Most cancer cells employ the Warburg effect to support anabolic growth and tumorigenesis. Here, we discovered a key link between Warburg effect and aberrantly activated Wnt/β-catenin

signalling, especially by pathologically significant APC loss, in CRC.

Proteomic analyses were performed to evaluate the global effects of KYA1797K, Wnt/β-catenin signalling inhibitor, on cellular proteins in CRC. The effects of APC-loss or Wnt ligand on the

identified enzymes, PKM2 and LDHA, as well as Warburg effects were investigated. A linkage between activation of Wnt/β-catenin signalling and cancer metabolism was analysed in tumour of

LDHA, and increased glucose consumption and lactate secretion. Pathological significance of this linkage was indicated by increased expression of glycolytic genes with Wnt target genes in

tumour of Apcmin/+ mice and CRC patients. Warburg effect and growth of xenografted tumours-induced by APC-mutated-CRC cells were suppressed by PKM2-depletion.

The β-catenin-PKM2 regulatory axis induced by APC loss activates the Warburg effect in CRC.

Cancer cells, in contrast with normal cells, exhibit an altered metabolism, increased aerobic glycolysis, and decreased oxidative phosphorylation.1,2 Elevated glucose uptake and lactate

production regardless of oxygen availability, which is referred to as aerobic glycolysis or the Warburg effect, are a dominant phenotype of most cancer cells.3 This rewired metabolism is

required for growth, proliferation, and survival of cancer cells and consequently promotes the initiation and progression of tumours.4 Recently, growing evidence has shown that the

reprogramming of cancer metabolism is directly regulated by activation of oncogenes or loss of function mutations of tumour suppressors such as phosphoinositide 3-kinase (PI3K), hypoxia

inducible factor-1 (HIF-1) and p53.1,5,6,7 In addition, many studies showed that mutations and abnormal levels of metabolic enzymes and altered amounts of several metabolites directly affect

tumorigenesis in various types of human cancers.8,9,10,11 These results renew interest in cancer metabolism and suggest the rewired cancer metabolism as a potential therapeutic target for

cancer therapy. However, the mechanistic basis by which cancer metabolism is controlled by oncogenes and tumour suppressor genes varies in different tumour types, and in most cases, the

molecular mechanisms that induce the alterations in the expression of the metabolic enzymes are poorly understood. Furthermore, although cellular metabolism reprogramming is considered as a

critical event during tumorigenesis, the regulatory mechanisms and signalling pathways that initiate and control it remain elusive.

The Wnt/β-catenin signalling pathway is a major pathway that regulates important biological processes including normal development and oncogenesis.12 Aberrant activation of Wnt/β-catenin

signalling has been frequently found in various cancers. Colorectal cancer (CRC) is the most notable cancer type associated with activation of Wnt/β-catenin signalling because loss of

function mutations of adenomatous polyposis coli (APC) occur in up to 90% of human CRC patients.13 APC plays a role as a gatekeeper in colorectal tumorigenesis, and APC loss initiates the

pathogenesis via stabilisation and subsequent nuclear translocation of β-catenin for transcriptional activation of target genes involving cell proliferation and transformation.13 CRC

development caused by this aberrant Wnt/β-catenin signalling is regarded to most likely result from the inappropriate activation of genes.14 Recently, speculative links between Wnt/β-catenin

signalling and cancer metabolism have been poised in several cancers.15,16,17,18 However, regulation of cancer metabolism by Wnt/β-catenin signalling, especially APC mutations, implying the

major pathological significance of activation of Wnt/β-catenin signalling and initiation of CRC, is poorly characterised.

