1450.Full-Text.Pdf

Published OnlineFirst September 23, 2016; DOI: 10.1158/1078-0432.CCR-16-0871

Personalized Medicine and Imaging Clinical Cancer Research Identiﬁcation of Keratin 19–Positive Cancer Stem Cells Associating Human Hepatocellular Carcinoma Using 18F-Fluorodeoxyglucose Positron Emission Tomography Takayuki Kawai1,2, Kentaro Yasuchika1, Satoru Seo1, Tatsuya Higashi3, Takamichi Ishii1, Yuya Miyauchi1, Hidenobu Kojima1, Ryoya Yamaoka1, Hokahiro Katayama1, Elena Yukie Yoshitoshi1,4, Satoshi Ogiso1, Sadahiko Kita1, Katsutaro Yasuda1,4, Ken Fukumitsu1, Yuji Nakamoto3, Etsuro Hatano1, and Shinji Uemoto1

Abstract

Purpose: The current lack of tools for easy assessment of cancer 18F-FDG uptake and GLUT1 expression were examined in FACS- þ stem cells (CSC) prevents the development of therapeutic strat- isolated K19 /K19 cells. egies for hepatocellular carcinoma (HCC). We previously Results: In hepatocellular carcinoma patients, K19 expression reported that keratin 19 (K19) is a novel HCC-CSC marker and was significantly correlated with GLUT1 expression and FDG that PET with 18F-fluorodeoxyglucose (18F-FDG) is an effective accumulation. ROC analyses revealed that among preoperative method for predicting postoperative outcome in hepatocellular clinical factors, TNR was the most sensitive indicator of K19 þ carcinoma. Herein, we examined whether K19 HCC-CSCs can be expression in hepatocellular carcinoma tumors. In hepatocellular þ tracked using 18F-FDG-PET. carcinoma cells, FACS-isolated K19 cells displayed significantly Experimental Design: K19 and glucose transporter-1 (GLUT1) higher 18F-FDG uptake than K19 cells. Moreover, gain/loss-of- expression was evaluated by IHC in 98 hepatocellular carcinoma function experiments confirmed that K19 regulates 18F-FDG patients who underwent 18F-FDG-PET scans before primary uptake through TGFb/Smad signaling, including Sp1 and its tumor resection. Standardized uptake values (SUV) for primary downstream target GLUT1. tumors and tumor-to-nontumor SUV ratios (TNR) were calculat- Conclusions:18F-FDG-PET can be used to predict K19 expres- ed using FDG accumulation levels, and values were compared sion in hepatocellular carcinoma and should thereby aid in þ þ among K19 /K19 patients. Using hepatocellular carcinoma the development of novel therapeutic strategies targeting K19 cell lines encoding with a K19 promoter–driven enhanced GFP, HCC-CSCs. Clin Cancer Res; 23(6); 1450–60. 2016 AACR.

Introduction cancer stem cells (CSC) is essential. CSCs possess the ability to self-renew and to generate heterogeneous populations of cancer Hepatocellular carcinoma (HCC), which accounts for a major- cells that exhibit high motility and proliferation rates (2, 3). These ity of liver cancers, is the sixth most common cancer and the stem cell–like features of CSCs contribute to rapid tumor growth, second leading cause of cancer-related deaths worldwide (1). the resistance of tumors to chemotherapy/radiotherapy, and the Although various therapies have been developed or are currently epithelial–mesenchymal transition (EMT; refs. 4–6). In hepato- being developed, the prognosis for patients with hepatocellular cellular carcinoma, various molecules expressed during liver carcinoma is far from satisfactory. To develop the new phase of development, such as epithelial cell adhesion molecule (EpCAM), hepatocellular carcinoma treatment, the clinical application of cluster of differentiation (CD) 90, CD133, and Sal-like protein 4 (SALL4), have been identified as CSC markers (7–11). Inde- 1Department of Surgery, Graduate School of Medicine, Kyoto University, Kyoto, pendently, we previously reported that keratin 19 (K19), a hepatic Japan. 2Department of Hepatobiliary Surgery and Liver Transplantation, Pitie- progenitor cell marker, is a novel HCC-CSC marker associated Salpetri ere Hospital, University of Pierre and Marie Curie (UPMC), Paris, France. with EMT and TGFb/Smad signaling. Furthermore, we proposed þ 3Department of Diagnostic Radiology, Graduate School of Medicine, Kyoto that K19 HCC-CSCs could be new therapeutic targets of TGFb 4 University, Kyoto, Japan. Center for iPS Cell Research and Application (CiRA), receptor 1 inhibitor in hepatocellular carcinoma (12). However, Kyoto University, Kyoto, Japan. the lack of a non- or less-invasive pretreatment evaluation method Note: Supplementary data for this article are available at Clinical Cancer for identifying HCC-CSCs has hindered our ability to fully predict Research Online (http://clincancerres.aacrjournals.org/). patients' outcomes and evaluate the therapeutic efficacy in Corresponding Author: Kentaro Yasuchika, Department of Surgery, Graduate patients with hepatocellular carcinoma. School of Medicine, Kyoto University, 54 Kawahara-cho, Shogoin, Sakyo-ku, On the other hand, PET with 18F-fluorodeoxyglucose (18F- Kyoto 606-8507, Japan. Phone: 81-75-751-3242; Fax: 81-75-751-4246; E-mail: FDG) analysis, which is based on the increased glucose uptake [email protected] in cancer cells, has been shown to be an effective method for doi: 10.1158/1078-0432.CCR-16-0871 diagnosis or assessment of treatment efficacies in various malig- 2016 American Association for Cancer Research. nancies (13–15). Indeed, we previously reported that the

