jx06 selectively inhibits pyruvate dehydrogenase kinase ... · corresponding authors: meiyu geng,...

Therapeutics, Targets, and Chemical Biology

JX06 Selectively Inhibits PyruvateDehydrogenase Kinase PDK1 by a CovalentCysteine ModificationWenyi Sun1, Zuoquan Xie1, Yifu Liu2, Dan Zhao3, Zhixiang Wu4, Dadong Zhang1,Hao Lv1, Shuai Tang1, Nan Jin1, Hualiang Jiang3, Minjia Tan4, Jian Ding1, Cheng Luo3,Jian Li2, Min Huang1, and Meiyu Geng1

Abstract

Pyruvate dehydrogenase kinase PDK1 is a metabolic enzymeresponsible for switching glucose metabolism from mitochon-drial oxidation to aerobic glycolysis in cancer cells, a generalhallmark of malignancy termed the Warburg effect. Herein wereport the identification of JX06 as a selective covalent inhibitor ofPDK1 in cells. JX06 forms a disulfide bondwith the thiol group ofa conserved cysteine residue (C240) based on recognition of ahydrophobic pocket adjacent to the ATP pocket of the PDK1enzyme. Our investigations of JX06 mechanism suggested that

covalent modification at C240 induced conformational changesat Arginine 286 through Van der Waals forces, thereby hinderingaccess of ATP to its binding pocket and in turn impairing PDK1enzymatic activity. Notably, cells with a higher dependency onglycolysis were more sensitive to PDK1 inhibition, reflecting ametabolic shift that promoted cellular oxidative stress and apo-ptosis. Our findings offer new mechanistic insights includinghow to therapeutically target PDK1 by covalently modifying theC240 residue. Cancer Res; 75(22); 4923–36. �2015 AACR.

IntroductionCancer cells feature a unique metabolic profile of high aerobic

glycolysis, also known as "Warburg effect," which describes thephenomenon of the enhanced conversion of glucose into lactateeven in the presence of oxygen (1). Aerobic glycolysis confers asignificant growth advantage of cancer cells by supplying essentialATP production, generating precursors for biosynthesis, and pro-viding reducing equivalents for antioxidant defense (2, 3).AlthoughWarburgeffect hasbeenobserved fordecades,howcancercells gain this unique metabolic profile remains unclear (4, 5).

Mitochondrial pyruvate dehydrogenase kinase (PDK) func-tions as a molecular switch that diminishes mitochondrial res-

piration and enhances aerobic glycolysis via phosphorylating andinactivating pyruvate dehydrogenase (PDH), a gate-keepingmito-chondrial enzyme (6, 7). PDH catalyzes the oxidative decarbox-ylation of pyruvate to acetyl-CoA, the only entry leading pyruvateinto tricarboxylic acid cycle inmitochondria (8). Phosphorylationof PDH by PDK, which occurs in the PDH complex (PDC),restricts the access of glycolytic products into the mitochondrialrespiration. To date, four PDK isoforms (PDK1-4) have beenidentified in humanmitochondriawith tissue-specific expression,all of which participate in the phosphorylation of serine 293, themain regulatory site of PDH activity (9–11).

Recently, an increasingly recognized linkage between PDK1and upstream oncogenic proteins has provided a glimpse into themolecular basis of the metabolic reprogramming in cancer cells.PDK1 is directly trans-activated by hypoxia-inducible transcrip-tion factor 1 (HIF-1) or the cooperation of dysregulated c-Myc andHIF-1 to promote glycolysis (12–14). Recent studies haverevealed that PDK1 is also regulated by oncogenic drivers at theposttranslational level. Diverse oncogenic tyrosine kinases,including FGFR1, BCR-ABL, JAK2 V617F, and FLT3-ITD, whichare frequently aberrant in human cancers, activate PDK1 activityby promoting its binding to ATP (15). Moreover, aberrant expres-sion of Lin28 facilitates aerobic glycolysis via targeting PDK1 bymicroRNA let-7 in a HIF-1–independent manner (16). All thisevidence suggests a model that oncogenic proteins rewire meta-bolic network via modulating PDK1 at multiple levels. PDK1appears to be the Achilles heel in the reprogrammed glucosemetabolism in cancer cells (17). Indeed, themetabolic alterationscaused by PDK1 inhibition in cancer cells provoke an increase inthe intracellular reactive oxygen species (ROS; ref. 18), whichfurther leads to mitochondria-dependent apoptosis (15, 19).Meanwhile, it has been also noticed that PDK1 inhibition causesdistinct responses in different cancer cells (20). It is thus crucial to

1Division of Antitumor Pharmacology, State Key Laboratory of DrugResearch, Shanghai Institute of Materia Medica, Chinese Academy ofSciences, Shanghai, China. 2Shanghai Key Laboratory of New DrugDesign, School of Pharmacy, East China University of Science andTechnology, Shanghai, China. 3Drug Discovery and Design Center,State Key Laboratory of Drug Research, Shanghai Institute of MateriaMedica, Chinese Academy of Sciences, Shanghai, China. 4The Chem-ical Proteomics Center, State Key Laboratory of Drug Research,Shanghai Institute of Materia Medica, Chinese Academy of Sciences,Shanghai, China.

Note: Supplementary data for this article are available at Cancer ResearchOnline (http://cancerres.aacrjournals.org/).

W. Sun, Z. Xie, Y. Liu, and D. Zhao contributed equally to this article.

Corresponding Authors: Meiyu Geng, Shanghai Institute of Materia Medica,Chinese Academy of Sciences, 555 Zuchongzhi Road, Shanghai 201203, China.Phone: 86-21-50806600-2426; Fax: 86-21-50807088; E-mail:[email protected]; Min Huang, [email protected]; Jian Li,[email protected]; and Cheng Luo, [email protected]

doi: 10.1158/0008-5472.CAN-15-1023

�2015 American Association for Cancer Research.

CancerResearch

www.aacrjournals.org 4923

on March 28, 2021. © 2015 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst October 19, 2015; DOI: 10.1158/0008-5472.CAN-15-1023

http://cancerres.aacrjournals.org/

know whether there exists PDK1 dependent cancer subset, whichwill facilitate the translation of PDK1 inhibition to therapeuticbenefits.

A few PDK1 inhibitors have been reported, including theATP competitive inhibitor radicicol, allosteric inhibitordichloroacetate (DCA), and AZD7545 that disrupts PDK1interaction with the PDC component (21–24). Among theseinhibitors, DCA is the only one showing preliminary clinicalefficacy in a small cohort of patients with glioblastoma (25).The therapeutic promise of PDK1 inhibitors in cancer treat-ment needs to be further explored. Recently, covalent inhibi-tors are believed to offer an alternative solution to kinaseinhibition (26). The pharmacologic advantage of covalentinhibitors, such as high selectivity across kinase family andsustained inhibitory effect, may provide new therapeuticopportunities for targeting PDK1.

In an effort to discover new PDK1 inhibitors, we carried out anenzymatic screen using a small-molecule library composed ofcommercially available known drugs (Supplementary Table S1and Supplementary Fig. S1). Thiram, a known pesticide withanticancer activity, emerged prominently due to the remarkableactivity against PDK1 (27, 28). Further chemical efforts based onthiram led to the discovery of a more potent new compounddesignated as JX06. In this study, the unique molecular mecha-nism of JX06 allowed us to probe the enzymatic regulation ofPDK1.We discovered a highly conserved cysteine, which offered atarget for covalent modulation of PDK1. Moreover, our resultsuncovered a responsive cancer subset to PDK1 inhibition, whichmight be translated to improve the clinical benefits of PDK1inhibitors.

