RESEARCH ARTICLE SPECIAL COLLECTION: CANCER METABOLISM

Synergistic antiproliferative effects of an mTOR inhibitor (rad001)plus gemcitabine on cholangiocarcinoma by decreasing cholinekinase activityGigin Lin*,1, Kun-Ju Lin*,2, Frank Wang4, Tse-Ching Chen3, Tzu-Chen Yen2 and Ta-Sen Yeh4,‡

ABSTRACTAlthough gemcitabine plus cisplatin is the gold standard chemotherapyregimen for advanced cholangiocarcinoma, the response rate hasbeen disappointing. This study aims to investigate a novel therapeuticregimen [gemcitabine plus everolimus (rad001), an mTOR inhibitor]for cholangiocarcinoma. Gemcitabine, oxaliplatin, cetuximab andrad001 in various combinations were first evaluated in vitro using sixcholangiocarcinoma cell lines. In vivo therapeutic efficacies ofgemcitabine and rad001 alone and their combination were furtherevaluated using a xenograft mouse model and a chemically inducedorthotopic cholangiocarcinoma rat model. In the in vitro study,gemcitabine plus rad001 exerted a synergistic therapeutic effect onthe cholangiocarcinoma cells, irrespective of theKRASmutation status.In the xenograft study, gemcitabine plus rad001 showed the besttherapeutic effect on tumor volume change, and was associated withincreased caspase-3 expression, decreased eIF4E expression, as wellas overexpression of both death receptor- and mitochondrial apoptoticpathway-related genes. In a chemically induced cholangiocarcinoma-afflicted rat model, the gemcitabine plus rad001 treatment suppressedtumor glycolysis as measured by 18F-fludeoxyglucose micro-positronemission tomography. Also, increased intratumoral free choline,decreased glycerophosphocholine and nearly undetectablephosphocholine levels were demonstrated by proton nuclear magneticresonance, supported by results of decreased choline kinaseexpression in western blotting. We concluded that gemcitabine plusrad001 has a synergistic antiproliferative effect on cholangiocarcinoma,irrespective of the KRAS mutation status. The antitumor effect isassociated with activation of both death receptor and mitochondrialpathways, as well as the downregulation of choline kinase activity,resulting in a characteristic change in choline metabolism.

KEY WORDS: Cholangiocarcinoma, Choline kinase, FAS,Gemcitabine, Rad001

INTRODUCTIONCholangiocarcinoma is a malignant biliary tract tumor that carries avery poor prognosis (de Groen et al., 1999; Khan et al., 2005).Although uncommon worldwide, its incidence reaches 87 per100,000 in endemic areas, including Israel, Japan and South EasternAsian countries such as Taiwan (Rajagopalan et al., 2004; Yehet al., 2005). Surgical resection is the only treatment modality thatoffers a potential cure. Unfortunately, resection rates are low at theinitial diagnosis because of early intrahepatic spread, nodalinvolvement, vascular invasion and distant metastasis. Othertherapeutic strategies, including chemotherapy, radiation therapy,transplantation and photodynamic therapy, have been developed,but none has shown significant survival benefit in patients withadvanced cholangiocarcinoma (de Groen et al., 1999; Khan et al.,2005). Gemcitabine, with or without platinum compounds, iscurrently the most effective chemotherapy regimen for advancedbiliary tract cancer, although the response rate is only ∼20% (Andréet al., 2004; Eckel and Schmid, 2007; Valle et al., 2010, 2009).In addition, epidermal growth factor receptor (EGFR) is activatedby bile acids and has a role in the carcinogenesis ofcholangiocarcinoma through the induction of cyclooxygenase-2expression via anMAPK protein cascade. Accordingly, neutralizingantibodies that target the extracellular domain of EGFR (e.g.cetuximab) have sporadically shown clinical efficacy in advancedcholangiocarcinoma (Paule et al., 2007).

Mammalian target of rapamycin (mTOR) is a conserved serine/threonine kinase that regulates cell growth and metabolism. ThemTOR pathway is dysregulated in various cancers includingcholangiocarcinoma (Chung et al., 2009; Ma and Blenis, 2009),making mTOR an important target for the development of newanticancer drugs (Faivre et al., 2006). Everolimus (rad001), a 40-O-(2-hydroxyethyl) derivative of rapamycin, conjugates with FKbinding protein-12 (FKBP-12; FKBP1A) to inhibit the activation ofmTOR. Although rad001 exhibits antitumor activity both as a singleoral agent and in combination with other anticancer agents(Mabuchi et al., 2007; Majumder et al., 2004; Mondesire et al.,2004; Motzer et al., 2008; Piguet et al., 2008; Yan et al., 2006;Zhang et al., 2009), its therapeutic effect on cholangiocarcinoma hasyet to be clarified.

Phosphocholine (PC) is both a precursor and a breakdown productof phosphatidylcholine, the most abundant phospholipid in thebiological membranes. Increased levels of PC and total choline, asdetected by 1H or 31P magnetic resonance spectroscopy (MRS), arecharacteristics of cancer cells, including breast, prostate, colon, lung,ovarian and brain tumors (Ackerstaff et al., 2003; Iorio et al., 2005;Venkatesh et al., 2012). The significance of these studies is thatspecific changes in the levels of the phospholipid precursors andcatabolites in the context of solid cancers might be a usefulbiochemical indicator of tumor progression and/or treatment response.Received 13 November 2017; Accepted 26 March 2018

1Department of Medical Imaging and Intervention, Imaging Core Lab, Institute forRadiological Research, Clinical Metabolomics Core Lab, Chang Gung MemorialHospital at Linkou, Chang Gung University, Taoyuan 333, Taiwan. 2Department ofNuclear Medicine and Molecular Imaging Center, Chang Gung Memorial Hospitalat Linkou, Chang Gung University, Taoyuan 333, Taiwan. 3Department ofPathology, Chang Gung Memorial Hospital at Linkou, Chang Gung University,Taoyuan 333, Taiwan. 4Department of Surgery, Chang Gung Memorial Hospital atLinkou, Chang Gung University, Taoyuan 333, Taiwan.*These authors contributed equally to this work

‡Author for correspondence (tsy471027@adm.cgmh.org.tw)

G.L., 0000-0001-7246-1058; T.-C.C., 0000-0001-5526-8424; T.-C.Y., 0000-0002-2165-8464; T.-S.Y., 0000-0002-1830-9466

This is an Open Access article distributed under the terms of the Creative Commons AttributionLicense (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use,distribution and reproduction in any medium provided that the original work is properly attributed.

