in the prostate cell lines is not caused by lower ex- fects redox balance in prostate cancer cells. pression of the ROS-detoxifying enzymes catalase and SOD2, Silencing of PFKFB4 in DU145 cells resulted in a signifi- which are expressed at similar or higher levels compared with cant decrease in the levels of NADPH (Fig. 6A). NADPH is RWPE1 (Fig. 6F). We also investigated the expression of several required to maintain cellular stores of reduced glutathione genes involved in the oxidative stress response and observed (GSH), an important antioxidant that prevents the accumu- greater levels of thioredoxin, an oxidoreductase involved in lation of reactive oxygen species (ROS). PFKFB4 silencing ROS detoxification (33, 34), in PFKFB4-depleted xenograft tu- caused a reduction in the GSH/oxidized glutathione (GSSG) mors (Supplementary Fig. S5D). Furthermore, treatment with ratio in DU145 cells (Fig. 6B). NADPH is also important the chemical antioxidant 4-hydroxy-TEMPO, a SOD2 mimetic to provide the reducing power for anabolic reactions, such ROS scavenger, fully rescued viability after PFKFB4 silencing as the de novo synthesis of fatty acids. Indeed, silencing of in the 3 prostate cancer cell lines (Fig. 6G), demonstrating that

APRIL 2012 CANCER DISCOVERY | 337

RESEARCH ARTICLE Ros et al.

A Figure 5. PFKFB4 supports prostate cancer cell survival by regulating Fru- 2,6-BP levels. A, schematic representation of the regulation of glycolytic flux by the allosteric activator Fru-2,6-BP. B, metabolic properties of the different PFK2 isoforms and their effect on prostate cell survival. C, prostate cells were grown in FM or LDM for 24 hours and PFKFB3 and PFKFB4 mRNA levels were determined. Graphs show means and SEM of 3 independent experiments. D, cells were transfected B with siRNA targeting PFKFB3 or PFKFB4 and cultured in KGM (RWPE1) or LDM (DU145, LNCaP and PC3) for 60 hours. Intracellular levels of Fru-2,6-BP were determined. Graphs show means and SEM of 6 independent experiments. In all cases, P values were determined by use of the Student t test; *P # 0.05; **P # 0.005. C

PFKFB4 silencing compromises viability of prostate cancer support a model in which PFKFB4 is required to maintain cells by decreasing their ability to manage ROS accumulation. the balance between energy production and cellular redox Finally, we hypothesized that the detrimental effect of Fru- control (Fig. 6J). 2,6-BP accumulation in response to PFKFB4 silencing could be rescued by cosilencing PFKFB3, the PFK2 isoform that strongly favors formation of Fru-2,6-BP. Indeed, cosilenc- DISCUSSION ing of PFKFB3 significantly reduced apoptosis after PFKFB4 In this study, we have applied a functional genetic ap- depletion in DU145 and PC3 cells (Fig. 6H) and blocked the proach to investigate metabolic adaptations that differen- induction of ROS levels (Fig. 6I), while not affecting PFKFB4 tiate prostate cancer cell lines from nonmalignant prostate silencing efficiency (Supplementary Fig. S5E). This finding epithelial cells. This strategy proved to be well-suited for the strongly suggests that PFK2 activity is crucial to control the identification of metabolic processes that are essential for balance between glycolytic flux and redox regulation by the prostate cancer cell survival. Overall, the overlap between the pentose phosphate pathway. 4 cell lines used in this study was relatively small. This find- Taken together, these results implicate PFKFB4 as an im- ing is most likely attributable to the diversity of the cell lines portant regulator of prostate cancer cell survival. Our results used. The 3 prostate cancer cell lines differ in the mutation

338 | CANCER DISCOVERY APRIL 2012 www.aacrjournals.org

Functional Metabolic Screen in Prostate Cancer RESEARCH ARTICLE ) A P 5 0.0068 B P 5 C 4 DU145 LNCaP PC3 D 0.0021 40 P 5 10 P 5 15 P 5 RWPE1 70 1.25 0.023 0.026 0.065 120000 FM 60 8 1.00 30 90000 10 LDM 50 6 40 0.75 20 60000 4 30 0.50 5 10 20 2 30000 0.25 NADPH levels (mean ﬂuorescence) GSH/GSSG ratio 10 Lipid synthesis rate DCFDA incorporation DCFDA (pmoles/ m g protein)

0 0 0 2 0 (cpm/cell mass unit 3 10

0 0 H O 2 siCtrl siCtrl siCtrl siCtrl siCtrl siCtrl siCtrl 2 siCtrl siCtrl siSOD2 siPFKFB4 siPFKFB4 siPFKFB4 siPFKFB4 siPFKFB4 siPFKFB4 siCtrl 1 H

