Combined preparations of PKM2 modulators and HMGB1

Abstract

The present invention relates to a combined preparation comprising (i) a modulator of pyruvate kinase M2 (PKM2) activity, and (ii) an agent providing high mobility group box 1 (HMGB1) polypeptide or a derivative thereof. The present invention also relates to the aforesaid combined preparation for use as a medicament and for use in the treatment of inappropriate cellular proliferation, preferably in the treatment of cancer. Moreover, the present invention relates to a method for determining whether a subject suffering from inappropriate cellular proliferation is amenable to a treatment comprising administration of a modulator of PKM2 activity as the only PKM2 inhibitor and to treatment methods related thereto.

Claims

1. A method of treating inappropriate cellular proliferation in a subject suffering from inappropriate cellular proliferation comprising administering an agent providing HMGB1 or a derivative thereof and a modulator of PKM2 activity selected from the group consisting of DASA, ML265, and P-M2tide, thereby treating inappropriate cellular proliferation.

2. The method of claim 1, wherein treating inappropriate cellular proliferation is treating cancer.

3. The method of claim 2, wherein said cancer is colorectal carcinoma or chronic lymphocytic leukemia (CLL).

4. The method of claim 1, wherein said inappropriate cellular proliferation is resistant to treatment with a modulator of PKM2 activity.

Description

FIGURE LEGENDS

(1) FIG. 1: HMGB1 from human NK cells induces cell death in colorectal cancer. (A) HMGB1 was purified from NK-92 Cl cells by chromatography. Arrow=HMGB1 containing fraction (eluate #38). (B) Immunoblot showing the membrane containing eluates 37-42. A specific HMGB1 band at 30 kD was detected only in eluate #38. (C) Cytotoxicity assay after 72 h incubation with HMGB1 (n=3). A 1:40 dilution of the purified HMGB1 (#38) was used (corresponding to approximately 16 nM). Recombinant human HMGB1 was used at 80 nM. (D) Supernatants from activated human peripheral blood NK cells (cross-linked with anti-Nkp30 antibody) from two donors were tested for their cytotoxic capacity in a crystal violet assay (72h, n=3). Glycyrrhizin (200 μM) was used as an inhibitor of HMGB1. (E) Immunoblot of the supernatants used in (D). HMGB1 was specifically secreted upon Nkp30 crosslinking. *=non-specific band. **p<0.002. (F) to (M) NK cell derived HMGB1 induces cell death in colorectal cancer. (F, G) HMGB1 from cytotoxic granules from NK-92 CI cell line was purified by reversed phase chromatography on a Resource RPC column (F) and on a Source 15RPC ST 4.6/100 column (G) before the final purification step. Details are given in the Experimental Procedures section. The dashed line indicates the acetonitrile gradient. HMGB1 containing fractions are indicated by arrows. (H) The purification yield was approx. 90% as determined by Commassie Blue staining of the corresponding gel. (I, left) Survival of HT29 cancer cells as assessed by crystal violet viability assay. Cells were treated with recombinant human HMGB1 (160 nM) or NK cell derived HMGB1 (160 nM); where indicated 200 μM glycyrrhizin was used (72h, n=3). (I, right) Side-by-side comparison of HMGB1 cytotoxicity was performed using 80 nM HMGB1 concentrations (72 h, n=3). (K) Silver gel showing purity of HMGB1 in eluate #38 (0.5 μg protein loaded). (L) HPLC-purified HMGB1 (80 nM, 24 h, n=3) from the supernatant of Nkp30 stimulated blood donor NK cells was diluted in the IgG1 control supernatant and shows substantial cytotoxicity. (M) Interferon gamma concentration was determined by ELISA (see Experimental Procedures) and used as a positive control for the activation of the cultured and stimulated NK cells derived from blood donors. Error bars represent the SD. **p<0.002.

(2) FIG. 2: HMGB1 inhibits mitochondrial respiration. (A) Activities of respiratory chain complexes measured in mitochondrial fractions of colorectal cancer cell lines after treatment with HMGB1 (80 nM, 24 h, n=5). (B) Tissue slices were generated from a fresh surgical human colon carcinoma specimen and treated with HMGB1 (160 nM, 72 h). After homogenization of tissue slices, COX activity was measured in the mitochondrial fractions (n=8). (C-F) Colorectal cancer cells and tissue slices were treated with HMGB1 as described in (A) and (B), respectively. Then, cyanide sensitive respiration was measured. *p<0.05, **p<0.001.

(3) FIG. 3: HMGB1 blocks glycolysis by interfering with PK M2. (A) Activities of glycolytic enzymes measured in cytosolic fractions after treatment with HMGB1 (80 nM, 24 h, n=5). HK=Hexokinase, PFK=Phosphofructokinase, GAPDH=Glyceraldehyde 3-phosphate dehydrogenase, TPI=Triose-phosphate isomerase, PGM=Phosphoglycerate mutase, ENO=Enolase, LDH=Lactate dehydrogenase. (B) Colon cancer tissue slices from a fresh surgical specimen were treated with HMGB1 (160 nM, 72 h). Tetrameric PK M2 activity was measured in eight homogenates. (C) PK M2 activity in colorectal cancer cells was measured after 24 h incubation with the supernatant derived from stimulated NK cells from blood donor #2 (see FIG. 1D). Glycyrrhizin (200 μM) was used as a HMGB1 inhibitor (n=3). (D) 5-3H-glucose turn-over was assessed after treatment with HMGB1 (80 nM, 24 h, n=3). (E) The experiment using SW480 cells was performed as outlined in (D). Glycyrrhizin (200 μM) was used as an inhibitor of HMGB1 (n=3). (F) Enrichment of 14C in the mRNA of crude extracts after treatment with HMGB1 (80 nM, 24 h, n=3). (G) Isolation of the PK M2-HMGB1 complex: ultrafiltration of a solution containing 2 μM human PK M2 and 2 μM human HMGB1. The filtrated PK M2-HMGB1 complex was exposed to Western Blotting. *p<0.05, **p<0.002, ***p<0.00008.

(4) FIG. 4: HMGB1 is an allosteric inhibitor of tetrameric PK M2. (A) The calculations were performed for the following HMGB1 constructs: leftmost: A box (dark), tyrosines unphosphorylated; center left: A box, tyrosines phosphorylated; center right: B box, tyrosines unphosphorylated; rightmost: B box, tyrosines phosphorylated. (B) Dark clouds=electrostatic potentials of HMGB1 A and B box, with: leftmost) A box, unphosphorylated tyrosines; center left) A box, phosphorylated tyrosines; center right) B box, unphosphorylated tyrosines; rightmost) B box, phosphorylated tyrosines. (C) Left: Distances from HMGB1 Box B residues pTyr 116 and pTyr 162 (numbering according to PDB file: 2YRQ, corresponding to residues 109 and 155, respectively, in the human sequence) to PK M2 K433 for the best ranked docked pose and of PK M2 Y105 to the nearest charged residue from the HMGB1 Box B. Right: A rotated view with the electrostatic potential (dark) of PK M2. (D) The experiment was performed as outlined in (A), here in the presence of FBP (left) or with Tyr105 phosphorylated (center) or in the absence of FBP and with unphosphorylated Tyr 105 (right).

