Introduction

One of the metabolic characteristics of cancer cells is Warburg effect, that is, even in the presence of ample oxygen, cancer cells use glucose through glycolysis and ferment the end product pyruvate to lactic acid rather than completely oxidizing it through Krebs cycle (Warburg, 1956; Gatenby and Gillies, 2004; DeBerardinis et al., 2008; Hsu and Sabatini, 2008; Vander Heiden et al., 2009). Thus, glycolysis, the major pathway for energy generation, is vital for the proliferation and survival of cancer cells (DeBerardinis et al., 2008; Vander Heiden et al., 2009). Why then cancer cells shift energy production from Krebs cycle to glycolysis? Although it is not completely understood, it is believed that pyruvate kinase (PK) has important roles in this shift (Mazurek et al., 2002, 2005; Mazurek and Eigenbrodt, 2003; Mazurek, 2007; Christofk et al., 2008).

In mammalian cells, there are two PK genes: one encodes pyruvate kinase-M1 (PKM1) and PKM2, and the other encodes PKL and PKR. Thus, PKM1 and PKM2, or PKL and PKR, are the isoforms resulting from alternative splicing (Noguchi et al., 1986, 1987). PKM1 and PKM2 are the primary isoforms in adult tissue and embryonic tissues, respectively, and PKL and PKR are exclusively expressed in liver and red blood cells, respectively (Imamura et al., 1973; Ibsen, 1977). Tumor cells almost ubiquitously express PKM2, which is also called tumor M2-PK (Altenberg and Greulich, 2004). PKM2 has been shown to have an important role in cancer cell metabolism and growth, because inhibition of PKM2 by peptide aptamer inhibited cell growth (Spoden et al., 2008, 2009) and PKM2 knockdown by RNA interference or displacement of PKM2 with PKM1 (Christofk et al., 2008) significantly reduced the ability of human cancer cell lines to form tumor in nude mice. 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 and Eigenbrodt, 2003; Gatenby and Gillies, 2004; Mazurek, 2007; Christofk et al., 2008; DeBerardinis et al., 2008; Hsu and Sabatini, 2008; Vander Heiden et al., 2009), PKM2 is a potential molecular target for disrupting glucose metabolism in cancer cells (Christofk et al., 2008; Spoden et al., 2008, 2009).

In the last four decades, shikonin, alkannin and their 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 et al., 2001; Masuda et al., 2003, 2004; Nakaya and Miyasaka, 2003; Han et al., 2007; Hu and Xuan, 2008; Xuan and Hu, 2009; Yang et al., 2009), including a clinical study (Guo et al., 1991), which indicated that a shikonin mixture was effective in treating 19 patients of later-stage lung cancer who were not suitable for operation, radiotherapy and chemotherapy. We previously reported that shikonin and its analogs induced cell death with characteristics of necroptosis, a basic cell death defined by Degterev et al. and further studied in depth by the same group of scientists (Degterev et al., 2005, 2008; Hitomi et al., 2008). We found that shikonin and its analogs showed similar cytotoxicity toward cancer cells regardless of P-gp, MRP1, BCRP1, Bcl-1 or Bcl-xL expression levels, indicating that these compounds were capable to bypass multiple defense lines of cancer cells against anticancer drugs (Han et al., 2007; Xuan and Hu, 2009). In this study, we tried to investigate the molecular targets of shikonin. Using solid-phase shikonin to extract cellular proteins that were subsequently analyzed by mass spectroscopy, we found that PK isozyme-M2 (PKM2) was one of the proteins potentially bound by shikonin. We then determined the inhibitory activity and specificity of shikonin and its analogs (Figure 1) toward PKM2, PKM1 and PKL. Our study shows that shikonin and alkannin are potent and specific inhibitors to PKM2, an enzyme that dictates the last rate-limiting step of glycolysis, which is essential for cancer cells’ proliferation and survival.

Figure 1
figure 1

(a) The chemical structures of shikonin and its analogs. (b) The enantiomeric shikonin and alkannin.