E DU145 LNCaP PC3 F KGM KGM 100000 40000 40000 FM FM 2.5 LDM 15 LDM 80000 30000 30000 2.0 10 60000 1.5 20000 20000 10000 10000 10000 1.0

CAT mRNA CAT 5 5000 5000 5000 SOD2 mRNA 0.5 (mean ﬂuorescence) DCFDA incorporation DCFDA 2 0 0 0 0 0 H

2 2 O O O 2 2 2 2 PC3 PC3 siCtrl siCtrl siCtrl 1 H siSOD2 siSOD2 1 H siSOD2 RWPE1DU145LNCaP RWPE1DU145LNCaP siPFKFB4 siPFKFB4 siPFKFB4 siCtrl siCtrl 1 H siCtrl G DU145 LNCaP PC3 H 2.5 2.0 3

2.0 1.5 2 1.5 1.0 1.0 1

(fold of siCtrl) 0.5 0.5 0 0 0 Caspase activity / CM M M mM 1 2 mM 1 m 2 mM 1 mM2 m 4 mM Solvent Solvent Solvent TEMPOL TEMPOL TEMPOL

I J Metabolic effects of PFKFB4 silencing Glucose Pentose Glucose-6-phosphate phosphate pathway PFKFB3 Fructose-6-phosphate PFK1 NADP1 NADPH Fru-2,6-BP Fru-1,6-BP ROS

Pyruvate Lactate

Figure 6. PFKFB4 is required to maintain NADPH levels and cellular redox balance in prostate cancer cells. A, DU145 cells were transfected with siRNA targeting PFKFB4 and grown in LDM for 60 hours. NADPH levels were determined. Graphs show means and SEM of 8 independent experiments. B, DU145 cells were transfected with siRNAs and grown in LDM for 60 hours. Intracellular levels of reduced GSH and GSSG were determined. Graphs show means and SD of 2 independent experiments, each performed in triplicate. C, acetate-dependent de novo lipid synthesis rates was determined in cells transfected as indicated in FM or LDM. Graphs show means and SD of 3 independent experiments. D, ROS levels in

RWPE1 cells after depletion of PFKFB4. Silencing of SOD2 or treatment with 1 mM H2O2 for 1 hour was used as positive controls. Graphs show means of 2 independent experiments. E, ROS levels in DU145, LNCaP, and PC3 cells incubated in LDM after depletion of PFKFB4 as in panel E. Graphs show means and SEM of 3 independent experiments. F, prostate cells were grown in FM or LDM for 24 hours. SOD2 and catalase (CAT) mRNA levels were determined by quantitative real-time PCR. Graphs show means and SEM of 3 independent experiments. G, prostate cancer cells were depleted of PFKFB4 and treated with the indicated concentrations of 4-hydroxy-TEMPO (TEMPOL). Induction of apoptosis was measured after 96 hours in LDM. Graphs show means and SEM of 3 independent experiments. H, cells were transfected as indicated and induction of apoptosis was measured after 96 hours of culture in LDM. PLK1 silencing was used as positive control. Graphs show the means and SEM of 3 independent experiments. I, ROS levels in cells treated as in panel H. Graphs show the means and SEM of 3 independent experiments. J, schematic model shows the metabolic effects of PFKFB4 silencing in prostate cancer cells. In all cases, P values were determined by use of the Student t test; *P # 0.05; **P # 0.005.

