Synergistic antiproliferative effects of an mTOR inhibitor (rad001) plus gemcitabine on cholangiocarcinoma by decreasing choline kinase activity

ABSTRACT Although gemcitabine plus cisplatin is the gold standard chemotherapy regimen for advanced cholangiocarcinoma, the response rate has been disappointing. This study aims to investigate a novel therapeutic regimen [gemcitabine plus everolimus (rad001), an mTOR inhibitor] for cholangiocarcinoma. Gemcitabine, oxaliplatin, cetuximab and rad001 in various combinations were first evaluated in vitro using six cholangiocarcinoma cell lines. In vivo therapeutic efficacies of gemcitabine and rad001 alone and their combination were further evaluated using a xenograft mouse model and a chemically induced orthotopic cholangiocarcinoma rat model. In the in vitro study, gemcitabine plus rad001 exerted a synergistic therapeutic effect on the cholangiocarcinoma cells, irrespective of the KRAS mutation status. In the xenograft study, gemcitabine plus rad001 showed the best therapeutic effect on tumor volume change, and was associated with increased caspase-3 expression, decreased eIF4E expression, as well as overexpression of both death receptor- and mitochondrial apoptotic pathway-related genes. In a chemically induced cholangiocarcinoma-afflicted rat model, the gemcitabine plus rad001 treatment suppressed tumor glycolysis as measured by 18F-fludeoxyglucose micro-positron emission tomography. Also, increased intratumoral free choline, decreased glycerophosphocholine and nearly undetectable phosphocholine levels were demonstrated by proton nuclear magnetic resonance, supported by results of decreased choline kinase expression in western blotting. We concluded that gemcitabine plus rad001 has a synergistic antiproliferative effect on cholangiocarcinoma, irrespective of the KRAS mutation status. The antitumor effect is associated with activation of both death receptor and mitochondrial pathways, as well as the downregulation of choline kinase activity, resulting in a characteristic change in choline metabolism. Summary: Rad001 plus gemcitabine exerts a synergistic antitumor effect on cholangiocarcinoma irrespective of KRAS mutation status, with underlying mechanisms involving activation of the death receptor, mitochondrial pathways and downregulated choline kinase activity.


INTRODUCTION
Cholangiocarcinoma is a malignant biliary tract tumor that carries a very poor prognosis (de Groen et al., 1999;Khan et al., 2005). Although uncommon worldwide, its incidence reaches 87 per 100,000 in endemic areas, including Israel, Japan and South Eastern Asian countries such as Taiwan (Rajagopalan et al., 2004;Yeh et al., 2005). Surgical resection is the only treatment modality that offers a potential cure. Unfortunately, resection rates are low at the initial diagnosis because of early intrahepatic spread, nodal involvement, vascular invasion and distant metastasis. Other therapeutic strategies, including chemotherapy, radiation therapy, transplantation and photodynamic therapy, have been developed, but none has shown significant survival benefit in patients with advanced cholangiocarcinoma (de Groen et al., 1999;Khan et al., 2005). Gemcitabine, with or without platinum compounds, is currently the most effective chemotherapy regimen for advanced biliary tract cancer, although the response rate is only ∼20% (André et al., 2004;Eckel and Schmid, 2007;Valle et al., 2010Valle et al., , 2009). In addition, epidermal growth factor receptor (EGFR) is activated by bile acids and has a role in the carcinogenesis of cholangiocarcinoma through the induction of cyclooxygenase-2 expression via an MAPK protein cascade. Accordingly, neutralizing antibodies that target the extracellular domain of EGFR (e.g. cetuximab) have sporadically shown clinical efficacy in advanced cholangiocarcinoma (Paule et al., 2007).
Phosphocholine (PC) is both a precursor and a breakdown product of phosphatidylcholine, the most abundant phospholipid in the biological membranes. Increased levels of PC and total choline, as detected by 1 H or 31 P magnetic resonance spectroscopy (MRS), are characteristics 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 that specific changes in the levels of the phospholipid precursors and catabolites in the context of solid cancers might be a useful biochemical indicator of tumor progression and/or treatment response.
This study aimed to evaluate the therapeutic effects of rad001 and gemcitabine on cholangiocarcinoma, and elucidate the molecular and metabolic mechanisms underlying their antitumor activity.