(5) FIG. 5: Glucose fermentation and glutaminolysis circumvent the HMGB1-triggered metabolic block in cancer cells. (A) Viability assay performed with respiratory chain deficient cells (ρ0) and control (wild type) cells treated with HMGB1 (160 nM, 72 h, n=3). **p<0.0001. (B) The amount of ATP production was calculated from 02 consumption and from 13C-lactate efflux derived from 13C labeled glucose. (C) The amount of ATP production was calculated from 13C-lactate efflux derived from 13C labeled glutamine. Cells were treated with HMGB1 (80 nM, 24 h, n=3, *p<0.02). (D) Survival of cells after treatment with oligomycin (10 ng/ml) and HMGB1 (80 nM) (both 72 h, n=3, **p<0.0001). (E) Crystal violet survival assay in glucose free medium after treatment with HMGB1 (80 nM, SW480 and HCT116; 160 nM, HT29; 24 h). L-DON (1 μM) was added as indicated (n=3, *p<0.02). (F) After siRNA-mediated knock-down of ME1 the HT29 cells were treated with HMGB1 (80 nM, 72 h, n=3, *p<0.0003, **p<0.000002). Right panel: immunoblot with anti-ME1 antibody to confirm the knock-down. (G) ME1 expression levels and local invasion depth of cancer tissue (pT stage) in patients with rectal carcinoma; +low, ++moderate, +++ strong expression of ME1. (H) Association of ME1 expression levels and lymph node metastasis (pN stage) in patients with rectal carcinoma. (I) to Inhibition of tetrameric PKM2 phenocopies the cytotoxicity of HMGB1. (I) Cells were treated with 100 μM P-M2tide (phosphotyrosine peptide) for 24 h (n=3). The PKM2 activating small molecule ML-265 that binds to the dimer-dimer interface distant from the P-M2tide binding site did not induce cytotoxicity. (K) Cytosolic fractions of the indicated cells treated with 125I-labelled HMGB1 (80 nM, 24 h, n=3). (L) PKM2 was down-regulated or overexpressed transiently and then treated with HMGB1 (80 nM, 72 h, n=3). Successful knock-down or overexpression of PKM2 was confirmed by western blotting (M, N). Error bars represent the SD. *p<0.05, **p<0.01.

(6) FIG. 6: Co-administration of HMGB1 and an activator (A left and right) or an inhibitor (B) of PMK2 has a synergistic cytotoxic effect, as determined in the crystal violet cytotoxicity assay (assays were performed in duplicate, * indicates p<0.05). (C) Immunoprecipitation with I125-labeled HMGB1 (PKM2 pull-down). HMGB1, like phosphotyrosine-peptides, binds close to the allosteric center of PKM2, leading to a competition in binding. In contrast, the activator ML265, which binds to the tetramer at the dimer-dimer interface, which is far off the allosteric center, does not compete with HMBG1.

(7) FIG. 7: Cell-death inducing effect of oligo-phosphorylated HMGB1: indicated cancer cell lines were incubated with two different concentrations of GInHMGB1 (SEQ ID NO:12) and cell viability was tested with the MTT assay; control: no addition of GInHMGB1; y-axis: optical density at 630 nm.

(8) FIG. 8: Cell-death inducing effect of oligo-phosphorylated HMGB1: indicated cancer cell lines were incubated with two different concentrations of GInHMGB1 (SEQ ID NO:12) and cell viability was tested with the LDH release assay; control: no addition of GInHMGB1; y-axis: fraction of LDH in the supernatant (%).

(9) The following Examples shall merely illustrate the invention. They shall not be construed, whatsoever, to limit the scope of the invention.

EXAMPLE 1

NK Cell Derived HMGB1 Protein Induces Cell Death in Colorectal Cancer

(10) Given the cytotoxic activity of recombinant human HMGB1 protein on cancer cells .sup.9 we sought to examine the cellular effects of immune cell derived endogenous HMGB1. To this end, we isolated HMGB1 from the cytosolic granules of the NK cell line NK-92 CI by HPLC (FIG. 1A). Elution of HMGB1 was confirmed by immunoblot analysis (FIG. 1B). Both NK cell derived HMGB1 and, as a comparison, recombinant human HMGB1 efficiently killed SW480 and HCT116 colorectal cancer cells (FIG. 1C). The observed cell death was specific for HMGB1 since glycyrrhizin, an inhibitor of HMGB1, significantly blocked its cytotoxic effects. In contrast, HT29 cells were resistant to low to intermediate HMGB1 concentrations (16 to 80 nM). Higher concentrations (80 nM or 160 nM) of NK cell derived HMGB1 exerted higher cytotoxicity than recombinant HMGB1 as assessed in side-by-side cytotoxicity experiments (FIG. 1I). A silver staining gel of eluate #38 confirmed that HMGB1 was isolated with high purity (single band at approximately 30 kD, FIG. 1K).

(11) Upon stimulation of activated human peripheral blood NK cells from healthy blood donors by agonistic anti-Nkp30 mAbs, the NK cell-dependent cytotoxic effect on HT29 and HCT116 colon cancer cells was diminished by the HMGB1-specific inhibitor glycyrrhizin, indicating that HMGB1 was partly mediating the NK cell-triggered tumor cell death (FIG. 1D). In contrast, cytotoxicity in SW480 cells was not changed by glycyrrhizin.

(12) Secretion of HMGB1 from NK cells was confirmed by immunoblot (FIG. 1E). Moreover, high levels of Interferon-γ were detected in the supernatant, indicating activation of NK cells by the agonistic anti-NKp30 mAb. Thus, NK cell derived HMGB1 protein induces cell death in colorectal cancer cells.

EXAMPLE 2

HMGB1 Inhibits Aerobic Respiration in Colorectal Carcinomas In Vitro and Ex Vivo

(13) The HMGB1 mediated cell death was characterized by formation of giant mitochondria and a substantial decrease of ATP in HMGB1-sensitive (SW480) and HMGB1 partly resistant (HCT116) cancer cells, but not in HMGB1 resistant HT29 cells. Due to the observed loss of energy equivalents and the altered mitochondrial morphology we examined whether HMGB1 affects the main ATP generating pathways, oxidative phosphorylation (OXPHOS) and glycolysis. HMGB1 treatment resulted in significantly lower activity levels of cytochrome c oxidase (COX) which is vital for oxygen derived ATP generation (FIG. 2A). Electron flow from complex I to III was unchanged, whereas coupled complex II and III activity was decreased in the HMGB1 sensitive cells (SW480) and maintained or even upregulated in the partly HMGB1-resistant cell line HCT116 and the HMGB1-resistant cell line HT29. ATP synthase activity was not diminished supporting the hypothesis that the decrease of intracellular ATP was caused by inhibition of energy metabolism up-stream of the respiratory chain. Next, we confirmed our in vitro monolayer cell culture based results in an alternative model accounting for the in vivo complexity of human colorectal cancer tissue using 300 μm thick slices from fresh tumor tissue of colorectal cancer patients. HMGB1 treatment decreased the turn-over of oxygen as demonstrated by a potent inhibition of COX activity in the primary tumor tissue (FIG. 2B). Consistently, HMGB1 strongly decreased mitochondrial oxygen consumption in colorectal cancer tissue (FIG. 2C). A similar effect was observed in cultured colon cancer cells, where the inhibition of mitochondrial oxygen consumption was pronounced in HMGB1-sensitive SW480 cells and in partly HMGB1-resistant HCT116 cells (FIGS. 2 D+E) whereas mitochondrial respiration of HMGB1-resistant HT29 cells was only slightly reduced by HMGB1 (FIG. 2F). These results indicate that HMGB1 inhibits aerobic respiration in colorectal carcinoma cells.