APRIL 2012 CANCER DISCOVERY | 339

RESEARCH ARTICLE Ros et al. status of cancer genes and have been isolated from metasta- can be induced by ROS after glucose starvation in an AMPK- ses in different organs. Notably, LNCaP is the only androgen- dependent manner (39), indicating that this enzyme can be dependent cell line in this study. Despite this diversity, the both upstream and downstream of oxidative stress. effect of siRNA mediated gene silencing of the panel of genes Induction of lipid biosynthesis increases cellular demand for used in this study clearly separated the cancer cell lines from NADPH and the dependency of prostate cancer cells on PFKFB4 the nonmalignant line, indicating that the different cancer was particularly prominent under conditions of lipoprotein cell lines share common metabolic features. starvation in vitro. The increased biosynthetic demand of cancer We only found 2 genes that showed similar effects in all cells renders them highly dependent on NADPH generation by cancer cell lines. This may be attributable to the stringent the pentose phosphate pathway and the importance of ROS de- selection criteria but could also indicate redundancy between toxification for tumorigenesis has been demonstrated recently different genes within each pathway. Interestingly, both of (40). The dramatic effect of PFKFB4 depletion on tumor growth these genes, PRKAB1 and PFKFB4, are expressed at greater in vivo suggests that the balance between glycolysis and pen- levels in metastatic prostate cancer compared with primary tose phosphate pathway is crucial to support tumor cells under site tumors in public data from human prostate cancer. physiologic conditions. Furthermore, depletion of PFKFB4 pre- PRKAB1 codes for one of the isoforms of the β subunits of vented the induction of lipid synthesis, required for growth of AMPK. AMPK is generally considered to be part of a tumor cancer cells, possibly by reducing NAPDH availability. suppressor pathway (21). Loss of an AMPK upstream kinase Other enzymes have been shown to play a role in pre- + LKB1 decreases the latency of prostate lesions in PTEN /– venting oxidative stress by balancing glycolytic and pentose mice (35). However, certain cancers may depend on AMPK to phosphate pathway flux. Depletion of TIGAR inhibits the downregulate energy-consuming processes under conditions pentose phosphate pathway and induces oxidative stress and of nutrient starvation. Despite the strong effect of PRKAB1 autophagy in cancer cells under metabolic stress (37, 41). In silencing on cancer cell survival in vitro, depletion of this gene contrast, reduction of glycolytic flux by inhibition of PKM2 only had a minor effect on tumor growth in vivo, suggesting is required to enhance antioxidant production by increasing that PRKAB1 could be dispensable under conditions encoun- pentose phosphate pathway activity in response to oxidative tered by prostate cancer cells in vivo. stress (42). The second gene identified in our study was PFKFB4, an In summary, this study demonstrates the usefulness of isoform of PFK2 (22). We found that silencing of PFKFB4 functional screens to identify metabolic weaknesses in cancer not only selectively induced apoptosis in prostate cancer cells cells and highlights the importance of metabolic regulation but also completely blocked tumor growth in vivo. Expression to maintain the balance between bioenergetics and antioxi- of PFKFB4 has been shown to be increased in breast, colon, dant production for cancer cell survival. Targeting this bal- gastric, and prostate cancer (28, 30). Our analysis of public ance may also provide novel strategies for cancer therapy. microarray data showed that PFKFB4 expression is greater in metastatic prostate cancer compared with primary site tumors. METHODS Indeed, the 3 prostate cancer cell lines used in this study, which Cell Culture and Reagents originate from metastases, showed increased expression of PFKFB4 RWPE1 and LNCaP (clone FGC) were obtained from the American compared with nonmalignant RWPE1 cells. Type Culture Collection. All other cells lines were from the LRI Cell Although the core structure is highly conserved between Services. RWPE1 cells were grown in Keratinocyte serum-free me- the different PFK2 isozymes, the relative kinase and phos- dium (Gibco) supplemented with epidermal growth factor and bo- phatase activity of these bifunctional enzymes shows large vine pituitary extract (KGM). All other cell lines were grown in RPMI, variation [Fig. 5B (23)], and all PFK2 isoforms can be subject DMEM, or DMEM/F12 supplemented with 10% fetal calf serum to posttranslational modifications, which can also modu- (FM; PAA Laboratories) or 1% fetal bovine lipoprotein-deficient se- late the relative activities of their catalytic domains (36). We rum (LDM; Intracel), glutamine, and penicillin/streptomycin. A full found that silencing of PFKFB4 increased levels of Fru-2,6-BP description of the additional cell lines used, where they were ob- levels selectively in the prostate cancer cells, suggesting that tained from, and the method and date of cell line authentication is detailed in the Supplementary Methods. it was mainly functioning as a fructose-2,6-bisphosphatase. Polyclonal PRKAB1 (A4856) and monoclonal PRKAB2 (clone This increase in Fru-2,6-BP, an allosteric activator of the 2G9) antibodies were from Sigma-Aldrich. Antibodies for PFKFB4 master regulator of glycolysis PFK1, should divert glucose were from Abgent (AP8154C). All other primary antibodies were from 6-phosphate towards the glycolytic pathway, thereby deplet- Cell Signaling. Horseradish peroxidase-conjugated anti-GAPDH ing the pentose phosphate pathway. This would explain why and anti-β-actin antibodies were from Abcam and Sigma-Aldrich, prostate cancer cells showed lower NADPH and reduced glu- respectively. ATP-citrate lyase inhibitor SB204990 was provided by tathione levels after PFKFB4 silencing, resulting in enhanced GlaxoSmithKline (Stevenage). All other reagents were from Sigma. oxidative stress and cell death (see model in Fig. 6J). Glucose Uptake, Lactate Secretion, A similar role in the regulation of cellular ROS levels has and Oxygen Consumption been demonstrated for TIGAR, a p53-inducible protein with fructose-2,6-bisphosphatase activity similar to PFK2 (37). Glucose and lactate concentrations in media incubated with or without cells were determined by the use of glucose and lactate assay Furthermore, degradation of PFKFB3 by the action of the kits from BioVision. Cell mass was used for normalization. Oxygen E3 ubiquitin ligase anaphase-promoting complex/cyclosome consumption was measured with the BD Oxygen Biosensor Systems (APC/C)-Cdh1 has been shown to promote antioxidant for- (OBS) from BD Bioscience with 1 × 106 cells in 200 μL of medium mation in neurons by enhancing the activity of the pentose in the presence or absence of 10 μM FCCP or 0.2 μg/m: oligomycin. phosphate pathway (38). Interestingly, PFKFB3 expression Oxygen consumption rate was normalized to protein concentration.

