Shikonin and Its Analogs Inhibit Cancer Cell Glycolysis by Targeting Tumor Pyruvate Kinase-M2

Oncogene (2011) 30, 4297–4306 & 2011 Macmillan Publishers Limited All rights reserved 0950-9232/11 www.nature.com/onc ORIGINAL ARTICLE Shikonin and its analogs inhibit cancer cell glycolysis by targeting tumor pyruvate kinase-M2

J Chen1, J Xie1, Z Jiang2, B Wang1, Y Wang2 and X Hu1

1Cancer Institute (a Key Laboratory For Cancer Prevention & Intervention, National Ministry of Education, China; a Key Laboratory of Molecular Biology in Medical Sciences, Zhejiang, China), The Second Afﬁliated Hospital, Zhejiang University School of Medicine, Zhejiang, China and 2Department of Chemistry, Zhejiang University, Hangzhou, China

We recently reported that shikonin and its analogs were et al., 2008; Hsu and Sabatini, 2008; Vander Heiden a class of necroptotic inducers that could bypass cancer et al., 2009). Thus, glycolysis, the major pathway for drug resistance. However, the molecular targets of energy generation, is vital for the proliferation and shikonin are not known. Here, we showed that shikonin survival of cancer cells (DeBerardinis et al.,2008;Vander and its analogs are inhibitors of tumor-specific pyruvate Heiden et al., 2009). Why then cancer cells shift energy kinase-M2 (PKM2), among which shikonin and its production from Krebs cycle to glycolysis? Although it is enantiomeric isomer alkannin were the most potent and not completely understood, it is believed that pyruvate showed promising selectivity, that is, shikonin and kinase (PK) has important roles in this shift (Mazurek alkannin at concentrations that resulted in over 50% et al., 2002, 2005; Mazurek and Eigenbrodt, 2003; inhibition of PKM2 activity did not inhibit PKM1 and Mazurek, 2007; Christofk et al.,2008). pyruvate kinase-L (PKL). Shikonin and alkannin signifi- In mammalian cells, there are two PK genes: one cantly inhibited the glycolytic rate, as manifested by encodes pyruvate kinase-M1 (PKM1) and PKM2, and cellular lactate production and glucose consumption in the other encodes PKL and PKR. Thus, PKM1 and drug-sensitive and resistant cancer cell lines (MCF-7, PKM2, or PKL and PKR, are the isoforms resulting MCF-7/Adr, MCF-7/Bcl-2, MCF-7/Bcl-xL and A549) from alternative splicing (Noguchi et al., 1986, 1987). that primarily express PKM2. HeLa cells transfected PKM1 and PKM2 are the primary isoforms in adult with PKM1 showed reduced sensitivity to shikonin- or tissue and embryonic tissues, respectively, and PKL and alkannin-induced cell death. To the best of our knowledge, PKR are exclusively expressed in liver and red blood shikonin and alkannin are the most potent and specific cells, respectively (Imamura et al., 1973; Ibsen, 1977). inhibitors to PKM2 reported so far. As PKM2 universally Tumor cells almost ubiquitously express PKM2, which is expresses in cancer cells and dictates the last rate-limiting also called tumor M2-PK (Altenberg and Greulich, step of glycolysis vital for cancer cell proliferation and 2004). PKM2 has been shown to have an important role survival, enantiomeric shikonin and alkannin may have in cancer cell metabolism and growth, because inhibition potential in future clinical application. of PKM2 by peptide aptamer inhibited cell growth Oncogene (2011) 30, 4297–4306; doi:10.1038/onc.2011.137; (Spoden et al., 2008, 2009) and PKM2 knockdown by published online 25 April 2011 RNA interference or displacement of PKM2 with PKM1 (Christofk et al., 2008) significantly reduced the ability Keywords: shikonin; alkannin; pyruvate kinase-M2; of human cancer cell lines to form tumor in nude mice. glycolysis; inhibitor As PKM2 is necessary for cancer cells’ aerobic glycolysis, which is a hallmark of cancer metabolism and the major energy source essential for cancer cell growth and survival (Warburg, 1956; Mazurek et al., 2002, 2005; Mazurek Introduction and Eigenbrodt, 2003; Gatenby and Gillies, 2004; Mazurek, 2007; Christofk et al., 2008; DeBerardinis One of the metabolic characteristics of cancer cells is et al., 2008; Hsu and Sabatini, 2008; Vander Heiden et al., Warburg effect, that is, even in the presence of ample 2009), PKM2 is a potential molecular target for oxygen, cancer cells use glucose through glycolysis and disrupting glucose metabolism in cancer cells (Christofk ferment the end product pyruvate to lactic acid rather et al., 2008; Spoden et al., 2008, 2009). than completely oxidizing it through Krebs cycle In the last four decades, shikonin, alkannin and their (Warburg, 1956; Gatenby and Gillies, 2004; DeBerardinis derivatives have been investigated as potential anticancer drugs for various aspects of cancer treatment (Ahn et al., 1995; Papageorgiou VP et al., 1999; Bailly, 2000; Kim Correspondence: Professor X Hu, Cancer Institute, The Second et al., 2001; Masuda et al., 2003, 2004; Nakaya and Affiliated Hospital, Zhejiang University School of Medicine, 88 Miyasaka, 2003; Han et al., 2007; Hu and Xuan, 2008; Jiefang Road, Hangzhou, Zhejiang 310009, China. E-mail: [email protected] Xuan and Hu, 2009; Yang et al.,2009),including Received 3 October 2010; revised 3 March 2011; accepted 23 March aclinicalstudy(Guoet al., 1991), which indicated 2011; published online 25 April 2011 that a shikonin mixture was effective in treating Shikonin–potent inhibitor to pyruvate kinase M2 J Chen et al 4298 19 patients of later-stage lung cancer who were not methods) to extract cellular proteins (MCF-7) of suitable for operation, radiotherapy and chemotherapy. interest. The proteins bound to crystal shikonin were We previously reported that shikonin and its analogs subjected to sodium dodecyl sulfate (SDS)–PAGE, and induced cell death with characteristics of necroptosis, a the protein bands were sliced for subsequent mass basic cell death defined by Degterev et al. and further spectroscopy. One of the bands was identified to be studied in depth by the same group of scientists (Degterev PKM2 (Figure 2); the other proteins potentially bound et al., 2005, 2008; Hitomi et al., 2008). We found that to crystal shikonin have been shown in Supplementary shikonin and its analogs showed similar cytotoxicity Figures 1–13. Alternatively, we used shikonin-conju- toward cancer cells regardless of P-gp, MRP1, BCRP1, gated Sepharose-6B to extract cellular proteins from Bcl-1 or Bcl-xL expression levels, indicating that these MCF-7 cell lysate, among which PKM2 was detected by compounds were capable to bypass multiple defense lines mass spectroscopy. Purified recombinant PKM2, PKM1 of cancer cells against anticancer drugs (Han et al.,2007; and PKL could also bind to shikonin-conjugated Xuan and Hu, 2009). In this study, we tried to investigate Sepharose-6B (Supplementary Figure 14). These results the molecular targets of shikonin. Using solid-phase showed that PKM2 was a potential target of shikonin. shikonin to extract cellular proteins that were subse- quently analyzed by mass spectroscopy, we found that Shikonin and its enantiomeric alkannin are potent PK isozyme-M2 (PKM2) was one of the proteins inhibitors of PKM2 but not PKM1 and PKL potentially bound by shikonin. We then determined the An enzyme assay was performed according to the reported inhibitory activity and specificity of shikonin and its method for identification of compound-3 (N-(3-carboxy-4- analogs (Figure 1) toward PKM2, PKM1 and PKL. Our hydroxy)phenyl-2,5,-dimethylpyrrole) as PKM2 inhibitor, study shows that shikonin and alkannin are potent and the most potent inhibitor discovered from a screening of specific inhibitors to PKM2, an enzyme that dictates the over 100 000 compounds (Vander Heiden et al., 2010). last rate-limiting step of glycolysis, which is essential for Shikonin and alkannin (the optical isomer of shikonin) cancer cells’ proliferation and survival. showed comparable inhibitory activities, more potent than other shikonin analogs (Figures3bandc).Intheabsence Results of D-fructose-1,6-bisphosphate (FBP), the IC50 of both shikonin and alkannin was 0.3 mM, about 100-fold lower PKM2 is a potential target of shikonin than that of compound-3 (37.4 mM) (Figure 3c). In the In order to find out the molecular targets of shikonin, presence of FBP (125 mM), the IC50s of shikonin and we used solid-phase shikonin (see Materials and alkannin were 0.8 and 0.9 mM, respectively, about 50-fold lower than that of compound-3 (50 mM) (Figure 3d). Although allosteric activation of enzyme activities by FBP is concentration dependent (Figure 3a), we used FBP (125 mM) for the enzyme assay. First, FBP (125 mM) significantly activated PKM2 (Figure 3a), so that the results of this study could be compared with the reported data (Vander Heiden et al., 2010). In fact, the IC50 of compound-3 determined in this study is similar to the reported data. Second, according to available data (Hirayama et al., 2009), intracellular FBP levels were lower than 125 mM. As we used a lactate dehydrogenase (LDH)-coupled enzyme assay for measurement of PKM2 activity, it is necessary to rule out that LDH activity is not affected by shikonin and alkannin. The results showed that LDH was not inhibited by shikonin, alkannin and the analogs (Figure 3e). Despite that the three isoenzymes could all bind to solid-phase shikonin, shikonin and alkannin showed promising selectivity toward PKM2, PKM1 and PKL (Figure 4). The IC50s of shikonin toward PKM1 and PKL were about 20- and 10-fold higher than that toward PKM2. The IC50s of alkannin toward PKM1 and PKL were also about 20- and 10-fold higher than that toward PKM2.