1450 Clin Cancer Res; 23(6) March 15, 2017

Identiﬁcation of K19þ HCC-CSCs Using 18F-FDG-PET

to the 18F-FDG-PET procedure (53 patients), or exhibited high Translational Relevance C-reactive protein (CRP > 5.0 mg/dL) and/or fasting blood Non/less-invasive methods for assessing cancer stem cells sugar (FBS > 150 mg/dL; 28 patients) levels (Supplementary (CSC) have yet to be developed, which has prevented the Fig. S1). In addition, of the 163 patients subjected to 18F-FDG- development of novel therapeutic strategies for hepatocellular PET analysis, 62 were subsequently excluded because they did carcinoma (HCC). We previously reported that keratin 19 not provide consent for the use of surgical tissues, two were (K19) is a novel HCC-CSC marker and that PET with 18F- excluded after diagnosis with cholangiocellular carcinoma by fluorodeoxyglucose (18F-FDG) is useful for predicting post- pathologic examination, and one was excluded due to difficul- operative outcomes in hepatocellular carcinoma. However, ties in pathologic assessment resulting from tumor necrosis. þ the relationship between K19 HCC-CSCs and 18F-FDG-PET Thus, a total of 98 patients were included in this retrospective þ remains unclear. Therefore, we hypothesized that K19 HCC- study (Supplementary Fig. S1). The clinical–pathologic char- CSCs could be tracked in hepatocellular carcinoma via 18F- acteristics of the patients are summarized in Supplementary FDG-PET. Our findings indicate that, among preoperative Table S1. Tumor recurrence was investigated until the death of clinical factors, the tumor-to-nontumor standardized uptake the patient or the end of the study (December 31, 2014). The value ratio is the most sensitive indicator of K19 expression follow-up period from surgery until death or the endpoint of in human hepatocellular carcinoma and that K19 regulates the study ranged from 157 to 2,891 days (mean 1,408 days). 18F-FDG uptake through TGFb/Smad signaling and its down- Written informed consent for the use of resected tissue samples stream target, glucose transporter 1 expression in hepatocel- was obtained from all 98 patients in accordance with the Decla- lular carcinoma cells. 18F-FDG-PET will contribute to the ration of Helsinki, and this study was approved by the Institu- development of new therapeutic strategies focusing on tional Review Committee of our hospital. þ K19 HCC-CSCs. 18F-FDG-PET and image analyses All PET imaging procedures and image analysis were performed as described previously (16–18). Briefly, maximum SUVs were standardized uptake value (SUV) for the primary tumor and the calculated for quantitative analyses of the levels of tumor 18F-FDG tumor-to-nontumor SUV ratio (TNR), as calculated from the uptake as follows: SUV ¼ C (kBq/mL)ID (kBq) 1body weight levels of 18F-FDG accumulation, in hepatocellular carcinoma is (kg) 1, where C represents the tissue activity concentration mea- correlated with the expression of glucose transporter-1 (GLUT1), a sured by PET and ID represents the injected dose. TNR values were key glucose transporter in both normal and cancer tissues (16). In calculated as follows: TNR ¼ tumor SUV/nontumor SUV, where addition, our previous study revealed that TNR is an independent the nontumor SUV is equal to the average of the SUVs at five predictor of postoperative recurrence and overall survival in points in nontumor liver tissues. hepatocellular carcinoma (17, 18). However, the relationship þ 18 between K19 HCC-CSCs and F-FDG-PET has yet to be eluci- IHC dated. Therefore, we hypothesized that K19 expression can be Immunohistochemical analyses were performed as reported predicted by 18F-FDG-PET in hepatocellular carcinoma. þ previously (19, 20). Mouse anti-human K19 (Dako) and rabbit The aims of this study were to demonstrate that K19 CSCs anti-human GLUT1 primary antibodies (sc-7903, Santa Cruz 18 can be tracked via F-FDG-PET in hepatocellular carcinoma Biotechnology Inc.) were used at 1:100 and 1:200 dilution, tumors. For these analyses, the expression of K19 and GLUT1 respectively. K19 and GLUT1 expression levels were semiquanti- was investigated in human hepatocellular carcinoma surgical tatively assessed. Samples were considered positive for a particular fi specimens. Furthermore, we examined the ef cacy of SUV, marker when expression of that marker was observed in greater TNR, and various preoperative clinical factors for detecting than 5% of the tumor cells examined. Each slide was evaluated by K19 expression in hepatocellular carcinoma. Moreover, a trans- two independent investigators (T. Kawai and K. Yasuchika) who gene vector that expressed enhanced GFP (EGFP) under the were blinded to the patient outcomes by anonymizing the sam- control of the human K19 promoter was transfected into four þ ples prior to assessment. hepatocellular carcinoma cell lines to visualize K19 cells as þ HCC-CSCs (12). Subsequently, FACS-isolated K19 /K19 cells were used to explore the underlying mechanism governing the Generation of transgenic hepatocellular carcinoma cell lines regulation of 18F-FDG uptake through TGFb/Smad signaling in The human hepatocellular carcinoma cell lines Huh7, HLF, þ K19 HCC-CSCs. and PLC/PRF/5 were obtained from the ATCC. All cells were authenticated by short tandem repeat profiling analysis before receipt and were propagated for less than 6 months after Materials and Methods resuscitation. The cells were cultured at 37 Cwith5%CO2 in Patients RPMI1640 medium (Invitrogen) supplemented with 10% FBS From January 2007 to December 2011, 316 consecutive (MP Biomedicals), and penicillin/streptomycin (Meiji Seika). patients diagnosed with hepatocellular carcinoma by CT We generated a transgene plasmid vector that expressed EGFP and/or MRI were subjected to curative resection at Kyoto Uni- under the control of the human K19 promoter as described versity Hospital (Kyoto, Japan). Of these, only 163 patients previously (12). In summary, the plasmid vector pHCK-2952, 18 underwent F-FDG-PET during preoperative examination. The which was constructed by inserting the human K19 promoter remaining 153 patients were excluded because they had previ- sequence into the pGL3-Basic (Promega) vector, was kindly ously undergone transarterial chemoembolization and/or provided by Professor Shuichi Kaneko (Kanazawa University, radiofrequency ablation (RFA; 72 patients), did not consent Kanazawa, Japan; ref. 21). The human K19 promoter region was

www.aacrjournals.org Clin Cancer Res; 23(6) March 15, 2017 1451

Kawai et al.