340 | CANCER DISCOVERY APRIL 2012 www.aacrjournals.org

Functional Metabolic Screen in Prostate Cancer RESEARCH ARTICLE

Lipid Synthesis 1–4, RISC-free siRNA, and a SMARTPool targeting PLK1 were used as negative and positive controls, respectively. A gene was considered Cells were incubated in medium containing 2.5 μCi/mL [2-14C]- validated if at least 2 individual siRNA reduced cell mass to less than pyruvate (166 μM final concentration) or 10 μCi/mL [1-14C] acetate 70% of the average of the negative controls or increased normalized (85 μM final concentration, Perkin Elmer) for 4 hours. After washing caspase activity more than 1.8-fold. 3 times in PBS, cells were lysed in 0.5% Triton X-100. Lipids were extracted by successive addition of 2 mL of methanol, 2 mL of chlo- Generation of Doxycycline-Inducible shRNA Cell Lines roform, and 1 mL of dH O. Phases were separated by centrifugation 2 shRNA sequences targeting PFKFB4 or PRKAB1 were cloned before the organic phase was dried and used for scintillation count- into the TetOnPLKO lentiviral vector (Addgene; refs. 47 and 48). ing. Results were normalized to cell mass. Lentiviruses were produced by cotransfecting HEK 293T with the shRNA plasmid and the packaging plasmids pCMVΔR8.91 (gag- Metabolite Extraction and Analysis by Mass Spectrometry pol) and pMD.G [VSV-G glycoprotein (49)]. Supernatants contain- Cells were grown in 6-well plates, washed with warm PBS, and fro- ing lentiviruses were collected 72 hours after transfection, mixed zen in liquid nitrogen. Intracellular metabolites were extracted by ad- with polybrene (8 μg/mL), and used to infect PC3 cells. Fresh me- 13 dition of a prewarmed 70% (v/v) ethanol aqueous at 75°C. C-labeled dium-containing puromycin (3 μg/mL) was added after 24 hours, biomass was added to serve as internal standard for metabolite analy- and cells were selected for at least 48 hours before being used in sis. The cell mixture was centrifuged, the supernatant collected, and experiments. dried. Metabolites were dissolved in ultrapure water. For the analysis of extracellular metabolites, the medium was rapidly collected and Xenograft Experiments frozen in liquid nitrogen. Internal standards (norvaline and glutaric PC3 cells stably expressing the ecotropic retrovirus receptor were acid) were added before drying. generated by transfection with the pWZLneoEcoR vector. To gen- Intracellular metabolites were analyzed by ion-pairing ultrahigh- erate PC3 expressing luciferase cells, the retroviral pBabe-Lucifer- performance liquid chromatography-tandem mass spectrometry (43). ase vector (a generous gift from M. Murillo, LRI) was packaged in Eco Extracellular metabolites were analyzed by gas chromatography-time Phoenix cells and used to infect the PC3 R-neo cells. PC3luc cells of flight-mass spectrometry after off-line methoxyamination and were then infected with lentiviral vectors and selected as described just-in-time N-tert-butyldimethylsilyl-N-methyltrifluoroacetamide previously. 6 derivatization. Metabolite quantification for both intracellular Nude mice (nu/nu) were injected subcutaneously with 10 PC3luc- and extracellular metabolites was obtained by means of calibration TetOnPLKO-PRKAB1 cells (sequence 70) or PC3luc-TetOnPLKO- curves of a mixture of standard compounds (44). Metabolite data PFKFB4 (sequences 68 and 64) cells into the dorsal flank. After 12 were analyzed by an in-house MATLAB 2010 (Mathworks) applica- days, animals were subdivided into 2 experimental groups, a doxy- tion for peak integration, internal standard correction, and absolute cycline-treated group and a nontreated group. Bioluminescence im- quantification of metabolites. Absolute amounts were normalized to aging was used to split mice with similar