Shikonin and alkannin inhibit the rates of cellular lactate production and glucose consumption Figure 1 (a) The chemical structures of shikonin and its analogs. The direct consequence of cellular PKM2 inhibition is (b) The enantiomeric shikonin and alkannin. slowdown of the rate from phosphoenolpyruvate (PEP)

Oncogene Shikonin–potent inhibitor to pyruvate kinase M2 J Chen et al 4299

Figure 2 Identification of PKM2 as a potential target of shikonin by solid-phase shikonin extraction and mass spectroscopy. (a) SDS–PAGE of cellular proteins extracted by solid-phase shikonin. (b) Mass spectroscopic analysis of a protein with a molecular weight of about 55 kDa (indicated by an arrow), which was identified as PKM2. For detailed procedures, refer to section Materials and methods. PKM2, pyruvate kinase-2; SDS–PAGE, sodium dodecyl sulfate–PAGE. to pyruvate catalyzed by PK, the enzyme dictating the total viable cell number. We thus checked the death last rate-limiting step in glycolysis (Warburg, 1956; rates of cells, which were negligible within 1-h incuba- Gatenby and Gillies, 2004; DeBerardinis et al., 2008; tion (Supplementary Figure 15). In addition, consistent Hsu and Sabatini, 2008; Vander Heiden et al., 2009), with Vander Heiden et al. (2010) who reported that cell ultimately leading to decrease of glycolytic flux, which lysate prepared from PKM2 inhibitor compound-3- can be reflected by the rates of cellular glucose treated cells showed significantly less PKM2 activity, consumption and lactate production. Cells were incu- cell lysate from shikonin-treated cells also showed bated with shikonin or alkannin (0–20 mM) for 1 h and significantly lower PKM2 activity (Supplementary cellular glucose consumption, lactate production and Figure 16). cell death were determined. A549 and MCF-7 expressed As shikonin showed similar toxicity toward drug- PKM2 but not PKM1 and PKL (Figure 5a). The results sensitive and drug-resistant cancer cells (Han et al., showed that shikonin and alkannin inhibited the cellular 2007; Xuan and Hu 2009), we determined whether or not glycolytic rate in a concentration-dependent manner shikonin also showed a similar potency inhibiting glyco- (Figure 5b). As shikonin and alkannin are cytotoxic, the lysis in MCF-7, MCF-7/Adr (an MDR sub-line over- reduced glycolysis rate can be due to the decrease of expressing P-gp), MCF-7/Bcl-xL (a sub-line transfected