EXAMPLE 3

HMGB1 Controls Glycolysis in Colorectal Carcinoma Cells by Specific Inhibition of Tetrameric PK M2

(14) Since aerobic respiration can be glucose-driven we studied the effect of HMGB1 on the activity of the major glycolytic enzymes. We observed a reduced activity of an isoform (M2) of pyruvate kinase (PK) after HMGB1 treatment (FIG. 3A) that is known to drive glucose-mediated respiration. HMGB1 specifically inhibited the tetrameric pyruvate kinase isoform PK M2 in all colorectal cancer cell lines tested as well as in ex vivo tissue slice cultures (FIGS. 3A+B). Further experiments showed that HMGB1-containing supernatants from Nkp30 stimulated NK cells from a blood donor (donor #2, see FIG. 1) also significantly inhibited the tetrameric PK M2 (FIG. 3C). Importantly, this inhibition was caused by HMGB1 since glycyrrhizin completely restored tetrameric PK M2 activity. Dimeric PK M2 activity was unchanged. Glucose flux in HMGB1-treated cells was reduced at the enolase reaction step (FIG. 3D). The observed metabolic shift was partly reversed by co-treatment with the HMGB1 inhibitor glycyrrhizin (FIG. 3E). Moreover, HMGB1 treatment resulted in an increased flux of glycolytic intermediates into the pentose phosphate shunt (FIG. 3F). Consistent with the accumulation of glucose intermediates up-stream of pyruvate kinase there was a strong increase in the hexokinase (HK) product glucose-6-phosphate that could explain the observed decrease in HK activity by product inhibition. Supporting the results from the enzymatic tests, HMGB1 physically interacted with PK M2 (FIG. 3G) in vitro. Using .sup.125l-labeled HMGB1 we could show specific binding of HMGB1 to PKM2 in vivo by immunoprecipitating PKM2 (FIG. 6C). Non-cytotoxic P-M2tide concentrations substantially inhibited binding of HMGB1 to PKM2 supporting our in silico results (FIG. 6C). These data implicate that HMGB1 binding competes with the P-M2tide PKM2 binding site, involving the K433 near in the FBP binding pocket of PKM2. Importantly, the small molecule ML-265, an activator of PKM2, previously identified to bind to the dimer-dimer-interface of PKM2.sup.10 (far away from the FBP binding pocket) did not compete with HMGB1 binding to PKM2 (FIG. 6C).

EXAMPLE 4

Binding and Allosteric Inhibition of Tetrameric PK M2 by HMGB1

(15) To characterize the inhibition of tetrameric PK M2 by HMGB1 in more detail we performed in silico protein docking studies. The polyphosphorylated HMGB1 B Box produced a single cluster of poses indicating specific binding to PK M2 (FIG. 4A). Specific binding was not observed when the same calculation procedure was applied to the unphosphorylated HMGB1 B Box, or the polyphosphorylated or unphosphorylated HMGB1 A Box (FIG. 4A). This was further supported by energetic analysis of the bound clusters which showed a strongly electrostatically driven binding for the polyphosphorylated HMGB1 B Box, which is in contrast to a mainly hydrophobically driven binding typical of non-specific binding in the three other test cases (FIG. 4B). For the phosphorylated B Box, where specific binding is observed, a large region of negative electrostatic potential (red isopotential) was present in the vicinity of the binding interface, whereas the non-specific binding cases lacked such a region (FIG. 4B). Furthermore, the interaction involves K433 of PK M2 (FIG. 4C), previously shown to be involved in phosphotyrosine (pTyr) peptide binding near the FBP binding pocket.sup.11, and in the regulation of PK M2 activity through controlling tetramerization.sup.12. There is variation of the electrostatic potential in the region surrounding the proposed HMGB1 binding site on binding of FBP or phosphorylation of Y105 (FIG. 4D). The decrease in the size of the positive electrostatic potential on binding FBP, or the introduction of negative electrostatic potential on phosphorylation of Y105 is likely to hinder binding of the negatively charged phosphate groups from the phosphorylated HMGB1 Box B (FIG. 4D). These results support that HMGB1 is an allosteric inhibitor of the PK M2 tetramer.

(16) Moreover, we could phenocopy the observed cell death using a known inhibitor of the PKM2 tetramer, a phosphotyrosine peptide called P-M2tide (FIG. 5L). P-M2tide has previously been shown to bind near the FBP binding pocket involving the interaction of K433 of PKM2.sup.11. Whereas P-M2tide induced substantial cell death, an activator of PKM2, the small molecule ML-265 was not able to induce cell death (FIG. 5L). Penetration of the cell membrane was confirmed using .sup.125l-labelled HMGB1.sup.9, showing a rapid (24 h) increase of cytosolic radioactivity (FIG. 5K). Gain- and loss-of-function experiments for PKM2 using PKM2 siRNA/plasmid showed that down-regulation of PKM2 sensitized the cells to HMGB1 whereas overexpression of PKM2 rendered them more resistant to HMGB1 (FIG. 5L-N).

EXAMPLE 5

HMGB1 Resistant Cancer Cells are Characterized by Enhanced Glucose Fermentation and Increased Glutaminolysis

(17) The observed cell death induced by specific inhibition of the PK M2 tetramer and consequent inhibition of glucose driven respiration should favour the survival of cancer cells performing mainly (anaerobic) glycolysis. To test this hypothesis we generated colorectal cancer cells devoid of an intact respiratory chain (ρ.sup.0 cells) from one HMGB1-sensitive (SW480) and one partly HMGB1-sensitive (HCT116) cell line. These modified cell lines, performing solely glycolysis, became almost completely resistant to HMGB1 (FIG. 5A). In order to assess the relative contributions of glycolysis, glutaminolysis, and aerobic respiration to cellular survival in presence of HMGB1 we calculated total ATP generation (FIG. 5B+C) from the lactate production rates and from the oxygen consumption. Both HMGB1 partly (HCT116) and highly resistant (HT29) cancer cells compensated the HMGB1-caused decline of ATP production efficiently by glycolysis whereas SW480 cells showed a strong decline of ATP production of ˜50% (FIG. 5B). However, after HMGB1 treatment, only HT29 cells could sustain ATP production from aerobic respiration by employing glutaminolysis, as ATP yield from glutaminolysis (FIG. 5C) was in good agreement with ATP produced by 02 utilization (FIG. 5B). Consistently, after HMGB1 treatment of HT29 cells glucose oxidation was strongly decreased (˜50%) and glutamine oxidation increased (˜35%) as assessed by measuring production of labeled CO.sub.2. Importantly, energy from aerobic respiration was critical for survival of SW480 and HCT116 cells as shown by induction of rapid cell death by oligomycin (FIG. 5D). Inhibition of glutaminolysis by L-DON resulted in synergistic cytotoxicity in both glucose deprived (FIG. 5E) and glucose supplemented medium. After down-regulation of ME1 we observed sensitization of HT29 cells towards HMGB1 cytotoxicity (FIG. 5F). Further HMGB1 inhibited HMGB1-sensitive SW480 xenograft tumor growth nude mice whereas treatment of HMGB1-resistant HT29 xenograft tumors with a combination therapy of HMGB1 and L-DON substantially inhibited tumor growth. Taken together, both enhanced glucose fermentation and increased glutaminolysis might render cancer cells resistant to HMGB1 and animal experiments suggest that treatment with recombinant HMGB1 could constitute a therapeutic option.