tumor burden equally into cell number or cell mass in silencing experiments. the different treatment groups. For induction of shRNA expression, mice were treated with doxycycline (doxycycline diet, 0.2 g/kg siRNA Screen and Data Analysis food pellets, TD.98186; Harlan Laboratories), and tumor growth Cells were reverse-transfected with 25 nM of Dharmacon was followed over the course of 23 days. Tumor volume was deter- SMARTpools in 96-well plates using Dharmafect 1 reagent. After mined by use of the ellipsoidal volume formula: 1/2 × length × 24 hours, culture medium was replaced with fresh medium. Then, width2. All animal experiments were performed according to U.K. 72 hours later, caspase 3/7 activity was measured with a fluorescent Home Office guidelines. substrate (Invitrogen). Cells were then fixed with trichloroacetic acid and stained with Sulforhodamine B assay (Sigma-Aldrich) to deter- Determination of Fru-2,6-BP Levels mine total protein content (cell mass). Two independent screens were Cells were homogenized in 0.1 M NaOH and 0.1% Triton X-100, performed, each with triplicate transfections. heated to 80°C for 10 minutes, and centrifuged for 5 minutes. The Caspase 3/7 activity was normalized to cell mass. Normalized supernatant was neutralized with acetic acid in 20 mM HEPES. Fru- caspase activity and cell mass values were then normalized to the 2,6-BP was determined as previously described (50). Protein concen- median value of all samples within each plate (plate centering), and tration was used for normalization. the average of the 6 replicates was calculated. Histograms determined by normalized caspase activity and cell mass data are shown Determination of NADPH Levels and GSH/GSSG Ratio in Supplementary Figure S2B. Z-scores were calculated from nor- For detection of NADPH, cells were lysed in buffer containing 20 malized caspase activity values using the median absolute deviation mM nicotinamide, 20 mM NaHCO3, and 100 mM Na2CO3. Cleared (45) as recommended for small-scale screens (46). Z-scores are pro- supernatants were incubated for 30 minutes at 60°C. Then, 20 μL vided in Supplementary Table S2. Cell mass results were expressed of supernatant was mixed with 160 μL of NADP-cycling buffer (100 relative to the average of the negative controls and are provided in mM Tris-Cl, pH 8.0; 0.5 mM thiazolyl blue; 2 mM phenazine ethosul- Supplementary Table S2. A 2-way cluster analysis was generated us- fate; 5 mM EDTA; 1.3 IU glucose-6-phosphate dehydrogenase). After ing Pearson correlation (centered) and average linkage clustering a 1-minute incubation in the dark at 30°C, 1 mM glucose-6-phos- with Cluster 3.0 and Java TreeView (1.1.3). phate was added, and the change in absorbance at 570 nm was mea- Gene expression data were downloaded from Gene Expression sured every 30 seconds for 4 minutes at 30°C. Protein concentration Omnibus and analyzed using Gene Set Enrichment Analysis (see was used for normalization. The GSH/GSSG ratio was determined Supplementary Methods). Normalized expression data and P val- with a kit (Biovision). ues for PRKAB1 and PFKFB4 were downloaded from Oncomine (Compendia Bioscience). Citations for individual studies are in- Determination of ROS Levels cluded in the Supplementary Methods. ROS levels were determined by incubating cells for 30 minutes ° μ at 37 C in medium with 3 M CM-H2-DCFDA (Molecular Probes) siRNA Screen Validation followed by fluorescence-activated cell-sorting analysis. 4′,6 Cells were transfected with the original SMARTPool or the 4 in- Diamidino 2 phenylindole (i.e., DAPI) was added before analysis to dividual siRNA oligonucleotides. siGENOME nontargeting controls exclude dead cells.

APRIL 2012 CANCER DISCOVERY | 341

RESEARCH ARTICLE Ros et al.