isolated from pHCK-2952 by restriction digestion with the XhoI Hepatocellular carcinoma cells were then transfected with this and HindIII enzymes (Takara Bio) and then ligated to the plasmid K19 expression vector or with mock vector (PS100020, OriGene) EGFP-1 (pEGFP1; BD Biosciences), which was linearized by using Lipofectamine LTX reagent (Invitrogen), according to the digestion with XhoI and HindIII. The transgenic vector was then manufacturer's protocol. K19 expression was significantly upre- introduced into hepatocellular carcinoma cells by transfection gulated by the K19 expression vector (Supplementary Fig. S2B). using Lipofectamine LTX reagent (Invitrogen), according to the For Western blot analysis, hepatocellular carcinoma cells were manufacturer's protocol. Stably transfected cells were selected harvested 48 hours posttransfection. for by cultivating in the presence of 200 mg/mL G418 (Sigma- Aldrich) over 30 days. We confirmed proper transgene insertion Reagents by PCR and immunocytochemistry analyses, as described pre- The TGFbR1 inhibitor LY2157299 was obtained from Axon viously (12). Medchem. The compound was dissolved in 100% DMSO (Sigma) and diluted with RPMI1640 to the desired concentration with a Flow cytometry and single-cell culture analysis final DMSO concentration of less than 0.5%. We prepared the cultured cells as described previously (12, 22, 23). Dead cells were eliminated via 7-aminoactinomycin D 18F-FDG accumulation in hepatocellular carcinoma cells (Beckman Coulter) staining. Single-cell culture analysis was To determine FDG uptake in vitro, freshly sorted hepatocel- then performed as described previously (12, 22, 23). Individual lular carcinoma cells were plated at a density of 2.0 105 cells isolated cells were each sorted into 96-well culture plates using per well in 12-well plates. In the experiments using TGFbR1 a FACSAria device (BD Biosciences), and wells were visualized inhibitor LY2157299, 0.5 mmol/L LY2157299 or DMSO control by light microscopy at 10 to 16 hours after sorting to confirm was added. After 24 hours of incubation, the medium was that each well contained only one cell. After isolation of each replaced by 1 mL of medium containing 555 kBq of 18F-FDG clone, the cells were expanded and subjected to flow cytometry and was added to each well. 18F-FDG was allowed to accumu- analysis. late in cells in the incubator over times ranging from 30 to 120 RT-PCR and qRT-PCR minutes. After incubation for the respective times, the medium The total RNA was extracted with RNeasy Mini Kit (Qiagen) was removed and washed three times with ice-cold PBS; the and RNase-free DNAse (Qiagen). Approximately 1 mgoftotal cells were then dissociated with trypsin and collected into 18 RNA was then reverse transcribed into cDNA using an Omnis- tubes. The F-FDG radioactivity was immediately determined cript Reverse Transcription Kit (Qiagen), according to the using a gamma counter (Cobra II Auto-gamma; Packard), with manufacturer's protocol. Primers were generated for the fol- nonspecific background assessed in identically treated cells that lowing genes: K19, GLUT1, K19 open reading frame, Sp1, and were not incubated. b-actin (Supplementary Table S2). RT-PCR analysis was performed as described previously (24). qPCR and qRT-PCR assays Statistical analyses were performed using SYBR Green PCR Master Mix (Applied Statistical analyses were performed using SPSS version 17.0 Biosystems) on the ABI 7500 system (Applied Biosystems). (SPSS Statistics, Inc.) and GraphPad Prism software Version 5.0 Each target was run in triplicate, and expression levels were (GraphPad Software Inc.). Data are presented as the mean SD of normalized to those of b-actin. three or more independent experiments. Student t tests, Mann– Whitney U tests, Fisher exact tests, or c2 tests, repeated measures Western blot analysis ANOVA, and log-rank tests were used for analyses of statistical Western blot analysis was performed as reported previously significance. (25). Primary antibodies specific to phospho-smad2 (pSmad2; Recurrence-free survival (RFS) and overall survival (OS) after Ser465/467, #3108; Cell Signaling Technology), Smad2 (#5339, the operation were calculated using the Kaplan–Meier method Cell Signaling Technology), Sp1 (ab133596, Abcam), GLUT1 (sc- and analyzed with the log-rank test. Significant variables from 7903, Santa Cruz Biotechnology), and GAPDH (sc-25778, Santa univariate analyses were entered in the multivariate analysis Cruz Biotechnology) were used at 1:1,000 dilutions. HRP-conju- using a Cox regression model with forward stepwise selection. gated bovine anti-rabbit IgG (1:2000; Molecular Probes) second- We plotted ROC curve for SUV, TNR, and preoperative clinical ary antibody was used for detection of pSmad2, Smad2, Sp1, factors and calculated the area under each ROC curve (AUC). GLUT1, and GAPDH. The optimal cut-off values for SUV and TNR were calculated using the maximum sum of sensitivity and specificity, as well as Knockdown and overexpression of K19 the minimum distance to the top left corner of the ROC curve. For K19 knockdown experiments, hepatocellular carcinoma The clinical cut-off value was used for the assessment of cells were transfected with a 10 nmol/L concentration of the K19- preoperative clinical factors tested. Statistical significance was siRNA (#4427037-s7998 or #4427037-s7999; Invitrogen) or defined as P < 0.05. control-siRNA (#4390843; Invitrogen) using Lipofectamine LTX reagent (Invitrogen), according to the manufacturer's protocol. K19 expression was significantly downregulated by both K19 Results siRNAs (Supplementary Fig. S2A). According to the same result K19 and GLUT1 expressions in human hepatocellular acquired with both siRNAs, K19-siRNA (#4427037-s7999) was carcinoma surgical specimens shown as representative data. 18F-FDG accumulation is reflected by GLUT1 expression in For K19 overexpression experiments, human K19 open reading various tumors, including hepatocellular carcinoma (16, 26–28). frame was amplified by RT-PCR and ligated to the CMV6-AC To detect the K19 and GLUT1 expression in human hepatocellular plasmid (PS100020; OriGene) digested with Sgf1 and RsrII. carcinoma clinical samples, 98 surgically resected primary

1452 Clin Cancer Res; 23(6) March 15, 2017 Clinical Cancer Research

Identiﬁcation of K19þ HCC-CSCs Using 18F-FDG-PET

Figure 1. K19 and glucose transporter 1 (GLUT1) expression in human hepatocellular carcinoma (HCC). A, Representative images of K19-expressing hepatocellular carcinoma (left) and GLUT1-expressing hepatocellular carcinoma (right). Scale bar, 100 mm. B, Correlation between K19 and GLUT1 expression (Fisher exact test, , P < 0.01). C, RFS rates according to K19 or GLUT1 expression in the hepatocellular carcinoma tissues [log-rank test; K19, , P < 0.01 (left); GLUT1, P ¼ 0.42 (right)]. D, OS rates according to K19 or GLUT1 expression in the hepatocellular carcinoma tissue [log-rank test; K19, , P < 0.01 (left); GLUT1, P ¼ 0.55 (right)].

www.aacrjournals.org Clin Cancer Res; 23(6) March 15, 2017 1453

Kawai et al.