EXAMPLE 6

Methods

(18) Cell Culture and Animal Studies

(19) Human colorectal carcinoma, human glioblastoma and human NK cell lines were purchased from ATCC. Human NK cells were purified out of leukocyte concentrates. All animal work was carried out in accordance with the NIH guidelines Guide for the Care and Use of Laboratory Animals. Cell lines, cell culture and generation of rho zero cells and animal studies

(20) Human colon carcinoma cell lines SW480, HCT116, HT29 and Caco2, the human glioblastoma cell line U251MG and the Natural killer cell line NK-92 Cl were purchased from ATCC. Cell lines were regularly tested for contamination by multiplex PCR performed in the Genomics and Proteomics Core Facility 2 (DKFZ, Heidelberg, Germany). For experiments, cells were cultured for no more than 10 passages. Rho zero cells were generated as described earlier 1. (Gdynia, G. et al. Danger signaling protein HMGB1 induces a distinct form of cell death accompanied by formation of giant mitochondria. Cancer research 70, 8558-8568). Cells used in the experiments were cultured in RPMI (#1640, colon carcinomas, NK cells) or DMEM high glucose (#41965-039, glioblastoma cells) medium. Rho zero cells were generated as described earlier (Gdynia, G. et al. ibd.). Briefly, cells were cultured in RPMI medium (10% FCS, 1% P/S) supplemented with 250 ng/ml ethidiumbromide, 50 μg/ml L-pyruvate and 5 mg/ml uridine over a period of 12 weeks. For cytotoxicity measurements, cells were cultured in 96-well plates, treated with recombinant human HMGB1 protein (Sigma-Aldrich®), glycyrrhizinic acid ((3β, 18α)-30-hydroxy-11, 30-dioxoolean-12-en-3-yl 2-O-β-D-glucopyranuronosyl-β-D-glucopyranosiduronic acid)) (Sigma-Aldrich®), 10 μM (non-toxic) or 100 μM (cytotoxic) P-M2tide (aa sequence: GGAVDDDpYAQFANGG; #BML-P239-0001; Enzo Life Sciences) or 100 μM ML-265 (Cayman Chemical), then cell viability was assessed by crystal violet staining (Gillies, R. J., Didier, N. & Denton, M. Determination of cell number in monolayer cultures. Analytical biochemistry 159, 109-113 (1986)). Malic enzyme 1 knock-down was performed with 40 nM siRNA using lipofectamine in 6-well plates followed by treatment with 80 nM HMGB1 for 72 h. Sequences of siRNA were: ME1, 5′-CCCUGUGGGUAAAUUGGCUCUAUAU-3′ and scrambled control 5′-CCUGCAGUACUUCAAGCGGtt-3′. PKM2 siRNA was from Santa Cruz. A nonspecific siRNA served as control (Dharmacon, Schwerte, Germany). For overexpression of PKM2 or ME1, cells were transfected with pCMV-PKM2 (Sino Biological Inc., Beijing, China) or pCMV-ME1 (OriGene, Rockville, Md., USA) using Lipofectamine 2000. For cytotoxicity measurements confluent cells were cultured in 96-well plates if not otherwise indicated. Cytotoxic activity of supernatants from stimulated NK cells was assessed in 96-well plates for 3 days with RPMI medium as reference. Recombinant human HMGB1 (10 ng, Sigma) suited as positive control. For mass isotopomer assays, cells were cultured in glucose- or glutamine-free medium supplemented with either uniformly labeled (U)-13C-D-glucose or U-13C-glutamine (Sigma-Aldrich®).

(21) For animal studies six-week-old female and male athymic CD1 nude mice (Charles River, n=40) were injected subcutaneously with 5×106 SW480 or HT29 cells in 100 μl PBS in the right flank using a 30-gauge needle. Treatment was started when tumors were palpable. Daily intraperitoneal injections at the contralateral side for 2 weeks were done with 10 μg rhHMGB1 in 500 μl PBS or PBS only (control group) and/or 12.5 mg/kg/injection L-DON (Ovejera, A. A., Houchens, D. P., Catane, R., Sheridan, M. A. & Muggia, F. M. Efficacy of 6-diazo-5-oxo-L-norleucine and N-[N-gamma-glutamyl-6-diazo-5-oxo-norleucinyl]-6-diazo-5-oxo-norleucine against experimental tumors in conventional and nude mice. Cancer research 39, 3220-3224 (1979).). Tumor volume was measured by a calliper using the ellipsoid formula (length×width×height×½) as described (Tomayko, M. M. & Reynolds, C. P. Determination of subcutaneous tumor size in athymic (nude) mice. Cancer Chemother Pharmacol 24, 148-154 (1989).). After 2 weeks of treatment the animals were sacrificed.

(22) Reversed-Phase HPLC Purification and Identification of HMGB1

(23) Reversed phase chromatography: HMGB1 was extracted and purified by reversed phase chromatography referring to Zetterstrom and coworkers (Zetterstrom, C. K. et al. High mobility group box chromosomal protein 1 (HMGB1) is an antibacterial factor produced by the human adenoid. Pediatric research 52, 148-154) with the exception that Source 15 media were applied for chromatography. For the first purification step a Resource RPC column (6.4×100 mm; GE Healthcare) was applied. Solvent A was water with 0.17% TFA, solvent B was acetonitrile with 0.15% TFA. Flow rate was 1 ml/min. The following elution program was performed: 5% solvent B isocratic for 10 min, 5-30% B linear for 15 min, 30-60% B linear for 45 min, 60-90% B for 5 min, 90% B isocratic for 5 min. The second purification step was conducted on a Source 15RPC ST 4.6/100 column applying the same elution conditions as described above. Final purification was achieved on the Source 15RPC ST 4.6/100 column by elution with 5% B isocratic for 10 min, 5-40% B linear for 15 min, 40-50% B linear for 45 min, 50-90% B for 5 min and 90% B isocratic for 5 min.

(24) NK-92 Cl cells were cultured in minimum essential medium (MEM) alpha (Gibco) supplemented with 12.5% fetal bovine serum (Gibco), 12.5% horse serum (Life technologies GmbH), 0.1% 2-mercaptoethanol (Gibco) and 100 IU/ml penicillin and 100 μg/ml streptomycin (both Sigma Aldrich). Cells were split and expanded by carefully rocking the culture flasks on a daily basis and adding fresh medium upon necessity. 24 hours before harvesting the cells, recombinant human IL-2 (Tecin™ from Roche, kindly provided by the NIH) was added to a concentration of 100 IU/ml. 6×10.sup.8 NK-92 Cl cells were harvested from 1.8 I of culture medium and used for purification of intracellular membraneous vesicles as described.sup.7. Coomassie blue staining of all eluates (80) was performed with Brilliant Blue R-250 dye (Sigma) according to standard protocols. HMGB1 was detected by immunoblot analysis using human anti-HMGB1 antibody (1:1,000, abcam). The gel was stained with the Pierce Silver Stain Kit (Thermo Scientific, Rockford, Ill.) according to the manufacturer's instructions.