Full size image

Results

PKM2 is a potential target of shikonin

In order to find out the molecular targets of shikonin, we used solid-phase shikonin (see Materials and methods) to extract cellular proteins (MCF-7) of interest. The proteins bound to crystal shikonin were subjected to sodium dodecyl sulfate (SDS)–PAGE, and the protein bands were sliced for subsequent mass spectroscopy. One of the bands was identified to be PKM2 (Figure 2); the other proteins potentially bound to crystal shikonin have been shown in Supplementary Figures 1–13. Alternatively, we used shikonin-conjugated Sepharose-6B to extract cellular proteins from MCF-7 cell lysate, among which PKM2 was detected by mass spectroscopy. Purified recombinant PKM2, PKM1 and PKL could also bind to shikonin-conjugated Sepharose-6B (Supplementary Figure 14). These results showed that PKM2 was a potential target of shikonin.

Figure 2
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.

Full size image

Shikonin and its enantiomeric alkannin are potent inhibitors of PKM2 but not PKM1 and PKL

An enzyme assay was performed according to the reported method for identification of compound-3 (N-(3-carboxy-4-hydroxy)phenyl-2,5,-dimethylpyrrole) as PKM2 inhibitor, the most potent inhibitor discovered from a screening of over 100 000 compounds (Vander Heiden et al., 2010). Shikonin and alkannin (the optical isomer of shikonin) showed comparable inhibitory activities, more potent than other shikonin analogs (Figures 3b and c). In the absence of D-fructose-1,6-bisphosphate (FBP), the IC50 of both shikonin and alkannin was 0.3 μM, about 100-fold lower than that of compound-3 (37.4 μM) (Figure 3c). In the presence of FBP (125 μM), the IC50s of shikonin and alkannin were 0.8 and 0.9 μM, respectively, about 50-fold lower than that of compound-3 (50 μM) (Figure 3d). Although allosteric activation of enzyme activities by FBP is concentration dependent (Figure 3a), we used FBP (125 μM) for the enzyme assay. First, FBP (125 μM) 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 μM.

Figure 3
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 μM 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.

Full size image

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.

Figure 4
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.

Full size image

Shikonin and alkannin inhibit the rates of cellular lactate production and glucose consumption

The direct consequence of cellular PKM2 inhibition is slowdown of the rate from phosphoenolpyruvate (PEP) to pyruvate catalyzed by PK, the enzyme dictating the last rate-limiting step in glycolysis (Warburg, 1956; Gatenby and Gillies, 2004; DeBerardinis et al., 2008; Hsu and Sabatini, 2008; Vander Heiden et al., 2009), ultimately leading to decrease of glycolytic flux, which can be reflected by the rates of cellular glucose consumption and lactate production. Cells were incubated with shikonin or alkannin (0–20 μM) for 1 h and cellular glucose consumption, lactate production and cell death were determined. A549 and MCF-7 expressed PKM2 but not PKM1 and PKL (Figure 5a). The results showed that shikonin and alkannin inhibited the cellular glycolytic rate in a concentration-dependent manner (Figure 5b). As shikonin and alkannin are cytotoxic, the reduced glycolysis rate can be due to the decrease of total viable cell number. We thus checked the death rates of cells, which were negligible within 1-h incubation (Supplementary Figure 15). In addition, consistent with Vander Heiden et al. (2010) who reported that cell lysate prepared from PKM2 inhibitor compound-3-treated cells showed significantly less PKM2 activity, cell lysate from shikonin-treated cells also showed significantly lower PKM2 activity (Supplementary Figure 16).

Figure 5
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.

Full size image

As shikonin showed similar toxicity toward drug-sensitive and drug-resistant cancer cells (Han et al., 2007; Xuan and Hu 2009), we determined whether or not shikonin also showed a similar potency inhibiting glycolysis in MCF-7, MCF-7/Adr (an MDR sub-line overexpressing P-gp), MCF-7/Bcl-xL (a sub-line transfected with Bcl-xL) and MCF-7/Bcl-2 (a sub-line transfected with Bcl-2) (Li et al., 2007; Xuan and Hu, 2009). As shown in Figure 5c, glycolysis in these cell lines was inhibited to a comparable extent by shikonin. Likewise, alkannin was comparable to shikonin in inhibiting glycolysis in these cell lines (Figure 5c).