hepatocellular carcinoma tumors were subjected to immunohis- (SUV, mean ¼ 4.2 2.2; TNR, mean ¼ 1.5 0.8; P < 0.01). þ tochemical analysis. K19 and GLUT1 expression was observed in However, no significant differences were observed between K19 14 of 98 (14%) cases and in 15 of 98 (15%) cases, respectively and K19 patients in preoperative a-fetoprotein (AFP) levels and (Fig. 1A). Notably, there was a significant correlation between K19 the amounts of protein induced by vitamin K antagonists-II þ expression and GLUT1 expression (Fig. 1B; P < 0.01). In K19 (PIVKA-II; Fig. 2B). Notably, ROC analysis revealed that among þ þ hepatocellular carcinoma, the majority of the K19 and GLUT1 the preoperative clinical factors, including AFP and PIVKA-II, TNR cells were found in the invasive front of hepatocellular carcinoma was the most sensitive indicator of K19 expression in hepatocel- þ and in the K19-expressing zone, respectively. In contrast, GLUT1 lular carcinoma tumors (Fig. 2C; Table 2). cells were distributed in various areas in K19 hepatocellular þ carcinoma. EGFP marking of K19 cell populations in heterogeneous þ The K19 patients exhibited significantly lower RFS and OS hepatocellular carcinoma cell lines þ rates, with the median RFS being 183 days for K19 patients and To investigate the underlying mechanism of high FDG accu- þ 1,070 days for K19 patients, and the median OS being 464 days mulation in K19 hepatocellular carcinoma, we generated trans- þ þ for K19 patients and 2,112 days for K19 patients, respectively genic hepatocellular carcinoma cell lines to visualize K19 cells, as (Fig. 1C and D; Table 1). In contrast, no significant differences described previously (12). Initial, our RT-PCR analyses demon- þ were detected between the RFS and OS rates of GLUT1 and strated that the Huh7 and PLC/PRF/5 cell lines expressed K19, GLUT1 patients (RFS, P ¼ 0.42; OS, P ¼ 0.55; Fig. 1C and whereas HLF cells did not (Supplementary Fig. S3A). Histologic D; Table 1). The significance of K19 expression in predicting analyses demonstrated that Huh7 and PLC/PRF/5 cell lines con- postoperative RFS and OS was confirmed by log-rank test and tained both K19-expressing and nonexpressing cells. Subsequent- multivariate analyses (Table 1 and Supplementary Table S3). ly, each of these three cell lines was transfected with the K19-EGFP Specifically, a log-rank test revealed that K19 expression, high reporter vector (Supplementary Fig. S3B), and double staining of preoperative total bilirubin concentrations (>1.0 mg/dL), tumor K19 and GFP confirmed that our selective GFP-labeling method size (5 cm), poor tumor differentiation, and microvascular resulted in successful (efficiency of >95%) marking of K19-expressing cells (Supplementary Fig. S3C and S3D). As indicated by our invasion were associated with worse RFS rates (Supplementary þ Table S3). Regarding OS rates, K19 expression, high preoperative FACS results, the ratio of K19 cells differed among the cell lines. aspartate aminotransferase concentrations (40 IU/L), tumor Specifically, 20.6% 3.9% of the Huh7 cells and 14.8% 2.7% size (5 cm), the presence of multiple tumors, poor tumor of the PLC/PRF/5 cells (n ¼ 3) exhibited EGFP expression (Sup- plementary Fig. S3E). PCR analyses confirmed that the sorted differentiation, and microvascular invasion were detected by þ log-rank test (Supplementary Table S3). Furthermore, multivar- EGFP and EGFP cells contained equal copy numbers of the reporter gene (Supplementary Fig. S3F). iate analysis demonstrated that K19 expression was an indepen- þ As we previously reported (12), FACS-isolated single K19 cells dent predictor of both postoperative recurrence and lower OS þ generated both K19 and K19 cell fractions during single-cell rates (Table 1). culture analyses. In contrast, single K19 cells produced only a fi K19 cell fraction (Supplementary Fig. S3G). Therefore, we fur- Ef cacy of SUV, TNR, and preoperative factors for the þ ther analyzed the K19 and K19 populations derived from a evaluation of K19 expression in hepatocellular carcinoma þ Our finding that K19 expression is significantly correlated with single K19 cell. GLUT1 expression in hepatocellular carcinoma surgical speci- 18 þ mens prompted us to examine whether SUV and TNR could be F-FDG accumulation in K19 and K19 hepatocellular utilized to identify K19 expression in human hepatocellular carcinoma cells carcinomas. First, consistent with the results of our previous study, To test the correlation between K19 expression and SUV/TNR in þ fi hepatocellular carcinoma patients, we first examined 18F-FDG (16) GLUT1 patients exhibited signi cantly higher SUV and TNR þ uptake in K19 and K19 hepatocellular carcinoma cells. The than GLUT1 patients (data not shown). Subsequently, as shown þ fi kinetics of 18F-FDG uptake in K19 and K19 Huh7 cells indi- in Fig. 2A, we found that SUV and TNR were signi cantly higher in þ þ cated that 18F-FDG uptake was significantly higher in K19 Huh7 K19 (SUV, 7.6 3.5; TNR, 2.8 0.8) than in K19 patients þ cells than in K19 Huh7 cells (Fig. 3A). In addition, in K19 Huh7 cells, siRNA-based K19 knockdown significantly decreased the Table 1. Multivariate analysis of factors predicting postoperative prognosis 18F-FDG uptake (Fig. 3B). On the other hand, K19 overexpression Postoperative recurrence in K19 Huh7 cells consistently and significantly increased the Variable HR (95% CI) P 18F-FDG uptake (Fig. 3C). Meanwhile, treatment with the TGFbR1 – K19 expression 4.06 (1.82 9.04) 0.001 inhibitor LY2157299 resulted in significant suppression of FDG Total bilirubin (>1.0 mg/dL) 1.55 (0.85–2.82) 0.154 þ Tumor size (5 cm) 1.50 (0.82–2.74) 0.190 accumulation in K19 Huh7 cells (Fig. 3D). Similar results were – fi Poorly differentiated 1.48 (0.82–2.70) 0.197 obtained in PLC/PRF/5 cells (Fig. 3A D). These ndings strongly 18 Microvascular invasion 1.35 (0.74–2.46) 0.332 suggest that K19 functions as a regulator of F-FDG accumulation in hepatocellular carcinoma cells and that evaluation of 18F-FDG Postoperative survival K19 expression 5.67 (2.39–13.44) <0.001 accumulation might therefore be useful for monitoring the þ b Poorly differentiated 2.08 (1.04–4.15) 0.038 response of K19 HCCs to TGF R1 inhibitors. Tumor number (multiple) 1.48 (0.79–2.78) 0.220 Microvascular invasion 1.33 (0.71–2.49) 0.379 K19 regulates GLUT1 expression through TGFb/Smad AST activity (40 IU/L) 1.29 (0.70–2.37) 0.415 signaling Tumor size (5 cm) 0.83 (0.42–1.65) 0.602 To elucidate the mechanism underlying K19 and 18F-FDG fi Abbreviations: AST, aspartate aminotransferase; CI, con dence interval. uptake, we focused on TGFb/Smad signaling, a signaling pathway