(25) Preparation and Culture of Human Natural Killer Cells

(26) Human NK cells were purified out of leukocyte concentrates from the blood bank in Mannheim (Germany). The concentrate was diluted with PBS and subjected to a centrifugation step on biocoll separation solution (Biochrom AG). The buffy coat was harvested and plastic adherence was carried out for 45 minutes. Out of the obtained peripheral blood leukocytes NK cells were isolated with the human NK cell isolation kit (Miltenyi) according to the manufacturer's instructions. Highly pure NK cells (95% CD3− CD56+ cells as determined by flow cytometry) were then cultured in CellGro stem cell growth medium (CellGenix) with 10% human AB serum (PAA Laboratories), 200 U/ml recombinant human interleukin-2 (IL-2, National Institutes of Health) and 100 U/mL penicillin and 100 mg/ml streptomycin (Sigma-Aldrich) at a density of 1×10.sup.6 cells/ml. After 6 days the NK cells were harvested, counted and re-seeded at a density of 2×10.sup.6 cells/ml in antibody pre-coated wells of a 96 well-plate in RPMI (Sigma-Aldrich) supplemented with 10% fetal calf serum (Invitrogen) and 100 U/ml IL-2. For the coating, one day before seeding the cells, the wells were incubated with 1 μg/ml of either mIgG1 (clone MOPC-21) or anti-NKp30 antibody (clone P30-15, both from BioLegend) in PBS over night at 4° C. After 2 days on the pre-coated plates, the supernatants were harvested and centrifuged to pellet potential cellular contaminants. Aliquots of the supernatants were used for performing an IFN-γ ELISA (BioLegend) according to the instructions provided by the manufacturer.

(27) Ex Vivo Colon Carcinoma Specimens, Colon Carcinoma Tissue Microarray

(28) Immediately after the surgical removal of the colon part containing the tumor, a fresh tumor biopsy was processed with a vibrating blade microtome (Vibratome™, Leica). Tissue slices of 300 μm were generated and incubated for the indicated times in RPMI cell culture medium. Control sections were fixed overnight in buffered 4% formalin (pH 7.4) solution, then paraffin embedded and hematoxylin and eosin (HE) staining was performed on an automated staining system (Techmate 500, DakoCytomation). HE-sections were reviewed by pathologists (WR, GG) for the presence of colorectal carcinoma. All surgical specimens were obtained from the Department of General, Visceral and Accident Surgery of the Heidelberg University Hospital (Germany). The use of the human tissue for study purposes was approved by the local ethics committee at the Heidelberg University Hospital.

(29) For creation of the tumor microarray (TMA), tissue samples from 1.260 colorectal carcinoma patients, included in the German DACHS (Darmkrebs: Chancen der Verhutung durch Screening; Colon Cancer: Chances of Prevention through Screening) case control study, were collected by the Tumor Tissue Bank of the NCT Heidelberg. The use of the human tissue was approved by the local ethics committee of the University of Heidelberg and the medical boards of Baden-Wuerttemberg and Rhineland-Palatinate. Written informed consent was obtained from each participant at baseline, including the assignment of tumor tissue from patients with CRC.

(30) Immunohistochemistry

(31) TMA sections were immunostained as described earlier (Gdynia, G. et al. Basal caspase activity promotes migration and invasiveness in glioblastoma cells. Molecular cancer research: MCR 5, 1232-1240, doi:10.1158/1541-7786.MCR-07-0343 (2007)) using an automated staining system (Techmate 500, DakoCytomation). Visualization was done with avidin-biotin-complex peroxidase, aminoethylcarbazole and hematoxylin. The sections were incubated with the rabbit polyclonal anti-malic enzyme 1 antibody (1:100, ab97445, abcam) and processed with the following kits: ChemMate Detection Kit (K5003, DakoCytomation), ChemMate Buffer Kit (K5006, DakoCytomation) and Avidin/Biotin Blocking Kit (SP-2001, Vector Laboratories). A product of the scores of staining intensity and quantity of positive cancer cells was assessed semiquantitatively and independently by two pathologists (WR, GG). Herein the intensity range was 0=negative; 1=low; 2=medium and 3=high and the quantity 0=no positivity; 1=positivity in 0-10%; 2=positivity in 11-50%; 3=positivity in 51-80%; 4=positivity in more than 80%. For few cases of discrepant validation a consensus score was determined. The staining and evaluation was additionally performed on a second TMA giving similar results. The final immunoreactive score (IRS, ranging from 0 to 12) is obtained by multiplication of the intensity score and the quantity score. For ME1 low, moderate and strong positive expression was defined as IRS<3, IRS between 3 and 6, and IRS >6, respectively. For HMGB1 strong and strong positive expression was defined as IRS between 3 and 6, and IRS >6, respectively. Only 5 tumors were completely HMGB1 negative, thus here statistical analysis could only be performed using strong and strong positive expression. ME1 antibody specifity: cells were plated and transfected on glass coverslips in 6-well plates. The coverslips were collected, fixed with paraformaldehyde and immunostained with ME1 antibody as described for the TMA sections.

(32) Electron Microscopy

(33) Cells were fixed (2.3% glutaraldehyde in 50 mM sodium cacodylate, pH 7.2) in situ for 30 min at 4° C., scraped, centrifuged at 200×g for 10 min at 4° C. and stained (2% osmium tetroxide, 5% uranyl acetate). Ultrathin sections from dehydrated and Epon embedded samples were microphotographed with a Zeiss EM-10A electron microscope at 80 kV. Grating replica suited as controls for the magnification indicator.

(34) Enzymatic Assays

(35) Enzymatic activities of respiratory chain complexes, glycolytic proteins and malic enzyme were determined in subcellular fractions as previously described (Kaminski, M. M. et al. T cell activation is driven by an ADP-dependent glucokinase linking enhanced glycolysis with mitochondrial reactive oxygen species generation. Cell reports 2, 1300-1315 (2012).Bruncko, M. et al. Naphthamidine urokinase plasminogen activator inhibitors with improved pharmacokinetic properties. Bioorganic & medicinal chemistry letters 15, 93-98, (2005)) using a computer-tuneable spectrophotometer (Spectramax Plus Microplate Reader, Molecular Devices; Sunny Vale, Calif., USA) operating in the dual wavelength mode; samples were analyzed in temperature-controlled 96-well plates in a final volume of 300 μl. Activity of ME1 was recorded in presence of increasing amounts of malic acid (0.02, 0.05, 0.1, 0.2, 0.5, 1, 2.5 mM). Vmax and Km were calculated using a Hanes-Woolf plot. In the presence of high substrate levels the Km for malic acid was similar in all three tested cell lines (SW480: 0.32 mM, HCT116: 0.30 mM, HT29: 0.31 mM). Vmax (mU/mg protein) values were 3.38 (SW480, 0.5-5.0 mM malic acid), 5.77 (HCT116, 0.5-5.0 mM malic acid) and 3.68 (HT29, 0.5-5.0 mM malic acid) and 1.68 (SW480, 0.02-0.2 mM malic acid), 3.24 (HCT116, 0.02-0.2 mM malic acid) and 1.89 (HT29, 0.02-0.2 mM malic acid). Two isoforms of ME1, mitochondrial (NAD(P)+dependent) ME3 and mitochondrial (NAD+dependent) ME2, had very low or no detectable activities (data not shown). Dimeric PK M2 is virtually inactive at physiological PEP levels allowing differentiation of both forms by using very high (10 mM) and low (100 μM) amounts of PEP in the enzymatic assay.