Materials and MethodsELISA-based kinase activity assay

The 6xHis-tagged full-length coding sequences of PDK1-4and PDHA1 were expressed in E. coli and purified with Ni-NTAcolumn following the manufacturer's instructions (Qiagen).Enzymatic reaction was carried out with indicated enzyme(0.25 mg/well), substrate (0.5 mg/well) and ATP (10 mmol/L)in 100 mL buffer (50 mmol/L HEPES, 10 mmol/L MgCl2 and1 mmol/L EGTA). After reaction, the plate was washed withTween-PBS, followed by the incubation with primary antibody(p-PDHA1, S293; Abgent) and horseradish peroxidase–conju-gated secondary antibody (Calbiochem). The plate was washedand visualized with citrate buffer containing 0.1% H2O2 and2 mg/mL o-phenylenediamine and the absorbance was mea-sured at 490 nm after termination.

Cell cultureEBC-1 cells were obtained from Japanese Research Resources

Bank (Tokyo, Japan). Kelly cells were purchased from DeutscheSammlung von Mikroorganismen und Zellkulturen (DSMZ).786-O, L02, and NCI-H460 cells were obtained from the cellbank of Chinese Academy of Sciences (Shanghai, China). SMMC-7721 cells were gifted by the Second Military Medical University(Shanghai, China). HeLa cells were obtained from ShanghaiCancer Institute (Shanghai, China). SKOV-3 cells were obtainedfrom Fudan University Shanghai Cancer Center (Shanghai, Chi-na). GM00637 cell line was gifted by Dr. Yves Pommier (NCI,Bethesda, MD). Other cells used in this study were obtained fromAmerican Type Culture Collection. Cell lines were authenticated

by analyzing the DNA profile of eight short tandem repeat (STR)loci plus amelogenin (Genesky Biotechnologies Inc.). Cells weremaintained in appropriate culture medium as the supplierssuggested.

Glucose uptake measurementAfter being treated with JX06 or vehicle control for approxi-

mately 18 hours, cells were washed with PBS and cultured inserum-free high-glucosemedium for 6more hours. Glucose levelsin the medium were measured using a glucose assay kit (Beyo-time) and the readouts were normalized by the correspondingprotein amounts of each sample.

Intracellular lactate measurementIntracellular lactate levels were measured using a Lactate Col-

orimetric/Fluorometric Assay Kit (Biovision). After exposure toJX06 or vehicle control for 24 hours, cells were lysed and cen-trifuged at 12,000 �g to collect the cell supernatant. The super-natant of cell lysates was mixed with the assay solution. Theabsorbance was measured at 570 nm and the readout was nor-malized by the protein amounts.

Intracellular ATP measurementIntracellular ATP levels were assessed using an ATP assay kit

(Beyotime). After exposure to JX06 or vehicle control for 24hours, cells were lysed and centrifuged at 12,000 �g to collectthe cell supernatant. An aliquot of ATP detection workingsolution was added to a black 96-well culture plate and wasincubated for 5 minutes at room temperature. Then, the celllysate was added to the wells, and the luminescence was mea-sured immediately. The readout was normalized by the proteinamounts of each well.

Biotinylated JX06 pull-down assayCells were lysed with NP-40 lysis buffer (Beyotime). Biotiny-

lated pull-down assay was performed with biotinylated JX06 atindicated dose and 50 mL streptavidin sepharose beads (LifeTechnologies) in NP-40 lysis buffer (Beyotime). Wash buffercontained extra 1% SDS in addition to the NP-40 buffer.

Oxygen consumption rate and extracellular acidification rateanalysis

Cells were planted into XF96 cell culture plates (SeahorseBioscience). Each XF96 assay well was equipped with a dispos-able sensor cartridge and embedded with 96 pairs of fluorescentbiosensors (oxygen and pH), coupled to fiber-optic waveguides.The measurement of oxygen consumption was expressedin pmol/min and extracellular acidification rate was expressedin mpH/min.

ROS measurementCells after JX06 treatment for 24 hours were incubated with

3 mmol/L dihydroethidium (Invitrogen) in PBS for 20 minutesat 37�C. Intracellular ROS was measured by FACS analysis(FACSCalibur flow cytometer; BD Biosciences).

Mitochondrial membrane potentialThe mitochondrial membrane potential was determined by

measuring TMRM retention (red fluorescence) using IN CellAnalyzer 2000 (GE Healthcare).

Sun et al.

Cancer Res; 75(22) November 15, 2015 Cancer Research4924




Apoptosis assayApoptotic cells were measured by Annexin V and propidium

iodide (PI) dual-staining using the Annexin V-FITC ApoptosisDetection Kit (Vazyme) followed by FACS analysis.

Cell viability assayCells were seeded into a 96-well plate. After attachment, cells

were treated with JX06. CCK8 assay (Life Technologies) wascarried out after incubation for 72 hours. Untreated cells servedas the indicator of 100% cell viability.

Molecular dockingMissing residues (residue 68–71, 168–169, 204–214, and

415–416) of PDK1 (PDB entry: 2Q8F1) were complementedusing "Build Homology Models" Module of Discovery Studio2.5 (Accelrys Software Inc.). Then molecular docking was per-formed with Autodock 4.0 software. The binding site was definedas a spherewith a radius of 10Åaround the sulfur atomofCys240.AutoDock Tools (ADT) was used to prepare the grid parameterfiles (gpf) and the docking parameter files (dpf). The Lamarckiangenetic algorithm (LGA) was employed to explore the optimalchemical space in the binding site (29). Docking parameters werekept at the default values.

Statistical analysisStatistical significance was analyzed using the Student t test or

one-way ANOVA, and P < 0.05 was considered significant.

ResultsJX06 is a potent and selective inhibitor of PDK

We first examined the activity of JX06 against all PDK isoformsusing a biochemistry enzymatic assay. JX06 dose-dependentlyinhibited PDK1, PDK2, and PDK3, with a half-maximal inhibi-tory concentration (IC50) value of 0.049� 0.003 mmol/L, 0.101�0.028 mmol/L, and 0.313 � 0.011 mmol/L, respectively (Fig. 1A).The potency of JX06 was more than 100-fold higher thanDCA (Supplementary Table S2), a pan-PDK inhibitor known forits benefits in glioblastoma patients (25). But JX06 barelyshowed inhibitory activity against PDK4 at a concentration upto 10mmol/L (Supplementary Table S2), suggesting JX06 as a pan-inhibitor of PDK1, 2, and 3. Owing to the most significant role ofPDK1 among the PDK subtypes in cancer therapy, our efforts werethus particularly devoted to understand the impact of JX06 onPDK1. Inhibition of PDK1by JX06 led to the functional activationof PDC (Fig. 1B). To examine whether JX06 specifically targetedPDKs, JX06 activity was profiled in a board panel of 323 kinases,which covered most known kinases implicated in cancer malig-nancy (Supplementary Table S3). Among the tested kinases, onlyPDKs and FAK were remarkably inhibited by JX06 at the concen-trationof 10mmol/L (Fig. 1C).However, FAK inhibition couldnotbe recapitulated at the cellular level (Supplementary Fig. S2).These data suggested that JX06was a selective andpotent inhibitorof PDK.