Selection of optimal drug regimens for cholangiocarcinoma cells
The corresponding proliferation indices of the six cholangiocarcinoma cell lines treated with standard IC 50 dosages of various therapeutic regimens are shown in Fig. 1A. Gemcitabine exhibited superior therapeutic efficacy against all cholangiocarcinoma cell lines as a single-agent therapy, compared with oxaliplatin, cetuximab and rad001. Gemcitabine plus rad001 was the most promising combination therapeutic regimen, in comparison with gemcitabine plus oxaliplatin and gemcitabine plus oxaliplatin plus cetuximab. The median effects analysis demonstrated that rad001 and gemcitabine had a synergistic effect on both KRAS mutation (HuCCT1, RBE) and wild-type (TFK-1, YSCCC) cell lines with a combination index (CI)<1 (Fig. 1B).

Rationale of mTOR inhibition for treatment of cholangiocarcinoma
To 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. In the in vitro analyses, phosphorylation of p70S6K (RPS6KB1), but not p38MAPK (MAPK14), was significantly decreased in the Rad and Gem+Rad groups in both HuCCT1and TFK-1 cells, indicating that Gem and Gem+Rad specifically inhibited the downstream mTOR signaling irrespective of the KRAS mutation status. Gem might not synergistically enhance the effect of Rad to inhibit mTOR (Fig. 1C). Regarding cell cycling, a smaller percentage of HuCCT1 cells entered the G2/M phase in the Gem+Rad group compared with the 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 Gem group (30.1% versus 4.4%, P<0.01; Fig. 1E).

Evaluation of therapeutic response using a xenograft model
Xenograft experiments were conducted to evaluate the measurable volumetric changes of the cholangiocarcinomas treated with various drug regimens. The representative therapeutic response of the HuCCT1 xenograft model after various treatments for 3 weeks is shown in Fig. 2A. The therapeutic responses of HuCCT1 and TFK-1 xenograft models treated for 3 weeks were analyzed (Fig. 2B). Gem+Rad regime conferred the most promising therapeutic response in both HuCCT1 and TFK-1 xenograft models. Representative images of caspase-3 and eIF4E expression in the HuCCT1 xenograft are shown in Fig. 2C. The expression of caspase-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 the control, 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 rad001 combination therapy
Based on the xenograft model, the percentages of FAS-positive HuCCT1 cells in the control, Gem, Rad and Gem+Rad groups, as detected by flow cytometry, were 13.7%, 30.6%, 23.6% and 44.5%, respectively (P<0.01), and the intensity of FAS-positive HuCCT1 cells was 7.2, 9.5, 8.4 and 10.5, respectively (P<0.01; Fig. 3A). The expression of six apoptotic genes, including FAS, CASP8, BID, APAF1, XIAP and CASP3, derived from the treated HuCCT1 xenograft using quantitative real-time polymerase chain reaction (qPCR), is shown in Fig. 3B. The expression of these six genes was modestly increased in the Gem and Rad groups compared with the control group, whereas the expression of all six genes was significantly increased in the Gem+Rad group compared with the other groups.
Metabolic response of orthotopic cholangiocarcinoma rat model as detected by 18 F-FDG microPET and 1 H NMR The representative in vivo metabolic response of cholangiocarcinomaafflicted rats to various treatments as detected by 18 F-fludeoxyglucose micro-positron emission tomography ( 18 F-FDG microPET) is shown in Fig. 4A. The Gem+Rad group had the most significant metabolic response after two cycles of treatment (Fig. 4B). The expression of Glut-1 (Slc2a1), Hk2, Hif1a and Vegf (Vegfa) mRNAs of the corresponding surgical specimens as detected by qPCR is shown in Fig. 4C. Of these, the expression of Hk2 and Glut-1 mRNA in the Gem +Rad group was reduced by half in comparison with the control group. The difference in the expression of these mRNAs correlated with the standardized uptake value (SUV) changes in FDG microPET. The tumor response in the cholangiocarcinoma-afflicted rats was confirmed by histopathological examination. Results derived from the histomorphological score and Ki67 (Mki67) labeling index of the corresponding surgical specimen also supported the findings in the FDG microPET study ( Table 2).
The alteration of choline-associated metabolites in the treated cholangiocarcinoma-afflicted rats was analyzed by 1 H nuclear magnetic resonance (NMR), with a particular emphasis on the free choline, PC and glycerophosphocholine (GPC) levels ( Fig. 5A). A significant increase in the free choline level of the Gem+Rad group was associated with dramatic disappearance of PC and a significantly reduced level of GPC (Fig. 5B). In contrast, the Gem group exhibited a significant increase in the free choline level without significant changes in PC and GPC levels. No significant change in the choline-associated metabolites was observed in the Rad group. Choline kinase (CK; CHKA), an enzyme that catalyzes the phosphorylation of free choline using ATP to produce PC and plays a rate-limiting regulatory role in the phosphatidylcholine biosynthesis, was assayed to determine the underlying mechanism that was responsible for the observed changes in the PC levels. Of the four treatment groups, the CK level detected by western blotting was lowest in the Gem+Rad group (Fig. 5C,D).