In this study, through systematic proteomic analysis, we identified significant decreases in the levels of glycolytic enzymes such as pyruvate kinase M2 (PKM2) and lactate dehydrogenase A

(LDHA) in DLD1 cells treated with KYA1797K, a small molecule that inhibits Wnt/β-catenin signalling by degradation of β-catenin.19 The Wnt/β-catenin-signalling-dependent regulation of the

metabolic enzyme expression was confirmed by various in vitro studies. The positive relationships between the expression of metabolic enzymes and β-catenin were further confirmed in tumour

tissues of CRC patients, indicating the pathological significance of the linkage. The APC mutation-induced Wnt/β-catenin signalling activation increased the PKM2 and LDHA along with the

Warburg effect in CRC cells. The induction of PKM2, a key enzyme mediating the Warburg effect,20,21 by APC loss occurs via β-catenin/Tcf4-mediated transcription. The role of Wnt/β-catenin

signalling in the pathogenesis of CRC related to the Warburg effect is shown by elevation of PKM2 and subsequent increment of the glycolytic genes including LDHA and glucose transporter

(GLUT1) due to APC loss. In addition, the critical role of PKM2 in tumour growth induced by APC-mutated CRC cells was confirmed in vivo by xenograft mouse model.

These findings provide a key link between aberrantly activated Wnt/β-catenin signalling, especially due to APC loss, and the Warburg effect in CRC. In addition, these results suggest that

PKM2 could be a potential therapeutic target for CRC caused by aberrant Wnt/β-catenin signalling activation.

Human CRC cells (DLD1, SW48, SW480, HCT15, HCT116, RKO, and WiDr), U87 cells, and human embryonic kidney (HEK) cell line 293T cells were obtained from the American Type Culture Collection

(ATCC, Manassas, VA). DLD1, SW48, SW480, and HCT15 cells were cultured in in RPMI 1640 medium (Gibco) supplemented with 10% foetal bovine serum (FBS; Gibco), and RKO, WiDr, U87, and HEK293T

cells were grown in Dulbecco’s modified Eagle’s medium (DMEM; Gibco) containing 10% FBS. HCT116 cells were maintained in McCoy’s 5a Medium (Gibco) containing 10% FBS. Mycoplasma

contamination tests were performed for all cells used in this study. KYA1797K was dissolved in dimethyl sulfoxide (DMSO; Sigma-Aldrich) for the in vitro studies. Recombinant wnt3a, EGF and

bFGF were purchased from PeproTech.

A total of 3 × 106 DLD1 cells per well were seeded in 100 cm2 plates and treated with DMSO or 25 μM KYA1797K for 3 h. The cells were harvested and lysed in the sample buffer (7 M urea, 2 M

thiourea, 4.5% CHAPS, 100 mM DTE, 40 mM Tris (pH 8.8)) complemented with complete protease inhibitors (Roche) on ice for 30 min by adding DNase I. The lysates were applied to immobilised pH

3–10 nonlinear gradient strips (Amersham Biosciences) and isoelectric focusing was performed at 80,000 Vh. The second-dimensional separation was performed in 9–16% linear gradient

polyacrylamide gels at a constant 40 mA per gel for ∼5 h. After protein fixation in 40% methanol and 5% phosphoric acid for 1 h, the gels were stained with Coomassie brilliant blue G-250 for

12 h. Stained gels were scanned using a GS-710 imaging densitometer (Bio-Rad) and analysed with an Image Master 2-DE Platinum image analysis program (Amersham Biosciences). Expression

levels of the spots were determined by the relative spot volume of proteins compared with the volume of a single spot in the gel using the Melanie II program (GenBio).22

For mass spectrometry fingerprinting, protein spots were cut out of the gel and digested using trypsin (Promega), as previously described.23 Aliquots of the peptide mixtures obtained from

digestion were applied onto a target disk and allowed to air-dry. Spectra were obtained using a Voyager DE PRO MALDI-TOF spectrometer (Applied Biosystems). Protein database searching was

performed with MS-Fit24 using monoisotopic peaks. A mass tolerance was first allowed within 50 ppm, and then recalibration was performed at 20 ppm after obtaining the protein lists.

A total of 5–7 × 104 cells per well were seeded in 12-well plates. Forty-eight hours after plating, the medium was collected, and the glucose and lactate levels were examined using a glucose

colorimetric/fluorometric assay kit (Biovision) and lactate colorimetric/fluorometric assay kit (Biovision), respectively, according to manufacturer’s instruction. Glucose and lactate were

calorimetrically measured at 590 nm using a FLUOstar OPTIMA (BMG LABTECH), and the glucose consumption and lactate production were normalised to cell number.