The activity of JX06 toward PDK1 suggested its potential valuein anticancer therapy. We then proceeded to assess the impacts ofJX06 in A549 lung cancer cells. JX06 decreased PDHA1 phos-phorylation at both serine 293 and serine 232 (S293 and S232) ina time- and dose-dependent manner (Fig. 1D and E). Glucoseuptake and intracellular ATP level in A549 cells were significantlyincreased by JX06 treatment (Fig. 1F and G), suggesting the

reactivation of mitochondrial respiration resulted from PDKinhibition. In the meanwhile, aerobic glycolysis determined bythe lactate production was significantly diminished by JX06treatment (Fig. 1H), which indicated a metabolic switch fromaerobic glycolysis to oxidative phosphorylation (19, 30). Further,JX06 dose-dependently suppressed the growth of A549 cells whileoverexpression of PDK1 rescued cells from JX06-caused prolifer-ative inhibition (Fig. 1I). Consistently, depletion of PDK1 com-promised cell sensitivity toward JX06 (Fig. 1J), suggesting thatPDK1 inhibition largely accounted for the growth impendence ofcancer cells caused by JX06.

These results together demonstrated that JX06 was a selectivePDK inhibitor. JX06 treatment led to metabolic alterations thatredirected glucose flux from aerobic glycolysis to mitochondrialoxidation. PDK1 inhibition largely accounted for the cancer cellproliferative suppression caused by JX06.

Cancer cells with high dependency on glycolysis are moresensitive to PDK1 inhibition

PDK1 inhibition is known to provoke an increase in intracel-lular ROS (18), which may result from decreased generation ofreducing substances, mainly NAPDH, for ROS detoxification(31). Mechanistically, ROS generation and consequent mito-chondrial membrane potential (MMP) reduction are believed toaccount for PDK1 inhibition caused apoptosis of cancer cells (32).In an effort to examine the cellular impacts of JX06 in humancancer cells, we unexpectedly discovered that cancer cell linesexhibited distinct outcomes in ROS generation caused by JX06,despite similarly inhibited PDK1 signaling (Fig. 2A and Supple-mentary Fig. S3). Moreover, the variation in the amount of ROSgeneration was closely associated with MMP reduction and apo-ptotic occurrence (Fig. 2B and C). For example, A549 and EBC-1cells, which were more responsive than the other two cell lines interms of ROS generation after exposure to JX06, showed a dra-matic decrease in MMP and an increase in the ratio of apoptoticcells. In contrast, JX06 treatment–triggered responses of ROSgeneration, MMP alteration, and cell apoptosis in H460 andHT29 cells were either marginal or undetectable (Fig. 2A–C).

Our results above suggested the existence of a responsive subsetto PDK1 inhibition among different cancer cell lines, which wasintriguing and barely explored yet. We first examined whether thevariable PDK1 abundance could explain the diverse sensitivity ofcancer cells to JX06, however, the apparent association betweenPDK1 protein level and the JX06 sensitivity was not observed(Supplementary Fig. S4). We then speculated that basal glucosemetabolic capacity of cancer cells might play a key role indetermining the outcomes of PDK1 inhibition in cancer cells. Totest this hypothesis, we used seahorse XF 96 Analyzer to measurethe two major glucose metabolic pathways of 20 cell lines,including various cancer and normal cells. The extracellularacidification rate (ECAR) indicated the glycolytic capacity, where-as the oxygen consumption rate (OCR) reflected the oxidativephosphorylation level of the tested cells (30, 33). The ratios ofECAR to OCR (ECAR/OCR), which indicated the relative relianceof the cell lines on glycolysis or mitochondria respiration, werecompared among the cell lines. As expected, the three normal celllines (L02, WI-38, and GM00637) all exhibited low ECAR/OCRratios, which was in agreement with the current knowledge thatmitochondrial respiration plays a dominant role in glucosemetabolismof normal cells.Meanwhile, we observed an apparentvariation in ECAR/OCR ratios among these cancer cells. Cells like

Selective and Covalent Inhibition of PDK1

www.aacrjournals.org Cancer Res; 75(22) November 15, 2015 4925




A2058, BXPC-3, and A549 exhibited high ECAR/OCR ratios,suggesting a dominant reliance on glycolysis (Fig. 2D). In con-trast, HeLa and HT-29 cells harbored low ECAR/OCR ratios,

suggesting they were more dependent onmitochondrial function(Fig. 2D).We then subgrouped these cancer cells at a cutoff ECAR/OCR value of 0.5 and examined their sensitivity to JX06. Indeed,

Figure 1.JX06specifically inhibitsPDK1andcausesmetabolicswitch fromaerobicglycolysis tooxidativephosphorylation incancercells.A,enzymaticactivityagainstPDK1,2,3.B,PDCreactivation. C, selectivity against a panel of kinases. D and E, time-, and dose-dependent inhibition of PDHA1 phosphorylation in A549 cells. Cells were treated withJX06 at 10 mmol/L for indicated durations (D) or indicated concentrations (E) for 24 hours. PDHA1 phosphorylation was detected using immunoblotting. F–H,metabolic alterations in glucose uptake (F), intracellular ATP amount (G), and extracellular lactate production (H). I, ectopic expression of PDK1 rescued JX06 inhibited cellgrowth. J, depletion of PDK1 compromised cell sensitivity toward JX06. All the resultswere normalized to the protein amounts. Mean� SE (n¼ 3); � , P < 0.05; �� , P < 0.01..

Sun et al.





cellswithhigher ECAR/OCR ratio tended tobemore responsive toJX06 treatment. Normal cells and cancer cells with higher capacityin mitochondrial respiration exhibited much less sensitivity to

JX06 (Fig. 2E). The variable dependency of these cells on PDK1activity was also confirmed by PDK1 depletion using PDK siRNA(Fig. 2F and Supplementary Fig. S5).

Figure 2.JX06 induces ROS generation and cell apoptosis in cancer cells with high ECAR/OCR. A, mitochondrial ROS generation. Cancer cells were treated with JX06 at10 mmol/L for 24 hours. B, mitochondrial membrane potential. Cells were treated with JX06 at 10 mmol/L for 24 hours. C, cell apoptosis. Cells were treatedwith JX06 at 10 mmol/L for 48 hours. D, ratio of ECAR andOCR in a panel of cancer and normal cells. E, cell sensitivity toward JX06. Cell viability wasmeasured usingthe CCK-8 assay. F, cell survival after depletion of PDK1. Mean � SE (n ¼ 3); �� , P < 0.01.






Our results above revealed an intriguing phenomenon thatcancer cells responded differently to PDK1 inhibition, and cellswith higher dependency on aerobic glycolysis, which could bemeasured by ECAR/OCR ratio, might be the responsive cancersubset to PDK1 inhibition.