1

© 2018. Published by The Company of Biologists Ltd | Disease Models & Mechanisms (2018) 11, dmm033050. doi:10.1242/dmm.033050

Disea

seModels&Mechan

isms

This study aimed to evaluate the therapeutic effects of rad001 andgemcitabine on cholangiocarcinoma, and elucidate the molecularand metabolic mechanisms underlying their antitumor activity.

RESULTSSelection of optimal drug regimens for cholangiocarcinomacellsThe corresponding proliferation indices of the six cholangiocarcinomacell lines treated with standard IC50 dosages of various therapeuticregimens are shown in Fig. 1A. Gemcitabine exhibited superiortherapeutic efficacy against all cholangiocarcinoma cell lines as asingle-agent therapy, comparedwithoxaliplatin, cetuximaband rad001.Gemcitabine plus rad001 was the most promising combinationtherapeutic regimen, in comparison with gemcitabine plus oxaliplatinand gemcitabine plus oxaliplatin plus cetuximab. The median effectsanalysis demonstrated that rad001 and gemcitabine had a synergisticeffect on bothKRASmutation (HuCCT1, RBE) andwild-type (TFK-1,YSCCC) cell lines with a combination index (CI)<1 (Fig. 1B).

Rationale of mTOR inhibition for treatment ofcholangiocarcinomaTo assess the pharmacological effects of rad001 and gemcitabine,four treatment groups were assigned for in vitro and in vivo

experiments: control, gemcitabine-treated (Gem), rad001-treated(Rad) and gemcitabine plus rad001-treated (Gem+Rad) groups. Inthe in vitro analyses, phosphorylation of p70S6K (RPS6KB1), butnot p38MAPK (MAPK14), was significantly decreased in the Radand Gem+Rad groups in both HuCCT1and TFK-1 cells, indicatingthat Gem and Gem+Rad specifically inhibited the downstreammTOR signaling irrespective of the KRAS mutation status. Gemmight not synergistically enhance the effect of Rad to inhibit mTOR(Fig. 1C). Regarding cell cycling, a smaller percentage of HuCCT1cells entered the G2/M phase in the Gem+Rad group compared withthe Gem group (10.2% versus 13.6%, P<0.05; Fig. 1D, Table 1).Furthermore, the percentage of apoptotic HuCCT1 cells in the Gem+Rad group was significantly increased compared with the Gemgroup (30.1% versus 4.4%, P<0.01; Fig. 1E).

Evaluation of therapeutic response using a xenograft modelXenograft experiments were conducted to evaluate the measurablevolumetric changes of the cholangiocarcinomas treated with variousdrug regimens. The representative therapeutic response of theHuCCT1 xenograft model after various treatments for 3 weeks isshown in Fig. 2A. The therapeutic responses of HuCCT1 and TFK-1xenograft models treated for 3 weeks were analyzed (Fig. 2B).Gem+Rad regime conferred the most promising therapeutic

Fig. 1. Optimization ofmTOR inhibition for treatment of cholangiocarcinoma. (A) Proliferation index of six cholangiocarcinoma cell lines treated with differentregimens (n=4 in each group). We examined the corresponding proliferation indices based on the HuCCT1 cell line and applied the dosage to the other fivecholangiocarcinoma cell lines treated with mono-therapy and combined therapy, respectively. OX, oxaliplatin; Gem, gemcitabine; Erb, cetuximab; Rad, rad001.(B) The interaction between rad001 and gemcitabine in cholangiocarcinoma cell lines was determined by median effects analysis to derive a combination index(CI). (C) Phosphorylation of p70S6K and p38MAPK in HuCCT1 and TFK-1 cells detected by the Bio-Plex phosphoprotein assay (n=4 in each group). *P<0.05versus control group. (D) Cell cycle changes in HuCCT1 cells detected by flow cytometry (n=4 in each group). (E) Percentage of apoptotic HuCCT1 cells detectedby an Annexin V-FITC Apoptosis Kit (n=4 in each group). Error bars indicate s.d.

2

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm033050. doi:10.1242/dmm.033050

Disea

seModels&Mechan

isms

response in both HuCCT1 and TFK-1 xenograft models.Representative images of caspase-3 and eIF4E expression in theHuCCT1 xenograft are shown in Fig. 2C. The expression ofcaspase-3 in the control, Gem, Rad and Gem+Rad groups was 2.1±0.7, 9.4±1.6, 5.1±1.1 and 14.5±2.3 per high-power field (HPF),respectively (P<0.01). In contrast, the expression of eIF4E in thecontrol, Gem, Rad and Gem+Rad groups was 14.2±1.3, 8.9±1.5,11.3±2.3 and 4.6±0.7 per HPF, respectively (P<0.01).

Altered apoptotic signaling after gemcitabine and rad001combination therapyBased on the xenograft model, the percentages of FAS-positiveHuCCT1 cells in the control, Gem, Rad and Gem+Rad groups, asdetected by flow cytometry, were 13.7%, 30.6%, 23.6% and 44.5%,respectively (P<0.01), and the intensity of FAS-positiveHuCCT1 cellswas 7.2, 9.5, 8.4 and 10.5, respectively (P<0.01; Fig. 3A). Theexpression of six apoptotic genes, including FAS, CASP8, BID,APAF1, XIAP and CASP3, derived from the treated HuCCT1xenograft using quantitative real-time polymerase chain reaction(qPCR), is shown in Fig. 3B. The expression of these six genes wasmodestly increased in the Gem and Rad groups compared with thecontrol group, whereas the expression of all six genes was significantlyincreased in the Gem+Rad group compared with the other groups.

Metabolic response of orthotopic cholangiocarcinoma ratmodel as detected by 18F-FDG microPET and 1H NMRThe representative in vivometabolic response of cholangiocarcinoma-afflicted rats to various treatments as detected by 18F-fludeoxyglucosemicro-positron emission tomography (18F-FDG microPET) is shownin Fig. 4A. The Gem+Rad group had the most significant metabolicresponse after two cycles of treatment (Fig. 4B). The expression ofGlut-1 (Slc2a1), Hk2, Hif1a and Vegf (Vegfa) mRNAs of thecorresponding surgical specimens as detected by qPCR is shown inFig. 4C. Of these, the expression ofHk2 andGlut-1mRNA in theGem+Rad groupwas reduced by half in comparison with the control group.The difference in the expression of these mRNAs correlated with thestandardized uptake value (SUV) changes in FDG microPET. Thetumor response in the cholangiocarcinoma-afflicted rats wasconfirmed by histopathological examination. Results derived fromthe histomorphological score and Ki67 (Mki67) labeling index of thecorresponding surgical specimen also supported the findings in theFDG microPET study (Table 2).The alteration of choline-associated metabolites in the treated

cholangiocarcinoma-afflicted rats was analyzed by 1H nuclearmagnetic resonance (NMR), with a particular emphasis on the freecholine, PC and glycerophosphocholine (GPC) levels (Fig. 5A). Asignificant increase in the free choline level of the Gem+Rad groupwas associated with dramatic disappearance of PC and asignificantly reduced level of GPC (Fig. 5B). In contrast, the Gemgroup exhibited a significant increase in the free choline levelwithout significant changes in PC and GPC levels. No significantchange in the choline-associated metabolites was observed in the