Disclosure of Potential Conflicts of Interest 14. Beckers A, Organe S, Timmermans L, Scheys K, Peeters A, Brusselmans K, et al. Chemical inhibition of acetyl-CoA carboxyl- A. Schulze provides minor consultancy work for Astra Zeneca. ase induces growth arrest and cytotoxicity selectively in cancer cells. Cancer research 2007;67:8180–7. Acknowledgments 15. Mycielska ME, Patel A, Rizaner N, Mazurek MP, Keun H, Ganapathy The authors thank J. Downward and A. Behrens for critical V, et al. Citrate transport and metabolism in mammalian cells: pros- discussion. They also thank B. Saunders (LRI High Throughput tate epithelial cells and prostate cancer. Bioessays 2009;31:10–20. Screening Service) for help with data analysis; Julie Bee and 16. Cairns RA, Harris IS, Mak TW. Regulation of cancer cell metabolism. Francois Lassailly (LRI) for help with in vivo experiments and Nat Rev Cancer 2011;11:85–95. in vivo imaging; and Emma Nye, Richard Stone, and Gordon 17. Bello D, Webber MM, Kleinman HK, Wartinger DD, Rhim JS. Androgen Stamp (LRI Histopathology Service) for tissue analysis. They responsive adult human prostatic epithelial cell lines immortalized by also thank the LRI Research Services for technical support and human papillomavirus 18. Carcinogenesis 1997;18:1215–23. B. Griffiths for help with tissue culture. The authors are grateful 18. DeBerardinis RJ, Mancuso A, Daikhin E, Nissim I, Yudkoff M, Wehrli to J.J. Guinovart (IRB, Barcelona) for help with the Fru-2,6-BP S, et al. Beyond aerobic glycolysis: transformed cells can engage in glutamine metabolism that exceeds the requirement for protein and measurements. nucleotide synthesis. Proc Natl Acad Sci U S A 2007;104:19345–50. Grant Support 19. Pizer ES, Pflug BR, Bova GS, Han WF, Udan MS, Nelson JB. Increased fatty acid synthase as a therapeutic target in androgen-independent This study was funded by Cancer Research UK. C.R. Santos was prostate cancer progression. Prostate 2001;47:102–10. supported by an EMBO long-term fellowship (ALTF 102-2007). 20. Seidlitz EP, Sharma MK, Saikali Z, Ghert M, Singh G. Cancer cell lines release glutamate into the extracellular environment. Clin Exp Received September 12, 2011; revised February 14, 2012; accepted Metastasis 2009;26:781–7. February 15, 2012; published OnlineFirst March 22, 2012. 21. Shackelford DB, Shaw RJ. The LKB1-AMPK pathway: metabolism and growth control in tumour suppression. Nat Rev Cancer 2009;9:563–75. REFERENCES 22. Sakata J, Abe Y, Uyeda K. Molecular cloning of the DNA and ex- 1. Vander Heiden MG, Cantley LC, Thompson CB. Understanding the pression and characterization of rat testes fructose-6-phosphate, Warburg effect: the metabolic requirements of cell proliferation. 2-kinase: fructose-2,6-bisphosphatase. J Biol Chem 1991;266:15764–70. Science 2009;324:1029–33. 23. Okar DA, Manzano A, Navarro-Sabate A, Riera L, Bartrons R, Lange 2. Deberardinis RJ, Sayed N, Ditsworth D, Thompson CB. Brick by AJ. PFK-2/FBPase-2: maker and breaker of the essential biofactor brick: metabolism and tumor cell growth. Curr Opin Genet Dev fructose-2,6-bisphosphate. Trends Biochem Sci 2001;26:30–5. 2008;18:54–61. 24. Atsumi T, Chesney J, Metz C, Leng L, Donnelly S, Makita Z, et al. 3. Kroemer G, Pouyssegur J. Tumor cell metabolism: cancer’s Achilles’ High expression of inducible 6-phosphofructo-2-kinase/fructose- heel. Cancer Cell 2008;13:472–82. 2,6-bisphosphatase (iPFK-2; PFKFB3) in human cancers. Cancer Res 4. Parsons DW, Jones S, Zhang X, Lin JC, Leary RJ, Angenendt P, et al. 2002;62:5881–7. An integrated genomic analysis of human glioblastoma multiforme. 