1454 Clin Cancer Res; 23(6) March 15, 2017 Clinical Cancer Research

Identiﬁcation of K19þ HCC-CSCs Using 18F-FDG-PET

Figure 2. Relationship between K19 expression and 18F-FDG accumulation in human hepatocellular carcinoma. A, SUV and TNR in K19/K19 hepatocellular carcinoma patients. K19þ patients had significantly higher SUV and TNR than K19 patients (Mann–Whitney U test, , P < 0.01). B, Preoperative serum levels of AFP and protein induced by vitamin K absence 2 (PIVKA-II) in K19/K19 hepatocellular carcinoma patients (Mann–Whitney U test, n.s., not significant). C, ROC curve evaluating the performance of SUV and TNR for predicting K19 expression in hepatocellular carcinoma. Note the high AUC, particularly for the TNR values. D, ROC curve evaluating the performance of preoperative serum levels of AFP and PIVKA-II for predicting K19 expression in hepatocellular carcinoma. Each line in the left panel indicates median levels with 95% confidence interval (CI; A and B). involved in CSC maintenance. As we previously showed, TGFb/ For these analyses, the Huh7 and PLC/PRF/5 cell lines were Smad signaling is activated at the steady state and suppressed by utilized due to their expression of K19, GLUT1, and Sp1 (Fig. 4A). þ K19 knockdown in K19 HCC-CSCs (12). Meanwhile, previous Compared with the K19 population, Sp1 and GLUT1 were þ studies demonstrated that the transcription factor Sp1, a down- detected at significantly higher levels in K19 Huh7 and PLC/ stream target of TGFb/Smad signaling, binds and transactivates PRF/5 cells (Fig. 4B). To assess whether K19 regulates GLUT1 the GLUT1 promoter (29, 30). A separate study reported that the expression through TGFb/Smad signaling, we performed gain/ þ activation of TGFb/Smad signaling plays a central role in glucose- loss of K19 function experiments. In K19 Huh7 and PLC/PRF/5 induced cell hypertrophy in fibroblasts and epithelial cells (31). cells, siRNA-mediated K19 knockdown resulted in decreased These notions prompted us to explore the mechanism governing pSmad2, Sp1, and GLUT1 expression (Fig. 4C and D). Further- the regulation of GLUT1 expression through TGFb/Smad signal- more, K19 overexpression resulted in increased pSmad2, Sp1, and þ ing in K19 HCC-CSCs. GLUT1 expression in K19 Huh7 and PLC/PRF/5 cells (Fig. 4C and D). On the basis of these results, we concluded that K19 regulates 18F-FDG accumulation through TGFb/Smad signaling Table 2. Efficacy of SUV, TNR, and preoperative factors for the evaluation of K19 via a mechanism that includes Sp1 and GLUT1. expression in hepatocellular carcinoma Factor (cut-off value) AUC (95% CI) Sensitivity (%) Specificity (%) Discussion TNR (2.0) 0.89 (0.83–0.96) 82.1 85.7 SUV (5.0) 0.84 (0.72–0.97) 81.0 71.4 Despite advances in treatments, such as surgical resection, liver Tumor size (5 cm) 0.81 (0.72–0.90) 58.3 92.9 transplantation, RFA, and regional/systemic chemotherapy, hepa- 4 m – Plt (10.0 10 / L) 0.70 (0.55 0.86) 26.2 92.9 tocellular carcinoma is associated with one of the poorest prog- – AFP (20 ng/mL) 0.65 (0.44 0.85) 48.8 71.4 fi PIVKA-II (40 mAU/mL) 0.63 (0.47–0.79) 24.4 85.7 noses among carcinomas due to the dif culty in controlling fi T-Bil (1.0 mg/dL) 0.62 (0.47–0.78) 67.9 50.0 tumor recurrence and metastasis. Therefore, the identi cation Alb (3.5 g/dL) 0.50 (0.31–0.68) 78.6 21.4 and clinical application of CSCs, which are involved in tumor Abbreviations: Alb, albumin; CI, confidence interval; PIVKA-II, protein induced by recurrence/metastasis, contribute to the improvement of hepato- vitamin K absence 2; Plt, platelet; T-Bil, total bilirubin. cellular carcinoma treatment outcomes. We previously reported

www.aacrjournals.org Clin Cancer Res; 23(6) March 15, 2017 1455

Kawai et al.

Figure 3. 18F-FDG accumulation in K19þ and K19 hepatocellular carcinoma cells. The kinetics of 18F-FDG uptake in K19þ and K19 Huh7 and PLC/PRF/5 cells. A, K19þ cells showed significantly higher 18F-FDG uptake than K19 cells (repeated measures ANOVA, Huh7, , P < 0.01; PLC/PRF/5, , P < 0.01). B,18F-FDG uptake was significantly suppressed by K19 knockdown in K19þ cells (KD-NC, steady-state K19þ cells; KD-K19, K19 knockdown K19þ cells by K19-siRNA; repeated measures ANOVA, Huh7, , P < 0.05; PLC/PRF/5, , P < 0.05). C,18F-FDG uptake was significantly elevated by K19 overexpression in K19 cells (EX-NC, steady-state K19 cells; EX-K19, K19 overexpression K19 cells; repeated-measures ANOVA, Huh7, , P < 0.05; PLC/PRF/5, , P < 0.05). D, TGFbR1 inhibitor LY2157299 significantly suppressed the 18F-FDG accumulation in K19þ cells (repeated measures ANOVA, Huh7, , P < 0.05; PLC/PRF/5, , P < 0.05).