Rad group. Choline kinase (CK; CHKA), an enzyme that catalyzesthe phosphorylation of free choline using ATP to produce PC andplays a rate-limiting regulatory role in the phosphatidylcholinebiosynthesis, was assayed to determine the underlying mechanismthat was responsible for the observed changes in the PC levels. Ofthe four treatment groups, the CK level detected by western blottingwas lowest in the Gem+Rad group (Fig. 5C,D).

DISCUSSIONThe presence of the tyrosine kinase-AKT-mTOR axis in thecarcinogenesis of cholangiocarcinoma (Chung et al., 2009) hasprovided us with an opportunity to evaluate the therapeutic role ofan mTOR inhibitor, rad001, in the management of this disease. Wehave shown in this study that rad001 treatment resulted in reductionof expression of the downstream molecules of mTOR, such asp70S6K-1, but not p38MAPK. Furthermore, we demonstrated, forthe first time, that gemcitabine plus rad001 exerted a synergisticantitumor effect on cholangiocarcinoma, independent of the KRASmutation status. In contrast, the EGFR monoclonal antibody,cetuximab, had a modest therapeutic effect on the KRAS wild-typecholangiocarcinoma xenograft (TFK-1), but not the KRAS mutatedxenograft (HuCCT1). This finding is consistent with previousclinical experience on colorectal and nonsmall cell lung cancers(NSCLCs) (Bardelli and Siena, 2010; Massarelli et al., 2007).The in vivo therapeutic effect of gemcitabine plus rad001 wasmacroscopically evaluated and confirmed by measuring changes inthe tumor volume in the xenograft model, and assessing themetabolic response in the orthotopic rat cholangiocarcinoma modelusing 18F-FDG microPET. The downstream molecules of mTOR(Ki67, Hif1a, Hk2, Glut-1 and Vegf) were all attenuated in the Rad+Gem group. This observation reflected the changes in SUVidentified in 18F-FDG microPET.

The available evidence in the literature suggests that gemcitabineprimarily acts on an intrinsic (mitochondrial) apoptotic pathway inNSCLCs and pancreatic cancer to achieve its pharmacological effect.Gemcitabine-mediated activation of caspase-8 is a late andmitochondrial-dependent event which is independent of FAS-FASL(FASLG) signaling (Ferreira et al., 2000; Salvesen and Duckett,2002). Que et al. reported that cholangiocarcinoma cells synthesizedFASL and FAS and that FASL expressed by these cells inducedhost lymphocyte cell death (Que et al., 1999). Shimonishi et al.further demonstrated upregulation of FASL in the early stagesand downregulation of FAS in the advanced stages ofcholangiocarcinoma, the latter of which was responsible for evasionfrom host immunity and disease progression (Shimonishi et al., 2000).To understand themechanism bywhich rad001 and gemcitabine actedsynergistically to enhance apoptosis in the cholangiocarcinoma cells,we examined six key apoptotic genes that were involved in theextrinsic and intrinsic cell death pathways. These included FAS,CASP8, BID, APAF1, XIAP and CASP3. Among these six genes, theformer two represent the death receptor (extrinsic) pathway, whereasthe latter three represent the mitochondrial (intrinsic) pathway,with both pathways intermediated by BID (Salvesen and Duckett,2002). Of particular importance, XIAP, an X-linked inhibitor ofapoptosis, is thought to directly inhibit certain caspases such ascaspase-3, caspase-7 and caspase-9. Shrikhande et al. showed thatXIAP was overexpressed in pancreatic cancer and contributed tochemoresistance (Shrikhande et al., 2006). Our data revealed that bothdeath receptor-related (FAS and CASP8) and mitochondrial pathway-related genes (APAF1 and CASP3) in the Gem+Rad group wereincreased 10-fold in comparison with those of the Gem group,indicating that both death receptor and mitochondrial pathways were

Table 1. Cell cycle distribution of HuCCT1 treated by different regimens

Phase Control (%) Gem (%) Rad (%) Gem+Rad (%)

G1 50.4±0.4 76.1±2.5* 45.1±2.4*† 71.9±1.6*‡

S 28.0±0.5 10.1±2.2* 24.9±2.5† 17.6±1.9*†‡

G2/M 21.6±0.4 13.6±0.9* 29.9±0.4*† 10.2±0.3*†‡

Gem, gemcitabine; Rad, rad001.*, † and ‡ representP<0.05 when comparedwith control, Gem and Rad groups,respectively.

3

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm033050. doi:10.1242/dmm.033050

Disea

seModels&Mechan

isms

recruited and co-activated by the combination therapy. It is noteworthythat XIAP was simultaneously increased by up to 10-fold in the Gem+Rad group, along with the pro-apoptotic genes, in comparison withthe Gem group. Our findings suggested that XIAP plays a major rolein arresting the apoptotic process and potentially contributing to drugresistance.Early detection of response to therapy is desirable but not easily

achieved using conventional imaging techniques such as computedtomography or magnetic resonance imaging. Biological cancertreatment might be associated with stagnation of cellular

proliferation or cellular clonal changes rather than obvious tumorshrinkage in the early stage of the therapeutic response. Cancermetabolism has emerged in recent studies as a powerful linkbetween deregulated signaling pathways and altered cellularmetabolism. Specific endogenous metabolites and metabolicfluxes within cells and tissues in vivo and in vitro can be readilyidentified using 31P, 1H or 13C MRS. Of them, derangement in thecholine metabolism during carcinogenesis and treatment has beenincreasingly recognized in various solid cancers (Ackerstaff et al.,2003). CK catalyzes the phosphorylation of choline to yield PC in

Fig. 2. Therapeutic response of HcCCT1 and TFK-1 xenograft models. (A) Representative features of tumor growth in HuCCT1 xenograft models after a3-week treatment (n=8 in each group). (B) Left: quantitative analysis of tumor growth in HuCCT1 xenograft models after a 3-week treatment. *P<0.05 and**P<0.01 versus control group, respectively; †P<0.05 versus Rad group; ‡P<0.05 versus Gem group. Right: quantitative analysis of tumor growth in TFK-1xenograft models after a 3-week treatment (n=8 in each group). *P<0.05 and **P<0.01 versus control group, respectively; †P<0.05 versus Gem group; ‡P<0.05versus Gem+cetuximab group. (C) Representative expression of caspase-3 and eIF4E in HuCCT1 xenograft model detected by immunohistochemistry (n=8 ineach group). Original magnification, 400× for caspase -3 and 40× for eIF4E. Error bars indicate s.d.