DISCUSSION
The presence of the tyrosine kinase-AKT-mTOR axis in the carcinogenesis of cholangiocarcinoma (Chung et al., 2009) has provided us with an opportunity to evaluate the therapeutic role of an mTOR inhibitor, rad001, in the management of this disease. We have shown in this study that rad001 treatment resulted in reduction of expression of the downstream molecules of mTOR, such as p70S6K-1, but not p38MAPK. Furthermore, we demonstrated, for the first time, that gemcitabine plus rad001 exerted a synergistic antitumor effect on cholangiocarcinoma, independent of the KRAS mutation status. In contrast, the EGFR monoclonal antibody, cetuximab, had a modest therapeutic effect on the KRAS wild-type cholangiocarcinoma xenograft (TFK-1), but not the KRAS mutated xenograft (HuCCT1). This finding is consistent with previous clinical 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 was macroscopically evaluated and confirmed by measuring changes in the tumor volume in the xenograft model, and assessing the metabolic response in the orthotopic rat cholangiocarcinoma model using 18 F-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 SUV identified in 18 F-FDG microPET.
The available evidence in the literature suggests that gemcitabine primarily acts on an intrinsic (mitochondrial) apoptotic pathway in NSCLCs and pancreatic cancer to achieve its pharmacological effect. Gemcitabine-mediated activation of caspase-8 is a late and mitochondrial-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 synthesized FASL and FAS and that FASL expressed by these cells induced host lymphocyte cell death (Que et al., 1999). Shimonishi et al. further demonstrated upregulation of FASL in the early stages and downregulation of FAS in the advanced stages of cholangiocarcinoma, the latter of which was responsible for evasion from host immunity and disease progression (Shimonishi et al., 2000). To understand the mechanism by which rad001 and gemcitabine acted synergistically to enhance apoptosis in the cholangiocarcinoma cells, we examined six key apoptotic genes that were involved in the extrinsic and intrinsic cell death pathways. These included FAS, CASP8, BID, APAF1, XIAP and CASP3. Among these six genes, the former two represent the death receptor (extrinsic) pathway, whereas the 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 of apoptosis, is thought to directly inhibit certain caspases such as caspase-3, caspase-7 and caspase-9. Shrikhande et al. showed that XIAP was overexpressed in pancreatic cancer and contributed to chemoresistance (Shrikhande et al., 2006). Our data revealed that both death receptor-related (FAS and CASP8) and mitochondrial pathwayrelated genes (APAF1 and CASP3) in the Gem+Rad group were increased 10-fold in comparison with those of the Gem group, indicating that both death receptor and mitochondrial pathways were recruited and co-activated by the combination therapy. It is noteworthy that XIAP was simultaneously increased by up to 10-fold in the Gem +Rad group, along with the pro-apoptotic genes, in comparison with the Gem group. Our findings suggested that XIAP plays a major role in arresting the apoptotic process and potentially contributing to drug resistance. Early detection of response to therapy is desirable but not easily achieved using conventional imaging techniques such as computed tomography or magnetic resonance imaging. Biological cancer treatment might be associated with stagnation of cellular proliferation or cellular clonal changes rather than obvious tumor shrinkage in the early stage of the therapeutic response. Cancer metabolism has emerged in recent studies as a powerful link between deregulated signaling pathways and altered cellular metabolism. Specific endogenous metabolites and metabolic fluxes within cells and tissues in vivo and in vitro can be readily identified using 31 P, 1 H or 13 C MRS. Of them, derangement