Oncogene Shikonin–potent inhibitor to pyruvate kinase M2 J Chen et al 4300

Figure 3 Shikonin and alkannin are potent inhibitors of recombinant human PKM2. (a) FBP activates the activity of recombinant PKM2. (b) The potency of shikonin analogs on the inhibition of recombinant human PKM2 in the absence of FBP. (c) Comparison of the potency of compound-3, shikonin and alkannin on inhibiting the activity of recombinant PKM2 in the absence of FBP. (d) Comparison of the potency of compound-3, shikonin and alkannin on inhibiting the activity of recombinant PKM2 in the presence of 125 mM FBP. (e) Shikonin (SK), alkannin (AK) and analogs show no signiﬁcant inhibition of LDH. For detailed procedures and statistics, refer section Materials and methods. FBP, D-fructose-1,6-bisphosphate; LDH, lactate dehydrogenase; PKM2, pyruvate kinase-2.

Figure 4 Shikonin and alkannin show selective inhibition of PKM2. (a) Determination of the IC50s of shikonin toward recombinant PKM2, PKM1 and PKL. (b) Determination of the IC50s of alkannin toward recombinant PKM2, PKM1 and PKL. For detailed procedures and statistics, refer section Materials and methods. PKL, pyruvate kinase-L; PKM2, pyruvate kinase-2.

Oncogene Shikonin–potent inhibitor to pyruvate kinase M2 J Chen et al 4301

Figure 5 Shikonin and alkannin inhibit the metabolic rate of glucose and lactate. (a) MCF-7 and A549 express PKM2 but not PKM1 and PKL, as determined by western blotting. Recombinant PKM2, PKM1 and PKL were used as control. The greater molecular weight of recombinant PKM2 than its native one is due to the His tag (840 Da). (b) Concentration-dependent inhibition of glucose consumption and lactate production in MCF-7 by shikonin and alkannin. (c) Inhibition of glucose consumption and lactate production in MCF-7, MCF-7/Adr, MCF-7/Neo, MCF-7/Bcl-2, MCF-7/Bcl-xL and A549 by shikonin and alkannin. For detailed procedures and statistics, refer section Materials and methods. PKL, pyruvate kinase-L; PKM2, pyruvate kinase-M2.

Oncogene Shikonin–potent inhibitor to pyruvate kinase M2 J Chen et al 4302 with Bcl-xL) and MCF-7/Bcl-2 (a sub-line transfected Previously, we found that the cell death induced by with Bcl-2) (Li et al., 2007; Xuan and Hu, 2009). As shikonin-class compounds was accompanied with a shown in Figure 5c, glycolysis in these cell lines was sharp reduction of cellular ATP (Han et al., 2009). This inhibited to a comparable extent by shikonin. Likewise, can be explained at least partially by the results from alkannin was comparable to shikonin in inhibiting this study–shikonin and its analogs are inhibitors to glycolysis in these cell lines (Figure 5c). PKM2, the last rate-limiting enzyme of glycolysis, which is the major energy source for cancer cells (Warburg, PKM1 attenuates shikonin-induced cell death 1956; Gatenby and Gillies, 2004; DeBerardinis et al., et al. To find out whether PKM2 inhibition is relevant to 2008; Hsu and Sabatini, 2008; Vander Heiden , shikonin-induced death, we tested the sensitivity of 2009). In previous reports, we showed that shikonin PKM1-transfected HeLa cells toward shikonin. Western could circumvent drug-transporter and apoptotic defect- et al. blotting was performed to confirm that PKM2 was mediated drug resistance (Han , 2007; Hu and expressed primarily in HeLa cells and that PKM2 and Xuan, 2008; Xuan and Hu, 2009). The results from this PKM1 were both expressed in transfected the HeLa cells study can also partially explain why. As PKM2 is (Figures 6a and b). Accordingly, cells transfected with essential for maintaining the homeostasis of cellular et al. PKM1 also showed higher PK activities (Figure 6e). energy for cancer cells (Mazurek , 2002, 2005; Both time- and concentration-dependent assays showed Mazurek and Eigenbrodt, 2003; Mazurek, 2007; Chris- that HeLa cells transfected with PKM1 were signifi- tofk et al., 2008), regardless of drug resistance or not, cantly more resistant to shikonin than the vector-only the consequences of PKM2 inhibition are similar. control (Figures 6c and d). Although the PK activities Besides inhibiting PKM2, shikonin and its analogs remaining in the cells after exposure to shikonin or may also interact with other important proteins such as alkannin decreased with increased concentration of the microtubules (Supplementary Figures 1, 10–12), the chemicals, cells transfected with PKM1 showed higher targets of the well-known anticancer agent taxanes. PK activities than the vector-only control. These results In summary, shikonin and its analogs are a novel class showed that PKM1 could attenuate shikonin- and of inhibitors of tumor PKM2 and can effectively inhibit alkannin-induced cell death. cancer cell glycolysis. As glycolysis is a core metabolism for cancer cells, shikonin and its analogs may have potential for cancer treatment.