þ that K19 is a new HCC-CSC marker associated with TGFb/Smad evaluated in our prior study. However, patients with K19 hepa- þ signaling and EMT and that K19 HCC-CSCs could be new tocellular carcinomas in this study exhibited poorer outcomes therapeutic targets for TGFb receptor 1 inhibitors. Consistent than those in our previous study, which is likely due to the with previous studies (32–34), K19 expression was found to correlation between K19 expression and tumor size/differentia- comprise an independent prognosticator of poor hepatocellular tion observed in the current, but not the prior analysis. Likewise, a carcinoma outcomes in both the current cohort, as well as those separate study reported that K19 expression is a prognosticator of

1456 Clin Cancer Res; 23(6) March 15, 2017 Clinical Cancer Research

Identiﬁcation of K19þ HCC-CSCs Using 18F-FDG-PET

A B Huh7 * *

K19

Sp1

GLUT1 PLC/PRF/5 * b-Actin *

RT-

C Huh7 PLC/PRF/5 Sp1 GLUT1 Sp1 GLUT1 * * * ** * * *

D Huh7 PLC/PRF/5

pSmad2 pSmad2

Smad2 Smad2

Sp1 Sp1

GLUT1 GLUT1

GAPDH GAPDH

Figure 4. K19 regulates GLUT1 expression through TGFb/Smad signaling in an Sp1-dependent manner in hepatocellular carcinoma cells. A, RT-PCR analysis of K19, Sp1, and GLUT1 expression in hepatocellular carcinoma cell lines. B, qRT-PCR analysis of Sp1 and GLUT1 expression in K19þ and K19 Huh7 and PLC/PRF/5 cells (Student t test, , P < 0.05). Data, mean SD. C, qRT-PCR analysis of Sp1 and GLUT1 expression in gain/loss of K19 function experiments. Sp1 and GLUT1 expression was suppressed by K19 knockdown and rescued by K19 overexpression (Student t test, , P < 0.05). Data, mean SD. D, Western blot analysis of cells used in gain/loss of K19 function experiments revealed that K19 regulates GLUT1 expression thorough TGFb/Smad signaling and its downstream transcription factor Sp1. KD-NC, control knockdown; KD-K19, K19 knockdown; EX-NC, control overexpression; EX-K19, K19 overexpression.

www.aacrjournals.org Clin Cancer Res; 23(6) March 15, 2017 1457

Kawai et al.

poor outcomes, is associated with tumor size/differentiation, and carcinoma. In the current study, we used K19 promoter–driven þ plays a key role in hepatocellular carcinoma invasion (35).For the EGFP-marked cells to isolate K19 populations of human þ clinical application of these highly malignant K19 HCC-CSCs as hepatocellular carcinoma cell lines. Our findings demonstrate þ a new therapeutic target, the development of tools for non/less- that K19 cells exhibited significantly higher 18F-FDG uptake invasive assessment of K19 expression in hepatocellular carcino- and significantly higher expression levels of Sp1, a downstream ma is essential. transcription factor of TGFb/Smad signaling, and GLUT1. In the clinical field of cancer treatment, 18F-FDG-PET has Moreover, our gain/loss of K19 function experiments clearly been established as a relatively noninvasive diagnostic tool for showed that K19 regulates 18F-FDG uptake through TGFb/ detecting/staging of cancers and for assessing the therapeutic Smad signaling pathway and that this regulation occurred in efficacy of chemotherapies for many cancers. This image inspec- an Sp1/GLUT1–dependent manner. We also observed that FDG þ tion technique can be used to evaluate glucose metabolism in accumulation was significantly suppressed in K19 cells treated vivo by measuring the uptake of the glucose analogue FDG. with the TGFbR1 inhibitor LY2157299, suggesting the potential Besides, GLUT1 expression is known to be a major factor that clinical application for monitoring the response of this com- affects FDG uptake (28). As for the role of 18F-FDG-PET in pound. Besides, among the human hepatocellular carcinoma þ hepatocellular carcinoma, we previously found that although specimens examined in this study, less than half of the GLUT1 the frequency of GLUT1 expression is low, GLUT1 expression is tumors exhibited K19 expression. GLUT1 expression was pre- correlated with elevated 18F-FDG accumulation in hepatocel- viously reported to be regulated by Notch signaling in breast lular carcinoma and that hepatocellular carcinoma with high cancer (39, 40). Given the significance of Notch signaling in TNR values (2.0) exhibits more aggressive malignancy poten- hepatocellular carcinoma (41), it is conceivable that this sig- tial (16–18). naling pathway is also involved in the regulation of GLUT1 Notably, our analyses using human samples showed the expression in these cells. relevance of K19 expression to 18F-FDG accumulation. By Various studies regarding the role of K19 in hepatocellular þ analyzing 98 hepatocellular carcinoma surgical specimens, we carcinoma show the aggressive properties of K19 hepatocel- þ demonstrated that patients with K19 tumors exhibited signif- lular carcinoma and prompt us to develop new therapeutic icantly higher SUV and TNR values and a higher incidence of strategies for it. Accompanied with the efficacy of TGFb receptor þ GLUT1expressionthanthosewithK19 tumors. Moreover, it 1 inhibitor against K19 HCC-CSCs, detecting K19 expression should be noted that among the preoperative clinical factors, using 18F-FDG-PET study before hepatocellular carcinoma þ TNR is the most sensitive predictor of K19 expression in treatment aids to enable individualized medicine for K19 hepa- hepatocellular carcinoma. In the current study, TNR was more tocellular carcinoma patients. Further prospective clinical trials sensitive than SUV for the prediction of K19 expression. Dif- examining whether 18F-FDG-PET comprises an effective method þ ferentiation between tumor tissue and physiologic noise in the for identifying K19 hepatocellular carcinoma patients, and for liver by 18F-FDG-PET can sometimes be difficult, as normal assessing the therapeutic efficacy of TGFb receptor 1 inhibitors in þ liver tissue exhibits heterogeneity in 18F-FDG uptake, and K19 hepatocellular carcinoma patients, will advance the clinical þ physiologic noise is therefore apparent. Background nontumor application of K19 HCC-CSCs. liver uptake varied to some extent, and the contrast between In conclusion, the results of our study indicate that K19 the liver tumor and background liver noise may be dependent regulates 18F-FDG uptake in an Sp1/GLUT1–dependent man- on the extent of tumor uptake itself as well as on the general ner through the TGFb/Smad signaling pathway and that 18F- levels of liver background. For these reasons, TNR would be FDG-PET is an effective method for identifying K19 expression expected to provide a better predictive value than SUV in this in hepatocellular carcinoma tissues. We believe that further þ study.Besides,inthisstudy,we included only patients with study of K19 HCC-CSCs and 18F-FDG-PET will contribute to blood glucose levels <150 mg/dL and CRP levels <5.0 mg/dL for the development of novel approaches for the treatment of 18F-FDG-PET analysis. However, no significant differences in hepatocellular carcinoma. clinical–pathologic backgrounds, RFS, or OS were observed between patients of the current study and those of our previous Disclosure of Potential Conflicts of Interest study (data not shown), which indicates that there was no No potential conflicts of interest were disclosed. apparent selection bias in this study. In clinical setting, CT and/ or MRI are routinely used for the diagnosis and monitoring of hepatocellular carcinoma. By combining 18F-FDG-PET with CT Authors' Contributions and/or MRI, a more precise prediction of K19 expression in Conception and design: T. Kawai, K. Yasuchika, T. Higashi, Y. Miyauchi, hepatocellular carcinoma might be achieved. S. Ogiso On the other hand, given the pivotal functions of various Development of methodology: T. Kawai, K. Yasuchika, T. Higashi signaling pathways in the maintenance and proliferation of Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): T. Kawai, S. Seo, K. Yasuda, E. Hatano embryonic stem/progenitor cells, including the Notch, Wnt/ Analysis and interpretation of data (e.g., statistical analysis, biostatistics, b-catenin, and TGFb/Smad signaling pathways, it is feasible computational analysis): T. Kawai, K. Yasuchika, T. Ishii, S. Ogiso, that such pathways also function in CSCs (36–38). Indeed, we K. Fukumitsu þ previously demonstrated that K19 HCC-CSCs, in which TGFb/ Writing, review, and/or revision of the manuscript: T. Kawai, K. Yasuchika, Smad signaling is active under steady-state conditions, exhibit S. Ogiso, E. Hatano high proliferative activity and a tendency to cause EMT (12). Administrative, technical, or material support (i.e., reporting or organizing fi data, constructing databases): T. Higashi, Y. Miyauchi, H. Kojima, R. Yamaoka, These ndings highlight the necessity for further investigation H. Katayama, E.Y. Yoshitoshi, S. Ogiso, S. Kita, K. Fukumitsu, Y. Nakamoto into the mechanism by which TGFb/Smad signaling contri- Study supervision: K. Yasuchika, T. Ishii, Y. Miyauchi, H. Kojima, R. Yamaoka, þ butes to the clinical application of K19 CSCs in hepatocellular H. Katayama, E.Y. Yoshitoshi, S. Ogiso, K. Fukumitsu, E. Hatano, S. Uemoto