PKM1 attenuates shikonin-induced cell death

To find out whether PKM2 inhibition is relevant to shikonin-induced death, we tested the sensitivity of PKM1-transfected HeLa cells toward shikonin. Western blotting was performed to confirm that PKM2 was expressed primarily in HeLa cells and that PKM2 and PKM1 were both expressed in transfected the HeLa cells (Figures 6a and b). Accordingly, cells transfected with PKM1 also showed higher PK activities (Figure 6e). Both time- and concentration-dependent assays showed that HeLa cells transfected with PKM1 were significantly more resistant to shikonin than the vector-only control (Figures 6c and d). Although the PK activities remaining in the cells after exposure to shikonin or alkannin decreased with increased concentration of the chemicals, cells transfected with PKM1 showed higher PK activities than the vector-only control. These results showed that PKM1 could attenuate shikonin- and alkannin-induced cell death.

Figure 6
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 μM 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.

Full size image

Discussion

Shikonin and its naturally occurring analogs represent a class of novel PKM2 inhibitors. First, shikonin and its analogs are naphthoquinones (Figure 1), structurally distinct from and functionally much more potent than the three classes of PKM2 inhibitors screened by Vander Heiden et al. (2010) from over 100 000 compounds. Second, shikonin and alkannin are the most potent PKM2 inhibitors among the tested shikonin analogs. Matching its nature as PKM2 inhibitor, shikonin and alkannin effectively inhibited the cellular glycolytic flux in cancer cells dominantly expressing PKM2. Finally, shikonin and alkannin showed a promising selectivity toward PKM2, that is, these compounds did not inhibit the activity of PKM1 and PKL at concentrations that resulted in over 50% inhibition of PKM2. How shikonin discriminates PKM2 from PKM1 is not known. One explanation is that shikonin may interact with the allosteric region of PKM2. PKM2 differed from PKM1 only in a region composed of 56 amino acids encoded by an alternatively spliced exon (Noguchi et al., 1986). This region does not contribute to catalytic activity, but rather to the site responsible for allosteric regulation by FBP (Jurica et al., 1998; Dombrauckas et al., 2005). PKM2, but not PKM1, showed a catalytic nature of allosteric regulation.

Intriguingly, PKL, PKM1 and PKM2 can all bind to solid-phase shikonin, suggesting there is a common region in these proteins to interact with shikonin. The identity of the region and how shikonin interacts with this region, however, await further studies.

Previously, we found that the cell death induced by shikonin-class compounds was accompanied with a sharp reduction of cellular ATP (Han et al., 2009). This can be explained at least partially by the results from this study–shikonin and its analogs are inhibitors to PKM2, the last rate-limiting enzyme of glycolysis, which is the major energy source for cancer cells (Warburg, 1956; Gatenby and Gillies, 2004; DeBerardinis et al., 2008; Hsu and Sabatini, 2008; Vander Heiden et al., 2009). In previous reports, we showed that shikonin could circumvent drug-transporter and apoptotic defect-mediated drug resistance (Han et al., 2007; Hu and Xuan, 2008; Xuan and Hu, 2009). The results from this study can also partially explain why. As PKM2 is essential for maintaining the homeostasis of cellular energy for cancer cells (Mazurek et al., 2002, 2005; Mazurek and Eigenbrodt, 2003; Mazurek, 2007; Christofk et al., 2008), regardless of drug resistance or not, the consequences of PKM2 inhibition are similar. Besides inhibiting PKM2, shikonin and its analogs may also interact with other important proteins such as microtubules (Supplementary Figures 1, 10–12), the targets of the well-known anticancer agent taxanes.

In summary, shikonin and its analogs are a novel class of inhibitors of tumor PKM2 and can effectively inhibit cancer cell glycolysis. As glycolysis is a core metabolism for cancer cells, shikonin and its analogs may have potential for cancer treatment.