(36) Glucose-6-phosphate levels in cells were measured using the Glucose-6-phosphate assay kit (Sigma) according to the manufacturer's protocol.

(37) Metabolic Assays

(38) Lactate derived from the metabolism of .sup.13C.sub.6-D-glucose or .sup.13C.sub.5-glutamine was determined by comparing the CH.sub.3 group intensities of labeled and non-labeled lactate in NMR. Mitochondrial respiration was measured using an Oroboros 1 oxygraph system. Glycolysis was measured by monitoring the conversion of 5-.sup.3H-Glucose to .sup.3H.sub.2O. Incorporation of .sup.14C into RNA ribose from U-.sup.14C-labelled glucose was taken as glucose utilization in the pentose phosphate shunt. Enzymatic activities were determined in subcellular fractions as previously described.sup.24.

(39) In Silico HMGB1—PK M2 Protein Docking Studies

(40) The individual HMGB1 Box domains (PDB: 1CKT, 2YRQ) were used rather than the complete HMGB1 structure due to the complexity of accurately accounting for the structural flexibility of the linker region (residues 79-94) between the two domains. Initial structures were taken from the Protein Data Bank (PDB), and X-ray structures were taken preferentially over NMR structures where possible. The PK M2 structure (PDB code: 3BJF), HMGB Box A (PDB code: 1CKT), HMGB Box B (PDB code: 2YRQ, residues 95-163) were used. All calculations used the chain A from 3BJF. For the calculation with FBP present, this residue was saved as a mol2 file in UCSF Chimera, and submitted to the PDB2PQR web server in addition to the modified PDB file. In all other calculations, all ligands were removed from the structures. The PDB2PQR web server was used to prepare all structures for simulation with SDA, using the AMBER force field parameters, and protonation states assigned at pH 7. Each HMGB1 structure, and the PK M2 phosphorylated at Y105, was phosphorylated using the build feature of Chimera. Charges and radii were manually added to the PQR files using the phosphotyrosine parameters as specified in the AMBER parameters database

(41) (http://personalpages.manchester.ac.uk/staff/Richard.Bryce/amber/pro/phos2_inf.html). APBS version 1.2.1 was used to solve the linearized Poisson-Boltzmann equation with simple Debye-Hückel boundary conditions, a protein dielectric constant of 1, and a solute dielectric of 78 to calculate the electrostatic potential for each protein on cubic grids of 129 points, with 1 Å grid spacing. The potential was calculated at 50 mM ionic strength, with positive and negative ions with 1.5 Å radius. Dielectric and ion-accessibility coefficients were calculated using the smoothed method (smol option), and the smoothing window was set to 0.3 Å. For the purposes of the effective charge calculations, test charges were placed as per the original SDA calculations, and additionally on the phosphorus and oxygen atoms of the phosphotyrosine residue. The SDA program was used to calculate in excess of 40,000 docked encounter complexes of each of the four HMGB1 models (unphosphorylated Box A, phosphorylated Box A, unphosphorylated Box B, phosphorylated Box B) with PK M2. The SDA calculations included electrostatic interaction, electrostatic desolvation and hydrophobic desolvation terms, with weighting factors 0.5, 1.0 and −0.013 respectively. The protein probe radius was set to 1.77 Å, solvent probe to 1.4 Å, and an exclusion grid spacing of 0.5 Å. Proteins were initially separated by 260 Å, and a simulation was stopped if the center-center distance exceeded 540 Å or the total simulation time exceeded 5,000 ps. The top 5,000 docked complexes, as ranked by favorable interaction energy, were retained for cluster analysis using the hierarchical clustering tool provided with SDA. For each simulation, the docked complexes were clustered to produce 10 clusters for quantitative and visual analysis. All images were prepared using the VMD visualization software.

(42) NMR Analysis of Metabolites

(43) For analysis of .sup.13C-lactate efflux 1 ml of the cellular supernatant was centrifuged (8,000×g, 10 min, 4° C.) to spin down cellular debris. To 500 μl of the supernatant 10% of D.sub.2O were added respectively and transferred to 5 mm NMR sample tubes. The samples were measured with a Bruker AvanceIII 600 NMR spectrometer, equipped with a cryogenically cooled detection probe (QNP-CryoProbe™).

(44) Parameters for Measurement:

(45) Magnetic Field 14.09 Tesla; sample temperature 295 K; pulse width 4.7 us (corresponding to 30° flip angle); Broadband Composite Pulse Decoupling (Waltz65) during acquisition and relaxation delay, 128K total acquisition data points; acquisition time 1.8 sec; relaxation delay 1.5 sec; 512 transients; total experiment time 30 min.

(46) Processing Parameters:

(47) Zero filling to 256K real data points, exponential multiplication (lb=1.0 Hz); Fourier transformation with backward linear prediction in order to compensate for base line artifacts.

(48) Data Analysis:

(49) The integral of the signal of the .sup.13CH.sub.3 group of lactate (singlet at δ=20.108 ppm for non labeled lactate and doublet for labeled lactate at δ=20.097 ppm (.sup.1J(.sup.13C.sup.13C)=36.8 Hz) respectively) was taken as the measure of lactate concentration. In order to get reliable quantitative results, the intensities were calibrated with standard samples containing known amounts of labeled and non-labeled lactate. This procedure also compensates errors due to incomplete relaxation of the .sup.13C nuclei within the chosen repetition time (3.3 sec) The determination of concentrations was performed by using the “ERETIC” functionality built in the Bruker NMR software (Topspin 3.2, Bruker BioSpin 2012). The concentrations obtained in this way were corrected for the incomplete degree of .sup.13C enrichments in .sup.13C.sub.6 glucose and .sup.13C.sub.5 glutamine respectively (98%).

(50) Immunoblot Analysis, Subcellular Fractionation, Ultrafiltration

(51) Cells were lysed in lysis buffer P (20 mM Tris-HCl (pH 7.4), 137 mM NaCl, 10% (v/v) glycerine, 1% Triton X-100, 2 mM EDTA, 100 mM phenylmethylsulfonyl fluoride and protease inhibitors (Complete mini from Roche). Lysates were centrifuged at 14,000×g (10 min) at 4° C. Total protein was measured by the Bradford (Bio-Rad) method. Soluble protein was resolved by SDS-PAGE, blotted onto nitrocellulose and incubated with one of the following antibodies: mouse monoclonal anti-β-actin (1:3,000, Sigma-Aldrich), rabbit anti PK M2 (1:1,000, Cell Signaling), rabbit anti-malic enzyme 1 (1:1,000, abcam). Appropriate secondary antibodies (1:3,000, horse-radish peroxidase-conjugated) were from Bio-Rad. Visualization was done by enhanced chemiluminescence technique (GE-Healthcare). Mitochondrial fractions were extracted using the ApoAlert Cell Fractionation Kit (Clontech) as described earlier (Gdynia, G. et al. BLOC1S2 interacts with the HIPPI protein and sensitizes NCH89 glioblastoma cells to apoptosis. Apoptosis: an international journal on programmed cell death 13, 437-447, (2008)).