1458 Clin Cancer Res; 23(6) March 15, 2017 Clinical Cancer Research

Identiﬁcation of K19þ HCC-CSCs Using 18F-FDG-PET

Acknowledgments The costs of publication of this article were defrayed in part by the The authors wish to thank Dr. Makiko Kagaya and Professor Shuichi Kaneko payment of page charges. This article must therefore be hereby marked advertisement (Kanazawa University, Kanazawa, Japan) for providing the plasmid vector in accordance with 18 U.S.C. Section 1734 solely to indicate pHCK-2952. this fact.

Grant Support This work was supported by grants from the Scientiﬁc Research Fund of the Received April 5, 2016; revised August 8, 2016; accepted September 8, 2016; Japan Science and Technology Agency (research project number: 24659606). published OnlineFirst September 23, 2016.

References 1.LlovetJM,Zucman-RossiJ,PikarskyE,SangroB,SchwartzM,Sher- 20. Ishii T, Yasuchika K, Machimoto T, Kamo N, Komori J, Konishi S, et al. man M, et al. Hepatocellular carcinoma. Nat Rev Dis Primers 2016; Transplantation of embryonic stem cell-derived endodermal cells into 2:16018. mice with induced lethal liver damage. Stem Cells 2007;25:3252–60. 2. Clarke MF, Dick JE, Dirks PB, Eaves CJ, Jamieson CH, Jones DL, et al. Cancer 21. Kagaya M, Kaneko S, Ohno H, Inamura K, Kobayashi K. Cloning stem cells–perspectives on current status and future directions: AACR and characterization of the 50-flanking region of human cytokeratin Workshop on cancer stem cells. Cancer Res 2006;66:9339–44. 19 gene in human cholangiocarcinoma cell line. J Hepatol 2001; 3. Pardal R, Clarke MF, Morrison SJ. Applying the principles of stem-cell 35:504–11. biology to cancer. Nat Rev Cancer 2003;3:895–902. 22. Ishii T, Yasuchika K, Suemori H, Nakatsuji N, Ikai I, Uemoto S. Alpha- 4. Dean M, Fojo T, Bates S. Tumour stem cells and drug resistance. Nat Rev fetoprotein producing cells act as cancer progenitor cells in human cho- Cancer 2005;5:275–84. langiocarcinoma. Cancer Lett 2010;294:25–34. 5. Phillips TM, McBride WH, Pajonk F. The response of CD24(-/low)/ 23. Sasaki N, Ishii T, Kamimura R, Kajiwara M, Machimoto T, Nakatsuji N, et al. CD44þ breast cancer-initiating cells to radiation. J Natl Cancer Inst Alpha-fetoprotein-producing pancreatic cancer cells possess cancer stem 2006;98:1777–85. cell characteristics. Cancer Lett 2011;308:152–61. 6. Pattabiraman DR, Weinberg RA. Tackling the cancer stem cells - what 24. Ishii T, Fukumitsu K, Yasuchika K, Adachi K, Kawase E, Suemori H, et al. challenges do they pose? Nat Rev Drug Discov 2014;13:497–512. Effects of extracellular matrixes and growth factors on the hepatic differ- 7. Yamashita T, Ji J, Budhu A, Forgues M, Yang W, Wang HY, et al. EpCAM- entiation of human embryonic stem cells. Am J Physiol Gastrointest Liver positive hepatocellular carcinoma cells are tumor-initiating cells with Physiol 2008;295:G313–21. stem/progenitor cell features. Gastroenterology 2009;136:1012–24. 25. Yamanaka K, Hatano E, Narita M, Kitamura K, Yanagida A, Asechi H, et al. 8. Ma S, Chan KW, Hu L, Lee TK, Wo JY, Ng IO, et al. Identification and Olprinone attenuates excessive shear stress through up-regulation of characterization of tumorigenic liver cancer stem/progenitor cells. Gastro- endothelial nitric oxide synthase in a rat excessive hepatectomy model. enterology 2007;132:2542–56. Liver Transpl 2011;17:60–9. 9. Ma S, Tang KH, Chan YP, Lee TK, Kwan PS, Castilho A, et al. miR-130b 26. Pedersen MW, Holm S, Lund EL, Højgaard L, Kristjansen PE. Coregulation Promotes CD133(þ) liver tumor-initiating cell growth and self-renewal of glucose uptake and vascular endothelial growth factor (VEGF) in two via tumor protein 53-induced nuclear protein 1. Cell Stem Cell small-cell lung cancer (SCLC) sublines in vivo and in vitro. Neoplasia 2010;7:694–707. 2001;3:80–7. 10. Yang ZF, Ho DW, Ng MN, Lau CK, Yu WC, Ngai P, et al. Significance of 27. Brown RS, Goodman TM, Zasadny KR, Greenson JK, Wahl RL. Expression CD90þ cancer stem cells in human liver cancer. Cancer Cell 2008;13: of hexokinase II and Glut-1 in untreated human breast cancer. Nucl Med 153–66. Biol 2002;29:443–53. 11. Oikawa T, Kamiya A, Zeniya M, Chikada H, Hyuck AD, Yamazaki Y, et al. 28. Higashi T, Saga T, Nakamoto Y, Ishimori T, Mamede MH, Wada M, et al. Sal-like protein 4 (SALL4), a stem cell biomarker in liver cancers. Hepatol- Relationship between retention index in dual-phase (18)F-FDG PET, and ogy 2013;57:1469–83. hexokinase-II and glucose transporter-1 expression in pancreatic cancer. 12. Kawai T, Yasuchika K, Ishii T, Katayama H, Yoshitoshi EY, Ogiso S, et al. J Nucl Med 2002;43:173–80. Keratin 19, a cancer stem cell marker in human hepatocellular carcinoma. 