in the choline metabolism during carcinogenesis and treatment has been increasingly recognized in various solid cancers (Ackerstaff et al., 2003). CK catalyzes the phosphorylation of choline to yield PC in the biosynthesis of phosphatidylcholine, a process known as the Kennedy pathway (Ackerstaff et al., 2003;Ramírez de Molina et al., 2008). The enzyme is modulated by RAS and interacts with HIF1A (Ramírez de Molina et al., 2002a). It might also play a role in the cell cycle regulation (Gruber et al., 2012), MAPK and PI3K (PIK3CA)/ AKT protein signaling pathways (Yalcin et al., 2010), as well as acting as a potential biomarker for various solid cancers (Ramírez de Molina et al., 2002b). Specific inhibitors against CK are to be anticipated soon in clinical trials as these agents have shown promising antitumor effects in preclinical studies (Ackerstaff et al., 2003;Al-Saffar et al., 2010;Ramírez de Molina et al., 2008, 2002a. In the present study, we have implemented 1 H NMR to detect alterations in choline-associated metabolites in the treated cholangiocarcinoma-afflicted rats. Downregulated CK activity, presenting as an accumulation of free choline and nearly undetectable levels of PC in the Gem+Rad group, was confirmed by significantly decreased expression of CK in western blotting. Such downregulation of CK following mTOR inhibition was associated with HIF1A and PI3K/AKT/mTOR pathways, with the downregulation of Hif1a mRNA correlating with the respective CK levels in the Gem+Rad, Gem and control groups. In line with our study results, Glunde et al. demonstrated a similar process of CK regulation by HIF1A signaling in human prostate cancer (Glunde et al., 2008). Furthermore, using a PI3K inhibitor, PI-103, Al-Saffar et al. were able to demonstrate downregulation of CK levels in prostate and colon cancers, with a corresponding decrease in PC levels (Al-Saffar et al., 2010). Again, by silencing CK using small interfering RNA (siRNA), Yalcin et al. showed that PC was reduced through the PI3K/AKT/mTOR pathways, leading to the inhibition of cervical cancer cell growth (Yalcin et al., 2010).
The change in GPC levels after biological agent treatment is another interesting issue. Al-Saffar et al. reported that blockade of PI3K with LY294002 (a relatively less potent, nonselective inhibitor of PI3K) was associated with a decrease in PC levels as well as an elevation in GPC content in human breast cancer cells (Beloueche-Babari et al., 2006). In contrast, the authors noted that both PC and GPC levels were below the detection limit of MRS when the HCT116 colon cancer cells were treated instead with PI-103, a potent selective inhibitor of class 1 PI3K (Al-Saffar et al., 2010). Accordingly, they concluded that the contradictory response of the GPC levels following LY294002 and PI-103 treatments was probably related to other off-target effects of the former inhibitor. In our study, the characteristic profile of the choline metabolites in the cholangiocarcinoma cells treated by gemcitabine plus rad001 was most likely a result of a potent and selective targeting against the PI3K/AKT/mTOR signaling pathway. As there is an increase in choline and a decrease in its downstream metabolite, PC, we expect to observe an accumulation of choline signal on 11 C-choline PET when the treatment is effective. A future in vivo imaging study using 11 C-choline PET or magnetic resonance spectroscopy imaging would better facilitate longitudinal monitoring of the relevant biological response.
In conclusion, our preclinical study shows that gemcitabine plus rad001 exerts a synergistic antitumor effect on cholangiocarcinoma irrespective of the KRAS mutation status. The underlying mechanisms of antitumor activity are associated with activation of the death receptor and mitochondrial pathways, as well as the downregulation of CK activity, resulting in a specific change in choline metabolism.