4

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm033050. doi:10.1242/dmm.033050

Disea

seModels&Mechan

isms

the biosynthesis of phosphatidylcholine, a process known as theKennedy pathway (Ackerstaff et al., 2003; Ramírez deMolina et al.,2008). The enzyme is modulated by RAS and interacts with HIF1A(Ramírez deMolina et al., 2002a). It might also play a role in the cellcycle regulation (Gruber et al., 2012), MAPK and PI3K (PIK3CA)/AKT protein signaling pathways (Yalcin et al., 2010), as well asacting as a potential biomarker for various solid cancers (Ramírez deMolina et al., 2002b). Specific inhibitors against CK are to beanticipated soon in clinical trials as these agents have shownpromising antitumor effects in preclinical studies (Ackerstaff et al.,2003; Al-Saffar et al., 2010; Ramírez deMolina et al., 2008, 2002a).In the present study, we have implemented 1H NMR to detectalterations in choline-associated metabolites in the treatedcholangiocarcinoma-afflicted rats. Downregulated CK activity,presenting as an accumulation of free choline and nearlyundetectable levels of PC in the Gem+Rad group, was confirmedby significantly decreased expression of CK in western blotting.Such downregulation of CK following mTOR inhibition wasassociated with HIF1A and PI3K/AKT/mTOR pathways, with thedownregulation of Hif1a mRNA correlating with the respective CKlevels in the Gem+Rad, Gem and control groups. In line with our

study results, Glunde et al. demonstrated a similar process of CKregulation by HIF1A signaling in human prostate cancer (Glundeet al., 2008). Furthermore, using a PI3K inhibitor, PI-103, Al-Saffaret al. were able to demonstrate downregulation of CK levels inprostate and colon cancers, with a corresponding decrease in PClevels (Al-Saffar et al., 2010). Again, by silencing CK using smallinterfering RNA (siRNA), Yalcin et al. showed that PC was reducedthrough the PI3K/AKT/mTOR pathways, leading to the inhibitionof cervical cancer cell growth (Yalcin et al., 2010).

The change inGPC levels after biological agent treatment is anotherinteresting issue. Al-Saffar et al. reported that blockade of PI3K withLY294002 (a relatively less potent, nonselective inhibitor of PI3K)was associated with a decrease in PC levels as well as an elevation inGPC content in human breast cancer cells (Beloueche-Babari et al.,2006). In contrast, the authors noted that both PC andGPC levels werebelow the detection limit of MRS when the HCT116 colon cancercells were treated instead with PI-103, a potent selective inhibitor ofclass 1 PI3K (Al-Saffar et al., 2010). Accordingly, they concluded thatthe contradictory response of the GPC levels followingLY294002 andPI-103 treatments was probably related to other off-target effects of theformer inhibitor. In our study, the characteristic profile of the cholinemetabolites in the cholangiocarcinoma cells treated by gemcitabineplus rad001 was most likely a result of a potent and selective targetingagainst the PI3K/AKT/mTOR signaling pathway. As there is anincrease in choline and a decrease in its downstream metabolite, PC,we expect to observe an accumulation of choline signal on 11C-cholinePET when the treatment is effective. A future in vivo imaging studyusing 11C-choline PET or magnetic resonance spectroscopy imagingwould better facilitate longitudinal monitoring of the relevantbiological response.

In conclusion, our preclinical study shows that gemcitabine plusrad001 exerts a synergistic antitumor effect on cholangiocarcinomairrespective of theKRASmutation status. The underlyingmechanismsof antitumor activity are associated with activation of the deathreceptor andmitochondrial pathways, aswell as the downregulation ofCK activity, resulting in a specific change in choline metabolism.

MATERIALS AND METHODSCholangiocarcinoma cell linesSix cholangiocarcinoma cell lines used in this study (HuCCT1, SSP-25,RBE, YSCCC, TGBC-24TKB and TFK-1) were purchased from RIKENCell Bank (Tsukuba, Japan). The HuCCT1, SSP-25 and RBE cell lines hada KRAS mutation, whereas the remaining three cell lines were KRAS wildtype. All in vitro studies were performed at least in triplicate.

In vitro drug cytotoxicity assayOxaliplatin (TTY Biopharm, Taiwan), gemcitabine (Gemmis, TTYBiopharm), cetuximab (Erbitux, Boehringer Ingelheim, Germany), andeverolimus (rad001, Novartis, Switzerland), alone or in combination, werefirst used to treat the cholangiocarcinoma cell lines in vitro. The half-maximal inhibitory concentration (IC50) was determined using a cellproliferation reagent, WST-1 (Roche Applied Science, Germany).

Analysis of interactionsA commercial software package (CalcuSyn, Biosoft, UK) was used toperformmedian effect analysis as described by Chou and Talalay (Chou andTalalay, 1984). The combination index (CI) was calculated using theformula: CI=(D)1/(Dx)1+(D)2/(Dx)2, where CI=1 indicated an additiveeffect, CI>1 an antagonistic effect, and CI<1 a synergistic effect.

Flow cytometryBoth adherent and floating cholangiocarcinoma cells were collected 72 hafter various drug treatments for cell cycle analysis. Cellswere passed through

Fig. 3. Altered apoptotic signaling after gemcitabine and rad001combination therapy. (A) Percentages (red digits) and intensities (blue digits)of FAS-positive HuCCT1 cells in control, gemcitabine-treated, rad001-treated,and gemcitabine plus rad001-treated groups detected by flow cytometry (n=4in each group). (B) The expression of FAS, CASP8, BID, APAF1, XIAP andCASP3 mRNA in the HuCCT1 xenograft model detected by quantitative real-time PCR (n=6 in each group). *P<0.05 and **P<0.01 versus control group,respectively. †P<0.05 and ††P<0.01 versus Gem group, respectively; ‡P<0.01versus Rad group. Error bars indicate s.d.