29. Santalucía T, Boheler KR, Brand NJ, Sahye U, Fandos C, Vinals~ F, et al. Clin Cancer Res 2015;21:3081–91. Factors involved in GLUT-1 glucose transporter gene transcription in 13. Rigo P, Paulus P, Kaschten BJ, Hustinx R, Bury T, Jerusalem G, cardiac muscle. J Biol Chem 1999;274:17626–34. et al. Oncological applications of positron emission tomography 30. Vinals~ F, Fandos C, Santalucia T, Ferre J, Testar X, Palacín M, et al. with fluorine-18 fluorodeoxyglucose. Eur J Nucl Med 1996;23: Myogenesis and MyoD down-regulate Sp1. A mechanism for the repres- 1641–74. sion of GLUT1 during muscle cell differentiation. J Biol Chem 14. Stokkel MP, ten Broek FW, van Rijk PP. The role of FDG PET in the 1997;272:12913–21. clinical management of head and neck cancer. Oral Oncol 1998; 31. Wu L, Derynck R. Essential role of TGF-beta signaling in glucose-induced 34:466–71. cell hypertrophy. Dev Cell 2009;17:35–48. 15. Eary JF, Conrad EU. Positron emission tomography in grading soft tissue 32. Uenishi T, Kubo S, Yamamoto T, Shuto T, Ogawa M, Tanaka H, et al. sarcomas. Semin Musculoskelet Radiol 1999;3:135–8. Cytokeratin 19 expression in hepatocellular carcinoma predicts early 16. SeoS,HatanoE,HigashiT,NakajimaA,NakamotoY,TadaM,etal.P- postoperative recurrence. Cancer Sci 2003;94:851–7. glycoprotein expression affects 18F-fluorodeoxyglucose accumulation 33. Durnez A, Verslype C, Nevens F, Fevery J, Aerts R, Pirenne J, et al. The in hepatocellular carcinoma in vivo and in vitro. Int J Oncol 2009; clinicopathological and prognostic relevance of cytokeratin 7 and 19 34:1303–12. expression in hepatocellular carcinoma. A possible progenitor cell origin. 17. Seo S, Hatano E, Higashi T, Hara T, Tada M, Tamaki N, et al. Histopathology 2006;49:138–51. Fluorine-18 fluorodeoxyglucose positron emission tomography pre- 34. Kim H, Choi GH, Na DC, Ahn EY, Kim GI, Lee JE, et al. Human hepato- dicts tumor differentiation, P-glycoprotein expression, and outcome cellular carcinomas with "Stemness"-related marker expression: keratin 19 after resection in hepatocellular carcinoma. Clin Cancer Res 2007; expression and a poor prognosis. Hepatology 2011;54:1707–17. 13:427–33. 35. Govaere O, Komuta M, Berkers J, Spee B, Janssen C, de Luca F, et al. Keratin 18. Kitamura K, Hatano E, Higashi T, Seo S, Nakamoto Y, Yamanaka K, et al. 19: a key role player in the invasion of human hepatocellular carcinomas. Preoperative FDG-PET predicts recurrence patterns in hepatocellular car- Gut 2014;63:674–85. cinoma. Ann Surg Oncol 2012;19:156–62. 36. Rizzo P, Osipo C, Foreman K, Golde T, Osborne B, Miele L. Rational 19. Ishii T, Yasuchika K, Fujii H, Hoppo T, Baba S, Naito M, et al. In vitro targeting of Notch signaling in cancer. Oncogene 2008;27:5124–31. differentiation and maturation of mouse embryonic stem cells into hepa- 37. Klaus A, Birchmeier W. Wnt signalling and its impact on development and tocytes. Exp Cell Res 2005;309:68–77. cancer. Nat Rev Cancer 2008;8:387–98.

www.aacrjournals.org Clin Cancer Res; 23(6) March 15, 2017 1459

Kawai et al.

38. Mishra L, Shetty K, Tang Y, Stuart A, Byers SW. The role of TGF-beta and 40. Landor SK, Mutvei AP, Mamaeva V, Jin S, Busk M, Borra R, et al. Wnt signaling in gastrointestinal stem cells and cancer. Oncogene 2005;24: Hypo- and hyperactivated Notch signaling induce a glycolytic switch 5775–89. through distinct mechanisms. Proc Natl Acad Sci U S A 2011;108: 39. Efferson CL, Winkelmann CT, Ware C, Sullivan T, Giampaoli S, Tammam J, 18814–9. et al. Downregulation of Notch pathway by a gamma-secretase inhibitor 41. Villanueva A, Alsinet C, Yanger K, Hoshida Y, Zong Y, Toffanin S, attenuates AKT/mammalian target of rapamycin signaling and glucose et al. Notch signaling is activated in human hepatocellular carcinoma uptake in an ERBB2 transgenic breast cancer model. Cancer Res 2010; and induces tumor formation in mice. Gastroenterology 2012;143: 70:2476–84. 1660–9.e7.

1460 Clin Cancer Res; 23(6) March 15, 2017 Clinical Cancer Research

Identification of Keratin 19−Positive Cancer Stem Cells Associating Human Hepatocellular Carcinoma Using 18F-Fluorodeoxyglucose Positron Emission Tomography

Takayuki Kawai, Kentaro Yasuchika, Satoru Seo, et al.

Clin Cancer Res 2017;23:1450-1460. Published OnlineFirst September 23, 2016.

Updated version Access the most recent version of this article at: doi:10.1158/1078-0432.CCR-16-0871

Supplementary Access the most recent supplemental material at: Material http://clincancerres.aacrjournals.org/content/suppl/2016/09/23/1078-0432.CCR-16-0871.DC1

Cited articles This article cites 41 articles, 9 of which you can access for free at: http://clincancerres.aacrjournals.org/content/23/6/1450.full#ref-list-1

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

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/23/6/1450. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.