JX06 attenuates tumor growth in A549 xenograft modelsTo further prove the therapeutic potential of JX06 via PDK1

inhibition in the responsive cancer subset, we tested its antitumorefficacy in A549 subcutaneous xenograft mice models. Tumor-bearing mice were randomly divided into three groups andintraperitoneally received vehicle control or JX06 at 40 and80 mg/kg per day, respectively. A 21-day continuous treatmentof 80 mg/kg JX06 considerably reduced 67.5% tumor volumecompared with the vehicle control (Fig. 3A). Endpoint tumorweights in the JX06 treated groupwere significantly less than thosetreated with vehicle control (Fig. 3B). In the meanwhile, we did

not observe the body weight loss in JX06-treated mice, suggestingthat JX06 was well tolerated at the administration dose (Fig. 3C).In agreement with the marked antitumor efficacy of JX06, theintratumoral PDK1 signaling was significantly inhibited, asshown by the blockage of the PDHA1 phosphorylation as wellas the decrease of plasma lactate level in JX06-treated mice (Fig.3D and E).

Meanwhile, JX06 showed no significant antitumor activity inHT-29 models despite of similar inhibition of PDK1 pathway,further validating our above observations in cell lines (Fig. 3F andG). These results also largely excluded the involvement of otherpotential targets in the mediating the anticancer efficacy of JX06.

Structure–activity relationship analysis identifies thepharmacophore of JX06

The remarkable selectivity and potency of JX06 against PDK1inspired us to further explore themolecular mechanisms of PDK1

Figure 3.JX06 inhibits tumor growth in vivo. A–E, A549 xenograft mouse models were treated with JX06 or vehicle control daily at indicated dosages for 21 consecutivedays. A, tumor volume change. B, tumor weight after last dosing. C, body weight change. D, intratumoral PDHA1 phosphorylation. E, serum lactate level.F and G, HT-29 xenograft mouse models were treated with JX06 or vehicle control daily at indicated dosages for 21 consecutive days. F, tumor volume change.G, intratumoral PDHA1 phosphorylation. One-way ANOVA analysis, �� , P < 0.01.

Sun et al.





inhibition by JX06. To this end, we employed the structure–activity relationship (SAR) analysis to identify the key moietiesof JX06 that might critically interact with PDK1. There are threecommon moieties included in the chemical structures of boththiramand JX06, namely a central disulfidebond, two thioamidesand symmetrical terminal N-methyl substituents. These moietieswere individually removed or modified to examine their impactson the activity against PDK1 (Fig. 4A). Compounds JX02, 03, 04,and 12, in which the disulfide bond was replaced, completely lostthe activity against PDK1, suggesting that the disulfide bond wasessential for PDK1 inhibitory activity. Likewise, replacing sym-metrical thioamides with amide (JX09, JX10) or ketones (JX11)was not beneficial aswell. However, it appeared that retaining onethioamide was sufficient to sustain most of PDK1 inhibitory

activity. In contrast, the replacement of the N-substituents withother alkyls, such as -C2H5 (JX05), - (CH2)4- (JX07), -(CH2)5-(JX08), and -(CH2)2O(CH2)2- (JX06), showed only a marginaleffect on the potency of the inhibitor, suggesting that the bindingpocket of N-substituents was more accommodative.

The SAR data aforementioned suggested a mechanism thatinvolved a disulfide bond and an adjacent thioamide in criticallyinteracting with PDK1. The susceptibility of this moiety to reduc-tion lightened us to propose a possibility of its covalent interac-tionwith the thiol side chain of cysteine residuals in PDK1. Likely,the thiol of cysteine attacks the chemically reactive thioamide andforms a new disulfide bond with JX compounds, resulting in thebreakage of the disulfide bond of JX06 (Fig. 4B). To prove thishypothesis, glutathione (GSH) that contains a reducing thiol

Figure 4.SAR analysis identifies pharmacophore of JX06. A, SAR analysis identified the key pharmacophore of JX06, i.e., a disulfide bonds (red) and two thioamides(magenta). B, proposed chemistry reaction between JX06 and sulfhydryl. C, inhibition curve of JX06 toward PDK1 under different concentrations of GSH.






group was introduced to interrupt the disulfide bonds possiblyformed between PDK1 and JX06. As expected, the inhibitoryactivity of JX06 was dramatically reduced with the augment ofGSH concentration in a dose-dependent manner, while in con-trast the efficacy of DCA was not affected (Fig. 4C).

JX06 inhibits PDK1 by covalently binding to a conservedcysteine residue

To further understand the binding and inactivating steps ofPDK1 inhibition by JX06, we examined the time-dependentinhibition of PDK1 by JX06. The enzymatic assay of PDK1 wasinitiated in the presence of JX06 at 50 nmol/L following30-minute preincubation. The enzymatic activity was measuredover time up to 60 minutes, and the noncovalent irreversibleinhibitor DCA was used as a control. The results suggested thatdistinct from DCA, preincubation of JX06 apparently influencedthe time-dependent inhibition curve of JX06, supporting thedisulfide bond formation between JX06 and PDK1 (Fig. 5A).

For further confirmation, we generated a biotin-conjugatedJX06 to test whether JX06 was able to covalently bind to endog-enous PDK1 derived from cancer cells. Biotin-conjugated JX06,which retained the activity against PDK1 (Supplementary Fig. S6),was preincubated with A549 cell lysates, and JX06-associatedproteins were enriched by affinity purification with streptavidinbeads. Afterwards, proteins possibly associated via noncovalentbonds were removed by sufficient wash using wash buffer con-taining 1% SDS. The presence of PDK1 in the resultant complexwas examined by immunoblotting. Indeed, PDK1 protein wasdetected to be binding to biotin-conjugated JX06 in a dose-dependentmanner (Fig. 5B), and this bindingwas largely reducedwhen biotin-conjugated JX06was premixedwith 30-fold of label-free JX06 during incubation with cell lysates (Fig. 5C). In contrast,the addition of label-free JX06 after a 12-hour incubation barelyaffected the binding between PDK1 and biotin-conjugated JX06(Fig. 5D), suggesting that regardless of the sufficient amount,label-free JX06 failed to interrupt the bond of biotin-labeled JX06and PDK1 once the chemical reaction was completed. These datatogether strongly suggested the covalent association between JX06and PDK1.

We then proceeded to identify the cysteine residues involved inthe formationof the covalent bondwith JX06. All the four cysteineresidues of PDK1 were individually mutated according to theprediction of Sorting Intolerant From Tolerant (SIFT) to retainPDK1 activity maximally. Four PDK1 mutants, namely C71S,C223T, C240L, and C421S, were generated. Enzyme amounttitration assay showed that C71S, C223T, and C421S maintainedtheir enzymatic activity as thewild-type PDK1. TheC240Lmutantlargely lost its catalytic capacity in an enzymatic assay (Fig. 5E),and ectopic expression of C240L mutant failed to rescue thedefects in PDHA1 phosphorylation or cell growth of PDK1depleted cells (Supplementary Fig. S7). These data showed thatC240 was crucial for the catalytic function of PDK1. We alsonoticed that C240 in PDK1 was highly conserved across evolu-tion, from Arabidopsis to human, supporting its critical role insustaining the catalytic function of PDK1 (Fig. 5F). We thendetermined the inhibitory activity of JX06 toward these fourmutants. JX06 largely lost its activity toward C240L but not toother three mutants (Fig. 5G), whereas all these mutantsresponded similarly to DCA inhibition. These data suggested thatJX06 inhibited PDK1 activity via interacting with the C240 res-idue. For further validation, the mixture of recombinant PDK1

protein and JX06 was submitted for mass spectrometry analysis.The results showed that C240 residue in PDK1 was modified andthemolecular weight shiftmatchedwith JX06 (Fig. 5H). Althoughwe could not rule out the possibility that similar modificationsoccurred on other cysteine, it largely appeared that covalentinteraction with C240 fundamentally accounted for the enzymat-ic inhibition by JX06.