Discussion Materials and methods Shikonin and its naturally occurring analogs represent a class of novel PKM2 inhibitors. First, shikonin and its Reagents, enzymes and antibodies analogs are naphthoquinones (Figure 1), structurally Shikonin analogs (Figure 1) were purchased from the Tokyo distinct from and functionally much more potent than Chemical Industry (Tokyo, Japan). Alkannin was purchased the three classes of PKM2 inhibitors screened by Vander from Wako Pure Chemical Industries Ltd., Chuo-ku, Okasa, Heiden et al. (2010) from over 100 000 compounds. Japan. Trypan blue was from Amresco Inc., Solon, OH, USA; Second, shikonin and alkannin are the most potent the mammalian protein extraction reagent, the protease inhibitor PKM2 inhibitors among the tested shikonin analogs. cocktail kit, the BCA protein assay kit and a non-reducing lane Matching its nature as PKM2 inhibitor, shikonin and marker sample buffer were from Thermo Fisher Scientiﬁc Inc., alkannin effectively inhibited the cellular glycolytic ﬂux Waltham, MA, USA; the EZ-ECL kit (Chemiluminescence Detection Kit for horseradish peroxidase (HRP)) was from in cancer cells dominantly expressing PKM2. Finally, Biological Industries, Kibbutz Beit-Haemek, Israel; b-NADH, shikonin and alkannin showed a promising selectivity ADP and PEP were from Roche, Basel, Switzerland; dimethyl- toward PKM2, that is, these compounds did not inhibit sulfoxide (DMSO), FBP, LDH and pyruvate were from Sigma, the activity of PKM1 and PKL at concentrations that St Louis, MO, USA; human recombinant PKM2 was from BPS resulted in over 50% inhibition of PKM2. How shikonin Bioscience Inc (San Diego, CA, USA); and the Molecular Imager discriminates PKM2 from PKM1 is not known. One FX was from Bio-Rad Laboratories (Hercules, CA, USA). The explanation is that shikonin may interact with the primary antibodies used were as follows: rabbit monoclonal allosteric region of PKM2. PKM2 differed from PKM1 anti-human PKM2 antibody (Cell Signaling, Boston, MA, USA), only in a region composed of 56 amino acids encoded by rabbit monoclonal anti-human PKM1 antibody (Sigma), goat an alternatively spliced exon (Noguchi et al., 1986). This anti-human PKL/R antibody (Santa Cruz Biotechnology, Santa Cruz, CA, USA), His-tagged mouse monoclonal antibody (Abmart, region does not contribute to catalytic activity, but Arlington, MA, USA) and mouse monoclonal anti-human b-actin rather to the site responsible for allosteric regulation by IgG (Biomedical Technologies Inc, Stoughton, MA, USA). The FBP (Jurica et al., 1998; Dombrauckas et al., 2005). secondary antibodies used were HRP-conjugated anti-mouse PKM2, but not PKM1, showed a catalytic nature of IgG, HRP-conjugated anti-rabbit IgG and HRP-conjugated allosteric regulation. anti-goat IgG (Santa Cruz Biotechnology). Intriguingly, PKL, PKM1 and PKM2 can all bind to solid-phase shikonin, suggesting there is a common Cell cultures region in these proteins to interact with shikonin. The MCF-7, A549 and HeLa cells were purchased from the identity of the region and how shikonin interacts with Chinese Academy of Sciences Cell Bank of Type Culture this region, however, await further studies. Collection (Shanghai, China). MCF-7/Neo, MCF-7/Bcl-2

Oncogene Shikonin–potent inhibitor to pyruvate kinase M2 J Chen et al 4303

Figure 6 PKM1 attenuates shikonin- and alkannin-induced cell death. (a) Expression of PK isoenzymes in HeLa cells as conﬁrmed by western blotting. (b) Western blot of PKM1 expression in HeLa cells transfected with PKM1 and vector-only control (Ctr). (c) Sensitivity of transfectants and vector-only cells (Ctr) toward a serial concentrations of shikonin or alkannin. (d) Time-dependent survival of transfectants and vector-only cells (Ctr) exposed to 10 mM shikonin or alkannin. For detailed procedures refer section Materials and methods. (e) Activities of PK in cells transfected with PKM1 or vector-only control (Ctr). The cells were treated with shikonin or alkannin for 1 h, and the cells were lysed for measurement of PK activity. PK, pyruvate kinase; PKM2, pyruvate kinase-M2.

and MCF-7/Bcl-xL cells were obtained by transfection with MCF-7/Adr was grown in RPMI-1640 containing 10% fetal pSFFV-Neo, pSFFV-Bcl-2 and pSFFV-Bcl-xL, respectively, calf serum and 500 ng/ml doxorubicin. The MCF-7/Neo, using Lipofectamine 2000, according to the manufacturer’s MCF-7/Bcl-2 and MCF-7/Bcl-xL cells were maintained in instructions. The MCF-7, A549 and HeLa cells were main- RPMI-1640 containing 10% fetal calf serum and 500 mg/ml G- tained in RPMI-1640 containing 10% fetal calf serum and 418 as described previously (Han et al., 2007; Li et al., 2007; 100 U/ml penicillin–streptomycin, and HEK293 was main- Xuan and Hu, 2009). The cells were grown in a humidiﬁed tained in Dulbecco’s Modiﬁed Eagle’s Medium containing CO2 incubator at 37 1C, and sub-cultured with 0.25% trypsin 10% fetal calf serum and 100 U/ml penicillin–streptomycin. containing 0.02% EDTA.