Cholangiocarcinoma cell lines
Six cholangiocarcinoma cell lines used in this study (HuCCT1, SSP-25, RBE, YSCCC, TGBC-24TKB and TFK-1) were purchased from RIKEN Cell Bank (Tsukuba, Japan). The HuCCT1, SSP-25 and RBE cell lines had a KRAS mutation, whereas the remaining three cell lines were KRAS wild type. All in vitro studies were performed at least in triplicate.

Flow cytometry
Both adherent and floating cholangiocarcinoma cells were collected 72 h after various drug treatments for cell cycle analysis. Cells were passed through a fluorescence activated cell sorter (FACS Caliber, BD Biosciences, USA), and data were acquired using CellQuest (BD Biosciences) software. The cell cycle was modeled using ModFit software (Venty Software, USA). The separation of cells into G0/G1, S and G2/M phases was based on linear fluorescence intensity after staining with propidium iodide. Cells were labeled with the Annexin V-FITC Apoptosis Kit (BD Biosciences-Clontech) and analyzed on a FACS Caliber to detect apoptosis. The cell surface expression of FAS after various drug treatments for 16 h was assessed. Indirect immunofluorescence staining for FAS was performed with an IgG1 mouse anti-FAS primary monoclonal antibody (DX2 clone, BD Biosciences) and a fluorescein isothiocyanate (FITC)-conjugated secondary antibody (goat anti-mouse Ig) (BD Biosciences). The percentage and intensity of FASpositive cells were analyzed by the FACS Caliber.

Bio-Plex phosphoprotein assay
Cholangiocarcinoma cells were treated with various drug regimens, and protein 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 the manufacturer's protocols.

Xenograft mice model
To initiate tumor xenograft, cholangiocarcinoma cells, HuCCT1 and TFK-1, were implanted subcutaneously (2×10 6 cells) in the flank of the Fox Chase severe combined immunodeficiency (SCID) mice (BioLASCO, Taiwan). Six therapeutic regimens were provided, as follows: gemcitabine 200 mg/kg by intraperitoneal injection (I.P.)+oxaliplatin 5 mg/kg I.P. every week (GEMOX); gemcitabine 200 mg/kg I.P.+oxaliplatin 5 mg/kg I.P.+cetuximab 2 mg/kg I.P. every week (GEMOX+cetuximab); gemcitabine 200 mg/kg I.P. every week (GEM); rad001 5 mg/kg orally twice every week (Rad); gemcitabine, 200 mg/kg I.P. every week+rad001 5 mg/kg orally twice every week (GEM+Rad); and saline with the same volume I.P. every week (control group). Tumor size did not exceed 1000 mm 3 (∼4% of mouse body weight) in compliance with animal welfare regulations. At the end of the third week, the xenografts were measured and retrieved. The surgical specimens were divided into two portions; half was kept frozen and stored at −80°C and the remaining half was prepared as a paraffin block. All animal experiments were performed according to a protocol approved by the Institutional Animal Care and Use Committee, Chang Gung University and Chang Gung Memorial Hospital, Taiwan (IACUC 2012092401). The care and use of experimental animals conformed with the regulatory standards.