Together, our data suggested that JX06 covalently bound to ahighly conserved cysteine residue, C240 in PDK1, which gave riseto strong inhibition of the enzymatic activity.

JX06 recognizes a hydrophobic pocket in PDK1The specificity of covalent inhibitors is known to stem from the

recognition of a binding pocket, which restricts its specific acces-sibility to the targeted residues to undergo a follow-up chemicalreaction. We next studied whether there existed a binding pocketin the surroundings of the targeted cysteine that allowed thespecific modification by JX06.

To address this question, we took the advantage of the previ-ously solved crystal structures of PDK1. Molecular dockingapproachwas used to analyze the bindingmode of JX06 to PDK1.JX06 was docked into PDK1 (PDB entry: 2Q8F1) with Autodock4.0 (34) and a spherewith a radius of 10 Å around the sulfur atomof C240 was defined as the binding site. JX06 suited quite well inthe binding pocket near C240 and formed hydrophobic interac-tions with the surrounding residues (Fig. 6A and B). Remarkably,the sulfur atom of JX06 was located within 3.8 Å distance of thesulfur atom of C240, offering a possibility to form a disulfidebond between them. To test the dependency of JX06 on thisbinding pocket, we designed a new JX06 derivative, named JX86,which contained the intact key moieties but was unable to fit intothis binding pocket. As expected, the inhibitory activity of JX86toward PDK1 was significantly reduced compared with JX06 (Fig.6C). Also, we introduced point mutations to the key surroundingresidues that were identified to interact with JX06 in the bindingpocket, aiming at interfering JX06's docking by changing the spaceof the pocket or disrupting the electrostatic interactions. Accord-ing to the sidechain conformations, M289F mutant was designedto reduce the space of the pocket, Y243A and Y289A mutants togive rise to a bigger pocket, and E290K mutant to impact theelectrostatic interactions. Indeed, all these point mutations moreor less affected the potency of JX06 toward PDK1 and dualmutations combining Y243A and M289A significantly reducedthe potency of JX06 (Fig. 6D), although thesemutations per se didnot affect the catalytic activity of PDK1 (Fig. 6E). These datademonstrated that the inhibitory effect of JX06 on PDK1 requiredthe specific recognition of the unique binding pocket.

Importantly, the identification of this pocket provided answersto a previous puzzle that JX06 was lacking the inhibitory capacityagainst PDK4 (Supplementary Table S2). PDK4 similarly pos-sessed a conserved cysteine C215, which was essentially requiredfor the enzymatic activity of PDK4 as well (Fig. 6F). Molecularmodeling of JX06 toPDK4 indicated that the sulfur atomsofC215and JX06 were in a distance of 8.8Å, which was beyond the rangepossibly to form a chemical bond (Fig. 6G and H). These dataexplained the selectivity of JX06 between PDK1 and PDK4.

Covalent modification by JX06 reduces ATP affinity of PDK1Our data demonstrated the essential role of C240 in modulat-

ing the enzymatic activity of PDK1. However, it remained unclearhow this residue was critically involved in the catalytic function of

Sun et al.





Figure 5.JX06 inhibits PDK1 activity via covalently binding to a cysteine residue in an irreversible manner. A, covalent inhibition of PDK1 by JX06. B–D, PDK1binding to biotin-conjugated JX06. PDK1 was copurified with biotin-conjugated JX06 in A549 cell lysate after incubation in the absence (B) or presence (C)of 30 folds of free JX06. D, coincubation with 30 folds of free JX06 was conducted after binding reaction. E, enzyme titration of indicated PDK1mutants. F, alignment of PDK1 C240 across diverse species. G, IC50 of JX06 and DCA against mutated PDK1. H, mass spectrometry identified PDK1C240 modification by JX06. Mean � SE (n ¼ 3).






PDK1. PDK1 has been known to interchange between two con-formational states (24, 35–37). The closed conformation, wherethe C-terminal tail of each PDK1 subunit loses its interaction withthe lipoyl-bearing domain (L2), has the active-site cleft closedinside. It exhibits a high binding affinity to ADP andhence stays inan inactive state (38). Upon the binding of L2 to PDK1, the C-terminal tails form a crossover configuration and thus the active-site cleft is widened, which leads to an open conformation. Theopen conformation preferentially binds to ATP rather than ADPand activates its catalytic function. Comparing the closed and theopen conformations might provide important clues about howC240 was involved in the regulation of PDK1 activity. We hence

compared the structures of open conformation of PDK1 (PDBentry: 2Q8F) and closed conformation of PDK2 (PDB entry:1JM6; Fig. 7A; refs. 24, 38). It was observed that the sidechainorientations around C240 barely differed in these two conforma-tions, suggesting that binding of JX06 with C240 unlikely inter-feredwith the transformation between the open and closed states.

We next attempted to seek for other possibilities. Notably,N283/R286 in PDK1 differed dramatically in the sidechain ori-entation between the open and closed conformations (Fig. 7B andC). Furthermore, we ran a 200-nanosecond molecular dynamicsimulation of the open conformation of PDK1 and found out thatthe sidechain of R286 swung flexibly (Fig. 7D and Supplementary

Figure 6.The binding pocket of JX06 in PDK1. A, the binding pocket of JX06 (white sticks) in PDK1 (green surface). C240 is shown as green sticks. B, detailedinteractions between JX06 (white sticks) and its surrounding residues in PDK1 (green sticks). PDK1 is shown as light green cartoon. C, superposed structures of JX86(orange sticks) with JX06 (white sticks). D, JX06 activity against PDK1 mutants. E, enzymatic activity of PDK1 mutants. F, enzyme titration of PDK4 C215L.G, molecular model of the PDK4–JX06 complex. The binding pocket of JX06 (white sticks) in PDK4 (green surface). C215 is shown as green sticks. H, detailedinteractions between JX06 (white sticks) and its surrounding residues in PDK4 (green sticks).

Sun et al.





Figure 7.JX06 binding reduces ATP affinity of PDK1. A, superposed structures of residues around C240 in PDK1 open conformation and PDK2 closed conformation. Residuesaround C240 in PDK1 are shown as green sticks and labeled in green letters. Residues around C204 in PDK2 are shown as magenta sticks. The residues shown assticks are R237/K201, R238/L202, I239/L203, C240/C204, D241/D205, L242/K206, Y243/Y207, Y244/Y208, I245/M209, N246/A210, S247/S211, P248/P212,M285/M249, and M289/V253 in PDK1/PDK2. PDK1 and PDK2 are shown as light green and magenta cartoons, respectively. B, the electrostatic surfaces of PDK1 (blue,positive charge; red, negative charge). JX06 and ATP are shown as sticks. Two residues (N283 and R286), which are likely involved in ATP binding are shown asgreen sticks. C, superposed structures of PDK1 open conformation andPDK2closed conformation. C240/N283/R286 of PDK1, C204/N247/R250of PDK2, JX06 andATPare shown as sticks. PDK1 and PDK2 are shown as light green and magenta cartoons, respectively. D, structures extracted from the trajectory of molecular dynamicsimulation. PDK1 at 0 nanoseconds is shown as green cartoon. C240, N283, and R286 at 0 nanoseconds are shown as green sticks and the ones extracted frommolecular dynamic simulation are shown as magenta sticks. JX06 and ATP binding sites are shown as dotted ovals. E, PDK1 R286A and R286I enzymatic activity indifferent amounts of ATP. F, JX06 activity against PDK1 R286A and R286I mutants. G, a scheme showing proposed mechanism of JX06 in PDK1 inhibition.