Oncogene Shikonin–potent inhibitor to pyruvate kinase M2 J Chen et al 4304 Preparation of cell lysate Visualization was performed with the Molecular Imager FX Cells were trypsinized and collected. The collected cells were using the Kodak ID imaging densitometry analysis software lysed using a mammalian protein extraction reagent containing on a personal computer. a protease inhibitor cocktail. After centrifugation at 17 000 g for 15 min, the protein content in the supernatant was determined using the BCA protein assay kit. Cloning and purification of recombinant PKM1, PKM2 and PKL The cDNA of PKM1, PKM2 and PKL was amplified from the Solid-phase shikonin extraction of cellular proteins cDNAs of the human breast cancer cell line MCF-7, human and mass spectroscopy muscle and human liver, respectively. Specimens of human Approach-1. Crystal shikonin (8 mg) was mixed with the cell muscle and human liver were obtained from the Tissue Bank lysate of MCF-7 (2 mg protein), rotated at 4 1C for 2 h and of the Second Affiliated Hospital, Zhejiang University. The use centrifuged to remove the supernatant. The precipitate (crystal of human tissue was approved by the hospital’s Institutional shikonin with bound cellular protein) was washed three times Review Board. The cDNA of PKM1, PKM2 and PKL was with chilled washing buffer (50 mM Tris–HCl (pH 7.5), 150 mM cloned into pQE-30 (Qiagen, Hamburg, Germany) with an NaCl, 2 mM EDTA, 10% glycerol, 1 mM NaF) by centrifuga- N-terminal 6 Â His tag and expressed in Escherichia coli– tion (5000 r.p.m. for 1 min at 4 1C). The crystal shikonin with strain XL-1 blue (Qiagen). Each clone was confirmed by bound cellular protein was dissolved and boiled in SDS– sequencing by Invitrogen Co., Carlsbad, CA, USA. When the PAGE sample buffer for 10 min. The samples were loaded and E. coli culture attained an OD (600 nm) of 0.7, expression separated on 10% SDS–polyacrylamide gel and stained by was induced by 1 mM isopropyl-b-D-thiogalactoside (Gibco, silver staining. The bands were cut, digested in gel and loaded Grand Island, NY, USA) for 6 h at room temperature. The for LC-MS-MS spectrometer analysis by Shanghai Applied cells were collected and lysed by a freeze/thaw cycle and Protein Technology Co Ltd. Proteins were identified by sonication. The lysate was passed through an Ni–Sepharose searching the databases of Swiss-Prot using SEQUEST. column (GE, Piscataway, NJ, USA); the protein not bound to Ni–Sepharose was washed with washing buffer (0.1 M Tris– Approach-2. Shikonin-conjugated Sepharose-6B was produced HCl (pH 7.8), containing 0.5 M NaCl and 40 mM imidazole) according to the instructions of epoxy-activated Sepharose-6B and PK was eluted by using 250 mM imidazole. The purity (Amersham Biosciences, Little Chalfont, Bucks, UK) (Supple- of recombinant PKs was determined by SDS–PAGE and mentary Figure 17). Sepharose-6B incubated with the vehicle only Coomassie blue staining (Supplementary Figures 18–20). The was used as negative control. Shikonin-conjugated Sepharose-6B kinetic characteristics of the PKs were determined by an LDH- was suspended in 0.1 M Tris–HCl buffer (pH 8.0) containing coupled activity assay. FBP (Sigma) could activate PKM2 and 0.5 M NaCl at a volume ratio of 1:1. The cell lysate (4 mg) PKL but not PKM1. The enzyme parameters are consistent of MCF-7 was mixed with 100 ml of shikonin-conjugated with the published data (Parkison et al., 1991; Jurica et al., Sepharose-6B to a total volume of 300 ml. The mixture was 1998; Dombrauckas et al., 2005; Vander Heiden et al., 2010) incubated for 4 h at 4 1C with constant and gentle stirring. (Supplementary Figures 18–20; Supplementary Table 1). The mixture was spun down at 5000 r.p.m. for 2 min at 4 1C. The supernatant was discarded and the precipitate was washed Preparation of compound-3 with chilled washing buffer (50 mM Tris–HCl (pH 8.0), 500 mM N-(3-carboxy-4-hydroxy)phenyl-2,5-dimethylpyrrole (com- NaCl, 10% glycerol, 1 mM DTT, 1 mM phenylmethylsulfonyl pound-3, the most potent inhibitor discovered by Vander fluoride) by centrifugation (5000 r.p.m. for 2 min at 4 1C) for Heiden et al. (2010) from screening over 100 000 compounds) three times. The precipitate was mixed with SDS–PAGE sample was prepared from 5-amino-2-hydroxybenzoic acid and buffer and boiled for 10 min. The samples were loaded and hexane-2,5-dione (Supplementary Figure 21A) by a modified separated on 10% SDS–polyacrylamide gel. The gel was dyed procedure of the published method (Liu et al., 2008). Thus, by silver staining. The whole lane was digested in gel and 5-aminosalicylic acid (5 mmol), hexane-2,5-dione (6 mmol) subjected for LC-MS-MS spectrometer analysis by Shanghai and 4-methylbenzenesulfonic acid (0.05 mmol) were added to Applied Protein Technology Co Ltd. Proteins were identified by a dried round-bottom flask. Toluene (10 ml) was added and searching the databases of Swiss-Prot using SEQUEST. the mixture was refluxed for 1 h. After cooling to room temperature, the solvent was removed in vacuum and the Binding of recombinant PK isozymes with solid-phase shikonin residue was subjected to flash chromatography on silica gel Purified recombinant His-tagged PKM1, PKM2 or PKL using petroleum ether/ethyl acetate (1:1) as eluent to yield the (0.2 mM) were mixed with 20 ml of a shikonin-conjugated pure product, which was re-crystalized from petroleum ether/ Sepharose-6B suspension to a total volume of 300 ml. His- ethyl acetate. Orange solid (melting point 167–169 1C o 1 tagged yellow fluorescence protein was used as negative (ref. 169–171 C) (Liu et al., 2008); H-NMR (CDCl3) d control to exclude nonspecific binding with the resin. The 10.45 (s, 1H), 7.83 (d, J ¼ 2.4 Hz, 1H), 7.42–7.39 (m, 1H), 7.13 mixture was incubated for 4 h at 4 1C with constant and gentle (d, J ¼ 8.8 Hz, 1H), 5.93 (s, 2H), 2.06 (s, 6H) p.p.m.; MS (ESI) stirring. The mixture was spun down at 5000 r.p.m. for 2 min m/z 232.0 ([M þ H] þ ) (Supplementary Figure 21B). at 4 1C. The supernatant was discarded and the precipitate was washed with chilled washing buffer (50 mM Tris–HCl (pH 8.0), 500 mM NaCl, 10% glycerol, 1 mM DTT, 1 mM phenylmethyl- Western blotting sulfonyl fluoride) by centrifugation (5000 r.p.m. for 2 min The cell lysate was mixed with a non-reducing lane marker at 4 1C) for three times. The precipitate was mixed with sample buffer followed by heat denaturation. The protein was SDS–PAGE sample buffer and boiled for 10 min. The samples applied to a 12% SDS–polyacrylamide gel, transferred to a were loaded and separated on 10% SDS–polyacrylamide gel. polyvinylidene fluoride membrane and then detected by the The proteins in gel were transferred to a polyvinylidene proper primary and secondary antibodies before visualization fluoride membrane and then detected using a His-tagged by EZ-ECL. Visualization was performed with the Molecular mouse monoclonal antibody (Abmart) and HRP-conjugated Imager FX using the Kodak ID imaging densitometry analysis anti-mouse IgG (Cell Signaling) before visualization by EZ-ECL. software on a personal computer.