(52) Ultrafiltration of the PK M2-HMGB1 complex: equimolar amounts of HMGB1 and PKM2 were mixed in a final volume of 300 μl and filtrated (14000 g, 4° C.) to a final volume of 15 μl in an Amicon Ultra 0.5 ml 30k device (Merck-Millipore, Darmstadt, Germany). The retentate was adjusted to the original volume after centrifugation. Then filtrate and retentate were analyzed by Western Blot. For controls HMGB1 and PKM2 were also analyzed alone. Pure HMGB1 (2 μM) suited as a negative control.

(53) Quantitative PCR Analysis

(54) Quantitative PCR analysis was performed as described previously (Fassl, A. et al. Notch1 signaling promotes survival of glioblastoma cells via EGFR-mediated induction of anti-apoptotic Mcl-1. Oncogene 31, 4698-4708, doi:10.1038/onc.2011.615 (2012)).

(55) Statistical Analysis

(56) We evaluated the association between ME1 or HMGB1 expression and local tumor extent (pT) and lymph node metastasis (pN) for all colorectal samples together as well as for the colon and rectal cancer subgroups using the linear by linear association test (Agresti A. Categorical Data Analysis. John Wiley & Sons. Hoboken, N.J., 2002). Overall survival time was defined as the time from diagnosis until death from any cause. Endpoints for progression-free survival were tumor recurrence, distant metastases or death from any cause, whatever occurred first. For the analysis of CRC (colorectal cancer)-related survival, deaths from unrelated causes were treated as competing events. Multivariate (cause-specific) proportional hazards regression models included ME1 or HMGB1 expression (IRS score), age, sex, grade, pT, pN, pM, tumor site, adjuvant and neoadjuvant chemo- and radiotherapy. The pT stadium is defined by the extent of tumor invasion into the colonic wall: submucosa (pT1), muscularis propria (pT2), subserosa/pericolic fat tissue (pT3), and perforation through peritoneum/invasion into other organs (pT4). The pN stadium is definied by the number of regional lymph node metastasis: metastasis in 1 regional lymph node (pN1a), metastasis in 2 to 3 regional lymph nodes (pN1b), tumor deposit(s) in the subserosa, or in the non-peritonealized pericolic or perirectal soft tissue without regional lymph node metastasis (pN1c), metastasis in 4 or more regional lymph nodes (pN2). The pM0 or pM1 stadium is definied by the absence or the occurrence of distant metastasis, respectively.

(57) Results of laboratory experiments were analyzed using paired t tests. Results were illustrated using means±SD. For all statistical tests a significance level of 5% was used. Significance in figures is shown by asterisks. Statistical analyses were performed using the statistical software environment R, version 2.15.3, and Microsoft Excel 2010 software.

EXAMPLE 7

(58) In a particular preferred embodiment of the invention the reagents needed for the enzyme activity determination are deposited as solids at the bottom and/or wall of the well-plates. Well-plates treated that way allow the addition of the sample in a sample buffer and the photometric measurement of the activity of different enzymes of the samples in the wells of the plate.

(59) Well-plates with the solid reagents deposited to the wall and/or bottom of the wells can be obtained, for example, by dry-freezing (lyophilization), e.g. treatment of the wells of a well plate with a certain amount of a buffer solution with the reagents needed for the determination of enzyme activity in a suited concentration. Under vacuum the wells treated that way can be dried at a low temperature to evaporate the water of the buffer solution. The dried reagents for the determination of enzyme activity of a specific enzyme adhere to the bottom and the wall of the well.

(60) TABLE-US-00002 TABLE 2 CLL cells used in this study with sensitivity towards PKM2 modulating drug candidates PM2-tide (inhibitor; 100 μM, 250 M) and DASA (activator; 100 μM) and predicted sensitivity by the anaerobic glycolysis predictor EnFin. Cells were classified sensitive to the drug when less than 60% were viable upon treatment (25,000 - 50,000 CLL cells, CellTiter-Glo ® Luminescent Cell Viability Assay, Promega). * = Highly sensitive leukemia cells responding to 100 μM PM2- tide. $ = Interval of 10 months between blood taking (same patient). ™ = Interval of 3 months between blood taking (same patient). The anaerobic glycolysis predictor values S for the samples were calculated according to eq. 1 as described herein above and were: 14PB0079: 1.30; 14PB0132: 2.33; 13PB0500: 1.97; 13PB0473: 2.03; 13PB0555: 2.04; 13PB0649: 2.15; 13PB0501: 1.51; 14PB0471: 1.22; 14PB0451: 1.68. Sensitivity Sensitivity classification prediction Viable cells Viable cells based on viable based on Patient Sample [PM2-tide] [DASA] cells EnFin P0074 14BP0079  46%* 37% Sensitive Sensitive P0010.sup.$ 14PB0132 75% 67% Resistant Resistant P0641 ™ 13PB0500 100%  100%  Resistant Resistant P0645 13PB0473 91% 90% Resistant Resistant P0369 13PB0555 100%  100%  Resistant Resistant P0641 ™ 13PB0649 93% 88% Resistant Resistant P0010.sup.$ 13PB0501 100%  100%  Resistant Resistant P0460 14PB0471  56%* 72% Sensitive Sensitive P0701 14PB0451 100%  100%  Resistant Resistant Sensitive Resistant Total Correct (no.) (no.) (no.) prediction [%] EnFin Sensitive 2 0 2 100 predicted EnFin Resistant 0 7 7 100 predicted Total 2 7 9

(61) TABLE-US-00003 TABLE 3 PM2-tide (inhibitor; 100 μM, 250 μM) and DASA (activator; 100 μM) response prediction of colorectal cancer and chronic lymphocytic leukemia samples using the anaerobic glycolysis predictor EnFin. CRC: 50 mg of colorectal primary cancer tissue (UICC stage I-IV) were used. CLL: a volume of 500 μl leukemic cells from the buffy coat was used for analysis. Tumor Total Predicted “PM2-tide/ Data set type number DASA - sensitive” no. (%) source CLL 209 42 (20) EnFin Colorectal 26  7 (27) EnFin Cancer

EXAMPLE 8

Non-Phosphorylatable HMGB1

(62) Generation of GInHMGB1

(63) A plasmid encoding a HMGB1 polypeptide with its B-Box domain tyrosine residues replaced by glutamine (SEQ ID NO:12) was transfected into HEK cells (serum-free suspension cell culture, 1,000 ml (app. 2.5×10.sup.6 cells/ml), then supplemented with Valproic Acid). The cell pellet was homogenized and purified via IMAC and TALON (Clontech) Resins and eluted using imidazole. Eluates were analyzed via SDS-PAGE (Coomassie staining). After pooling of positive eluates the protein was gel filtrated (Superdex) and finally analyzed by SDS-PAGE. The purified Protein was used in the concentrations indicated in FIGS. 7 and 8 in inhibition assays.