The promoter of PKM2 was subcloned into the pGL3-Basic vector (pGL3; Addgene) to obtain a pGL3-PKM2 promoter-LUC plasmid. The pGL3-PKM2 promoter-LUC plasmid was co-transfected with internal

control pCMV-β-galactosidase (β-gal) reporter plasmid (Clontech). When treating Wnt3a proteins, the cells were treated with 50 ng/ml of Wnt3a with fresh media for 24 h after 24 h of

transfection. Forty-eight hours after transfection, cells were harvested and lysed in Reporter Lysis Buffer (Promega) according to the manufacturer’s instructions. Luminescence and

β-galactosidase activity were measured with a FLUOstar OPTIMA (BMG LABTECH). Luciferase activity was normalised to the β-galactosidase activity. Relative luciferase activity was normalised

to the control for each experiment.

All animal experiments were performed in accordance with the Korean Food and Drug Administration guidelines. Protocols were reviewed and approved by the Institutional Review Board of

Severance Hospital, Yonsei University College of Medicine. C57BL/6J-ApcMin/+ (Apcmin/+) mice were obtained from Jackson Laboratory (Bar Harbor, ME). Mice were housed in micro-ventilation

cage system (MVCS) cages with a computerised environmental control system (Threeshine Inc.). The temperature was maintained at 24 °C with a relative humidity of 45–55%. Mice were euthanised

by carbon dioxide. To generate Apcmin/+ mice, Apcmin/+ mice were crossed with C57BL/6J+/+ (WT) mice. Mouse genotyping was performed using genomic DNA extracted from the tail, and we used

male mice for further animal experiments. To control genetic background effects, sex-matched littermates were always used. To investigate the expression level of glycolytic enzymes and

β-catenin in WT and Apcmin/+ mice, 12-week-old mice were used and were euthanised by carbon dioxide (WT, n = 5; Apcmin/+, n = 6). For the study of in vivo efficacy of KYA1797K, 5-week-old

Apcmin/+ mice were randomly assigned to 2 groups receiving either vehicle or KYA1797K. The mice were injected i.p. with vehicle (90% PBS and 10% Tween 80; n = 6) or KYA1797K (25 mg/kg; n = 

5) dissolved in the vehicle 5 days per week for 7 weeks. At that time, the mice were euthanised by carbon dioxide and sacrificed. Immediately after sacrifice, the abdomen of each mouse was

cut open longitudinally and cleaned by flushing with PBS. Proximal regions of the small intestine were dissected, stained with 0.025% methylene blue (Sigma-Aldrich) for 2 min, and fixed with

4% paraformaldehyde (PFA). Gross images of tumour tissues were captured, and the tissues were resected and embedded in paraffin according to standard procedures. The tumours were classified

according to standard World Health Organization histopathological criteria. For histopathologic analyses, a subset of freshly isolated tissues was snap frozen in liquid nitrogen and stored

at −80 °C.

All animal experiments were performed in accordance with the Korean Food and Drug Administration guidelines. Protocols were reviewed and approved by the Institutional Review Board of

Severance Hospital, Yonsei University College of Medicine. Four-week-old male athymic BalbC nu/nu mice were purchased from Joongabio Inc. Mice were housed in micro-ventilation cage system

(MVCS) cages with a computerised environmental control system (Threeshine Inc.). The temperature was maintained at 24 °C with a relative humidity of 45–55%. After acclimatisation for 1 week,

sixteen mice were randomly divided into two groups and injected subcutaneously in the dorsal flank with 2 × 107 of SW480 stably expressing control shRNA or shRNA targeting PKM2 (shRNA #1)

in 100 µl of PBS:Matrigel (BD Bioscience; 1:1) (n = 8 per each group) after anesthetised by i.p. injection of sterile avertin (tribromoethanol: 250 mg/Kg). The vital signs of mice were

observed each week after injection, and no death was observed in the all groups. Tumours were measured using Vernier callipers, and tumour volume was calculated according to the following

formula: π/6 × length × width × height. The mice were euthanised by carbon dioxide when the tumour volume exceeded 1500 mm3. Eighty-six days after injection, the mice were sacrificed, and

the tumours were excised, weighed, and fixed in 4% PFA or snap frozen in liquid nitrogen for further analysis.