Oncogene Shikonin–potent inhibitor to pyruvate kinase M2 J Chen et al 4305 Trypan blue exclusion assay Measurement of the metabolic rate of glucose and lactate Loss of integrity of the plasma membrane was determined by A total of 3 Â 106 cells were plated in a 25-cm2 flask. After trypan blue exclusion assay. After drug treatment, cells were attachment, the cells were treated with shikonin/alkannin collected and stained with 0.4% trypan blue for 5 min. Dead or vehicle (0.1% DMSO) for 1 h. The culture medium was cells were permeable and viable cells were impermeable to collected. The concentration of glucose in the medium was trypan blue. The viable and dead cells were counted micro- measured by using the AU2700 chemistry-immuno analyzer scopically. The viable rate of each data set was calculated. from Olympus Company, Center Valley, PA, USA, and the concentration of lactate in the medium was determined by PK activity assay using the Vitros 5,1 FS chemistry system from Johnson & The enzyme activity and inhibition assay was performed Johnson Company, New Brunswick, NJ, USA. Consumption according to the method reported by Vander Heiden et al. of glucose and production of lactate were calculated from the (2010). PK activity was measured by an LDH coupled assay. difference between the concentrations in the medium at the The reaction mixture contained 50 mM Tris–HCl (pH 7.5), beginning and at an appropriate time of culture. 100 mM KCl, 10 mM MgCl2, 0.6 mM PEP, 0.9 mM ADP, 0.12 mM b-NADH and 4.8 U/ml LDH. The chemicals were Transfection of HeLa with PKM1 and sensitivity dissolved in DMSO and DMSO was diluted to 1/10 (v/v) by of transfectants toward shikonin and alkannin adding cell lysis buffer. FBP was dissolved in ddH2O to the The cDNA of PKM1 was cloned into pcDNA 3.1 (Invitrogen), indicated concentration. In the inhibition assay, 1 ml of FBP which was transferred into E. coli strain XL-1 blue (Qiagen). solution or ddH2O was added to 8 ml of a 20-ng/ml cell lysate or The plasmids were harvested by using the plasmid miniprep system 50-pg/ml recombinant human PKM2, and the mixture was pre- (Promega, Madison, WI, USA) according to the manufacturer’s incubated for 30 min. Then 1 ml of chemical solution or DMSO instructions. PKM1-pCDNA3.1 and pcDNA 3.1 (negative control) solution was added to the 9-ml pre-incubated mixture. The were introduced into HeLa cells by using Lipofectamine LTX total 10-ml mixture was incubated for 25 min at 25 1C before (Invitrogen) according to the manufacturer’s instructions. Cells 125 ml of reaction solution was added to the mixture. PK were subjected to western blotting and shikonin/alkannin suscep- activity was calculated at 25 1C by monitoring change of tibility assays 48 h after the transfection. The transfectant and absorbance at 340 nm from 0 to 20 min. The activity of PK control cells (2.5 Â 105 per well in a 24-well plate) were treated here was defined as the quantity of NADH oxidized by 1 mg of with 10 mM shikonin (or alkannin) for a time course of 6 h, or lysate protein or recombinant human PKM2 per minute with a serial concentrations of shikonin (or alkannin) for 4 h. (Ikeda and Noguchi, 1998; Shimada et al., 2008; Vander The cell survival rate was determined by trypan blue exclusion Heiden et al., 2010). assay as described in the preceding section. Human recombinant PKM2 was either purchased from BPS Bioscience Inc, with an attached datasheet of characterization (Supplementary Figure 22), or from an in-house source as Statistical analyses described in the preceding sections. The enzyme (GenBank Unless otherwise stated, the data were expressed as the mean±s.d. accession no. NM_002654), full length with an N-terminal His from at least three independent experiments, and analyzed by tag (molecular weight ¼ 58.9 kDa), was expressed in an E. coli Student’s t-test. The IC50 was analyzed by using SigmaPlot 10.0. expression system. When pH reached 7.4, the Km for PEP was 0.4 mM in the presence of 2 mM ADP, and the Km for ADP was 0.38 mM in the presence of 10 mM PEP. In the presence of Conflict of interest 0.5 mM FBP, PKM2 activity increases 0.5-fold and the Km for PEP decreased to 0.25 mM. The parameters of recombinant The authors declare no conflict of interest. PKM2 are thus consistent with the published data in general (Jurica et al., 1998; Dombrauckas et al., 2005).