Immunohistochemical staining
Formalin-fixed, paraffin-embedded tissues were cut into 4-µm sections and mounted on glue-coated slides. A modification of the avidin-biotinperoxidase complex immunohistochemical method was performed. Various primary antibodies [human caspase-3 and eukaryotic initiating factor 4E (eIF4E); and rat Ki67] were applied. The expression of eIF4E was assessed  using the product of two scores (intensity×% positive), whereas the expression of caspase-3 was represented as positively stained cells per HPF. The Ki67 labeling index was calculated as positively stained cells per 100 cells counted×100%. Histopathological assessment of tumor response was performed by both histomorphological regression analysis (Schneider et al., 2005) and Ki67 labeling index. The extent of histomorphological regression was divided into four categories: grade 1, 75-100% viable residual tumor cells; grade 2, 50-75% viable residual tumor cells; grade 3, 25-50% viable residual tumor cells; and grade 4, 0-25% viable residual tumor cells.

Thioacetamide-induced cholangiocarcinoma-afflicted rat model
Male Sprague-Dawley rats weighing 180-200 g were housed in a temperature-and humidity-controlled environment with free access to water and rat chow. Cholangiocarcinoma was induced by oral administration of thioacetamide (0.03% w/w in water) for 24 weeks, as previously described (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 orally twice 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 FDG microPET pretreatment, and at 2 and 4 weeks after treatment, before being sacrificed for histological confirmation.

18
F-FDG microPET imaging to measure metabolic response Rats had full access to drinking water at all times and were fasted overnight for 8 h before radiotracer injection. A dose of 18.5 MBq (0.5 mCi) 18 F-FDG was administered and the liver region was scanned for 120 min continuously after intravenous radiotracer injection. Image analysis was carried out at an image analysis workstation (PMOD Technologies, Switzerland). Regions of interest 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 the highest T/L ratio was reached. SUVs were used to quantify tracer uptake according to the following formula: SUV = Decay corrected tissue activity(Bq/ml) Injected dose(Bq)/Body weight(g) The SUVmax and SUVmean are maximal and mean SUV, respectively, within the voxels of interest (VOIs). Tumor counts and FDG uptake values were obtained in all animals (Yeh et al., 2008).

Quantification of intratumoral metabolites by NMR
Thioacetamide-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). Watersoluble extracts were freeze-dried and reconstituted in 700 µl deuterated water (D 2 O, Sigma-Aldrich, USA), and the extracts (500 µl) were then placed in 5 mm NMR tubes. Fifty microliters of 0.75% sodium 3trimethylsilyl-2,2,3,3-tetradeuteropropionate (TSP) in D 2 O (Sigma-Aldrich) was added to the samples for chemical shift calibration and quantification. 1 H NMR measurement was performed on a 600 MHz spectrometer (Bruker, Germany): 7500 Hz spectral width, 16,384 time domain points, 128 scans, temperature 298 K, acquisition time ∼5 min. The water resonance was suppressed by a gated irradiation centered on the water frequency. Spectral processing was carried out using the Bruker Topspin-2 software package to quantify the metabolites.

Statistical analysis
All continuous variables are expressed as the mean (s.d.). Statistical analysis for continuous variables was performed using Student's t-test or one-way ANOVA test where appropriate. Intergroup comparisons of CK were otherwise performed using a box plot, where the data were expressed as the median value. A P-value <0.05 was considered statistically significant.
This article is part of a special subject collection 'Cancer Metabolism: models, mechanisms and targets', which was launched in a dedicated issue guest edited by Almut Schulze and Mariia Yuneva. See related articles in this collection at http:// dmm.biologists.org/collection/cancermetabolism.