Materials and methods

Reagents, enzymes and antibodies

Shikonin analogs (Figure 1) were purchased from the Tokyo Chemical Industry (Tokyo, Japan). Alkannin was purchased from Wako Pure Chemical Industries Ltd., Chuo-ku, Okasa, Japan. Trypan blue was from Amresco Inc., Solon, OH, USA; the mammalian protein extraction reagent, the protease inhibitor cocktail kit, the BCA protein assay kit and a non-reducing lane marker sample buffer were from Thermo Fisher Scientific Inc., Waltham, MA, USA; the EZ-ECL kit (Chemiluminescence Detection Kit for horseradish peroxidase (HRP)) was from Biological Industries, Kibbutz Beit-Haemek, Israel; β-NADH, ADP and PEP were from Roche, Basel, Switzerland; dimethyl-sulfoxide (DMSO), FBP, LDH and pyruvate were from Sigma, St Louis, MO, USA; human recombinant PKM2 was from BPS Bioscience Inc (San Diego, CA, USA); and the Molecular Imager FX was from Bio-Rad Laboratories (Hercules, CA, USA). The primary antibodies used were as follows: rabbit monoclonal anti-human PKM2 antibody (Cell Signaling, Boston, MA, USA), rabbit monoclonal anti-human PKM1 antibody (Sigma), goat anti-human PKL/R antibody (Santa Cruz Biotechnology, Santa Cruz, CA, USA), His-tagged mouse monoclonal antibody (Abmart, Arlington, MA, USA) and mouse monoclonal anti-human β-actin IgG (Biomedical Technologies Inc, Stoughton, MA, USA). The secondary antibodies used were HRP-conjugated anti-mouse IgG, HRP-conjugated anti-rabbit IgG and HRP-conjugated anti-goat IgG (Santa Cruz Biotechnology).

Cell cultures

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

Preparation of cell lysate

Cells were trypsinized and collected. The collected cells were lysed using a mammalian protein extraction reagent containing 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.

Solid-phase shikonin extraction of cellular proteins and mass spectroscopy

Approach-1

Crystal shikonin (8 mg) was mixed with the cell lysate of MCF-7 (2 mg protein), rotated at 4 °C for 2 h and centrifuged to remove the supernatant. The precipitate (crystal shikonin with bound cellular protein) was washed three times with chilled washing buffer (50 mM Tris–HCl (pH 7.5), 150 mM NaCl, 2 mM EDTA, 10% glycerol, 1 mM NaF) by centrifugation (5000 r.p.m. for 1 min at 4 °C). The crystal shikonin with bound cellular protein was dissolved and boiled in SDS–PAGE sample buffer for 10 min. The samples were loaded and separated on 10% SDS–polyacrylamide gel and stained by silver staining. The bands were cut, digested in gel and loaded for LC-MS-MS spectrometer analysis by Shanghai Applied Protein Technology Co Ltd. Proteins were identified by searching the databases of Swiss-Prot using SEQUEST.

Approach-2

Shikonin-conjugated Sepharose-6B was produced according to the instructions of epoxy-activated Sepharose-6B (Amersham Biosciences, Little Chalfont, Bucks, UK) (Supplementary Figure 17). Sepharose-6B incubated with the vehicle only was used as negative control. Shikonin-conjugated Sepharose-6B was suspended in 0.1 M Tris–HCl buffer (pH 8.0) containing 0.5 M NaCl at a volume ratio of 1:1. The cell lysate (4 mg) of MCF-7 was mixed with 100 μl of shikonin-conjugated Sepharose-6B to a total volume of 300 μl. The mixture was incubated for 4 h at 4 °C with constant and gentle stirring. The mixture was spun down at 5000 r.p.m. for 2 min at 4 °C. 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 phenylmethylsulfonyl fluoride) by centrifugation (5000 r.p.m. for 2 min at 4 °C) for three times. The precipitate was mixed with SDS–PAGE sample buffer and boiled for 10 min. The samples were loaded and separated on 10% SDS–polyacrylamide gel. The gel was dyed by silver staining. The whole lane was digested in gel and subjected for LC-MS-MS spectrometer analysis by Shanghai Applied Protein Technology Co Ltd. Proteins were identified by searching the databases of Swiss-Prot using SEQUEST.

Binding of recombinant PK isozymes with solid-phase shikonin

Purified recombinant His-tagged PKM1, PKM2 or PKL (0.2 μM) were mixed with 20 μl of a shikonin-conjugated Sepharose-6B suspension to a total volume of 300 μl. His-tagged yellow fluorescence protein was used as negative control to exclude nonspecific binding with the resin. The mixture was incubated for 4 h at 4 °C with constant and gentle stirring. The mixture was spun down at 5000 r.p.m. for 2 min at 4 °C. 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 phenylmethylsulfonyl fluoride) by centrifugation (5000 r.p.m. for 2 min at 4 °C) for three times. The precipitate was mixed with SDS–PAGE sample buffer and boiled for 10 min. The samples were loaded and separated on 10% SDS–polyacrylamide gel. The proteins in gel were transferred to a polyvinylidene fluoride membrane and then detected using a His-tagged mouse monoclonal antibody (Abmart) and HRP-conjugated anti-mouse IgG (Cell Signaling) before visualization by EZ-ECL. Visualization was performed with the Molecular Imager FX using the Kodak ID imaging densitometry analysis software on a personal computer.