LDH activity assay Acknowledgements Assay of LDH activity was performed at 25 1C in a mixture containing 100 mM Tris–HCl (pH 8.0), 100 mM KCl, 10 mM We thank BPS Bioscience Inc (San Diego, CA, USA) for MgCl2, 0.5 mM EDTA, 0.2 mM b-NADH and 5 mM pyruvate technical support of their product—recombinant human (Anderson et al., 1964). One unit of LDH activity here was PKM2. This work was supported in part by the China deﬁned as the quantity of enzyme required to oxidize 1 mM of National 863 project (2007AA02Z143) to XH; China Natural NADH per minute. In the LDH inhibition assay, the protein Sciences Foundation projects (30772544, 81071802) to XH and was pre-incubated with shikonin or its analogs for 30 min at the Fundamental Research Funds for the Central Universities, 25 1C before the LDH activity assay. National Ministry of Education, China, to XH.

References

Ahn BZ, Baik KU, Kweon GR, Lim K, Hwang BD. (1995). Christofk HR, Vander Heiden MG, Harris MH, Ramanathan A, Acylshikonin analogues: synthesis and inhibition of DNA topo- Gerszten RE, Wei R et al. (2008). The M2 splice isoform of isomerase-I. J Med Chem 38: 1044–1047. pyruvate kinase is important for cancer metabolism and tumour Altenberg B, Greulich KO. (2004). Genes of glycolysis are ubiquitously growth. Nature 452: 230–233. overexpressed in 24 cancer classes. Genomics 84: 1014–1020. DeBerardinis RJ, Lum JJ, Hatzivassiliou G, Thompson CB. (2008). Anderson SR, Florini JR, Vestling CS. (1964). Rat liver lactate The biology of cancer: metabolic reprogramming fuels cell growth dehydrogenase. 3. Kinetics and speciﬁcity. JBiolChem239: 2991–2997. and proliferation. Cell Metab 7: 11–20. Bailly C. (2000). Topoisomerase I poisons and suppressors as Degterev A, Huang Z, Boyce M, Li Y, Jagtap P, Mizushima N et al. anticancer drugs. Curr Med Chem 7: 20. (2005). Chemical inhibitor of nonapoptotic cell death with