25. Minchenko A, Leshchinsky I, Opentanova I, Sang N, Srinivas V, Science 2008;321:1807–12. Armstead V, et al. Hypoxia-inducible factor-1-mediated expression 5. Christofk HR, Vander Heiden MG, Harris MH, Ramanathan A, of the 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase-3 Gerszten RE, Wei R, et al. The M2 splice isoform of pyruvate kinase (PFKFB3) gene. Its possible role in the Warburg effect. J Biol Chem is important for cancer metabolism and tumour growth. Nature 2002;277:6183–7. 2008;452:230–3. 26. Telang S, Yalcin A, Clem AL, Bucala R, Lane AN, Eaton JW, et al. 6. Fantin VR, St-Pierre J, Leder P. Attenuation of LDH-A expression Ras transformation requires metabolic control by 6-phosphofructo- uncovers a link between glycolysis, mitochondrial physiology, and 2-kinase. Oncogene 2006;25:7225–34. tumor maintenance. Cancer Cell 2006;9:425–34. 27. Minchenko O, Opentanova I, Minchenko D, Ogura T, Esumi H. 7. Hatzivassiliou G, Zhao F, Bauer DE, Andreadis C, Shaw AN, Dhanak Hypoxia induces transcription of 6-phosphofructo-2-kinase/fruc- D, et al. ATP citrate lyase inhibition can suppress tumor cell growth. tose-2,6-biphosphatase-4 gene via hypoxia-inducible factor-1alpha Cancer Cell 2005;8:311–21. activation. FEBS Lett 2004;576:14–20. 8. Menendez JA, Vellon L, Mehmi I, Oza BP, Ropero S, Colomer R, et al. 28. Bobarykina AY, Minchenko DO, Opentanova IL, Moenner M, Caro Inhibition of fatty acid synthase (FAS) suppresses HER2/neu (erbB- J, Esumi H, et al. Hypoxic regulation of PFKFB-3 and PFKFB-4 2) oncogene overexpression in cancer cells. Proc Natl Acad Sci U S A gene expression in gastric and pancreatic cancer cell lines and 2004;101:10715–20. expression of PFKFB genes in gastric cancers. Acta Biochim Pol 9. Wang JB, Erickson JW, Fuji R, Ramachandran S, Gao P, Dinavahi R, 2006;53:789–99. et al. Targeting mitochondrial glutaminase activity inhibits onco- 29. Goidts V, Bageritz J, Puccio L, Nakata S, Zapatka M, Barbus S, et al. genic transformation. Cancer Cell 2010;18:207–19. RNAi screening in glioma stem-like cells identifies PFKFB4 as a key 10. Locasale JW, Grassian AR, Melman T, Lyssiotis CA, Mattaini KR, molecule important for cancer cell survival. Oncogene 2011 Nov 7. Bass AJ, et al. Phosphoglycerate dehydrogenase diverts glycolytic flux [Epub ahead of print]. and contributes to oncogenesis. Nat Genet 2011;43:869–74. 30. Minchenko OH, Ochiai A, Opentanova IL, Ogura T, Minchenko DO, 11. Possemato R, Marks KM, Shaul YD, Pacold ME, Kim D, Birsoy K, Caro J, et al. Overexpression of 6-phosphofructo-2-kinase/fructose- et al. Functional genomics reveal that the serine synthesis pathway is 2,6-bisphosphatase-4 in the human breast and colon malignant essential in breast cancer. Nature 2011;476:346–50. tumors. Biochimie 2005;87:1005–10. 12. Moon JS, Jin WJ, Kwak JH, Kim HJ, Yun MJ, Kim JW, et al. Androgen 31. Sakakibara R, Kato M, Okamura N, Nakagawa T, Komada Y, Tominaga stimulates glycolysis for de novo lipid synthesis by increasing the ac- N, et al. Characterization of a human placental fructose-6-phosphate, tivities of hexokinase 2 and 6-phosphofructo-2-kinase/fructose-2,6- 2-kinase/fructose-2,6-bisphosphatase. J Biochem 1997;122:122–8. bisphosphatase 2 in prostate cancer cells. Biochem J 2010;433:225–33. 32. Sakakibara R, Uemura M, Hirata T, Okamura N, Kato M. Human 13. De Schrijver E, Brusselmans K, Heyns W, Verhoeven G, Swinnen placental fructose-6-phosphate,2-kinase/fructose-2,6-bisphosphatase: JV. RNA interference-mediated silencing of the fatty acid syn- its isozymic form, expression and characterization. Biosci Biotechnol thase gene attenuates growth and induces morphological changes Biochem 1997;61:1949–52. and apoptosis of LNCaP prostate cancer cells. Cancer research 33. Taniguchi Y, Taniguchi-Ueda Y, Mori K, Yodoi J. A novel promoter 2003;63:3799–804. sequence is involved in the oxidative stress-induced expression of the