Cloning and purification of recombinant PKM1, PKM2 and PKL

The cDNA of PKM1, PKM2 and PKL was amplified from the cDNAs of the human breast cancer cell line MCF-7, human muscle and human liver, respectively. Specimens of human muscle and human liver were obtained from the Tissue Bank of the Second Affiliated Hospital, Zhejiang University. The use of human tissue was approved by the hospital’s Institutional Review Board. The cDNA of PKM1, PKM2 and PKL was cloned into pQE-30 (Qiagen, Hamburg, Germany) with an N-terminal 6 × His tag and expressed in Escherichia coli–strain XL-1 blue (Qiagen). Each clone was confirmed by sequencing by Invitrogen Co., Carlsbad, CA, USA. When the E. coli culture attained an OD (600 nm) of 0.7, expression was induced by 1 mM isopropyl-β-D-thiogalactoside (Gibco, Grand Island, NY, USA) for 6 h at room temperature. The cells were collected and lysed by a freeze/thaw cycle and sonication. The lysate was passed through an Ni–Sepharose column (GE, Piscataway, NJ, USA); the protein not bound to Ni–Sepharose was washed with washing buffer (0.1 M Tris–HCl (pH 7.8), containing 0.5 M NaCl and 40 mM imidazole) and PK was eluted by using 250 mM imidazole. The purity of recombinant PKs was determined by SDS–PAGE and Coomassie blue staining (Supplementary Figures 18–20). The kinetic characteristics of the PKs were determined by an LDH-coupled activity assay. FBP (Sigma) could activate PKM2 and PKL but not PKM1. The enzyme parameters are consistent with the published data (Parkison et al., 1991; Jurica et al., 1998; Dombrauckas et al., 2005; Vander Heiden et al., 2010) (Supplementary Figures 18–20; Supplementary Table 1).

Preparation of compound-3

N-(3-carboxy-4-hydroxy)phenyl-2,5-dimethylpyrrole (compound-3, the most potent inhibitor discovered by Vander Heiden et al. (2010) from screening over 100 000 compounds) was prepared from 5-amino-2-hydroxybenzoic acid and hexane-2,5-dione (Supplementary Figure 21A) by a modified procedure of the published method (Liu et al., 2008). Thus, 5-aminosalicylic acid (5 mmol), hexane-2,5-dione (6 mmol) and 4-methylbenzenesulfonic acid (0.05 mmol) were added to a dried round-bottom flask. Toluene (10 ml) was added and the mixture was refluxed for 1 h. After cooling to room temperature, the solvent was removed in vacuum and the residue was subjected to flash chromatography on silica gel using petroleum ether/ethyl acetate (1:1) as eluent to yield the pure product, which was re-crystalized from petroleum ether/ethyl acetate. Orange solid (melting point 167–169 °C (ref. 169–171 oC) (Liu et al., 2008); 1H-NMR (CDCl3) δ 10.45 (s, 1H), 7.83 (d, J=2.4 Hz, 1H), 7.42–7.39 (m, 1H), 7.13 (d, J=8.8 Hz, 1H), 5.93 (s, 2H), 2.06 (s, 6H) p.p.m.; MS (ESI) m/z 232.0 ([M+H] +) (Supplementary Figure 21B).

Western blotting

The cell lysate was mixed with a non-reducing lane marker sample buffer followed by heat denaturation. The protein was applied to a 12% SDS–polyacrylamide gel, transferred to a polyvinylidene fluoride membrane and then detected by the proper primary and secondary antibodies before visualization by EZ-ECL. Visualization was performed with the Molecular Imager FX using the Kodak ID imaging densitometry analysis software on a personal computer.

Trypan blue exclusion assay

Loss of integrity of the plasma membrane was determined by trypan blue exclusion assay. After drug treatment, cells were collected and stained with 0.4% trypan blue for 5 min. Dead cells were permeable and viable cells were impermeable to trypan blue. The viable and dead cells were counted microscopically. The viable rate of each data set was calculated.