5

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm033050. doi:10.1242/dmm.033050

Disea

seModels&Mechan

isms

a fluorescence activated cell sorter (FACS Caliber, BD Biosciences, USA),and data were acquired using CellQuest (BD Biosciences) software. The cellcycle was modeled using ModFit software (Venty Software, USA). Theseparation of cells into G0/G1, S and G2/M phases was based on linearfluorescence intensity after staining with propidium iodide. Cells werelabeled with the Annexin V-FITCApoptosis Kit (BDBiosciences-Clontech)and analyzed on a FACS Caliber to detect apoptosis. The cell surfaceexpression of FAS after various drug treatments for 16 h was assessed.Indirect immunofluorescence staining for FAS was performed with an IgG1mouse anti-FAS primarymonoclonal antibody (DX2 clone, BDBiosciences)and a fluorescein isothiocyanate (FITC)-conjugated secondaryantibody (goatanti-mouse Ig) (BD Biosciences). The percentage and intensity of FAS-positive cells were analyzed by the FACS Caliber.

Bio-Plex phosphoprotein assayCholangiocarcinoma cells were treated with various drug regimens, andprotein lysates were prepared using a cell lysis kit (Bio-Rad, USA).Phospho-p38MAPK and phospho-p70S6 kinase were detected by the Bio-

Plex phosphoprotein and total protein assay kits (Bio-Rad) according to themanufacturer’s protocols.

Xenograft mice modelTo initiate tumor xenograft, cholangiocarcinoma cells, HuCCT1 and TFK-1,were implanted subcutaneously (2×106 cells) in the flank of the Fox Chasesevere combined immunodeficiency (SCID) mice (BioLASCO, Taiwan).Six therapeutic regimens were provided, as follows: gemcitabine 200 mg/kgby intraperitoneal injection (I.P.)+oxaliplatin 5 mg/kg I.P. every week(GEMOX); gemcitabine 200 mg/kg I.P.+oxaliplatin 5 mg/kgI.P.+cetuximab 2 mg/kg I.P. every week (GEMOX+cetuximab);gemcitabine 200 mg/kg I.P. every week (GEM); rad001 5 mg/kg orallytwice every week (Rad); gemcitabine, 200 mg/kg I.P. every week+rad0015 mg/kg orally twice every week (GEM+Rad); and saline with the samevolume I.P. every week (control group). Tumor size did not exceed1000 mm3 (∼4% of mouse body weight) in compliance with animalwelfare regulations. At the end of the third week, the xenografts weremeasured and retrieved. The surgical specimens were divided into twoportions; half was kept frozen and stored at −80°C and the remaining halfwas prepared as a paraffin block. All animal experiments were performedaccording to a protocol approved by the Institutional Animal Care and UseCommittee, Chang Gung University and Chang Gung Memorial Hospital,Taiwan (IACUC 2012092401). The care and use of experimental animalsconformed with the regulatory standards.

Immunohistochemical stainingFormalin-fixed, paraffin-embedded tissues were cut into 4-µm sections andmounted on glue-coated slides. A modification of the avidin-biotin-peroxidase complex immunohistochemical method was performed. Variousprimary antibodies [human caspase-3 and eukaryotic initiating factor 4E(eIF4E); and rat Ki67] were applied. The expression of eIF4E was assessed

Fig. 4. Metabolic response of the orthotopic cholangiocarcinoma rat model, as detected by 18F-FDG microPET. (A) Representative image of 18F-FDGmicroPET on cholangiocarcinoma-afflicted rats (n=6 in each group). (B) Metabolic response of cholangiocarcinoma-afflicted rats detected by 18F-FDGmicroPETwas quantitatively analyzed (n=6 in each group). G, gemcitabine; R, Rad001. *P<0.05 and **P<0.01 versus control group, respectively; †P<0.05 versus Radgroup; ‡P<0.05 versus Gem group. (C) Expression ofGlut-1,Hk2,Hif1a and VegfmRNAs in rat cholangiocarcinoma detected by quantitative real-time PCR (n=6in each group). *P<0.05 versus control group; †P<0.05 versus Rad group; ‡P<0.05 versus Gem group. Error bars indicate s.d.

Table 2. Assessment of tumor response of rat cholangiocarcinoma tochemo- and/or target therapy

Groups Histomorphological regression score Ki67 labeling index

Control 3.6±0.5 62±13%Gem 2.7±0.7* 36±11%*Rad 2.8±0.7* 39±17%*Gem+Rad 1.7±0.7*†‡ 15±4%*†‡

Gem, gemcitabine; Rad, rad001.*, † and ‡ represent P<0.05 when compared with control, Gem- and Rad-groups, respectively.

6

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm033050. doi:10.1242/dmm.033050

Disea

seModels&Mechan

isms

using the product of two scores (intensity×% positive), whereas theexpression of caspase-3 was represented as positively stained cells per HPF.The Ki67 labeling index was calculated as positively stained cells per 100cells counted×100%. Histopathological assessment of tumor response wasperformed by both histomorphological regression analysis (Schneider et al.,2005) and Ki67 labeling index. The extent of histomorphological regressionwas divided into four categories: grade 1, 75-100% viable residual tumorcells; grade 2, 50-75% viable residual tumor cells; grade 3, 25-50% viableresidual tumor cells; and grade 4, 0-25% viable residual tumor cells.

Thioacetamide-induced cholangiocarcinoma-afflicted rat modelMale Sprague-Dawley rats weighing 180-200 g were housed in atemperature- and humidity-controlled environment with free access towater and rat chow. Cholangiocarcinomawas induced by oral administrationof thioacetamide (0.03% w/w in water) for 24 weeks, as previouslydescribed (Jan et al., 2004). The treatment regimens were as follows:(1) gemcitabine, 25 mg/kg I.P. every 2 weeks; (2) rad001, 5 mg/kg orallytwice every 2 weeks; (3) gemcitabine, 25 mg/kg I.P. every 2 weeks+rad001,5 mg/kg orally twice every 2 weeks; or (4) saline with the same volume I.P.every 2 weeks for the vehicle group. The animals underwent FDGmicroPET pretreatment, and at 2 and 4 weeks after treatment, beforebeing sacrificed for histological confirmation.