342 | CANCER DISCOVERY APRIL 2012 www.aacrjournals.org

Functional Metabolic Screen in Prostate Cancer RESEARCH ARTICLE

adult T-cell leukemia-derived factor (ADF)/human thioredoxin (Trx) 42. Anastasiou D, Poulogiannis G, Asara JM, Boxer MB, Jiang JK, Shen M, gene. Nucleic Acids Res 1996;24:2746–52. et al. Inhibition of pyruvate kinase M2 by reactive oxygen species con- 34. Kim YC, Masutani H, Yamaguchi Y, Itoh K, Yamamoto M, Yodoi J. tributes to cellular antioxidant responses. Science 2011;334:1278–83. Hemin-induced activation of the thioredoxin gene by Nrf2. A differ- 43. Buescher JM, Moco S, Sauer U, Zamboni N. Ultrahigh performance ential regulation of the antioxidant responsive element by a switch of liquid chromatography-tandem mass spectrometry method for fast its binding factors. J Biol Chem 2001;276:18399–406. and robust quantification of anionic and aromatic metabolites. Anal 35. Huang X, Wullschleger S, Shpiro N, McGuire VA, Sakamoto K, Woods Chem 2010;82:4403–12. YL, et al. Important role of the LKB1-AMPK pathway in suppressing 44. Buscher JM, Czernik D, Ewald JC, Sauer U, Zamboni N. Cross- tumorigenesis in PTEN-deficient mice. Biochem J 2008;412:211–21. platform comparison of methods for quantitative metabolomics of 36. Rider MH, Bertrand L, Vertommen D, Michels PA, Rousseau GG, primary metabolism. Anal Chem 2009;81:2135–43. Hue L. 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase: 45. Chung N, Zhang XD, Kreamer A, Locco L, Kuan PF, Bartz S, et al. head-to-head with a bifunctional enzyme that controls glycolysis. Median absolute deviation to improve hit selection for genome-scale Biochem J 2004;381:561–79. RNAi screens. J Biomol Screen 2008;13:149–58. 37. Bensaad K, Tsuruta A, Selak MA, Vidal MN, Nakano K, Bartrons R, 46. Birmingham A, Selfors LM, Forster T, Wrobel D, Kennedy CJ, Shanks et al. TIGAR, a p53-inducible regulator of glycolysis and apoptosis. E, et al. Statistical methods for analysis of high-throughput RNA Cell 2006;126:107–20. interference screens. Nat Methods 2009;6:569–75. 38. Herrero-Mendez A, Almeida A, Fernandez E, Maestre C, Moncada 47. Wiederschain D, Wee S, Chen L, Loo A, Yang G, Huang A, et al. S, Bolanos JP. The bioenergetic and antioxidant status of neurons is Single-vector inducible lentiviral RNAi system for oncology target controlled by continuous degradation of a key glycolytic enzyme by validation. Cell Cycle 2009;8:498–504. APC/C-Cdh1. Nat Cell Biol 2009;11:747–52. 48. Wee S, Wiederschain D, Maira SM, Loo A, Miller C, deBeaumont R, 39. Yun H, Lee M, Kim SS, Ha J. Glucose deprivation increases mRNA et al. PTEN-deficient cancers depend on PIK3CB. Proc Natl Acad Sci stability of vascular endothelial growth factor through activation of U S A 2008;105:13057–62. AMP-activated protein kinase in DU145 prostate carcinoma. J Biol 49. Zufferey R, Nagy D, Mandel RJ, Naldini L, Trono D. Multiply Chem 2005;280:9963–72. attenuated lentiviral vector achieves efficient gene delivery in vivo. 40. DeNicola GM, Karreth FA, Humpton TJ, Gopinathan A, Wei C, Nat Biotechnol 1997;15:871–5. Frese K, et al. Oncogene-induced Nrf2 transcription promotes ROS 50. Van Schaftingen E, Lederer B, Bartrons R, Hers HG. A kinetic study detoxification and tumorigenesis. Nature 2011;475:106–9. of pyrophosphate: fructose-6-phosphate phosphotransferase from 41. Bensaad K, Cheung EC, Vousden KH. Modulation of intracellular potato tubers. Application to a microassay of fructose 2,6-bisphos- ROS levels by TIGAR controls autophagy. EMBO J 2009;28:3015–26. phate. Eur J Biochem 1982;129:191–5.

APRIL 2012 CANCER DISCOVERY | 343

Functional Metabolic Screen Identifies 6-Phosphofructo-2-Kinase/Fructose-2,6-Biphosphatase 4 as an Important Regulator of Prostate Cancer Cell Survival

Susana Ros, Claudio R. Santos, Sofia Moco, et al.

Cancer Discovery 2012;2:328-343. Published OnlineFirst March 22, 2012.

Updated version Access the most recent version of this article at: doi:10.1158/2159-8290.CD-11-0234

Supplementary Access the most recent supplemental material at: Material http://cancerdiscovery.aacrjournals.org/content/suppl/2012/02/16/2159-8290.CD-11-0234.DC 1

Cited articles This article cites 49 articles, 14 of which you can access for free at: http://cancerdiscovery.aacrjournals.org/content/2/4/328.full#ref-list-1

Citing articles This article has been cited by 19 HighWire-hosted articles. Access the articles at: http://cancerdiscovery.aacrjournals.org/content/2/4/328.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerdiscovery.aacrjournals.org/content/2/4/328. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.