Movie S1; ref. 24). In addition, we noticed that R286 was locatedadjacent to both JX06-binding pocket and ATP-binding pocket(Fig. 7C). We hence speculated that conformational changesdisturbed by JX06 might be transferred by R286 to the ATP-binding site, thus affecting the PDK1 activity. To test this possi-bility, we generated PDK1 R286A and R286I mutants andobserved that both mutants exhibited compromised enzymaticactivity (Fig. 7E). In agreement with our speculation, the increaseof ATP amount was able to rescue the enzymatic activity of thesetwo mutants, suggesting that R286 might play a critical role inmodulating affinity between PDK1 and ATP (Fig. 7E). Moreimportantly, JX06 greatly lost its inhibitory activity towardR286 substitutionmutants (R286A, R286I), whereas the allostericinhibitor DCA exhibited similar activity against the mutant andwild-type PDK1 (Fig. 7F). These data together supported ourhypothesis that R286 residue played an irreplaceable role inmediating the inhibitory activity of JX06, and the conformationalchanges of R286 caused by JX06 impaired PDK1 enzymaticactivity via reducing the ATP affinity.

These data together provided important evidence to under-stand themolecular basis of JX06. Covalent modification by JX06on PDK1 C240 induced confirmation changes of R286 throughVan der Waals forces, which was further transported to the ATP-binding site and thus affected the binding affinity between ATPand PDK1 (Fig. 7G).

DiscussionPDK1 is one of the most extensively investigated anticancer

targets in the aerobic glycolysis pathway, although most inhi-bitors are still at early stage preclinical research. The discoveryof these PDK1 inhibitors has benefited from the mechanisticinsights into the enzymatic regulation of PDK1. Phosphoryla-tion of PDHA1 by PDK1 occurs in the PDC, where PDK1 isrecruited to the complex through binding to E2 subunit (8, 39).Disrupting the interaction between PDK1 and E2 via binding tothe lipoyl-binding pocket in the PDK1 N-terminal has given riseto the discovery of AZD7545 (23, 40). Despite its remarkableglucose-lowering effect, the anticancer efficacy of AZD7545seems obscure (23). Like other kinases, binding to ATP-bindingpocket of PDK1, also known as GHKL (gyrase, Hsp90, histidinekinase, and MutL) domain, to block ATP entry also results inPDK1 inhibition (41). However, the fact that GHKL domainwidely exists in several other kinases compromised the speci-ficity of ATP-completive PDK1 inhibitors such as radicicol(38, 42). DCA is the most advanced PDK1 inhibitor that hasshown preliminary benefits in glioblastoma patients (25).Mechanistically, DCA is a pyruvate analog that promotes con-formational changes at the active-site cleft of PDK1 and hindersthe dissociation of ADP from the active site (24, 43). However,its potency in PDK1 inhibition is relatively low. The IC50 valueof DCA for enzymatic inhibition is at millimole level (19, 22).Thus far, explorations into all these known mechanisms appearnot very successful.

Lately, covalent inhibitors have gained increasing attentionfor their pharmacologic advantages (44). Covalent inhibitorshave been demonstrated to hold the promise in overcoming thechallenges of potency, selectivity, efficacy, and resistance facedby noncovalent inhibitors (45–47). Currently, structure-basedapproaches are employed to design small molecules that targetkinases through covalent attachment to a specific cysteine.

Theoretically, the cysteine should be essential for the enzymaticactivity of the targeted protein. In the meanwhile, surroundingsfavorable for the inhibitors landing are required to allow thefollow-up chemical reaction and confer the selectivity of theinhibitor (48). Lack of this knowledge has largely restricted thediscovery of the covalent inhibitors for most kinases, which isalso the case for PDK1. In this study, we discovered a previouslyunappreciated cysteine, C240 in PDK1, which was essential forits enzymatic activity. Covalent modification on this residueinduced conformational changes of arginine 286 through Vander Waals forces, which likely interfered the access of ATP toPDK1 and in turn impaired the kinase activity of PDK1. To ourknowledge, this study provided the first evidence suggesting theessential role of C240 in PDK1. We have also noticed that aprevious study reported that oxidation of cysteine residues 45and 392 of PDK2 by hydrogen peroxide resulted in its catalyticsuppression, although the molecular mechanism remainedunknown (49). The counterpart residues of these two cysteinesin PDK1 are cysteine 71 and 421. Our data have shown thatPDK1 C71S and C421S mutants retained their enzymaticactivity, largely excluding their critical involvement in deter-mining the catalytic activity of PDK1 (Fig. 5E). Also, it would beinteresting to investigate whether cysteine 240 in PDK1 can beposttranslationally modified to fine tune the enzymatic regu-lation of PDK1. Moreover, we identified a unique hydrophobicpocket, which provided a venue for small molecules to access toC240 within a distance allowing chemical modification. Thisinformation could contribute to the rational design of covalentinhibitors of PDK1. Also, we proved that the first covalentPDK1 inhibitor JX06 elicits potent inhibition to PDK1, whichcould be translated to metabolic alterations and optimal anti-tumor efficacy.

A very recent study has shown that breast cancer T-47D andMDA-MB-231 cells exhibited distinct metabolic responses toDCA inhibition, suggesting the potential discrepancy in anti-cancer efficacy of PDK1 inhibition among different cancer cells(20). Our study extended this observation by proving thevariations in the antitumor efficacy of PDK1 inhibition in abroad panel of cancer cells, and identifying a subset cancer cellsmore responsive to PDK1 inhibition. These results intrigued usto investigate the molecular indicators that might possiblyallow us to identify the responsive subset. Indeed, the ratiosof ECAR to OCR were closely related to the sensitivities to PDK1inhibition. According to our results, equivalent PDK inhibitioncaused variable responses in ROS generation and MMP levelchange in cancer cells. In the subset cells with less reliance onglycolysis, such as H460 cells, the MMP level posttreatmentintended to remain at a hyperpolarizing state (SupplementaryFig. S8), which explained the resistance to the apoptosis induc-tion in these cells (19, 25). Encouragingly, technical campaignaiming at directly measuring the ECAR/OCR ratio of freshtumor tissue is undergoing at the moment. And its practicecan be expected in the very near future. This technical break-through will make it possible to test our observations in clinicalsettings.