18F-FDG microPET imaging to measure metabolic responseRats had full access to drinking water at all times and were fasted overnightfor 8 h before radiotracer injection. A dose of 18.5 MBq (0.5 mCi) 18F-FDGwas administered and the liver region was scanned for 120 min continuouslyafter intravenous radiotracer injection. Image analysis was carried out at animage analysis workstation (PMODTechnologies, Switzerland). Regions ofinterest were drawn over the liver tumor and nontumoral liver background.The time-activity curve and tumor-to-liver background (T/L) ratio were

calculated. The optimal scanning time point was determined when thehighest T/L ratio was reached. SUVs were used to quantify tracer uptakeaccording to the following formula:

SUV =Decay corrected tissue activity(Bq/ml)

Injected dose(Bq)/Bodyweight(g)

The SUVmax and SUVmean are maximal and mean SUV, respectively,within the voxels of interest (VOIs). Tumor counts and FDG uptake valueswere obtained in all animals (Yeh et al., 2008).

Quantification of intratumoral metabolites by NMRThioacetamide-induced rat cholangiocarcinoma tumor tissues (∼100 mg)were extracted by dual-phase procedures as previously described(Ackerstaff et al., 2003; Iorio et al., 2005; Venkatesh et al., 2012). Water-soluble extracts were freeze-dried and reconstituted in 700 µl deuteratedwater (D2O, Sigma-Aldrich, USA), and the extracts (500 µl) were thenplaced in 5 mm NMR tubes. Fifty microliters of 0.75% sodium 3-trimethylsilyl-2,2,3,3-tetradeuteropropionate (TSP) in D2O (Sigma-Aldrich) was added to the samples for chemical shift calibration andquantification. 1H NMR measurement was performed on a 600 MHzspectrometer (Bruker, Germany): 7500 Hz spectral width, 16,384 timedomain points, 128 scans, temperature 298 K, acquisition time ∼5 min. Thewater resonance was suppressed by a gated irradiation centered on the waterfrequency. Spectral processing was carried out using the Bruker Topspin-2software package to quantify the metabolites.

qPCR for human FAS, CASP8, BID, APAF1, XIAP and CASP3, andrat Glut-1, Hk2, Hif1a and VegfTotal RNA (2 µg) was treated with DNAse 1 (Invitrogen, USA) in a 50 µlreaction mixture. TaqMan primers and probes were used for quantitativedetection of human apoptotic genes (FAS, CASP8, BID, APAF1, XIAP andCASP3). Rat Glut-1 (ABI assay ID, Rn01417099), Hk2 (Rn00562457),Hif1a (Rn00577560), Vegf (Rn01511601) and Gapdh (Rn99999916) weredesigned with Primer Express (ABI/Perkin Elmer) using the GenBankaccession number. cDNA samples were mixed with 2× Universal TaqManbuffer containing the Taq enzyme, primers and probes to a total volume of25 µl. The thermal cycle conditions were 50°C for 2 min, 95°C for 10 min,and 42 cycles of 95°C for 15 s and 60°C for 1 min. All PCRs and analyseswere performed using an ABI PRISM 7700 Sequence Detection System(Applied Biosystems, USA). All samples were run in triplicate.

Statistical analysisAll continuous variables are expressed as the mean (s.d.). Statisticalanalysis for continuous variables was performed using Student’s t-test orone-way ANOVA test where appropriate. Intergroup comparisons of CKwere otherwise performed using a box plot, where the data were expressedas the median value. A P-value <0.05 was considered statisticallysignificant.

This article is part of a special subject collection ‘Cancer Metabolism: models,mechanisms and targets’, which was launched in a dedicated issue guest edited byAlmut Schulze and Mariia Yuneva. See related articles in this collection at http://dmm.biologists.org/collection/cancermetabolism.

AcknowledgementsWe thank the Metabolomics Core Laboratory, Healthy Aging Research Center(HARC), Chang Gung University and the Clinical Metabolomics Core Laboratory,Chang Gung Memorial Hospital for carrying out analyses using NMR spectroscopy.We are grateful to Center for AdvancedMolecular Imaging and Translation (CAMIT),Chang Gung Memorial Hospital for assisting with microPET experiments, andKuan-Ying Lu for help with manuscript preparation.

Competing interestsThe authors declare no competing or financial interests.

Author contributionsConceptualization: G.L., K.-J.L., F.W., T.-C.Y., T.-S.Y.; Methodology: G.L., K.-J.L.,F.W., T.-S.Y.; Software: G.L., K.-J.L., T.-S.Y.; Validation: G.L., K.-J.L., T.-C.C.,

Fig. 5. Alteration of choline-associated metabolites in the treatedcholangiocarcinoma-afflicted rats, as detected by 1H NMR.(A) Representative NMR spectrum of choline-associated metabolites incholangiocarcinoma-afflicted rats. Cho, choline; PC, phosphocholine; GPC,glycerophosphocholine. (B) Relative fold changes of intratumoral choline-associated metabolites after various treatment for 4 weeks (n=6 in eachgroup). G, gemcitabine; R, rad001. *P<0.05 versus control group.(C) Representative features of choline kinase of treated cholangiocarcinoma-afflicted rats detected by western blotting. Gem, gemcitabine; Rad, rad001.MCF-7, a breast cancer cell line, served as a positive control; actin served as aninternal control. (D) Quantitative analysis of CK in treated cholangiocarcinoma-afflicted rats detected by western blotting (n=6 in each group). G, gemcitabine;R, rad001. *P=0.021 versus control; †P=0.021 versus G; ‡P=0.043 versus R.Error bars indicate s.d.

7

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm033050. doi:10.1242/dmm.033050

Disea

seModels&Mechan

isms

T.-S.Y.; Formal analysis: G.L., K.-J.L., F.W., T.-C.C., T.-C.Y., T.-S.Y.; Investigation:G.L., K.-J.L., T. Yen, T.-S.Y.; Resources: G.L., T.-C.Y., T.-S.Y.; Data curation: G.L.,K.-J.L., F.W., T.-C.C., T.-C.Y., T.-S.Y.; Writing - original draft: G.L., K.-J.L., T.-S.Y.;Writing - review & editing: G.L., K.-J.L., F.W., T.-C.C., T.-C.Y., T.-S.Y.; Visualization:G.L., K.-J.L., T.-C.C., T.-S.Y.; Supervision: T.-C.Y., T.-S.Y.; Project administration:G.L., K.-J.L., T.-C.Y., T.-S.Y.; Funding acquisition: T.-C.Y., T.-S.Y.

FundingThis work was supported by Chang Gung Medical Foundation (CRRPG3F0021∼2,CMRPG3G1091∼2, CMRPG3E1321∼2), National Science Council (MOST 106-2314-B-182A-019-MY3) and Department of Health, Taiwan (DOH99-TD-C-111-006).

Data availabilityDetailed data are available at figshare (https://figshare.com/s/6e46c082a9a1f60d448b).