PK activity assay

The enzyme activity and inhibition assay was performed according to the method reported by Vander Heiden et al. (2010). PK activity was measured by an LDH coupled assay. The reaction mixture contained 50 mM Tris–HCl (pH 7.5), 100 mM KCl, 10 mM MgCl2, 0.6 mM PEP, 0.9 mM ADP, 0.12 mM β-NADH and 4.8 U/ml LDH. The chemicals were dissolved in DMSO and DMSO was diluted to 1/10 (v/v) by adding cell lysis buffer. FBP was dissolved in ddH2O to the indicated concentration. In the inhibition assay, 1 μl of FBP solution or ddH2O was added to 8 μl of a 20-ng/μl cell lysate or 50-pg/μl recombinant human PKM2, and the mixture was pre-incubated for 30 min. Then 1 μl of chemical solution or DMSO solution was added to the 9-μl pre-incubated mixture. The total 10-μl mixture was incubated for 25 min at 25 °C before 125 μl of reaction solution was added to the mixture. PK activity was calculated at 25 °C by monitoring change of absorbance at 340 nm from 0 to 20 min. The activity of PK here was defined as the quantity of NADH oxidized by 1 mg of lysate protein or recombinant human PKM2 per minute (Ikeda and Noguchi, 1998; Shimada et al., 2008; Vander Heiden et al., 2010).

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 described in the preceding sections. The enzyme (GenBank accession no. NM_002654), full length with an N-terminal His tag (molecular weight=58.9 kDa), was expressed in an E. coli 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 0.5 mM FBP, PKM2 activity increases 0.5-fold and the Km for PEP decreased to 0.25 mM. The parameters of recombinant PKM2 are thus consistent with the published data in general (Jurica et al., 1998; Dombrauckas et al., 2005).

LDH activity assay

Assay of LDH activity was performed at 25 °C in a mixture containing 100 mM Tris–HCl (pH 8.0), 100 mM KCl, 10 mM MgCl2, 0.5 mM EDTA, 0.2 mM β-NADH and 5 mM pyruvate (Anderson et al., 1964). One unit of LDH activity here was defined as the quantity of enzyme required to oxidize 1 μM of NADH per minute. In the LDH inhibition assay, the protein was pre-incubated with shikonin or its analogs for 30 min at 25 °C before the LDH activity assay.

Measurement of the metabolic rate of glucose and lactate

A total of 3 × 106 cells were plated in a 25-cm2 flask. After attachment, the cells were treated with shikonin/alkannin or vehicle (0.1% DMSO) for 1 h. The culture medium was collected. The concentration of glucose in the medium was measured by using the AU2700 chemistry-immuno analyzer from Olympus Company, Center Valley, PA, USA, and the concentration of lactate in the medium was determined by using the Vitros 5,1 FS chemistry system from Johnson & Johnson Company, New Brunswick, NJ, USA. Consumption of glucose and production of lactate were calculated from the difference between the concentrations in the medium at the beginning and at an appropriate time of culture.

Transfection of HeLa with PKM1 and sensitivity of transfectants toward shikonin and alkannin

The cDNA of PKM1 was cloned into pcDNA 3.1 (Invitrogen), which was transferred into E. coli strain XL-1 blue (Qiagen). The plasmids were harvested by using the plasmid miniprep system (Promega, Madison, WI, USA) according to the manufacturer's instructions. PKM1-pCDNA3.1 and pcDNA 3.1 (negative control) were introduced into HeLa cells by using Lipofectamine LTX (Invitrogen) according to the manufacturer's instructions. Cells were subjected to western blotting and shikonin/alkannin susceptibility assays 48 h after the transfection. The transfectant and control cells (2.5 × 105 per well in a 24-well plate) were treated with 10 μM shikonin (or alkannin) for a time course of 6 h, or with a serial concentrations of shikonin (or alkannin) for 4 h. The cell survival rate was determined by trypan blue exclusion assay as described in the preceding section.

Statistical analyses

Unless otherwise stated, the data were expressed as the mean±s.d. from at least three independent experiments, and analyzed by Student’s t-test. The IC50 was analyzed by using SigmaPlot 10.0.