(64) MTT-Assay

(65) Assays were performed at three time points (0, 24, 48 hours), and the experiment was repeated four times. At each time point the cell viability was measured using the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) assay. The MTT assay was performed as described by Mosmann (Mosmann T. Rapid colorimetric assay for cellular growth and survival: application to proliferation and cytotoxicity assays. J Immunol Methods. 1983;65:55-63.). In brief, at each time point, medium was aspirated from the wells and MTT (1 mg/ml) was gently added to each well. The cells were incubated for 3 hours at 37° C. in 5% CO.sub.2, after which the MTT was aspirated and acidified isopropanol (0.04 M HCl) added to solubilize the reduced blue formazan crystals. Aliquots were transferred to a 96-well plate and the absorbance measured at a test filter of 590 nm and a reference filter of 630 nm on a 96-well plate reader. In all colon carcinoma cell lines used in the experiment there was a significant decrease in proliferation compared to the untreated control (p<0.05, n=4).

(66) LDH-Assay

(67) Lactate dehydrogenase activity was assayed spectrophotometrically by measuring the oxidation of NADH with pyruvate substrate at 340 nm as described in Bergmeyer and Berndt E. (1974). Results were analyzed using the following equation:

(68) $\mathrm{Viability}=\frac{U\text{/}\mathrm{ml}\left[\mathrm{medium}\right]}{\left(U\text{/}\mathrm{ml}\left[\mathrm{cells}\right]+U\text{/}\mathrm{ml}\left[\mathrm{medium}\right]\right)}\times 100$

(69) In all colon carcinoma cell lines used in the experiment there was a significant release of LDH into the medium compared to the untreated control (p<0.05, n=4). Thus GInHMGB1 can induce cell death in different carcinoma cell lines.

REFERENCES

(70) 1 Wang, H. et al. HMG-1 as a late mediator of endotoxin lethality in mice. Science 285, 248-251 (1999). 2 Lotze, M. T. & Tracey, K. J. High-mobility group box 1 protein (HMGB1): nuclear weapon in the immune arsenal. Nature reviews. Immunology 5, 331-342, doi:10.1038/nri1594 (2005). 3 Apetoh, L. et al. Toll-like receptor 4-dependent contribution of the immune system to anticancer chemotherapy and radiotherapy. Nature medicine 13, 1050-1059, doi:10.1038/nm1622 (2007). 4 Saidi, H., Melki, M. T. & Gougeon, M. L. HMGB1-dependent triggering of HIV-1 replication and persistence in dendritic cells as a consequence of NK-DC cross-talk. PloS one 3, e3601, doi:10.1371/journal.pone.0003601 (2008). 5 Yang, D. et al. High mobility group box-1 protein induces the migration and activation of human dendritic cells and acts as an alarmin. Journal of leukocyte biology 81, 59-66, doi:10.1189/jIb.0306180 (2007). 6 Zetterstrom, C. K. et al. High mobility group box chromosomal protein 1 (HMGB1) is an antibacterial factor produced by the human adenoid. Pediatric research 52, 148-154, doi:10.1203/00006450-200208000-00004 (2002). 7 Calogero, S. et al. The lack of chromosomal protein Hmg1 does not disrupt cell growth but causes lethal hypoglycaemia in newborn mice. Nature genetics 22, 276-280, doi:10.1038/10338 (1999). 8 Grundtman, C. et al. Effects of HMGB1 on in vitro responses of isolated muscle fibers and functional aspects in skeletal muscles of idiopathic inflammatory myopathies. The FASEB journal: official publication of the Federation of American Societies for Experimental Biology 24, 570-578, doi:10.1096/fj.09-144782 (2010). 9 Gdynia, G. et al. Danger signaling protein HMGB1 induces a distinct form of cell death accompanied by formation of giant mitochondria. Cancer research 70, 8558-8568, doi:10.1158/0008-5472.CAN-10-0204 (2010). 10 Anastasiou, D. et al. Pyruvate kinase M2 activators promote tetramer formation and suppress tumorigenesis. Nature chemical biology 8, 839-847, doi:10.1038/nchembio.1060 (2012). 11 Christofk, H. R., Vander Heiden, M. G., Wu, N., Asara, J. M. & Cantley, L. C. Pyruvate kinase M2 is a phosphotyrosine-binding protein. Nature 452, 181-186, doi:10.1038/nature06667 (2008). 12 Hitosugi, T. et al. Tyrosine phosphorylation inhibits PKM2 to promote the Warburg effect and tumor growth. Science signaling 2, ra73, doi:10.1126/scisignal.2000431 (2009). 13 Bhat, R. & Rommelaere, J. NK-cell-dependent killing of colon carcinoma cells is mediated by natural cytotoxicity receptors (NCRs) and stimulated by parvovirus infection of target cells. BMC cancer 13, 367, doi:10.1186/1471-2407-13-367 (2013). 14 Vander Heiden, M. G., Cantley, L. C. & Thompson, C. B. Understanding the Warburg effect: the metabolic requirements of cell proliferation. Science 324, 1029-1033, doi:10.1126/science.1160809 (2009). 15 Ikeda, Y. & Noguchi, T. Allosteric regulation of pyruvate kinase M2 isozyme involves a cysteine residue in the intersubunit contact. The Journal of biological chemistry 273, 12227-12233 (1998). 16 Erlandsson Harris, H. & Andersson, U. Mini-review: The nuclear protein HMGB1 as a proinflammatory mediator. European journal of immunology 34, 1503-1512, doi:10.1002/eji.200424916 (2004). 17 Youn, J. H. & Shin, J. S. Nucleocytoplasmic shuttling of HMGB1 is regulated by phosphorylation that redirects it toward secretion. J Immunol 177, 7889-7897 (2006). 18 Lazzarino, G., Viola, A. R., Mulieri, L., Rotilio, G. & Mavelli, I. Prevention by fructose-1,6-bisphosphate of cardiac oxidative damage induced in mice by subchronic doxorubicin treatment. Cancer research 47, 6511-6516 (1987).

(71) 19 Sandev, S., Khattar, S. K. & Saini, K. S. Production of active eukaryotic proteins through bacterial expression systems: a review of the existing biotechnology strategies. Molecular and cellular biochemistry 307, 249-264, doi:10.1007/s11010-007-9603-6 (2008). 20 Veith, N. et al. Organism-adapted specificity of the allosteric regulation of pyruvate kinase in lactic acid bacteria. PLoS computational biology 9, e1003159, doi:10.1371/journal.pcbi.1003159 (2013). 21 Zhdanov, A. V., Waters, A. H., Golubeva, A. V., Dmitriev, R. I. & Papkovsky, D. B. Availability of the key metabolic substrates dictates the respiratory response of cancer cells to the mitochondrial uncoupling. Biochimica et biophysica acta 1837, 51-62, doi:10.1016/j.bbabio.2013.07.008 (2014). 22 Ren, J. G., Seth, P., Everett, P., Clish, C. B. & Sukhatme, V. P. Induction of erythroid differentiation in human erythroleukemia cells by depletion of malic enzyme 2. PloS one 5, doi:10.1371/journal.pone.0012520 (2010). 23Tennant, D. A., Duran, R. V. & Gottlieb, E. Targeting metabolic transformation for cancer therapy. Nature reviews. Cancer10, 267-277, doi:10.1038/nrc2817 (2010). 24 Kaminski, M. M. et al. T cell activation is driven by an ADP-dependent glucokinase linking enhanced glycolysis with mitochondrial reactive oxygen species generation. Cell reports 2, 1300-1315, doi:10.1016/j.celrep.2012.10.009 (2012).