In summary, our findings might contribute to the field in threeaspects: first, we gained new insights into the internal enzymaticregulation of PDK1, which involved a previously unappreciatedcysteine; second, we identified the first PDK1 covalent inhibitorand proved the possibility of the rationale design of PDK1covalent inhibitors; third, we showed that the intrinsic metabolic

Sun et al.





status of cancer cells varied among the cells and determined theirresponsiveness to PDK1 inhibition. ECAR/OCR ratio reflectedaddiction to aerobic glycolysis and might be used to select cancerpatients receiving metabolic-modulating drugs.

Disclosure of Potential Conflicts of InterestNo potential conflicts of interest were disclosed.

Authors' ContributionsConception and design: Z. Xie, D. Zhao, J. Li, M. Huang, M. GengDevelopment of methodology: W. Sun, Z. Xie, Y. Liu, D. Zhao, H. LvAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): W. Sun, Z. Xie, D. Zhao, Z. Wu, N. Jin, M. Tan, J. LiAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis): W. Sun, Z. Xie, Y. Liu, D. Zhao, Z. Wu, H. Jiang,M. Tan, C. Luo, J. LiWriting, review, and/or revision of the manuscript: W. Sun, Z. Xie, D. Zhao,Z. Wu, C. Luo, J. Li, M. Huang, M. Geng

Administrative, technical, or material support (i.e., reporting or organizingdata, constructing databases): D. Zhao, D. Zhang, S. TangStudy supervision: J. Ding, J. Li, M. Huang, M. Geng

Grant SupportThis work was supported by the National Program on Key Basic Research

Project of China (No.2012CB910704 to M. Geng), grants from NationalNatural Science Foundation of China (No. 81222049 to M. Huang, No.21222211 to J. Li, and No. 81202549 to Z. Xie), and the Natural ScienceFoundationofChina for InnovationResearchGroup (No. 81321092 to J.Ding).The PDK enzymatic assaywas establishedwith the financial support by ActelionPharmaceuticals Ltd.

The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby markedadvertisement in accordance with 18 U.S.C. Section 1734 solely to indicatethis fact.

Received April 16, 2015; revised August 2, 2015; accepted August 4, 2015;published OnlineFirst October 19, 2015.

References1. Warburg O. Origin of cancer cells. Science 1956;123:309–14.2. Vander Heiden MG, Cantley LC, Thompson CB. Understanding the War-

burg effect: the metabolic requirements of cell proliferation. Science2009;324:1029–33.

3. Mankoff DA, Eary JF, Link JM, Muzi M, Rajendran JG, Spence AM, et al.Tumor-specific positron emission tomography Imaging in patients: F-18fluorodeoxyglucose and beyond. Clin Cancer Res 2007;13:3460–9.

4. Vander Heiden MG. Targeting cancer metabolism: a therapeutic windowopens. Nat Rev Drug Discov 2011;10:671–84.

5. Sanchez-Arago M, Chamorro M, Cuezva JM. Selection of cancer cells withrepressed mitochondria triggers colon cancer progression. Carcinogenesis2010;31:567–76.

6. Harris RA, Huang BL, Wu PF. Control of pyruvate dehydrogenase kinasegene expression. In:Weber G, editor. Advances in enzyme regulation.Oxford: Pergamon Press 2001. p 269–88.

7. Holness MJ, Sugden MC. Regulation of pyruvate dehydrogenase complexactivityby reversible phosphorylation.BiochemSocTrans2003;31:1143–51.

8. ZhouZH,McCarthyDB,O'ConnorCM,Reed LJ, Stoops JK. The remarkablestructural and functional organization of the eukaryotic pyruvate dehy-drogenase complexes. Proc Natl Acad Sci U S A 2001;98:14802–7.

9. Popov KM, Hawes JW, Harris RA. Mitochondrial alpha-ketoacid dehydro-genase kinases - A new family of protein kinases. In: Corbin JD, Francis SH,editors. Advances in SecondMessenger andPhosphoproteinResearchNewYork: Raven Press; 1997. p. 105–11.

10. Gudi R, Bowkerkinley MM, Kedishvili NY, Zhao Y, Popov KM. Diversity ofthe pyruvate-dehydrogenase kinase gene family in humans. J Biol Chem1995;270:28989–94.

11. Bajotto G, Murakami T, Nagasaki M, Qin B, Matsuo Y, Maeda K, et al.Increased expression of hepatic pyruvate dehydrogenase kinases 2 and 4 inyoung and middle-aged Otsuka Long-Evans Tokushima Fatty rats: induc-tion by elevated levels of free fatty acids. Metabolism 2006;55:317–23.

12. Kim J-W,GaoP, Liu Y-C, SemenzaGL,DangCV.Hypoxia-inducible factor Iand dysregulated c-myc cooperatively induce vascular endothelial growthfactor and metabolic switches hexokinase 2 and pyruvate dehydrogenasekinase 1. Mol Cell Biol 2007;27:7381–93.

13. Kim JW, Tchernyshyov I, Semenza GL, Dang CV. Hif-1-mediated expres-sion of pyruvate dehydrogenase kinase: a metabolic switch required forcellular adaptation to hypoxia. Cell Metab 2006;3:177–85.

14. Papandreou I, Cairns RA, Fontana L, Lim AL, Denko NC. Hif-1 mediatesadaptation to hypoxia by actively downregulating mitochondrial oxygenconsumption. Cell Metab 2006;3:187–97.

15. Hitosugi T, Fan J, Chung T-W, Lythgoe K, Wang X, Xie J, et al. Tyrosinephosphorylation of mitochondrial pyruvate dehydrogenase kinase 1 isimportant for cancer metabolism. Mol Cell 2011;44:864–77.

16. Ma X, Li C, Sun L, Huang D, Li T, He X, et al. Lin28/let-7 axis regulatesaerobic glycolysis and cancer progression via PDK1. Nat Commun 2014;5:5212.

17. Kroemer G, Pouyssegur J. Tumor cell metabolism: Cancer's Achilles' heel.Cancer Cell 2008;13:472–82.

18. Sutendra G, Michelakis ED. Pyruvate dehydrogenase kinase as a noveltherapeutic target in oncology. Front Oncol 2013;3:38.

19. Bonnet S, Archer SL, Allalunis-Turner J, Haromy A, Beaulieu C, ThompsonR, et al. A mitochondria-Kþ channel axis is suppressed in cancer and itsnormalizationpromotes apoptosis and inhibits cancer growth. Cancer Cell2007;11:37–51.

20. Gang BP, Dilda PJ, Hogg PJ, Blackburn AC. Targeting of two aspectsof metabolism in breast cancer treatment. Cancer Biol Ther 2014;15:1533–41.

21. Besant PG, Lasker MV, Bui CD, Turck CW. Inhibition of branched-chainalpha-keto acid dehydrogenase kinase and Sln1 yeast histidine kinase bythe antifungal antibiotic radicicol. Mol Pharmacol 2002;62:289–96.

22. Bersin RM, Stacpoole PW. Dichloroacetate as metabolic therapy for myo-cardial ischemia and failure. Am Heart J 1997;134:841–55.

23. Mayers RM, Butlin RJ, Kilgour E, Leighton B, Martin D, Myatt J, et al.AZD7545, a novel inhibitor of pyruvate dehydrogenase kinase 2 (PDHK2),activates pyruvate dehydrogenase in vivo and improves blood glucosecontrol in obese (fa/fa) Zucker rats. Biochem Soc Trans 2003;31:1165–7.