ReferencesAckerstaff, E., Glunde, K. and Bhujwalla, Z. M. (2003). Choline phospholipidmetabolism: a target in cancer cells? J. Cell. Biochem. 90, 525-533.

Al-Saffar, N. M. S., Jackson, L. E., Raynaud, F. I., Clarke, P. A., Ramirez deMolina, A., Lacal, J. C., Workman, P. and Leach, M. O. (2010). Thephosphoinositide 3-kinase inhibitor PI-103 downregulates choline kinase alphaleading to phosphocholine and total choline decrease detected by magneticresonance spectroscopy. Cancer Res. 70, 5507-5517.

Andre, T., Tournigand, C., Rosmorduc, O., Provent, S., Maindrault-Goebel, F.,Avenin, D., Selle, F., Paye, F., Hannoun, L., Houry, S. et al. (2004). Gemcitabinecombined with oxaliplatin (GEMOX) in advanced biliary tract adenocarcinoma: aGERCOR study. Ann. Oncol. 15, 1339-1343.

Bardelli, A. and Siena, S. (2010). Molecular mechanisms of resistance tocetuximab and panitumumab in colorectal cancer. J. Clin. Oncol. 28, 1254-1261.

Beloueche-Babari, M., Jackson, L. E., Al-Saffar, N. M., Eccles, S. A., Raynaud,F. I., Workman, P., Leach, M. O. and Ronen, S. M. (2006). Identification ofmagnetic resonance detectable metabolic changes associated with inhibition ofphosphoinositide 3-kinase signaling in human breast cancer cells. Mol. CancerTher. 5, 187-196.

Chou, T.-C. and Talalay, P. (1984). Quantitative analysis of dose-effectrelationships: the combined effects of multiple drugs or enzyme inhibitors. Adv.Enzyme Regul. 22, 27-55.

Chung, J.-Y., Hong, S.-M., Choi, B. Y., Cho, H., Yu, E. and Hewitt, S. M. (2009).The expression of phospho-AKT, phospho-mTOR, and PTEN in extrahepaticcholangiocarcinoma. Clin. Cancer Res. 15, 660-667.

de Groen, P. C., Gores, G. J., LaRusso, N. F., Gunderson, L. L. and Nagorney,D. M. (1999). Biliary tract cancers. N. Engl. J. Med. 341, 1368-1378.

Eckel, F. and Schmid, R. M. (2007). Chemotherapy in advanced biliary tractcarcinoma: a pooled analysis of clinical trials. Br. J. Cancer 96, 896-902.

Faivre, S., Kroemer, G. and Raymond, E. (2006). Current development of mTORinhibitors as anticancer agents. Nat. Rev. Drug Discov. 5, 671-688.

Ferreira, C. G., Span, S. W., Peters, G. J., Kruyt, F. A. and Giaccone, G. (2000).Chemotherapy triggers apoptosis in a caspase-8-dependent and mitochondria-controlled manner in the non-small cell lung cancer cell line NCI-H460. CancerRes. 60, 7133-7141.

Glunde, K., Shah, T., Winnard, P. T., Jr, Raman, V., Takagi, T., Vesuna, F.,Artemov, D. and Bhujwalla, Z. M. (2008). Hypoxia regulates choline kinaseexpression through hypoxia-inducible factor-1 alpha signaling in a human prostatecancer model. Cancer Res. 68, 172-180.

Gruber, J., See Too, W. C., Wong, M. T., Lavie, A., McSorley, T. and Konrad, M.(2012). Balance of human choline kinase isoforms is critical for cell cycleregulation: implications for the development of choline kinase-targeted cancertherapy. FEBS J. 279, 1915-1928.

Iorio, E., Mezzanzanica, D., Alberti, P., Spadaro, F., Ramoni, C., D’Ascenzo, S.,Millimaggi, D., Pavan, A., Dolo, V., Canevari, S. et al. (2005). Alterations ofcholine phospholipid metabolism in ovarian tumor progression. Cancer Res. 65,9369-9376.

Jan, Y.-Y., Yeh, T.-S., Yeh, J.-N., Yang, H.-R. and Chen, M.-F. (2004). Expressionof epidermal growth factor receptor, apomucins, matrix metalloproteinases, andp53 in rat and human cholangiocarcinoma: appraisal of an animal model ofcholangiocarcinoma. Ann. Surg. 240, 89-94.

Khan, S. A., Thomas, H. C., Davidson, B. R. and Taylor-Robinson, S. D. (2005).Cholangiocarcinoma. Lancet 366, 1303-1314.

Ma, X. M. and Blenis, J. (2009). Molecular mechanisms of mTOR-mediatedtranslational control. Nat. Rev. Mol. Cell Biol. 10, 307-318.

Mabuchi, S., Altomare, D. A., Connolly, D. C., Klein-Szanto, A., Litwin, S.,Hoelzle, M. K., Hensley, H. H., Hamilton, T. C. and Testa, J. R. (2007). RAD001(Everolimus) delays tumor onset and progression in a transgenic mouse model ofovarian cancer. Cancer Res. 67, 2408-2413.

Majumder, P. K., Febbo, P. G., Bikoff, R., Berger, R., Xue, Q., McMahon, L. M.,Manola, J., Brugarolas, J., McDonnell, T. J., Golub, T. R. et al. (2004). mTOR

inhibition reverses Akt-dependent prostate intraepithelial neoplasia throughregulation of apoptotic and HIF-1-dependent pathways. Nat. Med. 10, 594-601.

Massarelli, E., Varella-Garcia, M., Tang, X., Xavier, A. C., Ozburn, N. C., Liu,D. D., Bekele, B. N., Herbst, R. S. andWistuba, I. I. (2007). KRASmutation is animportant predictor of resistance to therapy with epidermal growth factor receptortyrosine kinase inhibitors in non-small-cell lung cancer. Clin. Cancer Res. 13,2890-2896.

Mondesire, W. H., Jian, W., Zhang, H., Ensor, J., Hung, M. C., Mills, G. B. andMeric-Bernstam, F. (2004). Targeting mammalian target of rapamycinsynergistically enhances chemotherapy-induced cytotoxicity in breast cancercells. Clin. Cancer Res. 10, 7031-7042.

Motzer, R. J., Escudier, B., Oudard, S., Hutson, T. E., Porta, C., Bracarda, S.,Grunwald, V., Thompson, J. A., Figlin, R. A., Hollaender, N. et al. (2008).Efficacy of everolimus in advanced renal cell carcinoma: a double-blind,randomised, placebo-controlled phase III trial. Lancet 372, 449-456.