Oncogene Shikonin–potent inhibitor to pyruvate kinase M2 J Chen et al 4306 therapeutic potential for ischemic brain injury. Nat Chem Biol 1: Masuda Y, Shima G, Aiuchi T, Horie M, Hori K, Nakajo S et al. 112–119. (2004). Involvement of tumor necrosis factor receptor-associated Degterev A, Hitomi J, Germscheid M, Ch0en IL, Korkina O, Teng X protein 1 (TRAP1) in apoptosis induced by beta-hydroxyisovaler- et al. (2008). Identification of RIP1 kinase as a specific cellular ylshikonin. J Biol Chem 279: 42503–42515. target of necrostatins. Nat Chem Biol 4: 313–321. Mazurek S. (2007). Pyruvate kinase type M2: a key regulator within Dombrauckas JD, Santarsiero BD, Mesecar AD. (2005). Structural the tumour metabolome and a tool for metabolic profiling of basis for tumor pyruvate kinase M2 allosteric regulation and tumours. Ernst Schering Found Symp Proc 4: 99–124. catalysis. Biochemistry 44: 9417–9429. Mazurek S, Boschek CB, Hugo F, Eigenbrodt E. (2005). Pyruvate Gatenby RA, Gillies RJ. (2004). Why do cancers have high aerobic kinase type M2 and its role in tumor growth and spreading. Semin glycolysis? Nat Rev Cancer 4: 891–899. Cancer Biol 15: 300–308. Guo XP, Zhang XY, Zhang SD. (1991). [Clinical trial on the effects of Mazurek S, Eigenbrodt E. (2003). The tumor metabolome. Anticancer shikonin mixture on later stage lung cancer]. Zhong Xi Yi Jie He Za Res 23: 1149–1154. Zhi 11: 598–599, 580. Mazurek S, Grimm H, Boschek CB, Vaupel P, Eigenbrodt E. (2002). Han W, Li L, Qiu S, Lu Q, Pan Q, Gu Y et al. (2007). Shikonin Pyruvate kinase type M2: a crossroad in the tumor metabolome. Br circumvents cancer drug resistance by induction of a necroptotic J Nutr 87(Suppl 1): S23–S29. death. Mol Cancer Ther 6: 1641–1649. Nakaya K, Miyasaka T. (2003). A shikonin derivative, beta-hydro- Han W, Xie J, Li L, Liu Z, Hu X. (2009). Necrostatin-1 reverts xyisovalerylshikonin, is an ATP-non-competitive inhibitor of shikonin-induced necroptosis to apoptosis. Apoptosis 14: protein tyrosine kinases. Anticancer Drugs 14: 683–693. 674–686. Noguchi T, Inoue H, Tanaka T. (1986). The M1- and M2-type Hirayama A, Kami K, Sugimoto M, Sugawara M, Toki N, Onozuka isozymes of rat pyruvate kinase are produced from the same gene by H et al. (2009). Quantitative metabolome profiling of colon and alternative RNA splicing. J Biol Chem 261: 13807–13812. stomach cancer microenvironment by capillary electrophoresis Noguchi T, Yamada K, Inoue H, Matsuda T, Tanaka T. (1987). The time-of-flight mass spectrometry. Cancer Res 69: 4918–4925. L- and R-type isozymes of rat pyruvate kinase are produced from a Hitomi J, Christofferson DE, Ng A, Yao J, Degterev A, Xavier RJ single gene by use of different promoters. J Biol Chem 262: 14366– et al. (2008). Identification of a molecular signaling network that 14371. regulates a cellular necrotic cell death pathway. Cell 135: 1311–1323. Papageorgiou VP AA, Couladouros EA, Hepworth D, Nicolaou KC.. Hsu PP, Sabatini DM. (2008). Cancer cell metabolism: Warburg and (1999). The chemistry and biology of alkannin, shikonin, and beyond. Cell 134: 703–707. related naphthazarin natural products. Angewandte Chemie Inter- Hu X, Xuan Y. (2008). Bypassing cancer drug resistance by activating national Edition 38: 32. multiple death pathways—a proposal from the study of circumvent- Parkison C, Ashizawa K, McPhie P, Lin K, Cheng S. (1991). The ing cancer drug resistance by induction of necroptosis. Cancer Lett monomer of pyruvate kinase, subtype M1, is both a kinase and a 259: 127–137. cytosolic thyroid hormone binding protein. Biochem Biophys Res Ibsen KH. (1977). Interrelationships and functions of the pyruvate Commun 179: 668–674. kinase isozymes and their variant forms: a review. Cancer Res 37: Shimada N, Shinagawa T, Ishii S. (2008). Modulation of M2-type 341–353. pyruvate kinase activity by the cytoplasmic PML tumor suppressor Ikeda Y, Noguchi T. (1998). Allosteric regulation of pyruvate kinase protein. Genes Cells 13: 245–254. M2 isozyme involves a cysteine residue in the intersubunit contact. Spoden GA, Mazurek S, Morandell D, Bacher N, Ausserlechner MJ, J Biol Chem 273: 12227–12233. Jansen-Durr P et al. (2008). Isotype-specific inhibitors of the Imamura K, Tanaka T, Nishina T, Nakashima K, Miwa S. (1973). glycolytic key regulator pyruvate kinase subtype M2 moderately Studies on pyruvate kinase (PK) deficiency. II. Electrophoretic, decelerate tumor cell proliferation. Int J Cancer 123: 312–321. kinetic and immunological studies on pyruvate kinase of erythro- Spoden GA, Rostek U, Lechner S, Mitterberger M, Mazurek S, cytes and other tissues. J Biochem 74: 1165–1175. Zwerschke W. (2009). Pyruvate kinase isoenzyme M2 is a glycolytic Jurica MS, Mesecar A, Heath PJ, Shi W, Nowak T, Stoddard BL. sensor differentially regulating cell proliferation, cell size and (1998). The allosteric regulation of pyruvate kinase by fructose-1, apoptotic cell death dependent on glucose supply. Exp Cell Res 6-bisphosphate. Structure 6: 195–210. 315: 2765–2774. Kim SH, Kang IC, Yoon TJ, Park YM, Kang KS, Song GY et al. Vander Heiden MG, Cantley LC, Thompson CB. (2009). Under- (2001). Antitumor activities of a newly synthesized shikonin standing the Warburg effect: the metabolic requirements of cell derivative, 2-hyim-DMNQ-S-33. Cancer Lett 172: 171–175. proliferation. Science 324: 1029–1033. Li L, Han W, Gu Y, Qiu S, Lu Q, Jin J et al. (2007). Honokiol induces Vander Heiden MG, Christofk HR, Schuman E, Subtelny AO, a necrotic cell death through the mitochondrial permeability Sharfi H, Harlow EE et al. (2010). Identification of small mole- transition pore. Cancer Res 67: 4894–4903. cule inhibitors of pyruvate kinase M2. Biochem Pharmacol 79: Liu K, Lu H, Hou L, Qi Z, Teixeira C, Barbault F et al. (2008). 1118–1124. Design, synthesis, and biological evaluation of N-carboxyphenyl- Warburg O. (1956). On the origin of cancer cells. Science 123: 309–314. pyrrole derivatives as potent HIV fusion inhibitors targeting gp41. Xuan Y, Hu X. (2009). Naturally-occurring shikonin analogues—a J Med Chem 51: 7843–7854. class of necroptotic inducers that circumvent cancer drug resistance. Masuda Y, Nishida A, Hori K, Hirabayashi T, Kajimoto S, Nakajo S Cancer Lett 274: 233–242. et al. (2003). Beta-hydroxyisovalerylshikonin induces apoptosis in Yang H, Zhou P, Huang H, Chen D, Ma N, Cui QC et al. (2009). human leukemia cells by inhibiting the activity of a polo-like kinase Shikonin exerts antitumor activity via proteasome inhibition and 1 (PLK1). Oncogene 22: 1012–1023. cell death induction in vitro and in vivo. Int J Cancer 124: 2450–2459.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)

Oncogene