24. Kato M, Li J, Chuang JL, Chuang DT. Distinct structural mechanisms forinhibition of pyruvate dehydrogenase kinase isoforms by AZD7545,dichloroacetate, and radicicol. Structure 2007;15:992–1004.

25. Michelakis ED, Sutendra G, Dromparis P, Webster L, Haromy A, Niven E,et al. Metabolic modulation of glioblastoma with dichloroacetate. SciTransl Med 2010;2:31ra34.

26. Lannine BR, Whitby LR, Dix MM, Douhan J, Gilbert AM, Hett EC, et al. Aroad map to evaluate the proteome-wide selectivity of covalent kinaseinhibitors. Nat Chem Biol 2014;10:760–7.

27. Larson RH, Walker JC. Thiram for smut control in onion set plantings.Phytopathology 1953;43:596–7.

28. MarikovskyM. Thiram inhibits angiogenesis and slows the development ofexperimental tumours in mice. Br J Cancer 2002;86:779–87.

29. Morris GM, Goodsell DS, Halliday RS, Huey R, Hart WE, Belew RK, et al.Automated docking using a Lamarckian genetic algorithm and an empir-ical binding free energy function. J Comput Chem 1998;19:1639–62.

30. Wang Y-H, Israelsen WJ, Lee D, Yu VWC, Jeanson NT, Clish CB, et al. Cell-state-specific metabolic dependency in hematopoiesis and leukemogene-sis. Cell 2014;158:1309–23.

31. Singh R, Lemire J,Mailloux RJ, AppannaVD. Anovel strategy involved anti-oxidative defense: the conversion of NADH into NADPH by a metabolicnetwork. PLoS ONE 2008;3:e2682.

32. Fulda S, Galluzzi L, Kroemer G. Targetingmitochondria for cancer therapy.Nat Rev Drug Discov 2010;9:447–64.

33. Zhou Y, Tozzi F, Chen J, Fan F, Xia L, Wang J, et al. Intracellular ATP levelsare a pivotal determinant of chemoresistance in colon cancer cells. CancerRes 2012;72:304–14.






34. MorrisGM,HueyR, LindstromW, SannerMF, BelewRK,Goodsell DS, et al.AutoDock4 andAutoDockTools4: automateddockingwith selective recep-tor flexibility. J Comput Chem 2009;30:2785–91.

35. Green T, Grigorian A, Klyuyeva A, Tuganova A, Luo M, Popov KM.Structural and functional insights into the molecular mechanisms respon-sible for the regulation of pyruvate dehydrogenase kinase 2. J Biol Chem2008;283:15789–98.

36. Kato M, Chuang JL, Tso SC, Wynn RM, Chuang DT. Crystal structure ofpyruvate dehydrogenase kinase 3 bound to lipoyl domain 2 of humanpyruvate dehydrogenase complex. EMBO J 2005;24:1763–74.

37. Wynn RM, Kato M, ChuangO JL, Tso S-C, Li J, Chuang DT. Pyruvatedehydrogenase kinase-4 structures reveal a metastable open conformationfostering robust core-free basal activity. J Biol Chem 2008;283:25305–15.

38. SteussyCN, PopovKM,Bowker-KinleyMM, SloanRB,Harris RA,HamiltonJA. Structure of pyruvate dehydrogenase kinase - novel folding pattern for aserine protein kinase. J Biol Chem 2001;276:37443–50.

39. Hiromasa Y, Fujisawa T, Aso Y, Roche TE. Organization of the cores of themammalian pyruvate dehydrogenase complex formed by E2 and E2 plusthe E3-binding protein and their capacities to bind the E1 and E3 compo-nents. J Biol Chem 2004;279:6921–33.

40. Morrell JA, Orme J, Butlin RJ, Roche TE, Mayers RM, Kilgour E. AZD7545 isa selective inhibitor of pyruvate dehydrogenase kinase 2. Biochem SocTrans 2003;31:1168–70.

41. Dutta R, InouyeM.GHKL, an emergent ATPase/kinase superfamily. TrendsBiochem Sci 2000;25:24–8.

42. Bellon S, Parsons JD,Wei YY, HayakawaK, Swenson LL, Charifson PS, et al.Crystal structures of Escherichia coli topoisomerase IV ParE subunit(24 and 43 kilodaltons): a single residue dictates differences in novobiocinpotency against topoisomerase IV and DNA gyrase. Antimicrob AgentsChemother 2004;48:1856–64.

43. Whitehou S, Cooper RH, Randle PJ. Mechanism of activation of pyruvate-dehydrogenase by dichloroacetate and other halogenated carboxylic-acids.Biochem J 1974;141:761–74.

44. Liu Q, Sabnis Y, Zhao Z, Zhang T, Buhrlage SJ, Jones LH, et al. Developingirreversible inhibitors of the protein kinase cysteinome. Chem Biol 2013;20:146–59.

45. Singh J, Petter RC, Kluge AF. Targeted covalent drugs of the kinase family.Curr Opin Chem Biol 2010;14:475–80.

46. Carmi C, Mor M, Petronini PG, Alfieri RR. Clinical perspectives forirreversible tyrosine kinase inhibitors in cancer. Biochem Pharmacol 2012;84:1388–99.

47. Barf T, Kaptein A. Irreversible protein kinase inhibitors: balancing thebenefits and risks. J Med Chem 2012;55:6243–62.

48. Cross DAE, Ashton SE, Ghiorghiu S, Eberlein C, Nebhan CA, Spitzler PJ,et al. AZD9291, an irreversible EGFR tki, overcomes T790M-mediatedresistance to EGFR inhibitors in lung cancer. Cancer Discov 2014;4:1046–61.

49. Hurd TR, Collins Y, Abakumova I, Chouchani ET, Baranowski B, FearnleyIM, et al. Inactivation of pyruvate dehydrogenase kinase 2 by mitochon-drial reactive oxygen species. J Biol Chem 2012;287:35153–60.


Sun et al.




2015;75:4923-4936. Published OnlineFirst October 19, 2015.Cancer Res Wenyi Sun, Zuoquan Xie, Yifu Liu, et al. a Covalent Cysteine ModificationJX06 Selectively Inhibits Pyruvate Dehydrogenase Kinase PDK1 by

Updated version

10.1158/0008-5472.CAN-15-1023doi:

Access the most recent version of this article at:

Material

Supplementary

http://cancerres.aacrjournals.org/content/suppl/2015/10/17/0008-5472.CAN-15-1023.DC1

Access the most recent supplemental material at:

Cited articles

http://cancerres.aacrjournals.org/content/75/22/4923.full#ref-list-1

This article cites 47 articles, 21 of which you can access for free at:

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

Subscriptions

Reprints and

[email protected]

To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/75/22/4923To request permission to re-use all or part of this article, use this link



http://cancerres.aacrjournals.org/lookup/doi/10.1158/0008-5472.CAN-15-1023

http://cancerres.aacrjournals.org/content/suppl/2015/10/17/0008-5472.CAN-15-1023.DC1

http://cancerres.aacrjournals.org/content/75/22/4923.full#ref-list-1

http://cancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://cancerres.aacrjournals.org/content/75/22/4923


jx06 selectively inhibits pyruvate dehydrogenase kinase ... · corresponding authors: meiyu geng,...

Documents