Paule, B., Herelle, M.-O., Rage, E., Ducreux, M., Adam, R., Guettier, C. andBralet, M.-P. (2007). Cetuximab plus gemcitabine-oxaliplatin (GEMOX) inpatients with refractory advanced intrahepatic cholangiocarcinomas. Oncology72, 105-110.

Piguet, A.-C., Semela, D., Keogh, A., Wilkens, L., Stroka, D., Stoupis, C., St-Pierre, M. V. and Dufour, J.-F. (2008). Inhibition of mTOR in combination withdoxorubicin in an experimental model of hepatocellular carcinoma. J. Hepatol. 49,78-87.

Que, F. G., Phan, V. A., Phan, V. H., Celli, A., Batts, K., LaRusso, N. F. andGores,G. J. (1999). Cholangiocarcinomas express Fas ligand and disable the Fasreceptor. Hepatology 30, 1398-1404.

Rajagopalan, V., Daines, W. P., Grossbard, M. L. and Kozuch, P. (2004).Gallbladder and biliary tract carcinoma: a comprehensive update, Part 1.Oncology (Williston Park) 18, 889-896.

Ramırez deMolina, A., Penalva, V., Lucas, L. and Lacal, J. C. (2002a). Regulationof choline kinase activity by Ras proteins involves Ral-GDS and PI3K. Oncogene21, 937-946.

Ramırez deMolina, A., Rodrıguez-Gonzalez, A., Gutierrez, R., Martınez-Pin eiro,L., Sanchez, J., Bonilla, F., Rosell, R. and Lacal, J. C. (2002b). Overexpressionof choline kinase is a frequent feature in human tumor-derived cell lines and inlung, prostate, and colorectal human cancers. Biochem. Biophys. Res. Commun.296, 580-583.

Ramırez de Molina, A., Gallego-Ortega, D., Sarmentero-Estrada, J., Lagares,D., Gomez Del Pulgar, T., Bandres, E., Garcıa-Foncillas, J. and Lacal, J. C.(2008). Choline kinase as a link connecting phospholipid metabolism and cellcycle regulation: implications in cancer therapy. Int. J. Biochem. Cell Biol. 40,1753-1763.

Salvesen, G. S. and Duckett, C. S. (2002). IAP proteins: blocking the road todeath’s door. Nat. Rev. Mol. Cell Biol. 3, 401-410.

Schneider, P. M., Baldus, S. E., Metzger, R., Kocher, M., Bongartz, R.,Bollschweiler, E., Schaefer, H., Thiele, J., Dienes, H. P., Mueller, R. P. et al.(2005). Histomorphologic tumor regression and lymph node metastasesdetermine prognosis following neoadjuvant radiochemotherapy for esophagealcancer: implications for response classification. Ann. Surg. 242, 684-692.

Shimonishi, T., Isse, K., Shibata, F., Aburatani, I., Tsuneyama, K., Sabit, H.,Harada, K., Miyazaki, K. andNakanuma, Y. (2000). Up-regulation of fas ligand atearly stages and down-regulation of Fas at progressed stages of intrahepaticcholangiocarcinoma reflect evasion from immune surveillance. Hepatology 32,761-769.

Shrikhande, S. V., Kleeff, J., Kayed, H., Keleg, S., Reiser, C., Giese, T., Buchler,M. W., Esposito, I. and Friess, H. (2006). Silencing of X-linked inhibitor ofapoptosis (XIAP) decreases gemcitabine resistance of pancreatic cancer cells.Anticancer Res. 26, 3265-3273.

Valle, J. W., Wasan, H., Johnson, P., Jones, E., Dixon, L., Swindell, R., Baka, S.,Maraveyas, A., Corrie, P., Falk, S. et al. (2009). Gemcitabine alone or incombination with cisplatin in patients with advanced or metastaticcholangiocarcinomas or other biliary tract tumours: a multicentre randomisedphase II study - The UK ABC-01 Study. Br. J. Cancer 101, 621-627.

Valle, J. W., Wasan, H., Palmer, D. H., Cunningham, D., Anthoney, A.,Maraveyas, A., Madhusudan, S., Iveson, T., Hughes, S., Pereira, S. P. et al.(2010). Cisplatin plus gemcitabine versus gemcitabine for biliary tract cancer.N. Engl. J. Med. 362, 1273-1281.

Venkatesh, H. S., Chaumeil, M. M., Ward, C. S., Haas-Kogan, D. A., James, C. D.and Ronen, S. M. (2012). Reduced phosphocholine and hyperpolarized lactateprovide magnetic resonance biomarkers of PI3K/Akt/mTOR inhibition inglioblastoma. Neuro Oncol. 14, 315-325.

Yalcin, A., Clem, B., Makoni, S., Clem, A., Nelson, K., Thornburg, J., Siow, D.,Lane, A. N., Brock, S. E., Goswami, U. et al. (2010). Selective inhibition ofcholine kinase simultaneously attenuates MAPK and PI3K/AKT signaling.Oncogene 29, 139-149.

Yan, H., Frost, P., Shi, Y., Hoang, B., Sharma, S., Fisher, M., Gera, J. andLichtenstein, A. (2006). Mechanism by which mammalian target of rapamycininhibitors sensitize multiple myeloma cells to dexamethasone-induced apoptosis.Cancer Res. 66, 2305-2313.

8

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm033050. doi:10.1242/dmm.033050

Disea

seModels&Mechan

isms

Yeh, T.-S., Tseng, J.-H., Chen, T.-C., Liu, N.-J., Chiu, C.-T., Jan, Y.-Y.and Chen, M.-F. (2005). Characterization of intrahepatic cholangiocarcinomaof the intraductal growth-type and its precursor lesions. Hepatology 42,657-664.

Yeh, C.-N., Lin, K.-J., Hsiao, I.-T., Yen, T.-C., Chen, T.-W., Jan, Y.-Y., Chung, Y.-H., Lin, C.-F. and Chen, M.-F. (2008). Animal PET for thioacetamide-induced rat

cholangiocarcinoma: a novel and reliable platform. Mol. Imaging Biol. 10,209-216.

Zhang,W., Zhu, J., Efferson, C. L., Ware, C., Tammam, J., Angagaw,M., Laskey,J., Bettano, K. A., Kasibhatla, S., Reilly, J. F. et al. (2009). Inhibition of tumorgrowth progression by antiandrogens and mTOR inhibitor in a Pten-deficientmouse model of prostate cancer. Cancer Res. 69, 7466-7472.

9

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm033050. doi:10.1242/dmm.033050

Disea

seModels&Mechan

isms