Tissue microarrays (TMA) for normal and cancer tissues, colon disease spectrum tissue array (BC05002a, BC051110b), were purchased from US Biomax. IHC was performed with antibodies against

PKM2, LDHA, or β-catenin. From the TMA, 30 normal tissue samples and 104 CRC adenocarcinoma tissue samples were used for further analyses. Signals of the TMA slides were analysed using a

bright field microscope (Nikon TE-2000U). For quantitative analysis, the intensity of each staining was determined by IHC Profiler plugin.25 All signals were analysed in a double-blind

manner.

To generate luciferase reporter plasmid containing the promoter of PKM2, two DNA fragments (fragment 1, from the upstream region of exon1 to intron1 (−14762 to −10163); fragment 2, from −230

of intron 1 to 5′flanking region of exon2 (21 bp)) were obtained by PCR. The following primers were used: fragment 1, forward 5′-CCTACTATGCACCTAATGTGAGC-3′ and reverse 5′-

ACACTTACTGAGTGTGCCACATCC-3′ (including EcoRV restriction site); fragment 2, forward 5′-GTCTAGGTAGATGTCAGTCAGCC-3′ (including SmaI restriction site) and reverse

5′-TTCGAGATGGCTGCTGAGGTCCTGG-3′. The two DNA fragments were ligated and inserted into the pGL3-Basic (pGL3) vector. The cloned plasmid was verified by DNA sequencing (Cosmogenetech).

Myc-TCF4-pcDNA3.0 and Myc-ΔN-TCF4E-pcDNA3.0 (dnTCF4) vectors were obtained from Eric R. Fearon of the University of Michigan26 and pMD2G and psPAX2 were a gift from Dr. KunLiang Guan,

University of California.27 β-gal reporter plasmid and pLKO.1puro-β-catenin (β-catenin shRNA) plasmids were purchased from Clontech and Addgene, respectively.

APC knockout (KO) RKO or WiDr cell lines were generated using CRISPR/Cas9 methodology28. Briefly, annealed oligonucleotides (CRISPR single guide RNA sequences; sgRNA), targeting the APC exon

15 were cloned into the lentiCRISPRv2 (Addgene). The sequences of sgRNA were as follows: sgRNA-1 5′-CACCGTCGCTCTTCATGGATTTTTA-3′; sgRNA-2 5′-AAACTAAAAATCCATGAAGAGCGAC-3′. HEK293T cells were

transfected with the lentiCRISPRv2 containing the APC sgRNA and packing vectors pMD2G and psPAX2 at a 2:2:1 ratio for viral production. Then, RKO or WiDr cell lines were transduced with the

APC-lentivirus and selected with puromycin (Sigma-Aldrich) to generate the stable cell lines. For generation of PKM2 knockdown (KD) cell lines, PKM2 shRNAs were designed by Bionics (shPKM2

#1, GCCATCTACCACTTGCAATTA; shPKM2 #2, GCCATAATCGTCCTCACCAAG). APC KO-RKO and -WiDr cells were transfected with pGPU6/hygro (shCON) or two independent pGPU6/hygro-PKM2 (shPKM2) vectors, and

then selected in a medium containing hygromycin B (1 μg/ml; Duchefa). Of those, PKM2 shRNA #1 was used for real-time PCR analysis, glucose consumption, lactate secretion assays, and mouse

xenograft assay. SW480 cells were transfected with the control shRNA or PKM2 shRNA #1 and then selected in a medium containing hygromycin B (1 μg/ml; Duchefa).

All data are expressed as the mean ± standard deviation (s.d.), and the number of samples is indicated in each figure legend. The statistical significance of differences was assessed using

the Student’s t-test or Spearman correlation analysis. Results shown are representative of at least three independent experiments. Differences reached statistical significance with *P